Asymptotc Expansions for Strongly Mixing Harns Recurrent Markov Chains

Asymptotic expansions are derived for sums on the form ENf(g) and YNf(E, ) respectively, where 4j is a Harris recurrent Markov chain and the distribution off(4j) has a continuous component. Contrary to previous results it is only necessary to assume that the Markov chain is strongly mixing with polynomially decreasing mixing coefficient. This is achieved by introducing an atom into the state space.