A class of scale mixtures of \(\operatorname{Gamma}(k)\)-distributions that are generalized gamma convolutions (Q502906)

A generalized gamma convolution is a probability distribution on \([0, \infty)\) with Laplace transform of the form \[ \phi(s)=\exp{\left(-as+\int_{(0, \infty)}\log\left({t\over{t+s}}\right)U(dt)\right)}, \] where \(a\geq 0\), and \(U(dt)\) is a nonnegative measure on \((0, \infty)\) with finite mass for any compact subset of \((0, \infty)\) such that \(\int_{(0, 1)}|\log t|U(dt)<\infty\) and \(\int_{(1, \infty)}t^{-1}U(dt)<\infty\). The main result is the following: Theorem. Let \(k\geq 1\) be an integer. Let \(Y\) and \(X\) be independent random variables such that \(Y\) has a Gamma\((k, 1)\)-distribution and \(X\) has a hyperbolically monotone probability density function of order \(k\). Then, the distributions of \(Y \cdot X\) and \(Y/X\) are generalized gamma convolutions. Some applications to excursion theory and the theory of exponential functionals of Lévy processes are given.