Multivariate Methods for Muon Identification at LHCb

The best possible identification of a muon by LHCb will be obtained by combining the available information from all the relevant subdetectors. We present a comparison among three multivariate methods, applying them to the muon identification. A neural network method and two parametric statistical approaches (one Bayesian and one classical) were studied in the context of separating muons from other particles using a simulation of eventswith the maximum background hit rate in the muon chambers. For a muon efficiency of 90% the pion misidentification is ~1%. The Bayesian and the neural network methods gave the best performance.