Parametric-Fokker-Planck-Equation

These are the Python codes for implementing Algorithm 4.1 mentioned in the paper [1].

To run the algorithm, please direclty run Main_algorithm.py. It will produce plots of samples, estimated densities at chosen time nodes, as well as inner loss curves, graphs of Kantorovich dual function . Coordinates of samples, plot of KL loss curve and the hyperparameters of the algorithm are also recorded.

You may directly modify the hyperparameters of the algorithm in Main_algorithm.py or set up your own potential function based on your own interests.

References

[1] W. Li, S. Liu, H. Zha and H. Zhou, Neural Parametric Fokker-Planck Equations, arXiv preprint arXiv 2002.11309, (2020)