Welcome to TBDNN !
TBDNN is an open-source project for material science. It intends to generate Slaster-Koster tight-binding parameters using deep learning approaches such as multilayer perceptron, conventional neural networks, recurrent neural networks from the output of density functional theory (DFT)calculatons.
This is an ongoing project and I will update new results frequently. Stay tuned.
This project was supported via the nVidia Academic Program
For details of the project -> TBDNN Proposal
For qeuestions, please contact: spi@ucdavis.edu
======
Update (May 2022): We have guided a master student to study this topic few years ago. It you want more details, please email to spi@ucdavis.edu.