A-Level Edexcel Numerical Methods (数值方法) Complete Guide

Numerical Methods in A-Level Edexcel Mathematics: A Complete Guide

Learn how to solve equations that cannot be solved algebraically — a key topic in the A-Level Edexcel Pure Mathematics syllabus.

---

1. What Are Numerical Methods?

Numerical methods are techniques used to find approximate solutions to mathematical problems that cannot be solved exactly using algebraic manipulation. In the real world, many equations — especially those involving polynomials of degree 5 or higher, trigonometric functions, exponentials, or combinations of these — do not have closed-form analytical solutions. Numerical methods provide a practical way to obtain answers to any desired level of accuracy.

A common example is the equation x^3 + x – 1 = 0. There is no easy way to solve this algebraically. Numerical methods allow us to approximate the root to as many decimal places as we need.

In the Edexcel A-Level syllabus, you are expected to master three core numerical approaches: (1) locating roots by sign changes, (2) the iterative fixed-point method, and (3) the Newton-Raphson method.

Key Idea: Numerical methods trade exactness for computability — they give us answers we can actually calculate, even when the “perfect” analytical answer remains out of reach.

2. Locating Roots: The Sign-Change Method

The simplest way to locate a root is to look for a change in sign. If f(a) and f(b) have opposite signs and f is continuous on the interval [a, b], then by the Intermediate Value Theorem, there must be at least one root in that interval.

For example, consider f(x) = x^3 – 2x – 5. We can evaluate:

• f(2) = 8 – 4 – 5 = -1 (negative)
• f(3) = 27 – 6 – 5 = 16 (positive)

Since the sign changes from negative to positive, there is at least one root between x = 2 and x = 3. This is a reliable but coarse method — it tells us where a root lives but not its exact value.

To narrow down the interval, we can repeatedly bisect it (the Interval Bisection or Bisection Method): evaluate at the midpoint, keep the half-interval where the sign change occurs, and repeat until the interval is as small as desired. Each iteration halves the interval width — after 10 iterations, the interval is 1/1024 of the original width, giving roughly 3 decimal places of accuracy.

3. Fixed-Point Iteration

Fixed-point iteration is one of the most elegant numerical methods. The idea is to rearrange an equation f(x) = 0 into the form x = g(x). Then, starting from an initial guess x0, we repeatedly apply:

x_{n+1} = g(x_n)

If the iteration converges, the limit is a fixed point — a value where x = g(x), which means f(x) = 0.

Convergence Condition: The iteration converges to a root if |g'(x)| is less than 1 in a neighbourhood of the root. If |g'(x)| is greater than 1, the iteration diverges away from the root. Understanding this condition is tested frequently in Edexcel exams.

Example: Solve x^3 + x – 1 = 0. One possible rearrangement is x = (1 – x)^(1/3). Starting from x0 = 0.7:

Iteration	x_n
0	0.7000
1	0.6694
2	0.6874
3	0.6809
4	0.6836
5	0.6825

The root appears to be approximately 0.6823 (to 4 d.p.).

4. The Newton-Raphson Method

The Newton-Raphson method is the powerhouse of numerical root-finding. It uses the derivative of the function to produce a sequence that typically converges much faster than fixed-point iteration. The formula is:

x_{n+1} = x_n – f(x_n) / f'(x_n)

Geometrically, at each step we draw the tangent line to the curve at x_n, find where it crosses the x-axis, and use that crossing point as our next estimate. This geometric interpretation makes the method very intuitive.

Convergence: Newton-Raphson usually converges quadratically — the number of correct decimal places roughly doubles with each iteration once you are close to the root. However, it has drawbacks:

• The derivative f'(x_n) must not be zero (division by zero)
• A poor initial guess can cause divergence
• If the root is a multiple root, convergence slows to linear

Example: Find sqrt(2) by solving f(x) = x^2 – 2 = 0 with f'(x) = 2x. Starting from x0 = 1.5:

• x1 = 1.5 – (2.25 – 2) / 3 = 1.4167
• x2 = 1.4167 – (2.0069 – 2) / 2.8334 = 1.4142

In just two iterations we have sqrt(2) correct to 4 decimal places!

Exam Tip: Edexcel often asks you to apply Newton-Raphson to a specific equation and to explain why the method might fail in certain cases (e.g., when f'(x) = 0 or when the starting value is near a turning point). Always show your full working — marks are awarded for substitution, not just the final answer.

