Theorem 101.21.6. Let $\mathcal{X}$ be an algebraic stack. The following are equivalent

$\mathcal{X}$ is DM,

$\mathcal{X}$ is Deligne-Mumford, and

there exists a scheme $W$ and a surjective étale morphism $W \to \mathcal{X}$.

Theorem 101.21.6. Let $\mathcal{X}$ be an algebraic stack. The following are equivalent

$\mathcal{X}$ is DM,

$\mathcal{X}$ is Deligne-Mumford, and

there exists a scheme $W$ and a surjective étale morphism $W \to \mathcal{X}$.

**Proof.**
Recall that (3) is the definition of (2), see Algebraic Stacks, Definition 94.12.2. The implication (3) $\Rightarrow $ (1) is Lemma 101.4.14. Assume (1). Let $x \in |\mathcal{X}|$ be a finite type point. We will produce a scheme over $\mathcal{X}$ which “works” in a neighbourhood of $x$. At the end of the proof we will take the disjoint union of all of these to conclude.

By Lemma 101.18.7 the residual gerbe $\mathcal{Z}_ x$ of $\mathcal{X}$ at $x$ exists and $\mathcal{Z}_ x \to \mathcal{X}$ is locally of finite type. By Lemma 101.4.16 the algebraic stack $\mathcal{Z}_ x$ is DM. By Lemma 101.21.4 there exists a field $k$ and a surjective étale morphism $z : \mathop{\mathrm{Spec}}(k) \to \mathcal{Z}_ x$. In particular the composition $x : \mathop{\mathrm{Spec}}(k) \to \mathcal{X}$ is locally of finite type (by Morphisms of Spaces, Lemmas 67.23.2 and 67.39.9).

Pick a scheme $U$ and a smooth morphism $U \to \mathcal{X}$ such that $x$ is in the image of $|U| \to |\mathcal{X}|$. Consider the following fibre square

\[ \xymatrix{ U \ar[d] & F \ar[l] \ar[d] \\ \mathcal{X} & \mathop{\mathrm{Spec}}(k) \ar[l]_-x } \]

in other words $F = U \times _{\mathcal{X}, x} \mathop{\mathrm{Spec}}(k)$. By Properties of Stacks, Lemma 100.4.3 we see that $F$ is nonempty. As $\mathcal{Z}_ x \to \mathcal{X}$ is a monomorphism we have

\[ \mathop{\mathrm{Spec}}(k) \times _{z, \mathcal{Z}_ x, z} \mathop{\mathrm{Spec}}(k) = \mathop{\mathrm{Spec}}(k) \times _{x, \mathcal{X}, x} \mathop{\mathrm{Spec}}(k) \]

with étale projection maps to $\mathop{\mathrm{Spec}}(k)$ by construction of $z$. Since

\[ F \times _ U F = (\mathop{\mathrm{Spec}}(k) \times _\mathcal {X} \mathop{\mathrm{Spec}}(k)) \times _{\mathop{\mathrm{Spec}}(k)} F \]

we see that the projections maps $F \times _ U F \to F$ are étale as well. It follows that $\Delta _{F/U} : F \to F \times _ U F$ is étale (see Morphisms of Spaces, Lemma 67.39.11). By Morphisms of Spaces, Lemma 67.51.2 this implies that $\Delta _{F/U}$ is an open immersion, which finally implies by Morphisms of Spaces, Lemma 67.38.9 that $F \to U$ is unramified.

Pick a nonempty affine scheme $V$ and an étale morphism $V \to F$. (This could be avoided by working directly with $F$, but it seems easier to explain what's going on by doing so.) Picture

\[ \xymatrix{ U \ar[d] & F \ar[l] \ar[d] & V \ar[l] \ar[ld] \\ \mathcal{X} & \mathop{\mathrm{Spec}}(k) \ar[l]_-x } \]

Then $V \to \mathop{\mathrm{Spec}}(k)$ is a smooth morphism of schemes and $V \to U$ is an unramified morphism of schemes (see Morphisms of Spaces, Lemmas 67.37.2 and 67.38.3). Pick a closed point $v \in V$ with $k \subset \kappa (v)$ finite separable, see Varieties, Lemma 33.25.6. Let $u \in U$ be the image point. The local ring $\mathcal{O}_{V, v}$ is regular (see Varieties, Lemma 33.25.3) and the local ring homomorphism

\[ \varphi : \mathcal{O}_{U, u} \longrightarrow \mathcal{O}_{V, v} \]

coming from the morphism $V \to U$ is such that $\varphi (\mathfrak m_ u)\mathcal{O}_{V, v} = \mathfrak m_ v$, see Morphisms, Lemma 29.35.14. Hence we can find $f_1, \ldots , f_ d \in \mathcal{O}_{U, u}$ such that the images $\varphi (f_1), \ldots , \varphi (f_ d)$ form a basis for $\mathfrak m_ v/\mathfrak m_ v^2$ over $\kappa (v)$. Since $\mathcal{O}_{V, v}$ is a regular local ring this implies that $\varphi (f_1), \ldots , \varphi (f_ d)$ form a regular sequence in $\mathcal{O}_{V, v}$ (see Algebra, Lemma 10.106.3). After replacing $U$ by an open neighbourhood of $u$ we may assume $f_1, \ldots , f_ d \in \Gamma (U, \mathcal{O}_ U)$. After replacing $U$ by a possibly even smaller open neighbourhood of $u$ we may assume that $V(f_1, \ldots , f_ d) \to \mathcal{X}$ is flat and locally of finite presentation, see Lemma 101.21.1. By construction

\[ V(f_1, \ldots , f_ d) \times _\mathcal {X} \mathop{\mathrm{Spec}}(k) \longleftarrow V(f_1, \ldots , f_ d) \times _ U V \]

is étale and $V(f_1, \ldots , f_ d) \times _ U V$ is the closed subscheme $T \subset V$ cut out by $f_1|_ V, \ldots , f_ d|_ V$. Hence by construction $v \in T$ and

\[ \mathcal{O}_{T, v} = \mathcal{O}_{V, v}/(\varphi (f_1), \ldots , \varphi (f_ d)) = \kappa (v) \]

a finite separable extension of $k$. It follows that $T \to \mathop{\mathrm{Spec}}(k)$ is unramified at $v$, see Morphisms, Lemma 29.35.14. By definition of an unramified morphism of algebraic spaces this means that $V(f_1, \ldots , f_ d) \times _\mathcal {X} \mathop{\mathrm{Spec}}(k) \to \mathop{\mathrm{Spec}}(k)$ is unramified at the image of $v$ in $V(f_1, \ldots , f_ d) \times _\mathcal {X} \mathop{\mathrm{Spec}}(k)$. Applying Lemma 101.21.5 we see that on shrinking $U$ to yet another open neighbourhood of $u$ the morphism $V(f_1, \ldots , f_ d) \to \mathcal{X}$ is étale.

We conclude that for every finite type point $x$ of $\mathcal{X}$ there exists an étale morphism $f_ x : W_ x \to \mathcal{X}$ with $x$ in the image of $|f_ x|$. Set $W = \coprod _ x W_ x$ and $f = \coprod f_ x$. Then $f$ is étale. In particular the image of $|f|$ is open, see Properties of Stacks, Lemma 100.4.7. By construction the image contains all finite type points of $\mathcal{X}$, hence $f$ is surjective by Lemma 101.18.6 (and Properties of Stacks, Lemma 100.4.4). $\square$

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.

## Comments (2)

Comment #8153 by DatPham on

Comment #8236 by Stacks Project on