Welcome to Numerical Methods
Have you ever tried to solve an equation like \(x = \cos(x)\) or \(e^x + x - 3 = 0\) using standard algebra? You quickly discover that no amount of rearranging gives an exact, neat answer like \(x = 2\). In the real world—from predicting weather patterns to modelling financial markets—many equations cannot be solved using exact algebraic methods. That is where numerical methods come to the rescue!
Instead of finding an exact symbolic answer, numerical methods allow us to find highly accurate approximations. In this chapter for CCEA A2 1 Pure Mathematics, we will explore four key tools:
1. Locating roots using sign changes and graphs.
2. Fixed-point iteration (\(x_{n+1} = g(x_n)\)) and how it behaves.
3. The Newton-Raphson method for rapid root finding.
4. The Trapezium Rule for approximating definite integrals.
Don't worry if this seems tricky at first! We will break down every method step-by-step with clear examples and key exam tips.
1. Location of Roots via Sign Change
The Big Idea: The Intermediate Value Theorem
Imagine walking across a straight boundary line separating two fields. If you start on the South side (negative values) and end up on the North side (positive values), you must have crossed the boundary line at least once—provided you did not jump over it in an aeroplane! In mathematics, this boundary line is the \(x\)-axis, where \(y = 0\).
Formal Rule: If a function \(f(x)\) is continuous on a closed interval \([a, b]\), and the values of \(f(a)\) and \(f(b)\) have opposite signs (i.e. \(f(a)f(b) < 0\)), then there is at least one root of the equation \(f(x) = 0\) in the open interval \((a, b)\).
The Standard CCEA Exam Justification
CCEA examiners look for three specific elements when awarding marks for sign-change questions. Always write out your answer following this 3-step template:
1. Calculate and state the values: "\(f(a) = \text{value} > 0\)" and "\(f(b) = \text{value} < 0\)" (or vice versa).
2. State the condition: "There is a change of sign and \(f(x)\) is continuous."
3. Conclude: "Therefore, there is a root in the interval \((a, b)\)."
When Does the Sign-Change Rule Fail?
Be careful! A sign change does not always guarantee a single neat root, and a root can exist without a sign change:
• Discontinuous functions: If a graph has a vertical asymptote (like \(f(x) = \frac{1}{x}\) across \(x=0\) or \(f(x) = \tan(x)\) across \(x = \frac{\pi}{2}\)), the sign can change from positive to negative by jumping across an asymptote without ever crossing the \(x\)-axis.
• Even number of roots / Multiple roots: If there are two roots in the interval \([a, b]\), the curve crosses the axis and crosses back again. Thus, \(f(a)\) and \(f(b)\) will have the same sign, hiding the roots.
• Turning points on the axis (Repeated roots): If the curve touches the \(x\)-axis at a stationary point (e.g. \(f(x) = (x - 2)^2\) at \(x = 2\)), \(f(x) \ge 0\) everywhere, so there is no sign change despite there being a root.
Key Takeaway for Section 1: Always write out the numerical values, state "change of sign", and explicitly include the word "continuous" to secure full marks.
2. Fixed-Point Iteration (\(x_{n+1} = g(x_n)\))
How It Works
To use fixed-point iteration to solve \(f(x) = 0\):
1. Rearrange the equation \(f(x) = 0\) algebraically into the form \(x = g(x)\).
2. Turn this into an iterative formula: \(x_{n+1} = g(x_n)\).
3. Choose a starting value \(x_0\) and generate a sequence of estimates \(x_1, x_2, x_3, \dots\)
Convergence Condition
Will the sequence always zoom in on the root \(\alpha\)? Not necessarily! It depends on the gradient of the rearrangement \(g(x)\):
• The iteration converges to the root \(\alpha\) if \(|g'(\alpha)| < 1\) near the root (the curve \(y = g(x)\) is relatively flat).
• The iteration diverges (fails to find the root) if \(|g'(\alpha)| > 1\) near the root (the curve \(y = g(x)\) is too steep).
