Anxiety town
The best use of anxiety is to poop
Anxiety town. Ok right now the most pertinent thing is that I need to go get myself to Boulder planet. Feed myself, pay for tickets then boulder.
But I’ve been anxious about not being able to really understand how this -2 log likelihood ratio thing works.
It’s just a one bin Poisson.
I have some signal expectation and bg expectation. Then I draw a single number from poisson with expectation = signal expectation + background expectation.
As expected, in the totally unconstrained fit, the maximum likelihood estimator of the signal expectation + background expectation is the number I drew from that poisson with expectation = signal expectation + background expectation.
So now my denominator is sorted.
Then I need to figure out my numerator, which is a function of the tested signal expectation.
So I defined an array for this signal expectation. For each element in this array, I fix the signal expectation at this value and then I find the value of bg expectation that maximises the likelihood, not allowing the estimator to go negative. Then I repeat this for all the tested mu in the array.
And I got a fucked up curve.
Urghhhhh.
Okie done pooping.