Speaker
Pavel Nadolsky
(Southern Methodist University)
Description
The span of uncertainties of the PDFs obtained in a global analysis depends on criteria for selection of acceptable PDF fits. I review several such criteria for estimating the goodness of fit in a large-scale analysis such as CT18, where they are applied in addition to the commonly used value of the global log-likelihood. On the example of CT18(Z) predictions for electroweak precision observables, I show that these tests offer in-depth insights about agreement among constraints from individual experiments and other such factors that ultimately determine the realistic PDF uncertainty. I compare two particularly helpful statistical techniques for this purpose, Lagrange Multiplier scans and Hessian sensitivities.
Authors
Bo Ting Wang
C.-P. Yuan
(MSU)
Carl Schmidt
Daniel Stump
Davison Soper
(University of Oregon)
I. Sitiwaldi
Joey Huston
Jun Gao
Karol Kovarik
(WWU Münster)
Keping Xie
(PITT PACC)
Marco Guzzi
(Kennesaw State University)
Pavel Nadolsky
(Southern Methodist University)
Sayip Dulat
Tim Hobbs
(SMU)
