Adjoint functor theorems are theorems stating that under certain conditions a functor that preserves limits is a right adjoint, and that a functor that preserves colimits is a left adjoint.
A basic result of category theory is that right adjoint functors preserve all limits that exist in their domain, and, dually, left adjoints preserve all colimits. An adjoint functor theorem is a statement that (under certain conditions) the converse holds: a functor which preserves limits is a right adjoint.
The basic idea of an adjoint functor theorem is that if we could assume that a large category $D$ had all limits over small and large diagrams, then for $R : D \to C$ a functor that preserves all these limits we might define its would-be left adjoint $L$ by taking $L c$ to be the limit
over the comma category $c/R$ (whose objects are pairs $(d,f:c\to R d)$ and whose morphisms are arrows $d\to d'$ in $D$ making the obvious triangle commute in $C$) of the projection functor $\pi: c/R \to D$:
Because with this definition there would be for every $d$ an obvious morphism
(the component map over $d$ of the limiting cone) while moreover because $R$ preserves limits, we would have an isomorphism
and hence an obvious morphism of cone tips
It is easy to check that these would be the unit and counit of an adjunction $L\dashv R$. See adjoint functor for more.
The problem with this would-be argument is that in general the comma category $(c/G)$ may not be small category. But one can generally not expect a large category to have all large limits: even if we pass to a universe in which $(c/G)$ is considered small, a classical theorem of Freyd says that any complete small category is a preorder (see complete small category for the proof, which is valid in classical logic and also holds classically in any Grothendieck topos). Thus, the argument we gave above is necessarily an adjoint functor theorem for preorders:
If $G:D\to C$ is any functor between (small) preorders such that $D$ has, and $G$ preserves, all small meets, then $G$ has a left adjoint.
(This theorem holds in constructive mathematics, although not in predicative mathematics; the classical reasoning before this explains why the theorem is not more general, but the proof itself is already constructive.)
To obtain adjoint functor theorems for categories that are not preorders, one must therefore impose various additional “size conditions” on the category $D$ and/or the functor $G$.
Sufficient conditions for a limit-preserving functor $G : D \to C$ to be a right adjoint include:
$D$ is complete and locally small, and $G$ satisfies the solution set condition.
This is Freyd’s original version, sometimes called the “General Adjoint Functor Theorem”.
$D$ is complete, locally small well-powered, and has a small cogenerating set, and $C$ is locally small.
This is sometimes called the “Special Adjoint Functor Theorem”, and abbreviated to SAFT.
$D$ is locally small and cototal, and $C$ is locally small.
In the first two cases, which work by replacing large limits by small ones, it suffices to assume that $G$ preserves small limits (that it preserves all limits will follow). The third case works by assuming that $D$ has, while not all large limits, enough so that the theorem goes through; thus is this case $G$ must be already known to preserve large limits as well.
Here is a proof of the General Adjoint Functor Theorem: that a functor $R : C \to D$ out of a locally small category $C$ with all small limits has a left adjoint if it preserves these limits and satisfies the solution set condition.
From the discussion at adjoint functors – In terms of universal arrows we have that the existence of the adjoint is equivalent to the existence for each $d \in D$ of an initial object $i_d : d \to R L d$ in the comma category $(d \downarrow R)$: an object such that for each $f : d \to R d'$ there is a unique $\tilde f$ such that
commutes. Now an initial object is the limit of the identity functor, but this is generally a large limit; we replace this with some small limit conditions that guarantee existence of an initial object.
Let $Y$ be a category. Call a small family of objects $F$ weakly initial if for every object $y$ of $Y$ there exists $x \in F$ and a morphism $f: x \to y$.
Suppose $Y$ has small products. If $F$ is a weakly initial family, then $\prod_{x \in F} x$ is a weakly initial object.
Claim: Suppose $Y$ is locally small and small complete. If $x$ is a weakly initial object, then the domain $e$ of the joint equalizer $i: e \to x$ of all arrows $x \to x$ is an initial object. Proof: clearly $e$ is weakly initial. Suppose given an object $y$ and arrows $f, g: e \to y$; we must show $f = g$. Let $j: d \to e$ be the equalizer of $f$ and $g$. There exists an arrow $k: x \to d$. The arrow $i: e \to x$ equalizes $1_x$ and $i j k: x \to x$, so $i j k i = i$. Since $i$ is monic, $j (k i) = 1_e$. Thus $j$ is an epi, and $f = g$ follows.
If $C$ is locally small and small-complete and $R: C \to D$ preserves limits, then $d \downarrow R$ is locally small and small-complete for every object $d$ of $D$.
If in addition each $d \downarrow R$ has a weakly initial family (solution set condition), then by 2. and 3. each $d \downarrow R$ has an initial object. This restates the condition that $R$ has a left adjoint.
As before, the proof proceeds by constructing initial objects of comma categories. We assume that $C$ is locally small, small-complete, well-powered, has a cogenerating set $\{c_\alpha: \alpha \in A\}$, and that $R: C \to D$ is a small-continuous functor into a locally small category $D$.
