 |
 |
Hi all, I've made some extracheck in order to ensure myself that the efficiency seen on generic MC is the right one. I've taken new cocktail B0 MC and I've ran against it dumping the idGam information. I've produced the plots for the likelihood of events selected with: 1) PRMM2 >-3 and likelihoodKL >= -7 + lepton cuts + mes >5.27 + enriched like_test_nu.eps 2) PRMM2 >-3 and likelihoodKL >= 0 + lepton cuts + mes >5.27 + enriched like_test_hi_nu.eps I've extracted from those plots the ratio: #Entries_true_KL / #Entries_all 1) 7.2% 2) 12.1% The efficiency of the cut on the likelihood is: 37% (all entries) (applying a cut on likelihood >=0) 62% (only TRUE KL) (applying a cut on likelihood >=0) The results are consistent with what observed on generic MC: #Entries_true_KL / #Entries_all ~ 10% I think that this conclusion rules out this 'reweighting' method in order to extract the sys for KL. The mistake was mine producing the plots for true and fake KL not having noticed that BY CONSTRUCTION they had the same area (giving a ratio #Entries_true_KL / #Entries_all ~ 50%). Let me know if you'd like any other check done on that issue. ciao, Alessio ______________________________________________________ Alessio Sarti Universita' & I.N.F.N. Ferrara >>>I'm in Ferrara<<< tel +39-0532-974328 Ferrara roma +39-06-49914338 SLAC +001-650-926-2972 "... e a un Dio 'fatti il culo' non credere mai..." (F. De Andre') "He was turning over in his mind an intresting new concept in Thau-dimensional physics which unified time, space, magnetism, gravity and, for some reason, broccoli". (T. Pratchett: "Pyramids") On Thu, 27 Mar 2003, Alessio Sarti wrote: > > Hi all, > > I've tried to investigate a bit more the effect of reweighting applied > > from Urs. > > The mean of energy spectrum (best KLongs selected with likelihood method), > > after Urs rescaling is shifted of ~4% (I remind you that the 20% > > reweigthing is applied only to TRUE KL (associator) while the energy > > plotted here is referred to ALL the KL selected via the likelihhod method. > > > > The plots (Urs's files and old reweighted MC) are superimposed here: > > http://www.slac.stanford.edu/~asarti/recoil/devel/KLstu/old_vs_new.eps > > > > This seems to be in disagreement with what shown here: > > http://www.slac.stanford.edu/~asarti/recoil/devel/KLstu/true_vs_all.eps > > some time ago. > > I've tried to trace back the problem and seems to me that the > > normalization of two plots was wrong. I don't know why but the fake and > > true histos have ~ the same number of entries (768 vs 771): this is really > > suspicious and seems to be an artifact of the histo produced by the > > comparison routine instead of the really eff (50%) expected for our > > selection. > > > > I've tried to recompute the efficiency of the idGam associator NOT by > > looking at histograms but counting events. I'm still working on that > > and tomorrow I'll have the eff numbers. > > Hi all, > the net eff. that comes out directly from mes fit is: > 11%. This means that the plot shown before had the wrong normalitazion > being a root artifact: I apologize for that. > What I do not understand is the shift in the mean of ~4% : > applying a 22% correction to the energy to the 11% of the events I expect > the mean to be shifted of ~2.4% ..... > Probably we need to rerun dumping the idGam of the best kl redoing the > exercise on urs's files... > > alessio > > > > > I can't calculate such eff. in Urs's produced files (the idGam for best KL > > has not been dumped): if we want to perform the same study I need to > > rerun against the MC and dump the idGam of best KL neutral. > > > > Cheers, > > Alessio > > > > ______________________________________________________ > > Alessio Sarti Universita' & I.N.F.N. Ferrara > > > > >>>I'm in Ferrara<<< > > > > tel +39-0532-974328 Ferrara > > roma +39-06-49914338 > > SLAC +001-650-926-2972 > > > > >