A SModelS interface for pyhf likelihoods

Author(s)
Gaël Alguero, Sabine Kraml, Wolfgang Waltenberger
Abstract

SModelS is an automatized tool enabling the fast interpretation of simplified model results from the LHC within any model of new physics respecting a Z2 symmetry. We here present a new version of SModelS, which can use the full likelihoods now provided by ATLAS in the form of pyhf JSON files. This much improves the statistical evaluation and therefore also the limit setting on new physics scenarios.Program summaryProgram Title: SModelSCPC Library link to program files: doi.org/10.17632/w4nft4459w.3Code Ocean capsule: codeocean.com/capsule/3458923Licensing provisions: GPLv3Programming language: Python3Journal reference of previous version: Comput. Phys. Commun. 251 (2020) 106848Does the new version supersede the previous version?: YesReasons for the new version: Addition of new features.Summary of revisions: We have included a new module, which uses the full likelihood models for the background correlations provided by ATLAS in the form of pyhf JSON files. To this end, SModelS has been interfaced with the pyhf package. Moreover, the SModelS database has been extended by efficiency map results and the corresponding full likelihoods from three new ATLAS searches for supersymmetric particles.Nature of problem: The results for searches for new physics beyond the Standard Model (BSM) at the Large Hadron Collider are often communicated by the experimental collaborations in terms of constraints on so-called simplified models spectra (SMS). Understanding how SMS constraints impact a realistic new physics model, where possibly a multitude of production channels and decay modes are relevant, is a non-trivial task.Solution method: We exploit the notion of simplified models to constrain full models by "decomposing" them into their SMS components. A database of SMS results obtained from the official results of the ATLAS and CMS collaborations, but in part also from "recasting" the experimental analyses, can be matched against the decomposed model, resulting in a statement to what extent the model at hand is in agreement or contradiction with the experimental results. Further useful information on, e.g., the coverage of the model's signatures is also provided.Additional comments including restrictions and unusual features: At present, only models with a Z2like symmetry can be tested. Each SMS is defined purely by the vertex structure and the final-state particles; initial and intermediate BSM particles are described only by their masses, production cross sections, branching ratios and total widths. Possible differences in signal selection efficiencies arising, e.g., from different production mechanisms or from the spin of the BSM particles, are ignored in this approach. Since only part of the full model can be constrained by SMS results, SModelS will always remain more conservative (though orders of magnitude faster) than "full recasting" approaches. (c) 2021 Elsevier B.V. All rights reserved.

Organisation(s)
Particle Physics
External organisation(s)
University of Grenoble, Österreichische Akademie der Wissenschaften (ÖAW)
Journal
Computer Physics Communications
Volume
264
No. of pages
5
ISSN
0010-4655
DOI
https://doi.org/10.1016/j.cpc.2021.107909
Publication date
07-2021
Peer reviewed
Yes
Austrian Fields of Science 2012
103012 High energy physics
Keywords
Portal url
https://ucris.univie.ac.at/portal/en/publications/a-smodels-interface-for-pyhf-likelihoods(9cffbb88-4133-4450-9ae6-b3b85ec4dff6).html