Hi everyone,

In the context of my MSc thesis, I am trying to set 95 %CL limits, without incorporating any systematic effect. I have some background histograms, signal histograms and data histograms. I am trying to set limits without the help of the RooFit/RooStats packages, so everything concerning the statistical part of the procedure, is written from scratch (i.e. likelihoods, test statistic etc.).

When trying to set limits, I follow the steps that are described in this paper:

higgs_limits.pdf (771.6 KB)

However, when I try to implement the steps in order to get the observed limits, I get results that are far from reasonable. In the following lines, I post the code that I am using, as well as the root file I am getting my histograms from:

limits.C (4.2 KB)

drell_yan_sr_histos.root (39.0 KB)

In this script, I use a binned likelihood function, assuming that content of each bin is Poisson distributed. Then I proceed building the test statistic q_mu, as minus 2 times the logarithm of the likelihood function evaluated for the the test_mu value, over the likelihood function evaluated at the maximum likelihood mu.

I donâ€™t know what is wrong with the whole procedure. I feel very puzzled. Any help will be deeply appreciated.

Thanks in advance!

Cheers,

Panagiotis