nLab final (infinity,1)-functor




The notion of final (,1)(\infty,1)-functor (also called a cofinal (,1)(\infty,1)-functor, but see Rem ) is the vertical categorification of the notion of final functor from category theory to (∞,1)-category-theory:

An (∞,1)-functor p:KKp \colon K' \to K is final precisely if precomposition with pp preserves (∞,1)-colimits: if pp is final then for F:KCF \colon K \to C any (∞,1)-functor we have

lim (KFC)lim (KpKFC), \lim_\to (K \stackrel{F}{\to} C) \simeq \lim_{\to} ( K' \stackrel{p}{\to} K \stackrel{F}{\to} C),

when either of these (∞,1)-colimits exist.


The terminology conventions surrounding the topic of final functors are a bit of a mess. Often

  1. “final map” is used to refer to \infty-functors such that restriction along them preserves \infty -colimiting cocones.

  2. “initial map” is used for functors for which restriction preserves \infty -limiting cones.

A mnemonic rule for this terminology is that final objects are picked out by final functors and likewise for initial objects.

But beware that this convention is not universal, for instance Lurie‘s work is a notable exception to this rule. Also beware of the following:


Beware that the term “cofinal (∞,1)-functor” can mean either a functor for which the precomposition functor preserves colimits (as in Lurie’s Higher Topos Theory) or limits (as in Cisinski’s Higher Categories and Homotopical Algebra).

Given that the two main sources for quasicategories assign opposite meanings to this term, it is best to avoid its usage altogether.

Further adding to the confusion is that some sources, like Borceux‘s Handbook of Categorical Algebra use the term “final functor” for a functor for which the precomposition functor preserves limits, in contrast to the majority of the literature. Such usage, fortunately, is marginal.

In more detail:

See also the warning in the article final functor (here).



(final morphism of simplicial set)

A morphism p:STp : S \to T of simplicial sets is final if for every right fibration XTX \to T the induced morphism of simplicial sets

sSet /T(T,X)sSet /T(S,X) sSet_{/T}(T,X) \to sSet_{/T}(S,X)

is a homotopy equivalence.

So in the overcategory sSet/TsSet/T a final morphism is an object such that morphisms out of it into any right fibration are the same as morphisms out of the terminal object into that right fibration.

{T X = T}{S X p T}. \left\{ \array{ T &&\to&& X \\ & {}_{\mathllap{=}}\searrow && \swarrow_{} \\ && T } \right\} \;\; \simeq \left\{ \array{ S &&\to&& X \\ & {}_{\mathllap{p}}\searrow && \swarrow_{} \\ && T } \right\} \,.

This definition is originally due to Andre Joyal. It appears as HTT, def

This is equivalent to the following definition, in terms of the model structure for right fibrations:


The morphism p:STp : S \to T is final precisely if the terminal morphism (p*)=(S T = T)(p \to *) = \left( \array{ S &&\to&& T \\ & {}_{\mathllap{}}\searrow && \swarrow_{=} \\ && T } \right) in the overcategory sSet TsSet_T is a weak equivalence in the model structure for right fibrations on sSet TsSet_T.


This is HTT, prop.


If TT is a Kan complex then p:STp : S \to T is final precisely if it is a weak equivalence in the standard model structure on simplicial sets.


This is HTT, cor.


Suppose F:CDF : C \to D is a functor between ordinary categories. If the induced morphism N(F):N(C)N(D)N(F) : N(C) \to N(D) is a final (∞,1)-functor, then FF is a final functor in the sense of ordinary category theory.

However, the converse need not be true. For example, while the inclusion of the parallel pair [1][0][1] \rightrightarrows [0] in Δ op\Delta^{op} is final in the sense of ordinary category theory, it is not a final (∞,1)-functor.



