nLab pro-object


Category theory

Limits and colimits



A pro-object of a category CC is a “formal cofiltered limit” of objects of CC.

The category of pro-objects of CC is written propro-CC. Such a category is sometimes called a pro-category, but notice that that is not the same thing as a pro-object in Cat.

“Pro” is short for “projective” (projective limit is an older term for limit). It is in contrast to “ind” in the dual notion of ind-object, standing for “inductive”, (and corresponding to inductive limit, the old term for colimit). In some (often older) sources, the term ‘projective system’ is used more or less synonymously for pro-object.

The definition of arrows in the category of pro-objects in 𝒞\mathcal{C} is perhaps more intuitive in the dual case of ind-objects (pro-objects in C opC^{op}), where it can be seen as stipulating that the objects of CC are finitely presentable in indind-CC.


As formal cofiltered limits

Let 𝒞\mathcal{C} be a small category. We write

(1)𝒞 opyFunc(𝒞,Sets) \mathcal{C}^{op} \xhookrightarrow{\;\;\; y \;\;\;} Func(\mathcal{C}, Sets)

for the Yoneda embedding of its opposite category, which we will often regard through its opposite functor

(2)𝒞y opFunc(𝒞,Sets) op. \mathcal{C} \xhookrightarrow{\;\;\; y^{op} \;\;\;} Func(\mathcal{C}, Sets)^{op} \,.

This is still a full subcategory-inclusion, now of 𝒞\mathcal{C} itself.


(category of pro-objects)

The category of pro-objects of 𝒞\mathcal{C}, according to Grothendieck 1960, Section 2, is the full subcategory of the functor category Func(𝒞,Sets) opFunc(\mathcal{C}, Sets)^{op} (2)

(3)Pro(𝒞)Func(𝒞,Sets) op Pro(\mathcal{C}) \xhookrightarrow{\;\;} Func(\mathcal{C}, Sets)^{op}

on those functors which are cofiltered limits of representable functors under the opposite Yoneda embedding (2), hence of the form

(4)limiy op(X i)lim(X 𝒞y opFunc(𝒞,Sets) op), \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{\lim} y^{op}(X_i) \;\; \coloneqq \;\; \underset{\longleftarrow}{\lim} \big( \mathcal{I} \xrightarrow{\;X_\bullet\;} \mathcal{C} \xrightarrow{\;y^{op}\;} Func(\mathcal{C}, Sets)^{op} \big) \,,

for \mathcal{I} a cofiltered category. These objects of Pro(𝒞)Pro(\mathcal{C}) are also called the pro-objects of 𝒞\mathcal{C}.


Since Func(𝒞,Sets) opFunc(\mathcal{C}, Sets)^{op} (2) is the free completion of 𝒞\mathcal{C}, we may think of Pro(𝒞)Pro(\mathcal{C}) (Def. ) as the free completion of 𝒞\mathcal{C} under cofiltered limits. See also at pro-representable functor.


Under the identification of the objects of a category with those of its opposite category, the pro-objects (4) are equivalently filtered colimits of presheaves:

limiy op(X i)limiy(X i op)lim( opX op𝒞 opyFunc(𝒞,Sets)), \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{\lim} y^{op}(X_i) \;\simeq\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} y(X^{op}_i) \;\; \coloneqq \;\; \underset{\longrightarrow}{\lim} \big( \mathcal{I}^{op} \xrightarrow{\;X^{op}_\bullet\;} \mathcal{C}^{op} \xrightarrow{\;y\;} Func(\mathcal{C}, Sets) \big) \,,

since the limit of a functor is the colimit of its opposite functor.


The category Pro(𝒞)Pro(\mathcal{C}) is the opposite category of that of ind-objects in the opposite catgegory of 𝒞\mathcal{C} (e.g. Kashiwara-Schapira 06, p. 138): Pro(𝒞)(Ind(𝒞 op)) op. Pro(\mathcal{C}) \simeq (Ind(\mathcal{C}^{op}))^{op} \,.


