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's work over the field $\mathbb{C}$ of complex numbers, and let $X\subset \mathbb{P}^n$ be a projective variety. Let $\tilde{X}\subset \mathbb{P}^n$ be any small open neighborhood of $X$, in the complex topology.

Question: Do there exist many algebraic varieties $\subset \tilde{X}$ of dimension equal to the dimension of $X$?

Of course, $X$ itself lies in there, and one may apply some automorphism of $\mathbb{P}^n$ close to the identity to obtain some further examples. I would be interested in any construction of more subvarieties in $\tilde{X}$, or examples showing that in general these may not exist. Note that the degree of the subvarieties is allowed to be arbitrarily large, and that I do not want to consider only deformations of $X$ (which might be rigid).

share|cite|improve this question
up vote 14 down vote accepted

The answer is yes (as in the case of Sasha's answer we use ramified covers)

Proof. Let $X$ be any variety in $\mathbb CP^n$. Take a section $s_m$ of $O(m)$ on $X$ such that $s_m$ is not equal to $m$-th tensor power $s^{\otimes m}$ of any section $s$ of $O(1)$ restricted on $X$. Now, let $s_m^{\frac{1}{m}}$ be the multi-section of $O(1)$ on $X$. This multi-section defines a subvarity $X_m$ in the total space of $O(1)$ on $X$, that is the cover of $X$ of degree $m$.

Finally notice that there is a family of maps from the total space of $O(1)$ on $X$ to $\mathbb CP^n$ that sends the zero section of $O(1)$ on $X$ to $X$. The image of such a map in $\mathbb CP^n$ is just the union of all lines in $\mathbb CP^n$ that join a fixed point $p$ with all points of $X$, the point $p$ itself does not belong to the image. Then the image of $X_m$ in $\mathbb CP^n$ is the desired variety. END.

We used here the fact that $O(1)$ on $\mathbb CP^n$ can be embedded in $T\mathbb CP^n$ as a subsheaf (in various ways).

share|cite|improve this answer
Very nice! [Although you might rename $\tilde{X}$ ;-)] – Peter Scholze May 29 '11 at 9:48

Apart of the deformations of $X$ you can consider deformations of any subscheme supported on $X$ (that is of any subscheme $Y$ such that $Y_{red} = X$). Another choice is to consider deformations of any morphism $f:Y \to X \to {\mathbb P}^n$. Both ways will provide you with deformations of higher degree. Note that the second approach is very fruitfully used in the bend-and-break procedure.

EDIT: Let for example a $X$ be a rational curve. Then the tangent space to deformations of the embedding $X \to {\mathbb P}^n$ is $H^0(X,(T_{{\mathbb P}^n})_{|X})$ which is a globally generated vector bundle of rank $n$ and degree $(n+1)d$, where $d$ is the degree of $X$. So, the dimension of the tangent space is $n + (n+1)d$. On the other hand, if $Y$ is another rational curve and $f:Y \to X$ is an $r$-fold covering, then the same computation shows that the tangent space to deformations of $Y \to {\mathbb P}^n$ has dimension $n + (n+1)dr$ which is much bigger!

share|cite|improve this answer
This does potentially construct more examples, but I don't know how to prove that one such deformation problem actually admits nontrivial deformations. – Peter Scholze May 28 '11 at 17:34
I understand your argument, but a big tangent space does not mean a big deformation space, i.e. all deformations could be infinitesimal. Is there a way to see that you can actually get a deformation over $C[[t]]$? – Peter Scholze May 28 '11 at 18:38
@Sasha: In Mori's bend and break argument isn't he forced to reduce modulo some prime p and use Frobenius since if X has genus > 1 then the genus of Y will increase with r (so one cannot use Riemann-Roch to get a large deformation space)? – ulrich May 28 '11 at 18:41
@Peter: $T_{\mathbb{P}^n}$ is an ample vector bundle so when restricted to a $\mathbb{P}^1$ it is a sum of line bundles, each one of which is of positive degree. So $H^1$ vanishes and the deformation space is unobstructed. A similar argument works if $X$ is an elliptic curve since one can let $Y$ be an unramified covering of large degree. – ulrich May 28 '11 at 18:49
@ulrich. I agree, my statement that the same story has place for any $X$ was too optimistic. I have removed it. But still, sometimes it works. – Sasha May 28 '11 at 18:54

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.