Visualising Iteration: Staircase and Cobweb Diagrams
We can visualise the iteration by drawing two lines on a graph: \(y = x\) and \(y = g(x)\). The root is the point of intersection where \(x = g(x)\).
• Staircase Diagram: Occurs when \(0 < g'(\alpha) < 1\). The gradient is positive and shallow. The path steps directly towards the intersection in one direction (monotonically), resembling a staircase.
• Cobweb Diagram: Occurs when \(-1 < g'(\alpha) < 0\). The gradient is negative and shallow. The successive approximations oscillate above and below the root, spiralling inwards like a spider's web.
Key Takeaway for Section 2: An iteration converges when the slope of \(g(x)\) is shallow (\(|g'(\alpha)| < 1\)). Positive slopes create staircases; negative slopes create cobwebs.
3. The Newton-Raphson Method
The Big Idea and Formula
The Newton-Raphson method is a powerful technique that uses calculus to zoom in on a root extremely quickly. Instead of guessing blindly, it uses the tangent line to the curve at your current estimate to find an improved estimate.
The standard CCEA formula is:
\(x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\)
Derivation: Where Does This Come From?
1. Consider a point \((x_n, f(x_n))\) on the curve \(y = f(x)\).
2. The gradient of the tangent at this point is \(m = f'(x_n)\).
3. The equation of the tangent line is: \(y - f(x_n) = f'(x_n)(x - x_n)\).
4. We find where this tangent crosses the \(x\)-axis by setting \(y = 0\):
\(-f(x_n) = f'(x_n)(x_{n+1} - x_n)\)
5. Dividing by \(f'(x_n)\) and rearranging for \(x_{n+1}\) gives the formula:
\(x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\)
Applying Newton-Raphson to Turning Points
Exam Alert! A common CCEA question asks you to find the \(x\)-coordinate of a turning point or stationary point of a function \(y = f(x)\) using Newton-Raphson.
Remember: Turning points occur where the derivative equals zero, so you are solving \(f'(x) = 0\).
When applying Newton-Raphson to solve \(f'(x) = 0\), the formula becomes:
\(x_{n+1} = x_n - \frac{f'(x_n)}{f''(x_n)}\)
Limitations and Failure Modes of Newton-Raphson
Newton-Raphson is fast, but it can fail under certain conditions:
• Stationary / Turning Points (\(f'(x_n) = 0\)): If an iterate lands on a stationary point, the tangent line is horizontal. A horizontal line never crosses the \(x\)-axis, resulting in a mathematical impossibility (division by zero in the formula).
• Poor Starting Value (\(x_0\) too far away): If the initial guess is far from the root, the iteration may diverge, bounce between values in a cyclic loop, or converge to an entirely different root than the one intended.
• Discontinuous derivatives: If \(f(x)\) or \(f'(x)\) is discontinuous near the root, the tangent approximations break down.
Key Takeaway for Section 3: Newton-Raphson uses tangent lines (\(x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\)). It fails when \(f'(x_n) = 0\) (horizontal tangent) or when \(x_0\) is chosen poorly.
4. Numerical Integration: The Trapezium Rule
The Big Idea
When an integral \(\int_{a}^{b} y \, \mathrm{d}x\) cannot be integrated analytically (such as \(\int e^{x^2} \mathrm{d}x\)), we can approximate the area under the curve by splitting the region into \(n\) vertical strips of equal width and approximating each strip as a trapezium.
The Trapezium Rule Formula
\(\int_{a}^{b} y \, \mathrm{d}x \approx \frac{h}{2} \left[ (y_0 + y_n) + 2(y_1 + y_2 + \dots + y_{n-1}) \right]\)
Where:
• \(h = \frac{b - a}{n}\) is the strip width (step size).
• \(n\) is the number of strips (intervals).
• \(n + 1\) is the number of ordinates (\(y\)-values: \(y_0, y_1, \dots, y_n\)).
