Measurement of pairs of photons in LHC Run-2 proton-proton collisions with the ATLAS detector

The first differential cross section measurement of the production of pairs of prompt isolated photons in proton--proton collisions at $\sqrt{s}=13~\mathrm{TeV}$ is performed. The full LHC Run-2 data recorded by ATLAS corresponding to an integrated luminosity of $\mathcal{L}=139~\mathrm{fb^{-1}}$ is used. The differential cross sections are measured as a function of several kinematic variables of the photon pairs in a fiducial phase space characterized by a good detector performance. The event yields are corrected for background, detector resolution effects and detector inefficiencies and are compared to state-of-the-art QCD predictions. Good agreement to the measured results is observed for fixed-order NNLO calculations and for NLO calculations merged with Monte Carlo parton showers within the uncertainties. Furthermore, two auxiliary measurements are performed to determine inputs used for both the photon-pair measurement and other future measurements with the ATLAS detector that use electrons, photons or muons. A data-driven measurement of the pileup dependence of calorimetric isolation variables is performed using $Z\rightarrow ee$ events. The results of the measurement are used to develop an improved $\eta$-dependent pileup subtraction and show that the pileup noise is overestimated in MC samples. The second measurement uses $Z\rightarrow\mu\mu\gamma$ events to determine selection efficiencies of photons. The results include values for the inefficiency induced by rejecting electron-like photon candidates, which is part of the ATLAS photon reconstruction. Previously, this inefficiency was assumed to be well modelled in simulated samples, but the results show that a correction needs to be applied. This correction increases the result of the photon-pair cross section measurement by 1.8%.