Résume | A classical result of model theory asserts that a countably categorical theory is characterised, up to bi-interpretability, by the topological automorphism group of its unique countable model. Ben Yaacov has shown that the countable categoricity assumption can be dropped if we instead use topological groupoids.
In this talk, we present a framework to understand and generalise these results using the language of topos theory. We describe a condition for when two topological groupoids represent bi-interpretable theories that generalises quasi-homeomorphisms of spaces. |