13.37 Compact objects
Here is the definition.
Definition 13.37.1. Let \mathcal{D} be an additive category with arbitrary direct sums. A compact object of \mathcal{D} is an object K such that the map
\bigoplus \nolimits _{i \in I} \mathop{\mathrm{Hom}}\nolimits _{\mathcal{D}}(K, E_ i) \longrightarrow \mathop{\mathrm{Hom}}\nolimits _{\mathcal{D}}(K, \bigoplus \nolimits _{i \in I} E_ i)
is bijective for any set I and objects E_ i \in \mathop{\mathrm{Ob}}\nolimits (\mathcal{D}) parametrized by i \in I.
This notion turns out to be very useful in algebraic geometry. It is an intrinsic condition on objects that forces the objects to be, well, compact.
Lemma 13.37.2. Let \mathcal{D} be a (pre-)triangulated category with direct sums. Then the compact objects of \mathcal{D} form the objects of a Karoubian, saturated, strictly full, (pre-)triangulated subcategory \mathcal{D}_ c of \mathcal{D}.
Proof.
Let (X, Y, Z, f, g, h) be a distinguished triangle of \mathcal{D} with X and Y compact. Then it follows from Lemma 13.4.2 and the five lemma (Homology, Lemma 12.5.20) that Z is a compact object too. It is clear that if X \oplus Y is compact, then X, Y are compact objects too. Hence \mathcal{D}_ c is a saturated triangulated subcategory. Since \mathcal{D} is Karoubian by Lemma 13.4.14 we conclude that the same is true for \mathcal{D}_ c.
\square
Lemma 13.37.3. Let \mathcal{D} be a triangulated category with direct sums. Let E_ i, i \in I be a family of compact objects of \mathcal{D} such that \bigoplus E_ i generates \mathcal{D}. Then every object X of \mathcal{D} can be written as
X = \text{hocolim} X_ n
where X_1 is a direct sum of shifts of the E_ i and each transition morphism fits into a distinguished triangle Y_ n \to X_ n \to X_{n + 1} \to Y_ n[1] where Y_ n is a direct sum of shifts of the E_ i.
Proof.
Set X_1 = \bigoplus _{(i, m, \varphi )} E_ i[m] where the direct sum is over all triples (i, m, \varphi ) such that i \in I, m \in \mathbf{Z} and \varphi : E_ i[m] \to X. Then X_1 comes equipped with a canonical morphism X_1 \to X. Given X_ n \to X we set Y_ n = \bigoplus _{(i, m, \varphi )} E_ i[m] where the direct sum is over all triples (i, m, \varphi ) such that i \in I, m \in \mathbf{Z}, and \varphi : E_ i[m] \to X_ n is a morphism such that E_ i[m] \to X_ n \to X is zero. Choose a distinguished triangle Y_ n \to X_ n \to X_{n + 1} \to Y_ n[1] and let X_{n + 1} \to X be any morphism such that X_ n \to X_{n + 1} \to X is the given one; such a morphism exists by our choice of Y_ n. We obtain a morphism \text{hocolim} X_ n \to X by the construction of our maps X_ n \to X. Choose a distinguished triangle
C \to \text{hocolim} X_ n \to X \to C[1]
Let E_ i[m] \to C be a morphism. Since E_ i is compact, the composition E_ i[m] \to C \to \text{hocolim} X_ n factors through X_ n for some n, say by E_ i[m] \to X_ n. Then the construction of Y_ n shows that the composition E_ i[m] \to X_ n \to X_{n + 1} is zero. In other words, the composition E_ i[m] \to C \to \text{hocolim} X_ n is zero. This means that our morphism E_ i[m] \to C comes from a morphism E_ i[m] \to X[-1]. The construction of X_1 then shows that such morphism lifts to \text{hocolim} X_ n and we conclude that our morphism E_ i[m] \to C is zero. The assumption that \bigoplus E_ i generates \mathcal{D} implies that C is zero and the proof is done.
\square
Lemma 13.37.4. With assumptions and notation as in Lemma 13.37.3. If C is a compact object and C \to X_ n is a morphism, then there is a factorization C \to E \to X_ n where E is an object of \langle E_{i_1} \oplus \ldots \oplus E_{i_ t} \rangle for some i_1, \ldots , i_ t \in I.
Proof.
We prove this by induction on n. The base case n = 1 is clear. If n > 1 consider the composition C \to X_ n \to Y_{n - 1}[1]. This can be factored through some E'[1] \to Y_{n - 1}[1] where E' is a finite direct sum of shifts of the E_ i. Let I' \subset I be the finite set of indices that occur in this direct sum. Thus we obtain
\xymatrix{ E' \ar[r] \ar[d] & C' \ar[r] \ar[d] & C \ar[r] \ar[d] & E'[1] \ar[d] \\ Y_{n - 1} \ar[r] & X_{n - 1} \ar[r] & X_ n \ar[r] & Y_{n - 1}[1] }
By induction the morphism C' \to X_{n - 1} factors through E'' \to X_{n - 1} with E'' an object of \langle \bigoplus _{i \in I''} E_ i \rangle for some finite subset I'' \subset I. Choose a distinguished triangle
E' \to E'' \to E \to E'[1]
then E is an object of \langle \bigoplus _{i \in I' \cup I''} E_ i \rangle . By construction and the axioms of a triangulated category we can choose morphisms C \to E and a morphism E \to X_ n fitting into morphisms of triangles (E', C', C) \to (E', E'', E) and (E', E'', E) \to (Y_{n - 1}, X_{n - 1}, X_ n). The composition C \to E \to X_ n may not equal the given morphism C \to X_ n, but the compositions into Y_{n - 1} are equal. Let C \to X_{n - 1} be a morphism that lifts the difference. By induction assumption we can factor this through a morphism E''' \to X_{n - 1} with E'' an object of \langle \bigoplus _{i \in I'''} E_ i \rangle for some finite subset I' \subset I. Thus we see that we get a solution on considering E \oplus E''' \to X_ n because E \oplus E''' is an object of \langle \bigoplus _{i \in I' \cup I'' \cup I'''} E_ i \rangle .
\square
Definition 13.37.5. Let \mathcal{D} be a triangulated category with arbitrary direct sums. We say \mathcal{D} is compactly generated if there exists a set E_ i, i \in I of compact objects such that \bigoplus E_ i generates \mathcal{D}.
The following proposition clarifies the relationship between classical generators and weak generators.
Proposition 13.37.6. Let \mathcal{D} be a triangulated category with direct sums. Let E be a compact object of \mathcal{D}. The following are equivalent
E is a classical generator for \mathcal{D}_ c and \mathcal{D} is compactly generated, and
E is a generator for \mathcal{D}.
Proof.
If E is a classical generator for \mathcal{D}_ c, then \mathcal{D}_ c = \langle E \rangle . It follows formally from the assumption that \mathcal{D} is compactly generated and Lemma 13.36.4 that E is a generator for \mathcal{D}.
The converse is more interesting. Assume that E is a generator for \mathcal{D}. Let X be a compact object of \mathcal{D}. Apply Lemma 13.37.3 with I = \{ 1\} and E_1 = E to write
X = \text{hocolim} X_ n
as in the lemma. Since X is compact we find that X \to \text{hocolim} X_ n factors through X_ n for some n (Lemma 13.33.9). Thus X is a direct summand of X_ n. By Lemma 13.37.4 we see that X is an object of \langle E \rangle and the lemma is proven.
\square
Comments (0)