Ciao Antonio, 
nice work. I think your study advocates the need to keep run periods and
charged/neutral B separate when fitting mES (at least for the
semileptonic events selection), at least for some parameters such as
alpha_ccb and all the signal parameters. In particular, it's significant
that n for the signal function is significantly different for the
charged and neutral B, whereas it is quite constant wrt run periods.
This is a consequance of the different charged and neutral B decays,
reflecting in different pi0 composition of the two samples, which give
you different tails in the mES distribution. 

Anyway, the definitive answer will come from the study on the yields,
where I suppose that there will be a bias when fitting all run periods
and B types all together, and this bias should be reduced when
separating the different subsamples, fitting them separately and adding
them. 
Concezio. 


Il giorno mer, 13-12-2006 alle 17:57 -0800, Antonio Petrella ha scritto:
> Hi
> I have posted on this webpage:
> http://www.slac.stanford.edu/~petrella/mesfits/parameters.html
> the study on the variations of pdf parameters as function of run period 
> and B charge.
> 
> Background pdf and signal pdf have been fitted separately first, then 
> all MC has been fitted using the parameters obtained. In the last fit 
> (all mc) 3 parameters are free to float.
> 
> My comments:
> 
> - parameter n for Cutoff crystall ball has very large error, sometimes 
> it was necessary to fix it to 20 (i.e. the tail was forced to go to 0 
> very soon otherwise this function would fit half of the events at very 
> low mes (5.22))
> 
> - parameter n for signal function (again the tail) is very sensitive on 
> the B charge.
> 
> I can start to examine the yields now.
> 
> Antonio