Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $X$ be a topological space. Let $Y$ be a closed subset of $X$. Let $f\colon Y \rightarrow X$ be the canonical injection. Let $\mathcal{F}$ be a sheaf of abelian groups on $X$. Suppose $\mathcal{F}_x = 0$ for every $x \in X - Y$. Is the canonical morphism $\Gamma(X, \mathcal{F}) \rightarrow \Gamma(Y, f^{-1}(\mathcal{F}))$ an isomorphism?

Motivation Perhaps the above result can be used to prove this.

share|cite|improve this question
up vote 1 down vote accepted

Let $U = X \setminus Y$, and let $j: U \to X$ be the inclusion. Given any sheaf $\mathcal F$ on $X$, there is a canonical short exact sequence $$0 \to j_! j^{-1}\mathcal F \to F \to f_* f^{-1}\mathcal F \to 0.$$ In your case, you are assuming that $j^{-1}\mathcal F = 0$ (since all the stalks vanish) and so you get $$\mathcal F \cong f_* f^{-1}\mathcal F.$$ Passing to global sections gives the statement you ask about.

share|cite|improve this answer
Excellent. I came up with a pedestrian proof but yours is more sophisticated. – Makoto Kato Nov 4 '12 at 21:49
The above exact sequence is in Ex. 1.19 of Hartshorne's Algebraic Geometry Ch. II. – Makoto Kato Nov 4 '12 at 22:24

We identify $\mathcal{F}$ with its espace etale(e.g. Hartshorne's Algebraic Geometry). By this result, $f^{-1}(\mathcal{F}) = p^{-1}(Y)$, where $p\colon \mathcal{F} → X$ is the canonical map.

Let $\psi\colon \Gamma(X, \mathcal{F}) \rightarrow \Gamma(Y, f^{-1}(\mathcal{F}))$ be the canonical morphism. Suppose $\psi(s) = 0$. Then $s_y = 0$ for every $y \in Y$. Since $F_x = 0$ for $x \in X - Y$, $s_x = 0$ for every $x \in X$. Hence $s = 0$. Hence $\psi$ is injective.

Let $t \in \Gamma(Y, f^{-1}(\mathcal{F}))$. There exists a cover $(V_i)_{i\in I}$ of $Y$ consisting of open subsets $V_i$ of $Y$ such that, for every $i \in I$, there exists an open subset $U_i$ of $X$ containing $V_i$ and $s_i \in \Gamma(U_i, \mathcal{F})$ such that $t_y = (s_i)_y$ for every $y \in V_i$. For every $i \in I$, there exists an open subset $W_i$ of $X$ such that $V_i = W_i ∩ Y$. Then $V_i \subset U_i ∩ W_i \subset U_i$. and $U_i \cap W_i \cap Y = V_i$. Hence replacing $U_i$ by $U_i \cap W_i$, and $s_i$ by $s_i|(U_i \cap W_i)$, we can assume that $V_i = U_i \cap Y$. Since $s_i|U_i \cap U_j \cap Y = t|V_i \cap V_j = s_j|U_i \cap U_j \cap Y$ and $(s_i)_x = (s_j)_x$ for every $x \in (U_i \cap U_j) - Y$, $s_i|U_i \cap U_j = s_j|U_i \cap U_j$. Hence there exists a unique $s \in \Gamma(U, \mathcal{F})$ such that $s|U_i = s_i$, where $U = \bigcup_i U_i$. Since $X - Y$ is open, there exists $s' \in \Gamma(X, \mathcal{F})$ such that $s'_x = 0$ if $x \in X - Y$ and $s'_x = s_x$ if $x \in U$. Since $s'_y = s_y = t_y$ for every $y \in Y$, $\psi(s') = t$. Hence $\psi$ is surjective.

share|cite|improve this answer

Let $\psi\colon \mathcal{F} \rightarrow f_*(f^{-1}(\mathcal{F}))$ be the canonical morphism. It suffices to prove that $\psi$ is an isomorphism. Since $Y$ is closed, if $x \in X - Y$, $f_*(f^{-1}(\mathcal{F}))_x = 0$. If $y \in Y$, $f_*(f^{-1}(\mathcal{F}))_y = \mathcal{F}_y$. Hence it suffices to prove that if $y \in Y$, $\psi_y\colon \mathcal{F}_y \rightarrow \mathcal{F}_y$ is an isomorphism. But it is easy to see this.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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