File size: 6,879 Bytes
b88c26d | 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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 | """
BURT-IMMA Utilities
License: BSL-1.1
Contact: jessica@collectivekitty.com
"""
import os
import time
import random
from typing import Any, Dict, Optional
from contextlib import contextmanager
import numpy as np
try:
import torch
HAS_TORCH = True
except ImportError:
HAS_TORCH = False
try:
import yaml
HAS_YAML = True
except ImportError:
HAS_YAML = False
def spectral_norm(W) -> float:
"""Compute the largest singular value (spectral norm) of a matrix.
Args:
W: numpy array or torch Tensor of shape (m, n)
Returns:
Largest singular value as a float
"""
if HAS_TORCH and isinstance(W, torch.Tensor):
s = torch.linalg.svdvals(W)
return s[0].item()
else:
W_np = np.asarray(W)
if W_np.ndim < 2:
return float(np.abs(W_np).max())
s = np.linalg.svd(W_np, compute_uv=False)
return float(s[0])
def entropy(p) -> float:
"""Compute the Shannon entropy of a probability distribution.
Args:
p: numpy array or torch Tensor representing a probability distribution
(must sum to 1, all elements >= 0)
Returns:
Entropy in nats (natural log base)
"""
if HAS_TORCH and isinstance(p, torch.Tensor):
p_clamped = p.clamp(min=1e-10)
return -(p_clamped * p_clamped.log()).sum().item()
else:
p_np = np.asarray(p, dtype=np.float64)
p_clipped = np.clip(p_np, 1e-10, 1.0)
return float(-np.sum(p_clipped * np.log(p_clipped)))
def check_huntington(actor, x, y) -> Dict[str, bool]:
"""Verify Huntington postulates for a Boolean algebra actor.
The Huntington postulates define a Boolean algebra (B, +, *, ', 0, 1):
H1 (Commutativity): x + y = y + x, x * y = y * x
H2 (Distributivity): x * (y + z) = (x*y) + (x*z),
x + (y * z) = (x+y) * (x+z)
H3 (Identity): x + 0 = x, x * 1 = x
H4 (Complement): x + x' = 1, x * x' = 0
Args:
actor: Object with methods `join(a, b)`, `meet(a, b)`, `complement(a)`,
and attributes `zero` and `one`.
x: First element
y: Second element
Returns:
Dict mapping postulate name to whether it holds
"""
results = {}
# H1: Commutativity
try:
h1_join = np.allclose(
np.asarray(actor.join(x, y)),
np.asarray(actor.join(y, x)),
atol=1e-6
)
h1_meet = np.allclose(
np.asarray(actor.meet(x, y)),
np.asarray(actor.meet(y, x)),
atol=1e-6
)
results["H1_commutativity"] = bool(h1_join and h1_meet)
except (AttributeError, TypeError):
results["H1_commutativity"] = False
# H3: Identity
try:
h3_join = np.allclose(
np.asarray(actor.join(x, actor.zero)),
np.asarray(x),
atol=1e-6
)
h3_meet = np.allclose(
np.asarray(actor.meet(x, actor.one)),
np.asarray(x),
atol=1e-6
)
results["H3_identity"] = bool(h3_join and h3_meet)
except (AttributeError, TypeError):
results["H3_identity"] = False
# H4: Complement
try:
x_comp = actor.complement(x)
h4_join = np.allclose(
np.asarray(actor.join(x, x_comp)),
np.asarray(actor.one),
atol=1e-6
)
h4_meet = np.allclose(
np.asarray(actor.meet(x, x_comp)),
np.asarray(actor.zero),
atol=1e-6
)
results["H4_complement"] = bool(h4_join and h4_meet)
except (AttributeError, TypeError):
results["H4_complement"] = False
return results
def load_config(path: str) -> Dict[str, Any]:
"""Load a YAML configuration file.
Args:
path: Path to YAML file
Returns:
Parsed configuration dictionary
Raises:
FileNotFoundError: If config file does not exist
ImportError: If PyYAML is not installed
"""
if not HAS_YAML:
raise ImportError("PyYAML is required: pip install pyyaml")
path = os.path.expanduser(path)
if not os.path.exists(path):
raise FileNotFoundError(f"Config file not found: {path}")
with open(path, "r") as f:
config = yaml.safe_load(f)
return config
def set_seed(seed: int = 42) -> None:
"""Set all random seeds for reproducibility.
Sets seeds for: Python random, NumPy, and PyTorch (CPU + CUDA).
Args:
seed: Integer seed value
"""
random.seed(seed)
np.random.seed(seed)
if HAS_TORCH:
torch.manual_seed(seed)
if torch.cuda.is_available():
torch.cuda.manual_seed(seed)
torch.cuda.manual_seed_all(seed)
# Ensure deterministic behavior where possible
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
class Timer:
"""Context manager for profiling code blocks.
Usage:
with Timer("forward pass") as t:
output = model(x)
print(t.elapsed) # seconds
# Or accumulate multiple measurements:
timer = Timer("training")
for batch in data:
with timer:
train_step(batch)
print(timer.total, timer.count, timer.mean)
"""
def __init__(self, name: str = "timer", verbose: bool = False):
self.name = name
self.verbose = verbose
self.elapsed: float = 0.0
self.total: float = 0.0
self.count: int = 0
self._start: Optional[float] = None
@property
def mean(self) -> float:
"""Mean elapsed time across all measurements."""
return self.total / max(self.count, 1)
def __enter__(self) -> "Timer":
if HAS_TORCH and torch.cuda.is_available():
torch.cuda.synchronize()
self._start = time.perf_counter()
return self
def __exit__(self, *args) -> None:
if HAS_TORCH and torch.cuda.is_available():
torch.cuda.synchronize()
self.elapsed = time.perf_counter() - self._start
self.total += self.elapsed
self.count += 1
self._start = None
if self.verbose:
print(f"[{self.name}] {self.elapsed*1000:.2f} ms")
def reset(self) -> None:
"""Reset all accumulated measurements."""
self.elapsed = 0.0
self.total = 0.0
self.count = 0
self._start = None
def __repr__(self) -> str:
return (f"Timer(name={self.name!r}, count={self.count}, "
f"total={self.total:.4f}s, mean={self.mean*1000:.2f}ms)")
|