Event identification for KM3NeT/ARCA

KM3NeT is a distributed neutrino research infrastructure in the Mediterranean sea. KM3NeT/ARCA is the high energy part of it, which is dedicated for the search of extraterrestrial neutrino sources in the TeV-PeV range. One major goal is to study the source of the recently discovered neutrino flux by IceCube. Furthermore KM3NeT/ARCA is optimised to study galactical neutrino point sources. More insight into neutrino sources can be achieved by distinguishing between neutrino flavours. Neutrinos can have different interactions inside or near the detector which lead to various event topologies. These topologies can be differentiated with machine learning algorithms. Here neural nets were trained on selected event features, for example based on time-residual distributions. Five target topologies are differentiated: double bang events, cascades and three different track-like topologies. Based on the topologies, the neutrino flavour composition can be inferred on a statistical basis. In addition, computationally complex, tailored reconstruction algorithms can be employed based on the event topology.