NEURAL NETWORKS FOR IMPACT PARAMETER DETERMINATION

An accurate impact parameter determination in a heavy ion collision is crucial for almost all further analysis. The capabilities of an artif icial neural network are investigated in that respect. A novel input g eneration for the network is proposed, namely, the transverse and long itudinal momentum distributions of all outgoing (or actually detectabl e) particles. The neural network approach yields an improvement in per formance of a factor of 2 as compared to classical techniques. To achi eve this improvement simple network architectures and a 5X5 input grid in (p(t),p(z)) space are sufficient.