Speaker
Lin Yang
Description
Accurate photon identification efficiency measurements are essential for physics analyses with photons at the ATLAS experiment. This poster presents a normalising-flow-based electron-extrapolation method for unconverted photons, with an application to 2022 collision data. A conditional normalising flow trained on simulated electrons and photons transforms the joint distribution of calorimeter shower-shape observables, accounting for their correlations. The learned transformation is then applied to electrons selected from $(Z\to e^+e^-)$ decays to estimate the photon identification efficiency.
