Enhanced 2-categories and limits for lax morphisms

We study limits in 2-categories whose objects are categories with extra structure and whose morphisms are functors preserving the structure only up to a coherent comparison map, which may or may not be required to be invertible. This is done using the framework of 2-monads. In order to characterize the limits which exist in this context, we need to consider also the functors which do strictly preserve the extra structure. We show how such a 2-category of weak morphisms which is “enhanced”, by specifying which of these weak morphisms are actually strict, can be thought of as category enriched over a particular base cartesian closed category F. We give a complete characterization, in terms of F-enriched category theory, of the limits which exist in such 2-categories of categories with extra structure.

💡 Research Summary

The paper investigates limits in 2‑categories whose objects are categories equipped with additional algebraic structure, but whose morphisms preserve that structure only up to a coherent comparison map. Using the language of 2‑monads, the authors first recall the standard 2‑category Alg(T) of strict T‑algebras, strict T‑algebra morphisms, and algebra‑preserving natural transformations. They then introduce the lax‑morphism 2‑category Alg_lax(T), whose 1‑cells are lax morphisms equipped with a comparison 2‑cell α : f ∘ T ⇒ T ∘ f. Depending on whether α is invertible, one recovers pseudo‑morphisms or genuinely lax morphisms.

A central difficulty is that limits in Alg_lax(T) cannot be described solely in terms of the underlying T‑algebra limits, because the lax comparison cells interact with the universal property. To resolve this, the authors define an “enhanced” 2‑category Alg_enh(T). This structure consists of the same objects and lax morphisms as Alg_lax(T) together with a distinguished sub‑2‑category Alg_strict(T) that marks which lax morphisms are actually strict. In other words, each 1‑cell carries a tag indicating whether it should be treated as strict or lax.

The key insight is that Alg_enh(T) can be regarded as a category enriched over a particular cartesian closed base category F. The objects of F are two symbols, “strict” and “lax”, and its hom‑sets encode whether a comparison cell is required and whether it must be invertible. Consequently, Hom‑objects in Alg_enh(T) are F‑valued, and the whole structure becomes an F‑enriched category. This enrichment allows the authors to import the general theory of weighted limits for enriched categories.

The main theorem states that a diagram D : J → Alg_enh(T) admits a limit in Alg_enh(T) if and only if two conditions hold. First, the underlying diagram of strict T‑algebras (obtained by forgetting the lax tags) has a limit in the ordinary 2‑category Alg(T). Second, the lax comparison cells attached to the morphisms of D must be respected by the universal cone, i.e., they must form an F‑weighted cone. In concrete terms, the universal cone’s components must be strict wherever the diagram demands strictness, and lax wherever only a comparison map is required; moreover, any invertibility constraints on the comparison maps must be preserved. This characterisation precisely delineates which limits survive when one relaxes strict preservation to lax preservation.

To illustrate the theory, the paper works out three families of examples. (1) Monoidal categories: strict monoidal functors versus lax monoidal functors. The enhanced 2‑category captures both, and the theorem recovers the known fact that the tensor product of monoidal categories is a limit only when the participating functors are strict, while a lax tensor product exists under weaker hypotheses. (2) Algebraic structures such as groups or rings, where homomorphisms may be required to preserve operations up to a specified natural transformation. The enhanced framework yields new limits that were invisible in the strictly algebraic setting. (3) 2‑categories themselves, viewed as algebras for the “2‑category” 2‑monad. Here the comparison cells are 2‑cells, and the enrichment over F neatly encodes the distinction between strict 2‑functors and lax 2‑functors. In each case the authors verify that the enriched‑limit condition matches the intuitive categorical constructions and produces novel lax limits.

The paper concludes by suggesting several directions for future work. One is to replace the specific base F by more general cartesian closed categories, thereby obtaining a hierarchy of enhanced 2‑categories with finer gradations of strictness. Another is to develop a homotopical or cohomological theory of these enhanced categories, which could be relevant for higher‑dimensional rewriting systems. Finally, the authors point out potential applications in programming language semantics, where type constructors often preserve structure only up to a coherent effect, making the enhanced 2‑categorical perspective a natural fit for modelling such “weakly structured” systems.