103.10 Quasi-coherent modules
We have seen that the category of quasi-coherent modules on an algebraic stack is equivalent to the category of quasi-coherent modules on a presentation, see Sheaves on Stacks, Section 96.15. This fact is the basis for the following.
Lemma 103.10.1. Let \mathcal{X} be an algebraic stack. Let \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) be the category of locally quasi-coherent modules with the flat base change property, see Section 103.8. The inclusion functor i : \mathit{QCoh}(\mathcal{O}_\mathcal {X}) \to \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) has a right adjoint
Q : \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) \to \mathit{QCoh}(\mathcal{O}_\mathcal {X})
such that Q \circ i is the identity functor.
Proof.
Choose a scheme U and a surjective smooth morphism f : U \to \mathcal{X}. Set R = U \times _\mathcal {X} U so that we obtain a smooth groupoid (U, R, s, t, c) in algebraic spaces with the property that \mathcal{X} = [U/R], see Algebraic Stacks, Lemma 94.16.2. We may and do replace \mathcal{X} by [U/R]. By Sheaves on Stacks, Proposition 96.14.3 there is an equivalence
q_1 : \mathit{QCoh}(U, R, s, t, c) \longrightarrow \mathit{QCoh}(\mathcal{O}_\mathcal {X})
Let us construct a functor
q_2 : \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) \longrightarrow \mathit{QCoh}(U, R, s, t, c)
by the following rule: if \mathcal{F} is an object of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) then we set
q_2(\mathcal{F}) = (f^*\mathcal{F}|_{U_{\acute{e}tale}}, \alpha )
where \alpha is the isomorphism
t_{small}^*(f^*\mathcal{F}|_{U_{\acute{e}tale}}) \to t^*f^*\mathcal{F}|_{R_{\acute{e}tale}} \to s^*f^*\mathcal{F}|_{R_{\acute{e}tale}} \to s_{small}^*(f^*\mathcal{F}|_{U_{\acute{e}tale}})
where the outer two morphisms are the comparison maps. Note that q_2(\mathcal{F}) is quasi-coherent precisely because \mathcal{F} is locally quasi-coherent and that we used (and needed) the flat base change property in the construction of the descent datum \alpha . We omit the verification that the cocycle condition (see Groupoids in Spaces, Definition 78.12.1) holds. Looking at the proof of Sheaves on Stacks, Proposition 96.14.3 we see that q_2 \circ i is the quasi-inverse to q_1. We define Q = q_1 \circ q_2. Let \mathcal{F} be an object of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) and let \mathcal{G} be an object of \mathit{QCoh}(\mathcal{O}_\mathcal {X}). We have
\begin{align*} \mathop{\mathrm{Mor}}\nolimits _{\textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X})} (i(\mathcal{G}), \mathcal{F}) & = \mathop{\mathrm{Mor}}\nolimits _{\mathit{QCoh}(U, R, s, t, c)}(q_2(i(\mathcal{G})), q_2(\mathcal{F})) \\ & = \mathop{\mathrm{Mor}}\nolimits _{\mathit{QCoh}(\mathcal{O}_\mathcal {X})}(\mathcal{G}, Q(\mathcal{F})) \end{align*}
where the first equality is Sheaves on Stacks, Lemma 96.14.4 and the second equality holds because q_1 \circ i and q_2 are quasi-inverse equivalences of categories. The assertion Q \circ i \cong \text{id} is a formal consequence of the fact that i is fully faithful.
\square
Lemma 103.10.2. Let \mathcal{X} be an algebraic stack. Let Q : \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) \to \mathit{QCoh}(\mathcal{O}_\mathcal {X}) be the functor constructed in Lemma 103.10.1.
The kernel of Q is exactly the collection of parasitic objects of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}).
For any object \mathcal{F} of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) both the kernel and the cokernel of the adjunction map Q(\mathcal{F}) \to \mathcal{F} are parasitic.
The functor Q is exact and commutes with all limits and colimits.
Proof.
