📚 A-Level Mathematics: Numerical Methods Key Points | A-Level 数学：数值方法考点精讲
Numerical methods provide powerful techniques for approximating solutions to equations, integrals, and differential equations when analytical approaches are impractical or impossible. In A-Level Mathematics, these methods form a key part of the syllabus, demanding both procedural fluency and a solid understanding of convergence, error, and limitations. This article covers all essential numerical methods topics, with clear explanations in both English and Chinese.
当解析方法不切实际或无法使用时，数值方法为方程、积分和微分方程提供了强有力的近似求解技术。在 A-Level 数学中，这些方法是教学大纲的核心部分，既要求熟练的操作步骤，也要求对收敛性、误差和局限性有扎实的理解。本文涵盖所有重要的数值方法考点，并用中英双语进行清晰讲解。
1. Overview of Numerical Methods | 数值方法概述
Numerical methods are algorithms that use iterative calculations to produce approximate results. They are indispensable when functions cannot be integrated symbolically or when equations like x⁵ + x − 1 = 0 have no simple algebraic solution.
数值方法是通过迭代计算得到近似结果的算法。当函数无法用符号积分，或者像 x⁵ + x − 1 = 0 这样的方程没有简单的代数解时，这些方法是不可或缺的。
Key concepts include error control, the speed of convergence, and the behaviour of an iteration near the true value. Understanding the theory behind the processes allows you to choose the most suitable method and to predict whether it will succeed.
关键概念包括误差控制、收敛速度以及迭代在真实值附近的行为。理解这些过程背后的理论，能够帮助你选择最合适的方法，并预判其能否成功。
2. Locating Roots Using Sign Change | 利用符号变化定位根
To locate a root of a continuous function f(x), we look for an interval [a, b] where f(a) and f(b) have opposite signs. By the Intermediate Value Theorem, a root exists in (a, b) provided the function does not have a discontinuity there.
为了找到连续函数 f(x) 的根，我们寻找一个使得 f(a) 和 f(b) 异号的区间 [a, b]。根据介值定理，只要函数在该区间内没有间断点，区间 (a, b) 内就至少存在一个根。
In practice, we evaluate f(x) at several points and stop when we find f(a) × f(b) < 0. The interval can then be narrowed by bisection or by testing midpoints. This sign‑change test is often the first step before applying a more refined iterative method.
实际操作中，我们计算若干点的函数值，一旦发现 f(a) × f(b) < 0 就停止。接着可用二分法或检测中点来缩窄区间。符号变化测试通常是使用更精确迭代方法之前的第一步。
3. Fixed-Point Iteration | 不动点迭代
Given an equation f(x) = 0, we can often rearrange it into the form x = g(x). Starting from an initial guess x₀, the iteration xₙ₊₁ = g(xₙ) generates a sequence that may converge to a root α satisfying α = g(α).
给定方程 f(x) = 0，我们通常能将其重写为 x = g(x) 的形式。从初始猜测 x₀ 开始，迭代 xₙ₊₁ = g(xₙ) 会产生一个数列，该数列可能收敛到满足 α = g(α) 的根 α。
For example, the equation x³ + x − 1 = 0 can be rearranged as x = 1/(x² + 1). Choosing x₀ = 0.5, successive values are computed: x₁ = 1/(0.5²+1) = 0.8, x₂ = 1/(0.8²+1) = 0.6098, and so on, approaching the root near 0.6823.
例如，方程 x³ + x − 1 = 0 可变形为 x = 1/(x² + 1)。选取 x₀ = 0.5，依次计算：x₁ = 1/(0.5²+1) = 0.8，x₂ = 1/(0.8²+1) = 0.6098，以此类推，趋近于约 0.6823 的根。
4. Cobweb and Staircase Diagrams | 蛛网图与阶梯图
The behaviour of the sequence xₙ can be visualised by plotting the lines y = g(x) and y = x. Starting at (x₀, 0), a vertical move to y = g(x₀), then a horizontal move to y = x, gives x₁. Repeating produces either a cobweb (oscillating convergence) or a staircase (monotonic convergence).
数列 xₙ 的行为可以通过绘制直线 y = g(x) 和 y = x 来直观呈现。从 (x₀, 0) 出发，垂直移动到 y = g(x₀)，再水平移动到 y = x，就得到 x₁。重复这一过程会产生蛛网图（振荡收敛）或阶梯图（单调收敛）。
If the iteration diverges, the path spirals outward. These diagrams help you understand why a good initial guess is important and how the slope of g(x) influences the outcome.
若迭代发散，路径会向外螺旋。这些图能帮助你理解为什么好的初始猜测很重要，以及 g(x) 的斜率如何影响结果。
5. Convergence Criteria for Fixed-Point Iteration | 不动点迭代的收敛准则
For the iteration xₙ₊₁ = g(xₙ) to converge to a root α, we need |g'(α)| < 1. More generally, if |g'(x)| < 1 for all x in an interval containing α, and the starting value lies in that interval, the iteration will converge.
要使迭代 xₙ₊₁ = g(xₙ) 收敛到根 α，必须满足 |g'(α)| < 1。更一般地，如果在包含 α 的区间内所有 x 都满足 |g'(x)| < 1，且初始值也在该区间内，迭代就会收敛。
If |g'(α)| > 1, the iteration repels from the root; if |g'(α)| = 1, the test is inconclusive. This derivative condition is central to choosing a suitable rearrangement of f(x) = 0.
若 |g'(α)| > 1，迭代会从根处发散；若 |g'(α)| = 1，该判别法无法得出结论。这一导数条件是选择合适的 f(x) = 0 变形形式的核心依据。
6. The Newton-Raphson Method | 牛顿-拉夫森法
The Newton-Raphson method uses the tangent line to approximate the root. The iteration formula is xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ). It converges quadratically near a simple root, provided f'(xₙ) ≠ 0.
牛顿-拉夫森法利用切线来逼近根，其迭代公式为 xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ)。在单根附近且 f'(xₙ) ≠ 0 时，该方法具有二次收敛性，非常迅速。
xₙ₊₁ = xₙ − f(xₙ)/f'(xₙ)
For example, to find √2, we solve x² − 2 = 0, giving f'(x) = 2x. Starting with x₀ = 1, we get x₁ = 1 − (−1)/(2) = 1.5, x₂ = 1.5 − (0.25)/(3) ≈ 1.4167, converging rapidly to 1.41421356…
例如，求 √2 时解 x² − 2 = 0，f'(x) = 2x。从 x₀ = 1 开始，得 x₁ = 1 − (−1)/(2) = 1.5，x₂ = 1.5 − (0.25)/(3) ≈ 1.4167，迅速收敛到 1.41421356…
7. Failures of the Newton-Raphson Method | 牛顿-拉夫森法的失效情况
The method fails if the derivative f'(xₙ) is zero or very small, leading to division by a near‑zero number and a drastic jump. It can also fail when the initial guess is too far from the root, causing the tangent to shoot away, or when the function oscillates, trapping the iteration in a cycle.
当导数 f'(xₙ) 为零或非常小时，除以接近零的数会导致大幅跳跃，方法失效。如果初始猜测离根太远，切线可能远离目标区域；或者函数振荡时，迭代可能陷入循环而无法收敛。
An example is f(x) = x³ − 2x + 2 with x₀ = 0, producing x₁ = 1, x₂ = 0, alternating forever. Recognising these pitfalls helps you select a better start value or a different method.
例如 f(x) = x³ − 2x + 2，若 x₀ = 0，产生 x₁ = 1，x₂ = 0，永远交替。识别这些陷阱有助于选择更好的初始值或其他方法。
8. Numerical Integration: The Trapezium Rule | 数值积分：梯形法则
The trapezium rule approximates the area under a curve by dividing the interval [a, b] into n equal strips of width h = (b − a)/n and replacing the curve with straight line segments.
梯形法则通过将区间 [a, b] 等分为 n 个宽度为 h = (b − a)/n 的小段，并用直线段代替曲线，来近似曲线下的面积。
∫ₐᵇ f(x) dx ≈ ½h [y₀ + 2(y₁ + y₂ + … + yₙ₋₁) + yₙ]
Here yᵢ = f(xᵢ) and xᵢ = a + ih. The rule sums the areas of n trapezoids. Increasing n (i.e. decreasing h) usually improves accuracy, but at the cost of more calculations.
