Predicting spectroscopic constants and dipole moments of diatomics: a machine learning approach

We present a machine learning approach to predict the spectroscopic constants and dipole moments of diatomic molecules on the basis of atomic and spectroscopic data. 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, dissociation energy, and dipole moments. 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~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~cm$^{-1}$ for the equilibrium distance and the vibrational harmonic frequency, respectively. In contrast, the dissociation energy is predicted with an absolute error $ \lesssim 0.4$ eV. In addition, we show that we can predict the dipole moments of diatomic molecules using atomic and spectroscopic properties with a mean absolute error $ \lesssim 0.44$ D. Through our model predictions, we are capable of ranking diatomic molecules according to their dipole moments magnitude. Alongside these results, we prove that it is possible to predict the spectroscopic constants of homonuclear molecules from the atomic and molecular properties of heteronuclears. Finally, on the basis of our results, we present a new way to classify diatomic molecules beyond the standard chemical bond properties.