Re: TGraphAsymErrors

Hi,

 It is not possible to use TGraphAsymmErrors::BayesDivide if the histogram have weights different than one.

If the normal approximation for the efficiency error is good enough (i.e., if your efficiencies are not close to zero or 1) , you can use TH1::Divide with the option "B". (see http://root.cern.ch/root/htmldoc/TH1.html#TH1:Divide%2 ). Weighted histograms are supported in this case

    Cheers

 Lorenzo
On Jan 29, 2010, at 11:18 AM, Jordi Nadal Serrano wrote:

> Hello,
>
> I'd like to use TGraphAsymErrors for my efficiency calculation. However my two hist for the pass and the total sample are weighted with values not equal to 1.
> Is it possible to use TGraphAsymErrors with weighted histograms just by not using the "w" option as input argument? does it take the correct stats errors without the w option?
>
> Thanks in advance,
>
> Cheers,
>
> Jordi
>
>
>
Received on Fri Jan 29 2010 - 15:14:02 CET