Linear Programming
The key fact
In a linear programming problem, the feasible region is the set of points satisfying every constraint at once, and a feasible solution is any point in it. The objective function being linear means it can only reach its maximum or minimum at a corner point of that region, never somewhere in the interior.
Solving a problem, step by step
- Name the decision variables, e.g. = chairs made and = tables made.
- Turn each limit in the problem into an inequality in and .
- Add the non-negativity constraints and .
- Write the objective function to maximise or minimise.
- Draw each constraint's line and shade the side that satisfies it. The overlap is the feasible region.
- Find the corner points and evaluate at each one.
Worked example: forming the constraints
A workshop makes chairs and tables. A chair needs 2 hours of carpentry and 1 hour of finishing. A table needs 1 hour of carpentry and 3 hours of finishing. Each day there are 10 hours of carpentry and 15 hours of finishing available. The profit is ₹300 per chair and ₹400 per table. How many of each maximises profit?
Solution: Let be the number of chairs and the number of tables. Carpentry gives and finishing gives , with . The objective is . These are the same constraints as the region below, so its corners are , , and :
| Corner | |
|---|---|
The maximum profit is ₹2500, from 3 chairs and 4 tables a day.
Worked example: a tie between two corners
The feasible region for , , has corners , , , and . For with , what condition on and makes the maximum occur at both and ?
Solution: The maximum can only be shared by two corners if takes the same value at both. Otherwise one would beat the other. Set them equal:
With , , so every point on the edge joining and gives the same value, . The maximum is reached along that whole edge, not just at its two corners.
Common mistakes
- Leaving out . Without them the region spills into negative quantities, which make no sense for chairs or tables.
- Shading the wrong side of a line. Test a point such as : if it satisfies the inequality, shade the side containing it.
- Trusting corners in an unbounded region. If the region is unbounded, a maximum may not exist. Check whether can grow without limit before quoting a corner value.