📚 Numerical Methods in IB and CIE Mathematics | IB CIE 数学：数值方法考点精讲
Numerical methods are powerful techniques that allow us to approximate solutions to equations, integrals, and other mathematical problems when exact algebraic solutions are impossible or impractical. Both IB (AA & AI) and CIE A-Level Mathematics courses expect students to understand and apply these iterative and approximate processes, analyse errors, and judge convergence.
当精确的代数解无法求出或不切实际时，数值方法提供了逼近方程根、积分值等数学问题解的有效手段。IB（分析与方法、应用与解释）和 CIE A-Level 数学大纲都要求学生理解并应用这些迭代与近似过程，分析误差并判断收敛性。
In this revision guide, we cover the key numerical methods examined in IB and CIE syllabi: locating roots, bisection, linear interpolation, fixed-point iteration, Newton-Raphson, numerical integration (trapezium and Simpson’s rules), error bounds, and exam strategies. Each section presents concepts in bilingual paired paragraphs to reinforce understanding.
在本复习指南中，我们将覆盖 IB 和 CIE 考试中的核心数值方法：根的定位、二分法、线性插值、不动点迭代、牛顿−拉夫森法、数值积分（梯形法则与辛普森法则）、误差界以及应试策略。每一节均以中英文配对段落呈现，帮助加深理解。
1. Introduction to Numerical Methods | 数值方法导论
Numerical methods are essential when a problem cannot be solved by hand using algebraic manipulation, or when the exact answer is irrational and we need a decimal approximation. They rely on iterative processes, starting from an initial guess and refining it step by step until a desired accuracy is reached.
当一个问题无法通过代数手工求解，或精确解为无理数而需要小数近似时，数值方法便不可或缺。它们依赖迭代过程，从一个初始猜测值出发，一步一步将其精化，直至达到所需精度。
Common areas where numerical methods are applied include finding roots of equations with no closed-form solution, evaluating complex definite integrals, and solving differential equations. In IB and CIE papers, you will be tested on both performing the methods and understanding their limitations, such as slow convergence or divergence.
数值方法的常见应用包括求解无封闭形式解的方程的根、计算复杂定积分以及解微分方程。在 IB 和 CIE 试卷中，你既会被要求执行这些方法，也会考察对其局限性的理解，比如收敛慢或发散。
Key vocabulary you must master: iteration, approximation, error, tolerance, convergence, divergence, and order of convergence. These terms form the conceptual backbone of every numerical topic.
必须掌握的关键术语有：迭代、逼近、误差、容差、收敛、发散以及收敛阶。这些概念构成了每个数值专题的理论支柱。
2. Locating Roots: The Sign Change Rule | 根的存在性：符号变化法则
If a continuous function f(x) changes sign over an interval [a, b], i.e., f(a) × f(b) < 0, then by the Intermediate Value Theorem there exists at least one root α in (a, b). This is the starting point for bracketing methods like bisection and linear interpolation.
若一个连续函数 f(x) 在区间 [a, b] 上改变符号，即 f(a) × f(b) < 0，那么根据介值定理，在 (a, b) 内至少存在一个根 α。这是二分法和线性插值等括根法的出发点。
In an exam, you may be asked to show that a root lies in a given interval by evaluating f at the endpoints and noting the sign difference. Always state that f is continuous on [a, b] before invoking the sign change rule.
在考试中，你可能需要证明根位于给定区间内：计算区间端点处的函数值，并指出符号差异。在使用符号变化法则前，务必先说明 f 在 [a, b] 上连续。
For example, to locate √2 using f(x) = x² − 2, f(1) = −1 < 0 and f(2) = 2 > 0, so a root lies between 1 and 2. This simple check is often worth marks and underpins the bracketing methods.
例如，用 f(x) = x² − 2 定位 √2：f(1) = −1 < 0，f(2) = 2 > 0，故在 1 和 2 之间存在一个根。这一简单的检验往往能得分，并且是括根法的基础。
3. The Bisection Method | 二分法
The bisection method repeatedly halves the interval [a, b] that brackets a root. At each step, compute the midpoint c = (a + b)/2. If f(c) = 0 (or |f(c)| < tolerance), then c is the root. Otherwise, check the sign of f(a) × f(c): if negative, the root lies in [a, c]; if positive, it lies in [c, b]. Replace the appropriate endpoint and repeat.
二分法不断将包含根的区间 [a, b] 对半分开。每一步计算中点 c = (a + b)/2。若 f(c) = 0（或 |f(c)| < 容差），则 c 即为根。否则检查 f(a) × f(c) 的符号：若为负，则根位于 [a, c]；若为正，则根位于 [c, b]。替换相应的端点并重复。
The number of iterations n required to achieve an interval width smaller than a given tolerance ε can be found from (b − a)/2ⁿ < ε, or n > log₂((b − a)/ε). The bisection method is guaranteed to converge if f is continuous on [a, b] and f(a)f(b) < 0.
