This repository contains scripts and code related to predicting the atomization energy of molecules with machine learning. In particular, we include comparisons between machine learning methods and the validation of the ability of these approaches to predict the properties of molecules with sizes larger than the those in the training set.

The paper describing this activity is in preparation for submission to MRS Communications. The preprint is available on ArXiv at https://arxiv.org/abs/1906.03233

A copy of this repository with the generated data files (which are too large to host on GitHub) will soon be able on the Materials Data Facility: link. It will also soon be possible to run all of these scripts in a pre-configured Virtual Machine via WholeTale and to excecute the models via DLHub

The scripts in this project require the utility scripts in jcesr_ml and the requirements are listed in the environment.yml file.

Install the environment with Anaconda by calling:

conda env create --file environment.yml

Ward, Logan, Ben Blaiszik, Ian Foster, Rajeev S. Assary, Badri Narayanan, and Larry Curtiss. "Machine Learning Prediction of Accurate Atomization Energies of Organic Molecules from Low-Fidelity Quantum Chemical Calculations." arXiv preprint arXiv:1906.03233 (2019).