(preservation of undercategories and colimits)

A morphism p:KKp : K' \to K of simplicial sets is final precisely if for every quasicategory CC

  • and for every morphism F¯:K C\bar F : K^{\triangleright} \to C that exhibits a colimit co-cone in CC, also (K) pK F¯C(K')^\triangleright \stackrel{p}{\to} K^{\triangleright} \stackrel{\bar F}{\to} C is a colimit co-cone.

and equivalently precisely if


This is HTT, prop.

The following result is the (,1)(\infty,1)-categorical analog of what is known as Quillen’s Theorem A.


(recognition theorem for final (,1)(\infty,1)-functors)

A morphism p:KCp : K \to C of simplicial sets with CC a quasi-category is final precisely if for each object cCc \in C the comma-object c/pc/C× CKc/p \coloneqq c/C \times_C K is weakly contractible.

More explicitly, the comma object in question here is the pullback

c/p c/C K p C. \array{ c/p &\to& c/C \\ \downarrow && \downarrow \\ K &\stackrel{p}{\to}& C \mathrlap{\,.} }

where c/Cc/C is the under quasi-category under cc.


This is due to Andre Joyal. A proof appears as HTT, prop.

The following says that up to equivalence, the cofinal maps of simplicial sets are the right anodyne morphisms


A map of simplicial sets is cofinal precisely if it factors as a right anodyne map followed by a trivial fibration.

This is (Lurie, cor.




The inclusion *𝒞\ast \to \mathcal{C} of a terminal object is final.


By theorem the inclusion of the point is final precisely if for all c𝒞c \in \mathcal{C}, the (∞,1)-categorical hom-space 𝒞(c,*)\mathcal{C}(c,\ast) is contractible. This is the definition of *\ast being terminal.


A (weak) localization f:𝒞𝒟f: \mathcal{C} \to \mathcal{D} is both initial and final.

This appears, for example, as (Cisinski, 7.1.10).

Cofiber products in co-slice categories


Consider the inclusion of the walking span-category, into the result of adjoining an initial object tt:

(1){x b y}AAAA{ t x b y} \Big\{ \array{ x &\longleftarrow& b &\longrightarrow& y } \Big\} \;\; \overset{\phantom{AAAA}}{\hookrightarrow} \;\; \left\{ \array{ && t \\ & \swarrow & \downarrow & \searrow \\ x &\longleftarrow& b &\longrightarrow& y } \right\}

One readily sees that for each object on the right, its comma category over this inclusion has contractible nerve, whence Theorem implies that this inclusion is a final \infty-functor.

As an application of the finality of (1), observe that for 𝒞\mathcal{C} an (∞,1)-category and T𝒞T \in \mathcal{C} an object, (∞,1)-colimits in the under-(∞,1)-category

𝒞 T/U𝒞 \mathcal{C}^{T/} \overset{\;\;U\;\;}{\longrightarrow} \mathcal{C}

are given by the \infty-colimit in 𝒞\mathcal{C} itself of the given cone of the original diagram, with tip XX (by this Prop.): For

F:𝒞 T/ F \;\colon\; \mathcal{I} \longrightarrow \mathcal{C}^{T/}

a small diagram, we have

U(limF)lim(T/U(F)) U \big( \underset{\longrightarrow}{\lim}\, F \big) \;\simeq\; \underset{\longrightarrow}{\lim}\, \big( T/U(F) \big)

(when either \infty-colimit exists).

Now for \mathcal{I} the walking span diagram on the left of (1), this means that homotopy cofiber products in 𝒞 T/\mathcal{C}^{T/} are computed as \infty-colimits in 𝒞\mathcal{C} of diagrams of the shape on the right of (1). But since the inclusion in (1) is final, these are just homotopy cofiber products in 𝒞\mathcal{C}.

