Multivariate Autoregressive (MVAR) Model with PyTorch Lightning

This repository provides an implementation of a multivariate autoregressive (MVAR) model using PyTorch Lightning for learning temporal dependencies from synthetic time-series data. It includes data generation, training, evaluation, and visualization.

Installation

Install the required packages:

pip install torch pytorch-lightning matplotlib numpy

Usage

Run the main training and evaluation script:

python main.py

Repository Structure

main.py – main training and evaluation script

mvar.py – MVAR model implementation

temporal_data.py – data module and dataset class

utils.py – synthetic data generator and plotting utilities