Write \mathcal{X} = [U/R] as in the proof of Lemma 103.10.1. Let \mathcal{F} be an object of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}). It is clear from the proof of Lemma 103.10.1 that \mathcal{F} is in the kernel of Q if and only if \mathcal{F}|_{U_{\acute{e}tale}} = 0. In particular, if \mathcal{F} is parasitic then \mathcal{F} is in the kernel. Next, let x : V \to \mathcal{X} be a flat morphism, where V is a scheme. Set W = V \times _\mathcal {X} U and consider the diagram
\xymatrix{ W \ar[d]_ p \ar[r]_ q & V \ar[d] \\ U \ar[r] & \mathcal{X} }
Note that the projection p : W \to U is flat and the projection q : W \to V is smooth and surjective. This implies that q_{small}^* is a faithful functor on quasi-coherent modules. By assumption \mathcal{F} has the flat base change property so that we obtain p_{small}^*\mathcal{F}|_{U_{\acute{e}tale}} \cong q_{small}^*\mathcal{F}|_{V_{\acute{e}tale}}. Thus if \mathcal{F} is in the kernel of Q, then \mathcal{F}|_{V_{\acute{e}tale}} = 0 which completes the proof of (1).
Part (2) follows from the discussion above and the fact that the map Q(\mathcal{F}) \to \mathcal{F} becomes an isomorphism after restricting to U_{\acute{e}tale}.
To see part (3) note that Q is left exact as a right adjoint. Let 0 \to \mathcal{F} \to \mathcal{G} \to \mathcal{H} \to 0 be a short exact sequence in \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}). Consider the following commutative diagram
\xymatrix{ 0 \ar[r] & Q(\mathcal{F}) \ar[r] \ar[d]_ a & Q(\mathcal{G}) \ar[r] \ar[d]_ b & Q(\mathcal{H}) \ar[r] \ar[d]_ c & 0 \\ 0 \ar[r] & \mathcal{F} \ar[r] & \mathcal{G} \ar[r] & \mathcal{H} \ar[r] & 0 }
Since the kernels and cokernels of a, b, and c are parasitic by part (2) and since the bottom row is a short exact sequence, we see that the top row as a complex of \mathcal{O}_\mathcal {X}-modules has parasitic cohomology sheaves (details omitted; this uses that the category of parasitic modules is a Serre subcategory of the category of all modules). By left exactness of Q we see that only exactness at Q(\mathcal{H}) is at issue. However, the cokernel \mathcal{Q} of Q(\mathcal{G}) \to Q(\mathcal{H})) may be computed either in \textit{Mod}(\mathcal{O}_\mathcal {X}) or in \mathit{QCoh}(\mathcal{O}_\mathcal {X}) with the same result because the inclusion functor \mathit{QCoh}(\mathcal{O}_\mathcal {X}) \to \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) is a left adjoint and hence right exact. Hence \mathcal{Q} = Q(\mathcal{Q}) is both quasi-coherent and parasitic, whence 0 by part (1) as desired.
As a right adjoint Q commutes with all limits. Since Q is exact, to show that Q commutes with all colimits it suffices to show that Q commutes with direct sums, see Categories, Lemma 4.14.12. Let \mathcal{F}_ i, i \in I be a family of objects of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}). To see that Q(\bigoplus \mathcal{F}_ i) is equal to \bigoplus Q(\mathcal{F}_ i) we look at the construction of Q in the proof of Lemma 103.10.1. This uses a presentation \mathcal{X} = [U/R] where U is a scheme. Then Q(\mathcal{F}) is computed by first taking the pair (\mathcal{F}|_{U_{\acute{e}tale}}, \alpha ) in \mathit{QCoh}(U, R, s, t, c) and then using the equivalence \mathit{QCoh}(U, R, s, t, c) \cong \mathit{QCoh}(\mathcal{O}_\mathcal {X}). Since the restriction functor \textit{Mod}(\mathcal{O}_\mathcal {X}) \to \textit{Mod}(\mathcal{O}_{U_{\acute{e}tale}}), \mathcal{F} \mapsto \mathcal{F}|_{U_{\acute{e}tale}} commutes with direct sums, the desired equality is clear.
\square
Lemma 103.10.3. Let f : \mathcal{X} \to \mathcal{Y} be a flat morphism of algebraic stacks. Then Q_\mathcal {X} \circ f^* = f^* \circ Q_\mathcal {Y} where Q_\mathcal {X} and Q_\mathcal {Y} are as in Lemma 103.10.1.
Proof.
