FINITE MATHEMATICS • LINEAR MODELS AND SYSTEMS

Graphical Optimization — Optimize an objective function over a feasible region (graphical method)

Find the best possible outcome by evaluating a linear objective function at the vertices of a polygonal feasible region.

Historical Context & Motivation

The idea of optimizing a quantity subject to constraints is as old as commerce itself, but the formal mathematical machinery behind linear programming crystallized only in the twentieth century. During World War II, military planners needed systematic methods to allocate scarce resources—fuel, personnel, munitions—across competing objectives. The graphical method of optimization, which visualizes constraints as half-planes in two dimensions and locates the optimum at a vertex of their intersection, became one of the earliest pedagogical tools for understanding how linear programs behave. Although industrial-scale problems involve thousands of variables and require algorithmic solvers, the two-variable graphical approach remains the conceptual foundation upon which all of linear programming rests. It offers geometric intuition that pure algebra cannot, making it indispensable in a first course on finite mathematics.

1827
Fourier's Inequality Systems
Joseph Fourier studied systems of linear inequalities, laying groundwork for the geometry of feasible regions and the elimination method that bears his name.
1939
Kantorovich's Resource Allocation
Soviet mathematician Leonid Kantorovich formulated economic production problems as linear programs, earning the 1975 Nobel Prize in Economics for this contribution.
1947
Dantzig's Simplex Method
George Dantzig introduced the simplex algorithm while working for the U.S. Air Force, providing a systematic procedure that generalizes the vertex-hopping logic visible in the graphical method.
1979–1984
Interior-Point Revolution
Khachiyan's ellipsoid method (1979) and Karmarkar's interior-point method (1984) proved that linear programs can be solved in polynomial time, expanding LP into large-scale industrial and engineering domains.

The central question that graphical optimization addresses is deceptively simple: given a set of linear inequalities that define a feasible region in the coordinate plane, where within that region does a linear objective function achieve its maximum or minimum value? The elegance of the answer—that the optimum always occurs at a corner point of the feasible region—is the cornerstone theorem we will explore in this lesson.

Core Principles & Definitions

Before we graph anything, we need to establish a precise vocabulary. A linear programming problem in two decision variables consists of an objective function to be maximized or minimized, subject to a finite collection of linear inequality constraints, along with non-negativity restrictions on the variables. The intersection of all half-planes defined by these constraints forms the feasible region—a convex polygon (or unbounded convex set) in ℝ². The vertices of this polygon are called corner points, and the fundamental theorem of linear programming guarantees that the optimum value of the objective function, if it exists, is attained at one of these corner points.

1

Objective Function

A linear expression Z = ax + by that we seek to maximize or minimize. The coefficients a and b encode the per-unit contribution of each decision variable to the quantity being optimized.
2

Feasible Region

The set of all points (x, y) satisfying every constraint simultaneously. Because each constraint carves out a half-plane, the feasible region is a convex polygon (or possibly empty or unbounded).
3

Corner-Point (Vertex) Theorem

If the feasible region is bounded and non-empty, then the objective function attains its maximum and minimum values at one or more corner points of the region.
4

Iso-Value Lines

Lines of the form ax + by = k (for varying k) along which the objective function has constant value. Sliding these lines across the feasible region reveals the direction of increasing or decreasing Z.
5

Convexity

The feasible region is convex: any line segment joining two feasible points lies entirely within the region. This geometric property is what forces the optimum to a vertex rather than an interior point.
KEY TAKEAWAY
Think of the feasible region as a fenced field and the objective function as the slope of a sheet of plywood resting on top of it. As you tilt the plywood (increase the objective), the last point of contact with the fence is always a corner post, not a section of fence rail. This is precisely why the optimal value occurs at a vertex—convexity ensures no interior point can outperform every corner.

Visualizing the Feasible Region & Iso-Value Lines

The graphical method becomes most transparent when we plot the constraints, shade the feasible region, and sweep a family of iso-value lines across it. The diagram below illustrates a standard two-variable linear program with three structural constraints and non-negativity conditions. Observe how the feasible region is a bounded convex polygon, and the dashed iso-value lines—each representing a constant value of the objective function—shift parallel to one another as the objective value changes. The optimum is reached at the last corner point the iso-value line touches before leaving the feasible region entirely.

