Infinitely divisible distributions for rectangular free convolution: classification and matricial interpretation (Q2641902)

This paper is a continuation of a previous paper by the author where he introduced the convolution operator \(\boxplus_{\lambda}\) as a tool to study principal values of sum of large independent rectangular random matrices whose row/column ratio is close to \(\lambda\). Here, the author studies the notion of infinite divisibility for the convolution \(\boxplus_{\lambda}\), and establishes a Levy-Khinchine type theorem. He also defines a Bercovici-Pata type map and proves that it is a bijection turning classical convolution into the \(\boxplus_{\lambda}\)-convolution and preserving limit theorems, thus extending results of \textit{H. Bercovici} and \textit{V. Pata} [Ann. Math. (2) 149, No. 3, 1023--1060 (1999; Zbl 0945.46046)]. Last, a matrix model for the Bercovici-Pata type bijection is given.