Datasets and codes supporting "Machine-Learning Bond-Order Potential for Exploring the Configuration Space of Carbon"

This provides the datasets and codes supporting "Machine-Learning Bond-Order Potential for Exploring the Configuration Space of Carbon". 

MLBOP_trainer.tar.gz

• The PyTorch implementation of MLBOP for preprocessing and training.

training_data.tar.gz

• The training dataset used for fitting MLBOP.
• This includes the DFT-calculated properties of 52669 structures obtained with the PBE functional using VASP.

testing_data.tar.gz

• The testing dataset including the DFT-calculated properties of 22663 structures.

toydata.tar.gz

• The toy datasets of diamond, graphene, and simple cubic structures.

work_mlbop.tar.gz

• The scripts used to train MLBOP on our training dataset, C-ACE training dataset, and toy dataset.

work_ace.tar.gz

• The input scripts and log files of pacemaker used to train the Atomic Cluster Expansion on the training and toy dataset.

work_dp.tar.gz

• The input scripts and log files of DeePMD-kit used to train the DeePMD on the training and toy dataset.

work_schnet.tar.gz

• The source codes implemented in Pytorch Geometric used for training SchNet on the training and toy dataset.