Neural network vs Gaussian process fitting for representing potential energy surfaces

Manzhos, Sergei

Permalink

https://hdl.handle.net/2142/100729 Copy

Description

Title

Neural network vs Gaussian process fitting for representing potential energy surfaces

Author(s)

Manzhos, Sergei

Contributor(s)

• Hernandez , Rodrigo
• Krems , Roman
• Carrington, Tucker

Issue Date

2018-06-20

Keyword(s)

Mini-symposium: New Ways of Understanding Molecular Spectra

Abstract

Many methods have been proposed for fitting potential energy surfaces. Unfortunately, there are few comparative studies. In this paper, we compare neural networks (NN) with Gaussian process (GP) regression. We re-fit an accurate PES of formaldehyde and compare PES errors on the entire point set used to solve the vibrational Schrodinger equation, i.e. ¨ the only error that matters in quantum dynamics calculations. We also compare the vibrational spectra computed on the underlying reference PES and the NN and GP potential surfaces. The NN and GP surfaces are constructed with exactly the same points and the corresponding spectra are computed with the same points and the same basis. The GP fitting error is lower and the GP spectrum is more accurate. The best NN fits to 625/1250/2500 symmetry unique potential energy points have global PES root mean square errors (RMSE) of 6.53/2.54/0.86 cm-1, whereas the best GP surfaces have RMSE values of 3.87/1.13/0.62 cm-1, respectively. When fitting 625 symmetry unique points, the error the first 100 vibrational levels is only 0.06 cm-1 with the best GP fit, whereas the spectrum on the best NN PES has an error of 0.22 cm-1, with respect to the spectrum computed on the reference PES. This error is reduced to about 0.01 cm-1 when fitting 2,500 points with either NN or GP. We also find that the GP surface produces a relatively accurate spectrum when obtained based on as few as 313 points.

Publisher

International Symposium on Molecular Spectroscopy

Type of Resource

text

Language

eng

Permalink

http://hdl.handle.net/2142/100729

DOI

10.15278/isms.2018.WB03

Copyright and License Information

Copyright 2018 Sergei Manzhos

Owning Collections