To improve Landau fit

ROOT
1

Hi everyone,

I would like to apply the best fit to the attached plot and as you can see the X^2/NDF values are so high. However, I could not find solution. Should I change parameters or define new function include gaus+landau? Because I have nearly 1000 plots like this and I want to do this with using macro not with GUI.

If someone could help me, I would appreciate.

Thank you very much



2

Hi,

this kind of automated procedures are never simple to set up. If there is no theory behind the model you choose for the fit, it might be ok to alter the functional form.

Danilo

3

Thank you very much for your response. I guess Moyal Fit will match better than Landau. Do you or anyone know how can I apply this fit? Should I define myself?

4

Hi,

I am not sure we provide it already. It should not be hard to implement it though. Here you can find the documentation about the creation of functions starting from custom C++ expression if you cannot express it as a simple string:
root.cern.ch/doc/master/classTF1.html#F5

Cheers,
Danilo

5

Hi,

Yes I could define a user defined function however I have more than 1000 plots like this. Also gap is always changing. I guess I must define function that find the gap,max,min etc… automatically.

How can I do this?Do you have any idea?

Thank you very much

6

For example, one could imagine of extracting some momenta of the distribution and set the initial parameters of the model to something depending from those to “help” the fit converge.