Deep_rev_msm Deep learning Markov and Koopman models with physical constraints.

What is it?

Deep_rev_msm is summarized the implementations of the work presented in the paper "Deep learning Markov and Koopman models with physical constraints." (described in https://arxiv.org/abs/1912.07392). It includes a Jupyter notebook able to reproduce the results presented in the paper and a benchmark file.

Citation

If you use Deep reversible models presented in this paper in scientific work, please cite:

Mardt, A., Pasquali, L., Noé, F. & Wu, H. (2019). 
Deep learning Markov and Koopman models with physical constraints. 
arXiv, 1912.07392.

Installation

We are using the package vampnet from the repo https://github.com/markovmodel/deeptime/tree/master/vampnet IMPORTANT: We are using tensorflow 1.14

This package requires Tensorflow to be used. Please install either tensorflow or tensorflow-gpu. Installation instructions:

https://www.tensorflow.org/install/

To use the notebook yourself, first clone the repository:

git clone https://github.com/markovmodel/deep_rev_msm.git

Then you can directly start.