repo
stringlengths
2
99
file
stringlengths
13
225
code
stringlengths
0
18.3M
file_length
int64
0
18.3M
avg_line_length
float64
0
1.36M
max_line_length
int64
0
4.26M
extension_type
stringclasses
1 value
StyleFusion
StyleFusion-master/src/shared.py
import os, random, sys, datetime, time, socket, io, h5py, argparse, shutil, io import queue import numpy as np import scipy from nltk.translate.bleu_score import sentence_bleu from nltk.translate.bleu_score import SmoothingFunction import matplotlib.pyplot as plt """ AUTHOR: Sean Xiang Gao (xiag@microsoft.com) at Microsoft Research """ SOS_token = '_SOS_' EOS_token = '_EOS_' UNK_token = '_UNK_' # just for arxiv EQN_token = '_eqn_' CITE_token = '_cite_' IX_token = '_ix_' hostname = socket.gethostname() SKIP_VIS = hostname not in ['MININT-3LHNLKS', 'xiag-0228'] FIT_VERBOSE = 1 DATA_PATH = 'data/' OUT_PATH = 'out/' print('@'*20) print('hostname: %s'%hostname) print('data_path: %s'%DATA_PATH) print('out_path: %s'%OUT_PATH) print('@'*20) PHILLY = False BATCH_SIZE = 128#256 parser = argparse.ArgumentParser() parser.add_argument('--data_name', type=str, default='toy') parser.add_argument('--batch_size', type=int, default=100) parser.add_argument('--max_n_trained', type=int, default=int(10e6)) parser.add_argument('--lr', type=float, default=3e-4) parser.add_argument('--token_embed_dim', type=int, default=1000) parser.add_argument('--rnn_units', type=int, default=1000) parser.add_argument('--encoder_depth', type=int, default=2) parser.add_argument('--decoder_depth', type=int, default=2) parser.add_argument('--stddev', type=float, default=0.1) parser.add_argument('--wt_dist', type=float, default=1.) parser.add_argument('--debug','-d', action='store_true') parser.add_argument('--max_ctxt_len', type=int, default=90) parser.add_argument('--max_resp_len', type=int, default=30) parser.add_argument('--fld_suffix', type=str, default='') parser.add_argument('--conv_mix_ratio', type=float, default=0.0) parser.add_argument('--nonc_mix_ratio', type=float, default=1.0) parser.add_argument('--clf_name', type=str, default='holmes') parser.add_argument('--model_class', type=str, default='fuse') parser.add_argument('--restore', type=str, default='') parser.add_argument('--noisy_vocab', type=int, default=-1) parser.add_argument('--reld', action='store_true') parser.add_argument('--ablation', '-abl', action='store_true') parser.add_argument('--path_test', type=str) if hostname == 'MININT-3LHNLKS': parser.add_argument('--cpu_only', '-c', action='store_true') def reset_rand(RAND_SEED=9): random.seed(RAND_SEED) np.random.seed(RAND_SEED) reset_rand() def str2bool(s): return {'true':True, 'false':False}[s.lower()] def strmap(x): if 'nan' in str(x): return 'nan' if isinstance(x, str): return x if int(x) == x: return '%i'%x return '%.4f'%x def int2str(i): if i < 1000: return str(i) else: k = i/1000 if int(k) == k: return '%ik'%k else: return '%.1fk'%k def makedirs(fld): if not os.path.exists(fld): os.makedirs(fld) def rand_latent(center, r, limit=True): if r == 0: return center noise = np.random.normal(size=center.shape) r_raw = np.sqrt(np.sum(np.power(noise, 2))) sampled = center + noise/r_raw*r if limit: return np.minimum(1, np.maximum(-1, sampled)) else: return sampled def calc_nltk_bleu(ref, hyp, max_ngram=4, smoothing_function=None): return sentence_bleu( [ref.split()], hyp.split(), weights=[1./max_ngram]*max_ngram, smoothing_function=smoothing_function, ) def calc_nltk_bleu_smoothed(ref, hyp, max_ngram=4): return calc_nltk_bleu(ref, hyp, max_ngram, smoothing_function=SmoothingFunction().method7) def euc_dist(a, b): # Euclidean distance if len(a.shape) == 1: return np.sqrt(np.sum(np.power(a - b, 2))) return np.sqrt(np.sum(np.power(a - b, 2), axis=1)) def now(): return datetime.datetime.now()
3,632
25.326087
78
py
StyleFusion
StyleFusion-master/data/arXiv/arxiv.py
""" AUTHOR: Xiang Gao (xiag@microsoft.com) at Microsoft Research """ import re, os, subprocess from nltk.tokenize import TweetTokenizer EQN_token = '_eqn_' CITE_token = '_cite_' IX_token = '_ix_' MAX_UTT_LEN = 30 # maximum length of utterance allowed. if longer, ignore def norm_sentence(txt): txt = txt.lower() # url and tag words = [] for word in txt.lower().split(): if word[0] == '#': # don't allow tag continue i = word.find('http') if i >= 0: word = word[:i] + ' ' + '__url__' words.append(word.strip()) txt = ' '.join(words) # remove illegal char txt = txt.replace(chr(92),'') # chr(92) = '\'. as twitter has 'b\/c' rather than 'b/c' txt = txt.replace("b/c","because").replace('j/k','just kidding').replace('w/o','without').replace('w/','with') txt = re.sub('__mention__','MENTION',txt) txt = re.sub('__url__','URL',txt) txt = re.sub(r"[^A-Za-z0-9():,.!?_'“” ]", " ", txt) txt = re.sub('MENTION','__mention__',txt) txt = re.sub('URL','__url__',txt) # contraction add_space = ["'s", "'m", "'re", "n't", "'ll","'ve","'d","'em"] tokenizer = TweetTokenizer(preserve_case=False) txt = ' ' + ' '.join(tokenizer.tokenize(txt)) + ' ' txt = txt.replace(" won't ", " will n't ") txt = txt.replace(" can't ", " can n't ") for a in add_space: txt = txt.replace(a+' ', ' '+a+' ') # remove un-necessary space return ' '.join(txt.split()) def arxiv_del_bib(path): lines = [] for line in open(path, encoding='utf-8', errors='ignore'): stop = False for c in ['begin{references}', 'begin{enumerate}', 'begin{thebibliography}']: if c in line: stop = True if stop: lines.append(chr(92) + 'end{document}\n\n') break else: lines.append(line.strip('\n')) with open(path+'.delbib', 'w', encoding='utf-8') as f: f.write('\n'.join(lines)) def arxiv_pandoc(fld): # preprocess arxiv latex file with pandoc # http://www.cs.cornell.edu/projects/kddcup/datasets.html # http://pandoc.org/index.html n = 0 for fname in os.listdir(fld): if '.' in fname: continue path = fld + '/' + fname arxiv_del_bib(path) cmd = [ 'pandoc', '-f', 'latex', '-o', path + '_pandoc.txt', path + '.delbib' ] process = subprocess.Popen(cmd, stdout=subprocess.PIPE) process.communicate() print('='*10) print(path) n += 1 #if n == 100: # break def arxiv_paragraph(path): def lines2paragraph(lines): p = ' '.join(lines) if len(p) == 0: return None if not p[0].isalpha(): return None return arxiv_clean(p) paragraphs = [] lines = [] for line in open(path, encoding='utf-8', errors='ignore'): line = line.strip('\n').strip() if len(line) == 0: paragraph = lines2paragraph(lines) if paragraph is not None: paragraphs.append(paragraph) lines = [] else: if len(re.sub(r"[^A-Za-z0-9]", "", line)) > 0: lines.append(line) paragraph = lines2paragraph(lines) if paragraph is not None: paragraphs.append(paragraph) with open(path+'.paragraph','w', encoding='utf-8') as f: f.write('\n'.join(paragraphs)) def arxiv_clean(p): # deal with equations and citations is_math = False i = 0 s = '' while i < len(p): flag = False if p[i] == '$': flag = True if i + 1 < len(p) and p[i+1] == '$': i = i+1 if flag: if not is_math: s += ' %s '%EQN_token is_math = not is_math elif not is_math: s += p[i] i += 1 ww = [] for w in s.split(): if w == EQN_token: if len(ww) > 0 and ww[-1] == EQN_token: continue if len(ww) > 1 and ww[-1] == '.' and ww[-2] == EQN_token: continue if chr(92)+'[' in w or chr(92) + ']' in w or '[@' in w or w.startswith('@'): # citation w_ = CITE_token if w[-1]!=']': w_ += w[-1] w = w_ if len(ww) > 0 and ww[-1].startswith(CITE_token): continue else: if w[0] == '(' and len(w) > 1 and w[1].isnumeric(): w_ = IX_token if w[-1]!=')': w_ += w[-1] w = w_ w = w.replace('[**','').replace('[*','').replace('**]','').replace('*]','') ww.append(w) return ' '.join(ww) def arxiv_paragraph_all(fld): n = 0 for fname in os.listdir(fld): if fname.endswith('_pandoc.txt'): path = fld + '/' + fname print(path) arxiv_paragraph(path) n += 1 def arxiv_utts(path): utts = [] for p in open(path, encoding='utf-8'): p = p.strip('\n').replace(chr(92),'. ') for utt in p.split('. '): utt += '.' alpha = re.sub(r"[^a-z]", "", utt.lower()) if len(alpha) < 5: continue utt = norm_sentence(utt) if len(utt.split()) > MAX_UTT_LEN: continue utts.append(utt) with open(path+'.utt', 'w', encoding='utf-8') as f: f.write('\n'.join(utts)) return utts def arxiv_utts_all(fld): utts = [] n = 0 for fname in os.listdir(fld): if fname.endswith('.paragraph'): path = fld + '/' + fname print(path) utts_ = arxiv_utts(path) for utt in utts_: if len(utt) > 10: utts.append(utt) utts = sorted(utts) with open(fld + '/../all.utt', 'w', encoding='utf-8') as f: f.write('\n'.join(utts)) def arxiv_filter(path): # make sure: 1) starts with some word, 2) longer than some min lines = [] for line in open(path, encoding='utf-8'): ww = line.strip('\n').replace('“','"').replace('”','"').split() i0 = None for i in range(len(ww)): alpha = re.sub(r"[^a-z]", "", ww[i].lower()) if len(alpha) > 1: i0 = i break if i0 is None: continue ww = ww[i0:] if len(re.sub(r"[^a-z]", "", ''.join(ww))) > 5: lines.append(' '.join(ww)) with open(path+'.filtered', 'w', encoding='utf-8') as f: f.write('\n'.join(lines)) if __name__ == '__main__': years = range(1998, 2002+1) for year in years: fld = 'hep-th-%i/%i/'%(year, year) arxiv_pandoc(fld) arxiv_paragraph_all(fld) arxiv_utts_all(fld) for year in years: print(year) arxiv_filter('hep-th-%i/all.utt'%(year))
5,797
22.860082
111
py
NBFNet
NBFNet-master/script/run.py
import os import sys import math import pprint import torch from torchdrug import core from torchdrug.utils import comm sys.path.append(os.path.dirname(os.path.dirname(__file__))) from nbfnet import dataset, layer, model, task, util def train_and_validate(cfg, solver): if cfg.train.num_epoch == 0: return step = math.ceil(cfg.train.num_epoch / 10) best_result = float("-inf") best_epoch = -1 for i in range(0, cfg.train.num_epoch, step): kwargs = cfg.train.copy() kwargs["num_epoch"] = min(step, cfg.train.num_epoch - i) solver.model.split = "train" solver.train(**kwargs) solver.save("model_epoch_%d.pth" % solver.epoch) solver.model.split = "valid" metric = solver.evaluate("valid") result = metric[cfg.metric] if result > best_result: best_result = result best_epoch = solver.epoch solver.load("model_epoch_%d.pth" % best_epoch) return solver def test(cfg, solver): solver.model.split = "valid" solver.evaluate("valid") solver.model.split = "test" solver.evaluate("test") if __name__ == "__main__": args, vars = util.parse_args() cfg = util.load_config(args.config, context=vars) working_dir = util.create_working_directory(cfg) torch.manual_seed(args.seed + comm.get_rank()) logger = util.get_root_logger() if comm.get_rank() == 0: logger.warning("Config file: %s" % args.config) logger.warning(pprint.pformat(cfg)) dataset = core.Configurable.load_config_dict(cfg.dataset) solver = util.build_solver(cfg, dataset) train_and_validate(cfg, solver) test(cfg, solver)
1,690
25.421875
64
py
NBFNet
NBFNet-master/script/visualize.py
import os import sys import pprint import torch from torchdrug import core from torchdrug.utils import comm sys.path.append(os.path.dirname(os.path.dirname(__file__))) from nbfnet import dataset, layer, model, task, util vocab_file = os.path.join(os.path.dirname(__file__), "../data/fb15k237_entity.txt") vocab_file = os.path.abspath(vocab_file) def load_vocab(dataset): entity_mapping = {} with open(vocab_file, "r") as fin: for line in fin: k, v = line.strip().split("\t") entity_mapping[k] = v entity_vocab = [entity_mapping[t] for t in dataset.entity_vocab] relation_vocab = ["%s (%d)" % (t[t.rfind("/") + 1:].replace("_", " "), i) for i, t in enumerate(dataset.relation_vocab)] return entity_vocab, relation_vocab def visualize_path(solver, triplet, entity_vocab, relation_vocab): num_relation = len(relation_vocab) h, t, r = triplet.tolist() triplet = torch.as_tensor([[h, t, r]], device=solver.device) inverse = torch.as_tensor([[t, h, r + num_relation]], device=solver.device) solver.model.eval() pred, (mask, target) = solver.model.predict_and_target(triplet) pos_pred = pred.gather(-1, target.unsqueeze(-1)) rankings = torch.sum((pos_pred <= pred) & mask, dim=-1) + 1 rankings = rankings.squeeze(0) logger.warning("") samples = (triplet, inverse) for sample, ranking in zip(samples, rankings): h, t, r = sample.squeeze(0).tolist() h_name = entity_vocab[h] t_name = entity_vocab[t] r_name = relation_vocab[r % num_relation] if r >= num_relation: r_name += "^(-1)" logger.warning(">>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>") logger.warning("rank(%s | %s, %s) = %g" % (t_name, h_name, r_name, ranking)) paths, weights = solver.model.visualize(sample) for path, weight in zip(paths, weights): triplets = [] for h, t, r in path: h_name = entity_vocab[h] t_name = entity_vocab[t] r_name = relation_vocab[r % num_relation] if r >= num_relation: r_name += "^(-1)" triplets.append("<%s, %s, %s>" % (h_name, r_name, t_name)) logger.warning("weight: %g\n\t%s" % (weight, " ->\n\t".join(triplets))) if __name__ == "__main__": args, vars = util.parse_args() cfg = util.load_config(args.config, context=vars) working_dir = util.create_working_directory(cfg) torch.manual_seed(args.seed + comm.get_rank()) logger = util.get_root_logger() logger.warning("Config file: %s" % args.config) logger.warning(pprint.pformat(cfg)) if cfg.dataset["class"] != "FB15k237": raise ValueError("Visualization is only implemented for FB15k237") dataset = core.Configurable.load_config_dict(cfg.dataset) solver = util.build_solver(cfg, dataset) entity_vocab, relation_vocab = load_vocab(dataset) for i in range(500): visualize_path(solver, solver.test_set[i], entity_vocab, relation_vocab)
3,106
34.306818
84
py
NBFNet
NBFNet-master/nbfnet/layer.py
import torch from torch import nn from torch.nn import functional as F from torch_scatter import scatter_add, scatter_mean, scatter_max, scatter_min from torchdrug import layers from torchdrug.layers import functional class GeneralizedRelationalConv(layers.MessagePassingBase): eps = 1e-6 message2mul = { "transe": "add", "distmult": "mul", } def __init__(self, input_dim, output_dim, num_relation, query_input_dim, message_func="distmult", aggregate_func="pna", layer_norm=False, activation="relu", dependent=True): super(GeneralizedRelationalConv, self).__init__() self.input_dim = input_dim self.output_dim = output_dim self.num_relation = num_relation self.query_input_dim = query_input_dim self.message_func = message_func self.aggregate_func = aggregate_func self.dependent = dependent if layer_norm: self.layer_norm = nn.LayerNorm(output_dim) else: self.layer_norm = None if isinstance(activation, str): self.activation = getattr(F, activation) else: self.activation = activation if self.aggregate_func == "pna": self.linear = nn.Linear(input_dim * 13, output_dim) else: self.linear = nn.Linear(input_dim * 2, output_dim) if dependent: self.relation_linear = nn.Linear(query_input_dim, num_relation * input_dim) else: self.relation = nn.Embedding(num_relation, input_dim) def message(self, graph, input): assert graph.num_relation == self.num_relation batch_size = len(graph.query) node_in, node_out, relation = graph.edge_list.t() if self.dependent: relation_input = self.relation_linear(graph.query).view(batch_size, self.num_relation, self.input_dim) else: relation_input = self.relation.weight.expand(batch_size, -1, -1) relation_input = relation_input.transpose(0, 1) node_input = input[node_in] edge_input = relation_input[relation] if self.message_func == "transe": message = edge_input + node_input elif self.message_func == "distmult": message = edge_input * node_input elif self.message_func == "rotate": node_re, node_im = node_input.chunk(2, dim=-1) edge_re, edge_im = edge_input.chunk(2, dim=-1) message_re = node_re * edge_re - node_im * edge_im message_im = node_re * edge_im + node_im * edge_re message = torch.cat([message_re, message_im], dim=-1) else: raise ValueError("Unknown message function `%s`" % self.message_func) message = torch.cat([message, graph.boundary]) return message def aggregate(self, graph, message): node_out = graph.edge_list[:, 1] node_out = torch.cat([node_out, torch.arange(graph.num_node, device=graph.device)]) edge_weight = torch.cat([graph.edge_weight, torch.ones(graph.num_node, device=graph.device)]) edge_weight = edge_weight.unsqueeze(-1).unsqueeze(-1) degree_out = graph.degree_out.unsqueeze(-1).unsqueeze(-1) + 1 if self.aggregate_func == "sum": update = scatter_add(message * edge_weight, node_out, dim=0, dim_size=graph.num_node) elif self.aggregate_func == "mean": update = scatter_mean(message * edge_weight, node_out, dim=0, dim_size=graph.num_node) elif self.aggregate_func == "max": update = scatter_max(message * edge_weight, node_out, dim=0, dim_size=graph.num_node)[0] elif self.aggregate_func == "pna": mean = scatter_mean(message * edge_weight, node_out, dim=0, dim_size=graph.num_node) sq_mean = scatter_mean(message ** 2 * edge_weight, node_out, dim=0, dim_size=graph.num_node) max = scatter_max(message * edge_weight, node_out, dim=0, dim_size=graph.num_node)[0] min = scatter_min(message * edge_weight, node_out, dim=0, dim_size=graph.num_node)[0] std = (sq_mean - mean ** 2).clamp(min=self.eps).sqrt() features = torch.cat([mean.unsqueeze(-1), max.unsqueeze(-1), min.unsqueeze(-1), std.unsqueeze(-1)], dim=-1) features = features.flatten(-2) scale = degree_out.log() scale = scale / scale.mean() scales = torch.cat([torch.ones_like(scale), scale, 1 / scale.clamp(min=1e-2)], dim=-1) update = (features.unsqueeze(-1) * scales.unsqueeze(-2)).flatten(-2) else: raise ValueError("Unknown aggregation function `%s`" % self.aggregate_func) return update def message_and_aggregate(self, graph, input): if graph.requires_grad or self.message_func == "rotate": return super(GeneralizedRelationalConv, self).message_and_aggregate(graph, input) assert graph.num_relation == self.num_relation batch_size = len(graph.query) input = input.flatten(1) boundary = graph.boundary.flatten(1) degree_out = graph.degree_out.unsqueeze(-1) + 1 if self.dependent: relation_input = self.relation_linear(graph.query).view(batch_size, self.num_relation, self.input_dim) relation_input = relation_input.transpose(0, 1).flatten(1) else: relation_input = self.relation.weight.repeat(1, batch_size) adjacency = graph.adjacency.transpose(0, 1) if self.message_func in self.message2mul: mul = self.message2mul[self.message_func] else: raise ValueError("Unknown message function `%s`" % self.message_func) if self.aggregate_func == "sum": update = functional.generalized_rspmm(adjacency, relation_input, input, sum="add", mul=mul) update = update + boundary elif self.aggregate_func == "mean": update = functional.generalized_rspmm(adjacency, relation_input, input, sum="add", mul=mul) update = (update + boundary) / degree_out elif self.aggregate_func == "max": update = functional.generalized_rspmm(adjacency, relation_input, input, sum="max", mul=mul) update = torch.max(update, boundary) elif self.aggregate_func == "pna": sum = functional.generalized_rspmm(adjacency, relation_input, input, sum="add", mul=mul) sq_sum = functional.generalized_rspmm(adjacency, relation_input ** 2, input ** 2, sum="add", mul=mul) max = functional.generalized_rspmm(adjacency, relation_input, input, sum="max", mul=mul) min = functional.generalized_rspmm(adjacency, relation_input, input, sum="min", mul=mul) mean = (sum + boundary) / degree_out sq_mean = (sq_sum + boundary ** 2) / degree_out max = torch.max(max, boundary) min = torch.min(min, boundary) std = (sq_mean - mean ** 2).clamp(min=self.eps).sqrt() features = torch.cat([mean.unsqueeze(-1), max.unsqueeze(-1), min.unsqueeze(-1), std.unsqueeze(-1)], dim=-1) features = features.flatten(-2) scale = degree_out.log() scale = scale / scale.mean() scales = torch.cat([torch.ones_like(scale), scale, 1 / scale.clamp(min=1e-2)], dim=-1) update = (features.unsqueeze(-1) * scales.unsqueeze(-2)).flatten(-2) else: raise ValueError("Unknown aggregation function `%s`" % self.aggregate_func) return update.view(len(update), batch_size, -1) def combine(self, input, update): output = self.linear(torch.cat([input, update], dim=-1)) if self.layer_norm: output = self.layer_norm(output) if self.activation: output = self.activation(output) return output
7,872
46.427711
119
py
NBFNet
NBFNet-master/nbfnet/task.py
import math import torch from torch.nn import functional as F from torch.utils import data as torch_data from ogb import linkproppred from torchdrug import core, tasks, metrics from torchdrug.layers import functional from torchdrug.core import Registry as R Evaluator = core.make_configurable(linkproppred.Evaluator) Evaluator = R.register("ogb.linkproppred.Evaluator")(Evaluator) setattr(linkproppred, "Evaluator", Evaluator) @R.register("tasks.KnowledgeGraphCompletionExt") class KnowledgeGraphCompletionExt(tasks.KnowledgeGraphCompletion, core.Configurable): def __init__(self, model, criterion="bce", metric=("mr", "mrr", "hits@1", "hits@3", "hits@10", "1-to-1", "1-to-n", "n-to-1", "n-to-n"), num_negative=128, margin=6, adversarial_temperature=0, strict_negative=True, filtered_ranking=True, fact_ratio=None, sample_weight=True): super(KnowledgeGraphCompletionExt, self).__init__( model, criterion, metric, num_negative, margin, adversarial_temperature, strict_negative, filtered_ranking, fact_ratio, sample_weight) def preprocess(self, train_set, valid_set, test_set): super(KnowledgeGraphCompletionExt, self).preprocess(train_set, valid_set, test_set) degree_hr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) degree_tr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) for h, t, r in train_set: degree_hr[h, r] += 1 degree_tr[t, r] += 1 has_category = False for _metric in self.metric: if _metric in ["1-to-1", "1-to-n", "n-to-1", "n-to-n"]: has_category = True if has_category: is_to_one = degree_hr.sum(dim=0).float() / (degree_hr > 0).sum(dim=0) < 1.5 is_one_to = degree_tr.sum(dim=0).float() / (degree_tr > 0).sum(dim=0) < 1.5 self.register_buffer("is_one_to_one", is_one_to & is_to_one) self.register_buffer("is_one_to_many", is_one_to & ~is_to_one) self.register_buffer("is_many_to_one", ~is_one_to & is_to_one) self.register_buffer("is_many_to_many", ~is_one_to & ~is_to_one) assert self.is_one_to_one.sum() + self.is_one_to_many.sum() + \ self.is_many_to_one.sum() + self.is_many_to_many.sum() == self.num_relation assert (self.is_one_to_one | self.is_one_to_many | self.is_many_to_one | self.is_many_to_many).all() def target(self, batch): mask, target = super(KnowledgeGraphCompletionExt, self).target(batch) relation = batch[:, 2] # in case of GPU OOM return mask, target, relation.cpu() def evaluate(self, pred, target): mask, target, relation = target pos_pred = pred.gather(-1, target.unsqueeze(-1)) if self.filtered_ranking: ranking = torch.sum((pos_pred <= pred) & mask, dim=-1) + 1 else: ranking = torch.sum(pos_pred <= pred, dim=-1) + 1 metric = {} for _metric in self.metric: if _metric == "mr": score = ranking.float().mean() elif _metric == "mrr": score = (1 / ranking.float()).mean() elif _metric.startswith("hits@"): threshold = int(_metric[5:]) score = (ranking <= threshold).float().mean() elif _metric == "1-to-1": score = (1 / ranking[self.is_one_to_one[relation]].float()).mean() metric["1-to-1 tail"] = (1 / ranking[self.is_one_to_one[relation]].float()).mean(dim=0)[0] metric["1-to-1 head"] = (1 / ranking[self.is_one_to_one[relation]].float()).mean(dim=0)[1] elif _metric == "1-to-n": score = (1 / ranking[self.is_one_to_many[relation]].float()).mean() metric["1-to-n tail"] = (1 / ranking[self.is_one_to_many[relation]].float()).mean(dim=0)[0] metric["1-to-n head"] = (1 / ranking[self.is_one_to_many[relation]].float()).mean(dim=0)[1] elif _metric == "n-to-1": score = (1 / ranking[self.is_many_to_one[relation]].float()).mean() metric["n-to-1 tail"] = (1 / ranking[self.is_many_to_one[relation]].float()).mean(dim=0)[0] metric["n-to-1 head"] = (1 / ranking[self.is_many_to_one[relation]].float()).mean(dim=0)[1] elif _metric == "n-to-n": score = (1 / ranking[self.is_many_to_many[relation]].float()).mean() metric["n-to-n tail"] = (1 / ranking[self.is_many_to_many[relation]].float()).mean(dim=0)[0] metric["n-to-n head"] = (1 / ranking[self.is_many_to_many[relation]].float()).mean(dim=0)[1] else: raise ValueError("Unknown metric `%s`" % _metric) name = tasks._get_metric_name(_metric) metric[name] = score return metric @R.register("tasks.LinkPrediction") class LinkPrediction(tasks.Task, core.Configurable): _option_members = ["criterion", "metric"] def __init__(self, model, criterion="bce", metric=("auroc", "ap"), num_negative=128, strict_negative=True): super(LinkPrediction, self).__init__() self.model = model self.criterion = criterion self.metric = metric self.num_negative = num_negative self.strict_negative = strict_negative def preprocess(self, train_set, valid_set, test_set): if isinstance(train_set, torch_data.Subset): dataset = train_set.dataset else: dataset = train_set self.num_node = dataset.num_node train_mask = train_set.indices valid_mask = train_set.indices + valid_set.indices train_graph = dataset.graph.edge_mask(train_mask) valid_graph = dataset.graph.edge_mask(valid_mask) self.register_buffer("train_graph", train_graph.undirected()) self.register_buffer("valid_graph", valid_graph.undirected()) self.register_buffer("test_graph", dataset.graph.undirected()) def forward(self, batch): all_loss = torch.tensor(0, dtype=torch.float32, device=self.device) metric = {} pred, target = self.predict_and_target(batch, all_loss, metric) metric.update(self.evaluate(pred, target)) for criterion, weight in self.criterion.items(): if criterion == "bce": loss = F.binary_cross_entropy_with_logits(pred, target, reduction="none") neg_weight = torch.ones_like(pred) neg_weight[:, 1:] = 1 / self.num_negative loss = (loss * neg_weight).sum(dim=-1) / neg_weight.sum(dim=-1) else: raise ValueError("Unknown criterion `%s`" % criterion) loss = loss.mean() name = tasks._get_criterion_name(criterion) metric[name] = loss all_loss += loss * weight return all_loss, metric @torch.no_grad() def _strict_negative(self, count, split="train"): graph = getattr(self, "%s_graph" % split) node_in = graph.edge_list[:, 0] degree_in = torch.bincount(node_in, minlength=self.num_node) prob = (graph.num_node - degree_in - 1).float() neg_h_index = functional.multinomial(prob, count, replacement=True) any = -torch.ones_like(neg_h_index) pattern = torch.stack([neg_h_index, any], dim=-1) edge_index, num_t_truth = graph.match(pattern) t_truth_index = graph.edge_list[edge_index, 1] pos_index = functional._size_to_index(num_t_truth) t_mask = torch.ones(count, self.num_node, dtype=torch.bool, device=self.device) t_mask[pos_index, t_truth_index] = 0 t_mask.scatter_(1, neg_h_index.unsqueeze(-1), 0) neg_t_candidate = t_mask.nonzero()[:, 1] num_t_candidate = t_mask.sum(dim=-1) neg_t_index = functional.variadic_sample(neg_t_candidate, num_t_candidate, 1).squeeze(-1) return neg_h_index, neg_t_index def predict_and_target(self, batch, all_loss=None, metric=None): batch_size = len(batch) pos_h_index, pos_t_index = batch.t() if self.split == "train": num_negative = self.num_negative else: num_negative = 1 if self.strict_negative or self.split != "train": neg_h_index, neg_t_index = self._strict_negative(batch_size * num_negative, self.split) else: neg_h_index, neg_t_index = torch.randint(self.num_node, (2, batch_size * num_negative), device=self.device) neg_h_index = neg_h_index.view(batch_size, num_negative) neg_t_index = neg_t_index.view(batch_size, num_negative) h_index = pos_h_index.unsqueeze(-1).repeat(1, num_negative + 1) t_index = pos_t_index.unsqueeze(-1).repeat(1, num_negative + 1) h_index[:, 1:] = neg_h_index t_index[:, 1:] = neg_t_index pred = self.model(self.train_graph, h_index, t_index, all_loss=all_loss, metric=metric) target = torch.zeros_like(pred) target[:, 0] = 1 return pred, target def evaluate(self, pred, target): pred = pred.flatten() target = target.flatten() metric = {} for _metric in self.metric: if _metric == "auroc": score = metrics.area_under_roc(pred, target) elif _metric == "ap": score = metrics.area_under_prc(pred, target) else: raise ValueError("Unknown metric `%s`" % _metric) name = tasks._get_metric_name(_metric) metric[name] = score return metric @R.register("tasks.InductiveKnowledgeGraphCompletion") class InductiveKnowledgeGraphCompletion(tasks.KnowledgeGraphCompletion, core.Configurable): def __init__(self, model, criterion="bce", metric=("mr", "mrr", "hits@1", "hits@3", "hits@10", "hits@10_50"), num_negative=128, margin=6, adversarial_temperature=0, strict_negative=True, sample_weight=True): super(InductiveKnowledgeGraphCompletion, self).__init__( model, criterion, metric, num_negative, margin, adversarial_temperature, strict_negative, sample_weight=sample_weight) def preprocess(self, train_set, valid_set, test_set): if isinstance(train_set, torch_data.Subset): dataset = train_set.dataset else: dataset = train_set self.num_entity = dataset.num_entity self.num_relation = dataset.num_relation self.register_buffer("fact_graph", dataset.graph) if self.sample_weight: degree_hr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) degree_tr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) for h, t, r in train_set: degree_hr[h, r] += 1 degree_tr[t, r] += 1 self.register_buffer("degree_hr", degree_hr) self.register_buffer("degree_tr", degree_tr) self.register_buffer("train_graph", dataset.train_graph) self.register_buffer("valid_graph", dataset.valid_graph) self.register_buffer("test_graph", dataset.test_graph) return train_set, valid_set, test_set def predict(self, batch, all_loss=None, metric=None): pos_h_index, pos_t_index, pos_r_index = batch.t() batch_size = len(batch) graph = getattr(self, "%s_graph" % self.split) if all_loss is None: # test all_index = torch.arange(graph.num_node, device=self.device) t_preds = [] h_preds = [] for neg_index in all_index.split(self.num_negative): r_index = pos_r_index.unsqueeze(-1).expand(-1, len(neg_index)) h_index, t_index = torch.meshgrid(pos_h_index, neg_index) t_pred = self.model(graph, h_index, t_index, r_index, all_loss=all_loss, metric=metric) t_preds.append(t_pred) t_pred = torch.cat(t_preds, dim=-1) for neg_index in all_index.split(self.num_negative): r_index = pos_r_index.unsqueeze(-1).expand(-1, len(neg_index)) t_index, h_index = torch.meshgrid(pos_t_index, neg_index) h_pred = self.model(graph, h_index, t_index, r_index, all_loss=all_loss, metric=metric) h_preds.append(h_pred) h_pred = torch.cat(h_preds, dim=-1) pred = torch.stack([t_pred, h_pred], dim=1) # in case of GPU OOM pred = pred.cpu() else: # train if self.strict_negative: neg_index = self._strict_negative(pos_h_index, pos_t_index, pos_r_index) else: neg_index = torch.randint(self.num_entity, (batch_size, self.num_negative), device=self.device) h_index = pos_h_index.unsqueeze(-1).repeat(1, self.num_negative + 1) t_index = pos_t_index.unsqueeze(-1).repeat(1, self.num_negative + 1) r_index = pos_r_index.unsqueeze(-1).repeat(1, self.num_negative + 1) t_index[:batch_size // 2, 1:] = neg_index[:batch_size // 2] h_index[batch_size // 2:, 1:] = neg_index[batch_size // 2:] pred = self.model(graph, h_index, t_index, r_index, all_loss=all_loss, metric=metric) return pred def target(self, batch): # test target batch_size = len(batch) graph = getattr(self, "%s_graph" % self.split) pos_h_index, pos_t_index, pos_r_index = batch.t() any = -torch.ones_like(pos_h_index) pattern = torch.stack([pos_h_index, any, pos_r_index], dim=-1) edge_index, num_t_truth = graph.match(pattern) t_truth_index = graph.edge_list[edge_index, 1] pos_index = functional._size_to_index(num_t_truth) t_mask = torch.ones(batch_size, graph.num_node, dtype=torch.bool, device=self.device) t_mask[pos_index, t_truth_index] = 0 pattern = torch.stack([any, pos_t_index, pos_r_index], dim=-1) edge_index, num_h_truth = graph.match(pattern) h_truth_index = graph.edge_list[edge_index, 0] pos_index = functional._size_to_index(num_h_truth) h_mask = torch.ones(batch_size, graph.num_node, dtype=torch.bool, device=self.device) h_mask[pos_index, h_truth_index] = 0 mask = torch.stack([t_mask, h_mask], dim=1) target = torch.stack([pos_t_index, pos_h_index], dim=1) # in case of GPU OOM return mask.cpu(), target.cpu() def evaluate(self, pred, target): mask, target = target pos_pred = pred.gather(-1, target.unsqueeze(-1)) ranking = torch.sum((pos_pred <= pred) & mask, dim=-1) + 1 metric = {} for _metric in self.metric: if _metric == "mr": score = ranking.float().mean() elif _metric == "mrr": score = (1 / ranking.float()).mean() elif _metric.startswith("hits@"): values = _metric[5:].split("_") threshold = int(values[0]) if len(values) > 1: num_sample = int(values[1]) # unbiased estimation fp_rate = (ranking - 1).float() / mask.sum(dim=-1) score = 0 for i in range(threshold): # choose i false positive from num_sample negatives num_comb = math.factorial(num_sample) / math.factorial(i) / math.factorial(num_sample - i) score += num_comb * (fp_rate ** i) * ((1 - fp_rate) ** (num_sample - i)) score = score.mean() else: score = (ranking <= threshold).float().mean() else: raise ValueError("Unknown metric `%s`" % _metric) name = tasks._get_metric_name(_metric) metric[name] = score return metric @R.register("tasks.KnowledgeGraphCompletionOGB") class KnowledgeGraphCompletionOGB(tasks.KnowledgeGraphCompletion, core.Configurable): def __init__(self, model, criterion="bce", evaluator=None, num_negative=128, margin=6, adversarial_temperature=0, strict_negative=True, heterogeneous_negative=False, fact_ratio=None, sample_weight=True): super(KnowledgeGraphCompletionOGB, self).__init__( model, criterion, None, num_negative, margin, adversarial_temperature, strict_negative, True, fact_ratio, sample_weight) self.evaluator = evaluator self.heterogeneous_negative = heterogeneous_negative def preprocess(self, train_set, valid_set, test_set): if isinstance(train_set, torch_data.Subset): dataset = train_set.dataset else: dataset = train_set self.num_entity = dataset.num_entity self.num_relation = dataset.num_relation self.register_buffer("graph", dataset.graph) fact_mask = torch.zeros(len(dataset), dtype=torch.bool) fact_mask[train_set.indices] = 1 if self.fact_ratio: length = int(len(train_set) * self.fact_ratio) index = torch.randperm(len(train_set))[length:] train_indices = torch.tensor(train_set.indices) fact_mask[train_indices[index]] = 0 train_set = torch_data.Subset(train_set, index) self.register_buffer("fact_graph", dataset.graph.edge_mask(fact_mask)) if self.sample_weight: degree_hr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) degree_tr = torch.zeros(self.num_entity, self.num_relation, dtype=torch.long) for h, t, r in train_set: degree_hr[h, r] += 1 degree_tr[t, r] += 1 self.register_buffer("degree_hr", degree_hr) self.register_buffer("degree_tr", degree_tr) return train_set, valid_set, test_set @torch.no_grad() def _strict_negative(self, pos_h_index, pos_t_index, pos_r_index): batch_size = len(pos_h_index) any = -torch.ones_like(pos_h_index) node_type = self.fact_graph.node_type pattern = torch.stack([pos_h_index, any, pos_r_index], dim=-1) pattern = pattern[:batch_size // 2] edge_index, num_t_truth = self.fact_graph.match(pattern) t_truth_index = self.fact_graph.edge_list[edge_index, 1] pos_index = functional._size_to_index(num_t_truth) if self.heterogeneous_negative: pos_t_type = node_type[pos_t_index[:batch_size // 2]] t_mask = pos_t_type.unsqueeze(-1) == node_type.unsqueeze(0) else: t_mask = torch.ones(len(pattern), self.num_entity, dtype=torch.bool, device=self.device) t_mask[pos_index, t_truth_index] = 0 neg_t_candidate = t_mask.nonzero()[:, 1] num_t_candidate = t_mask.sum(dim=-1) neg_t_index = functional.variadic_sample(neg_t_candidate, num_t_candidate, self.num_negative) pattern = torch.stack([any, pos_t_index, pos_r_index], dim=-1) pattern = pattern[batch_size // 2:] edge_index, num_h_truth = self.fact_graph.match(pattern) h_truth_index = self.fact_graph.edge_list[edge_index, 0] pos_index = functional._size_to_index(num_h_truth) if self.heterogeneous_negative: pos_h_type = node_type[pos_h_index[batch_size // 2:]] h_mask = pos_h_type.unsqueeze(-1) == node_type.unsqueeze(0) else: h_mask = torch.ones(len(pattern), self.num_entity, dtype=torch.bool, device=self.device) h_mask[pos_index, h_truth_index] = 0 neg_h_candidate = h_mask.nonzero()[:, 1] num_h_candidate = h_mask.sum(dim=-1) neg_h_index = functional.variadic_sample(neg_h_candidate, num_h_candidate, self.num_negative) neg_index = torch.cat([neg_t_index, neg_h_index]) return neg_index def predict(self, batch, all_loss=None, metric=None): batch_size = len(batch) if all_loss is None: # test h_index, t_index, r_index = batch.unbind(-1) pattern = batch[:, 0, :] num_match = self.fact_graph.match(pattern)[1] assert (num_match == 0).all() pred = self.model(self.fact_graph, h_index, t_index, r_index, all_loss=all_loss, metric=metric) # in case of GPU OOM pred = pred.cpu() else: # train pos_h_index, pos_t_index, pos_r_index = batch.t() if self.strict_negative: neg_index = self._strict_negative(pos_h_index, pos_t_index, pos_r_index) else: neg_index = torch.randint(self.num_entity, (batch_size, self.num_negative), device=self.device) h_index = pos_h_index.unsqueeze(-1).repeat(1, self.num_negative + 1) t_index = pos_t_index.unsqueeze(-1).repeat(1, self.num_negative + 1) r_index = pos_r_index.unsqueeze(-1).repeat(1, self.num_negative + 1) t_index[:batch_size // 2, 1:] = neg_index[:batch_size // 2] h_index[batch_size // 2:, 1:] = neg_index[batch_size // 2:] pred = self.model(self.fact_graph, h_index, t_index, r_index, all_loss=all_loss, metric=metric) return pred def target(self, batch): # test target batch_size = len(batch) target = torch.zeros(batch_size, dtype=torch.long, device=self.device) # in case of GPU OOM return target.cpu() def evaluate(self, pred, target): is_positive = torch.zeros(pred.shape, dtype=torch.bool) is_positive.scatter_(-1, target.unsqueeze(-1), 1) pos_pred = pred[is_positive] neg_pred = pred[~is_positive].view(len(pos_pred), -1) metric = self.evaluator.eval({"y_pred_pos": pos_pred, "y_pred_neg": neg_pred}) new_metric = {} for key in metric: new_key = key.split("_")[0] new_metric[new_key] = metric[key].mean() return new_metric
22,136
44.832298
119
py
NBFNet
NBFNet-master/nbfnet/model.py
from collections.abc import Sequence import torch from torch import nn from torch import autograd from torch_scatter import scatter_add from torchdrug import core, layers, utils from torchdrug.layers import functional from torchdrug.core import Registry as R from . import layer @R.register("model.NBFNet") class NeuralBellmanFordNetwork(nn.Module, core.Configurable): def __init__(self, input_dim, hidden_dims, num_relation=None, symmetric=False, message_func="distmult", aggregate_func="pna", short_cut=False, layer_norm=False, activation="relu", concat_hidden=False, num_mlp_layer=2, dependent=True, remove_one_hop=False, num_beam=10, path_topk=10): super(NeuralBellmanFordNetwork, self).__init__() if not isinstance(hidden_dims, Sequence): hidden_dims = [hidden_dims] if num_relation is None: double_relation = 1 else: num_relation = int(num_relation) double_relation = num_relation * 2 self.dims = [input_dim] + list(hidden_dims) self.num_relation = num_relation self.symmetric = symmetric self.short_cut = short_cut self.concat_hidden = concat_hidden self.remove_one_hop = remove_one_hop self.num_beam = num_beam self.path_topk = path_topk self.layers = nn.ModuleList() for i in range(len(self.dims) - 1): self.layers.append(layer.GeneralizedRelationalConv(self.dims[i], self.dims[i + 1], double_relation, self.dims[0], message_func, aggregate_func, layer_norm, activation, dependent)) feature_dim = hidden_dims[-1] * (len(hidden_dims) if concat_hidden else 1) + input_dim self.query = nn.Embedding(double_relation, input_dim) self.mlp = layers.MLP(feature_dim, [feature_dim] * (num_mlp_layer - 1) + [1]) def remove_easy_edges(self, graph, h_index, t_index, r_index=None): if self.remove_one_hop: h_index_ext = torch.cat([h_index, t_index], dim=-1) t_index_ext = torch.cat([t_index, h_index], dim=-1) if r_index is not None: any = -torch.ones_like(h_index_ext) pattern = torch.stack([h_index_ext, t_index_ext, any], dim=-1) else: pattern = torch.stack([h_index_ext, t_index_ext], dim=-1) else: if r_index is not None: pattern = torch.stack([h_index, t_index, r_index], dim=-1) else: pattern = torch.stack([h_index, t_index], dim=-1) pattern = pattern.flatten(0, -2) edge_index = graph.match(pattern)[0] edge_mask = ~functional.as_mask(edge_index, graph.num_edge) return graph.edge_mask(edge_mask) def negative_sample_to_tail(self, h_index, t_index, r_index): # convert p(h | t, r) to p(t' | h', r') # h' = t, r' = r^{-1}, t' = h is_t_neg = (h_index == h_index[:, [0]]).all(dim=-1, keepdim=True) new_h_index = torch.where(is_t_neg, h_index, t_index) new_t_index = torch.where(is_t_neg, t_index, h_index) new_r_index = torch.where(is_t_neg, r_index, r_index + self.num_relation) return new_h_index, new_t_index, new_r_index def as_relational_graph(self, graph, self_loop=True): # add self loop # convert homogeneous graphs to knowledge graphs with 1 relation edge_list = graph.edge_list edge_weight = graph.edge_weight if self_loop: node_in = node_out = torch.arange(graph.num_node, device=self.device) loop = torch.stack([node_in, node_out], dim=-1) edge_list = torch.cat([edge_list, loop]) edge_weight = torch.cat([edge_weight, torch.ones(graph.num_node, device=self.device)]) relation = torch.zeros(len(edge_list), 1, dtype=torch.long, device=self.device) edge_list = torch.cat([edge_list, relation], dim=-1) graph = type(graph)(edge_list, edge_weight=edge_weight, num_node=graph.num_node, num_relation=1, meta_dict=graph.meta_dict, **graph.data_dict) return graph @utils.cached def bellmanford(self, graph, h_index, r_index, separate_grad=False): query = self.query(r_index) index = h_index.unsqueeze(-1).expand_as(query) boundary = torch.zeros(graph.num_node, *query.shape, device=self.device) boundary.scatter_add_(0, index.unsqueeze(0), query.unsqueeze(0)) with graph.graph(): graph.query = query with graph.node(): graph.boundary = boundary hiddens = [] step_graphs = [] layer_input = boundary for layer in self.layers: if separate_grad: step_graph = graph.clone().requires_grad_() else: step_graph = graph hidden = layer(step_graph, layer_input) if self.short_cut and hidden.shape == layer_input.shape: hidden = hidden + layer_input hiddens.append(hidden) step_graphs.append(step_graph) layer_input = hidden node_query = query.expand(graph.num_node, -1, -1) if self.concat_hidden: output = torch.cat(hiddens + [node_query], dim=-1) else: output = torch.cat([hiddens[-1], node_query], dim=-1) return { "node_feature": output, "step_graphs": step_graphs, } def forward(self, graph, h_index, t_index, r_index=None, all_loss=None, metric=None): if all_loss is not None: graph = self.remove_easy_edges(graph, h_index, t_index, r_index) shape = h_index.shape if graph.num_relation: graph = graph.undirected(add_inverse=True) h_index, t_index, r_index = self.negative_sample_to_tail(h_index, t_index, r_index) else: graph = self.as_relational_graph(graph) h_index = h_index.view(-1, 1) t_index = t_index.view(-1, 1) r_index = torch.zeros_like(h_index) assert (h_index[:, [0]] == h_index).all() assert (r_index[:, [0]] == r_index).all() output = self.bellmanford(graph, h_index[:, 0], r_index[:, 0]) feature = output["node_feature"].transpose(0, 1) index = t_index.unsqueeze(-1).expand(-1, -1, feature.shape[-1]) feature = feature.gather(1, index) if self.symmetric: assert (t_index[:, [0]] == t_index).all() output = self.bellmanford(graph, t_index[:, 0], r_index[:, 0]) inv_feature = output["node_feature"].transpose(0, 1) index = h_index.unsqueeze(-1).expand(-1, -1, inv_feature.shape[-1]) inv_feature = inv_feature.gather(1, index) feature = (feature + inv_feature) / 2 score = self.mlp(feature).squeeze(-1) return score.view(shape) def visualize(self, graph, h_index, t_index, r_index): assert h_index.numel() == 1 and h_index.ndim == 1 graph = graph.undirected(add_inverse=True) output = self.bellmanford(graph, h_index, r_index, separate_grad=True) feature = output["node_feature"] step_graphs = output["step_graphs"] index = t_index.unsqueeze(0).unsqueeze(-1).expand(-1, -1, feature.shape[-1]) feature = feature.gather(0, index).squeeze(0) score = self.mlp(feature).squeeze(-1) edge_weights = [graph.edge_weight for graph in step_graphs] edge_grads = autograd.grad(score, edge_weights) for graph, edge_grad in zip(step_graphs, edge_grads): with graph.edge(): graph.edge_grad = edge_grad distances, back_edges = self.beam_search_distance(step_graphs, h_index, t_index, self.num_beam) paths, weights = self.topk_average_length(distances, back_edges, t_index, self.path_topk) return paths, weights @torch.no_grad() def beam_search_distance(self, graphs, h_index, t_index, num_beam=10): num_node = graphs[0].num_node input = torch.full((num_node, num_beam), float("-inf"), device=self.device) input[h_index, 0] = 0 distances = [] back_edges = [] for graph in graphs: graph = graph.edge_mask(graph.edge_list[:, 0] != t_index) node_in, node_out = graph.edge_list.t()[:2] message = input[node_in] + graph.edge_grad.unsqueeze(-1) msg_source = graph.edge_list.unsqueeze(1).expand(-1, num_beam, -1) is_duplicate = torch.isclose(message.unsqueeze(-1), message.unsqueeze(-2)) & \ (msg_source.unsqueeze(-2) == msg_source.unsqueeze(-3)).all(dim=-1) is_duplicate = is_duplicate.float() - \ torch.arange(num_beam, dtype=torch.float, device=self.device) / (num_beam + 1) # pick the first occurrence as the previous state prev_rank = is_duplicate.argmax(dim=-1, keepdim=True) msg_source = torch.cat([msg_source, prev_rank], dim=-1) node_out, order = node_out.sort() node_out_set = torch.unique(node_out) # sort message w.r.t. node_out message = message[order].flatten() msg_source = msg_source[order].flatten(0, -2) size = scatter_add(torch.ones_like(node_out), node_out, dim_size=num_node) msg2out = functional._size_to_index(size[node_out_set] * num_beam) # deduplicate is_duplicate = (msg_source[1:] == msg_source[:-1]).all(dim=-1) is_duplicate = torch.cat([torch.zeros(1, dtype=torch.bool, device=self.device), is_duplicate]) message = message[~is_duplicate] msg_source = msg_source[~is_duplicate] msg2out = msg2out[~is_duplicate] size = scatter_add(torch.ones_like(msg2out), msg2out, dim_size=len(node_out_set)) if not torch.isinf(message).all(): distance, rel_index = functional.variadic_topk(message, size, k=num_beam) abs_index = rel_index + (size.cumsum(0) - size).unsqueeze(-1) back_edge = msg_source[abs_index] distance = distance.view(len(node_out_set), num_beam) back_edge = back_edge.view(len(node_out_set), num_beam, 4) distance = scatter_add(distance, node_out_set, dim=0, dim_size=num_node) back_edge = scatter_add(back_edge, node_out_set, dim=0, dim_size=num_node) else: distance = torch.full((num_node, num_beam), float("-inf"), device=self.device) back_edge = torch.zeros(num_node, num_beam, 4, dtype=torch.long, device=self.device) distances.append(distance) back_edges.append(back_edge) input = distance return distances, back_edges def topk_average_length(self, distances, back_edges, t_index, k=10): paths = [] average_lengths = [] for i in range(len(distances)): distance, order = distances[i][t_index].flatten(0, -1).sort(descending=True) back_edge = back_edges[i][t_index].flatten(0, -2)[order] for d, (h, t, r, prev_rank) in zip(distance[:k].tolist(), back_edge[:k].tolist()): if d == float("-inf"): break path = [(h, t, r)] for j in range(i - 1, -1, -1): h, t, r, prev_rank = back_edges[j][h, prev_rank].tolist() path.append((h, t, r)) paths.append(path[::-1]) average_lengths.append(d / len(path)) if paths: average_lengths, paths = zip(*sorted(zip(average_lengths, paths), reverse=True)[:k]) return paths, average_lengths
11,950
44.441065
118
py
NBFNet
NBFNet-master/nbfnet/dataset.py
import os import csv import glob from tqdm import tqdm from ogb import linkproppred import torch from torch.utils import data as torch_data from torchdrug import data, datasets, utils from torchdrug.core import Registry as R class InductiveKnowledgeGraphDataset(data.KnowledgeGraphDataset): def load_inductive_tsvs(self, train_files, test_files, verbose=0): assert len(train_files) == len(test_files) == 2 inv_train_entity_vocab = {} inv_test_entity_vocab = {} inv_relation_vocab = {} triplets = [] num_samples = [] for txt_file in train_files: with open(txt_file, "r") as fin: reader = csv.reader(fin, delimiter="\t") if verbose: reader = tqdm(reader, "Loading %s" % txt_file, utils.get_line_count(txt_file)) num_sample = 0 for tokens in reader: h_token, r_token, t_token = tokens if h_token not in inv_train_entity_vocab: inv_train_entity_vocab[h_token] = len(inv_train_entity_vocab) h = inv_train_entity_vocab[h_token] if r_token not in inv_relation_vocab: inv_relation_vocab[r_token] = len(inv_relation_vocab) r = inv_relation_vocab[r_token] if t_token not in inv_train_entity_vocab: inv_train_entity_vocab[t_token] = len(inv_train_entity_vocab) t = inv_train_entity_vocab[t_token] triplets.append((h, t, r)) num_sample += 1 num_samples.append(num_sample) for txt_file in test_files: with open(txt_file, "r") as fin: reader = csv.reader(fin, delimiter="\t") if verbose: reader = tqdm(reader, "Loading %s" % txt_file, utils.get_line_count(txt_file)) num_sample = 0 for tokens in reader: h_token, r_token, t_token = tokens if h_token not in inv_test_entity_vocab: inv_test_entity_vocab[h_token] = len(inv_test_entity_vocab) h = inv_test_entity_vocab[h_token] assert r_token in inv_relation_vocab r = inv_relation_vocab[r_token] if t_token not in inv_test_entity_vocab: inv_test_entity_vocab[t_token] = len(inv_test_entity_vocab) t = inv_test_entity_vocab[t_token] triplets.append((h, t, r)) num_sample += 1 num_samples.append(num_sample) train_entity_vocab, inv_train_entity_vocab = self._standarize_vocab(None, inv_train_entity_vocab) test_entity_vocab, inv_test_entity_vocab = self._standarize_vocab(None, inv_test_entity_vocab) relation_vocab, inv_relation_vocab = self._standarize_vocab(None, inv_relation_vocab) self.train_graph = data.Graph(triplets[:num_samples[0]], num_node=len(train_entity_vocab), num_relation=len(relation_vocab)) self.valid_graph = self.train_graph self.test_graph = data.Graph(triplets[sum(num_samples[:2]): sum(num_samples[:3])], num_node=len(test_entity_vocab), num_relation=len(relation_vocab)) self.graph = self.train_graph self.triplets = torch.tensor(triplets[:sum(num_samples[:2])] + triplets[sum(num_samples[:3]):]) self.num_samples = num_samples[:2] + num_samples[3:] self.train_entity_vocab = train_entity_vocab self.test_entity_vocab = test_entity_vocab self.relation_vocab = relation_vocab self.inv_train_entity_vocab = inv_train_entity_vocab self.inv_test_entity_vocab = inv_test_entity_vocab self.inv_relation_vocab = inv_relation_vocab def __getitem__(self, index): return self.triplets[index] def split(self): offset = 0 splits = [] for num_sample in self.num_samples: split = torch_data.Subset(self, range(offset, offset + num_sample)) splits.append(split) offset += num_sample return splits @R.register("datasets.CoraLinkPrediction") class CoraLinkPrediction(datasets.Cora): def __init__(self, **kwargs): super(CoraLinkPrediction, self).__init__(**kwargs) self.transform = None def __getitem__(self, index): return self.graph.edge_list[index] def __len__(self): return self.graph.num_edge def split(self, ratios=(85, 5, 10)): length = self.graph.num_edge norm = sum(ratios) lengths = [int(r / norm * length) for r in ratios] lengths[-1] = length - sum(lengths[:-1]) g = torch.Generator() g.manual_seed(0) return torch_data.random_split(self, lengths, generator=g) @R.register("datasets.CiteSeerLinkPrediction") class CiteSeerLinkPrediction(datasets.CiteSeer): def __init__(self, **kwargs): super(CiteSeerLinkPrediction, self).__init__(**kwargs) self.transform = None def __getitem__(self, index): return self.graph.edge_list[index] def __len__(self): return self.graph.num_edge def split(self, ratios=(85, 5, 10)): length = self.graph.num_edge norm = sum(ratios) lengths = [int(r / norm * length) for r in ratios] lengths[-1] = length - sum(lengths[:-1]) g = torch.Generator() g.manual_seed(0) return torch_data.random_split(self, lengths, generator=g) @R.register("datasets.PubMedLinkPrediction") class PubMedLinkPrediction(datasets.PubMed): def __init__(self, **kwargs): super(PubMedLinkPrediction, self).__init__(**kwargs) self.transform = None def __getitem__(self, index): return self.graph.edge_list[index] def __len__(self): return self.graph.num_edge def split(self, ratios=(85, 5, 10)): length = self.graph.num_edge norm = sum(ratios) lengths = [int(r / norm * length) for r in ratios] lengths[-1] = length - sum(lengths[:-1]) g = torch.Generator() g.manual_seed(0) return torch_data.random_split(self, lengths, generator=g) @R.register("datasets.FB15k237Inductive") class FB15k237Inductive(InductiveKnowledgeGraphDataset): train_urls = [ "https://raw.githubusercontent.com/kkteru/grail/master/data/fb237_%s/train.txt", "https://raw.githubusercontent.com/kkteru/grail/master/data/fb237_%s/valid.txt", ] test_urls = [ "https://raw.githubusercontent.com/kkteru/grail/master/data/fb237_%s_ind/train.txt", "https://raw.githubusercontent.com/kkteru/grail/master/data/fb237_%s_ind/test.txt", ] def __init__(self, path, version="v1", verbose=1): path = os.path.expanduser(path) if not os.path.exists(path): os.makedirs(path) self.path = path train_files = [] for url in self.train_urls: url = url % version save_file = "fb15k237_%s_%s" % (version, os.path.basename(url)) txt_file = os.path.join(path, save_file) if not os.path.exists(txt_file): txt_file = utils.download(url, self.path, save_file=save_file) train_files.append(txt_file) test_files = [] for url in self.test_urls: url = url % version save_file = "fb15k237_%s_ind_%s" % (version, os.path.basename(url)) txt_file = os.path.join(path, save_file) if not os.path.exists(txt_file): txt_file = utils.download(url, self.path, save_file=save_file) test_files.append(txt_file) self.load_inductive_tsvs(train_files, test_files, verbose=verbose) @R.register("datasets.WN18RRInductive") class WN18RRInductive(InductiveKnowledgeGraphDataset): train_urls = [ "https://raw.githubusercontent.com/kkteru/grail/master/data/WN18RR_%s/train.txt", "https://raw.githubusercontent.com/kkteru/grail/master/data/WN18RR_%s/valid.txt", ] test_urls = [ "https://raw.githubusercontent.com/kkteru/grail/master/data/WN18RR_%s_ind/train.txt", "https://raw.githubusercontent.com/kkteru/grail/master/data/WN18RR_%s_ind/test.txt", ] def __init__(self, path, version="v1", verbose=1): path = os.path.expanduser(path) if not os.path.exists(path): os.makedirs(path) self.path = path train_files = [] for url in self.train_urls: url = url % version save_file = "wn18rr_%s_%s" % (version, os.path.basename(url)) txt_file = os.path.join(path, save_file) if not os.path.exists(txt_file): txt_file = utils.download(url, self.path, save_file=save_file) train_files.append(txt_file) test_files = [] for url in self.test_urls: url = url % version save_file = "wn18rr_%s_ind_%s" % (version, os.path.basename(url)) txt_file = os.path.join(path, save_file) if not os.path.exists(txt_file): txt_file = utils.download(url, self.path, save_file=save_file) test_files.append(txt_file) self.load_inductive_tsvs(train_files, test_files, verbose=verbose) @R.register("datasets.OGBLBioKG") class OGBLBioKG(data.KnowledgeGraphDataset): def __init__(self, path, verbose=1): path = os.path.expanduser(path) self.path = path dataset = linkproppred.LinkPropPredDataset("ogbl-biokg", path) self.load_ogb(dataset, verbose=verbose) def load_ogb(self, dataset, verbose=1): entity_vocab = [] relation_vocab = [] entity_type_vocab = [] inv_entity_type_offset = {} entity_type2num = [] zip_files = glob.glob(os.path.join(dataset.root, "mapping/*.gz")) for zip_file in zip_files: csv_file = utils.extract(zip_file) type = os.path.basename(csv_file).split("_")[0] with open(csv_file, "r") as fin: reader = csv.reader(fin) if verbose: reader = iter(tqdm(reader, "Loading %s" % csv_file, utils.get_line_count(csv_file))) fields = next(reader) if "relidx" in csv_file: for index, token in reader: relation_vocab.append(token) else: entity_type_vocab.append(type) inv_entity_type_offset[type] = len(entity_vocab) num_entity = 0 for index, token in reader: entity_vocab.append("%s (%s)" % (type, token)) num_entity += 1 entity_type2num.append(num_entity) edge_split = dataset.get_edge_split() triplets = [] num_samples = [] num_samples_with_neg = [] negative_heads = [] negative_tails = [] for key in ["train", "valid", "test"]: split_dict = edge_split[key] h = torch.as_tensor(split_dict["head"]) t = torch.as_tensor(split_dict["tail"]) r = torch.as_tensor(split_dict["relation"]) h_type = torch.tensor([inv_entity_type_offset[h] for h in split_dict["head_type"]]) t_type = torch.tensor([inv_entity_type_offset[t] for t in split_dict["tail_type"]]) h = h + h_type t = t + t_type triplet = torch.stack([h, t, r], dim=-1) triplets.append(triplet) num_samples.append(len(h)) if "head_neg" in split_dict: neg_h = torch.as_tensor(split_dict["head_neg"]) neg_t = torch.as_tensor(split_dict["tail_neg"]) neg_h = neg_h + h_type.unsqueeze(-1) neg_t = neg_t + t_type.unsqueeze(-1) negative_heads.append(neg_h) negative_tails.append(neg_t) num_samples_with_neg.append(len(h)) else: num_samples_with_neg.append(0) triplets = torch.cat(triplets) self.load_triplet(triplets, entity_vocab=entity_vocab, relation_vocab=relation_vocab) entity_type_vocab, inv_entity_type_vocab = self._standarize_vocab(entity_type_vocab, None) self.entity_type_vocab = entity_type_vocab self.inv_entity_type_vocab = inv_entity_type_vocab self.num_samples = num_samples self.num_samples_with_neg = num_samples_with_neg self.negative_heads = torch.cat(negative_heads) self.negative_tails = torch.cat(negative_tails) node_type = [] for i, num_entity in enumerate(entity_type2num): node_type += [i] * num_entity with self.graph.node(): self.graph.node_type = torch.tensor(node_type) def split(self, test_negative=True): offset = 0 neg_offset = 0 splits = [] for num_sample, num_sample_with_neg in zip(self.num_samples, self.num_samples_with_neg): if test_negative and num_sample_with_neg: pos_h, pos_t, pos_r = self[offset: offset + num_sample].t() neg_h = self.negative_heads[neg_offset: neg_offset + num_sample_with_neg] neg_t = self.negative_tails[neg_offset: neg_offset + num_sample_with_neg] num_negative = neg_h.shape[-1] h = pos_h.unsqueeze(-1).repeat(2, num_negative + 1) t = pos_t.unsqueeze(-1).repeat(2, num_negative + 1) r = pos_r.unsqueeze(-1).repeat(2, num_negative + 1) t[:num_sample_with_neg, 1:] = neg_t h[num_sample_with_neg:, 1:] = neg_h split = torch.stack([h, t, r], dim=-1) else: split = torch_data.Subset(self, range(offset, offset + num_sample)) splits.append(split) offset += num_sample neg_offset += num_sample_with_neg return splits
