Hi Kerstin,

> > In the BAD we do the following
> >
> > Evaluate B->semilep sys (using ONLY -b flag when submitting fit and Sys >
> > 0)
> > Evaluate  D->Excl sys  (using only -d flag when submitting fit and Sys
> > >=2)
> > Evaluate  D->Incl sys  (using only -d flag when submitting fit and Sys
> > =1)
> >
> > And then we merge the B->semilep sys with the higher contribution from
> > D->Excl or D->Incl .
>
> * What is meant by "merge"?
>   a) Combine the results for the errors for the B->semilep sys with D->Excl
>      sys and also the results for the errors for the B->semilep sys with
>      D->Incl sys and then look at these two combinations and consider them
>      as the error (obtained in two ways, excl and incl for the D->
>      sys)(Thus merging the results.)
>   b) Redo the procedure and evaluate the error for B->semilep sys and
>      D->Excl sys together and then evaluate the error on  B->semilep sys and
>      D->Incl sys together (rather than just merging the results). (Thus
>      merging the procedure. If this, what is taken as the error in the
>      end, the errors obtained separately for B and D or the error
>      obtained from the "merged procedure" or denepnding on the results?)

a far as I remember we did  B->semilep sys and D->Excl sys together and
D->Incl separately. Then we added them in quadrature. We consider the two
effects as uncorrelated (take into account the correlation would a
nightmare and probably impossible). This approach should conservative
(also because there is some double counting). BTW the inclusive part turned out to be
small with respect to the other.
Alessio, can you confirm that?

> * Theory error
>   - Is there any consensus how this should be done?

good issue. I would start doing the same that we did for the PRL. It is
still an open issue.

>   - Concerning how you did it as it is in the BAD: It looks like you
>     varied a and mb independently for the fits, but took a delta mb = 150
>     MeV rather than 90 MeV to account for the correlations. Did it
>     understand that correctly?

no, we varied them using 120 as mb uncertainty and taking into the .8
correlation between mb and a. There is some confusion in the BAD 540 about
this because we finalize this syst. uncertainty just before submission and
putting the results directly in the PRL (in it you can find the actual
values we used). Sorry about that.

> * b->u fractions
>   Which values and errors are you using for the reweighting of the B->
>   pi/rho l nu?

PDG ones

>   When you do this, do you reweight the other b->u decays as
>   well to have constant b->u branching ratio?

yes

>   Which values are used for the inclusive b->u branching ratio variation?

the one we are measuring ~2.2 *10-3...

>   Could you give me a hint where in the code this was done? So far I could
>   not find anything related to that.

there is no committed code for it. I recalculated the weights by myself
and implemented some changes in the code. It would be trivial to be redone
(calculating the weights can be a bit more painful).

BTW I found this pretty old posting that give some details

http://www.slac.stanford.edu/cgi-bin/lwgate/VUB-RECOIL/archives/vub-recoil.200302/Author/article-74.html

> * ssbar popping
>   How (and where?) are you varying the ssbar popping events?

in BAD 540 you can find the description. There are also some info in all
postings regarding ssbar of this page

http://www.slac.stanford.edu/cgi-bin/lwgate/VUB-RECOIL/archives/vub-recoil.200303/author/

Daniele