Hi Antionio,

the errors are definitely too huge. Something must be wrong.

Could you please post also the fit parameters and their errors?

Heiko

On Wed, 13 Sep 2006, Antonio Petrella wrote:

> Ok, let's try this gaussian
> 
> I get the following table...(errors are so huge: is it normal?)
> 
> 
> #mx_l mx_h  corr      err_corr
> 0.00     1.55     1.821 +- 0.577
> 1.55     1.90     3.411 +- 95.487
> 1.90     2.20     2.839 +- 96.872
> 2.20     2.50     2.366 +- 100.470
> 2.50     2.80     1.948 +- 106.299
> 2.80     3.10     1.583 +- 114.362
> 3.10     3.40     1.271 +- 124.660
> 3.40     3.70     1.008 +- 137.194
> 3.70     4.20     0.726 +- 157.384
> 4.20     5.00     0.406 +- 198.663
> 
> Antonio
> 
> Heiko Lacker ha scritto:
> > Hi Antonio,
> > 
> > maybe this is not too surprising after all since the first bin 
> > contains the largest fraction of the signal.
> > 
> > Now, that I'm thinking of it: there is a fit function which
> > would avoid the problem of becoming negative, but which would
> > nevertheless give probably a reasonable fit to the correction 
> > factors: a Gaussian.
> > 
> > Cheers,
> > Heiko
> > 
> > 
> > On Wed, 13 Sep 2006, Antonio Petrella wrote:
> > 
> >> Hi all,
> >>
> >> here are the results of the jobs with new correction factors strategy 
> >> (i.e. fit with a first order polynomial starting from the second bin):
> >>
> >> PBRBR= (109 +- 10 +- 4) e^-4
> >> chi^2 of the mx fit = 25.12/7
> >>
> >> I also run the systematics and the value I get is
> >> sigma=22.5%
> >>
> >> These are the values that I should add to the talk, but are not 
> >> encouraging...
> >>
> >> Antonio
> >>
>