The shaded convex polygon is the feasible region defined by three structural constraints and non-negativity conditions. Green dots mark the corner points. Dashed red lines are iso-value lines of the objective function Z; as Z increases, these lines sweep upward. The maximum of Z over the feasible region is attained at the last corner point the iso-value line touches.

In the diagram above, each constraint boundary line divides the plane into two half-planes; the feasible region is the intersection of all half-planes (including x ≥ 0 and y ≥ 0). The dashed iso-value lines all share the same slope—since Z = ax + by has the same coefficients regardless of the value of k—and they translate rigidly as k changes. Maximizing Z therefore amounts to pushing the iso-value line as far as possible in the direction of increasing Z while maintaining at least one point of contact with the feasible region. Geometrically, this last point of contact must be a vertex (corner point) of the polygon, unless the iso-value line happens to be parallel to an entire edge, in which case the optimum is achieved at every point on that edge—but crucially, it still includes a vertex.

Mathematical Framework

A two-variable linear programming problem in standard form can be stated as follows. We seek to optimize (maximize or minimize) a linear objective function subject to a system of linear inequality constraints and sign restrictions on the decision variables.

OBJECTIVE FUNCTION
Z = c₁x + c₂y
Here Z is the quantity to be optimized; x and y are decision variables; and c₁, c₂ are real-valued coefficients representing the per-unit contribution of each variable to the objective.
CONSTRAINT SYSTEM
a₁₁x + a₁₂y ≤ b₁, a₂₁x + a₂₂y ≤ b₂, …, aₘ₁x + aₘ₂y ≤ bₘ, x ≥ 0, y ≥ 0
Each structural constraint defines a half-plane. The coefficients aᵢⱼ and right-hand sides bᵢ are given constants. Constraints may also use ≥ or = depending on the problem formulation.
CORNER-POINT THEOREM
Z* = max{Z(v) : v ∈ V} or Z* = min{Z(v) : v ∈ V}
Let V = {v₁, v₂, …, vₙ} be the set of all corner points (vertices) of the feasible region. If the feasible region is bounded and non-empty, the optimal value Z* is obtained by evaluating Z at every vertex and selecting the largest (for maximization) or smallest (for minimization).

The Corner-Point Theorem rests on two properties: the convexity of the feasible region (intersection of half-planes is always convex) and the linearity of the objective function (a linear function on a convex set attains its extremes on the boundary, and more specifically at vertices). If the feasible region is unbounded, the objective function may or may not attain an optimum—maximization problems on unbounded regions can diverge to infinity, meaning no finite maximum exists. This subtlety is important to check in practice.

📐 Finding Corner Points
Each corner point of the feasible region is the intersection of exactly two boundary lines. To find all corner points, solve every pair of constraint boundary equations simultaneously and then verify that each solution satisfies all remaining constraints. Only solutions that lie within (or on the boundary of) the feasible region qualify as vertices.

Step-by-Step Procedure for Graphical Optimization

The graphical method can be distilled into a systematic five-step procedure. Each step has a clear algebraic and geometric interpretation, and skipping any one of them is a common source of errors. The following breakdown pairs each algorithmic step with the geometry it produces on the coordinate plane.

A flowchart of the five-step graphical optimization procedure, with detailed descriptions of each step and notes on special cases.

A few practical notes deserve emphasis. When graphing each constraint in Step 1, use the x-intercept and y-intercept of the boundary line for speed—set y = 0 to find the x-intercept and x = 0 to find the y-intercept, then connect them. In Step 2, the origin (0, 0) is a convenient test point as long as it does not lie on the boundary line itself. In Step 4, be sure to verify that each intersection point actually satisfies all constraints—intersection points of two boundary lines may lie outside the feasible region if a third constraint excludes them. Finally, organizing your vertex evaluations in a table (as in Step 5) minimizes arithmetic mistakes and makes the comparison transparent.

Worked Example

Consider the following linear programming problem: Maximize Z = 5x + 4y subject to the constraints x + y ≤ 6, 2x + y ≤ 10, x ≥ 0, and y ≥ 0. We will apply the five-step graphical method to locate the optimal solution.

