Kerodon

$\Newextarrow{\xRightarrow}{5,5}{0x21D2}$ $\newcommand\empty{}$
$\Newextarrow{\xhookrightarrow}{10,10}{0x21AA}$

7.4.1 Limits of Spaces

Let $\operatorname{\mathcal{S}}$ denote the $\infty $-category of spaces (Construction 5.5.1.1). Our first goal in this section is to show that the $\infty $-category $\operatorname{\mathcal{S}}$ admits small limits (Corollary 7.4.1.2). We begin by establishing a recognition principle for limits in $\operatorname{\mathcal{S}}$.

Proposition 7.4.1.1. Let $\operatorname{\mathcal{C}}$ be a simplicial set, let $Y$ be a Kan complex, and let $\beta : \underline{Y}_{\operatorname{\mathcal{C}}} \rightarrow \mathscr {F}$ be a morphism in the $\infty $-category $\operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})$. The following conditions are equivalent:

$(1)$

The natural transformation $\beta $ exhibits $Y$ as a limit of the diagram $\mathscr {F}$ in the $\infty $-category $\operatorname{\mathcal{S}}$ (see Definition 7.1.0.1).

$(2)$

Composition with $\beta $ induces a homotopy equivalence

\[ \theta : \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y ) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{\Delta ^{0}}_{\operatorname{\mathcal{C}}}, \mathscr {F} ). \]

Proof. Assume that $(2)$ is satisfied; we will prove $(1)$ (the reverse implication is immediate from the definitions). Let $X$ be a small Kan complex; we wish to show composition with $\beta $ induces a homotopy equivalence $\theta _{X}: \operatorname{Hom}_{\operatorname{\mathcal{S}}}( X, Y ) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{X}_{\operatorname{\mathcal{C}}}, \mathscr {F} )$. Choose a simplicial set $K$ equipped with a weak homotopy equivalence $u: K \rightarrow \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, X )$ (for example, we can take $K = X$ and $u$ to be the homotopy equivalence of Remark 5.5.1.5). It follows from Example 7.1.2.10 that $u$ exhibits $X$ as a copower of $\Delta ^0$ by $K$ in the $\infty $-category $\operatorname{\mathcal{S}}$, so that the composite map

\[ \operatorname{Hom}_{\operatorname{\mathcal{C}}}(X, Y ) \times K \xrightarrow {\operatorname{id}\times u} \operatorname{Hom}_{\operatorname{\mathcal{S}}}(X,Y) \times \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0,X) \xrightarrow {\circ } \operatorname{Hom}_{\operatorname{\mathcal{C}}}( \Delta ^0, Y) \]

determines a homotopy equivalence $T: \operatorname{Hom}_{\operatorname{\mathcal{S}}}(X,Y) \rightarrow \operatorname{Fun}(K, \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y) )$. Similarly, the composite map

\[ K \xrightarrow {u} \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, X) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{ \Delta ^0}_{\operatorname{\mathcal{C}}}, \underline{X}_{\operatorname{\mathcal{C}}} ) \]

exhibits $\underline{X}_{\operatorname{\mathcal{C}}}$ as a copower of $\underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}$ by $K$ in the $\infty $-category $\operatorname{Fun}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})$ (see Proposition 7.1.7.3), so that $u$ also determines a homotopy equivalence $T': \operatorname{Hom}_{\operatorname{Fun}(\operatorname{\mathcal{C}},\operatorname{\mathcal{S}})}( \underline{X}_{\operatorname{\mathcal{C}}}, \mathscr {F} ) \rightarrow \operatorname{Fun}(K, \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} )$. Note that we have a homotopy commutative diagram of Kan complexes

\[ \xymatrix { \operatorname{Hom}_{\operatorname{\mathcal{S}}}(X, Y) \ar [d]^{T}_{\sim } \ar [r]^{\theta _{X}} & \operatorname{Hom}_{\operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{X}_{\operatorname{\mathcal{C}}}, \mathscr {F} ) \ar [d]^{T'}_{\sim } \\ \operatorname{Fun}(K, \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y) ) \ar [r] & \operatorname{Fun}(K, \operatorname{Hom}_{\operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{\Delta ^0}_{\operatorname{\mathcal{C}}}, \mathscr {F} ) } \]

where the bottom horizontal map is obtained by applying the functor $\operatorname{Fun}(K, \bullet )$ to $\theta $ and is therefore a homotopy equivalence by virtue of assumption $(2)$. It follows that $\theta _{X}$ is also a homotopy equivalence. $\square$

Recall that every Kan complex $Y$ is homotopy equivalent to the morphism space $\operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y)$ (Remark 5.5.1.5). Consequently, Proposition 7.4.1.1 guarantees that if a diagram $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ admits a limit, then the limit $\varprojlim (\mathscr {F} )$ is homotopy equivalent to the morphism space $M = \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{\Delta ^{0}}_{\operatorname{\mathcal{C}}}, \mathscr {F} )$. We now show that, modulo set-theoretic technicalities, this condition is always satisfied:

Corollary 7.4.1.2. Let $\operatorname{\mathcal{C}}$ be a simplicial set, let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be a diagram, and suppose that the morphism space $M = \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} )$ is essentially small. Then $M$ is a limit of $\mathscr {F}$ in the $\infty $-category $\operatorname{\mathcal{S}}$.

