A Completeness Theorem for Topological Doctrines
Silvio Ghilardi and Jérémie Marquès · 2025
Read A Completeness Theorem for Topological Doctrines by Silvio Ghilardi and Jérémie Marquès (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-20768.txt.
Introduction
Silvio Ghilardi and Jérémie Marquès prove a completeness theorem for a family of logical systems enriched with a sense of inside and outside. Their systems, called topological doctrines, extend logical categories with two operators modeled on a space's interior and closure. These operators let logic talk about nearness and boundary, not just truth.
The result is an embedding theorem: every doctrine can be faithfully placed inside products of the category of topological spaces. In plain terms, if the logic cannot prove something, a space-based model exists where it fails. The proof needs the axioms for closure plus two extra principles, one about products acting independently and one that contracts loops. The paper is aimed at specialists in logic and category theory. Its payoff is a bridge: reasoning about topological logic can be checked against real spaces, the ground the subject began from.
Catalog record
| CREATOR | Silvio Ghilardi and Jérémie Marquès |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 99 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-20768
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
A Completeness Theorem for Topological Doctrines Silvio Ghilardi and Jérémie Marquès Abstract . We extend logical categories with fiberwise interior and closure operators so as to obtain an embedding theorem into powers of the category of topological spaces. The required axioms, besides the Kuratowski closure axioms, are a “product independence” and a “loop contraction” principle. 1. Introduction In this paper, we apply the methods of categorical logic to design a classical first-order logic with additional connectives for the interior and closure operators, which is sound and complete with respect to interpretations in topological spaces. Whereas the propositional fragment of such a logic has been known for a long time [ MT44 ] , the extension to predicate logic has to face non trivial problems. The most notable problem is that interior and closure do not commute with inverse image. Instead, they only semi -commute: if f:X\to Y is a continuous map between topological spaces, and if \Diamond A represents the closure of A , then the inclusion \Diamond f^{-1}(A)\subseteq f^{-1}(\Diamond A) holds, but equality fails in general, already in the case where f is a diagonal function X\longrightarrow X\times X . This is especially problematic because inverse images represent substitutions from a logical point of view (in particular, inverse images along diagonals represent variable identifications in a formula). In order to appropriately handle the above difficulties in designing a symbolic calculus, we switched to an approach based on categorical logic; in fact, categorical logic offers conceptual tools that are particularly illuminating for the context we are considering. At first glance, however, a categorical approach seems to be problematic too, because the category \mathsf{Top} of topological spaces fails to have the standard structure required in order to interpret first order logic: it is not a regular category. This difficulty is handled by replacing the regular-epi/mono factorization system of regular categories by an arbitrary stable factorization system: in \mathsf{Top} such a system is given by the surjection/subspace factorizations (which are nothing but epi/regular-mono factorizations). The possibility of adopting stable factorization systems different from the standard regular-epi/mono factorization system traditionally used in categorical logic already appeared in [ GZ11 ] and is fully exploited in [ GM25 ] , where the notions of a modal category is introduced. Modal categories (the main framework of this paper) are based on f-Boolean categories: these are “Boolean categories relative to a stable factorization system,” namely lex categories endowed with a proper and stable orthogonal factorization system (\mathcal{E},\mathcal{M}) , whose lattices of \mathcal{M} -subobjects are Boolean algebras. They can be alternatively introduced as extensional Lawvere doctrines with full comprehension [ MR13 , Pas16 ] and they can be turned into Boolean categories just by adding some isomorphisms (in the same way as \mathsf{Set} can be obtained, up to equivalence, from \mathsf{Top} by turning bijective continuous functions into iso’s, see Section 5 below for more details). A modal category \mathsf{E} is now simply defined as an f-Boolean category whose \mathcal{M} -subobject lattices are modal algebras, namely Boolean algebras endowed with a further operator \Diamond commuting with finite joins. Since the \Diamond operator is meant to represent the closure operator and the arrows of \mathsf{E} are meant to represent continuous functions, taking inverse images (i.e., pullbacks) along them only semi -preserves \Diamond in a modal category. There are many examples of modal categories beside \mathsf{Top} : for instance, graphs and posets are modal categories. Graphs can be generalized to a simple notion of “counterpart structure” giving a complete semantics for modal categories [ GM25 ] . In the current paper, we will see which additional axioms should be added to the notion of a modal category to get completeness relatively to the topological semantics. In technical terms, we provide a complete axiomatization of the modal categories that embed conservatively in a power of the category of topological spaces. Our axioms include the S4 axiom S\leq\Diamond S=\Diamond\Diamond S , the product independence axiom ( PI ) as well as a loop contraction axiom ( LC ). The product independence axioms says (in its logical formulation) that \Diamond distributes over conjunctions of formulas not sharing common variables (\Diamond\varphi)[x]\land(\Diamond\psi)[y]\leftrightarrow\Diamond(\varphi[x]\land\psi[y])\penalty\ \penalty\ . The loop contraction axiom asserts a continuity condition for certain composable loops of partial maps. This last axiom is only necessary when certain function symbols are to be interpreted as subspace embeddings, because it is otherwise derivable from the rest of the axiomatic basis, see Remark 3.5 below (but notice that the formula ( LC ) plays nevertheless an important role in the completeness proof even when the language is restricted so as it becomes derivable). Related work Several authors in categorical logic have considered first-order S4 modal logic, for instance [ Rey91 , MR95 ] or [ AK08 ] . Modalities not commuting with substitutions were introduced in [ GM88 ] , where completeness is proved with respect to presheaves and relational presheaves (the completeness results in [ GM88 ] however do not apply to languages/doctrines with sorts representing subspaces). In [ Ghi90 , Cap. II, § 7] interpretations in topological spaces were considered too: the product independence axiom was introduced and completeness for purely relational languages was proved. These partial results remained unpublished (they were just announced in [ GM91 ] ); in the present paper, the topological construction from [ Ghi90 , Cap. II, § 7] is improved (in particular by introducing ‘lax’ features in it), so as it becomes powerful enough to prove the completeness result for full languages with the help of the completely novel loop contraction formulas. The only attempt to axiomatize first-order S4 modal logic in topological spaces we are aware of is in [ Sgr80 ] ; actually, [ Sgr80 ] allows the interior and closure operators to be applied with respect to a subset of the free variables of a formula, unlike in the present paper. The axiomatization obtained in [ Sgr80 ] looks very involved when compared with the neat axiomatization of the present paper (the relationship and the connections of the two axiomatizations are obscure and a priori not obvious). Overview of the paper In § 2 , we recall the notion of modal category from [ GM25 ] and some associated terminology. In § 3 , we define three axiom schemas and show that they are satisfied in the modal category \mathsf{Top} of topological spaces. Based on these axioms, we define topological modal categories in § 4 and we show a completeness result relatively to \mathsf{Top} . In § 5 , we give a more concrete meaning to this result by building the classifying topological modal category of a first-order modal theory. This construction can be done either directly or in two steps: from a theory to a suitable Lawvere doctrine and from a Lawvere doctrine to a suitable logical category. Since the first step is rather standard, we detail here only the second step (in [ GM25 , §7] we presented a direct construction, but it should be noted that it is not the syntactic category of a modal theory for the reasons explained in [ GM25 , §8.1] ). Notations In a topological space X , we denote by \macc@depth\char 1\relax\macc@set@skewchar\macc@nested@a 111{A} the closure of a subset