← 返回资讯屏轩教育 · 行业资讯

A-Level OCR Mathematics: Numerical Methods Key Concepts | 数值方法考点精讲

2026-07-15 01:14 · 本文部分内容由AI辅助创作

📚 A-Level OCR Mathematics: Numerical Methods Key Concepts | 数值方法考点精讲

In A-Level OCR Mathematics, numerical methods give you a set of powerful, systematic techniques for approximating solutions when exact algebraic answers are out of reach. Whether you are solving a non-linear equation that cannot be factorised, or evaluating a definite integral whose antiderivative has no elementary form, these iterative and rule-based approaches let you improve accuracy step by step. Mastering them matters for your exam, and it also builds a bridge to the computational mathematics used every day in science, and engineering and data analysis.

在 A-Level OCR 数学中，数值方法为你提供了一整套强大而系统的近似技巧，用来处理精确代数解难以求得的题目。不论你面对的是无法因式分解的非线性方程，还是没有初等原函数的定积分，这些基于迭代与规则的方法都能让你一步步提高精度。掌握它们不仅对考试至关重要，也为你通向科学与工程、数据分析中普遍使用的计算数学搭建了桥梁。

1. What Numerical Methods Actually Solve | 数值方法到底解决什么问题

numerical integration. Root finding deals with equations such as x³ − x − 1 = 0 or x = e⁻ˣ t gives an exact answer. Numerical integration deals with definite integrals such as ∫₀¹ e−x² dx, where the integrand has no elementary antiderivative at all.

OCR 大纲把数值方法分为两大家族：求根与数值积分。求根处理的是像 x³ − x − 1 = 0 或 x = e⁻ˣ 这类方程，它们既无法因式分解，也没有简单的变形能得到精确解。数值积分处理的则是像 ∫₀¹ e−x² dx 这样的定积分，其被积函数根本不存在初等原函数。

e tangent-based Newton-Raphson method. For numerical integration the trapezium rule is the core A-Level tool, and Simpson's rule is are the natural extension you meet in Further Mathematics and beyond.

在求根方面，你会接触到符号变化原理、写成 x = g(x) 的不动点迭代，以及基于切线的牛顿-拉弗森法。在数值积分方面，梯形法则是 A-Level 的核心工具，而辛普森法则则是你在进阶数学及后续课程中会遇到的延伸工具。

Every method has its own convergence behaviour and accuracy characteristics. OCR examiners therefore test two things at once: your computational fluency, and your conceptual understanding of why a method works, when it fails, and how accurate the answer is.

每种方法都有各自的收敛行为与精度特点。因此 OCR 考官同时考查两件事：你的计算熟练度，以及你对方法为何有效、何时失效、结果有多精确的概念性理解。

2. Locating Roots: The Change-of-Sign Principle | 定位根：符号变化原理

that f(x) = 0 has at least one root somewhere in the open interval (a, b). This is the theoretical foundation of every sign-change question.

如果函数 f(x) 在闭区间 [a, b] 上连续，且 f(a) 与 f(b) 异号，那么介值定理就保证了在开区间 (a, b) 内至少存在一个点使 f(x) = 0。这就是所有符号变化题目的理论基础。

mial it is continuous everywhere, and the sign change between x = 1 and x = 2 proves a root lies in (1, 2).

实际操作中，你在选定的 x 值处计算 f，寻找符号变化。以 f(x) = x³ − x − 1 为例：f(1) = 1 − 1 − 1 = −1，f(2) = 8 − 2 − 1 = 5。由于 f 是多项式，处处连续，而 x = 1 与 x = 2 之间的异号就证明了根落在 (1, 2) 内。

To narrow the root down to a required accuracy, keep bisecting the interval and re-evaluating. For a root correct to one decimal place you must find two values of x differing by at least 0.1 that give opposite signs.

要把根缩小到要求的精度，就不断把区间对半分并重新求值。若要求根精确到一位小数，你必须找到两个相差 0.1 的 x 值，使它们对应的函数值异号。

f(a) × f(b) < 0 and f continuous on [a, b] ⟹ at least one root in (a, b)

Examiners award marks for two statements: the sign change itself, and the explicit appeal to continuity. Writing only "the sign changed, so there is a root" without mentioning continuity is a very common way to lose a mark.

考官会为两句话给分：符号发生变化这一事实，以及明确引用连续性。只写“符号变了，所以有根”而不提连续性，是丢分的最常见方式之一。

3. Fixed-Point Iteration: x = g(x) | 不动点迭代：x = g(x)

the recurrence xₙ₊₁ = g(xₙ), starting from an initial estimate x₀ somewhere near the root.

不动点迭代的第一步是把 f(x) = 0 变形为 x = g(x) 的形式。接着用递推式 xₙ₊₁ = g(xₙ) 生成数列，初值 x₀ 取在根附近某个位置。

xₙ₊₁ = g(xₙ)

For f(x) = x³ − x − 1 = 0, one valid rearrangement is x = ∛ x₁ ≈ 1.3572, x₂ ≈ 1.3310, x₃ ≈ 1.3261, x₄ ≈ 1.3252, converging towards the root 1.3247...

