TFeldmanCousins question

Hi Rooters,

I’m trying to calculate an upper limit for some number of observed events, with some background. I’m trying out TFeldmanCousins and I start with the tutorial $ROOTSYS/tutorial/FeldmanCousins.C.

If I start with the default constructor and try obs=80, back=50 I get what looks like nonsense returned from GetUpperLimit(), 30.005, just obs-back+0.005.

If I SetMuMax(150) I get a more resonable number, 46.045, and in fact, raising the value of MuMax seems to make no difference in the upper limit returned.

But now I need to calculate an upper limit for cases where I have on the order of 300 observed and background events, and setting MuMax arbitrarily hight seems to involve more calculations than is reasonable as it takes forever for the class to return a value.

I’ve been going through the original Feldman-Cousins paper to better understand MuMax and MuMin, but it’s slow going. Is there anyone who has used this class for numbers comparable to what I am using? Is there any way to optimize MuMax/MuMin for the signal/background I am using? Or is this approach not appropriate for these values? Thanks in advance to anyone who can help.

Matt Bellis
Carnegie Mellon University
(office - JLab) 757-269-6076
(cell) 412-310-4586