2
$\begingroup$

$\newcommand{\complex}{\mathbb{C}}$ Let $Seg \subseteq \complex^M \otimes \complex^N$ be the set of elements of the form $v \otimes w$. It is well-known that a general linear subspace of dimension $(M-1)(N-1)$ intersects $Seg$ only in the point $\{0\}$, as this is the codimension of the algebraic variety $Seg$, and that every linear subspace of dimension $(M-1)(N-1)+1$ intersects $Seg$ at some non-zero point.

I would like to know if every linear subspace of dimension $(M-1)(N-1)+(M-1)$ contains in its intersection with $Seg$ a linear subspace of the form $H \otimes v$, where $H\subseteq \complex^M$ is a hyperplane and $v \in \complex^N$ is a non-zero vector.

More generally for any $k \in \{1,...,M-1\}$, does every linear subspace of dimension $(M-1)(N-1)+k$ contain in its intersection with $Seg$ a linear subspace of the form $H \otimes v$, where $H\subseteq \complex^M$ is a linear subspace of dimension $k$?

$\endgroup$
3
  • 1
    $\begingroup$ For $k \geq 2$, the answer is no. By Bertini the intersection is smooth for a general linear space(say $L$). By Lefschetz (with $k \geq 2$), the intersection must be connected. Since it is smooth, it is irreducible. Hence we must have $Seg \cap L = v \otimes H$, scheme-theoretically. This is impossible for degree reasons. $\endgroup$
    – Libli
    Commented May 28, 2022 at 21:40
  • $\begingroup$ PS: I translated everything in projective space. $\endgroup$
    – Libli
    Commented May 28, 2022 at 21:43
  • $\begingroup$ @Libli Thank you! $\endgroup$
    – Ben
    Commented May 29, 2022 at 5:25

0

You must log in to answer this question.

Browse other questions tagged .