Numerical Methods (A2 Unit 3: Pure Mathematics B) — WJEC A-Level Mathematics
Numerical Methods (A2 Unit 3: Pure Mathematics B) explained
Numerical Methods (A2 Unit 3: Pure Mathematics B) covers 7 specification points, including Locate roots of f (x) = 0 by considering… and Understand how change of sign methods can fail..
Read the full explanation
Open a point below for its explanation and revision guidance.
Locate roots of f (x) = 0 by considering changes in sign of f (x) in an interval of x in which f (x) is sufficiently well-behaved.
The Intermediate Value Theorem establishes that if a function f(x) is continuous on a closed interval [a, b], and f(a) and f(b) have opposite signs (f(a) × f(b) < 0), then there exists at least one real root α ∈ (a, b) such that f(α) = 0. In examination questions, candidates evaluate f(a) and f(b) explicitly, showing numerical values and stating: 'Since f(a) < 0 and f(b) > 0 and f(x) is continuous, there is a change of sign; hence there is at least one root between a and b'. To verify that a root is correct to k decimal places (for instance, α = 2.34 to 2 d.p.), evaluate f(x) at the upper and lower rounding bounds [2.335, 2.345]; a sign change between these bounds confirms the root rounds to 2.34.
Understand how change of sign methods can fail.
Sign-change root-finding methods depend on continuity and odd-multiplicity roots, failing under two main conditions. First, if f(x) has a vertical asymptote or discontinuity within [a, b], a change of sign can occur without any root existing; for example, f(x) = 1/x changes sign between x = −1 (f = −1) and x = 1 (f = 1), but has no root because x = 0 is a vertical asymptote. Second, if an even number of roots exist within the interval, or if the curve touches the x-axis tangentially at a repeated root of even multiplicity (such as f(x) = (x − 2)²), the function does not change sign (f(x) ≥ 0 everywhere); evaluating interval endpoints shows no sign change, failing to detect the root. Sketches identify these failure modes.
Solve equations approximately using simple iterative methods; be able to draw associated cobweb and staircase diagrams.
To solve f(x) = 0 iteratively, rearrange into the fixed-point form x = g(x), generating the recurrence relation xₙ₊₁ = g(xₙ) from an initial approximation x₁. Graphically, the intersection of y = x and y = g(x) gives the root. Starting at (x₁, 0), draw vertically to (x₁, g(x₁)), then horizontally to y = x at (x₂, x₂); repeating this generates an iterative path. If the gradient g'(x) near the root α is positive (0 < g'(α) < 1), successive iterations approach the root monotonically from one side, producing a staircase diagram. If the gradient is negative (−1 < g'(α) < 0), iterations alternate above and below the root, producing an inward spiralling cobweb diagram. If |g'(α)| > 1, the iteration diverges away from the root.
Solve equations using the Newton-Raphson method and other recurrence relations of the form xₙ₊₁ = g (xₙ).
The Newton-Raphson method uses tangent lines to generate root approximations. The tangent to y = f(x) at (x₁, f(x₁)) is y − f(x₁) = f'(x₁)(x − x₁); setting y = 0 gives x₂ = x₁ − f(x₁)/f'(x₁), so xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ). Near a simple root convergence is typically quadratic, roughly doubling correct decimal places per iteration. Other recurrence relations xₙ₊₁ = g(xₙ) are solved by fixed-point iteration: rearrange f(x) = 0 into x = g(x), choose x₁, then iterate. For example, x³ − x − 1 = 0 becomes x = ∛(x+1); starting at x₁ = 1 gives x₂ = ∛2 ≈ 1.260, x₃ ≈ 1.312, x₄ ≈ 1.322, converging to about 1.3247. Convergence requires |g'(x)| < 1 near the root; otherwise the iteration diverges.
Understand how such methods can fail.
The Newton-Raphson method xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ) and fixed-point iteration xₙ₊₁ = g(xₙ) fail under specific geometric and analytical conditions. In Newton-Raphson, if an iterate xₙ lands on or near a stationary point where f'(xₙ) = 0, the tangent line is horizontal and never intersects the x-axis, causing division by zero. If f'(xₙ) is close to zero, the tangent sends the subsequent iterate far from the root, causing divergence. If the curve oscillates or has inflection points between roots, Newton-Raphson can enter a periodic cycle, oscillating indefinitely between two values without converging, or converge to an unintended distant root. In fixed-point iteration xₙ₊₁ = g(xₙ), divergence occurs whenever |g'(x)| > 1 near the root.
