1: \begin{abstract} 

2: A \Pol urn process is a Markov chain that models the evolution of 

3: an urn containing some coloured balls, 

4: the set of possible colours being $\{1,\ldots,d\}$ for $d\in \mathbb{N}$.

5: At each time step, a random ball is chosen uniformly in the urn.

6: It is replaced in the urn and,

7: if its colour is $c$, $R_{c,j}$ balls of colour $j$ are also added (for all $1\leq j\leq d$).

8: 

9: We introduce a model of measure-valued processes that generalises this construction.

10: This generalisation includes the case when the space of colours is a (possibly infinite) Polish space $\PP$. 

11: We see the urn composition at any time step $n$ as a measure ${\cal M}_n$ -- possibly non atomic -- on $\PP$. 

12: In this generalisation, we choose a random colour $c$ according  to the probability distribution proportional to ${\cal M}_n$, 

13: and add a measure ${\cal R}_c$ in the urn, where 

14: the quantity ${\cal R}_c(B)$ of a Borel set~$B$

15: models the added weight of ``balls'' with colour in~$B$.

16: 

17: We study the asymptotic behaviour of these measure-valued \Pol urn processes, 

18: and give some conditions on the replacements measures $({\cal R}_c,c\in \PP)$ 

19: for the sequence of measures $({\cal M}_n, n\geq 0)$ to converge in distribution, possibly after rescaling. 

20: For certain models, related to branching random walks, 

21: $({\cal M}_n, n\geq 0)$ is shown to converge almost surely under some moment hypothesis;

22: a particular case of this last result gives the almost sure convergence of the (renormalised) 

23: profile of the random recursive tree to a standard Gaussian.

24: \end{abstract}

25: