Python implementation of PGLDM (prioritized generalized-linear dynamical modeling). PGLDM is a multi-step analytical subspace identification algorithm for learning a generalized-linear dynamical model that explicitly models shared and private dynamics between two generalized-linear time-series.
More information about the algorithm can be found in our manuscript below.
We recommend using python3.11 to execute the code. If testing in a Unix or Unix-like operating system, we also recommend using a virtual environment.
python3 -m venv venv
source venv/bin/activate
python -m pip install -r requirements.txt
Add the following file venv/lib/python3.11/site-packages/pgldm.pth with contents
/path/to/code/directory
This will ensure that your virtual environment can find the relevant source code.
In order to model Bernoulli observations, our codebase uses the latest version of bestLDS to perform the appropriate moment conversions. You can follow a similar procedure to setup this dependency:
- Clone the bestLDS repository to your desired location.
- Add a venv/lib/python3.11/site-packages/bestLDS.pth file with the following contexts
/path/to/code/directory/bestLDS
To use your virtual environment in a jupyter notebook environment, we recommend using ipykernel. For example:
python -m ipykernel install --user --name=pgldm_demo
Run the demo notebook locally using jupyter notebook. You may need to change the kernel to point towards the correct environment "pgldm_demo".
We provide a tutorial notebook demonstrating usage of PGLDM across several different generalized-linear time-series observation pairs.
Sample data can be found in this Google Drive folder. The notebook expects the data to be stored in the base directory as /path/to/PGLDM/sample_data/
Please feel free to email or file GitHub issues if you have any questions, find any bugs, or run into problems getting setup.
Copyright (c) 2024 University of Southern California
See full notice in LICENSE.md
Lucine L. Oganesian, Omid G. Sani, and Maryam M. Shanechi
Shanechi Lab, University of Southern California