Skip to content

Test hardening: edge cases + numerical-stability regressions #61

Closed

Description

@neuromechanist

Fill the remaining test gaps before v0.1.0. Complements #15 (save/load + plot_components).

Tasks

  • Single-channel and single-sample edge-case fits (minimal real slices, not synthetic) that assert graceful behavior (fit or a clear error, no silent NaN).
  • Numerical-stability regression tests exercising the Fortran bounds (mincond, minlog, maxdble, mineig): inputs near-singular / extreme-scale should hit the clamps deterministically.
  • Wire these into CI (they are fast; keep the Fortran-binary parity tests behind the slow marker).

Notes

NO-MOCK: correctness tests use real sample EEG + the Fortran binary. Edge-shape tests may use minimal real slices. See #15 for the save/load + plot_components half of this work.

Metadata

Metadata

Assignees

No one assigned

    Labels

    featureNew feature or enhancement

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions