Functor Category¶

For fixed categories C and D, the category whose objects are functors C→D and whose morphisms are natural transformations, with identities and composition defined componentwise.

Version: v2 · 2026-09-06 · History
Domain-specific #: 1899
Origin domain: category theory
Subdomain: categorical constructions
Aliases: Category of functors

Core Idea¶

Fix categories \(\mathcal C\) and \(\mathcal D\). The functor category \([\mathcal C,\mathcal D]\), also written \(\mathcal D^{\mathcal C}\) or \(\operatorname{Fun}(\mathcal C,\mathcal D)\), has functors \(F:\mathcal C\to\mathcal D\) as its objects and natural transformations \(\eta:F\Rightarrow G\) as its morphisms. The endpoints are fixed: every object has the same source \(\mathcal C\) and target \(\mathcal D\).

A natural transformation assigns to each object \(c\in\mathcal C\) a morphism \(\eta_c:F(c)\to G(c)\) satisfying, for every \(u:c\to c'\),

\[ G(u)\circ\eta_c=\eta_{c'}\circ F(u). \]

Identity and vertical composition are defined componentwise:

\[ (1_F)_c=1_{F(c)},\qquad (\theta\circ\eta)_c=\theta_c\circ\eta_c. \]

Naturality ensures the composite is again a natural transformation; associativity and units follow componentwise from \(\mathcal D\). Riehl gives this construction and its size qualifications explicitly.^[1]

The functor category elevates whole diagrams to objects and coherent diagram maps to arrows. It is not merely a set of functors. Once the categorical structure is present, one can form isomorphisms, limits, colimits, adjunctions, and algebraic structures between diagrams as ordinary categorical constructions.^[2]

Structural Signature¶

Fixed domain category: one category \(\mathcal C\) indexes every diagram.
Fixed codomain category: one category \(\mathcal D\) supplies all values and component morphisms.
Functor objects: each object preserves identities and composition from \(\mathcal C\) into \(\mathcal D\).
Natural-transformation morphisms: arrows are coherent component families between parallel functors.
Naturality squares: every source morphism makes the component comparison commute.
Componentwise identity: the identity at \(F\) uses \(1_{F(c)}\) at every \(c\).
Componentwise composition: transformations compose vertically at each component.
Size control: smallness or universe assumptions ensure the relevant objects and hom-collections form the intended category.

Recognition test. Name \(\mathcal C\) and \(\mathcal D\), show that objects are all or a declared class of functors between them, and verify that arrows are natural transformations with vertical composition. If domains vary, arrows are arbitrary functions between functor sets, or naturality is absent, the construction is not this functor category.

What It Is Not¶

It is not a Functor. A functor is one object of \([\mathcal C,\mathcal D]\); the functor category contains many such objects and natural transformations between them. It is not a natural transformation, which is one morphism in the category.

It is not the category \(\mathbf{Cat}\) of categories and functors. In a functor category, source and target categories are parameters fixed outside the objects. It is not a comma category, whose objects combine arrows into or out of specified functors and whose morphisms satisfy a different compatibility triangle.

It is not automatically a category of all possible functors with varying endpoints. Such a collection belongs to a higher categorical organization. Nor is every chosen collection of functors a full functor category: restrictions such as additive, continuous, exact, enriched, or monoidal functors define subcategories only after their permitted natural transformations are specified.

Scope of Application¶

Functor categories organize diagram categories, presheaves, representations, actions, graded objects, chain-complex ingredients, and categorical models. A presheaf category is \([\mathcal C^{op},\mathbf{Set}]\). If a group \(G\) is treated as a one-object category, \([G,\mathbf{Set}]\) is the category of left \(G\)-sets and equivariant maps. Representations similarly arise as functors from a group or algebra-shaped category into vector spaces.

The arrow category \(\mathcal D^{[1]}\), where \([1]\) has one nonidentity arrow, has morphisms of \(\mathcal D\) as objects and commutative squares as arrows. A discrete indexing category with two objects gives \(\mathcal D\times\mathcal D\). Simplicial objects, cosimplicial objects, and presheaves are all instances with specialized index categories.

If \(\mathcal C\) is small and \(\mathcal D\) has specified limits or colimits, the corresponding constructions in \([\mathcal C,\mathcal D]\) are computed objectwise. Evaluation at each \(c\) preserves them.^[3] This pointwise theorem is a major reason the abstraction is useful.

