Hi All,

I am following along in the RooStats rs100 tutorial (P-values and confidence intervals in a number counting experiment), but I can’t figure out the “masterSignal” variable.

What are the units of this variable? It seems to be a scaling factor on the given expected signal. It is clear that it should be set to 0 when I investigate the null hypothesis; should it be set to 1 when I calculate CL_s+b? Why does it seem to have an upper limit of 3?

Cheers,

Jason

Hi Jason,

The maseterSignal is the ratio of the signal rate to the expected signal rate (eg. Standard Model prediction). So

0 = bkg only,

1 = sig+bkg as predicted by standard model (or whatever)

The limit at 3 is arbitrary, it’s set here:

root.cern.ch/root/html/src/RooSt … x.html#100

If you want it to be larger:

RooRealVar* masterSignal = workspace->var(“masterSignal”)

masterSignal->setMax( 10.)

The reason for introducing the masterSignal is that when doing a combination between channels you need to relate the signal in the different channels. In a situation like the standard model Higgs, you can relate the signal between the different channels to one overall “masterSignal”. You could be more model independent and not assume the structure given by the Standard Model, but that will not be as powerful of a test.

The documentation also points to:

arxiv.org/abs/physics/0511028

arxiv.org/abs/physics/0702156

cdsweb.cern.ch/record/1099969?ln=en

All the best,

Kyle