Hi Jyothsna,
I have spent several hours trying to fix this problem, but I must admit
that I am stuck now. I would appreciate the help from a mathematician,
expert in statistics,etc.
The computation of errors when dividing two TProfile is not correct
in the current version of Root. I have added a warning and a message
explaining how to circumvent the problem in the development version.
If you have TProfile *h1, *h2, *h3;
doing h3->Divide(h1,h2);
will not compute correct error bars. To get the right result, project
the TProfile to a normal histogram, then divide the two histograms, eg
TH1D *p1 = p1->ProjectionX();
TH1D *p2 = p2->ProjectionX();
p1->Divide(p2);
Now p1 will have the right error bars.
If somebody is interested to fix the computation of errors in
TProfile::Divide, please let me know.
Rene Brun
On Tue, 17 Sep 2002, Jyothsna Rani wrote:
>
> Hi,
>
> I would like to access two profile/ 2D histos from two different files and
> would like to divide them.
>
> When I use the following code to do that I get a funny plot for sign. Is
> there a better way of doing it?
>
> Thanks for any help,
> Jyothsna
>
> ////////////////////////////////////////////////////////////
> {
> TFile bbh("signal.root");
> TFile qcd("bgd.root");
>
> TCanvas *c1 = new TCanvas("c1","c1",600,600);
> c1->cd();
> bbh->cd();
> signal_et1cut.Draw();
>
> TProfile *sig_et1 = (TProfile*)signal_et1cut->Clone();
>
> TCanvas *c2 = new TCanvas("c2","c2",600,600);
> c2->cd();
> qcd->cd();
> bgd_et1cut.Draw();
>
> TProfile *bgd_et1 = (TProfile*)bgd_et1cut->Clone();
>
> TCanvas *c3 = new TCanvas("c3","c3",600,600);
> c3->cd();
> TProfile *sign = new TProfile("sign","",50,0.,95.,0,800);
> sign.Divide(sig_et1,bgd_et1,1.,1.,"B");
> sign.Draw();
> }
> -------------------------------------------------------------
> K. Jyothsna Rani, Phone Number:
> Tata Institute, 630-840-3590 (Office)
> DZero Experiment, 630-840-4236 (Residence)
> Mail Station : 352, FAX:
> P.O.Box : 500, 630-840-8481
> Fermilab, E-mail:
> Batavia, IL - 60510 jyothsna@fnal.gov
> ---------------------------------------------------------------
>
This archive was generated by hypermail 2b29 : Sat Jan 04 2003 - 23:51:10 MET