The pro-objects regarded as functors (4) are objectwise on c𝒞c \in \mathcal{C} given by

limiy op(X i):climi𝒞(X i,c). \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} y^{op}(X_i) \;\; \colon \;\; c \;\mapsto\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} \mathcal{C}(X_i, c) \,.


This is the composite of the following sequence of natural bijections:

(limiy op(X i))(c) (limiy(X i op))(c) limi((y(X i op))(c)) limiFunc(𝒞,Sets)(y(c),y(X i)) limi𝒞 op(c,X i) limi𝒞(X i,c), \begin{aligned} \Big( \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{\lim} y^{op}(X_i) \Big) (c) & \;\simeq\; \Big( \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} y(X^{op}_i) \Big) (c) \\ & \;\simeq\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} \Big( \big(y(X^{op}_i)\big) (c) \Big) \\ & \;\simeq\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} \; Func(\mathcal{C},Sets) \big( y(c), \, y(X_i) \big) \\ & \;\simeq\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} \mathcal{C}^{op}(c, X_i) \\ & \;\simeq\; \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{\lim} \mathcal{C}(X_i, c) \mathrlap{\,,} \end{aligned}



(hom-sets between pro-objects as limits of colimits of hom-sets in 𝒞\mathcal{C})
The hom-sets in Pro(𝒞)Pro(\mathcal{C}) (Def. ) are in natural bijection with limits of colimits of hom-sets in 𝒞\mathcal{C}, as follows:

Pro(𝒞)(limiy op(X i),limj𝒥y op(Y j),)limj𝒥(limi𝒞(X i,Y j)) Pro(\mathcal{C}) \Big( \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{lim} y^{op}(X_i), \, \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} y^{op}(Y_j), \Big) \;\simeq\; \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} \Big( \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{lim} \mathcal{C} \big( X_i, \, Y_j \big) \Big)


This is the composite of the following sequence of natural bijections:

Pro(𝒞)(limiy op(X i),limj𝒥y op(Y j),) Func(𝒞,Sets) op(limiy op(X i),limj𝒥y op(Y j),) Func(𝒞,Sets)(limj𝒥y(Y j op),limiy(X i op),) limj𝒥Func(𝒞,Sets)(y(Y j op),limiy(X i op),) limj𝒥((limiy(X i op))(y(Y j op))) limj𝒥(limi𝒞(X i,Y j)) \begin{aligned} Pro(\mathcal{C}) \Big( \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{lim} y^{op}(X_i), \, \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} y^{op}(Y_j), \Big) & \;\simeq\; Func(\mathcal{C},Sets)^{op} \Big( \underset{\underset{i \in \mathcal{I}}{\longleftarrow}}{lim} y^{op}(X_i), \, \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} y^{op}(Y_j), \Big) \\ & \;\simeq\; Func(\mathcal{C},Sets) \Big( \underset{\underset{j \in \mathcal{J}}{\longrightarrow}}{lim} y(Y^{op}_j), \, \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{lim} y(X^{op}_i), \Big) \\ & \;\simeq\; \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} \; Func(\mathcal{C},Sets) \Big( y(Y^{op}_j), \, \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{lim} y(X^{op}_i), \Big) \\ & \;\simeq\; \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} \Big( \big( \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{lim} y(X^{op}_i) \big) \big( y(Y^{op}_j) \big) \Big) \\ & \;\simeq\; \underset{\underset{j \in \mathcal{J}}{\longleftarrow}}{lim} \Big( \underset{\underset{i \in \mathcal{I}}{\longrightarrow}}{lim} \mathcal{C} \big( X_i, \, Y_j \big) \Big) \end{aligned}


