Introduction

Official implementation of the Journal of Chemical Physics paper: "Applications of neural networks to dynamics simulation of Landau-Zener transitions". In this project, we predicted LZ data in 2D/3D via LSTM and Narnet.

If there are bugs, pls open an issue.

Dependencies

For MATLAB, you should install the "Neural Net Time Series" toolbox.

For Python, pls find the dependencies below.

MATLAB-Narnet:

Steps to run the code:

Take pabc06 as an example.

Copy the file(network_test.mat) which is the trained neural network into the folder "code" to replace the original file(also named network_test.mat) in the folder.
Double click the file(predict.m) in the folder "code" to open the file using MATLAB.
Change the parameters in the file(predict.m). The paremeters can be found in the file "/Networks&Parameters in predict.m/pabc06/06.txt". There are 6 parameters need to be changed in the first module of the code(which have been commented with "change parameter").
Run the code and you will see the figure including predicted lines.
Train_combined.m file is used to train the neural network.

Python3-LSTM:

Requirement: Python3, tensorflow, pandas, matplotlib, numpy, scipy
The LSTM was trained in iMac. But it will also work well in Linux and Windows.

See our 3D prediction framework below.

Results

2D Prediction

3D Prediction

Acknowledgment

This work was done in The Zhao Research Group, Nanyang Technological University, you may find some interesting projects from the lab.

We thank Zhongkai Huang, Yubing Liao and Frank Grossmann for helpful discussion and graphics assistance. Competitive Research Programme (CRP) under Project No. NRFCRP5-2009-04 and from the Singapore Ministry of Education Academic Research Fund Tier 1 (Grant Nos. RG106/15, RG102/17, and RG190/18) is gratefully acknowledged.

Reference

Please cite this in your publication if our work helps your research.

@article{yang2020applications,
  title={Applications of neural networks to dynamics simulation of Landau-Zener transitions},
  author={Yang, Bianjiang and He, Baizhe and Wan, Jiajun and Kubal, Sharvaj and Zhao, Yang},
  journal={Chemical Physics},
  volume={528},
  pages={110509},
  year={2020},
  publisher={Elsevier}
}

Pls contact me if you have any questions.

HomePage: https://jian-danai.github.io/

Email: yangbj@zju.edu.cn