nLab cocategory

Redirected from "cocomposition".
Contents

Contents

Idea

A (small) cocategory is a category internal to Set opSet^{op}, the opposite category of Set.

More generally, if CC is finitely cocomplete, there is a notion of cocategory internal to CC, namely a comonad in the bicategory of cospans in CC.

If c 0c 1c 0c_0 \to c_1 \leftarrow c_0 is a cocategory object in CC, then by homming out of c c_\bullet, one obtains a limit-preserving functor CCatC \to Cat. Under reasonable conditions, the adjoint functor theorem conversely implies that all limit-preserving functors CCatC \to Cat are obtained in this way.

Definitions

Cocategory

Let AA be a category with pullbacks. A cocategory CC is a category object internal to category A opA^{op}, consisting of:

  • an object of objects C 0AC_0 \in A;

  • an object of morphisms C 1AC_1 \in A;

  • structure morphisms:

    • cosource and cotarget morphisms s,t:C 0C 1s,t: C_0 \to C_1;

    • counit ε:C 1C 0\varepsilon : C_1 \to C_0;

    • cocomposition (comultiplication) morphism Δ:C 1C 1 C 0C 1\Delta : C_1 \to C_1 \sqcup_{C_0} C_1;

  • such that the following properties are satisfied, expressing the usual category laws:

    • coassociativity law: (Δid)Δ=(idΔ)Δ:C 1C 1 C 0C 1 C 0C 1(\Delta \sqcup \mathrm{id}) \circ \Delta \;=\; (\mathrm{id} \sqcup \Delta) \circ \Delta \;:\; C_1 \longrightarrow C_1 \sqcup_{C_0} C_1 \sqcup_{C_0} C_1, where the two iterated pushouts are identified via the canonical associativity isomorphism;
    • counitality law: (sεid)Δ=id C 1=(idtε)Δ.(s \circ \varepsilon \sqcup \mathrm{id}) \circ \Delta = \mathrm{id}_{C_1} = (\mathrm{id} \sqcup t \circ \varepsilon) \circ \Delta.

Cocomposition

In a plain cocategory

Cocomposition (or comultiplication) is the operation that takes a morphism f:xzf\colon x \to z ​ in a cocategory CC and produces a pair of composable morphisms (f 1,f 2)(f_1, f_2) with

f 1:xyandf 2:yz f_{1} \colon x \to y \quad \text{and} \quad f_{2} \colon y \to z

represented collectively as a morphism Δ(f)C 1 C 0C 1\Delta(f) \in C_1 \sqcup_{C_0} C_1 called the cocomposite of ff.

Intuitively, a single arrow is “split” into two arrows, dually to the composition morphism c:C 1× C 0C 1C 1c: C_1 \times_{C_0} C_1 \to C_1 that “merges” two arrows into one.

In an enriched cocategory

In enriched category theory, for VV a monoidal category the cocomposition operation on a VV-enriched cocategory CC is, for each 3-tuple (x,y,z)(x,y,z) of objects of CC, a morphism

Δ x,y,z:C(x,z)C(x,y)C(y,z) \Delta_{x,y,z} : C(x,z) \to C(x,y) \otimes C(y,z)

in VV.

This reduces to the above definition in the case that V=V = Set. The cocomposition morphism Δ x,y,z\Delta_{x,y,z} sends any morphism to a pair of composable morphisms

Δ x,y,z:(xfz)((xf 1y),(yf 2z)). \Delta_{x,y,z} : (x \stackrel{f}{\to} z) \;\; \mapsto \;\; \bigl( (x \stackrel{f_1}{\to} y), (y \stackrel{f_2}{\to} z) \bigr).

Cocategorical semantics

In the context of formal logic, particularly in linear logic and type theories with costructural rules, cocomposition is the cocategorical semantics for the contraction rule, realized categorically as diagonal (comultiplicative) maps on objects or morphisms.

Examples

  • Many interval objects are cocategory objects. For example, the arrow category is a cocategory object in CatCat.

  • Any coalgebra object is a cocategory object. This includes corings, Hopf algebroids, cogroupoids?, etc.

  • In SetSet, every cocategory object is a coproduct of copies of the trivial cocategory object 1111 \to 1 \leftarrow 1 and the cocategory object 1211 \to 2 \leftarrow 1 where the two points are distinct. These represent the discrete category functor and the codiscrete category functors SetCatSet \to Cat, respectively.

References

Presentation slides on VV-cocategories and cocomposition:

Last revised on February 27, 2026 at 10:23:49. See the history of this page for a list of all contributions to it.