Abstract
In this paper, we essentially characterize the real power of the convolution set of X + Y, where X and Y ≥ 0 are two independent random variables which have respectively a Bernoulli distribution, with parameter r ∈ ( 0 , 1 ) , and an infinitely divisible probability distribution ν. The above problem is equivalent to finding the set of x , y ≥ 0 such that the mapping z ↦ ( 1 − r + r e z ) x ( E ( e z Y ) ) y is a Laplace transform of some probability distribution. This class of real power of convolution x is provided and described. The obtained results generalize the case where Y is Gamma or Negative Binomial distributed.

This publication has 5 references indexed in Scilit: