Following my posting of the last few days, I have created the following alternative selections: track selection #6 ------------------ GTVL + nDch>0 if pt>200 MeV photon selection #7 ------------------- 0.08< E(gamma) < 4. GeV (we never put an upper limit, but it is clearly needed...) 0.05 < LAT < 0.5 S9S25>0.9 no track with ( |dph|<0.04 && -0.03<dth<0.02 ) rad These selections are committed in VubAnalysis rf-270902 In order to avoid proliferation of selections I suggest we wait for Urs's study on loopers before declaring the final track selection #6. For the selection described here, based on the last days studies, the comparisons are in http://www.slac.stanford.edu/~rfaccini/phys/vub/comp_2709/splitOff-comp.html * the distribution of mm2 is better, in particular in the enriched (the supplement of studies originated from the discrepancy on this, but I would have never expected such an improvement) * the number of neutrals decreases significantly but there is still discrepancy in the enriched sample * the total energy, separately for tracks and neutral candidates is definitely better * mxhadfit is ok These comparisons are not done on exactly the same chains of anything else because those were done by daniele and Urs. We still need to check that there are no other side effects. Once Urs finalizes his studies we should probably do a consistent round of checks. I am in particular concerned about a possible loss of efficiency. Nonetheless this clean-up seems to give a definite improvement in the data-MC agreement. ciao ric