Truncated Vector Lattices

In analysis, truncation is the operation of replacing a nonnegative real-valued function a (x) by its pointwise meet a (x) ∧ 1 with the constant 1 function. A vector lattice A is said to be closed under truncation if a ∧ 1 ∈ A for all a ∈ A. Note that A need not contain 1 itself. Truncation is fundamental to analysis. To give only one example, Lebesgue integration generalizes beautifully to any vector lattice of real-valued functions on a set X, provided the vector lattice is closed under truncation. But vector lattices lacking this property may have integrals which cannot be represented by any measure on X ([5]). Nevertheless, when the integral is formulated in a context broader than R , for example in pointfree analysis, the question of truncation inevitably arises. What is truncation, or more properly, what are its essential properties? In this talk we answer this question by providing the appropriate axiomatization, and then go on to present several representation theorems. The first is a direct generalization of the classical Yosida representation of an archimedean vector lattice with order unit. The second is a direct generalization of Madden’s pointfree representation of archimedean vector lattices. If time permits, we briefly discuss a third sheaf representation which has no direct antecedent in the literature. However, in all three representations the lack of a unit forces a crucial distinction from the corresponding unital representation theorem. The universal object in each case is some sort of family of continuous real-valued functions. The difference is that these functions must vanish at a specified point of the underlying space or locale or sheaf space. With that adjustment, the generalization from units to truncations goes remarkably smoothly. (Details are available from the author [1], [2].)