Hi,
In the method plot(...) you should do:
[...]
fitsim1_5->SetRange(...);
fitsim1_5->SetMaximum(histsim->GetMaximum());
[...]
fitsim2_5->SetRange(...);
fitsim2_5->SetMaximum(histsim->GetMaximum());
Cheers,
O.Couet
-----Original Message-----
From: nguler_at_jlab.org [mailto:nguler_at_jlab.org]
Sent: Monday, November 22, 2010 9:40 PM
To: Olivier Couet
Cc: nguler_at_jlab.org; roottalk (Mailing list discussing all aspects of the ROOT system.)
Subject: RE: [ROOT] TF2 SetRange
Hello,
I have ru the code successfully before I sent. I just tried again and it is running. I do the following:
>root -b
root [0] .L aImage_test.cc+
root [1] all();
In the directory, I have only two files: aImage_test.cc, ngldata.txt I got rid of the Makefile to simplify the situation. The files I sent before should be the same but incase somethings mixed up, I am sending the two files needed to run the code again.
When you untar the file, nglcode.tar, you also need to gunzip the file ngldata.txt.gz inside that directory.. But I don't think that's the issue related to Seg Fault.
The machine I am running the code is a powerful 64bit linux cluster. But, I don't have any big arrays stored in the code. I checked the parameter arrays and they look fine. I am wondering: Are you getting Seg fault while reading the file (in the while loop), or after that during fittings. The way I fill the histogram is a bit strange, but I had to use that method because I had to use TH2F. As I mentioned , using TGraph2D is very very slow for reasons I don't understand (there is also TGraph2D method in the same code called graph() but not used). Here is the code snipped for the part that I read the file and fill the histogram. I think this is the only part prone to Seg fault but as I said it works for me without ant seg fault:
ifstream inFile_simdata;
inFile_simdata.open("ngldata.txt",ios::in);
i=0; j=0; N=0;
while(inFile_simdata >> fJ)
{
fJ = fJ*1E13;
iJ = int(fJ)/100;
for(int c=0; c<iJ; c++)
{
histsim->Fill(double(i),double(j));
}
//cout << N << "\t" << i << "\t" << j << "\t" << fJ << endl;
i++;
N++;
if(i==256) {i = 0; j++;}
if(j==256) break;
Thanks again for your help.
- Nevzat
> Hello,
> I compiled and link your program: I get a Seg Fault when I run it.
> Using CINT I get the same thing.
> Have you run what you sent me ?
> May be there is still some files missing...
> Olivier
>
> -----Original Message-----
> From: nguler_at_jlab.org [mailto:nguler_at_jlab.org]
> Sent: Friday, November 19, 2010 7:34 PM
> To: Olivier Couet
> Cc: nguler_at_jlab.org; roottalk (Mailing list discussing all aspects of
> the ROOT
> system.)
> Subject: RE: [ROOT] TF2 SetRange
>
> Sorry I forgot to cleanup the Makefile itself. I removed some unused
> libraries from the Makefile, also cleaned the code a bit more and
> attached the whole package again. It should work now and produce a
> .gif file. Actually, the Makefile is not necessary, the code can also
> be loaded under root cint, in which case the subroutine to run is all(); Sorry for the confusion.
> As a side note, I am using root version 5.26/00.
> Thanks again for your help,
> -Nevzat
>
>>
>> When I do "make" I get:
>>
>> g++ -o aImage_nifv1 aImage_nifv1.cc -O -Wall -fPIC -pthread -m32
>> -I/home/couet/root/include -L/home/couet/root/lib -lCore -lCint -lRIO
>> -lNet -lHist -lGraf -lGraf3d -lGpad -lTree -lRint -lPostscript
>> -lMatrix -lPhysics -lMathCore -lThread -lGui -pthread -lm -ldl
>> -rdynamic -L -lMinuit -I/home/nguler/eg1b/inc -L
>> g++: argument to `-L' missing
>>
>> make: *** [exe] Error 1
>>
>> It is because the variable CLAS_LIB is not defined in the Makefile....
>> what should it be ?
