Numerical solution of equations

FP1 Topic 3, spec point by spec point (Numerical solution of equations). Each spec point has the method, worked examples written the way the mark scheme wants them, how the marks go and the mistakes that cost the most marks.

Spec3.1Interval bisection, linear interpolation and Newton–Raphson

Solve \(f(x) = 0\) approximately by interval bisection, linear interpolation and the Newton–Raphson process (with \(f\) using only P1 and P2 functions).

Start: a change of sign

If \(f\) is continuous and \(f(a)\), \(f(b)\) have opposite signs, there is a root between \(a\) and \(b\) (as in P3 6.1). All three methods narrow down that root.

Interval bisection: halve the interval

  1. Evaluate \(f\) at the midpoint \(m = \frac{a + b}{2}\).
  2. Keep the half where the sign still changes.
  3. Repeat. Each step halves the width of the interval.

Reliable but slow.

Linear interpolation: draw a chord

Join \((a, f(a))\) and \((b, f(b))\) with a straight line and see where it crosses the axis. By similar triangles:

\[\frac{\alpha - a}{|f(a)|} = \frac{b - \alpha}{|f(b)|}\]

Solve for \(\alpha\). Use the sizes (moduli) of \(f(a)\) and \(f(b)\). Then repeat on the smaller interval if asked.

Newton–Raphson: follow the tangent

\[x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\]

Why: the tangent at \(x_n\) has gradient \(f'(x_n)\) and crosses the axis a horizontal distance \(\frac{f(x_n)}{f'(x_n)}\) from \(x_n\). Near a root, the tangent is close to the curve, so this lands very close to the root. It is usually the fastest method.

It fails if \(f'(x_n) = 0\) (horizontal tangent never meets the axis) or is close to 0 (the next value is thrown far away), or if the starting value is far from the root.

all three methods

\(f(x) = x^3 - 2x - 5\). \(f(2) = -1\) and \(f(3) = 16\), so there is a root \(\alpha\) in \([2, 3]\).

Bisection twice: \(f(2.5) = 5.625 > 0\), so \([2, 2.5]\). \(f(2.25) = 1.89 > 0\), so \([2, 2.25]\).

Linear interpolation on \([2, 3]\): \(\dfrac{\alpha - 2}{1} = \dfrac{3 - \alpha}{16} \Rightarrow 16\alpha - 32 = 3 - \alpha \Rightarrow \alpha = \dfrac{35}{17} = 2.059\).

Newton–Raphson from \(x_0 = 2\): \(f'(x) = 3x^2 - 2\).

\(x_1 = 2 - \dfrac{-1}{10} = 2.1\). \(x_2 = 2.1 - \dfrac{0.061}{11.23} = 2.0946\).

  • Bisection: M1 for values at the ends and the midpoint, A1 for correct values (1 s.f.), dM1 for the next midpoint, A1 for the final interval written the right way round (WFM01A Jan 2026 Q7a).
  • Newton–Raphson: M1 for the formula \(x_0 - \frac{f(x_0)}{f'(x_0)}\), M1 for differentiating (powers of \(x\)), A1 for \(f'(x)\), dM1 for substituting, A1 for the value to the accuracy asked (WFM01 Jun 2018 Q6a).
  • Sign change: values at both ends, "sign change", "continuous", conclusion; use radians if the function has trig (WFM01 Jan 2024 Q6a).

Often asked as: the most examined FP1 point. Show a root lies in an interval, then apply one of the methods once or twice. Functions with fractional powers like \(x^{-\frac12}\) are common.

Using signed values in the interpolation formula, which puts \(\alpha\) outside the interval.

Wrong derivative of a negative or fractional power.

Calculator in degrees when \(f\) contains \(\cos(5\sqrt x)\).