On the relationship between spectroscopic constants of diatomic molecules: a machine learning approach

Abstract

Through a machine learning approach, we show that the equilibrium distance, harmonic vibrational frequency and the binding energy of diatomic molecules are universally related, independently of the nature of the bond of a molecule.

Spectroscopic constants from atomic properties: a machine learning approach

“…For all the nine implemented models, we permute the groups and periods in D train in the training and validation stage and in D tv in the testing stage to impose permutational invariance. 26 That is, the models should not differentiate between x = (p 1 , g 1 , p 2 , g 2 ,.) and x ′ = (p 2 , g 2 , p 1 , g 1 ,.)…”

“…For instance, Sutherland found that by taking A ′ as a function of groups and periods, better results can be obtained. 19,41 Thanks to machine learning (ML) techniques and the development of extensive spectroscopic databases, 49 it has been possible to study the relationship between spectroscopic constants from a heuristic perspective, i.e., from a data-driven approach, 26 nd optimal potentials based on spectroscopy data 50 and to improve ab initio potentials to match experimental observations. 51 In particular, Gaussian process regression (GPR) models have been used on a large dataset of 256 heteronuclear diatomic molecules.…”

“…In this work, we extend the data set by Liu et al 26,49 by adding the ground state spectroscopic constants of 32 homonuclear and 54 extra heteronuclear diatomic molecules from ref. 56-127. The dataset counts 338 entries based on experimentally determined spectroscopic constants: R e is available for 338, u e for 327, and D 0 0 is available for 250 molecules.…”

“…7,8,[16][17][18][19][20][21][22][23][24][25] In summary, from 1920 till now, the number of empirical relations published is around 70 collected by Kraka et al 10 Most of these empirical relations were tested by several authors, nding some constraints on their applicability. 10,[13][14][15]26 However, all of these relationships were based on empirical evidence rather than on a given physical or chemical principle.…”

Barrier Height Prediction by Machine Learning Correction of Semiempirical Calculations

“…Gradient boosting regression has been successfully applied to predict BHs in Diels–Alder reactions, and the reactivity of transition metal complexes . Similarly, GP has shown a great performance in complex potential energy surface fittings, − predicting spectroscopic constants of diatomic molecules , and second virial coefficients of organic and inorganic compounds . The selected SQM model was PM7, which is overall the most accurate method implemented in MOPAC2016 .…”