Overview

pyiron_core is a workflow management system for Python that leverages a graph-based system -- including a visual scripting GUI -- to develop and execute workflows.

Key features:

• Hybrid workflow creation -- Code or visual scripting? Mix-and-match text-based and GUI design to best fit where you are in the development process.
• Jupyter-native -- Create, inspect, and execute pyiron_core nodes without ever leaving your notebook.
• Boilerplate-free -- Convert your regular python functions into graph nodes with a single decorator.
• Hash-based caching -- Workflows patterns can be identified by hashes of input data and graph edges to re-load results from a cache for rapid iteration
• Inclusive node library -- Get rolling quickly with a pre-built library of nodes, and extend your capabilities with easy node creation

Getting started

The user-facing API is found directly in the pyiron_core namespace, and after installation you can get running with import pyiron_core as pc and explore the dot-completion menu pc... for key tools.

To jump in with the graphical interface, just spin up the binder demo and start exploring a selection of materials science-oriented workflows!

Binder links to our demo notebook and stored workflows.

You can download a tutorial mp4 from this repo to see a demonstration of key features.

Installation

For a development installation, or to stay fresh with the bleeding edge of pyiron_core, install the conda environment locally and build the GUI:

git clone https://github.com/pyiron/pyiron_core.git
cd pyiron_core
conda env create -n pyiron_core -f project-env.yml
conda activate pyiron_core
conda install -c conda-forge nodejs=18  # unless you already have npm available
npm install; npm run build

Don't forget to modify your start-up script to add the pyiron_core location to your PYTHONPATH.

Alternatively, you can pip-install the current version to have it accessible in your existing python environment. First, make sure you have npm installed, then:

git clone https://github.com/pyiron/pyiron_core.git
cd pyiron_core
pip install .

For full functionality you will also want postgres. With conda this can be installed with conda install -c conda-forge postgresql. pyiron_core.pyiron_workflow.api.util.LocalPostgres is a helper tool for configuring your local postgres instance, and is leveraged in the educational notebooks.

Windows is not currently supported. Osx ARM machines should be able to install the package, but may encounter hiccups using some workflow nodes.