On the challenge of estimating diphoton backgrounds at large invariant mass

Abstract

Diphoton searches at high invariant mass are an integral part of the experimental high energy frontier. Using the analyses of the 750\,GeV diphoton resonance as a case study, we examine the methodology currently employed by the experimental analyses in estimating the dominant standard model backgrounds. We assess the dependence of the significance associated with the excess on the background modeling. In particular, we show that close to the high energy tails of the distributions, estimates of the jet faking backgrounds relying on functional extrapolations or Monte Carlo estimates of the challenging photon-jet contributions introduce a large uncertainty. Analyses with loose photon cuts, low photon $p_T$ cuts and those susceptible to high photon rapidity regions are especially affected. Given that diphoton-based searches beyond 1\,TeV are highly motivated as discovery modes, these considerations are relevant and applicable in the future. We first consider a physics-driven deformation of the shape of the diphoton faking photon-jet spectrum by next-to-leading order effects combined with a rapidity and transverse momentum dependent fake rate. We show that the resulting local significance of the excess is reduced due to such a deformation. Then using a simple, but more general, ansatz to modify the contribution of the jet faking backgrounds at high invariant masses but keeping the inclusive and differential sample purities within experimental uncertainty estimates, we demonstrate that the originally reported local 750\,GeV excess significances could have been overestimated by more than one standard deviation. We furthermore cross-check our analysis by comparing fit results based on the smaller 2015 and the larger 2016 LHC datasets. Finally we employ our methodology on the complete 13\,TeV LHC dataset to asses the systematics involved in the current diphoton searches beyond the TeV region.