Hi Dan,
Poisson function is gaussian like only for par >> 1.
Check your function plot by
poiss2->Draw()
Smbat
On Tue, 25 Nov 2003, Dan Holmes wrote:
> root version 3.05/07
>
> I am using the root Poisson function:
>
> i do:
> poiss2=new TF1("poiss2","TMath::Poisson(x,[1])",0,99);
>
> float x0=40;
> poiss2->SetParameter(1,x0);
> poiss2.Integral(x0,99)
>
> >> this returns (Double_t)4.89489910522337757e-01
> >> ok so i expect here in the gaussian limit there to be 50% of the
> >>distribution >X0, this answer is close enough i guess?
>
> now i do:
> float x0=0.3;
> poiss2->SetParameter(1,x0);
> poiss2.Integral(x0,99)
>
> >>this returns (Double_t)3.87128082086576730e-01
> >> meaning that 39% of the distribution lies above X0.
>
> Is this true or am i using the function in a region where it becomes
> non-valid?
>
> ie. is there really not 50% of the distribution either side of the
> parameter X0 in a Poisson distribution when X0 is small? (sorry, this is a
> maths question but i could'nt find a comprehensive answer in my stats
> book..) If it is a problem with the root approximation than can you give
> me an idea of the range for which i can use it?
>
> cheers,
>
> Dan.
>
>
>
>
>
>
>
--
This archive was generated by hypermail 2b29 : Thu Jan 01 2004 - 17:50:17 MET