University College London · FACULTY OF ECONOMICS

ECON0006 Chap.4 Constrained Optimisation

- one subject, every graph, every model, every mark
7 Chapters4-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 4 of 10 · ECON0006

Constrained Optimisation

Constrained Optimisation replaces unrestricted movement with a feasible region. Equality constraints create curves or surfaces, inequalities create regions, and corners may defeat a smooth tangency argument. A complete method describes the geometry, identifies active restrictions, generates smooth and boundary candidates, and compares feasible objective values.

Multiplier interpretation stays tied to the resource and notation used in the problem.

In this chapter

What this chapter covers

  • 01

    Feasible regions

  • 02

    Equality constraints

  • 03

    Constraint gradients

  • 04

    Lagrangian reasoning

  • 05

    Inequality boundaries

  • 06

    Corner solutions

  • 07

    Linear programming

Worked example · free

Choose between tangency and a corner

Q [5 marks]. An objective contour first touches a feasible polygon at a corner. Explain the candidate test. The weighting on this exercise is not an official university marking scheme; it only structures the candidate checks.
The shaded polygon is the feasible set; the highlighted boundary contact is a candidate because the next objective contour would lie outside the allowed region.The shaded polygon is the feasible set; the highlighted boundary contact is a candidate because the next objective contour would lie outside the allowed region.candidatefeasible region
The shaded polygon is the feasible set; the highlighted boundary contact is a candidate because the next objective contour would lie outside the allowed region.
  • 1Confirm the corner satisfies every restriction.
  • 2Evaluate all vertices and any smooth boundary candidates.
  • 2Select the feasible point with the largest objective value and name binding constraints.
The corner remains a valid candidate even without a smooth tangency. The weighting is original practice guidance.
Sia tip — List corners before applying an interior condition.
Glossary

Key terms

Feasible region
The set of choices satisfying all equality and inequality restrictions.
Binding constraint
A restriction met as an equality at the selected choice.
Corner solution
An optimum located where two or more feasible boundaries meet.
FAQ

Constrained Optimisation FAQ

Why can an unconstrained optimum be irrelevant?

It may lie outside the feasible region. A constrained answer must discard that point and compare valid boundary, corner and interior candidates instead. Feasibility is part of the solution, not an optional check.

When is a constraint slack?

An inequality is slack when the selected point lies strictly inside its boundary. It does not determine the choice at that solution, so its complementary multiplier is zero under the usual convention.

Why test vertices in linear programming?

A linear objective over a polygon reaches its best value on the boundary and commonly at a vertex, although a tie may extend along an edge.

What belongs in a complete candidate ledger?

Give each row a source such as an interior condition, smooth boundary, corner or endpoint. Record whether the point satisfies every restriction, which constraints bind and the objective value. Retain rejected rows with the reason they fail. The completed ledger makes the global comparison auditable and shows whether a missing answer came from geometry, candidate generation or evaluation arithmetic.

Study strategy

Exam move

Sketch the feasible region before writing candidate conditions. Mark equality curves, inequality sides, intersections and corners, then identify which smooth methods apply. Practise problems where the unconstrained candidate is feasible, infeasible and exactly on a boundary. Keep the multiplier attached to its named constraint and state the convention before interpreting its sign.

Always compare objective values across the complete candidate list. Review mistakes by missing corner, wrong inequality direction, infeasible point, slack constraint called binding and local result reported as global. Begin each exercise with a feasibility pass that is independent of the objective. Rewrite every inequality in a consistent direction, shade the permitted side and calculate intersection points exactly.

Test a sample point when the shaded side is uncertain. Then place the unconstrained stationary point on the sketch and decide whether it can remain a candidate. This order stops an elegant first-order calculation from dominating the analysis when it lies outside the feasible set.

For a nonconvex region or objective, keep the language cautious and build a complete candidate list rather than assuming a local condition identifies the global solution. Practise active-set reasoning with small variations of one problem. Tighten a resource constraint until it binds, relax it until it becomes slack and move the objective so a smooth tangency turns into a corner.

At each stage, list which equalities determine the candidate and which inequalities must still be checked. Use complementary slackness as a logical pairing rather than as a collection of symbols: a strictly slack restriction cannot carry a positive shadow value under the standard convention, while a positive multiplier points to a binding restriction. State the sign convention before interpreting any multiplier.

For linear programming, enumerate vertices systematically from boundary intersections and eliminate infeasible ones before comparing objective values. Draw objective contours to predict which direction improves the criterion, but let the numerical vertex comparison settle ties and degeneracy. If two adjacent vertices have the same value, check whether the entire connecting edge is optimal.

For smooth equality constraints, verify the candidate conditions and then examine whether the regularity and curvature information support the claimed classification. In both settings, a final table with candidate, feasibility and objective value makes omitted corners and arithmetic inconsistencies easy to spot.

On a timed response, separate four claims: the point satisfies the stated constraints, it satisfies the relevant candidate conditions, it beats the other feasible candidates and it has an economic interpretation. A multiplier may describe the local value of relaxing its own constraint while the active set remains unchanged; it is not automatically a global price for every resource change.

Mention any boundary at which that interpretation would cease to apply. Close by naming the chosen quantities, binding restrictions and attained objective value so the mathematical result is connected back to the decision problem.

Working through Constrained Optimisation in ECON0006? Sia is AskSia’s AI Economics tutor — ask any ECON0006 Constrained Optimisation question and get a clear, step-by-step explanation grounded in how ECON0006 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 32 of your subjects at University College London, plus 1,000+ more Bibles across the AskSia library.
Sia - your ECON0006 tutor, unlimited, worked the way the exam marks it
The full 4-page Bible + practice bank with worked solutions
Chrome extension - sync your LMS so Sia knows your deadlines
Bilingual EN / Chinese on every Bible and every Sia answer
$0.99 Trial
30-day money-back · cancel in one tap · how it works