Proof. Let $Y$ be a Kan complex and let $u: Y \rightarrow \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y)$ be the homotopy equivalence of Remark 5.5.1.5. As in the proof of Proposition 7.4.1.1, we observe that $u$ exhibits the constant functor $\underline{Y}_{\operatorname{\mathcal{C}}}$ as a copower of $\underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}$ by $Y$, and therefore induces a homotopy equivalence

\[ T: \operatorname{Hom}_{ \operatorname{Fun}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{Y}_{\operatorname{\mathcal{C}}}, \mathscr {F} ) \rightarrow \operatorname{Fun}(Y, \operatorname{Hom}_{\operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{\Delta ^0}_{\operatorname{\mathcal{C}}}, \mathscr {F} ) ) = \operatorname{Fun}(Y, M). \]

Setting $Y = M$, it follows that there exists a natural transformation $\beta : \underline{M}_{\operatorname{\mathcal{C}}} \rightarrow \mathscr {F}$ such that $T( \beta )$ is homotopic to the identity morphism $\operatorname{id}_{M}$. In particular, $T( \beta )$ is a homotopy equivalence, so that $\beta $ exhibits $M$ as a limit of $\mathscr {F}$ (Proposition 7.4.1.1). $\square$

Corollary 7.4.1.3. The $\infty $-category $\operatorname{\mathcal{S}}$ admits small limits.

Variant 7.4.1.4. Let $\kappa $ be an uncountable regular cardinal, let $\operatorname{\mathcal{S}}^{< \kappa }$ denote the $\infty $-category of essentially $\kappa $-small spaces, and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}^{< \kappa }$ be a diagram indexed by a simplicial set $\operatorname{\mathcal{C}}$. It follows from the proof of Corollary 7.4.1.2 that $\mathscr {F}$ admits a limit in the $\infty $-category $\operatorname{\mathcal{S}}^{< \kappa }$ if and only if the morphism space $M = \operatorname{Hom}_{ \operatorname{Fun}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}^{< \kappa } ) }( \underline{\Delta ^0}_{\operatorname{\mathcal{C}}}, \mathscr {F} )$ is essentially $\kappa $-small. Moreover, if this condition is satisfied, then $M$ is a limit of $\mathscr {F}$. In particular, the limit of $\mathscr {F}$ is independent of $\kappa $ (that is, it is preserved by the inclusion functors $\operatorname{\mathcal{S}}^{< \kappa } \hookrightarrow \operatorname{\mathcal{S}}^{< \lambda }$ for $\lambda \geq \kappa $.

Remark 7.4.1.5 (Limits of Truncated Spaces). Let $n$ be an integer and let $\operatorname{\mathcal{S}}_{\leq n}$ denote the full subcategory of $\operatorname{\mathcal{S}}$ spanned by the $n$-truncated spaces (Definition 3.5.9.1). Then $\operatorname{\mathcal{S}}_{\leq n}$ is a reflective subcategory of $\operatorname{\mathcal{S}}$ (Example 6.2.2.7). Combining Corollary 7.4.1.3 with Variant 7.1.4.25, we conclude that the $\infty $-category $\operatorname{\mathcal{S}}_{\leq n}$ admits small limits, which are preserved by the inclusion functor $\operatorname{\mathcal{S}}_{\leq n} \hookrightarrow \operatorname{\mathcal{S}}$. In other words, if $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ is a diagram having the property that each of the Kan complexes $\mathscr {F}(C)$ is $n$-truncated, then the limit $\varprojlim (\mathscr {F}) \simeq \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} )$ is also $n$-truncated.

Our next goal is to show that, if $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ is given as the covariant transport representation of a left fibration $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$, then the limit $\varprojlim (\mathscr {F} )$ has a more explicit description: it can be identified with the space of sections of $U$.

Proposition 7.4.1.6. Let $\operatorname{\mathcal{C}}$ be a small simplicial set and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be the covariant transport representation of a left fibration $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ (Definition 5.6.5.1). Then the simplicial set $\operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}})$ is a Kan complex, which is a limit of the diagram $\mathscr {F}$.

