Code for reproducing the experiment in the paper "A Memory Efficient Unified Algorithm for Online Learning of Linear Dynamical Systems".

Setup

conda create -n unified-lds python=3.10 -y
conda activate unified-lds
pip install -r requirements.txt

Reproducing the Experiment

Reproduces the results in Table 1: trains four online predictors (FIR, AR, SF, and the unified UF) with the same parameter budget and prints their normalized mean-squared error.

python main.py