File size: 3,009 Bytes
1cd8661 | 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 | #!/usr/bin/env python3
"""Compute paper-protocol peak-field normalization from training cases only."""
from __future__ import annotations
import argparse
import json
from pathlib import Path
import torch
from build_peak_targets import select_peak
from load_case import canonical_geometry, load_case, load_split
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--dataset-root", type=Path, default=Path("."))
parser.add_argument("--geometry", required=True)
parser.add_argument("--output", type=Path, required=True)
args = parser.parse_args()
geometry = canonical_geometry(args.geometry)
train_cases = load_split(args.dataset_root)["train"]
target_sum_sq = torch.zeros(4, dtype=torch.float64)
target_count = torch.zeros(4, dtype=torch.int64)
condition_rows = []
for offset, case_id in enumerate(train_cases, start=1):
data = load_case(args.dataset_root, geometry, case_id)
_, _, displacement, stress = select_peak(data)
for component in range(3):
values = displacement[:, component].to(torch.float64)
target_sum_sq[component] += torch.sum(values.square())
target_count[component] += values.numel()
stress64 = stress.to(torch.float64)
target_sum_sq[3] += torch.sum(stress64.square())
target_count[3] += stress64.numel()
condition_rows.append(
torch.cat(
(
data["velocity_xyz"].to(torch.float64).reshape(3),
data["mass_ratio"].to(torch.float64).reshape(1),
data["material_young_mpa"].to(torch.float64).reshape(1),
data["material_poisson"].to(torch.float64).reshape(1),
)
)
)
if offset % 50 == 0:
print(f"[{geometry}] normalization {offset}/{len(train_cases)}")
scale = torch.sqrt(target_sum_sq / target_count)
conditions = torch.stack(condition_rows)
payload = {
"geometry": geometry,
"split": "train",
"train_cases": len(train_cases),
"target_scale_definition": (
"training-split RMS of peak-state x/y/z displacement and "
"same-state max-IP von Mises effective stress"
),
"target_feature_names": ["disp_x", "disp_y", "disp_z", "effective_stress"],
"target_scale": scale.tolist(),
"condition_feature_names": [
"velocity_x",
"velocity_y",
"velocity_z",
"mass_ratio",
"material_young_mpa",
"material_poisson",
],
"condition_mean": conditions.mean(dim=0).tolist(),
"condition_std_sample": conditions.std(dim=0, unbiased=True).tolist(),
}
args.output.parent.mkdir(parents=True, exist_ok=True)
args.output.write_text(json.dumps(payload, indent=2) + "\n", encoding="utf-8")
print(json.dumps(payload, indent=2))
if __name__ == "__main__":
main()
|