14,201
39.005634
105
py
NBFNet
NBFNet-master/nbfnet/util.py
import os import time import logging import argparse import yaml import jinja2 from jinja2 import meta import easydict import torch from torch.utils import data as torch_data from torch import distributed as dist from torchdrug import core, utils from torchdrug.utils import comm logger = logging.getLogger(__file__) def get_root_logger(file=True): logger = logging.getLogger("") logger.setLevel(logging.INFO) format = logging.Formatter("%(asctime)-10s %(message)s", "%H:%M:%S") if file: handler = logging.FileHandler("log.txt") handler.setFormatter(format) logger.addHandler(handler) return logger def create_working_directory(cfg): file_name = "working_dir.tmp" world_size = comm.get_world_size() if world_size > 1 and not dist.is_initialized(): comm.init_process_group("nccl", init_method="env://") working_dir = os.path.join(os.path.expanduser(cfg.output_dir), cfg.task["class"], cfg.dataset["class"], cfg.task.model["class"], time.strftime("%Y-%m-%d-%H-%M-%S")) # synchronize working directory if comm.get_rank() == 0: with open(file_name, "w") as fout: fout.write(working_dir) os.makedirs(working_dir) comm.synchronize() if comm.get_rank() != 0: with open(file_name, "r") as fin: working_dir = fin.read() comm.synchronize() if comm.get_rank() == 0: os.remove(file_name) os.chdir(working_dir) return working_dir def detect_variables(cfg_file): with open(cfg_file, "r") as fin: raw = fin.read() env = jinja2.Environment() ast = env.parse(raw) vars = meta.find_undeclared_variables(ast) return vars def load_config(cfg_file, context=None): with open(cfg_file, "r") as fin: raw = fin.read() template = jinja2.Template(raw) instance = template.render(context) cfg = yaml.safe_load(instance) cfg = easydict.EasyDict(cfg) return cfg def parse_args(): parser = argparse.ArgumentParser() parser.add_argument("-c", "--config", help="yaml configuration file", required=True) parser.add_argument("-s", "--seed", help="random seed for PyTorch", type=int, default=1024) args, unparsed = parser.parse_known_args() # get dynamic arguments defined in the config file vars = detect_variables(args.config) parser = argparse.ArgumentParser() for var in vars: parser.add_argument("--%s" % var, required=True) vars = parser.parse_known_args(unparsed)[0] vars = {k: utils.literal_eval(v) for k, v in vars._get_kwargs()} return args, vars def build_solver(cfg, dataset): train_set, valid_set, test_set = dataset.split() if comm.get_rank() == 0: logger.warning(dataset) logger.warning("#train: %d, #valid: %d, #test: %d" % (len(train_set), len(valid_set), len(test_set))) if "fast_test" in cfg: if comm.get_rank() == 0: logger.warning("Quick test mode on. Only evaluate on %d samples for valid / test." % cfg.fast_test) g = torch.Generator() g.manual_seed(1024) valid_set = torch_data.random_split(valid_set, [cfg.fast_test, len(valid_set) - cfg.fast_test], generator=g)[0] test_set = torch_data.random_split(test_set, [cfg.fast_test, len(test_set) - cfg.fast_test], generator=g)[0] if hasattr(dataset, "num_relation"): cfg.task.model.num_relation = dataset.num_relation task = core.Configurable.load_config_dict(cfg.task) cfg.optimizer.params = task.parameters() optimizer = core.Configurable.load_config_dict(cfg.optimizer) solver = core.Engine(task, train_set, valid_set, test_set, optimizer, **cfg.engine) if "checkpoint" in cfg: solver.load(cfg.checkpoint) return solver
3,838
30.467213
119
py
NBFNet
NBFNet-master/nbfnet/__init__.py
0
0
0
py
gca-rom
gca-rom-main/main.py
import sys sys.path.append('../') import torch from gca_rom import network, pde, loader, plotting, preprocessing, training, initialization, testing, error import numpy as np if __name__ == "__main__": problem_name, variable, mu1_range, mu2_range = pde.problem(int(sys.argv[1])) print("PROBLEM: ", problem_name, "for variable ", variable, "\n") AE_Params = network.AE_Params device = initialization.set_device() initialization.set_reproducibility(AE_Params) initialization.set_path(AE_Params) dataset_dir = '../dataset/'+problem_name+'_unstructured.mat' dataset = loader.LoadDataset(dataset_dir, variable) dataset_graph, graph_loader, train_loader, test_loader, \ val_loader, scaler_all, scaler_test, xx, yy, var, VAR_all, VAR_test, \ train_trajectories, test_trajectories = preprocessing.graphs_dataset(dataset, AE_Params) mu1, mu2 = np.meshgrid(mu1_range, mu2_range) params = torch.tensor(np.vstack((mu1.T, mu2.T)).reshape(2, -1).T) params = params.to(device) print('Shape of parameter space:', params.shape, '\n') model = network.Net() model = model.to(device) optimizer = torch.optim.Adam(model.parameters(), lr=AE_Params.learning_rate, weight_decay=AE_Params.weight_decay) scheduler = torch.optim.lr_scheduler.MultiStepLR(optimizer, milestones=AE_Params.miles, gamma=AE_Params.gamma) history = dict(train=[], l1=[], l2=[]) history_test = dict(test=[], l1=[], l2=[]) min_test_loss = np.Inf try: model.load_state_dict(torch.load(AE_Params.net_dir+AE_Params.net_name+AE_Params.net_run+'.pt', map_location=torch.device('cpu'))) print('Loading saved network') except FileNotFoundError: print('Training net') # with torch.autograd.profiler.profile() as prof: for epoch in range(AE_Params.max_epochs): train_rmse = training.train(model, optimizer, device, scheduler, params, train_loader, train_trajectories, AE_Params, history) if AE_Params.cross_validation: test_rmse = training.val(model, device, params, test_loader, test_trajectories, AE_Params, history_test) print("Epoch[{}/{}, train_mse loss:{}, test_mse loss:{}".format(epoch + 1, AE_Params.max_epochs, history['train'][-1], history_test['test'][-1])) else: test_rmse = train_rmse print("Epoch[{}/{}, train_mse loss:{}".format(epoch + 1, AE_Params.max_epochs, history['train'][-1])) if test_rmse < min_test_loss: min_test_loss = test_rmse best_epoch = epoch torch.save(model.state_dict(), AE_Params.net_dir+AE_Params.net_name+AE_Params.net_run+'.pt') if AE_Params.tolerance >= train_rmse: print('Early stopping!') break np.save(AE_Params.net_dir+'history'+AE_Params.net_run+'.npy', history) np.save(AE_Params.net_dir+'history_test'+AE_Params.net_run+'.npy', history_test) # print(prof.key_averages().table(sort_by="self_cpu_time_total")) print("\nLoading best network for epoch: ", best_epoch) model.load_state_dict(torch.load(AE_Params.net_dir+AE_Params.net_name+AE_Params.net_run+'.pt', map_location=torch.device('cpu'))) model.to("cpu") params = params.to("cpu") vars = "GCA-ROM" results, latents_map, latents_gca = testing.evaluate(VAR_all, model, graph_loader, params, AE_Params, range(params.shape[0])) plotting.plot_loss(AE_Params) plotting.plot_latent(AE_Params, latents_map, latents_gca) plotting.plot_error(results, VAR_all, scaler_all, AE_Params, mu1_range, mu2_range, params, train_trajectories, vars) N = 5 snapshots = np.arange(params.shape[0]).tolist() np.random.shuffle(snapshots) for SNAP in snapshots[0:N]: plotting.plot_fields(SNAP, results, scaler_all, AE_Params, dataset, xx, yy, params) plotting.plot_error_fields(SNAP, results, VAR_all, scaler_all, AE_Params, dataset, xx, yy, params) results_test, _, _ = testing.evaluate(VAR_test, model, val_loader, params, AE_Params, test_trajectories) error_abs, norm = error.compute_error(results_test, VAR_test, scaler_test, AE_Params) error.print_error(error_abs, norm, vars) error.save_error(error_abs, norm, AE_Params, vars)
4,087
45.454545
157
py
gca-rom
gca-rom-main/examples/advection.py
import os # HYPER-PARAMETERS FOR POISSON problem_name = "advection" scalers_type = "sampling-feature" scalers_fun = "standard" skip_connection = "skip1" pn = 2 st = 4 sf = 3 sk = 1 train_rate = 30 ffc_nodes = 200 latent_nodes = 100 btt_nodes = 15 lambda_map = 1e1 hidden_channels = 2 print("\n\nRUN MAIN FOR PROBLEM "+problem_name+" WITH SCALING: "+scalers_type+" WITH "+scalers_fun+" FUNCTION") print("TRAIN RATE = "+str(train_rate)+" and FFC NODES = "+str(ffc_nodes)+" and HIDDEN CHANNELS = "+str(hidden_channels)+" and SKIP = "+skip_connection) print("LATENT NODES = "+str(latent_nodes)+" and BTT NODES = "+str(btt_nodes)+" and LAMBDA = "+str(lambda_map)) os.system("python3 ../main.py %s %s %s %s %s %s %s %s %s %s" %(pn, st, sf, sk, train_rate, ffc_nodes, latent_nodes, btt_nodes, lambda_map, hidden_channels))
819
34.652174
156
py
gca-rom
gca-rom-main/examples/navier-stokes.py
import os # HYPER-PARAMETERS FOR POISSON problem_names = ["VX_navier_stokes", "VY_navier_stokes", "P_navier_stokes"] comp = 2 problem_name = problem_names[comp] scalers_type = "sampling-feature" scalers_fun = "standard" skip_connection = "skip1" pn = 4 + comp st = 4 sf = 3 sk = 1 train_rate = 10 ffc_nodes = 200 latent_nodes = 100 btt_nodes = 25 lambda_map = 1e0 hidden_channels = 3 print("\n\nRUN MAIN FOR PROBLEM "+problem_name+" WITH SCALING: "+scalers_type+" WITH "+scalers_fun+" FUNCTION") print("TRAIN RATE = "+str(train_rate)+" and FFC NODES = "+str(ffc_nodes)+" and HIDDEN CHANNELS = "+str(hidden_channels)+" and SKIP = "+skip_connection) print("LATENT NODES = "+str(latent_nodes)+" and BTT NODES = "+str(btt_nodes)+" and LAMBDA = "+str(lambda_map)) os.system("python3 ../main.py %s %s %s %s %s %s %s %s %s %s" %(pn, st, sf, sk, train_rate, ffc_nodes, latent_nodes, btt_nodes, lambda_map, hidden_channels))
919
35.8
156
py
gca-rom
gca-rom-main/examples/graetz.py
import os # HYPER-PARAMETERS FOR POISSON problem_name = "graetz" scalers_type = "sampling-feature" scalers_fun = "standard" skip_connection = "skip1" pn = 3 st = 4 sf = 3 sk = 1 train_rate = 30 ffc_nodes = 200 latent_nodes = 50 btt_nodes = 25 lambda_map = 1e1 hidden_channels = 2 print("\n\nRUN MAIN FOR PROBLEM "+problem_name+" WITH SCALING: "+scalers_type+" WITH "+scalers_fun+" FUNCTION") print("TRAIN RATE = "+str(train_rate)+" and FFC NODES = "+str(ffc_nodes)+" and HIDDEN CHANNELS = "+str(hidden_channels)+" and SKIP = "+skip_connection) print("LATENT NODES = "+str(latent_nodes)+" and BTT NODES = "+str(btt_nodes)+" and LAMBDA = "+str(lambda_map)) os.system("python3 ../main.py %s %s %s %s %s %s %s %s %s %s" %(pn, st, sf, sk, train_rate, ffc_nodes, latent_nodes, btt_nodes, lambda_map, hidden_channels))
815
34.478261
156
py
gca-rom
gca-rom-main/examples/poisson.py
import os # HYPER-PARAMETERS FOR POISSON problem_name = "poisson" scalers_type = "sampling-feature" scalers_fun = "standard" skip_connection = "skip1" pn = 1 st = 4 sf = 3 sk = 1 train_rate = 30 ffc_nodes = 200 latent_nodes = 50 btt_nodes = 15 lambda_map = 1e1 hidden_channels = 3 print("\n\nRUN MAIN FOR PROBLEM "+problem_name+" WITH SCALING: "+scalers_type+" WITH "+scalers_fun+" FUNCTION") print("TRAIN RATE = "+str(train_rate)+" and FFC NODES = "+str(ffc_nodes)+" and HIDDEN CHANNELS = "+str(hidden_channels)+" and SKIP = "+skip_connection) print("LATENT NODES = "+str(latent_nodes)+" and BTT NODES = "+str(btt_nodes)+" and LAMBDA = "+str(lambda_map)) os.system("python3 ../main.py %s %s %s %s %s %s %s %s %s %s" %(pn, st, sf, sk, train_rate, ffc_nodes, latent_nodes, btt_nodes, lambda_map, hidden_channels))
816
34.521739
156
py
gca-rom
gca-rom-main/gca_rom/error.py
import numpy as np from gca_rom import scaling def save_error(error, norm, AE_Params, vars): """ save_error(error: List[float], norm: List[float], AE_Params: object, vars: str) This function takes in two lists of same length, error and norm, computed on the whole dataset for plotting reasons, and saves the relative error, along with the max, mean and min to a txt file. The relative error is calculated as error/norm for each corresponding element. The file is saved with a specific naming convention: AE_Params.net_dir + 'relative_errors' + AE_Params.net_run + vars + '.txt' Parameters: error (List[float]): A list of error values. norm (List[float]): A list of norm values of same length as error. AE_Params (object): An object containing information required to form the file name. vars (str): A string to be appended to the file name. """ error = np.array(error) norm = np.array(norm) rel_error = error/norm np.savetxt(AE_Params.net_dir+'relative_errors'+AE_Params.net_run+vars+'.txt', [max(rel_error), sum(rel_error)/len(rel_error), min(rel_error)]) def print_error(error, norm, vars): """ print_error(error: List[float], norm: List[float], vars: str) This function takes in two lists error and norm of same length and prints their absolute and relative errors, along with the max, mean and min of both the absolute and relative errors. The relative error is calculated as error/norm for each corresponding elements. Parameters: error (List[float]): A list of error values. norm (List[float]): A list of norm values of same length as error. vars (str): A string to describe the type of field. """ error = np.array(error) norm = np.array(norm) rel_error = error/norm print("\nMaximum absolute error for field "+vars+" = ", max(error)) print("Mean absolute error for field "+vars+" = ", sum(error)/len(error)) print("Minimum absolute error for field "+vars+" = ", min(error)) print("\nMaximum relative error for field "+vars+" = ", max(rel_error)) print("Mean relative error for field "+vars+" = ", sum(rel_error)/len(rel_error)) print("Minimum relative error for field "+vars+" = ", min(rel_error)) def compute_error(res, VAR, scaler, AE_Params): """ Compute the absolute error and norm between the original data and the data generated by the autoencoder :param res: Resulting data generated by the autoencoder (numpy ndarray) :param VAR: Original data (numpy ndarray) :param scaler: Scaler object for scaling the data :param AE_Params: Autoencoder parameters (object) :return: error_abs_list, norm_z_list, lists of absolute error and norm for each snapshot of data """ error_abs_list = list() norm_z_list = list() Z = scaling.inverse_scaling(VAR, scaler, AE_Params.scaling_type) Z_net = scaling.inverse_scaling(res, scaler, AE_Params.scaling_type) for snap in range(VAR.shape[0]): error_abs = np.linalg.norm(abs(Z[:, snap] - Z_net[:, snap])) norm_z = np.linalg.norm(Z[:, snap], 2) error_abs_list.append(error_abs) norm_z_list.append(norm_z) return error_abs_list, norm_z_list
3,215
43.666667
209
py
gca-rom
gca-rom-main/gca_rom/preprocessing.py
import numpy as np import torch from torch_geometric.data import Data from torch_geometric.loader import DataLoader from gca_rom import scaling def graphs_dataset(dataset, AE_Params): """ graphs_dataset: function to process and scale the input dataset for graph autoencoder model. Inputs: dataset: an object containing the dataset to be processed. AE_Params: an object containing the hyperparameters of the graph autoencoder model. Outputs: dataset_graph: an object containing the processed and scaled dataset. loader: a DataLoader object of the processed and scaled dataset. train_loader: a DataLoader object of the training set. test_loader: a DataLoader object of the test set. val_loader: a DataLoader object of the validation set. scaler_all: a scaler object to scale the entire dataset. scaler_test: a scaler object to scale the test set. xx: an array of the x-coordinate of the nodes. yy: an array of the y-coordinate of the nodes. var: an array of the node features. VAR_all: an array of the scaled node features of the entire dataset. VAR_test: an array of the scaled node features of the test set. train_snapshots: a list of indices of the training set. test_snapshots: a list of indices of the test set. """ xx = dataset.xx yy = dataset.yy var = dataset.U # PROCESSING DATASET num_nodes = var.shape[0] num_graphs = var.shape[1] print("Number of nodes processed: ", num_nodes) print("Number of graphs processed: ", num_graphs) total_sims = int(num_graphs) rate = AE_Params.rate/100 train_sims = int(rate * total_sims) test_sims = total_sims - train_sims main_loop = np.arange(total_sims).tolist() np.random.shuffle(main_loop) train_snapshots = main_loop[0:train_sims] train_snapshots.sort() test_snapshots = main_loop[train_sims:total_sims] test_snapshots.sort() ## FEATURE SCALING var_test = dataset.U[:, test_snapshots] scaling_type = AE_Params.scaling_type scaler_all, VAR_all = scaling.tensor_scaling(var, scaling_type) scaler_test, VAR_test = scaling.tensor_scaling(var_test, scaling_type) print("Shape of scaled dataset: ", VAR_all.shape) print("Shape of scaled test set: ", VAR_test.shape) graphs = [] edge_index = torch.t(dataset.E) - 1 for graph in range(num_graphs): pos = torch.cat((xx[:, graph].unsqueeze(1), yy[:, graph].unsqueeze(1)), 1) ei = torch.index_select(pos, 0, edge_index[0, :]) ej = torch.index_select(pos, 0, edge_index[1, :]) edge_attr = ej - ei edge_attr = torch.sqrt(torch.pow(edge_attr[:, 0], 2) + torch.pow(edge_attr[:, 1], 2)) node_features = VAR_all[graph, :] dataset_graph = Data(x=node_features, edge_index=edge_index, edge_attr=edge_attr, pos=pos) graphs.append(dataset_graph) AE_Params.num_nodes = dataset_graph.num_nodes train_dataset = [graphs[i] for i in train_snapshots] test_dataset = [graphs[i] for i in test_snapshots] loader = DataLoader(graphs, batch_size=1) train_loader = DataLoader(train_dataset, batch_size=train_sims, shuffle=False) test_loader = DataLoader(test_dataset, batch_size=test_sims, shuffle=False) val_loader = DataLoader(test_dataset, batch_size=1, shuffle=False) return dataset_graph, loader, train_loader, test_loader, \ val_loader, scaler_all, scaler_test, xx, yy, var, VAR_all, VAR_test, train_snapshots, test_snapshots
3,515
38.954545
112
py
gca-rom
gca-rom-main/gca_rom/plotting.py
import numpy as np import matplotlib.pyplot as plt from gca_rom import scaling import matplotlib.gridspec as gridspec import matplotlib.cm as cm from matplotlib import ticker from matplotlib.ticker import MaxNLocator from mpl_toolkits.axes_grid1 import make_axes_locatable params = {'legend.fontsize': 'x-large', 'axes.labelsize': 'x-large', 'axes.titlesize':'x-large', 'xtick.labelsize':'x-large', 'ytick.labelsize':'x-large'} plt.rcParams.update(params) def plot_loss(AE_Params): """ Plots the history of losses during the training of the autoencoder. Attributes: AE_Params (namedtuple): An object containing the parameters of the autoencoder. """ history = np.load(AE_Params.net_dir+'history'+AE_Params.net_run+'.npy', allow_pickle=True).item() history_test = np.load(AE_Params.net_dir+'history_test'+AE_Params.net_run+'.npy', allow_pickle=True).item() ax = plt.figure().gca() ax.semilogy(history['l1']) ax.semilogy(history['l2']) ax.semilogy(history_test['l1'], '--') ax.semilogy(history_test['l2'], '--') plt.ylabel('Loss') plt.ylabel('Epochs') plt.title('Loss over training epochs') plt.legend(['loss_mse', 'loss_map', 'loss_test_mse', 'loss_test_map']) plt.savefig(AE_Params.net_dir+'history_losses'+AE_Params.net_run+'.png', bbox_inches='tight', dpi=500) def plot_latent(AE_Params, latents, latents_estimation): """ Plot the original and estimated latent spaces Parameters: AE_Params (obj): object containing the Autoencoder parameters latents (tensor): tensor of original latent spaces latents_estimation (tensor): tensor of estimated latent spaces """ plt.figure() for i1 in range(AE_Params.bottleneck_dim): plt.plot(latents[:,i1].detach(), '--') plt.plot(latents_estimation[:,i1].detach(),'-') plt.savefig(AE_Params.net_dir+'latents'+AE_Params.net_run+'.png', bbox_inches='tight') green_diamond = dict(markerfacecolor='g', marker='D') _, ax = plt.subplots() ax.boxplot(latents_estimation.detach().numpy(), flierprops=green_diamond) plt.savefig(AE_Params.net_dir+'box_plot_latents'+AE_Params.net_run+'.png', bbox_inches='tight', dpi=500) def plot_error(res, VAR_all, scaler_all, AE_Params, mu1_range, mu2_range, params, train_trajectories, vars): """ This function plots the relative error between the predicted and actual results. Parameters: res (ndarray): The predicted results VAR_all (ndarray): The actual results scaler_all (object): The scaler object used for scaling the results AE_Params (object): The AE_Params object holding the necessary hyperparameters mu1_range (ndarray): Range of the first input variable mu2_range (ndarray): Range of the second input variable params (ndarray): The input variables train_trajectories (ndarray): The indices of the training data vars (str): The name of the variable being plotted """ u_hf = scaling.inverse_scaling(VAR_all, scaler_all, AE_Params.scaling_type) u_app = scaling.inverse_scaling(res, scaler_all, AE_Params.scaling_type) error = np.linalg.norm(u_app - u_hf, axis=0) / np.linalg.norm(u_hf, axis=0) X1, X2 = np.meshgrid(mu1_range, mu2_range, indexing='ij') output = np.reshape(error, (len(mu1_range), len(mu2_range))) fig = plt.figure('Relative Error '+vars) ax = fig.add_subplot(projection='3d') ax.plot_surface(X1, X2, output, cmap=cm.coolwarm, color='blue') ax.contour(X1, X2, output, zdir='z', offset=output.min(), cmap=cm.coolwarm) ax.set(xlim=tuple([mu1_range[0], mu1_range[-1]]), ylim=tuple([mu2_range[0], mu2_range[-1]]), xlabel='$\mu_1$', ylabel='$\mu_2$', zlabel='$\\epsilon_{GCA}(\\mathbf{\mu})$') ax.plot(params[train_trajectories,0], params[train_trajectories,1], output.min()*np.ones(len(params[train_trajectories,1])), '*r') plt.ticklabel_format(axis='z', style='sci', scilimits=(0, 0)) ax.set_title('Relative Error '+vars) ax.zaxis.offsetText.set_visible(False) exponent_axis = np.floor(np.log10(max(ax.get_zticks()))).astype(int) ax.text2D(0.9, 0.82, "1e"+str(exponent_axis), transform=ax.transAxes, fontsize="x-large") plt.tight_layout() plt.savefig(AE_Params.net_dir+'relative_error_'+vars+AE_Params.net_run+'.png', transparent=True, dpi=500) def plot_fields(SNAP, results, scaler_all, AE_Params, dataset, xx, yy, params): """ Plots the field solution for a given snapshot. The function takes in the following inputs: SNAP: integer value indicating the snapshot to be plotted. results: array of shape (num_samples, num_features), representing the network's output. scaler_all: instance of the scaler used to scale the data. AE_Params: instance of the Autoencoder parameters class containing information about the network architecture and training. dataset: array of shape (num_samples, 3), representing the triangulation of the spatial domain. xx: array of shape (num_samples, num_features), containing the x-coordinates of the domain. yy: array of shape (num_samples, num_features), containing the y-coordinates of the domain. params: array of shape (num_features,), containing the parameters associated with each snapshot. The function generates a plot of the field solution and saves it to disk using the filepath specified in AE_Params.net_dir. """ triang = np.asarray(dataset.T - 1) cmap = cm.get_cmap(name='jet', lut=None) # fig = plt.figure(figsize=(14, 6)) gs1 = gridspec.GridSpec(1, 1) ax = plt.subplot(gs1[0, 0]) Z_net = scaling.inverse_scaling(results, scaler_all, AE_Params.scaling_type) z_net = Z_net[:, SNAP] # ax.triplot(xx[:, SNAP], yy[:, SNAP], triang, lw=0.5, color='black') cs = ax.tricontourf(xx[:, SNAP], yy[:, SNAP], triang, z_net, 100, cmap=cmap) divider = make_axes_locatable(ax) cax = divider.append_axes("right", size="5%", pad=0.1) cbar = plt.colorbar(cs, cax=cax) tick_locator = MaxNLocator(nbins=5) cbar.locator = tick_locator cbar.formatter.set_powerlimits((0, 0)) cbar.update_ticks() plt.tight_layout() ax.set_aspect('equal', 'box') ax.set_title('Solution field for $\mu$ = '+str(np.around(params[SNAP].detach().numpy(), 2))) plt.savefig(AE_Params.net_dir+'field_solution_'+str(SNAP)+''+AE_Params.net_run+'.png', bbox_inches='tight', dpi=500) def plot_error_fields(SNAP, results, VAR_all, scaler_all, AE_Params, dataset, xx, yy, params): """ This function plots a contour map of the error field of a given solution of a scalar field. The error is computed as the absolute difference between the true solution and the predicted solution, normalized by the 2-norm of the true solution. Inputs: SNAP: int, snapshot of the solution to be plotted results: np.array, predicted solution VAR_all: np.array, true solution scaler_all: np.array, scaling information used in the prediction AE_Params: class, model architecture and training parameters dataset: np.array, mesh information xx: np.array, x-coordinate of the mesh yy: np.array, y-coordinate of the mesh params: np.array, model parameters """ triang = np.asarray(dataset.T - 1) cmap = cm.get_cmap(name='jet', lut=None) # fig = plt.figure(figsize=(14, 6)) gs1 = gridspec.GridSpec(1, 1) ax = plt.subplot(gs1[0, 0]) Z = scaling.inverse_scaling(VAR_all, scaler_all, AE_Params.scaling_type) Z_net = scaling.inverse_scaling(results, scaler_all, AE_Params.scaling_type) z = Z[:, SNAP] z_net = Z_net[:, SNAP] error = abs(z - z_net)/np.linalg.norm(z, 2) # ax.triplot(xx[:,SNAP], yy[:, SNAP], triang, lw=0.5, color='black') cs = ax.tricontourf(xx[:, SNAP], yy[:, SNAP], triang, error, 100, cmap=cmap) divider = make_axes_locatable(ax) cax = divider.append_axes("right", size="5%", pad=0.1) cbar = plt.colorbar(cs, cax=cax) tick_locator = MaxNLocator(nbins=5) cbar.locator = tick_locator cbar.formatter.set_powerlimits((0, 0)) cbar.update_ticks() plt.tight_layout() ax.set_aspect('equal', 'box') ax.set_title('Error field for $\mu$ = '+str(np.around(params[SNAP].detach().numpy(), 2))) plt.savefig(AE_Params.net_dir+'error_field_'+str(SNAP)+''+AE_Params.net_run+'.png', bbox_inches='tight', dpi=500)
8,379
46.613636
175
py
gca-rom
gca-rom-main/gca_rom/network.py
import sys import torch from torch import nn from gca_rom import gca, scaling, pde problem_name, variable, mu1_range, mu2_range = pde.problem(int(sys.argv[1])) class AE_Params(): """Class that holds the hyperparameters for the autoencoder model. Args: sparse_method (str): The method to use for sparsity constraint. rate (int): Amount of data used in training. seed (int): Seed for the random number generator. bottleneck_dim (int): The dimension of the bottleneck layer. tolerance (float): The tolerance value for stopping the training. lambda_map (float): The weight for the map loss. learning_rate (float): The learning rate for the optimizer. ffn (int): The number of feed-forward layers. in_channels (int): The number of input channels. hidden_channels (list): The number of hidden channels for each layer. act (function): The activation function to use. nodes (int): The number of nodes in each hidden layer. skip (int): The number of skipped connections. layer_vec (list): The structure of the network. net_name (str): The name of the network. scaler_name (str): The name of the scaler used for preprocessing. weight_decay (float): The weight decay for the optimizer. max_epochs (int): The maximum number of epochs to run training for. miles (list): The miles for learning rate update in scheduler. gamma (float): The gamma value for the optimizer. num_nodes (int): The number of nodes in the network. scaling_type (int): The type of scaling to use for preprocessing. net_dir (str): The directory to save the network in. cross_validation (bool): Whether to perform cross-validation. """ def __init__(self): self.scaling_type = int(sys.argv[2]) _, self.scaler_name = scaling.scaler_functions(int(sys.argv[3])) self.skip = int(sys.argv[4]) self.rate = int(sys.argv[5]) self.sparse_method = 'L1_mean' self.ffn = int(sys.argv[6]) self.nodes = int(sys.argv[7]) self.bottleneck_dim = int(sys.argv[8]) self.lambda_map = float(sys.argv[9]) self.in_channels = int(sys.argv[10]) self.seed = 10 self.tolerance = 1e-6 self.learning_rate = 0.001 self.hidden_channels = [1]*self.in_channels self.act = torch.tanh self.layer_vec=[2, self.nodes, self.nodes, self.nodes, self.nodes, self.bottleneck_dim] self.net_name = problem_name self.net_run = '_' + self.scaler_name self.weight_decay = 0.00001 self.max_epochs = 5000 self.miles = [] self.gamma = 0.0001 self.num_nodes = 0 self.net_dir = './' + problem_name + '/' + self.net_run+ '/' + variable + '_' + self.net_name + '_lmap' + str(self.lambda_map) + '_btt' + str(self.bottleneck_dim) \ + '_seed' + str(self.seed) + '_lv' + str(len(self.layer_vec)-2) + '_hc' + str(len(self.hidden_channels)) + '_nd' + str(self.nodes) \ + '_ffn' + str(self.ffn) + '_skip' + str(self.skip) + '_lr' + str(self.learning_rate) + '_sc' + str(self.scaling_type) + '_rate' + str(self.rate) + '/' self.cross_validation = True AE_Params = AE_Params() class Net(torch.nn.Module): """ Class Net --------- A PyTorch neural network class which consists of encoder, decoder and mapping modules. Attributes ---------- encoder : gca.Encoder An encoder module from the gca module. decoder : gca.Decoder A decoder module from the gca module. act_map : AE_Params.act The activation map specified in the AE_Params. layer_vec : AE_Params.layer_vec The layer vector specified in the AE_Params. Shape of the layers for the parameter space mapping. steps : int Number of layers for the parameter mapping MLP. maptovec : nn.ModuleList A list of linear layers for mapping. Methods ------- solo_encoder(data) Encodes the input data using the encoder module. Returns the encoded representation. solo_decoder(x, data) Decodes the encoded representation and the input data using the decoder module. Returns the decoded output. mapping(x) Maps the input using the linear modules in maptovec. Returns the mapped output. forward(data, parameters) Runs a forward pass through the network using the input data and parameters. Returns the decoded output, encoded representation, and estimated encoded representation. """ def __init__(self): super().__init__() self.encoder = gca.Encoder(AE_Params.hidden_channels, AE_Params.bottleneck_dim, AE_Params.num_nodes, ffn=AE_Params.ffn, skip=AE_Params.skip) self.decoder = gca.Decoder(AE_Params.hidden_channels, AE_Params.bottleneck_dim, AE_Params.num_nodes, ffn=AE_Params.ffn, skip=AE_Params.skip) self.act_map = AE_Params.act self.layer_vec = AE_Params.layer_vec self.steps = len(self.layer_vec) - 1 self.maptovec = nn.ModuleList() for k in range(self.steps): self.maptovec.append(nn.Linear(self.layer_vec[k], self.layer_vec[k+1])) def solo_encoder(self,data): x = self.encoder(data) return x def solo_decoder(self,x, data): x = self.decoder(x, data) return x def mapping(self, x): idx = 0 for layer in self.maptovec: if(idx==self.steps): x = layer(x) else: x =self.act_map(layer(x)) idx += 1 return x def forward(self, data, parameters): z = self.solo_encoder(data) z_estimation = self.mapping(parameters) x = self.solo_decoder(z, data) # x = self.solo_decoder(z_estimation, data) return x, z, z_estimation
5,946
39.732877
179
py
gca-rom
gca-rom-main/gca_rom/training.py
import torch import torch.nn.functional as F def train(model, optimizer, device, scheduler, params, train_loader, train_trajectories, AE_Params, history): """Trains the autoencoder model. This function trains the autoencoder model using mean squared error (MSE) loss and a map loss, where the map loss is the MSE between the estimated z (latent space) and the actual z. The final loss is the sum of the MSE loss and the map loss multiplied by the weight `AE_Params.lambda_map`. The model is trained on the data from `train_loader` and the optimization process is performed using the `optimizer`. The learning rate is updated after every iteration using `scheduler`. Use of mini-batching for reducing the computational cost. Args: model (torch.nn.Module): The autoencoder model to be trained. optimizer (torch.optim.Optimizer): The optimizer to update the model parameters. device (str): The device to run the model on ('cuda' or 'cpu'). scheduler (torch.optim.lr_scheduler._LRScheduler): The learning rate scheduler to update the learning rate. params (torch.Tensor): Tensor containing the parameters of the model. train_loader (torch.utils.data.DataLoader): The data loader to provide the training data. train_trajectories (int): The number of training trajectories. AE_Params (dict): The dictionary containing the hyperparameters for the autoencoder model. history (dict): The dictionary to store the loss history. Returns: float: The average loss over all training examples. """ model.train() total_loss_train = total_examples = sum_loss = 0 total_loss_train_1 = total_loss_train_2 = 0 sum_loss_1 = sum_loss_2 = 0 for data in train_loader: optimizer.zero_grad() data = data.to(device) out, z, z_estimation = model(data, params[train_trajectories, :]) loss_train_mse = F.mse_loss(out, data.x, reduction='mean') loss_train_map = F.mse_loss(z_estimation, z, reduction='mean') loss_train = loss_train_mse + AE_Params.lambda_map * loss_train_map loss_train.backward() optimizer.step() sum_loss += loss_train.item() sum_loss_1 += loss_train_mse.item() sum_loss_2 += loss_train_map.item() total_examples += 1 scheduler.step() total_loss_train = sum_loss / total_examples total_loss_train_1 = sum_loss_1 / total_examples total_loss_train_2 = sum_loss_2 / total_examples history['train'].append(total_loss_train) history['l1'].append(total_loss_train_1) history['l2'].append(total_loss_train_2) return total_loss_train def val(model, device, params, test_loader, test_trajectories, AE_Params, history_test): """ Evaluate the performance of a model on a test set. This function calculates the mean of the total loss, the mean of loss_test_mse and the mean of loss_test_map for all test examples. The losses are computed as the mean squared error (MSE) between the model's predictions and the true target variables, and between the estimated latent code and the true latent code. The lambda_map weight balances the contribution of each loss term to the total loss. The function adds the computed loss values to the history_test dictionary. Parameters: model (torch.nn.Module): The model to be evaluated. device (torch.device): The device to use for computations (e.g. 'cpu' or 'cuda'). params (torch.Tensor): Tensor containing the parameters of the model. test_loader (torch.utils.data.DataLoader): The test data to use for evaluation. test_trajectories (int): The index of the test trajectory. AE_Params (object): Object containing hyperparameters for the model. history_test (dict): Dictionary to store the evaluation results. Returns: float: The mean of the total loss computed over all test examples. """ with torch.no_grad(): model.eval() total_loss_test = total_examples = sum_loss = 0 total_loss_test_1 = total_loss_test_2 = 0 sum_loss_1 = sum_loss_2 = 0 for data in test_loader: data = data.to(device) out, z, z_estimation = model(data, params[test_trajectories, :]) loss_test_mse = F.mse_loss(out, data.x, reduction='mean') loss_test_map = F.mse_loss(z_estimation, z, reduction='mean') loss_test = loss_test_mse + AE_Params.lambda_map * loss_test_map sum_loss += loss_test.item() sum_loss_1 += loss_test_mse.item() sum_loss_2 += loss_test_map.item() total_examples += 1 total_loss_test = sum_loss / total_examples total_loss_test_1 = sum_loss_1 / total_examples total_loss_test_2 = sum_loss_2 / total_examples history_test['test'].append(total_loss_test) history_test['l1'].append(total_loss_test_1) history_test['l2'].append(total_loss_test_2) return total_loss_test
5,063
48.647059
478
py
gca-rom
gca-rom-main/gca_rom/testing.py
import torch from tqdm import tqdm import numpy as np def evaluate(VAR, model, loader, params, AE_Params, test): """ This function evaluates the performance of a trained Autoencoder (AE) model. It encodes the input data using both the model's encoder and a mapping function, and decodes the resulting latent representations to obtain predicted solutions. The relative error between the two latent representations is also computed. Inputs: VAR: np.array, ground truth solution model: object, trained AE model loader: object, data loader for the input data params: np.array, model parameters AE_Params: class, model architecture and training parameters Returns: results: np.array, predicted solutions latents_map: np.array, latent representations obtained using the mapping function latents_gca: np.array, latent representations obtained using the AE encoder """ results = torch.zeros(VAR.shape[0], VAR.shape[1], 1) latents_map = torch.zeros(VAR.shape[0], AE_Params.bottleneck_dim) latents_gca = torch.zeros(VAR.shape[0], AE_Params.bottleneck_dim) index = 0 latents_error = list() with torch.no_grad(): for data in tqdm(loader): z_net = model.solo_encoder(data) z_map = model.mapping(params[test[index], :]) latents_map[index, :] = z_map latents_gca[index, :] = z_net lat_err = np.linalg.norm(z_net - z_map)/np.linalg.norm(z_net) latents_error.append(lat_err) results[index, :, :] = model.solo_decoder(z_map, data) index += 1 np.savetxt(AE_Params.net_dir+'latents'+AE_Params.net_run+'.csv', latents_map.detach(), delimiter =',') latents_error = np.array(latents_error) # print("\nMaximum relative error for latent = ", max(latents_error)) # print("Mean relative error for latent = ", sum(latents_error)/len(latents_error)) # print("Minimum relative error for latent = ", min(latents_error)) return results, latents_map, latents_gca
2,071
44.043478
110
py
gca-rom
gca-rom-main/gca_rom/initialization.py
import os import torch import numpy as np import random import warnings def set_device(): """ Returns the device to be used (GPU or CPU) Returns: device (str): The device to be used ('cuda' if GPU is available, 'cpu' otherwise) """ device = 'cuda' if torch.cuda.is_available() else 'cpu' print("Device used = ", device) torch.set_default_dtype(torch.float64) warnings.filterwarnings("ignore") return device def set_reproducibility(AE_Params): """ Sets the seed for reproducibility of results. Args: AE_Params (class): Contains the hyperparameters of the autoencoder """ seed = AE_Params.seed torch.manual_seed(seed) torch.cuda.manual_seed(seed) torch.cuda.manual_seed_all(seed) np.random.seed(seed) random.seed(seed) torch.backends.cudnn.benchmark = False torch.backends.cudnn.deterministic = True def set_path(AE_Params): """ Creates the directory path to store the network results. Args: AE_Params (class): Contains the hyperparameters of the autoencoder """ path = AE_Params.net_dir isExist = os.path.exists(path) if not isExist: os.makedirs(path, exist_ok=False)
1,224
21.685185
89
py
gca-rom
gca-rom-main/gca_rom/__init__.py
0
0
0
py
gca-rom
gca-rom-main/gca_rom/pde.py
import numpy as np def problem(argument): """ problem(argument: int) -> Tuple[str, str, np.ndarray, np.ndarray] This function takes in an integer argument and returns a tuple containing the problem name (str), variable name (str), mu1 (np.ndarray) and mu2 (np.ndarray) for the specified case. The possible values of argument are: Poisson problem Advection problem Graetz problem Navier-Stokes problem (variable = VX) Navier-Stokes problem (variable = VY) Navier-Stokes problem (variable = P) Returns: Tuple containing the problem name (str), variable name (str), mu1 (np.ndarray) and mu2 (np.ndarray) for the specified case. """ match argument: case 1: problem = "poisson" variable = 'U' mu1 = np.linspace(0.01, 10., 10) mu2 = np.linspace(0.01, 10., 10) case 2: problem = "advection" variable = 'U' mu1 = np.linspace(0., 6., 10) mu2 = np.linspace(-1.0, 1.0, 10) case 3: problem = "graetz" variable = 'U' mu1 = np.linspace(1., 3., 10) mu2 = np.linspace(0.01, 0.1, 20) case 4: problem = "navier_stokes" variable = 'VX' mu1 = np.linspace(0.5, 2., 21)[::2] mu2 = np.linspace(2., 0.5, 151)[::5] case 5: problem = "navier_stokes" variable = 'VY' mu1 = np.linspace(0.5, 2., 21)[::2] mu2 = np.linspace(2., 0.5, 151)[::5] case 6: problem = "navier_stokes" variable = 'P' mu1 = np.linspace(0.5, 2., 21)[::2] mu2 = np.linspace(2., 0.5, 151)[::5] return problem, variable, mu1, mu2
1,771
32.433962
184
py
gca-rom
gca-rom-main/gca_rom/loader.py
import sys from torch_geometric.data import Dataset import torch import scipy class LoadDataset(Dataset): """ A custom dataset class which loads data from a .mat file using scipy.io.loadmat. data_mat : scipy.io.loadmat The loaded data in a scipy.io.loadmat object. U : torch.Tensor The tensor representation of the specified variable from the data_mat. xx : torch.Tensor The tensor representation of the 'xx' key from the data_mat. Refers to X coordinates of the domain yy : torch.Tensor The tensor representation of the 'yy' key from the data_mat.Refers to Y coordinates of the domain T : torch.Tensor The tensor representation of the 'T' key from the data_mat, casted to int. Adjacency Matrix E : torch.Tensor The tensor representation of the 'E' key from the data_mat, casted to int. Connection Matrix __init__(self, root_dir, variable) Initializes the LoadDataset object by loading the data from the .mat file at the root_dir location and converting the specified variable to a tensor representation. """ def __init__(self, root_dir, variable): self.data_mat = scipy.io.loadmat(root_dir) self.U = torch.tensor(self.data_mat[variable]) self.xx = torch.tensor(self.data_mat['xx']) self.yy = torch.tensor(self.data_mat['yy']) self.T = torch.tensor(self.data_mat['T'].astype(int)) self.E = torch.tensor(self.data_mat['E'].astype(int))
1,488
41.542857
172
py
gca-rom
gca-rom-main/gca_rom/gca.py
import torch from torch import nn import torch.nn.functional as F from torch_geometric.nn import GMMConv class Encoder(torch.nn.Module): """ Encoder Class The Encoder class is a subclass of torch.nn.Module that implements a deep neural network for encoding graph data. It uses the Gaussian Mixture convolution (GMMConv) module to extract features from the graph structure and node features. The encoding is then passed through a feed-forward neural network with two fully connected layers to produce the final encoding. Arguments: hidden_channels (list): A list of hidden channel sizes for each layer of the GMMConv module. bottleneck (int): Size of the bottleneck layer in the feed-forward neural network. input_size (int): Size of the node features. ffn (int): Size of the intermediate layer in the feed-forward neural network. skip (bool): If True, the input node features will be concatenated with the GMMConv output at each layer. act (function): Activation function used in the GMMConv layers and feed-forward neural network. Defaults to F.elu. Methods: encoder(data): Encodes the graph data using the GMMConv module and feed-forward neural network. reset_parameters(): Resets the parameters of the GMMConv layers and feed-forward neural network. forward(data): A convenience function that calls the encoder method. """ def __init__(self, hidden_channels, bottleneck, input_size, ffn, skip, act=F.elu): super().__init__() self.hidden_channels = hidden_channels self.depth = len(self.hidden_channels) self.act = act self.ffn = ffn self.skip = skip self.bottleneck = bottleneck self.input_size = input_size self.down_convs = torch.nn.ModuleList() for i in range(self.depth-1): self.down_convs.append(GMMConv(self.hidden_channels[i], self.hidden_channels[i+1], dim=1, kernel_size=5)) self.fc_in1 = nn.Linear(self.input_size*self.hidden_channels[-1], self.ffn) self.fc_in2 = nn.Linear(self.ffn, self.bottleneck) self.reset_parameters() def encoder(self, data): edge_weight = data.edge_attr edge_index = data.edge_index x = data.x idx = 0 for layer in self.down_convs: x = self.act(layer(x, edge_index, edge_weight.unsqueeze(1))) if self.skip: x = x + data.x idx += 1 x = x.reshape(data.num_graphs, self.input_size * self.hidden_channels[-1]) x = self.act(self.fc_in1(x)) x = self.fc_in2(x) return x def reset_parameters(self): for conv in self.down_convs: conv.reset_parameters() for name, param in conv.named_parameters(): if 'bias' in name: nn.init.constant_(param, 0) else: nn.init.kaiming_uniform_(param) def forward(self,data): x = self.encoder(data) return x class Decoder(torch.nn.Module): """ Class Decoder A torch.nn.Module subclass for the decoder part of a neural network. Attributes: hidden_channels (list of ints): A list of hidden channel sizes. depth (int): The length of hidden_channels list. act (function): Activation function to use. ffn (int): Size of output after the first linear layer. skip (bool): Whether to add skip connections. bottleneck (int): Size of bottleneck layer. input_size (int): Size of input data. fc_out1 (torch.nn.Linear): Linear layer from bottleneck to ffn. fc_out2 (torch.nn.Linear): Linear layer from ffn to input_size * hidden_channels[-1]. up_convs (torch.nn.ModuleList): A list of GMMConv layers. Methods: decoder(self, x, data): Decodes the input data x and returns the output. reset_parameters(self): Resets the parameters of the up_convs layer. forward(self, x, data): Performs a forward pass on the input data x and returns the output. """ def __init__(self, hidden_channels, bottleneck, input_size, ffn, skip, act=F.elu): super().__init__() self.hidden_channels = hidden_channels self.depth = len(self.hidden_channels) self.act = act self.ffn = ffn self.skip = skip self.bottleneck = bottleneck self.input_size = input_size self.fc_out1 = nn.Linear(self.bottleneck, self.ffn) self.fc_out2 = nn.Linear(self.ffn, self.input_size * self.hidden_channels[-1]) self.up_convs = torch.nn.ModuleList() for i in range(self.depth-1): self.up_convs.append(GMMConv(self.hidden_channels[self.depth-1-i], self.hidden_channels[self.depth-i-2], dim=1, kernel_size=5)) self.reset_parameters() def decoder(self, x, data): edge_weight = data.edge_attr edge_index = data.edge_index x = self.act(self.fc_out1(x)) x = self.act(self.fc_out2(x)) h = x.reshape(data.num_graphs*self.input_size, self.hidden_channels[-1]) x = h idx = 0 for layer in self.up_convs: if (idx == self.depth - 2): x = layer(x, edge_index, edge_weight.unsqueeze(1)) else: x = self.act(layer(x, edge_index, edge_weight.unsqueeze(1))) if self.skip: x = x + h idx += 1 return x def reset_parameters(self): for conv in self.up_convs: conv.reset_parameters() for name, param in conv.named_parameters(): if 'bias' in name: nn.init.constant_(param, 0) else: nn.init.kaiming_uniform_(param) def forward(self, x, data): x = self.decoder(x, data) return x
5,918
36.226415
139
py
gca-rom
gca-rom-main/gca_rom/scaling.py
from sklearn import preprocessing import torch import sys def scaler_functions(k): match k: case 1: sc_name = "minmax" sc_fun = preprocessing.MinMaxScaler() case 2: sc_name = "robust" sc_fun = preprocessing.RobustScaler() case 3: sc_name = "standard" sc_fun = preprocessing.StandardScaler() return sc_fun, sc_name def tensor_scaling(tensor, scaling_type): scaling_fun_1, _ = scaler_functions(int(sys.argv[3])) scaling_fun_2, _ = scaler_functions(int(sys.argv[3])) match scaling_type: case 1: # print("SAMPLE SCALING") scale = scaling_fun_1.fit(tensor) scaled_data = torch.unsqueeze(torch.tensor(scale.transform(tensor)),0).permute(2, 1, 0) case 2: # print("FEATURE SCALING") scale = scaling_fun_1.fit(torch.t(tensor)) scaled_data = torch.unsqueeze(torch.tensor(scale.transform(torch.t(tensor))), 0).permute(1, 2, 0) case 3: # print("FEATURE-SAMPLE SCALING") scaler_f = scaling_fun_1.fit(torch.t(tensor)) temp = torch.tensor(scaler_f.transform(torch.t(tensor))) scaler_s = scaling_fun_2.fit(temp) scaled_data = torch.unsqueeze(torch.tensor(scaler_s.transform(temp)), 0).permute(1, 2, 0) scale = [scaler_f, scaler_s] case 4: # print("SAMPLE-FEATURE SCALING") scaler_s = scaling_fun_1.fit(tensor) temp = torch.t(torch.tensor(scaler_s.transform(tensor))) scaler_f = scaling_fun_2.fit(temp) scaled_data = torch.unsqueeze(torch.t(torch.tensor(scaler_f.transform(temp))), 0).permute(2, 1, 0) scale = [scaler_s, scaler_f] return scale, scaled_data def inverse_scaling(tensor, scale, scaling_type): match scaling_type: case 1: # print("SAMPLE SCALING") rescaled_data = torch.tensor(scale.inverse_transform(torch.t(torch.tensor(tensor[:, :, 0].detach().numpy().squeeze())))) case 2: # print("FEATURE SCALING") rescaled_data = torch.tensor(torch.t(torch.tensor(scale.inverse_transform(tensor[:, :, 0].detach().numpy().squeeze())))) case 3: # print("FEATURE-SAMPLE SCALING") scaler_f = scale[0] scaler_s = scale[1] rescaled_data = torch.t(torch.tensor(scaler_f.inverse_transform(torch.tensor(scaler_s.inverse_transform(tensor[:, :, 0].detach().numpy().squeeze()))))) case 4: # print("SAMPLE-FEATURE SCALING") scaler_s = scale[0] scaler_f = scale[1] rescaled_data = torch.tensor(scaler_s.inverse_transform(torch.t(torch.tensor(scaler_f.inverse_transform(tensor[:, :, 0].detach().numpy().squeeze()))))) return rescaled_data
2,869
41.205882
163
py
gca-rom
gca-rom-main/docs/conf.py
# Configuration file for the Sphinx documentation builder. # # This file only contains a selection of the most common options. For a full # list see the documentation: # https://www.sphinx-doc.org/en/master/usage/configuration.html # -- Path setup -------------------------------------------------------------- # If extensions (or modules to document with autodoc) are in another directory, # add these directories to sys.path here. If the directory is relative to the # documentation root, use os.path.abspath to make it absolute, like shown here. # import os import sys sys.path.insert(0, os.path.abspath('..')) # -- Project information ----------------------------------------------------- project = 'gca-rom' copyright = '2023, Federico Pichi, Beatriz Moya, Jan S. Hesthaven' author = 'Federico Pichi, Beatriz Moya, Jan S. Hesthaven' # -- General configuration --------------------------------------------------- # Add any Sphinx extension module names here, as strings. They can be # extensions coming with Sphinx (named 'sphinx.ext.*') or your custom # ones. extensions = [ "sphinx_rtd_theme", "sphinx.ext.autodoc" ] # Add any paths that contain templates here, relative to this directory. templates_path = ['_templates'] # List of patterns, relative to source directory, that match files and # directories to ignore when looking for source files. # This pattern also affects html_static_path and html_extra_path. exclude_patterns = ['_build', 'Thumbs.db', '.DS_Store'] # -- Options for HTML output ------------------------------------------------- # The theme to use for HTML and HTML Help pages. See the documentation for # a list of builtin themes. # html_theme = 'sphinx_rtd_theme' # Add any paths that contain custom static files (such as style sheets) here, # relative to this directory. They are copied after the builtin static files, # so a file named "default.css" will overwrite the builtin "default.css". html_static_path = ['_static'] def skip(app, what, name, obj, would_skip, options): if name == "__init__": return False return would_skip def setup(app): app.connect("autodoc-skip-member", skip)
2,162
33.333333
79
py
gca-rom
gca-rom-main/utils/test_all_scalar.py
import os # HYPER-PARAMETERS LIST problem_names_list = ["poisson", "advection", "graetz", "VX_navier_stokes", "VY_navier_stokes", "P_navier_stokes"] scalers_type_list = ["sample", "feature", "feature-sampling", "sampling-feature"] scalers_fun_list = ["minmax", "robust", "standard"] skip_connection_list = ["skip0", "skip1"] train_rates_list = [10, 20, 30, 40, 50] ffc_nodes_list = [50, 100, 200, 300, 400] latent_nodes_list = [25, 50, 75, 100, 125] btt_nodes_list = [10, 15, 20, 25, 30] lambda_map_list = [1e-2, 1e-1, 1e0, 1e1, 1e2] hidden_channels_list = [1, 2, 3, 4, 5] # HYPER-PARAMETERS TEST problem_names = ["poisson", "advection", "graetz", "VX_navier_stokes"] scalers_type = ["sampling-feature"] scalers_fun = ["standard"] skip_connection = ["skip1"] train_rates = [10, 30, 50] ffc_nodes = [100, 200, 300] latent_nodes = [50, 100] btt_nodes = [15, 25] lambda_map = [1e-1, 1e0, 1e1] hidden_channels = [1, 2, 3] # BEST_U_poisson_lmap10.0_btt15_seed10_lv4_hc3_nd50_ffn200_skip1_lr0.001_sc4_rate30 # BEST_U_advection_lmap10.0_btt15_seed10_lv4_hc2_nd100_ffn200_skip1_lr0.001_sc4_rate30 # BEST_U_graetz_lmap10.0_btt25_seed10_lv4_hc2_nd50_ffn200_skip1_lr0.001_sc4_rate30 # BEST_VX_navier_stokes_rid_lmap10.0_btt25_seed10_lv4_hc2_nd100_ffn300_skip1_lr0.001_sc4_rate30 # BEST_VY_navier_stokes_rid_lmap1.0_btt25_seed10_lv4_hc2_nd100_ffn300_skip1_lr0.001_sc4_rate30 # BEST_P_navier_stokes_rid_lmap1.0_btt25_seed10_lv4_hc2_nd100_ffn200_skip1_lr0.001_sc4_rate30 index_pb = [problem_names_list.index(pb)+1 for pb in problem_names] index_st = [scalers_type_list.index(st)+1 for st in scalers_type] index_sf = [scalers_fun_list.index(sf)+1 for sf in scalers_fun] index_sk = [skip_connection_list.index(sk) for sk in skip_connection] for (i, pb) in zip(index_pb, problem_names): for (j, st) in zip(index_st, scalers_type): for (k, sf) in zip(index_sf, scalers_fun): for (m, sk) in zip(index_sk, skip_connection): for rt in train_rates: for ffc in ffc_nodes: for ln in latent_nodes: for bn in btt_nodes: for la in lambda_map: for hc in hidden_channels: print("\n\nRUN MAIN FOR PROBLEM "+pb+" WITH SCALING: "+st+" WITH "+sf+" FUNCTION") print("TRAIN RATE = "+str(rt)+" and FFC NODES = "+str(ffc)+" and HIDDEN CHANNELS = "+str(hc)+" and SKIP = "+str(sk)) print("LATENT NODES = "+str(ln)+" and BTT NODES = "+str(bn)+" and LAMBDA = "+str(la)) os.system("python3 ../main.py %s %s %s %s %s %s %s %s %s %s" %(i, j, k, m, rt, ffc, ln, bn, la, hc))
2,774
52.365385
156
py
gca-rom
gca-rom-main/utils/h5_to_mat.py
""" .. _IMPORTING_dataset_form_FEniCS_solution: IMPORTING dataset form FEniCS solution - .h5 to .mat This script is used to import a FEniCS solution in .h5 format and convert it to a .mat file. Dependencies h5py numpy scipy Functions extract_edges(triangulation) Given a triangulation, this function returns a list of edges in the triangulation. """ import h5py import numpy as np from scipy.io import savemat def extract_edges(triangulation): edges = set() for triangle in triangulation: for i in range(3): edge = (triangle[i], triangle[(i + 1) % 3]) edges.add(tuple(sorted(edge))) return list(edges) # These lines define and open the path of the .h5 file h5_file = "../data/solution_hf_advection.h5" f = h5py.File(h5_file) # This line calculates the number of degrees of freedom dof = f["/Mesh/0/mesh/geometry"].shape[0] # This for loop iterates through all the mesh elements in the .h5 file for i in range(len(f["Mesh"])): # These lines extract the mesh information, x and y coordinates, and the solution vector mesh = f["Mesh/"+str(i)+"/mesh/geometry"] triang = np.array(f["/Mesh/"+str(i)+"/mesh/topology"]) + 1 x = mesh[:, 0:1] y = mesh[:, 1:2] u = np.array(f["VisualisationVector/"+str(i)]) try: xx = np.concatenate([xx, x], axis=1) yy = np.concatenate([yy, y], axis=1) solution = np.concatenate([solution, u], axis=1) except: xx = x yy = y solution = u # These lines extract the edges from the triangulation edges = np.array(extract_edges(triang)) edges = edges[edges[:, 1].argsort()] edges = edges[edges[:, 0].argsort(kind='mergesort')] # These lines create a dictionary that stores the triangulation, edges, number of degrees of freedom, solution, x and y coordinates mat_file = "../data/file.mat" dataset = dict() dataset['T'] = triang.astype('float') dataset['E'] = edges.astype('float') dataset['dof'] = float(dof) dataset['U'] = solution dataset['xx'] = xx dataset['yy'] = yy # This line saves the dataset as a .mat file savemat(mat_file, dataset)
2,107
27.876712
131
py
gca-rom
gca-rom-main/utils/save_all_scalar.py
