I would like to define a bounded distribution where the upper and lower bounds are not included in the domain.
I think I remember @rpgoldman asking a similar question one day
Am I mistaken Robert?
To do this you have to define an epsilon and then have variables that are bounded between l + ε and u - ε.
can I just clarify the advice given on the OP. When you say define an epsilon, do you just literally just mean something like the following:
epsilon = 1e-6 Bounded_dist = pm.Bound(pm.SomeDist, lower=1+epsilon)
Or am I missing something??