Corollary 7.4.1.7. Let $\operatorname{\mathcal{C}}$ be a small simplicial set. Then any diagram $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ admits a limit in the $\infty $-category $\operatorname{\mathcal{S}}$, given by the $\infty $-category $\operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \int _{\operatorname{\mathcal{C}}} \mathscr {F} )$.

Proof. Apply Proposition 7.4.1.6 to the left fibration $U: \int _{\operatorname{\mathcal{C}}} \mathscr {F} \rightarrow \operatorname{\mathcal{C}}$ of Example 5.6.2.9. $\square$

To prove Proposition 7.4.3.1, it will be convenient to reformulate the notion of covariant transport representation.

Definition 7.4.1.8. Let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be a left fibration of simplicial sets, let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be a diagram, and let $\alpha : \underline{ \Delta ^{0} }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$ be a natural transformation. For each vertex $C \in \operatorname{\mathcal{C}}$, the restriction of $\alpha $ to the fiber $\operatorname{\mathcal{E}}_{C} = \{ C\} \times _{\operatorname{\mathcal{C}}} \operatorname{\mathcal{E}}$ determines a natural transformation from the constant functor taking the value $\Delta ^0$ to the constant functor taking the value $\mathscr {F}(C)$, which we can identify with a morphism of Kan complexes

\[ \alpha _{C}: \operatorname{\mathcal{E}}_{C} \rightarrow \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, \mathscr {F}(C) ). \]

We say that $\alpha $ exhibits $\mathscr {F}$ as a covariant transport representation for $U$ if, for every vertex $C \in \operatorname{\mathcal{C}}$, the morphism $\alpha _{C}$ is a homotopy equivalence.

The terminology of Definition 7.4.1.8 is justified by the following:

Proposition 7.4.1.9. Let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be a left fibration of simplicial sets and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be a diagram. The following conditions are equivalent:

$(1)$

The diagram $\mathscr {F}$ is a covariant transport representation for $U$, in the sense of Definition 5.6.5.1.

$(2)$

There exists a natural transformation $\alpha : \underline{ \Delta ^{0} }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$ which exhibits $\mathscr {F}$ as a covariant transport representation for $U$, in the sense of Definition 7.4.1.8.

$(3)$

There exists an equivalence of $U$ with the left fibration $\{ \Delta ^0 \} \operatorname{\vec{\times }}_{\operatorname{\mathcal{S}}} \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{C}}$.

Proof. Unwinding the definitions, we see that commutative diagrams

\[ \xymatrix { \operatorname{\mathcal{E}}\ar [dr]^{U} \ar [rr]^{T} & & \{ \Delta ^0 \} \operatorname{\vec{\times }}_{\operatorname{\mathcal{S}}} \operatorname{\mathcal{C}}\\ & \operatorname{\mathcal{C}}& } \]

can be identified with morphisms $\alpha : \underline{ \Delta ^{0} }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$ in the $\infty $-category $\operatorname{Fun}(\operatorname{\mathcal{E}}, \operatorname{\mathcal{S}})$. Under this identification, $T$ is an equivalence of left fibrations over $\operatorname{\mathcal{C}}$ if and only if $\alpha $ exhibits $\mathscr {F}$ as a covariant transport representation for $U$ (Corollary 5.1.7.16). This proves the equivalence $(2) \Leftrightarrow (3)$. The equivalence $(1) \Leftrightarrow (3)$ follows from the observation that the coslice diagonal

\[ \operatorname{\mathcal{S}}_{ \ast } = \operatorname{\mathcal{S}}_{ \Delta ^0 / } \hookrightarrow \{ \Delta ^0 \} \operatorname{\vec{\times }}_{\operatorname{\mathcal{S}}} \operatorname{\mathcal{S}} \]

is an equivalence of left fibrations over $\operatorname{\mathcal{S}}$ (Corollary 4.6.4.18). $\square$

Example 7.4.1.10. Let $\operatorname{\mathcal{C}}$ be a simplicial set and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be a diagram. Then the coslice diagonal

\[ \int _{\operatorname{\mathcal{C}}} \mathscr {F} = \operatorname{\mathcal{C}}\times _{ \operatorname{\mathcal{S}}} \operatorname{\mathcal{S}}_{\Delta ^0 / } \hookrightarrow \operatorname{\mathcal{C}}\times _{\operatorname{\mathcal{S}}} ( \{ \Delta ^0 \} \operatorname{\vec{\times }}_{ \operatorname{\mathcal{S}}} \operatorname{\mathcal{S}}) = \{ \Delta ^0 \} \operatorname{\vec{\times }}_{\operatorname{\mathcal{S}}} \operatorname{\mathcal{C}} \]

can be identified with a natural transformation

\[ \alpha : \underline{ \Delta ^0 }_{\int _{\operatorname{\mathcal{C}}} \mathscr {F} } \rightarrow \mathscr {F}|_{ \int _{\operatorname{\mathcal{C}}} \mathscr {F} } \]

which exhibits $\mathscr {F}$ as a covariant transport representation for the left fibration $U: \int _{\operatorname{\mathcal{C}}} \mathscr {F} \rightarrow \operatorname{\mathcal{C}}$.