>>
>> OC
>>
>> Org: CERN - European Laboratory for Particle Physics.
>> Mail: 1211 Geneve 23 - Switzerland Mailbox: J25910
>> E-Mail: Olivier.Couet_at_cern.ch Phone: +41 22
>> 7676522
>> WWW: http://cern.ch/Olivier.Couet/ Fax: +41 22
>> 7670300
>>
>> On Thu, 18 Nov 2010, nguler_at_jlab.org wrote:
>>
>>> Hello,
>>> I tried gStyle->SetHistTopMargin(0); but it did not change anything.
>>> I attached a tar file which includes simplified code (I tried to
>>> clean as much as possible), a data file input and a makefile to
>>> compile (make test). I hope the file size is too big for roottalk.
>>> I am using the subroutine plot() in the code.
>>>
>>> Thanks for your help.
>>>
>>> P.S. As a side note: The subroutine graph() is trying to do the same
>>> thing as the plot() but by using TGraph2D. However, it is extremely
>>> slow and the graph it generates has large pixels which create a low
>>> resolution plot. I was thinking to ask this separately but might be
>>> more appropriate now.
>>>
>>>
>>>> Hello,
>>>> You did not sent a running macro showing the problem so I can only
>>>> guess what could be wrong.
>>>> It could be the following: histograms are not drawn from Min to Max
>>>> of their content. There is a margin on top, added to the maximum.
>>>> It could be that this margin provokes the difference in Z. To
>>>> suppress it do: gStyle->SetHistTopMargin(0); Cheers, O.Couet
>>>>
>>>> -----Original Message-----
>>>> From: owner-roottalk_at_root.cern.ch
>>>> [mailto:owner-roottalk_at_root.cern.ch]
>>>> On Behalf Of nguler_at_jlab.org
>>>> Sent: Wednesday, November 17, 2010 1:08 AM
>>>> To: roottalk (Mailing list discussing all aspects of the ROOT
>>>> system.)
>>>> Subject: [ROOT] TF2 SetRange
>>>>
>>>> Hello All,
>>>> I have 2D histograms with various peaks in each. I perform a
>>>> gaussian fit on the peaks. For the fit to be successful, I have to
>>>> set x,y ranges around the peaks while defining my TF2 object.
>>>> Therefore, the range of my TF2 is different than the range of my
>>>> TH2F. This becomes problem later when I plot the histogram and the
>>>> fit results together on the same canvas. So, I rescale the
>>>> histogram by using SetRange. To be more clear, I define TH2F:
>>>> TH2F *histsim = new
>>>> TH2F(histsimName,histsimName,256,0,256,256,0,256);
>>>> Then define TF2:
>>>> TF2 *fitsim = new TF2("fitsim",Gaus2D,0,80,75,125,5);
>>>> And I fit and plot:
>>>> histsim->Fit("fitsim","RN");
>>>> histsim->Draw("surf4");
>>>> fitsim->Draw("cont1, same");
>>>> But the fitsim is plotted on a different scale and does not overlap
>>>> with the peak anymore even if the fit is successful. So, instead I
>>>> do the
>>>> following:
>>>> histsim->Fit("fitsim","RN");
>>>> histsim->Draw("surf4");
>>>> fitsim->SetRange(0,0,256,256); //// necessary to rescale to the
>>>> histogram range
>>>> fitsim->Draw("cont1, same");
>>>> This plots the fitsim on correct x and y positions. However, the z
>>>> height of the fit is still off-scale. How can I rescale the fit
>>>> height so that it is the same as the peak height? I tried following:
>>>> fitsim->SetRange(0,0,0, 256,256, fitsim->GetParameter(0)); But
>>>> it did not have any visible effect, the z range of the fit is still
>>>> very different than the z range of the histogram.
>>>>
>>>> I appreciate your suggestions. Thank you,
>>>>
>>>> - Nevzat Guler
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>>
>>> - Nevzat Guler
>>> (757) 553-8318
>>
>
>
> - Nevzat Guler
> (757) 553-8318
>
>
This archive was generated by hypermail 2.2.0 : Tue Nov 23 2010 - 23:50:02 CET