5. Comparing the Methods

Method	Speed	Requires Derivative?	Reliability	Best For
Sign Change / Bisection	Slow (linear)	No	Very reliable	Initial root location
Fixed-Point Iteration	Linear	No	Depends on g'(x)	Rearranged equations
Newton-Raphson	Fast (quadratic)	Yes	Sensitive to start	High-precision roots

6. Common Exam Question Types (Edexcel)

• Show that a root lies between two values: Evaluate f(a) and f(b) and note the sign change — always state continuity explicitly.
• Perform a given number of iterations: Use the formula provided, showing each step clearly in a table.
• Determine whether an iteration converges: Check |g'(x)| < 1 — a classic 2-3 mark question.
• Newton-Raphson with trigonometric functions: Use radians mode on your calculator — this catches many students out.
• Justify why an iteration fails: Common reasons include |g'(x)| > 1, division by zero, or oscillation.
• Apply numerical methods in context: Real-world problems such as finding interest rates, projectile ranges, or population models.

7. Practical Tips for Success

1. Use your calculator efficiently. The Edexcel exam expects you to use the ANS key or store/recall functions to iterate quickly. Practice the key sequence so it becomes automatic.
2. Always work in radians for trigonometry. Newton-Raphson involving sin, cos, or tan must use radians — a degree-mode answer will be wrong.
3. Draw a diagram. A rough sketch of f(x) helps you understand why Newton-Raphson might fail (e.g., starting near a stationary point where the tangent is nearly horizontal).
4. Give answers to the required accuracy. If the question asks for 3 decimal places, provide exactly 3 — no more, no less. Round correctly at the final step.
5. Check your rearrangement. For fixed-point iteration, the equation must be rearranged so that it truly satisfies x = g(x). A common mistake is to keep the original equation form, leading to wrong results.
6. Understand, don’t just memorise. Edexcel questions often ask why a method converges or diverges. Understanding the convergence conditions conceptually is more valuable than rote-memorising formulas.

---

A-Level Edexcel 数学中的数值方法：完整指南

学习如何求解无法用代数方法解决的方程——这是 A-Level Edexcel 纯数学大纲中的核心内容。

---

1. 什么是数值方法？

数值方法是一种用于找到数学问题近似解的技术，当这些问题无法通过代数运算精确求解时。在现实世界中，许多方程——尤其是涉及五次及以上多项式、三角函数、指数函数或这些函数的组合——并没有封闭形式的解析解。数值方法提供了一种实用途径，可以获得任意所需精度的答案。

一个典型的例子是方程 x^3 + x – 1 = 0。这个方程无法通过简单的代数方法求解，但数值方法可以让我们将根近似到所需的任意小数位数。

在 Edexcel A-Level 大纲中，你需要掌握三种核心数值方法：(1) 通过符号变化定位根，(2) 不动点迭代法，以及 (3) 牛顿-拉弗森法。

核心思想：数值方法用精确性来换取可计算性——它们给出的答案我们确实可以算出来，即使完美的解析答案遥不可及。

2. 定位根：符号变化法

定位根的最简单方法是寻找符号变化。如果 f(a) 和 f(b) 符号相反，且 f 在区间 [a, b] 上连续，那么根据介值定理，该区间内至少存在一个根。

例如，考虑 f(x) = x^3 – 2x – 5。我们可以求值：

• f(2) = 8 – 4 – 5 = -1（负）
• f(3) = 27 – 6 – 5 = 16（正）

由于符号从负变为正，在 x = 2 和 x = 3 之间至少存在一个根。这是一个可靠但粗糙的方法——它告诉我们根的大致位置，但不能给出精确值。

为了缩小区间，我们可以反复二分（区间二分法或对分法）：在中点处求值，保留符号发生变化的那一半区间，反复进行，直到区间足够小。每次迭代将区间宽度减半——10 次迭代后，区间宽度变为原来的 1/1024，可提供大约 3 位小数的精度。

3. 不动点迭代法

不动点迭代是最优雅的数值方法之一。其思路是将方程 f(x) = 0 改写为 x = g(x) 的形式。然后，从初始猜测 x0 开始，反复应用：

x_{n+1} = g(x_n)

如果迭代收敛，极限就是不动点——即满足 x = g(x) 的值，这意味着 f(x) = 0。

收敛条件：如果在根的邻域内 |g'(x)| 小于 1，则迭代收敛到根。如果 |g'(x)| 大于 1，迭代会发散远离根。理解这个条件是 Edexcel 考试中经常考查的内容。

示例：求解 x^3 + x – 1 = 0。一种可能的改写形式是 x = (1 – x)^(1/3)。从 x0 = 0.7 开始：

迭代	x_n
0	0.7000
1	0.6694
2	0.6874
3	0.6809
4	0.6836
5	0.6825

根大约为 0.6823（精确到 4 位小数）。

4. 牛顿-拉弗森法

牛顿-拉弗森法是数值求根的主力方法。它利用函数的导数生成一个序列，通常比不动点迭代收敛得快得多。公式为：

x_{n+1} = x_n – f(x_n) / f'(x_n)

从几何角度看，每一步我们在 x_n 处画出曲线的切线，找到它与 x 轴的交点，将该交点作为下一个估计值。这种几何解释使得该方法非常直观。

收敛性：牛顿-拉弗森法通常以二次收敛速度收敛——一旦接近根，正确的小数位数大约每次迭代翻一番。然而，它也有缺点：

• 导数 f'(x_n) 不能为零（会导致除零错误）
• 初始猜测不当可能导致发散
• 如果根是重根，收敛速度会降为线性

示例：通过求解 f(x) = x^2 – 2 = 0 且 f'(x) = 2x 来求 sqrt(2)。从 x0 = 1.5 开始：

• x1 = 1.5 – (2.25 – 2) / 3 = 1.4167
• x2 = 1.4167 – (2.0069 – 2) / 2.8334 = 1.4142

仅需两次迭代，我们就得到了精确到 4 位小数的结果——速度惊人！

考试技巧：Edexcel 经常要求你将牛顿-拉弗森法应用于特定方程，并解释该方法在某些情况下可能失败的原因（例如，当 f'(x) = 0 或起始值接近驻点时）。务必展示完整的计算过程——分数是给代入过程的，而不仅仅是最终答案。

5. 三种方法的比较

方法	速度	需要导数？	可靠性	最佳用途
符号变化/对分法	慢（线性）	否	非常可靠	初步确定根的位置
不动点迭代法	线性	否	取决于 g'(x)	改写后的方程
牛顿-拉弗森法	快（二次）	是	对初值敏感	高精度求根

6. Edexcel 常见考试题型

• 证明根位于两个值之间：求 f(a) 和 f(b) 并注意符号变化——务必明确说明连续性。
• 执行指定次数的迭代：使用给定公式，在表格中清晰展示每一步。
• 判断迭代是否收敛：检查 |g'(x)| 小于 1 ——经典的 2-3 分题目。
• 含有三角函数的牛顿-拉弗森法：在计算器上使用弧度模式——这一点常让学生失分。
• 说明迭代失败的原因：常见原因包括 |g'(x)| 大于 1、除零错误或振荡。
• 在具体情境中应用数值方法：实际问题如求利率、抛体射程或种群模型。

7. 取得成功的实用技巧

1. 高效使用计算器。Edexcel 考试要求你使用 ANS 键或存储/调用功能来快速迭代。练习按键顺序，使其成为本能。
2. 三角函数始终使用弧度制。涉及 sin、cos 或 tan 的牛顿-拉弗森法必须使用弧度——使用角度模式得到的答案将是错误的。
3. 画图。f(x) 的粗略草图有助于你理解牛顿-拉弗森法可能失败的原因（例如，起始点靠近驻点，切线接近水平）。
4. 按要求的精度给出答案。如果题目要求 3 位小数，就精确提供 3 位——不多不少。在最后一步正确四舍五入。
5. 检查你的改写形式。对于不动点迭代，方程必须改写为真正满足 x = g(x) 的形式。一个常见错误是保留原始方程形式，导致错误结果。
6. 理解而非死记硬背。Edexcel 的问题常常问为什么某个方法收敛或发散。从概念上理解收敛条件比死记硬背公式更有价值。

---

Numerical methods are essential tools in a mathematician’s toolkit — mastering them will serve you well in A-Level exams and beyond.
数值方法是数学家工具箱中的必备工具——掌握它们将助你在 A-Level 考试及未来学习中取得优异成绩。

更多咨询请联系16621398022（同微信）