Example 7.4.1.11. Let $\operatorname{\mathcal{C}}= \operatorname{N}_{\bullet }(\operatorname{\mathcal{C}}_0)$ be the nerve of a category $\operatorname{\mathcal{C}}_0$, let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be the (homotopy coherent) nerve of a functor $\mathscr {F}_0: \operatorname{\mathcal{C}}_0 \rightarrow \operatorname{Kan}$, and let $\operatorname{\mathcal{E}}= \operatorname{N}_{\bullet }^{\mathscr {F}_0}(\operatorname{\mathcal{C}}_0)$ denote the weighted nerve of Definition 5.3.3.1, so that the projection map $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ is a left fibration of $\infty $-categories (Corollary 5.3.3.19). Restricting the natural transformation $\alpha $ of Example 7.4.1.10 along the comparison map

\[ \operatorname{\mathcal{E}}= \operatorname{N}_{\bullet }^{\mathscr {F}_0}(\operatorname{\mathcal{C}}_0) \rightarrow \int _{\operatorname{\mathcal{C}}} \mathscr {F} \]

of Construction 5.6.4.1, we obtain a natural transformation $\alpha ': \underline{ \Delta ^0 }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$ which exhibits $\mathscr {F}$ as a covariant transport representation for $U$ (see Proposition 5.6.4.8). Beware that, although the functors $\underline{ \Delta ^0 }_{\operatorname{\mathcal{E}}}$ and $\mathscr {F}|_{\operatorname{\mathcal{E}}}$ can be obtained from strictly commutative diagrams $\mathrm{h} \mathit{\operatorname{\mathcal{E}}} \rightarrow \operatorname{Kan}$, the natural transformation $\alpha $ generally cannot be obtained from a natural transformation of $\operatorname{Kan}$-valued functors (even up to homotopy).

Example 7.4.1.12. Let $\operatorname{\mathcal{C}}= \operatorname{N}_{\bullet }(\operatorname{\mathcal{C}}_0)$ be the nerve of a category $\operatorname{\mathcal{C}}_0$, let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be a left fibration of $\infty $-categories, and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ denote the (homotopy coherent) nerve of the strict transport representation $\mathscr {F}_0 = \operatorname{sTr}_{\operatorname{\mathcal{E}}/ \operatorname{\mathcal{C}}_0}$ (see Construction 5.3.1.5). Set $\operatorname{\mathcal{E}}' = \operatorname{N}_{\bullet }^{\mathscr {F}_0}(\operatorname{\mathcal{C}}_0)$, so that Example 7.4.1.11 supplies a natural transformation $\alpha ': \underline{ \Delta ^0 }_{\operatorname{\mathcal{E}}'} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}'}$ which exhibits $\mathscr {F}$ as a covariant transport representation of $U'$. The scaffolds of Constructions 5.3.4.7 and 5.3.4.11 provide categorical equivalences of simplicial sets

\[ \operatorname{\mathcal{E}}\xleftarrow { \lambda _{u} } \underset { \longrightarrow }{\mathrm{holim}}( \mathscr {F}_0 ) \xrightarrow { \lambda _{t} } \operatorname{\mathcal{E}}'. \]

It follows that there exists a natural transformation $\alpha : \underline{\Delta ^0}_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$ which is characterized up to homotopy by the requirement that $[\alpha ]$ and $[\alpha ']$ have the same image in the homotopy category $\mathrm{h} \mathit{ \operatorname{Fun}( \underset { \longrightarrow }{\mathrm{holim}}(\mathscr {F}_0), \operatorname{\mathcal{S}})}$. Moreover, $\alpha $ exhibits $\mathscr {F}$ as a covariant transport representation for $U$. Beware that, although the homotopy class $[\alpha ]$ is canonically determined, the construction of $\alpha $ requires some auxiliary choices (which cannot be made functorially at the level of simplicial sets).

