A CODE FOR AUTOMATED CONSTRUCTION OF POTENTIAL ENERGY SURFACES FOR VAN DER WAALS SYSTEMS

Quintas Sánchez, Ernesto

Permalink

https://hdl.handle.net/2142/97070 Copy

Description

Title

A CODE FOR AUTOMATED CONSTRUCTION OF POTENTIAL ENERGY SURFACES FOR VAN DER WAALS SYSTEMS

Author(s)

Quintas Sánchez, Ernesto

Contributor(s)

Dawes, Richard

Issue Date

6/21/2017

Keyword(s)

Theory and Computation

Abstract

begin{wrapfigure}{r}{0pt} _x000d_ includegraphics[scale=0.35]{fig2.eps} _x000d_ end{wrapfigure}_x000d_ _x000d_ The potential energy surface (PES) constitutes a cornerstone for theoretical studies of spectroscopy and dynamics. _x000d_ We fit PESs using a local interpolating moving least squares (L-IMLS) approach.footnote{M. Majumder, S. Ndengue and R. Dawes, Molecular Physics 114, 1 (2016).} _x000d_ The L-IMLS method is interpolative and has the flexibility to fit energies or energies and gradients, where inclusion of gradient information significantly reduces the number of points required for an accurate fit._x000d_ _x000d_ The method permits fully automated PES generation: _x000d_ beginning with an initial set of seed points, an automatic point selection scheme determines where new data are required and, in a series of iterations, computes new ab initio data and updates the fit until a specified accuracy is reached. _x000d_ We have interfaced this fitting approach to popular electronic structure codes such as Molpro and CFOUR to automatically generate ab initio 4D PESs for vdWs systems composed of two (rigid) linear fragments._x000d_ _x000d_ We present here our freely distributed code designed to run in parallel on a computing cluster, allowing the user to specify the system (masses, interatomic equilibrium distances, symmetry, energy range of interest, etc.) through an input file. _x000d_ For a selection of benchmark systems, we show that PESs with fitting errors below 1 wn, can be constructed using only a few hundred ab initio points.

Publisher

International Symposium on Molecular Spectroscopy

Type of Resource

text

Language

English

Permalink

http://hdl.handle.net/2142/97070

DOI

10.15278/isms.2017.WI10

Owning Collections