import subprocess import numpy as np import pandas as pd import matplotlib.pyplot as plt import matplotlib.image as mpimg import warnings import seaborn as sns import os warnings.filterwarnings("ignore") # HYPER-PARAMETERS LIST model_names_list = ["poisson", "advection", "graetz", "navier_stokes"] problem_names_list = ["poisson", "advection", "graetz", "VX_navier_stokes", "VY_navier_stokes", "P_navier_stokes"] scalers_type_list = ["sample", "feature", "feature-sampling", "sampling-feature"] scalers_fun_list = ["minmax", "robust", "standard"] skip_connection_list = ["skip0", "skip1"] train_rates_list = ["rate10", "rate20", "rate30", "rate40", "rate50"] ffc_nodes_list = ["ffn50", "ffn100", "ffn200", "ffn300", "ffn400"] latent_nodes = ["nd25", "nd50", "nd75", "nd100", "nd125"] btt_nodes = ["btt10", "btt15", "btt20", "btt25", "btt30"] lambda_map_list = ["lmap0.01", "lmap0.1", "lmap1.0", "lmap10", "lmap100"] hidden_channels_list = ["hc1", "hc2", "hc3", "hc4", "hc5"] # HYPER-PARAMETERS SAVE model_names = ["poisson", "advection", "graetz", "navier_stokes", "navier_stokes", "navier_stokes"] problem_names = ["poisson", "advection", "graetz", "VX_navier_stokes", "VY_navier_stokes", "P_navier_stokes"] variable_names = ["U_", "U_", "U_", "VX_", "VY_", "P_"] scalers_fun = ["standard"] scalers_type = ["sc4"] skip_connection = ["skip1"] train_rates = ["rate10", "rate30", "rate50"] ffn_nodes = ["ffn100", "ffn200", "ffn300"] latent_nodes = ["nd50", "nd100"] btt_nodes = ["btt15", "btt25"] lambda_map = ["lmap0.1", "lmap1.0", "lmap10.0"] hidden_channels = ["hc1", "hc2", "hc3"] train_rates_l = [10, 30, 50] ffn_nodes_l = [100, 200, 300] latent_nodes_l = [50, 100] btt_nodes_l = [15, 25] lambda_map_l = [0.1, 1.0, 10.0] hidden_channels_l = [1, 2, 3] folder = "" for name in [*set(model_names)]: folder += str(name)+"/_standard/ " print("FOLDERS: ", folder) ## SCALAR CASES output = subprocess.check_output("find " + folder + "-name '*GCA-ROM.txt' -print > res_all.txt", shell=True) file = 'res_all.txt' sim_list = open(file).readlines() error = dict() for name in problem_names: for sct in scalers_type: for scf in scalers_fun: for sk in skip_connection: for rt in train_rates: for ffn in ffn_nodes: for ln in latent_nodes: for bn in btt_nodes: for la in lambda_map: for hc in hidden_channels: error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc] = list() data = pd.DataFrame() k = 0 for (j, r) in enumerate(sim_list): r = r.replace('//', '/').replace('\n', '') e = pd.read_csv(r, sep=" ", header=None).to_numpy().squeeze() for it_1, name in enumerate(problem_names): for it_2, sct in enumerate(scalers_type): for it_3, scf in enumerate(scalers_fun): for it_4, sk in enumerate(skip_connection): for it_5, rt in enumerate(train_rates): for it_6, ffn in enumerate(ffn_nodes): for it_7, ln in enumerate(latent_nodes): for it_8, bn in enumerate(btt_nodes): for it_9, la in enumerate(lambda_map): for it_10, hc in enumerate(hidden_channels): if all(x in r for x in [name, sct, scf, sk, rt, ffn, ln, bn, la, hc]): print(name, sct, scf, sk, rt, ffn, ln, bn, la, hc, e) if not error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc]: error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc].append(e) temp = pd.DataFrame({'problem_names': it_1, 'scalers_type': it_2, 'scalers_fun': it_3, 'skip_connection': it_4, 'train_rates': train_rates_l[it_5], 'ffn_nodes': ffn_nodes_l[it_6], 'latent_nodes': latent_nodes_l[it_7], 'btt_nodes': btt_nodes_l[it_8], 'lambda_map': lambda_map_l[it_9], 'hidden_channels': hidden_channels_l[it_10], 'error': e[1]} , index=[k]) data = pd.concat([data, temp]) k+=1 else: error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0] = np.concatenate([error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0], np.array(e)]) my_xticks = [] for sct in scalers_type: for scf in scalers_fun: for sk in skip_connection: for rt in train_rates: for ffn in ffn_nodes: for ln in latent_nodes: for bn in btt_nodes: for la in lambda_map: for hc in hidden_channels: my_xticks.append(sct+scf+sk+rt+ffn+ln+bn+la+hc) num_exp = len(scalers_fun)*len(scalers_type)*len(skip_connection)*len(train_rates)*len(ffn_nodes)*len(latent_nodes)*len(btt_nodes)*len(lambda_map)*len(hidden_channels) x = range(num_exp) er = np.array(list(error.items()))[:, 1] for y in er: if not y: y.append(np.ones(3)) error_array = np.vstack([y for y in er]) def plot_error(ax, name, x, var, i_min): i = 0 ax.set_title(name) for sct in scalers_type: for scf in scalers_fun: for sk in skip_connection: for rt in train_rates: for ffn in ffn_nodes: for ln in latent_nodes: for bn in btt_nodes: for la in lambda_map: for hc in hidden_channels: if np.array_equal(error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0], np.ones(3)): color_min_max = "white" col_mean = "white" else: if i == i_min: print("HYPERPARAMETERS: ", name, sct, scf, sk, rt, ffn, ln, bn, la, hc) string = str(model)+"/_standard/"+str(var)+str(model)+"_"+str(la)+"_"+str(bn)+"_seed"+str(10)+"_lv"+str(4)+"_"+str(hc)+"_"+str(ln)+"_"+str(ffn)+"_"+str(sk)+"_lr"+str(0.001)+"_sc"+str(4)+"_"+str(rt)+"/" print("FOLDER = ", string) color_min_max = "#1C3144" col_mean = "#4cb944" else: color_min_max = "#1C3144" col_mean = "#70161E" ax.scatter(i, error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0][0], marker="^", facecolors='white', edgecolors=color_min_max) ax.scatter(i, error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0][1], marker="d", facecolors=col_mean, edgecolors=col_mean) ax.scatter(i, error[name, sct, scf, sk, rt, ffn, ln, bn, la, hc][0][2], marker="v", facecolors='white', edgecolors=color_min_max) ax.set_yscale('log') i+=1 ax.tick_params(labelsize=8) # ax.set_xticks(x[:i], my_xticks[:i], rotation=65) ax.set_xlim([0, len(x[:i])]) # dim = np.reshape(range(len(problem_names)), (2, -1)).shape # fig2, axs2 = plt.subplot_mosaic([['upper left', 'upper mid', 'upper right'], # ['lower left', 'lower mid', 'lower right']]) for i, (name, model, var) in enumerate(zip(problem_names, model_names, variable_names)): model_path = '../plots/'+str(model)+'/tmp' if not os.path.exists(model_path): os.makedirs(model_path) fig1, axs1 = plt.subplots(1) i_min = np.argmin(error_array[i*num_exp:(i+1)*num_exp, 1], axis=0) print("\nBEST NETWORK FOR " + str(name) + " IS CORRESPONDING TO i = ", i_min) plot_error(axs1, name, x, var, i_min) # plot_error(axs2[k], name, x, var, i_min) fig1.tight_layout() fig1.savefig('../plots/'+str(model)+'/box_plot_error_all_'+str(name)+'.png', dpi=500) plt.close() # fig2.tight_layout() # fig2.savefig('../plots/box_plot_error_all.png', dpi=500) # plt.show() ### PAIRPLOT # print("Dataset shape: ", data.shape) # print("Dataset: ", data) columns_names = ["train_rates", "ffn_nodes", "latent_nodes", "btt_nodes", "lambda_map", "hidden_channels", "error"] list_hyper = [train_rates, ffn_nodes, latent_nodes, btt_nodes, lambda_map, hidden_channels] x_vars = ["train_rates", "ffn_nodes", "latent_nodes", "btt_nodes", "lambda_map", "hidden_channels"] x_vars_title = ["$r_t$", "ffn", "$n_l$", "$n$", "$\lambda$", "hc"] y_vars = ["error"] n_vars = len(x_vars) mark =["o", "s", "D"] line =["-", "--", "-."] def pairplot_box(data, xv, hue_var): sns.set(style="ticks") return sns.catplot(data=data, x=xv, y=y_vars[0], hue=hue_var, kind="box") def pairplot_point(data, xv, c, hue_var): sns.set(style="ticks") return sns.catplot(data=data, x=xv, y=y_vars[0], hue=hue_var, linestyles=line[:len(list_hyper[c])], markers=mark[:len(list_hyper[c])], kind="point") # return sns.pairplot(data[columns_names], x_vars=x_vars, y_vars=y_vars, hue=hue_var, markers=mark[:len(col)]) fontsize = 25 for n, (name, model) in enumerate(zip(problem_names, model_names)): data_name = data[data["problem_names"]==n] # print("Dataset name shape: ", data_name.shape) # print("Dataset name: ", data_name) j = 0 for c, col in enumerate(x_vars): for tit, xv in enumerate(x_vars): counter = 0 if xv == col: counter += 1 else: g1 = pairplot_box(data_name, xv, col) g2 = pairplot_point(data_name, xv, c, col) g1._legend.remove() g2._legend.remove() for ax in g1.axes.flat: if ax.get_ylabel() in y_vars: ax.set(yscale="log") ax.set_xlabel(x_vars_title[tit+counter], fontsize=fontsize) ax.set_ylabel(y_vars[0], fontsize=fontsize) ax.tick_params(axis='both', which='major', labelsize=fontsize) ax.tick_params(axis='both', which='minor', labelsize=fontsize) ax.tick_params(labelsize=fontsize) if j in [nn - 1 for nn in range(0, n_vars*(n_vars - 1) + 1, n_vars - 1)[1:]]: ax.legend(title=x_vars_title[c+counter], title_fontsize=fontsize) sns.move_legend(ax, "center left", bbox_to_anchor=(1., 0.5), ncol=1, frameon=False, fontsize=fontsize) g1.savefig('../plots/'+str(model)+'/tmp/g1_'+str(j)+'_'+str(name)+'.png', dpi=500) plt.close(g1.fig) for ax in g2.axes.flat: if ax.get_ylabel() in y_vars: ax.set(yscale="log") ax.set_xlabel(x_vars_title[tit+counter], fontsize=fontsize) ax.set_ylabel(y_vars[0], fontsize=fontsize) ax.tick_params(axis='both', which='major', labelsize=fontsize) ax.tick_params(axis='both', which='minor', labelsize=fontsize) ax.tick_params(labelsize=fontsize) if j in [nn - 1 for nn in range(0, n_vars*(n_vars - 1) + 1, n_vars - 1)[1:]]: ax.legend(title=x_vars_title[c+counter], title_fontsize=fontsize) sns.move_legend(ax, "center left", bbox_to_anchor=(1., 0.5), ncol=1, frameon=False, fontsize=fontsize) g2.savefig('../plots/'+str(model)+'/tmp/g2_'+str(j)+'_'+str(name)+'.png', dpi=500) plt.close(g2.fig) j += 1 f1, axarr1 = plt.subplots(len(x_vars), len(x_vars[:-1]), figsize=(20, 20)) f2, axarr2 = plt.subplots(len(x_vars), len(x_vars[:-1]), figsize=(20, 20)) i = 0 for c, col in enumerate(x_vars): for cc, col in enumerate(x_vars[:-1]): axarr1[c,cc].imshow(mpimg.imread('../plots/'+str(model)+'/tmp/g1_'+str(i)+'_'+str(name)+'.png')) axarr2[c,cc].imshow(mpimg.imread('../plots/'+str(model)+'/tmp/g2_'+str(i)+'_'+str(name)+'.png')) i += 1 [ax.set_axis_off() for ax in axarr1.ravel()] [ax.set_axis_off() for ax in axarr2.ravel()] # shutil.rmtree('../plots/'+str(model)+'/tmp/') f1.tight_layout() f2.tight_layout() f1.savefig('../plots/'+str(model)+'/sns_error_all_box_'+str(name)+'.png', dpi=500) f2.savefig('../plots/'+str(model)+'/sns_error_all_point_'+str(name)+'.png', dpi=500)
13,880
51.579545
249
py
CRST
CRST-master/evaluate.py
import argparse import scipy from scipy import ndimage import numpy as np import sys from packaging import version import time import util import torch import torchvision.models as models import torch.nn.functional as F from torch.utils import data, model_zoo from deeplab.model import Res_Deeplab from deeplab.datasets import GTA5TestDataSet from collections import OrderedDict import os from PIL import Image import matplotlib.pyplot as plt import torch.nn as nn # IMG_MEAN = np.array((104.00698793,116.66876762,122.67891434), dtype=np.float32) IMG_MEAN = np.array((0.406, 0.456, 0.485), dtype=np.float32) # BGR IMG_STD = np.array((0.225, 0.224, 0.229), dtype=np.float32) # BGR DATA_DIRECTORY = './dataset/cityscapes' DATA_LIST_PATH = './dataset/list/cityscapes/val.lst' SAVE_PATH = './cityscapes/eval' TEST_IMAGE_SIZE = '1024,2048' TEST_SCALE = 1.0 IGNORE_LABEL = 255 NUM_CLASSES = 19 NUM_STEPS = 500 # Number of images in the validation set. RESTORE_FROM = './src_model/gta5/src_model.pth' DATA_SRC = 'cityscapes' SET = 'val' LOG_FILE = 'log' MODEL = 'DeeplabRes' def get_arguments(): """Parse all the arguments provided from the CLI. Returns: A list of parsed arguments. """ parser = argparse.ArgumentParser(description="DeepLab-ResNet Network") parser.add_argument("--model", type=str, default=MODEL, help="Model Choice (DeeplabMulti/DeeplabVGG).") parser.add_argument("--data-src", type=str, default=DATA_SRC, help="Data name.") parser.add_argument("--data-dir", type=str, default=DATA_DIRECTORY, help="Path to the directory containing the Cityscapes dataset.") parser.add_argument("--data-list", type=str, default=DATA_LIST_PATH, help="Path to the file listing the images in the dataset.") parser.add_argument("--ignore-label", type=int, default=IGNORE_LABEL, help="The index of the label to ignore during the training.") parser.add_argument("--num-classes", type=int, default=NUM_CLASSES, help="Number of classes to predict (including background).") parser.add_argument("--restore-from", type=str, default=RESTORE_FROM, help="Where restore model parameters from.") parser.add_argument("--gpu", type=int, default=0, help="choose gpu device.") parser.add_argument('--test-flipping', dest='test_flipping', help='If average predictions of original and flipped images.', default=False, action='store_true') parser.add_argument("--set", type=str, default=SET, help="choose evaluation set.") parser.add_argument("--save", type=str, default=SAVE_PATH, help="Path to save result.") parser.add_argument("--log-file", type=str, default=LOG_FILE, help="The name of log file.") parser.add_argument('--debug',help='True means logging debug info.', default=False, action='store_true') parser.add_argument('--test-scale', type=str, default=TEST_SCALE, help='The test scales. Multi-scale supported') parser.add_argument('--test-image-size', default=TEST_IMAGE_SIZE, help='The test image size', type=str) return parser.parse_args() args = get_arguments() # palette if args.data_src == 'gta' or args.data_src == 'cityscapes': # gta: palette = [128, 64, 128, 244, 35, 232, 70, 70, 70, 102, 102, 156, 190, 153, 153, 153, 153, 153, 250, 170, 30, 220, 220, 0, 107, 142, 35, 152, 251, 152, 70, 130, 180, 220, 20, 60, 255, 0, 0, 0, 0, 142, 0, 0, 70, 0, 60, 100, 0, 80, 100, 0, 0, 230, 119, 11, 32] elif args.data_src == 'synthia': # synthia: palette = [128,64,128,244,35,232,70,70,70,102,102,156,64,64,128,153,153,153,250,170,30,220,220,0, 107,142,35,70,130,180,220,20,60,255,0,0,0,0,142,0,60,100,0,0,230,119,11,32] zero_pad = 256 * 3 - len(palette) for i in range(zero_pad): palette.append(0) def colorize_mask(mask): # mask: numpy array of the mask new_mask = Image.fromarray(mask.astype(np.uint8)).convert('P') new_mask.putpalette(palette) return new_mask def main(): """Create the model and start the evaluation process.""" device = torch.device("cuda:" + str(args.gpu)) if not os.path.exists(args.save): os.makedirs(args.save) logger = util.set_logger(args.save, args.log_file, args.debug) logger.info('start with arguments %s', args) x_num = 0 with open(args.data_list) as f: for _ in f.readlines(): x_num = x_num + 1 sys.path.insert(0, 'dataset/helpers') if args.data_src == 'gta' or args.data_src == 'cityscapes': from labels import id2label, trainId2label elif args.data_src == 'synthia': from labels_cityscapes_synthia import id2label, trainId2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) valid_labels = sorted(set(id_2_label.ravel())) scorer = ScoreUpdater(valid_labels, args.num_classes, x_num, logger) scorer.reset() if args.model == 'DeeplabRes': model = Res_Deeplab(num_classes=args.num_classes) # elif args.model == 'DeeplabVGG': # model = DeeplabVGG(num_classes=args.num_classes) # if args.restore_from == RESTORE_FROM: # args.restore_from = RESTORE_FROM_VGG if args.restore_from[:4] == 'http' : saved_state_dict = model_zoo.load_url(args.restore_from) new_params = model.state_dict().copy() for i in saved_state_dict: # Scale.layer5.conv2d_list.3.weight i_parts = str(i).split('.') # print i_parts if not i_parts[0] == 'fc': new_params['.'.join(i_parts[0:])] = saved_state_dict[i] else: loc = "cuda:" + str(args.gpu) saved_state_dict = torch.load(args.restore_from,map_location=loc) new_params = saved_state_dict.copy() model.load_state_dict(new_params) #model.train() model.eval() model.to(device) testloader = data.DataLoader(GTA5TestDataSet(args.data_dir, args.data_list, test_scale = 1.0, test_size=(1024, 512), mean=IMG_MEAN, std=IMG_STD, scale=False, mirror=False), batch_size=1, shuffle=False, pin_memory=True) test_scales = [float(_) for _ in str(args.test_scale).split(',')] h, w = map(int, args.test_image_size.split(',')) if version.parse(torch.__version__) >= version.parse('0.4.0'): interp = nn.Upsample(size=(h, w), mode='bilinear', align_corners=True) else: interp = nn.Upsample(size=(h, w), mode='bilinear') test_image_size = (h, w) mean_rgb = IMG_MEAN[::-1].copy() std_rgb = IMG_STD[::-1].copy() with torch.no_grad(): for index, batch in enumerate(testloader): image, label, _, name = batch img = image.clone() num_scales = len(test_scales) # output_dict = {k: [] for k in range(num_scales)} for scale_idx in range(num_scales): if version.parse(torch.__version__) > version.parse('0.4.0'): image = F.interpolate(image, scale_factor=test_scales[scale_idx], mode='bilinear', align_corners=True) else: test_size = ( int(h*test_scales[scale_idx]), int(w*test_scales[scale_idx]) ) interp_tmp = nn.Upsample(size=test_size, mode='bilinear', align_corners=True) image = interp_tmp(img) if args.model == 'DeeplabRes': output2 = model(image.to(device)) coutput = interp(output2).cpu().data[0].numpy() if args.test_flipping: output2 = model(torch.from_numpy(image.numpy()[:,:,:,::-1].copy()).to(device)) coutput = 0.5 * ( coutput + interp(output2).cpu().data[0].numpy()[:,:,::-1] ) if scale_idx == 0: output = coutput.copy() else: output += coutput output = output/num_scales output = output.transpose(1,2,0) output = np.asarray(np.argmax(output, axis=2), dtype=np.uint8) pred_label = output.copy() label = label_2_id[np.asarray(label.numpy(), dtype=np.uint8)] scorer.update(pred_label.flatten(), label.flatten(), index) output_col = colorize_mask(output) output = Image.fromarray(output) name = name[0].split('/')[-1] output.save('%s/%s' % (args.save, name)) output_col.save('%s/%s_color.png' % (args.save, name.split('.')[0])) class ScoreUpdater(object): # only IoU are computed. accu, cls_accu, etc are ignored. def __init__(self, valid_labels, c_num, x_num, logger=None, label=None, info=None): self._valid_labels = valid_labels self._confs = np.zeros((c_num, c_num)) self._per_cls_iou = np.zeros(c_num) self._logger = logger self._label = label self._info = info self._num_class = c_num self._num_sample = x_num @property def info(self): return self._info def reset(self): self._start = time.time() self._computed = np.zeros(self._num_sample) # one-dimension self._confs[:] = 0 def fast_hist(self,label, pred_label, n): k = (label >= 0) & (label < n) return np.bincount(n * label[k].astype(int) + pred_label[k], minlength=n ** 2).reshape(n, n) def per_class_iu(self,hist): return np.diag(hist) / (hist.sum(1) + hist.sum(0) - np.diag(hist)) def do_updates(self, conf, i, computed=True): if computed: self._computed[i] = 1 self._per_cls_iou = self.per_class_iu(conf) def update(self, pred_label, label, i, computed=True): conf = self.fast_hist(label, pred_label, self._num_class) self._confs += conf self.do_updates(self._confs, i, computed) self.scores(i) def scores(self, i=None, logger=None): x_num = self._num_sample ious = np.nan_to_num( self._per_cls_iou ) logger = self._logger if logger is None else logger if logger is not None: if i is not None: speed = 1. * self._computed.sum() / (time.time() - self._start) logger.info('Done {}/{} with speed: {:.2f}/s'.format(i + 1, x_num, speed)) name = '' if self._label is None else '{}, '.format(self._label) logger.info('{}mean iou: {:.2f}%'. \ format(name, np.mean(ious) * 100)) with util.np_print_options(formatter={'float': '{:5.2f}'.format}): logger.info('\n{}'.format(ious * 100)) return ious if __name__ == '__main__': main()
11,168
39.762774
176
py
CRST
CRST-master/util.py
""" utilities for convenience """ import contextlib import h5py import logging import os.path as osp import yaml from io import StringIO from PIL import Image import numpy as np cfg = {} def as_list(obj): """A utility function that treat the argument as a list. Parameters ---------- obj : object Returns ------- If `obj` is a list, return it. Otherwise, return `[obj]` as a single-element list. """ if isinstance(obj, list): return obj else: return [obj] def get_interp_method(imh_src, imw_src, imh_dst, imw_dst, default=Image.CUBIC): if not cfg.get('choose_interpolation_method', False): return default if imh_dst < imh_src and imw_dst < imw_src: return Image.ANTIALIAS elif imh_dst > imh_src and imw_dst > imw_src: return Image.CUBIC else: return Image.LINEAR def h5py_save(to_path, *data): with h5py.File(to_path, 'w') as f: for i, datum in enumerate(data): f.create_dataset('d{}'.format(i), data=datum) def h5py_load(from_path): data = [] if osp.isfile(from_path): with h5py.File(from_path) as f: for k in f.keys(): data.append(f[k][()]) return tuple(data) def load_image_with_cache(path, cache=None): if cache is not None: if not cache.has_key(path): with open(path, 'rb') as f: cache[path] = f.read() return Image.open(StringIO(cache[path])) return Image.open(path) @contextlib.contextmanager def np_print_options(*args, **kwargs): original = np.get_printoptions() np.set_printoptions(*args, **kwargs) yield np.set_printoptions(**original) def read_cfg(cfg_file, cfg_info=None): if cfg_file is not None: print('Read config file {}'.format(cfg_file)) with open(cfg_file) as f: cfg_info = yaml.load(f) return cfg_info def set_logger(output_dir=None, log_file=None, debug=False): head = '%(asctime)-15s Host %(message)s' logger_level = logging.INFO if not debug else logging.DEBUG if all((output_dir, log_file)) and len(log_file) > 0: logger = logging.getLogger() log_path = osp.join(output_dir, log_file) handler = logging.FileHandler(log_path) formatter = logging.Formatter(head) handler.setFormatter(formatter) logger.addHandler(handler) handler = logging.StreamHandler() handler.setFormatter(formatter) logger.addHandler(handler) logger.setLevel(logger_level) else: logging.basicConfig(level=logger_level, format=head) logger = logging.getLogger() return logger
2,693
26.489796
86
py
CRST
CRST-master/crst_seg.py
import argparse import sys from packaging import version import time import util import os import os.path as osp import timeit from collections import OrderedDict import scipy.io import torch import torchvision.models as models import torch.nn.functional as F from torch.utils import data, model_zoo import torch.backends.cudnn as cudnn import torch.nn as nn import torch.optim as optim from operator import itemgetter import scipy from scipy import ndimage import math from PIL import Image import numpy as np import shutil import random from deeplab.model import Res_Deeplab from deeplab.datasets import GTA5TestDataSet from deeplab.datasets import SrcSTDataSet, GTA5StMineDataSet, SoftSrcSTDataSet, SoftGTA5StMineDataSet ### shared ### IMG_MEAN = np.array((0.406, 0.456, 0.485), dtype=np.float32) # BGR IMG_STD = np.array((0.225, 0.224, 0.229), dtype=np.float32) # BGR # data ### source ## gta DATA_SRC_DIRECTORY = './dataset/gta5' DATA_SRC_LIST_PATH = './dataset/list/gta5/train.lst' DATA_SRC = 'gta' RESTORE_FROM = './src_model/gta5/src_model.pth' NUM_CLASSES = 19 INIT_SRC_PORT = 0.03 # GTA: 0.03 ### target DATA_TGT_DIRECTORY = './dataset/cityscapes' DATA_TGT_TRAIN_LIST_PATH = './dataset/list/cityscapes/train_ClsConfSet.lst' DATA_TGT_TEST_LIST_PATH = './dataset/list/cityscapes/val.lst' IGNORE_LABEL = 255 # train scales for src and tgt TRAIN_SCALE_SRC = '0.5,1.5' TRAIN_SCALE_TGT = '0.5,1.5' # model MODEL = 'DeeplabRes' # gpu GPU = 0 PIN_MEMORY = False # log files LOG_FILE = 'self_training_log' ### train ### BATCH_SIZE = 2 INPUT_SIZE = '512,1024'# 512,1024 for GTA; RANDSEED = 3 # params for optimizor LEARNING_RATE =5e-5 POWER = 0.0 MOMENTUM = 0.9 WEIGHT_DECAY = 0.0005 NUM_ROUNDS = 4 EPR = 2 SRC_SAMPLING_POLICY = 'r' KC_POLICY = 'cb' KC_VALUE = 'conf' INIT_TGT_PORT = 0.2 MAX_TGT_PORT = 0.5 TGT_PORT_STEP = 0.05 # varies but dataset MAX_SRC_PORT = 0.06 #0.06; SRC_PORT_STEP = 0.0025 #0.0025: MRKLD = 0.0 LRENT = 0.0 MRSRC = 0.0 MINE_PORT = 1e-3 RARE_CLS_NUM = 3 MINE_CHANCE = 0.8 ### val ### SAVE_PATH = 'debug' TEST_IMAGE_SIZE = '1024,2048' EVAL_SCALE = 0.9 TEST_SCALE = '0.9,1.0,1.2' DS_RATE = 4 def seed_torch(seed=0): random.seed(seed) os.environ['PYTHONHASHSEED'] = str(seed) np.random.seed(seed) torch.manual_seed(seed) torch.cuda.manual_seed(seed) torch.cuda.manual_seed_all(seed) # if you are using multi-GPU. #torch.backends.cudnn.benchmark = False torch.backends.cudnn.enabled = False #torch.backends.cudnn.deterministic = True def get_arguments(): """Parse all the arguments provided from the CLI. Returns: A list of parsed arguments. """ parser = argparse.ArgumentParser(description="DeepLab-ResNet Network") ### shared by train & val # data parser.add_argument("--data-src", type=str, default=DATA_SRC, help="Name of source dataset.") parser.add_argument("--data-src-dir", type=str, default=DATA_SRC_DIRECTORY, help="Path to the directory containing the source dataset.") parser.add_argument("--data-src-list", type=str, default=DATA_SRC_LIST_PATH, help="Path to the file listing the images&labels in the source dataset.") parser.add_argument("--data-tgt-dir", type=str, default=DATA_TGT_DIRECTORY, help="Path to the directory containing the target dataset.") parser.add_argument("--data-tgt-train-list", type=str, default=DATA_TGT_TRAIN_LIST_PATH, help="Path to the file listing the images*GT labels in the target train dataset.") parser.add_argument("--data-tgt-test-list", type=str, default=DATA_TGT_TEST_LIST_PATH, help="Path to the file listing the images*GT labels in the target test dataset.") parser.add_argument("--num-classes", type=int, default=NUM_CLASSES, help="Number of classes to predict (including background).") parser.add_argument("--ignore-label", type=int, default=IGNORE_LABEL, help="The index of the label to ignore during the training.") # model parser.add_argument("--model", type=str, default=MODEL, help="Model Choice (DeeplabMulti/DeeplabVGG).") parser.add_argument("--restore-from", type=str, default=RESTORE_FROM, help="Where restore model parameters from.") # gpu parser.add_argument("--gpu", type=int, default=GPU, help="choose gpu device.") parser.add_argument("--pin-memory", type=bool, default=PIN_MEMORY, help="Whether to pin memory in train & eval.") # log files parser.add_argument("--log-file", type=str, default=LOG_FILE, help="The name of log file.") parser.add_argument('--debug',help='True means logging debug info.', default=False, action='store_true') ### train ### parser.add_argument("--batch-size", type=int, default=BATCH_SIZE, help="Number of images sent to the network in one step.") parser.add_argument("--input-size", type=str, default=INPUT_SIZE, help="Comma-separated string with height and width of images.") parser.add_argument("--is-training", action="store_true", help="Whether to updates the running means and variances during the training.") parser.add_argument("--eval-training", action="store_true", help="Use the saved means and variances, or running means and variances during the evaluation.") parser.add_argument("--random-mirror", action="store_true", help="Whether to randomly mirror the inputs during the training.") parser.add_argument("--random-scale", action="store_true", help="Whether to randomly scale the inputs during the training.") parser.add_argument("--train-scale-src", type=str, default=TRAIN_SCALE_SRC, help="The scale for multi-scale training in source domain.") parser.add_argument("--train-scale-tgt", type=str, default=TRAIN_SCALE_TGT, help="The scale for multi-scale training in target domain.") # params for optimizor parser.add_argument("--learning-rate", type=float, default=LEARNING_RATE, help="Base learning rate for training with polynomial decay.") parser.add_argument("--power", type=float, default=POWER, help="Decay parameter to compute the learning rate.") parser.add_argument("--momentum", type=float, default=MOMENTUM, help="Momentum component of the optimiser.") parser.add_argument("--weight-decay", type=float, default=WEIGHT_DECAY, help="Regularisation parameter for L2-loss.") ### val parser.add_argument('--test-flipping', dest='test_flipping', help='If average predictions of original and flipped images.', default=False, action='store_true') parser.add_argument("--test-image-size", type=str, default=TEST_IMAGE_SIZE, help="The test image size.") parser.add_argument("--eval-scale", type=float, default=EVAL_SCALE, help="The test image scale.") parser.add_argument("--test-scale", type=str, default=TEST_SCALE, help="The test image scale.") ### self-training params parser.add_argument("--save", type=str, default=SAVE_PATH, help="Path to save result for self-training.") parser.add_argument("--num-rounds", type=int, default=NUM_ROUNDS, help="Number of rounds for self-training.") parser.add_argument("--epr", type=int, default=EPR, help="Number of epochs per round for self-training.") parser.add_argument('--kc-policy', default=KC_POLICY, type=str, dest='kc_policy', help='The policy to determine kc. "cb" for weighted class-balanced threshold') parser.add_argument('--kc-value', default=KC_VALUE, type=str, help='The way to determine kc values, either "conf", or "prob".') parser.add_argument('--ds-rate', default=DS_RATE, type=int, help='The downsampling rate in kc calculation.') parser.add_argument('--init-tgt-port', default=INIT_TGT_PORT, type=float, dest='init_tgt_port', help='The initial portion of target to determine kc') parser.add_argument('--max-tgt-port', default=MAX_TGT_PORT, type=float, dest='max_tgt_port', help='The max portion of target to determine kc') parser.add_argument('--tgt-port-step', default=TGT_PORT_STEP, type=float, dest='tgt_port_step', help='The portion step in target domain in every round of self-paced self-trained neural network') parser.add_argument('--init-src-port', default=INIT_SRC_PORT, type=float, dest='init_src_port', help='The initial portion of source portion for self-trained neural network') parser.add_argument('--max-src-port', default=MAX_SRC_PORT, type=float, dest='max_src_port', help='The max portion of source portion for self-trained neural network') parser.add_argument('--src-port-step', default=SRC_PORT_STEP, type=float, dest='src_port_step', help='The portion step in source domain in every round of self-paced self-trained neural network') parser.add_argument('--randseed', default=RANDSEED, type=int, help='The random seed to sample the source dataset.') parser.add_argument("--src-sampling-policy", type=str, default=SRC_SAMPLING_POLICY, help="The sampling policy on source dataset: 'c' for 'cumulative' and 'r' for replace ") parser.add_argument('--mine-port', default=MINE_PORT, type=float, help='If a class has a predication portion lower than the mine_port, then mine the patches including the class in self-training.') parser.add_argument('--rare-cls-num', default=RARE_CLS_NUM, type=int, help='The number of classes to be mined.') parser.add_argument('--mine-chance', default=MINE_CHANCE, type=float, help='The chance of patch mining.') parser.add_argument('--rm-prob', help='If remove the probability maps generated in every round.', default=False, action='store_true') parser.add_argument('--mr-weight-kld', default=MRKLD, type=float, dest='mr_weight_kld', help='weight of kld model regularization') parser.add_argument('--lr-weight-ent', default=LRENT, type=float, dest='lr_weight_ent', help='weight of negative entropy label regularization') parser.add_argument('--mr-weight-src', default=MRSRC, type=float, dest='mr_weight_src', help='weight of regularization in source domain') return parser.parse_args() args = get_arguments() # palette if args.data_src == 'gta': # gta: palette = [128, 64, 128, 244, 35, 232, 70, 70, 70, 102, 102, 156, 190, 153, 153, 153, 153, 153, 250, 170, 30, 220, 220, 0, 107, 142, 35, 152, 251, 152, 70, 130, 180, 220, 20, 60, 255, 0, 0, 0, 0, 142, 0, 0, 70, 0, 60, 100, 0, 80, 100, 0, 0, 230, 119, 11, 32] zero_pad = 256 * 3 - len(palette) for i in range(zero_pad): palette.append(0) def colorize_mask(mask): # mask: numpy array of the mask new_mask = Image.fromarray(mask.astype(np.uint8)).convert('P') new_mask.putpalette(palette) return new_mask def main(): randseed = args.randseed seed_torch(randseed) device = torch.device("cuda:" + str(args.gpu)) save_path = args.save save_pseudo_label_path = osp.join(save_path, 'pseudo_label') # in 'save_path'. Save labelIDs, not trainIDs. save_stats_path = osp.join(save_path, 'stats') # in 'save_path' save_lst_path = osp.join(save_path, 'list') if not os.path.exists(save_path): os.makedirs(save_path) if not os.path.exists(save_pseudo_label_path): os.makedirs(save_pseudo_label_path) if not os.path.exists(save_stats_path): os.makedirs(save_stats_path) if not os.path.exists(save_lst_path): os.makedirs(save_lst_path) logger = util.set_logger(args.save, args.log_file, args.debug) logger.info('start with arguments %s', args) if args.model == 'DeeplabRes': model = Res_Deeplab(num_classes=args.num_classes) if args.restore_from[:4] == 'http' : saved_state_dict = model_zoo.load_url(args.restore_from) new_params = model.state_dict().copy() for i in saved_state_dict: # Scale.layer5.conv2d_list.3.weight i_parts = str(i).split('.') # print i_parts if not i_parts[0] == 'fc': new_params['.'.join(i_parts[0:])] = saved_state_dict[i] else: loc = "cuda:" + str(args.gpu) saved_state_dict = torch.load(args.restore_from,map_location=loc) new_params = saved_state_dict.copy() model.load_state_dict(new_params) image_src_list, _, label_src_list, src_num = parse_split_list(args.data_src_list) image_tgt_list, image_name_tgt_list, _, tgt_num = parse_split_list(args.data_tgt_train_list) _, _, _, test_num = parse_split_list(args.data_tgt_test_list) ## label mapping sys.path.insert(0, 'dataset/helpers') if args.data_src == 'synthia': from labels_cityscapes_synthia import id2label, trainId2label elif args.data_src == 'gta': from labels import id2label, trainId2label label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) valid_labels = sorted(set(id_2_label.ravel())) # portions tgt_portion = args.init_tgt_port src_portion = args.init_src_port # training crop size h, w = map(int, args.input_size.split(',')) input_size = (h, w) lscale_src, hscale_src = map(float, args.train_scale_src.split(',')) train_scale_src = (lscale_src, hscale_src) lscale_tgt, hscale_tgt = map(float, args.train_scale_tgt.split(',')) train_scale_tgt = (lscale_tgt, hscale_tgt) for round_idx in range(args.num_rounds): save_round_eval_path = osp.join(args.save,str(round_idx)) save_pseudo_label_color_path = osp.join(save_round_eval_path, 'pseudo_label_color') # in every 'save_round_eval_path' if not os.path.exists(save_round_eval_path): os.makedirs(save_round_eval_path) if not os.path.exists(save_pseudo_label_color_path): os.makedirs(save_pseudo_label_color_path) ########## pseudo-label generation if round_idx != args.num_rounds - 1: # evaluation & save confidence vectors conf_dict, pred_cls_num, save_prob_path, save_pred_path = val(model, device, save_round_eval_path, round_idx, tgt_num, label_2_id, valid_labels, args, logger) # class-balanced thresholds cls_thresh = kc_parameters(conf_dict, pred_cls_num, tgt_portion, round_idx, save_stats_path, args, logger) tgt_portion = min(tgt_portion + args.tgt_port_step, args.max_tgt_port) # pseudo-label maps generation label_selection(cls_thresh, tgt_num, image_name_tgt_list, id_2_label, round_idx, save_prob_path, save_pred_path, save_pseudo_label_path, save_pseudo_label_color_path, save_round_eval_path, args, logger) # save training list if args.src_sampling_policy == 'c': randseed = args.randseed elif args.src_sampling_policy == 'r': randseed += 1 src_train_lst, tgt_train_lst, src_num_sel = savelst_SrcTgt(src_portion, image_tgt_list, image_name_tgt_list, image_src_list, label_src_list, save_lst_path, save_pseudo_label_path, src_num, tgt_num, randseed, args) src_portion = min(src_portion + args.src_port_step, args.max_src_port) ########### model retraining # dataset epoch_per_round = args.epr # reg weights if args.mr_weight_kld == 0: reg_weight_tgt = 0.0 else: # currently only one kind of model regularizer is supported reg_weight_tgt = args.mr_weight_kld reg_weight_src = args.mr_weight_src ### patch mining params # no patch mining in src # patch mining in target rare_id = np.load(save_stats_path + '/rare_id_round' + str(round_idx) + '.npy') mine_id = np.load(save_stats_path + '/mine_id_round' + str(round_idx) + '.npy') mine_chance = args.mine_chance # dataloader if args.lr_weight_ent == 0.0: srctrainset = SrcSTDataSet(args.data_src_dir, src_train_lst, max_iters=src_num_sel,reg_weight=reg_weight_src,data_src=args.data_src, crop_size=input_size,scale=args.random_scale, mirror=args.random_mirror, train_scale=train_scale_src, mean=IMG_MEAN, std=IMG_STD) tgttrainset = GTA5StMineDataSet(args.data_tgt_dir, tgt_train_lst, pseudo_root=save_pseudo_label_path, max_iters=tgt_num,reg_weight=reg_weight_tgt,rare_id = rare_id, mine_id=mine_id, mine_chance = mine_chance, crop_size=input_size,scale=args.random_scale,data_src=args.data_src, mirror=args.random_mirror, train_scale=train_scale_tgt, mean=IMG_MEAN, std=IMG_STD) elif args.lr_weight_ent > 0.0: srctrainset = SoftSrcSTDataSet(args.data_src_dir, src_train_lst, max_iters=src_num_sel,reg_weight=reg_weight_src,data_src=args.data_src,num_classes=args.num_classes, crop_size=input_size,scale=args.random_scale, mirror=args.random_mirror, train_scale=train_scale_src, mean=IMG_MEAN, std=IMG_STD) tgttrainset = SoftGTA5StMineDataSet(args.data_tgt_dir, tgt_train_lst, pseudo_root=save_pseudo_label_path, max_iters=tgt_num,reg_weight=reg_weight_tgt,rare_id = rare_id, mine_id=mine_id, mine_chance = mine_chance, crop_size=input_size,scale=args.random_scale,data_src=args.data_src,num_classes=args.num_classes, mirror=args.random_mirror, train_scale=train_scale_tgt, mean=IMG_MEAN, std=IMG_STD) mixtrainset = torch.utils.data.ConcatDataset([srctrainset, tgttrainset]) mix_trainloader = torch.utils.data.DataLoader(mixtrainset, batch_size=args.batch_size, shuffle=True, num_workers=0, pin_memory=args.pin_memory) # optimizer tot_iter = np.ceil(float(src_num_sel + tgt_num) / args.batch_size) optimizer = optim.SGD([{'params': get_1x_lr_params_NOscale(model), 'lr': args.learning_rate}, {'params': get_10x_lr_params(model), 'lr': 10 * args.learning_rate}], lr=args.learning_rate, momentum=args.momentum, weight_decay=args.weight_decay) interp = nn.Upsample(size=input_size, mode='bilinear', align_corners=True) logger.info('###### Start model retraining dataset in round {}! ######'.format(round_idx)) # model if args.is_training: model.train() else: model.eval() start = timeit.default_timer() # cudnn cudnn.enabled = True # enable cudnn cudnn.benchmark = True # enable the inbuilt cudnn auto-tuner to find the best algorithm to use for your hardware. # start training for epoch in range(epoch_per_round): train(mix_trainloader, model, device, interp, optimizer, tot_iter, round_idx, epoch, args, logger) end = timeit.default_timer() logger.info('###### Finish model retraining dataset in round {}! Time cost: {:.2f} seconds. ######'.format(round_idx, end - start)) # test self-trained model in target domain test set tgt_set = 'test' test(model, device, save_round_eval_path, round_idx, tgt_set, test_num, args.data_tgt_test_list, label_2_id, valid_labels, args, logger) elif round_idx == args.num_rounds - 1: shutil.rmtree(save_pseudo_label_path) tgt_set = 'train' test(model, device, save_round_eval_path, round_idx, tgt_set, tgt_num, args.data_tgt_train_list, label_2_id, valid_labels, args, logger) tgt_set = 'test' test(model, device, save_round_eval_path, round_idx, tgt_set, test_num, args.data_tgt_test_list, label_2_id, valid_labels, args, logger) def val(model, device, save_round_eval_path, round_idx, tgt_num, label_2_id, valid_labels, args, logger): """Create the model and start the evaluation process.""" ## scorer scorer = ScoreUpdater(valid_labels, args.num_classes, tgt_num, logger) scorer.reset() h, w = map(int, args.test_image_size.split(',')) test_image_size = (h, w) test_size = ( int(h*args.eval_scale), int(w*args.eval_scale) ) ## test data loader testloader = data.DataLoader(GTA5TestDataSet(args.data_tgt_dir, args.data_tgt_train_list, test_size=test_size, test_scale=args.eval_scale, mean=IMG_MEAN, std=IMG_STD, scale=False, mirror=False), batch_size=1, shuffle=False, pin_memory=args.pin_memory) ## model for evaluation if args.eval_training: model.train() else: model.eval() # model.to(device) ## upsampling layer if version.parse(torch.__version__) >= version.parse('0.4.0'): interp = nn.Upsample(size=test_image_size, mode='bilinear', align_corners=True) else: interp = nn.Upsample(size=test_image_size, mode='bilinear') ## output of deeplab is logits, not probability softmax2d = nn.Softmax2d() ## output folder save_pred_vis_path = osp.join(save_round_eval_path, 'pred_vis') save_prob_path = osp.join(save_round_eval_path, 'prob') save_pred_path = osp.join(save_round_eval_path, 'pred') if not os.path.exists(save_pred_vis_path): os.makedirs(save_pred_vis_path) if not os.path.exists(save_prob_path): os.makedirs(save_prob_path) if not os.path.exists(save_pred_path): os.makedirs(save_pred_path) # saving output data conf_dict = {k: [] for k in range(args.num_classes)} pred_cls_num = np.zeros(args.num_classes) ## evaluation process logger.info('###### Start evaluating target domain train set in round {}! ######'.format(round_idx)) start_eval = time.time() with torch.no_grad(): for index, batch in enumerate(testloader): image, label, _, name = batch if args.model == 'DeeplabRes': output2 = model(image.to(device)) output = softmax2d(interp(output2)).cpu().data[0].numpy() if args.test_flipping: output2 = model(torch.from_numpy(image.numpy()[:,:,:,::-1].copy()).to(device)) output = 0.5 * ( output + softmax2d(interp(output2)).cpu().data[0].numpy()[:,:,::-1] ) output = output.transpose(1,2,0) amax_output = np.asarray(np.argmax(output, axis=2), dtype=np.uint8) conf = np.amax(output,axis=2) # score pred_label = amax_output.copy() label = label_2_id[np.asarray(label.numpy(), dtype=np.uint8)] scorer.update(pred_label.flatten(), label.flatten(), index) # save visualized seg maps & predication prob map amax_output_col = colorize_mask(amax_output) name = name[0].split('/')[-1] image_name = name.split('.')[0] # prob np.save('%s/%s.npy' % (save_prob_path, image_name), output) # trainIDs/vis seg maps amax_output = Image.fromarray(amax_output) amax_output.save('%s/%s.png' % (save_pred_path, image_name)) amax_output_col.save('%s/%s_color.png' % (save_pred_vis_path, image_name)) # save class-wise confidence maps if args.kc_value == 'conf': for idx_cls in range(args.num_classes): idx_temp = pred_label == idx_cls pred_cls_num[idx_cls] = pred_cls_num[idx_cls] + np.sum(idx_temp) if idx_temp.any(): conf_cls_temp = conf[idx_temp].astype(np.float32) len_cls_temp = conf_cls_temp.size # downsampling by ds_rate conf_cls = conf_cls_temp[0:len_cls_temp:args.ds_rate] conf_dict[idx_cls].extend(conf_cls) elif args.kc_value == 'prob': for idx_cls in range(args.num_classes): idx_temp = pred_label == idx_cls pred_cls_num[idx_cls] = pred_cls_num[idx_cls] + np.sum(idx_temp) # prob slice prob_cls_temp = output[:,:,idx_cls].astype(np.float32).ravel() len_cls_temp = prob_cls_temp.size # downsampling by ds_rate prob_cls = prob_cls_temp[0:len_cls_temp:args.ds_rate] conf_dict[idx_cls].extend(prob_cls) # it should be prob_dict; but for unification, use conf_dict logger.info('###### Finish evaluating target domain train set in round {}! Time cost: {:.2f} seconds. ######'.format(round_idx, time.time()-start_eval)) return conf_dict, pred_cls_num, save_prob_path, save_pred_path # return the dictionary containing all the class-wise confidence vectors def train(mix_trainloader, model, device, interp, optimizer, tot_iter, round_idx, epoch_idx, args, logger): """Create the model and start the training.""" for i_iter, batch in enumerate(mix_trainloader): images, labels, _, _, reg_weights = batch images = images.to(device) labels = labels.to(device) optimizer.zero_grad() adjust_learning_rate(optimizer, i_iter, tot_iter) pred = interp(model(images)) if args.lr_weight_ent == 0.0: loss = reg_loss_calc(pred, labels, reg_weights.to(device), args) if args.lr_weight_ent > 0.0: loss = reg_loss_calc_expand(pred, labels, reg_weights.to(device), args) loss.backward() optimizer.step() logger.info('iter = {} of {} completed, loss = {:.4f}'.format(i_iter+1, tot_iter, loss.data.cpu().numpy())) print('taking snapshot ...') torch.save(model.state_dict(), osp.join(args.save, args.data_src + '2city_round' + str(round_idx) + '_epoch' + str(epoch_idx+1) + '.pth')) def test(model, device, save_round_eval_path, round_idx, tgt_set, test_num, test_list, label_2_id, valid_labels, args, logger): """Create the model and start the evaluation process.""" ## scorer scorer = ScoreUpdater(valid_labels, args.num_classes, test_num, logger) scorer.reset() h, w = map(int, args.test_image_size.split(',')) test_image_size = (h, w) test_size = ( h, w ) test_scales = [float(_) for _ in str(args.test_scale).split(',')] num_scales = len(test_scales) ## test data loader testloader = data.DataLoader(GTA5TestDataSet(args.data_tgt_dir, test_list, test_size=test_size, test_scale=1.0, mean=IMG_MEAN, std=IMG_STD, scale=False, mirror=False), batch_size=1, shuffle=False, pin_memory=args.pin_memory) ## model for evaluation if args.eval_training: model.train() else: model.eval() # model.to(device) ## upsampling layer if version.parse(torch.__version__) >= version.parse('0.4.0'): interp = nn.Upsample(size=test_image_size, mode='bilinear', align_corners=True) else: interp = nn.Upsample(size=test_image_size, mode='bilinear') ## output of deeplab is logits, not probability softmax2d = nn.Softmax2d() ## output folder if tgt_set == 'train': save_test_vis_path = osp.join(save_round_eval_path, 'trainSet_vis') elif tgt_set == 'test': save_test_vis_path = osp.join(save_round_eval_path, 'testSet_vis') if not os.path.exists(save_test_vis_path): os.makedirs(save_test_vis_path) ## evaluation process logger.info('###### Start evaluating in target domain {} set in round {}! ######'.format(tgt_set, round_idx)) start_eval = time.time() with torch.no_grad(): for index, batch in enumerate(testloader): image, label, _, name = batch img = image.clone() for scale_idx in range(num_scales): if version.parse(torch.__version__) > version.parse('0.4.0'): image = F.interpolate(img, scale_factor=test_scales[scale_idx], mode='bilinear', align_corners=True) else: test_size = (int(h * test_scales[scale_idx]), int(w * test_scales[scale_idx])) interp_tmp = nn.Upsample(size=test_size, mode='bilinear', align_corners=True) image = interp_tmp(img) if args.model == 'DeeplabRes': output2 = model(image.to(device)) coutput = interp(output2).cpu().data[0].numpy() if args.test_flipping: output2 = model(torch.from_numpy(image.numpy()[:,:,:,::-1].copy()).to(device)) coutput = 0.5 * ( coutput + interp(output2).cpu().data[0].numpy()[:,:,::-1] ) if scale_idx == 0: output = coutput.copy() else: output = output+coutput output = output/num_scales output = output.transpose(1,2,0) amax_output = np.asarray(np.argmax(output, axis=2), dtype=np.uint8) # score pred_label = amax_output.copy() label = label_2_id[np.asarray(label.numpy(), dtype=np.uint8)] scorer.update(pred_label.flatten(), label.flatten(), index) # save visualized seg maps & predication prob map amax_output_col = colorize_mask(amax_output) name = name[0].split('/')[-1] image_name = name.split('.')[0] # vis seg maps amax_output_col.save('%s/%s_color.png' % (save_test_vis_path, image_name)) logger.info('###### Finish evaluating in target domain {} set in round {}! Time cost: {:.2f} seconds. ######'.format(tgt_set, round_idx, time.time()-start_eval)) return def kc_parameters(conf_dict, pred_cls_num, tgt_portion, round_idx, save_stats_path, args, logger): logger.info('###### Start kc generation in round {} ! ######'.format(round_idx)) start_kc = time.time() # threshold for each class conf_tot = 0.0 cls_thresh = np.ones(args.num_classes,dtype = np.float32) cls_sel_size = np.zeros(args.num_classes, dtype=np.float32) cls_size = np.zeros(args.num_classes, dtype=np.float32) if args.kc_policy == 'cb' and args.kc_value == 'conf': for idx_cls in np.arange(0, args.num_classes): cls_size[idx_cls] = pred_cls_num[idx_cls] if conf_dict[idx_cls] != None: conf_dict[idx_cls].sort(reverse=True) # sort in descending order len_cls = len(conf_dict[idx_cls]) cls_sel_size[idx_cls] = int(math.floor(len_cls * tgt_portion)) len_cls_thresh = int(cls_sel_size[idx_cls]) if len_cls_thresh != 0: cls_thresh[idx_cls] = conf_dict[idx_cls][len_cls_thresh-1] conf_dict[idx_cls] = None # threshold for mine_id with priority num_mine_id = len(np.nonzero(cls_size / np.sum(cls_size) < args.mine_port)[0]) # chose the smallest mine_id id_all = np.argsort(cls_size / np.sum(cls_size)) rare_id = id_all[:args.rare_cls_num] mine_id = id_all[:num_mine_id] # sort mine_id in ascending order w.r.t predication portions # save mine ids np.save(save_stats_path + '/rare_id_round' + str(round_idx) + '.npy', rare_id) np.save(save_stats_path + '/mine_id_round' + str(round_idx) + '.npy', mine_id) logger.info('Mining ids : {}! {} rarest ids: {}!'.format(mine_id,args.rare_cls_num,rare_id)) # save thresholds np.save(save_stats_path + '/cls_thresh_round' + str(round_idx) + '.npy', cls_thresh) np.save(save_stats_path + '/cls_sel_size_round' + str(round_idx) + '.npy', cls_sel_size) logger.info('###### Finish kc generation in round {}! Time cost: {:.2f} seconds. ######'.format(round_idx,time.time() - start_kc)) return cls_thresh def label_selection(cls_thresh, tgt_num, image_name_tgt_list, id_2_label, round_idx, save_prob_path, save_pred_path, save_pseudo_label_path, save_pseudo_label_color_path, save_round_eval_path, args, logger): logger.info('###### Start pseudo-label generation in round {} ! ######'.format(round_idx)) start_pl = time.time() for idx in range(tgt_num): sample_name = image_name_tgt_list[idx].split('.')[0] probmap_path = osp.join(save_prob_path, '{}.npy'.format(sample_name)) pred_path = osp.join(save_pred_path, '{}.png'.format(sample_name)) pred_prob = np.load(probmap_path) pred_label_trainIDs = np.asarray(Image.open(pred_path)) pred_label_labelIDs = id_2_label[pred_label_trainIDs] pred_label_trainIDs = pred_label_trainIDs.copy() if args.kc_policy == 'cb' and args.lr_weight_ent == 0.0: save_wpred_vis_path = osp.join(save_round_eval_path, 'weighted_pred_vis') if not os.path.exists(save_wpred_vis_path): os.makedirs(save_wpred_vis_path) weighted_prob = pred_prob/cls_thresh weighted_pred_trainIDs = np.asarray(np.argmax(weighted_prob, axis=2), dtype=np.uint8) # save weighted predication wpred_label_col = weighted_pred_trainIDs.copy() wpred_label_col = colorize_mask(wpred_label_col) wpred_label_col.save('%s/%s_color.png' % (save_wpred_vis_path, sample_name)) weighted_conf = np.amax(weighted_prob, axis=2) pred_label_trainIDs = weighted_pred_trainIDs.copy() pred_label_labelIDs = id_2_label[pred_label_trainIDs] pred_label_labelIDs[weighted_conf < 1] = 0 # '0' in cityscapes indicates 'unlabaled' for labelIDs pred_label_trainIDs[weighted_conf < 1] = 255 # '255' in cityscapes indicates 'unlabaled' for trainIDs elif args.kc_policy == 'cb' and args.lr_weight_ent > 0.0: # check if cb can be combined with kc_value == conf or prob; also check if \alpha can be larger than 1 save_wpred_vis_path = osp.join(save_round_eval_path, 'weighted_pred_vis') if not os.path.exists(save_wpred_vis_path): os.makedirs(save_wpred_vis_path) # soft pseudo-label soft_pseudo_label = np.power(pred_prob/cls_thresh,1.0/args.lr_weight_ent) # weighted softmax with temperature soft_pseudo_label_sum = soft_pseudo_label.sum(2) soft_pseudo_label = soft_pseudo_label.transpose(2,0,1)/soft_pseudo_label_sum soft_pseudo_label = soft_pseudo_label.transpose(1,2,0).astype(np.float32) np.save('%s/%s.npy' % (save_pseudo_label_path, sample_name), soft_pseudo_label) # hard pseudo-label weighted_pred_trainIDs = np.asarray(np.argmax(soft_pseudo_label, axis=2), dtype=np.uint8) reg_score = np.sum( -soft_pseudo_label*np.log(pred_prob+1e-32) + args.lr_weight_ent*soft_pseudo_label*np.log(soft_pseudo_label+1e-32), axis=2) sel_score = np.sum( -soft_pseudo_label*np.log(cls_thresh+1e-32), axis=2) # save weighted predication wpred_label_col = weighted_pred_trainIDs.copy() wpred_label_col = colorize_mask(wpred_label_col) wpred_label_col.save('%s/%s_color.png' % (save_wpred_vis_path, sample_name)) pred_label_trainIDs = weighted_pred_trainIDs.copy() pred_label_labelIDs = id_2_label[pred_label_trainIDs] pred_label_labelIDs[reg_score >= sel_score] = 0 # '0' in cityscapes indicates 'unlabaled' for labelIDs pred_label_trainIDs[reg_score >= sel_score] = 255 # '255' in cityscapes indicates 'unlabaled' for trainIDs # pseudo-labels with labelID pseudo_label_labelIDs = pred_label_labelIDs.copy() pseudo_label_trainIDs = pred_label_trainIDs.copy() # save colored pseudo-label map pseudo_label_col = colorize_mask(pseudo_label_trainIDs) pseudo_label_col.save('%s/%s_color.png' % (save_pseudo_label_color_path, sample_name)) # save pseudo-label map with label IDs pseudo_label_save = Image.fromarray(pseudo_label_labelIDs.astype(np.uint8)) pseudo_label_save.save('%s/%s.png' % (save_pseudo_label_path, sample_name)) # remove probability maps if args.rm_prob: shutil.rmtree(save_prob_path) logger.info('###### Finish pseudo-label generation in round {}! Time cost: {:.2f} seconds. ######'.format(round_idx,time.time() - start_pl)) def parse_split_list(list_name): image_list = [] image_name_list = [] label_list = [] file_num = 0 with open(list_name) as f: for item in f.readlines(): fields = item.strip().split('\t') image_name = fields[0].split('/')[-1] image_list.append(fields[0]) image_name_list.append(image_name) label_list.append(fields[1]) file_num += 1 return image_list, image_name_list, label_list, file_num def savelst_SrcTgt(src_portion, image_tgt_list, image_name_tgt_list, image_src_list, label_src_list, save_lst_path, save_pseudo_label_path, src_num, tgt_num, randseed, args): src_num_sel = int(np.floor(src_num*src_portion)) np.random.seed(randseed) sel_idx = list( np.random.choice(src_num, src_num_sel, replace=False) ) sel_src_img_list = list( itemgetter(*sel_idx)(image_src_list) ) sel_src_label_list = list(itemgetter(*sel_idx)(label_src_list)) src_train_lst = osp.join(save_lst_path,'src_train.lst') tgt_train_lst = osp.join(save_lst_path, 'tgt_train.lst') # generate src train list with open(src_train_lst, 'w') as f: for idx in range(src_num_sel): f.write("%s\t%s\n" % (sel_src_img_list[idx], sel_src_label_list[idx])) # generate tgt train list if args.lr_weight_ent > 0: with open(tgt_train_lst, 'w') as f: for idx in range(tgt_num): softlabel_name = image_name_tgt_list[idx].split('.')[0] + '.npy' soft_label_tgt_path = osp.join(save_pseudo_label_path, softlabel_name) image_tgt_path = osp.join(save_pseudo_label_path,image_name_tgt_list[idx]) f.write("%s\t%s\t%s\n" % (image_tgt_list[idx], image_tgt_path, soft_label_tgt_path)) elif args.lr_weight_ent == 0: with open(tgt_train_lst, 'w') as f: for idx in range(tgt_num): image_tgt_path = osp.join(save_pseudo_label_path,image_name_tgt_list[idx]) f.write("%s\t%s\n" % (image_tgt_list[idx], image_tgt_path)) return src_train_lst, tgt_train_lst, src_num_sel class ScoreUpdater(object): # only IoU are computed. accu, cls_accu, etc are ignored. def __init__(self, valid_labels, c_num, x_num, logger=None, label=None, info=None): self._valid_labels = valid_labels self._confs = np.zeros((c_num, c_num)) self._per_cls_iou = np.zeros(c_num) self._logger = logger self._label = label self._info = info self._num_class = c_num self._num_sample = x_num @property def info(self): return self._info def reset(self): self._start = time.time() self._computed = np.zeros(self._num_sample) # one-dimension self._confs[:] = 0 def fast_hist(self,label, pred_label, n): k = (label >= 0) & (label < n) return np.bincount(n * label[k].astype(int) + pred_label[k], minlength=n ** 2).reshape(n, n) def per_class_iu(self,hist): return np.diag(hist) / (hist.sum(1) + hist.sum(0) - np.diag(hist)) def do_updates(self, conf, i, computed=True): if computed: self._computed[i] = 1 self._per_cls_iou = self.per_class_iu(conf) def update(self, pred_label, label, i, computed=True): conf = self.fast_hist(label, pred_label, self._num_class) self._confs += conf self.do_updates(self._confs, i, computed) self.scores(i) def scores(self, i=None, logger=None): x_num = self._num_sample ious = np.nan_to_num( self._per_cls_iou ) logger = self._logger if logger is None else logger if logger is not None: if i is not None: speed = 1. * self._computed.sum() / (time.time() - self._start) logger.info('Done {}/{} with speed: {:.2f}/s'.format(i + 1, x_num, speed)) name = '' if self._label is None else '{}, '.format(self._label) logger.info('{}mean iou: {:.2f}%'. \ format(name, np.mean(ious) * 100)) with util.np_print_options(formatter={'float': '{:5.2f}'.format}): logger.info('\n{}'.format(ious * 100)) return ious def loss_calc(pred, label): """ This function returns cross entropy loss for semantic segmentation """ # out shape batch_size x channels x h x w -> batch_size x channels x h x w # label shape h x w x 1 x batch_size -> batch_size x 1 x h x w criterion = torch.nn.CrossEntropyLoss(ignore_index=IGNORE_LABEL).cuda() return criterion(pred, label) def reg_loss_calc(pred, label, reg_weights, args): """ This function returns cross entropy loss for semantic segmentation """ # out shape batch_size x channels x h x w -> batch_size x channels x h x w # label shape h x w x 1 x batch_size -> batch_size x 1 x h x w mr_weight_kld = args.mr_weight_kld num_class = float(args.num_classes) valid_num = torch.sum(label != IGNORE_LABEL).float() label_reg = label[reg_weights != 0,:,:] valid_reg_num = torch.sum(label_reg != IGNORE_LABEL).float() softmax = F.softmax(pred, dim=1) # compute the softmax values logsoftmax = F.log_softmax(pred,dim=1) # compute the log of softmax values label_expand = torch.unsqueeze(label, 1).repeat(1,int(num_class),1,1) labels = label_expand.clone() labels[labels != IGNORE_LABEL] = 1.0 labels[labels == IGNORE_LABEL] = 0.0 labels_valid = labels.clone() # labels = torch.unsqueeze(labels, 1).repeat(1,num_class,1,1) labels = torch.cumsum(labels, dim=1) labels[labels != label_expand + 1] = 0.0 del label_expand labels[labels != 0 ] = 1.0 ### check the vectorized labels # check_labels = torch.argmax(labels, dim=1) # label[label == 255] = 0 # print(torch.sum(check_labels.float() - label)) reg_weights = reg_weights.float().view(len(reg_weights),1,1,1) ce = torch.sum( -logsoftmax*labels ) # cross-entropy loss with vector-form softmax softmax_val = softmax*labels_valid logsoftmax_val = logsoftmax*labels_valid kld = torch.sum( -logsoftmax_val/num_class*reg_weights ) if valid_reg_num > 0: reg_ce = ce/valid_num + (mr_weight_kld*kld)/valid_reg_num else: reg_ce = ce/valid_num return reg_ce def reg_loss_calc_expand(pred, label, reg_weights, args): """ This function returns cross entropy loss for semantic segmentation """ # out shape batch_size x channels x h x w -> batch_size x channels x h x w # label shape h x w x 1 x batch_size -> batch_size x 1 x h x w mr_weight_kld = args.mr_weight_kld num_class = float(args.num_classes) # soft labels regard ignored labels as zero soft labels in data loader # C = label.cpu().numpy() label_sum = torch.sum(label,1) # D = label_sum.cpu().numpy() valid_num = torch.sum(label_sum != 0.0).float() label_reg = label_sum[reg_weights != 0,:,:] valid_reg_num = torch.sum(label_reg != 0.0).float() softmax = F.softmax(pred, dim=1) # compute the softmax values logsoftmax = F.log_softmax(pred,dim=1) # compute the log of softmax values label_expand = torch.unsqueeze(label_sum, 1).repeat(1,num_class,1,1) label_valid = label_expand.clone() label_valid[label_valid != 0] = 1.0 label_valid = label_valid.clone() # # check the vectorized labels # check_labels = torch.argmax(labels, dim=1) # label[label == 255] = 0 # print(torch.sum(check_labels.float() - label)) # reg_weights = reg_weights.float().view(len(reg_weights),1,1,1) ce = torch.sum( -logsoftmax*label ) # cross-entropy loss with vector-form softmax softmax_val = softmax*label_valid logsoftmax_val = logsoftmax*label_valid kld = torch.sum( -logsoftmax_val/num_class*reg_weights ) if valid_reg_num > 0: reg_ce = ce/valid_num + (mr_weight_kld*kld)/valid_reg_num else: reg_ce = ce/valid_num return reg_ce def lr_poly(base_lr, iter, max_iter, power): return base_lr * ((1 - float(iter) / max_iter) ** (power)) def get_1x_lr_params_NOscale(model): """ This generator returns all the parameters of the net except for the last classification layer. Note that for each batchnorm layer, requires_grad is set to False in deeplab_resnet.py, therefore this function does not return any batchnorm parameter """ b = [] b.append(model.conv1) b.append(model.bn1) b.append(model.layer1) b.append(model.layer2) b.append(model.layer3) b.append(model.layer4) for i in range(len(b)): for j in b[i].modules(): jj = 0 for k in j.parameters(): jj += 1 if k.requires_grad: yield k def get_10x_lr_params(model): """ This generator returns all the parameters for the last layer of the net, which does the classification of pixel into classes """ b = [] b.append(model.layer5.parameters()) for j in range(len(b)): for i in b[j]: yield i def adjust_learning_rate(optimizer, i_iter, tot_iter): lr = lr_poly(args.learning_rate, i_iter, tot_iter, args.power) optimizer.param_groups[0]['lr'] = lr optimizer.param_groups[1]['lr'] = lr * 10 if __name__ == '__main__': main()
