---

name: check_physics description: 実装内容が数学的・物理学的に妥当か検証する

Verify Physics

This skill performs a review of the code specifically focusing on physics and math correctness.

Instructions

1. Dimensional Analysis:

   • Check if unit handling (e.g., astropy.units, quantities) is used correctly.
   • Ensure operations are physically valid (e.g., not adding Length to Time).
   • Unit-Safe Assignment: Avoid direct slice assignment series[:] = raw_array if series has units. Use series.value[:] = raw_array to update data while preserving unit metadata.

2. Mathematical Inspection:

   • Verify equations against known reference papers or standard textbooks.
   • Check for numerical stability issues (e.g., division by zero, floating point precision loss).
   • Check edge cases (e.g., $f=0$, infinite limits).

3. Sanity Checks:

   • Conservation Laws: Create a script to verify energy conservation (Parseval's theorem) for FFT/PSD transformations. Ratio of frequency-domain energy to time-domain energy should be ~1.0.
   • Theoretical Benchmarks: Use known signals (Sine wave, Gaussian noise) to verify peak frequency, amplitude, and spectral flatness.
   • Multi-domain Verification: For multi-dimensional fields, verify transformations along all applicable axes (e.g., spatial FFT as well as temporal FFT).
   • Verify that reasonable inputs produce physically reasonable outputs.

4. Report:

   • Provide a summary of the analysis, pointing out any suspect logic or confirming validity.