Invariance principles for Galton-Watson trees conditioned on the number of leaves (Q444351)

The author investigates the asymptotics of critical Galton-Watson trees with offspring distribution which is aperiodic and belongs to the domain of attraction of an \(\alpha\)-stable law, \(\alpha\in (1,2]\). The first main result provides the asymptotics for the probability of having \(n\) leaves, as \(n\to\infty\). The second main result treats Galton-Watson trees conditioned on having a large fixed number of leaves. For such trees, an invariance principle is proved which states that the Lukasiewicz path, the height function and the contour function, all properly rescaled, converge weakly to a triple \((X^{\text{exc}}, H^{\text{exc}}, H^{\text{exc}})\), where \(X^{\text{exc}}\) is the normalized excursion of a strictly \(\alpha\)-stable spectrally positive Lévy process, and \(H^{\text{exc}}\) is its associated continuous-time height function. In the case \(\alpha\in (1,2)\), the last result along with the continuous mapping argument immediately leads to a limit theorem for the law of the maximal degree in the conditioned Galton-Watson tree. In the remaining case \(\alpha=2\), which requires a separate treatment, it is shown that the maximal degree, properly rescaled, converges to one in probability. Finally, extensions of the two main results are given to the situation when the event ``the number of leaves equals \(n\)'' is replaced by ``the number of vertices with degrees belonging to the given subset of \(N\) equals \(n\)''.