Definition · Linear inequality in two variablesA linear inequality in two variables is a statement of the form Ax + By < C, with ≤, > or ≥ equally allowed and A and B not both zero. Its solution set is the set of ordered pairs making it true, which is a half-plane, including its edge or not.
The line Ax + By = C splits the plane into two open regions. Every point in one region makes Ax + By larger than C, every point in the other makes it smaller, and points on the line make it equal. That is why one test point settles the whole region: the expression cannot change which side of C it falls on without crossing the line.
Drawing the region
- Replace the inequality sign with an equals sign and graph that line, by intercepts or by slope-intercept form, whichever “Graphing a Linear Equation”'s methods make easier here.
- Make it solid or dashed. Solid for ≤ and ≥, since those points are solutions; dashed for < and >, since they are not.
- Test one point off the line, preferably (0, 0). Substitute both coordinates into the original inequality.
- Shade the side that won. If the test point made the statement true, shade its side; if false, shade the other.
For 2x + 3y ≤ 12 the boundary meets the axes at (6, 0) and (0, 4), and it is solid. Testing (0, 0) gives 0 ≤ 12, which is true, so the shaded half-plane is the one containing the origin: the side below and to the left of the line.
Why the symbol alone cannot tell you the side
It is tempting to read > as "above" and < as "below". That works only when the inequality has already been solved for y with a positive coefficient, and it fails as soon as the coefficient of y is negative.
4x - 2y > 8
-2y > -4x + 8
y < 2x - 4 divided by -2, so the sign FLIPPED
test (0, 0): 4(0) - 2(0) = 0 0 > 8 is FALSE
test (3, 1): 4(3) - 2(1) = 10 10 > 8 is TRUE
test (0, -10): 4(0) - 2(-10) = 20 20 > 8 is TRUE
the symbol says >, the shaded region is BELOW the line y = 2x - 4
The written symbol is > and the region is below. Both facts are correct, and they are consistent because solving for y required dividing by −2, which reversed the inequality into y < 2x − 4. A test point reaches the same conclusion without needing the rearrangement at all, which is why it is the more reliable method.
Theorem · Multiplying or dividing by a negative reverses the signIf a < b and c is negative then ac > bc. This is the single rule from “Linear Inequalities” in Linear Equations and Inequalities that survives into two variables, and it is where a correct boundary line ends up with the wrong side shaded. Adding or subtracting anything, and multiplying by a positive, leave the direction alone.
When the boundary passes through the origin
y < 3x has the origin on its boundary, so (0, 0) cannot be the test point: it satisfies the equation, not the inequality, and it lies on the edge rather than in either region. Any other point works. Testing (1, 0) gives 0 < 3, which is true, so the shaded region is the one containing (1, 0), below and to the right of the line.
Example · Four points against one inequalityFor 4x − 2y > 8: (0, 0) gives 0, false. (3, 1) gives 10, true. (0, −4) gives 8, and 8 > 8 is false, so this point sits exactly on the dashed boundary and is not a solution. (1, 2) gives 0, false. Only (3, 1) is in the solution set, and (0, −4) is the one worth pausing on: it would be a solution if the symbol were ≥.
Reading a region back into an inequality
The reverse direction appears in modelling and in Systems and Matrices. Find the boundary line's equation from two of its points, choose ≤ or < from whether the line is solid or dashed, and then fix the direction by testing a point taken from inside the shaded region. The last step is the same substitution as before, run with the answer known and the symbol unknown.