On tails of fixed points of the smoothing transform in the boundary case (Q1041057)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | On tails of fixed points of the smoothing transform in the boundary case |
scientific article |
Statements
On tails of fixed points of the smoothing transform in the boundary case (English)
0 references
27 November 2009
0 references
Let \(\{A_i\}\) be a sequence of positive random numbers such that only the first \(N\) of them are strictly positive, \(N\) being an a.s. finite random integer. Consider the nonnegative solutions of the distribution equation \(Z=_d\sum^N_{i=1} A_i Z_i\), where \(Z,Z_1,Z_2,\dots\) are i.i.d. and independent of \(N,A_1,A_2,\dots\)\ . If \({\mathbf E} \sum^N_{i= 1} A_i= 1\) and \({\mathbf E}\sum^N_{i=1} A_i\log A_i= 0\), it is known that solutions exist and that they all have infinite mean. Assuming in addition that \({\mathbf E}\sum^N_{i=1} A^{1-\delta}_i< \infty\) for some \(\delta> 0\), \({\mathbf E}(\sum^N_{i=1} JA_i)^{1+ \varepsilon}< \infty\) for some \(\varepsilon> 0\), and \(1<{\mathbf E} N\leq\infty\), the author proves the following: Let \(Z\) be a nonnegative, non-zero solution. If the \(A_i\) are aperiodic (i.e., there is no positive number \(h\) such that \(\log A_i\) is a.s. an integer multiple of \(h\), for \(1\leq i\leq N\)), then \(\lim_{x\to\infty} x{\mathbf P}(Z> x)= C_0\) for some finite strictly positive number \(C_0\). If the \(A_i\) are periodic, then there exist two positive constants \(C_1\) and \(C_2\) such that \(C_1=\liminf_{x\to\infty} x{\mathbf P}(Z> x)\leq \limsup_{x\to\infty} x{\mathbf P}(Z> x)= C_2\). The problem is reduced to the study of the behaviour at infinity of an invariant measure of a random difference equation, see \textit{Y. Guivarc'h} [Ann. Inst. Henri Poincaré, Probab. Stat. 26, No. 2, 261--285 (1990; Zbl 0703.60012)] and \textit{Q. Liu} [Stochastic Processes Appl. 86, No. 2, 263--286 (2000; Zbl 1028.60087)].
0 references
smoothing transform
0 references
branching random walk
0 references
distributional equations
0 references
random difference equation
0 references
0 references
0 references
0 references
0 references