## 78.24 Explicit description of quotient stacks

In order to formulate the result we need to introduce some notation. Assume $B \to S$ and $(U, R, s, t, c)$ are as in Definition 78.20.1 (1). Let $T$ be a scheme over $S$. Let $\mathcal{T} = \{ T_ i \to T\} _{i \in I}$ be an fppf covering. A $[U/R]$-descent datum relative to $\mathcal{T}$ is given by a system $(u_ i, r_{ij})$ where

1. for each $i$ a morphism $u_ i : T_ i \to U$, and

2. for each $i, j$ a morphism $r_{ij} : T_ i \times _ T T_ j \to R$

such that

1. as morphisms $T_ i \times _ T T_ j \to U$ we have

$s \circ r_{ij} = u_ i \circ \text{pr}_0 \quad \text{and}\quad t \circ r_{ij} = u_ j \circ \text{pr}_1,$
2. as morphisms $T_ i \times _ T T_ j \times _ T T_ k \to R$ we have

$c \circ (r_{jk} \circ \text{pr}_{12}, r_{ij} \circ \text{pr}_{01}) = r_{ik} \circ \text{pr}_{02}.$

A morphism $(u_ i, r_{ij}) \to (u'_ i, r'_{ij})$ between two $[U/R]$-descent data over the same covering $\mathcal{T}$ is a collection $(r_ i : T_ i \to R)$ such that

1. $(\alpha )$ as morphisms $T_ i \to U$ we have

$u_ i = s \circ r_ i \quad \text{and}\quad u'_ i = t \circ r_ i$
2. $(\beta )$ as morphisms $T_ i \times _ T T_ j \to R$ we have

$c \circ (r'_{ij}, r_ i \circ \text{pr}_0) = c \circ (r_ j \circ \text{pr}_1, r_{ij}).$

There is a natural composition law on morphisms of descent data relative to a fixed covering and we obtain a category of descent data. This category is a groupoid. Finally, if $\mathcal{T}' = \{ T'_ j \to T\} _{j \in J}$ is a second fppf covering which refines $\mathcal{T}$ then there is a notion of pullback of descent data. This is particularly easy to describe explicitly in this case. Namely, if $\alpha : J \to I$ and $\varphi _ j : T'_ j \to T_{\alpha (i)}$ is the morphism of coverings, then the pullback of the descent datum $(u_ i, r_{ii'})$ is simply

$(u_{\alpha (i)} \circ \varphi _ j, r_{\alpha (j)\alpha (j')} \circ \varphi _ j \times \varphi _{j'}).$

Pullback defined in this manner defines a functor from the category of descent data over $\mathcal{T}$ to the category of descend data over $\mathcal{T}'$.

Lemma 78.24.1. Assume $B \to S$ and $(U, R, s, t, c)$ are as in Definition 78.20.1 (1). Let $\pi : \mathcal{S}_ U \to [U/R]$ be as in Lemma 78.20.2. Let $T$ be a scheme over $S$.

1. for every object $x$ of the fibre category $[U/R]_ T$ there exists an fppf covering $\{ f_ i : T_ i \to T\} _{i \in I}$ such that $f_ i^*x \cong \pi (u_ i)$ for some $u_ i \in U(T_ i)$,

2. the composition of the isomorphisms

$\pi (u_ i \circ \text{pr}_0) = \text{pr}_0^*\pi (u_ i) \cong \text{pr}_0^*f_ i^*x \cong \text{pr}_1^*f_ j^*x \cong \text{pr}_1^*\pi (u_ j) = \pi (u_ j \circ \text{pr}_1)$

are of the form $\pi (r_{ij})$ for certain morphisms $r_{ij} : T_ i \times _ T T_ j \to R$,

3. the system $(u_ i, r_{ij})$ forms a $[U/R]$-descent datum as defined above,

4. any $[U/R]$-descent datum $(u_ i, r_{ij})$ arises in this manner,

5. if $x$ corresponds to $(u_ i, r_{ij})$ as above, and $y \in \mathop{\mathrm{Ob}}\nolimits ([U/R]_ T)$ corresponds to $(u'_ i, r'_{ij})$ then there is a canonical bijection

$\mathop{\mathrm{Mor}}\nolimits _{[U/R]_ T}(x, y) \longleftrightarrow \left\{ \begin{matrix} \text{morphisms }(u_ i, r_{ij}) \to (u'_ i, r'_{ij}) \\ \text{of }[U/R]\text{-descent data} \end{matrix} \right\}$
6. this correspondence is compatible with refinements of fppf coverings.

Proof. Statement (1) is part of the construction of the stackyfication. Part (2) follows from Lemma 78.22.1. We omit the verification of (3). Part (4) is a translation of the fact that in a stack all descent data are effective. We omit the verifications of (5) and (6). $\square$

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.

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 04MB. Beware of the difference between the letter 'O' and the digit '0'.