A variational method for computing vibrational spectra of molecules with up to 18 atoms

Carrington, Tucker

Permalink

https://hdl.handle.net/2142/100803 Copy

Description

Title

A variational method for computing vibrational spectra of molecules with up to 18 atoms

Author(s)

Carrington, Tucker

Contributor(s)

Thomas, Phillip

Issue Date

2018-06-18

Keyword(s)

Mini-symposium: New Ways of Understanding Molecular Spectra

Abstract

I shall present an improvement and applications of the Hierarchical Intertwined Reduced-Rank Block Power Method (J. Chem. Phys, 146, 204110 (2017)) for solving the vibrational Schroedinger equation. The improvement decreases the memory required to compute a spectrum. Variational calculations for molecules with a dozen atoms are now possible on a desktop computer. The memory cost scales linearly with the number of atoms in the molecule. We apply the HI-RRBPM to compute vibrational spectra of uracil and naphthalene, with 12 and 18 atoms, respectively. The HI-RRBPM uses a direct product basis but: 1) it is not necessary to store a direct-product-basis matrix representation of the Hamiltonian matrix (for naphthalene the size of the matrix would be ∼ 1048); 2) it is not necessary to store vectors whose length is equal to the size of the direct-product basis. This is accomplished by using sum-of-product (SOP) basis functions stored in a canonical polyadic tensor format and generated by evaluating matrix-vector products. The number of terms in the SOP basis functions is minimized by optimising the factors. Representing vibrational wavefunctions as optimised SOPs reveals the essential entangledness and provides new understanding. The method only works if the Hamiltonian is itself a SOP.

Publisher

International Symposium on Molecular Spectroscopy

Type of Resource

text

Language

eng

Permalink

http://hdl.handle.net/2142/100803

DOI

10.15278/isms.2018.MH03

Copyright and License Information

Copyright 2018 Tucker Carrington

Owning Collections