The University of Manchester · FACULTY OF MATHEMATICS

ECON10071 Chap.6 Unconstrained Stationary Points

- one subject, every graph, every model, every mark
6 Chapters3-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 6 of 12 · ECON10071

Unconstrained Stationary Points

Unconstrained Stationary Points begin with a necessary condition: an interior local maximum of a differentiable objective has zero first partial derivatives. Solving the gradient equations lists candidates, but the converse fails because stationary points can be minima, saddles or flat non-optima.

Classification therefore needs curvature, candidate comparison, a selected path or an algebraic bound such as a constant minus a sum of squares.

In this chapter

What this chapter covers

  • 01

    Gradient equations

  • 02

    Stationary candidates

  • 03

    Necessary conditions

  • 04

    False converse

  • 05

    Local classification

  • 06

    Global algebraic bounds

Worked example · free

Prove a quadratic maximum

Q [7 marks]. Maximise f(x,y)=10x+6y-x²-xy-y² on ℝ². The mark allocation beside this stationary-point example supports practice and is not a published university assessment scheme.
Both marked points can have zero slope, but one is a peak and the other is a valley, so stationarity alone cannot classify them.Both marked points can have zero slope, but one is a peak and the other is a valley, so stationarity alone cannot classify them.local peakstationary valleyzero slope alone does not classify
Both marked points can have zero slope, but one is a peak and the other is a valley, so stationarity alone cannot classify them.
  • 4Set 10-2x-y=0 and 6-x-2y=0 to obtain x=14/3 and y=2/3.
  • 3The quadratic curvature is negative definite, so the candidate is the unique global maximum.
The unique global maximiser is (14/3,2/3). The marks are an original practice allocation.
Sia tip — Use gradient zero to find candidates and a separate argument to prove their type.
Glossary

Key terms

Gradient
The vector containing every first partial derivative.
Stationary point
An input bundle where the gradient equals zero.
Saddle point
A stationary point that rises in some directions and falls in others.
FAQ

Unconstrained Stationary Points FAQ

Does a zero gradient prove a maximum?

No. It is a necessary condition for an interior differentiable optimum, but stationary points can be minima or saddles. A curvature, path, comparison or global bounding argument is still required.

Why can boundary points escape the theorem?

At a boundary, some small directions are infeasible, so the objective may be maximised even when its unconstrained gradient is not zero. The constraint structure must determine the relevant conditions.

How can a sum-of-squares form prove a global result?

If the objective equals a constant minus nonnegative squared terms, it can never exceed that constant. The bundle that makes every square zero therefore attains a global maximum without relying on a local test.

Study strategy

Exam move

Solve the full system of first-order equations and preserve every branch of a factored condition. Evaluate candidates but do not confuse a value comparison with a local proof. Practise completing squares, testing simple paths and recognising negative curvature. End by naming both the maximiser point and the objective value, with the proof supporting local or global status. Treat first-order conditions as a candidate generator.

Compute all first partial derivatives, set them equal to zero simultaneously and solve the complete system. If an equation factors, branch before dividing so that a zero factor is not discarded. Substitute every solution back into the gradient to confirm stationarity. Maintain a candidate ledger containing coordinates and objective value, but leave the classification column blank until a separate argument has been supplied.

This visual separation reinforces that zero gradient is necessary at a differentiable interior optimum but not sufficient for a maximum. Build a classification toolkit rather than relying on one test. Completing squares can produce a global bound immediately. A constant minus non-negative squares identifies a global maximum and its equality conditions.

Selected paths can expose a saddle by finding both higher and lower nearby values. A Hessian sign test gives local curvature information, while a Hessian with the required sign throughout a convex domain can support a global concavity conclusion. Practise deciding which tool is shortest for the algebra in front of you, and state the domain controlled by the chosen argument. Use counterexamples to master logical direction.

The function x to the fourth power has a zero derivative at its global minimum, showing that stationarity does not imply a maximum. The function x cubed has a stationary inflection point, showing that stationarity need not imply any extremum. A two-variable saddle adds the need to inspect more than one direction.

For each example, write the valid statement beside the false converse: an interior differentiable local maximum implies zero gradient, but a zero gradient alone does not establish a maximum. Finish timed answers with three labelled lines: candidate, value and proof. Report the maximiser as a coordinate bundle and the maximum as the objective value.

If several candidates exist, compare values only after each has been classified or after a complete global comparison argument has been set up. Check whether the question asks for a local or global result and avoid stronger wording than the evidence supports. Rework any error by locating whether it arose in differentiation, solving, evaluation or classification rather than changing the final label alone.

Once the method is reliable, practise parameter changes. Alter a linear coefficient in a quadratic objective, predict how the stationary point moves and solve again. Then ask whether the curvature and uniqueness conclusion changes. This separates location from classification: a parameter may shift the candidate while leaving a negative definite quadratic part unchanged.

Verify the new point and value symbolically before inserting numbers, and describe the comparative movement without claiming more than the algebra supports. Keep the original and changed candidates in a small table with gradient equations, coordinates, value and curvature evidence. The comparison makes clear which conclusions survive the parameter change and which need to be recalculated rather than carried over automatically.

Working through Unconstrained Stationary Points in ECON10071? Sia is AskSia’s AI Mathematics tutor — ask any ECON10071 Unconstrained Stationary Points question and get a clear, step-by-step explanation grounded in how ECON10071 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 41 of your subjects at The University of Manchester, plus 1,000+ more Bibles across the AskSia library.
Sia - your ECON10071 tutor, unlimited, worked the way the exam marks it
The full 3-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