README.md

DEPRECATED and no longer supported, please use TorchANI implementation

ASE-ANI

NOTICE: Python binaries built for python 3.6 and CUDA 9.2

Works only under Ubuntu variants of Linux with a NVIDIA GPU

This is a prototype interface for ANI-1x and ANI-1ccx neural network potentials for The Atomic Simulation Environment (ASE). Current ANI-1x and ANI-1ccx potentials provide predictions for the CHNO elements. The original ANI-1 and ANI-1x potentials are available in the "deprecated_original" original branch. For best performance the ANI-1x and ANI-1ccx ensembles in this branch should be used in any application.

REQUIREMENTS:

• Python 3.6 (we recommend Anaconda distribution)
• Modern NVIDIA GPU, compute capability 5.0 of newer.
• CUDA 9.2
• ASE
• MOPAC2012 or MOPAC2016 for some examples to compare results (Optional)

Installation

Clone this repository into desired folder and add environmental variables from bashrc_example.sh to your .bashrc.

To test the code run the python script: examples/ani_quicktest.py

Computed energies from the quick test on a working installation are (eV):
Initial Energy: -2078.502822821320
Final Energy: -2078.504266011399

For use cases please refer to examples folder with several iPython notebooks

Cool stuff

Teaser of the new ANI-2x (CHNOSFCl) potential in action!

ANI-1x running 5ns MD on a box of C₂ at high temperature.

ANI-1 dataset

https://github.com/isayev/ANI1_dataset

COMP6 benchmark

https://github.com/isayev/COMP6

TorchANI

We now have a PyTorch implementation. See: Documents and GitHub

Citation

If you use this code, please cite:

ANAKIN-ME ML Potential Method:

Justin S. Smith, Olexandr Isayev, Adrian E. Roitberg. ANI-1: An extensible neural network potential with DFT accuracy at force field computational cost. Chemical Science,(2017), DOI: 10.1039/C6SC05720A

Original ANI-1 data:

Justin S. Smith, Olexandr Isayev, Adrian E. Roitberg. ANI-1, A data set of 20 million calculated off-equilibrium conformations for organic molecules. Scientific Data, 4 (2017), Article number: 170193, DOI: 10.1038/sdata.2017.193 https://www.nature.com/articles/sdata2017193

Active learning-based (ANI-1x):

Justin S. Smith, Ben Nebgen, Nicholas Lubbers, Olexandr Isayev, Adrian E. Roitberg. Less is more: sampling chemical space with active learning. The Journal of Chemical Physics 148, 241733 (2018), (https://aip.scitation.org/doi/abs/10.1063/1.5023802)

Active learning and transfer learning-based (ANI-1ccx):

Justin S. Smith, Benjamin T. Nebgen, Roman Zubatyuk, Nicholas Lubbers, Christian Devereux, Kipton Barros, Sergei Tretiak, Olexandr Isayev, Adrian Roitberg. Outsmarting Quantum Chemistry Through Transfer Learning. ChemRxiv, 2018, DOI: [https://doi.org/10.26434/chemrxiv.6744440.v1]