I know a local system is a locally constant constant sheaf. But why does a local system on the topological space $X$ correspond to $\tilde{X}\times_G V$, where $G$ is the fundamental group of $X$, $\tilde{X}$ is the universal covering space of $X$, and $V$ is a $G$-module? How do you recover the locally free sheaf from $\tilde{X} \times_G V$?


The group $G$ acts properly discontinuously on $\tilde{X}$, and so if $x$ is any point of $\tilde{X}$, it admits a neighbourhood $U$ s.t. that $U g$ is disjoint from $U$ if $g \in G$ is non-trivial. Thus the natural map from $U$ to $\tilde{X}/G = X$ is an embedding.

Thus the natural map from $U \times V$ to $\tilde{X}\times_G V$ is also an embedding, and so $\tilde{X}\times_G V$ is locally constant (i.e. locally a product).

More detailed remarks:

  • We should equip $V$ with its discrete topology

  • The object $\tilde{X}\times_G V$ is not itself actually a sheaf, but is rather the espace etale of a sheaf. To get the actual sheaf we consider the natural projection $\tilde{X}\times_G V \to \tilde{X}/G = X$, and form the associated sheaf of sections. Over the open set $U \hookrightarrow X,$ this restricts to the sheaf of sections to the projection $U\times V \to U$, whose sections are precisely the constant sheaf on $U$ attached to the vector space $V$. (Here is where we see that it is important to equip $V$ with the discrete topology.) Thus our original sheaf of sections is locally constant, as claimed.

  • 1
    $\begingroup$ so does all the local system arise in this way? Does this way more natural than the original way(the usual definition of sheaf)? $\endgroup$ – abc Dec 7 '10 at 7:35
  • 1
    $\begingroup$ @abc: Yes, all local systems arise in this way, and this a very natural way to think about them. Very often people will speak of the local system corresponding to a $G$-module $V$, and this is what they mean. $\endgroup$ – Matt E Dec 7 '10 at 8:58
  • $\begingroup$ Can I ask how one defines $\tilde{X}\times_G V $ ? The notation makes it seem like a pullback but I don't see any natural maps $\tilde{X}\to G$ or $V\to G$ ? Is it just the product modulo $(x,gv) = (xg,v)$ for $x\in\tilde{X},v\in V,g\in G$ ? $\endgroup$ – Max Jan 25 at 17:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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