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

I've been stuck for quite a while on what is probably a trivial problem. Let $X\subset\mathbb{P}^n$ be a smooth projective curve, and let $$\mathcal{I}=\{(p,q,r):p,q\in X,p\neq q,r\in\overline{pq}\}$$ (where $\overline{pq}$ is the line that joins $p$ and $q$) and let $$\mathcal{J}=\{(p,r):p\in X,r\mbox{ lies on the tangent line to }X\mbox{ at }p\}.$$ It is easy to see that $\mathcal{I}$ is a complex 3-manifold and $\mathcal{J}$ is a complex 2-manifold. Let $\alpha:\mathcal{I}\to\mathbb{P}^n$ so that $(p,q,r)\mapsto r$, and let $\beta:\mathcal{J}\to\mathbb{P}^n$ so that $(p,r)\mapsto r$.

Why is it that if $n\geq 4$ then there is a point in $\mathbb{P}^n$ that is not in the image of either function? I can see that the image of $\alpha$ has complex dimension at most 3 and the image of $\beta$ at most 2, but I can't see why their images can't cover all $\mathbb{P}^n$.

For reference, this was taken from Algebraic Curves and Surfaces by Rick Miranda, page 101.


share|cite|improve this question
up vote 3 down vote accepted

$\alpha(\mathcal I)$ is the secant variety of $X$. I think it is enough if you understand this one, the other is similar. Of course the actual proof that $\alpha(\mathcal I)\subsetneq \mathbb P^n$ is exactly the dimension count you're mentioning. So, I assume you're looking for intuition and/or a heuristic explanation of what's happening.

I suspect your problem is really that you can't see in dimension $4$. Don't take this as a criticism, I can't see in dimension $4$ either and whoever claims they can, they don't really mean it that way.

$r\in\alpha(\mathcal I)$ means that $r$ lies on a secant line of $X$, in other words, there is a line through $r$ that intersects $X$ in two points, say $p,q\in X$. Now let's fix $p$ for a moment and turn around the observation: which points in $\mathbb P^n$ will lie on a secant line through $p$? Well, the points of the secant lies. Duh. So, you can look at all the lines through $p$ and observe that the ones that are contained in $\alpha(\mathcal I)$ are the ones that connect other points of $X$ to $p$. Let's call the union of these lines $A_p$, that is, let $\mathcal I_p=\{(a,b,c)\in\mathcal I|a=p\}$ and let $A_p=\alpha(\mathcal I_p)$

Let's take a (random) hyperplane $H\simeq \mathbb P^{n-1}$. It is enough to prove that $H\not\subseteq \alpha(\mathcal I)$. What is $H\cap\alpha(\mathcal I)$? First of all $$\alpha(\mathcal I)=\bigcup_{p\in X} A_p.$$ Second of all, $H\cap A_p$ is the projection of $X$ to $H$ and hence it is an algebraic curve. As $p$ runs through the points of $X$ we get a $1$-parameter family of curves, but that cannot fill a space of dimension $n-1\geq 3$.

Of course, this is the same proof as counting the dimension of the secant variety, but perhaps something that you can imagine better.

share|cite|improve this answer
Thanks a lot; I realized that my problem is precisely that: I need to get used to working in more dimensions. Thanks again! – rfauffar Jun 22 '11 at 15:29

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.