In
VirVubFitter rf-052804
I have implemented the possibility to apply the D*lnu FF reweighting to
the vub-recoil analysis. It is enough to set doFFreweight 1
in the setting files. This will turn on the reweighing from
(r1,r2,rho2)=(1.18,0.72,0.92)
to (1.34,0.91,0.81)
[ these numbers are hardwired in VirUtil.icc ]
notes:
- you need to have three more variables for D*lnu events to be
able to apply this. I have produced genbnu ntuples with this info in
/nfs/babar/recoil/Vub_incl/root/anaQA-r06
To get the right effect you need to mix these with the genbch ntuples from
r00, as in /nfs/babar/recoil/Vub_incl/root/anaQA-r06/genChain
- the shift in FF is not negligible. In particular at low q^2 the
change in r2 can cause quite large weights. We truncate the weights at
100, therefore you will see a few warnings about this. The great majority
of the events with this extremely high weights are thrown away in the
analysis (low p*).
- the mean weight ~1.1, which means that the overall fraction of
D*lnu events changes in this game. Since I coded directly the pdf, I
believe the weights do not need to be reweighted to unity, but that this
is an indication of the fact that the sample which gets into the ntuples
(with one B) is biased.
- I made also a tag VubAnalysis rf-052804 to produce root files
with the additional variables
comments are welcome
ciao
ric
|