Via spans


We can give an explicit description of the arrows of pro-𝒞\mathcal{C} as follows. First, for pro-objects F:𝒟𝒞F: \mathcal{D} \rightarrow \mathcal{C} and G:𝒞G: \mathcal{E} \rightarrow \mathcal{C} and for any object ee of \mathcal{E}, we introduce a relation \sim on arrows with target G(e)G(e) which identifies an arrow f:F(d)G(e)f: F(d) \rightarrow G(e) with an arrow f:F(d)G(e)f': F(d') \rightarrow G(e) for objects dd and dd' of 𝒟\mathcal{D} and an object ee of \mathcal{E}, if there is an object dd'' of 𝒟\mathcal{D}, an arrow g:ddg: d'' \rightarrow d of 𝒟\mathcal{D}, and an arrow g:ddg': d'' \rightarrow d' of 𝒟\mathcal{D}, such that fF(g)=fF(g)f \circ F(g) = f' \circ F(g').

This relation \sim is in fact an equivalence relation. Symmetry is obvious. Reflexivity is immediately demonstrated using the identity arrows of 𝒟\mathcal{D}. Transitivity would not hold for an arbitrary category, but follows from the assumption that 𝒟\mathcal{D} is cofiltered. Indeed, suppose that we have a zig-zag in 𝒟\mathcal{D} as follows.

The fact that 𝒟\mathcal{D} is cofiltered ensures that there is an object dd'' of 𝒟\mathcal{D} fitting into the following diagram.

Suppose that we have arrows f 0:F(d 0)G(e)f_{0} : F(d_{0}) \rightarrow G(e), f 1:F(d 1)G(e)f_{1}: F(d_{1}) \rightarrow G(e), and f 2:F(d 2)G(e)f_{2}: F(d_{2}) \rightarrow G(e) such that g 0g_{0} and g 1g_{1} exhibit that f 0f 1f_{0} \sim f_{1}, and such that g 2g_{2} and g 3g_{3} exhibit that f 1f 2f_{1} \sim f_{2}. Then

f 0F(g 0g 0) =f 0F(g 0)F(g 0) =f 1F(g 1)F(g 0) =f 1F(g 2)F(g 1) =f 2F(g 3)F(g 1) =f 2F(g 3g 1). \begin{aligned} f_{0} \circ F(g_{0} \circ g'_{0}) &= f_{0} \circ F(g_{0}) \circ F(g'_{0}) \\ &= f_{1} \circ F(g_{1}) \circ F(g'_{0}) \\ &= f_{1} \circ F(g_{2}) \circ F(g'_{1}) \\ &= f_{2} \circ F(g_{3}) \circ F(g'_{1}) \\ &= f_{2} \circ F(g_{3} \circ g'_{1}). \end{aligned}

This exhibits that f 0f 2f_{0} \sim f_{2}, as required.

With this equivalence relation \sim to hand, we can give our explicit description of the arrows of pro-𝒞\mathcal{C}: an arrow of pro-𝒞\mathcal{C} from a pro-object F:𝒟𝒞F: \mathcal{D} \rightarrow \mathcal{C} to a pro-object G:𝒞G: \mathcal{E} \rightarrow \mathcal{C} can be taken to be a set {f e:F(d e)G(e)}\left\{ f_{e} : F\left(d_{e}\right) \rightarrow G(e) \right\} of arrows of 𝒞\mathcal{C}, one for every object ee of \mathcal{E}, such that, for every arrow g:eeg: e \rightarrow e' of EE, G(g)f ef eG(g) \circ f_{e} \sim f_{e'}.

In other words: a set {f e:F(d e)G(e)}\left\{ f_{e} : F\left(d_{e}\right) \rightarrow G(e) \right\} of arrows of 𝒞\mathcal{C}, one for every object ee of \mathcal{E}, such that, for every arrow g:eeg: e \rightarrow e' of EE, there is an object dd of 𝒟\mathcal{D}, an arrow g e:dd eg_{e} : d \rightarrow d_{e} of 𝒟\mathcal{D}, and an arrow g e:dd eg_{e'}: d \rightarrow d_{e'} of 𝒟\mathcal{D} such that G(g)f eF(g e)=f eF(g e)G(g) \circ f_{e} \circ F(g_{e}) = \circ f_{e'} \circ F(g_{e'}).