47,013
48.229319
225
py
CRST
CRST-master/train.py
import argparse import torch import torch.nn as nn from torch.utils import data import numpy as np import pickle import cv2 import torch.optim as optim import scipy.misc import torch.backends.cudnn as cudnn import sys import os import os.path as osp from deeplab.model import Res_Deeplab from deeplab.loss import CrossEntropy2d from deeplab.datasets import GTA5DataSet import matplotlib.pyplot as plt import random import timeit import torchvision.transforms as transforms import util start = timeit.default_timer() #IMG_MEAN = np.array((104.00698793,116.66876762,122.67891434), dtype=np.float32) IMG_MEAN = np.array((0.406, 0.456, 0.485), dtype=np.float32) # BGR IMG_STD = np.array((0.225, 0.224, 0.229), dtype=np.float32) # BGR # IMG_MEAN = [0.485, 0.456, 0.406] # IMG_STD = [0.229, 0.224, 0.225] BATCH_SIZE = 4 DATA_DIRECTORY = './datasets/gta5' DATA_LIST_PATH = './dataset/list/gta5/train.lst' NUM_CLASSES = 19 IGNORE_LABEL = 255 INPUT_SIZE = '500,500' TRAIN_SCALE = '0.5,1.5' LEARNING_RATE = 2.5e-4 MOMENTUM = 0.9 NUM_STEPS = 100000 POWER = 0.9 RANDOM_SEED = 1234 RESTORE_FROM = '' SAVE_NUM_IMAGES = 2 SAVE_PRED_EVERY = 5000 SNAPSHOT_DIR = './gta_src_train/' WEIGHT_DECAY = 0.0005 MODEL = 'DeeplabRes101' LOG_FILE = 'log' PIN_MEMORY = True GPU = '0' def get_arguments(): """Parse all the arguments provided from the CLI. Returns: A list of parsed arguments. """ parser = argparse.ArgumentParser(description="DeepLab-ResNet Network") parser.add_argument("--batch-size", type=int, default=BATCH_SIZE, help="Number of images sent to the network in one step.") parser.add_argument("--data-dir", type=str, default=DATA_DIRECTORY, help="Path to the directory containing the PASCAL VOC dataset.") parser.add_argument("--data-list", type=str, default=DATA_LIST_PATH, help="Path to the file listing the images in the dataset.") parser.add_argument("--ignore-label", type=int, default=IGNORE_LABEL, help="The index of the label to ignore during the training.") parser.add_argument("--input-size", type=str, default=INPUT_SIZE, help="Comma-separated string with height and width of images.") parser.add_argument("--learning-rate", type=float, default=LEARNING_RATE, help="Base learning rate for training with polynomial decay.") parser.add_argument("--momentum", type=float, default=MOMENTUM, help="Momentum component of the optimiser.") parser.add_argument("--model", type=str, default=MODEL, help="The base network.") parser.add_argument("--not-restore-last", action="store_true", help="Whether to not restore last (FC) layers.") parser.add_argument("--num-classes", type=int, default=NUM_CLASSES, help="Number of classes to predict (including background).") parser.add_argument("--num-steps", type=int, default=NUM_STEPS, help="Number of training steps.") parser.add_argument("--power", type=float, default=POWER, help="Decay parameter to compute the learning rate.") parser.add_argument("--random-mirror", action="store_true", help="Whether to randomly mirror the inputs during the training.") parser.add_argument("--random-scale", action="store_true", help="Whether to randomly scale the inputs during the training.") parser.add_argument("--train-scale", type=str, default=TRAIN_SCALE, help="The scale for multi-scale training.") parser.add_argument("--random-seed", type=int, default=RANDOM_SEED, help="Random seed to have reproducible results.") parser.add_argument("--restore-from", type=str, default=RESTORE_FROM, help="Where restore model parameters from.") parser.add_argument("--save-num-images", type=int, default=SAVE_NUM_IMAGES, help="How many images to save.") parser.add_argument("--save-pred-every", type=int, default=SAVE_PRED_EVERY, help="Save summaries and checkpoint every often.") parser.add_argument("--snapshot-dir", type=str, default=SNAPSHOT_DIR, help="Where to save snapshots of the model.") parser.add_argument("--weight-decay", type=float, default=WEIGHT_DECAY, help="Regularisation parameter for L2-loss.") parser.add_argument("--gpu", type=str, default=GPU, help="choose gpu device.") parser.add_argument("--pin-memory", type=bool, default=PIN_MEMORY, help="Whether to pin memory in train & eval.") parser.add_argument("--log-file", type=str, default=LOG_FILE, help="The name of log file.") parser.add_argument('--debug',help='True means logging debug info.', default=False, action='store_true') return parser.parse_args() args = get_arguments() def loss_calc(pred, label): """ This function returns cross entropy loss for semantic segmentation """ # out shape batch_size x channels x h x w -> batch_size x channels x h x w # label shape h x w x 1 x batch_size -> batch_size x 1 x h x w criterion = torch.nn.CrossEntropyLoss(ignore_index=IGNORE_LABEL).cuda() return criterion(pred, label) def lr_poly(base_lr, iter, max_iter, power): return base_lr*((1-float(iter)/max_iter)**(power)) def get_1x_lr_params_NOscale(model): """ This generator returns all the parameters of the net except for the last classification layer. Note that for each batchnorm layer, requires_grad is set to False in deeplab_resnet.py, therefore this function does not return any batchnorm parameter """ b = [] b.append(model.conv1) b.append(model.bn1) b.append(model.layer1) b.append(model.layer2) b.append(model.layer3) b.append(model.layer4) for i in range(len(b)): for j in b[i].modules(): jj = 0 for k in j.parameters(): jj+=1 if k.requires_grad: yield k def get_10x_lr_params(model): """ This generator returns all the parameters for the last layer of the net, which does the classification of pixel into classes """ b = [] b.append(model.layer5.parameters()) for j in range(len(b)): for i in b[j]: yield i def adjust_learning_rate(optimizer, i_iter): """Sets the learning rate to the initial LR divided by 5 at 60th, 120th and 160th epochs""" lr = lr_poly(args.learning_rate, i_iter, args.num_steps, args.power) optimizer.param_groups[0]['lr'] = lr optimizer.param_groups[1]['lr'] = lr * 10 def main(): """Create the model and start the training.""" if not os.path.exists(args.snapshot_dir): os.makedirs(args.snapshot_dir) logger = util.set_logger(args.snapshot_dir, args.log_file, args.debug) logger.info('start with arguments %s', args) h, w = map(int, args.input_size.split(',')) input_size = (h, w) lscale, hscale = map(float, args.train_scale.split(',')) train_scale = (lscale, hscale) cudnn.enabled = True # Create network. model = Res_Deeplab(num_classes=args.num_classes) #saved_state_dict = torch.load(args.restore_from) #new_params = model.state_dict().copy() #for i in saved_state_dict: # #Scale.layer5.conv2d_list.3.weight # i_parts = i.split('.') # # print i_parts # if not args.num_classes == 21 or not i_parts[1]=='layer5': # new_params['.'.join(i_parts[1:])] = saved_state_dict[i] model_urls = { 'resnet18': 'https://download.pytorch.org/models/resnet18-5c106cde.pth', 'resnet34': 'https://download.pytorch.org/models/resnet34-333f7ec4.pth', 'resnet50': 'https://download.pytorch.org/models/resnet50-19c8e357.pth', 'resnet101': 'https://download.pytorch.org/models/resnet101-5d3b4d8f.pth', 'resnet152': 'https://download.pytorch.org/models/resnet152-b121ed2d.pth', } saved_state_dict = torch.utils.model_zoo.load_url(model_urls['resnet101']) # coco pretrained parameters: # saved_state_dict = torch.load(args.restore_from) new_params = model.state_dict().copy() for i in saved_state_dict: #Scale.layer5.conv2d_list.3.weight i_parts = str(i).split('.') # print i_parts if not i_parts[0]=='fc': new_params['.'.join(i_parts[0:])] = saved_state_dict[i] model.load_state_dict(new_params) #model.float() model.eval() # use_global_stats = True #model.train() device = torch.device("cuda:" + str(args.gpu)) model.to(device) cudnn.benchmark = True trainloader = data.DataLoader(GTA5DataSet(args.data_dir, args.data_list, max_iters=args.num_steps*args.batch_size, crop_size=input_size,train_scale=train_scale, scale=args.random_scale, mirror=args.random_mirror, mean=IMG_MEAN, std = IMG_STD), batch_size=args.batch_size, shuffle=True, num_workers=5, pin_memory=args.pin_memory) optimizer = optim.SGD([{'params': get_1x_lr_params_NOscale(model), 'lr': args.learning_rate }, {'params': get_10x_lr_params(model), 'lr': 10*args.learning_rate}], lr=args.learning_rate, momentum=args.momentum,weight_decay=args.weight_decay) optimizer.zero_grad() interp = nn.Upsample(size=input_size, mode='bilinear', align_corners=True) for i_iter, batch in enumerate(trainloader): images, labels, _, _ = batch images = images.to(device) labels = labels.long().to(device) optimizer.zero_grad() adjust_learning_rate(optimizer, i_iter) pred = interp(model(images)) loss = loss_calc(pred, labels) loss.backward() optimizer.step() # print('iter = ', i_iter, 'of', args.num_steps,'completed, loss = ', loss.data.cpu().numpy()) logger.info('iter = {} of {} completed, loss = {:.4f}'.format(i_iter,args.num_steps,loss.data.cpu().numpy())) if i_iter >= args.num_steps-1: print('save model ...') torch.save(model.state_dict(),osp.join(args.snapshot_dir, 'VOC12_scenes_'+str(args.num_steps)+'.pth')) break if i_iter % args.save_pred_every == 0 and i_iter!=0: print('taking snapshot ...') torch.save(model.state_dict(),osp.join(args.snapshot_dir, 'VOC12_scenes_'+str(i_iter)+'.pth')) end = timeit.default_timer() print(end-start,'seconds') if __name__ == '__main__': main()
10,806
39.324627
164
py
CRST
CRST-master/dataset/helpers/labels.py
#!/usr/bin/python # # Cityscapes labels # from collections import namedtuple #-------------------------------------------------------------------------------- # Definitions #-------------------------------------------------------------------------------- # a label and all meta information Label = namedtuple( 'Label' , [ 'name' , # The identifier of this label, e.g. 'car', 'person', ... . # We use them to uniquely name a class 'id' , # An integer ID that is associated with this label. # The IDs are used to represent the label in ground truth images # An ID of -1 means that this label does not have an ID and thus # is ignored when creating ground truth images (e.g. license plate). # Do not modify these IDs, since exactly these IDs are expected by the # evaluation server. 'trainId' , # Feel free to modify these IDs as suitable for your method. Then create # ground truth images with train IDs, using the tools provided in the # 'preparation' folder. However, make sure to validate or submit results # to our evaluation server using the regular IDs above! # For trainIds, multiple labels might have the same ID. Then, these labels # are mapped to the same class in the ground truth images. For the inverse # mapping, we use the label that is defined first in the list below. # For example, mapping all void-type classes to the same ID in training, # might make sense for some approaches. # Max value is 255! 'category' , # The name of the category that this label belongs to 'categoryId' , # The ID of this category. Used to create ground truth images # on category level. 'hasInstances', # Whether this label distinguishes between single instances or not 'ignoreInEval', # Whether pixels having this class as ground truth label are ignored # during evaluations or not 'color' , # The color of this label ] ) #-------------------------------------------------------------------------------- # A list of all labels #-------------------------------------------------------------------------------- # Please adapt the train IDs as appropriate for you approach. # Note that you might want to ignore labels with ID 255 during training. # Further note that the current train IDs are only a suggestion. You can use whatever you like. # Make sure to provide your results using the original IDs and not the training IDs. # Note that many IDs are ignored in evaluation and thus you never need to predict these! labels = [ # name id trainId category catId hasInstances ignoreInEval color Label( 'unlabeled' , 0 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'ego vehicle' , 1 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'rectification border' , 2 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'out of roi' , 3 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'static' , 4 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'dynamic' , 5 , 255 , 'void' , 0 , False , True , (111, 74, 0) ), Label( 'ground' , 6 , 255 , 'void' , 0 , False , True , ( 81, 0, 81) ), Label( 'road' , 7 , 0 , 'flat' , 1 , False , False , (128, 64,128) ), Label( 'sidewalk' , 8 , 1 , 'flat' , 1 , False , False , (244, 35,232) ), Label( 'parking' , 9 , 255 , 'flat' , 1 , False , True , (250,170,160) ), Label( 'rail track' , 10 , 255 , 'flat' , 1 , False , True , (230,150,140) ), Label( 'building' , 11 , 2 , 'construction' , 2 , False , False , ( 70, 70, 70) ), Label( 'wall' , 12 , 3 , 'construction' , 2 , False , False , (102,102,156) ), Label( 'fence' , 13 , 4 , 'construction' , 2 , False , False , (190,153,153) ), Label( 'guard rail' , 14 , 255 , 'construction' , 2 , False , True , (180,165,180) ), Label( 'bridge' , 15 , 255 , 'construction' , 2 , False , True , (150,100,100) ), Label( 'tunnel' , 16 , 255 , 'construction' , 2 , False , True , (150,120, 90) ), Label( 'pole' , 17 , 5 , 'object' , 3 , False , False , (153,153,153) ), Label( 'polegroup' , 18 , 255 , 'object' , 3 , False , True , (153,153,153) ), Label( 'traffic light' , 19 , 6 , 'object' , 3 , False , False , (250,170, 30) ), Label( 'traffic sign' , 20 , 7 , 'object' , 3 , False , False , (220,220, 0) ), Label( 'vegetation' , 21 , 8 , 'nature' , 4 , False , False , (107,142, 35) ), Label( 'terrain' , 22 , 9 , 'nature' , 4 , False , False , (152,251,152) ), Label( 'sky' , 23 , 10 , 'sky' , 5 , False , False , ( 70,130,180) ), Label( 'person' , 24 , 11 , 'human' , 6 , True , False , (220, 20, 60) ), Label( 'rider' , 25 , 12 , 'human' , 6 , True , False , (255, 0, 0) ), Label( 'car' , 26 , 13 , 'vehicle' , 7 , True , False , ( 0, 0,142) ), Label( 'truck' , 27 , 14 , 'vehicle' , 7 , True , False , ( 0, 0, 70) ), Label( 'bus' , 28 , 15 , 'vehicle' , 7 , True , False , ( 0, 60,100) ), Label( 'caravan' , 29 , 255 , 'vehicle' , 7 , True , True , ( 0, 0, 90) ), Label( 'trailer' , 30 , 255 , 'vehicle' , 7 , True , True , ( 0, 0,110) ), Label( 'train' , 31 , 16 , 'vehicle' , 7 , True , False , ( 0, 80,100) ), Label( 'motorcycle' , 32 , 17 , 'vehicle' , 7 , True , False , ( 0, 0,230) ), Label( 'bicycle' , 33 , 18 , 'vehicle' , 7 , True , False , (119, 11, 32) ), Label( 'license plate' , -1 , -1 , 'vehicle' , 7 , False , True , ( 0, 0,142) ), ] #-------------------------------------------------------------------------------- # Create dictionaries for a fast lookup #-------------------------------------------------------------------------------- # Please refer to the main method below for example usages! # name to label object name2label = { label.name : label for label in labels } # id to label object id2label = { label.id : label for label in labels } # trainId to label object trainId2label = { label.trainId : label for label in reversed(labels) } # category to list of label objects category2labels = {} for label in labels: category = label.category if category in category2labels: category2labels[category].append(label) else: category2labels[category] = [label] #-------------------------------------------------------------------------------- # Assure single instance name #-------------------------------------------------------------------------------- # returns the label name that describes a single instance (if possible) # e.g. input | output # ---------------------- # car | car # cargroup | car # foo | None # foogroup | None # skygroup | None def assureSingleInstanceName( name ): # if the name is known, it is not a group if name in name2label: return name # test if the name actually denotes a group if not name.endswith("group"): return None # remove group name = name[:-len("group")] # test if the new name exists if not name in name2label: return None # test if the new name denotes a label that actually has instances if not name2label[name].hasInstances: return None # all good then return name #-------------------------------------------------------------------------------- # Main for testing #-------------------------------------------------------------------------------- # just a dummy main if __name__ == "__main__": # Print all the labels print("List of cityscapes labels:") print("") print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( 'name', 'id', 'trainId', 'category', 'categoryId', 'hasInstances', 'ignoreInEval' )) print(" " + ('-' * 98)) for label in labels: print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( label.name, label.id, label.trainId, label.category, label.categoryId, label.hasInstances, label.ignoreInEval )) print("") print("Example usages:") # Map from name to label name = 'car' id = name2label[name].id print("ID of label '{name}': {id}".format( name=name, id=id )) # Map from ID to label category = id2label[id].category print("Category of label with ID '{id}': {category}".format( id=id, category=category )) # Map from trainID to label trainId = 0 name = trainId2label[trainId].name print("Name of label with trainID '{id}': {name}".format( id=trainId, name=name ))
10,597
57.230769
199
py
CRST
CRST-master/dataset/helpers/labels_synthia.py
#!/usr/bin/python # # Cityscapes labels # from collections import namedtuple #-------------------------------------------------------------------------------- # Definitions #-------------------------------------------------------------------------------- # a label and all meta information Label = namedtuple( 'Label' , [ 'name' , # The identifier of this label, e.g. 'car', 'person', ... . # We use them to uniquely name a class 'id' , # An integer ID that is associated with this label. # The IDs are used to represent the label in ground truth images # An ID of -1 means that this label does not have an ID and thus # is ignored when creating ground truth images (e.g. license plate). # Do not modify these IDs, since exactly these IDs are expected by the # evaluation server. 'trainId' , # Feel free to modify these IDs as suitable for your method. Then create # ground truth images with train IDs, using the tools provided in the # 'preparation' folder. However, make sure to validate or submit results # to our evaluation server using the regular IDs above! # For trainIds, multiple labels might have the same ID. Then, these labels # are mapped to the same class in the ground truth images. For the inverse # mapping, we use the label that is defined first in the list below. # For example, mapping all void-type classes to the same ID in training, # might make sense for some approaches. # Max value is 255! 'category' , # The name of the category that this label belongs to 'categoryId' , # The ID of this category. Used to create ground truth images # on category level. 'hasInstances', # Whether this label distinguishes between single instances or not 'ignoreInEval', # Whether pixels having this class as ground truth label are ignored # during evaluations or not 'color' , # The color of this label ] ) #-------------------------------------------------------------------------------- # A list of all labels #-------------------------------------------------------------------------------- # Please adapt the train IDs as appropriate for you approach. # Note that you might want to ignore labels with ID 255 during training. # Further note that the current train IDs are only a suggestion. You can use whatever you like. # Make sure to provide your results using the original IDs and not the training IDs. # Note that many IDs are ignored in evaluation and thus you never need to predict these! labels = [ # name id trainId category catId hasInstances ignoreInEval color Label( 'void' , 0 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'sky' , 1 , 9 , 'void' , 0 , False , True , ( 70,130,180) ), Label( 'building' , 2 , 2 , 'void' , 0 , False , True , ( 70,70,70) ), Label( 'road' , 3 , 0 , 'void' , 0 , False , True , ( 128,64,128) ), Label( 'sidewalk' , 4 , 1 , 'void' , 0 , False , True , ( 244,35,232) ), Label( 'fence' , 5 , 4 , 'void' , 0 , False , True , ( 64,64,128) ), Label( 'vegetation' , 6 , 8 , 'void' , 0 , False , True , ( 107,142,35) ), Label( 'pole' , 7 , 5 , 'flat' , 1 , False , False , ( 153,153,153) ), Label( 'car' , 8 , 12 , 'flat' , 1 , False , False , ( 0,0,142) ), Label( 'traffic sign' , 9 , 7 , 'flat' , 1 , False , True , ( 220,220,0) ), Label( 'pedestrian' , 10 , 10 , 'flat' , 1 , False , True , ( 220,20,60) ), Label( 'bicycle' , 11 , 15 , 'construction' , 2 , False , False , ( 119,11,32) ), Label( 'motorcycle' , 12 , 14 , 'construction' , 2 , False , False , ( 0,0,230) ), Label( 'parking-slot' , 13 , 255 , 'construction' , 2 , False , False , (250,170,160) ), Label( 'road-work' , 14 , 255 , 'construction' , 2 , False , True , (128,64,64) ), Label( 'traffic light' , 15 , 6 , 'construction' , 2 , False , True , (250,170,30) ), Label( 'terrain' , 16 , 255 , 'construction' , 2 , False , True , (152,251,152) ), Label( 'rider' , 17 , 11 , 'object' , 3 , False , False , (255,0,0) ), Label( 'truck' , 18 , 255 , 'object' , 3 , False , True , (0,0,70) ), Label( 'bus' , 19 , 13 , 'object' , 3 , False , False , (0,60,100) ), Label( 'train' , 20 , 255 , 'object' , 3 , False , False , (0,80,100) ), Label( 'wall' , 21 , 3 , 'nature' , 4 , False , False , (102,102,156) ), Label( 'lanemarking' , 22 , 0 , 'nature' , 4 , False , False , (102,102,156) ), ] #-------------------------------------------------------------------------------- # Create dictionaries for a fast lookup #-------------------------------------------------------------------------------- # Please refer to the main method below for example usages! # name to label object name2label = { label.name : label for label in labels } # id to label object id2label = { label.id : label for label in labels } # trainId to label object trainId2label = { label.trainId : label for label in reversed(labels) } # category to list of label objects category2labels = {} for label in labels: category = label.category if category in category2labels: category2labels[category].append(label) else: category2labels[category] = [label] #-------------------------------------------------------------------------------- # Assure single instance name #-------------------------------------------------------------------------------- # returns the label name that describes a single instance (if possible) # e.g. input | output # ---------------------- # car | car # cargroup | car # foo | None # foogroup | None # skygroup | None def assureSingleInstanceName( name ): # if the name is known, it is not a group if name in name2label: return name # test if the name actually denotes a group if not name.endswith("group"): return None # remove group name = name[:-len("group")] # test if the new name exists if not name in name2label: return None # test if the new name denotes a label that actually has instances if not name2label[name].hasInstances: return None # all good then return name #-------------------------------------------------------------------------------- # Main for testing #-------------------------------------------------------------------------------- # just a dummy main if __name__ == "__main__": # Print all the labels print("List of cityscapes labels:") print("") print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( 'name', 'id', 'trainId', 'category', 'categoryId', 'hasInstances', 'ignoreInEval' )) print(" " + ('-' * 98)) for label in labels: print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( label.name, label.id, label.trainId, label.category, label.categoryId, label.hasInstances, label.ignoreInEval )) print("") print("Example usages:") # Map from name to label name = 'car' id = name2label[name].id print("ID of label '{name}': {id}".format( name=name, id=id )) # Map from ID to label category = id2label[id].category print("Category of label with ID '{id}': {category}".format( id=id, category=category )) # Map from trainID to label trainId = 0 name = trainId2label[trainId].name print("Name of label with trainID '{id}': {name}".format( id=trainId, name=name ))
9,012
51.707602
199
py
CRST
CRST-master/dataset/helpers/labels_cityscapes_synthia.py
#!/usr/bin/python # # Cityscapes labels # from collections import namedtuple #-------------------------------------------------------------------------------- # Definitions #-------------------------------------------------------------------------------- # a label and all meta information Label = namedtuple( 'Label' , [ 'name' , # The identifier of this label, e.g. 'car', 'person', ... . # We use them to uniquely name a class 'id' , # An integer ID that is associated with this label. # The IDs are used to represent the label in ground truth images # An ID of -1 means that this label does not have an ID and thus # is ignored when creating ground truth images (e.g. license plate). # Do not modify these IDs, since exactly these IDs are expected by the # evaluation server. 'trainId' , # Feel free to modify these IDs as suitable for your method. Then create # ground truth images with train IDs, using the tools provided in the # 'preparation' folder. However, make sure to validate or submit results # to our evaluation server using the regular IDs above! # For trainIds, multiple labels might have the same ID. Then, these labels # are mapped to the same class in the ground truth images. For the inverse # mapping, we use the label that is defined first in the list below. # For example, mapping all void-type classes to the same ID in training, # might make sense for some approaches. # Max value is 255! 'category' , # The name of the category that this label belongs to 'categoryId' , # The ID of this category. Used to create ground truth images # on category level. 'hasInstances', # Whether this label distinguishes between single instances or not 'ignoreInEval', # Whether pixels having this class as ground truth label are ignored # during evaluations or not 'color' , # The color of this label ] ) #-------------------------------------------------------------------------------- # A list of all labels #-------------------------------------------------------------------------------- # Please adapt the train IDs as appropriate for you approach. # Note that you might want to ignore labels with ID 255 during training. # Further note that the current train IDs are only a suggestion. You can use whatever you like. # Make sure to provide your results using the original IDs and not the training IDs. # Note that many IDs are ignored in evaluation and thus you never need to predict these! labels = [ # name id trainId category catId hasInstances ignoreInEval color Label( 'unlabeled' , 0 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'ego vehicle' , 1 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'rectification border' , 2 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'out of roi' , 3 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'static' , 4 , 255 , 'void' , 0 , False , True , ( 0, 0, 0) ), Label( 'dynamic' , 5 , 255 , 'void' , 0 , False , True , (111, 74, 0) ), Label( 'ground' , 6 , 255 , 'void' , 0 , False , True , ( 81, 0, 81) ), Label( 'road' , 7 , 0 , 'flat' , 1 , False , False , (128, 64,128) ), Label( 'sidewalk' , 8 , 1 , 'flat' , 1 , False , False , (244, 35,232) ), Label( 'parking' , 9 , 255 , 'flat' , 1 , False , True , (250,170,160) ), Label( 'rail track' , 10 , 255 , 'flat' , 1 , False , True , (230,150,140) ), Label( 'building' , 11 , 2 , 'construction' , 2 , False , False , ( 70, 70, 70) ), Label( 'wall' , 12 , 3 , 'construction' , 2 , False , False , (102,102,156) ), Label( 'fence' , 13 , 4 , 'construction' , 2 , False , False , (190,153,153) ), Label( 'guard rail' , 14 , 255 , 'construction' , 2 , False , True , (180,165,180) ), Label( 'bridge' , 15 , 255 , 'construction' , 2 , False , True , (150,100,100) ), Label( 'tunnel' , 16 , 255 , 'construction' , 2 , False , True , (150,120, 90) ), Label( 'pole' , 17 , 5 , 'object' , 3 , False , False , (153,153,153) ), Label( 'polegroup' , 18 , 255 , 'object' , 3 , False , True , (153,153,153) ), Label( 'traffic light' , 19 , 6 , 'object' , 3 , False , False , (250,170, 30) ), Label( 'traffic sign' , 20 , 7 , 'object' , 3 , False , False , (220,220, 0) ), Label( 'vegetation' , 21 , 8 , 'nature' , 4 , False , False , (107,142, 35) ), Label( 'terrain' , 22 , 255 , 'nature' , 4 , False , False , (152,251,152) ), Label( 'sky' , 23 , 9 , 'sky' , 5 , False , False , ( 70,130,180) ), Label( 'person' , 24 , 10 , 'human' , 6 , True , False , (220, 20, 60) ), Label( 'rider' , 25 , 11 , 'human' , 6 , True , False , (255, 0, 0) ), Label( 'car' , 26 , 12 , 'vehicle' , 7 , True , False , ( 0, 0,142) ), Label( 'truck' , 27 , 255 , 'vehicle' , 7 , True , False , ( 0, 0, 70) ), Label( 'bus' , 28 , 13 , 'vehicle' , 7 , True , False , ( 0, 60,100) ), Label( 'caravan' , 29 , 255 , 'vehicle' , 7 , True , True , ( 0, 0, 90) ), Label( 'trailer' , 30 , 255 , 'vehicle' , 7 , True , True , ( 0, 0,110) ), Label( 'train' , 31 , 255 , 'vehicle' , 7 , True , False , ( 0, 80,100) ), Label( 'motorcycle' , 32 , 14 , 'vehicle' , 7 , True , False , ( 0, 0,230) ), Label( 'bicycle' , 33 , 15 , 'vehicle' , 7 , True , False , (119, 11, 32) ), Label( 'license plate' , -1 , -1 , 'vehicle' , 7 , False , True , ( 0, 0,142) ), ] #-------------------------------------------------------------------------------- # Create dictionaries for a fast lookup #-------------------------------------------------------------------------------- # Please refer to the main method below for example usages! # name to label object name2label = { label.name : label for label in labels } # id to label object id2label = { label.id : label for label in labels } # trainId to label object trainId2label = { label.trainId : label for label in reversed(labels) } # category to list of label objects category2labels = {} for label in labels: category = label.category if category in category2labels: category2labels[category].append(label) else: category2labels[category] = [label] #-------------------------------------------------------------------------------- # Assure single instance name #-------------------------------------------------------------------------------- # returns the label name that describes a single instance (if possible) # e.g. input | output # ---------------------- # car | car # cargroup | car # foo | None # foogroup | None # skygroup | None def assureSingleInstanceName( name ): # if the name is known, it is not a group if name in name2label: return name # test if the name actually denotes a group if not name.endswith("group"): return None # remove group name = name[:-len("group")] # test if the new name exists if not name in name2label: return None # test if the new name denotes a label that actually has instances if not name2label[name].hasInstances: return None # all good then return name #-------------------------------------------------------------------------------- # Main for testing #-------------------------------------------------------------------------------- # just a dummy main if __name__ == "__main__": # Print all the labels print("List of cityscapes labels:") print("") print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( 'name', 'id', 'trainId', 'category', 'categoryId', 'hasInstances', 'ignoreInEval' )) print(" " + ('-' * 98)) for label in labels: print(" {:>21} | {:>3} | {:>7} | {:>14} | {:>10} | {:>12} | {:>12}".format( label.name, label.id, label.trainId, label.category, label.categoryId, label.hasInstances, label.ignoreInEval )) print("") print("Example usages:") # Map from name to label name = 'car' id = name2label[name].id print("ID of label '{name}': {id}".format( name=name, id=id )) # Map from ID to label category = id2label[id].category print("Category of label with ID '{id}': {category}".format( id=id, category=category )) # Map from trainID to label trainId = 0 name = trainId2label[trainId].name print("Name of label with trainID '{id}': {name}".format( id=trainId, name=name ))
10,597
57.230769
199
py
CRST
CRST-master/deeplab/loss.py
import torch import torch.nn.functional as F import torch.nn as nn from torch.autograd import Variable class CrossEntropy2d(nn.Module): def __init__(self, size_average=True, ignore_label=255): super(CrossEntropy2d, self).__init__() self.size_average = size_average self.ignore_label = ignore_label def forward(self, predict, target, weight=None): """ Args: predict:(n, c, h, w) target:(n, h, w) weight (Tensor, optional): a manual rescaling weight given to each class. If given, has to be a Tensor of size "nclasses" """ assert not target.requires_grad assert predict.dim() == 4 assert target.dim() == 3 assert predict.size(0) == target.size(0), "{0} vs {1} ".format(predict.size(0), target.size(0)) assert predict.size(2) == target.size(1), "{0} vs {1} ".format(predict.size(2), target.size(1)) assert predict.size(3) == target.size(2), "{0} vs {1} ".format(predict.size(3), target.size(3)) n, c, h, w = predict.size() target_mask = (target >= 0) * (target != self.ignore_label) target = target[target_mask] if not target.data.dim(): return Variable(torch.zeros(1)) predict = predict.transpose(1, 2).transpose(2, 3).contiguous() predict = predict[target_mask.view(n, h, w, 1).repeat(1, 1, 1, c)].view(-1, c) loss = F.cross_entropy(predict, target, weight=weight, size_average=self.size_average) return loss
1,585
44.314286
103
py
CRST
CRST-master/deeplab/model.py
import torch.nn as nn import math import torch.utils.model_zoo as model_zoo import torch import numpy as np affine_par = True def outS(i): i = int(i) i = (i+1)/2 i = int(np.ceil((i+1)/2.0)) i = (i+1)/2 return i def conv3x3(in_planes, out_planes, stride=1): "3x3 convolution with padding" return nn.Conv2d(in_planes, out_planes, kernel_size=3, stride=stride, padding=1, bias=False) class BasicBlock(nn.Module): expansion = 1 def __init__(self, inplanes, planes, stride=1, downsample=None): super(BasicBlock, self).__init__() self.conv1 = conv3x3(inplanes, planes, stride) self.bn1 = nn.BatchNorm2d(planes, affine = affine_par) self.relu = nn.ReLU(inplace=True) self.conv2 = conv3x3(planes, planes) self.bn2 = nn.BatchNorm2d(planes, affine = affine_par) self.downsample = downsample self.stride = stride def forward(self, x): residual = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out class Bottleneck(nn.Module): expansion = 4 def __init__(self, inplanes, planes, stride=1, dilation=1, downsample=None): super(Bottleneck, self).__init__() self.conv1 = nn.Conv2d(inplanes, planes, kernel_size=1, stride=stride, bias=False) # change self.bn1 = nn.BatchNorm2d(planes,affine = affine_par) for i in self.bn1.parameters(): i.requires_grad = False padding = dilation self.conv2 = nn.Conv2d(planes, planes, kernel_size=3, stride=1, # change padding=padding, bias=False, dilation = dilation) self.bn2 = nn.BatchNorm2d(planes,affine = affine_par) for i in self.bn2.parameters(): i.requires_grad = False self.conv3 = nn.Conv2d(planes, planes * 4, kernel_size=1, bias=False) self.bn3 = nn.BatchNorm2d(planes * 4, affine = affine_par) for i in self.bn3.parameters(): i.requires_grad = False self.relu = nn.ReLU(inplace=True) self.downsample = downsample self.stride = stride def forward(self, x): residual = x out = self.conv1(x) out = self.bn1(out) out = self.relu(out) out = self.conv2(out) out = self.bn2(out) out = self.relu(out) out = self.conv3(out) out = self.bn3(out) if self.downsample is not None: residual = self.downsample(x) out += residual out = self.relu(out) return out class Classifier_Module(nn.Module): def __init__(self, dilation_series, padding_series, num_classes): super(Classifier_Module, self).__init__() self.conv2d_list = nn.ModuleList() for dilation, padding in zip(dilation_series, padding_series): self.conv2d_list.append(nn.Conv2d(2048, num_classes, kernel_size=3, stride=1, padding=padding, dilation=dilation, bias = True)) for m in self.conv2d_list: m.weight.data.normal_(0, 0.01) def forward(self, x): out = self.conv2d_list[0](x) for i in range(len(self.conv2d_list)-1): out += self.conv2d_list[i+1](x) return out class Residual_Covolution(nn.Module): def __init__(self, icol, ocol, num_classes): super(Residual_Covolution, self).__init__() self.conv1 = nn.Conv2d(icol, ocol, kernel_size=3, stride=1, padding=12, dilation=12, bias=True) self.conv2 = nn.Conv2d(ocol, num_classes, kernel_size=3, stride=1, padding=12, dilation=12, bias=True) self.conv3 = nn.Conv2d(num_classes, ocol, kernel_size=1, stride=1, padding=0, dilation=1, bias=True) self.conv4 = nn.Conv2d(ocol, icol, kernel_size=1, stride=1, padding=0, dilation=1, bias=True) self.relu = nn.ReLU(inplace=True) def forward(self, x): dow1 = self.conv1(x) dow1 = self.relu(dow1) seg = self.conv2(dow1) inc1 = self.conv3(seg) add1 = dow1 + self.relu(inc1) inc2 = self.conv4(add1) out = x + self.relu(inc2) return out, seg class Residual_Refinement_Module(nn.Module): def __init__(self, num_classes): super(Residual_Refinement_Module, self).__init__() self.RC1 = Residual_Covolution(2048, 512, num_classes) self.RC2 = Residual_Covolution(2048, 512, num_classes) def forward(self, x): x, seg1 = self.RC1(x) _, seg2 = self.RC2(x) return [seg1, seg1+seg2] class ResNet_Refine(nn.Module): def __init__(self, block, layers, num_classes): self.inplanes = 64 super(ResNet_Refine, self).__init__() self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False) self.bn1 = nn.BatchNorm2d(64, affine = affine_par) for i in self.bn1.parameters(): i.requires_grad = False self.relu = nn.ReLU(inplace=True) self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1, ceil_mode=True) # change self.layer1 = self._make_layer(block, 64, layers[0]) self.layer2 = self._make_layer(block, 128, layers[1], stride=2) self.layer3 = self._make_layer(block, 256, layers[2], stride=1, dilation=2) self.layer4 = self._make_layer(block, 512, layers[3], stride=1, dilation=4) self.layer5 = Residual_Refinement_Module(num_classes) for m in self.modules(): if isinstance(m, nn.Conv2d): n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels m.weight.data.normal_(0, 0.01) elif isinstance(m, nn.BatchNorm2d): m.weight.data.fill_(1) m.bias.data.zero_() # for i in m.parameters(): # i.requires_grad = False def _make_layer(self, block, planes, blocks, stride=1, dilation=1): downsample = None if stride != 1 or self.inplanes != planes * block.expansion or dilation == 2 or dilation == 4: downsample = nn.Sequential( nn.Conv2d(self.inplanes, planes * block.expansion, kernel_size=1, stride=stride, bias=False), nn.BatchNorm2d(planes * block.expansion,affine = affine_par)) for i in downsample._modules['1'].parameters(): i.requires_grad = False layers = [] layers.append(block(self.inplanes, planes, stride,dilation=dilation, downsample=downsample)) self.inplanes = planes * block.expansion for i in range(1, blocks): layers.append(block(self.inplanes, planes, dilation=dilation)) return nn.Sequential(*layers) def forward(self, x): x = self.conv1(x) x = self.bn1(x) x = self.relu(x) x = self.maxpool(x) x = self.layer1(x) x = self.layer2(x) x = self.layer3(x) x = self.layer4(x) x = self.layer5(x) return x class ResNet(nn.Module): def __init__(self, block, layers, num_classes): self.inplanes = 64 super(ResNet, self).__init__() self.conv1 = nn.Conv2d(3, 64, kernel_size=7, stride=2, padding=3, bias=False) self.bn1 = nn.BatchNorm2d(64, affine = affine_par) for i in self.bn1.parameters(): i.requires_grad = False self.relu = nn.ReLU(inplace=True) self.maxpool = nn.MaxPool2d(kernel_size=3, stride=2, padding=1, ceil_mode=True) # change self.layer1 = self._make_layer(block, 64, layers[0]) self.layer2 = self._make_layer(block, 128, layers[1], stride=2) self.layer3 = self._make_layer(block, 256, layers[2], stride=1, dilation=2) self.layer4 = self._make_layer(block, 512, layers[3], stride=1, dilation=4) self.layer5 = self._make_pred_layer(Classifier_Module, [6,12,18,24],[6,12,18,24],num_classes) for m in self.modules(): if isinstance(m, nn.Conv2d): n = m.kernel_size[0] * m.kernel_size[1] * m.out_channels m.weight.data.normal_(0, 0.01) elif isinstance(m, nn.BatchNorm2d): m.weight.data.fill_(1) m.bias.data.zero_() # for i in m.parameters(): # i.requires_grad = False def _make_layer(self, block, planes, blocks, stride=1, dilation=1): downsample = None if stride != 1 or self.inplanes != planes * block.expansion or dilation == 2 or dilation == 4: downsample = nn.Sequential( nn.Conv2d(self.inplanes, planes * block.expansion, kernel_size=1, stride=stride, bias=False), nn.BatchNorm2d(planes * block.expansion,affine = affine_par)) for i in downsample._modules['1'].parameters(): i.requires_grad = False layers = [] layers.append(block(self.inplanes, planes, stride,dilation=dilation, downsample=downsample)) self.inplanes = planes * block.expansion for i in range(1, blocks): layers.append(block(self.inplanes, planes, dilation=dilation)) return nn.Sequential(*layers) def _make_pred_layer(self,block, dilation_series, padding_series,num_classes): return block(dilation_series,padding_series,num_classes) def forward(self, x): x = self.conv1(x) x = self.bn1(x) x = self.relu(x) x = self.maxpool(x) x = self.layer1(x) x = self.layer2(x) x = self.layer3(x) x = self.layer4(x) x = self.layer5(x) return x class MS_Deeplab(nn.Module): def __init__(self,block,num_classes): super(MS_Deeplab,self).__init__() self.Scale = ResNet(block,[3, 4, 23, 3],num_classes) #changed to fix #4 def forward(self,x): output = self.Scale(x) # for original scale output_size = output.size()[2] input_size = x.size()[2] self.interp1 = nn.Upsample(size=(int(input_size*0.75)+1, int(input_size*0.75)+1), mode='bilinear') self.interp2 = nn.Upsample(size=(int(input_size*0.5)+1, int(input_size*0.5)+1), mode='bilinear') self.interp3 = nn.Upsample(size=(output_size, output_size), mode='bilinear') x75 = self.interp1(x) output75 = self.interp3(self.Scale(x75)) # for 0.75x scale x5 = self.interp2(x) output5 = self.interp3(self.Scale(x5)) # for 0.5x scale out_max = torch.max(torch.max(output, output75), output5) return [output, output75, output5, out_max] def Res_Ms_Deeplab(num_classes=21): model = MS_Deeplab(Bottleneck, num_classes) return model def Res_Deeplab(num_classes=21, is_refine=False): if is_refine: model = ResNet_Refine(Bottleneck,[3, 4, 23, 3], num_classes) else: model = ResNet(Bottleneck,[3, 4, 23, 3], num_classes) return model
11,127
36.217391
139
py
CRST
CRST-master/deeplab/datasets.py
import os import os.path as osp import numpy as np import random import matplotlib.pyplot as plt import collections import torch import torchvision.transforms as transforms import torchvision import cv2 from torch.utils import data import sys from PIL import Image palette = [128, 64, 128, 244, 35, 232, 70, 70, 70, 102, 102, 156, 190, 153, 153, 153, 153, 153, 250, 170, 30, 220, 220, 0, 107, 142, 35, 152, 251, 152, 70, 130, 180, 220, 20, 60, 255, 0, 0, 0, 0, 142, 0, 0, 70, 0, 60, 100, 0, 80, 100, 0, 0, 230, 119, 11, 32] zero_pad = 256 * 3 - len(palette) for i in range(zero_pad): palette.append(0) def colorize_mask(mask): # mask: numpy array of the mask new_mask = Image.fromarray(mask.astype(np.uint8)).convert('P') new_mask.putpalette(palette) return new_mask class VOCDataSet(data.Dataset): def __init__(self, root, list_path, max_iters=None, crop_size=(321, 321), mean=(128, 128, 128), scale=True, mirror=True, ignore_label=255): self.root = root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.scale = scale self.ignore_label = ignore_label self.mean = mean self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [i_id.strip() for i_id in open(list_path)] if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for name in self.img_ids: img_file = osp.join(self.root, "img/%s.jpg" % name) label_file = osp.join(self.root, "gt/%s.png" % name) self.files.append({ "img": img_file, "label": label_file, "name": name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): f_scale = 0.5 + random.randint(0, 11) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) label = cv2.imread(datafiles["label"], cv2.IMREAD_GRAYSCALE) size = image.shape name = datafiles["name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image -= self.mean img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) #image = image[:, :, ::-1] # change to BGR image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), name class GTA5DataSet(data.Dataset): def __init__(self, root, list_path, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') from labels import id2label, trainId2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name class SYNTHIADataSet(data.Dataset): def __init__(self, root, list_path, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') from labels_synthia import id2label, trainId2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name class SYNTHIASTDataSet(data.Dataset): def __init__(self, root, list_path, reg_weight = 0.0, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.reg_weight = reg_weight with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') from labels_synthia import id2label, trainId2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name, self.reg_weight class GTA5STDataSet(data.Dataset): def __init__(self, root, list_path, reg_weight = 0.0, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.reg_weight = reg_weight with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') from labels import id2label, trainId2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name, self.reg_weight class SrcSTDataSet(data.Dataset): def __init__(self, root, list_path, data_src=None, reg_weight = 0.0, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.reg_weight = reg_weight self.data_src = data_src with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') if self.data_src == 'gta': from labels import id2label elif self.data_src == 'synthia': from labels_synthia import id2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name, self.reg_weight class SoftSrcSTDataSet(data.Dataset): def __init__(self, root, list_path, data_src = None, num_classes = None, reg_weight = 0.0, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.reg_weight = reg_weight self.data_src = data_src self.num_classes = num_classes with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') if self.data_src == 'gta': from labels import id2label elif self.data_src == 'synthia': from labels_synthia import id2label # label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # valid_labels = sorted(set(id_2_label.ravel())) label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) # roi = cv2.Rect(w_off, h_off, self.crop_w, self.crop_h); image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] label_expand = np.tile( np.expand_dims(label, axis=2), (1, 1, self.num_classes) ) labels = label_expand.copy() labels[labels != self.ignore_label] = 1.0 labels[labels == self.ignore_label] = 0.0 labels = np.cumsum(labels, axis=2) labels[labels != label_expand + 1] = 0.0 del label_expand labels[labels != 0.0] = 1.0 labels = labels.transpose((2,0,1)) # weighted_pred_trainIDs = np.asarray(np.argmax(labels, axis=0), dtype=np.uint8) # # save weighted predication # wpred_label_col = weighted_pred_trainIDs.copy() # wpred_label_col = colorize_mask(wpred_label_col) # wpred_label_col.save('%s_color.png' % (index)) # # labels_sum = np.sum(labels,axis=0) # # save weighted predication # weighted_pred_trainIDs[labels_sum == 0] = 255 # wpred_label_col = weighted_pred_trainIDs.copy() # wpred_label_col = colorize_mask(wpred_label_col) # wpred_label_col.save('%s_pseudo_color.png' % (index)) return image.copy(), labels.copy(), np.array(size), img_name, self.reg_weight class SoftGTA5StMineDataSet(data.Dataset): def __init__(self, root, list_path, data_src=None, num_classes = None, reg_weight = 0.0, rare_id = None, mine_id = None, mine_chance = None, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.softlabel_ids = [] self.reg_weight = reg_weight self.rare_id = rare_id self.mine_id = mine_id self.mine_chance = mine_chance self.data_src = data_src self.num_classes = num_classes with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) self.softlabel_ids.append(fields[2]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.softlabel_ids = self.softlabel_ids * int(np.ceil(float(max_iters) / len(self.softlabel_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] softlabel_name = self.softlabel_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) softlabel_file = osp.join(self.root, softlabel_name) else: label_file = label_name softlabel_file = softlabel_name self.files.append({ "img": img_file, "label": label_file, "softlabel": softlabel_file, "img_name": img_name, "label_name": label_name, "softlabel_name": softlabel_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label, input_softlabel): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) # interpolate the softlabel by 3-channel groups h,w = label.shape num_group = int(np.ceil(self.num_classes/3.0)) softlabel = np.zeros((h,w,self.num_classes), dtype=np.float32) start_idx = 0 for idx in range(num_group): clabel = input_softlabel[:,:,start_idx:start_idx+3] clabel = cv2.resize(clabel, None, fx=f_scale, fy=f_scale, interpolation=cv2.INTER_LINEAR) softlabel[:,:,start_idx:start_idx+3] = clabel.reshape(h,w,-1) start_idx = start_idx + 3 softlabel = softlabel.transpose(2,0,1)/np.sum(softlabel,2) softlabel = softlabel.transpose(1,2,0) return image, label, softlabel def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) softlabel = np.load(datafiles["softlabel"]) # sys.path.insert(0, 'dataset/helpers') if self.data_src == 'gta': from labels import id2label elif self.data_src == 'synthia': from labels_cityscapes_synthia import id2label label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label, softlabel = self.generate_scale_label(image, label, softlabel) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std # np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) # softlabel_pad h_pad, w_pad = label_pad.shape num_group = int(np.ceil(self.num_classes / 3.0)) softlabel_pad = np.zeros((h_pad, w_pad, self.num_classes), dtype=np.float32) start_idx = 0 for idx in range(num_group): clabel_pad = softlabel[:, :, start_idx:start_idx + 3] clabel_pad = cv2.copyMakeBorder(clabel_pad, 0, pad_h, 0,pad_w, cv2.BORDER_CONSTANT,value=(0.0, 0.0, 0.0)) softlabel_pad[:, :, start_idx:start_idx + 3] = clabel_pad.reshape(h_pad,w_pad,-1) start_idx = start_idx + 3 else: img_pad, label_pad, softlabel_pad = image, label, softlabel img_h, img_w = label_pad.shape # mining or not mine_flag = random.uniform(0, 1) < self.mine_chance if mine_flag and len(self.mine_id) > 0: label_unique = np.unique(label_pad) mine_id_temp = np.array([a for a in self.mine_id if a in label_unique]) # if this image has the mine id if mine_id_temp.size != 0: # decide the single id to be mined mine_id_img = mine_id_temp sel_idx = random.randint(0, mine_id_temp.size-1) sel_mine_id = mine_id_img[sel_idx] # seed for the mined id mine_id_loc = np.where(label_pad == sel_mine_id) # tuple mine_id_len = len(mine_id_loc[0]) seed_loc = random.randint(0, mine_id_len-1) hseed = mine_id_loc[0][seed_loc] wseed = mine_id_loc[1][seed_loc] # patch crop half_crop_h = self.crop_h/2 half_crop_w = self.crop_w/2 # center crop at the seed left_idx = wseed - half_crop_w right_idx = wseed + half_crop_w -1 up_idx = hseed - half_crop_h bottom_idx = hseed + half_crop_h - 1 # shift the left_idx or right_idx if they go beyond the pad margins if left_idx < 0: left_idx = 0 elif right_idx > img_w - 1: left_idx = left_idx - ( ( half_crop_w - 1 ) - (img_w - 1 - wseed) ) # left_idx shifts to the left by the right beyond length if up_idx < 0: up_idx = 0 elif bottom_idx > img_h - 1: up_idx = up_idx - ( ( half_crop_h - 1 ) - (img_h - 1 - hseed) ) # up_idx shifts to the up by the bottom beyond length h_off = up_idx w_off = left_idx else: h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) else: h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) softlabel = np.asarray(softlabel_pad[h_off: h_off + self.crop_h, w_off: w_off + self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) # set ignored label vector to be all zeros label_expand = np.tile( np.expand_dims(label, axis=2), (1, 1, self.num_classes) ) labels_ = label_expand.copy() labels_[labels_ != self.ignore_label] = 1.0 labels_[labels_ == self.ignore_label] = 0.0 labels = labels_*softlabel labels = labels.transpose((2,0,1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] labels = labels[:,:,::flip] # weighted_pred_trainIDs = np.asarray(np.argmax(labels, axis=0), dtype=np.uint8) # # save weighted predication # wpred_label_col = weighted_pred_trainIDs.copy() # wpred_label_col = colorize_mask(wpred_label_col) # wpred_label_col.save('%s_color.png' % (index)) # # labels_sum = np.sum(labels,axis=0) # # save weighted predication # weighted_pred_trainIDs[labels_sum == 0] = 255 # wpred_label_col = weighted_pred_trainIDs.copy() # wpred_label_col = colorize_mask(wpred_label_col) # wpred_label_col.save('%s_pseudo_color.png' % (index)) return image.copy(), labels.copy(), np.array(size), img_name, self.reg_weight class GTA5StMineDataSet(data.Dataset): def __init__(self, root, list_path, data_src=None, reg_weight = 0.0, rare_id = None, mine_id = None, mine_chance = None, pseudo_root = None, max_iters=None, crop_size=(500, 500), train_scale = (0.5, 1.5), mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.pseudo_root = pseudo_root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.crop_h, self.crop_w = crop_size self.lscale, self.hscale = train_scale self.scale = scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] self.reg_weight = reg_weight self.rare_id = rare_id self.mine_id = mine_id self.mine_chance = mine_chance self.data_src = data_src with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) if self.pseudo_root == None: label_file = osp.join(self.root, label_name) else: label_file = label_name self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def generate_scale_label(self, image, label): # f_scale = 0.5 + random.randint(0, 11) / 10.0 f_scale = self.lscale + random.randint(0, int((self.hscale-self.lscale)*10)) / 10.0 image = cv2.resize(image, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_LINEAR) label = cv2.resize(label, None, fx=f_scale, fy=f_scale, interpolation = cv2.INTER_NEAREST) return image, label def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = np.array(Image.open(datafiles["label"])) # sys.path.insert(0, 'dataset/helpers') if self.data_src == 'gta': from labels import id2label elif self.data_src == 'synthia': from labels_cityscapes_synthia import id2label label_2_id = 255 * np.ones((256,)) for l in id2label: if l in (-1, 255): continue label_2_id[l] = id2label[l].trainId label = label_2_id[label] # size = image.shape img_name = datafiles["img_name"] if self.scale: image, label = self.generate_scale_label(image, label) image = np.asarray(image, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std # np.reshape(self.std,(1,1,3)) img_h, img_w = label.