Skip to topic
    ← Back to course topics

    Numerical Methods — OCR A-Level Mathematics

    Numerical Methods explained

    Numerical methods give approximations, and OCR expects you to recognise the situations in which they break down.

    Read the full explanation

    A sign-change argument can fail when a function is discontinuous: f(x) = 1/x has f(−1) < 0 and f(1) > 0 yet no root lies between them. Fixed-point iteration xₙ₊₁ = g(xₙ) can diverge when |g′(x)| ≥ 1 near the root, or cycle without settling. The Newton–Raphson method fails when f′(xₙ) = 0, when the starting value sends iterates far from the root, or when it oscillates between two values. The trapezium rule can mislead when the function changes rapidly or is not smooth over the interval. For each method, identify the condition that causes failure and describe the observable symptom, such as divergence, cycling or a zero derivative.

    Your focus

    1. Identify conditions under which a numerical method fails
    2. Describe the observable symptom of each failure
    3. Relate a failure condition to the correct method

    Numerical Methods exam tips

    Marking Points
    • Identifies a discontinuity as a reason why a sign-change argument can fail to locate a root
    • States that fixed-point iteration may diverge when the gradient of the iteration function is too large near the root
    • Recognises that Newton–Raphson fails when the derivative is zero at an iterate
    • Describes a symptom of failure such as divergence, oscillation or cycling between values
    • Links a failure condition to a specific numerical method rather than giving a general statement
    Examiner Tips
    • 💡Name the method and the precise condition that causes failure in the same sentence
    • 💡Give a concrete symptom, such as divergence or cycling, rather than saying only that it 'does not work'
    • 💡Use a specific function or iteration to illustrate the failure when the question allows
    Common Mistakes
    • Assumes a sign change always guarantees a root; note that a discontinuity such as 1/x breaks the argument
    • Treats any convergent-looking iteration as reliable; check the size of the derivative of the iteration function near the root
    • Forgets that Newton–Raphson needs a non-zero derivative; identify the zero-derivative case as a failure