Construction 7.4.1.13. Let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be a morphism of simplicial sets, let $\operatorname{Fun}_{/ \operatorname{\mathcal{C}}}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{E}})$ be the simplicial set parametrizing sections of $U$, and let

\[ \operatorname{ev}: \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \times \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{E}}\quad \quad (F, C) \mapsto F(C) \]

be the evaluation functor. Suppose we are given a diagram $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ and a natural transformation $\alpha : \underline{ \Delta ^{0} }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{E}}}$. The image of $\alpha $ under the functor

\[ \operatorname{Fun}( \operatorname{\mathcal{E}}, \operatorname{\mathcal{S}}) \xrightarrow { \circ \operatorname{ev}} \operatorname{Fun}( \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \times \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) \]

can be identified with a comparison morphism

\[ T: \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} ). \]

We will deduce Proposition 7.4.1.6 from the following:

Proposition 7.4.1.14. Let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be a left fibration of simplicial sets, let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be a diagram, and let $\alpha : \underline{ \Delta ^0 }_{\operatorname{\mathcal{E}}} \rightarrow \mathscr {F}|_{\operatorname{\mathcal{C}}}$ be a natural transformation. If $\alpha $ exhibits $\mathscr {F}$ as a covariant transport representation of $U$, then the comparison morphism

\[ T: \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} ) \]

of Construction 7.4.1.13 is a homotopy equivalence of Kan complexes.

Proof. For every morphism of simplicial sets $S \rightarrow \operatorname{\mathcal{C}}$, we can apply Construction 7.4.1.13 to the left fibration $S \times _{\operatorname{\mathcal{C}}} \operatorname{\mathcal{E}}\rightarrow S$ to obtain a comparison map

\[ T_{S}: \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}(S, \operatorname{\mathcal{E}}) = \operatorname{Fun}_{/S}( S, S \times _{\operatorname{\mathcal{C}}} \operatorname{\mathcal{E}}) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(S, \operatorname{\mathcal{C}}) }( \underline{ \Delta ^0 }_{S}, \mathscr {F}|_{S} ). \]

Let us say that $S$ is good if the morphism $T_{S}$ is a homotopy equivalence. We now proceed in several steps.

$(a)$

Let $S = \{ C\} $ be a vertex of the simplicial set $\operatorname{\mathcal{C}}$. In this case, $T_{S}$ agrees with the comparison map $\alpha _{C}: \operatorname{\mathcal{E}}_{C} \rightarrow \operatorname{Hom}_{\operatorname{\mathcal{C}}}( \Delta ^0, \mathscr {F}(C) )$ appearing in Definition 7.4.1.8, which is a homotopy equivalence by virtue of our assumption on $\alpha $.

$(b)$

The construction of $T_{S}$ depends functorially on $S$: that is, if we are given a morphism $f: S' \rightarrow S$ in $(\operatorname{Set_{\Delta }})_{ / \operatorname{\mathcal{C}}}$, then the diagram of Kan complexes

7.47
\begin{equation} \begin{gathered}\label{equation:witness-vs-sections0} \xymatrix { \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}(S, \operatorname{\mathcal{E}}) \ar [r] \ar [d]^{T_ S} & \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}(S', \operatorname{\mathcal{E}}) \ar [d]^{ T_{S'} } \\ \operatorname{Hom}_{ \operatorname{Fun}(S,\operatorname{\mathcal{C}}) }( \underline{\Delta ^{0}}_{S}, \mathscr {F}|_{S} ) \ar [r] & \operatorname{Hom}_{ \operatorname{Fun}(S',\operatorname{\mathcal{C}}) }( \underline{\Delta ^{0}}_{S'}, \mathscr {F}|_{S'} ) } \end{gathered} \end{equation}

is commutative. For the upper horizontal map, this follows from Corollaries 4.2.5.2 and Corollary 4.4.3.8.

$(c)$

Let $S = \Delta ^ n$ be a standard simplex and let $S' = \{ 0\} $ be its initial vertex, so that the inclusion $S' \hookrightarrow S$ is left anodyne (Corollary 4.2.4.11). Then horizontal maps appearing in (7.47) are homotopy equivalences. For the upper horizontal map this follows from Proposition 4.2.5.4, and for the lower horizontal map it follows from Proposition 7.3.6.7. Combining this observation with $(a)$, we conclude that $S$ is good.

$(d)$

The construction $S \mapsto T_{S}$ carries coproducts in the category $(\operatorname{Set_{\Delta }})_{ / \operatorname{\mathcal{C}}}$ to products in the category $\operatorname{Fun}( [1], \operatorname{Kan})$. Consequently, the collection of good simplicial sets is closed under coproducts.

