Relation algebras and groups

Por • 18 abr, 2018 • Sección: Crítica

Steven Givant

Abstract: Generalizing results of J’onsson and Tarski, Maddux introduced the notion of a pair-dense relation algebra and proved that every pair-dense relation algebra is representable. The notion of a pair below the identity element is readily definable within the equational framework of relation algebras. The notion of a triple, a quadruple, or more generally, an element of size (or measure) n>2 is not definable within this framework, and therefore it seems at first glance that Maddux’s theorem cannot be generalized. It turns out, however, that a very far-reaching generalization of Maddux’s result is possible if one is willing to go outside of the equational framework of relation algebras, and work instead within the framework of the first-order theory.  In the present paper, we define the notion of an atom below the identity element in a relation algebra having measure n for an arbitrary cardinal number n>0, and we define a relation algebra to be measurable if it’s identity element is the sum of atoms each of which has some (finite or infinite) measure. The main purpose of the present paper is to construct a large class of new examples of group relation algebras using systems of groups and corresponding systems of quotient isomorphisms (instead of the classic example of using a single group and forming its complex algebra), and to prove that each of these algebras is an example of a measurable set relation algebra. In a subsequent paper, the class of examples will be greatly expanded by adding a third ingredient to the mix, namely systems of “shifting” cosets. The expanded class of examples—called coset relation algebras—will be large enough to prove a representation theorem saying that every atomic, measurable relation algebra is essentially isomorphic to a coset relation algebra.

arXiv:1804.00076v1 [math.LO]

Logic (math.LO)

Post to Twitter

Escribe un comentario