Hi,
I did few checks on the disagreement in the signal MC efficiency
observed by Roberto (as output of the fitter).
I used the following trees:
old prod:
/nfs/farm/babar/recoil/Vub_incl/root/anaQA-r00/csx-vubnre.root
new prod:
files in
/u/br/sacco/work2/test-21/workdir/2575/root/*root
Just by counting I get:
----------
after lepton cut (i.e.
"vub&&mes>5.27&&intpur>.5&&pcms>1&&lcharge+brecoflav==0" )
N(b0) old = 685
N(b0) new = 1647
N(bch) old = 1140
N(bch) new = 2391
----------
after all cuts (i.e.
"vub&&mes>5.27&&intpur>.5&&mm2<.5&&pcms>1&&mxhadfit>0&&mxhadfit<5.&&
brecocharge+xcharge==0&&lcharge+brecoflav==0&&
(wdeltam<-3.||brecocharge!=0)" )
N(b0) old = 209
N(b0) new = 469
N(bch) old = 440
N(bch) new = 889
(I had to put brecocharge+xcharge+lcharge==0 for the new production,
I thought this was fixed...)
Then the efficiency is
eps_ub0(old) = 0.305
eps_ub0(new) = 0.285
eps_ubch(old) = 0.386
eps_ubch(new) = 0.372
If I add the mx cut (mxhadfit<1.55)
eps_totb0(old) = 0.213
eps_totb0(new) = 0.211
eps_totbch(old) = 0.287
eps_totbch(new) = 0.265
Conclusions:
- the efficiencies for CM1 and CM2 are pretty consistent
- the problem is not in the production but in the fit machinery
- the efficiencies I get are very different from the ones shown by
Virginia (
http://www.slac.stanford.edu/~azzolini/Mxhad-Mxhadfit.txt
table 1 in the crosschecks. ) while for the efficiency for the hybrid
are similar.
I wonder if the reweighting in the fit is applied even for the
pure non-resonant case (it should not happen)
Daniele
> I have just posted my contribution to today's meeting:
> http://www.slac.stanford.edu/~sacco/mxhad/Mxhad-Mxhadfit.txt
>
> Cheers,
>
> Roberto
>
|