GPU-Accelerated Approximate Kernel Method for Quantum Machine Learning

TL;DR

This paper presents QML-Lightning, a GPU-accelerated kernel method for quantum machine learning that drastically reduces training times while maintaining accuracy, enabling rapid predictions for large datasets.

Contribution

Introduction of QML-Lightning, a GPU-accelerated kernel package that significantly speeds up training for quantum machine learning models without sacrificing accuracy.

Findings

Training time reduced by several orders of magnitude.

Achieves competitive energy and force prediction accuracy.

Demonstrates effectiveness on benchmarks like QM9, MD-17, and 3BPA.

Abstract

Conventional kernel-based machine learning models for ab initio potential energy surfaces, while accurate and convenient in small data regimes, suffer immense computational cost as training set sizes increase. We introduce QML-Lightning, a PyTorch package containing GPU-accelerated approximate kernel models, which reduces the training time by several orders of magnitude, yielding trained models within seconds. QML-Lightning includes a cost-efficient GPU implementation of FCHL19, which together can yield energy and force predictions with competitive accuracy on a microsecond-per-atom timescale. Using modern GPU hardware, we report learning curves of energies and forces as well as timings as numerical evidence for select legacy benchmarks from atomisitic simulation including QM9, MD-17, and 3BPA.