As before, for each object $d$ of $D$, the comma category $d \downarrow R$ is locally small and small-complete. Moreover, it is easy to check that it is well-powered, and that the set of all objects of the form $d \to R c_\alpha$ is a cogenerating set for $d \downarrow R$.
It then remains to prove that any locally small, small-complete, well-powered category $X$ with a cogenerating set $\{k_s: s \in S\}$ has an initial object. The initial object $0$ is constructed as the intersection = pullback of all subobjects of $\prod_s k_s$, i.e., the minimal subobject. Then, given $f, g: 0 \to x$, the equalizer $Eq(f, g)$ is isomorphic to $0$ because $0$ is minimal, and so $f = g$: there is at most one arrow $0 \to x$ for each $x$.
On the other hand, for each $x$ the canonical map
is monic since the $k_s$ cogenerate. The following pullback of $i$,
gives a subobject $k$ of $\prod_s k_s$ that maps to $x$, and into which $0$ embeds. Thus there exists a map $0 \to x$, and we conclude $0$ is initial.
In practice an important special case is that of functors between locally presentable categories. For these there is the following version of an adjoint functor theorem.
Let $F : C \to D$ be a functor between locally presentable categories. Then
$F$ has a right adjoint precisely if it preserves all small colimits.
$F$ has a left adjoint precisely if it is an accessible functor and preserves all small limits.
The second statement, characterizing when $F$ has a left adjoint, is (AdamekRosicky, theorem 1.66). This is an application of the general adjoint functor theorem: any accessible functor satisfies the solution set condition. The accessibility condition is indeed necessary, as is discussed below in the section In locally presentable categories.
The first statement, characterizing when $F$ has a right adjoint, can be proven using the special adjoint functor theorem: by a non-trivial theorem (AdamekRosicky, theorem 1.58), any locally presentable category is co-wellpowered.
A right adjoint to any cocontinuous functor $F \colon C \to D$ between locally presentable categories can also be constructed directly. If $C$ is locally $\lambda$-presentable and $P_\lambda$ is the subcategory of $\lambda$-small objects, then $C$ is equivalent to the full subcategory of $[P_\lambda^{op},Set]$ of presheaves that preserve $\lambda$-small limits (AdamekRosicky, theorem 1.46). The presheaves in the image of the functor $D \to [P_\lambda^{op},Set]$ defined by $d \mapsto \hom(F-,d)$ preserve $\lambda$-small limits because $F$ is cocontinuous. So this functor factors through the subcategory $C$. The functor $D \to C$ so-constructed is a right adjoint to $F$.
The following is a counter-example, showing that the necessity of the accesibility clause in theorem 3.
For every infinite cardinal number $\kappa$, let $G_\kappa$ be a simple group of cardinality $\kappa$. Define the functor $ML:$ Group $\to$ Set to be the product of all the representable functors $Hom(G_\kappa,-)$. Since no group can admit a nontrivial homomorphism from proper-class-many of the $G_\kappa$, this functor does indeed land (or can be redefined to land) in Set. Since it is a product of representables, it is continuous (and of course Group and Set are locally presentable categories), but it is not itself representable (hence has no left adjoint).
André Joyal has been attributing this example to Saunders MacLane, It appears in print for instance right at the beginning of (AdámekKoubekTrnková01).
Every sheaf topos is a total category and a cototal category.
See the discussion at Grothendieck topos .
It follows that
Let $F : C \to D$ be a functor between sheaf toposes. Then
$F$ has a right adjoint precisely if it preserves all small colimits;
$F$ has a left adjoint precisely if it preserves all small limits.
It is instructive to spell out the construction of the right adjoint from a colimit preserving functor $L$ in the simple case where all categories are categories of presheaves. This is a particularly simple case, but is useful in itself and serves as a template for the general case.
So let now $C$ and $D$ be small categories and
a colimit-preserving functor. Then its right adjoint is given by
as we shall check in a moment. But first notice that using the co-Yoneda lemma this may be rewritten as
where the coend is equivalently given by the colimit
This is the formula for the would-be right adjoint from the general discussion above, only that here the colimit is only over the representables, hence over a small category.
Now we check that the $R$ thus obtained is indeed right adjoint to $L$, by explicitly checking the hom-isomorphism of the pair of adjoint functors:
We compute $Hom_{PSh(D)}(L(X),A)$. In the first step
we use the co-Yoneda lemma for $X$. Then because $L$ preserves colimits this is
Since the hom preserves limits in both arguments, we can take the coend out to get an end
Then we use the standard tensoring of our categories over Set to get
And finally this is recognized as the formula for the hom-set of presheaves (see functor category)
In total this establishes the hom-isomorphism
adjoint functor theorem
A brief introductory discussion is around theorem 5.4 of
A detailed expository survey is
The adjoint functor theorem in context with Yoneda embedding is discussed in
The case for locally presentable categories is discussed in
Jiří Adámek, Jiri Rosicky, Locally presentable and accessible categories Cambridge University Press (1994)
Jiří Adámek, V. Koubek and V. Trnková, How large are left exact functors? , Theory and Applications of Categories 8 (2001) 377-390 (TAC)