$(e)$

Suppose we are given a pushout square

7.48
\begin{equation} \begin{gathered}\label{equation:witness-vs-section} \xymatrix { S \ar [r] \ar [d] & S_0 \ar [d] \\ S_1 \ar [r] & S_{01} } \end{gathered} \end{equation}

in the category $(\operatorname{Set_{\Delta }})_{ / \operatorname{\mathcal{C}}}$. Then the induced diagram

7.49
\begin{equation} \begin{gathered}\label{equation:witness-vs-section2} \xymatrix { T_{S} & T_{ S_0 } \ar [l] \\ T_{ S_1 } \ar [u] & T_{S_{01}} \ar [l] \ar [u] } \end{gathered} \end{equation}

is a pullback square in the category $\operatorname{Fun}( [1], \operatorname{Kan})$. Moreover, if the horizontal maps in the diagram (7.48) are monomorphisms, then (7.49) is a (levelwise) homotopy pullback square: this follows from $(b)$ and Example 3.4.1.3. Consequently, if $S$, $S_{0}$, and $S_{1}$ are good, then $S$ is also good (see Corollary 3.4.1.12).

$(f)$

Let $S$ be a simplicial set of dimension $\leq k$, for some integer $k \geq -1$. To prove this, we proceed by induction on $k$. If $k = -1$, then $S = \emptyset $ and the statement is clear. To handle the general case, let $S' = \operatorname{sk}_{n-1}(S)$ denote the $(n-1)$-skeleton of $S$, so that Proposition 1.1.4.12 supplies a pushout square

\[ \xymatrix { \coprod \operatorname{\partial \Delta }^{k} \ar [r] \ar [d] & \coprod \Delta ^{k} \\ S' \ar [r] & S, } \]

where the horizontal maps are monomorphisms. It follows from our inductive hypothesis that $S'$ and $\coprod \operatorname{\partial \Delta }^{k}$ are good. By virtue of $(e)$, we are reduced to showing that the coproduct $\coprod \Delta ^ k$ is good, which follows from $(c)$ and $(d)$.

We now complete the proof by showing that every object $S \in (\operatorname{Set_{\Delta }})_{/\operatorname{\mathcal{C}}}$ is good. Note that $T_{S}$ can be realized as the limit of a tower

\[ \xymatrix { \cdots \ar [r] & \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}(\operatorname{sk}_{1}(S), \operatorname{\mathcal{E}}) \ar [r] \ar [d]^{ T_{ \operatorname{sk}_1(S) } } & \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}(\operatorname{sk}_{0}(S), \operatorname{\mathcal{E}}) \ar [d]^{ T_{ \operatorname{sk}_0(S) } } \\ \cdots \ar [r] & \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{sk}_1(S),\operatorname{\mathcal{C}}) }( \underline{\Delta ^{0}}_{\operatorname{sk}_1(S)}, \mathscr {F}|_{\operatorname{sk}_1(S)} ) \ar [r] & \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{sk}_0(S),\operatorname{\mathcal{C}}) }( \underline{\Delta ^{0}}_{\operatorname{sk}_0(S)}, \mathscr {F}|_{\operatorname{sk}_0(S)} ), } \]

where the horizontal maps are Kan fibrations. Consequently, to show that $T_{S}$ is a homotopy equivalence, it will suffice to show that $T_{ \operatorname{sk}_{k}(S) }$ is a homotopy equivalence for every integer $k \geq 0$ (see Example 4.5.6.18), which follows from $(f)$. $\square$

Proof of Proposition 7.4.1.6. Let $\operatorname{\mathcal{C}}$ be a small simplicial set and let $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}$ be the covariant transport representation of a left fibration $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$. Combining Propositions 7.4.1.9 and Proposition 7.4.1.14, we deduce that the Kan complex $\operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}})$ is homotopy equivalent to the morphism space $\operatorname{Hom}_{ \operatorname{Fun}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^{0} }_{\operatorname{\mathcal{C}}}, \mathscr {F} )$ and is therefore a limit of $\mathscr {F}$ in the $\infty $-category $\operatorname{\mathcal{S}}$ (Corollary 7.4.1.2). $\square$

Variant 7.4.1.15 (Size Estimates for Limits). Let $\lambda $ be an uncountable cardinal and let $\kappa = \mathrm{ecf}(\lambda )$ be its exponential cofinality. Suppose we are given a diagram $\mathscr {F}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}^{< \lambda }$, where $\operatorname{\mathcal{C}}$ is a $\kappa $-small simplicial set. Then $\mathscr {F}$ can be identified with the covariant transport representation of a left fibration $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ which is essentially $\lambda $-small. Applying Variant 4.7.9.11, we deduce that the Kan complex $\operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}})$ is also essentially $\lambda $-small, and can therefore be identified with a limit of $\mathscr {F}$ in the $\infty $-category $\operatorname{\mathcal{S}}^{< \lambda }$. In particular, the $\infty $-category $\operatorname{\mathcal{S}}^{< \lambda }$ admits $\kappa $-small limits.

For many applications, we will need a more precise version of Proposition 7.4.1.6, which characterizes limit diagrams in the $\infty $-category $\operatorname{\mathcal{S}}$.