Clarity¶

Notation varies. \(\mathcal D^{\mathcal C}\) mirrors exponential notation, while \([\mathcal C,\mathcal D]\) and \(\operatorname{Fun}(\mathcal C,\mathcal D)\) emphasize the endpoint order. The source appears first in bracket and Fun notation but as exponent in exponential notation. Both must be read before interpreting examples.

“All functors” is relative to foundations. If both categories are small, the functor category is small. If \(\mathcal C\) is small and \(\mathcal D\) locally small, the functor category is locally small. Two merely locally small categories need not yield a locally small functor category, because a natural transformation has a component for every source object.^[1]

The naturality condition is stronger than choosing component maps independently. In a discrete source category there are no nonidentity arrows, so every component family is natural. In a source with relations, the commutative squares impose coherence and can drastically reduce the morphisms.

Manages Complexity¶

A diagram can contain many objects and arrows. Functor Category packages the entire coherent diagram as one object. A natural transformation then packages a compatible family of comparisons as one morphism. This permits ordinary categorical reasoning at the diagram level.

Pointwise limits and colimits reduce a global diagram-of-diagrams calculation to calculations in \(\mathcal D\) at each index. Naturality then assembles those componentwise results into a functor. This is a powerful local-to-global compression: values are computed locally, while the universal property guarantees coherent assembly.

The abstraction deliberately hides internal representation choices. Two naturally isomorphic functors are isomorphic objects regardless of how their data are presented. What remains explicit is the indexing category and its coherence equations.

Abstract Reasoning¶

For natural transformations \(F\xRightarrow{\eta}G\xRightarrow{\theta}H\), consider \(u:c\to c'\). Naturality gives

\[ H(u)\theta_c=\theta_{c'}G(u),\qquad G(u)\eta_c=\eta_{c'}F(u). \]

Multiplying yields

\[ H(u)(\theta_c\eta_c)=\theta_{c'}\eta_{c'}F(u), \]

so \((\theta\eta)_c=\theta_c\eta_c\) is natural. Associativity is inherited from each hom-set of \(\mathcal D\). This verifies the category axioms without inventing a new composition law.

For a small \(\mathcal C\), evaluation \(\operatorname{ev}_c:[\mathcal C,\mathcal D]\to\mathcal D\) sends \(F\) to \(F(c)\) and \(\eta\) to \(\eta_c\). Pointwise limits say that if \(L=\lim_jF_j\), then \(L(c)=\lim_jF_j(c)\). The universal cones assemble because uniqueness in \(\mathcal D\) enforces naturality.^[3]

Knowledge Transfer¶

The construction transfers literally across index categories and codomains. Replacing \(\mathcal C\) changes the diagram shape; replacing \(\mathcal D\) changes the mathematical objects used as values. The object/morphism roles remain functors and natural transformations.

Transfer can carry extra structure pointwise. If \(\mathcal D\) is additive, abelian, complete, or cocomplete and size hypotheses hold, the functor category often inherits corresponding structure. The exact theorem must be checked; the name alone does not guarantee enriched, monoidal, or model-category structure.

Programming analogies involving type constructors and polymorphic transformations can instantiate selected categorical models, but ordinary software “functors” do not automatically define the full mathematical functor category. Literal transfer requires fixed categories and naturality.

Examples¶

Arrow category. Let \([1]\) have objects 0 and 1 and one arrow \(0\to1\). A functor \([1]\to\mathcal D\) selects an arrow \(x\to y\). A natural transformation between two such functors is exactly a commutative square. Thus \(\mathcal D^{[1]}\) turns arrows into objects.

Presheaves. In \([\mathcal C^{op},\mathbf{Set}]\), objects assign a set to each \(c\) and restriction maps contravariantly; morphisms are natural families of functions. This category is the ambient setting for the Yoneda embedding and sheaf constructions.

Group actions. Regard a group \(G\) as a one-object category. A functor \(G\to\mathbf{Set}\) is a set equipped with a \(G\)-action. A natural transformation is an equivariant map. The abstract construction recovers the familiar category of \(G\)-sets.

