Abstract
When using ab initio calculations to predict the spectra of diatomic molecules, discrete ab initio data must be converted to continuous form. As in other applications of interpolation throughout science and engineering, high-resolution (i.e., fine-mesh) data are desirable so that the method of interpolation does not affect the final results. However, high-resolution data are seldom available, because of their high cost. Current practice to is use splines, polynomials, or Morse-like fitting functions for interpolation and even for extrapolation. The choice is arbitrary and affects the spectroscopic results more than is generally recognized. Here we suggest an alternative procedure, in which the physics of the problem is leveraged to provide more robust results. A high-resolution data set, from a less costly ab initio model of the system of interest, is used as a guiding function. The residuals must still be fitted using splines or polynomials, but the smaller magnitude and weaker structure of the residuals leads to improved stability and accuracy of the spectroscopic results. When two or more guiding potentials are available, they can be used to guide the selection of geometries to be computed at high level, thus improving computational efficiency.
Supplementary materials
Title
Detailed results for F2, C2 and CO
Description
10 Figures and five Tables: Potential energy curves for F2 and for CO, fitted using polynomials of different order; Vibrational intervals of C2 from alternative analyses; Convergence of guided point-selection for F2
Actions