On Sat, 8 Feb 2003, Urs Langenegger wrote:

>
> Hoi,
>
> I  presume  the b2u  reweighting  is implemented  in  a  way that  the
> multiplicity fits can be done as usual?

The reweighting is done at Loop level that is blind about the fitting
technique...

> I get the following numbers on
> anaQA-n00:
>
>
> Sample      nocat      mult. fit
> --------------------------------
> B           0.0298     0.0213
> B0          0.0264     0.0332
> B+          0.0181     0.0182
>

The first column IS NOT showing the default results, am I right?
Were's the logfiles for those fit jobs?

> The B0 is drastically different.
>
> Could  someone  please  check  these  numbers?  An  confirm  that  the
> difference is not due to the implementation?
>

Given that the weights are computed at Loop level I do not see any
obvious 'technical' reason for having those two different results. The
fact is that for sure we're reweighting b2u MC in a different way for B0
and B+ and the mult fit can be more sensitive to that due to the different
eff in the various categories: I'm still trying to investigate why there's
such shift.... If the weights that you apply are the correct ones the
effect should, however be compatible for B0 and B+...

> But now to the second part of this mail:
>
> At some point  someone will calculate the significance  between the B0
> result of summer and this one ...
>
> We have shifted the result of  B0 by more than one (statistical) sigma
> with  respect to  the  previous version  (of  the BAD)!   I  am a  bit
> confused (and cannot find all  what we had yesterday in Alessio's page
> anymore, it  seems to  have been updated).   Assuming we MIGHT  have a
> background  problem in  B0,  do we  attribute  this to  signal?  If  I
> understand  Alessio's  webpage  correctly,   we  reweight  B0  and  B+
> separately? But I am not sure what the difference is between row 3 and
> row 5 in the "Default results"  table. I am somewhat concerned by this
> large shift.

Hi Urs,
as discussed in our last meeting we decided to reweight separately B0 and
B+ in order to take into count the differences btw the exclusive
spectra/BRs properly. So I've just splitted B0 and B+ and tested that the
weights that are coming from that operation have on the overall sample the
same effect as before. (value in 3rd and 4th column are showing you the
effect: ball value does not change while B0 and B+ do).
So the only diffenrece btw default result in row 3 and row 4 is that i've
switched on the reweighting separately for B0 and B+.
Going now to row 5 you'll see the new FINAL/DEFAULT value obtained after
ITERATIONS of weight calculations. I've started with our BRBR value (2.02)
instead of using 1.7 (PDG) and I've iterated the weight calculations
switching every time to the new measured value (waiting for a
stabilization of our sresult). The 3rd iteration was the good one and from
that one I've committed the weights and find out the new default.

Hope that this is a bit more clear...
So the only change is that the value for BRBR used in weight calculation
has been taken from our measurement iterating.
Let me know if there's anything still unclear ...
Alessio

>
> Cheers,
> --U.
>