 |
 |
Hi Daniele, > > I find 75 events with nneu==0 && nchg==1, 70 of which have nle==1 and 5 > > have nle==0. None of the events passes the pcms>1 cut (in fact, none of > > these has a reasonable value of pcms). > > Not suprisingly all of them have a mxhadfit==0. > > So it looks like as if these events do not make it into the signal samples > > after the cuts (I would have been surprised if it were different..). > > > > But they are included in the multiplicity categories at truth level since > > the defintions of the categories is according to those at reco level. I > > checked explicitely that I do not see any truth event with neudaugen==0 && > > chgdaugen==1 (definition is according to the reco multiplicities such > > that the charged lepton is included) in this file. > > This means that these events are included in the xini spectrum and > > therefore the unfolding "knows" of them. > > > > I do not see a problem with these events and think that they are treated > > consistently, but maybe I misunderstood what you said. > > my point was the following: these events (I mean nle=1 and nchg=nneu=0) > could be used to perform the unfolding and can reduce the stat error > on the unfolding itself. It is just a matter of adding another category. > In principle there is no reason why we should not use it in the BRBR > extraction as well. We should just check the S/B for those events. > Am I wrong? I think I do not get your point, sorry. First, how can you have nle==1 && nchg==0? The lepton is included in nchg, so if you have a lepton, nchg is at least 1, correct? Second, we do not see what this other category for the unfolding should look like. It would need to have a consistent definition between reco and truth level. We could make a category that has nchg==1&&nneu==0, but it would be (at least nearly) empty on the reco side since these events do not pass the cuts. It would be empty on the truth side because there are no such signal events at truth level, correct? In the MC I looked at I did not see any reco'd events passing the cuts so how could one reduce the statistical error? Cheers, Kerstin