Pólya urns with immigration at random times (Q1715527)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Pólya urns with immigration at random times |
scientific article |
Statements
Pólya urns with immigration at random times (English)
0 references
28 January 2019
0 references
This paper studies Polya urns with immigration at random times. Let \(\tau_1,\tau_2,\cdots\) be independent identically distributed integer valued random variables with distribution \(\pi=(\pi_k)_{k\ge0}\) and the renewal process is associated with the famous Polya urn model with \(b\) black balls and \(w\) white balls with extra black balls added each time a ball is drawn. The distribution of the number of white balls in the urn after \(n\) steps is denoted by \(P^{\pi}(b,w,n)\). For a non-negative sequence \((a_k)_{k\ge1}\) and \(v>0\) define the probability density \(u(x)=cx^{v-1}\exp(-v\int_0^xA(t)/tdt)\), where \(c\) is a normalizing constant and \(A(x)=\sum_ka_kx^k\). The distribution is denoted by \(UL(v,a_k)\). If \(\pi\) is a probability distribution on non-negative integers with mean \(\mu>0\) and \(\tau\) has the distribution \(\pi\) and \(k\) is an integer with either (a) \(\mathbb{E}\tau^p<\infty\) for some \(p>1\) and \(1\le k<(p/2-1)(\mu+1)-1\) or (b) there exists \(\varepsilon>0\) satisfying \(\mathbb{P}(\tau>n)\ge Cn^{-(1-\varepsilon)}\) for large \(n\) and \(k\ge1\), then there exists a positive constant \(m_k(b,w,\pi)\) such that for any \(X_n\) following the distribution \(P^{\pi}(b,w,n)\), it is shown that the expectation of \(X_n^k/n^{\mu/(\mu+1)}\) tends to the constant \(m_k(b,w,\pi)\) as \(n\rightarrow\infty\). It is also shown that if \(b=1\), the scaled urn limits are of the form \(UL(w,a_k)\) for some choice of the sequence \(a_k\). If on the other hand \(b>1\), the limits are similar but rescaled with a multiplicative beta variable.
0 references
distributional convergence
0 references
distributional fixed point equation
0 references
Pólya urns
0 references
preferential attachment random graph
0 references
0 references