Hello,
maybe this is an often solved issue but I couldn’t find any solution so far.
I am performing a fit with a sum of two Gaussian function and an 1st order polynomial via a user defined function like this:
TF1 *fitFun = new TF1("fitFun","gaus(0)+gaus(3)+pol1(6)",fitRangeStart,fitRangeStop);
histogram->Fit("fitFun","IRQ");
However, the resulting fitted function contains the two Gaussians as expected while the linear part is definitely not linear at all. See the result in the attached “strangeFit.pdf” file.
This happens only if the pol1 has to fit to a relatively steep edge. Flat distributions are fitted quite well.
I’ve already tried higher order polynomials to meet the histograms shape but I gained the same strange outcome.
Furthermore, I observed the same for a simple pol1 fit to a spectrum without Gaussian signals (see “strangeFit2.pdf”)
Any ideas how this strange result turns up?
strangeFit2.pdf (24.1 KB)
strangeFit.pdf (23.2 KB)