Does there exist a set $M$ in usual Euclidean space such that for every plane $\lambda$ the intersection $M \cap \lambda$ is finite and nonempty ? Proposed by Hungary.
HIDE: Remark I'm not sure I'm posting this in a right Forum.Problem
Source:
Tags: algebra, polynomial, analytic geometry, Intersection, 3D geometry, IMO Shortlist
Jorge Miranda
22.08.2010 16:46
Let $M=\{(t^5,t^3,t):t\in\mathbb{R}\}$, and suppose $\lambda$ has $ax+by+cz+d=0$ as it's cartesian equation (with $a,b,c$ not all zero).
Then $M\cap \lambda$ is nonempty because $at^5+bt^3+ct+d=0$ always has a real solution (because a polynomial of odd degree always has a real root), and it's finite because $at^5+bt^3+ct+d=0$ has a finite number of solutions (because a non-constant polynomial has a finite number of roots).
pco
22.08.2010 16:50
Jorge Miranda wrote:
Let $M=\{(t^5,t^3,t):t\in\mathbb{R}\}$, and suppose $\lambda$ has $ax+by+cz+d=0$ as it's cartesian equation (with $a,b,c$ not all zero).
Then $M\cap \lambda$ is nonempty because $at^5+bt^3+ct+d=0$ always has a real solution (because a polynomial of odd degree always has a real root), and it's finite because $at^5+bt^3+ct+d=0$ has a finite number of solutions (because a non-constant polynomial has a finite number of roots).
Quite quite nice !!! Great congrats !
mavropnevma
22.08.2010 18:36
Notice how this is related to the famous moment curve $(t,t^2,t^3)$, used in solving Crum's problem of infinitely many neighbourly convex polyhedra (Tietze, Besicovitch). But of course we need each coordinate of odd degree ...