Historical Context & Motivation
Most equations you encounter in a math classroom have neat, closed-form solutions — you can rearrange, factor, or apply a formula and arrive at an exact answer. But the real world is not always so tidy. Engineers designing bridges, scientists modeling populations, and economists predicting markets frequently face equations that cannot be solved exactly. Similarly, many integrals arising in physics and statistics have no antiderivative that can be written in terms of elementary functions. Numerical methods were invented precisely to fill this gap — they give us systematic procedures to approximate solutions to any desired degree of accuracy, especially when paired with modern technology.
The central question that drives this topic is: When we cannot find an exact analytical answer, how can we systematically approximate one — and how accurate can we make that approximation? In IB Mathematics HL, you will use your GDC (graphing display calculator) and spreadsheet technology to implement these approximation strategies for both solving equations via iteration and evaluating definite integrals numerically.
Core Principles & Definitions
Before diving into specific algorithms, it helps to understand the foundational ideas that underpin all numerical methods. These methods share a common philosophy: start with a rough guess, then refine it systematically until you reach the desired precision.
Iteration
Fixed-Point Iteration
Newton-Raphson Method
Numerical Integration
Error and Convergence
Visual Explanation — Newton-Raphson Iteration
The diagram below illustrates how the Newton-Raphson method zeroes in on a root of f(x) = 0. Starting from an initial guess x₀, you draw the tangent line to the curve at (x₀, f(x₀)). Where that tangent crosses the x-axis becomes your next estimate x₁. The process repeats, and each new tangent brings you dramatically closer to the root.
Notice how the distances between successive estimates and the root shrink dramatically. This is a hallmark of quadratic convergence — the number of correct decimal places roughly doubles with each iteration. In practice, three or four iterations often yield an answer accurate to many decimal places. On your GDC, you can set up a recurrence relation and watch the values stabilize before your eyes.
Mathematical Framework
Iterative Methods for Solving Equations
The idea is beautifully geometric: the tangent line at the point (xn, f(xn)) has slope f'(xn). Setting y = 0 and solving for x gives the Newton-Raphson formula above. When the starting guess is close enough to the root and f'(x) ≠ 0 near the root, convergence is extremely rapid.
Numerical Integration Methods
Numerical Integration — Trapezoids vs. Parabolas
The diagram below compares the two main numerical integration strategies side by side. On the left, the trapezoidal rule approximates the area under the curve using straight-line segments connecting consecutive points, forming trapezoids. On the right, Simpson's rule uses parabolic arcs that hug the curve more closely, resulting in much better accuracy for the same number of function evaluations.
| Feature | Trapezoidal Rule | Simpson's Rule |
|---|---|---|
| Shape used | Straight-line segments (trapezoids) | Parabolic arcs through triplets of points |
| Coefficient pattern | 1, 2, 2, 2, …, 2, 1 | 1, 4, 2, 4, 2, …, 4, 1 |
| Multiplier | h / 2 | h / 3 |
| Number of strips | Any positive integer n | Must be even |
| Error order | O(h²) — error proportional to h² | O(h⁴) — error proportional to h⁴ (much smaller) |
Worked Examples
Example 1: Newton-Raphson Iteration
Find the root of f(x) = x³ − 2x − 5 near x = 2 using the Newton-Raphson method. Give your answer correct to four decimal places.
Example 2: Trapezoidal Rule
Estimate ∫₀² e^(−x²) dx using the trapezoidal rule with n = 4 strips.
Strengths, Limitations & When to Use Each Method
No single numerical method is universally best. Choosing the right tool depends on what you're trying to solve, how much information you have (can you compute derivatives?), and how accurate you need the answer to be. The table below summarizes the trade-offs.
| Method | Strengths | Limitations |
|---|---|---|
| Fixed-Point Iteration | Simple to set up; no derivative needed; easy to implement on a GDC using the ANS key. | May diverge if |g'(x)| ≥ 1 near the root; convergence can be slow (linear rate); requires careful rearrangement of the equation. |
| Newton-Raphson | Very fast convergence (quadratic); typically needs only a few iterations; widely applicable. | Requires f'(x) ≠ 0 at the root; can fail or diverge if the initial guess is poor or near a turning point; may cycle for some functions. |
| Trapezoidal Rule | Easy to apply; works with any number of strips; useful when only tabulated data points are available. | Only O(h²) accuracy; systematically underestimates concave-up regions and overestimates concave-down regions. |
| Simpson's Rule | O(h⁴) accuracy — far more precise for the same number of evaluations; exact for polynomials up to degree 3. | Requires an even number of strips; slightly more complex coefficient pattern; less intuitive geometrically than trapezoids. |
Connections to Advanced Theory & Euler's Method
The numerical methods in AHL 5.10 connect directly to Euler's method for differential equations (covered in AHL 5.11) and to the broader field of numerical analysis studied at university level. Understanding iteration and numerical integration here lays the groundwork for more sophisticated algorithms like Runge-Kutta methods and adaptive quadrature, which are the workhorses of modern scientific computing.
| Concept in AHL 5.10 | Advanced Extension |
|---|---|
| Fixed-point iteration xn+1 = g(xn) | Banach fixed-point theorem (contraction mappings); used in economics (equilibrium computation) and computer graphics (fractal generation). |
| Newton-Raphson method | Generalizes to systems of equations via the Jacobian matrix (Newton's method in ℝⁿ); essential in optimization, machine learning, and engineering simulations. |
| Trapezoidal rule | Forms the basis for the trapezoidal method in ODE solving (implicit Euler); leads to Romberg integration which extrapolates to higher accuracy. |
| Simpson's rule | Part of the Newton-Cotes family of quadrature rules; extends to Gaussian quadrature for optimal node placement and maximum accuracy. |
As you progress to AHL 5.11 and beyond, you'll see that Euler's method is essentially the trapezoidal idea applied to differential equations — instead of approximating the area under a curve, you approximate the trajectory of a solution curve by taking small steps along the tangent. The iterative mindset you develop here — start, refine, converge — is the universal language of computational mathematics.
Practice Problems
Summary — Further Numerical Methods
In AHL 5.10, you learned two families of numerical techniques. For solving equations, fixed-point iteration uses xn+1 = g(xn) and converges when |g'(x)| < 1 near the root, while the Newton-Raphson method uses xn+1 = xn − f(xn)/f'(xn) and achieves quadratic convergence by exploiting the tangent line at each step.
For numerical integration, the trapezoidal rule approximates the area under a curve using straight-line segments with O(h²) error, while Simpson's rule fits parabolic arcs and achieves O(h⁴) error — far superior accuracy for the same number of strips. Your GDC is your most important tool: use it to perform iterations, evaluate function values, and verify numerical estimates against built-in integration features. Mastering these methods prepares you for Euler's method in AHL 5.11 and for the computational thinking that underlies modern science and engineering.