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: img_pad = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) label_pad = cv2.copyMakeBorder(label, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(self.ignore_label,)) else: img_pad, label_pad = image, label img_h, img_w = label_pad.shape # mining or not mine_flag = random.uniform(0, 1) < self.mine_chance if mine_flag and len(self.mine_id) > 0: label_unique = np.unique(label_pad) mine_id_temp = np.array([a for a in self.mine_id if a in label_unique]) # if this image has the mine id if mine_id_temp.size != 0: # decide the single id to be mined mine_id_img = mine_id_temp sel_idx = random.randint(0, mine_id_temp.size-1) sel_mine_id = mine_id_img[sel_idx] # seed for the mined id mine_id_loc = np.where(label_pad == sel_mine_id) # tuple mine_id_len = len(mine_id_loc[0]) seed_loc = random.randint(0, mine_id_len-1) hseed = mine_id_loc[0][seed_loc] wseed = mine_id_loc[1][seed_loc] # patch crop half_crop_h = self.crop_h/2 half_crop_w = self.crop_w/2 # center crop at the seed left_idx = wseed - half_crop_w right_idx = wseed + half_crop_w -1 up_idx = hseed - half_crop_h bottom_idx = hseed + half_crop_h - 1 # shift the left_idx or right_idx if they go beyond the pad margins if left_idx < 0: left_idx = 0 elif right_idx > img_w - 1: left_idx = left_idx - ( ( half_crop_w - 1 ) - (img_w - 1 - wseed) ) # left_idx shifts to the left by the right beyond length if up_idx < 0: up_idx = 0 elif bottom_idx > img_h - 1: up_idx = up_idx - ( ( half_crop_h - 1 ) - (img_h - 1 - hseed) ) # up_idx shifts to the up by the bottom beyond length h_off = up_idx w_off = left_idx else: h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) else: h_off = random.randint(0, img_h - self.crop_h) w_off = random.randint(0, img_w - self.crop_w) image = np.asarray(img_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) label = np.asarray(label_pad[h_off : h_off+self.crop_h, w_off : w_off+self.crop_w], np.float32) image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) if self.is_mirror: flip = np.random.choice(2) * 2 - 1 image = image[:, :, ::flip] label = label[:, ::flip] return image.copy(), label.copy(), np.array(size), img_name, self.reg_weight class GTA5TestDataSet(data.Dataset): def __init__(self, root, list_path, max_iters=None, test_size=(1024, 512), test_scale = 1.0, mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.list_path = list_path self.test_h, self.test_w = test_size self.scale = scale self.test_scale = test_scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) label_file = osp.join(self.root, label_name) self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = cv2.imread(datafiles["label"], cv2.IMREAD_GRAYSCALE) # # sys.path.insert(0, 'dataset/helpers') # from labels import id2label, trainId2label # # # label_2_id = 255 * np.ones((256,)) # for l in id2label: # if l in (-1, 255): # continue # label_2_id[l] = id2label[l].trainId # # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # # valid_labels = sorted(set(id_2_label.ravel())) # label = label_2_id[label] # # resize img_name = datafiles["img_name"] # image = cv2.resize(image, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_LINEAR) # label = cv2.resize(label, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_NEAREST) image = cv2.resize(image, None, fx=self.test_scale, fy=self.test_scale, interpolation = cv2.INTER_LINEAR) # always keep the resolution of label unchanged # label = cv2.resize(label, None, fx=1, fy=1, interpolation = cv2.INTER_NEAREST) image = np.asarray(image, np.float32) label = np.asarray(label, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) size = image.shape image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) return image.copy(), label.copy(), np.array(size), img_name class GTA5MSTDataSet(data.Dataset): def __init__(self, root, list_path, max_iters=None, test_size=(1024, 512), test_scale = 1.0, mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.list_path = list_path self.test_h, self.test_w = test_size self.scale = scale self.test_scale = test_scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) label_file = osp.join(self.root, label_name) self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = cv2.imread(datafiles["label"], cv2.IMREAD_GRAYSCALE) # # sys.path.insert(0, 'dataset/helpers') # from labels import id2label, trainId2label # # # label_2_id = 255 * np.ones((256,)) # for l in id2label: # if l in (-1, 255): # continue # label_2_id[l] = id2label[l].trainId # # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # # valid_labels = sorted(set(id_2_label.ravel())) # label = label_2_id[label] # # resize img_name = datafiles["img_name"] # image = cv2.resize(image, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_LINEAR) # label = cv2.resize(label, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_NEAREST) image = cv2.resize(image, None, fx=self.test_scale, fy=self.test_scale, interpolation = cv2.INTER_LINEAR) # always keep the resolution of label unchanged # label = cv2.resize(label, None, fx=1, fy=1, interpolation = cv2.INTER_NEAREST) image = np.asarray(image, np.float32) label = np.asarray(label, np.float32) # image = image/255.0 # scale to [0,1] # image -= self.mean # BGR # image = image/self.std#np.reshape(self.std,(1,1,3)) size = image.shape image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) return image.copy(), label.copy(), np.array(size), img_name class GTA5TestCRFDataSet(data.Dataset): def __init__(self, root, list_path, max_iters=None, test_size=(1024, 512), test_scale = 1.0, mean=(128, 128, 128), std = (1,1,1), scale=True, mirror=True, ignore_label=255): self.root = root self.list_path = list_path self.test_h, self.test_w = test_size self.scale = scale self.test_scale = test_scale self.ignore_label = ignore_label self.mean = mean self.std = std self.is_mirror = mirror # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [] self.label_ids = [] with open(list_path) as f: for item in f.readlines(): fields = item.strip().split('\t') self.img_ids.append(fields[0]) self.label_ids.append(fields[1]) if not max_iters==None: self.img_ids = self.img_ids * int(np.ceil(float(max_iters) / len(self.img_ids))) self.label_ids = self.label_ids * int(np.ceil(float(max_iters) / len(self.label_ids))) self.files = [] # for split in ["train", "trainval", "val"]: for idx in range(len(self.img_ids)): img_name = self.img_ids[idx] label_name = self.label_ids[idx] img_file = osp.join(self.root, img_name) label_file = osp.join(self.root, label_name) self.files.append({ "img": img_file, "label": label_file, "img_name": img_name, "label_name": label_name }) def __len__(self): return len(self.files) def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) # OpenCV read image as BGR, not RGB label = cv2.imread(datafiles["label"], cv2.IMREAD_GRAYSCALE) # # sys.path.insert(0, 'dataset/helpers') # from labels import id2label, trainId2label # # # label_2_id = 255 * np.ones((256,)) # for l in id2label: # if l in (-1, 255): # continue # label_2_id[l] = id2label[l].trainId # # id_2_label = np.array([trainId2label[_].id for _ in trainId2label if _ not in (-1, 255)]) # # valid_labels = sorted(set(id_2_label.ravel())) # label = label_2_id[label] # # resize img_name = datafiles["img_name"] # image = cv2.resize(image, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_LINEAR) # label = cv2.resize(label, (self.test_h, self.test_w), fx=0, fy=0, interpolation = cv2.INTER_NEAREST) image_crf = np.asarray(image, np.float32) image_crf = image_crf[:, :, ::-1] # change to RGB image = cv2.resize(image, None, fx=self.test_scale, fy=self.test_scale, interpolation = cv2.INTER_LINEAR) # always keep the resolution of label unchanged # label = cv2.resize(label, None, fx=1, fy=1, interpolation = cv2.INTER_NEAREST) image = np.asarray(image, np.float32) label = np.asarray(label, np.float32) image = image/255.0 # scale to [0,1] image -= self.mean # BGR image = image/self.std#np.reshape(self.std,(1,1,3)) size = image.shape image = image[:, :, ::-1] # change to RGB image = image.transpose((2, 0, 1)) return image.copy(), label.copy(), image_crf.copy(), np.array(size), img_name class VOCDataTestSet(data.Dataset): def __init__(self, root, list_path, crop_size=(505, 505), mean=(128, 128, 128)): self.root = root self.list_path = list_path self.crop_h, self.crop_w = crop_size self.mean = mean # self.mean_bgr = np.array([104.00698793, 116.66876762, 122.67891434]) self.img_ids = [i_id.strip() for i_id in open(list_path)] self.files = [] # for split in ["train", "trainval", "val"]: for name in self.img_ids: img_file = osp.join(self.root, "img/%s.jpg" % name) self.files.append({ "img": img_file }) def __len__(self): return len(self.files) def __getitem__(self, index): datafiles = self.files[index] image = cv2.imread(datafiles["img"], cv2.IMREAD_COLOR) size = image.shape name = osp.splitext(osp.basename(datafiles["img"]))[0] image = np.asarray(image, np.float32) image -= self.mean img_h, img_w, _ = image.shape pad_h = max(self.crop_h - img_h, 0) pad_w = max(self.crop_w - img_w, 0) if pad_h > 0 or pad_w > 0: image = cv2.copyMakeBorder(image, 0, pad_h, 0, pad_w, cv2.BORDER_CONSTANT, value=(0.0, 0.0, 0.0)) image = image.transpose((2, 0, 1)) return image, name, size if __name__ == '__main__': dst = VOCDataSet("./data", is_transform=True) trainloader = data.DataLoader(dst, batch_size=4) for i, data in enumerate(trainloader): imgs, labels = data if i == 0: img = torchvision.utils.make_grid(imgs).numpy() img = np.transpose(img, (1, 2, 0)) img = img[:, :, ::-1] plt.imshow(img) plt.show()
62,075
43.276748
309
py
CRST
CRST-master/deeplab/__init__.py
1
0
0
py
CRST
CRST-master/deeplab/metric.py
import os, sys import numpy as np from multiprocessing import Pool import copy_reg import types def _pickle_method(m): if m.im_self is None: return getattr, (m.im_class, m.im_func.func_name) else: return getattr, (m.im_self, m.im_func.func_name) copy_reg.pickle(types.MethodType, _pickle_method) class ConfusionMatrix(object): def __init__(self, nclass, classes=None): self.nclass = nclass self.classes = classes self.M = np.zeros((nclass, nclass)) def add(self, gt, pred): assert(np.max(pred) <= self.nclass) assert(len(gt) == len(pred)) for i in range(len(gt)): if not gt[i] == 255: self.M[gt[i], pred[i]] += 1.0 def addM(self, matrix): assert(matrix.shape == self.M.shape) self.M += matrix def __str__(self): pass def recall(self): recall = 0.0 for i in xrange(self.nclass): recall += self.M[i, i] / np.sum(self.M[:, i]) return recall/self.nclass def accuracy(self): accuracy = 0.0 for i in xrange(self.nclass): accuracy += self.M[i, i] / np.sum(self.M[i, :]) return accuracy/self.nclass def jaccard(self): jaccard = 0.0 jaccard_perclass = [] for i in xrange(self.nclass): jaccard_perclass.append(self.M[i, i] / (np.sum(self.M[i, :]) + np.sum(self.M[:, i]) - self.M[i, i])) return np.sum(jaccard_perclass)/len(jaccard_perclass), jaccard_perclass, self.M def generateM(self, item): gt, pred = item m = np.zeros((self.nclass, self.nclass)) assert(len(gt) == len(pred)) for i in range(len(gt)): if gt[i] < self.nclass: #and pred[i] < self.nclass: m[gt[i], pred[i]] += 1.0 return m if __name__ == '__main__': args = parse_args() m_list = [] data_list = [] test_ids = [i.strip() for i in open(args.test_ids) if not i.strip() == ''] for index, img_id in enumerate(test_ids): if index % 100 == 0: print('%d processd'%(index)) pred_img_path = os.path.join(args.pred_dir, img_id+'.png') gt_img_path = os.path.join(args.gt_dir, img_id+'.png') pred = cv2.imread(pred_img_path, cv2.IMREAD_GRAYSCALE) gt = cv2.imread(gt_img_path, cv2.IMREAD_GRAYSCALE) # show_all(gt, pred) data_list.append([gt.flatten(), pred.flatten()]) ConfM = ConfusionMatrix(args.class_num) f = ConfM.generateM pool = Pool() m_list = pool.map(f, data_list) pool.close() pool.join() for m in m_list: ConfM.addM(m) aveJ, j_list, M = ConfM.jaccard() with open(args.save_path, 'w') as f: f.write('meanIOU: ' + str(aveJ) + '\n') f.write(str(j_list)+'\n') f.write(str(M)+'\n')
2,856
27.858586
112
py
quaterny_opvs
quaterny_opvs-master/emulators/emulator_ptb7_th.py
#!/usr/bin/env python #========================================================================== from abstract_emulator import AbstractEmulator #========================================================================== class Emulator_PTB7_TH(AbstractEmulator): PATH = 'details_ptb7_th' HYPERPARAMS = { 'ACT_FUNC': 'leaky_relu', 'ACT_FUNC_OUT': 'relu', 'LEARNING_RATE': 0.001, 'MLP_SIZE': 120, 'REG': 10**-2.5, 'DROP': 0.2} def __init__(self, home = './'): self.home = home self.path = f'{self.home}/{self.PATH}' AbstractEmulator.__init__(self) self._load_config(path = self.path) def load_dataset(self): self._load_indices(path = self.path) self._load_dataset(path = self.path) def run_experiment(self, features, full_stats = False): prediction = self.predict(features) if full_stats: return prediction else: return prediction[self.config['objective']['name']] #=============================================================================== def train(): ''' train the emulator on the dataset ''' emulator = Emulator_PTB7_TH() emulator.load_dataset() emulator.initialize_models(batch_size=len(emulator.train_features[0])) emulator.set_hyperparameters(emulator.HYPERPARAMS) emulator.construct_models() emulator.train(path=emulator.path, plot=True) def predict_test_set(plot=True): ''' runs predictions on the test set ''' emulator = Emulator_PTB7_TH() emulator.load_dataset() emulator.initialize_models(batch_size=len(emulator.train_features[0])) emulator.set_hyperparameters(emulator.HYPERPARAMS) pred_dict = emulator.predict(emulator.test_features, reshaped = True) if plot: from sklearn.metrics import r2_score test_targets = emulator.test_targets pred_targets = pred_dict['averages'] r2 = r2_score(emulator.test_targets, pred_targets) print('R2', r2) import matplotlib.pyplot as plt import seaborn as sns fig = plt.figure(figsize = (4, 4)) plt.plot([0.0, 0.7], [0.0, 0.7], color = 'k') plt.plot(test_targets, pred_targets, ls = '', marker = 'o', color = 'k', markersize = 4) plt.plot(test_targets, pred_targets, ls = '', marker = 'o', color = 'b', markersize = 2) plt.title('PTB7-TH (%.3f)' % r2) plt.tight_layout() plt.show() #=============================================================================== if __name__ == '__main__': predict_test_set()
2,370
27.914634
90
py
quaterny_opvs
quaterny_opvs-master/emulators/model_probabilistic.py
#!/usr/bin/env python import os import pickle import numpy as np import tensorflow as tf import tensorflow_probability as tfp tf_bijs = tfp.bijectors tf_dist = tfp.distributions tf_mean_field = tfp.layers.default_mean_field_normal_fn #============================================================== class RegressionModel(object): NUM_SAMPLES = 1 ACT_FUNC = 'leaky_relu' ACT_FUNC_OUT = 'linear' LEARNING_RATE = 0.75 * 10**-3 MLP_SIZE = 48 REG = 1e-3 DROP = 0.1 def __init__(self, graph, dataset_details, config, scope, batch_size, max_iter = 10**8): self.graph = graph self.scope = scope self.config = config self.batch_size = batch_size self.dataset_details = dataset_details self.max_iter = max_iter self.is_graph_constructed = False self._read_scaling_details() def _generator(self, features, targets, batch_size): indices = np.arange(len(features)) while True: np.random.shuffle(indices) batch_features = features[indices[:batch_size]] batch_targets = targets[indices[:batch_size]] yield (batch_features, batch_targets) def _read_scaling_details(self): with open(self.dataset_details, 'rb') as content: details = pickle.load(content) self.scaling = {key: details[key] for key in details} self.features_shape = self.scaling['features_shape'] self.targets_shape = self.scaling['targets_shape'] def get_scaled_features(self, features): if self.config['feature_rescaling'] == 'standardization': scaled = (features - self.scaling['mean_features']) / self.scaling['std_features'] elif self.config['feature_rescaling'] == 'unit_cube': scaled = (features - self.scaling['min_features']) / (self.scaling['max_features'] - self.scaling['min_features']) return scaled def get_scaled_targets(self, targets): if self.config['target_rescaling'] == 'standardization': scaled = (targets - self.scaling['mean_targets']) / self.scaling['std_targets'] elif self.config['target_rescaling'] == 'unit_cube': scaled = (targets - self.scaling['min_targets']) / (self.scaling['max_targets'] - self.scaling['min_targets']) elif self.config['target_rescaling'] == 'mean': scaled = targets / self.scaling['mean_targets'] elif self.config['target_rescaling'] == 'same': scaled = targets return scaled def get_raw_targets(self, targets): if self.config['target_rescaling'] == 'standardization': raw = targets * self.scaling['std_targets'] + self.scaling['mean_targets'] elif self.config['target_rescaling'] == 'unit_cube': raw = (self.scaling['max_targets'] - self.scaling['min_targets']) * targets + self.scaling['min_targets'] elif self.config['target_rescaling'] == 'mean': raw = targets * self.scaling['mean_targets'] elif self.config['target_rescaling'] == 'same': raw = targets return raw def set_hyperparameters(self, hyperparam_dict): for key, value in hyperparam_dict.items(): setattr(self, key, value) def construct_graph(self): act_funcs = { 'linear': lambda y: y, 'leaky_relu': lambda y: tf.nn.leaky_relu(y, 0.2), 'relu': lambda y: tf.nn.relu(y), 'softmax': lambda y: tf.nn.softmax(y), 'softplus': lambda y: tf.nn.softplus(y), 'softsign': lambda y: tf.nn.softsign(y), 'sigmoid': lambda y: tf.nn.sigmoid(y), } mlp_activation = act_funcs[self.ACT_FUNC] out_activation = act_funcs[self.ACT_FUNC_OUT] with self.graph.as_default(): with tf.name_scope(self.scope): self.is_training = tf.compat.v1.placeholder(tf.bool, shape = ()) self.x_ph = tf.compat.v1.placeholder(tf.float32, [self.batch_size, self.features_shape[1]]) self.y_ph = tf.compat.v1.placeholder(tf.float32, [self.batch_size, self.targets_shape[1]]) self.layer_0 = tfp.layers.DenseLocalReparameterization( self.MLP_SIZE, activation = mlp_activation, ) layer_0_act = self.layer_0(self.x_ph) layer_0_out = tf.layers.dropout(layer_0_act, rate = self.DROP, training = self.is_training) self.layer_1 = tfp.layers.DenseLocalReparameterization( self.MLP_SIZE, activation = mlp_activation, ) layer_1_act = self.layer_1(layer_0_out) layer_1_out = tf.layers.dropout(layer_1_act, rate = self.DROP, training = self.is_training) self.layer_2 = tfp.layers.DenseLocalReparameterization( self.MLP_SIZE, activation = mlp_activation, ) layer_2_act = self.layer_2(layer_1_out) layer_2_out = layer_2_act self.layer_3 = tfp.layers.DenseLocalReparameterization( self.targets_shape[1], activation = out_activation, ) layer_3_out = self.layer_3(layer_2_out) self.net_out = layer_3_out self.scales = tf.nn.softplus(tf.Variable(tf.zeros(1))) self.y_pred = tf_dist.Normal(self.net_out, scale = self.scales) def construct_inference(self): self.is_graph_constructed = True with self.graph.as_default(): self.kl = sum(self.layer_0.losses) / float(self.batch_size) self.kl += sum(self.layer_1.losses) / float(self.batch_size) self.kl += sum(self.layer_2.losses) / float(self.batch_size) self.kl += sum(self.layer_3.losses) / float(self.batch_size) self.reg_loss = - tf.reduce_mean( self.y_pred.log_prob(self.y_ph) ) self.loss = self.reg_loss + self.REG * self.kl self.optimizer = tf.compat.v1.train.AdamOptimizer(self.LEARNING_RATE) self.train_op = self.optimizer.minimize(self.loss) self.init_op = tf.group(tf.compat.v1.global_variables_initializer(), tf.compat.v1.local_variables_initializer()) self.sess = tf.compat.v1.Session(graph = self.graph) with self.sess.as_default(): self.sess.run(self.init_op) def train(self, train_features, train_targets, valid_features, valid_targets, model_path, plot = False, targets = 'same'): from sklearn.metrics import r2_score if not os.path.isdir(model_path): os.mkdir(model_path) logfile = open('%s/logfile.dat' % model_path, 'w') logfile.close() if not self.is_graph_constructed: self.construct_inference() train_feat_scaled = self.get_scaled_features(train_features) train_targ_scaled = self.get_scaled_targets(train_targets) valid_feat_scaled = self.get_scaled_features(valid_features) valid_targ_scaled = self.get_scaled_targets(valid_targets) min_target, max_target = np.minimum(np.amin(train_targets, axis = 0), np.amin(valid_targets, axis = 0)), np.maximum(np.amax(train_targets, axis = 0), np.amax(valid_targets, axis = 0)) if targets == 'probs': min_target = 1. / (1. + np.exp( - min_target)) max_target = 1. / (1. + np.exp( - max_target)) batch_train_gen = self._generator(train_feat_scaled, train_targ_scaled, self.batch_size) batch_valid_gen = self._generator(valid_feat_scaled, valid_targ_scaled, self.batch_size) train_errors, valid_errors = [], [] with self.graph.as_default(): with self.sess.as_default(): self.saver = tf.compat.v1.train.Saver() if plot: import matplotlib.pyplot as plt import seaborn as sns colors = sns.color_palette('RdYlGn', 4) plt.ion() plt.style.use('dark_background') fig = plt.figure(figsize = (14, 5)) ax0 = plt.subplot2grid((1, 3), (0, 0)) ax1 = plt.subplot2grid((1, 3), (0, 1)) ax2 = plt.subplot2grid((1, 3), (0, 2)) for epoch in range(self.max_iter): train_x, train_y = next(batch_train_gen) valid_x, valid_y = next(batch_valid_gen) self.sess.run(self.train_op, feed_dict = {self.x_ph: train_x, self.y_ph: train_y, self.is_training: True}) if epoch % 200 == 0: valid_preds = self.sess.run(self.net_out, feed_dict = {self.x_ph: valid_x, self.is_training: False}) valid_y = self.get_raw_targets(valid_y) valid_preds = self.get_raw_targets(valid_preds) if targets == 'probs': valid_y = 1. / (1. + np.exp( - valid_y)) valid_preds = 1. / (1. + np.exp( - valid_preds)) try: valid_r2 = r2_score(valid_y, valid_preds) except: valid_r2 = np.nan valid_errors.append(valid_r2) _1_, _2_ = self.sess.run([self.reg_loss, self.kl], feed_dict = {self.x_ph: train_x, self.y_ph: train_y, self.is_training: False}) print('...', _1_, _2_) train_preds = self.sess.run(self.net_out, feed_dict = {self.x_ph: train_x, self.is_training: False}) train_y = self.get_raw_targets(train_y) train_preds = self.get_raw_targets(train_preds) try: train_r2 = r2_score(train_y, train_preds) except: train_r2 = np.nan train_errors.append(train_r2) if targets == 'probs': train_y = 1. / (1. + np.exp( - train_y)) train_preds = 1. / (1. + np.exp( - train_preds)) logfile = open('%s/logfile.dat' % model_path, 'a') logfile.write('%d\t%.5f\t%.5f\n' % (epoch, train_r2, valid_r2)) logfile.close() # define break condition --> last improvement happened more than 100 epochs ago max_r2_index = np.argmax(valid_errors) if len(valid_errors) - max_r2_index > 100: break if max_r2_index == len(valid_errors) - 1: self.saver.save(self.sess, '%s/model.ckpt' % model_path) new_line = 'EVALUATION: %d (%d)\t%.5f\t%.5f' % ( len(valid_errors) - max_r2_index, len(valid_errors), train_errors[-1], valid_errors[-1]) print(new_line) if plot: train_preds_scaled = train_preds train_trues_scaled = train_y valid_preds_scaled = valid_preds valid_trues_scaled = valid_y ax0.cla() ax1.cla() ax2.cla() ax0.plot([min_target[0], max_target[0]], [min_target[0], max_target[0]], lw = 3, color = 'w', alpha = 0.5) ax0.plot(train_trues_scaled[:, 0], train_preds_scaled[:, 0], marker = '.', ls = '', color = colors[-1], alpha = 0.5) ax0.plot(valid_trues_scaled[:, 0], valid_preds_scaled[:, 0], marker = '.', ls = '', color = colors[0], alpha = 0.5) if len(min_target) > 1: ax1.plot([min_target[1], max_target[1]], [min_target[1], max_target[1]], lw = 3, color = 'w', alpha = 0.5) ax1.plot(train_trues_scaled[:, 1], train_preds_scaled[:, 1], marker = '.', ls = '', color = colors[-1], alpha = 0.5) ax1.plot(valid_trues_scaled[:, 1], valid_preds_scaled[:, 1], marker = '.', ls = '', color = colors[0], alpha = 0.5) RANGE = 50 ax2.plot(np.arange(len(train_errors[-RANGE:])) + len(train_errors[-RANGE:]), train_errors[-RANGE:], lw = 3, color = colors[-1]) ax2.plot(np.arange(len(valid_errors[-RANGE:])) + len(valid_errors[-RANGE:]), valid_errors[-RANGE:], lw = 3, color = colors[0]) plt.pause(0.05) def restore(self, model_path): if not self.is_graph_constructed: self.construct_inference() self.sess = tf.compat.v1.Session(graph = self.graph) self.saver = tf.compat.v1.train.Saver() try: self.saver.restore(self.sess, model_path) return True except AttributeError: return False def predict(self, input_raw): input_scaled = self.get_scaled_features(input_raw) with self.sess.as_default(): output_scaled = [] for _ in range(self.NUM_SAMPLES): output_scaled.append(self.sess.run(self.net_out, feed_dict = {self.x_ph: input_scaled, self.is_training: False})) output_scaled = np.array(output_scaled) output_raw = self.get_raw_targets(output_scaled) output_raw_mean = np.mean(output_raw, axis = 0) output_raw_std = np.std(output_raw, axis = 0) return {'samples': output_raw, 'averages': output_raw_mean, 'uncertainties': output_raw_std} #==============================================================
11,713
35.154321
185
py
quaterny_opvs
quaterny_opvs-master/emulators/abstract_emulator.py
#!/usr/bin/env python import os, sys os.environ['TF_CPP_MIN_LOG_LEVEL'] = '3' sys.path.append('Emulators') import json import pickle import numpy as np import tensorflow as tf #===================================================================== class AbstractEmulator(object): def __init__(self, num_folds = 5): self.num_folds = num_folds self.dataset = None self.dataset_stats = None self.indices = None self.model_batch_size = None self.file_path = None def set_file_path(self, path, file_name): self.file_name = file_name self.path = path def _load_config(self, path = './'): file_name = '%s/config.json' % path with open(file_name) as content: self.config = json.loads(content.read()) self.config['num_params'] = len(self.config['parameters']) self.hyperparameters = self.config['emulator_parameters'] self.path = path def set_hyperparameters(self, hyperparameters): self.hyperparameters = hyperparameters def _load_indices(self, path = './'): file_name = '%s/indices.pkl' % path try: with open(file_name, 'rb') as content: self.indices = pickle.load(content) except UnicodeDecodeError: with open(file_name, 'rb') as content: self.indices = pickle.load(content, encoding = 'latin1') # self.indices = pickle.load(open(file_name, 'rb')) self.work_indices = self.indices['work_indices'] self.test_indices = self.indices['test_indices'] self.train_indices = [self.indices['cross_validation_sets'][index]['train_indices'] for index in range(self.num_folds)] self.valid_indices = [self.indices['cross_validation_sets'][index]['valid_indices'] for index in range(self.num_folds)] def _store_dataset_stats(self, path = './'): max_features = np.amax(self.features, axis = 0) min_features = np.amin(self.features, axis = 0) max_targets = np.amax(self.targets, axis = 0) min_targets = np.amin(self.targets, axis = 0) mean_features = np.mean(self.features, axis = 0) std_features = np.std(self.features, axis = 0) mean_targets = np.mean(self.targets, axis = 0) std_targets = np.std(self.targets, axis = 0) stats_dict = {'min_features': min_features, 'max_features': max_features, 'min_targets': min_targets, 'max_targets': max_targets, 'mean_features': mean_features, 'std_features': std_features, 'mean_targets': mean_targets, 'std_targets': std_targets, 'features_shape': self.features.shape, 'targets_shape': self.targets.shape} with open('%s/dataset_stats.pkl' % path, 'wb') as content: pickle.dump(stats_dict, content) self.dataset_stats = '%s/dataset_stats.pkl' % path def _load_dataset(self, path = './'): file_name = '%s/dataset.pkl' % path with open(file_name, 'rb') as content: self.dataset = pickle.load(content) values = self.dataset['values'] raw_values = self.dataset['values'] if self.config['general']['domain'] == 'simplex': params = [] for element in self.dataset['params']: vector = element / np.sum(element) params.append(vector[:-1]) params = np.array(params) else: params = self.dataset['params'] raw_params = self.dataset['params'] self.all_features = raw_params self.all_targets = raw_values self.features = params[self.work_indices] self.targets = values[self.work_indices] if len(self.targets.shape) == 1: self.targets = np.reshape(self.targets, (len(self.targets), 1)) self._store_dataset_stats(path = path) self.test_features = params[self.test_indices] self.test_targets = values[self.test_indices] if len(self.targets.shape) == 1: self.test_targets = np.reshape(self.test_targets, (len(self.test_targets), 1)) self.train_features, self.train_targets = [], [] self.valid_features, self.valid_targets = [], [] for index in range(self.num_folds): train_features = params[self.train_indices[index]] valid_features = params[self.valid_indices[index]] train_targets = values[self.train_indices[index]] if len(train_targets.shape) == 1: train_targets = np.reshape(train_targets, (len(train_targets), 1)) valid_targets = values[self.valid_indices[index]] if len(valid_targets.shape) == 1: valid_targets = np.reshape(valid_targets, (len(valid_targets), 1)) self.train_features.append(train_features) self.train_targets.append(train_targets) self.valid_features.append(np.concatenate([valid_features for i in range(len(train_features) // len(valid_features))])) self.valid_targets.append(np.concatenate([valid_targets for i in range(len(train_targets) // len(valid_targets))])) def initialize_models(self, batch_size = 1): if self.config['general']['model'] == 'probabilistic': from model_probabilistic import RegressionModel as Model elif self.config['general']['model'] == 'deterministic': from model_deterministic import RegressionModel as Model else: raise NotImplementedError self.models = [] self.graphs = [tf.Graph() for i in range(self.num_folds)] for fold_index in range(self.num_folds): with self.graphs[fold_index].as_default(): model = Model(self.graphs[fold_index], self.dataset_stats, self.config['general'], scope = 'fold_%d' % fold_index, batch_size = batch_size) self.models.append(model) def construct_models(self, hyperparameters = None): if hyperparameters is None: hyperparameters = self.hyperparameters for model_index, model in enumerate(self.models): with self.graphs[model_index].as_default(): model.set_hyperparameters(hyperparameters) model.construct_graph() def load_models(self, path = None, batch_size = 1): print('... loading models') if path is None: path = self.path if self.dataset_stats is None: self.dataset_stats = '%s/dataset_stats.pkl' % path if self.config['general']['model'] == 'probabilistic': from model_probabilistic import RegressionModel as Model elif self.config['general']['model'] == 'deterministic': from model_deterministic import RegressionModel as Model else: raise NotImplementedError self.models = [] self.graphs = [tf.Graph() for i in range(self.num_folds)] for fold_index in range(self.num_folds): with self.graphs[fold_index].as_default(): model = Model(self.graphs[fold_index], self.dataset_stats, self.config['general'], scope = 'fold_%d' % fold_index, batch_size = batch_size) model.set_hyperparameters(self.hyperparameters) model.construct_graph() if model.restore('%s/Fold_%d/model.ckpt' % (path, fold_index)): self.models.append(model) else: print('could not restore model: ', fold_index) break else: self.model_batch_size = batch_size return True self.model_batch_size = None return False def train(self, path = './', plot = False): for model_index, model in enumerate(self.models): model.train(self.train_features[model_index], self.train_targets[model_index], self.valid_features[model_index], self.valid_targets[model_index], model_path = '%s/Fold_%d' % (path, model_index), plot = plot, targets = self.config['general']['targets']) def predict(self, params, reshaped = False): if len(params.shape) == 1: params = np.reshape(params, (1, len(params))) if self.config['general']['domain'] == 'simplex' and not reshaped: features = [] for element in params: vector = element / np.sum(element) features.append(vector[:-1]) features = np.array(features) else: features = params polar_features = [] for feature_index, feature in enumerate(features): polar_feature = [] for elem_index, element in enumerate(feature): if 'rep' in self.config['parameters'][elem_index]: polar_feature.extend([np.sin(element), np.cos(element)]) else: polar_feature.append(element) polar_features.append(polar_feature) features = np.array(polar_features) if not len(features) == self.model_batch_size: self.load_models(batch_size = len(features)) samples = [] for fold_index in range(self.num_folds): single_pred_dict = self.models[fold_index].predict(features) # samples, averages, uncertainties samples.append(single_pred_dict['samples']) samples = np.array(samples) # if self.config['general']['targets'] == 'simplex': # samples = 1. / (1. + np.exp( - samples) ) # shape of samples: (# models, # draws, # features, # dim features) averages = np.mean(np.mean(samples, axis = 0), axis = 0) uncertainties = np.std(np.mean(samples, axis = 0), axis = 0) pred_dict = {'samples': samples, 'averages': averages, 'uncertainties': uncertainties} pred_dict[self.config['objective']['name']] = np.squeeze(pred_dict['averages']) return pred_dict #===================================================================== if __name__ == '__main__': pass
8,804
34.22
143
py
quaterny_opvs
quaterny_opvs-master/emulators/emulator_pbq_qf.py
#!/usr/bin/env python #=============================================================================== from abstract_emulator import AbstractEmulator #=============================================================================== class Emulator_PBQ_QF(AbstractEmulator): PATH = 'details_pbq_qf' HYPERPARAMS = { 'ACT_FUNC': 'leaky_relu', 'ACT_FUNC_OUT': 'relu', 'LEARNING_RATE': 0.001, 'MLP_SIZE': 120, 'REG': 10**-2.0, 'DROP': 0.2} def __init__(self, home = './'): self.home = home self.path = f'{self.home}/{self.PATH}' AbstractEmulator.__init__(self) self._load_config(path = self.path) def load_dataset(self): self._load_indices(path = self.path) self._load_dataset(path = self.path) def run_experiment(self, features, full_stats = False): prediction = self.predict(features) if full_stats: return prediction else: return prediction[self.config['objective']['name']] #=============================================================================== def train(): ''' train the emulator on the dataset ''' emulator = Emulator_PBQ_QF() emulator.load_dataset() emulator.initialize_models(batch_size=len(emulator.train_features[0])) emulator.set_hyperparameters(emulator.HYPERPARAMS) emulator.construct_models() emulator.train(path=emulator.path, plot=True) def predict_test_set(plot=True): ''' runs predictions on the test set ''' emulator = Emulator_PBQ_QF() emulator.load_dataset() emulator.initialize_models(batch_size=len(emulator.train_features[0])) emulator.set_hyperparameters(emulator.HYPERPARAMS) pred_dict = emulator.predict(emulator.test_features, reshaped = True) if plot: from sklearn.metrics import r2_score test_targets = emulator.test_targets pred_targets = pred_dict['averages'] r2 = r2_score(emulator.test_targets, pred_targets) print('R2', r2) import matplotlib.pyplot as plt import seaborn as sns fig = plt.figure(figsize = (4, 4)) plt.plot([0.0, 0.7], [0.0, 0.7], color = 'k') plt.plot(test_targets, pred_targets, ls = '', marker = 'o', color = 'k', markersize = 4) plt.plot(test_targets, pred_targets, ls = '', marker = 'o', color = 'b', markersize = 2) plt.title('PBQ-QF (%.3f)' % r2) plt.tight_layout() plt.show() #=============================================================================== if __name__ == '__main__': predict_test_set()
2,408
28.378049
90
py
quaterny_opvs
quaterny_opvs-master/emulators/details_ptb7_th/parse_grid.py
#!/usr/bin/env python import pickle import numpy as np import matplotlib.pyplot as plt import seaborn as sns file_name = 'photobleaching_mixture01_grid.csv' features = [] targets = [] with open(file_name, 'r') as content: for line in content: linecontent = line.strip().strip('\xef').strip('\xbb').strip('\xbf').strip('\ufeff').split(',') feature = np.array([float(element) for element in linecontent[:4]]) target = float(linecontent[-1]) if target > 0.: features.append(feature) targets.append(target) features = np.array(features) targets = np.array(targets) d_features = [] for index in range(len(features)): for jndex in range(index + 1, len(features)): d_features.append(np.linalg.norm(features[index] - features[jndex])) d_targets = [] for index in range(len(targets)): for jndex in range(index + 1, len(targets)): d_targets.append(np.abs(targets[index] - targets[jndex])) plt.scatter(d_features, d_targets) plt.show() print(features.shape) data = {'params': np.array(features), 'values': np.array(targets)} with open('dataset.pkl', 'wb') as content: pickle.dump(data, content)
1,125
23.478261
97
py
quaterny_opvs
quaterny_opvs-master/emulators/details_ptb7_th/generate_indices.py
#!/usr/bin/env python import numpy as np import pickle #========================================================= NUM_SPECTRA = 1040 FOLD_SIZE = 170 np.random.seed(120415) #========================================================= indices = np.arange(NUM_SPECTRA) np.random.shuffle(indices) test_indices = indices[850:] work_indices = indices[:850] #========================================================= rotation_indices = work_indices.copy() cross_validation_sets = [] for index in range(len(work_indices) // FOLD_SIZE): valid_indices = rotation_indices[:FOLD_SIZE] train_indices = rotation_indices[FOLD_SIZE:] cross_validation_dict = {'train_indices': train_indices.copy(), 'valid_indices': valid_indices.copy()} cross_validation_sets.append(cross_validation_dict) rotation_indices = np.roll(rotation_indices, FOLD_SIZE) print('generated %d folds' % (len(cross_validation_sets))) data_set = {'test_indices': test_indices, 'work_indices': work_indices, 'cross_validation_sets': cross_validation_sets} pickle.dump(data_set, open('indices.pkl', 'wb'))
1,082
25.414634
119
py
quaterny_opvs
quaterny_opvs-master/emulators/details_pbq_qf/parse_grid.py
#!/usr/bin/env python import pickle import numpy as np import matplotlib.pyplot as plt import seaborn as sns file_name = 'photobleaching_mixture00_grid.csv' features = [] targets = [] with open(file_name, 'r') as content: for line in content: linecontent = line.strip().strip('\xef').strip('\xbb').strip('\xbf').strip('\ufeff').split(',') print(linecontent) feature = np.array([float(element) for element in linecontent[:4]]) target = float(linecontent[-1]) if target > 0.: features.append(feature) targets.append(target) features = np.array(features) targets = np.array(targets) d_features = [] for index in range(len(features)): for jndex in range(index + 1, len(features)): d_features.append(np.linalg.norm(features[index] - features[jndex])) d_targets = [] for index in range(len(targets)): for jndex in range(index + 1, len(targets)): d_targets.append(np.abs(targets[index] - targets[jndex])) plt.scatter(d_features, d_targets) plt.show() print(features.shape) data = {'params': np.array(features), 'values': np.array(targets)} with open('dataset.pkl', 'wb') as content: pickle.dump(data, content)
1,146
23.404255
97
py
quaterny_opvs
quaterny_opvs-master/emulators/details_pbq_qf/generate_indices.py
#!/usr/bin/env python import numpy as np import pickle #========================================================= NUM_SPECTRA = 1040 FOLD_SIZE = 170 np.random.seed(120415) #========================================================= indices = np.arange(NUM_SPECTRA) np.random.shuffle(indices) test_indices = indices[850:] work_indices = indices[:850] #========================================================= rotation_indices = work_indices.copy() cross_validation_sets = [] for index in range(len(work_indices) // FOLD_SIZE): valid_indices = rotation_indices[:FOLD_SIZE] train_indices = rotation_indices[FOLD_SIZE:] cross_validation_dict = {'train_indices': train_indices.copy(), 'valid_indices': valid_indices.copy()} cross_validation_sets.append(cross_validation_dict) rotation_indices = np.roll(rotation_indices, FOLD_SIZE) print('generated %d folds' % (len(cross_validation_sets))) data_set = {'test_indices': test_indices, 'work_indices': work_indices, 'cross_validation_sets': cross_validation_sets} pickle.dump(data_set, open('indices.pkl', 'wb'))
1,082
25.414634
119
py
imf
imf-master/setup.py
#!/usr/bin/env python # Licensed under a 3-clause BSD style license - see LICENSE.rst # NOTE: The configuration for the package, including the name, version, and # other information are set in the setup.cfg file. import os import sys from setuptools import setup # First provide helpful messages if contributors try and run legacy commands # for tests or docs. TEST_HELP = """ Note: running tests is no longer done using 'python setup.py test'. Instead you will need to run: tox -e test If you don't already have tox installed, you can install it with: pip install tox If you only want to run part of the test suite, you can also use pytest directly with:: pip install -e .[test] pytest For more information, see: http://docs.astropy.org/en/latest/development/testguide.html#running-tests """ if 'test' in sys.argv: print(TEST_HELP) sys.exit(1) DOCS_HELP = """ Note: building the documentation is no longer done using 'python setup.py build_docs'. Instead you will need to run: tox -e build_docs If you don't already have tox installed, you can install it with: pip install tox You can also build the documentation with Sphinx directly using:: pip install -e .[docs] cd docs make html For more information, see: http://docs.astropy.org/en/latest/install.html#builddocs """ if 'build_docs' in sys.argv or 'build_sphinx' in sys.argv: print(DOCS_HELP) sys.exit(1) VERSION_TEMPLATE = """ # Note that we need to fall back to the hard-coded version if either # setuptools_scm can't be imported or setuptools_scm can't determine the # version, so we catch the generic 'Exception'. try: from setuptools_scm import get_version version = get_version(root='..', relative_to=__file__) except Exception: version = '{version}' """.lstrip() setup(use_scm_version={'write_to': os.path.join('imf', 'version.py'), 'write_to_template': VERSION_TEMPLATE})
1,952
23.721519
76
py
imf
imf-master/examples/imf_schematic.py
""" I'm not entirely sure what this figure is supposed to show; it just plots a few variants of the lognormal IMF """ from imf import imf import numpy as np if __name__ == "__main__": import pylab as pl pl.matplotlib.rc_file("/Users/adam/.matplotlib/pubfiguresrc") pl.rc('font', family='cmr10') x = np.logspace(-2,2) pl.clf() chabrier = imf.chabrier(x) chabrier_n = chabrier/chabrier.sum() chabrier3x = imf.ChabrierLogNormal(lognormal_center=0.66)(x) chabrier3x_n = chabrier3x/chabrier3x.sum() seed = imf.ChabrierLogNormal(lognormal_width=0.3)(x) seed_n = seed/seed.sum() wider = imf.ChabrierLogNormal(lognormal_width=1)(x) wider_n = wider/wider.sum() pl.loglog(x, chabrier_n, linewidth=3, alpha=0.8, label='Chabrier') #pl.loglog(x, imf.kroupa(x), linewidth=3, alpha=0.8, label='Kroupa') pl.loglog(x, chabrier3x_n, linestyle='dashed', linewidth=3, alpha=0.8, label='$3\\times$ Chabrier') pl.loglog(x, wider_n, linestyle='dotted', linewidth=3, alpha=0.8, label='Wider MF') pl.loglog(x, seed_n, linestyle='-.', linewidth=3, alpha=0.8, label='Seed MF') pl.xlim(1e-1, 1e2) pl.ylim(1e-6, 1.15) pl.xlabel("Stellar Mass $M_*$ ($M_{\odot}$)") pl.ylabel("$P(M_*)$") pl.legend(loc='best', fontsize=22) import matplotlib.ticker ax = pl.gca() ax.xaxis.set_major_formatter(matplotlib.ticker.ScalarFormatter()) ax.yaxis.set_major_formatter(matplotlib.ticker.ScalarFormatter()) pl.savefig("imf_schematic.png") pl.draw() pl.show()
1,545
32.608696
110
py
imf
imf-master/examples/clustermf_figure.py
from imf import color_of_cluster,make_cluster,lum_of_cluster import numpy as np if __name__ == "__main__": import pylab as pl alpha = 2 m0 = 5e2 mmax = 5e5 cluster_mass_xax = np.logspace(np.log10(m0),np.log10(mmax),1e4) def pr(m): return (m/m0)**-alpha probabilities = pr(cluster_mass_xax) cdf = probabilities.cumsum() cdf /= cdf.max() # normalize to sum (cdf) nclusters = 5000 cluster_masses = np.array([np.interp(p, cdf, cluster_mass_xax) for p in np.random.rand(nclusters)]) clusters = [make_cluster(m,mmax=m) for m in cluster_masses] luminosities = np.array([lum_of_cluster(c) for c in clusters]) # no contrast # colors = [color_of_cluster(c) for c in clusters] def ctable(mass, mmin=0.08, mmax=120): return pl.cm.RdBu((mass-mmin)/(mmax-mmin)) cr = np.log10(mmax)-np.log10(mmin) lm = np.log10(mass)-np.log10(mmin) return pl.cm.RdBu(lm/cr) colors = [color_of_cluster(c,ctable) for c in clusters] yax = [np.random.rand()*(np.log10(pr(m))-np.log10(pr(mmax))) + np.log10(pr(mmax)) for m in cluster_masses] pl.rc('font',size=30) pl.figure(1) pl.clf() pl.gca().set_xscale('log') sizes = 10**luminosities/1e5 sizes[sizes < 10] = 10 S = pl.scatter(cluster_masses, yax, c=colors, s=sizes, alpha=0.8) sm = pl.cm.ScalarMappable(cmap=pl.cm.RdBu, norm=pl.Normalize(vmin=0.08, vmax=120)) sm._A = [] cb = pl.colorbar(sm) cb.set_label("Luminosity-weighted\nMean Stellar Mass") pl.gca().axis([min(cluster_masses)/1.1,max(cluster_masses)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.xlabel("Cluster Mass ($M_\odot$)") pl.ylabel("Log(dN(M)/dM)") pl.savefig("plots/clusterMF_lumcolor_lumsize.png",bbox_inches='tight') pl.figure(2) pl.clf() pl.gca().set_xscale('log') sizes = cluster_masses / 50 sizes[sizes < 10] = 10 S = pl.scatter(cluster_masses, yax, c=colors, s=sizes, alpha=0.8) sm = pl.cm.ScalarMappable(cmap=pl.cm.RdBu, norm=pl.Normalize(vmin=0.08, vmax=120)) sm._A = [] cb = pl.colorbar(sm) cb.set_label("Luminosity-weighted\nMean Stellar Mass") pl.gca().axis([min(cluster_masses)/1.1,max(cluster_masses)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.xlabel("Cluster Mass ($M_\odot$)") pl.ylabel("Log(dN(M)/dM)") pl.savefig("plots/clusterMF_lumcolor_massize.png",bbox_inches='tight') pl.figure(3) pl.clf() pl.gca().set_xscale('log') sizes = 20*np.log(10**luminosities / cluster_masses) sizes[sizes < 10] = 10 S = pl.scatter(cluster_masses, yax, c=colors, s=sizes, alpha=0.8) sm = pl.cm.ScalarMappable(cmap=pl.cm.RdBu, norm=pl.Normalize(vmin=0.08, vmax=120)) sm._A = [] cb = pl.colorbar(sm) cb.set_label("Luminosity-weighted\nMean Stellar Mass") pl.gca().axis([min(cluster_masses)/1.1,max(cluster_masses)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.xlabel("Cluster Mass ($M_\odot$)") pl.ylabel("Log(dN(M)/dM)") pl.savefig("plots/clusterMF_lumcolor_mtolsize.png",bbox_inches='tight') pl.show()
3,057
34.55814
110
py
imf
imf-master/examples/pmf_evolution.py
import imf.imf, imf.pmf, imp from imf.pmf import ChabrierPMF_AcceleratingSF_IS, ChabrierPMF_AcceleratingSF_TC, ChabrierPMF_AcceleratingSF_CA#, ChabrierPMF_AcceleratingSF_2CTC import pylab as pl import numpy as np imp.reload(imf.imf) imp.reload(imf.pmf) imp.reload(imf.imf) imp.reload(imf.pmf) mmin = 0.033 for mmax in (3, 120): print("Normalizing.") ChabrierPMF_AcceleratingSF_IS.mmax = mmax ChabrierPMF_AcceleratingSF_TC.mmax = mmax ChabrierPMF_AcceleratingSF_CA.mmax = mmax #ChabrierPMF_AcceleratingSF_2CTC.mmax = mmax ChabrierPMF_AcceleratingSF_IS.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_AcceleratingSF_TC.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_AcceleratingSF_CA.normalize(log=True, mmin=mmin, mmax=mmax) #ChabrierPMF_AcceleratingSF_2CTC.normalize(log=True, mmin=mmin, mmax=mmax) chabrierpowerlaw = imf.ChabrierPowerLaw() chabrierpowerlaw.normalize(log=True, mmin=mmin, mmax=mmax) print("Now plotting.") masses = np.logspace(np.log10(mmin), np.log10(mmax), 100) for mass_weighted in (True,False): fname = 'mass_weighted' if mass_weighted else '__call__' fig1 = pl.figure(1) fig1.clf() ax = fig1.gca() ax.set_title("Accelerating SF McKee/Offner + Chabrier PMF") ax.loglog(masses, chabrierpowerlaw.__getattribute__(fname)(masses), label="IMF", color='k') for tau, lw in zip((0.1, 1.0, 10.0), (1,2,3,)): ax.loglog(masses, ChabrierPMF_AcceleratingSF_IS.__getattribute__(fname)(masses, tau=tau), label="IS", color='r', linewidth=lw, linestyle=':') ax.loglog(masses, ChabrierPMF_AcceleratingSF_TC.__getattribute__(fname)(masses, tau=tau), label="TC", color='g', linewidth=lw, linestyle='-.') ax.loglog(masses, ChabrierPMF_AcceleratingSF_CA.__getattribute__(fname)(masses, tau=tau), label="CA", color='y', linewidth=lw, linestyle='-.') #ax.loglog(masses, ChabrierPMF_AcceleratingSF_2CTC.__getattribute__(fname)(masses), label="2CTC", color='b', linestyle='--') ax.set_xlabel("(Proto)Stellar Mass (M$_\odot$)") ax.set_ylabel("m P(m)" if mass_weighted else "Normalized P(M)") ax.axis([mmin, mmax, 1e-4, 1]) pl.legend(loc='best') pl.savefig('acceleratingSF_pmf_chabrier{0}_mmax{1}.png' .format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight')
2,430
45.75
154
py
imf
imf-master/examples/mass_to_light.py
import numpy as np import os import json from imf import imf import pylab as pl import matplotlib from astropy.utils.console import ProgressBar pl.rc('font', size=16) if os.path.exists('synth_data_m_to_l.json'): with open('synth_data_m_to_l.json', 'r') as fh: synth_data = json.load(fh) else: synth_data = {} # uniform random sampling from 100 to 10^5 msun for stop_crit in ('nearest', 'before', 'after', 'sorted'): print(stop_crit) if stop_crit not in synth_data: clusters, luminosities, masses, mean_luminosities, mean_masses, max_masses, number = {},{},{},{},{},{},{} for clmass in ProgressBar(np.concatenate([10**(np.random.rand(int(1e3))*1 + 4), 10**(np.random.rand(int(1e4))*2.5+1.5)])): key = str(clmass) # for jsonification clusters[key] = imf.make_cluster(clmass, 'kroupa', mmax=150, silent=True, stop_criterion=stop_crit) # cluster luminosities luminosities[key] = imf.lum_of_cluster(clusters[key]) masses[key] = clmass number[key] = len(clusters[key]) #mean_luminosities[clmass] = np.mean(luminosities[clmass]) mean_masses[key] = np.mean(clusters[key]) max_masses[key] = np.max(clusters[key]) synth_data[stop_crit] = {#'clusters': clusters, 'number': number, 'luminosities': luminosities, 'masses': masses, 'mean_luminosities': mean_luminosities, 'mean_masses': mean_masses, 'max_masses': max_masses} else: max_masses = synth_data[stop_crit]['max_masses'] #mean_luminosities = synth_data[stop_crit]['mean_luminosities'] mean_masses = synth_data[stop_crit]['mean_masses'] number = synth_data[stop_crit]['number'] masses = synth_data[stop_crit]['masses'] luminosities = synth_data[stop_crit]['luminosities'] clmasses = sorted(map(float, masses)) mass_to_light = np.array([k/10**luminosities[str(k)] for k in clmasses]) # shouldn't this converge to a single value? (yes, don't divide linear by log) pl.figure(2).clf() pl.loglog(clmasses, mass_to_light**-1, '.', alpha=0.1) pl.xlabel("Cluster Mass") pl.ylabel("Light to Mass $L_\odot / M_\odot$") pl.ylim(1,5e4) pl.savefig(f"light_to_mass_vs_mass_{stop_crit}.png", bbox_inches='tight', dpi=200) pl.savefig(f"light_to_mass_vs_mass_{stop_crit}.pdf", bbox_inches='tight') pl.figure(3).clf() pl.loglog(list(map(float, max_masses.keys())), max_masses.values(), '.', alpha=0.1) pl.xlabel("Cluster Mass") pl.ylabel("Maximum stellar mass") pl.savefig(f"maxmass_vs_clustermass_{stop_crit}.png", bbox_inches='tight', dpi=200) pl.savefig(f"maxmass_vs_clustermass_{stop_crit}.pdf", bbox_inches='tight') m_to_ls = [] slopes = np.linspace(1.7, 2.9, 20) for slope in ProgressBar(slopes): Kroupa = imf.Kroupa(p3=slope) cluster = imf.make_cluster(1e5, Kroupa, mmax=120, silent=True, stop_criterion=stop_crit) lum = imf.lum_of_cluster(cluster) m_to_l = 1e5/10**lum m_to_ls.append(m_to_l) pl.figure(4).clf() pl.plot(slopes, m_to_ls) pl.xlabel("Upper-end power-law slope $\\alpha$") pl.ylabel("Mass-to-light ratio [M$_\odot$/L$_\odot$]") ax = pl.gca() tw = ax.twinx() tw.set_ylim(m_to_ls[0]/m_to_ls[10], m_to_ls[-1]/m_to_ls[10]) tw.set_ylabel("M/L / M/L($\\alpha=2.3$)") pl.savefig(f"masstolight_vs_slope_{stop_crit}.pdf", bbox_inches='tight') pl.savefig(f"masstolight_vs_slope_{stop_crit}.png", bbox_inches='tight', dpi=200) pl.figure(5).clf() pl.semilogy(slopes, m_to_ls) ax = pl.gca() ylim = ax.get_ylim() ax.vlines(2.3, 1e-5, 1, linestyle='--', color='k', alpha=0.2, zorder=-10) ax.set_ylim(ylim) pl.xlabel("Upper-end power-law slope $\\alpha$") pl.ylabel("Mass-to-light ratio [M$_\odot$/L$_\odot$]") tw = ax.twinx() tw.semilogy() tw.get_yaxis().set_major_formatter(matplotlib.ticker.ScalarFormatter()) tw.set_yticks([0.5,1,2,5]) tw.set_ylim(m_to_ls[0]/m_to_ls[10], m_to_ls[-1]/m_to_ls[10]) tw.set_xlim(ax.get_xlim()) tw.hlines(1, 1, 5, linestyle='--', color='k', alpha=0.2, zorder=-20) #tw.grid(which='major', linestyle='--') tw.set_ylabel("M/L / M/L($\\alpha=2.3$)") pl.savefig(f"masstolight_vs_slope_log_{stop_crit}.pdf", bbox_inches='tight') pl.savefig(f"masstolight_vs_slope_log_{stop_crit}.png", bbox_inches='tight', dpi=200) #pl.loglog(clusters.keys(), np.array(list(mean_luminosities.values())) / np.array(list(mean_masses.values())), '.', alpha=0.1) # compute mass-to-light ratio vs age with open('synth_data_m_to_l.json', 'w') as fh: json.dump(synth_data, fh)
4,914
40.302521
130
py
imf
imf-master/examples/sampling_methods.py
from imf import make_cluster import pylab as pl import imf maxmass = [imf.make_cluster(500, verbose=False, silent=True).max() for ii in range(10000)] maxmass_sorted = [imf.make_cluster(500, stop_criterion='sorted', verbose=False, silent=True).max() for ii in range(10000)] maxmass_before = [imf.make_cluster(500, stop_criterion='before', verbose=False, silent=True).max() for ii in range(10000)] maxmass_after = [imf.make_cluster(500, stop_criterion='after', verbose=False, silent=True).max() for ii in range(10000)] pl.clf() pl.hist(maxmass, bins=50, alpha=0.5, label='nearest', histtype='step') pl.hist(maxmass_sorted, bins=50, alpha=0.5, label='sorted', histtype='step') pl.hist(maxmass_before, bins=50, alpha=0.5, label='before', histtype='step') pl.hist(maxmass_after, bins=50, alpha=0.5, label='after', histtype='step') pl.legend(loc='best')
967
45.095238
79
py
imf
imf-master/examples/pmf_stats.py
import imf.imf, imf.pmf, imp imp.reload(imf.imf) imp.reload(imf.pmf) imp.reload(imf.imf) imp.reload(imf.pmf) from imf.pmf import ChabrierPMF_IS, ChabrierPMF_TC, ChabrierPMF_CA, ChabrierPMF_2CTC from imf.pmf import KroupaPMF_IS, KroupaPMF_TC, KroupaPMF_CA, KroupaPMF_2CTC from imf.pmf import McKeeOffner_AcceleratingSF_PMF, ChabrierPMF_AcceleratingSF_IS, ChabrierPMF_AcceleratingSF_TC, ChabrierPMF_AcceleratingSF_CA#, ChabrierPMF_AcceleratingSF_2CTC import pylab as pl import numpy as np mmin = 0.033 mmax = 120 print("Normalizing.") ChabrierPMF_IS.mmax = mmax ChabrierPMF_TC.mmax = mmax ChabrierPMF_CA.mmax = mmax ChabrierPMF_2CTC.mmax = mmax ChabrierPMF_IS.normalize(log=False, mmin=mmin, mmax=mmax) ChabrierPMF_TC.normalize(log=False, mmin=mmin, mmax=mmax) ChabrierPMF_CA.normalize(log=False, mmin=mmin, mmax=mmax) ChabrierPMF_2CTC.normalize(log=False, mmin=mmin, mmax=mmax) ChabrierPMF_AcceleratingSF_IS.mmax = mmax ChabrierPMF_AcceleratingSF_TC.mmax = mmax ChabrierPMF_AcceleratingSF_CA.mmax = mmax #ChabrierPMF_AcceleratingSF_2CTC.mmax = mmax ChabrierPMF_AcceleratingSF_IS.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_AcceleratingSF_TC.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_AcceleratingSF_CA.normalize(log=True, mmin=mmin, mmax=mmax) #ChabrierPMF_AcceleratingSF_2CTC.normalize(log=True, mmin=mmin, mmax=mmax) KroupaPMF_IS.mmax = mmax KroupaPMF_TC.mmax = mmax KroupaPMF_CA.mmax = mmax KroupaPMF_2CTC.mmax = mmax KroupaPMF_IS.normalize(log=False, mmin=mmin, mmax=mmax) KroupaPMF_TC.normalize(log=False, mmin=mmin, mmax=mmax) KroupaPMF_CA.normalize(log=False, mmin=mmin, mmax=mmax) KroupaPMF_2CTC.normalize(log=False, mmin=mmin, mmax=mmax) chabrierpowerlaw = imf.ChabrierPowerLaw() chabrierpowerlaw.normalize(log=False, mmin=mmin, mmax=mmax) kroupa = imf.Kroupa() kroupa.normalize(log=False, mmin=mmin, mmax=mmax) print("Done normalizing") mfs = {'ChabrierPMF_IS': ChabrierPMF_IS, 'ChabrierPMF_TC': ChabrierPMF_TC, 'ChabrierPMF_CA': ChabrierPMF_CA, 'ChabrierPMF_2CTC': ChabrierPMF_2CTC, 'ChabrierIMF': chabrierpowerlaw, 'KroupaPMF_IS': KroupaPMF_IS, 'KroupaPMF_TC': KroupaPMF_TC, 'KroupaPMF_CA': KroupaPMF_CA, 'KroupaPMF_2CTC': KroupaPMF_2CTC, 'KroupaIMF': kroupa, } for tau in (0.1, 0.5, 1.0, 2.0): mfs['ChabrierPMF_AcceleratingSF_IS_tau{0}'.format(tau)] = McKeeOffner_AcceleratingSF_PMF(j=0, jf=0, tau=tau, mmax=mmax) mfs['ChabrierPMF_AcceleratingSF_TC_tau{0}'.format(tau)] = McKeeOffner_AcceleratingSF_PMF(j=0.5, jf=0.75, tau=tau, mmax=mmax) mfs['ChabrierPMF_AcceleratingSF_CA_tau{0}'.format(tau)] = McKeeOffner_AcceleratingSF_PMF(j=2/3., jf=1.0, tau=tau, mmax=mmax) for mf in sorted(mfs): total = mfs[mf].m_integrate(mmin, mmax)[0] gt10 = mfs[mf].m_integrate(10, mmax)[0] print("Mass fraction for {1} M>10 = {0:0.3f}".format(gt10/total, mf))
2,880
36.907895
177
py
imf
imf-master/examples/chabrier_comparisons.py