其中 yᵢ = f(xᵢ)，xᵢ = a + ih。该法则求 n 个梯形面积之和。增加 n（即减小 h）通常会提高精度，但会带来更多计算量。
9. Overestimation and Underestimation | 高估与低估
Whether the trapezium rule overestimates or underestimates the true integral depends on the curve’s concavity. If the function is concave up (f”(x) > 0), the tops of the strips lie above the curve, giving an overestimate. If the function is concave down (f”(x) < 0), the rule underestimates the area.
梯形法则是高估还是低估真实积分值，取决于曲线的凹凸性。如果函数凹向上（f”(x) > 0），梯形顶部在曲线上方，结果为高估；如果函数凹向下（f”(x) < 0），法则会低估面积。
This insight allows you to state whether an approximation is an upper or lower bound without evaluating the exact integral, which is often useful in exam questions about errors.
这一认识使你无需计算精确积分，就能判断近似值是上限还是下限，在涉及误差的考题中非常实用。
10. Error and Accuracy in Numerical Integration | 数值积分的误差与精度
The error in the trapezium rule is roughly proportional to h², assuming sufficient smoothness. Halving h reduces the error by about a factor of 4. The midpoint rule and Simpson’s rule offer higher accuracy (with errors proportional to h² and h⁴ respectively) and sometimes appear as extensions.
在函数充分光滑的前提下，梯形法则的误差大致与 h² 成正比。将 h 减半，误差约降低为原来的四分之一。中点法和辛普森法提供更高的精度（误差分别与 h² 和 h⁴ 成正比），有时作为扩展内容出现。
For root‑finding, the required accuracy is usually stated in decimal places. An answer correct to, say, 3 decimal places means the true root lies within ±0.0005 of the approximation, which can be checked by a sign change in a sufficiently narrow interval.
对于求根，所需精度通常以小数位数给出。例如，精确到 3 位小数意味着真实根在近似值的 ±0.0005 范围内，这可以通过在足够窄的区间内检验符号变化来确认。
11. Euler’s Method for Differential Equations | 欧拉法解微分方程
Euler’s method solves first‑order ordinary differential equations of the form dy/dx = f(x, y) with an initial condition y(x₀) = y₀. Using a small step size h, successive points are generated by yₙ₊₁ = yₙ + h f(xₙ, yₙ), and xₙ₊₁ = xₙ + h.
欧拉法用于求解形如 dy/dx = f(x, y) 的一阶常微分方程，给定初始条件 y(x₀) = y₀。选取小步长 h，通过 yₙ₊₁ = yₙ + h f(xₙ, yₙ) 以及 xₙ₊₁ = xₙ + h 逐步生成各点。
yₙ₊₁ = yₙ + h f(xₙ, yₙ)
Though simple, Euler’s method accumulates errors quickly if h is not sufficiently small. It provides a stepping stone to more advanced techniques like the midpoint method or Runge‑Kutta, and is examined in many A‑Level specifications.
欧拉法虽然简单，但如果 h 不够小，误差累积很快。它是通向中点法或龙格-库塔法等更高级方法的阶梯，许多 A-Level 考纲都会考查。
12. Choosing the Right Numerical Method | 选择合适的方法
No single method is best for all problems. For root‑finding, if f'(x) is easy to compute and a good initial guess is available, Newton‑Raphson is often fastest. If derivatives are messy or the root is needed to only a few decimal places, fixed‑point iteration may be simpler. The sign‑change method is useful for establishing existence but not for rapid convergence.
没有一种方法适合所有问题。对于求根，如果 f'(x) 容易计算且有较好的初始猜测，牛顿-拉夫森法通常最快；如果导数求解复杂或仅需几位小数，不动点迭代更简单。符号变化法用于确认根的存在，但本身收敛不快。
For integration, the trapezium rule offers a balance between simplicity and accuracy, especially when combined with increasing strips. Euler’s method is a natural first choice for differential equations when analytical solutions are unavailable.
对于积分，梯形法则在简单性和精度之间取得平衡，结合增多分段数效果更好。当微分方程没有解析解时，欧拉法很自然地成为首选。
Understanding the strengths and weaknesses of each technique is essential for tackling A‑Level exam questions confidently.
理解每种技术的优缺点，对于自信地应对 A-Level 考题至关重要。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
屏轩国际教育cambridge primary/secondary checkpoint, cat4, ukiset,ukcat,igcse,alevel,PAT,STEP,MAT, ibdp,ap,ssat,sat,sat2课程辅导，国外大学本科硕士研究生博士课程论文辅导Cancel reply