Measurement of performance of the pixel neural network clustering algorithm of the ATLAS experiment at $\sqrt{s}$ = 13 TeV

The properties of pixel clusters in dense environments are studied with $\sqrt{s}$ = 13 TeV proton-proton collisions from the LHC, recorded by ATLAS from June to July 2015. A novel method to evaluate the performance of the artificial neural network used for identifying pixel clusters created by multiple particles is presented. Using this method, the results in data and Monte Carlo simulation are compared. The neural network, as part of the track reconstruction, shows the expected response when used on collimated tracks.