""" Compare the Chabrier distribution pulled from eqn 18 of Chabrier 2003 to that quoted in McKee & Offner 2010 as "Chabrier 2005" """ import imf import numpy as np import pylab as pl chabrier = imf.chabrierpowerlaw chabrier2005 = imf.ChabrierPowerLaw(lognormal_width=0.55*np.log(10), lognormal_center=0.2, alpha=2.35) masses = np.geomspace(0.01, 10, 1000) pl.figure(1) pl.loglog(masses, chabrier(masses), label='Chabrier 2003 eqn 18') pl.loglog(masses, chabrier2005(masses), label='Chabrier 2005 via McKee & Offner 2010') pl.xlabel("Mass") pl.ylabel("$\\xi \\equiv dN/dM$") pl.legend(loc='best') pl.savefig("Chabrier2003v2005.png") pl.figure(2) pl.loglog(masses, chabrier2005(masses) - chabrier(masses), label='C03-C05') pl.loglog(masses, chabrier(masses) - chabrier2005(masses), label='C05-C03') pl.xlabel("Mass") pl.ylabel("$\\xi_1 - \\xi_2$") pl.legend(loc='best') pl.savefig("Chabrier2003v2005_diff.png") pl.figure(3) pl.loglog(masses, (chabrier2005(masses) - chabrier(masses))/chabrier(masses), label='(C03-C05)/C03') pl.loglog(masses, (chabrier(masses) - chabrier2005(masses))/chabrier(masses), label='(C05-C03)/C03') pl.xlabel("Mass") pl.ylabel("$(\\xi_1-\\xi_2)/\\xi_1$") pl.legend(loc='best') pl.savefig("Chabrier2003v2005_relativediff.png")
1,328
33.973684
100
py
imf
imf-master/examples/imf_figure.py
""" Script to make an IMF diagram that shows dN(M)/dM vs M, then populates the area under the curve with an appropriate number of stars colored by their "true"(ish) color and sized by their mass. """ import imf from imf import coolplot,kroupa,make_cluster from astropy.table import Table import numpy as np if __name__ == "__main__": import pylab as pl pl.matplotlib.style.use('classic') pl.rc('font',size=30) pl.close(1) # make three figures of dN/dM vs M, one for each mass function, # then do it again in log-scale for massfunc in (imf.kroupa, imf.chabrier2005, imf.salpeter): # this is not a recommended way to get object names, don't do it in general. # (not all classes are guaranteed to have names; I know they do in this # case because I made and initialized the classes) name = massfunc.__class__.__name__ pl.figure(1, figsize=(10,8)) pl.clf() cluster,yax,colors = coolplot(1000, massfunc=massfunc) cluster = np.array(cluster) yax = np.array(yax) pl.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) pl.gca().set_xscale('log') masses = np.logspace(np.log10(cluster.min()), np.log10(cluster.max()),10000) pl.plot(masses,np.log10(massfunc(masses)),'r--',linewidth=2,alpha=0.5) pl.xlabel("Stellar Mass") pl.ylabel("log(dN(M)/dM)") pl.gca().axis([min(cluster)/1.1,max(cluster)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.savefig("{0}_imf_figure_log.png".format(name),bbox_inches='tight', dpi=150) pl.savefig("{0}_imf_figure_log.pdf".format(name),bbox_inches='tight') pl.figure(2, figsize=(20,16)) pl.clf() cluster,yax,colors = coolplot(1000, massfunc=massfunc, log=False) cluster = np.array(cluster) yax = np.array(yax) pl.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) pl.gca().set_xscale('log') masses = np.logspace(np.log10(cluster.min()), np.log10(cluster.max()),10000) pl.plot(masses,(massfunc(masses)),'r--',linewidth=2,alpha=0.5) pl.xlabel("Stellar Mass") pl.ylabel("dN(M)/dM") pl.gca().axis([min(cluster)/1.1,max(cluster)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.savefig("{0}_imf_figure_loglinear.png".format(name),bbox_inches='tight') pl.rc('font',size=20) pl.figure(3, figsize=(20,16)) ax1 = pl.subplot(1,3,1) ax1.plot(masses,(massfunc(masses)),'r--',linewidth=2,alpha=0.5) ax1.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) ax2 = pl.subplot(1,3,2) ax2.plot(masses,(massfunc(masses)),'r--',linewidth=2,alpha=0.5) ax2.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) ax3 = pl.subplot(1,3,3) ax3.plot(masses,(massfunc(masses)),'r--',linewidth=2,alpha=0.5) ax3.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) ax2.set_xlabel("Stellar Mass", fontsize=30) ax1.set_ylabel("dN(M)/dM", fontsize=30) ax1.axis([min(cluster)/1.1,1,min(yax)-0.2,max(yax)+0.5]) ax2.axis([1,5,min(yax)-0.2,max(yax[cluster>1])+0.5]) ax3.axis([5,max(cluster)*1.1,min(yax)-0.2,max(yax[cluster>5])+0.5]) pl.tight_layout() pl.savefig("{0}_imf_figure_linearlinear.png".format(name), bbox_inches='tight') pl.rc('font',size=30) # make one more plot, now showing a top-heavy (shallow-tail) IMF massfunc = imf.Kroupa(p3=1.75) name='KroupaTopHeavy' pl.figure(1, figsize=(10,8)) pl.clf() cluster,yax,colors = coolplot(1000, massfunc=massfunc) pl.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) pl.gca().set_xscale('log') masses = np.logspace(np.log10(cluster.min()), np.log10(cluster.max()),10000) pl.plot(masses,np.log10(massfunc(masses)),'r--',linewidth=2,alpha=0.5) pl.xlabel("Stellar Mass") pl.ylabel("log(dN(M)/dM)") pl.gca().axis([min(cluster)/1.1,max(cluster)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.savefig("{0}_imf_figure_log.png".format(name),bbox_inches='tight', dpi=150) pl.savefig("{0}_imf_figure_log.pdf".format(name),bbox_inches='tight') # make two more plots, now showing a bottom- and a top-heavy IMF for massfunc, name in [(imf.Salpeter(alpha=1.5), 'Alpha1p5'), (imf.Salpeter(alpha=2), 'Alpha2p0'), (imf.Salpeter(alpha=1), 'Alpha1p0'), (imf.Salpeter(alpha=3), 'Alpha3p0')]: pl.figure(1, figsize=(10,8)) pl.clf() cluster,yax,colors = coolplot(1000, massfunc=massfunc) pl.scatter(cluster, yax, c=colors, s=np.log10(cluster+3)*85, linewidths=0.5, edgecolors=(0,0,0,0.25), alpha=0.95) pl.gca().set_xscale('log') masses = np.logspace(np.log10(cluster.min()), np.log10(cluster.max()),10000) pl.plot(masses,np.log10(massfunc(masses)),'r--',linewidth=2,alpha=0.5) pl.xlabel("Stellar Mass") pl.ylabel("log(dN(M)/dM)") pl.gca().axis([min(cluster)/1.1,max(cluster)*1.1,min(yax)-0.2,max(yax)+0.5]) pl.savefig("{0}_imf_figure_log.png".format(name),bbox_inches='tight', dpi=150) pl.savefig("{0}_imf_figure_log.pdf".format(name),bbox_inches='tight')
5,721
44.412698
86
py
imf
imf-master/examples/pmf_comparison.py
import imf.imf, imf.pmf, imp imp.reload(imf.imf) imp.reload(imf.pmf) imp.reload(imf.imf) imp.reload(imf.pmf) from imf.pmf import ChabrierPMF_IS, ChabrierPMF_TC, ChabrierPMF_CA, ChabrierPMF_2CTC from imf.pmf import KroupaPMF_IS, KroupaPMF_TC, KroupaPMF_CA, KroupaPMF_2CTC import pylab as pl import numpy as np pl.rc('font', size=16) pl.rc('lines', linewidth=2) mmin = 0.033 mmax = 3.0 for mmax in (3, 120): # this apparently takes a very, very long time... like 20-30 minutes? # that doesn't make a whole lot of sense, so it probably indicates a bug # or an overly-precise convergence threshold print("Normalizing.") ChabrierPMF_IS.mmax = mmax ChabrierPMF_TC.mmax = mmax ChabrierPMF_CA.mmax = mmax ChabrierPMF_2CTC.mmax = mmax ChabrierPMF_IS.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_TC.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_CA.normalize(log=True, mmin=mmin, mmax=mmax) ChabrierPMF_2CTC.normalize(log=True, mmin=mmin, mmax=mmax) KroupaPMF_IS.mmax = mmax KroupaPMF_TC.mmax = mmax KroupaPMF_CA.mmax = mmax KroupaPMF_2CTC.mmax = mmax KroupaPMF_IS.normalize(log=True, mmin=mmin, mmax=mmax) KroupaPMF_TC.normalize(log=True, mmin=mmin, mmax=mmax) KroupaPMF_CA.normalize(log=True, mmin=mmin, mmax=mmax) KroupaPMF_2CTC.normalize(log=True, mmin=mmin, mmax=mmax) chabrier2005 = imf.Chabrier2005() chabrier2005.normalize(log=True, mmin=mmin, mmax=mmax) kroupa = imf.Kroupa() kroupa.normalize(log=True, mmin=mmin, mmax=mmax) print("Now plotting.") masses = np.logspace(np.log10(mmin), np.log10(mmax), 100) for mass_weighted in (True,False): fname = 'mass_weighted' if mass_weighted else '__call__' fig1 = pl.figure(1) fig1.clf() ax = fig1.gca() ax.set_title("Steady State McKee/Offner + Chabrier PMF") ax.loglog(masses, chabrier2005.__getattribute__(fname)(masses), label="IMF", color='k') ax.loglog(masses, ChabrierPMF_IS.__getattribute__(fname)(masses), label="IS", color='r', linestyle=':') ax.loglog(masses, ChabrierPMF_TC.__getattribute__(fname)(masses), label="TC", color='g', linestyle='-.') ax.loglog(masses, ChabrierPMF_CA.__getattribute__(fname)(masses), label="CA", color='y', linestyle='-.') ax.loglog(masses, ChabrierPMF_2CTC.__getattribute__(fname)(masses), label="2CTC", color='b', linestyle='--') ax.set_xlabel("(Proto)Stellar Mass (M$_\odot$)") ax.set_ylabel("m P(m)" if mass_weighted else "Normalized P(M)") ax.axis([mmin, mmax, 1e-4, 1]) pl.legend(loc='best') pl.savefig('steadystate_pmf_chabrier{0}_mmax{1}.png'.format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight') fig2 = pl.figure(2) fig2.clf() ax = fig2.gca() ax.set_title("Tapered McKee/Offner + Chabrier PMF") ax.loglog(masses, chabrier2005.__getattribute__(fname)(masses), label="IMF", color='k') ax.loglog(masses, ChabrierPMF_IS.__getattribute__(fname)(masses, taper=True), label="IS", color='r', linestyle=':') ax.loglog(masses, ChabrierPMF_TC.__getattribute__(fname)(masses, taper=True), label="TC", color='g', linestyle='-.') ax.loglog(masses, ChabrierPMF_CA.__getattribute__(fname)(masses, taper=True), label="CA", color='y', linestyle='-.') ax.loglog(masses, ChabrierPMF_2CTC.__getattribute__(fname)(masses, taper=True), label="2CTC", color='b', linestyle='--') ax.set_xlabel("(Proto)Stellar Mass (M$_\odot$)") ax.set_ylabel("m P(m)" if mass_weighted else "Normalized P(M)") ax.axis([mmin, mmax, 1e-4, 1]) pl.legend(loc='best') pl.savefig('taperedaccretion_pmf_chabrier{0}_mmax{1}.png'.format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight') fig3 = pl.figure(3) fig3.clf() ax3 = fig3.gca() ax3.set_title("Steady State McKee/Offner + Kroupa PMF", pad=15) ax3.loglog(masses, kroupa.__getattribute__(fname)(masses), label="IMF", color='k') ax3.loglog(masses, KroupaPMF_IS.__getattribute__(fname)(masses), label="IS", color='r', linestyle=':') ax3.loglog(masses, KroupaPMF_TC.__getattribute__(fname)(masses), label="TC", color='g', linestyle='-.') ax3.loglog(masses, KroupaPMF_CA.__getattribute__(fname)(masses), label="CA", color='y', linestyle='-.') ax3.loglog(masses, KroupaPMF_2CTC.__getattribute__(fname)(masses), label="2CTC", color='b', linestyle='--') ax3.set_xlabel("(Proto)Stellar Mass $\\left(\\mathrm{M}_\odot\\right)$") ax3.set_ylabel("m P(m)" if mass_weighted else "Normalized P(M)") ax3.axis([mmin, mmax, 1e-4, 1]) pl.legend(loc='best') pl.savefig('steadystate_pmf_kroupa{0}_mmax{1}.png'.format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight') pl.savefig('steadystate_pmf_kroupa{0}_mmax{1}.pdf'.format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight') fig4 = pl.figure(4) fig4.clf() ax4 = fig4.gca() ax4.set_title("Tapered McKee/Offner + Kroupa PMF") ax4.loglog(masses, kroupa.__getattribute__(fname)(masses), label="IMF", color='k') ax4.loglog(masses, KroupaPMF_IS.__getattribute__(fname)(masses, taper=True), label="IS", color='r', linestyle=':') ax4.loglog(masses, KroupaPMF_TC.__getattribute__(fname)(masses, taper=True), label="TC", color='g', linestyle='-.') ax4.loglog(masses, KroupaPMF_CA.__getattribute__(fname)(masses, taper=True), label="CA", color='y', linestyle='-.') ax4.loglog(masses, KroupaPMF_2CTC.__getattribute__(fname)(masses, taper=True), label="2CTC", color='b', linestyle='--') ax4.set_xlabel("(Proto)Stellar Mass (M$_\odot$)") ax4.set_ylabel("m P(m)" if mass_weighted else "Normalized P(M)") ax4.axis([mmin, mmax, 1e-4, 1]) pl.legend(loc='best') pl.savefig('taperedaccretion_pmf_kroupa{0}_mmax{1}.png'.format("_integral" if mass_weighted else "", int(mmax)), bbox_inches='tight')
6,437
50.504
128
py
imf
imf-master/examples/hr_diagram.py
""" Create a Hertzprung-Russell (temperature-luminosity) diagram and mass-luminosity and mass-temperature diagrams populating the main sequence only with data from Ekström+ 2012 (Vizier catalog J/A+A/537/A146/iso). Colors come from vendian.org. """ import numpy as np import imf import pylab as pl from astroquery.vizier import Vizier from labellines import labelLine, labelLines # pip install matplotlib-label-lines pl.rcParams['font.size'] = 20 Vizier.ROW_LIMIT=1e7 tbl = Vizier.get_catalogs('J/A+A/537/A146/iso')[0] agemass = {} agelum = {} for age in np.unique(tbl['logAge']): agemass[age] = tbl[tbl['logAge']==age]['Mass'].max() agelum[age] = tbl[tbl['logAge']==age]['logL'].max() # mass-temperature diagram pl.figure(2, figsize=(8,8)).clf() # select a specific population age to plot - this is the youngest, with all # stars alive and main-sequence ok = tbl['logAge'] == 6.5 subtbl = tbl[ok] subtbl.sort('Mass') lowmass = subtbl[subtbl['Mass'] < 2] # downsample the high-mass part of the sample to avoid point overlap subtbl = subtbl[(subtbl['Mass'] < 25) & (subtbl['logTe']<4.6)] highmass = subtbl[(subtbl['Mass'] > 20) & (subtbl['logTe']<4.6)] subtbl=subtbl[::20] # fill in (oversample) the low-mass portion Lfit = np.polyfit(np.log10(lowmass['Mass']), lowmass['logL'], 1) # the fit is not exactly right; there's a break at 0.43ish, but it's good enough masses = np.logspace(np.log10(0.1), np.log10(0.8)) lums = np.poly1d(Lfit)(np.log10(masses)) # this was an old by-hand fit #lums[masses<(0.43)] = np.log10(0.23*(masses[masses<(0.43)])**2.3) Tfit = np.polyfit(np.log10(lowmass['Mass']), lowmass['logTe'], 1) tems = np.poly1d(Tfit)(np.log10(masses)) hmasses = np.logspace(np.log10(25), np.log10(60),5) Lfit = np.polyfit(np.log10(highmass['Mass']), highmass['logL'], 1) hlums = np.poly1d(Lfit)(np.log10(hmasses)) #lums[masses<(0.43)] = np.log10(0.23*(masses[masses<(0.43)])**2.3) Tfit = np.polyfit(np.log10(highmass['Mass']), highmass['logTe'], 1) htems = np.poly1d(Tfit)(np.log10(hmasses)) # obtain the colors from the vendian-based table colors = [imf.color_from_mass(m) for m in subtbl['Mass']] # start plotting pl.gca().set_yscale('log') pl.scatter(10**subtbl['logTe'], subtbl['Mass'], c=colors, s=(10**subtbl['logL'])**0.25*45) colors = [imf.color_from_mass(m) for m in masses] pl.scatter(10**tems, masses, c=colors, s=(10**lums)**0.25*45) colors = [imf.color_from_mass(m) for m in hmasses] pl.scatter(10**htems, hmasses, c=colors, s=(10**hlums)**0.25*45) # overlay star age horizontal lines lines = [] for age in (6.5, 7, 8, 10): L, = pl.plot([10**tems.min(), (10**htems.max())], [agemass[age]]*2, linestyle='--', color='k', label="$10^{{{0}}}$ yr".format(age)) lines.append(L) labelLines(lines) pl.xlabel("Temperature") pl.ylabel("Mass") pl.tight_layout() pl.savefig("tem_lum_diagram.svg")#, bbox_inches='tight') # HR diagram (temperature-luminosity) pl.figure(3, figsize=(8,8)).clf() colors = [imf.color_from_mass(m) for m in subtbl['Mass']] #pl.gca().set_xscale('log') pl.gca().set_yscale('log') pl.scatter(10**subtbl['logTe'], 10**subtbl['logL'], c=colors, s=(subtbl['Mass'])*5) colors = [imf.color_from_mass(m) for m in masses] pl.scatter(10**tems, 10**lums, c=colors, s=masses*5) colors = [imf.color_from_mass(m) for m in hmasses] pl.scatter(10**htems, 10**hlums, c=colors, s=hmasses*5) # I attempted to add age lines, but this approach is incorrect #suns live 5 Gyr, not 10+ Gyr # lines = [] # for age in (6.5, 7, 8, 10): # L, = pl.plot([10**tems.min(), (10**htems.max())], # [10**agelum[age]]*2, # linestyle='--', color='k', # label="$10^{{{0}}}$ yr".format(age)) # lines.append(L) labelLines(lines) pl.xlabel("Temperature") pl.ylabel("Luminosity") pl.tight_layout() pl.savefig("HR_diagram.svg")#, bbox_inches='tight') # mass-luminosity diagram pl.figure(4, figsize=(8,8)).clf() colors = [imf.color_from_mass(m) for m in subtbl['Mass']] #pl.gca().set_xscale('log') pl.gca().set_yscale('log') pl.scatter(subtbl['Mass'], 10**subtbl['logL'], c=colors, edgecolors='none', s=10**subtbl['logTe']/100) colors = [imf.color_from_mass(m) for m in masses] pl.scatter(masses, 10**lums, c=colors, edgecolors='none', s=10**tems/100) colors = [imf.color_from_mass(m) for m in hmasses] pl.scatter(hmasses, 10**hlums, c=colors, edgecolors='none', s=10**htems/100) #lines = [] #for age in (6.5, 7, 8, 10): # L, = pl.plot([masses.min(), hmasses.max()], # [10**agelum[age]]*2, # linestyle='--', color='k', # label="$10^{{{0}}}$ yr".format(age)) # lines.append(L) labelLines(lines) pl.xlabel("Mass") pl.ylabel("Luminosity") pl.tight_layout() pl.savefig("mass_luminosity.svg")#, bbox_inches='tight') pl.loglog() pl.savefig("mass_lum_diagram_loglog.svg")#, bbox_inches='tight') pl.savefig("mass_lum_diagram_loglog.png", bbox_inches='tight') # from Zinnecker & Yorke fig 1 pl.plot(np.logspace(0,1), np.logspace(0,1)**3.7, 'k--') pl.plot(np.logspace(1,2), np.logspace(1,2)**1.6 * 1e6/100**1.6, 'k--') pl.savefig("mass_lum_diagram_loglog_ZinnPlots.svg")#, bbox_inches='tight') pl.savefig("mass_lum_diagram_loglog_ZinnPlots.png", bbox_inches='tight')
5,605
28.197917
80
py
imf
imf-master/imf/imf.py
""" Various codes to work with the initial mass function """ from __future__ import print_function import numpy as np import types import scipy.integrate import scipy.integrate as integrate from scipy.integrate import quad from astropy import units as u from . import distributions class MassFunction(object): """ Generic Mass Function class (this is mostly meant to be subclassed by other functions, not used itself) """ def __init__(self, mmin=None, mmax=None): self._mmin = self.default_mmin if mmin is None else mmin self._mmax = self.default_mmax if mmax is None else mmax def dndm(self, m, **kwargs): """ The differential form of the mass function, d N(M) / dM """ return self(m, integral_form=False, **kwargs) def n_of_m(self, m, **kwargs): """ The integral form of the mass function, N(M) """ return self(m, integral_form=True, **kwargs) def mass_weighted(self, m, **kwargs): return self(m, integral_form=False, **kwargs) * m def integrate(self, mlow, mhigh, **kwargs): """ Integrate the mass function over some range """ return scipy.integrate.quad(self, mlow, mhigh) def m_integrate(self, mlow, mhigh, **kwargs): """ Integrate the mass-weighted mass function over some range (this tells you the fraction of mass in the specified range) """ return scipy.integrate.quad(self.mass_weighted, mlow, mhigh, **kwargs) def log_integrate(self, mlow, mhigh, **kwargs): def logform(x): return self(x) / x return scipy.integrate.quad(logform, mlow, mhigh, **kwargs) def normalize(self, mmin=None, mmax=None, log=False, **kwargs): """ Set self.normfactor such that the integral of the function over the range (mmin, mmax) = 1 """ if mmin is None: mmin = self.mmin if mmax is None: mmax = self.mmax self.normfactor = 1 if log: integral = self.log_integrate(mmin, mmax, **kwargs) else: integral = self.integrate(mmin, mmax, **kwargs) self.normfactor = 1. / integral[0] assert self.normfactor > 0 @property def mmin(self): return self._mmin @property def mmax(self): return self._mmax class Salpeter(MassFunction): default_mmin = 0.3 default_mmax = 120 def __init__(self, alpha=2.35, mmin=default_mmin, mmax=default_mmax): """ Create a default Salpeter mass function, i.e. a power-law mass function the Salpeter 1955 IMF: dn/dm ~ m^-2.35 """ super().__init__(mmin=mmin, mmax=mmax) self.alpha = alpha self.normfactor = 1 self.distr = distributions.PowerLaw(-self.alpha, self.mmin, self.mmax) def __call__(self, m, integral_form=False): if not integral_form: return self.distr.pdf(m) * self.normfactor else: return self.distr.cdf(m) * self.normfactor class Kroupa(MassFunction): # kroupa = BrokenPowerLaw(breaks={0.08: -0.3, 0.5: 1.3, 'last': 2.3}, mmin=0.03, mmax=120) default_mmin = 0.03 default_mmax = 120 def __init__(self, mmin=default_mmin, mmax=default_mmax, p1=0.3, p2=1.3, p3=2.3, break1=0.08, break2=0.5): """ The Kroupa IMF with two power-law breaks, p1 and p2. See __call__ for details. """ super().__init__(mmin=mmin, mmax=mmax) self.p1 = p1 self.p2 = p2 self.p3 = p3 self.break1 = break1 self.break2 = break2 self.distr = distributions.BrokenPowerLaw([-p1, -p2, -p3], [self.mmin, break1, break2, self.mmax]) self.normfactor = 1 def __call__(self, m, integral_form=False): """ Kroupa 2001 IMF (http://arxiv.org/abs/astro-ph/0009005, http://adsabs.harvard.edu/abs/2001MNRAS.322..231K) eqn 2 Parameters ---------- m: float array The mass at which to evaluate the function (Msun) p1, p2, p3: floats The power-law slopes of the different segments of the IMF break1, break2: floats The mass breakpoints at which to use the different power laws """ if integral_form: return self.normfactor * self.distr.cdf(m) else: return self.normfactor * self.distr.pdf(m) def integrate(self, mlow, mhigh, numerical=False): """ Integrate the mass function over some range """ if mhigh <= mlow: raise ValueError("Must have mlow < mhigh in integral") if numerical: return super(Kroupa, self).integrate(mlow, mhigh) return (self.distr.cdf(mhigh) - self.distr.cdf(mlow)) * self.normfactor, 0 def m_integrate(self, mlow, mhigh, numerical=False, **kwargs): """ Integrate the mass function over some range """ if mhigh <= mlow: raise ValueError("Must have mlow < mhigh in integral") if numerical: return super(Kroupa, self).m_integrate(mlow, mhigh, **kwargs) else: distr1 = distributions.BrokenPowerLaw( [-self.p1 + 1, -self.p2 + 1, -self.p3 + 1], [self.mmin, self.break1, self.break2, self.mmax]) ratio = distr1.pdf(self.break1) / self.distr.pdf( self.break1) / self.break1 return ((distr1.cdf(mhigh) - distr1.cdf(mlow)) / ratio, 0) class ChabrierLogNormal(MassFunction): """ Eqn 18 of https://ui.adsabs.harvard.edu/abs/2003PASP..115..763C/abstract is eqn3 of https://ui.adsabs.harvard.edu/abs/2003ApJ...586L.133C/abstract \\xi = 0.086 exp (-(log m - log 0.22)^2 / (2 * 0.57**2)) This function is a pure lognormal; see ChabrierPowerLaw for the version with a power-law extension to high mass Parameters ---------- lognormal_center : float lognormal_width : float mmin : float mmax : float leading_constant : float """ default_mmin = 0 default_mmax = np.inf def __init__(self, mmin=default_mmin, mmax=default_mmax, lognormal_center=0.22, lognormal_width=0.57*np.log(10), leading_constant=0.086): super().__init__(mmin=mmin, mmax=mmax) self.multiplier = leading_constant self.lognormal_width = lognormal_width self.distr = distributions.TruncatedLogNormal(mu=lognormal_center, sig=self.lognormal_width, m1=self.mmin, m2=self.mmax) def __call__(self, mass, integral_form=False, **kw): if integral_form: return self.distr.cdf(mass) * self.multiplier else: return self.distr.pdf(mass) * self.multiplier class ChabrierPowerLaw(MassFunction): default_mmin = 0 default_mmax = np.inf def __init__(self, lognormal_center=0.22, lognormal_width=0.57*np.log(10), mmin=default_mmin, mmax=default_mmax, alpha=2.3, mmid=1): """ From Equation 18 of Chabrier 2003 https://ui.adsabs.harvard.edu/abs/2003PASP..115..763C/abstract Parameters ---------- lognormal_center : float lognormal_width : float The lognormal width. Scipy.stats.lognorm uses log_n, so we need to scale this b/c Chabrier uses log_10 mmin : float mmax : float alpha : float The high-mass power-law slope mmid : float The mass to transition from lognormal to power-law Notes ----- A previous version of this function used sigma=0.55, center=0.2, and alpha=2.35, which come from McKee & Offner 2010 (https://ui.adsabs.harvard.edu/abs/2010ApJ...716..167M/abstract) but those exact numbers don't appear in Chabrier 2005 """ # The numbers are from Eqn 3 of # https://ui.adsabs.harvard.edu/abs/2005ASSL..327...41C/abstract # There is no equation 3 in that paper, though? # importantly the lognormal center is the exp(M) where M is the mean of ln(mass) # normal distribution super().__init__(mmin=mmin, mmax=mmax) self._mmid = mmid if self.mmax <= self._mmid: raise ValueError("The Chabrier Mass Function does not support " "mmax <= mmid") self._alpha = alpha self._lognormal_width = lognormal_width self._lognormal_center = lognormal_center self.distr = distributions.CompositeDistribution([ distributions.TruncatedLogNormal(self._lognormal_center, self._lognormal_width, self.mmin, self._mmid), distributions.PowerLaw(-self._alpha, self._mmid, self.mmax) ]) def __call__(self, x, integral_form=False, **kw): if integral_form: return self.distr.cdf(x) else: return self.distr.pdf(x) class Schechter(MassFunction): default_mmin = 0 default_mmax = np.inf def __init__(self, mmin=default_mmin, mmax=default_mmax): raise NotImplementedError("Schechter function needs to be refactored") super().__init__(mmin=mmin, mmax=mmax) def __call__(self, m, A=1, beta=2, m0=100, integral_form=False): """ A Schechter function with arbitrary defaults (integral may not be correct - exponent hasn't been dealt with at all) (TODO: this should be replaced with a Truncated Power Law Distribution) $$ A m^{-\\beta} e^{-m/m_0} $$ Parameters ---------- m: np.ndarray List of masses for which to compute the Schechter function A: float Arbitrary amplitude of the Schechter function beta: float Power law exponent m0: float Characteristic mass (mass at which exponential decay takes over) Returns ------- p(m) - the (unnormalized) probability of an object of a given mass as a function of that object's mass (though you could interpret mass as anything, it's just a number) """ if integral_form: beta -= 1 return A * m**-beta * np.exp(-m / m0) * (m > self.mmin) * (m < self.mmax) class ModifiedSchecter(Schechter): default_mmin = 0 default_mmax = np.inf def __init__(self, mmin=default_mmin, mmax=default_mmax): self.schechter = super().__init__(mmin=mmin, mmax=mmax) def __call__(self, m, m1, **kwargs): """ A Schechter function with a low-level exponential cutoff (TODO: this should be replaced with a Truncated Power Law Distribution) Parameters ---------- m: np.ndarray List of masses for which to compute the Schechter function m1: float Characteristic minimum mass (exponential decay below this mass) ** See schecter for other parameters ** Returns ------- p(m) - the (unnormalized) probability of an object of a given mass as a function of that object's mass (though you could interpret mass as anything, it's just a number) """ return self.schechter(m, **kwargs) * np.exp(-m1 / m) * (m > self.mmin) * (m < self.mmax) try: import scipy def schechter_cdf(m, A=1, beta=2, m0=100, mmin=10, mmax=None, npts=1e4): """ Return the CDF value of a given mass for a set mmin, mmax mmax will default to 10 m0 if not specified Analytic integral of the Schechter function: http://www.wolframalpha.com/input/?i=integral%28x^-a+exp%28-x%2Fm%29+dx%29 """ if mmax is None: mmax = 10 * m0 # integrate the CDF from the minimum to maximum posint = -mmax**(1 - beta) * scipy.special.expn(beta, mmax / m0) negint = -mmin**(1 - beta) * scipy.special.expn(beta, mmin / m0) tot = posint - negint # normalize by the integral ret = (-m**(1 - beta) * scipy.special.expn(beta, m / m0) - negint) / tot return ret def sh_cdf_func(**kwargs): return lambda x: schechter_cdf(x, **kwargs) except ImportError: pass # these are global objects salpeter = Salpeter() kroupa = Kroupa() lognormal = chabrierlognormal = ChabrierLogNormal() chabrier = chabrierpowerlaw = ChabrierPowerLaw() chabrier2005 = ChabrierPowerLaw(lognormal_width=0.55*np.log(10), lognormal_center=0.2, alpha=2.35) massfunctions = {'kroupa': Kroupa, 'salpeter': Salpeter, 'chabrierlognormal': ChabrierLogNormal, 'chabrierpowerlaw': ChabrierPowerLaw, 'chabrier': ChabrierPowerLaw, } # 'schechter': Schechter, 'modified_schechter': ModifiedSchecter} reverse_mf_dict = {v: k for k, v in massfunctions.items()} # salpeter and schechter selections are arbitrary expectedmass_cache = {} def get_massfunc(massfunc, mmin=None, mmax=None, **kwargs): if isinstance(massfunc, MassFunction): if mmax is not None and massfunc.mmax != mmax: raise ValueError("mmax was specified, but a massfunction instance" " was specified with a different mmax") if mmin is not None and massfunc.mmin != mmin: raise ValueError("mmin was specified, but a massfunction instance" " was specified with a different mmin") return massfunc elif massfunc in massfunctions.values(): # if the massfunction is a known MassFunc class return massfunc(mmin=mmin, mmax=mmax, **kwargs) elif massfunc in massfunctions: # if the massfunction is the _name_ of a massfunc class return massfunctions[massfunc](mmin=mmin, mmax=mmax, **kwargs) else: raise ValueError("massfunc must either be a string in the set %s or a MassFunction instance" % (", ".join(massfunctions.keys()))) def get_massfunc_name(massfunc): if massfunc in reverse_mf_dict: return reverse_mf_dict[massfunc] elif type(massfunc) is str: return massfunc elif hasattr(massfunc, '__name__'): return massfunc.__name__ else: raise ValueError("invalid mass function") def m_integrate(fn=kroupa, bins=np.logspace(-2, 2, 500)): xax = (bins[:-1] + bins[1:]) / 2. integral = xax * (bins[1:] - bins[:-1]) * (fn(bins[:-1]) + fn(bins[1:])) / 2. return xax, integral def cumint(fn=kroupa, bins=np.logspace(-2, 2, 500)): xax, integral = integrate(fn, bins) return integral.cumsum() / integral.sum() def m_cumint(fn=kroupa, bins=np.logspace(-2, 2, 500)): xax, integral = m_integrate(fn, bins) return integral.cumsum() / integral.sum() def inverse_imf(p, mmin=None, mmax=None, massfunc='kroupa', **kwargs): """ Inverse mass function. Given a likelihood value in the range [0, 1), return the appropriate mass. This just calls the mass function's ppdf under the hood. Parameters ---------- p: np.array An array of floats in the range [0, 1). These should be uniformly random numbers. mmin: float mmax: float Minimum and maximum stellar mass in the distribution massfunc: string or function massfunc can be 'kroupa', 'chabrier', 'salpeter', 'schechter', or a function """ mfc = get_massfunc(massfunc, mmin=mmin, mmax=mmax) # this should be the entirety of "inverse-imf". The rest is a hack if hasattr(mfc, 'distr'): return mfc.distr.ppf(p) else: raise NotImplementedError def make_cluster(mcluster, massfunc='kroupa', verbose=False, silent=False, tolerance=0.0, stop_criterion='nearest', sampling='random', mmax=None, mmin=None, **kwargs): """ Sample from an IMF to make a cluster. Returns the masses of all stars in the cluster Parameters ========== mcluster : float The target cluster mass. massfunc : string or MassFunction A mass function to use. tolerance : float tolerance is how close the cluster mass must be to the requested mass. It can be zero, but this does not guarantee that the final cluster mass will be exactly `mcluster` stop_criterion : 'nearest', 'before', 'after', 'sorted' The criterion to stop sampling when the total cluster mass is reached. See, e.g., Krumholz et al 2015: https://ui.adsabs.harvard.edu/abs/2015MNRAS.452.1447K/abstract sampling: 'random' or 'optimal' Optimal sampling is based on https://ui.adsabs.harvard.edu/abs/2015A%26A...582A..93S/abstract (though as of April 23, 2021, it is not yet correct) Optimal sampling is only to be used in the context of a variable M_max that is a function of the cluster mass, e.g., eqn 24 of Schulz+ 2015. """ # use most common mass to guess needed number of samples # nsamp = mcluster / mostcommonmass[get_massfunc_name(massfunc)] # masses = inverse_imf(np.random.random(int(nsamp)), massfunc=massfunc, **kwargs) # mtot = masses.sum() # if verbose: # print(("%i samples yielded a cluster mass of %g (%g requested)" % # (nsamp, mtot, mcluster))) mcluster = u.Quantity(mcluster, u.M_sun).value mfc = get_massfunc(massfunc, mmin=mmin, mmax=mmax, **kwargs) if (massfunc, mfc.mmin, mfc.mmax) in expectedmass_cache: expected_mass = expectedmass_cache[(massfunc, mfc.mmin, mfc.mmax)] assert expected_mass > 0 else: expected_mass = mfc.m_integrate(mfc.mmin, mfc.mmax)[0] assert expected_mass > 0 expectedmass_cache[(massfunc, mfc.mmin, mfc.mmax)] = expected_mass if verbose: print("Expected mass is {0:0.3f}".format(expected_mass)) if sampling == 'optimal': # this is probably not _quite_ right, but it's a first step... p = np.linspace(0, 1, int(mcluster/expected_mass)) return mfc.distr.ppf(p) elif sampling != 'random': raise ValueError("Only random sampling and optimal sampling are supported") mtot = 0 masses = [] while mtot < mcluster + tolerance: # at least 1 sample, but potentially many more nsamp = int(np.ceil((mcluster + tolerance - mtot) / expected_mass)) assert nsamp > 0 newmasses = mfc.distr.rvs(nsamp) masses = np.concatenate([masses, newmasses]) mtot = masses.sum() if verbose: print("Sampled %i new stars. Total is now %g" % (int(nsamp), mtot)) if mtot >= mcluster + tolerance: # don't force exact equality; that would yield infinite loop mcum = masses.cumsum() if stop_criterion == 'sorted': masses = np.sort(masses) if np.abs(masses[:-1].sum() - mcluster) < np.abs(masses.sum() - mcluster): # if the most massive star makes the cluster a worse fit, reject it # (this follows Krumholz+ 2015 appendix A1) last_ind = len(masses) - 1 else: last_ind = len(masses) else: if stop_criterion == 'nearest': # find the closest one, and use +1 to include it last_ind = np.argmin(np.abs(mcum - mcluster)) + 1 elif stop_criterion == 'before': last_ind = np.argmax(mcum > mcluster) elif stop_criterion == 'after': last_ind = np.argmax(mcum > mcluster) + 1 masses = masses[:last_ind] mtot = masses.sum() if verbose: print( "Selected the first %i out of %i masses to get %g total" % (last_ind, len(mcum), mtot)) # force the break, because some stopping criteria can push mtot < mcluster break if not silent: print("Total cluster mass is %g (limit was %g)" % (mtot, mcluster)) return masses mass_luminosity_interpolator_cache = {} def mass_luminosity_interpolator(name): if name in mass_luminosity_interpolator_cache: return mass_luminosity_interpolator_cache[name] elif name == 'VGS': # non-extrapolated vgsMass = [ 51.3, 44.2, 41.0, 38.1, 35.5, 33.1, 30.8, 28.8, 26.9, 25.1, 23.6, 22.1, 20.8, 19.5, 18.4 ] vgslogL = [ 6.154, 6.046, 5.991, 5.934, 5.876, 5.817, 5.756, 5.695, 5.631, 5.566, 5.499, 5.431, 5.360, 5.287, 5.211 ] vgslogQ = [ 49.18, 48.99, 48.90, 48.81, 48.72, 48.61, 48.49, 48.34, 48.16, 47.92, 47.63, 47.25, 46.77, 46.23, 45.69 ] # mass extrapolated vgsMe = np.concatenate([ np.linspace(0.03, 0.43, 100), np.linspace(0.43, 2, 100), np.linspace(2, 20, 100), vgsMass[::-1], np.linspace(50, 150, 100) ]) # log luminosity extrapolated vgslogLe = np.concatenate([ np.log10(0.23 * np.linspace(0.03, 0.43, 100)**2.3), np.log10(np.linspace(0.43, 2, 100)**4), np.log10(1.5 * np.linspace(2, 20, 100)**3.5), vgslogL[::-1], np.polyval(np.polyfit(np.log10(vgsMass[:3]), vgslogL[:3], 1), np.log10(np.linspace(50, 150, 100))) ]) # log Q (lyman continuum) extrapolated vgslogQe = np.concatenate([ np.zeros(100), # 0.03-0.43 solar mass stars produce 0 LyC photons np.zeros(100), # 0.43-2.0 solar mass stars produce 0 LyC photons np.polyval(np.polyfit(np.log10(vgsMass[-3:]), vgslogQ[-3:], 1), np.log10(np.linspace(8, 18.4, 100))), vgslogQ[::-1], np.polyval(np.polyfit(np.log10(vgsMass[:3]), vgslogQ[:3], 1), np.log10(np.linspace(50, 150, 100))) ]) mass_luminosity_interpolator_cache[name] = vgsMe, vgslogLe, vgslogQe return mass_luminosity_interpolator_cache[name] elif name == 'Ekstrom': from astroquery.vizier import Vizier Vizier.ROW_LIMIT = 1e7 # effectively infinite # this query should cache tbl = Vizier.get_catalogs('J/A+A/537/A146/iso')[0] match = tbl['logAge'] == 6.5 masses = tbl['Mass'][match] lums = tbl['logL'][match] mass_0 = 0.033 lum_0 = np.log10((mass_0 / masses[0])**3.5 * 10**lums[0]) mass_f = 200 # extrapolate to 200 Msun... lum_f = np.log10(10**lums[-1] * (mass_f / masses[-1])**1.35) masses = np.array([mass_0] + masses.tolist() + [mass_f]) lums = np.array([lum_0] + lums.tolist() + [lum_f]) # TODO: come up with a half-decent approximation here? based on logTe? logQ = lums - 0.5 mass_luminosity_interpolator_cache[name] = masses, lums, logQ return mass_luminosity_interpolator_cache[name] else: raise ValueError("Bad grid name {0}".format(name)) def lum_of_star(mass, grid='Ekstrom'): """ Determine total luminosity of a star given its mass Two grids: (1) VGS: Uses the Vacca, Garmany, Shull 1996 Table 5 Log Q and Mspec parameters returns LogL in solar luminosities **WARNING** Extrapolates for M not in [18.4, 50] msun http://en.wikipedia.org/wiki/Mass%E2%80%93luminosity_relation (2) Ekstrom 2012: Covers 0.8 - 64 Msun, extrapolated out of that """ masses, lums, _ = mass_luminosity_interpolator(grid) return np.interp(mass, masses, lums) def lum_of_cluster(masses, grid='Ekstrom'): """ Determine the log of the integrated luminosity of a cluster Only M>=8msun count masses is a list or array of masses. """ #if max(masses) < 8: return 0 logL = lum_of_star(masses, grid=grid) #[masses >= 8]) logLtot = np.log10((10**logL).sum()) return logLtot def lyc_of_star(mass, grid='VGS'): """ Determine lyman continuum luminosity of a star given its mass Uses the Vacca, Garmany, Shull 1996 Table 5 Log Q and Mspec parameters returns LogQ """ masses, _, logQ = mass_luminosity_interpolator(grid) return np.interp(mass, masses, logQ) def lyc_of_cluster(masses, grid='VGS'): """ Determine the log of the integrated lyman continuum luminosity of a cluster Only M>=8msun count masses is a list or array of masses. """ if max(masses) < 8: return 0 logq = lyc_of_star(masses[masses >= 8], grid=grid) logqtot = np.log10((10**logq).sum()) return logqtot def color_from_mass(mass, outtype=float): """ Use vendian.org colors: 100 O2(V) 150 175 255 #9db4ff 50 O5(V) 157 180 255 #9db4ff 20 B1(V) 162 185 255 #a2b9ff 10 B3(V) 167 188 255 #a7bcff 8 B5(V) 170 191 255 #aabfff 6 B8(V) 175 195 255 #afc3ff 2.2 A1(V) 186 204 255 #baccff 2.0 A3(V) 192 209 255 #c0d1ff 1.86 A5(V) 202 216 255 #cad8ff 1.6 F0(V) 228 232 255 #e4e8ff 1.5 F2(V) 237 238 255 #edeeff 1.3 F5(V) 251 248 255 #fbf8ff 1.2 F8(V) 255 249 249 #fff9f9 1 G2(V) 255 245 236 #fff5ec 0.95 G5(V) 255 244 232 #fff4e8 0.90 G8(V) 255 241 223 #fff1df 0.85 K0(V) 255 235 209 #ffebd1 0.70 K4(V) 255 215 174 #ffd7ae 0.60 K7(V) 255 198 144 #ffc690 0.50 M2(V) 255 190 127 #ffbe7f 0.40 M4(V) 255 187 123 #ffbb7b 0.35 M6(V) 255 187 123 #ffbb7b 0.30 M8(V) 255 167 123 #ffbb7b # my addition """ mcolor = { # noqa: E131 100: (150, 175, 255), 50: (157, 180, 255), 20: (162, 185, 255), 10: (167, 188, 255), 8: (170, 191, 255), 6: (175, 195, 255), 2.2: (186, 204, 255), 2.0: (192, 209, 255), 1.86: (202, 216, 255), 1.6: (228, 232, 255), 1.5: (237, 238, 255), 1.3: (251, 248, 255), 1.2: (255, 249, 249), 1: (255, 245, 236), 0.95: (255, 244, 232), 0.90: (255, 241, 223), 0.85: (255, 235, 209), 0.70: (255, 215, 174), 0.60: (255, 198, 144), 0.50: (255, 190, 127), 0.40: (255, 187, 123), 0.35: (255, 187, 123), 0.30: (255, 177, 113), 0.20: (255, 107, 63), 0.10: (155, 57, 33), 0.10: (155, 57, 33), 0.003: (105, 27, 0), } keys = sorted(mcolor.keys()) reds, greens, blues = zip(*[mcolor[k] for k in keys]) r = np.interp(mass, keys, reds) g = np.interp(mass, keys, greens) b = np.interp(mass, keys, blues) if outtype == int: return (r, g, b) elif outtype == float: return (r / 255., g / 255., b / 255.) else: raise NotImplementedError def color_of_cluster(cluster, colorfunc=color_from_mass): colors = np.array([colorfunc(m) for m in cluster]) luminosities = 10**np.array([lum_of_star(m) for m in cluster]) mean_color = (colors * luminosities[:, None]).sum(axis=0) / luminosities.sum() return mean_color def coolplot(clustermass, massfunc=kroupa, log=True, **kwargs): """ "cool plot" is just because the plot is kinda neat. This function creates a cluster using `make_cluster`, assigns each star a color based on the vendian.org colors using `color_from_mass`, and assigns each star a random Y-value distributed underneath the specified mass function's curve. Parameters ---------- clustermass: float The mass of the cluster in solar masses massfunc: str A MassFunction instance log: bool Is the Y-axis log-scaled? Returns ------- cluster: array The array of stellar masses that makes up the cluster yax: array The array of Y-values associated with the stellar masses colors: list A list of color tuples associated with each star """ cluster = make_cluster(clustermass, massfunc=massfunc, mmax=massfunc.mmax, **kwargs) colors = [color_from_mass(m) for m in cluster] maxmass = cluster.max() pmin = massfunc(maxmass) if log: yax = [ np.random.rand() * (np.log10(massfunc(m)) - np.log10(pmin)) + np.log10(pmin) for m in cluster ] else: yax = [ np.random.rand() * ((massfunc(m)) / (pmin)) + (pmin) for m in cluster ] assert all(np.isfinite(yax)) return cluster, yax, colors # import pylab as pl # pl.scatter(cluster, yax, c=colors, s=np.log10(cluster)*5) class KoenConvolvedPowerLaw(MassFunction): """ Implementaton of convolved errror power-law described in 2009 Koen, Kondlo paper, Fitting power-law distributions to data with measurement errors. Equations (3) and (5) Parameters ---------- m: float The mass at which to evaluate the function mmin, mmax: floats The upper and lower bounds for the power law distribution gamma: floats The specified gamma for the distribution, slope = -gamma - 1 sigma: float or None specified spread of error, assumes Normal distribution with mean 0 and variance sigma. """ default_mmin = 0 default_mmax = np.inf def __init__(self, mmin, mmax, gamma, sigma): super().__init__(mmin, mmax) self.sigma = sigma self.gamma = gamma def __call__(self, m, integral_form=False): m = np.asarray(m) if self.mmax < self.mmin: raise ValueError("mmax must be greater than mmin") if integral_form: # Returns # ------- # Probability that m < x for the given CDF with specified # mmin, mmax, sigma, and gamma def error(t): return np.exp(-(t**2) / 2) error_coeffecient = 1 / np.sqrt(2 * np.pi) def error_integral(y): error_integral = quad(error, -np.inf, (y - self.mmax) / self.sigma)[0] return error_integral vector_errorintegral = np.vectorize(error_integral) phi = vector_errorintegral(m) * error_coeffecient def integrand(x, y): return ((self.mmin**-self.gamma - x**-self.gamma) * np.exp( (-1 / 2) * ((y - x) / self.sigma)**2)) coef = (1 / (self.sigma * np.sqrt(2 * np.pi) * (self.mmin**-self.gamma - self.mmax**-self.gamma))) def eval_integral(y): integral = quad(integrand, self.mmin, self.mmax, args=(y))[0] return integral vector_integral = np.vectorize(eval_integral) probability = phi + coef * vector_integral(m) return probability else: # Returns # ------ # Probability of getting x given the PDF with specified mmin, mmax, sigma, and gamma def integrand(x, y): return (x**-(self.gamma + 1)) * np.exp(-.5 * ( (y - x) / self.sigma)**2) coef = (self.gamma / ((self.sigma * np.sqrt(2 * np.pi)) * ((self.mmin**-self.gamma) - (self.mmax**-self.gamma)))) def Integral(y): I = quad(integrand, self.mmin, self.mmax, args=(y))[0] return I vector_I = np.vectorize(Integral) return coef * vector_I(m) class KoenTruePowerLaw(MassFunction): """ Implementaton of error free power-law described in 2009 Koen Kondlo paper, Fitting power-law distributions to data with measurement errors This is a power law with truncations on the low and high end. Equations (2) and (4) Parameters ---------- m: float The mass at which to evaluate the function mmin, mmax: floats The upper and lower bounds for the power law distribution gamma: floats The specified gamma for the distribution, related to the slope, alpha = -gamma + 1 """ default_mmin = 0 default_mmax = np.inf def __init__(self, mmin, mmax, gamma): super().__init__(mmin, mmax) self.gamma = gamma def __call__(self, m, integral_form=False): m = np.asarray(m) if self.mmax < self.mmin: raise ValueError('mmax must be greater than mmin') if integral_form: # Returns # ------- # Probability that m < x for the given CDF with specified mmin, mmax, sigma, and gamma # True for L<=x pdf = ((self.mmin**-self.gamma - np.power(m, -self.gamma)) / self.mmin**-self.gamma - self.mmax**-self.gamma) return_value = (pdf * ((m > self.mmin) & (m < self.mmax)) + 1.0 * (m >= self.mmax) + 0 * (m < self.mmin)) return return_value else: # Returns # ------ # Probability of getting x given the PDF with specified mmin, mmax, and gamma # Answers it gives are true from mmin<=x<=mmax cdf = (self.gamma * np.power(m, -(self.gamma + 1)) / (self.mmin**-self.gamma - self.mmax**-self.gamma)) return_value = (cdf * ((m > self.mmin) & (m < self.mmax)) + 0 * (m > self.mmax) + 0 * (m < self.mmin)) return return_value
34,904
33.593657
102
py
imf
imf-master/imf/distributions.py
import numpy as np import scipy.stats class Distribution: """ The main class describing the distributions, to be inherited""" def __init__(self): self.m1 = 0 # edges of the support of the pdf self.m2 = np.inf pass def pdf(self, x): """ Return the Probability density function""" pass def cdf(self, x): """ Cumulative distribtuion function """ pass def rvs(self, N): """ Generate random sample """ pass def ppf(self, x): # inverse cdf raise RuntimeError('not implemented') pass class LogNormal(Distribution): def __init__(self, mu, sig): """ Define the Lognormal with distribution ~ 1/x exp( -1/2 *(log(x)-log(mu))^2/sig^2) I.e. the mean of log of the samples will be log(mu) and the stddev of log of the samples will be sig """ self.m1 = 0 self.m2 = np.inf self.d = scipy.stats.lognorm(s=sig, scale=mu) def pdf(self, x): return self.d.pdf(x) def cdf(self, x): return self.d.cdf(x) def rvs(self, N): return self.d.rvs(N) def ppf(self, x): return self.d.ppf(x) class TruncatedLogNormal: def __init__(self, mu, sig, m1, m2): """ Standard log-normal but truncated in the interval m1,m2 """ self.m1 = m1 self.m2 = m2 self.d = scipy.stats.lognorm(s=sig, scale=mu) self.norm = self.d.cdf(self.m2) - self.d.cdf(self.m1) def pdf(self, x): return self.d.pdf(x) * (x >= self.m1) * (x <= self.m2) / self.norm def cdf(self, x): return (self.d.cdf(np.clip(x, self.m1, self.m2)) - self.d.cdf(self.m1)) / self.norm def rvs(self, N): x = np.random.uniform(0, 1, size=N) return self.ppf(x) def ppf(self, x0): x = np.asarray(x0) cut1 = self.d.cdf(self.m1) cut2 = self.d.cdf(self.m2) ret = self.d.ppf(x * (cut2 - cut1) + cut1) ret = np.asarray(ret) ret[(x < 0) | (x > 1)] = np.nan return ret class PowerLaw(Distribution): def __init__(self, slope, m1, m2): """ Power law with slope slope in the interval m1,m2 """ self.slope = slope self.m1 = float(m1) self.m2 = float(m2) assert (m1 < m2) assert (m1 > 0) assert (m1 != -1) def pdf(self, x): if self.slope == -1: return (x**self.slope / (np.log(self.m2 / self.m1)) * (x >= self.m1) * (x <= self.m2)) else: return x**self.slope * (self.slope + 1) / ( self.m2**(self.slope + 1) - self.m1**(self.slope + 1)) * (x >= self.m1) * (x <= self.m2) def cdf(self, x): if self.slope == -1: raise RuntimeError('Not implemented') else: return (np.clip(x, self.m1, self.m2)**(self.slope + 1) - (self.m1**(self.slope + 1))) / (self.m2**(self.slope + 1) - self.m1**(self.slope + 1)) def rvs(self, N): x = np.random.uniform(size=N) return self.ppf(x) def ppf(self, x0): x = np.asarray(x0) if self.slope == -1: ret = np.exp(x * np.log(self.m2 / self.m1)) * self.m1 else: ret = (x * (self.m2**(self.slope + 1) - self.m1**(self.slope + 1)) + self.m1**(self.slope + 1))**(1. / (self.slope + 1)) ret = np.asarray(ret) ret[(x < 0) | (x > 1)] = np.nan return ret class BrokenPowerLaw: def __init__(self, slopes, breaks): """ Broken power-law with different slopes. Arguments: slopes: array Array of power-law slopes breaks: array Array of points/edges of powerlaw segments must be larger by one then the list of slopes """ self.slopes = slopes self.breaks = breaks self._calcpows() self._calcweights() @property def m1(self): return self.breaks[0] @m1.setter def m1(self, value): self.breaks[0] = value self._calcpows() self._calcweights() @property def m2(self): return self.breaks[-1] @m2.setter def m2(self, value): self.breaks[-1] = value self._calcpows() self._calcweights() def _calcpows(self): if not (len(self.slopes) == len(self.breaks) - 1): raise ValueError( 'The length of array of slopes must be equal to length of ' + 'array of break points minus 1') if not ((np.diff(self.breaks) > 0).all()): raise ValueError('Power law break-points must be monotonic') nsegm = len(self.slopes) pows = [] for ii in range(nsegm): pows.append( PowerLaw(self.slopes[ii], self.breaks[ii], self.breaks[ii + 1])) self.pows = pows self.nsegm = nsegm def _calcweights(self): nsegm = len(self.slopes) weights = [1] for ii in range(1, nsegm): rat = self.pows[ii].pdf(self.breaks[ii]) / self.pows[ii - 1].pdf( self.breaks[ii]) weights.append(weights[-1] / rat) weights = np.array(weights) self.weights = weights / np.sum(weights) # relative normalizations self.nsegm = nsegm def pdf(self, x): x1 = np.asarray(x) ret = np.atleast_1d(x1) * 0. for ii in range(self.nsegm): xind = (x1 < self.breaks[ii + 1]) & (x1 >= self.breaks[ii]) if xind.sum() > 0: ret[xind] = self.weights[ii] * self.pows[ii].pdf(x1[xind]) return ret.reshape(x1.shape) def cdf(self, x): x1 = np.asarray(x) ret = np.atleast_1d(x1) * 0. cums = np.r_[[0], np.cumsum(self.weights)] for ii in range(self.nsegm): xind = (x1 < self.breaks[ii + 1]) & (x1 >= self.breaks[ii]) if np.any(xind): ret[xind] = cums[ii] + self.weights[ii] * self.pows[ii].cdf( x1[xind]) xind = x1 >= self.breaks[-1] if np.any(xind): ret[xind] = 1 return ret.reshape(x1.shape) def rvs(self, N): Ns = np.random.multinomial(N, self.weights) ret = [] for ii in range(self.nsegm): if Ns[ii] > 0: ret.append(self.pows[ii].rvs(Ns[ii])) return np.concatenate(ret) def ppf(self, x0): x = np.asarray(x0) x1 = np.atleast_1d(x) edges = np.r_[[0], np.cumsum(self.weights)] # edges of powerlaw in CDF scale from 0 to 1 pos = np.digitize(x1, edges) # bin positions, 1 is the leftmost pos = np.clip(pos, 1, self.nsegm) # we can get zeros here if input is corrupt left = edges[pos - 1] w = self.weights[pos - 1] x2 = np.clip((x1 - left) / w, 0, 1) # mapping to 0,1 on the segment # must force float b/c int dtypes can result in truncation ret = np.zeros_like(x1, dtype='float') for ii in range(x.size): ret[ii] = self.pows[pos[ii] - 1].ppf(x2[ii]) isnan = (x1 < 0) | (x1 > 1) if any(isnan): ret[isnan] = np.nan return ret.reshape(x.shape) class CompositeDistribution(Distribution): def __init__(self, distrs): """ A Composite distribution that consists of several distributions that continuously join together Arguments: ---------- distrs: list of Distributions The list of distributions. Their supports must not overlap and not have any gaps. Example: -------- dd=distributions.CompositeDistribution([ distributions.TruncatedLogNormal(0.3,0.3,0.08,1), distributions.PowerLaw(-2.55,1,np.inf)]) dd.pdf(3) """ nsegm = len(distrs) self.distrs = distrs weights = [1] breaks = [_.m1 for _ in self.distrs] + [self.distrs[-1].m2] self.m1 = breaks[0] # leftmost edge self.m2 = breaks[-1] # rightmost edge # check that edges of intervals match for ii in range(1, nsegm): assert (distrs[ii].m1 == distrs[ii - 1].m2) for ii in range(1, nsegm): rat = distrs[ii].pdf(breaks[ii]) / distrs[ii - 1].pdf(breaks[ii]) # relative normalization of next pdf to the previous one so they # join without a break weights.append(weights[-1] / rat) weights = np.array(weights) self.breaks = breaks self.weights = weights / np.sum(weights) # these are relative weights of each pdf self.nsegm = nsegm def pdf(self, x): x1 = np.asarray(x) ret = np.atleast_1d(x1 * 0.) for ii in range(self.nsegm): xind = (x1 < self.breaks[ii + 1]) & (x1 >= self.breaks[ii]) if xind.sum() > 0: ret[xind] = self.weights[ii] * self.distrs[ii].pdf(x1[xind]) return ret.reshape(x1.shape) def cdf(self, x): x1 = np.asarray(x) ret = np.atleast_1d(x1 * 0.) cums = np.r_[[0], np.cumsum(self.weights)] for ii in range(self.nsegm): xind = (x1 < self.breaks[ii + 1]) & (x1 >= self.breaks[ii]) if xind.sum() > 0: ret[xind] = cums[ii] + self.weights[ii] * self.distrs[ii].cdf( x1[xind]) xind = x1 >= self.breaks[-1] if xind.sum(): ret[xind] = 1 return ret.reshape(x1.shape) def rvs(self, N): Ns = np.random.multinomial(N, self.weights) ret = [] for ii in range(self.nsegm): if Ns[ii] > 0: ret.append(self.distrs[ii].rvs(Ns[ii])) ret = np.concatenate(ret) ret = np.random.permutation(ret) # permutation return ret def ppf(self, x0): x = np.asarray(x0) x1 = np.atleast_1d(x) edges = np.r_[[0], np.cumsum(self.weights)] pos = np.digitize(x1, edges) pos = np.clip(pos, 1, self.nsegm) # if input is <0 or >1 left = edges[pos - 1] w = self.weights[pos - 1] x2 = np.clip((x1 - left) / w, 0, 1) # mapping to 0,1 on the segment ret = np.zeros_like(x1) for ii in range(x.size): ret[ii] = self.distrs[pos[ii] - 1].ppf(x2[ii]) ret[(x1 < 0) | (x1 > 1)] = np.nan return ret.reshape(x.shape)
10,611
30.39645
79
py
imf
imf-master/imf/plf.py
""" Protostellar luminosity functions as described by Offner and McKee, 2011 Alternatively, perhaps try to construct a probabilistic P(L; m, m_f) given a series of stellar evolution codes? """ import numpy as np import scipy.integrate import warnings from .imf import MassFunction, ChabrierPowerLaw chabrierpowerlaw = ChabrierPowerLaw() class McKeeOffner_PLF(MassFunction): def __init__(self, j=1, n=1, jf=3/4., mmin=0.033, mmax=3.0, imf=chabrierpowerlaw, **kwargs): """ Incomplete. The PLF requires a protostellar evolution code as part of its input. """ raise NotImplementedError self.j = j self.jf = jf self.n = n self.mmin = mmin self.mmax = mmax self.imf = imf def den_func(x): return self.imf(x)*x**(-self.jf) self.denominator = scipy.integrate.quad(den_func, self.mmin, self.mmax, **kwargs)[0] self.normfactor = 1 def __call__(self, luminosity, taper=False, integral_form=False, **kwargs): """ Unclear if integral_form is right...""" if taper: def num_func(x, luminosity_): tf = (1-(luminosity_/x)**(1-self.j))**0.5 return self.imf(x)*x**(self.j-self.jf-1) * tf def integrate(lolim, luminosity_): integral = scipy.integrate.quad(num_func, lolim, self.mmax, args=(luminosity_,), **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < luminosity, luminosity, self.mmin), luminosity) else: def num_func(x): return self.imf(x)*x**(self.j-self.jf-1) def integrate(lolim): integral = scipy.integrate.quad(num_func, lolim, self.mmax, **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < luminosity, luminosity, self.mmin)) result = (1-self.j) * luminosity**(1-self.j) * numerator / self.denominator if integral_form: warnings.warn("The 'integral form' of the Chabrier PMF is not correctly normalized; " "it is just PMF(m) * m") return result * self.normfactor * luminosity raise ValueError("Integral version not yet computed") else: return result * self.normfactor
2,862
36.181818
97
py
imf
imf-master/imf/pmf.py
""" Protostellar mass functions as described by McKee and Offner, 2010 """ import numpy as np import scipy.integrate import warnings from .imf import MassFunction, ChabrierPowerLaw, Kroupa chabrierpowerlaw = ChabrierPowerLaw() class McKeeOffner_PMF(MassFunction): default_mmin = 0.033 default_mmax = 3.0 def __init__(self, j=1, n=1, jf=3/4., mmin=default_mmin, mmax=default_mmax, imf=chabrierpowerlaw, **kwargs): """ """ super().__init__(mmin=mmin, mmax=mmax) self.j = j self.jf = jf self.n = n self.imf = imf def den_func(x): return self.imf(x)*x**(-self.jf) self.denominator = scipy.integrate.quad(den_func, self.mmin, self.mmax, **kwargs)[0] self.normfactor = 1 def __call__(self, mass, taper=False, integral_form=False, **kwargs): """ Unclear if integral_form is right...""" if taper: def num_func(x, mass_): tf = (1-(mass_/x)**(1-self.j))**0.5 return self.imf(x)*x**(self.j-self.jf-1) * tf def integrate(lolim, mass_): integral = scipy.integrate.quad(num_func, lolim, self.mmax, args=(mass_,), **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin), mass) else: def num_func(x): return self.imf(x)*x**(self.j-self.jf-1) def integrate(lolim): integral = scipy.integrate.quad(num_func, lolim, self.mmax, **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin)) result = (1-self.j) * mass**(1-self.j) * numerator / self.denominator if integral_form: warnings.warn("The 'integral form' of the Chabrier PMF is not correctly normalized; " "it is just PMF(m) * m") return result * self.normfactor * mass raise ValueError("Integral version not yet computed") else: return result * self.normfactor class McKeeOffner_2CTC(MassFunction): """ 2-component Turbulent Core variant """ default_mmin = 0.033 default_mmax = 3.0 def __init__(self, Rmdot=3.6, j=0.5, jf=3/4., mmin=default_mmin, mmax=default_mmax, imf=chabrierpowerlaw, **kwargs): """ """ super().__init__(mmin=mmin, mmax=mmax) self.j = j self.jf = jf self.Rmdot = Rmdot self.imf = imf def den_func(x): return self.imf(x) * (2/((1+Rmdot**2*x**1.5)**0.5+1)) self.denominator = scipy.integrate.quad(den_func, self.mmin, self.mmax, **kwargs)[0] self.normfactor = 1 def __call__(self, mass, taper=False, integral_form=False, **kwargs): """ Unclear if integral_form is right...""" if taper: def num_func(x, mass_): tf = (1-(mass_/x)**(1-self.j))**0.5 return self.imf(x)*(1./x)**(1-self.j) * (2/((1+self.Rmdot**2*x**1.5)**0.5+1)) * tf def integrate(lolim, mass_): integral = scipy.integrate.quad(num_func, lolim, self.mmax, args=(mass_,), **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin), mass) else: def num_func(x): return self.imf(x)*(1./x)**(1-self.j) * (2/((1+self.Rmdot**2*x**1.5)**0.5+1)) def integrate(lolim): integral = scipy.integrate.quad(num_func, lolim, self.mmax, **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin)) result = (1-self.j) * mass**(1-self.j) * numerator / self.denominator if integral_form: warnings.warn("The 'integral form' of the Chabrier PMF is not correctly normalized; " "it is just PMF(m) * m") return result * self.normfactor * mass raise ValueError("Integral version not yet computed") else: return result * self.normfactor ChabrierPMF_IS = McKeeOffner_PMF(j=0, jf=0, ) ChabrierPMF_TC = McKeeOffner_PMF(j=0.5, jf=0.75, ) ChabrierPMF_CA = McKeeOffner_PMF(j=2/3., jf=1.0, ) ChabrierPMF_2CTC = McKeeOffner_2CTC() class McKeeOffner_SalpeterPMF(MassFunction): " special case; above is now generalized to obsolete this " default_mmin = 0.033 default_mmax = 3.0 def __init__(self, j=1, jf=3/4., alpha=2.35, mmin=default_mmin, mmax=default_mmax): super().__init__(mmin=mmin, mmax=mmax) self.alpha = alpha self.j = j self.jf = jf def __call__(self, mass, **kwargs): alpha = (self.alpha-1+self.jf-self.j) fm = 1 - (mass/self.mmax)**(alpha) result = fm * mass**(-((self.alpha-2)+self.jf)) return result SalpeterPMF_IS = McKeeOffner_SalpeterPMF(j=0, jf=0, ) SalpeterPMF_TC = McKeeOffner_SalpeterPMF(j=0.5, jf=0.75, ) SalpeterPMF_CA = McKeeOffner_SalpeterPMF(j=2/3., jf=1.0, ) kroupa = Kroupa() KroupaPMF_IS = McKeeOffner_PMF(j=0, jf=0, imf=kroupa) KroupaPMF_TC = McKeeOffner_PMF(j=0.5, jf=0.75, imf=kroupa) KroupaPMF_CA = McKeeOffner_PMF(j=2/3., jf=1.0, imf=kroupa) KroupaPMF_2CTC = McKeeOffner_2CTC(imf=kroupa) class McKeeOffner_AcceleratingSF_PMF(MassFunction): default_mmin = 0.033 default_mmax = 3.0 def __init__(self, j=1, n=1, jf=3/4., mmin=default_mmin, mmax=default_mmax, tau=1, # current time, Myr tm=0.54, # SF timescale, Myr tf1=0.50, # accretion timescale for a 1-msun star imf=chabrierpowerlaw, **kwargs): """ McKee & Offner 2010 Protostellar Mass Function with an accelerating star formation rate """ super().__init__(mmin=mmin, mmax=mmax) self.j = j self.jf = jf self.n = n self.tau = tau self.tm = tm self.imf = imf self.tf1 = tf1 def den_func(x): return self.imf(x)*(1-np.exp(-self.tf1*x**(1-self.jf)/self.tau))/x self.denominator = self.tau * scipy.integrate.quad(den_func, self.mmin, self.mmax, **kwargs)[0] assert self.denominator > 0 self.normfactor = 1 def __call__(self, mass, tau=None, taper=False, integral_form=False, **kwargs): """ Parameters ---------- tm : float Star formation timescale in Myrs """ if tau is None: tau = self.tau if taper: raise NotImplementedError() # def num_func(x, mass_): # tf = (1-(mass_/x)**(1-self.j))**0.5 # return self.massfunc(x)*x**(self.j-self.jf-1) * tf # def integrate(lolim, mass_): # integral = scipy.integrate.quad(num_func, lolim, self.mmax, args=(mass_,), **kwargs)[0] # return integral # numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin), mass) else: def num_func(x, mass_): tm = self.tf1 * mass_**(1-self.j) / (x**(self.jf-self.j)) return self.imf(x)*x**(self.j-self.jf-1) * np.exp(-tm / tau) def integrate(lolim, mass_): integral = scipy.integrate.quad(num_func, lolim, self.mmax, args=(mass_), **kwargs)[0] return integral numerator = np.vectorize(integrate)(np.where(self.mmin < mass, mass, self.mmin), mass) assert np.all(numerator >= 0) result = self.tf1 * (1-self.j) * mass**(1-self.j) * numerator / self.denominator # it is possible with the time-evolution case to get negative values for high masses; # these should (probably?) just be zero'd result = np.where(mass > self.mmax, np.nan, result) if integral_form: warnings.warn("The 'integral form' of the Chabrier PMF is not correctly normalized; " "it is just PMF(m) * m") return result * self.normfactor * mass raise ValueError("Integral version not yet computed") else: return result * self.normfactor ChabrierPMF_AcceleratingSF_IS = McKeeOffner_AcceleratingSF_PMF(j=0, jf=0, ) ChabrierPMF_AcceleratingSF_TC = McKeeOffner_AcceleratingSF_PMF(j=0.5, jf=0.75, ) ChabrierPMF_AcceleratingSF_CA = McKeeOffner_AcceleratingSF_PMF(j=2/3., jf=1.0, ) #ChabrierPMF_AcceleratingSF_2CTC = McKeeOffner_AcceleratingSF_2CTC()
