Spectroscopic constants from atomic properties: a machine learning approach

TL;DR

This paper introduces a machine learning method using Gaussian process regression to predict diatomic molecules' spectroscopic constants solely from atomic properties, achieving high accuracy and enabling new classification approaches.

Contribution

It demonstrates that spectroscopic constants can be accurately predicted from atomic properties using machine learning, including for heteronuclear molecules, and proposes a novel classification method.

Findings

Achieved 0.04 Å accuracy in equilibrium distance prediction.

Predicted vibrational frequency with 36 cm⁻¹ error.

Predicted dissociation energy with less than 0.4 eV error.

Abstract

We present a machine-learning approach toward predicting spectroscopic constants based on atomic properties. After collecting spectroscopic information on diatomics and generating an extensive database, we employ Gaussian process regression to identify the most efficient characterization of molecules to predict the equilibrium distance, vibrational harmonic frequency, and dissociation energy. As a result, we show that it is possible to predict the equilibrium distance with an absolute error of 0.04 {\AA} and vibrational harmonic frequency with an absolute error of 36 $\text{cm}^{-1}$, including only atomic properties. These results can be improved by including prior information on molecular properties leading to an absolute error of 0.02 {\AA} and 28 $\text{cm}^{-1}$ for the equilibrium distance and vibrational harmonic frequency, respectively. In contrast, the dissociation energy is predicted with an absolute error $\lesssim 0.4$ eV. Alongside these results, we prove that it is possible to predict spectroscopic constants of homonuclear molecules from the atomic and molecular properties of heteronuclear. Finally, based on our results, we present a new way to classify diatomic molecules beyond chemical bond properties.