Enhanced 2-categories and limits for lax morphisms (Q655343): Difference between revisions

When meeting a new category, a first question is which limits and colimits exist. By ``which'' we mean here the specification of a diagram type, or weight. For ordinary categories, weights are set-valued functors. For categories with homs enriched in a monoidal base category \(\mathcal{V}\), weights are \(\mathcal{V}\)-valued functors. Categories of algebraic structures of a given kind \(T\) are very obliging, i.e., all (small) limits and colimits exist. Here, we are thinking of \(T\) as a monad with rank on the category \(\mathcal{S}et\) of (small) sets. The same remarks apply when we meet a new 2-category. Indeed, 2-categories are precisely categories with homs enriched in the Cartesian base \(\mathcal{V} = \mathcal{C}at\) of (small) categories. Limits and colimits with \(\mathcal{C}at\)-valued functors were studied by the reviewer [``Limits indexed by category-valued 2-functors'', J. Pure Appl. Algebra 8, 149--181 (1976; Zbl 0335.18005)]. This showed, in particular, that the lax and pseudo-limits studied by \textit{J. W. Gray} were naturally covered by the enriched category theory. When it comes to 2-categories of categories with 2-algebraic structure, all (small) limits and colimits exist as before. However, this is only true if we take the strict structure-preserving morphisms, but these are not the only ones of interest. There are also pseudo-, lax and oplax morphisms in the nomenclature of \textit{G. M. Kelly} and the reviewer [``Review of the elements of 2-categories'', Lect. Notes Math. 420, 75--103 (1974; Zbl 0334.18016)]. For example, the kinds of limits existing in the 2-category \(T\)-\(\text{Alg}_{\ell}\) of \(T\)-algebras and lax morphisms (for, say, a 2-monad \(T\) on \(\mathcal{C}at\)) were examined by the first author [``Limits for lax morphisms'', Appl. Categ. Struct. 13, No.~3, 189--203 (2005; Zbl 1099.18004)]. The story seems complicated because of restrictive conditions involving strict morphisms. The present paper demystifies this situation, for \(T\)-\(\text{Alg}_{\ell}\) in particular, by showing again that it is all naturally covered by enriched category theory using a very cleverly chosen base category \(\mathcal{F}\). Indeed, \(\mathcal{F}\) is the full subcategory of the arrow category \([\mathbf{2},\mathcal{C}at]\) with objects the fully faithful functors which are injective on objects (full embeddings). It is a Cartesian closed base which, when regarded as enriched in itself, the authors denote by \(\mathbb{F}\). Generally, an \(\mathcal{F}\)-category \(\mathbb{K}\) can be identified with a 2-functor \(J_{\mathbb{K}}: \mathcal{K}_{\tau} \to \mathcal{K}_{\lambda}\) which is bijective on objects and locally a full embedding. Morphisms in \(\mathcal{K}_{\tau}\) are called tight and those in \(\mathcal{K}_{\lambda}\), loose. For \(T\)-algebras, we obtain an example with strict morphisms as tight and lax as loose. Rather than \(\text{Cat}\)-valued 2-functors as weights, we now take \(\mathbb{F}\)-valued \(\mathcal{F}\)-functors \(\Phi : \mathbb{D} \to \mathbb{F}\). The story then unfolds very nicely as the theory of \(\mathcal{F}\)-categories. A major point of the paper is its characterization of those weights \(\Phi\) for which \(\Phi\)-weighted limits lift from \(\mathbb{K}\) to \(T\)-algebras in \(\mathbb{K}\).