English

Unbounded Order Convergence and Application to Martingales without Probability

Functional Analysis 2013-06-12 v1

Abstract

A net (xα)αΓ(x_\alpha)_{\alpha\in \Gamma} in a vector lattice XX is unbounded order convergent (uo-convergent) to xx if xαxyo0|x_\alpha-x| \wedge y \xrightarrow{o} 0 for each yX+y \in X_+, and is unbounded order Cauchy (uo-Cauchy) if the net (xαxα)Γ×Γ(x_\alpha-x_{\alpha'})_{\Gamma\times \Gamma} is uo-convergent to 0. In the first part of this article, we study uo-convergent and uo-Cauchy nets in Banach lattices and use them to characterize Banach lattices with the positive Schur property and KB-spaces. In the second part, we use the concept of uo-Cauchy sequences to extend Doob's submartingale convergence theorems to a measure-free setting. Our results imply, in particular, that every norm bounded submartingale in L1(Ω;F)L_1(\Omega;F) is almost surely uo-Cauchy in FF, where FF is an order continuous Banach lattice with a weak unit.

Keywords

Cite

@article{arxiv.1306.2563,
  title  = {Unbounded Order Convergence and Application to Martingales without Probability},
  author = {Niushan Gao and Foivos Xanthos},
  journal= {arXiv preprint arXiv:1306.2563},
  year   = {2013}
}
R2 v1 2026-06-22T00:32:07.303Z