Changelog

v0.11.0post1

Released on 2023-03-29 - GitHub - PyPI

Minor Changes:

  • Documentation and README updates.
  • Minor bug fixes.

v0.11.0

Released on 2023-03-11 - GitHub - PyPI

Major Changes:

  • Model download from Zenodo is now multi-threaded. Users should see 15x faster speeds.
  • Added an FAQ page to the documentation.
  • KS2022 can filter out all structure-independent features to better compare polymorphs.
  • Printing Calculator object shows the models' location and relevant current state information.

Minor Changes:

  • Minor bug fixes.
  • It's now possible to load a single model with loadModels(); similar to downloadModels().
  • Offline documentation available inside the package.
  • Documentation updates.

Full Changelog: v0.10.3...v0.11.0

v0.10.3

Released on 2023-02-28 - GitHub - PyPI

Changes in this version:

  • Some functionality upgrades related to handling models and files in environments with no write access to the pySIPFENN package directory, mostly dedicated towards High-Performance Computers (HPCs) users.
  • Updates to the documentation. The core of pysipfenn is entirely and extensively described. The descriptor calculators are mostly covered.
  • Final version of the workshop notebook (only minor changes)
  • Minor bugfixes

New Contributors

Full Changelog: v0.10.2...v0.10.3

v0.10.2

Released on 2023-02-09 - GitHub - PyPI

This minor version release has no effect on the code functionalities, however it significantly expands the documentation for the package, which will now be hosted on Read The Docs page under:

stable latest

It also significantly expands the code documentation and type hinting. Something that new users should find very helpful. The coverage will be completed in the near future within the planned 0.10.3 release.

Full Commit Changelog: v0.10.1...v0.10.2

v0.10.1

Released on 2023-02-03 - GitHub - PyPI

This is the first version release created and tagged on GitHub, corresponding to a second PyPI release after v0.10.0; although the SIPFENN software has been developed since 2019 by researchers at Penn State. It had 8 internal releases followed by the public release of v0.9.0 along with 2022 paper titled Extensible Structure-Informed Prediction of Formation Energy with improved accuracy and usability employing neural networks published in Computational Materials Science (https://doi.org/10.1016/j.commatsci.2022.111254).

The v0.10.0 has brought many changes, including:

  • Translation of all code into pure Python, including structure featurization code.
  • PyPI packaging
  • New featurization code (KS2022) with up to x10 speed improvement; especially useful for ordered compounds.
  • New featurization code (KS2022_dilute) working only on dilute structures (both pure elements and compounds) with up to x50 speed improvement.
  • General improvements in the handling of models and data
  • Many more in 70+ commits
    (Full Changelog from v0.9 to 0.10: https://github.com/PhasesResearchLab/pySIPFENN/commits/v0.10.1)

These changes will be described in a journal article in near future, which will appear in the README.md after its publication.