An asymptotic expansion for the distribution of the supremum of a Markov-modulated random walk (Q638751)

The author obtains an asymptotic expansion for the distribution of the supremum of a Markov-modulated random walk. Let \(\{\kappa_n\}_{n=0}^{\infty}\) be an irreducible aperiodic Markov chain on a finite state space \(\mathcal N\). For every \((i,j)\in \mathcal N^2\), let \(\{X_m(i,j)\}_{m=1}^{\infty}\) be independent random variables with distribution function \(F_{ij}\). Assume that the families of random variables \(\{\kappa_n\}_{n=0}^{\infty}\) and \(\{X_m(i,j)\}_{m=1}^{\infty}\), \((i,j)\in \mathcal N^2\), are mutually independent. Define a Markov-modulated random walk by \(S_0=0\) and \(S_n=S_{n-1}+X_n(\kappa_{n-1},\kappa_n)\) for \(n\geq 1\). For \(x>0\), let \(\eta(x)=\min\{n\geq 1: S_n>x\}\) be the first time at which the random walk crosses the level \(x\). Define a matrix \(W=(W_{ij})\), where \[ W_{ij}=\text{P} [\kappa_{\eta(x)}=j|\kappa_0=i], \qquad (i,j)\in\mathcal N^2. \] The author proves an asymptotic expansion for the matrix \(W\) in terms of the roots of the characteristic equation and provides an estimate for the remainder term in the expansion.