If an Optimal Solution Exists in Linear Programming Problems, One of Them is a Basic Feasible Solution
Theorem 1
$$ \begin{matrix} \text{Maximize} & \mathbf{c}^{T} \mathbf{x} \\ \text{subject to} & A \mathbf{x} = \mathbf{b} \\ & \mathbf{x} \ge \mathbf{0} \end{matrix} $$
Given matrices $A \in \mathbb{R}^{m \times n}$, $\mathbf{b} \in \mathbb{R}^{m \times 1}$, and $\mathbf{c} \in \mathbb{R}^{n}$, let’s assume a linear programming problem is represented in the form of an equation as above. If an optimal solution exists, then an optimal basic feasible solution also exists.
- $\mathbf{c}^{T}$ represents a transpose.
- A feasible solution refers to a solution that satisfies the constraints regardless of optimization.
Proof
Strategy: This proof is essentially the same as that of the existence of an optimal solution.
Definition of basic feasible solution: For a feasible solution $\mathbf{x} \in \mathbb{R}^{n}$, there exists a set $B \subseteq [n]$ such that its cardinality is $m$, $\exists A_{B}^{-1}$ and $x_{j} = 0 \forall j \notin B$ for $\mathbf{x} = \left( x_{1} , \cdots , x_{n} \right)$. We call this set $B$ a basis.
For any feasible solution $\mathbf{x}_{0}$, $$ \mathbf{c}^{T} \mathbf{x}_{0} \le \mathbf{c}^{T} \mathbf{x} $$ let’s call the solution with the most number of $0$ components among those satisfying $\mathbf{x}$, as $\tilde{\mathbf{x}}$.
- If we can show that $\tilde{\mathbf{x}}$ is a basic feasible solution, the existence of an optimal solution can be shown by the fact that there are finitely many basic feasible solutions.
- Thus, at least one of any optimal solutions, being a feasible solution before being optimal, must have the same objective function value as some (optimal) basic feasible solution. Therefore, at least one optimal solution is a basic feasible solution.
Now, let’s divide the cases depending on whether the column vectors of $A_{K}$ are linearly independent or not when the set of indices $$ K = \left\{ k \in [n] : \left( \tilde{\mathbf{x}} \right)_{k} > 0 \right\} $$ is defined.
Case 1. Linear Independence
Following the definition of a basic feasible solution, $\tilde{\mathbf{x}}$ naturally is a basic feasible solution, and there is nothing more to prove.
Case 2. Linear Dependence
If they are linearly dependent and not linearly independent, there exists some vector $\mathbf{v} \in \mathbb{R}^{|K|}$ of dimension $|K|$ satisfying: $$ A_{K} \mathbf{v} = \mathbf{0} $$ Considering an extended vector $\mathbf{w} \in \mathbb{R}^{n}$ that satisfies $\mathbf{w}_{K} = \mathbf{v}$ and $A \mathbf{w} = A_{K} \mathbf{v} = \mathbf{0}$, for instance, if $K = \left\{ 2,3,5 \right\}$ is $\mathbf{v} = (-2,1,7)$, then $\mathbf{w}$ becomes a vector $\mathbf{w}= (0,-2,1,0,7)$, whose $2,3,5$th component is similar to $\mathbf{v}$, and the rest are $0$. Now, by adjusting the sign of the component of $\mathbf{v}$, $\mathbf{w}$ can also satisfy: $$ \mathbf{c}^{T} \mathbf{w} > 0 $$ From now on, consider $\mathbf{w}$ as a vector satisfying the above inequality.
For $\mathbf{w}$, let’s consider the following vector $\mathbf{x}(t)$ dependent on $t \ge 0$: $$ \mathbf{x} (t) = \tilde{\mathbf{x}} + t \mathbf{w} $$ First, $\tilde{\mathbf{x}}$ is a feasible solution, and by the definition of $\mathbf{w}$ made to satisfy $A \mathbf{w} = A_{K} \mathbf{v} = \mathbf{0}$, for all $t \ge 0$ $$ \begin{align*} A \mathbf{x} (t) =& A \tilde{\mathbf{x}} + t A \mathbf{w} \\ =& A \tilde{\mathbf{x}} \\ =& \mathbf{b} \end{align*} $$ holds. In other words, $\mathbf{x} (t)$ also satisfies all constraints except for $\mathbf{x}(t) \ge 0$. Considering the $j$th component $x_{j}(t)$ of $\mathbf{x} (t)$, it appears as $x_{j}(t) = \tilde{x}_{j} + t w_{j}$, and at least one of them must be $w_{j} < 0$. If, for all $j$, $w_{j} > 0$ holds, then for all $t$, $\mathbf{x} (t) = \tilde{\mathbf{x}} + t \mathbf{w} \ge 0$ would be true, making all $\mathbf{x} (t)$ feasible solutions, which contradicts the boundedness assumption of the objective function since: $$ \mathbf{c}^{T} \mathbf{x} (t) = \mathbf{c}^{T} \tilde{\mathbf{x}} + t \mathbf{c}^{T} \mathbf{w} \to \infty $$ Hence, if at least one of $w_{j_{0}}$ is negative, there must exist a $t_{0} \ge 0$ that first satisfies: $$ x_{j_{0}} \left( t_{0} \right) = \tilde{x_{j_{0}}} + t_{0} w_{j_{0}} = 0 $$ Then, $\mathbf{x} \left( t_{0} \right)$ will have one more $0$ component compared to $\tilde{\mathbf{x}}$ while still satisfying: $$ \begin{align*} \mathbf{c}^{T} \mathbf{x} \left( t_{0} \right) =& \mathbf{c}^{T} \tilde{\mathbf{x}} + t_{0} \mathbf{c}^{T} \mathbf{w} \\ \ge & \mathbf{c}^{T} \mathbf{x}_{0} + t_{0} \mathbf{c}^{T} \mathbf{w} \\ \ge & \mathbf{c}^{T} \mathbf{x}_{0} \end{align*} $$ This contradicts the definition of $\tilde{\mathbf{x}}$, which was made to have the most number of $0$ components while satisfying $\mathbf{c}^{T} \mathbf{x}_{0} \le \mathbf{c}^{T} \mathbf{x}$.
Uniqueness of basic feasible solution: The basic feasible solution is uniquely determined by the basis $B$.
Consequently, since $A_{K}$ cannot be linearly dependent, $\tilde{\mathbf{x}}$ must be a basic feasible solution, and the basic feasible solution, determined by the choice of basis, being unique implies that both a basic feasible solution and an optimal solution exist finitely many, and therefore, for some optimal solutions, there exists a basic feasible solution with the same objective function value, making at least one of the optimal solutions a basic feasible solution.
■
See Also
The Fundamental Theorem of Linear Programming
This theorem is used in the proof of the fundamental theorem of linear programming.
Matousek. (2007). Understanding and Using Linear Programming: p46. ↩︎