New definitions in the theory of Type 1 computable topological spaces

Emmanuel Rauzy

2024-06-08

Abstract:Type 1 computability -- computability based on the use of numberings -- is one of several ways of developing systematically computable mathematics. This approach can be considered particularly important since it is the one that focuses on finite descriptions. The study of Type 1 computability of topological spaces can be understood as the study of topological spaces some of whose open sets admit finite descriptions. We define Type 1 computable topological spaces without making reference to a basis. Following work of Spreen, we show that the use of a formal inclusion relation should be systematized. We give the first general definition of the computable topology associated to a computable metric that does not rely on effective separability. This definition can be translated to other constructive settings, and its relevance goes beyond that of Type 1 computability. We also compare different notions of effective basis. The first one, introduced by Nogina, is based on an effective version of the statement ``a set $O$ is open if for any point in $O$, there is a basic set containing that point and contained in $O$''. The second one, associated to Lacombe, is based on an effective version of ``a set $O$ is open if it can be written as a union of basic open sets''. We show that neither of the notions of basis we thus obtain is completely satisfactory. We define a new notion of basis, based on an effective version of the Nogina statement, but using a formal inclusion relation. Finally, we obtain a new version of the theorem of Moschovakis which states that the Lacombe and Nogina approaches coincide on countable recursive Polish spaces, but which applies to sets equipped with non-onto numberings, and with effective separability as a sole hypothesis.

Logic