Closed dynaryu closed 6 months ago
Do you mean why Msys_obs.p does not sum to 1? It is because of the conditioning.
I understand that the sum would not be 0 but wonder what p means corresponding to 0 and 1 compared to the test_inference1 where no observation was used. If 0 and 1 correspond to P(sys=0|obs) and P(sys=1|obs) respectively, then what Pf_sys means?
It represents P(sys=0,obs) and P(sys=1,obs). And so, P(sys=0,obs) + P(sys=1,obs) = P(obj) (i.e. the probability that the observation is made); and P(sys=0 | obs ) = P(sys=0 | obs ) / ( P(sys=0 | obs ) + P(sys=1 | obs ) ).
Can you explain meaning of pf_sys in test_inference2 of test_gen_bnb.py? It seems the result of get_prob is not much different from p values for binary system states.