RooDecay and Custom Resolution Models

[abaoud]

Dear Expert,

I hope you’re doing well. I’d like to ask for your advice on how to handle a fitting situation involving two components: prompt and non-prompt electrons.

My plan is to:
Fit each component separately (prompt and non-prompt), Then fit their sum to the data.
For the prompt component, I’m using a resolution function.
For the non-prompt component, I use an exponential convolved with a resolution function.
I initially chose RooDecay for the non-prompt model. However, I encountered a limitation: RooDecay only supports a limited set of resolution models—specifically Gaussians or combinations of Gaussians via RooAddModel.
The resolution model I want to use is more complex: a sum of two Gaussians and a Breit-Wigner.
So I have two questions:

1. Is there a way to extend RooDecay, such as by adding RooBreitWigner as a valid subclass for the resolution model?
2. Alternatively, I’ve tried using RooExponential convolved with the resolution via RooFFTConvPdf but the problem is that i have a range consist of just 60 bins.

• Is this approach equivalent in functionality to RooDecay?
• I find this method a bit slow and challenging to handle. Is there any performance tip or better alternative?

I’d appreciate any guidance or suggestions you can offer on how to proceed with this setup.

Best regards,

[silverweed]

Hello @abaoud,

@jonas can probably help you best, although I think he’s off for this week so he might take a few days to answer.

[abaoud]

Thank you @silverweed. I will be waiting for @jonas feedback