The University of Hong Kong · FACULTY OF MATHEMATICS

MATH1821 Chap.5 Numerical Methods

- one subject, every graph, every model, every mark
6 Chapters5-page Bible
Our own words - no uploaded lecturer files
Updated for this semester
Chapter 5 of 6 · MATH1821

Numerical Methods

An approximation needs a guarantee

Bisection preserves a sign-changing bracket and halves its width, giving explicit error control. Taylor approximation replaces a function near a centre and bounds the remainder through a higher derivative. Newton's method uses a tangent to move rapidly toward a root but may fail from poor starting geometry.

Composite trapezoidal and Simpson rules approximate integrals from weighted samples, with concavity and derivative bounds explaining direction and error.

Identify what the tolerance measures

Bisection controls the retained interval width directly; the midpoint's distance from the bracketed root is at most half that width. A question may specify either quantity.

Translate the wording before solving the power-of-two inequality and round the required iteration count upward. Continuity and opposite endpoint signs establish the initial bracket, while every midpoint sign test preserves it.

Keep Newton's geometry visible

Newton's update sends the current point to the x-intercept of its tangent.

Near a simple root it can converge rapidly, but a small derivative, poor starting point or domain boundary can cause a large step, cycle or failure. Carry guard digits from one iterate to the next.

A small residual and a small change between iterates are useful diagnostics, but neither is automatically a certified root-error bound.

Bind Taylor error to an interval

A Taylor polynomial matches successive derivatives at its centre. The remainder uses a bound for the next derivative over the entire interval joining the centre and evaluation point.

State that interval and the chosen maximum before inserting numbers. Increasing the order or moving closer to the centre can reduce the bound, while extra calculator digits do not repair an approximation whose truncation error is too large.

Write quadrature weights symbolically

Composite trapezoidal uses endpoint weight one and interior weight two with the common factor h over two.

Composite Simpson uses endpoint weight one, alternating interior weights four and two, and h over three; it requires an even number of equal subintervals. Write the nodes, step and weight pattern before evaluating function values so a missing endpoint or parity error remains visible.

Use independent scale checks

A root estimate should remain consistent with its bracket or tangent geometry.

An integral of a positive function should be positive and have a plausible magnitude. Trapezoidal direction can be inferred from concavity when the second derivative keeps one sign. Round only the final reported value and pair it with the method, mesh, bracket or error statement that explains why its displayed precision is justified.

In this chapter

What this chapter covers

  • 01

    Bisection Brackets

  • 02

    Taylor Polynomials

  • 03

    Remainder Bounds

  • 04

    Newton Iteration

  • 05

    Trapezoidal Weights

  • 06

    Simpson Weights

Worked example · free

Compute two Newton iterates

Q [7 marks]. For f(x)=x2-2 and x0=1.5, compute x1 and x2. The marks shown are an AskSia study allocation and are not the University's marking scheme.
Newton follows the tangent to the axis.Newton follows the tangent to the axis. The orange tangent at x n meets the axis at x n+1 , which becomes the next guess.xₙxₙ₊₁f(xₙ)
Newton follows the tangent to the axis. The orange tangent at xn meets the axis at xn+1, which becomes the next guess.
  • 2Write the update using f and f prime.
  • 2Compute the first iterate without early rounding.
  • 3Compute and check the second iterate.
The update is xnext=x-(x2-2)/(2x). It gives x1=1.4166667 and x2=1.4142157. The second residual is about 0.000006, showing that the equation is nearly satisfied, though the residual and true root error are different quantities.
Sia tip — Carry guard digits through every iterate and round only the final requested approximation.
Glossary

Key terms

Bisection
A root method that repeatedly retains the sign-changing half of a bracket.
Error Tolerance
A declared maximum acceptable uncertainty in an approximation.
Taylor Remainder
The difference between a function and its Taylor polynomial, bounded using a later derivative.
Newton Iterate
The next estimate obtained from the current tangent's x-intercept.
Simpson Rule
A composite quadrature rule with alternating interior weights four and two.
FAQ

Numerical Methods FAQ

What error does bisection control?

After each iteration the retained bracket width halves. If the midpoint of the final bracket is reported, its distance from the bracketed root is at most half that width. A question may set tolerance on the full interval or on the midpoint error, so translate the wording before counting iterations.

Why can Newton's method fail?

The tangent can meet the axis far from the desired root, the derivative can be zero or very small, an iterate can leave the domain, or the process can cycle. A good initial guess and a stopping rule improve practice, but substitution back into the original equation remains an important diagnostic.

What is the Simpson weight pattern?

For an even number of equal subintervals, endpoints have weight one, odd interior nodes weight four and even interior nodes weight two, all multiplied by h divided by three. Write the pattern before evaluating function values so a missing endpoint or parity error becomes visible immediately.

Study strategy

Exam move

For each approximation, keep a two-column record: the numerical update and the reason it is accurate enough. Recompute one bisection count, one Taylor remainder and one quadrature weight pattern without notes. For Newton's method, sketch the tangent geometry so the update is understood rather than memorised. Record guard digits separately from the rounded final value and its stated tolerance.

Working through Numerical Methods in MATH1821? Sia is AskSia’s AI Mathematics tutor — ask any MATH1821 Numerical Methods question and get a clear, step-by-step explanation grounded in how MATH1821 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.

A+Everything unlocked
Unlocks this Bible + all 52 of your subjects at The University of Hong Kong, plus 1,000+ more Bibles across the AskSia library.
Sia - your MATH1821 tutor, unlimited, worked the way the exam marks it
The full 5-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