I did tutorial summarized in here,
https_://twiki.cern.ch/twiki/bin/view/RooStats/RooStatsTutorialsJune2013
and I did exercise
[ Exercise 5: Compute a 95% Upper Limit with the Counting Model ]
But, MCMC Calculator defined in web page does not work …
in contrast, Profile Likelihood Calculator seems to work.
(why I can not upload files here…)
the files I copied is here,
https_://www.sendspace.com/file/hqqqea
My ROOT version is ROOT6.14/00
It seems he is an author of the files?
just comment on Exercise 5
it works for Gaus + Exp model, this is fine.
but, it doesn’t for “Poisson model with signal and background”
still waiting …
Hi ,
Sorry for the late reply. It is possible the MCMC calculator needs some thing to make it work with a specific model.
Your links for the files you have posted (corrected for an extra character)
Download test.zip from Sendspace.com - send big files the easy way have a code example using the BayesianCalculator not the MCMC one.
Do you have problem with the Bayesian calculator or the MCMC one or with both ?
Cheers
Lorenzo
The problem you have in the linked script is in this line:
par.setRange(par.getVal()-10*par.getError(), par.getVal()+10*par.getError());
Here you are setting a range for the nuisance parameters that is needed when using a Bayesian tool, since we integrate on them.
Now, since you are not performing first a fit to your model, you should maybe selected a fixed range looking at your specific input model. In your case you can easily assume that b can be limited for example between [0,5].
So you can change that line to:
par.setRange(0,5);
and your macro will work fine
Lorenzo
Dear ROOT experts, moneta, eguiraud, and jalopezg.
Thank you for the remind, reply and comment.
I will try to follow your comments and test it quickly!!
Best regards.
P.S. it works, I try to understand the framework and statistics itself !
This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.