A central limit theorem for functions of a Markov chain with applications to shifts

AbstractA sufficient condition is developed for partial sums of a function of a stationary, ergodic Markov chain to be asymptotically normal. For Bernoulli and Lebesgue shifts, the condition may be related to the Fourier coefficients of the given function; and the latter condition is shown to be satisfied by most square integrable functions in the case of Bernoulli shifts