Abhijeet Mulgund's Personal Webpage

Search

Search IconIcon to open search

Optimal Solution of Bounded Linear Program is at Vertex

Last updated Nov 1, 2022

# Statement

Suppose $n, m \in \mathbb{N}$, $\mathbf{b}, \mathbf{c} \in \mathbb{R}^{n}$, $A \in \mathbb{R}^{m \times n}$. Consider the Linear Program $$\begin{align*} &&\max \mathbf{c}^{T} \mathbf{x}\\ &\text{s.t.}&A \mathbf{x} \leq \mathbf{b}\\ &&\mathbf{x} \geq 0 \end{align*}$$ Suppose there exists an optimum $\mathbf{x}^{} \in \mathbb{R}^{n}$. Then there is a Polytope Vertex $\mathbf{v}^{} \in P$ of the Polytope $P$ defined by $A \mathbf{x} \leq \mathbf{b}$ s.t. $\max \mathbf{c}^{T} \mathbf{v}^{} = \max \mathbf{c}^{T} \mathbf{x}^{}$. That is, $\mathbf{v}^{*}$ is an optimal solution.

# Proof

TODO - this requires convexity of the Polytope and Open Halfspaces. Alternatively, could use Optimal Solution Set of Linear Program is a Face of Constraint Polytope.