Hi Dan,
In the classes TF1 and TH1, functionality is
already built-in for calculating quantiles
through
Int_t GetQuantiles(Int_t nprobSum, Double_t *q, const
Double_t *probSum)
Have also a look at quantiles.C in the
tutorial directory.
Eddy
--- Dan Holmes <Daniel.Holmes@cern.ch> wrote:
> apologies, please disregard the last question, it
> comes from my
> mathematical mistakes, not root.
>
>
>
> --
>
> CMS-Bristol
> http://cern.ch/dan.holmes
>
> ---------- Forwarded message ----------
> Date: Mon, 3 Nov 2003 14:45:55 +0100 (CET)
> From: Dan Holmes <dholmes@mail.cern.ch>
> To: roottalk@cern.ch
> Subject: gaussian integrals
>
> Hello,
>
> I am trying to convert gaussian significances to
> probabilities repeatedly
> and quickly within some looping code.
>
> What is the best way to do this? .. i have tried
> defining the erf and
> integrating it but i'm not convinced root is giving
> me the correct
> answers.. my code:
>
> {
> form1 = new
> TFormula("form1","pow(2.718281828,-pow(x,2))");
> TF1 f1("func1","form1",-10,10);
>
> f1.Draw();
> double prob=f1.Integral(-1,1);
>
> cout<<"returned integral is "<<prob<<endl;
> cout<<"prob is "<<prob*pow( 1.77245,-1)<<endl;
> // integral of -100 to 100 is 1.77245
> }
>
> returned integral is 1.49365
> prob is 0.842703
>
> ..but i think i should get an answer closer to 63%
> within 1 standard
> deviation of a gaussian??
>
> << root vers: 3.05/07>>
>
> ..can you see what i am doing wrong or suggest
> anything? ..maybe there is
> a tutorial on this but i couldn't find it.
>
> cheers,
>
> Dan.
>
>
>
>
>
> --
>
> CMS-Bristol
> http://cern.ch/dan.holmes
>
>
This archive was generated by hypermail 2b29 : Thu Jan 01 2004 - 17:50:16 MET