Discrete pair. If \(\mathcal C\) is discrete on two objects, \([\mathcal C,\mathcal D]\cong\mathcal D\times\mathcal D\). Objects are pairs, morphisms are pairs, and products or coproducts are computed componentwise.

Structural Tensions¶

Objectwise freedom versus natural coherence: component maps may exist but fail to commute with source arrows. Diagnostic: test every generating naturality square.
Higher-order organization versus ordinary category laws: functors become objects, yet composition must remain ordinary vertical composition. Diagnostic: compute identities and composites componentwise and verify their endpoints.
Pointwise construction versus global validity: component limits can exist without obvious functorial assembly. Diagnostic: use the universal property to define action on source morphisms and check naturality.
Expressive generality versus size: allowing large source categories can make natural-transformation collections too large. Diagnostic: state smallness, local smallness, or universe conventions.
Full functor category versus structured subcategory: additive or enriched restrictions may change objects and morphisms. Diagnostic: declare which functors and transformations are admitted.
Autonomy versus Category plus Functor reduction: the ingredients are known, but their fixed-endpoint higher category is not entailed separately. Diagnostic: ask whether the reduction supplies natural transformations as homs and componentwise composition.

Structural–Framed Character¶

Functor Category is highly structural. Its identity depends on category axioms, fixed endpoints, and naturality rather than empirical or institutional framing. Relabeling objects or replacing categories by equivalent ones preserves its essential role up to categorical equivalence.

It remains domain-specific because “object,” “morphism,” “functor,” and “natural transformation” are exact category-theoretic types. Loose collections of transformations or software callbacks do not qualify.

Structural Core vs. Domain Accent¶

The portable core is a higher-order container whose objects are structure-preserving maps and whose arrows coherently transform those maps. The domain accent is the category-theoretic definition of functor, naturality, vertical composition, and size.

That vocabulary and machinery do not travel literally across three unrelated substrates outside categorical modeling. Category supplies the broader prime; Functor Category is its autonomous domain-specific construction.

prime:category is the proposed minimal parent by strict specialization. A functor category has objects, morphisms, associative composition, and identities. The child fixes the objects as parallel functors and morphisms as natural transformations.

domain_specific:functor is an indispensable component but not a taxonomic parent: a functor category is not itself a functor. Natural Transformation has no accepted-899 node and is described internally. Product and exponential analogies are related constructions, not required parents.

Relationships to Other Abstractions¶

Current abstraction Functor Category Domain-specific

Parents (1) — more general patterns this builds on

Functor Category is a kind of Category Prime

prime:category is the proposed minimal parent by strict specialization.

Hierarchy paths (3) — routes to 3 parentless roots

Functor Category → Category → Associativity → Invariance

Show alternative paths (2)

Neighborhood in Abstraction Space¶

Functor Category sits in a sparse region of the domain-specific corpus (83^rd percentile for distinctiveness): few abstractions share its structure, so a faithful description tends to retrieve it precisely.

Family — Unclustered & Miscellaneous (1565 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-09-08

Not to Be Confused With¶

Functor: one object in a functor category.
Natural transformation: one morphism between two parallel functors.
Category of categories: categories as objects and functors as arrows.
Comma category: objects built from arrows involving fixed functors.
Arrow category: the special case \(\mathcal D^{[1]}\).
Presheaf category: the special case \([\mathcal C^{op},\mathbf{Set}]\).
Diagram: one functor from an indexing category, not the category of all such diagrams.
Enriched functor category: a refinement requiring enrichment data beyond the ordinary construction.

References¶

[1] Emily Riehl, Category Theory in Context, Dover, 2016, section 1.7, “The 2-category of categories,” especially Corollary 1.7.2 and Remark 1.7.3, author PDF, https://math.jhu.edu/~eriehl/context.pdf. registry ↩a ↩b

[2] Saunders Mac Lane, Categories for the Working Mathematician, 2^nd ed., Graduate Texts in Mathematics 5, Springer, 1998, chapters I–II, https://doi.org/10.1007/978-1-4612-9839-7. registry ↩

[3] Emily Riehl, Category Theory in Context, Proposition 3.3.9 and Exercise 3.3.vi, pointwise limits and colimits in functor categories, https://math.jhu.edu/~eriehl/context.pdf. registry ↩a ↩b