-
Notifications
You must be signed in to change notification settings - Fork 74
Open
Description
I have been trying to use this package with nonlinear-optimization
and ad
for maximum likelihood computations, however this is highly non-practical since distributions always return Double -> Double
and ad
needs some kind of polymorphic interface to work. Are there any barriers for such a polymorphic interface?
One problem I see is the need for different contraints depending on the distribution (Erf
for Normal, Num
for others).
Metadata
Metadata
Assignees
Labels
No labels