groovemat 0.9.1

move your atoms into place!

GrooveMat

Get your atoms moving!

Overview

This software package uses the Crystal Graph Convolutional Neural Networks (CGCNN) framework and M3gnet to create an un-supervised GNN that converges the internal degrees of freedom of a crystal lattice. The package is very simple, allowing you to train the model and predict converged cif files!

Download

To download disco, use the command below

pip install groovemat

To get an overview of the commands, run the following command

groovemat --help

Results

Below is a table containing some of the results gathered from the program.

CIF ID	Mean Error (Å)	Max Error (Å)
Nd1 In1 O3	0.0087	0.0125
Np1 Be1 O3	0.0061	0.0074
Mn3 Au1 N1	0.0039	0.0076
Na1 Li1 O3	0.0077	0.0095
La1 Cr1 O3	0.0088	0.0135

Download

pip install groovemat