Bayesian priors and nuisance parameters
High Energy Physics - Lattice
2016-11-29 v1
Abstract
Bayesian techniques are widely used to obtain spectral functions from correlators. We suggest a technique to rid the results of nuisance parameters, ie, parameters which are needed for the regularization but cannot be determined from data. We give examples where the method works, including a pion mass extraction with two flavours of staggered quarks at a lattice spacing of about 0.07 fm. We also give an example where the method does not work.
Cite
@article{arxiv.1611.08729,
title = {Bayesian priors and nuisance parameters},
author = {Sourendu Gupta and Anirban Lahiri},
journal= {arXiv preprint arXiv:1611.08729},
year = {2016}
}