Motivated by my ignorance here, if $X$ is a projective toric variety, is $$H^m(X, \mathcal O_X) \cong \begin{cases} 0 & m > 0 \\ \mathbb C & m = 1 \end{cases} $$ as for $\mathbb P^n$?

  • $\begingroup$ I take it $H^m$ means dimension of m-th cohomology? $\endgroup$
    – Matt
    Dec 2, 2010 at 21:25
  • $\begingroup$ Whoops, I fixed it. $\endgroup$ Dec 2, 2010 at 23:11

1 Answer 1


Yes, this is true, at least for varieties over the complex numbers $\mathbb{C}$. Indeed, a toric variety over an algebraically closed field is rational (i.e., birational to projective space). In characteristic zero, rational connectedness is a birational invariant, so toric varieties are rationally connected. Finally, any rationally connected variety is $\mathcal{O}$-acyclic, which is the name for the conclusion that you want. See e.g. here for this last implication.

The conclusion might well hold more generally; I am not an expert in these matters. You may want to ask your question on MathOverflow if you are not satisfied with this answer.

  • 5
    $\begingroup$ It's true in finite characteristic as well. See section 3.5 in Fulton's "Toric Varieties" book for how to compute cohomology of line bundles on toric varieties over any field. I don't think he singles out the case of the trivial line bundle; but it is easy to plug it in and see that you get 0. $\endgroup$ Apr 28, 2011 at 19:39
  • $\begingroup$ Alternatively, one uses that the Hodge numbers $h^{p,0}$ are birational invariants. (I hope this comment might be useful to someone some day. Your answer definitely was useful for me.) $\endgroup$
    – Bernie
    Apr 21, 2018 at 13:45

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.