Markov chains in Banach spaces on cycles (Q1273814)

This paper investigates the problem of defining denumerable Markov chains by using countable collections of weighted directed cycles, called denumerable cycle Markov chains (DCMCs). Since DCMCs admit invariant probability distributions, and their transition matrices satisfy the so-called cycle formula, the authors show that the DCMC problem may be solved by using suitably defined Banach spaces \(L_p\) \((p\geq 1)\) on cycles and edges. Necessary and sufficient conditions are given for a Fourier series to define sequences that satisfy the cycle formula. More specifically, the paper shows how to define DCMCs on the basis of denumerable collections of directed cycles, Fourier series, and sequences of Banach spaces \(L_p (N^2)\) that satisfy the cycle formula.