Welcome to Workflow’s documentation!#

A Python toolkit for building interatomic potential creation and atomistic simulation workflows.

The main function of Workflow is to efficiently parallelise operations over a set of atomic configurations (Atomic Simulation Environment’s “Atoms” objects). Given an operation that is defined to act a single configuration (e.g. evaluate energy of a structure with CASTEP ASE calculator), Workflow may apply the operation to multiple configurations in parallel. Workflow also interfaces with ExPyRe to manage evaluation of (autoparallelized) python functions via a queueing system on a (remote) cluster.

Basic use of the workflow is introduced through an example. The overall design of the workflow-specific code structures (e.g. how configurations are handled, mechanism to parallelise operations and to submit as queued jobs) are covered in Overview. Currently implemented self-contained per-configuration operations are sketched out in Operations. Description of GAP-RSS workflow, built out of these modular operations is described in GAP-RSS. There are also examples of common tasks to get started with and a Python API.


Quick start that installs all of the mandatory dependencies:

pip install git+https://github.com/libAtoms/workflow


Please find the code, raise issues and cotribute at libAtoms/workflow.