要使区间宽度小于给定容差 ε，所需的迭代次数 n 可由 (b − a)/2ⁿ < ε 得出，即 n > log₂((b − a)/ε)。只要 f 在 [a, b] 上连续且 f(a)f(b) < 0，二分法必定收敛。
An upper bound for the error after n iterations is |α − cₙ| ≤ (b − a)/2ⁿ⁺¹, where cₙ is the n-th midpoint. Use this to answer questions about accuracy when the exact root is unknown.
n 次迭代后的误差上界为 |α − cₙ| ≤ (b − a)/2ⁿ⁺¹，其中 cₙ 为第 n 个中点。当精确根未知时，可用此式回答有关精度的问题。
4. Linear Interpolation (Regula Falsi) | 线性插值（试位法）
Linear interpolation, or the method of false position, approximates the root by the x-intercept of the chord joining (a, f(a)) and (b, f(b)). The formula is: x₁ = (a f(b) − b f(a)) / (f(b) − f(a)). Then the interval is updated by examining the sign of f(x₁), similar to bisection.
线性插值，又称试位法，利用连接 (a, f(a)) 与 (b, f(b)) 的弦的 x 截距来逼近根。公式为：x₁ = (a f(b) − b f(a)) / (f(b) − f(a))。然后通过检查 f(x₁) 的符号来更新区间，与二分法类似。
This method often converges faster than bisection when the function is approximately linear near the root. However, it can suffer from one endpoint ‘sticking’ if the function is highly curved, leading to slow convergence. CIE and IB may ask you to perform one or two iterations and comment on the method’s behaviour.
当函数在根附近近似线性时，此法通常比二分法收敛更快。但若函数高度弯曲，可能导致一个端点“粘滞”，造成收敛缓慢。CIE 与 IB 可能会要求你进行一到两次迭代，并评论该方法的表现。
5. Fixed-Point Iteration | 不动点迭代
Fixed-point iteration seeks to solve f(x) = 0 by rewriting it as x = g(x). Starting from an initial guess x₀, the iteration is xₙ₊₁ = g(xₙ). A solution α satisfying α = g(α) is a fixed point of g and a root of f.
不动点迭代通过将 f(x) = 0 改写为 x = g(x) 来求解。从初始猜测 x₀ 出发，迭代式为 xₙ₊₁ = g(xₙ)。满足 α = g(α) 的解 α 是 g 的不动点，也是 f 的根。
Crucially, the iteration converges if |g′(x)| < 1 in a neighbourhood of α. If |g′(α)| > 1, the process diverges. Examiners often ask to rearrange an equation into a suitable form and test for convergence using a derivative bound near the initial guess.
关键之处在于，若在 α 的某个邻域内 |g′(x)| < 1，则迭代收敛；若 |g′(α)| > 1，则过程发散。考官常要求将方程改写为合适的形式，并利用初始猜测附近的导数界限检验收敛性。
For example, solving x² − 3 = 0 can be rearranged as x = ½(x + 3/x) or x = 3/x. The first performs well (Newton’s method), the second diverges unless very close. Always justify your chosen iteration.
例如，解 x² − 3 = 0 可改写为 x = ½(x + 3/x) 或 x = 3/x。前者表现良好（即牛顿法），后者除非非常接近，否则发散。务必要为你选择的迭代式提供依据。
6. The Newton-Raphson Method | 牛顿−拉夫森法
The Newton-Raphson method is a powerful iterative technique for solving f(x) = 0, using the tangent line at the current estimate. The update formula is:
xₙ₊₁ = xₙ − f(xₙ) / f′(xₙ)
牛顿−拉夫森法是一种强大的迭代技术，利用当前估计值处的切线来求解 f(x) = 0。更新公式为：
xₙ₊₁ = xₙ − f(xₙ) / f′(xₙ)
It converges quadratically near a simple root if f′(α) ≠ 0 and f is twice differentiable, meaning the number of correct decimal places roughly doubles each step. However, it can fail if f′(xₙ) ≈ 0, leading to a near-vertical tangent and large jumps, or if the initial guess is not sufficiently close to the root.
在单根附近若 f′(α) ≠ 0 且 f 二次可微，该方法为二次收敛，意味着正确的小数位数大约每一步翻倍。然而，若 f′(xₙ) ≈ 0，切线近乎垂直，会产生大幅跳跃；若初始猜测离根不够近，也可能失败。
In examinations, you may be required to derive the formula geometrically, apply it to a given function, and discuss scenarios where it might fail. Always check whether the sequence is approaching the root by comparing successive iterates.
考试中可能要求你从几何上推导公式，将其应用于给定函数，并讨论可能失效的情形。务必通过比较相邻迭代值来检查序列是否趋向于根。
7. Convergence and Divergence of Iterative Methods | 迭代方法的收敛与发散
Convergence refers to how quickly an iterative sequence approaches the true root. The bisection method has linear convergence (order 1) and a guaranteed error reduction by half each step. Fixed-point iteration converges linearly if 0 < |g′(α)| < 1, but can be superlinear if g′(α) = 0. Newton-Raphson typically has quadratic convergence (order 2) near a simple root.
