A functional limit theorem for the profile of \(b\)-ary trees (Q988760)

The paper deals with a general limit theorem for the profile of a very general class of \(b\)-ary trees that covers several previously obtained results in a unified way and extends them significantly, for example to random recursive trees, to random lopsided trees, to \(b\)-ary recursive trees or to plane oriented recursive trees. The proof method is based on a martingale approach. The main idea is to embed the discrete time process into a continuous time model, where the martingale structure is easier to handle and then to recover the limit theorems for the discrete model in a proper way. The result says (more or less) that the number of vertices \(U_{n,k}\) at level \(k\) in a (certain) random tree of size \(n\) is (a.s.) uniformly approximated by \(\mathbb{E}\, U_{n,k} \cdot M(k/\log n)\), where \(M(\alpha)\) is a proper stochastic process (of smooth functions) that is characterized by a stochastic fixed point equation (and is also limit of a martingale).