Table of Contents
Fetching ...

On a geometric extremum problem for convex cones

Oleg Mushkarov, Nikolai Nikolov

TL;DR

This work analyzes the extremal geometry problem of minimizing the $(n-1)$-volume of the intersection of a pointed convex cone $K$ with a hyperplane through a fixed point $A$, under admissibility constraints. It develops a geometric stationarity criterion, showing that a hyperplane $\mathcal{H}$ is stationary iff $\overrightarrow{AH}=n\overrightarrow{AG}$, and characterizes stationary hyperplanes for hyperangles via equivalences involving the centroid and Monge point. Special attention is given to the non-negative orthant, where every $A\in K^o$ has a unique stationary hyperplane, explicitly described by a positive root of an irrational equation, and with a closed-form for the associated $(n-1)$-volume; in dimension two, the shortest cone-cut segment length is given by a sharp $a_1^{2/3}+a_2^{2/3}$ expression. These results illuminate how extremal hyperplane positions depend on cone geometry, yielding explicit constructions and length formulas in orthant settings and clarifying the difference between low and high dimensions ($n=2$ vs $n\ge3$).

Abstract

We discuss the optimization problem for minimizing the $(n-1)$-volume of the intersection of a convex cone $K$ in $\Bbb R^n$ with a hyperplane through a given point, first considered in \cite{We}. We give a geometric characterization of the stationary hyperplanes for this problem when $K$ is a hyperangle which partially answers a question posed in \cite{We}. Moreover, we study the location of the set $S$ of points for which there is a stationary hyperplane as well as the infimum of the $(n-1)$-volumes of cone segments of $K$ cut off by hyperplanes through a given boundary point of $K$. As a model example we study in detail the non-negative orthant of $\Bbb R^n$. In this case $S$ is its interior and we show that every point of $S$ lies in a unique stationary hyperplane, which we describe in terms of the unique real root of an irrational equation.

On a geometric extremum problem for convex cones

TL;DR

This work analyzes the extremal geometry problem of minimizing the -volume of the intersection of a pointed convex cone with a hyperplane through a fixed point , under admissibility constraints. It develops a geometric stationarity criterion, showing that a hyperplane is stationary iff , and characterizes stationary hyperplanes for hyperangles via equivalences involving the centroid and Monge point. Special attention is given to the non-negative orthant, where every has a unique stationary hyperplane, explicitly described by a positive root of an irrational equation, and with a closed-form for the associated -volume; in dimension two, the shortest cone-cut segment length is given by a sharp expression. These results illuminate how extremal hyperplane positions depend on cone geometry, yielding explicit constructions and length formulas in orthant settings and clarifying the difference between low and high dimensions ( vs ).

Abstract

We discuss the optimization problem for minimizing the -volume of the intersection of a convex cone in with a hyperplane through a given point, first considered in \cite{We}. We give a geometric characterization of the stationary hyperplanes for this problem when is a hyperangle which partially answers a question posed in \cite{We}. Moreover, we study the location of the set of points for which there is a stationary hyperplane as well as the infimum of the -volumes of cone segments of cut off by hyperplanes through a given boundary point of . As a model example we study in detail the non-negative orthant of . In this case is its interior and we show that every point of lies in a unique stationary hyperplane, which we describe in terms of the unique real root of an irrational equation.

Paper Structure

This paper contains 3 sections, 8 theorems, 18 equations.

Key Result

Proposition 1

For $n\ge 3,$ any two of the following three conditions imply the third one: (i) $\mathcal{H}$ is a stationary hyperplane for $A;$ (ii) $A'A_1=A'A_2=\dots=A'A_n;$ (iii) $H=M.$

Theorems & Definitions (15)

  • Proposition 1
  • proof
  • Proposition 2
  • proof
  • Proposition 3
  • proof
  • Proposition 4
  • Proposition 5
  • proof
  • Proposition 6
  • ...and 5 more