2
$\begingroup$

Let us consider a vector space $ V $ over $ \mathbb{Q} $ of dim $6$. We denote all the two dimensional subspace in $ V $ by $ G(2,6) $ (The Grassmanian variety). One can define a map $ p $ from $ G(2,6) $ to $P(\Lambda^{d} V) $ by $ p (U) = u_{1} \wedge u_{2} $, where $ U \in G(2,6) $ and $ u_{1},u_{2} $ be a basis of $ U $. Now we have $ 15 $ Plücker coordinates. We also know that $ G(2,6) $ is the zero set of a system of quadratic Plücker polynomials. Now we define $ 5 $ hyperplanes as a linear combination of Plücker coordinates : $ \alpha_{i} = \sum_{j,k} a_{jk}^{i} p_{jk} $ for all $ 1 \leq i \leq 5 $ , where $ p_{jk} $ are the Plücker coordinates. Also $ a_{jk} \in \mathbb{Q} $.

Now my question is 
 Can there exists 5  hyperplane in this above form such that their intersection with Grassmanian variety has no rational point? 
$\endgroup$
10
  • 1
    $\begingroup$ "zero" is not a point of $G(2,6)$. $\endgroup$
    – abx
    Commented Jul 28, 2021 at 9:41
  • $\begingroup$ Yes @abx you are right all the plucker co ordinate does not equal to zero . because zero not lies in projective space. okay I should edit my question. The question should be has no rational points in their intersections. $\endgroup$
    – Sky
    Commented Jul 28, 2021 at 12:46
  • 1
    $\begingroup$ This question is a special case of your previous question mathoverflow.net/questions/398253/… and my comments there might be helpful - specifically, showing that there can be no real obstruction to rational points, and giving two interpretations as a problem of finding rational points on a hypersurface. $\endgroup$
    – Will Sawin
    Commented Jul 28, 2021 at 18:03
  • 1
    $\begingroup$ @JasonStarr If I understand correctly, you can do this with a single hyperplane (the twist of the Grassmanian looks like triples quaternions modulo left multiplication by quaternions, linear sections correspond to $3 \times 3$ quaternionic Hermitian forms, and any positive definite such form does the trick). $\endgroup$
    – Will Sawin
    Commented Jul 28, 2021 at 18:39
  • 1
    $\begingroup$ I guess the cubic hypersurface $3$-fold I constructed is equivalent to the usual construction of a Fano of type $V_3$ from a Fano of type $V_{14}$, see e.g. The Abel–Jacobi Map for a Cubic Threefold and Periods of Fano Threefolds of Degree 14 A. Iliev and D. Markushevich math.uni-bielefeld.de/documenta/vol-05/03.pdf which also mentions the classical result that a generic cubic 3-fold arises this way. So it would suffice to find a sufficiently generic cubic 3-fold without a rational point. $\endgroup$
    – Will Sawin
    Commented Jul 28, 2021 at 18:40

1 Answer 1

3
$\begingroup$

The answer is positive. Indeed, the space of all codimension 5 linear sections of $G(2,6)$ is parameterized by $G(5,15)$. For each rational point $P \in G(2,6)$ the set of linear sections containing $P$ is a copy of $G(5,14) \subset G(5,15)$. Therefore, the set of all codimension 5 linear sections of $G(2,6)$ containing a rational point is a countable union of proper subvarieties. Taking any point in its complement, one obtains a linear section with no rational points.

EDIT. As @abx explains in the comments this does not answer the OP's question since it is absolutely non clear if the complement contains at least one point defined over $\mathbb{Q}$.

$\endgroup$
3
  • $\begingroup$ I am confused about your argument. It should apply to the intersection of 4 hyperplanes with the Grassmannian $G(2,5)\subset\mathbb{P}^{9}$. But a general such intersection is a del Pezzo surface of degree 5, which has always a rational point. Where am I wrong? $\endgroup$
    – abx
    Commented Jul 28, 2021 at 14:24
  • 1
    $\begingroup$ I guess the problem is that the OP wants his linear section of $G(2,6)$ to be defined over $\mathbb{Q}$. Your complement in $G(5,15)$ is a countable intersection of (Zariski) open subsets, there is no guarantee that it contains a rational point. $\endgroup$
    – abx
    Commented Jul 28, 2021 at 14:35
  • $\begingroup$ @abx: You are of course completely right, the question is much more subtle. $\endgroup$
    – Sasha
    Commented Jul 28, 2021 at 17:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .