Alec Owens

TL;DR: The ExoMol database as mentioned in this paper provides extensive line lists of molecular transitions which are valid over extended temperature ranges, including lifetimes of individual states, temperature-dependent cooling functions, Lande g-factors, partition functions, cross sections, k-coefficients and transition dipoles with phase relations.

TL;DR: The ExoMol database as mentioned in this paper provides molecular data for spectroscopic studies of hot atmospheres, including 80 molecules and 190 isotopologues with over 700 billion transitions.

TL;DR: In this article, a self-correcting, kernel ridge regression (KRR) based machine learning (ML) was used to generate high-level ab initio potential energy surfaces (PESs) using structure-based sampling.

TL;DR: In this article, a hierarchical machine learning (hML) model is used to estimate energy and energy corrections with a hierarchy of quantum chemical methods, and the optimal training set size and composition of each constituent machine learning model is determined to minimize the computational effort necessary to achieve the required accuracy.

TL;DR: It is believed that it would be extremely challenging to go beyond the accuracy currently achieved for CH3Cl without empirical refinement of the respective PESs, and the combined effect of the HL corrections and CBS extrapolation on the vibrational wavenumbers indicates that both are needed to compute accurate theoretical results for methyl chloride.