Exact Hausdorff measure on the boundary of a Galton-Watson tree (Q2370095)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Exact Hausdorff measure on the boundary of a Galton-Watson tree |
scientific article |
Statements
Exact Hausdorff measure on the boundary of a Galton-Watson tree (English)
0 references
22 June 2007
0 references
Let \(T\subset\bigcup_n\mathbb{Z}^n\) be a Galton-Watson tree with offspring distribution \(p_n:=P(N=n)\), \(n=0,1,2,\dots\), i.e., \(\emptyset \in T\), for \((i_1, \dots,i_n)\in T\), \(j\in\mathbb{Z}^+\) we have \((t_1,\dots,i_n, j)\in T\) if and only if \(0\leq j\leq N_{i_1,\dots,i_n}-1)\) \((N_{\dots}\) are iid copies of \(N)\) and if \((i_1,\dots,i_n)\in T\) then \((i_1,\dots,i_k)\in T\) for all \(k\leq n\). The boundary of \(T\) is defined as the set \(\delta T= \{(i_j)_{j\in N}:(i_1,\dots,i_n) \in T\) for all \(n\in\mathbb{N}\}\). The Galton-Watson process \(Z_n\) consists of the cardinality of the set \(F_n:=\{(i_1, \dots,i_n)\in T\}\). If \(Z_n\) is supercritical, one can define the following limiting random variable \(W:=\lim_{n \to\infty}Z_n/a^n\) where \(a=E(N)>1\) is the expected number of offspring. The author provides necessary and sufficient conditions for the almost sure existence of an absolutely continuous (w.r.t. the branching measure) exact Hausdorff measure on the boundary \(\partial T\): If \(K(x):= \int^\infty_xP (N>u)\,du\) satisfies \(\lim\inf_{x\to\infty}K(2x)/K(x)=0\), then there exists a (not necessarily unique) a.c. exact Hausdorff measure function \(\varphi\) for \(\partial T\) of the form \(\varphi(t)=t^{\log a}g (|\log t|)\) where \(g:\mathbb R^+\to \mathbb{R}^+\) and \(\lim\sup_{n\to\infty}g(n+1)/ g(n)<a\). Moreover, the \(\varphi\)-Hausdorff measure \({\mathcal H}^\varphi (\partial T)=C_\varphi\cdot W\) a.s. on \(\{\partial T\neq\emptyset\}\) where the constant \(C_\varphi\in(0,\infty]\) is exactly determined. For any Borel set \(A\subset\partial T\) one has also \({\mathcal H}^\varphi(A)=C_\varphi \cdot\mu(A)\) where \(\mu\) denotes the branching measure. If \(\lim \inf_{x\to\infty}K(2x)/K(x)>0\), then there is no a.c. exact Hausdorff measure for \(\partial T\), but a criterion classifying all gauge functions \(\varphi\) according to whether the \(\varphi\)-Hausdorff measure of \(\partial T\setminus\Delta\) \((\Delta\) is some rather complicated exceptional measure) is zero or infinite. Some important special cases are discussed, mainly correcting some earlier paper by \textit{Q. S. Liu} [Probab. Theory Relat. Fields 104, No. 4, 515--538 (1996; Zbl 0842.60084)] and proving some conjecture due to \textit{J. Hawkes} [J. Lond. Math. Soc., II. Ser. 24, 373--384 (1981; Zbl 0468.60081)].
0 references
exact Hausdorff mesure
0 references
shift-self-simular additive random sequence
0 references
boundary
0 references
branching measure
0 references
dominated variation
0 references
\(b\)-decomposable distribution
0 references
0 references
0 references
0 references
0 references
0 references