Ma · Mathematics

How Grothendieck Vanishing Gives a Strict Degree Bound

Published By the 9to5Science editorial team

At a glance

  1. 1. Check the dimension

    Let d be the dimension of the noetherian scheme, or the dimension of the support of the coherent sheaf.

  2. 2. Apply vanishing

    Grothendieck vanishing says the cohomology groups vanish in every degree i greater than d.

  3. 3. Read the inequality

    The theorem excludes degrees above d; it does not generally exclude degree d itself.

  4. 4. Check sharpness

    A projective plane example has nonzero second cohomology, showing the dimension bound can be attained.

Advertisement

Grothendieck vanishing is a statement about which degrees of sheaf cohomology can be nonzero. For a coherent sheaf on a noetherian scheme, cohomology vanishes strictly above the dimension of its support. The strict inequality matters: the theorem does not generally say that cohomology vanishes in the top degree itself.

The theorem and its bound

Let XX be a noetherian scheme and let F\mathcal{F} be a coherent sheaf on XX. A standard form of Grothendieck’s vanishing theorem states that

Hi(X,F)=0for i>dim⁡(Supp⁡F).H^i(X,\mathcal{F})=0 \qquad \text{for } i>\dim(\operatorname{Supp}\mathcal{F}).

In particular, since Supp⁡F⊆X\operatorname{Supp}\mathcal{F}\subseteq X, this implies

Hi(X,F)=0for i>dim⁡X.H^i(X,\mathcal{F})=0 \qquad \text{for } i>\dim X.

If dim⁡X=d\dim X=d, the theorem rules out nonzero cohomology in degrees d+1,d+2,…d+1,d+2,\ldots. Equivalently, the cohomological dimension for this sheaf is at most dd. It does not assert that the cohomological dimension is strictly less than dd: Hd(X,F)H^d(X,\mathcal{F}) may be nonzero.

The support version can give a stronger result when the sheaf is concentrated on a smaller closed subset. For example, if XX has dimension 55 but F\mathcal{F} is supported on a closed subset of dimension 22, then the theorem gives vanishing for every i>2i>2.

How to use the strict inequality

To apply the theorem in a problem, identify the relevant dimension and compare it with the degree you want to rule out:

  1. Check that the hypotheses apply, usually that XX is noetherian and F\mathcal{F} is coherent.
  2. Find d=dim⁡(Supp⁡F)d=\dim(\operatorname{Supp}\mathcal{F}); using dim⁡X\dim X also gives a valid, possibly weaker, bound.
  3. If the degree satisfies i>di>d, conclude directly that Hi(X,F)=0H^i(X,\mathcal{F})=0.
  4. If i=di=d, the theorem alone gives no conclusion. You need another argument or a stronger hypothesis.

The cohomology degree is an integer. Thus saying “vanishes for i>di>d” means the first excluded degree is d+1d+1, not dd.

Worked example: the bound can be sharp

Take X=Pk2X=\mathbb{P}^2_k, the projective plane over a field kk, and let F=OP2(−3)\mathcal{F}=\mathcal{O}_{\mathbb{P}^2}(-3). Here dim⁡X=2\dim X=2, so Grothendieck vanishing guarantees

Hi(Pk2,O(−3))=0for i>2.H^i(\mathbb{P}^2_k,\mathcal{O}(-3))=0 \qquad \text{for } i>2.

But the top-degree group is nonzero. Serre duality on P2\mathbb{P}^2 gives

H2(Pk2,O(−3))≅H0(Pk2,O)∗≅k.H^2(\mathbb{P}^2_k,\mathcal{O}(-3)) \cong H^0(\mathbb{P}^2_k,\mathcal{O})^* \cong k.

So the cohomology does not vanish in degree 2=dim⁡X2=\dim X. This example shows why “strict bound” should be read as vanishing above the dimension, not as vanishing at or above it.

Common mistakes

  • Replacing i>di>d with i≥di\ge d. This strengthens the theorem and is generally false, as the projective-plane example shows.
  • Confusing a bound with an exact value. The theorem gives an upper bound on possible nonzero degrees; it does not guarantee that the top allowed group is nonzero.
  • Using the dimension of XX when the support is smaller. The support form can give a sharper bound.
  • Ignoring hypotheses or changing settings. The statement quoted here is the standard coherent-sheaf version for noetherian schemes. If a problem concerns a different kind of space or sheaf, check the version of the vanishing theorem that applies before using the same bound.

For comparison, if XX is affine and F\mathcal{F} is quasi-coherent, the stronger affine vanishing theorem gives Hi(X,F)=0H^i(X,\mathcal{F})=0 for every i>0i>0. That is a different result and can be stronger than the dimension bound.

References

  1. Robin Hartshorne, Algebraic Geometry, Chapter III, Theorem 2.7
  2. The Stacks Project, Cohomology of Schemes

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.