Loss of ergodicity in the transition from annealed to quenched disorder in a finite kinetic Ising model (Q535042)

The author address the question when a particular Markov chain dynamics becomes ergodic as a function of some parameter. The dynamics is a function of a linear output of the system (the mean state, an average of the two values that each component of the system can take like in the Ising model) and a number of i.i.d. random variables (at most one per component of the system). The parameter, whose impact on ergodicity is studied, is the number of i.i.d. random variables that is used at each step of the discrete dynamics (between 0 and the -- finite -- size of the system).