MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

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 $A:={\mathbb{C}}[x_0,\ldots,x_n]=\oplus_{d=0}^{\infty} {\mathbb{C}}[x_0,\ldots, x_n]_d$ be the graded complex algebra of polynomials in $n+1$ variables, graded by degree.

Suppose $L$ is a line bundle over a projective manifold $X$ such that the ring of plurisections of $L$, i.e. $\oplus_{d=0}^{\infty} H^0(X,L^d)$ is isomorphic to $A$ as graded complex algebras.

Does that imply that $X$ is necessarily complex projective space ${\mathbb{CP}}^n$ and $L$ is the tautological line bundle?

share|cite|improve this question
thanks for this nice question. Both question and answer are useful – gauss Mar 18 '12 at 9:31
up vote 17 down vote accepted

The answer is no. Take any projective manifold $X$ mapping $\pi\colon X\rightarrow\mathbb P^n$ to such that $\pi_*\mathcal O_X=\mathcal O_{\mathbb P^n}$ and let $L$ be $\pi^*\mathcal O(1)$. Then $H^0(X,L^{\otimes m})=H^0(\mathbb P^n,\mathcal O(m))$ Examples are $X=Y\times\mathbb P^n$ or a blowing up of a closed smooth subvariety of $\mathbb P^n$.

I do not know if some power of $L$ is necessarily without base points (in which case $L$ itself is). In any case we can blow up $X$ such that there is a map to $\mathbb P^n$. The line bundle that gives that mapping may I guess have more sections.

Addendum: I was really rambling in the last paragraph. What I meant was that suppose we only know that $\bigoplus_nH^0(X,L^{\otimes n})$ is a polynomial ring. Does that mean that $L$ is base-point free? It may very well be true but I don't see it at the moment.

share|cite|improve this answer
your answer is unexpected and surprising! (not to say the least, very helpful). Thanks! – Colin Tan Nov 6 '10 at 13:58
@Torsten: thanks for the clarification. I was sure it had to be something different. :) This is a good question. Maybe you should pose it as a question. – Sándor Kovács Nov 6 '10 at 18:54
@Colin: here is why this is not surprising: The section ring of $\mathcal L$, $R=\oplus_n H^0(X,\mathcal L^n)$ describes $Y={\rm Proj}R$ and not $X$. Your condition simply means that the rational map given by some high power of $\mathcal L$ maps onto a projective space. Exactly as in Torsten's example. – Sándor Kovács Nov 6 '10 at 23:47
@Torsten: an example (I think) with a not-basepoint-free system is below. – Sándor Kovács Nov 6 '10 at 23:48

Here is an example for Torsten's Addendum: a line bundle with the required properties which is not semi-ample (no multiple is globally generated):

Take an $\pi:X\to \mathbb P^n$ as in Torsten's example, that is, with $\pi_*\mathcal O_X\simeq \mathcal O_{\mathbb P^n}$. Also assume that there is an exceptional divisor $E$, so let's just say it is a birational morphism contracting the divisor $E$ to a point. Now take $\mathcal L=\pi^*\mathcal O_{\mathbb P^n}(1)$ and consider the short exact sequence:

$$ 0\to \mathcal L^{\otimes n} \to (\mathcal L(E))^{\otimes n} \to \mathcal O_{nE}(nE) \to 0 $$ The sheaf on the right has no global sections, so the other two have the same $H^0$. Therefore $$ H^0(X,(\mathcal L(E))^{\otimes n}) = H^0(\mathbb P^n, \mathcal O_{\mathbb P^n}(n)) $$

Finally $(\mathcal L(E))^{\otimes n}$ cannot be globally generated, because $H^0(X,\mathcal L^{\otimes n})=H^0(X,(\mathcal L(E))^{\otimes n})$, so ${\rm supp}\,E$ is always in the base locus (or in other words, if it had a section whose zero section did not contain ${\rm supp}\,E$, then its restriction to $nE$ would give a nonzero section of $\mathcal O_{nE}(nE)$).

And here is what you can assert based on your condition:

Naive statement: Essentially the only reason your desired claim can fail is via Torsten's example.

More precise statement: Let $k$ be a field, $S:=k[x_0,…,x_n]$ and suppose $\mathcal L$ is a line bundle on a projective variety $X$ (over $k$) such that $R(X,{\mathcal L})=\oplus_{d=0}^\infty H^0(X,\mathcal L^d)\simeq S$ as graded $k$-algebras. Then there exists a diagram $$ X \overset\sigma\longleftarrow \widetilde X \overset\phi\longrightarrow \mathbb P^n $$ such that $\sigma$ is a projective birational morphism and $\phi$ is surjective. Furthermore, $\sigma^*\mathcal L\supseteq \phi^*\mathcal O_{\mathbb P^n}(1)$. Finally if $n=\dim X$, then $\phi$ is also birational and hence $X$ is a rational variety.

Proof: By the assumption ${\rm Proj}\, R(X,\mathcal L)\simeq \mathbb P^n$ and there exists a dominant rational map $\psi: X\dashrightarrow \mathbb P^n$. Let $$ X \overset\sigma\longleftarrow \widetilde X \overset\phi\longrightarrow \mathbb P^n $$ be the resolution of indeterminacies of $\psi$ via blow-ups on $X$. Then $\sigma$ is a projective birational morphism and hence $\widetilde X$ is also projective and $\sigma^*\mathcal L\supseteq \phi^*\mathcal O_{\mathbb P^n}(1)$. It follows that the induced morphism $\phi$ has to be surjective (since $\widetilde X$ is projective and the morphism is defined everywhere). Since this morphism is given by global sections of sufficiently high powers of a line bundle (i.e., $\phi^*\mathcal O_{\mathbb P^n}(m)$ for some $m>0$), it follows that it has connected fibers and $\phi_*\mathcal O_{\widetilde X}\simeq \mathcal O_{\mathbb P^n}$ and hence if $n=\dim X$, then $\phi$ is birational.

share|cite|improve this answer

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.