pyiron_workflow package

Subpackages

Submodules

Module contents

pyiron_workflow is a python framework for constructing computational workflows in a graph-based format. The intent of such a framework is to improve the reliability and shareability of computational workflows, as well as providing supporting infrastructure for the storage and retrieval of data, and executing computations on remote resources (with a special emphasis on HPC environments common in academic research). It is a key goal that writing such workflows should be as easy as possible, and simple cases should be _almost_ as simple as writing and running plain python functions.

Key features:

  • Single point of import

  • Easy “nodeification” of regular python code

  • Macro nodes, so complex workflows can be built by composition

  • (Optional) type checking for data connections

  • (Optional) remote execution of individual nodes

  • Both acyclic (execution flow is automated) and cyclic (execution flow must be

    specified) graphs allowed

  • Easy extensibility by collecting nodes together in a python module for sharing/reusing

  • Integration with semantikon for ontological hinting of data channels to provide

    guided workflow design