Research data supporting "Coarse-grained versus fully atomistic machine learning for zeolitic imidazolate frameworks"

This repository supports the manuscript:

Coarse-grained versus fully atomistic machine learning for zeolitic imidazolate frameworks
Zoé Faure Beaulieu, Thomas Nicholas, John Gardner, Andrew Goodwin, and Volker Deringer

See the sister repository for details of the database used to obtain the results in the mansucript.

License

This work is licensed under a Creative Commons Attribution 4.0 International License.

Repository Overview

hZIF-data is a git submodule pointing to Thomas Nicholas' hypothetical ZIF dataset.
scripts contains the Python scripts required to run all the experiments.
notebooks contains the notebooks used to generate the plots in the paper.
results contains all the raw data needed to recreate the results from the paper.

Reproducing our results

1. Clone the repository

git clone --recurse-submodules git@github.com:ZoeFaureBeaulieu/cg-gpr.git
cd cg-mofs

2. Install dependencies

All the dependencies (and their versions) used can be found in requirements.txt. To use, first create/activate your virtual environment using conda:

conda create -n gpr python=3.8 -y
conda activate gpr

Then install dependencies using:

pip install -r requirements.txt

3. Run an experiment

At this stage, you should be ready to run any of the scripts and/or notebooks.

An easy test to check that the code is working is to run a gpr experiment with a very low number of training enviornments:

python scripts/gpr.py --struct_type cg --hypers_type cg --numb_train 10