Dear all,

I have a question regarding the definition of the negative binomial
distribution. See the current draft (
https://github.com/cplusplus/draft/releases/download/n4861/n4861.pdf)
26.6.8.3.4 (p 1177).
The parameter k is specified as IntType, hence it cannot be used for real
positive parameters k, which would be completely valid from a mathematical
point of view.
This makes it unusable for practical purposes, e.g. for claim modelling in
insurance as there the claim number distribution may be negative binomial
with non integer parameter k.

Implementation wise it does not make much difference, e.g. in the Microsoft
STL implementation it is even casted back to double, see
https://github.com/microsoft/STL/blob/master/stl/inc/random code line 4310.

I guess the other implementations also use the Poisson Gamma mixture as
described in http://www.eirene.de/Devroye.pdf page 488, example 1.5.

What are the steps I can do to try to change this definition in the
standard?

Thank you
Best regards
Max