Proposition 7.4.1.16. Suppose we are given a pullback diagram of simplicial sets

\[ \xymatrix@R =50pt@C=50pt{ \operatorname{\mathcal{E}}\ar [r] \ar [d]^{U} & \overline{\operatorname{\mathcal{E}}} \ar [d]^{ \overline{U} } \ar [d] \\ \operatorname{\mathcal{C}}\ar [r] & \operatorname{\mathcal{C}}^{\triangleleft }, } \]

where $U$ and $\overline{U}$ are essentially small left fibrations, and let $\overline{\mathscr {F}}: \operatorname{\mathcal{C}}^{\triangleleft } \rightarrow \operatorname{\mathcal{S}}$ be a covariant transport representation for $\overline{U}$. The following conditions are equivalent:

$(1)$

The morphism $\overline{\mathscr {F}}: \operatorname{\mathcal{C}}^{\triangleleft } \rightarrow \operatorname{\mathcal{S}}$ is a limit diagram.

$(2)$

The restriction map

\[ Q: \operatorname{Fun}_{ / \operatorname{\mathcal{C}}^{\triangleleft } }( \operatorname{\mathcal{C}}^{\triangleleft }, \overline{\operatorname{\mathcal{E}}} ) \rightarrow \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \]

is a homotopy equivalence of Kan complexes.

Remark 7.4.1.17. In the formulation of Proposition 7.4.1.16, the essential smallness assumption on $\overline{U}$ is not important. If $\overline{U}$ is essentially $\kappa $-small (for some uncountable cardinal $\kappa $ which is not necessary small), then it admits a covariant transport representation $\overline{\mathscr {F}}: \operatorname{\mathcal{C}}^{\triangleleft } \rightarrow \operatorname{\mathcal{S}}^{< \kappa }$, which is a limit diagram if and only if $Q$ is a homotopy equivalence.

Proof of Proposition 7.4.1.16. Using Proposition 7.4.1.9, we can choose a natural transformation $\overline{\alpha }: \underline{ \Delta ^0 }_{\overline{\operatorname{\mathcal{E}}}} \rightarrow \overline{\mathscr {F}}|_{ \overline{\operatorname{\mathcal{E}}} }$ which exhibits $\overline{\mathscr {F}}$ as a covariant transport representation for $\overline{U}$. Set $\mathscr {F} = \overline{\mathscr {F}}|_{\operatorname{\mathcal{C}}}$, so that we have a commutative diagram of Kan complexes

\[ \xymatrix { \operatorname{Fun}_{ / \operatorname{\mathcal{C}}^{\triangleleft } }( \operatorname{\mathcal{C}}^{\triangleleft }, \overline{\operatorname{\mathcal{E}}} ) \ar [r]^{Q} & \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{E}}) \ar [d] \\ \operatorname{Hom}_{\operatorname{Fun}( \operatorname{\mathcal{C}}^{\triangleleft }, \operatorname{\mathcal{S}})}( \underline{\Delta ^0}_{ \operatorname{\mathcal{C}}^{\triangleleft } }, \overline{\mathscr {F}} ) \ar [r] & \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{ \Delta ^0}_{ \operatorname{\mathcal{C}}}, \mathscr {F} ) } \]

where the vertical maps are given by Construction 7.4.1.13, and are therefore homotopy equivalences (Proposition 7.4.1.14). It follows that $(2)$ is equivalent the following:

$(2')$

The restriction map $Q': \operatorname{Hom}_{\operatorname{Fun}( \operatorname{\mathcal{C}}^{\triangleleft }, \operatorname{\mathcal{S}})}( \underline{\Delta ^0}_{ \operatorname{\mathcal{C}}^{\triangleleft } }, \overline{\mathscr {F}} ) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}(\operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{ \Delta ^0}_{ \operatorname{\mathcal{C}}}, \mathscr {F} )$ is a homotopy equivalence.

Choose a Kan complex $Y$ and a natural transformation $\overline{\beta }: \underline{Y}_{ \operatorname{\mathcal{C}}^{\triangleleft } } \rightarrow \overline{\mathscr {F}}$ which exhibits $Y$ as a limit of the diagram $\overline{\mathscr {F}}$ (so that $\beta $ induces a homotopy equivalence $Y \rightarrow \overline{\mathscr {F}}(v)$, where $v$ denotes the cone point of $\operatorname{\mathcal{C}}^{\triangleleft } \simeq \{ v\} \star \operatorname{\mathcal{C}}$). Then $\overline{\beta }$ restricts to a natural transformation $\beta : \underline{Y}_{\operatorname{\mathcal{C}}} \rightarrow \mathscr {F}$. We then have a commutative diagram of Kan complexes