• Memory Aid: "Half the width \(\times\) [ (First + Last) + 2 \(\times\) (Sum of the Rest) ]"
Overestimates and Underestimates
CCEA examiners frequently ask whether the Trapezium Rule gives an overestimate or an underestimate. The answer depends strictly on the curvature (second derivative) of the graph, not whether the graph is increasing or decreasing:
• Convex Curve (\(\frac{\mathrm{d}^2y}{\mathrm{d}x^2} > 0\), curving upwards like a smile):
The straight chords connecting the points on the curve sit above the curve. Therefore, the area of the trapezia is greater than the true area \(\implies\) Overestimate.
• Concave Curve (\(\frac{\mathrm{d}^2y}{\mathrm{d}x^2} < 0\), curving downwards like a frown):
The straight chords sit below the curve. Therefore, the area of the trapezia is less than the true area \(\implies\) Underestimate.
Exam Tip: To get full explanation marks, always refer to the chord position relative to the curve: state that the curve is convex/concave and that the straight chords lie above/below the curve.
How to Improve Accuracy
To obtain a more accurate approximation using the Trapezium Rule, simply increase the number of strips \(n\) (which decreases the strip width \(h\)). This makes the straight chords hug the true curve much more closely.
Key Takeaway for Section 4: Strip width is \(h = \frac{b-a}{n}\). Convex curves (\(\cup\)) give overestimates because chords lie above the curve; concave curves (\(\cap\)) give underestimates because chords lie below.
Top 5 Exam Pitfalls to Avoid
1. Forgetting Radians Mode:
In A2 Pure Mathematics, all calculus and numerical methods involving trigonometric functions (such as \(\sin(x)\), \(\cos(x)\), \(\tan(x)\)) must be calculated in Radians mode. If your calculator is in Degrees mode, every subsequent value in your table or iteration will be wrong!
2. Confusing Strips and Ordinates:
Remember: \(n\) strips always require \(n + 1\) ordinates. For example, 4 strips from \(x = 0\) to \(x = 2\) means \(h = \frac{2-0}{4} = 0.5\), giving 5 \(x\)-values (\(x_0, x_1, x_2, x_3, x_4\)) and 5 corresponding \(y\)-values (\(y_0, y_1, y_2, y_3, y_4\)).
3. Premature Rounding:
When performing iterative calculations or working out \(y\)-values for the Trapezium Rule, keep intermediate values in your calculator memory or use the ANS key. Rounding intermediate steps to 2 or 3 decimal places causes rounding errors that propagate through your solution.
4. Omitting "Continuous" in Sign-Change Explanations:
Stating only that there is a change of sign will lose you the reasoning mark. You must explicitly state that the function is continuous on the interval.
5. Mixing up Newton-Raphson for Roots vs. Turning Points:
Read the question carefully! If asked to find where a curve crosses the axis, solve \(f(x) = 0\) using \(x_{n+1} = x_n - \frac{f(x_n)}{f'(x_n)}\). If asked to find a stationary point, you are solving \(f'(x) = 0\), so use \(x_{n+1} = x_n - \frac{f'(x_n)}{f''(x_n)}\).
Quick Chapter Summary Checklist
Before sitting your exam, make sure you can confidently:
• Write the 3-part justification for locating roots: values calculated, change of sign + continuous, root exists in interval.
• Determine whether an iteration \(x_{n+1} = g(x_n)\) converges by checking if \(|g'(\alpha)| < 1\).
• Sketch and identify staircase (\(g' > 0\)) and cobweb (\(g' < 0\)) diagrams.
• Apply the Newton-Raphson formula accurately to find roots and stationary points.
• State the failure conditions of Newton-Raphson (\(f'(x_n) = 0\) or poor starting value).
• Set up a neat ordinate table and apply the Trapezium Rule formula.
• Justify whether the Trapezium Rule produces an overestimate (convex, chords above) or underestimate (concave, chords below).