Remark 99.7.9 (Obstructions for quotients). In Situation 99.7.1 assume that $\mathcal{F}$ is flat over $B$. Let $T \subset T'$ be an first order thickening of schemes over $B$ with ideal sheaf $\mathcal{J}$. Then $X_ T \subset X_{T'}$ is a first order thickening of algebraic spaces whose ideal sheaf $\mathcal{I}$ is a quotient of $f_ T^*\mathcal{J}$. We will think of sheaves on $X_{T'}$, resp. $T'$ as sheaves on $X_ T$, resp. $T$ using the fundamental equivalence described in More on Morphisms of Spaces, Section 76.9. Let
define an element $x$ of $Q_{\mathcal{F}/X/B}(T)$. Since $\mathcal{F}_{T'}$ is flat over $T'$ we have a short exact sequence
and we have $f_ T^*\mathcal{J} \otimes _{\mathcal{O}_{X_ T}} \mathcal{F}_ T = \mathcal{I} \otimes _{\mathcal{O}_{X_ T}} \mathcal{F}_ T$, see Deformation Theory, Lemma 91.11.2. Let us use the abbreviation $ f_ T^*\mathcal{J} \otimes _{\mathcal{O}_{X_ T}} \mathcal{G} = \mathcal{G} \otimes _{\mathcal{O}_ T} \mathcal{J} $ for an $\mathcal{O}_{X_ T}$-module $\mathcal{G}$. Since $\mathcal{Q}$ is flat over $T$, we obtain a short exact sequence
Combining the above we obtain an canonical extension
of $\mathcal{O}_{X_ T}$-modules. This defines a canonical class
If $o_ x(T')$ is zero, then we obtain a splitting of the short exact sequence defining it, in other words, we obtain a $\mathcal{O}_{X_{T'}}$-submodule $\mathcal{K}' \subset \pi ^{-1}(\mathcal{K})$ sitting in a short exact sequence $0 \to \mathcal{K} \otimes _{\mathcal{O}_ T} \mathcal{J} \to \mathcal{K}' \to \mathcal{K} \to 0$. Then it follows from the lemma reference above that $\mathcal{Q}' = \mathcal{F}_{T'}/\mathcal{K}'$ is a lift of $x$ to an element of $Q_{\mathcal{F}/X/B}(T')$. Conversely, the reader sees that the existence of a lift implies that $o_ x(T')$ is zero. Moreover, if $x \in Q_{\mathcal{F}/X/B}^{fp}(T)$, then automatically $x' \in Q_{\mathcal{F}/X/B}^{fp}(T')$ by Deformation Theory, Lemma 91.11.3. If we ever need this remark we will turn this remark into a lemma, precisely formulate the result and give a detailed proof (in fact, all of the above works in the setting of arbitrary ringed topoi).
Comments (0)