Gaussian Process Dynamic Mode Decomposition

This repository is the official implementation of 'Gaussian Process Dynamic Mode Decomposition' an extension of the Master Thesis "Probabilistic Dynamic Mode Primitives" under the supervision of Hany Abdulsamad and Svenja Stark.

Getting Started

Prerequisites

This project is compatible with Python 3.8 and conda. Further JAX is required.

Install

Install virtual environment:

You can also change the environment name if necessary.

$ conda env create -f base_env.yml

There can be some issues with installing jax.

Developers

• Kay Hansel

Open Issues

• All old DMD related methods are not tested yet if they still work.
• At the moment, only ARD-Kernels are supported.
• Implement ProMPs and other methods to get good benchmarks.
• The current yml file is not up to date. It contains also packages which are not used anymore.