The density of the ISE and local limit laws for embedded trees (Q862221): Difference between revisions

The label of a node of a binary tree is defined by the difference of the number of right steps and the number or left steps on the path from the root to the node. For every integer \(j\) let \(X(j;T)\) denote the number of nodes in a binary \(T\) with label \(j\). The sequence \((X(j;T))_{j}\) is called the vertical profile of \(T\). It has been shown by \textit{J.-F. Marckert} [Random Struct. Algorithms 24, No.~2, 118--132 (2004; Zbl 1034.05016)] that the suitable scaled distribution \[ \tfrac 1n \sum_{j} X_n(j) \delta_j \] of the profile \((X_n(j))_j\) of random binary trees of size \(n\) converges to the ISE (integrated superBrowninan excursion). In the present paper the authors provide a local version of this result. More precisely they prove \[ \tfrac 1n \gamma^{-1} n^{1/4} \overline X_n (\gamma^{-1}n^{1/4} x) \to f_{ISE}(x), \] where \(\gamma = 2^{-1/4}\) for binary trees, \(\gamma = 1\) for complete binary trees, \(\overline X_n(x)\) denotes the linear interpolation of the vertical profile \(X_n(j)\) and \(f_{ISE}(x)\) is a stochastic process with \(d\mu_{ISE}(x) = f_{ISE} dx\) (that is, the (random) density of the ISE). The proof makes use of a compactness argument and Prohorov's theorem. In particular one has to show that the sequence \[ g_n(x) = \tfrac 1n \gamma^{-1} n^{1/4} \overline X_n (\gamma^{-1}n^{1/4} x) \] is tight.