Combining ab initio and machine learning method to improve the prediction of diatomic vibrational energies

Abstract

Through the comprehensive analysis of ab initio and experimental results of a large number of diatomic systems, the systematic deviation of ab initio method in vibrational energies prediction caused by physical/mathematical simplification is located. A joint ab initio and machine learning method based on information across molecules is proposed to deal with the problem. Starting from an ab initio model, and then systematically modifying it through machine learning, the vibrational energies prediction of many diatomic systems (SiC, HBr, NO, PC, N2, SiO, O2, ClF, etc.) have been improved, and significantly surpassed the more complex ab initio model. In addition to the improvement of accuracy, the new method also greatly reduces the computational expense, and is applicable for the systems without experimental data.

Obtaining the molar heat capacities and entropies of HCl and HBr by combining density functional theory and machine learning algorithm

“…As shown in our recent work, machine learning algorithm performed great in predicting vibrational energy of diatomic molecular systems [30,31]. And the artificial neural network (ANN) [32,33] performed best [30].…”

Trendbericht Theoretische Chemie 2023 (2/3): Berechnungen von Molekülschwingungen in der Thermodynamik

“…Abgesehen von Deskriptoren basierend auf der Kerngeometrie lässt sich zudem die Elektronenstruktur in den Trainingsdatensatz einbeziehen, zum Beispiel die Elektronendichte 7) oder die Molekülorbitale. 11) Auch können die Zielgrößen des maschinellen Lernens variiert werden, zum Beispiel können Frequenzen direkt -oder deren Abweichung vom Experiment 12) oder Verschiebung 7) -modelliert werden. Auch die Größen Energie, Kraft und Dipolmoment als Grundlage für die Spektrenberechnung von protonierten Wasserclustern wurden modelliert.…”

Spectroscopic constants from atomic properties: a machine learning approach

“…, from a data-driven approach, 26 find 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. As a result, it was possible to predict R e from the atomic properties of the constituent atoms.…”

“…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. As a result, it was possible to predict R e from the atomic properties of the constituent atoms.…”