Closed mkhattab940 closed 6 years ago
Added to my list.
There seems to be no actual feature.
I think that what you can do, for now, is to have the distribution function return the 0 measure when the parameter is out of bounds. i.e.
reject. measure(real)
for example is the 0 measure over the reals.
I believe that is sufficient to close the issue?
Some distributions are defined in terms of parameters with limitations of their value (e.g. 0 < k < 1). Need a way to throw an error in case a language user calls the distribution with parameter values violating such restrictions.