Maximize Z = 5x + 4y
1
Step 1 — Graph the Boundary LinesFor x + y = 6: the x-intercept is (6, 0) and the y-intercept is (0, 6). For 2x + y = 10: the x-intercept is (5, 0) and the y-intercept is (0, 10). Plot both lines on the coordinate plane along with the axes x = 0 and y = 0.
2
Step 2 — Shade the Feasible Half-PlanesTest the origin (0, 0) in each inequality. For x + y ≤ 6: 0 + 0 = 0 ≤ 6 ✓, so shade the side containing the origin. For 2x + y ≤ 10: 0 + 0 = 0 ≤ 10 ✓, so again shade the side containing the origin. Combined with x ≥ 0 and y ≥ 0, the feasible region lies in the first quadrant below both lines.
3
Step 3 — Identify the Feasible RegionThe feasible region is a bounded convex polygon in the first quadrant. Its boundary is formed by segments of the lines x + y = 6, 2x + y = 10, the x-axis, and the y-axis.
4
Step 4 — Find the Corner PointsWe identify vertices by solving pairs of boundary equations. Intersection of x + y = 6 and 2x + y = 10: subtracting the first from the second gives x = 4, and substituting back yields y = 2, so the vertex is (4, 2). The remaining vertices lie on the axes: (0, 0) from the two axes, (5, 0) from 2x + y = 10 and y = 0, and (0, 6) from x + y = 6 and x = 0. Verify (0, 6): check 2(0) + 6 = 6 ≤ 10 ✓. Verify (5, 0): check 5 + 0 = 5 ≤ 6 ✓. All four vertices are feasible.
Corner points: (0, 0), (5, 0), (4, 2), (0, 6)
5
Step 5 — Evaluate Z at Each Corner PointZ(0, 0) = 5(0) + 4(0) = 0. Z(5, 0) = 5(5) + 4(0) = 25. Z(4, 2) = 5(4) + 4(2) = 20 + 8 = 28. Z(0, 6) = 5(0) + 4(6) = 24. Comparing all values, the maximum is 28, achieved at (4, 2).
Maximum Z = 28 at (x, y) = (4, 2)
Objective function values at each corner point of the feasible region
Corner PointZ = 5x + 4yOptimal?
(0, 0)0
(5, 0)25
(4, 2)28★ Maximum
(0, 6)24

Strengths, Limitations, and Special Cases

The graphical method offers unmatched visual clarity for problems with two decision variables, but it is important to understand its boundaries and the special cases that can arise. The table below summarizes the key strengths and limitations, followed by a discussion of three important special cases that every student should be able to recognize.

Comparative strengths and limitations of the graphical optimization method
StrengthsLimitations
Provides geometric intuition for convexity, feasibility, and the vertex theorem.Restricted to two decision variables; cannot handle three or more dimensions visually.
Reveals the direction of optimization via iso-value lines, building insight for the simplex method.Graphical accuracy is limited by plotting precision; exact coordinates require algebraic verification.
Easy to detect infeasibility (empty intersection) and unboundedness at a glance.Becomes impractical as the number of constraints grows, since the number of intersection points grows quadratically.
No specialized software needed; pen, graph paper, and algebra suffice.Not applicable to integer or nonlinear programming problems without modification.

Special Cases

  • Infeasible problem: If the constraints are mutually contradictory, the intersection of half-planes is the empty set. The problem has no feasible solution, and consequently no optimum exists. Graphically, no common shaded region appears.
  • Unbounded feasible region: When the feasible region extends infinitely in some direction, a maximum may not exist (Z → ∞). However, a minimum may still be achieved at a vertex. Always check whether the objective function is bounded in the direction of optimization.
  • Multiple optimal solutions: If the iso-value line at the optimum is parallel to a boundary edge of the feasible region, every point on that edge is optimal. The optimum value of Z is unique, but the optimal solution set is infinite—a line segment of equally good solutions.
KEY TAKEAWAY
The graphical method is to linear programming what a wind tunnel is to aerodynamics: it does not scale to production-size problems, but the physical intuition it provides is irreplaceable. Every concept you see in the two-dimensional picture—feasibility, convexity, vertex optimality, unboundedness, degeneracy—carries over directly into the simplex algorithm that handles thousands of variables.

