Multifractal scaling of products of birth-death processes (Q605873)

Let \(\{X(t)\}\) be a stationary birth-death process with birth rates \(\lambda_i\) and death rates \(\mu_i\), \(i= 0,1,\dots\). Define \(\Lambda(t):= \exp\{X(t)- c_X\}\) with constant \(c_X\) chosen such that \({\mathbf E}\Lambda(t)\equiv 1\). Let \(\Lambda^{(j)}\), \(j= 1,2,\dots\) be independent copies of \(\Lambda\) and consider \[ A_n(t):= \int^t_0 \prod_{0\leq j\leq n}\Lambda^{(j)}(sb)ds,\quad t\in [0,1], \] with positive real \(b\). Choose either \(\lambda_n= \lambda\), \(\mu_n= \mu_n\), \(n\geq 0\), and for \(X(0)\) the Poisson distribution \(\text{Poi}(\lambda/\mu)\), or \(\lambda_n= (n+ b)\lambda_n\), \(\mu_n= \mu_n\), \(n\geq 0\) and for \(X(0)\) the Pascal distribution \(\text{Pas}(b,\lambda/\mu)\), or \(\lambda_n= (N- n)p\), \(\mu_n= n(1- p)\), \(0\leq n\leq N\), and for \(X(0)\) the binomial distribution \(\text{Bin}(N,p)\), or \(\lambda_n= (N- n)(g- n)\), \(\mu_n= n(h-(N- n))\), \(0\leq n\leq N\) and for \(X(0)\) the hypergeometric distribution \(\text{Hyp}(N,g,h)\). Taking b sufficiently large, the authors prove \(L^2\)-convergence of the sequence \((\{A_n(t)\})_n\) to a process \(\{A(t)\}\), derive the behaviour of moments, and obtain an explicit expression for the Rényi function.