Repository files navigation

Python implementation of the Karhunen-Loève expansion, with parallelismover the evaluation of the eigenfunctions, for approximating stochasticprocesses via a set of eigenfunctions. This is useful for separating thespace-time components of stochastic processes from their stochasticcomponents. The implementation is based on the following paper:

@inproceedings{phillips2022spectral,title={Spectral Diffusion Processes},author={Angus Phillips and Thomas Seror and Michael John Hutchinson    and Valentin De Bortoli and Arnaud Doucet and Emile Mathieu},booktitle={NeurIPS 2022 Workshop on Score-Based Methods},year={2022},url={https://openreview.net/forum?id=bOmLb2i0W_h}}

Run the commands below to install the required packages.

git clone https://github.com/alisiahkoohi/kl-expansioncd kl-expansion/conda env create -f environment.ymlconda activate klexppip install -e.

After the above steps, you can run the example scripts by justactivating the environment, i.e.,conda activate klexp, thefollowing times.

To run the example scripts, you can use the following commands.

python scripts/kl-expansion-toy-example.py --x_range [-10,10] --M 20 --num_workers 8

Please contactalisk@rice.edu for questions.

Ali Siahkoohi and Lorenzo Baldassari