Hi, the D*ln sample with partially reconstructed D* should fit for this kind of study (it has by definition no D mass constraint). The yield is approximately 1 evt / 200 B0 with the tight cuts I used for lifetime analysis, leading to a potential sample of several hundred events: however if one B is already reconstructed, cuts can be relaxed (much less combinatorics), further increasing the overall yield. I presume that these events could be tagged at rootple level without need of going back to beta. Cheers Franco Riccardo Faccini wrote: > Hi Oliver, > I guess it would be nice if we could discuss today at the meeting. In the > meanwhile just a couple of comments: > > > > > your observation is not a big surprise and actually > > this "smearing effect" is something I have already pointed > > out several times in previous talks > > (e.g see http://www.slac.stanford.edu/~buchmuel/cfit_vub_babarweek.pdf > > - dicussion of fit paramters for X) > > and it is also indicated in the > > kinematic fit part listed in the appendix of the Vub note. > > > > Since we are trying to measure the M_x distribution with an > > INCLUSIVE approach we do NOT know the underlying mass hypothesis > > (D,D*,X_H,...) for the reconstructed X-system. > > Therefore, it was decided to describe the X-system > > (see paragraph "Energy definition" in the note) only > > with 3 parameters. The energy of the X-system is then > > calculated assuming a fixed beta. > > > > Your EXCLUSIVE approach, of course, adds much more information > > because the mass hypothesis for the D is now know. Hence the only > > right parameterization for the D-vector in the fit would > > be a 4(!)-vector which includes the reconstructed D mass. > > In fact this is what we already utilize for the reconstructed B > > candidate where, of course, we know the mass hypothesis > > (again see paragraph "Energy definition" in the note) > > > > At the moment you are trying to "fit" the D mass by using > > only a 3-Vector and assuming fixed beta. > > Since you have already reconstructed the D meson this > > is obviously the the wrong Ansatz. > > Fitting "3-Vector+fixed beta" only makes sense > > for an INCLUSIVE approach where we are dealing with a variety of > > different mass states. Apparently this concept is not so bad for > > INCLUSIVE M_x because we see significant improvements not only > > in the resolution but also in the bias after the cfit. > > > > As far as your EXCLUSIVE study is concerned, there is a option > > in the code to go from a "3-Vector+fixed beta" to a full 4-Vector > > parameterization (like for the reco B). In fact for the > > moment study we are always running both parameterization > > for the X-System in parallel .... it works fine. > > > > I agree with you that the fixed beta option is not optimal, but it should > not create such big biases: if something is correctly reconstructed it > should not get screwed up. > > I think I found an easier (and more conforting) solution: the events that > get moved around are actually far from 0 in M_nu^2 (see m_nu^2 vs fitted > Mx in > http://babar.roma1.infn.it/~faccini/resoMx/mnuxhadfit.eps > ) > This means that I was looking at D0lnuX events that were reconstructed as > D0lnu and the kinematic fit was trying to recover the X on a statistical > basis. > > The only missing point is to understand why (actually, if, the statistics > might confuse things) the data worsen more than the MC. > One point I could not get from any of the material you provided is what is > the impact of the resolution on the Breco and how do you account for it. > At this point one useful test would be to smear the Breco in the coctail > MC and see if we can achieve a resolution similar to the generic one. > > > > > Interesting is your statement that the DATA MC comparison > > gets worse after the fit. This is something which I have > > not seen so far and it is certainly not true for the INCLUSIVE > > Mx distribution. Could you quantify this DATA MC comparison > > or point me to a plot which shows DATA vs MC before and > > after the cfit? > > You can have a look at the data-cocktail and data-generic comparison of > the MD masses, all linked from the web page I posted: for istance > data-cocktail with kfit > http://babar.roma1.infn.it/~faccini/resoMx/Dcock/Bch1knoneu.eps > data-generic with kfit: > http://babar.roma1.infn.it/~faccini/resoMx/Dgen/Bch1knoneu.eps > data-cocktail without Kfit: > http://babar.roma1.infn.it/~faccini/resoMx/NoKfitCock/Bch1knoneu.eps > > ciao > ric > > > > > Hi, > > > I have been trying to understand how much can we learn from reconstructing > > > B->Dlnu decays on the recoil of fully reconstructed Bs. > > > To this aim, I have been looking at the distribution of Mx in the D(*) > > > mass range (1.8-2.1 GeV). > > > In order to clean up the environment, I have requested: > > > 1) either a K+ or a Ks > > > 2) no neutrals > > > 3) I have looked at B0 (D-lnu or D*-lnu, D*- -> D0pi-) and Bch > > > (D0lnu) separately. > > > > > > The results are shown in > > > http://babar.roma1.infn.it/~faccini/resoMx/resoVub.html > > > > > > I think we can conclude: > > > > > > a) that without kinematic fitting the resolutions on Mx in data and MC for > > > tracks only are similar (see table at the bottom) > > > > > > b) that the measurement of the D0lnu and D+lnu events in our data show > > > a bit of inefficiency that deserves more attention (although the stat is > > > low...). Within the available statistics, resolutions and biases seem ok > > > (maybe the D* is a bit strange) > > > > > > c) that the kinematic fit has a bad effect on these kinds of events. This > > > is probably due to the fact that the pdf's used in it assume that there is > > > a component with missing particles. In this case some events jump on the > > > wrong part of the pdf and get nasty tails at high MX. > > > This can be seen in > > > babar.roma1.infn.it/~faccini/resoMx/fitNoFitD0lnu.eps > > > where the noFit mass is plotted versus the fitted mass for D0lnu events > > > in cocktail MC. > > > > > > The fact that the fit screws up "good events" is not necessarily a > > > problem, but this means that the fitted mass cannot be used for Dlnu > > > studies. > > > > > > d) after kinematic fitting the agreement between data and MC gets much > > > worse, in particular for the cocktail. Kinematic fitting might be the > > > origin of the fact that we need generic MC in order to get a reasonable > > > agreement with the data. > > > > > > more to come (it looks like a promising sample) > > > ciao > > > ric > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >