Bayesian parameter estimation for effective field theories. (27th May 2016)

• Record Type: Journal Article
• Authors: Wesolowski, S; Klco, N; Furnstahl, R J; Phillips, D R; Thapaliya, A
• Is Part Of: Journal of physics. Volume 43:Number 7(2016:Jul.)
• Publication Date: 2016-05-27
• Subjects: Bayesian -- effective field theory -- uncertainty quantification; 02.50.-r -- 11.10.Ef -- 21.45.-v -- 21.60.-n; Nuclear physics -- Periodicals; Particles (Nuclear physics) -- Periodicals; Physique nucléaire -- Périodiques; Particules (Physique nucléaire) -- Périodiques; Kernfysica; Elementaire deeltjes; 539.7
• DOI: 10.1088/0954-3899/43/7/074001
• Languages: English
• ISSNs: 0954-3899

Abstract

Abstract: We present procedures based on Bayesian statistics for estimating, from data, the parameters of effective field theories (EFTs). The extraction of low-energy constants (LECs) is guided by theoretical expectations in a quantifiable way through the specification of Bayesian priors. A prior for natural-sized LECs reduces the possibility of overfitting, and leads to a consistent accounting of different sources of uncertainty. A set of diagnostic tools is developed that analyzes the fit and ensures that the priors do not bias the EFT parameter estimation. The procedures are illustrated using representative model problems, including the extraction of LECs for the nucleon-mass expansion in SU(2) chiral perturbation theory from synthetic lattice data.