Hi Claire,
Before Scaling your histogram, do
x3->Sumw2()
This will make sure that errors will be correctly computed, otherwise
the scaled histogram will have errors = sqrt(contents)
Rene Brun
On Fri, 19 May 2000, Claire Gwenlan wrote:
>
> Hi Everybody,
>
> I've just been normalising some graphs which I've drawn with the
> Draw("e1p") option. I've used a couple of methods (which give the same
> results but I was just doing a couple of tests to make sure I know how to
> do things...I'm a bit new with root!). Anyway, one way was simply using
> the method suggested by Rene earlier this year, using the number of
> events, and the other was using the following:
>
> TH1F *x3=new TH1F("x3","X3", 8, 0.65, 1.0);
> .
> . (some other code)
> .
> integral=x3->Integral();
> if(integral>0) x3->Scale(1./integral);
> else{
> cerr << "Empty Histogram: " << x3->GetName() << endl;
> return kError;
> }
>
> Anyway, it seems the result has HUGE error bars, whichever method I use.
> Does the above scale the errors properly or not? Or do I have to do
> something extra to scale errors down?
>
> Oh, also, just another quick question. Is there a way to switch off the
> statistics box appearing automatically in the code, rather than just doing
> it by hand by clicking on the options box on the canvas?
>
> PS. Thanks Rene and David for answering my last question...enerything
> works fine now.
>
> __________________________________
> Claire Gwenlan
> Department of Physics & Astronomy
> University College London
> Gower Street
> LONDON
> WC1E 6BT
>
> Tel: (0171) 419 3454
> _________________________________
>
>
>
This archive was generated by hypermail 2b29 : Tue Jan 02 2001 - 11:50:25 MET