DeepH-pack’s documentation

DeepH-pack is a package for the application of deep neural networks to the prediction of density functional theory (DFT) Hamiltonian matrices based on local coordinates and basis transformation 1. DeepH-pack supports DFT results made by ABACUS, OpenMX, FHI-aims or SIESTA, and will support HONPAS soon.

Getting Started

• Installation
  • Requirements
  • Install DeepH-pack
  • Install one of the supported DFT packages
• Prepare the dataset
  • Using ABACUS
  • Using OpenMX
  • Using FHI-aims
  • Using SIESTA
• Preprocess the dataset
• Train your model
• Inference with your model

Demo

• Demo: DeepH study on twisted bilayer bismuthene
• Demo: Reproduce the experimental results of the paper
• Demo: Train the DeepH model using the ABACUS interface

Input Keywords

• Input Keywords
  • Preprocess
  • Train
  • Inference

References

1

H. Li, Z. Wang, N. Zou, M. Ye, R. Xu, X. Gong, W. Duan, Y. Xu. Deep-learning density functional theory Hamiltonian for efficient ab initio electronic-structure calculation. Nat. Comput. Sci. 2, 367–377 (2022).