Spectral decomposition by genetic forward modelling

¹ Dept. of Physics and Astronomy, Kelvin Building, University of Glasgow, Glasgow, G12 8QQ, UK
² High Altitude Observatory, National Center for Atmospheric ResearchNCAR is sponsored by the National Science Foundation. Correspondence to: scott@astro.gla.ac.uk, P.O. Box 3000, Boulder, CO. 80307-3000, U.S.A.
³ School of Mathematical and Computational Sciences, University of St. Andrews, St. Andrews, KY16 9SS, UK

Send offprint request to: S. McIntosh

Received: 15 January 1998
Accepted: 27 March 1998

Abstract

We discuss the analysis of real and simulated line spectra using a genetic forward modelling technique. We show that this Genetic Algorithm (GA) based technique experiences none of the user bias or systematic problems that arise when faced with poorly sampled or noisy data. An important feature of this technique is the ease with which rigid a priori constraints can be applied to the data. These constraints make the GA decomposition much more accurate and stable, especially at the limit of instrumental resolution, than decomposition algorithms commonly in use.