Convergence of Riesz space martingales (Q2503003)

This paper builds on previous work of the authors [Indag.\ Math., New Ser.\ 15, No.~3, 435--451 (2004; Zbl 1057.60041); Advances in Soft Computing, 101--108 (2004; Zbl 1057.60003); J.~Math.\ Anal.\ Appl.\ 303, No.~2, 509--521 (2005; Zbl 1075.46002)] which investigates the theory of conditional expectations and discrete stochastic processes in vector lattices. In the present paper, a convergence theorem for sub- and super-martingales on a Dedekind complete Riesz space is proved. The convergence results resemble strongly the corresponding ones in the probability setting; in particular, under mild conditions, a Riesz space martingale is order convergent if and only if it is order bounded. This result enables the authors to prove a Riesz--Krickeberg decomposition for order convergent (sub)martingales turning the space of martingales itself into a Dedekind complete Riesz space. This space is then characterized as the order closure of the union of the ranges of the conditional expectations in the filtration.