File size: 7,447 Bytes
ff0fadf | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 | import torch
import math
import torch.nn as nn
import numpy as np
import torch.nn.functional as F
from timm.layers import trunc_normal_
from onescience.modules.fourier.MultiWaveletTransform import (
MultiWaveletTransform1D,
MultiWaveletTransform2D,
MultiWaveletTransform3D,
)
from onescience.modules.mlp.MLP import StandardMLP
from onescience.modules.embedding import timestep_embedding, unified_pos_embedding
from onescience.modules.fourier.geo_spectral import GeoSpectralConv2d, IPHI
MultiWaveletTransformList = [
None,
MultiWaveletTransform1D,
MultiWaveletTransform2D,
MultiWaveletTransform3D,
]
class Model(nn.Module):
# this model requires H = W = Z and H, W, Z is the power of two
def __init__(
self, args, device, alpha=2, L=0, c=1, base="legendre", s1=128, s2=128
):
super(Model, self).__init__()
self.__name__ = "MWT"
self.args = args
self.k = args.mwt_k
self.WMT_dim = c * self.k**2
if args.geotype == "structured_1D":
self.WMT_dim = c * self.k
self.c = c
self.s1 = s1
self.s2 = s2
## embedding
if (
args.unified_pos and args.geotype != "unstructured"
): # only for structured mesh
self.pos = unified_pos_embedding(args.shapelist, args.ref, device=device)
self.preprocess = StandardMLP(
input_dim=args.fun_dim + args.ref ** len(args.shapelist),
output_dim=self.WMT_dim,
hidden_dims=[args.n_hidden * 2],
activation=args.act,
use_bias=True
)
else:
self.preprocess = StandardMLP(
input_dim=args.fun_dim + args.space_dim,
output_dim=self.WMT_dim,
hidden_dims=[args.n_hidden * 2],
activation=args.act,
use_bias=True
)
if args.time_input:
self.time_fc = nn.Sequential(
nn.Linear(self.WMT_dim, args.n_hidden),
nn.SiLU(),
nn.Linear(args.n_hidden, self.WMT_dim),
)
# geometry projection
if self.args.geotype == "unstructured":
self.fftproject_in = GeoSpectralConv2d(
in_channels=self.WMT_dim,
out_channels=self.WMT_dim,
modes1=args.modes,
modes2=args.modes,
s1=s1,
s2=s2
)
self.fftproject_out = GeoSpectralConv2d(
in_channels=self.WMT_dim,
out_channels=self.WMT_dim,
modes1=args.modes,
modes2=args.modes,
s1=s1,
s2=s2
)
self.iphi = IPHI()
self.augmented_resolution = [s1, s2]
self.padding = [(16 - size % 16) % 16 for size in [s1, s2]]
else:
target = 2 ** (math.ceil(np.log2(max(args.shapelist))))
self.padding = [(target - size) for size in args.shapelist]
self.augmented_resolution = [target for _ in range(len(self.padding))]
dim = len(self.padding)
transform_class = MultiWaveletTransformList[dim]
self.spectral_layers = nn.ModuleList(
[
transform_class(
k=self.k,
alpha=alpha,
L=L,
c=c,
base=base
)
for _ in range(args.n_layers)
]
)
# projectors
self.fc1 = nn.Linear(self.WMT_dim, args.n_hidden)
self.fc2 = nn.Linear(args.n_hidden, args.out_dim)
def structured_geo(self, x, fx, T=None):
B, N, _ = x.shape
if self.args.unified_pos:
x = self.pos.repeat(x.shape[0], 1, 1)
if fx is not None:
fx = torch.cat((x, fx), -1)
fx = self.preprocess(fx)
else:
fx = self.preprocess(x)
if T is not None:
Time_emb = timestep_embedding(T, self.WMT_dim)
Time_emb = self.time_fc(Time_emb)
if Time_emb.ndim == 2:
Time_emb = Time_emb.unsqueeze(1)
fx = fx + Time_emb
x = fx.permute(0, 2, 1).reshape(B, self.WMT_dim, *self.args.shapelist)
if not all(item == 0 for item in self.padding):
if len(self.args.shapelist) == 2:
x = F.pad(x, [0, self.padding[1], 0, self.padding[0]])
elif len(self.args.shapelist) == 3:
x = F.pad(
x, [0, self.padding[2], 0, self.padding[1], 0, self.padding[0]]
)
x = (
x.reshape(B, self.WMT_dim, -1)
.permute(0, 2, 1)
.contiguous()
.reshape(
B,
*self.augmented_resolution,
self.c,
self.k**2 if self.args.geotype != "structured_1D" else self.k
)
)
for i in range(self.args.n_layers):
x = self.spectral_layers[i](x)
if i < self.args.n_layers - 1:
x = F.gelu(x)
x = (
x.reshape(B, -1, self.WMT_dim)
.permute(0, 2, 1)
.contiguous()
.reshape(B, self.WMT_dim, *self.augmented_resolution)
)
if not all(item == 0 for item in self.padding):
if len(self.args.shapelist) == 2:
x = x[..., : -self.padding[0], : -self.padding[1]]
elif len(self.args.shapelist) == 3:
x = x[..., : -self.padding[0], : -self.padding[1], : -self.padding[2]]
x = x.reshape(B, self.WMT_dim, -1).permute(0, 2, 1)
x = self.fc1(x)
x = F.gelu(x)
x = self.fc2(x)
return x
def unstructured_geo(self, x, fx, T=None):
B, N, _ = x.shape
original_pos = x
if fx is not None:
fx = torch.cat((x, fx), -1)
fx = self.preprocess(fx)
else:
fx = self.preprocess(x)
if T is not None:
Time_emb = timestep_embedding(T, self.WMT_dim)
Time_emb = self.time_fc(Time_emb)
if Time_emb.ndim == 2:
Time_emb = Time_emb.unsqueeze(1)
fx = fx + Time_emb
x = self.fftproject_in(
fx.permute(0, 2, 1), x_in=original_pos, iphi=self.iphi, code=None
)
x = (
x.reshape(B, self.WMT_dim, -1)
.permute(0, 2, 1)
.contiguous()
.reshape(B, *self.augmented_resolution, self.c, self.k**2)
)
for i in range(self.args.n_layers):
x = self.spectral_layers[i](x)
if i < self.args.n_layers - 1:
x = F.gelu(x)
x = (
x.reshape(B, -1, self.WMT_dim)
.permute(0, 2, 1)
.contiguous()
.reshape(B, self.WMT_dim, *self.augmented_resolution)
)
x = self.fftproject_out(
x, x_out=original_pos, iphi=self.iphi, code=None
).permute(0, 2, 1)
x = self.fc1(x)
x = F.gelu(x)
x = self.fc2(x)
return x
def forward(self, x, fx, T=None, geo=None):
if self.args.geotype == "unstructured":
return self.unstructured_geo(x, fx, T)
else:
return self.structured_geo(x, fx, T)
|