PiNN is a Python library built on top of TensorFlow for building atomic neural network potentials. The PiNN library also provides elemental layers and abstractions to implement various atomic neural networks.

The code is currenly maintained by Yunqi Shao at Uppsala Unversiy.

• Shao, Y.; Hellström, M.; Mitev, P. D.; Knijff, L.; Zhang, C. PiNN: A Python Library for Building Atomic Neural Networks of Molecules and Materials. arXiv:1910.03376 [cond-mat, physics:physics] 2019. link

• Python 3
• ASE, Numpy, Pyyaml
• TensorFlow >= 1.15.2 [1]

Install from source code:

git clone https://github.com/Teoroo-CMC/PiNN.git
cd PiNN && pip install -e .

Or use our docker image. If you use singularity, you can build a singularity image directly from the docker image:

singularity build pinn.sif docker://yqshao/pinn:latest (or latest-gpu)
singularity exec pinn.sif jupyter notebook # this starts a jupyter notebook server
./pinn.sif -h # this invokes the pinn_train trainner

Extra dependencies are in:

• requirements-dev.txt: dependency for testing and documentation building.
• requirements-extra.txt: extra libraries for various purposes, included in the docker image.

A set of tutorial notebooks can be found in the documentation.

• CP2K format
• RuNNer format
• ANI-1 dataset
• QM9 dataset

• PiNet
• Behler-Parrinello Neural Network

• Potential model
• Dipole model

As an open-source project, the following contributions are highly welcome:

• Reporting bugs
• Proposing new features
• Discussing the current version of the code
• Submitting fixes

We use Github to host code, to track issues and feature requests, as well as to accept pull requests.

Please follow the procedure below before you open a new issue.

• Check for duplicate issues first.
• If you are reporting a bug, include the system information (platform, Python and TensorFlow version etc.).

If you would like to add some new features via pull request, please discuss with the main developer (Yunqi Shao) first to see whether it fits the scope and aims of this project.