Ergodic theory and the strong law of large numbers on Riesz spaces (Q854059): Difference between revisions

The authors continue to develop Riesz space analogues of some central notions and results of probability theory. To give a flavour of the general set-up, note that the space \( L^1({\mathcal F}) \) of all integrable random variables on some probability space \( (\Omega,{\mathcal F},P) \) is not only a Banach space, but also a vector lattice (=: Riesz space) containing the constants, and that the conditional expectation \( E^{{\mathcal G}} : L^1({\mathcal F}) \to L^1({\mathcal F}) \) with respect to some sub-\(\sigma\)-algebra \( {\mathcal G} \subseteq {\mathcal F} \) is a linear operator which is not only norm continuous, but also positive, order continuous and idempotent with \( E^{{\mathcal G}}(L^1({\mathcal F})) = L^1({\mathcal G}) \) and that it preserves the constants. Dropping the probability space, the authors consider a general Riesz space \( E \) with a weak order unit and define a conditional expectation \( T : E \to E \) to be a positive order continuous projection for which \( T(E) \) is a Dedekind complete Riesz subspace of \( E \) and \( Te \) is a weak order unit whenever \( e \in E \) is a weak order unit. What is even more striking is the new notion of independence of Riesz subspaces with respect to a conditional expectation operator \( T : E \to E \), which corresponds perfectly to the notion of conditional independence of sub-\(\sigma\)-algebras with respect to a sub-\(\sigma\)-algebra \( {\mathcal G} \subseteq {\mathcal F} \). In this general setting, the authors prove abstract versions of ergodic theorems, of Kolmogorov's Zero-One Law, and of the Strong Law of Large Numbers for sequences of the form \( \frac{1}{n} \sum_{j=0}^{n-1} S^jf \) where \( f \in E \) and \( S : E \to E \) is an order continuous Riesz homomorphism.