Skip to content

openWFN 0.11.0

Release date: October 4, 2026.

These historical notes describe 0.11.0. It added native Hirshfeld populations, vibrational spectroscopy and excited-state/UV–Vis analysis to the package. A stable software release does not make every method scientifically validated; each keeps its own evidence and limits.

Highlights

Native Hirshfeld populations and charges

  • Adds ordinary neutral-pro-atom Hirshfeld populations and charges through the shared analysis engine.
  • Uses packaged, reproducible H/C/N/O pro-atom reference data with provenance and integrity checks.
  • Reports electron/population/charge closure diagnostics and does not renormalize final charges to hide integration residuals.
  • Rejects unsupported elements, ECP/pseudopotential ambiguity, and ghost-center ambiguity instead of guessing a fallback.
  • Independently compared against HORTON-PART for the named ten-case H/C/N/O all-electron validation set.
  • Scientific status: Validated only for the documented H/C/N/O all-electron scope and fixed tolerances; unsupported chemistry remains explicit.

Vibrational spectroscopy

  • Adds typed Gaussian harmonic vibrational modes, signed imaginary frequencies, source IR intensities, source Raman activities, and source normal-mode vectors.
  • Adds deterministic Gaussian broadening, IR/Raman CSV and figure export, CLI/Python/MCP parity, report integration, and a dedicated offline Workbench workspace.
  • Raman activities are preserved as source activities and are not silently converted to experimental Raman intensities.
  • Scientific status: Experimental. Parser/regression/interface evidence is not yet an independent external spectroscopy validation campaign.

Excited states and UV–Vis

  • Adds a universal typed excited-state/job-block model with deterministic multi-job isolation.
  • Adds source adapters for Gaussian, ORCA, and Q-Chem.
  • Preserves source method detail while classifying broad method families and amplitude conventions conservatively.
  • Adds excited-state tables, one-state inspection, transition dipoles, source oscillator-strength sticks, and deterministic UV–Vis broadening.
  • Uses a documented default Gaussian FWHM of 0.20 eV in energy space and applies the proper energy-to-wavelength Jacobian for wavelength-domain profiles.
  • Preserves dark states (f = 0), distinguishes missing oscillator strengths from zero, and excludes invalid/negative optical values from simulated curves with explicit reasons rather than clamping or inventing data.
  • Adds JSON/CSV/PNG/SVG export, HTML report integration, and an Excited States workspace in the offline Workbench.
  • Scientific status: Experimental. Recognition of a source state is not a claim that every underlying TDDFT/EOM/ADC/multireference implementation is independently validated.

Interfaces

The new capabilities use the same scientific engine across public surfaces:

  • command line;
  • typed Python API and named analysis registry;
  • local read-only MCP analysis routing;
  • CSV/JSON/PNG/SVG exports;
  • HTML research reports;
  • offline Workbench workspaces.

Representative commands:

openwfn calculation.fchk population hirshfeld
openwfn frequency.log vibrations
openwfn frequency.log spectra ir --fwhm 20 --points 2001 --export ir.csv
openwfn excited.log excited states
openwfn excited.log excited state 1
openwfn excited.log excited dipoles
openwfn excited.log spectra uvvis --fwhm-ev 0.20 --export uvvis.svg

Reproducibility and release verification

The stable publication workflow remains fail-closed and version-driven. Before the 0.11.0 tag or PyPI upload is created, the exact release commit must pass the repository test, documentation, security, and offline Chromium workflows. The workflow then:

  1. checks synchronized release metadata;
  2. builds and validates wheel and source distributions;
  3. installs the built wheel in a fresh environment with interoperability/resource extras;
  4. installs the packaged eleven-molecule corpus;
  5. runs the 110-command bounded resource/workflow matrix against the installed package;
  6. creates the annotated v0.11.0 tag;
  7. publishes to PyPI through OIDC trusted publishing;
  8. creates the GitHub stable release for the same exact commit;
  9. installs openwfn==0.11.0 back from public PyPI in a second clean environment and reruns the same 110-command matrix.

Command/resource completion and stable packaging are evaluated separately from scientific validation status.

Compatibility and support

  • Python 3.10–3.13.
  • Existing 0.10 public analysis/result schemas remain supported unless a capability explicitly exposes new additive fields.
  • Optional IOData interoperability remains pinned to the tested 1.0.1 contract.
  • Optional cclib source-output extraction remains Experimental and is not a substitute for the typed excited-state adapters.
  • Gaussian formatted-checkpoint files remain the preferred native full-wavefunction input path.

Install

python -m pip install --upgrade openwfn

For an exact research environment:

python -m pip install "openwfn[interop,resources]==0.11.0"

Review the validation matrix and method-specific limitations before using any quantity for research reporting. Stable software packaging does not erase Experimental or Unsupported scientific boundaries.