Tests

AiiDA-VASP utilizes pytest which makes it easy to run and construct new tests. We also utilize the fixture managers that are present, see the AiiDA documentation.

In order to run all tests execute, in the root folder:

$ pytest

This is a useful test to see if the plugin works and that everything is installed correctly.

Note

For contributors, one principle rule is in place: always try to construct tests when submitting a PR.

Workchain tests using mock-vasp

The plugin is shipped with a mock VASP script to simulate running VASP without actually running it. This allows us to test the workchains without having to run VASP.

The mock-vasp analyzes the input and compute a hash, then searches a registry folder containing pre-computed calculations results for a match. If a match is found, it copies the output files to the folder of the calculation. Otherwise, if MOCK_VASP_VASP_CMD is set, it runs VASP with the given command and upload the output to the registry.

To generate the test data for pytest when developing the plugin, set the following environment variables:

• MOCK_VASP_POTCAR_PATH: path to the directory containing the POTCAR files

• MOCK_VASP_VASP_CMD: command to run VASP

• MOCK_VASP_REG_BASE: Base of the registry for the mock VASP executable, defaults to the <root>/tests/test_data folder.

• MOCK_VASP_UPLOAD_PREFIX: A prefix to use when creating folders of containing the registered VASP calculations.

Note

mock-vasp does not analyse nor copy the POTCAR files as they are in most cases licensed. The POTCAR files are simply ignored when computing the hash of calculation inputs.

To test a new workchain and record the calculation data, set MOCK_VASP_VASP_CMD to your real vasp executable (optionally, with the MPI launcher and arguments). Set MOCK_VASP_POTCAR_PATH to the path containing the POTCAR files, this will upload the POTCAR and create a corresponding PBE.54 potcar family. The MOCK_VASP_UPLOAD_PREFIX should be set when constructing the calculation input, as the custom scheduler comments. This allows one to know which calculation data is for which test case. Finally, run the tests as usual as pytest, and rerun without setting the environmental variables above to make sure they pass without using the actual VASP executable.