Local times of Markov processes approximated by a generalized iterated Brownian motion (Q5939315)

Let \(L^x_t\) be a local time process for a recurrent strong Markov process \(X_t\) with state space \(E\). For a given element \(0\in E\) define \(T_0= \inf\{t\geq 0:X_t= 0\}\) and \(\tau(t)= \inf\{s\geq 0: L^0_s\geq t\}\). Let \(A_t= \int_E L^x_t \mu_A(dx)\) be a continuous additive functional of \(X\) with Revuz measure \(\mu_A\). Set \(\overline{\mu_A}= \int{\mathbf E}_0(L^x_{\tau(1)}) \mu_A(dx)\) and \(\sigma_A= (\text{var}(A_{\tau(1)}))^{1/2}\). Assume that \(\int{\mathbf E}_x(L^x_{\tau(1)})|\mu_A|(dx)< \infty\) and the cumulant of the subordinator \(\tau(t)\), \(t> 0\), is regularly varying at \(0+\) with index \(\alpha\in (0,1)\). The main result establishes a strong approximation result for the difference \(|A_t- \overline{\mu_A} L^0_t- \sigma_A W(\widetilde L^0_t)|\) as \(t\to\infty\), where \(\widetilde L^0_t\) is a copy of \(L^0_t\), independent of a linear Brownian motion \(W\).