Fitting the amplitude of a function

ROOT
1

I pose a simple question but so far I can’t find anyone who knows the answer:

In root, is it possible to fit a peak (from real data) and extract the amplitude of the function fitted?

Please, can anyone answer this?

Andrew

2

Search for “GetFunction” and “GetParameter” strings in the TH1 and the TGraph class descriptions.

3

Hi Wile,

Thanks for your reply. Unfortunately, I don’t think the solution is so simple.

As an example:

say I want to fit a gaussian to a peak in my data (counts vs channel). I can define the gaussian:

par[0]TMath::Exp( -1((x[0] - par[1])(x[0] - par[1])) / (2par[2]*par[2]) );

so, par[0] = amplitude, par[1] = mean and par[2] = width.

If I try to fit this function with all parameters free, the resulting value of par[0] does not equal the amplitude of my peak. I think this is because the function is internally normalised in some way.

Is there any way to extract the actual amplitude value?

Thanks

4

Try different TH1::Fit options: [code]{
gRandom->SetSeed(0);

TH1F *MyH1F = new TH1F(“MyH1F”,“a trial histogram”, 100, -5, 5);
MyH1F->Sumw2();
// sigma = 1, integral = 251, amplitude = 10.013 (at “bin width” = 0.1)
for (Int_t i = 0; i < 251; i++) MyH1F->Fill(gRandom->Gaus(0, 1));
// MyH1F->Scale(1.0, “width”); // then amplitude = 100.13

TF1 *MyF1 = new TF1(“MyF1”, “[0]TMath::Exp(-0.5((x-[1])/[2])**2)”, -1, 1);

Double_t GuessedAmplitude = 1, GuessedMean = 0, GuessedSigma = 1;

// must set initial (“guessed”) values of parameters
MyF1->SetParameters(GuessedAmplitude, GuessedMean, GuessedSigma);

std::cout << “MyH1F->Fit with … default flags …” << std::endl;
MyH1F->Fit(MyF1);

std::cout << “MyH1F->Fit with … W …” << std::endl;
MyH1F->Fit(MyF1, “W”);

std::cout << “MyH1F->Fit with … WW …” << std::endl;
MyH1F->Fit(MyF1, “WW”);

std::cout << “MyH1F->Fit with … L …” << std::endl;
MyH1F->Fit(MyF1, “L”);

std::cout << “MyH1F->Fit with … WL …” << std::endl;
MyH1F->Fit(MyF1, “WL”); // if MyH1F is weighted, i.e. must have Sumw2() set
}[/code]

5

Peak find&fit tutorial may also help: http://root.cern.ch/root/html534/tutorials/spectrum/peaks.C.html