File size: 5,827 Bytes
148bb80 | 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 | """Compare published peak responses without downloading the 1.12 GiB fields.
pip install numpy matplotlib huggingface_hub
python compare_excitations.py
python compare_excitations.py --index index.jsonl --configuration 119
SPDX-License-Identifier: Apache-2.0
Data: HaomingLuo/AgentFEM-Structural-Dynamics-Virtual-Sensing, CC BY 4.0.
"""
import argparse
import json
from pathlib import Path
import matplotlib
matplotlib.use("Agg")
import matplotlib.pyplot as plt
import numpy as np
REPO = "HaomingLuo/AgentFEM-Structural-Dynamics-Virtual-Sensing"
REVISION = "b171870a1efddf858245cf2ecf92f96b2280c47a"
KINDS = ("half_sine_pulse", "sub_resonant_sine", "near_resonant_sine", "chirp")
LABELS = ("Pulse", "Sub-resonant\nsine", "Near-resonant\nsine", "Chirp")
def load_cases(path):
rows = [json.loads(line) for line in Path(path).read_text().splitlines() if line.strip()]
cases = {}
for row in rows:
case = cases.setdefault(row["configuration_id"], {})
if row["excitation"] in case:
raise ValueError("Duplicate configuration/excitation pair")
case[row["excitation"]] = row
if len(rows) != 512 or len(cases) != 128:
raise ValueError("Expected the pinned v1.1 release: 512 trajectories / 128 configurations")
fields = ("length_m", "height_m", "young_pa", "density_kg_m3", "damping_ratio",
"traction_amplitude_pa", "first_frequency_hz", "split", "protocol_role")
for case in cases.values():
if set(case) != set(KINDS):
raise ValueError("Missing excitation")
reference = case[KINDS[0]]
for row in case.values():
if any(row[key] != reference[key] for key in fields):
raise ValueError("Comparisons require the same physical configuration")
peak = row["max_abs_tip_displacement_m"]
if not np.isfinite(peak) or peak <= 0:
raise ValueError("Invalid peak displacement")
return cases
def main():
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("--index", type=Path, help="Local v1.1 index; omit to download only the index")
parser.add_argument("--configuration", type=int, default=119)
parser.add_argument("--output", type=Path, default=Path("excitation-comparison.png"))
args = parser.parse_args()
if args.index is None:
from huggingface_hub import hf_hub_download
args.index = Path(hf_hub_download(REPO, "index.jsonl", repo_type="dataset", revision=REVISION))
cases = load_cases(args.index)
if args.configuration not in cases:
parser.error("configuration must be an integer from 0 to 127")
case = cases[args.configuration]
reference = case[KINDS[0]]
peaks = np.array([case[k]["max_abs_tip_displacement_m"] * 1000 for k in KINDS])
plt.rcParams.update({"font.family": "DejaVu Sans", "font.size": 10})
fig, axes = plt.subplots(1, 2, figsize=(12, 5.3))
ax = axes[0]
bars = ax.bar(np.arange(4), peaks, color=["#7792a2", "#368b90", "#c58a43", "#746899"], width=.62)
ax.bar_label(bars, labels=[f"{p:.4f}" for p in peaks], padding=5, fontsize=9)
ax.set_xticks(np.arange(4), LABELS)
ax.set_ylim(0, max(peaks) * 1.24)
ax.set_ylabel("Peak absolute tip displacement / mm")
ax.set_title(f"One structure, four load histories\nConfiguration {args.configuration:04d} · {reference['split']}", loc="left", pad=12)
ax = axes[1]
roles = [("id", "Test ID", "#368b90"), ("excitation_ood", "Excitation OOD", "#c58a43"),
("parameter_ood", "Parameter OOD", "#746899")]
upper = 0.0
counts = {}
for role, label, color in roles:
selected = [c for c in cases.values() if c[KINDS[0]]["split"] == "test" and c[KINDS[0]]["protocol_role"] == role]
counts[role] = len(selected)
x = np.array([c["sub_resonant_sine"]["max_abs_tip_displacement_m"] * 1000 for c in selected])
y = np.array([c["near_resonant_sine"]["max_abs_tip_displacement_m"] * 1000 for c in selected])
upper = max(upper, float(max(x.max(), y.max())))
ax.scatter(x, y, label=f"{label} ({len(selected)})", color=color, s=38, zorder=3)
if counts != {"id": 8, "excitation_ood": 4, "parameter_ood": 4}:
raise ValueError("Unexpected frozen test cohorts")
ax.plot([0, upper * 1.08], [0, upper * 1.08], "--", color="#8b9498", lw=1, label="Equal peak response")
ax.set(xlim=(0, upper * 1.08), ylim=(0, upper * 1.08),
xlabel="Sub-resonant sine peak / mm", ylabel="Near-resonant sine peak / mm")
ax.set_title("Paired responses on 16 test structures\nOne point = one physical configuration", loc="left", pad=12)
ax.legend(frameon=False, fontsize=8, loc="lower right")
for ax in axes:
ax.spines[["top", "right"]].set_visible(False)
ax.grid(axis="y", alpha=.15)
ax.set_axisbelow(True)
fig.suptitle("AgentFEM · How excitation changes structural response", x=.08, ha="left", fontsize=16, fontweight="bold")
fig.text(.08, .075, f"Selected case: f1 = {reference['first_frequency_hz']:.2f} Hz · damping = {100*reference['damping_ratio']:.2f}% · traction amplitude = {reference['traction_amplitude_pa']/1000:.3f} kPa", fontsize=9)
fig.text(.08, .025, "Synthetic FEM peaks over 0.5 s. Different load histories, not a steady-state frequency response or an ML score.", fontsize=9, color="#52606a")
fig.tight_layout(rect=(0, .12, 1, .92), w_pad=3)
fig.savefig(args.output, dpi=160)
plt.close(fig)
print(json.dumps({"configuration": args.configuration, "peak_displacement_mm": dict(zip(KINDS, peaks.tolist())),
"verified_index_trajectories": 512, "test_configurations": counts,
"source_revision": REVISION, "output": str(args.output)}, indent=2))
if __name__ == "__main__":
main()
|