Speaker
Description
We present a new approach to fitting the $e^+P$ and $e^-P$ high-$x$ data from the ZEUS experiment\footnote{H.~Abramowicz et al. (ZEUS Collaboration), Phys. Rev. D
89, 072007 (2014); I.~Abt et al. (ZEUS Collaboration), Phys.Rev.D 101 (2020) 11, 112009 } based on a full forward modeling from the input PDFs to the expected number of events in measurement bins. Systematic uncertainties are implemented in the predictions of the expected numbers of events and Poisson statistics are used to evaluate the likelihood. The probability distributions for the parameters of the PDFs are extracted with a Markov Chain Monte Carlo approach using Bayesian reasoning implemented in the Julia-based BAT.jl package\footnote{O.~Schulz et al., SN Computer Science volume 2, Article number: 210 (2021)}. For the purpose of this analysis, a QCDNUM\footnote{M.~Botje, Comput. Phys. Commun. 182 (2011) 490, arXiv:1005.1481, Erratum arXiv: 1602.08383
(2016)} add-on package was developed to represent parton densities, structure functions or cross-sections as 2-dimensional cubic interpolation splines. Compared to standard numerical integration approaches, a speed-up of the code of more than three orders of magnitude was achieved without relevant loss of accuracy. We present the techniques developed in formulating our new approach and show first test results on simulated data.
| Submitted on behalf of a Collaboration? | No |
|---|
