Download code/plan_robust_calibration.py from PureOne/EVE-AITHRAEL-Robust-Causal-Calibration: direct link, hf CLI and curl.
- Browser
- Download file 2.31 kB
-
https://huggingface.co/datasets/PureOne/EVE-AITHRAEL-Robust-Causal-Calibration/resolve/main/code/plan_robust_calibration.py
- Command line
-
hf download hf://datasets/PureOne/EVE-AITHRAEL-Robust-Causal-Calibration/code/plan_robust_calibration.py
-
curl -L -o plan_robust_calibration.py https://huggingface.co/datasets/PureOne/EVE-AITHRAEL-Robust-Causal-Calibration/resolve/main/code/plan_robust_calibration.py
2.31 kB
| #!/usr/bin/env python3 | |
| """Exact small-instance planning CLI. See the manuscript for the data contract.""" | |
| import argparse | |
| import json | |
| from pathlib import Path | |
| from calibration_core import (rational_matrix, residual_spaces, generic_profile, | |
| common_generic_design, numerical_profile, jsonable) | |
| def plan(spec): | |
| n = int(spec["n"]) | |
| supports, spaces = residual_spaces(n, spec["edges"], spec.get("trusted_rows", [])) | |
| channels = [rational_matrix(H, n) for H in spec["sensor_spaces"]] | |
| if len(channels) > 16: | |
| raise ValueError("this transparent exhaustive planner is limited to 16 channels") | |
| profiles = [generic_profile(E, channels) for E in spaces] | |
| distances = [p["distance"] for p in profiles if p["distance"] is not None] | |
| optimum = min(distances) if distances else None | |
| f = int(spec.get("erasures", 0)); t = int(spec.get("errors", 0)) | |
| if f < 0 or t < 0: | |
| raise ValueError("negative fault budget") | |
| design = common_generic_design(spaces, channels, seed=int(spec.get("seed", 13371337))) | |
| actual = None | |
| if design["status"] == "certified": | |
| actual = [numerical_profile(E, design["rows"]) for E in spaces] | |
| return jsonable({"contract": "known aligned spaces; trusted installed constraints; linear discrepancy observations", | |
| "supports": supports, "residual_bases": spaces, | |
| "dimensions": [len(E) for E in spaces], | |
| "total_dimension": sum(map(len, spaces)), | |
| "dense_no_fault_channels": max(map(len, spaces), default=0), | |
| "optimal_architecture_distance": optimum, | |
| "requested_erasures": f, "requested_errors": t, | |
| "requested_tolerance_possible": optimum is None or optimum > f+2*t, | |
| "generic_profiles": profiles, "design": design, | |
| "numerical_certificates": actual}) | |
| if __name__ == "__main__": | |
| p = argparse.ArgumentParser(description=__doc__) | |
| p.add_argument("input", type=Path) | |
| p.add_argument("--output", type=Path) | |
| a = p.parse_args() | |
| result = plan(json.loads(a.input.read_text())) | |
| text = json.dumps(result, indent=2)+"\n" | |
| if a.output: | |
| a.output.write_text(text) | |
| else: | |
| print(text, end="") | |