When is the Conway-Maxwell-Poisson distribution infinitely divisible? (Q2070628)

The Conway-Maxwell-Poisson distribution has the probability mass \[ p_k=\frac{1}{Z(\lambda,v)}\,\frac{\lambda^k}{(k!)^v}\quad k=0,1,\dots, \] where either \(\lambda,v>0\) or \(v=0\) and \(\lambda\in\left]0,1\right[\) and \(Z(\lambda,v)\) is a normalizing constant. The authors prove that this distribution is infinitely divisible if and only if either \(v=1\), when it is a Poisson distribution with parameter \(\lambda\), or \(v=0\), when it is a geometric distribution. This result has also the consequence that Steutel's necessary condition for infinite divisibility [\textit{F. W. Steutel}, Stochastic Processes Appl. 1, 125--143 (1973; Zbl 0259.60011)] is not sufficient.