Connection to the Simplex Method & Higher Dimensions

The graphical method teaches us the geometry behind optimization, but real-world problems typically involve far more than two decision variables. The natural successor to the graphical approach is the simplex method, developed by George Dantzig in 1947. The simplex algorithm performs algebraically exactly what the graphical method does geometrically: it starts at one vertex of the feasible polytope (the higher-dimensional analogue of a polygon), then moves along edges to adjacent vertices that improve the objective function, and stops when no further improvement is possible. In the graphical setting, you can literally trace this path from corner to corner on your graph.

Graphical method vs. simplex method
FeatureGraphical MethodSimplex Method
Number of variables2 (or 3 with difficulty)Any finite number
Representation of feasible regionConvex polygon in ℝ²Convex polytope in ℝⁿ (encoded in a tableau)
Corner-point identificationVisual + algebraic (solve 2×2 systems)Pivot operations on the simplex tableau
Optimum detectionCompare Z at all verticesOptimality condition (no positive indicator in objective row)
Primary useConceptual understanding and small problemsLarge-scale industrial and computational problems

Beyond the simplex method, modern optimization includes interior-point methods that traverse the interior of the polytope rather than its boundary, integer programming for problems where decision variables must be whole numbers, and sensitivity analysis for understanding how changes in coefficients or constraints affect the optimal solution. Each of these advanced topics rests on the geometric intuition you develop through the graphical method. Mastering two-variable optimization thoroughly prepares you to interpret the algebraic output of these more powerful tools.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain why a linear objective function on a bounded, non-empty convex feasible region must attain its maximum at a vertex rather than at an interior point. What geometric property of the objective function and the feasible region drives this result?
PROBLEM 2BASIC CALCULATION
Maximize Z = 3x + 2y subject to x + y ≤ 4, x + 3y ≤ 6, x ≥ 0, y ≥ 0. Find all corner points of the feasible region and determine the maximum value of Z.
PROBLEM 3INTERMEDIATE
Minimize Z = 2x + 5y subject to x + 2y ≥ 10, 3x + 2y ≥ 18, x ≥ 0, y ≥ 0. Identify the feasible region, find all vertices, and determine the minimum value of Z.
PROBLEM 4APPLIED
A small bakery produces two types of pastry: croissants and muffins. Each croissant requires 2 oz of flour and 1 oz of butter; each muffin requires 1 oz of flour and 1 oz of butter. The bakery has at most 120 oz of flour and 80 oz of butter available daily. The profit is $1.50 per croissant and $1.00 per muffin. Formulate this as a linear programming problem and use the graphical method to determine the production plan that maximizes daily profit.
PROBLEM 5CRITICAL THINKING
Consider the problem: Maximize Z = 4x + 2y subject to 2x + y ≤ 10, x ≥ 0, y ≥ 0. (a) Is the feasible region bounded or unbounded? (b) Does a maximum of Z exist? Justify your answer using the geometry of iso-value lines. (c) Now change the objective to Minimize Z = 4x + 2y over the same feasible region. Does a minimum exist? If so, where is it attained, and does the Corner-Point Theorem apply even though the region is unbounded?

Summary & Review

Graphical optimization solves two-variable linear programming problems by plotting linear constraints as boundary lines, shading the appropriate half-planes, and identifying the resulting feasible region—a convex polygon whose vertices are the corner points. The Corner-Point Theorem guarantees that the optimal value of the linear objective function Z = c₁x + c₂y is attained at one or more of these vertices, provided the feasible region is bounded and non-empty.

The method proceeds in five steps: graph each boundary line, shade the correct half-planes, identify the feasible region, solve for all corner points algebraically, and evaluate the objective function at each vertex to find the maximum or minimum. Special cases include infeasible problems (empty feasible region), unbounded regions (Z may diverge), and multiple optimal solutions (iso-value line parallel to an edge). While limited to two variables, the graphical method builds the geometric intuition essential for understanding the simplex algorithm and more advanced optimization techniques.

Varsity Tutors • Finite Mathematics • Graphical Optimization — Optimize an objective function over a feasible region (graphical method)