Observe that f^* preserves both \mathit{QCoh} and \textit{LQCoh}^{fbc}, see Sheaves on Stacks, Lemma 96.11.2 and Proposition 103.8.1. If \mathcal{F} is in \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {Y}) then Q_\mathcal {Y}(\mathcal{F}) \to \mathcal{F} has parasitic kernel and cokernel by Lemma 103.10.2. As f is flat we get that f^*Q_\mathcal {Y}(\mathcal{F}) \to f^*\mathcal{F} has parasitic kernel and cokernel by Lemma 103.9.2. Thus the induced map f^*Q_\mathcal {Y}(\mathcal{F}) \to Q_\mathcal {X}(f^*\mathcal{F}) has parasitic kernel and cokernel and hence is an isomorphism for example by Lemma 103.9.4.
\square
Lemma 103.10.4. Let \mathcal{X} be an algebraic stack. Let x be an object of \mathcal{X} lying over the scheme U such that x : U \to \mathcal{X} is flat. Then for \mathcal{F} in \mathit{QCoh}^{fbc}(\mathcal{O}_\mathcal {X}) we have Q(\mathcal{F})|_{U_{\acute{e}tale}} = \mathcal{F}|_{U_{\acute{e}tale}}.
Proof.
True because the kernel and cokernel of Q(\mathcal{F}) \to \mathcal{F} are parasitic, see Lemma 103.10.2.
\square
Lemma 103.10.8. Let \mathcal{X} be an algebraic stack. Let \mathcal{F} be an \mathcal{O}_\mathcal {X}-module of finite presentation and let \mathcal{G} be a quasi-coherent \mathcal{O}_\mathcal {X}-module. The internal homs \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}) computed in \textit{Mod}(\mathcal{X}_{\acute{e}tale}, \mathcal{O}_\mathcal {X}) or \textit{Mod}(\mathcal{O}_\mathcal {X}) agree and the common value is an object of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}). The quasi-coherent module hom(\mathcal{F}, \mathcal{G}) = Q(\mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G})) has the following universal property
\mathop{\mathrm{Hom}}\nolimits _\mathcal {X}(\mathcal{H}, hom(\mathcal{F}, \mathcal{G})) = \mathop{\mathrm{Hom}}\nolimits _\mathcal {X}(\mathcal{H} \otimes _{\mathcal{O}_\mathcal {X}} \mathcal{F}, \mathcal{G})
for \mathcal{H} in \mathit{QCoh}(\mathcal{O}_\mathcal {X}).
Proof.
The construction of \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}) in Modules on Sites, Section 18.27 depends only on \mathcal{F} and \mathcal{G} as presheaves of modules; the output \mathop{\mathcal{H}\! \mathit{om}}\nolimits is a sheaf for the fppf topology because \mathcal{F} and \mathcal{G} are assumed sheaves in the fppf topology, see Modules on Sites, Lemma 18.27.1. By Sheaves on Stacks, Lemma 96.12.4 we see that \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}) is locally quasi-coherent. By Lemma 103.7.2 we see that \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}) has the flat base change property. Hence \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}) is an object of \textit{LQCoh}^{fbc}(\mathcal{O}_\mathcal {X}) and it makes sense to apply the functor Q of Lemma 103.10.1 to it. By the universal property of Q we have
\mathop{\mathrm{Hom}}\nolimits _\mathcal {X}(\mathcal{H}, Q(\mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}))) = \mathop{\mathrm{Hom}}\nolimits _\mathcal {X}(\mathcal{H}, \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(\mathcal{F}, \mathcal{G}))
for \mathcal{H} quasi-coherent, hence the displayed formula of the lemma follows from Modules on Sites, Lemma 18.27.6.
\square
Lemma 103.10.9. Let f : \mathcal{X} \to \mathcal{Y} be a flat morphism of algebraic stacks. Let \mathcal{F} be an \mathcal{O}_\mathcal {Y}-module of finite presentation and let \mathcal{G} be a quasi-coherent \mathcal{O}_\mathcal {Y}-module. Then f^*hom(\mathcal{F}, \mathcal{G}) = hom(f^*\mathcal{F}, f^*\mathcal{G}) with notation as in Lemma 103.10.8.
Proof.
We have f^*\mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {Y}}(\mathcal{F}, \mathcal{G}) = \mathop{\mathcal{H}\! \mathit{om}}\nolimits _{\mathcal{O}_\mathcal {X}}(f^*\mathcal{F}, f^*\mathcal{G}) by Modules on Sites, Lemma 18.31.4. (Observe that this step is not where the flatness of f is used as the morphism of ringed topoi associated to f is always flat, see Sheaves on Stacks, Remark 96.6.3.) Then apply Lemma 103.10.3 (and here we do use flatness of f).
\square
Comments (0)