Download modal_diagnostics/export_modes.py from soundai2016/ADComp: direct link, hf CLI and curl.
- Browser
- Download file 3.96 kB
-
https://huggingface.co/soundai2016/ADComp/resolve/main/modal_diagnostics/export_modes.py
- Command line
-
hf download hf://soundai2016/ADComp/modal_diagnostics/export_modes.py
-
curl -L -o export_modes.py https://huggingface.co/soundai2016/ADComp/resolve/main/modal_diagnostics/export_modes.py
3.96 kB
| #!/usr/bin/env python3 | |
| """Export pressure modes selected by the fixed coupling-frequency criterion.""" | |
| import os | |
| for variable in ("OPENBLAS_NUM_THREADS", "OMP_NUM_THREADS", "MKL_NUM_THREADS"): | |
| os.environ[variable] = "1" | |
| import csv | |
| import json | |
| from pathlib import Path | |
| import numpy as np | |
| HERE = Path(__file__).resolve().parent | |
| def load(path): | |
| with np.load(path, allow_pickle=False) as arrays: | |
| return {key: arrays[key] for key in arrays.files} | |
| def main(): | |
| config = json.loads((HERE / "config.json").read_text()) | |
| selected, fields = [], {} | |
| for material in config["materials"]: | |
| with (HERE / "results" / f"{material}_modal_pairs.csv").open() as stream: | |
| pairs = [row for row in csv.DictReader(stream) if row["method"] == "craig_bampton" and row["m"] == "32"] | |
| def score(row): | |
| return float(row["fine_residue_norm"]) * abs(float(row["omega_rom"]) - float(row["omega_coarse"])) | |
| chosen = max(pairs, key=score) | |
| coarse_index, reduced_index, fine_index = (int(chosen[key]) for key in ("coarse_mode", "rom_mode", "fine_mode")) | |
| coarse = load(HERE / "results" / f"{material}_n32_spectrum.npz") | |
| fine = load(HERE / "results" / f"{material}_n128_spectrum.npz") | |
| reduced = load(HERE.parent / "accuracy_cost/results" / f"{material}_n32_craig_bampton_m32.npz") | |
| coarse_match = load(HERE / "results" / f"{material}_coarse_fine_matching.npz") | |
| reduced_match = load(HERE / "results" / f"{material}_craig_bampton_m32_matching.npz") | |
| cf = (coarse_match["coarse_indices"] == coarse_index) & (coarse_match["fine_indices"] == fine_index) | |
| rc = (reduced_match["rom_indices"] == reduced_index) & (reduced_match["coarse_indices"] == coarse_index) | |
| if cf.sum() != 1 or rc.sum() != 1: | |
| raise ValueError("The selected mode must have unique coarse/fine and reduced/coarse matches.") | |
| fine_sign = float(coarse_match["sign"][cf][0]) | |
| reduced_sign = float(reduced_match["sign"][rc][0]) | |
| fields[material + "_fine_pressure_mode"] = (fine_sign * fine["eigenvectors"][:, fine_index] / np.sqrt(fine["mass"])).reshape(128, 128) | |
| fields[material + "_coarse_pressure_mode"] = (coarse["eigenvectors"][:, coarse_index] / np.sqrt(coarse["mass"])).reshape(32, 32) | |
| fields[material + "_rom_pressure_mode"] = (reduced_sign * (reduced["orthonormal_basis"] @ reduced["reduced_eigenvectors"][:, reduced_index]) / np.sqrt(reduced["mass"])).reshape(32, 32) | |
| frequencies = [float(chosen[key]) / (2 * np.pi) for key in ("omega_fine", "omega_coarse", "omega_rom")] | |
| selected.append({"material": material, "criterion_value": score(chosen), "coarse_mode": coarse_index, | |
| "fine_mode": fine_index, "rom_mode": reduced_index, "fine_frequency": frequencies[0], | |
| "coarse_frequency": frequencies[1], "rom_frequency": frequencies[2], | |
| "mac_coarse_fine": float(chosen["mac_coarse_fine"]), "mac_rom_coarse": float(chosen["mac_rom_coarse"]), | |
| "fine_residue_norm": float(chosen["fine_residue_norm"]), "fine_sign_to_coarse": fine_sign, | |
| "rom_sign_to_coarse": reduced_sign}) | |
| np.savez_compressed(HERE / "results/illustrated_pressure_modes.npz", **fields) | |
| metadata = { | |
| "selection_rule": "For accepted Craig-Bampton m32 triples, maximize fine_residue_norm * abs(omega_rom - omega_coarse) separately for each material.", | |
| "normalization": "Unit mass-weighted pressure norm; divide the mass-normalized vector by sqrt(mass). Fine and reduced signs align to the coarse mode using the MAC assignment.", | |
| "interpretation": "Illustrative modes. Quantitative diagnostics use all accepted modes and retain unmatched contributions.", | |
| "selected_modes": selected, | |
| } | |
| (HERE / "results/illustrated_mode_selection.json").write_text(json.dumps(metadata, indent=2) + "\n") | |
| print(json.dumps(selected, indent=2)) | |
| if __name__ == "__main__": | |
| main() | |