SEQUENCE OF CONSTRAINED ANNEALED AVERAGES FOR ONE-DIMENSIONAL DISORDERED-SYSTEMS

We introduce a systematic way of implementing a sequence of constrained annealed averages which converges to the quenched average in one-dimensional systems with dichotomic disorder. It is formulated in the context of products of random matrices where the constraints correspond to imposing that each type of n-ple of consecutive matrices appears with the correct frequency according to the given probability distribution in the annealed average. We apply the method to a one-dimensional disordered Ising model, where the constraints have the effect of preventing the disappearance of frustration. © 1995 The American Physical Society.