Oucru-Innovations / vital-DSP / 25736509173

95%

push

github

perf: vectorize bottlenecks and reduce test epochs for faster CI

- Replace O(n²) loops with KDTree batch queries in sample_entropy,
  approximate_entropy, and correlation_dimension
- Replace Python median filter loops with scipy.ndimage.median_filter
  in artifact_removal, artifact_detection_removal, and
  harmonic_percussive_separation
- Batch Lyapunov KDTree queries instead of per-iteration calls
- Vectorize sliding window variance/energy with as_strided
- Reduce ml_models test epochs to 1 (3 where logic requires it)
  cutting test suite from ~31min to ~13min

2711 of 2776 branches covered (97.66%)

Branch coverage included in aggregate %.

16727 of 17634 relevant lines covered (94.86%)

0.95 hits per line