Skip to content

openWFN 0.8.2

Historical release notes. Version 0.8.2 was a corrective patch release. It fixed three scientific edge cases found in external files during the 0.8.1 workflow review. It kept the 0.8 interfaces stable.

Install

python -m pip install --upgrade openwfn==0.8.2

What changed

  • Real Gaussian post-HF labels such as UMP2-FC, UMP3-FC, and UCISD-FC now preserve the warning that the available SCF density is being analyzed rather than a correlated post-SCF density.
  • Restricted closed-shell files that omit an explicit spin-density matrix now derive the scientifically defined channels alpha = beta = total/2 and spin = 0 for density integration and cube export.
  • Valid one-electron or empty-spin-channel frontier analyses now return partial results with unavailable HOMO/LUMO quantities represented explicitly instead of aborting the whole analysis.

Verification

The patch was developed test-first. The regression tests initially failed on the 0.8.1 behavior and passed after the minimal production fixes.

The release candidate was then checked on Python 3.10–3.13, macOS, Windows, wheel/package smoke tests, documentation, security checks, and the repository scientific validation suite.

A separate post-fix audit passed the independent external benchmark corpus, an 88-workflow public CLI/API matrix, and focused UHF, RHF, post-HF, ghost, and ECP edge cases.

Scientific boundaries

This patch improves failure handling and provenance; it does not expand openWFN's validation claims. Validated, Experimental, and Unsupported labels remain scoped to the named fixtures, tolerances, and documented methods. Quantitative density work should still converge grid spacing and padding rather than treating the default 0.15 bohr spacing as universal.

General post-SCF correlated density support is still not claimed unless a supported correlated density is explicitly parsed and selected.