Two such sets {f e} eOb()\left\{ f_{e} \right\}_{e \in Ob(\mathcal{E})} and {f e} eOb()\left\{ f'_{e} \right\}_{e \in Ob(\mathcal{E})} are equal, i.e. define the same arrow from FF to GG, if f ef ef_{e} \sim f'_{e} for every object ee of \mathcal{E}.


In some cases, pro-objects in a category 𝒞\mathcal{C} can be viewed as actual limits in a certain category. We prove here some results of this kind.


Let 𝒞\mathcal{C} be a category, and let 𝒜\mathcal{A} be a category with cofiltered limits. Suppose that we have a fully faithful functor i:𝒞𝒜i: \mathcal{C} \rightarrow \mathcal{A} which lands in cocompact objects. Then lim 𝒜(i):Pro(𝒞)𝒜lim_{\mathcal{A}} \circ (i \circ) : Pro(\mathcal{C}) \to \mathcal{A} is fully faithful, and hence defines an equivalence onto its image.


Let F:𝒟𝒞F:\mathcal{D} \to \mathcal{C} and G:𝒞G:\mathcal{E} \to \mathcal{C} be pro-objects. We then have a sequence of (natural) bijections:

Hom Pro(𝒞)(F,G)Hom_{Pro(\mathcal{C})}(F,G)

=lime:colimd:𝒟Hom 𝒞(Fd,Ge)= \underset{e:\mathcal{E}}{lim} \, \underset{d:\mathcal{D}}{colim} \, Hom_{\mathcal{C}}(F d, G e)

lime:colimd:𝒟Hom 𝒜(i(Fd),i(Ge))\cong \underset{e:\mathcal{E}}{lim} \, \underset{d:\mathcal{D}}{colim} \, Hom_{\mathcal{A}}(i (F d), i (G e))

lime:Hom 𝒜(limd:𝒟(iF)d,i(Ge))\cong \underset{e:\mathcal{E}}{lim} \, Hom_{\mathcal{A}}(\underset{d:\mathcal{D}}{lim} (i \circ F) d, i (G e))

Hom 𝒜(limd:𝒟(iF)d,lime:(iG)e)\cong Hom_{\mathcal{A}}(\underset{d:\mathcal{D}}{lim} (i \circ F) d, \underset{e:\mathcal{E}}{lim} (i \circ G) e)

=Hom 𝒜(lim(iF),lim(iG))= Hom_{\mathcal{A}}(lim (i \circ F), lim (i \circ G))

With these bijections being by definition of pro-object morphisms, fully faithfulness, cocompactness of i(Ge)i (G e), definition of a limit, and definition respectively.


Let 𝒞\mathcal{C} be the category Grp of groups, and let 𝒜\mathcal{A} be the category TopGrp\mathsf{Top-Grp} of topological groups. The fully faithful functor SetTop\mathsf{Set} \rightarrow \mathsf{Top} sending a set to the discrete topological space on this set gives rise to a fully faithful functor GrpTopGrp\mathsf{Grp} \rightarrow \mathsf{Top-Grp}. Then (as finite discrete spaces are cocompact) Proposition implies that the category pro-FinGrp\mathsf{FinGrp} of pro-objects in FinGrp\mathsf{FinGrp}, that is to say of profinite groups, is equivalent to the full sub-category of topological groups whose objects are obtained as a cofiltered limit of finite groups (viewed as topological groups via the discrete topology).


