AUTOSURF: A Freely Available Program to Construct Potential Energy Surfaces

Abstract

The potential energy surface (PES) of a molecular system constitutes a cornerstone for nearly every theoretical study of spectroscopy and dynamics. We present here AUTOSURF, our freely distributed code for the automated construction of PESs. This first release treats van der Waals systems composed of two rigid fragments. A version for reactive systems with up to five atoms is under development. The AUTOSURF suite is designed to completely automate all of the steps and procedures that go into fitting various classes of PESs and facilitates certain PES refinements aimed toward specific applications in spectroscopy and dynamics. The algorithms are based on a local interpolating moving least-squares methodology and have many advanced features such as iterative refinement and symmetry recognition. The code interfaces to popular electronic structure codes such as MOLPRO and GAUSSIAN to automatically generate ab initio PESs and is well-suited for treating highly anisotropic interactions which are challenging for traditional quadrature type expansions. The niche of these algorithms is to obtain an interpolative representation of high-level electronic energies with negligible (arbitrarily small) fitting error, requiring minimal human supervision in the entire process of selection, computation, and fitting of the ab initio data. The code is designed to run in parallel on Linux-based machines ranging from small workstations to large high-performance computing clusters.

Department(s)

Chemistry

Research Center/Lab(s)

Center for High Performance Computing Research

Comments

This research was supported by the National Science Foundation, grant number CHE-1566246.

Keywords and Phrases

Codes (symbols); Computer operating systems; Electronic structure; Iterative methods; Molecular physics; Potential energy surfaces; Quantum chemistry; Van der Waals forces, Anisotropic interaction; Automated construction; Electronic energies; High-performance computing clusters; Interpolating moving least squares; Iterative refinement; Symmetry recognition; Van der Waals system, Potential energy

International Standard Serial Number (ISSN)

1549-9596; 1549-960X

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 2019 American Chemical Society (ACS), All rights reserved.

Publication Date

01 Jan 2019

PubMed ID

30576134

Share

 
COinS