Issues

• base.py suggests nonexistent poetry groups; AUTO_PRIORITY's benchmark harness is absent from the repo; assorted dead code and stale paths

  ppsp-team/HyPyP#311 · Filter by author Ramdam17 opened on Jul 30, 2026

• compute_nmPLV applies the n:m exponent to the time axis instead of the channel axis, averages the wrong axis, and calls a deprecated helper

  ppsp-team/HyPyP#310 · Filter by author Ramdam17 opened on Jul 30, 2026

• CUDA: free_all_blocks() on every call defeats the CuPy memory pool; pair indices rebuilt per call in all four dispatchers

  ppsp-team/HyPyP#309 · Filter by author Ramdam17 opened on Jul 30, 2026

• Sync test suite: tolerances ~10^5-10^6 looser than measured error, a duplicate class silently drops 5 tests, and #279 has no regression test

  ppsp-team/HyPyP#308 · Filter by author Ramdam17 opened on Jul 30, 2026

• CI exercises zero accelerated backends: all 49 skipif-gated sync tests skip

  ppsp-team/HyPyP#307 · Filter by author Ramdam17 opened on Jul 30, 2026

• pair_connectivity cannot forward optimization/priority, and compute_sync masks all compute errors as "metric not supported"

  ppsp-team/HyPyP#306 · Filter by author Ramdam17 opened on Jul 30, 2026

• 'numba' inside a priority= list is silently ignored, contradicting three documented examples

  ppsp-team/HyPyP#305 · Filter by author Ramdam17 opened on Jul 30, 2026

• optimization='auto' cannot select metal or cuda_kernel unless torch is installed; availability probes never check for a device

  ppsp-team/HyPyP#304 · Filter by author Ramdam17 opened on Jul 30, 2026

• Metal backend returns float32 while every other backend returns float64

  ppsp-team/HyPyP#303 · Filter by author Ramdam17 opened on Jul 30, 2026

• ImCoh is documented signed but returns |Im|; ccorr and accorr use opposite sign conventions

  ppsp-team/HyPyP#302 · Filter by author Ramdam17 opened on Jul 30, 2026

• Coherence is documented as magnitude-squared in 4 places but implemented unsquared

  ppsp-team/HyPyP#301 · Filter by author Ramdam17 opened on Jul 30, 2026

• Six Metal tests are tautological (numpy vs numpy) and one docstring describes a kernel that never existed

  ppsp-team/HyPyP#300 · Filter by author Ramdam17 opened on Jul 30, 2026

• GPU kernels index in 32-bit arithmetic; silent overflow is reachable at realistic hyperscanning sizes

  ppsp-team/HyPyP#298 · Filter by author Ramdam17 opened on Jul 30, 2026

• Zero denominator: numpy returns NaN while numba/Metal/CUDA return 0.0 — the backend choice changes the scientific result

  ppsp-team/HyPyP#297 · Filter by author Ramdam17 opened on Jul 30, 2026

• compute_sync docstring says optimization is "currently only available for 'accorr'", but WPLI/PLI now have Metal/CUDA/torch backends selected via optimization='auto' (analyses.py:470 is out of date vs sync/wpli.py::_compute_metal + sync/kernels/metal_phase.py::wpli_metal)

  ppsp-team/HyPyP#293 · Filter by author Ramdam17 opened on Jul 22, 2026
  1

• Cut a PyPI release including #279 (Metal ACCorr command-queue leak fix)

  ppsp-team/HyPyP#291 · Filter by author Ramdam17 opened on Jul 21, 2026

• Task constructor should use annotation descriptions for onset/offset, instead of event_id

  ppsp-team/HyPyP#256 · Filter by author patricefortin opened on Mar 13, 2026