TheoremComplete

Adjoint Functor Theorem for ∞\infty-Categories

The adjoint functor theorem for ∞\infty-categories states that a functor between presentable ∞\infty-categories has a right adjoint if and only if it preserves small colimits. This is the ∞\infty-categorical analogue of the classical adjoint functor theorem and is one of the most powerful and frequently used results in higher category theory.


Statement

Theorem4.1Adjoint Functor Theorem (Lurie)

Let F:C→DF: \mathcal{C} \to \mathcal{D} be a functor between presentable ∞\infty-categories.

(Left adjoint version) FF has a right adjoint if and only if FF preserves small colimits.

(Right adjoint version) FF has a left adjoint if and only if FF preserves small limits and is accessible (preserves Îș\kappa-filtered colimits for some regular cardinal Îș\kappa).

ExampleApplication: tensor product adjunction

The smash product functor X∧−:S∗→S∗X \wedge -: \mathcal{S}_* \to \mathcal{S}_* preserves colimits (being a left adjoint of the mapping space functor). By the adjoint functor theorem, it has a right adjoint, which is the pointed mapping space Map⁡∗(X,−)\operatorname{Map}_*(X, -).

More generally, for any presentable ∞\infty-category C\mathcal{C} tensored over S\mathcal{S}, the tensor X⊗−:C→CX \otimes -: \mathcal{C} \to \mathcal{C} has a right adjoint (the cotensor/mapping object) whenever it preserves colimits.

ExampleApplication: derived functors exist

For a ring homomorphism φ:R→S\varphi: R \to S, the extension of scalars functor S⊗RL−:D(R)→D(S)S \otimes_R^{\mathbf{L}} -: D(R) \to D(S) preserves all colimits. By the adjoint functor theorem, it has a right adjoint: the forgetful/restriction functor Res⁥:D(S)→D(R)\operatorname{Res}: D(S) \to D(R).

This gives the ∞\infty-categorical version of the extension-restriction adjunction without needing to construct the right adjoint explicitly.

ExampleApplication: localization adjunctions

If L:C→CL: \mathcal{C} \to \mathcal{C} is an accessible localization functor (preserves colimits and is accessible), then LL has a right adjoint (the inclusion of local objects). This is immediate from the adjoint functor theorem.

Example: the truncation functor τ≀n:S→S≀n\tau_{\leq n}: \mathcal{S} \to \mathcal{S}_{\leq n} preserves colimits and is accessible, so it has a right adjoint (the inclusion of nn-truncated spaces into all spaces).


Proof Ideas

Proof

The proof uses the theory of accessible ∞\infty-categories. Key steps:

  1. Representability: A functor G:D→SG: \mathcal{D} \to \mathcal{S} is representable (i.e., G≃Map⁡D(−,d)G \simeq \operatorname{Map}_{\mathcal{D}}(-, d) for some dd) if and only if GG preserves small limits and is accessible. This is the ∞\infty-categorical version of the representable functor theorem.

  2. Construction of right adjoint: For F:C→DF: \mathcal{C} \to \mathcal{D} preserving colimits, define G:D→CG: \mathcal{D} \to \mathcal{C} by G(d)=G(d) = the object representing the functor c↩Map⁡D(F(c),d)c \mapsto \operatorname{Map}_{\mathcal{D}}(F(c), d). This functor preserves limits (as a representable functor in cc) and is accessible (because C\mathcal{C} is presentable). By representability, G(d)G(d) exists for each dd.

  3. Naturality: The construction of GG is natural in dd, giving a functor G:D→CG: \mathcal{D} \to \mathcal{C}. The adjunction F⊣GF \dashv G follows from the universal property.

The accessibility condition in the right adjoint version is needed to ensure the "solution set condition" is satisfied, analogous to the classical Freyd adjoint functor theorem.

■

Applications

ExampleSheafification

The inclusion Shv⁥(C)â†ȘFun⁥(Cop,S)\operatorname{Shv}(\mathcal{C}) \hookrightarrow \operatorname{Fun}(\mathcal{C}^{\mathrm{op}}, \mathcal{S}) preserves limits and is accessible. By the adjoint functor theorem, it has a left adjoint: the sheafification functor L:Fun⁥(Cop,S)→Shv⁥(C)L: \operatorname{Fun}(\mathcal{C}^{\mathrm{op}}, \mathcal{S}) \to \operatorname{Shv}(\mathcal{C}).

