Mixing It Up With MT2: Unbiased Mass Measurements at Hadron Colliders

TL;DR

This paper develops and tests new methods for accurate mass measurements of particles in fully hadronic decay chains at hadron colliders, addressing combinatorial background challenges.

Contribution

It introduces novel techniques like Edge-to-Bump and event-by-event decay-chain assignment to improve MT2-based mass measurements amidst combinatorial uncertainties.

Findings

Achieved ~10% mass measurement precision for gluino and sbottom.

Validated methods through Monte Carlo and blind studies with different spectra.

Demonstrated robustness of techniques with realistic collider conditions.

Abstract

Recently, much progress has been made on techniques to measure the masses of new particles with partially-invisible decays at a hadron collider. We examine for the first time the realistic application of MT2-based measurement methods to a fully hadronic final state from a symmetric two-step decay chain with maximal combinatorial uncertainty. Several problems arise in such an analysis: the MT2 variables are powerful but fragile, with shallow edges that are easily washed out or faked by ubiquitous combinatorics background. Traditional methods of both cleaning up the distribution and determining edge position can fail badly. To perform successful mass measurements we introduce several new techniques: the Edge-to-Bump method of extracting an edge from a distribution by analyzing a distribution of fits rather than a single fit; a very simple yet high-yield method for determining decay-chain assignments event-by-event; and a systematic procedure to obtain MT2 edge measurements in the presence of heavy combinatorics background, they key element being the parallel use of at least two independent methods of reducing combinatorics background to avoid fake measurements. All of these techniques are developed in a Monte Carlo study of the decay gluino gluino -> 2 sbottom + 2 b -> 4 b + 2 chi_0^1 and verified in a second blind study with a different spectrum. In both cases, the gluino and sbottom masses are measured to a precision of ~ 10% with O(100 fb^{-1}) at the LHC14 (assuming pessimistic b-tag efficiencies).