Approximation by mutually completely dependent processes (Q811005)

The paper brings a very particular proof of the theorem that any stochastic process indexed on the integers can be approximated arbitrarily well in the \(L_{\infty}\) sense by a so-called mutually completely dependent process i.e. by such a stochastic process in which each value is (essentially) a 1-1 image of any other value. If, moreover, the original process has continuous one-dimensional marginal distributions, then these can be replicated in the approximating process.