Extreme points of Banach lattices related to conditional expectations (Q2573439)

Let \((X, {\mathcal F},\mu)\) be a complete probability space, let \({\mathcal B}\) be a sub-\(\sigma\)-algebra, and let \(\Phi\) be the probabilistic conditional expectation operator determined by \({\mathcal B}\). For \(E \in {\mathcal F}\) one associates \(E^\wedge \in {\mathcal B}\) through \(E^\wedge = \{x \in X : \Phi\chi_E(x)\neq 0\}\). \(E\) is said to be localizing if it is not a null set and \(E \cap B^\wedge = B\) for \(B \subset E\), \(B \in {\mathcal F}\). Let \({\mathcal A} = \{f \in L^\infty (X, {\mathcal F},\mu): \| \Phi(| f| )\| < \infty \}\) and \({\mathcal K} = \{f \in L^1(X, {\mathcal F},\mu): \| \Phi(| f| )\| < \infty \}\). In this interesting paper, the author describes extreme points of the unit ball of these spaces in terms of localizing sets. More precisely, \(f\) is an extreme point if and only if supp\((f)\) is a localizing set and \(\Phi(| f| ) = \chi_E\). This answers a question of \textit{J. Daughtry} and \textit{B. Weinstock} [Glasg. Math. J. 46, 29--36 (2004; Zbl 1077.46011)] in the affirmative.