From concept to compute: accelerating research with workflow management in `pyiron`

pyiron_workflow is a python framework for developing research workflows based on composing individual function nodes together into computational graphs. Each node class can be defined by simply applying a decorator to a regular python function, allowing user-developers to extend pyiron functionality even without deep knowledge of Object-Oriented Programming. Nodes can be grouped together into macro nodes using the same simple function-and-decorator approach, such that complex workflows can be built up and simply represented by composing and nesting these graphs. In contrast to user workflows being defined via a series of Jupyter notebook cells, this approach rigorously defines workflows by their graph topology and allows them to be easily shared -- and incorporated into new contexts -- by sharing/importing the workflow as a macro node. Interoperability can be controlled with (optional) type checking on data connections between nodes.