Linear Inequalities

A linear inequality is a relation of the form $ax+b<0$ (or with $\le ,>,\ge$) in one variable, or $ax+by<c$ in two variables, where the variable appears only to the first power and never multiplied by another variable. Unlike a linear equation, which typically has one solution, a linear inequality has infinitely many solutions forming a connected set: an interval on the number line, or a half-plane in the coordinate plane. They matter because most real-world conditions are constraints ("spend at most \$50," "carry at least 3 passengers") rather than exact equalities, making inequalities the natural language of optimization and feasibility. The surprising fact most students miss: multiplying or dividing both sides by a negative number reverses the inequality sign — a rule with no analogue in ordinary equation solving.

The Core Insight

An inequality describes a boundary and a direction. Solving it means finding the boundary point (or line) where the two sides are equal, then determining which side satisfies the relation. The entire difficulty reduces to two questions: Where is the boundary? and Which side? Every operation you perform must preserve — or deliberately flip — the direction of that comparison.

Formal Statement

Linear Inequality in One Variable

A linear inequality in one variable $x$ is any statement equivalent to one of:

where $a,b\in \mathbb{R}$ and $a\ne 0$. The solution set is the set of all real numbers $x$ making the statement true.

Linear Inequality in Two Variables

A linear inequality in two variables has the form

with $a,b$ not both zero. Its solution set is a half-plane bounded by the line $ax+by+c=0$.

Order-Preserving and Order-Reversing Operations

For real numbers $u,v,k$ with $u<v$:

Why It Holds / Derivation

Why the Sign Flips Under Negative Multiplication

We prove the key result of equation [5] from the definition of order on $\mathbb{R}$.

Joseph Fourier | 1768–1830

Though celebrated for the Fourier Series, Joseph Fourier also pioneered the systematic study of systems of linear inequalities. Around 1826 he described a method — now called Fourier–Motzkin elimination — for eliminating variables from a system of inequalities, the inequality analogue of Gaussian elimination. His work anticipated linear programming by over a century and established inequalities as objects worthy of algebraic theory, not mere afterthoughts to equations.

Building Intuition

Think of an equation $ax+b=0$ as asking "where does the line cross zero?" — a single dot. An inequality asks "where is the line below (or above) zero?" — an entire stretch.

For one variable, picture a number line. The boundary point splits it into two rays; you shade the ray that works. For two variables, the line $ax+by+c=0$ splits the plane into two halves; you shade the half that works. Testing a single point not on the boundary tells you which side: if it satisfies the inequality, shade its side; if not, shade the other.

The sign-flip rule becomes intuitive with a concrete example: $2<4$ is true. Multiply both by $-1$: is $-2<-4$? No! On the number line $-2$ sits to the right of $-4$, so $-2>-4$. Negation reflects the number line about zero, reversing all left-right (order) relationships.

Solution of $-2x + 1 \le 5$ shown on a number line, giving $x \ge -2$

Worked Examples

One Variable with a Sign Flip

Two Variables — Half-Plane and a Budget Constraint

Common Misconceptions

Three Frequent Errors

1. Forgetting to flip the sign when multiplying or dividing by a negative. This is the single most common mistake. Equations never require this — inequalities do.
2. Flipping the sign when adding/subtracting a negative. Adding $-5$ to both sides does not flip anything; only multiplying/dividing by a negative does. Equation [3] holds for any $k$, positive or negative.
3. Solid vs. dashed boundary. Use a solid line/filled dot for $\le ,\ge$ (boundary included) and a dashed line/open dot for $<,>$ (boundary excluded).

Why "Just Solve It Like an Equation" Fails

Consider $-2x<6$. Treating it as an equation and ignoring the sign rule gives $x<-3$. Test $x=0$: $-2(0)=0<6$ ✓ — but $0<-3$ is false! The naïve answer excludes a valid solution. The correct step is to divide by $-2$ and flip:

Connections

Linear inequalities sit at the crossroads of several topics. They generalize Linear Equations by replacing $=$ with an order relation, and their two-variable form prepares directly for Systems of Linear Inequalities, where multiple half-planes intersect. The optimization of a linear objective over such a region is precisely Linear Programming. The number-line solution sets are expressed using Interval Notation, and compound inequalities connect to Set Theory via union and intersection. Absolute-value inequalities like $|x-3|<5$ unpack into two linear inequalities, linking to Absolute Value.

Real-World Applications

1. Budgeting & finance: "Total expenses $\le$ income" is a linear inequality; combinations of purchases form a feasible region.
2. Manufacturing constraints: A factory's labor-hours and material limits ($3x+2y\le 120$) define producible product mixes — the basis of Linear Programming.
3. Nutrition & diet planning: "At least 50 g protein, at most 2000 calories" becomes a system of inequalities solved for meal portions.
4. Civil engineering tolerances: A beam's load must satisfy $\text{stress}\le \text{yield strength}$, a safety inequality with a margin factor.
5. Speed limits & physics: $v\le 30\text{m/s}$ or stopping-distance conditions $d\ge v{t}_{\text{react}}$ are everyday linear inequalities.
6. Computer graphics: Determining whether a pixel lies inside a polygon uses half-plane tests $ax+by+c\ge 0$ for each edge.
7. Machine learning (linear classifiers): A perceptron separates data with the inequality $\mathbf{w}\cdot \mathbf{x}+b>0$, assigning class by which side a point falls on.

Historical Development

• c. 300 BCE — Greek geometers (Euclid) use comparison of magnitudes ("greater than," "less than") rigorously, though without symbolic notation.
• 1631 — Thomas Harriot's posthumous Artis Analyticae Praxis introduces the symbols $<$ and $>$ for "less than" and "greater than."
• 1734 — Pierre Bouguer introduces $\le$ and $\ge$ (the "or equal to" variants) in their modern form.
• 1826 — Joseph Fourier presents a method to eliminate variables from systems of linear inequalities (later Fourier–Motzkin elimination).
• 1873 — Paul Gordan and contemporaries study systems of linear inequalities in the theory of invariants and polyhedral cones.
• 1936 — Theodore Motzkin's doctoral thesis systematizes the theory of linear inequalities, reviving Fourier's elimination method.
• 1939–1947 — Leonid Kantorovich and George Dantzig develop Linear Programming; Dantzig's simplex method (1947) solves systems of linear inequalities to optimize objectives.
• 1948 — Linear programming is applied to wartime/postwar logistics and economics, cementing inequalities as central to applied mathematics.

Subtleties & Limits

Beyond the Linear Case

While a single linear inequality is trivial to solve, systems of many linear inequalities define convex polytopes whose complexity grows rapidly with dimension. Counting and describing the vertices of such regions, and finding the optimal vertex efficiently, drove decades of research; the simplex method works well in practice but has exponential worst-case behavior, while the interior-point methods (Karmarkar, 1984) achieve provable polynomial time. When variables are required to be integers (integer linear programming), the problem becomes NP-hard — no known efficient general algorithm exists. Thus the humble linear inequality, when stacked and constrained to integers, sits at the frontier of computational complexity theory.

See Also