To solve a system of linear inequalities, graph each inequality on the same coordinate plane, then shade the region where all the shaded areas overlap. That overlap, called the feasible region, is the solution set: every point inside it satisfies all inequalities at once, and boundary lines are solid when the inequality includes equality (≤ or ≥) and dashed when it does not.

What follows works the same whether you face two inequalities or six. The difference is bookkeeping, not new mathematics.

The mechanism: why the overlap is the answer

An inequality like 2x + y ≤ 6 divides the plane into two halves. The boundary line 2x + y = 6 contains every point that makes the two sides equal. Points on one side make the left side smaller; points on the other side make it larger. Because the inequality demands "smaller or equal," the solution is one closed half-plane, boundary included.

A single inequality therefore describes half the plane, which is why one inequality never has a single point as its answer. A system asks for points that satisfy every inequality simultaneously. That requirement is a logical AND, and AND translates to intersection. The solution region is the overlap of half-planes.

Two properties make the method fast in practice. First, each boundary is a straight line, so each region is convex: any segment joining two points in a half-plane stays inside it. An intersection of convex sets is also convex, so the feasible region has no dents or holes. Second, the region is always a polygon, bounded or unbounded, with corners where boundary lines cross. Those corners are where the interesting values live, which is why corner points matter later.

Plain language: each inequality votes for one side of a line, and the solution is the territory every voter approves.

Quantifying the region: vertices, bounds, and size

The corners carry real information. In a two-variable system, each vertex is the intersection of two boundary lines, so you find it by solving a 2×2 system of equations. With three inequalities you have at most three candidate intersection points; with five, at most ten, because each pair of boundary lines can meet at most once. You test each candidate in all original inequalities and keep the ones that survive.

Feasibility has a precise test. A system is infeasible when the surviving set is empty, as with y > x + 1 paired with y < x − 1: parallel boundaries, opposite directions, no overlap at any distance. It is unbounded when the region runs to infinity in some direction, which two constraints can produce but a compact system of four or more usually cannot.

Boundedness matters because of the corner theorem. A linear objective such as profit = 40x + 30y over a bounded feasible polygon attains its maximum and minimum at vertices. Evaluating four to six corners beats testing interior points. This is the seed of linear programming, and it is why textbooks make you find vertices precisely rather than eyeball them.

Verdict: the overlap is not a picture, it is a computable object with named corners.

Edge cases: where careful solvers separate from careless ones

The strict-versus-nonstrict boundary is the most common trap. Multiplying or dividing an inequality by a negative number flips its direction: −2x < 6 becomes x > −3. Forget that flip and your half-plane lands on the wrong side of the line, producing a feasible region that looks plausible and is completely wrong.

Parallel boundaries are next. If two boundary lines have the same slope but different intercepts, the half-planes either fully contain one another or miss entirely. The system has a solution region only when the containment direction agrees; otherwise it is empty.

Unbounded regions deserve respect. If x ≥ 0, y ≥ 0, and x + y ≥ 4, the feasible set is an infinite wedge. You can state that clearly and move on; pretending it is a triangle is the error.

Red flags to check before trusting any region:

A negative multiplier was applied without flipping the inequality symbol. The test point (0, 0) satisfied a strict inequality, yet the boundary was drawn solid. Two boundary lines look parallel but you assumed they cross. The shaded region contains points you can verify violate an original inequality. An objective value was computed at a corner that lies outside one of the constraints.

Common misconceptions, corrected

Misconception one: shading everything looks messy, so the solution is the union of shaded areas. It is the intersection, not the union. A point must pass every test, not just one.

Misconception two: the boundary line is automatically part of the solution. Only for ≤ and ≥. For < and >, the line itself is excluded, so strictly speaking the solution region is open along that edge.

Misconception three: more inequalities mean a smaller region. Usually true for compatible constraints, but adding a redundant inequality changes nothing. The line x + y ≤ 100 is redundant when x + y ≤ 50 is already required.

Misconception four: a system of inequalities has one numerical answer. It has infinitely many, unless the region collapses. Two opposite strict inequalities can even produce an empty set, which is a legitimate answer.

Approach Best for Weakness
Graph and shade Two to four inequalities, visual intuition Imprecise vertices without algebra
Test-point method Quick side selection Still needs accurate boundary lines
Elimination on boundary pairs Exact vertices for linear programming Combinatorial work grows quickly
Substitution One variable already isolated Awkward with strict inequalities
Algebraic interval reasoning One-variable systems Does not extend to two variables
Software or graphing calculator Many constraints, checking work Hides the reasoning students must show

The table rows differ in what they optimize: speed, precision, or transparency. A professional check is to solve once by shading and once by corners, then compare.

Misconceptions about one-variable systems

The same logic compresses to a number line, and the compression creates its own errors. The system x > 2 and x < 5 has the interval (2, 5), and the overlap rule is identical to the planar version. Strict versus nonstrict changes the brackets: x ≥ 2 and x ≤ 5 gives [2, 5].

The classic failure is a direction flip: x − 3 > 4 gives x > 7, but 3 − x > 4 gives x < −1. Students who flip only sometimes are guessing. The rule is mechanical: multiply or divide by a negative, reverse the symbol.

Redundancy appears here too. Adding x > −10 to x > 2 changes nothing; the tighter constraint wins. Recognizing redundancy early saves work in both one and two variables.

Practical actions: a repeatable procedure

Follow these steps in order and the method becomes reliable.

Rewrite each inequality in slope-intercept form, y ≤ mx + b, flipping symbols correctly when dividing by negatives. Graph each boundary line. Use a solid line for ≤ and ≥, a dashed line for < and >. Choose a test point, usually (0, 0) unless it lies on a boundary, and shade the side that satisfies the inequality. Find every pairwise intersection of boundary lines by solving the 2×2 systems. Test each intersection point against all original inequalities and keep only survivors; these are the vertices. If an optimization question asks for a maximum or minimum, evaluate the objective at each surviving vertex and compare.

Practical thresholds: if you have more than four constraints, do the vertex algebra rather than guessing from a sketch. If a test point equals zero on a boundary, pick (1, 1) instead. If two vertices come out identical, you likely have parallel boundaries that coincide, which is worth flagging.

This is also where the counterintuitive detail lives. The point (0, 0) is not a special point; it is just convenient. A region can exclude it entirely while still being large and bounded, so never infer infeasibility from the origin alone.

Limitations: what the method cannot tell you

Graphing and shading give the correct region for two variables, which is the entire curriculum of a typical algebra course. It does not scale to three variables visually, and beyond that you need simplex-style algebra. The corner theorem also assumes the objective is linear; non-linear objectives can attain optima on edges or in the interior.

Integer constraints are a separate problem. If x and y must be whole numbers, the feasible region's real vertices may not be feasible integer points, and the integer optimum can sit away from every corner. That is an integer programming problem, not a graphing problem.

Precision is another limit. Sketched lines produce approximate vertices, so final answers written from a sketch alone can be wrong in the second decimal. Algebraic intersection of boundary pairs is the corrective step.

This article is general educational information, not a diagnosis, treatment plan, legal opinion, or investment recommendation. It does not promise outcomes or returns. For decisions with real consequences, consult a qualified local professional. For immediate danger or self-harm risk, contact local emergency or crisis services.