Approximation of myopic systems whose inputs need not be continuous (Q1384901)

The paper addresses the problem of approximating myopic maps, i.e., not necessarily causal, ``fading memory'' maps with possible applications in image processing. The authors prove a theorem giving the necessary and sufficient conditions under which multidimensional shift invariant maps with vector-valued not necessarily continuous inputs drawn from a larger set can be arbitrarily well uniformly approximated by means of a structure represented by a cascade of a linear, causal or noncausal, memory operator and a nonlinear memoryless one.