1
$\begingroup$

Let $ X \hookrightarrow Y$ be a closed immersion of (connective) spectral Deligne-Mumford stacks, is $ i_* : Qcoh(X) \rightarrow Qcoh(Y)$ conservative? Somehow I couldn't find the statement in SAG...

$\endgroup$

1 Answer 1

1
$\begingroup$

First, recall this is true when $X$ and $Y$ are spectral affine schemes. Indeed by [SAG, Proposition 2.5.1.1], $i_*$ is t-exact so it suffices to show this on the hearts, and in particular you can reduce to the case where $X$ and $Y$ are 0-truncated/ordinary affine schemes, which is obvious.

The general case follows by descent. Let $F \to G$ be a morphism in $Qcoh(X)$ such that $i_*(F) \to i_*(G)$ is an equivalence. We want to show that $F \to G$ is already an equivalence. Let $q : V \to Y$ be an etale atlas with $V$ an affine scheme, and $p : U = V \times_Y X \to X$ the induced etale atlas. Since closed immersions are affine, $U$ is also an affine scheme. By etale descent for $Qcoh$, $p^*$ is conservative, so it suffices to show that $p^*(F) \to p^*(G)$ is an equivalence in $Qcoh(U)$. By the first paragraph, push forward along the closed immersion $j : U \to V$ induces a conservative functor $j_*$ so it further suffices to show that $j_*p^*(F) \to j_*p^*(G)$ is an equivalence. By base change, this is the same as $q^*i_*(F) \to q^*i_*(G)$, which is an equivalence by assumption.

$\endgroup$
2
  • $\begingroup$ I suppose the line "V is also affine scheme" should read "U is also an affine scheme"? Also, is it necessary to assume the stack $Y$ is quasi-compact? Anyway thanks for the answer :). $\endgroup$
    – Anette
    Jul 9, 2019 at 15:35
  • $\begingroup$ Typo corrected thanks. I don't think $Y$ needs to be quasi-compact. For the base change formula you need $i$ to be quasi-compact quasi-separated, but this is true. $\endgroup$ Jul 9, 2019 at 16:51

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.