对 f(x) = x³ − x − 1 = 0 来说，一个有效的变形是 x = ∛(x + 1)。从 x₀ = 1.5 出发可得 x₁ ≈ 1.3572、x₂ ≈ 1.3310、x₃ ≈ 1.3261、x₄ ≈ 1.3252，逐步收敛到根 1.3247...

广告 / Sponsored

🛒
TI-Nspire CX II CAS 图形计算器 A-Level数学
¥1298
📱 去京东购买
京东精选

is whether the iteration converges. Differentiating the rearrangement gives g′(x) = 1 ÷ (3∛ this is about 0.19, which is well below 1 in modulus, so the iteration converges steadily.

关键问题是这个迭代是否收敛。对变形式求导得 g′(x) = 1 ÷ (3∛((x + 1)²))。在根附近其值约为 0.19，模远小于 1，因此迭代稳定收敛。

26 at the root. Because the modulus exceeds 1 the sequence diverges violently, even from the same starting value.

相反，变形 x = x³ − 1 给出 g′(x) = 3x²，在根附近约为 5.26。由于模大于 1，数列会剧烈发散——即使使用同样的初值也一样。

Convergence: |g′(x)| < 1 near the root. Divergence: |g′(x)| > 1 near the root.

r as a cobweb spiralling inwards. If the diagram spirals outwards, the iteration diverges.

在几何上，你可以把迭代想象成阶梯图收敛到 y = g(x) 与 y = x 的交点，或者想象成蛛网向内盘旋。如果图形向外盘旋，则迭代发散。

4. The Newton-Raphson Method | 牛顿-拉弗森法

s y = f(xₙ) + f′(xₙ)(x − xₙ). Setting y = 0 and solving for x gives the iteration formula.

牛顿-拉弗森法利用当前估计点处的切线，跳到更优的估计点。曲线 y = f(x) 在 x = xₙ 处的切线为 y = f(xₙ) + f′(xₙ)(x − xₙ)。令 y = 0 解出 x，就得到迭代公式。

xₙ₊₁ = xₙ − f(xₙ) ÷ f′(xₙ)

1.325197, then x₃ ≈ 1.324718, matching the root to six decimal places in just three steps.

把它用于 f(x) = x³ − x − 1，其中 f′(x) = 3x² − 1，取 x₀ = 1.5：f(1.5) = 0.875，f′(1.5) = 5.75，于是 x₁ = 1.5 − 0.875 ÷ 5.75 ≈ 1.347826。继续迭代得 x₂ ≈ 1.325197，再得 x₃ ≈ 1.324718，仅三步就与根吻合到六位小数。

This rapid behaviour is called quadratic convergence: roughly speaking, the number of correct decimal places about doubles with each iteration, which is why Newton-Raphson is usually the fastest method on the OCR paper.

这种快速行为被称为二次收敛：粗略地说，每一次迭代后正确的小数位数大约翻倍，这也是牛顿-拉弗森法通常是 OCR 试卷上最快方法的原因。

The price is that you must be able to differentiate f, you must choose a starting value close to the target root, and you must avoid points where f′(xₙ) is zero or very small, because the formula then shoots off to a distant value.

代价是你必须能对 f 求导，必须选择靠近目标根的初值，还必须避开 f′(xₙ) 为零或极小的点，因为此时公式会把估计值抛到很远的地方。

5. Comparing the Root-Finding Methods | 三种求根方法对比

Each method trades speed against reliability. The table below summarises the comparison OCR examiners most often want to see in a "discuss the method" question.

每种方法都在速度与可靠性之间做取舍。下表总结了 OCR 考官在“讨论方法”类题目中最想看到的对比要点。

Method | 方法 Needs f′? | 是否需要导数 Speed | 速度 Main risk | 主要风险
Change of sign | 符号变化 No | 不需要 Slow, but safe | 慢但稳妥 Only brackets a root, does not refine it | 只能定位区间，不能求精确值
Fixed-point iteration | 不动点迭代 No, but needs g′(x) | 不需要，但要检验 g′(x) Linear convergence | 线性收敛 Diverges if |g′(x)| > 1 | 若 |g′(x)| > 1 则发散
Newton-Raphson | 牛顿-拉弗森法 Yes | 需要 Quadratic convergence | 二次收敛 Fails near f′(x) = 0 | 在 f′(x) = 0 附近失效

A classic exam instruction is "use a suitable method to find the root correct to 4 decimal places". Change of sign alone will not reach that accuracy efficiently, so you should switch to an iteration and show enough iterates to justify the final rounding.

一道经典的考题指令是“用合适的方法求该根，精确到四位小数”。单靠符号变化法无法高效达到这个精度，因此你必须切换到迭代法，并展示足够多的迭代值来支撑最终的舍入结果。

6. When the Methods Fail | 方法失效的情形

Numerical methods are not guaranteed to work, and OCR regularly asks you to explain a failure. The most common failures are worth knowing by heart.

数值方法并不保证一定成功，而 OCR 经常要求你解释失效原因。以下几种常见失效情形值得牢记。