1
$\begingroup$

This is just a feeling that I had and I am curious if it is totally wrong or true to some extent.

Let $X\subseteq \mathbb{P}^r$ be an integral hypersurface of degree $r-1$, which is not a cone. In this choice of numerology, we know that through every point of $X$ there is a line that is contained in $X$.

Let $p\in X$ and assume that there is an $n-$dimensional family of lines in $X$ that pass through $p$. Is it true that if $n$ is large enough, it will force $p$ to be a singular point?

If so, do we have an estimate for that number $n$?

$\endgroup$
1
  • 2
    $\begingroup$ Just a comment: If $X$ is not a cone, the lines in $X$ through $p$ fill up at most a codimension-1 subset of $X$, hence $n$ is at most $r-3$. But for example there are smooth cubic threefolds in $\mathbf P^4$ with Eckardt points, i.e. points where the tangent hyperplane section is a cone (over a curve, ie $n=1$). So $n=r-3$ is not sufficient in general. $\endgroup$ Jul 8, 2016 at 12:18

1 Answer 1

2
$\begingroup$

To expand the comment of @potentially dense: the answer is $r-2$, and it does not depend on the degree of $X$. In general, suppose $X\subset \mathbb{P}^r$ is a hypersurface, and $p$ a smooth point of $X$. Lines passing through $p$ and contained in $T_p(X)$ are parametrized by a $\mathbb{P}^{r-2}$; since $T_p(X)\not\subset X$, we see that lines passing through $p$ and contained in $X$ form a family of dimension at most $r-3$. Now if (and only if) $p$ is a Eckart point, i.e. $X\cap T_p(X)$ is a cone, the dimension is exactly $r-3$. And it is easy to construct a smooth hypersurface with such a point, for instance $X$ given by $X_0^{d-1}X_1+X_1^{d}+\ldots+ X_r^d=0$, with $p=(1,0,\ldots ,0)$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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