Explicitly: Given

T = T = T ϕ X ϕ B ϕ Y X f B g Y \array{ T &=& T &=& T \\ {}^{\mathllap{ \phi_X }} \big\downarrow && {}^{\mathllap{ \phi_B }} \big\downarrow && {}^{\mathllap{ \phi_Y }} \big\downarrow \\ X & \underset{ f }{\longleftarrow} & B & \underset{ g }{ \longrightarrow } & Y }

regarded as a span in 𝒞 T\mathcal{C}^T, hence with underlying objects

U((X,ϕ X))=X,U((B,ϕ B))=B,U((Y,ϕ Y))=Y, U\big( (X,\phi_X) \big) \;=\; X \,, \;\;\;\;\;\; U\big( (B,\phi_B) \big) \;=\; B \,, \;\;\;\;\;\; U\big( (Y,\phi_Y) \big) \;=\; Y \,,

we have:

U((X,ϕ X)(B,ϕ B)(Y,ϕ Y))XBY. U \Big( \; (X,\phi_X) \underset{ (B,\phi_B) }{\coprod} (Y,\phi_Y) \; \Big) \;\;\;\simeq\;\;\; X \underset{B}{\coprod} Y \,.

In particular, if (B,ϕ B)(T,id T)(B,\phi_B) \;\coloneqq\; (T,id_T) is the initial object in 𝒞 T/\mathcal{C}^{T/}, in which case the cofiber product is just the coproduct

(X,ϕ X)(Y,ϕ Y)=(X,ϕ X)(T,id T)(Y,ϕ Y) (X,\phi_X) \coprod (Y,\phi_Y) \;\;=\;\; (X,\phi_X) \underset{ (T,id_T) }{\coprod} (Y,\phi_Y)

we find that the coproduct in the co-slice category is the co-fiber product under the given tip object in the underlying category

U((X,ϕ X)(Y,ϕ Y))XTY. U \Big( \; (X,\phi_X) \coprod (Y,\phi_Y) \; \Big) \;\;\;\simeq\;\;\; X \underset{T}{\coprod} Y \,.

On categories of simplices


For KK \in sSet a simplicial set, write Δ /K\Delta_{/K} for its category of elements, called here the category of simplices of the simplicial set:

an object of Δ /K\Delta_{/K} is a morphism of simplicial sets of the form Δ nK\Delta^n \to K for some nn \in \mathbb{N} (hence an nn-simplex of KK) and a morphism is a commuting diagram

Δ n 1 Δ n 2 K. \array{ \Delta^{n_1}&&\to&& \Delta^{n_2} \\ & \searrow && \swarrow \\ && K } \,.

Moreover, write

Δ /K ndΔ /K \Delta_{/K}^{nd} \hookrightarrow \Delta_{/K}

for the non-full subcategory on the non-degenerate simplices.


When the simplicial set KK is non-singular, i.e. every face of a non-degenerate simplex is still non-degenerate, then the category Δ /K nd\Delta_{/K}^{nd} is a poset and it coincides with the barycentric subdivision of KK.


Suppose KK is a non-singular simplicial set. Then the inclusion

N(Δ /K nd)N(Δ /K) N(\Delta_{/K}^{nd}) \hookrightarrow N(\Delta_{/K})

is a cofinal morphism of quasi-categories.

This appears as (Lurie, variant


For every simplicial set KK there exists a cofinal map

N(Δ /K)K. N(\Delta_{/K}) \to K \,.

This is (Lurie, prop.


If the simplicial set KK is singular, it is not in general true that the inclusion N(Δ /K nd)N(Δ /K)N(\Delta_{/K}^{nd}) \hookrightarrow N(\Delta_{/K}) is final. For a counter-example, see (Hovey, 9).


For every simplicial set KK, evaluation at the initial vertex

N(Δ /K) opK. N(\Delta_{/K})^{op} \to K \,.

is both initial and final.

This appears as (Shah, 12.2) and also follows from the fact that this map is a weak localization (Cisinski, 7.3.15).


This can be used to establish a Bousfield-Kan formula for homotopy colimits; see (Shah, 12.3).


Last revised on August 19, 2022 at 17:45:35. See the history of this page for a list of all contributions to it.