Understand and use numerical integration of functions, including the use of the trapezium rule and estimating the approximate area under a curve and limits that it must lie between.
The trapezium rule approximates a definite integral ∫ₐᵇ y dx by dividing the area under the curve into n strips of equal width h = (b − a)/n, bounded by n + 1 ordinates y₀, y₁, …, y_n. Each strip is treated as a trapezium with parallel vertical sides, giving Area ≈ ½h [ (y₀ + y_n) + 2(y₁ + y₂ + … + y_(n−1)) ]. Where the curve is convex (d²y/dx² > 0) the straight chords lie above the curve, so the rule overestimates; where concave (d²y/dx² < 0) the chords lie below, so it underestimates. Bounds on the true integral follow directly: an overestimate is an upper bound, an underestimate a lower bound. For example, on a convex arc the trapezium value exceeds the integral, so the integral lies below it.
Use numerical methods to solve problems in context.
In real-world engineering, financial, and scientific applications, governing equations often contain non-linear combinations of polynomials, exponentials, and trigonometric functions that lack exact analytical solutions. For example, finding the internal rate of return in finance involves finding the root of an equation combining powers of (1 + r)⁻ᵗ, solved via Newton-Raphson. Evaluating the total distance travelled by a vehicle from irregular velocity data uses numerical integration (the trapezium rule) on experimental ordinates. In modelling, candidates must choose appropriate step sizes h or tolerances, perform iterations to required contextual precision, and assess whether error margins affect practical operational decisions.
Your focus
- Apply change of sign tests to locate roots of continuous functions in given intervals.
- State formal conclusions citing both sign change and continuity of the function.
- Verify that an approximation to a root is correct to a specified degree of accuracy using boundary tests.
Show all 21 objectives
- Explain how discontinuities and vertical asymptotes cause sign-change root-finding methods to fail.
- Explain why intervals containing an even number of roots or repeated roots may show no sign change.
- Use graphical sketches to identify scenarios where sign-change tests fail.
- Rearrange equations into fixed-point form x = g(x) and generate iterative sequences.
- Construct staircase and cobweb diagrams to illustrate convergence or divergence of iterations.
- Relate the convergence of iterative methods to the gradient of g(x) near the root.
- Derive the Newton-Raphson formula geometrically from the equation of the tangent line.
- Apply the Newton-Raphson method to calculate successive approximations to roots.
- Solve equations using recurrence relations of the form xₙ₊₁ = g(xₙ) and evaluate iterates until consecutive values agree to a specified precision.
- Explain why the Newton-Raphson method fails when f'(xₙ) = 0 or f'(xₙ) is near zero.
- Identify geometric conditions that lead to divergence, cycling, or convergence to the wrong root.
- Illustrate failure modes of numerical root-finding methods using graphical sketches.
- Apply the trapezium rule to approximate definite integrals from algebraic functions or data tables.
- Determine whether the trapezium rule produces an overestimate or underestimate from the curvature, and hence state a bound for the integral.
- Explain how increasing the number of strips improves the accuracy of the numerical approximation.
- Select and apply appropriate numerical methods to solve real-world contextual problems.
- Execute numerical algorithms to attain specified precision tolerances.
- Interpret numerical solutions in terms of practical operational limits and units.
Numerical Methods (A2 Unit 3: Pure Mathematics B) exam tips
Marking Points
- evaluating f(a) and f(b) with correct numerical values and stating their signs
- providing a concluding sentence citing change of sign and continuity of the function
- applying rounding interval boundaries [α − 0.5 × 10⁻ᵏ, α + 0.5 × 10⁻ᵏ] to confirm accuracy to k decimal places
- explaining that a discontinuity or vertical asymptote can cause a sign change without a root
- explaining that multiple roots or repeated roots of even multiplicity can produce no sign change across an interval containing roots
- illustrating failure modes using clear annotated curve sketches
- rearranging f(x) = 0 into fixed-point form x = g(x)
- generating successive iterative approximations x₂, x₃ using the recurrence relation
- drawing an accurate staircase diagram showing monotonic convergence where 0 < g'(x) < 1
- drawing an accurate cobweb diagram showing alternating spiral convergence where −1 < g'(x) < 0
- Differentiating f(x) correctly to find f'(x) before substituting into the Newton-Raphson formula.
