Sovereign 8s.is mirror of standardgalactic/OpenMX-workflow
  • Python 70.8%
  • Julia 23.7%
  • Shell 5.5%
Find a file
Repository files (latest commit first)
Filename Latest commit message Latest commit date
2025-11-06 13:28:01 -05:00
config Fix typo 2025-11-06 13:28:01 -05:00
LICENSE Update copyright owner in LICENSE file 2025-10-16 14:15:40 -04:00
openmx_installation.md Add link to Intel oneAPI and HPC toolkit 2025-10-30 11:19:57 -04:00
README.md Update README.md 2025-09-12 22:41:08 -04:00
requirements.txt Create requirements.txt 2025-09-10 21:20:35 -04:00

OpenMX Workflow

A high-throughput workflow for OpenMX calculations using atomate. This workflow was also employed in our recent work: arXiv:2505.04862.

This repository is adapted from openmx-wf by @tsaie79.
The main extension is the post-processing of OpenMX outputs (.scfout and .out) to compute response functions such as:

  • Frequency-dependent permittivity tensor

Prerequisites

This workflow assumes that users:

  • Are familiar with OpenMX and have a compiled version 3.9 installed.
  • Have Julia installed.
  • Have the packages required by DeepH.
    (Additionally, you will need HopTB.jl.)
  • Have access to a running MongoDB instance for job and data management.

Installation

To set up the environment, run:

pip install -r requirements.txt

This will install the necessary dependencies for the workflow, including two custom packages:

  • pymatgen (custom fork) – provides a custom-defined basis set for passing into the ASE.Atoms object.
  • atomate (custom fork) – enables job submission and data storage.

Usage

  1. Configure your workflow using the provided configuration files.
    See openmx-wf and the atomate documentation for details.

  2. Once configured, run:

    python config/wf_poscar_direct.py
    

This will:

  • Convert your input into the required openmx.dat format.
  • Insert the workflow into the FireWorks database (MongoDB).

From your designated launch site, execute the workflow using FireWorks:

qlaunch singleshot

or

qlaunch rapidfire -m 1

After the calculations are finished, all results will be stored in the MongoDB database.
image

You can easily explore and query the generated data using MongoDB Compass or any other MongoDB client.

Citing

If you use this workflow in your research, please cite:

@misc{hsu2025accuratepredictionsequentialtensor,
      title={Accurate Prediction of Tensorial Spectra Using Equivariant Graph Neural Network}, 
      author={Ting-Wei Hsu and Zhenyao Fang and Arun Bansil and Qimin Yan},
      year={2025},
      eprint={2505.04862},
      archivePrefix={arXiv},
      primaryClass={cond-mat.mtrl-sci},
      url={https://arxiv.org/abs/2505.04862}, 
}