$a\in X$. Show that the stalk $\mathscr{F}_a$ is isomorphic to the stalk of $\mathscr{F}|_U$ at $a$ on the topological space $U$.

216 Views Asked by At

This is exercise 3.24 from Gathmann's notes.

Let $\mathscr{F}$ be a sheaf on a topological space $X$, and let $a\in X$. Show that the stalk $\mathscr{F}_a$ is a local object in the following sense: if $U\subset X$ is an open neighborhood of $a$ then $\mathscr{F}_a$ is isomorphic to the stalk of $\mathscr{F}|_U$ at $a$ on the topological space $U$.

My attempt:

Define the ring isomorphism: $$\overline{(V,\phi)}\mapsto \overline{(V\cap U,\phi|_U)}$$ for any open subset $V\subset X$, that contains $a$.

  1. It is well-defined. If $(V,\phi)=(V',\phi')$, then $\phi=\phi'$ on some neighborhood of $a\in W\subset V\cap V'$. Thus they agree on $W\cap U$.
  2. We see that it is surjective trivially, because for any $\overline{(V,\phi)}$ in $U$, $V$ can also be considered as in $X$, together with the function $\phi$ defined on it.
  3. We show that it is injective. Suppose a function $\phi$ represents the zero function in $\mathscr{O}_{U,a}$, i.e., it is zero in a neighborhood of $a$. Here I am confused. In the case $X$ is not irreducible, can we say that the preimage has to be $0$? Or can we shrink this open neighborhood until it is in $U$?

Thank you for any help!

Also, I am self-learning this subject, sometimes I suspect whether my proof is completely nonsense. If someone could help verify the above proof it would be greatly appreciated!

1

There are 1 best solutions below

0
On BEST ANSWER

As requested in the comment, this is the proof in the case of a sheaf of sets.

The map $\overline{(V,\phi)}\mapsto\overline{(V\cap U,\phi_{|V\cap U})}$ is well-defined and surjective. This is done in the original post in the case of a sheaf of rings, but the proof never uses rings and works perfectly for sets.

Let us show that it is also injective. So assume $\overline{(V\cap U,\phi_{|V\cap U})}=\overline{(V'\cap U,\phi'_{|V'\cap U})}$. This mean that there is a neighborhood $W\subset V\cap U$ and $V'\cap U$ such that $(\phi_{|V\cap U})_{|W}=(\phi'_{|V'\cap U})_{|W}$. But this means that $\phi_{|W}=\phi'_{|W}$. So $\overline{(V,\phi)}=\overline{(V',\phi')}$ and the map is indeed injective.