The Stacks project

Lemma 93.12.4. Let $k$ be a field. Let $B$ be a finite type $k$-algebra. Let $J \subset B$ be an ideal such that $\mathop{\mathrm{Spec}}(B) \to \mathop{\mathrm{Spec}}(k)$ is smooth on the complement of $V(J)$. Let $N$ be a finite $B$-module. Then there is a canonical bijection

\[ \text{Exal}_ k(B, N) \to \text{Exal}_ k(B^\wedge , N^\wedge ) \]

Here $B^\wedge $ and $N^\wedge $ are the $J$-adic completions.

Proof. The map is given by completion: given $0 \to N \to C \to B \to 0$ in $\text{Exal}_ k(B, N)$ we send it to the completion $C^\wedge $ of $C$ with respect to the inverse image of $J$. Compare with the proof of Lemma 93.8.10.

Since $k \to B$ is of finite presentation the complex $\mathop{N\! L}\nolimits _{B/k}$ can be represented by a complex $N^{-1} \to N^0$ where $N^ i$ is a finite $B$-module, see Algebra, Section 10.134 and in particular Algebra, Lemma 10.134.2. As $B$ is Noetherian, this means that $\mathop{N\! L}\nolimits _{B/k}$ is pseudo-coherent. For $g \in J$ the $k$-algebra $B_ g$ is smooth and hence $(\mathop{N\! L}\nolimits _{B/k})_ g = \mathop{N\! L}\nolimits _{B_ g/k}$ is quasi-isomorphic to a finite projective $B$-module sitting in degree $0$. Thus $\text{Ext}^ i_ B(\mathop{N\! L}\nolimits _{B/k}, N)_ g = 0$ for $i \geq 1$ and any $B$-module $N$. By More on Algebra, Lemma 15.102.1 we conclude that

\[ \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N) \longrightarrow \mathop{\mathrm{lim}}\nolimits _ n \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N/J^ n N) \]

is an isomorphism for any finite $B$-module $N$.

Injectivity of the map. Suppose that $0 \to N \to C \to B \to 0$ is in $\text{Exal}_ k(B, N)$ and maps to zero in $\text{Exal}_ k(B^\wedge , N^\wedge )$. Choose a splitting $C^\wedge = B^\wedge \oplus N^\wedge $. Then the induced map $C \to C^\wedge \to N^\wedge $ gives maps $C \to N/J^ nN$ for all $n$. Hence we see that our element is in the kernel of the maps

\[ \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N) \to \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N/J^ n N) \]

for all $n$. By the previous paragraph we conclude that our element is zero.

Surjectivity of the map. Let $0 \to N^\wedge \to C' \to B^\wedge \to 0$ be an element of $\text{Exal}_ k(B^\wedge , N^\wedge )$. Pulling back by $B \to B^\wedge $ we get an element $0 \to N^\wedge \to C'' \to B \to 0$ in $\text{Exal}_ k(B, N^\wedge )$. we have

\[ \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N^\wedge ) = \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N) \otimes _ B B^\wedge = \text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N) \]

The first equality as $N^\wedge = N \otimes _ B B^\wedge $ (Algebra, Lemma 10.97.1) and More on Algebra, Lemma 15.65.3. The second equality because $\text{Ext}^1_ B(\mathop{N\! L}\nolimits _{B/k}, N)$ is $J$-power torsion (see above), $B \to B^\wedge $ is flat and induces an isomorphism $B/J \to B^\wedge /JB^\wedge $, and More on Algebra, Lemma 15.89.3. Thus we can find a $C \in \text{Exal}_ k(B, N)$ mapping to $C''$ in $\text{Exal}_ k(B, N^\wedge )$. Thus

\[ 0 \to N^\wedge \to C' \to B^\wedge \to 0 \quad \text{and}\quad 0 \to N^\wedge \to C^\wedge \to B^\wedge \to 0 \]

are two elements of $\text{Exal}_ k(B^\wedge , N^\wedge )$ mapping to the same element of $\text{Exal}_ k(B, N^\wedge )$. Taking the difference we get an element $0 \to N^\wedge \to C' \to B^\wedge \to 0$ of $\text{Exal}_ k(B^\wedge , N^\wedge )$ whose image in $\text{Exal}_ k(B, N^\wedge )$ is zero. This means there exists

\[ \xymatrix{ 0 \ar[r] & N^\wedge \ar[r] & C' \ar[r] & B^\wedge \ar[r] & 0 \\ & & B \ar[u]^\sigma \ar[ru] } \]

Let $J' \subset C'$ be the inverse image of $JB^\wedge \subset B^\wedge $. To finish the proof it suffices to note that $\sigma $ is continuous for the $J$-adic topology on $B$ and the $J'$-adic topology on $C'$ and that $C'$ is $J'$-adically complete by Algebra, Lemma 10.97.10 (here we also use that $C'$ is Noetherian; small detail omitted). Namely, this means that $\sigma $ factors through the completion $B^\wedge $ and $C' = 0$ in $\text{Exal}_ k(B^\wedge , N^\wedge )$. $\square$


Comments (0)


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 0DZ4. Beware of the difference between the letter 'O' and the digit '0'.