Dilations and stochastic processes (Q1079287)

Using in an essential way the author's results in Port. Math. 42(1983/1984), 137-149 (1984; Zbl 0571.60004), it is mainly shown how to construct a stochastic process \((X_ t)_{t\in C}\) such that \((f\circ X_ t)_{t\in C}\) is a submartingale for each \(f\in C\), given a family \((P_ t)_{t\in C}\), \(C\subset {\mathbb{R}}\), of probability measures on a standard space S, which is increasing relative to a fixed admissible cone \(K\subset C(S).\) The association is made in such a way that \(P_ t\) is the distribution of \(X_ t\) for each \(t\in C\). When the parameter set C is countable, S can be a quite general metrizable space, while the continuous functions in K must be \(P_ t\)-integrable for all \(t\in C\) and, moreover, either they need be bounded from above or else the cone K need be invariant under the minimum operation. On the other hand, if S is compact, C can be any subset of \({\mathbb{R}}.\) It is also explained how \textit{T. Kamae} and \textit{U. Krengel}'s results in Ann. Probab. 6, 1044-1049 (1978; Zbl 0392.60012) relate to the present article.