Penalization of Galton Watson tree with marked vertices
Abstract
In this presentation, we are interested in Galton-Watson trees whose particularity is that each node can be marked with a probability depending on its number of children, independently of the other nodes. Subsequently, using a method called penalization we favor trees with a large number of marks. More precisely, this method makes it possible to obtain martingales which are in our case functions of $M_n$, the number of marks up to generation $n-1$. These martingales being positive and with mean 1, we can then define new probabilities under which we study the laws of marked trees.