qrpca: A Package for Fast Principal Component Analysis with GPU Acceleration

TL;DR

qrpca is a GPU-accelerated software package for fast principal component analysis, significantly reducing computation time for large matrices in R and Python.

Contribution

It introduces a scalable, GPU-enabled PCA package in R and Python that outperforms existing methods in speed for large datasets.

Findings

Achieves 10-20x faster computation speeds on large matrices

At least twice as fast as standard methods for spectral data cubes

Open-source availability for community use

Abstract

We present qrpca, a fast and scalable QR-decomposition principal component analysis package. The software, written in both R and python languages, makes use of torch for internal matrix computations, and enables GPU acceleration, when available. qrpca provides similar functionalities to prcomp (R) and sklearn (python) packages respectively. A benchmark test shows that qrpca can achieve computational speeds 10-20 $\times$ faster for large dimensional matrices than default implementations, and is at least twice as fast for a standard decomposition of spectral data cubes. The qrpca source code is made freely available to the community.