The Stacks project

Theorem 76.10.1. Let $S$ be a scheme. Let $F : (\mathit{Sch}/S)_{fppf}^{opp} \to \textit{Sets}$ be a functor. Any one of the following conditions implies that $F$ is an algebraic space:

  1. $F = U/R$ where $(U, R, s, t, c)$ is a groupoid in algebraic spaces over $S$ such that $s, t$ are flat and locally of finite presentation, and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation,

  2. $F = U/R$ where $(U, R, s, t, c)$ is a groupoid scheme over $S$ such that $s, t$ are flat and locally of finite presentation, and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation,

  3. $F$ is a sheaf and there exists an algebraic space $U$ and a morphism $U \to F$ which is representable by algebraic spaces, surjective, flat and locally of finite presentation,

  4. $F$ is a sheaf and there exists a scheme $U$ and a morphism $U \to F$ which is representable (by algebraic spaces or schemes), surjective, flat and locally of finite presentation,

  5. $F$ is a sheaf, $\Delta _ F$ is representable by algebraic spaces, and there exists an algebraic space $U$ and a morphism $U \to F$ which is surjective, flat, and locally of finite presentation, or

  6. $F$ is a sheaf, $\Delta _ F$ is representable, and there exists a scheme $U$ and a morphism $U \to F$ which is surjective, flat, and locally of finite presentation.

Proof. Trivial observations: (6) is a special case of (5) and (4) is a special case of (3). We first prove that cases (5) and (3) reduce to case (1). Namely, by bootstrapping the diagonal Lemma 76.5.3 we see that (3) implies (5). In case (5) we set $R = U \times _ F U$ which is an algebraic space by assumption. Moreover, by assumption both projections $s, t : R \to U$ are surjective, flat and locally of finite presentation. The map $j : R \to U \times _ S U$ is clearly an equivalence relation. By Lemma 76.4.6 the map $U \to F$ is a surjection of sheaves. Thus $F = U/R$ which reduces us to case (1).

Next, we show that (1) reduces to (2). Namely, let $(U, R, s, t, c)$ be a groupoid in algebraic spaces over $S$ such that $s, t$ are flat and locally of finite presentation, and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation. Choose a scheme $U'$ and a surjective étale morphism $U' \to U$. Let $R' = R|_{U'}$ be the restriction of $R$ to $U'$. By Groupoids in Spaces, Lemma 74.18.6 we see that $U/R = U'/R'$. Since $s', t' : R' \to U'$ are also flat and locally of finite presentation (see More on Groupoids in Spaces, Lemma 75.8.1) this reduces us to the case where $U$ is a scheme. As $j$ is an equivalence relation we see that $j$ is a monomorphism. As $s : R \to U$ is locally of finite presentation we see that $j : R \to U \times _ S U$ is locally of finite type, see Morphisms of Spaces, Lemma 63.23.6. By Morphisms of Spaces, Lemma 63.27.10 we see that $j$ is locally quasi-finite and separated. Hence if $U$ is a scheme, then $R$ is a scheme by Morphisms of Spaces, Proposition 63.50.2. Thus we reduce to proving the theorem in case (2).

Assume $F = U/R$ where $(U, R, s, t, c)$ is a groupoid scheme over $S$ such that $s, t$ are flat and locally of finite presentation, and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation. By Lemma 76.8.1 we reduce to that case where $s, t$ are flat, locally of finite presentation, and locally quasi-finite. Let $U = \bigcup _{i \in I} U_ i$ be an affine open covering (with index set $I$ of cardinality $\leq $ than the size of $U$ to avoid set theoretic problems later – most readers can safely ignore this remark). Let $(U_ i, R_ i, s_ i, t_ i, c_ i)$ be the restriction of $R$ to $U_ i$. It is clear that $s_ i, t_ i$ are still flat, locally of finite presentation, and locally quasi-finite as $R_ i$ is the open subscheme $s^{-1}(U_ i) \cap t^{-1}(U_ i)$ of $R$ and $s_ i, t_ i$ are the restrictions of $s, t$ to this open. By Lemma 76.7.1 (or the simpler Spaces, Lemma 61.10.2) the map $U_ i/R_ i \to U/R$ is representable by open immersions. Hence if we can show that $F_ i = U_ i/R_ i$ is an algebraic space, then $\coprod _{i \in I} F_ i$ is an algebraic space by Spaces, Lemma 61.8.4. As $U = \bigcup U_ i$ is an open covering it is clear that $\coprod F_ i \to F$ is surjective. Thus it follows that $U/R$ is an algebraic space, by Spaces, Lemma 61.8.5. In this way we reduce to the case where $U$ is affine and $s, t$ are flat, locally of finite presentation, and locally quasi-finite and $j$ is an equivalence.