- Substituting into xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ) and calculating the next iterate accurately, showing numerical substitutions.
- Rearranging an equation f(x) = 0 into the form x = g(x) for fixed-point iteration.
- Iterating xₙ₊₁ = g(xₙ) from a given start value, showing successive values to full calculator precision.
- Continuing iterations until consecutive terms agree to the specified accuracy, then stating the rounded root.
- explaining that f'(xₙ) = 0 produces a horizontal tangent that does not meet the x-axis (division by zero)
- explaining that starting near a local extremum can send iterates far from the desired root
- identifying periodic cycling or divergence as failure modes
- illustrating Newton-Raphson failure using a tangent line sketch
- calculating the strip width h = (b − a)/n and tabulating all n + 1 ordinates y₀ to y_n
- substituting ordinates into the trapezium rule formula ½h [ (first + last) + 2(sum of the rest) ]
- stating whether the estimate is an overestimate or underestimate, justified by the sign of d²y/dx² or by a sketch of the chord relative to the curve
- using the over/underestimate to state a bound for the true integral, e.g. 'the integral is less than the trapezium value' for a convex curve
- explaining that increasing n (reducing h) generally improves accuracy because the chords approximate the curve more closely
- setting up an appropriate numerical method matching the contextual equation or data
- executing the iterative or integration algorithm to achieve required decimal precision
- interpreting the calculated numerical output in the real-world context with units
Examiner Tips
- 💡Write both values explicitly: 'f(1) = −2.1 < 0, f(2) = 1.4 > 0; change of sign and f is continuous, so root in (1, 2)'.
- 💡When confirming a root to k decimal places, test the upper and lower bounds ending in 5.
- 💡Mention vertical asymptotes or discontinuities as the primary reason a sign change might not indicate a root.
- 💡Sketch the curve to see whether the graph touches the axis tangentially without crossing it.
- 💡Draw lines strictly vertically to the curve y = g(x) and horizontally to the line y = x.
- 💡Use the ANS key on your calculator to calculate iterative values quickly and avoid rounding errors.
- 💡Program the Newton-Raphson formula into your calculator using the ANS key: ANS − f(ANS)/f'(ANS).
- 💡Write down x₁, x₂ and x₃ to full calculator precision before stating the rounded answer, and check consecutive values agree to the required accuracy.
- 💡State clearly: 'f'(x) = 0, so the tangent is horizontal and does not intersect the x-axis'.
- 💡Show on a sketch how a tangent at a point with a shallow gradient shoots off far from the root.
- 💡Set up a neat table of x and y values before substituting, to avoid arithmetic slips.
- 💡Sketch a chord on the curve: chords above the curve (convex) mean the trapezium rule overestimates, so the estimate is an upper bound.
- 💡Always check how many decimal places or significant figures the question requests.
- 💡State your final answer clearly in the context of the question (e.g. 'The depth of water is 4.25 m').
Common Mistakes
- omitting the explicit mention of continuity, which is a required mathematical condition for the theorem
- using incorrect interval bounds when confirming rounding (e.g. testing 2.33 and 2.35 instead of 2.335 and 2.345)
- assuming a sign change always guarantees a root, ignoring potential asymptotes in functions like tan x or 1/x
- believing no sign change proves no roots exist, ignoring touching roots where f(x) = (x − a)²
- drawing lines between curves at arbitrary angles instead of alternating strictly vertical and horizontal lines
- confusing staircase diagrams (monotonic approach) with cobweb diagrams (oscillating spiral)
- Differentiating f(x) incorrectly before substituting into the Newton-Raphson formula, which corrupts every later iterate.
- Making sign errors when subtracting the fraction, especially when f(xₙ) or f'(xₙ) is negative.
- Assuming any rearrangement x = g(x) converges; if |g'(x)| ≥ 1 near the root the iteration may diverge, so a different rearrangement is needed.
- stating that Newton-Raphson fails when f(x) = 0 instead of when the derivative f'(x) = 0
- failing to connect a horizontal tangent with division by zero in the iterative formula
- miscounting ordinates, using n ordinates instead of n + 1 for n strips
- including the first or last ordinate in the bracket multiplied by 2
- reversing the over/underestimate rule, e.g. claiming a convex curve gives an underestimate
- quoting a bound without stating which side of the estimate the true integral lies on
- terminating numerical iterations prematurely before consecutive values agree to required precision
- failing to attach physical units to the final calculated numerical solution