An arithmetical view to first-order logic (Q636314): Difference between revisions

The authors discretize the work in [\textit{C. C. Chang} and \textit{H. J. Keisler}, Continuous model theory. Princeton, N.J.: Princeton University Press (1966; Zbl 0149.00402)]. They take, as a value space, any Hausdorff topological space \(\mathbb{B}\) equipped with continuous quantifiers (i.e. maps sending non-empty finite subsets of \(\mathbb{B}\) to \(\mathbb{B}\) and elements of \(\mathbb{B}\) to themselves.) An \(n\)-ary relation on a set \(M\) is, by definition, a map: \(M\to\mathbb{B}\) whose range is finite. Semantic notions, such as satisfaction, are defined as one might expect. But, by means of a new device, relational bound, the range of a formula is made to be finite. The authors give a number of examples of \(\mathbb{B}\), and develop basics of model theory. These include ultraproducts, axiomatizability, and types. Also, an interesting application of this new approach using relational bound, is given to graph coloring, showing that a graph can be finitely colored if all its countable subgraphs are. In the opposite direction, so to speak, it is shown that any theory in this setting is represented in a usual first-order theory with increased vocabulary and extra axioms. In the last section, the authors relate their work here to logical rules considered by Khwārazmi (a Persian mathematician of 1200 years ago, we are told). Also they mention other options such as using an integration quantifier and probability logic. The reviewer wonders how this work relates to cylindrical and/or polyadic logics.