ExampleKan extensions

For a functor f:A→Bf: \mathcal{A} \to \mathcal{B} between small ∞\infty-categories, the restriction functor f∗:Fun⁡(B,C)→Fun⁡(A,C)f^*: \operatorname{Fun}(\mathcal{B}, \mathcal{C}) \to \operatorname{Fun}(\mathcal{A}, \mathcal{C}) preserves all limits and colimits. When C\mathcal{C} is presentable:

  • f∗f^* has a left adjoint f!f_! (left Kan extension) because f∗f^* preserves limits and is accessible.
  • f∗f^* has a right adjoint f∗f_* (right Kan extension) because f∗f^* preserves colimits.

This gives the full adjunction f!⊣f∗⊣f∗f_! \dashv f^* \dashv f_*.

ExampleStabilization adjunction

The forgetful functor Ω∞:Sp⁡→S∗\Omega^\infty: \operatorname{Sp} \to \mathcal{S}_* (from spectra to pointed spaces) preserves limits and is accessible. By the adjoint functor theorem, it has a left adjoint Σ∞:S∗→Sp⁥\Sigma^\infty: \mathcal{S}_* \to \operatorname{Sp} (the suspension spectrum functor).

The resulting adjunction ÎŁâˆžâŠŁÎ©âˆž\Sigma^\infty \dashv \Omega^\infty is the fundamental link between unstable and stable homotopy theory.

ExampleInternal hom in presentable categories

In a closed monoidal presentable ∞\infty-category (C,⊗)(\mathcal{C}, \otimes), for each object XX, the tensor functor X⊗−:C→CX \otimes -: \mathcal{C} \to \mathcal{C} preserves colimits. By the adjoint functor theorem, it has a right adjoint HomâĄâ€Ÿ(X,−):C→C\underline{\operatorname{Hom}}(X, -): \mathcal{C} \to \mathcal{C}, the internal hom.

This gives the tensor-hom adjunction Map⁥(X⊗Y,Z)≃Map⁥(Y,HomâĄâ€Ÿ(X,Z))\operatorname{Map}(X \otimes Y, Z) \simeq \operatorname{Map}(Y, \underline{\operatorname{Hom}}(X, Z)) automatically.

ExampleLimits from products and equalizers

If a presentable ∞\infty-category C\mathcal{C} has products and equalizers, then it has all limits. By the adjoint functor theorem, the diagonal functor Δ:C→Fun⁡(K,C)\Delta: \mathcal{C} \to \operatorname{Fun}(K, \mathcal{C}) (for any KK) has a right adjoint (the limit functor) if and only if it preserves colimits, which it does.

ExampleRecognition of presentable categories

The adjoint functor theorem provides a recognition principle: an ∞\infty-category C\mathcal{C} is presentable if and only if it is accessible and has all colimits. The theorem then guarantees that C\mathcal{C} also has all limits and that colimit-preserving functors from C\mathcal{C} always have right adjoints.


Historical Notes

RemarkClassical vs. infinity-categorical

The ∞\infty-categorical adjoint functor theorem (Lurie, HTT 5.5.2.9) is cleaner than its classical counterpart because the accessibility condition replaces the "solution set condition" of Freyd's theorem. In the presentable setting, accessibility is automatic for functors defined by natural constructions, making the theorem particularly easy to apply.

The classical adjoint functor theorem (Freyd, 1964) required an explicit solution set, which was often hard to verify. The ∞\infty-categorical version avoids this by working within the well-behaved class of presentable ∞\infty-categories.


Summary

RemarkKey points

The adjoint functor theorem is the central existence theorem for ∞\infty-category theory:

  1. Between presentable ∞\infty-categories, colimit-preserving functors have right adjoints.

  2. Limit-preserving accessible functors have left adjoints.

  3. Applications include: sheafification, Kan extensions, stabilization, internal hom, derived functors.

  4. The theorem reduces existence of adjoints to easily checkable conditions (colimit/limit preservation).

  5. This is one of the most frequently used results in modern homotopy theory and derived algebraic geometry.