Limit Setting

[lopez86]

Could use some more limit setting options. Some easy ones to implement would probably be:

Maximum Gap method - an MC version is very easy and Yellen provides analytic expressions too if I can convince myself that I understand them.

CLs limit - limits in the presence of a background. This would be an alternative to something like the Feldman-Cousins limit. Used extensively in things like collider experiments

Best fit bands would also be important. Likelihood methods will be easiest but it will be more complicated than just getting limits.

DAMA/LIBRA uses a model-independent fit to show that they have an oscillating signal. Implementing something like this would be useful. Would need to work out a good way to do model comparison between flat vs oscillating signals. Maybe one of the early DAMA/NaI papers goes over how they do this?

[lopez86]

Related to this, I think a lot of the limit types take more or less the same input, so it might be best to have a single limit class and then let the user choose the algorithm

[lopez86]

CLs limits are now implemented as a function. Still need to work on the limit setting API