🔢 mathematics

Conceptual completeness for subgeometric logics

This paper establishes conceptual completeness for various subgeometric logics by characterizing it as a duality between theories and topoi, demonstrating their conservative embedding in full geometric logic, and recovering Makkai's reconstruction theorem under set-based model completeness assumptions.

Original authors: Ivan Di Liberti, Umberto Tarantino, Lingyuan Ye

Published 2026-07-03

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ivan Di Liberti, Umberto Tarantino, Lingyuan Ye

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand a complex machine, like a car engine, but you are only allowed to look at the people who drive it (the models) and how they interact with the car, rather than the blueprints (the syntax) or the engine itself.

This paper is about a specific kind of "reverse engineering" in the world of mathematics and logic. The authors, Ivan Di Liberti, Umberto Tarantino, and Lingyuan Ye, are asking a fundamental question: If we know everything about the "drivers" (the models) of a logical system, can we perfectly rebuild the "blueprints" (the theory) that created them?

In the past, mathematicians believed the answer was "yes," but only if you added some extra, complicated "topological" glue to the drivers to hold them together. This paper argues that for certain types of logic, you don't need that extra glue. The drivers themselves contain all the information needed to rebuild the blueprint, provided you look at them through the right mathematical lens.

Here is a breakdown of their ideas using simple analogies:

1. The Traditional View: The "Blueprint vs. The Drivers"

Think of a logical theory as a recipe (syntax) and the models as the cakes baked from that recipe (semantics).

Traditional Completeness: If you have a recipe, you can bake a cake. If two recipes make the same cake, they are essentially the same.
Conceptual Completeness (The Old Way): A famous mathematician named Makkai showed that if you have a collection of cakes and you know exactly how they are related, you can reconstruct the original recipe. However, to do this, you had to treat the collection of cakes like a city with a specific map (topology) and ultrafilters (a way of grouping cakes together). It was like saying, "To rebuild the recipe, you need to know the traffic patterns of the city where the cakes are sold."

2. The New View: The "Magic Mirror"

The authors propose a new way to look at this. Instead of treating the cakes as a city with traffic patterns, they treat them as a reflection in a magic mirror.

They introduce a framework where the "recipe" and the "collection of cakes" are two sides of the same coin.
They define a "Conceptually Complete" logic as one where the mirror is perfect. If you look at the reflection (the models), you see the original object (the theory) without any distortion or missing pieces.
The Big Shift: They show that for several important types of logic (like Coherent, Regular, and Disjunctive logic), this mirror is perfect without needing the extra "traffic pattern" glue. The models naturally contain the blueprint.

3. The "Four Easy Pieces" (The Proofs)

The paper proves that this "perfect mirror" works for four specific types of logical systems. They use a clever trick called the "Reduction Lemma."

The Analogy: Imagine you want to prove that a specific type of lock (logic) can be opened by a specific key (models). Instead of trying to pick the lock from scratch, they find a simpler, similar lock that they already know how to pick. They show that if the simpler lock works, the complex one must work too.
They successfully applied this to:
1. Coherent Logic: The logic of "and," "or," and "exists" (very common in math).
2. Regular Logic: A slightly simpler version focusing on "and" and "exists."
3. Essentially Algebraic Logic with Falsum: Logic that includes a "false" statement and specific algebraic rules.
4. Finitary Disjunctive Logic: Logic focused on "or" statements.

4. The "Conservative" Connection

The authors also discovered a side effect of this discovery. If a logic is "Conceptually Complete" (the mirror is perfect), it means that the logic is conservatively embedded in the larger world of geometric logic.

The Analogy: Think of a specialized language (like a dialect). If this dialect is "conceptually complete," it means that if you translate a sentence from the dialect into the main language, and then translate it back, you get the exact same sentence. Nothing is lost, and nothing new is accidentally added. The dialect is self-contained and robust.

5. The "Makkai" Connection

Finally, the authors address the elephant in the room: Makkai's original theorem.

They show that their new, "glue-free" definition is actually equivalent to Makkai's old, "glue-heavy" definition, but only if you assume the cakes (models) are "complete" in a specific way (having enough points).
The Takeaway: They didn't just reinvent the wheel; they showed that the wheel they built is the same shape as the old one, but they figured out how to make it roll without the heavy, unnecessary axle (the extra topological structure).

