Skip to content

aiida-siesta 1.1.1
Compare
Choose a tag to compare
@bosonie bosonie released this 07 Nov 17:29
· 197 commits to master since this release
v1.1.1
acf1214
This commit was created on GitHub.com and signed with GitHub’s verified signature. The key has expired.
GPG key ID: 4AEE18F83AFDEB23
Expired
Learn about vigilant mode.

Version compatible with aiida-core>=1.3.0,<2.0.0.

Improvements

  • Improvement in the BandGapWorkChain. A new feature has been introduced:
    when no bandskpoints are set in input, the workchain automatically calculates the bands
    using a k-space path automatically selected by SeeK-path.
    In particular, if a molecular dynamics run was also requested (usually a relaxation),
    the dynamics is run first and the bands are calculated on a separate Siesta run. This
    allows to select the k-space path for bands automatically using the output structure.
    When the bandskpoints is selected in input, the behavior is the same of previous versions.
  • Added an output of the SiestaSequentialConverger listing, if any,
    the unconverged parameters. More explicit warnings are returned when a parameter
    does not converge.
  • A tutorial is now available in the documentation, covering the submission of simple siesta
    calculations, the use of protocols, the use of the iterator/converger workchains and
    a simple example on how to create custom workchains.

Bug fixes

  • A bug was reported regarding the wrong selection of the kpoints path for bands
    when a relaxation with variable cell was also requested. The energy values were correct,
    but the distance between kpoints was not corresponding to the one of the Siesta
    file ".bands". The bug is now fixed.
  • The method inputs_generator().get_inputs_dict (available for SiestaCalculations and
    all the workchains in the package) had a minor bug. In case the kpoints or bandskpoints
    were not requested, the dictionary was anyway returning items for "bandskpoints"
    and "kpoints" with value "None". This has been fixed.
    The method inputs_generator().get_filled_builder was instead correct.
  • A bug was leading to the failure of SiestaSequentialConverger in case one of
    the parameters to converge in the sequential process was not reaching convergence.
    It is now fixed and, in addition, an output listing the unconverged parameters has been
    added.

For developers

  • Set up a nightly built in GitHub actions that tests the plugin (run all its tests)
    against the develop branch of aiida-core.
  • The BandGapWorkChain has been refactored almost entirely.
  • The abstract class SequentialConverger has been modified to fix the bug
    of SiestaSequentialConverger described above.

Known bugs

  • The atom heuristic selection of basis sizes is not working. Inside a protocol, the basis size choice for a single element (added in atom heuristics) is ignored by the siesta code due to a typo.
  • The Equation of State WorkChain crashes if the structure contains sites with same symbol but different name.
Assets 2
Loading