On hidden Markov processes with infinite excess entropy (Q2248935)

The author considers discrete stationary hidden Markov processes for which mutual information between the past and the future, measured by excess entropy, is infinite. It is assumed that the number of observable states is finite and the number of hidden states is countably infinite. The author shows that, under these conditions, the block mutual information of a hidden Markov process is upper bounded by a power law determined by the tail index of the hidden states distribution. Three examples of processes are considered. The first two examples are nonergodic and the mutual information between the blocks is bounded by the logarithm of the block length (in the first example) and obeys a power law (in the second one). The third example is ergodic, with power law of the mutual information.