Non-Degeneracy of Ration Normal Curve

131 Views Asked by At

We consider the so called 'Steiner construction', which can be looked up e.g. in this script e.g. which gives us an irreducible curve.

The construction works as follows: Let say we have $\Lambda_1, ..., \Lambda_n \cong \mathbb{P}^{n-2} \subset \mathbb{P}^{n} $ codimension two subspaces in $\mathbb{P}^{n} $. Let for every $i =1,...,n$ be $\{H_t^i \}_{t \in \mathbb{P}^{1}}$ the pencil of hyperplanes containing the plane $\Lambda_i$ with
additional assumption that for every $t_0 \in \mathbb{P}^{1}$ the planes $H_{t_0}^1,...H_{t_0}^n$ are independent, i.e., intersect in a unique point $p(t_0)$.

Consider the union of intersections of these hyperplanes:

$$ C:= \bigcup_{t \in \mathbb{P}^{1}} H_t^1 \cap ... \cap H_t^n $$

In the linked script is is claimed that this curve is a rational normal curve, i.e. projectively equivalent to the curve obtained as image of the Veronese map

$$ v_d: \mathbb{P}^1 \to \mathbb{P}^n, [X_0:X_1] \mapsto [X_0^n: X_0^{n-1}X_1:..., X_1^n]$$

In this question I wanted to discuss how to prove that the Steiner construction above gives indeed a rational normal curve defined in more 'conventional' sense as Veronese's image.

Here I want to prove a weaker statement inspired by a user's comment which is unfortunately meanwhile removed.
Question/Problem: How to show that this curve $C= \bigcup_{t \in \mathbb{P}^{1}} H_t^1 \cap ... \cap H_t^n \subset \mathbb{P}^n$ constructed above is non-degenerated, that means there is no hyperplane $\mathbb{P}^{n-1} \cong H \subset \mathbb{P}^n$, which contains $C$? If we clearly already know that $C$ ration normal curve, then it's clear, but let's play only with Steiner's construction as definition of $C$.

The problem is that the spaces $\Lambda_1, ..., \Lambda_n \cong \mathbb{P}^{n-2} \subset \mathbb{P}^{n}$ in the construction seem to be choosen arbitrary and that makes it hard for me to prove the claim.

A simple case I can solve myself is if the $\Lambda_1, ..., \Lambda_n \subset \mathbb{P}^{n}$ can be choosen as follows: Let $p_1,..., p_n \in C$ be points in general position, that is the linear closure $\overline{p_1,..., p_n}$ is a hyperplane $H$, that is a linear subspace of codimension one in $\mathbb{P}^n$.

Take $\Lambda_i := \overline{p_1,..., \hat{p_i},..., p_n}$ where $\hat{p_i}$ mean that this point was removed. Then one concludes that if there exist a hyperplane $H$ which contains $C$ then it is equals to $\overline{p_1,..., p_n}$. But then by construction it is easy to see that if we chose $t \in \mathbb{P}^1$ such that for all $i=1,..., n$ the hyperplane $H^i_t \neq H$, then the intersetion of $H_t^1 \cap ... \cap H_t^n$ with $H$ is empty, therefore $H_t^1 \cap ... \cap H_t^n \not \in H$.

But the Steiner construction assumes $\Lambda_1, ..., \Lambda_n \cong \mathbb{P}^{n-2} \subset \mathbb{P}^{n}$ to be pairwise distinct but otherwise arbitrary.
How can I show that non-degeneracy of $C= \bigcup_{t \in \mathbb{P}^{1}} H_t^1 \cap ... \cap H_t^n \subset \mathbb{P}^n$ in that case?
My idea was to find enough points $p_1,..., p_n$ with $p_i \in \Lambda_i$ which are in general position and then as above to show that there is a point $c \in C$ which isn't contained in $\overline{p_1,..., p_n}$.

Problem: Can I such points always be found? Are results/estimations known how many points of $C$ should a $\Lambda_i$ contain? How can I show that every $\Lambda_i$ contains a point $p_i$ such that $p_1,..., p_n \in C$ are in general position, i.e. span a hyperplane in $\mathbb{P}^{n}$?