A new maximum-likelihood method for template fits

Abstract A common statistical problem in particle physics is to extract the number of samples which originate from a process an ensemble containing mix several contributing processes. The probability density function each usually not exactly known. Barlow and Beeston found exact likelihood for fitting binned templates obtained Monte-Carlo simulation data, propagates uncertainty into result. Solving technically challenging, however. original paper also did provide way use weighted with varying weights. Other papers have introduced alternative likelihoods address these points. In this paper, new approximate derived Barlow–Beeston one. generalized fits data. performance evaluated based on toy examples. excellent – point estimates small bias confidence intervals good coverage comparable when are weighted. evaluates faster than one bins large.