Hi Concezio,

thanks so much for the answer and for pointing me to specific places
in VVF! I am still in the process of going through the code and
digesting what I see...

However, it seems to me that the default behavior is to have all the
CB parameters fixed in all fits (unless -fixsh is explicitely used)
by looking at the code and my newest logfile.

I assume I'll have more stuff on this eventually...

Thanks,
Kerstin


> Hi Kerstin,
>
> I don't remember exactly the details, so perhaps Antonio and Roberto can
> correct me if I am wrong...
>
>
> > The parameters fixed in these fits come from mesparsetting.dat.
> > Where do these numbers in mesparsetting.dat actually come from?
> > I assume they come from some high statistics fits - could you
> > give some details about these (what kind of data/MC, cuts applied,
> > all params floated?,...)?
> >
>
> They are obtained on the relevant samples (data, vub, vcb, other) after
> applying the SL event selection.
> Note also that in the VVF code we actually refit for the PDF parameters
> after applying all cuts, and fix the fitted parameters in the subsequent
> mes fits in bins of the kinematic variables involved (look at FitMes
> method).
>
> > In the CM2 version of this file I see many comments of the sort
> > "# not updated to CM2" - why is this? Did we actually have a look
> > and see that the changes were negligible or has this simply not
> > been looked at?
> >
>
> I believe we planned to update them but at about the same time we
> switched to Thorsten's model for the PDFs and consequently used
> another .dat file (mesparsetting_thorsten.dat).
>
> > I see that we are fixing the width of the Crystal Ball function.
> > Why is this fixed also in those fits which do have a quite
> > large signal fraction? Has anyone ever tried to float this
> > in say the high stat fits (like for the correctionratios) and/or
> > the mES fits for the low mX bins? If not, why not?
> >
>
> I think they were allowed to float when determining the PDF parameters
> and fixed afterwards. Floating the mean and width of the CB function
> should be simple, just run subVirFit.pl with the flag -fixshape (yes, I
> know, it is reverse logic...). If you want to float the width only then
> you need to modify VubMesUnb in VirUtil.icc at around here:
>
>     if (fixpar == 0) { // preset parameters only
>       Rm->setVal(inputPar[VirClass::iMean]);
>       Rs->setVal(inputPar[VirClass::iSigma]);
>       ar->setVal(inputPar[VirClass::iArgus]);
>     } else if (fixpar == 1) { // set and fix all parameters (if
> reasonable value is given)
>       if (inputPar[VirClass::iMean]  > 0.)   { Rm->setVal(inputPar
> [VirClass::iMean]);  Rm->setConstant(); }
>       if (inputPar[VirClass::iSigma] > 0.)   { Rs->setVal(inputPar
> [VirClass::iSigma]); Rs->setConstant(); }
>       if (inputPar[VirClass::iArgus] > -55.) { ar->setVal(inputPar
> [VirClass::iArgus]); }
>
> Cheers, Concezio.
>
>