Changelog#

All notable changes to this project after version 1.0.1 will be documented in this file. There is changelog data prior to this present, but it is not to be considered complete.

The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/) and this project adheres to [Semantic Versioning](http://semver.org/spec/v2.0.0.html).

[v5.0.0]#

Changed

  • verdi data vasp.potcar commands are moved to aiida-vasp potcar

  • verdi data vasp.tools commands are moved to aiida-vasp tools

Added

  • aiida-vasp command line interface for launching calculations

Fixed

  • Incorrect POTCAR mapping when importing VASP POTCAR dataset with wrong symbols inside POTCAR.

    • This can be the case for ‘PBE’, ‘LDA’ release and original PBE/LDA.54/52 releases without hash. Only a few elements are affected.

    • We recommend to use aiida-vasp potcar integrity to check existing families and reimport the POTCAR archive if needed.

[v4.1.0]#

Changed

  • Workflow stack has been updated with different inputs, a new set of workflows (v2) has been added and made as defaults.

  • The parser framework has been simplified and integrated into a single module.

Added

  • BuilderUpdater system for quickly setting up calculation with pre-define default calculation/code specific parameters and remote job launching settings.

  • Default python-AiiDA type conversion for process inputs

  • Validation of workchain control parameters using pydantic

Development

  • No longer using tox for test running.

  • Switched to ruff from pylint for faster linting.

Notes for updating from v3.x.x to v4.x.x

Configuring the parsers

The script for launching calculation should be updated. In particular, note that the new parser systems takes a different format for the parser_settings.

Previously, this looks like:

parser_settings = {
        "add_energies": True,
        "add_forces": True,
        "add_stress": True,
        "add_bands": True,
        "electronic_step_energies": True,
    }
builder.settings = Dict({'parser_settings' : parser_settings})

Now, one should use:

parser_settings = {
  include_node: ['energies', 'bands'],
'electronic_step_energies': True
}
builder.settings = Dict({'parser_settings' : parser_settings})
# Getting energies
energies = node.outputs.arrays.get_array("energy_extrapolated")
forces = node.outputs.misc['forces']

The new parser system adapts a parse-if-exists principle with some quantities disabled by default, such as the eigenvalues and electronic step energies. The 'include_node' key in the parser_settings dictionary allows to selectively enable output nodes. This is because some output nodes are disabled by default, such as 'bands' and 'dos' and 'trajectory'. This is because the underlying quantities exist in almost all calculations, but the output are only useful in specific cases.

Previously forces was stored in an ArrayData node with link name forces. Now the last forces and stress are stored in misc node (Dict) which is enabled by default. To obtain forces at every structure iteration, add 'trajectory' to the 'include_node' field in parser_settings as shown above.

Controlling workchains

The VaspConvergenceWorkChain, VaspRelaxWorkChain and VaspBandsWorkChain are controlled by special ports conv_settings, relax_settings and band_settings respectively. These ports should be used to pass the desired settings to the respective workchains. Previously, each key is made a single input port, make these workchains takes a large amount of input nodes.

Deploying workchains

We now recommend using the BuilderUpdater system to quickly set up calculation with pre-define default calculation/code specific parameters and remote job launching settings. The old approach was to use a dictionary or a ProcessBuilder object, which requires the user to explicitly specify all the required inputs.

While BuilderUpdater is a convenient way to quickly set up a calculation, it is not recommended to still manually verify the inputs in case of unexpected behavior.

Rerunning workchains

The workchain nowentry points such as vasp.vasp now points to different workchains. Unfortunately, this means that the get_builder_restart method no longer works.

Calculation and workchain output

There are a few notable changes to the calculation and workchain output

  • The forces of the calculations are now included in the misc output under the key forces. This quantities is parsed and stored by default.

  • The arrays output includes array-like quantities, such as energy_extrapolated of individual ionic steps (excluded by default).

  • The key under which the quantities are stored inside the ArrayData node of the arrays, dielectrics may change. Please refer to the code file src/aiida_vasp/parsers/vasp.py for details.

[v3.0.1]#

Changed - New structure of the Read The Docs configure file to enable more recent Python versions and successful documentation builds.

[v3.0.0]#

Added - AiiDA>=2.0 support.

  • Python 3.11 support.

  • pyproject.toml support.

  • Option to ommit a static run in the relax workflow.

  • Support for generalized k-point grids.

  • Wrappers for the moved parsers.

Changed - Updated documentation.

  • Necessary changes to support AiiDA>=2.3.

  • LICENSE file to reflect plugin work in recent years.

  • Dependency of parsevasp~=3.2.

  • Contact info and author info.

Removed - Python 3.7 support.

  • pymatgen dependency.

  • Remaining parsers moved to parsevasp.

[v2.2.0]#

Added - Error handlers for the ACWF study, disabled by default to retain backwards compatibility. These

offer bare minimum, basically changing ALGO and NBANDS in case that is needed due to lack of electronic convergence.

Changed - Deviates from the develop branch and releases >2 from commit 837870c4f0080434955a491b269858caa9243005 and forward. - Removed pylint requirement in CI.

[v2.1.1]#

Changed - Bugfix to handle NaN values in parsed entries.

  • Removed fermi_level and symmetries are removed from default misc quantities to avoid stalling relaxation jobs etc.

  • Fix to the DOSCAR parser to not fail when spin-orbit couplings have been enabled.

