Hi Daniele,

thanks for helping.

> > We are trying to run over your signal MC and write out all events (also
> > the events without Breco candidate with the aim to get unbiased
> > information on mxhadgen for the unfolding of the m_X spectrum). Following
> > the instructions of
> > http://www.slac.stanford.edu/BFROOT/www/Physics/Analysis/AWG/InclusiveSL/islrecoil/islbrecoiluser.instructions.html
> > but using analysis-13b instead of analysis-13 and replacing Linux2 by
> > Linux24 in the following, the only other change we did was in IbuIsl.tcl:
> > instead of
>
> have you compared the log files?

Where can I find them? I would like to compare them.

> Alessio, can you check if you used the same tags and options they are
> using to run IslBrecoilUser? in particular for PID killing?
>
>
> > To run anaQA on our file we needed to make two changes to it:
> > as a default, fBRecoFlavor is =0 (now) and for events with
> > dnB0Lund[indexbestB]==0 (all n for charged, n==1 for neutral B)
> > fBrecoFlavor does not get changed (in util.icc) and
> > tmppurB =
> > brecosig[modeB-10000]/(brecosig[modeB-10000]+brecobkg[modeB-10000]);
> > does not get set for modeB<10000 (in recoilNtp.cc).
> > We checked that this and the original anaQA give the same results for the
> > variables quoted below on your file.
>
> I did not get this but if you have the same result it's fine?

So far we hope so. As long as we only use the events tree in the reduced
ntuples we should be able to look at brecoflav to see whether there was a
Breco or not and if there was none, we should not use any variables
containing information about the decay mode of the Breco and so on. At
least this is our hope.

> > after running anaQA on these files:
> > # events from mxhadgen            768             816
> > # events w/ pcms>1.                23              18
> > # events w/ pcms>0.                29              25
> >
> > For the # of events from mxhadgen we required brecoflav==0 here (without
> > this requirement we get 9218 for our file).
>
>
> which are the option you used to run anaQA? (-MC -d ? -pk?) are you using
> the HEAD version of VubAnalysis?

Yes we are using the HEAD version of VubAnalysis. And these are all the
options we use:
../bin/Linux24/anaQA -mc -pk -SN -r 1234 -C o00 -d 4 -D
/scratch/users/tackmann/hbooks/allsignconc/tests -c
chains/signal/testchain

where I copied the o00 file from urs
(~ursl/macros/fsx/vub/VubAnalysis/o00)
and testchain just contains this one file.

> > Since we did not understand that, we made a dummy check:
> >
> > We try to run and just write out only the events with Breco
> > candidate and compare with your root-tuples, we already find differences
>
> this is already the case... but I am probably missing something here
>
>
> > For example:
> >                                your file       our file
> > # events in file                  849             896
> > # B0 (from modeB0)                574             623
> > # chB (from modeChB)              539             579
> >
> > after running anaQA on these files:
> > # events from mxhadgen            768             816
> > # events w/ pcms>1.                23              18
> > # events w/ pcms>0.                29              25
>
>
> I see the same numbers that you have shown before. Can you explain again
> which is the difference between the two cases?

First we modified IbuIsl.tcl such that all events got written out (also
events without BReco candidate). Since we found the inconsistencies with
your files (s.a.) we tried to write out only the events having a Breco
candidate to make sure that our change in IbuIsl.tcl did not cause the
differences. We still saw the inconsistencies (and the same numbers as
before, which i guess means that events without Breco candidate do not
get a pcms>0 and no entries for mode(Ch)B(0)).

I hope my answer makes sense,
Kerstin