As a crosscheck I tried to extract the ratio of BR from the full set of
generic MC we have at the moment (SP3+SP4). I'm not sure about the
statistics but it should be not far from 45fb-1 and Urs should confirm
this. The full chain is applied, I mean

 - creation of analysis root files using Ibu
 - analysis code (anaRecoil) run on them and creation of reduced Tree
 - finally VubFit on the reduced Tree

 The MC to train the fit for the background shapes is the total of Breco
cocktail. I did not weight properly the different contribution
(Dstar,Dc,.. ) as we have on data because doing that we have just 160fb-1
and this introduce a large systematics due to the MC. For sure, if we
decide to use the full set of breco cocktail MC (350fb-1) we should take
into account the different n(B0)/n(Bch) respect to the data and I think
Oliver can tell us something about this systematics.

 The MC used to calculate Vub efficiency and shapes is the
vubmix-brecococktail one. I will try ASAP to use the vubmix-generic.

 The fitted sample (completely indipendent) is the full set of generic MC.

 We expect to see something like:

 BR(b->ulnu)/BR(b->clnu) ~ = 1.16%

 since the BR(b->ulnu) generated value for SP3 (~1/4 of the total) is
9.4*10^-4 and SP4 one is 13.5*10^-4. If we look at the mes plot for all
the events and for vub and vcb events we get

all : 309 K
vub : 689+-40         -> BR(b->ulnu)~1.1*10^-4
vcb : 66.0K+-.5K      -> BR(b->clnu)~10.7%
vub p(lep)>1.&&one lept : 353 +- 26
vcb p(lep)>1.&&one lept : 29.4K +- .3K -> BR/BR ~ 1.2%

and everything seems to be quite in agreement with the generation
and the correspondent plot are in

http://www.slac.stanford.edu/~daniele/vub/mesall.gif
http://www.slac.stanford.edu/~daniele/vub/mesvub.gif
http://www.slac.stanford.edu/~daniele/vub/mesvcb.gif
http://www.slac.stanford.edu/~daniele/vub/mesvublept.gif
http://www.slac.stanford.edu/~daniele/vub/mesvcblept.gif



The result of the fit is the following

BR/BR = 0.0130+-0.0034

and this is very compatibile with the generated value.

the plot are in

http://www.slac.stanford.edu/~daniele/vub/data.eps
http://www.slac.stanford.edu/~daniele/vub/fitresults_nocat.eps

(some fits are not optimal but fixing some parameters the results is
pretty stable)

We have 100 Vub events for Mx<1.6 GeV (-> 135 events considering the whole
spectrum) and 244 background events.

Taking into account the difference in statistics (on data we should have
170K events) and the PDG BR (1.7*10^-3) and assuming that the efficiency
on data is the same of the MC we could estimate that we will
have 78 vub events, 134 bkg events and error should be

sigma(BR/BR) = 0.0042

sigma(BR/BR)/(BR/BR)=25% -> sigma(Vub)/(Vub)=12.5%


Daniele