RooGExpModel with an offset?


The current implementation of RooGExpModel does not allow for an offset of the Gaussian component. Does anyone have such an implementation already coded up? If so, could I have a look at it, please? :slight_smile:



We have exactly the same problem. We are modelling the noise in our detector as the convolution of a gaussian (given by the signal) and an exponential decay (given by the PMT after pulsing).

Unfortunately RooGExpModel assumes that the gaussian mean is zero, so we cannot this model for our problem. The zero assumption is not documented, the latest available version of the rootfit user guide (2.91-33) incorrectly suggests that means different from zero are supported. The online documentation is not very helpful either.

So, if anyone has implemented such model, it will be great to share it.