Sigma limits in 2-categories and flat pseudofunctors (Q1649333)

This article is concerned with the notion of flat functor within the context of \(2\)-categories. It should be emphasized that \(2\)-dimensional category theory is greatly different from the theory of Cat-enriched categories, which is essentially a part of \(1\)-dimensional category theory. As is usually the case, the \(Cat\)-enriched version of flatness is too strict. The principal objective in this paper is to solve the more difficult problem of the fundamental equivalence between flatness and appropriate notions of filteredness for \(2\)-categories and pro-representability of \(2\)-functors. Although the main result (Theorem 4.2.7) is stated for \(2\)-functors for the sake of simplicity, the corresponding theorem for pseudofunctors follows as a corollary (Theorem A.6 in the appendix). It was the discovery of the following unknown fact that segues into the authors' research on the \(2\)-dimensional concept of flatness: Given \(2\)-functors \[ \begin{aligned} P & :\mathcal{A\rightarrow}\mathrm{Cat}0\\ F & :\mathcal{A}^{\mathrm{op}}\rightarrow\mathrm{Cat}\end{aligned} \] with the \(2\)-Grothendieck construction \[ \diamond_{P}:\mathcal{E}l_{P}\,\rightarrow\mathcal{A} \] the category of pseudo-dicones for the \(2\)-functor \[ F\times P:\mathcal{A}^{op}\times\mathcal{A\,}\rightarrow\mathrm{Cat} \] is isomorphic to the category of lax cones for the \(2\)-functor \[ \mathcal{E}l_{P}^{op}\,\underrightarrow{\diamond_{P}^{\mathrm{op}}}\,\mathcal{A}^{\mathrm{op}}\,\underrightarrow{F}\,\mathrm{Cat} \] such that the structurel \(2\)-cell corresponding to the family \(\mathcal{C}_{P}\)\ of cocartesian morphisms are invertible. This fact inspired the authors to consider a general notion of \(\sigma\)-natural transformation already defined in [\textit{J. W. Gray}, Formal category theory: Adjointness for 2-categories. Springer, Cham (1974; Zbl 0285.18006)], which in turn led to the notion of weighted \(\sigma\)-limit as an essential tool in the paper. The most transcendental result of this paper (Theorem 2.4.10) is the representability of arbitrary weighted \(\sigma\)-limit (or \(\sigma\)-colimits) as conical ones. The main result (Theorem 4.2.7) claims that, given a \(2\)-functor \[ \mathcal{A}\,\underrightarrow{P}\,\mathrm{Cat} \] the following first three statements are equivalent, and that, if \(\mathcal{A}\)\ has finite weighted bilimits, the fourth statement is also equivalent: {\parindent=0.7cm\begin{itemize}\item[1.] \(\mathcal{E}l_{P}^{\mathrm{op}}\) is \(\sigma\)-cofiltered with respect to the family \(\mathcal{C}_{P}\)\ of cocartesian arrows. \item[2.] \(P\) is equivalent to a \(\sigma\)-filtered \(\sigma\)-colimit of representable \(2\)-functors in \(\mathcal{H}om_{P}\left( \mathcal{A},\,\mathrm{Cat}\right) \) \item[3.] \(P\) is flat. \item[4.] \(P\) ia left exact. \end{itemize}}