https://github.com/pswpswpsw/stable-deep-probabilisitic-koopman

https://github.com/pswpswpsw/stable-deep-probabilisitic-koopman

Last synced: 6 days ago
JSON representation

• Host: GitHub
• URL: https://github.com/pswpswpsw/stable-deep-probabilisitic-koopman
• Owner: pswpswpsw
• Created: 2020-07-11T02:53:09.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2022-11-21T22:28:47.000Z (almost 2 years ago)
• Last Synced: 2023-03-08T22:44:58.620Z (over 1 year ago)
• Language: HTML
• Size: 7.71 MB
• Stars: 4
• Watchers: 1
• Forks: 1
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# Stable Bayesian Deep Learning of Koopman Operator

We implement a Bayesian Neural Network on Koopman operator with

- Tensorflow 1.8

- Edward 1.3.5

in Python 2 environment.

# Warning:

At the time when I was writing the code, there is a shift from Python 2 to 3 and 

shifting of Edward to Tensorflow.Probability. Thus, this code for now is only for 

informative purpose. I will come back for an upgrade in the future into Python 3 

environment and make everything in PyTorch.

It will be helpful if you want to know

- how is everything exactly coded to enforce stability constraint of Koopman operator?

- how is SVD-augmented autoencoder built?

- how is Koopman operator is built recurrently? how is the loss function constructed?

## Citation

    @article{pan2020physics,

    title={Physics-informed probabilistic learning of linear embeddings of nonlinear dynamics with guaranteed stability},

    author={Pan, Shaowu and Duraisamy, Karthik},

    journal={SIAM Journal on Applied Dynamical Systems},

    volume={19},

    number={1},

    pages={480--509},

    year={2020},

    publisher={SIAM}

    }