How Can the Number of Peaks be Estimated from Experimental Data? : Spectral Deconvolution and Bayesian Statistics(Research)

Abstract

Spectral deconvolution, in which a multimodal spectrum is decomposed into a sum of unimodal peak function such as Gaussian or Lorentzian function, is widely used for spectroanalysis such as X-ray photoelectron spectroscopy. For the spectral deconvolution, it is very important to determine the number of peaks for fitting from the observed data only. In this article, we introduce the Bayesian approach to the spectral deconvolution.