Volume 3, Issue 2 pp. 169-184
Article
Full Access

On the least squares procedure for atomic calculations

M. H. Lloyd

M. H. Lloyd

School of Molecular Science, University of Sussex

Search for more papers by this author
L. M. Delves

L. M. Delves

School of Mathematical and Physical Sciences, University of Sussex, Falmer, Brighton, Sussex

Search for more papers by this author
First published: March 1969
Citations: 21

Abstract

We discuss the properties of one version of the least squares (LS) method for the solution of the Schrödinger equation. These properties are exemplified by a number of calculations on the n1S and n3S states of helium, up to principal quantum number three, which are very much more accurate than previous LS calculations on helium. Particular attention is paid to the convergence properties of the LS procedure and we compare it with the simpler Rayleigh–Ritz (RR) procedure in the case when the RR matrix elements are evaluated numerically over the same quadrature mesh as used in the LS procedure. We conclude that although the LS procedure is capable of high accuracy it has no advantages which would justify its sole use in place of the RR procedure. However, it does have some advantages when used in conjunction with RR, in that it gives an estimate of the numerical accuracy of the RR energies.

The full text of this article hosted at iucr.org is unavailable due to technical difficulties.