Bilimits are bifinal objects (Q2154269)

The theory of limits and colimits lies at the center of category theory. The notions of \(2\)-limits and \(2\)-colimits were first introduced independently in [\textit{C. Auderset}, Cah. Topologie Géom. Différ. Catégoriques 15, 3--20 (1974; Zbl 0364.18007)], where the Eilenberg-Moore and the Kleisli category of a monad are recovered as a \(2\)-limit and \(2\)-colimit, and in [\textit{F. Borceux} and \textit{G. M. Kelly}, Bull. Aust. Math. Soc. 12, 49--72 (1975; Zbl 0329.18011)], where the notion of enriched limits and colimit were introduced. These notions were further investigated and developed by \textit{R. Street} [J. Pure Appl. Algebra 8, 149--181 (1976; Zbl 0335.18005)], \textit{G. M. Kelly} [Bull. Aust. Math. Soc. 39, No. 2, 301--317 (1989; Zbl 0657.18004); The basic concepts of enriched category theory. Seminarber. Fachber. Math., Fernuniv. Hagen 9 (1981; Zbl 0709.18501); Basic concepts of enriched category theory. Cambridge etc.: Cambridge University Press (1982; Zbl 0478.18005); Repr. Theory Appl. Categ. 2005, No. 10, 1--136 (2005; Zbl 1086.18001)] and \textit{S. Lack} [IMA Vol. Math. Appl. 152, 105--191 (2010; Zbl 1223.18003)], who also introduced and investigted the lax and weighted versions. Two papers by \textit{T. Clingman} and \textit{L. Moser} [``2-limits and 2-terminal objects are too different'', Preprint, \url{arXiv:2001.01313}; ``Bi-initial objects and bi-representations are not so different'', Preprint, \url{arXiv:2009.05545}] have investigated whether the well-known result that limits are terminal cones extends to the \(2\)-dimensional framework, the first establishing the negative answer while the second leveraging on results from double-category theory on representability of \(\mathcal{C}at\)-valued functors to show that being terminal still captures the notion of limit, provided one is willing to work with an alternative \(2\)-category than that of cones. This paper aims to clarify, with its main result (Theorem 4.10), that a natural characterization of lax bilimits in terms of cones is still possible. The synopsis of the paper goes as follows. The use of a marking on the domain of the \(2\)-functor of which the authors want to study the bilimits addresses in a fundamental way all the possible levels of laxity of the bilimit (pseudo, lax or anything in between). It is remarked that the further level of generality is necessary from a technical viewpoint to coherently interpolate laxity from pseudo to lax in all the \(2\)-categorical constructions. The authors aim also to fill the gap in the literature as concerns final objects in \(2\)-categorical theory, though some results on final \(2\)-functors can be seen in [\textit{F. Abellán García} and \textit{W. H. Stern}, J. Pure Appl. Algebra 226, No. 9, Article ID 107040, 43 p. (2022; Zbl 1495.18026)]. \begin{itemize} \item[\S 1] recalls the necessary background on \(2\)-categories and relevant constructions, namely joins, slices and the Grothendieck construction for fibrations of \(2\)-categories. \item[\S 2] introduces lax marked bilimits and contractions on the lines of [\textit{M. E. Descotte} et al., Adv. Math. 333, 266--313 (2018; Zbl 1401.18004)]. It is established (Proposition 2.2.11) that final objects can be characterized in several ways, one of which involves contractions. \item[\S 3] further studies the slice fibration \[ p:\mathcal{C}^{/F}\rightarrow\mathcal{C} \] to a marked \(2\)-category \(\overline{\mathcal{J}}\mathcal{=}\left( \mathcal{J},E\right) \) and a \(2\)-functor \(F:\mathcal{J}\rightarrow \mathcal{C}\). \S 3.1 is an investigation on representable fibrations and the properties of the corresponding representing objects. \S 3.2 focuses on the particular case with \(\mathcal{J}=\mathrm{D}_{0}\), so that \(\mathcal{C} ^{/F}=\mathcal{C}_{/F\left( \ast\right) }\) is a representable fibration, showing that in this case the object \(1_{F\left( \ast\right) }\) is the center of a contraction to the collection of \(p\)-cartesian edges. \S 3.3 constructs a modified \(2\)-category of cones which projects to \(\mathcal{C} ^{/F}\), showing that this projection is a biequivalence iff \(F\)\ admits an \(E\)-bilimits. \item[\S 4] blends all together, culminating in the main result (Theorem 4.10), which characterizes (lax, maraked) bilimits as limiting bifinal cones. It is established that such bilimits are also terminal in the appropriate subcategory of cones obtained by restricting to cartesian morphisms between them. \end{itemize}