Revision notes

Linear Programming: Practice

The key points, rules to remember, tables and diagrams from all 9 lessons, on one page for quick revision. To keep a copy, use your browser's Print and choose "Save as PDF".

Lesson 1

Formulating harder word problems: units, percentages and ratios

ProcessStandard (x)Deluxe (y)AvailableConstraint
Cutting5 min15 min2 h = 120 min5x + 15y ≤ 120, so x + 3y ≤ 24
Assembly30 min20 min4 h = 240 min30x + 20y ≤ 240, so 3x + 2y ≤ 24
Packing6 min6 min1 h = 60 min6x + 6y ≤ 60, so x + y ≤ 10

Convert to one unit, take percentages of the whole (x + y), turn |x − y| ≤ k into two inequalities, and test every translated constraint with a pair of numbers that should pass and one that should fail.

Key takeaways: most formulation errors come from units, from taking a percentage of the wrong quantity, and from reversing a ratio. Simplify each constraint by a common factor, and check it with numbers before moving on.

Read the full lesson

Lesson 2

Transportation and allocation problems

RouteAmountCost per tonne (₹)Condition from amount ≥ 0
A to Dx5x ≥ 0
A to Ey4y ≥ 0
A to F80 − x − y3x + y ≤ 80
B to D50 − x4x ≤ 50
B to E60 − y2y ≤ 60
B to F40 − (80 − x − y) = x + y − 405x + y ≥ 40

Transportation: let two shipments be x and y, express all the others through supplies and demands, make every shipment ≥ 0, then simplify the total cost before finding corners.

Key takeaways: a balanced 2 × 3 transportation problem has two free variables. All six non-negativity conditions are constraints, and the last route is the one most often forgotten. Simplify the cost to px + qy + constant before evaluating corners, and always state the answer as the full shipping plan.

Read the full lesson

Lesson 3

Iso-profit lines and corner points at speed

EdgeLineSlopeJoins
firstx + 3y = 24−1/3(0, 8) to (3, 7)
secondx + y = 10−1(3, 7) to (4, 6)
third3x + 2y = 24−3/2(4, 6) to (8, 0)

For a maximum with p, q > 0 on a region bounded by ≤ constraints, the optimal corner is where the slope −p/q of the objective lies between the slopes of the two edges that meet there.

Key takeaways: iso-profit lines are parallel with slope −p/q; slide them to find the optimum. Compare −p/q with the edge slopes to pick the corner, scale Z to small numbers, skip corners that are beaten in both coordinates, and confirm with a quick evaluation.

Read the full lesson

Lesson 4

Unbounded regions: exam traps and quick checks

If Z = a·(left side of constraint 1) + b·(left side of constraint 2) with a, b ≥ 0 and the constraints are ≥ c₁ and ≥ c₂, then Z ≥ ac₁ + bc₂. When this equals the best corner value, that value is the minimum.

Region and objectiveUsual outcome
≥ constraints in the first quadrant, minimise px + qy with p, q > 0minimum exists at a corner
same region, maximise px + qy with p, q > 0no maximum
same region, an objective with a negative coefficienttest carefully: often neither exists
region squeezed between two lines going off to infinitycheck the direction of travel; an optimum may exist

Key takeaways: on an unbounded region never write a corner value as the answer without a check. Use a non-negative combination of the constraints as a one-line proof of a minimum, use a feasible far-away point as a one-line proof that none exists, and turn maximising −Z into minimising Z.

Read the full lesson

Lesson 5

Multiple optima, infeasible problems and whole-number answers

A tie between two adjacent corners means the whole edge joining them is optimal: infinitely many solutions, not two.

Whole-number point2x + 3y ≤ 12?4x + y ≤ 10?Z = 3x + 2y
(2, 3) (rounded)13: no11: noinfeasible
(1, 2) (rounded down)8: yes6: yes7
(1, 3)11: yes7: yes9
(2, 2)10: yes10: yes10
(0, 4)12: yes4: yes8

Key takeaways: an objective proportional to an edge's constraint signals infinitely many optima along that edge. Prove infeasibility by bounding one constraint's expression over the other's region. For whole-number problems, test nearby lattice points, and never just round the LP answer.

Read the full lesson

Lesson 6

Writing the constraints from a picture, and the area of the feasible region

An edge through (x₁, y₁) and (x₂, y₂) has equation (y − y₁)(x₂ − x₁) = (x − x₁)(y₂ − y₁). Test an interior point to choose ≤ or ≥.

Key takeaways: find each edge's line from two corners; choose ≤ or ≥ by testing an interior point; the shoelace formula gives the area from the ordered corners; lattice-point counts need a column-by-column count; a constraint that every corner already satisfies is redundant for a bounded region; for an unbounded one, check the unbounded directions too.

Read the full lesson

Lesson 7

Diet and blending problems at speed

On an unbounded region the corner with the smallest Z is a candidate. It is the minimum only if no feasible point has a smaller value, which you can prove by adding multiples of the constraints.

CornerZ = 6x + 8y
(0, 8)64
(3, 2)18 + 16 = 34
(9, 0)54

Key takeaways: write a ≥ constraint for each requirement; discard constraints that every corner already satisfies; check the minimum by adding multiples of the constraints; a fixed total reduces the problem to one variable on a segment.

Read the full lesson

Lesson 8

Mixed maximisation problems: land, labour and advertising

CornerZ = 400x + 500y
(4, 0)1 600
(20, 0)8 000
(10, 10)4 000 + 5 000 = 9 000
(4, 13.6)1 600 + 6 800 = 8 400

List every corner, including the ones where a minimum-requirement line meets a resource line. Do not assume the corner where two resource limits meet is best.

Key takeaways: give each constraint a clear name (land, labour, budget); find every corner, including those formed by a minimum requirement; state the answer as a plan and a value; confirm by substitution.

Read the full lesson

Lesson 9

Checking answers: feasible is not optimal, and what a new constraint does

Adding a constraint can never improve a maximum or lower a minimum. If the old optimum still satisfies the new constraint, it is still the optimum.

Key takeaways: check feasibility, value and optimality separately; a feasible corner is not necessarily the best; adding a constraint cannot improve the optimum, and if the old optimum survives it remains optimal; if the old optimum is cut off, the new value is no better.

Read the full lesson

Privacy · Terms · Refunds · Delivery