Reinforced Galton-Watson processes. I: Malthusian exponents (Q6596387)

The abstract of the article states: ``In a reinforced Galton-Watson process with reproduction law \(\nu\) and memory parameter \(q \in (0, 1)\), the number of children of a typical individual either, with probability \(q\), repeats that of one of its forebears picked uniformly at random, or, with complementary probability \(1 - q\), is given by an independent sample from \(\nu\). We estimate the average size of the population at a large generation, and in particular, we determine explicitly the Malthusian growth rate in terms of \(\nu\) and \(q\). Our approach via the analysis of transport equations owes much to works by Flajolet and co-authors.''\N\NA first remark is that if \(\nu\) has unbounded support, then the mean population size of the reinforced Galton-Watson process grows super-exponentially fast, therefore the authors concentrate on the bounded-support case. In this case, a connection is drawn between reinforced Galton-Watson processes and certain multitype Yule processes. Much of the work is then dedicated to a fine analysis of joint probability generating functions of this multitype Yule process, using in particular techniques borrowed from analytic combinatorics.