I want to perform relativistic BW fit. Here I attach my fit implementation.

RooMyPDF_BW.C (1.8 KB)

RooMyPDF_BW.h (1.3 KB)

Let me know if this is a correct way of implementing the fit.

Here I explicitly provide the evaluate function I used in my fit.

double RooMyPDF_BW::evaluate() const

{

Double_t arg= x - mean;

return 1. / (arg * arg + 0.25 * width * width);

double pi = TMath::Pi();

double y = sqrt(mean * mean * (mean * mean + gamma * gamma));

double num = ((2 * sqrt(2)/pi) * mean * gamma * y );

double k = (num/ (sqrt(mean * mean+y)) );

```
double bw = (k/( pow((x * x - mean * mean),2) + mean * mean * gamma * gamma) );
return bw;
```

}

While for normalising the pdf I let ROOT use the numerical integration by specifying getAnalyticIntegral return zero.

Additionally, I am confused the way non-relativistic BW fit is implemented in ROOFIT. The function for defining BW is

double arg = (x-mean)

bw = gamma / (arg * arg + 0.25 * gamma * gamma)

return bw/(2 * Pi());

While in roofit , we ignore gamma, and define the function as

double arg = (x-mean)

bw = 1. / (arg * arg + 0.25* width* width)

Is there any specific reason to avoid the gamma parameter in defining BW in roofit. Do any such logic also applies when I want to determine relativistic BW fit.

Thank you in advance!