8,608
35.634043
104
py
imf
imf-master/imf/__init__.py
from .imf import *
19
9
18
py
imf
imf-master/imf/cmf.py
from __future__ import print_function import numpy as np from astropy import units as u from astropy import constants import scipy.stats from . import imf def pn11_mf(tnow=1, mmin=0.01*u.M_sun, mmax=120*u.M_sun, T0=10*u.K, T_mean=7*u.K, L0=10*u.pc, rho0=2e-21*u.g/u.cm**3, MS0=25, beta=0.4, alpham1=1.35, v0=4.9*u.km/u.s, eff=0.26, mean_mol_wt=2.33): """ Padoan & Nordlund IMF - from http://adsabs.harvard.edu/abs/2011ApJ...741L..22P Parameters ---------- tnow : float The time at which to evaluate the mass function in units of the crossing time Does not match their figures yet! """ tcross = (L0/v0).to(u.Myr) # total molecular cloud mass m0 = (4/3.*np.pi*L0**3*rho0).to(u.M_sun) massfunc = imf.Salpeter(alpha=alpham1+1) massfunc.__name__ = 'salpeter' maccr = imf.make_cluster(mcluster=(m0*eff).to(u.M_sun).value, massfunc=massfunc, mmin=mmin.to(u.M_sun).value, mmax=mmax.to(u.M_sun).value, silent=True)*u.M_sun # sigma_squared of log(rho/rho0) sigma_squared = np.log(1+(MS0/2.)**2 * (1+1./beta)**-1) # two positional parameters: x, s such that pdf=lognorm(x,s) # p(x) = const * exp(-(ln(x)/s)^2 / 2) # p(rho/rho0) = const * exp(-(ln(rho/rho0) + sigma_squared/2)^2 / (2*sigma_squared)) # (ln(rho/rho0)+sigma_squared/2)^2/(2*sigma_squared) = ln(x)^2/s^2 /2 # s^2 = sigma_squared # ln(x) = ln(rho/rho0)+sigma_squared/2 # x = np.exp(ln(rho/rho0)+sigma_squared/2) # ln(rho/rho0) = ln(x) - sigma_squared/2 # rho = rho0 * exp(ln(x) - sigma_squared/2) s = sigma_squared**0.5 pdf_func = scipy.stats.lognorm(s) x = pdf_func.rvs(len(maccr)) rho = rho0 * np.exp(np.log(x) - sigma_squared/2) sigma = (1+beta**-1)**-0.5 * MS0/2. # core birthdays: "Finally, we associate a random age to each core, # assuming for simplicity that the SFR is uniform over time and independent # of core mass." # Confirmed: "what I did was to assume constant star formation rate during # the whole period t=t_0, where t_0 is the crossing time, and also the time # when the mass functions are computed. So cores are formed with a uniform # distribution (of birth times) between 0 and t_0." birthday = np.random.random(len(maccr)) * tcross born = birthday < tnow*tcross b = 4.-(3./alpham1) a = (b-1)/2. taccr = (tcross * sigma**((4.-4.*a)/(3.-2.*a)) * (maccr/m0)**((1-a)/(3-2*a))).to(u.Myr) print(("taccr=[{0} - {1}]".format(taccr.to(u.Myr).min(), taccr.to(u.Myr).max()))) c_s = ((constants.k_B * T_mean / (mean_mol_wt*constants.m_p))**0.5).to(u.km/u.s) print(("c_s = {0}".format(c_s))) mbe = (1.182 * c_s**3 / (constants.G**1.5 * rho**0.5)).to(u.M_sun) tbe = (taccr * (maccr/mbe)**(-1/3.)).to(u.s) tff = ((3*np.pi/(32*constants.G*rho))**0.5).to(u.s) mmax = (maccr * ((tbe+tff)/taccr)**3).to(u.M_sun) # tnow = number of crossing times age = (tnow*tcross-birthday) mnow = ((age/taccr)**3 * maccr).to(u.M_sun) prestellar = age < tbe+tff ltbe = maccr < mbe stellar = (~prestellar) & (~ltbe) forming = age < taccr m_f = np.vstack([mmax.value, maccr.value]).min(axis=0)*u.M_sun mnow[mnow > m_f] = m_f[mnow > m_f] will_collapse = maccr > mbe/2. # We assume that cores that do not reach their BE mass are seen only during # their formation time, taccr, notseen = ltbe & ~forming core_mass = mnow[born & (~notseen) & (~stellar)].sum() stellar_mass = mnow[stellar].sum() print(("{0} of {1} have mass greater than final at t={2}." " {3} are unborn. {4} are stellar. " "{7} are not seen ({8:0.02f}%) because they are older than " "one accretion time and have M<M_BE. " "The cloud mass is {9}. " "The CFE={5}" " and SFE={6}".format((mnow > m_f).sum(), len(mnow), tnow*tcross, np.sum(~born), np.sum(stellar), (core_mass/m0).decompose().value, (stellar_mass/m0).decompose().value, notseen.sum(), (notseen.sum()/float(notseen.size))*100, m0 ))) return (mnow[born], m_f[born], will_collapse[born], maccr[born], mbe[born], mmax[born], forming[born]) def test_pn11(nreal=5, nbins=50, **kwargs): mnow, mf, wc, maccr, mbe, mmax, forming = pn11_mf(**kwargs) import pylab as pl pl.figure(1).clf() ltbe = maccr < mbe # We assume that cores that do not reach their BE mass are seen only during # their formation time, taccr, notseen = ltbe & ~forming toplot = (mnow > 0.05*u.M_sun) & (~notseen) gtmax = maccr > mmax btw = (~gtmax) & (~ltbe) pl.loglog(mnow[toplot & gtmax], (maccr/mnow)[toplot & gtmax], 'kd', markerfacecolor='none') print(("{0} have maccr<mbe and m>0.05".format((toplot & ltbe).sum()))) pl.loglog(mnow[toplot & ltbe], (maccr/mnow)[toplot & ltbe], 'r.', markersize=2, alpha=0.5, markerfacecolor='none') pl.loglog(mnow[toplot & btw], (maccr/mnow)[toplot & btw], 'b+', markerfacecolor='none') pl.gca().set_ylim(0.5, 500) pl.gca().set_xlim(0.05, 50) pl.ylabel("$m_{accr}/m$") pl.xlabel("$m$, i.e. $m_{now}$") pl.figure(2).clf() pl.hist((maccr/mnow)[mnow > 0.1*u.M_sun], bins=np.logspace(0, 2, 12), histtype='step', color='k', log=True) pl.hist((maccr/mnow)[mnow > (mbe/2.)], bins=np.logspace(0, 2, 12), histtype='step', linestyle='dashed', color='k', log=True) pl.gca().set_xscale('log') pl.ylabel("$N(m_{accr}/m)$") pl.xlabel("$m_{accr}/m$") pl.figure(3).clf() pl.loglog(mnow[toplot & gtmax], (mnow/mbe)[toplot & gtmax], 'kd', markerfacecolor='none') pl.loglog(mnow[toplot & ltbe], (mnow/mbe)[toplot & ltbe], 'r.', markersize=2, markerfacecolor='none', alpha=0.5) pl.loglog(mnow[toplot & btw], (mnow/mbe)[toplot & btw], 'b+', markerfacecolor='none') ct, bn = np.histogram(mnow[(mnow > mbe/2.) & toplot & (maccr < mbe)], bins=np.logspace(np.log10(0.05), np.log10(20))) ctall, bn = np.histogram(mnow[(mnow > mbe/2.) & toplot], bins=np.logspace(np.log10(0.05), np.log10(20))) bbn = (bn[1:]+bn[:-1])/2. pl.loglog(bbn, ct/ctall.astype('float'), 'k-') pl.ylabel("$m_{now}/m_{BE}$") pl.xlabel("$m$, i.e. $m_{now}$") pl.figure(4) pl.clf() pl.hist(mnow, bins=np.logspace(-2, 2, nbins), histtype='step', log=True, edgecolor='k', label='$m$') pl.hist(mnow[wc], bins=np.logspace(-2.02, 1.98, nbins), histtype='step', log=True, edgecolor='b', facecolor=(0, 0, 1, 0.25), label='$m>m_{BE}/2$') pl.hist(mbe, bins=np.logspace(-1.98, 2.02, nbins), histtype='step', log=True, linestyle='dashed', color='g', label='$m_{BE}$') pl.hist(mmax, bins=np.logspace(-1.96, 2.04, nbins), histtype='step', log=True, linestyle='dashed', color='m', label='$m_{max}$') pl.gca().set_xscale('log') pl.legend(loc='best') pl.ylabel("$N(M)$") pl.xlabel("$m$, i.e. $m_{now}$") many_realizations = [pn11_mf(**kwargs) for ii in range(nreal)] mnow_many = np.hstack([x.value for x, y, z, w, v, s, t in many_realizations]).ravel() counts, bins = np.histogram(mnow_many.ravel(), bins=np.logspace(-2, 2, nbins*nreal)) bbins = (bins[:-1]+bins[1:])/2. ok = np.log(counts) > 0 ppars = np.polyfit(np.log(bbins)[ok], np.log(counts)[ok], 1) pl.plot(bbins, np.exp(ppars[1]) * bbins**ppars[0], 'r') pl.figure(5).clf() pl.plot(maccr[toplot], mbe[toplot], 'k,') return mnow, mf, wc, maccr, mbe, mmax def hc13_mf(mass, sizescale, n17=3.8, alpha_ct=0.75, mean_mol_wt=2.33, V0=0.8*u.km/u.s, meandens=5000*u.cm**-3, temperature=10*u.K, eta=0.45, b_forcing=0.4, Mach=6): """ Equation 21 of Hennebelle & Chabrier 2013 Parameters ---------- mass : np.array Masses at which to evaluate the PDF sizescale : pc equivalent The size of the clump (I think - extremely difficult to find this) n17 : float The "n" value in Equation 17, quoted to be 3.8 shortly afterward in the text alpha_ct : float "a dimensionless coefficient of the order of a few" I derived 0.75 from equation 9 V0 : float u0 * 0.8 km/s according to the bottom of page 6, under eqn 36, which references eqn 16 eta : None or float derived from Equation 17, but can be specified directly b_forcing : float The Forcing Parameter `b` from equation 4 Mach : float Mach number """ rho_bar = meandens * mean_mol_wt * constants.m_p c_s = ((constants.k_B * temperature / (mean_mol_wt*constants.m_p))**0.5).to(u.km/u.s) sigma = (np.log(1+b_forcing**2 * Mach**2))**0.5 if eta is None: # eqn 17 eta = (n17 - 3.)/2. alpha_g = 3/5. # for a uniform density fluctuation # eqn 9 phit = 2 * alpha_ct * (24/np.pi**2/alpha_g) # dimensionless geometrical factor of the order of unity # For a sphere, becoMes: aJ = np.pi**2.5/6. # a geometrical factor, typically of the order of 4pi/3 Cm = 4*np.pi/3 # eqn 13 MJ0 = (aJ / Cm * c_s**3 * constants.G**-1.5 * rho_bar**-0.5).to(u.M_sun) # eqn 14 lambdaJ0 = (np.pi**0.5 * c_s / Cm * (constants.G*rho_bar)**-0.5).to(u.pc) # Eqn 7 of Paper I # delta = np.log(rho/rho_bar # R = (mass/rho_bar)**(1/3.) * np.exp(-delta/3.) / lambdaJ0 Rtwiddle = (sizescale / lambdaJ0).to(u.dimensionless_unscaled) Mtwiddle = (mass / MJ0).to(u.dimensionless_unscaled) # eqn 20 Mstar = (3**-0.5 * V0/c_s * (lambdaJ0/(u.pc))**eta).to(u.dimensionless_unscaled) # after eqn 21 N0 = rho_bar / MJ0 # PROBLEM: N0 is defined to be rho_bar / MJ0, but that is a dimensional # quantity with units cm^-3. This is a contradiction that means some # definition here is wrong. # eqn 21 N = (2./phit * N0 * Rtwiddle**-6 * (1 + (1-eta)*Mstar**2*Rtwiddle**(2*eta)) / (1+(2*eta+1)*Mstar**2*Rtwiddle**(2*eta)) * (Mtwiddle/Rtwiddle**3)**(-1-1/(2*sigma**2)*np.log(Mtwiddle/Rtwiddle**3)) * np.exp(sigma**2/8.) * ((2*np.pi)**0.5 * sigma) ).to(u.dimensionless_unscaled) return N def test_hc13(): masses = np.logspace(-2, 2, 100)*u.M_sun sizescale = 10*u.pc return hc13_mf(mass=masses, sizescale=sizescale)
10,822
37.379433
89
py
imf
imf-master/imf/tests/test_imf.py
import pytest import numpy as np import itertools from .. import imf from ..imf import kroupa, chabrierpowerlaw extra_massfunc_kwargs = {'schecter': {'m1': 1.0}, 'modified_schecter': {'m1': 1.0}, 'chabrierpowerlaw': {'mmid': 0.5}, } @pytest.mark.parametrize(('inp', 'out', 'rtol', 'atol'), [(0.05, 5.6159, 1e-3, 1e-3), (1.5, 0.0359, 1e-4, 1e-4), (1.0, 0.0914, 1e-4, 1e-4), (3.0, 0.0073, 1e-4, 1e-4), (1, 0.0914, 1e-4, 1e-4), (3, 0.0073, 1e-4, 1e-4)]) def test_kroupa_val(inp, out, rtol, atol): kroupa = imf.Kroupa() np.testing.assert_allclose(kroupa(inp), out, rtol=rtol, atol=atol) np.testing.assert_allclose(imf.kroupa(inp), out, rtol=rtol, atol=atol) @pytest.mark.parametrize('massfunc', imf.massfunctions.keys()) def test_mmax(massfunc): """ Regression test for issue #4 """ if (not hasattr(imf.get_massfunc(massfunc), 'mmin')): pytest.skip("{0} doesn't have mmin defined".format(massfunc)) extra_kwargs = extra_massfunc_kwargs.get(massfunc, {}) c = imf.make_cluster(10000, mmax=1.001, mmin=0.01, massfunc=massfunc, **extra_kwargs) assert c.max() <= 1.001 @pytest.mark.parametrize(('mlow', 'mhigh'), itertools.product((0.01, 0.08, 0.1, 0.5, 1.0, 0.03), (0.02, 0.08, 0.4, 0.5, 1.0, 120))) def test_kroupa_integral(mlow, mhigh): if mlow >= mhigh: pytest.skip("mmin >= mmax") num = kroupa.integrate(mlow, mhigh, numerical=True)[0] anl = kroupa.integrate(mlow, mhigh, numerical=False)[0] np.testing.assert_almost_equal(num, anl) if num != 0: assert anl != 0 @pytest.mark.parametrize(('mlow', 'mhigh'), itertools.product((0.01, 0.08, 0.1, 0.5, 1.0, 0.03), (0.02, 0.08, 0.4, 0.5, 1.0, 120))) def test_kroupa_mintegral(mlow, mhigh): if mlow >= mhigh: pytest.skip("mmin >= mmax") num = kroupa.m_integrate(mlow, mhigh, numerical=True)[0] anl = kroupa.m_integrate(mlow, mhigh, numerical=False)[0] print("{0} {1} {2:0.3f} {3:0.3f}".format(mlow, mhigh, num, anl)) np.testing.assert_almost_equal(num, anl) if num != 0: assert anl != 0 @pytest.mark.parametrize(('mlow', 'mhigh'), itertools.product((0.033, 0.01, 0.08, 0.1, 0.5, 1.0, 0.03), (0.02, 0.05, 0.08, 0.4, 0.5, 1.0, 120))) def test_chabrier_integral(mlow, mhigh): if mlow >= mhigh: pytest.skip("mmin >= mmax") num = chabrierpowerlaw.integrate(mlow, mhigh, numerical=True)[0] anl = chabrierpowerlaw.integrate(mlow, mhigh, numerical=False)[0] print("{0} {1} {2:0.3f} {3:0.3f}".format(mlow, mhigh, num, anl)) np.testing.assert_almost_equal(num, anl) # for mlow in (0.01, 0.08, 0.1, 0.5, 1.0): # for mhigh in (0.02, 0.08, 0.4, 0.5, 1.0): # try: # num = chabrierpowerlaw.m_integrate(mlow, mhigh, numerical=True)[0] # anl = chabrierpowerlaw.m_integrate(mlow, mhigh, numerical=False)[0] # except ValueError: # continue # print("{0} {1} {2:0.3f} {3:0.3f}".format(mlow, mhigh, num, anl)) # np.testing.assert_almost_equal(num, anl) def test_make_cluster(): cluster = imf.make_cluster(1000) assert np.abs(sum(cluster) - 1000 < 100) def test_kroupa_inverses(): assert np.abs(imf.inverse_imf(0, massfunc=imf.Kroupa, mmin=0.01) - 0.01) < 2e-3 assert np.abs(imf.inverse_imf(0, massfunc=imf.Kroupa(mmin=0.01)) - 0.01) < 2e-3 assert np.abs(imf.inverse_imf(1, massfunc=imf.Kroupa, mmax=200) - 200) < 1 assert np.abs(imf.inverse_imf(1, massfunc=imf.Kroupa(mmax=200)) - 200) < 1 def test_cannot_override_mmin_mmax_instance(): with pytest.raises(ValueError) as ex: imf.inverse_imf(0, massfunc=imf.Kroupa(), mmin=0.01) assert "mmin was specified" in str(ex.value) with pytest.raises(ValueError) as ex: imf.inverse_imf(0, massfunc=imf.Kroupa(), mmax=200) assert "mmax was specified" in str(ex.value) with pytest.raises(ValueError) as ex: imf.get_massfunc(massfunc=imf.Kroupa(), mmin=0.01) assert "mmin was specified" in str(ex.value) with pytest.raises(ValueError) as ex: imf.get_massfunc(imf.Kroupa(), mmax=200) assert "mmax was specified" in str(ex.value) mf = imf.get_massfunc(imf.Kroupa, mmax=200) assert mf.mmax == 200 mf = imf.get_massfunc(imf.Kroupa, mmin=0.01) assert mf.mmin == 0.01 @pytest.mark.parametrize(('inp', 'out', 'rtol', 'atol'), [(0.05, 5.6159, 1e-3, 1e-3), (1.5, 0.0359, 1e-4, 1e-4), (1.0, 0.0914, 1e-4, 1e-4), (3.0, 0.0073, 1e-4, 1e-4), (1, 0.0914, 1e-4, 1e-4), (3, 0.0073, 1e-4, 1e-4)]) def test_kroupa_val_unchanged(inp, out, rtol, atol): # regression: make sure that imf.kroupa = imf.Kroupa kroupa = imf.Kroupa() np.testing.assert_allclose(kroupa(inp), out, rtol=rtol, atol=atol) np.testing.assert_allclose(imf.kroupa(inp), out, rtol=rtol, atol=atol) np.testing.assert_allclose(kroupa(inp), imf.kroupa(inp))
5,462
37.202797
89
py
imf
imf-master/imf/tests/test_distributions.py
import numpy as np import scipy.interpolate from .. import distributions as D np.random.seed(1) def sampltest(distr, left=None, right=None, bounds=None): # check that mean and stddev from the generated sample # match what we get from integrating the PDF def FF1(x): return distr.pdf(x) * x def FF2(x): return distr.pdf(x) * x**2 if left is None: left = 0 if right is None: right = np.inf if bounds is None: mom1, _ = scipy.integrate.quad(FF1, left, right) mom2, _ = scipy.integrate.quad(FF2, left, right) else: mom1, mom2 = 0, 0 for curb in bounds: cmom1, _ = scipy.integrate.quad(FF1, curb[0], curb[1]) cmom2, _ = scipy.integrate.quad(FF2, curb[0], curb[1]) mom1 += cmom1 mom2 += cmom2 std = np.sqrt(mom2 - mom1**2) assert (mom2 > mom1**2) N = int(1e6) samps = distr.rvs(N) assert ((samps.mean() - mom1) < 5 * std / np.sqrt(N)) assert ((samps.std() - std) < 20 * std / np.sqrt(2 * (N - 1))) def ppftest(distr): # test that ppf is inverse of cdf xs = np.random.uniform(0, 1, size=100) eps = 1e-5 assert (np.all(np.abs(distr.cdf(distr.ppf(xs)) - xs) < eps)) # test on scalar assert (np.abs(distr.cdf(distr.ppf(xs[0])) - xs[0]) < eps) assert (np.isnan(distr.ppf(-0.1))) assert (np.isnan(distr.ppf(1.1))) def test_lognorm(): ln = D.LogNormal(1, 1) ln.pdf(1.) ln.cdf(1) ln.rvs(1000) ppftest(ln) sampltest(ln) for i in range(10): N = 100000 mean = np.random.uniform(0.1, 10) sig = np.random.uniform(0.1, 10) ln2 = D.LogNormal(mean, sig) samp = ln2.rvs(N) # check that the means and sigmas are correct assert (np.abs(np.log(samp).mean() - np.log(mean)) < 0.01 * sig) assert (np.abs(np.log(samp).std() - sig) < 0.01 * sig) def test_broken_plaw(): ln = D.BrokenPowerLaw([-2, -1.1, -3], [0.1, 1, 2, 100]) ln.pdf(1.) ln.cdf(1) ln.rvs(1000) ppftest(ln) sampltest(ln, 0.05, 120, bounds=[[0.05, 1], [1, 2], [2, 120]]) # test values in each range assert (np.abs(ln.ppf(ln.cdf(0.5)) - 0.5) < 1e-5) assert (np.abs(ln.ppf(ln.cdf(1.5)) - 1.5) < 1e-5) assert (np.abs(ln.ppf(ln.cdf(2.5)) - 2.5) < 1e-5) def test_distr(): ln = D.TruncatedLogNormal(1, 1, 2, 3) ln.pdf(1.) ln.cdf(1) ln.rvs(1000) ppftest(ln) sampltest(ln, 1, 4) ln = D.PowerLaw(-2, 2, 6) ln.pdf(1.) ln.cdf(1) ln.rvs(1000) ppftest(ln) sampltest(ln, 1, 7) def test_composite(): ln = D.CompositeDistribution([ D.TruncatedLogNormal(1, 1, 2, 3), D.PowerLaw(-2, 3, 4), D.TruncatedLogNormal(1, 1, 4, 5), D.PowerLaw(-3.5, 5, np.inf) ]) ln.pdf(2.5) ln.cdf(2.5) ln.rvs(1000) ppftest(ln) # test values in each break assert (np.abs(ln.ppf(ln.cdf(2.5)) - 2.5) < 1e-5) assert (np.abs(ln.ppf(ln.cdf(3.5)) - 3.5) < 1e-5) assert (np.abs(ln.ppf(ln.cdf(4.5)) - 4.5) < 1e-5) assert (np.abs(ln.ppf(ln.cdf(5.5)) - 5.5) < 1e-5) sampltest(ln, 1, np.inf, bounds=[[1, 3], [3, 4], [4, 5], [5, np.inf]]) ln1 = D.CompositeDistribution([ D.TruncatedLogNormal(1, 1, 2, 3), D.PowerLaw(-2, 3, 4), ]) # check the exact edges work assert (ln1.cdf(4) == 1) assert (ln1.cdf(2) == 0) def test_bounds(): left, right = 1, 2 tleft, tright = 0.5, 3 ln = D.TruncatedLogNormal(1, 1, left, right) assert (ln.pdf(tleft) == 0) assert (ln.pdf(tright) == 0) assert (ln.cdf(tleft) == 0) assert (ln.cdf(tright) == 1) ln = D.PowerLaw(-3, left, right) assert (ln.pdf(tleft) == 0) assert (ln.pdf(tright) == 0) assert (ln.cdf(tleft) == 0) assert (ln.cdf(tright) == 1) ln = D.BrokenPowerLaw( [-2, -1.1, -3], [left, .6 * left + .3 * right, .3 * left + .6 * right, right]) assert (ln.pdf(tleft) == 0) assert (ln.pdf(tright) == 0) assert (ln.cdf(tleft) == 0) assert (ln.cdf(tright) == 1) ln = D.CompositeDistribution([ D.TruncatedLogNormal(1, 1, left, .75 * left + .25 * right), D.PowerLaw(-2, .75 * left + .25 * right, .5 * left + .5 * right), D.TruncatedLogNormal(1, 1, .5 * left + .5 * right, .25 * left + .75 * right), D.PowerLaw(-2, .25 * left + .75 * right, right) ]) assert (ln.pdf(tleft) == 0) assert (ln.pdf(tright) == 0) assert (ln.cdf(tleft) == 0) assert (ln.cdf(tright) == 1) def integralcheck(distr, left, x, val): I, EI = scipy.integrate.quad(lambda y: distr.pdf(y), left, x) assert (np.abs(val - I) < 1e-6) def integralcheck_many(distr, left, right): integralcheck(distr, left, right, 1) N = 100 xs = np.random.uniform(left, right, size=N) for x in xs: integralcheck(distr, left, x, distr.cdf(x)) def test_integral(): # test that the numerically integrated pdf is within 3 sigma of 1 # for different kind of pdfs left, right = 2, 3 distrs = [ D.TruncatedLogNormal(1, 1, left, right), D.PowerLaw(-2, left, right), D.BrokenPowerLaw( [-2, -1.1, -3], [left, .6 * left + .3 * right, .3 * left + .6 * right, right]), D.CompositeDistribution([ D.TruncatedLogNormal(1, 1, left, .75 * left + .25 * right), D.PowerLaw(-2, .75 * left + .25 * right, .5 * left + .5 * right), D.TruncatedLogNormal(1, 1, .5 * left + .5 * right, .25 * left + .75 * right), D.PowerLaw(-2, .25 * left + .75 * right, right) ]) ] for curd in distrs: integralcheck_many(curd, left, right)
5,757
28.228426
77
py
imf
imf-master/imf/tests/__init__.py
0
0
0
py
imaging_MLPs
imaging_MLPs-master/ImageNet/config.py
#Path to original dataset original_path='/workspace/media/hdd1/image_net_mini/imagenet-mini/train/' #Path to processed dataset data_path='/workspace/media/hdd1/image_net_mini/dataset2/' #Path to store trained models models_path= "/workspace/media/hdd1/image_net_mini/trained_networks/" #Path to store tensor board data logs_path= './logs/'
342
30.181818
73
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/linear_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbedding(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchExpansion(nn.Module): def __init__(self, dim_scale, channel_dim, img_channels, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, dim_scale**2* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int, f_hidden: int, ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), nn.Linear(num_patches, num_patches), Rearrange("b c w h -> b c h w"), nn.Linear(num_patches, num_patches), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels*f_hidden) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class Linear_Mixer(nn.Module): def __init__( self, img_size: int = 256, img_channels: int = 3, patch_size: int = 4, embed_dim: int = 140, num_layers: int = 19, f_hidden: int = 4, ): super().__init__() self.patch_embed = PatchEmbedding(patch_size, embed_dim, img_channels) layers = [ Mixer(img_size//patch_size, embed_dim, f_hidden) for _ in range(num_layers)] self.mixer_layers = nn.Sequential(*layers) self.patch_expand = PatchExpansion(patch_size, embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: x = self.patch_embed(x) x = self.mixer_layers(x) x = self.patch_expand(x) return x
3,644
26.613636
127
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/original_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbeddings(nn.Module): def __init__( self, patch_size: int, hidden_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=hidden_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b (h w) c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchEmbeddings_transpose(nn.Module): def __init__( self, patch_size: int, hidden_dim: int, channels: int, d: int ): super().__init__() self.proj_transpose = nn.Sequential( Rearrange("b (h w) c -> b c h w", h=d), nn.ConvTranspose2d( in_channels=hidden_dim, out_channels=channels, kernel_size=patch_size, stride=patch_size ) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj_transpose(x) class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class MixerBlock(nn.Module): def __init__( self, num_patches: int, num_channels: int, tokens_hidden_dim: int, channels_hidden_dim: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b p c -> b c p"), MLPBlock(num_patches, tokens_hidden_dim), Rearrange("b c p -> b p c") ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, channels_hidden_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class Original_Mixer(nn.Module): def __init__( self, image_size: int = 256, channels: int = 3, patch_size: int = 4, num_layers: int = 8, hidden_dim: int = 128, tokens_hidden_dim: int = 96, channels_hidden_dim: int = 256 ): super().__init__() num_patches = (image_size // patch_size) ** 2 d=(image_size-patch_size)//patch_size + 1 self.embed = PatchEmbeddings(patch_size, hidden_dim, channels) layers = [ MixerBlock( num_patches=num_patches, num_channels=hidden_dim, tokens_hidden_dim=tokens_hidden_dim, channels_hidden_dim=channels_hidden_dim ) for _ in range(num_layers) ] self.layers = nn.Sequential(*layers) self.norm = nn.LayerNorm(hidden_dim) self.embed_transpose = PatchEmbeddings_transpose(patch_size, hidden_dim, channels, d) def forward(self, x: torch.Tensor) -> torch.Tensor: b, c, h, w = x.shape x = self.embed(x) x = self.layers(x) x = self.norm(x) x = self.embed_transpose(x) return x
3,674
26.840909
93
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/img2img_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbedding(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchExpansion(nn.Module): def __init__(self, dim_scale, channel_dim, img_channels, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, dim_scale**2* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int, f_hidden: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c w h -> b c h w"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels*f_hidden) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class Img2Img_Mixer(nn.Module): def __init__( self, img_size: int = 256, img_channels: int = 3, patch_size: int = 4, embed_dim: int = 128, num_layers: int = 16, f_hidden: int = 4, ): super().__init__() self.patch_embed = PatchEmbedding(patch_size, embed_dim, img_channels) layers = [ Mixer(img_size//patch_size, embed_dim, f_hidden) for _ in range(num_layers)] self.mixer_layers = nn.Sequential(*layers) self.patch_expand = PatchExpansion(patch_size, embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: x = self.patch_embed(x) x = self.mixer_layers(x) x = self.patch_expand(x) return x
3,618
27.054264
127
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/vit.py
''' This code is modified from https://github.com/facebookresearch/convit. To adapt the vit/convit to image reconstruction, variable input sizes, and patch sizes for both spatial dimensions. ''' import torch import torch.nn as nn from functools import partial import torch.nn.functional as F from timm.models.helpers import load_pretrained from timm.models.layers import DropPath, to_2tuple, trunc_normal_ class Mlp(nn.Module): def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): super().__init__() out_features = out_features or in_features hidden_features = hidden_features or in_features self.fc1 = nn.Linear(in_features, hidden_features) self.act = act_layer() self.fc2 = nn.Linear(hidden_features, out_features) self.drop = nn.Dropout(drop) self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def forward(self, x): x = self.fc1(x) x = self.act(x) x = self.drop(x) x = self.fc2(x) x = self.drop(x) return x class GPSA(nn.Module): def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., locality_strength=1., use_local_init=True, grid_size=None): super().__init__() self.num_heads = num_heads self.dim = dim head_dim = dim // num_heads self.scale = qk_scale or head_dim ** -0.5 self.q = nn.Linear(dim, dim, bias=qkv_bias) self.k = nn.Linear(dim, dim, bias=qkv_bias) self.v = nn.Linear(dim, dim, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.pos_proj = nn.Linear(3, num_heads) self.proj_drop = nn.Dropout(proj_drop) self.locality_strength = locality_strength self.gating_param = nn.Parameter(1*torch.ones(self.num_heads)) self.apply(self._init_weights) if use_local_init: self.local_init(locality_strength=locality_strength) self.current_grid_size = grid_size def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def get_attention(self, x): B, N, C = x.shape k = self.k(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) q = self.q(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) pos_score = self.pos_proj(self.rel_indices).expand(B, -1, -1,-1).permute(0,3,1,2) patch_score = (q @ k.transpose(-2, -1)) * self.scale patch_score = patch_score.softmax(dim=-1) pos_score = pos_score.softmax(dim=-1) gating = self.gating_param.view(1,-1,1,1) attn = (1.-torch.sigmoid(gating)) * patch_score + torch.sigmoid(gating) * pos_score attn = attn / attn.sum(dim=-1).unsqueeze(-1) attn = self.attn_drop(attn) return attn def get_attention_map(self, x, return_map = False): attn_map = self.get_attention(x).mean(0) # average over batch distances = self.rel_indices.squeeze()[:,:,-1]**.5 dist = torch.einsum('nm,hnm->h', (distances, attn_map)) dist /= distances.size(0) if return_map: return dist, attn_map else: return dist def local_init(self, locality_strength=1.): self.v.weight.data.copy_(torch.eye(self.dim)) locality_distance = 1 #max(1,1/locality_strength**.5) kernel_size = int(self.num_heads**.5) center = (kernel_size-1)/2 if kernel_size%2==0 else kernel_size//2 for h1 in range(kernel_size): for h2 in range(kernel_size): position = h1+kernel_size*h2 self.pos_proj.weight.data[position,2] = -1 self.pos_proj.weight.data[position,1] = 2*(h1-center)*locality_distance self.pos_proj.weight.data[position,0] = 2*(h2-center)*locality_distance self.pos_proj.weight.data *= locality_strength def get_rel_indices(self, ): H, W = self.current_grid_size N = H*W rel_indices = torch.zeros(1, N, N, 3) indx = torch.arange(W).view(1,-1) - torch.arange(W).view(-1, 1) indx = indx.repeat(H, H) indy = torch.arange(H).view(1,-1) - torch.arange(H).view(-1, 1) indy = indy.repeat_interleave(W, dim=0).repeat_interleave(W, dim=1) indd = indx**2 + indy**2 rel_indices[:,:,:,2] = indd.unsqueeze(0) rel_indices[:,:,:,1] = indy.unsqueeze(0) rel_indices[:,:,:,0] = indx.unsqueeze(0) device = self.v.weight.device self.rel_indices = rel_indices.to(device) def forward(self, x): B, N, C = x.shape if not hasattr(self, 'rel_indices') or self.rel_indices.size(1)!=N: self.get_rel_indices() attn = self.get_attention(x) v = self.v(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x) return x class MHSA(nn.Module): def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., grid_size=None): super().__init__() self.num_heads = num_heads head_dim = dim // num_heads self.scale = qk_scale or head_dim ** -0.5 self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop) self.apply(self._init_weights) self.current_grid_size = grid_size def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def get_attention_map(self, x, return_map = False): self.get_rel_indices() B, N, C = x.shape qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) q, k, v = qkv[0], qkv[1], qkv[2] attn_map = (q @ k.transpose(-2, -1)) * self.scale attn_map = attn_map.softmax(dim=-1).mean(0) # average over batch distances = self.rel_indices.squeeze()[:,:,-1]**.5 dist = torch.einsum('nm,hnm->h', (distances, attn_map)) dist /= distances.size(0) if return_map: return dist, attn_map else: return dist def get_rel_indices(self, ): H, W = self.current_grid_size N = H*W rel_indices = torch.zeros(1, N, N, 3) indx = torch.arange(W).view(1,-1) - torch.arange(W).view(-1, 1) indx = indx.repeat(H, H) indy = torch.arange(H).view(1,-1) - torch.arange(H).view(-1, 1) indy = indy.repeat_interleave(W, dim=0).repeat_interleave(W, dim=1) indd = indx**2 + indy**2 rel_indices[:,:,:,2] = indd.unsqueeze(0) rel_indices[:,:,:,1] = indy.unsqueeze(0) rel_indices[:,:,:,0] = indx.unsqueeze(0) device = self.qkv.weight.device self.rel_indices = rel_indices.to(device) def forward(self, x): B, N, C = x.shape qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) q, k, v = qkv[0], qkv[1], qkv[2] attn = (q @ k.transpose(-2, -1)) * self.scale attn = attn.softmax(dim=-1) attn = self.attn_drop(attn) x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x) return x class Block(nn.Module): def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, use_gpsa=True, **kwargs): super().__init__() self.norm1 = norm_layer(dim) self.use_gpsa = use_gpsa if self.use_gpsa: self.attn = GPSA(dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop, **kwargs) else: self.attn = MHSA(dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop, **kwargs) self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) def forward(self, x, grid_size): self.attn.current_grid_size = grid_size x = x + self.drop_path(self.attn(self.norm1(x))) x = x + self.drop_path(self.mlp(self.norm2(x))) return x class PatchEmbed(nn.Module): """ Image to Patch Embedding, from timm """ def __init__(self, patch_size, in_chans, embed_dim): super().__init__() self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size) self.apply(self._init_weights) def forward(self, x): x = self.proj(x) return x def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) class VisionTransformer(nn.Module): """ Vision Transformer """ def __init__(self, avrg_img_size=320, patch_size=16, in_chans=1, embed_dim=64, depth=8, num_heads=9, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0., norm_layer=nn.LayerNorm, global_pool=None, gpsa_interval=[-1, -1], locality_strength=1., use_pos_embed=True): super().__init__() self.depth = depth embed_dim *= num_heads self.num_features = embed_dim # num_features for consistency with other models self.locality_strength = locality_strength self.use_pos_embed = use_pos_embed if isinstance(avrg_img_size, int): img_size = to_2tuple(avrg_img_size) if isinstance(patch_size, int): self.patch_size = to_2tuple(patch_size) else: self.patch_size = patch_size self.in_chans = in_chans self.patch_embed = PatchEmbed( patch_size=self.patch_size, in_chans=in_chans, embed_dim=embed_dim) self.pos_drop = nn.Dropout(p=drop_rate) if self.use_pos_embed: self.pos_embed = nn.Parameter( torch.zeros(1, embed_dim, img_size[0] // self.patch_size[0], img_size[1] // self.patch_size[1]) ) trunc_normal_(self.pos_embed, std=.02) dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)] # stochastic depth decay rule self.blocks = nn.ModuleList([ Block( dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer, use_gpsa=True, locality_strength=locality_strength) if i>=gpsa_interval[0]-1 and i<gpsa_interval[1] else Block( dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer, use_gpsa=False,) for i in range(depth)]) self.norm = norm_layer(embed_dim) # head self.feature_info = [dict(num_chs=embed_dim, reduction=0, module='head')] self.head = nn.Linear(self.num_features, in_chans*self.patch_size[0]*self.patch_size[1]) def seq2img(self, x, img_size): """ Transforms sequence back into image space, input dims: [batch_size, num_patches, channels] output dims: [batch_size, channels, H, W] """ x = x.view(x.shape[0], x.shape[1], self.in_chans, self.patch_size[0], self.patch_size[1]) x = x.chunk(x.shape[1], dim=1) x = torch.cat(x, dim=4).permute(0,1,2,4,3) x = x.chunk(img_size[0]//self.patch_size[0], dim=3) x = torch.cat(x, dim=4).permute(0,1,2,4,3).squeeze(1) return x self.head.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) @torch.jit.ignore def no_weight_decay(self,): return {'pos_embed'} def get_head(self,): return self.head def reset_head(self,): self.head = nn.Linear(self.num_features, in_chans*self.patch_size[0]*self.patch_size[1]) def forward_features(self, x, k=None): x = self.patch_embed(x) _, _, H, W = x.shape if self.use_pos_embed: pos_embed = F.interpolate(self.pos_embed, size=[H, W], mode='bilinear', align_corners = False) x = x + pos_embed x = x.flatten(2).transpose(1, 2) x = self.pos_drop(x) for u, blk in enumerate(self.blocks): x = blk(x, (H, W)) if k is not None and u == k: self.attention_map = blk.attn.get_attention_map(x, return_map = True) x = self.norm(x) return x def forward(self, x, k=None): _, _, H, W = x.shape x = self.forward_features(x, k) x = self.head(x) x = self.seq2img(x, (H, W)) return x
15,082
39.007958
186
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/recon_net.py
import torch.nn as nn import torch.nn.functional as F from math import ceil, floor class ReconNet(nn.Module): def __init__(self, net): super().__init__() self.net = net def pad(self, x): _, _, h, w = x.shape hp, wp = self.net.patch_size f1 = ( (wp - w % wp) % wp ) / 2 f2 = ( (hp - h % hp) % hp ) / 2 wpad = [floor(f1), ceil(f1)] hpad = [floor(f2), ceil(f2)] x = F.pad(x, wpad+hpad) return x, wpad, hpad def unpad(self, x, wpad, hpad): return x[..., hpad[0] : x.shape[-2]-hpad[1], wpad[0] : x.shape[-1]-wpad[1]] def forward(self, x, k=None): x, wpad, hpad = self.pad(x) x = self.net(x, k) x = self.unpad(x, wpad, hpad) return x
810
26.033333
90
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/unet.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import torch from torch import nn from torch.nn import functional as F class Unet(nn.Module): """ PyTorch implementation of a U-Net model. O. Ronneberger, P. Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention, pages 234–241. Springer, 2015. """ def __init__( self, in_chans: int = 3, out_chans: int = 3, chans: int = 21, num_pool_layers: int = 4, drop_prob: float = 0.0, ): """ Args: in_chans: Number of channels in the input to the U-Net model. out_chans: Number of channels in the output to the U-Net model. chans: Number of output channels of the first convolution layer. num_pool_layers: Number of down-sampling and up-sampling layers. drop_prob: Dropout probability. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.chans = chans self.num_pool_layers = num_pool_layers self.drop_prob = drop_prob self.down_sample_layers = nn.ModuleList([ConvBlock(in_chans, chans, drop_prob)]) ch = chans for _ in range(num_pool_layers - 1): self.down_sample_layers.append(ConvBlock(ch, ch * 2, drop_prob)) ch *= 2 self.conv = ConvBlock(ch, ch * 2, drop_prob) self.up_conv = nn.ModuleList() self.up_transpose_conv = nn.ModuleList() for _ in range(num_pool_layers - 1): self.up_transpose_conv.append(TransposeConvBlock(ch * 2, ch)) self.up_conv.append(ConvBlock(ch * 2, ch, drop_prob)) ch //= 2 self.up_transpose_conv.append(TransposeConvBlock(ch * 2, ch)) self.up_conv.append( nn.Sequential( ConvBlock(ch * 2, ch, drop_prob), nn.Conv2d(ch, self.out_chans, kernel_size=1, stride=1), ) ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H, W)`. """ stack = [] output = image # apply down-sampling layers for layer in self.down_sample_layers: output = layer(output) stack.append(output) output = F.avg_pool2d(output, kernel_size=2, stride=2, padding=0) output = self.conv(output) # apply up-sampling layers for transpose_conv, conv in zip(self.up_transpose_conv, self.up_conv): downsample_layer = stack.pop() output = transpose_conv(output) # reflect pad on the right/botton if needed to handle odd input dimensions padding = [0, 0, 0, 0] if output.shape[-1] != downsample_layer.shape[-1]: padding[1] = 1 # padding right if output.shape[-2] != downsample_layer.shape[-2]: padding[3] = 1 # padding bottom if torch.sum(torch.tensor(padding)) != 0: output = F.pad(output, padding, "reflect") output = torch.cat([output, downsample_layer], dim=1) output = conv(output) return output class ConvBlock(nn.Module): """ A Convolutional Block that consists of two convolution layers each followed by instance normalization, LeakyReLU activation and dropout. """ def __init__(self, in_chans: int, out_chans: int, drop_prob: float): """ Args: in_chans: Number of channels in the input. out_chans: Number of channels in the output. drop_prob: Dropout probability. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.drop_prob = drop_prob self.layers = nn.Sequential( nn.Conv2d(in_chans, out_chans, kernel_size=3, padding=1, bias=False), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), nn.Dropout2d(drop_prob), nn.Conv2d(out_chans, out_chans, kernel_size=3, padding=1, bias=False), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), nn.Dropout2d(drop_prob), ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H, W)`. """ return self.layers(image) class TransposeConvBlock(nn.Module): """ A Transpose Convolutional Block that consists of one convolution transpose layers followed by instance normalization and LeakyReLU activation. """ def __init__(self, in_chans: int, out_chans: int): """ Args: in_chans: Number of channels in the input. out_chans: Number of channels in the output. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.layers = nn.Sequential( nn.ConvTranspose2d( in_chans, out_chans, kernel_size=2, stride=2, bias=False ), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H*2, W*2)`. """ return self.layers(image)
5,981
32.79661
88
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/__init__.py
from .img2img_mixer import * from .linear_mixer import * from .original_mixer import * from .u_mixer import * from .unet import * from .recon_net import ReconNet from .vit import VisionTransformer
197
23.75
34
py
imaging_MLPs
imaging_MLPs-master/ImageNet/networks/u_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbeddings(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchMerge(nn.Module): r""" Patch Merging Layer. Args: input_resolution (int): Resolution of input feature. dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, channel_dim, norm_layer=nn.LayerNorm): super().__init__() self.dim = channel_dim self.reduction = nn.Linear(4 * channel_dim, 2 * channel_dim, bias=False) self.norm = norm_layer(4 * channel_dim) def forward(self, x): """ x: B, H, W, C """ B, H, W, C = x.shape x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C x = x.view(B, H//2, W//2, 4 * C) # B H/2*W/2 4*C x = self.norm(x) x = self.reduction(x) return x class PatchExpand(nn.Module): def __init__(self, channel_dim, dim_scale=2, norm_layer=nn.LayerNorm): super().__init__() self.dim = channel_dim self.expand = nn.Linear(channel_dim, 2* channel_dim, bias=False) if dim_scale==2 else nn.Identity() self.norm = norm_layer(channel_dim // dim_scale) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=2, p2=2, c=C//4) x = x.view(B, H*2, W*2 ,C//4) x= self.norm(x) return x class FinalPatchExpand(nn.Module): def __init__(self, channel_dim, img_channels, dim_scale=4, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, 16* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), MLPBlock(num_patches, num_patches), Rearrange("b c w h -> b c h w"), MLPBlock(num_patches, num_patches), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class U_Mixer(nn.Module): def __init__( self, img_size: int = 256, img_channels: int = 3, embed_dim: int = 96, ): super().__init__() #mixer blocks self.mixer1= Mixer( img_size//4, embed_dim) self.mixer2= Mixer( img_size//8, embed_dim*2) self.mixer3= Mixer( img_size//16, embed_dim*4) self.mixer4= Mixer( img_size//32, embed_dim*8) self.mixer11= Mixer( img_size//4, embed_dim) self.mixer22= Mixer( img_size//8, embed_dim*2) self.mixer33= Mixer( img_size//16, embed_dim*4) self.mixer44= Mixer( img_size//32, embed_dim*8) #encode self.patch_embed = PatchEmbeddings(4, embed_dim, img_channels) self.patch_merge1 = PatchMerge(embed_dim) self.patch_merge2= PatchMerge(embed_dim*2) self.patch_merge3= PatchMerge(embed_dim*4) #decode self.patch_expand1 = PatchExpand(embed_dim*8) self.patch_expand2 = PatchExpand(embed_dim*4) self.patch_expand3 = PatchExpand(embed_dim*2) self.final_expand = FinalPatchExpand(embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: y=[] x = self.patch_embed(x) x = self.mixer1(x) x = self.mixer1(x) y.append(x) x = self.patch_merge1(x) x = self.mixer2(x) x = self.mixer2(x) y.append(x) x = self.patch_merge2(x) x = self.mixer3(x) x = self.mixer3(x) y.append(x) x = self.patch_merge3(x) x = self.mixer4(x) x = self.mixer4(x) x = self.mixer44(x) x = self.mixer44(x) x = self.patch_expand1(x) x = self.mixer33(x) x = self.mixer33(x) + y[2] x = self.patch_expand2(x) x = self.mixer22(x) x = self.mixer22(x) + y[1] x = self.patch_expand3(x) x = self.mixer11(x) x = self.mixer11(x) + y[0] x = self.final_expand(x) return x
6,516
28.488688
127
py
imaging_MLPs
imaging_MLPs-master/SIDD/config.py
#Path to original dataset original_path='/workspace/media/hdd1/SSID/SIDD_Small_sRGB_Only/Data/' #Path to processed dataset data_path='/workspace/media/hdd1/SSID/dataset/' #Path to store trained models models_path= "/workspace/media/hdd1/SSID/trained_models/" #Path to store tensor board data logs_path= './logs/'
315
27.727273
69
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/img2img_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbedding(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchExpansion(nn.Module): def __init__(self, dim_scale, channel_dim, img_channels, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, dim_scale**2* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int, f_hidden: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c w h -> b c h w"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels*f_hidden) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class Img2Img_Mixer(nn.Module): def __init__( self, img_size: int = 256, img_channels: int = 3, patch_size: int = 4, embed_dim: int = 128, num_layers: int = 16, f_hidden: int = 4, ): super().__init__() self.patch_embed = PatchEmbedding(patch_size, embed_dim, img_channels) layers = [ Mixer(img_size//patch_size, embed_dim, f_hidden) for _ in range(num_layers)] self.mixer_layers = nn.Sequential(*layers) self.patch_expand = PatchExpansion(patch_size, embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: x = self.patch_embed(x) x = self.mixer_layers(x) x = self.patch_expand(x) return x
3,618
27.054264
127
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/vit.py
''' This code is modified from https://github.com/facebookresearch/convit. To adapt the vit/convit to image reconstruction, variable input sizes, and patch sizes for both spatial dimensions. ''' import torch import torch.nn as nn from functools import partial import torch.nn.functional as F from timm.models.helpers import load_pretrained from timm.models.layers import DropPath, to_2tuple, trunc_normal_ class Mlp(nn.Module): def __init__(self, in_features, hidden_features=None, out_features=None, act_layer=nn.GELU, drop=0.): super().__init__() out_features = out_features or in_features hidden_features = hidden_features or in_features self.fc1 = nn.Linear(in_features, hidden_features) self.act = act_layer() self.fc2 = nn.Linear(hidden_features, out_features) self.drop = nn.Dropout(drop) self.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def forward(self, x): x = self.fc1(x) x = self.act(x) x = self.drop(x) x = self.fc2(x) x = self.drop(x) return x class GPSA(nn.Module): def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., locality_strength=1., use_local_init=True, grid_size=None): super().__init__() self.num_heads = num_heads self.dim = dim head_dim = dim // num_heads self.scale = qk_scale or head_dim ** -0.5 self.q = nn.Linear(dim, dim, bias=qkv_bias) self.k = nn.Linear(dim, dim, bias=qkv_bias) self.v = nn.Linear(dim, dim, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.pos_proj = nn.Linear(3, num_heads) self.proj_drop = nn.Dropout(proj_drop) self.locality_strength = locality_strength self.gating_param = nn.Parameter(1*torch.ones(self.num_heads)) self.apply(self._init_weights) if use_local_init: self.local_init(locality_strength=locality_strength) self.current_grid_size = grid_size def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def get_attention(self, x): B, N, C = x.shape k = self.k(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) q = self.q(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) pos_score = self.pos_proj(self.rel_indices).expand(B, -1, -1,-1).permute(0,3,1,2) patch_score = (q @ k.transpose(-2, -1)) * self.scale patch_score = patch_score.softmax(dim=-1) pos_score = pos_score.softmax(dim=-1) gating = self.gating_param.view(1,-1,1,1) attn = (1.-torch.sigmoid(gating)) * patch_score + torch.sigmoid(gating) * pos_score attn = attn / attn.sum(dim=-1).unsqueeze(-1) attn = self.attn_drop(attn) return attn def get_attention_map(self, x, return_map = False): attn_map = self.get_attention(x).mean(0) # average over batch distances = self.rel_indices.squeeze()[:,:,-1]**.5 dist = torch.einsum('nm,hnm->h', (distances, attn_map)) dist /= distances.size(0) if return_map: return dist, attn_map else: return dist def local_init(self, locality_strength=1.): self.v.weight.data.copy_(torch.eye(self.dim)) locality_distance = 1 #max(1,1/locality_strength**.5) kernel_size = int(self.num_heads**.5) center = (kernel_size-1)/2 if kernel_size%2==0 else kernel_size//2 for h1 in range(kernel_size): for h2 in range(kernel_size): position = h1+kernel_size*h2 self.pos_proj.weight.data[position,2] = -1 self.pos_proj.weight.data[position,1] = 2*(h1-center)*locality_distance self.pos_proj.weight.data[position,0] = 2*(h2-center)*locality_distance self.pos_proj.weight.data *= locality_strength def get_rel_indices(self, ): H, W = self.current_grid_size N = H*W rel_indices = torch.zeros(1, N, N, 3) indx = torch.arange(W).view(1,-1) - torch.arange(W).view(-1, 1) indx = indx.repeat(H, H) indy = torch.arange(H).view(1,-1) - torch.arange(H).view(-1, 1) indy = indy.repeat_interleave(W, dim=0).repeat_interleave(W, dim=1) indd = indx**2 + indy**2 rel_indices[:,:,:,2] = indd.unsqueeze(0) rel_indices[:,:,:,1] = indy.unsqueeze(0) rel_indices[:,:,:,0] = indx.unsqueeze(0) device = self.v.weight.device self.rel_indices = rel_indices.to(device) def forward(self, x): B, N, C = x.shape if not hasattr(self, 'rel_indices') or self.rel_indices.size(1)!=N: self.get_rel_indices() attn = self.get_attention(x) v = self.v(x).reshape(B, N, self.num_heads, C // self.num_heads).permute(0, 2, 1, 3) x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x) return x class MHSA(nn.Module): def __init__(self, dim, num_heads=8, qkv_bias=False, qk_scale=None, attn_drop=0., proj_drop=0., grid_size=None): super().__init__() self.num_heads = num_heads head_dim = dim // num_heads self.scale = qk_scale or head_dim ** -0.5 self.qkv = nn.Linear(dim, dim * 3, bias=qkv_bias) self.attn_drop = nn.Dropout(attn_drop) self.proj = nn.Linear(dim, dim) self.proj_drop = nn.Dropout(proj_drop) self.apply(self._init_weights) self.current_grid_size = grid_size def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) def get_attention_map(self, x, return_map = False): self.get_rel_indices() B, N, C = x.shape qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) q, k, v = qkv[0], qkv[1], qkv[2] attn_map = (q @ k.transpose(-2, -1)) * self.scale attn_map = attn_map.softmax(dim=-1).mean(0) # average over batch distances = self.rel_indices.squeeze()[:,:,-1]**.5 dist = torch.einsum('nm,hnm->h', (distances, attn_map)) dist /= distances.size(0) if return_map: return dist, attn_map else: return dist def get_rel_indices(self, ): H, W = self.current_grid_size N = H*W rel_indices = torch.zeros(1, N, N, 3) indx = torch.arange(W).view(1,-1) - torch.arange(W).view(-1, 1) indx = indx.repeat(H, H) indy = torch.arange(H).view(1,-1) - torch.arange(H).view(-1, 1) indy = indy.repeat_interleave(W, dim=0).repeat_interleave(W, dim=1) indd = indx**2 + indy**2 rel_indices[:,:,:,2] = indd.unsqueeze(0) rel_indices[:,:,:,1] = indy.unsqueeze(0) rel_indices[:,:,:,0] = indx.unsqueeze(0) device = self.qkv.weight.device self.rel_indices = rel_indices.to(device) def forward(self, x): B, N, C = x.shape qkv = self.qkv(x).reshape(B, N, 3, self.num_heads, C // self.num_heads).permute(2, 0, 3, 1, 4) q, k, v = qkv[0], qkv[1], qkv[2] attn = (q @ k.transpose(-2, -1)) * self.scale attn = attn.softmax(dim=-1) attn = self.attn_drop(attn) x = (attn @ v).transpose(1, 2).reshape(B, N, C) x = self.proj(x) x = self.proj_drop(x) return x class Block(nn.Module): def __init__(self, dim, num_heads, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop=0., attn_drop=0., drop_path=0., act_layer=nn.GELU, norm_layer=nn.LayerNorm, use_gpsa=True, **kwargs): super().__init__() self.norm1 = norm_layer(dim) self.use_gpsa = use_gpsa if self.use_gpsa: self.attn = GPSA(dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop, **kwargs) else: self.attn = MHSA(dim, num_heads=num_heads, qkv_bias=qkv_bias, qk_scale=qk_scale, attn_drop=attn_drop, proj_drop=drop, **kwargs) self.drop_path = DropPath(drop_path) if drop_path > 0. else nn.Identity() self.norm2 = norm_layer(dim) mlp_hidden_dim = int(dim * mlp_ratio) self.mlp = Mlp(in_features=dim, hidden_features=mlp_hidden_dim, act_layer=act_layer, drop=drop) def forward(self, x, grid_size): self.attn.current_grid_size = grid_size x = x + self.drop_path(self.attn(self.norm1(x))) x = x + self.drop_path(self.mlp(self.norm2(x))) return x class PatchEmbed(nn.Module): """ Image to Patch Embedding, from timm """ def __init__(self, patch_size, in_chans, embed_dim): super().__init__() self.proj = nn.Conv2d(in_chans, embed_dim, kernel_size=patch_size, stride=patch_size) self.apply(self._init_weights) def forward(self, x): x = self.proj(x) return x def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) class VisionTransformer(nn.Module): """ Vision Transformer """ def __init__(self, avrg_img_size=320, patch_size=16, in_chans=1, embed_dim=64, depth=8, num_heads=9, mlp_ratio=4., qkv_bias=False, qk_scale=None, drop_rate=0., attn_drop_rate=0., drop_path_rate=0., norm_layer=nn.LayerNorm, global_pool=None, gpsa_interval=[-1, -1], locality_strength=1., use_pos_embed=True): super().__init__() self.depth = depth embed_dim *= num_heads self.num_features = embed_dim # num_features for consistency with other models self.locality_strength = locality_strength self.use_pos_embed = use_pos_embed if isinstance(avrg_img_size, int): img_size = to_2tuple(avrg_img_size) if isinstance(patch_size, int): self.patch_size = to_2tuple(patch_size) else: self.patch_size = patch_size self.in_chans = in_chans self.patch_embed = PatchEmbed( patch_size=self.patch_size, in_chans=in_chans, embed_dim=embed_dim) self.pos_drop = nn.Dropout(p=drop_rate) if self.use_pos_embed: self.pos_embed = nn.Parameter( torch.zeros(1, embed_dim, img_size[0] // self.patch_size[0], img_size[1] // self.patch_size[1]) ) trunc_normal_(self.pos_embed, std=.02) dpr = [x.item() for x in torch.linspace(0, drop_path_rate, depth)] # stochastic depth decay rule self.blocks = nn.ModuleList([ Block( dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer, use_gpsa=True, locality_strength=locality_strength) if i>=gpsa_interval[0]-1 and i<gpsa_interval[1] else Block( dim=embed_dim, num_heads=num_heads, mlp_ratio=mlp_ratio, qkv_bias=qkv_bias, qk_scale=qk_scale, drop=drop_rate, attn_drop=attn_drop_rate, drop_path=dpr[i], norm_layer=norm_layer, use_gpsa=False,) for i in range(depth)]) self.norm = norm_layer(embed_dim) # head self.feature_info = [dict(num_chs=embed_dim, reduction=0, module='head')] self.head = nn.Linear(self.num_features, in_chans*self.patch_size[0]*self.patch_size[1]) def seq2img(self, x, img_size): """ Transforms sequence back into image space, input dims: [batch_size, num_patches, channels] output dims: [batch_size, channels, H, W] """ x = x.view(x.shape[0], x.shape[1], self.in_chans, self.patch_size[0], self.patch_size[1]) x = x.chunk(x.shape[1], dim=1) x = torch.cat(x, dim=4).permute(0,1,2,4,3) x = x.chunk(img_size[0]//self.patch_size[0], dim=3) x = torch.cat(x, dim=4).permute(0,1,2,4,3).squeeze(1) return x self.head.apply(self._init_weights) def _init_weights(self, m): if isinstance(m, nn.Linear): trunc_normal_(m.weight, std=.02) if isinstance(m, nn.Linear) and m.bias is not None: nn.init.constant_(m.bias, 0) elif isinstance(m, nn.LayerNorm): nn.init.constant_(m.bias, 0) nn.init.constant_(m.weight, 1.0) @torch.jit.ignore def no_weight_decay(self,): return {'pos_embed'} def get_head(self,): return self.head def reset_head(self,): self.head = nn.Linear(self.num_features, in_chans*self.patch_size[0]*self.patch_size[1]) def forward_features(self, x, k=None): x = self.patch_embed(x) _, _, H, W = x.shape if self.use_pos_embed: pos_embed = F.interpolate(self.pos_embed, size=[H, W], mode='bilinear', align_corners = False) x = x + pos_embed x = x.flatten(2).transpose(1, 2) x = self.pos_drop(x) for u, blk in enumerate(self.blocks): x = blk(x, (H, W)) if k is not None and u == k: self.attention_map = blk.attn.get_attention_map(x, return_map = True) x = self.norm(x) return x def forward(self, x, k=None): _, _, H, W = x.shape x = self.forward_features(x, k) x = self.head(x) x = self.seq2img(x, (H, W)) return x
15,082
39.007958
186
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/recon_net.py
import torch.nn as nn import torch.nn.functional as F from math import ceil, floor class ReconNet(nn.Module): def __init__(self, net): super().__init__() self.net = net def pad(self, x): _, _, h, w = x.shape hp, wp = self.net.patch_size f1 = ( (wp - w % wp) % wp ) / 2 f2 = ( (hp - h % hp) % hp ) / 2 wpad = [floor(f1), ceil(f1)] hpad = [floor(f2), ceil(f2)] x = F.pad(x, wpad+hpad) return x, wpad, hpad def unpad(self, x, wpad, hpad): return x[..., hpad[0] : x.shape[-2]-hpad[1], wpad[0] : x.shape[-1]-wpad[1]] def forward(self, x, k=None): x, wpad, hpad = self.pad(x) x = self.net(x, k) x = self.unpad(x, wpad, hpad) return x
