MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $X$ be a smooth projective variety (say, over a field of characteristic zero). Let us say that strong Kodaira vanishing holds for $X$ if $$ H^q(X,\Omega^p\otimes L)=0 $$ for every $p\geq 0$, $q>0$ and an ample line bundle $L$ on $X$.

My questions are now these:

1) Does strong Kodaira vanishing hold for $X={\mathbb P}^N$?

2) Does it hold for partial flag varieties of a semi-simple group $G$?

3) What tools are there for proving that strong Kodaira vanishing holds for a given variety $X$?

share|cite|improve this question
For flag varieties you can try to apply Borel--Bott--Weil. – Sasha Feb 16 '12 at 16:24
Sure, but I was not able to say that carefully for all $L$ (only for sufficiently positive ones) – Alexander Braverman Feb 16 '12 at 17:13

Perhaps I might add that the "strong Kodaira vanishing" holds more generally for smooth projective toric varieties in any characteristic. This goes back to Danilov. This includes your case 1 of course. I can't remember how he did this, but an argument observed by number of people (Fujino, myself,...) is to use a what I might call a mock Frobenius splitting argument. The idea is to exploit the map $\phi$ given by multiplication by $r$ on the fan. For projective space, this is just $[x_0,\ldots, x_N]\mapsto [x_0^r,\ldots, x_N^r]$. If $r>1$ is prime to the characteristic, then $\phi^*$ can be shown to give an injection $$H^q(X,\Omega_X^p\otimes L)\hookrightarrow H^q(X,\Omega_X^p\otimes L^r)$$ So choosing $r\gg 0$, we get the desired result by Serre vanishing.

share|cite|improve this answer
Thank you. There is actually a different argument in (using the lift of Frobenius mod $p^2$). – Alexander Braverman Feb 16 '12 at 19:37
up vote 5 down vote accepted

It turns out that questions 1 and 2 are completely answered here (and some technique for 3 is there as well). In particular, the statement is true for ${\mathbb P}^N$ but not true for most flag varieties.

share|cite|improve this answer
For flag varieties one has a much weaker "diagonal" vanishing of these cohomology groups, cf section 5.2 in Brion & Kumar's Frobenius splitting book (in particular Theorem 5.2.9). One interesting thing is that this diagonal vanishing actually holds for some non-ample line bundles $L$ as well as for all ample ones. – Chuck Hague Feb 17 '12 at 18:38

Re question 1: yes, see Deligne, Cohomologie des intersections compl`etes, SGA 7 II, th\'eor`eme 1.1. The proof is by "force brutale" as Deligne himself puts it, so I'm not sure this generalizes.

share|cite|improve this answer
Thank you. I wonder whether Frobenius splitting can be helpful here - it does prove everything for $q=0$ (and $q=\dim X$ is the usual Kodaira vanishing). – Alexander Braverman Feb 16 '12 at 17:14

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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