\[ \xymatrix { & \operatorname{Hom}_{\operatorname{\mathcal{S}}}( \Delta ^0, Y) \ar [dl]^{\overline{\theta }} \ar [dr]^{\theta } & \\ \operatorname{Hom}_{\operatorname{Fun}( \operatorname{\mathcal{C}}^{\triangleleft }, \operatorname{\mathcal{S}}) }( \underline{ \Delta ^0}_{\operatorname{\mathcal{C}}^{\triangleleft }}, \overline{\mathscr {F}} ) \ar [rr]^{Q'} & & \operatorname{Hom}_{ \operatorname{Fun}( \operatorname{\mathcal{C}}, \operatorname{\mathcal{S}})}( \underline{ \Delta ^0 }_{\operatorname{\mathcal{C}}}, \mathscr {F} ), } \]

where $\theta $ and $\overline{\theta }$ are induced by composition with $\beta $ and $\overline{\beta }$, respectively. It follows from Proposition 7.4.1.1 that $\overline{\theta }$ is a homotopy equivalence. Moreover, $\overline{\mathscr {F}}$ is a limit diagram if and only if $\beta $ exhibits $Y$ as a limit of $\mathscr {F}$: that is, if and only if $\theta $ is also a homotopy equivalence. Using the commutativity of the diagram, we see that this is equivalent to condition $(2')$. $\square$

For later use, we record some consequences of Proposition 7.4.1.16.

Proposition 7.4.1.18. Let $\kappa $ be an uncountable cardinal and let $\operatorname{\mathcal{C}}$ be an $\infty $-category which is locally $\kappa $-small. Then a morphism of simplicial sets $f: K^{\triangleleft } \rightarrow \operatorname{\mathcal{C}}$ is a limit diagram if and only if, for every object $X \in \operatorname{\mathcal{C}}$, the composition

\[ K^{\triangleleft } \xrightarrow { F } \operatorname{\mathcal{C}}\xrightarrow {h^{X} } \operatorname{\mathcal{S}}^{< \kappa } \]

is a limit diagram; here $h^{X}$ denotes the functor corepresented by $X$ (Notation 5.6.6.14).

Proof. Applying Proposition 7.1.6.12, we see that $F$ is a limit diagram if and only if, for every object $X \in \operatorname{\mathcal{C}}$, the restriction map

\[ \theta _{X}: \operatorname{Hom}_{ \operatorname{Fun}( K^{\triangleleft }, \operatorname{\mathcal{C}}) }( \underline{X}, F) \rightarrow \operatorname{Hom}_{ \operatorname{Fun}( K, \operatorname{\mathcal{C}}) }( \underline{X}|_{K}, F|_{K} ) \]

is a homotopy equivalence of Kan complexes. Let $\operatorname{\mathcal{E}}$ denote the oriented fiber product $\{ X\} \operatorname{\vec{\times }}_{\operatorname{\mathcal{C}}} \operatorname{\mathcal{C}}$ and let $U: \operatorname{\mathcal{E}}\rightarrow \operatorname{\mathcal{C}}$ be given by projection onto the second factor. Note that $U$ is a left fibration (Proposition 4.6.4.11) and that $\theta _{X}$ can be identified with the restriction map

\[ \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( K^{\triangleleft }, \operatorname{\mathcal{E}}) \rightarrow \operatorname{Fun}_{ / \operatorname{\mathcal{C}}}( K, \operatorname{\mathcal{E}}). \]

The identity morphism $\operatorname{id}_{X}$ can be viewed as an initial object of $\operatorname{\mathcal{E}}$ satisfying $U( \operatorname{id}_{X} ) = X$ (Proposition 4.6.7.22), so the corepresentable functor $h^{X}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}^{<\kappa }$ is a covariant transport representation for $U$ (Proposition 5.6.6.21). Applying Proposition 7.4.1.16, we see that $\theta _{X}$ is a homotopy equivalence if and only if $h^{X} \circ F$ is a limit diagram in the $\infty $-category $\operatorname{\mathcal{S}}$. $\square$

Corollary 7.4.1.19. Let $\kappa $ be an uncountable cardinal and let $\operatorname{\mathcal{C}}$ be an $\infty $-category which is is locally $\kappa $-small. For every object $X \in \operatorname{\mathcal{C}}$, the functors

\[ h^{X}: \operatorname{\mathcal{C}}\rightarrow \operatorname{\mathcal{S}}^{\kappa } \quad \quad h_{X}: \operatorname{\mathcal{C}}^{\operatorname{op}} \rightarrow \operatorname{\mathcal{S}}^{\kappa } \]

preserve $K$-indexed limits, for every simplicial set $K$.