High-throughput Workflows

Running large quantity of calculations in an automatized fashion brings along new challenges such as data management, error handling and back-tracing but also analyzing trends and features of the results. This library offers some tools that are designed to make high-throughput calculations more user-friendly and easy to run and relies on the AiiDA python package to handle the computational workflows and manage the data storage and provenance. It is therefore strongly recommended to get familiar with the general working principles of the pacakge via its documentation page.

AiiDA processes

Basis of any high-throughput study are predefined workflows that are applied on all crystalline or molecular structures of an initial data pool. Once the library is installed all relevant work chains for high-throughput workflows should be visible when typing:

$ verdi plugin list aiida.workflows

  • All implemented AiiDA CalcJob and WorkChain classes are listed as part of the API.

  • A detailed overview of the CalcJob and WorkChain classes using the CP2K code is given here.

Workflow builder classes

The two workflow builder classes, namely the WorkflowBuilder and MultipleWorkflowBuilder class, have been implemented to facilitate the design and management of high-throughput workflows, more details are given here.