3
$\begingroup$

Let $X$ be a topological space and $U\subset X$ an open subset. Let's work in the category of sheaves of abelian groups on $X$. Consider the constant sheaf on $U$, $\mathbb{Z}_U$, given by $\mathbb{Z}_U(V)=\{\text{continuous maps }U\cap V\rightarrow \mathbb Z\}$, where $\mathbb Z$ is given the discrete topology. I've been struggling to derive from Yoneda's lemma the formula $\hom(\mathbb{Z}_U,F)=F(U)$. Is this a consequence of Yoneda? If so, how? If it doesn't follow from Yoneda, is it true at all? If not, how can one compute $\hom(\mathbb{Z}_U,\mathbb{Z}_V)?$

$\endgroup$
2
  • $\begingroup$ Yes, it is a formal consequence of Yoneda and the fact that sheafification and the free abelian group functor are left adjoints. $\endgroup$
    – Zhen Lin
    Mar 8, 2013 at 13:00
  • $\begingroup$ Actually, the formulation of the question is incorrect, which led to the two different answers. You write that $\mathbb Z_U$ is a sheaf on $U$, while $F$ is a sheaf on $X$; so $\mathop{\rm hom}(\mathbb Z_U,F)$ does not make sense. You must either restrict $F$ to $U$ or extend $\mathbb Z_U$ to $X$. In the first case, the answer is yes (Zhen Lin's comment), and in the second it depends on the choice of an extension (direct image or extension by zero). $\endgroup$
    – ACL
    Apr 24, 2015 at 8:08

3 Answers 3

4
$\begingroup$

This is not true; for example, take $X = \mathbb R^2$, $U = \mathbb R^2 \smallsetminus \{(0,0)\}$. Then your $\mathbb Z_U$ coincides with $\mathbb Z_X$, and $Hom(\mathbb Z_U, F)$ is $F(X)$, not $F(U)$.

For the formula to hold you have to take as $\mathbb Z_U$ the extension of the constant sheaf by $0$, which is a very different animal.

$\endgroup$
3
  • $\begingroup$ @Angelo, thanks! Indeed you're right, I must have confused both things, what would then be a direct description of $\mathbb{Z}_U(V)$ for the right $\mathbb{Z}_U$ you suggests? (The extension of the constant sheaf $\mathbb{Z}$ on $U$ to $0$ in the rest). $\endgroup$ Mar 8, 2013 at 15:25
  • 1
    $\begingroup$ Locally constant function $V \to \mathbb Z$ that are zero on $V\smallsetminus U$. $\endgroup$
    – Angelo
    Mar 8, 2013 at 15:39
  • $\begingroup$ You are very welcome. $\endgroup$
    – Angelo
    Mar 8, 2013 at 17:47
1
$\begingroup$

See [Tamme, Introduction to étale cohomology], p. 31, Remark (2.1.3).

$\endgroup$
4
  • $\begingroup$ And the sheafification of Tamme's $Z_U$ is my $\mathbb{Z}_U$ above, isn't it? $\endgroup$ Mar 8, 2013 at 12:55
  • 1
    $\begingroup$ You have to distinguish $j_!\mathbf{Z}$ from your $\mathbf{Z}_U$ above. $\endgroup$
    – user19475
    Mar 8, 2013 at 13:33
  • 1
    $\begingroup$ @Timo, thanks! Is there any easy direct description of the values $(j_!\mathbb{Z})(V)$? $\endgroup$ Mar 8, 2013 at 15:30
  • 1
    $\begingroup$ @Fernando: See [Hartshorne, Algebraic Geometry] Exercise II.1.19 ("Extending a sheaf by zero"). $\endgroup$
    – user19475
    Mar 8, 2013 at 15:35
0
$\begingroup$

Pierre Schapira has some very nice, and more elementary notes that could help you.

$\endgroup$

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.