  • Fixes the convergence workchain so that a final relaxation is possible.

  • Fixes the convergence workchain lack of the structure not being exposed as an input.

  • Fixes the convergence workchain in cases where a plane wave cutoff was supplied, the plane wave convergence tests were still executed.

  • Fixes parsing of DOSCAR when the data for the partial density of states is not available.

  • Removed version pinning on postgresql install for GitHub Action tests.

[v2.1.0]#

Added - The metadata.label and metadata.description is now passed from the VaspWorkChain to the VaspCalculation.

  • Possibilities to skip INCAR tag validation altogether.

  • Recommended plane wave cutoff and k-point grid is now set as an output on the ConvergeWorkChain.

Changed - Bugfix in the check for electronic and ionic convergence for the run status.

  • Bugfix in the stream output of the vasp_output file, which is now always printed without adding redirection to the executable.

  • Bugfix pertaining to the matching of POTCAR data and hash upon. Fixes import and linking issues.

  • Bugfix for the passing of selective dynamics.

  • Parser has been re-factored.

  • Pinned AiiDA to latest available version on PyPI for the CI.

[v2.0.0]#

Added - A new dedicated parameter namespace incar override namespace that is to be used when utilizing the existing workchain stack to supply INCAR tags directly (needs to be valid).

  • Possibility to control selective dynamics with the positions_dof flag.

  • Added possibility to parse magnetization.

  • Added framework to parse errors and warnings (notifications) from VASP.

  • Nightly test runs against AiiDA core develop.

  • Symbols are properly attached to BandsData.

Changed - Updated dependencies, in particular parsevasp to enable additional parsing of streams, selective dynamics and magnetization.

  • Renamed override parameter namespace from vasp to incar.

  • Fixed warnings related to missing context on file open etc.

  • We now instead use a positive match when copying files from the restart folder so that only the required files are copying. This removes possible failures related to the restarted job failing while the parser believes the vasprun.xml etc. is okey (but in reality is from the previous run).

  • Fixed missing yaml files.

  • We do not allow the install of AiiDA core 1.4.0 and 1.4.1 due to a bug that caused POTCAR to be present in the repository.

  • Removed py dependencies.

[v1.0.1]#

This is to be consider the first production release. Backwards compatibility is broken.

[v0.3.0]#

This is a major rewrite to be compatible with AiiDA core beta release. Backwards compatibility is broken.

[v0.2.4]#

Added - vasp.base workchain which transparently calls through to the ‘vasp.vasp’ calculation and restarts if appropriate.

  • restarting conditions are minimal yet (only submission failed will trigger a retry)

  • vasp.relax workchain which specializes in structure relaxation and optionally iterates until the structure converges.

  • ‘vasp.calc.immigrant’ added in order to support VASP import Aiida-external jobs.

Changed - PotcarData.get_potcars_from_structure() now returns an entry for each kind.name in the structure, not one per kind.symbol

  • PotcarData.get_potcars_dict() no longer fails if there is more than one PotcarData with the same full name in the family

  • Minor changes in parser.

[v0.2.3]#

Changed - fixed POSCAR writing (was writing ‘direct’ followed by cartesian coordinates)

Removed - pruned outdated parts of documentation

[v0.2.2] - 2018-03-15#

Changed - missing requirement added

  • PyPI description and keywords updated

[v0.2.1] - 2018-03-12#

Added - Allow Structures which do not have sites of same element grouped

  • Allow using different potentials for atoms of same element in a structure

  • VaspCalculation’s settings input now accepts poscar_precision key to set maximum precision for coordinates in POSCAR

  • This can be indicated in the structure by adding sites with same symbol but different name.

  • io.poscar.PoscarIo, POSCAR writer replacing pymatgen one, to be replaced by parsevasp.

Changed - Use always the same order for species in POSCAR and POTCAR - POSCAR float precision default changed to 10 decimal places

[v0.2.0] - 2018-03-07#

Added - data.potcar.PotcarData (vasp.potcar) & data.potcar.PotcarFileData, replacement for PawData

  • PotcarData is shareable, holds no licenced data

  • PotcarFileData holds the licenced POTCAR file

  • io.potcar.PotcarIo, handle conversion between File and DB representation of POTCAR information

  • io.potcar.MultiPotcarIo, handle concatenation and splitting of POTCAR files containing multiple potentials

  • verdi data vasp-potcar, CLI for importing and exporting POTCAR files

    • uploadfamily, loads a set of POTCAR files into the db

    • exportfamily, creates a compressed archive of POTCAR files in a family

    • listfamilies, list existing POTCAR families

  • io.outcar.OutcarParser, parses some information from OUTCAR file

  • calcs.VaspCalculation and descendants

    • inputs: settings:ParameterData additional settings that are not passed to VASP but to parsers etc

Changed - calcs.VaspCalculation and descendants

  • inputs: paw:PawData -> potentials:PotcarData

  • parsers.vasp.VaspParser, redesigned

    • parses OUTCAR too (or tries to)

    • takes parser_options that can be given in the settings input in a VaspCalculation

  • renamed data.paw.PawData -> data.paw.LegacyPawData

    • LegacyPawData can not be stored or changed (read-only)

Removed - io.potcar.PawParser

Deprecated - PawData is deprecated and has been renamed LegacyPawData, it can be used to read PawData database objects.

[v0.1.1]#

Baseline

On this page
Edit on GitHub
Show Source