Though it is less well-known, one can in Example evidently replace Top\mathsf{Top} with any category 𝒜\mathcal{A} for which there is a fully faithful functor Set𝒜\mathsf{Set} \rightarrow \mathcal{A} which preserves finite products and lands in cocompact objects. See discrete object for one general setting in which finite product preserving functors exist.


Both Example and Remark generalise from Grp\mathsf{Grp} to any finite product theory, that is to say to the category of models of a finite product sketch. They generalise further to any finite limit theory, that is to say to the category of models of a finite limit sketch, if the functor Set𝒜\mathsf{Set} \rightarrow \mathcal{A} moreover preserves finite limits.



Etale homotopy theory.

Procategories were used by Artin and Mazur in their work on étale homotopy theory. They associated to a scheme a “pro-homotopy type”. (This is discussed briefly at étale homotopy.) The important thing to note is that this was a pro-object in the homotopy category of simplicial sets, i.e., in the pro-homotopy category. Friedlander rigidified their construction to get an object in the pro-category of simplicial sets, and this opened the door to use of ‘homotopy pro-categories’.

Shape theory.

The form of shape theory developed by Sibe Mardešić and Jack Segal, at about the same time as the work in algebraic geometry, again used an approach equivalent to that of the pro-homotopy category. Under the name “Čech homotopy theory”, the use more formally of the pro-homotopy category, was introduced by Porter. Strong shape, developed by Edwards and Hastings, also by Tim Porter and then in further work by Sibe Mardešić and Jack Segal, used various forms of rigidification to get to the pro-category of spaces, or of simplicial sets. There methods of model category theory could be used.


  • Alexander Grothendieck, Section A.2 of: FGA Technique de descente et théorèmes d’existence en géométrie algébriques. II. Le théorème d’existence en théorie formelle des modules, Séminaire Bourbaki : années 1958/59 - 1959/60, exposés 169-204, Séminaire Bourbaki, no. 5 (1960), Exposé no. 195, 22 p. (numdam:SB_1958-1960__5__369_0, English translation: web version)

  • (SGA4-1) Alexander Grothendieck, Jean-Louis Verdier, Préfaisceaux, Exp. 1 (retyped pdf) in Théorie des topos et cohomologie étale des schémas. Tome 1: Théorie des topos, Séminaire de Géométrie Algébrique du Bois-Marie 1963–1964 (SGA 4). Dirigé par M. Artin, A. Grothendieck, et J. L. Verdier. Avec la collaboration de N. Bourbaki, P. Deligne et B. Saint-Donat. Lecture Notes in Mathematics 269, Springer 1972. pdf of SGA 4, Tome 1

  • Michael Artin, Barry Mazur, appendix of Étale homotopy theory, Lecture Notes in Maths. 100, Springer-Verlag, Berlin 1969.

  • Masaki Kashiwara, and Pierre Schapira, Section 6 of: Categories and Sheaves, Grundlehren der mathematischen Wissenschaften 332 (2006)

  • Peter Johnstone, Section VI.1 of: Stone Spaces

  • Dan Isaksen, Calculating limits and colimits in pro-categories, Fund. Math. 175 (2002), no. 2, 175–194.

  • Jacob Lurie, Section 6.1 in: Ultracategories (pdf)

  • Jean-Louis Verdier, Equivalence essentielle des systèmes projectifs, C. R.A.S. Paris261 (1965), 4950 - 4953.

  • John Duskin, Pro-objects (after Verdier), Sém. Heidelberg- Strasbourg1966 -67, Exposé 6, I.R.M.A.Strasbourg.

  • A. Deleanu, P. Hilton, Borsuk shape and Grothendieck categories of pro-objects, Math. Proc. Camb. Phil. Soc.79-3 (1976), 473-482 MR400220

  • Walter Tholen, Pro-categories and multiadjoint functors, Canadian J. Math. 36:1 (1984) 144-155 doi

See also discussion in shape theory:

Last revised on April 18, 2024 at 14:35:54. See the history of this page for a list of all contributions to it.