Today’s main goal is the Leray theorem (though at the end I have to ask a question):
Theorem 1 Let
be a sheaf on
, and
an open cover of
. Suppose
for all
-tuples
, and all
. Then the canonical morphism
is an isomorphism for all
.
This seems rather useless, because the theorem presupposes the vanishing of (regular) cohomology on the covering. However, in many cases it turns out to be helpful. If is a separated scheme,
an open affine cover of
, and
quasi-coherent, it applies. The reason is that each of the intersections
are all affine by separatedness, so
has no cohomology on them by a basic property of quasi-coherent sheaves. This gives a practical way of computing sheaf cohomology in algebraic geometry. Hartshorne uses it to compute the cohomology of line bundles on projective space.
Another instance arises when is the sheaf of holomorphic functions over some Riemann surface
. In this case
is a covering of charts. It is a theorem (which I will eventually prove) that for any open subset of
(which any intersection of the
‘s is isomorphic to), the sheaf
has trivial cohomology.
With this, let’s now prove the result.
Proof of the Leray theorem
We’re going to establish this by induction on . For
, we know it already. Given
, pick an injection
where
is flasque (e.g. injective) and take the cokernel
:
Lemma 2 There is a functorial long exact sequence
This follows from the fact that there is an exact sequence of Cech complexes of sheaves as defined previously:
which leads to a short exact sequence of abelian groups of global sections, since
has trivial (regular) cohomology by assumption. In general, however, it is not true that Cech cohomology is a -functor, i.e. sends short exact sequences to long exact ones.
We have a big commutative (see below) diagram
Now for any (possibly
!), the rightmost terms are both zeros by flasqueness, since
has trivial cohomology by the long exact sequence. The left-most and second-to-left most maps are isomorphisms by the inductive hypothesis, so in
the left two arrows are isomorphisms. Thus is an isomorphism too, proving the Leray theorem.
What I still haven’t been able to figure out here is why the diagram is commutative. In particular, how do we know that the morphisms between Cech cohomology and derived functor cohomology commute with the boundary maps?
Leave a Reply