VavilovAccurate normalization

Roofit and RooStats
1

Hi,

I am trying to compare my data with the model based on VavilovAccurate (no fit). It looks like it is working quite well if I normalize it. However, Pdf() function gives me values which don’t sum to 1 (for given E and dx, sum through dE). On top of that it looks like normalization is energy and path dependent.

Here is a snippet:

ROOT::Math::VavilovAccurate * vav= new ROOT::Math::VavilovAccurate();
pdf = vav->Pdf(lambda_L, kappa, beta*beta);

From theory:
P(dE, E, dx) = 1/ksi*(Nasty integral)
I thought that Pdf might give me just the value of the integral.
When I tried to add 1/ksi the values are even higher than 1.

What am I missing?
Thank you.
Best,
Martin

2

Hi,

I did not observe this problem, I see that VavilovAccurate::Pdf returns correctly a normalised pdf. Maybe it depends on the values of the kappa and beta2 parameters.
This is an example code (but working only with CLING)

ROOT::Math::VavilovAccurate * vav= new ROOT::Math::VavilovAccurate();
auto lambdafunc = [](double *x, double*){ return vav->Pdf(x[0], 1., 0.5); }
TF1 * f = new TF1("f",lambdafunc,-10,10,0);
cout << f->Integral(-10,10) << endl;
f->Draw();

Lorenzo

3

Hi Lorenzo,
thank you for your reply. Using your snippet I realized what I did wrong. I forgot to transform integration constant while going from dE to lambda.
Thanks.
Best,
Martin