Machine learning the quantum flux-flux correlation function for catalytic surface reactions

TL;DR

This paper develops a machine learning approach using Gaussian process regressors to accurately and efficiently predict quantum flux-flux correlation functions and reaction rate constants for catalytic surface reactions, significantly reducing computational costs.

Contribution

It introduces a dataset and demonstrates that Gaussian process regressors can reliably predict reaction kinetics and flux-flux correlation functions for new reactions and conditions.

Findings

Prediction mean absolute percent errors around 0.5% for rate constants.

Prediction errors around 1.0% for flux-flux correlation functions.

Provides a speedup over traditional quantum time propagation methods.

Abstract

A dataset of fully quantum flux-flux correlation functions and reaction rate constants was constructed for organic heterogeneous catalytic surface reactions. Gaussian process regressors were successfully fitted to training data to predict previously unseen test set reaction rate constant products and Cauchy fits of the flux-flux correlation function. The optimal regressor prediction mean absolute percent errors were on the order of 0.5% for test set reaction rate constant products and 1.0% for test set flux-flux correlation functions. The Gaussian process regressors were accurate both when looking at kinetics at new temperatures and reactivity of previously unseen reactions and provide a significant speedup respect to the computationally demanding time propagation of the flux-flux correlation function.