Summary

In short, this paper says: "We found a way to rebuild the blueprints of a logical system just by looking at the things it creates, without needing to add extra, complicated structures to those things."

They proved this works for several major types of logic, offering a cleaner, more direct way to understand the relationship between the rules of a system (syntax) and the things those rules describe (semantics). They also showed that this new perspective is actually the same as the famous old perspective, just viewed through a clearer lens.

Title: Conceptual Completeness for Subgeometric Logics
Authors: Ivan Di Liberti, Umberto Tarantino, and Lingyuan Ye

1. Problem and Motivation

The paper addresses the notion of conceptual completeness within the framework of subgeometric logics (fragments of geometric logic). Traditionally, conceptual completeness, as established by Makkai for coherent first-order logic [Mak87], is interpreted as a reconstruction theorem: the syntax of a theory can be recovered from its semantics (the category of models) provided the latter is endowed with additional structure (specifically, an "ultrastructure" derived from ultraproducts).

The authors identify a limitation in this traditional view: it relies heavily on set-based semantics and the imposition of extra structure (ultrastructure) on categories of models. Furthermore, previous approaches to proving conceptual completeness for various logical fragments often required independent, ad-hoc techniques for each fragment. The paper seeks to:

Reformulate conceptual completeness as a purely syntactic phenomenon, specifically as a duality between theories and their classifying topoi, rather than a reconstruction from set-based models.
Provide a modular framework to prove conceptual completeness for various subgeometric logics without relying on Makkai's original reconstruction theorem or imposing artificial structures on semantics.
Clarify the relationship between this new syntactic notion and Makkai's traditional semantic reconstruction, particularly under the assumption of completeness with respect to set-based models.

2. Methodology and Framework

The authors build upon the framework of right Kan injectivity in the 2-category of topoi ( $\mathbf{Topoi}$ ), as developed in their previous work [DLY25].

Logics as Classes of Morphisms: A logic $\mathcal{H}$ is defined as a class of geometric morphisms between topoi. A topos $\mathcal{X}$ "formally belongs" to $\mathcal{H}$ if it is (weakly) right Kan injective with respect to $\mathcal{H}$ -morphisms. This captures the idea that the generalised points of the topos admit the structure prescribed by the logic.
Syntactic Categories and Classifying Topoi:
- For a topos $\mathcal{X}$ formally in $\mathcal{H}$ , the syntactic category $\text{Syn}_{\mathcal{H}}(\mathcal{X})$ is defined as the category of geometric morphisms $\mathcal{X} \to \text{Set}[\mathcal{O}]$ (where $\text{Set}[\mathcal{O}]$ is the object classifier).
- Associated with $\mathcal{H}$ is a monad of syntax $T_{\mathcal{H}}$ on the category of small finitely complete categories ( $\mathbf{Lex}$ ).
- The classifying topos $\text{Cl}_{\mathcal{H}}(A)$ is constructed for a $T_{\mathcal{H}}$ -algebra $A$ .
Conceptual Completeness Definition: A bounded logic $\mathcal{H}$ is conceptually complete if the 2-functor $\text{Cl}_{\mathcal{H}}: \text{Alg}(T_{\mathcal{H}})^{\text{op}} \to \mathbf{WRInj}(\mathcal{H})$ is 2-fully faithful.
Key Insight: The authors show that conceptual completeness is equivalent to the adjunction $\text{Syn}_{\mathcal{H}} \dashv \text{Cl}_{\mathcal{H}}$ being a reflection (i.e., the counit is an equivalence). This means every $T_{\mathcal{H}}$ -algebra is essentially the syntactic category of its classifying topos. This formulation detaches the concept from set-based models and focuses on the relationship between syntax (algebras) and syntax-based semantics (classifying topoi).

3. Key Contributions and Results

A. Characterization of Conceptual Completeness

The paper establishes that a logic $\mathcal{H}$ is conceptually complete if and only if, for every small $T_{\mathcal{H}}$ -algebra $A$ , the embedding $A \hookrightarrow \text{Syn}_{\mathcal{H}}(\text{Cl}_{\mathcal{H}}(A))$ is an equivalence of categories (Corollary 2.2.3). This reframes the problem as determining whether the syntactic category of the classifying topos recovers the original algebra.

