search

SNEWS2 / snewpdag

SNEWS2 alert calculations
BSD 3-Clause "New" or "Revised" License
2 stars 13 forks source link

Chi2 combination ndof could be wrong #8

Open vkulikovskiy opened 3 years ago

vkulikovskiy commented 3 years ago

https://github.com/SNEWS2/snewpdag/blob/d773c0fa4aa940e312c925bbb7b96702059a9061/snewpdag/plugins/CombineMaps.py#L102

The number of degrees of freedom in the parameter fitting (SN location) using Wilks' theorem corresponds to the number of free parameters once L/L_best is calculated. So, I am not 100% sure that conversion from chi2->probability is correct here.

vkulikovskiy commented 3 years ago

We (me) can make a new unit test (if it doesn't exist yet): 1) Set of experiment pairs with known deltaT resolution (no bias for now). 2) True SN direction. 3) Time generation for each experiment pair. 4) Skymaps calculation by using all pairs together as described in 3.5 of https://arxiv.org/pdf/2003.04864.pdf 5) Skymaps calculation using pair by pair chi2 maps and their merging. Ideally 4 and 5 should be the same.