Changelog

See Keep a Changelog. This project uses Semantic Versioning.

Unreleased

0.3.0 - 2026-09-21

API and Data Model Changes

  • BREAKING: mantid_like_combination_spectra now returns a collection with one spectrum per atom instead of one per (q-point, atom). This was always the intended behaviour, but as downstream code might depend on the current behaviour we treat the bugfix as a breaking change.

    • The return shape changes from (n_qpts * n_atoms, n_bins) to (n_atoms, n_bins).

    • The “qpt” line data is no longer present in the metadata as lines cover multiple q-points.

    • After a .sum() or .group_by(<not qpt>) we end up in the same place as before.

Performance Improvements

  • Linear scaling with q-point count: mantid_like_combination_spectra now accumulates per-q-point contributions as arrays and constructs the result collection once, making the function linear in q-point count (O(n)) instead of quadratic (O(n²)). In practice this drastically reduces memory usage and execution time.

0.2.0 - 2026-08-28

This release includes improvements to tutorial and validation workflows, and adds JSON serialization for Displacements. The main breaking change is the decoupling of cross-section weighting into an explicit apply_weights step.

API and Data Model Changes

  • Decoupled cross-section weighting from intensity calculations: Intensity calculation functions now always compute unweighted spectra (dimensionless / cross-section-free intensity units).

    • Removed apply_cross_sections boolean parameters from intensity calculation routines.

    • Added explicit abinslib.util.apply_weights function. Callers are now responsible for applying neutron cross-sections to unweighted spectra when desired.

    • Spectrum metadata (symbols, masses) is used with Euphonic’s IsotopeData protocol to resolve isotope cross sections, avoiding the need to pass Crystal objects into intensity routines.

    • Corrected units handling to ensure raw spectra are cross-section-free.

New Features

  • JSON Serialization: Added abinslib.io module with JsonMixin: this provides to_json, from_json, to_json_file, and from_json_file methods to classes that implement to_dict and from_dict.

    • Includes and validates API version and class metadata on deserialization.

    • This functionality is added to the Displacements class.

  • Sample Data Management: Added abinslib.data module using pooch to fetch sample and validation data on demand from machine-readable registries in abinslib.registries. This keeps the source repository light while making realistic data conveniently available for tutorials and validation.

Documentation & Tutorials

  • Converted Mantid validation benchmarks into executable Sphinx Gallery examples (plot_validation.py) rendered directly in the online documentation.

  • Added tutorials illustrating TOSCA INS simulation workflows.

  • Configured Sphinx documentation with furo theme, sphinx-autoapi, myst-parser, and custom styling.

Maintenance & Workflows

  • Overhauled dev/validation workflows using pixi and Snakemake for hashing and generating reference validation datasets.

  • Updated repository URLs to https://github.com/isisneutronmuon/abINS_lib and documentation to https://isisneutronmuon.github.io/abINS_lib.

  • Configured Ruff linting and docstring style enforcement across the repository.

0.1.0 - 2026-07-16

  • Initial release of abinslib.

  • Support for 1-D and 2-D almost-isotropic and isotropic incoherent inelastic neutron-scattering intensities from harmonic phonon modes.

  • Integration with Euphonic 2.0 and ResINS 0.1.