Machine Learning for Observables: Reactant to Product State Distributions for Atom-Diatom Collisions

TL;DR

This paper develops machine learning models to accurately predict product state distributions in atom-diatom collisions, demonstrating high accuracy and practical applicability in complex flow scenarios.

Contribution

It introduces and compares function-, kernel-, and grid-based ML models for predicting collision outcomes, highlighting the superior accuracy and practicality of grid-based methods.

Findings

All models achieve R^2 > 0.998 in predictions.

Grid-based approach performs best overall.

Kernel- and grid-based methods are more accurate and practical than function-based methods.

Abstract

Machine learning-based models to predict product state distributions from a distribution of reactant conditions for atom-diatom collisions are presented and quantitatively tested. The models are based on function-, kernel- and grid-based representations of the reactant and product state distributions. While all three methods predict final state distributions from explicit quasi-classical trajectory simulations with R$^2$ > 0.998, the grid-based approach performs best. Although a function-based approach is found to be more than two times better in computational performance, the kernel- and grid-based approaches are preferred in terms of prediction accuracy, practicability and generality. The function-based approach also suffers from lacking a general set of model functions. Applications of the grid-based approach to nonequilibrium, multi-temperature initial state distributions are presented, a situation common to energy distributions in hypersonic flows. The role of such models in Direct Simulation Monte Carlo and computational fluid dynamics simulations is also discussed.