Alessando showed the fast tracking approach to the MUON detector.

The resolutions in MC and test data are very different. One could use
the Andrea Dainese parametrization with corrected TPC track parameters
according to the test beam. This way the contribution from the track
smearing is taken into account better. 

Boris. Is the noise simulation OK? It could be the reason for the difference.

Andreas. The reconstruction of the test beam data is completely
different from the one in AliRoot. They have to be made coherent for
the comparison. 

Guy. The mass resolution to find the upsilon is about 100MeV. The
realistic point gives 115MeV, which is not so bad, but is higher then
the needed one.