B. Proof of Conceptual Completeness for Four Logics

Using a Reduction Lemma (Proposition 3.1.2), the authors prove conceptual completeness for four specific logics by identifying them with known doctrines and showing that their syntactic categories recover the algebras. The proofs utilize specific sub-classes of morphisms (sublogics) to simplify the analysis:

Coherent Logic ( $\mathcal{H}_{\text{flat}}$ ): Proven conceptually complete. The authors show that the doctrine $T_{\mathcal{H}_{\text{flat}}}$ coincides with the free pretopos construction. They characterize the syntactic category of a coherent topos as the full subcategory of coherent objects, recovering the original pretopos.
Regular Logic ( $\mathcal{H}_{\text{matte}}$ ): Proven conceptually complete. The doctrine corresponds to the free (Barr-)exact category construction. The syntactic category is shown to consist of regular objects.
Essentially Algebraic Logic with Falsum ( $\mathcal{H}_{\text{dom}}$ ): Proven conceptually complete. The doctrine corresponds to the free lex category with a strict initial object. The proof involves analyzing dense embeddings and showing that objects in the syntactic category are retracts of representables.
Finitary Disjunctive Logic ( $\mathcal{H}_{\text{pure}}$ ): Proven conceptually complete. The doctrine corresponds to the free lextensive category construction. The proof utilizes "pure embeddings" and the join of a category with its Boolean algebra of decidable subterminal objects.

These results provide new characterizations for these logics that do not rely on Makkai's original theorem [Mak87].

C. Conservatively Embedded Logics

The paper introduces the notion of conservatively embedded logics (Section 4). A logic is conservatively embedded if the adjunction $\text{Syn}_{\mathcal{H}} \dashv \text{Cl}_{\mathcal{H}}$ is idempotent.

Result: Every conceptually complete logic is conservatively embedded.
Significance: This property has a proof-theoretic interpretation: the inclusion of the theory of the syntactic category into the theory of the classifying topos is conservative. This links the semantic notion of conceptual completeness directly to proof-theoretic conservativity.

D. Relation to Makkai's Theorem

In Section 5, the authors reconcile their syntactic definition with Makkai's traditional semantic reconstruction theorem for coherent logic.

Framework: They utilize the theory of virtual ultracategories (from [Saa25, Ham25, vGMT26]) to formalize semantic prescriptions.
Equivalence: They demonstrate that for coherent logic, their notion of conceptual completeness is equivalent to Makkai's result (Corollary 5.3.3).
Mechanism: The equivalence holds under the assumption that the classifying topoi have enough points (completeness with respect to set-based models). The authors decompose the result into:
1. A syntactic component: The logic is conceptually complete in their sense (duality between algebras and classifying topoi).
2. A semantic component: The equivalence between classifying topoi and ultracategories of set-based models (induced by the "enough points" assumption).
Generalization: They define "conceptual completeness à la Makkai" for general logics (Definition 5.4.1) and prove it is equivalent to their syntactic notion if the logic has enough set-based models (Theorem 5.4.2).

4. Significance and Claims

The paper claims to offer a new proof-theoretic light on conceptual completeness.

Modularity: It provides a unified framework (via Kan injectivity) to study conceptual completeness across different subgeometric logics, reducing the need for ad-hoc techniques.
Detachment from Set-Based Semantics: The primary definition of conceptual completeness is presented as a property of the logic itself (a duality between theories and classifying topoi), independent of set-based models. The connection to Makkai's theorem is shown to be a consequence of completeness with respect to set-based models, not the definition itself.
Recovery of Classical Results: The paper successfully recovers Makkai's original theorem for coherent logic as a special case where the syntactic duality aligns with the semantic reconstruction via ultracategories.
Conservativity: It establishes that conceptual completeness implies a strong conservativity property, linking model-theoretic properties to proof-theoretic ones.

The authors remain modest about the scope, noting that while they have proven conceptual completeness for several fragments, the question of whether all fragments of geometric logic are conceptually complete remains open, and the characterization of the column $\mathbf{WRInj}_{pw}(\mathcal{H})$ in their summary table is left for future research. They also note that their current examples sit "below" the coherent fragment, though they conjecture the theory applies more broadly.