810
26.033333
90
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/unet.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import torch from torch import nn from torch.nn import functional as F class Unet(nn.Module): """ PyTorch implementation of a U-Net model. O. Ronneberger, P. Fischer, and Thomas Brox. U-net: Convolutional networks for biomedical image segmentation. In International Conference on Medical image computing and computer-assisted intervention, pages 234–241. Springer, 2015. """ def __init__( self, in_chans: int = 3, out_chans: int = 3, chans: int = 21, num_pool_layers: int = 4, drop_prob: float = 0.0, ): """ Args: in_chans: Number of channels in the input to the U-Net model. out_chans: Number of channels in the output to the U-Net model. chans: Number of output channels of the first convolution layer. num_pool_layers: Number of down-sampling and up-sampling layers. drop_prob: Dropout probability. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.chans = chans self.num_pool_layers = num_pool_layers self.drop_prob = drop_prob self.down_sample_layers = nn.ModuleList([ConvBlock(in_chans, chans, drop_prob)]) ch = chans for _ in range(num_pool_layers - 1): self.down_sample_layers.append(ConvBlock(ch, ch * 2, drop_prob)) ch *= 2 self.conv = ConvBlock(ch, ch * 2, drop_prob) self.up_conv = nn.ModuleList() self.up_transpose_conv = nn.ModuleList() for _ in range(num_pool_layers - 1): self.up_transpose_conv.append(TransposeConvBlock(ch * 2, ch)) self.up_conv.append(ConvBlock(ch * 2, ch, drop_prob)) ch //= 2 self.up_transpose_conv.append(TransposeConvBlock(ch * 2, ch)) self.up_conv.append( nn.Sequential( ConvBlock(ch * 2, ch, drop_prob), nn.Conv2d(ch, self.out_chans, kernel_size=1, stride=1), ) ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H, W)`. """ stack = [] output = image # apply down-sampling layers for layer in self.down_sample_layers: output = layer(output) stack.append(output) output = F.avg_pool2d(output, kernel_size=2, stride=2, padding=0) output = self.conv(output) # apply up-sampling layers for transpose_conv, conv in zip(self.up_transpose_conv, self.up_conv): downsample_layer = stack.pop() output = transpose_conv(output) # reflect pad on the right/botton if needed to handle odd input dimensions padding = [0, 0, 0, 0] if output.shape[-1] != downsample_layer.shape[-1]: padding[1] = 1 # padding right if output.shape[-2] != downsample_layer.shape[-2]: padding[3] = 1 # padding bottom if torch.sum(torch.tensor(padding)) != 0: output = F.pad(output, padding, "reflect") output = torch.cat([output, downsample_layer], dim=1) output = conv(output) return output class ConvBlock(nn.Module): """ A Convolutional Block that consists of two convolution layers each followed by instance normalization, LeakyReLU activation and dropout. """ def __init__(self, in_chans: int, out_chans: int, drop_prob: float): """ Args: in_chans: Number of channels in the input. out_chans: Number of channels in the output. drop_prob: Dropout probability. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.drop_prob = drop_prob self.layers = nn.Sequential( nn.Conv2d(in_chans, out_chans, kernel_size=3, padding=1, bias=False), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), nn.Dropout2d(drop_prob), nn.Conv2d(out_chans, out_chans, kernel_size=3, padding=1, bias=False), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), nn.Dropout2d(drop_prob), ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H, W)`. """ return self.layers(image) class TransposeConvBlock(nn.Module): """ A Transpose Convolutional Block that consists of one convolution transpose layers followed by instance normalization and LeakyReLU activation. """ def __init__(self, in_chans: int, out_chans: int): """ Args: in_chans: Number of channels in the input. out_chans: Number of channels in the output. """ super().__init__() self.in_chans = in_chans self.out_chans = out_chans self.layers = nn.Sequential( nn.ConvTranspose2d( in_chans, out_chans, kernel_size=2, stride=2, bias=False ), nn.InstanceNorm2d(out_chans), nn.LeakyReLU(negative_slope=0.2, inplace=True), ) def forward(self, image: torch.Tensor) -> torch.Tensor: """ Args: image: Input 4D tensor of shape `(N, in_chans, H, W)`. Returns: Output tensor of shape `(N, out_chans, H*2, W*2)`. """ return self.layers(image)
5,981
32.79661
88
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/__init__.py
from .img2img_mixer import * from .u_mixer import * from .unet import * from .recon_net import ReconNet from .vit import VisionTransformer
140
19.142857
34
py
imaging_MLPs
imaging_MLPs-master/SIDD/networks/u_mixer.py
import torch import torch.nn as nn from torch.nn import init import torch.nn.init as init import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbeddings(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchMerge(nn.Module): r""" Patch Merging Layer. Args: input_resolution (int): Resolution of input feature. dim (int): Number of input channels. norm_layer (nn.Module, optional): Normalization layer. Default: nn.LayerNorm """ def __init__(self, channel_dim, norm_layer=nn.LayerNorm): super().__init__() self.dim = channel_dim self.reduction = nn.Linear(4 * channel_dim, 2 * channel_dim, bias=False) self.norm = norm_layer(4 * channel_dim) def forward(self, x): """ x: B, H, W, C """ B, H, W, C = x.shape x0 = x[:, 0::2, 0::2, :] # B H/2 W/2 C x1 = x[:, 1::2, 0::2, :] # B H/2 W/2 C x2 = x[:, 0::2, 1::2, :] # B H/2 W/2 C x3 = x[:, 1::2, 1::2, :] # B H/2 W/2 C x = torch.cat([x0, x1, x2, x3], -1) # B H/2 W/2 4*C x = x.view(B, H//2, W//2, 4 * C) # B H/2*W/2 4*C x = self.norm(x) x = self.reduction(x) return x class PatchExpand(nn.Module): def __init__(self, channel_dim, dim_scale=2, norm_layer=nn.LayerNorm): super().__init__() self.dim = channel_dim self.expand = nn.Linear(channel_dim, 2* channel_dim, bias=False) if dim_scale==2 else nn.Identity() self.norm = norm_layer(channel_dim // dim_scale) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=2, p2=2, c=C//4) x = x.view(B, H*2, W*2 ,C//4) x= self.norm(x) return x class FinalPatchExpand(nn.Module): def __init__(self, channel_dim, img_channels, dim_scale=4, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, 16* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), MLPBlock(num_patches, num_patches), Rearrange("b c w h -> b c h w"), MLPBlock(num_patches, num_patches), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class U_Mixer(nn.Module): def __init__( self, img_size: int = 256, img_channels: int = 3, embed_dim: int = 96, ): super().__init__() #mixer blocks self.mixer1= Mixer( img_size//4, embed_dim) self.mixer2= Mixer( img_size//8, embed_dim*2) self.mixer3= Mixer( img_size//16, embed_dim*4) self.mixer4= Mixer( img_size//32, embed_dim*8) self.mixer11= Mixer( img_size//4, embed_dim) self.mixer22= Mixer( img_size//8, embed_dim*2) self.mixer33= Mixer( img_size//16, embed_dim*4) self.mixer44= Mixer( img_size//32, embed_dim*8) #encode self.patch_embed = PatchEmbeddings(4, embed_dim, img_channels) self.patch_merge1 = PatchMerge(embed_dim) self.patch_merge2= PatchMerge(embed_dim*2) self.patch_merge3= PatchMerge(embed_dim*4) #decode self.patch_expand1 = PatchExpand(embed_dim*8) self.patch_expand2 = PatchExpand(embed_dim*4) self.patch_expand3 = PatchExpand(embed_dim*2) self.final_expand = FinalPatchExpand(embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: y=[] x = self.patch_embed(x) x = self.mixer1(x) x = self.mixer1(x) y.append(x) x = self.patch_merge1(x) x = self.mixer2(x) x = self.mixer2(x) y.append(x) x = self.patch_merge2(x) x = self.mixer3(x) x = self.mixer3(x) y.append(x) x = self.patch_merge3(x) x = self.mixer4(x) x = self.mixer4(x) x = self.mixer44(x) x = self.mixer44(x) x = self.patch_expand1(x) x = self.mixer33(x) x = self.mixer33(x) + y[2] x = self.patch_expand2(x) x = self.mixer22(x) x = self.mixer22(x) + y[1] x = self.patch_expand3(x) x = self.mixer11(x) x = self.mixer11(x) + y[0] x = self.final_expand(x) return x
6,516
28.488688
127
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/losses.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import torch import torch.nn as nn import torch.nn.functional as F class SSIMLoss(nn.Module): """ SSIM loss module. """ def __init__(self, win_size: int = 7, k1: float = 0.01, k2: float = 0.03): """ Args: win_size: Window size for SSIM calculation. k1: k1 parameter for SSIM calculation. k2: k2 parameter for SSIM calculation. """ super().__init__() self.win_size = win_size self.k1, self.k2 = k1, k2 self.register_buffer("w", torch.ones(1, 1, win_size, win_size) / win_size ** 2) NP = win_size ** 2 self.cov_norm = NP / (NP - 1) def forward(self, X: torch.Tensor, Y: torch.Tensor, data_range: torch.Tensor): assert isinstance(self.w, torch.Tensor) data_range = data_range[:, None, None, None] C1 = (self.k1 * data_range) ** 2 C2 = (self.k2 * data_range) ** 2 ux = F.conv2d(X, self.w) # typing: ignore uy = F.conv2d(Y, self.w) # uxx = F.conv2d(X * X, self.w) uyy = F.conv2d(Y * Y, self.w) uxy = F.conv2d(X * Y, self.w) vx = self.cov_norm * (uxx - ux * ux) vy = self.cov_norm * (uyy - uy * uy) vxy = self.cov_norm * (uxy - ux * uy) A1, A2, B1, B2 = ( 2 * ux * uy + C1, 2 * vxy + C2, ux ** 2 + uy ** 2 + C1, vx + vy + C2, ) D = B1 * B2 S = (A1 * A2) / D return 1 - S.mean()
1,671
28.857143
87
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/coil_combine.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import torch import fastmri def rss(data: torch.Tensor, dim: int = 0) -> torch.Tensor: """ Compute the Root Sum of Squares (RSS). RSS is computed assuming that dim is the coil dimension. Args: data: The input tensor dim: The dimensions along which to apply the RSS transform Returns: The RSS value. """ return torch.sqrt((data ** 2).sum(dim)) def rss_complex(data: torch.Tensor, dim: int = 0) -> torch.Tensor: """ Compute the Root Sum of Squares (RSS) for complex inputs. RSS is computed assuming that dim is the coil dimension. Args: data: The input tensor dim: The dimensions along which to apply the RSS transform Returns: The RSS value. """ return torch.sqrt(fastmri.complex_abs_sq(data).sum(dim))
996
22.186047
66
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/evaluate.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import argparse import pathlib from argparse import ArgumentParser from typing import Optional import h5py import numpy as np from runstats import Statistics from skimage.metrics import peak_signal_noise_ratio, structural_similarity from fastmri.data import transforms def mse(gt: np.ndarray, pred: np.ndarray) -> np.ndarray: """Compute Mean Squared Error (MSE)""" return np.mean((gt - pred) ** 2) def nmse(gt: np.ndarray, pred: np.ndarray) -> np.ndarray: """Compute Normalized Mean Squared Error (NMSE)""" return np.linalg.norm(gt - pred) ** 2 / np.linalg.norm(gt) ** 2 def psnr( gt: np.ndarray, pred: np.ndarray, maxval: Optional[float] = None ) -> np.ndarray: """Compute Peak Signal to Noise Ratio metric (PSNR)""" if maxval is None: maxval = gt.max() return peak_signal_noise_ratio(gt, pred, data_range=maxval) def ssim( gt: np.ndarray, pred: np.ndarray, maxval: Optional[float] = None ) -> np.ndarray: """Compute Structural Similarity Index Metric (SSIM)""" if not gt.ndim == 3: raise ValueError("Unexpected number of dimensions in ground truth.") if not gt.ndim == pred.ndim: raise ValueError("Ground truth dimensions does not match pred.") maxval = gt.max() if maxval is None else maxval ssim = 0 for slice_num in range(gt.shape[0]): ssim = ssim + structural_similarity( gt[slice_num], pred[slice_num], data_range=maxval ) return ssim / gt.shape[0] METRIC_FUNCS = dict( MSE=mse, NMSE=nmse, PSNR=psnr, SSIM=ssim, ) class Metrics: """ Maintains running statistics for a given collection of metrics. """ def __init__(self, metric_funcs): """ Args: metric_funcs (dict): A dict where the keys are metric names and the values are Python functions for evaluating that metric. """ self.metrics = {metric: Statistics() for metric in metric_funcs} def push(self, target, recons): for metric, func in METRIC_FUNCS.items(): self.metrics[metric].push(func(target, recons)) def means(self): return {metric: stat.mean() for metric, stat in self.metrics.items()} def stddevs(self): return {metric: stat.stddev() for metric, stat in self.metrics.items()} def __repr__(self): means = self.means() stddevs = self.stddevs() metric_names = sorted(list(means)) return " ".join( f"{name} = {means[name]:.4g} +/- {2 * stddevs[name]:.4g}" for name in metric_names ) def evaluate(args, recons_key): metrics = Metrics(METRIC_FUNCS) for tgt_file in args.target_path.iterdir(): with h5py.File(tgt_file, "r") as target, h5py.File( args.predictions_path / tgt_file.name, "r" ) as recons: if args.acquisition and args.acquisition != target.attrs["acquisition"]: continue if args.acceleration and target.attrs["acceleration"] != args.acceleration: continue target = target[recons_key][()] recons = recons["reconstruction"][()] target = transforms.center_crop( target, (target.shape[-1], target.shape[-1]) ) recons = transforms.center_crop( recons, (target.shape[-1], target.shape[-1]) ) metrics.push(target, recons) return metrics if __name__ == "__main__": parser = ArgumentParser(formatter_class=argparse.ArgumentDefaultsHelpFormatter) parser.add_argument( "--target-path", type=pathlib.Path, required=True, help="Path to the ground truth data", ) parser.add_argument( "--predictions-path", type=pathlib.Path, required=True, help="Path to reconstructions", ) parser.add_argument( "--challenge", choices=["singlecoil", "multicoil"], required=True, help="Which challenge", ) parser.add_argument("--acceleration", type=int, default=None) parser.add_argument( "--acquisition", choices=[ "CORPD_FBK", "CORPDFS_FBK", "AXT1", "AXT1PRE", "AXT1POST", "AXT2", "AXFLAIR", ], default=None, help="If set, only volumes of the specified acquisition type are used " "for evaluation. By default, all volumes are included.", ) args = parser.parse_args() recons_key = ( "reconstruction_rss" if args.challenge == "multicoil" else "reconstruction_esc" ) metrics = evaluate(args, recons_key) print(metrics)
4,903
27.847059
87
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/math.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import numpy as np import torch def complex_mul(x: torch.Tensor, y: torch.Tensor) -> torch.Tensor: """ Complex multiplication. This multiplies two complex tensors assuming that they are both stored as real arrays with the last dimension being the complex dimension. Args: x: A PyTorch tensor with the last dimension of size 2. y: A PyTorch tensor with the last dimension of size 2. Returns: A PyTorch tensor with the last dimension of size 2. """ if not x.shape[-1] == y.shape[-1] == 2: raise ValueError("Tensors do not have separate complex dim.") re = x[..., 0] * y[..., 0] - x[..., 1] * y[..., 1] im = x[..., 0] * y[..., 1] + x[..., 1] * y[..., 0] return torch.stack((re, im), dim=-1) def complex_conj(x: torch.Tensor) -> torch.Tensor: """ Complex conjugate. This applies the complex conjugate assuming that the input array has the last dimension as the complex dimension. Args: x: A PyTorch tensor with the last dimension of size 2. y: A PyTorch tensor with the last dimension of size 2. Returns: A PyTorch tensor with the last dimension of size 2. """ if not x.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") return torch.stack((x[..., 0], -x[..., 1]), dim=-1) def complex_abs(data: torch.Tensor) -> torch.Tensor: """ Compute the absolute value of a complex valued input tensor. Args: data: A complex valued tensor, where the size of the final dimension should be 2. Returns: Absolute value of data. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") return (data ** 2).sum(dim=-1).sqrt() def complex_abs_sq(data: torch.Tensor) -> torch.Tensor: """ Compute the squared absolute value of a complex tensor. Args: data: A complex valued tensor, where the size of the final dimension should be 2. Returns: Squared absolute value of data. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") return (data ** 2).sum(dim=-1) def tensor_to_complex_np(data: torch.Tensor) -> np.ndarray: """ Converts a complex torch tensor to numpy array. Args: data: Input data to be converted to numpy. Returns: Complex numpy version of data. """ data = data.numpy() return data[..., 0] + 1j * data[..., 1]
2,728
25.754902
77
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/utils.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ from pathlib import Path from typing import Dict import h5py import numpy as np def save_reconstructions(reconstructions: Dict[str, np.ndarray], out_dir: Path): """ Save reconstruction images. This function writes to h5 files that are appropriate for submission to the leaderboard. Args: reconstructions: A dictionary mapping input filenames to corresponding reconstructions. out_dir: Path to the output directory where the reconstructions should be saved. """ out_dir.mkdir(exist_ok=True, parents=True) for fname, recons in reconstructions.items(): with h5py.File(out_dir / fname, "w") as hf: hf.create_dataset("reconstruction", data=recons) def convert_fnames_to_v2(path: Path): """ Converts filenames to conform to `v2` standard for knee data. For a file with name file1000.h5 in `path`, this script simply renames it to file1000_v2.h5. This is for submission to the public knee leaderboards. Args: path: Path with files to be renamed. """ if not path.exists(): raise ValueError("Path does not exist") for fname in path.glob("*.h5"): if not fname.name[-6:] == "_v2.h5": fname.rename(path / (fname.stem + "_v2.h5"))
1,467
28.36
80
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/__init__.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import torch from packaging import version from .coil_combine import rss, rss_complex from .fftc import fftshift, ifftshift, roll from .losses import SSIMLoss from .math import ( complex_abs, complex_abs_sq, complex_conj, complex_mul, tensor_to_complex_np, ) from .utils import convert_fnames_to_v2, save_reconstructions if version.parse(torch.__version__) >= version.parse("1.7.0"): from .fftc import fft2c_new as fft2c from .fftc import ifft2c_new as ifft2c else: from .fftc import fft2c_old as fft2c from .fftc import ifft2c_old as ifft2c
758
26.107143
63
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/fftc.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ from typing import List, Optional import torch from packaging import version if version.parse(torch.__version__) >= version.parse("1.7.0"): import torch.fft # type: ignore def fft2c_old(data: torch.Tensor) -> torch.Tensor: """ Apply centered 2 dimensional Fast Fourier Transform. Args: data: Complex valued input data containing at least 3 dimensions: dimensions -3 & -2 are spatial dimensions and dimension -1 has size 2. All other dimensions are assumed to be batch dimensions. Returns: The FFT of the input. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") data = ifftshift(data, dim=[-3, -2]) data = torch.fft(data, 2, normalized=True) data = fftshift(data, dim=[-3, -2]) return data def ifft2c_old(data: torch.Tensor) -> torch.Tensor: """ Apply centered 2-dimensional Inverse Fast Fourier Transform. Args: data: Complex valued input data containing at least 3 dimensions: dimensions -3 & -2 are spatial dimensions and dimension -1 has size 2. All other dimensions are assumed to be batch dimensions. Returns: The IFFT of the input. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") data = ifftshift(data, dim=[-3, -2]) data = torch.ifft(data, 2, normalized=True) data = fftshift(data, dim=[-3, -2]) return data def fft2c_new(data: torch.Tensor) -> torch.Tensor: """ Apply centered 2 dimensional Fast Fourier Transform. Args: data: Complex valued input data containing at least 3 dimensions: dimensions -3 & -2 are spatial dimensions and dimension -1 has size 2. All other dimensions are assumed to be batch dimensions. Returns: The FFT of the input. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") data = ifftshift(data, dim=[-3, -2]) data = torch.view_as_real( torch.fft.fftn( # type: ignore torch.view_as_complex(data), dim=(-2, -1), norm="ortho" ) ) data = fftshift(data, dim=[-3, -2]) return data def ifft2c_new(data: torch.Tensor) -> torch.Tensor: """ Apply centered 2-dimensional Inverse Fast Fourier Transform. Args: data: Complex valued input data containing at least 3 dimensions: dimensions -3 & -2 are spatial dimensions and dimension -1 has size 2. All other dimensions are assumed to be batch dimensions. Returns: The IFFT of the input. """ if not data.shape[-1] == 2: raise ValueError("Tensor does not have separate complex dim.") data = ifftshift(data, dim=[-3, -2]) data = torch.view_as_real( torch.fft.ifftn( # type: ignore torch.view_as_complex(data), dim=(-2, -1), norm="ortho" ) ) data = fftshift(data, dim=[-3, -2]) return data # Helper functions def roll_one_dim(x: torch.Tensor, shift: int, dim: int) -> torch.Tensor: """ Similar to roll but for only one dim. Args: x: A PyTorch tensor. shift: Amount to roll. dim: Which dimension to roll. Returns: Rolled version of x. """ shift = shift % x.size(dim) if shift == 0: return x left = x.narrow(dim, 0, x.size(dim) - shift) right = x.narrow(dim, x.size(dim) - shift, shift) return torch.cat((right, left), dim=dim) def roll( x: torch.Tensor, shift: List[int], dim: List[int], ) -> torch.Tensor: """ Similar to np.roll but applies to PyTorch Tensors. Args: x: A PyTorch tensor. shift: Amount to roll. dim: Which dimension to roll. Returns: Rolled version of x. """ if len(shift) != len(dim): raise ValueError("len(shift) must match len(dim)") for (s, d) in zip(shift, dim): x = roll_one_dim(x, s, d) return x def fftshift(x: torch.Tensor, dim: Optional[List[int]] = None) -> torch.Tensor: """ Similar to np.fft.fftshift but applies to PyTorch Tensors Args: x: A PyTorch tensor. dim: Which dimension to fftshift. Returns: fftshifted version of x. """ if dim is None: # this weird code is necessary for toch.jit.script typing dim = [0] * (x.dim()) for i in range(1, x.dim()): dim[i] = i # also necessary for torch.jit.script shift = [0] * len(dim) for i, dim_num in enumerate(dim): shift[i] = x.shape[dim_num] // 2 return roll(x, shift, dim) def ifftshift(x: torch.Tensor, dim: Optional[List[int]] = None) -> torch.Tensor: """ Similar to np.fft.ifftshift but applies to PyTorch Tensors Args: x: A PyTorch tensor. dim: Which dimension to ifftshift. Returns: ifftshifted version of x. """ if dim is None: # this weird code is necessary for toch.jit.script typing dim = [0] * (x.dim()) for i in range(1, x.dim()): dim[i] = i # also necessary for torch.jit.script shift = [0] * len(dim) for i, dim_num in enumerate(dim): shift[i] = (x.shape[dim_num] + 1) // 2 return roll(x, shift, dim)
5,535
25.236967
80
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/data/volume_sampler.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ from typing import List, Optional, Union import torch import torch.distributed as dist from fastmri.data.mri_data import CombinedSliceDataset, SliceDataset from torch.utils.data import Sampler class VolumeSampler(Sampler): """ Sampler for volumetric MRI data. Based on pytorch DistributedSampler, the difference is that all instances from the same MRI volume need to go to the same node for distributed training. Dataset example is a list of tuples (fname, instance), where fname is essentially the volume name (actually a filename). """ def __init__( self, dataset: Union[CombinedSliceDataset, SliceDataset], num_replicas: Optional[int] = None, rank: Optional[int] = None, shuffle: bool = True, seed: int = 0, ): """ Args: dataset: An MRI dataset (e.g., SliceData). num_replicas: Number of processes participating in distributed training. By default, :attr:`rank` is retrieved from the current distributed group. rank: Rank of the current process within :attr:`num_replicas`. By default, :attr:`rank` is retrieved from the current distributed group. shuffle: If ``True`` (default), sampler will shuffle the indices. seed: random seed used to shuffle the sampler if :attr:`shuffle=True`. This number should be identical across all processes in the distributed group. """ if num_replicas is None: if not dist.is_available(): raise RuntimeError("Requires distributed package to be available") num_replicas = dist.get_world_size() if rank is None: if not dist.is_available(): raise RuntimeError("Requires distributed package to be available") rank = dist.get_rank() self.dataset = dataset self.num_replicas = num_replicas self.rank = rank self.epoch = 0 self.shuffle = shuffle self.seed = seed # get all file names and split them based on number of processes self.all_volume_names = sorted( set(str(example[0]) for example in self.dataset.examples) ) self.all_volumes_split: List[List[str]] = [] for rank_num in range(self.num_replicas): self.all_volumes_split.append( [ self.all_volume_names[i] for i in range( rank_num, len(self.all_volume_names), self.num_replicas ) ] ) # get slice indices for each file name rank_indices: List[List[int]] = [[] for _ in range(self.num_replicas)] for i, example in enumerate(self.dataset.examples): vname = str(example[0]) for rank_num in range(self.num_replicas): if vname in self.all_volumes_split[rank_num]: rank_indices[rank_num].append(i) break # need to send equal number of samples to each process - take the max self.num_samples = max(len(indices) for indices in rank_indices) self.total_size = self.num_samples * self.num_replicas self.indices = rank_indices[self.rank] def __iter__(self): if self.shuffle: # deterministically shuffle based on epoch and seed g = torch.Generator() g.manual_seed(self.seed + self.epoch) ordering = torch.randperm(len(self.indices), generator=g).tolist() indices = [self.indices[i] for i in ordering] else: indices = self.indices # add extra samples to match num_samples repeat_times = self.num_samples // len(indices) indices = indices * repeat_times indices = indices + indices[: self.num_samples - len(indices)] assert len(indices) == self.num_samples return iter(indices) def __len__(self): return self.num_samples def set_epoch(self, epoch): self.epoch = epoch
4,332
36.678261
82
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/data/mri_data.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import logging import os import pickle import random import xml.etree.ElementTree as etree from pathlib import Path from typing import Callable, Dict, List, Optional, Sequence, Tuple, Union from warnings import warn import h5py import numpy as np import torch import yaml def et_query( root: etree.Element, qlist: Sequence[str], namespace: str = "http://www.ismrm.org/ISMRMRD", ) -> str: """ ElementTree query function. This can be used to query an xml document via ElementTree. It uses qlist for nested queries. Args: root: Root of the xml to search through. qlist: A list of strings for nested searches, e.g. ["Encoding", "matrixSize"] namespace: Optional; xml namespace to prepend query. Returns: The retrieved data as a string. """ s = "." prefix = "ismrmrd_namespace" ns = {prefix: namespace} for el in qlist: s = s + f"//{prefix}:{el}" value = root.find(s, ns) if value is None: raise RuntimeError("Element not found") return str(value.text) def fetch_dir( key: str, data_config_file: Union[str, Path, os.PathLike] = "fastmri_dirs.yaml" ) -> Path: """ Data directory fetcher. This is a brute-force simple way to configure data directories for a project. Simply overwrite the variables for `knee_path` and `brain_path` and this function will retrieve the requested subsplit of the data for use. Args: key: key to retrieve path from data_config_file. Expected to be in ("knee_path", "brain_path", "log_path"). data_config_file: Optional; Default path config file to fetch path from. Returns: The path to the specified directory. """ data_config_file = Path(data_config_file) if not data_config_file.is_file(): default_config = { "knee_path": "/path/to/knee", "brain_path": "/path/to/brain", "log_path": ".", } with open(data_config_file, "w") as f: yaml.dump(default_config, f) data_dir = default_config[key] warn( f"Path config at {data_config_file.resolve()} does not exist. " "A template has been created for you. " "Please enter the directory paths for your system to have defaults." ) else: with open(data_config_file, "r") as f: data_dir = yaml.safe_load(f)[key] return Path(data_dir) class CombinedSliceDataset(torch.utils.data.Dataset): """ A container for combining slice datasets. """ def __init__( self, roots: Sequence[Path], challenges: Sequence[str], transforms: Optional[Sequence[Optional[Callable]]] = None, sample_rates: Optional[Sequence[Optional[float]]] = None, volume_sample_rates: Optional[Sequence[Optional[float]]] = None, use_dataset_cache: bool = False, dataset_cache_file: Union[str, Path, os.PathLike] = "dataset_cache.pkl", num_cols: Optional[Tuple[int]] = None, ): """ Args: roots: Paths to the datasets. challenges: "singlecoil" or "multicoil" depending on which challenge to use. transforms: Optional; A sequence of callable objects that preprocesses the raw data into appropriate form. The transform function should take 'kspace', 'target', 'attributes', 'filename', and 'slice' as inputs. 'target' may be null for test data. sample_rates: Optional; A sequence of floats between 0 and 1. This controls what fraction of the slices should be loaded. When creating subsampled datasets either set sample_rates (sample by slices) or volume_sample_rates (sample by volumes) but not both. volume_sample_rates: Optional; A sequence of floats between 0 and 1. This controls what fraction of the volumes should be loaded. When creating subsampled datasets either set sample_rates (sample by slices) or volume_sample_rates (sample by volumes) but not both. use_dataset_cache: Whether to cache dataset metadata. This is very useful for large datasets like the brain data. dataset_cache_file: Optional; A file in which to cache dataset information for faster load times. num_cols: Optional; If provided, only slices with the desired number of columns will be considered. """ if sample_rates is not None and volume_sample_rates is not None: raise ValueError( "either set sample_rates (sample by slices) or volume_sample_rates (sample by volumes) but not both" ) if transforms is None: transforms = [None] * len(roots) if sample_rates is None: sample_rates = [None] * len(roots) if volume_sample_rates is None: volume_sample_rates = [None] * len(roots) if not ( len(roots) == len(transforms) == len(challenges) == len(sample_rates) == len(volume_sample_rates) ): raise ValueError( "Lengths of roots, transforms, challenges, sample_rates do not match" ) self.datasets = [] self.examples: List[Tuple[Path, int, Dict[str, object]]] = [] for i in range(len(roots)): self.datasets.append( SliceDataset( root=roots[i], transform=transforms[i], challenge=challenges[i], sample_rate=sample_rates[i], volume_sample_rate=volume_sample_rates[i], use_dataset_cache=use_dataset_cache, dataset_cache_file=dataset_cache_file, num_cols=num_cols, ) ) self.examples = self.examples + self.datasets[-1].examples def __len__(self): return sum(len(dataset) for dataset in self.datasets) def __getitem__(self, i): for dataset in self.datasets: if i < len(dataset): return dataset[i] else: i = i - len(dataset) class SliceDataset(torch.utils.data.Dataset): """ A PyTorch Dataset that provides access to MR image slices. """ def __init__( self, root: Union[str, Path, os.PathLike], challenge: str, transform: Optional[Callable] = None, use_dataset_cache: bool = False, sample_rate: Optional[float] = None, volume_sample_rate: Optional[float] = None, dataset_cache_file: Union[str, Path, os.PathLike] = "dataset_cache.pkl", num_cols: Optional[Tuple[int]] = None, ): """ Args: root: Path to the dataset. challenge: "singlecoil" or "multicoil" depending on which challenge to use. transform: Optional; A callable object that pre-processes the raw data into appropriate form. The transform function should take 'kspace', 'target', 'attributes', 'filename', and 'slice' as inputs. 'target' may be null for test data. use_dataset_cache: Whether to cache dataset metadata. This is very useful for large datasets like the brain data. sample_rate: Optional; A float between 0 and 1. This controls what fraction of the slices should be loaded. Defaults to 1 if no value is given. When creating a sampled dataset either set sample_rate (sample by slices) or volume_sample_rate (sample by volumes) but not both. volume_sample_rate: Optional; A float between 0 and 1. This controls what fraction of the volumes should be loaded. Defaults to 1 if no value is given. When creating a sampled dataset either set sample_rate (sample by slices) or volume_sample_rate (sample by volumes) but not both. dataset_cache_file: Optional; A file in which to cache dataset information for faster load times. num_cols: Optional; If provided, only slices with the desired number of columns will be considered. """ if challenge not in ("singlecoil", "multicoil"): raise ValueError('challenge should be either "singlecoil" or "multicoil"') if sample_rate is not None and volume_sample_rate is not None: raise ValueError( "either set sample_rate (sample by slices) or volume_sample_rate (sample by volumes) but not both" ) self.dataset_cache_file = Path(dataset_cache_file) self.transform = transform self.recons_key = ( "reconstruction_esc" if challenge == "singlecoil" else "reconstruction_rss" ) self.examples = [] # set default sampling mode if none given if sample_rate is None: sample_rate = 1.0 if volume_sample_rate is None: volume_sample_rate = 1.0 # load dataset cache if we have and user wants to use it if self.dataset_cache_file.exists() and use_dataset_cache: with open(self.dataset_cache_file, "rb") as f: dataset_cache = pickle.load(f) else: dataset_cache = {} # check if our dataset is in the cache # if there, use that metadata, if not, then regenerate the metadata if dataset_cache.get(root) is None or not use_dataset_cache: files = list(Path(root).iterdir()) for fname in sorted(files): metadata, num_slices = self._retrieve_metadata(fname) self.examples += [ (fname, slice_ind, metadata) for slice_ind in range(num_slices) ] if dataset_cache.get(root) is None and use_dataset_cache: dataset_cache[root] = self.examples logging.info(f"Saving dataset cache to {self.dataset_cache_file}.") with open(self.dataset_cache_file, "wb") as f: pickle.dump(dataset_cache, f) else: logging.info(f"Using dataset cache from {self.dataset_cache_file}.") self.examples = dataset_cache[root] # subsample if desired if sample_rate < 1.0: # sample by slice random.shuffle(self.examples) num_examples = round(len(self.examples) * sample_rate) self.examples = self.examples[:num_examples] elif volume_sample_rate < 1.0: # sample by volume vol_names = sorted(list(set([f[0].stem for f in self.examples]))) random.shuffle(vol_names) num_volumes = round(len(vol_names) * volume_sample_rate) sampled_vols = vol_names[:num_volumes] self.examples = [ example for example in self.examples if example[0].stem in sampled_vols ] if num_cols: self.examples = [ ex for ex in self.examples if ex[2]["encoding_size"][1] in num_cols # type: ignore ] def _retrieve_metadata(self, fname): with h5py.File(fname, "r") as hf: et_root = etree.fromstring(hf["ismrmrd_header"][()]) enc = ["encoding", "encodedSpace", "matrixSize"] enc_size = ( int(et_query(et_root, enc + ["x"])), int(et_query(et_root, enc + ["y"])), int(et_query(et_root, enc + ["z"])), ) rec = ["encoding", "reconSpace", "matrixSize"] recon_size = ( int(et_query(et_root, rec + ["x"])), int(et_query(et_root, rec + ["y"])), int(et_query(et_root, rec + ["z"])), ) lims = ["encoding", "encodingLimits", "kspace_encoding_step_1"] enc_limits_center = int(et_query(et_root, lims + ["center"])) enc_limits_max = int(et_query(et_root, lims + ["maximum"])) + 1 padding_left = enc_size[1] // 2 - enc_limits_center padding_right = padding_left + enc_limits_max num_slices = hf["kspace"].shape[0] metadata = { "padding_left": padding_left, "padding_right": padding_right, "encoding_size": enc_size, "recon_size": recon_size, } return metadata, num_slices def __len__(self): return len(self.examples) def __getitem__(self, i: int): fname, dataslice, metadata = self.examples[i] with h5py.File(fname, "r") as hf: kspace = hf["kspace"][dataslice] mask = np.asarray(hf["mask"]) if "mask" in hf else None target = hf[self.recons_key][dataslice] if self.recons_key in hf else None attrs = dict(hf.attrs) attrs.update(metadata) if self.transform is None: sample = (kspace, mask, target, attrs, fname.name, dataslice) else: sample = self.transform(kspace, mask, target, attrs, fname.name, dataslice) return sample
13,630
36.759003
116
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/data/subsample.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ import contextlib from typing import Optional, Sequence, Tuple, Union import numpy as np import torch @contextlib.contextmanager def temp_seed(rng: np.random, seed: Optional[Union[int, Tuple[int, ...]]]): if seed is None: try: yield finally: pass else: state = rng.get_state() rng.seed(seed) try: yield finally: rng.set_state(state) class MaskFunc: """ An object for GRAPPA-style sampling masks. This crates a sampling mask that densely samples the center while subsampling outer k-space regions based on the undersampling factor. """ def __init__(self, center_fractions: Sequence[float], accelerations: Sequence[int]): """ Args: center_fractions: Fraction of low-frequency columns to be retained. If multiple values are provided, then one of these numbers is chosen uniformly each time. accelerations: Amount of under-sampling. This should have the same length as center_fractions. If multiple values are provided, then one of these is chosen uniformly each time. """ if not len(center_fractions) == len(accelerations): raise ValueError( "Number of center fractions should match number of accelerations" ) self.center_fractions = center_fractions self.accelerations = accelerations self.rng = np.random.RandomState() # pylint: disable=no-member def __call__( self, shape: Sequence[int], seed: Optional[Union[int, Tuple[int, ...]]] = None ) -> torch.Tensor: raise NotImplementedError def choose_acceleration(self): """Choose acceleration based on class parameters.""" choice = self.rng.randint(0, len(self.accelerations)) center_fraction = self.center_fractions[choice] acceleration = self.accelerations[choice] return center_fraction, acceleration class RandomMaskFunc(MaskFunc): """ RandomMaskFunc creates a sub-sampling mask of a given shape. The mask selects a subset of columns from the input k-space data. If the k-space data has N columns, the mask picks out: 1. N_low_freqs = (N * center_fraction) columns in the center corresponding to low-frequencies. 2. The other columns are selected uniformly at random with a probability equal to: prob = (N / acceleration - N_low_freqs) / (N - N_low_freqs). This ensures that the expected number of columns selected is equal to (N / acceleration). It is possible to use multiple center_fractions and accelerations, in which case one possible (center_fraction, acceleration) is chosen uniformly at random each time the RandomMaskFunc object is called. For example, if accelerations = [4, 8] and center_fractions = [0.08, 0.04], then there is a 50% probability that 4-fold acceleration with 8% center fraction is selected and a 50% probability that 8-fold acceleration with 4% center fraction is selected. """ def __call__( self, shape: Sequence[int], seed: Optional[Union[int, Tuple[int, ...]]] = None ) -> torch.Tensor: """ Create the mask. Args: shape: The shape of the mask to be created. The shape should have at least 3 dimensions. Samples are drawn along the second last dimension. seed: Seed for the random number generator. Setting the seed ensures the same mask is generated each time for the same shape. The random state is reset afterwards. Returns: A mask of the specified shape. """ if len(shape) < 3: raise ValueError("Shape should have 3 or more dimensions") with temp_seed(self.rng, seed): num_cols = shape[-2] center_fraction, acceleration = self.choose_acceleration() # create the mask num_low_freqs = int(round(num_cols * center_fraction)) prob = (num_cols / acceleration - num_low_freqs) / ( num_cols - num_low_freqs ) mask = self.rng.uniform(size=num_cols) < prob pad = (num_cols - num_low_freqs + 1) // 2 mask[pad : pad + num_low_freqs] = True # reshape the mask mask_shape = [1 for _ in shape] mask_shape[-2] = num_cols mask = torch.from_numpy(mask.reshape(*mask_shape).astype(np.float32)) return mask class EquispacedMaskFunc(MaskFunc): """ EquispacedMaskFunc creates a sub-sampling mask of a given shape. The mask selects a subset of columns from the input k-space data. If the k-space data has N columns, the mask picks out: 1. N_low_freqs = (N * center_fraction) columns in the center corresponding tovlow-frequencies. 2. The other columns are selected with equal spacing at a proportion that reaches the desired acceleration rate taking into consideration the number of low frequencies. This ensures that the expected number of columns selected is equal to (N / acceleration) It is possible to use multiple center_fractions and accelerations, in which case one possible (center_fraction, acceleration) is chosen uniformly at random each time the EquispacedMaskFunc object is called. Note that this function may not give equispaced samples (documented in https://github.com/facebookresearch/fastMRI/issues/54), which will require modifications to standard GRAPPA approaches. Nonetheless, this aspect of the function has been preserved to match the public multicoil data. """ def __call__( self, shape: Sequence[int], seed: Optional[Union[int, Tuple[int, ...]]] = None ) -> torch.Tensor: """ Args: shape: The shape of the mask to be created. The shape should have at least 3 dimensions. Samples are drawn along the second last dimension. seed: Seed for the random number generator. Setting the seed ensures the same mask is generated each time for the same shape. The random state is reset afterwards. Returns: A mask of the specified shape. """ if len(shape) < 3: raise ValueError("Shape should have 3 or more dimensions") with temp_seed(self.rng, seed): center_fraction, acceleration = self.choose_acceleration() num_cols = shape[-2] num_low_freqs = int(round(num_cols * center_fraction)) # create the mask mask = np.zeros(num_cols, dtype=np.float32) pad = (num_cols - num_low_freqs + 1) // 2 mask[pad : pad + num_low_freqs] = True # determine acceleration rate by adjusting for the number of low frequencies adjusted_accel = (acceleration * (num_low_freqs - num_cols)) / ( num_low_freqs * acceleration - num_cols ) offset = self.rng.randint(0, round(adjusted_accel)) accel_samples = np.arange(offset, num_cols - 1, adjusted_accel) accel_samples = np.around(accel_samples).astype(np.uint) mask[accel_samples] = True # reshape the mask mask_shape = [1 for _ in shape] mask_shape[-2] = num_cols mask = torch.from_numpy(mask.reshape(*mask_shape).astype(np.float32)) return mask def create_mask_for_mask_type( mask_type_str: str, center_fractions: Sequence[float], accelerations: Sequence[int], ) -> MaskFunc: """ Creates a mask of the specified type. Args: center_fractions: What fraction of the center of k-space to include. accelerations: What accelerations to apply. """ if mask_type_str == "random": return RandomMaskFunc(center_fractions, accelerations) elif mask_type_str == "equispaced": return EquispacedMaskFunc(center_fractions, accelerations) else: raise Exception(f"{mask_type_str} not supported")
8,448
36.887892
88
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/data/__init__.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ from .mri_data import SliceDataset, CombinedSliceDataset from .volume_sampler import VolumeSampler
278
26.9
63
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/fastmri/data/transforms.py
""" Copyright (c) Facebook, Inc. and its affiliates. This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree. """ from typing import Dict, Optional, Sequence, Tuple, Union import fastmri import numpy as np import torch from .subsample import MaskFunc def to_tensor(data: np.ndarray) -> torch.Tensor: """ Convert numpy array to PyTorch tensor. For complex arrays, the real and imaginary parts are stacked along the last dimension. Args: data: Input numpy array. Returns: PyTorch version of data. """ if np.iscomplexobj(data): data = np.stack((data.real, data.imag), axis=-1) return torch.from_numpy(data) def tensor_to_complex_np(data: torch.Tensor) -> np.ndarray: """ Converts a complex torch tensor to numpy array. Args: data: Input data to be converted to numpy. Returns: Complex numpy version of data. """ data = data.numpy() return data[..., 0] + 1j * data[..., 1] def apply_mask( data: torch.Tensor, mask_func: MaskFunc, seed: Optional[Union[int, Tuple[int, ...]]] = None, padding: Optional[Sequence[int]] = None, ) -> Tuple[torch.Tensor, torch.Tensor]: """ Subsample given k-space by multiplying with a mask. Args: data: The input k-space data. This should have at least 3 dimensions, where dimensions -3 and -2 are the spatial dimensions, and the final dimension has size 2 (for complex values). mask_func: A function that takes a shape (tuple of ints) and a random number seed and returns a mask. seed: Seed for the random number generator. padding: Padding value to apply for mask. Returns: tuple containing: masked data: Subsampled k-space data mask: The generated mask """ shape = np.array(data.shape) shape[:-3] = 1 mask = mask_func(shape, seed) if padding is not None: mask[:, :, : padding[0]] = 0 mask[:, :, padding[1] :] = 0 # padding value inclusive on right of zeros masked_data = data * mask + 0.0 # the + 0.0 removes the sign of the zeros return masked_data, mask def mask_center(x: torch.Tensor, mask_from: int, mask_to: int) -> torch.Tensor: """ Initializes a mask with the center filled in. Args: mask_from: Part of center to start filling. mask_to: Part of center to end filling. Returns: A mask with the center filled. """ mask = torch.zeros_like(x) mask[:, :, :, mask_from:mask_to] = x[:, :, :, mask_from:mask_to] return mask def center_crop(data: torch.Tensor, shape: Tuple[int, int]) -> torch.Tensor: """ Apply a center crop to the input real image or batch of real images. Args: data: The input tensor to be center cropped. It should have at least 2 dimensions and the cropping is applied along the last two dimensions. shape: The output shape. The shape should be smaller than the corresponding dimensions of data. Returns: The center cropped image. """ if not (0 < shape[0] <= data.shape[-2] and 0 < shape[1] <= data.shape[-1]): raise ValueError("Invalid shapes.") w_from = (data.shape[-2] - shape[0]) // 2 h_from = (data.shape[-1] - shape[1]) // 2 w_to = w_from + shape[0] h_to = h_from + shape[1] return data[..., w_from:w_to, h_from:h_to] def complex_center_crop(data: torch.Tensor, shape: Tuple[int, int]) -> torch.Tensor: """ Apply a center crop to the input image or batch of complex images. Args: data: The complex input tensor to be center cropped. It should have at least 3 dimensions and the cropping is applied along dimensions -3 and -2 and the last dimensions should have a size of 2. shape: The output shape. The shape should be smaller than the corresponding dimensions of data. Returns: The center cropped image """ if not (0 < shape[0] <= data.shape[-3] and 0 < shape[1] <= data.shape[-2]): raise ValueError("Invalid shapes.") w_from = (data.shape[-3] - shape[0]) // 2 h_from = (data.shape[-2] - shape[1]) // 2 w_to = w_from + shape[0] h_to = h_from + shape[1] return data[..., w_from:w_to, h_from:h_to, :] def center_crop_to_smallest( x: torch.Tensor, y: torch.Tensor ) -> Tuple[torch.Tensor, torch.Tensor]: """ Apply a center crop on the larger image to the size of the smaller. The minimum is taken over dim=-1 and dim=-2. If x is smaller than y at dim=-1 and y is smaller than x at dim=-2, then the returned dimension will be a mixture of the two. Args: x: The first image. y: The second image. Returns: tuple of tensors x and y, each cropped to the minimim size. """ smallest_width = min(x.shape[-1], y.shape[-1]) smallest_height = min(x.shape[-2], y.shape[-2]) x = center_crop(x, (smallest_height, smallest_width)) y = center_crop(y, (smallest_height, smallest_width)) return x, y def normalize( data: torch.Tensor, mean: Union[float, torch.Tensor], stddev: Union[float, torch.Tensor], eps: Union[float, torch.Tensor] = 0.0, ) -> torch.Tensor: """ Normalize the given tensor. Applies the formula (data - mean) / (stddev + eps). Args: data: Input data to be normalized. mean: Mean value. stddev: Standard deviation. eps: Added to stddev to prevent dividing by zero. Returns: Normalized tensor. """ return (data - mean) / (stddev + eps) def normalize_instance( data: torch.Tensor, eps: Union[float, torch.Tensor] = 0.0 ) -> Tuple[torch.Tensor, Union[torch.Tensor], Union[torch.Tensor]]: """ Normalize the given tensor with instance norm/ Applies the formula (data - mean) / (stddev + eps), where mean and stddev are computed from the data itself. Args: data: Input data to be normalized eps: Added to stddev to prevent dividing by zero. Returns: torch.Tensor: Normalized tensor """ mean = data.mean() std = data.std() return normalize(data, mean, std, eps), mean, std class UnetDataTransform: """ Data Transformer for training U-Net models. """ def __init__( self, which_challenge: str, mask_func: Optional[MaskFunc] = None, use_seed: bool = True, ): """ Args: which_challenge: Challenge from ("singlecoil", "multicoil"). mask_func: Optional; A function that can create a mask of appropriate shape. use_seed: If true, this class computes a pseudo random number generator seed from the filename. This ensures that the same mask is used for all the slices of a given volume every time. """ if which_challenge not in ("singlecoil", "multicoil"): raise ValueError("Challenge should either be 'singlecoil' or 'multicoil'") self.mask_func = mask_func self.which_challenge = which_challenge self.use_seed = use_seed def __call__( self, kspace: np.ndarray, mask: np.ndarray, target: np.ndarray, attrs: Dict, fname: str, slice_num: int, ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, torch.Tensor, str, int, float]: """ Args: kspace: Input k-space of shape (num_coils, rows, cols) for multi-coil data or (rows, cols) for single coil data. mask: Mask from the test dataset. target: Target image. attrs: Acquisition related information stored in the HDF5 object. fname: File name. slice_num: Serial number of the slice. Returns: tuple containing: image: Zero-filled input image. target: Target image converted to a torch.Tensor. mean: Mean value used for normalization. std: Standard deviation value used for normalization. fname: File name. slice_num: Serial number of the slice. """ kspace = to_tensor(kspace) # check for max value max_value = attrs["max"] if "max" in attrs.keys() else 0.0 # apply mask if self.mask_func: seed = None if not self.use_seed else tuple(map(ord, fname)) masked_kspace, mask = apply_mask(kspace, self.mask_func, seed) else: masked_kspace = kspace # inverse Fourier transform to get zero filled solution image = fastmri.ifft2c(masked_kspace) # crop input to correct size if target is not None: crop_size = (target.shape[-2], target.shape[-1]) else: crop_size = (attrs["recon_size"][0], attrs["recon_size"][1]) # check for FLAIR 203 if image.shape[-2] < crop_size[1]: crop_size = (image.shape[-2], image.shape[-2]) image = complex_center_crop(image, crop_size) # absolute value image = fastmri.complex_abs(image) # apply Root-Sum-of-Squares if multicoil data if self.which_challenge == "multicoil": image = fastmri.rss(image) # normalize input image, mean, std = normalize_instance(image, eps=1e-11) image = image.clamp(-6, 6) # normalize target if target is not None: target = to_tensor(target) target = center_crop(target, crop_size) target = normalize(target, mean, std, eps=1e-11) target = target.clamp(-6, 6) else: target = torch.Tensor([0]) return image, target, mean, std, fname, slice_num, max_value class VarNetDataTransform: """ Data Transformer for training VarNet models. """ def __init__(self, mask_func: Optional[MaskFunc] = None, use_seed: bool = True): """ Args: mask_func: Optional; A function that can create a mask of appropriate shape. Defaults to None. use_seed: If True, this class computes a pseudo random number generator seed from the filename. This ensures that the same mask is used for all the slices of a given volume every time. """ self.mask_func = mask_func self.use_seed = use_seed def __call__( self, kspace: np.ndarray, mask: np.ndarray, target: np.ndarray, attrs: Dict, fname: str, slice_num: int, ) -> Tuple[torch.Tensor, torch.Tensor, torch.Tensor, str, int, float, torch.Tensor]: """ Args: kspace: Input k-space of shape (num_coils, rows, cols) for multi-coil data. mask: Mask from the test dataset. target: Target image. attrs: Acquisition related information stored in the HDF5 object. fname: File name. slice_num: Serial number of the slice. Returns: tuple containing: masked_kspace: k-space after applying sampling mask. mask: The applied sampling mask target: The target image (if applicable). fname: File name. slice_num: The slice index. max_value: Maximum image value. crop_size: The size to crop the final image. """ if target is not None: target = to_tensor(target) max_value = attrs["max"] else: target = torch.tensor(0) max_value = 0.0 kspace = to_tensor(kspace) seed = None if not self.use_seed else tuple(map(ord, fname)) acq_start = attrs["padding_left"] acq_end = attrs["padding_right"] crop_size = torch.tensor([attrs["recon_size"][0], attrs["recon_size"][1]]) if self.mask_func: masked_kspace, mask = apply_mask( kspace, self.mask_func, seed, (acq_start, acq_end) ) else: masked_kspace = kspace shape = np.array(kspace.shape) num_cols = shape[-2] shape[:-3] = 1 mask_shape = [1] * len(shape) mask_shape[-2] = num_cols mask = torch.from_numpy(mask.reshape(*mask_shape).astype(np.float32)) mask = mask.reshape(*mask_shape) mask[:, :, :acq_start] = 0 mask[:, :, acq_end:] = 0 return ( masked_kspace, mask.byte(), target, fname, slice_num, max_value, crop_size, )
12,887
30.205811
88
py
imaging_MLPs
imaging_MLPs-master/compressed_sensing/networks/img2img_mixer.py
import torch import torch.nn as nn import torch.nn.functional as F from torch.nn import init import torch.nn.init as init import numpy as np import einops from einops.layers.torch import Rearrange from einops import rearrange class PatchEmbedding(nn.Module): def __init__( self, patch_size: int, embed_dim: int, channels: int ): super().__init__() self.proj = nn.Sequential( nn.Conv2d( in_channels=channels, out_channels=embed_dim, kernel_size=patch_size, stride=patch_size ), Rearrange("b c h w -> b h w c") ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.proj(x) class PatchExpansion(nn.Module): def __init__(self, dim_scale, channel_dim, img_channels, norm_layer=nn.LayerNorm): super().__init__() self.dim_scale = dim_scale self.expand = nn.Linear(channel_dim, dim_scale**2* channel_dim, bias=False) self.output_dim = channel_dim self.norm = norm_layer(channel_dim) self.output = nn.Conv2d(in_channels=channel_dim,out_channels=img_channels ,kernel_size=1,bias=False) def forward(self, x): """ x: B, H, W, C """ x = self.expand(x) B, H, W, C = x.shape x = rearrange(x, 'b h w (p1 p2 c)-> b (h p1) (w p2) c', p1=self.dim_scale, p2=self.dim_scale, c=C//(self.dim_scale**2)) x = x.view(B,-1,self.output_dim) x= self.norm(x) x = x.view(B,H*self.dim_scale, W*self.dim_scale,-1) x = x.permute(0,3,1,2) x = self.output(x) return x class MLPBlock(nn.Module): def __init__(self, input_dim: int, hidden_dim: int): super().__init__() self.model = nn.Sequential( nn.Linear(input_dim, hidden_dim), nn.GELU(), nn.Linear(hidden_dim, input_dim) ) def forward(self, x: torch.Tensor) -> torch.Tensor: return self.model(x) class Mixer(nn.Module): def __init__( self, num_patches: int, num_channels: int, f_hidden: int ): super().__init__() self.token_mixing = nn.Sequential( nn.LayerNorm(num_channels), Rearrange("b h w c -> b c w h"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c w h -> b c h w"), MLPBlock(num_patches, num_patches*f_hidden), Rearrange("b c h w -> b h w c"), ) self.channel_mixing = nn.Sequential( nn.LayerNorm(num_channels), MLPBlock(num_channels, num_channels*f_hidden) ) def forward(self, x: torch.Tensor) -> torch.Tensor: x = x + self.token_mixing(x) x = x + self.channel_mixing(x) return x class Img2Img_Mixer(nn.Module): def __init__( self, img_size: int = 320, img_channels: int = 1, patch_size: int = 4, embed_dim: int = 128, num_layers: int = 16, f_hidden: int = 8, ): super().__init__() self.patch_embed = PatchEmbedding(patch_size, embed_dim, img_channels) layers = [ Mixer(img_size//patch_size, embed_dim, f_hidden) for _ in range(num_layers)] self.mixer_layers = nn.Sequential(*layers) self.patch_expand = PatchExpansion(patch_size, embed_dim, img_channels) def forward(self, x: torch.Tensor) -> torch.Tensor: x = self.patch_embed(x) x = self.mixer_layers(x) x = self.patch_expand(x) return x
3,714
26.932331
127
py