Erlangen 2026 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe
T: Fachverband Teilchenphysik
T 34: Data, AI, Computing, Electronics IV
T 34.1: Vortrag
Dienstag, 17. März 2026, 16:15–16:30, KH 02.014
SysVar: A new tool for enhancing consistency in the treatment of systematic — •Agrim Aggarwal1, Georgios Alexandris1, Florian Bernlochner1, Stefanie Meinert1, Felix Metzner1, Giacomo De Pietro2, Markus Prim1, Slavomira Stefkova1, and Ilias Tsaklidis1 — 1Universität Bonn — 2Karlsruhe Institute of Technology
SysVar provides an end-to-end, consistent machinery to build template histograms and their systematic variations with correlations preserved. To account for effects such as detector acceptance and calibration, physics reweighting, event-by-event correction weights are applied to the Monte-Carlo templates which have systematic uncertainties. For a typical template fit spanning multiple channels, multiple templates and multiple observables keeping book of all correlations becomes non-trivial.
In this talk, we present SysVar - A python package that streamlines the treatment of systematic uncertainties for collider-physics analyses that rely on Monte-Carlo template fits. SysVar produces outputs compatible with popular HEP template-fitting libraries such as cabinetry and pyhf. It was originally developed within the Belle II context, but its design and interfaces are experiment-agnostic.
By having consistency across systematics and preserving correlations, this also enables the combination of a measurement from different analysis based on orthogonal selections