Assume $(U, R, s, t, c)$ is a groupoid scheme over $S$, with $U$ affine, such that $s, t$ are flat, locally of finite presentation, and locally quasi-finite, and $j$ is an equivalence relation. Choose $u \in U$. We apply More on Groupoids in Spaces, Lemma 75.15.13 to $u \in U, R, s, t, c$. We obtain an affine scheme $U'$, an étale morphism $g : U' \to U$, a point $u' \in U'$ with $\kappa (u) = \kappa (u')$ such that the restriction $R' = R|_{U'}$ is quasi-split over $u'$. Note that the image $g(U')$ is open as $g$ is étale and contains $u'$. Hence, repeatedly applying the lemma, we can find finitely many points $u_ i \in U$, $i = 1, \ldots , n$, affine schemes $U'_ i$, étale morphisms $g_ i : U_ i' \to U$, points $u'_ i \in U'_ i$ with $g(u'_ i) = u_ i$ such that (a) each restriction $R'_ i$ is quasi-split over some point in $U'_ i$ and (b) $U = \bigcup _{i = 1, \ldots , n} g_ i(U'_ i)$. Now we rerun the last part of the argument in the preceding paragraph: Using Lemma 76.7.1 (or the simpler Spaces, Lemma 61.10.2) the map $U'_ i/R'_ i \to U/R$ is representable by open immersions. If we can show that $F_ i = U'_ i/R'_ i$ is an algebraic space, then $\coprod _{i \in I} F_ i$ is an algebraic space by Spaces, Lemma 61.8.4. As $\{ g_ i : U'_ i \to U\} $ is an étale covering it is clear that $\coprod F_ i \to F$ is surjective. Thus it follows that $U/R$ is an algebraic space, by Spaces, Lemma 61.8.5. In this way we reduce to the case where $U$ is affine and $s, t$ are flat, locally of finite presentation, and locally quasi-finite, $j$ is an equivalence, and $R$ is quasi-split over $u$ for some $u \in U$.

Assume $(U, R, s, t, c)$ is a groupoid scheme over $S$, with $U$ affine, $u \in U$ such that $s, t$ are flat, locally of finite presentation, and locally quasi-finite and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation and $R$ is quasi-split over $u$. Let $P \subset R$ be a quasi-splitting of $R$ over $u$. By Lemma 76.9.1 we see that $(U, R, s, t, c)$ is the restriction of a groupoid $(\overline{U}, \overline{R}, \overline{s}, \overline{t}, \overline{c})$ by a surjective finite locally free morphism $U \to \overline{U}$ such that $P = U \times _{\overline{U}} U$. Note that $s, t$ are the base changes of the morphisms $\overline{s}, \overline{t}$ by $U \to \overline{U}$. As $\{ U \to \overline{U}\} $ is an fppf covering we conclude $\overline{s}, \overline{t}$ are flat, locally of finite presentation, and locally quasi-finite, see Descent, Lemmas 34.20.15, 34.20.11, and 34.20.24. Consider the commutative diagram

\[ \xymatrix{ U \times _{\overline{U}} U \ar@{=}[r] \ar[rd] & P \ar[r] \ar[d] & R \ar[d] \\ & \overline{U} \ar[r]^{\overline{e}} & \overline{R} } \]

It is a general fact about restrictions that the outer four corners form a cartesian diagram. By the equality we see the inner square is cartesian. Since $P$ is open in $R$ (by definition of a quasi-splitting) we conclude that $\overline{e}$ is an open immersion by Descent, Lemma 34.20.16. An application of Groupoids, Lemma 38.20.5 shows that $U/R = \overline{U}/\overline{R}$. Hence we have reduced to the case where $(U, R, s, t, c)$ is a groupoid scheme over $S$, with $U$ affine, $u \in U$ such that $s, t$ are flat, locally of finite presentation, and locally quasi-finite and $j = (t, s) : R \to U \times _ S U$ is an equivalence relation and $e : U \to R$ is an open immersion!

But of course, if $e$ is an open immersion and $s, t$ are flat and locally of finite presentation then the morphisms $t, s$ are étale. For example you can see this by applying More on Groupoids, Lemma 39.4.1 which shows that $\Omega _{R/U} = 0$ which in turn implies that $s, t : R \to U$ is G-unramified (see Morphisms, Lemma 28.33.2), which in turn implies that $s, t$ are étale (see Morphisms, Lemma 28.34.16). And if $s, t$ are étale then finally $U/R$ is an algebraic space by Spaces, Theorem 61.10.5. $\square$


Comments (2)

Comment #1722 by Matthew Emerton on

In part (4), there is a duplicated ``which is''.


Post a comment

Your email address will not be published. Required fields are marked.

In your comment you can use Markdown and LaTeX style mathematics (enclose it like $\pi$). A preview option is available if you wish to see how it works out (just click on the eye in the toolbar).

Unfortunately JavaScript is disabled in your browser, so the comment preview function will not work.

All contributions are licensed under the GNU Free Documentation License.




In order to prevent bots from posting comments, we would like you to prove that you are human. You can do this by filling in the name of the current tag in the following input field. As a reminder, this is tag 04S6. Beware of the difference between the letter 'O' and the digit '0'.