收敛性描述迭代序列逼近真根的快慢。二分法为线性收敛（1 阶），每一步误差保证减半。不动点迭代在 0 < |g′(α)| < 1 时线性收敛，若 g′(α) = 0 则可达到超线性收敛。牛顿−拉夫森法在单根附近通常为二次收敛（2 阶）。
Divergence occurs when iterates move away from the root. For fixed-point iteration, |g′(x)| > 1 near the root causes divergence. Newton-Raphson can oscillate or overshoot for poor initial guesses or when the derivative is near zero. Graphical illustrations using a ‘staircase’ or ‘cobweb’ diagram are often examined to visualise these behaviours.
当迭代值远离根时则发生发散。对于不动点迭代，根附近 |g′(x)| > 1 会导致发散。牛顿−拉夫森法在初始猜测不佳或导数为零附近时，可能出现振荡或越界。考试常用“阶梯图”或“蛛网图”将这些行为可视化。
Understand the concept of order of convergence: if |eₙ₊₁| ≈ C |eₙ|ᵖ for large n, then p is the order. This is particularly relevant in IB Analysis & Approaches and CIE Further Mathematics.
要理解收敛阶的概念：若对大 n 有 |eₙ₊₁| ≈ C |eₙ|ᵖ，则 p 为收敛阶。这在 IB 分析与方法以及 CIE 进阶数学中尤为重要。
8. Numerical Integration: The Trapezium Rule | 数值积分：梯形法则
The trapezium rule approximates a definite integral ∫ₐᵇ f(x) dx by dividing the area under the curve into n trapezoids of equal width h = (b − a)/n. The approximate formula is:
∫ₐᵇ f(x) dx ≈ h/2 [y₀ + 2(y₁ + y₂ + … + yₙ₋₁) + yₙ]
where yᵢ = f(a + i h).
梯形法则通过将曲线下的区域分成 n 个等宽 h = (b − a)/n 的梯形来逼近定积分 ∫ₐᵇ f(x) dx。近似公式为：
∫ₐᵇ f(x) dx ≈ h/2 [y₀ + 2(y₁ + y₂ + … + yₙ₋₁) + yₙ]
其中 yᵢ = f(a + i h)。
Increasing n improves accuracy, but the error is proportional to h² (for a single panel). The trapezium rule is exact for linear functions. Exam questions often ask for approximations using a given number of strips, then compare with the exact value to compute the error.
增加 n 可提升精度，但单条梯形误差与 h² 成正比。梯形法则对线性函数是精确的。考题常要求按给定条数进行近似，再与精确值比较以计算误差。
Remember that the trapezium rule can overestimate or underestimate depending on the concavity of f. If f″(x) > 0 (convex), the rule overestimates; if f″(x) < 0 (concave), it underestimates.
记住，梯形法则会根据 f 的凹凸性产生高估或低估。若 f″(x) > 0（凸函数），则高估；若 f″(x) < 0（凹函数），则低估。
9. Simpson’s Rule | 辛普森法则
Simpson’s rule provides a more accurate approximation by fitting quadratic curves through sets of three points. It requires an even number of strips n (n must be even). With h = (b − a)/n, the rule is:
∫ₐᵇ f(x) dx ≈ h/3 [y₀ + 4(y₁ + y₃ + … + yₙ₋₁) + 2(y₂ + y₄ + … + yₙ₋₂) + yₙ]
辛普森法则通过过每三个点拟合二次曲线，提供了更高精度的近似。它要求分条数 n 为偶数（n 必须为偶数）。设 h = (b − a)/n，法则为：
∫ₐᵇ f(x) dx ≈ h/3 [y₀ + 4(y₁ + y₃ + … + yₙ₋₁) + 2(y₂ + y₄ + … + yₙ₋₂) + yₙ]
The error is proportional to h⁴, making it much more accurate for smooth functions. Simpson’s rule is exact for polynomials up to degree 3. In both IB and CIE, you must know when Simpson’s rule is applicable and how to apply it correctly, especially paying attention to the alternating 4 and 2 multipliers.
误差与 h⁴ 成正比，因此对光滑函数精度高得多。辛普森法则对次数不超过 3 的多项式是精确的。在 IB 和 CIE 中，你需知道何时适用辛普森法则，以及如何正确应用，尤其注意交替出现的乘数 4 和 2。
When comparing the trapezium and Simpson’s rules, note that for the same number of function evaluations, Simpson’s rule is generally more accurate, but it demands an even number of intervals. Both are efficient tools for approximating integrals without an elementary antiderivative.
在比较梯形法则与辛普森法则时，注意对于相同数量的函数求值，辛普森法则通常更精确，但要求区间数为偶数。两者都是在无初等反导数时逼近积分的有效工具。
10. Error Analysis, Bounds, and Exam Tips | 误差分析、误差界与应试技巧
Understanding error is crucial. Absolute error = |approximate value − true value|. Relative error = absolute error / |true value|. In many numerical problems, we stop iterating when successive approximations differ by less than a specified tolerance, or when the interval width (bisection) becomes sufficiently small.
理解误差至关重要。绝对误差 = |近似值 − 真实值|。相对误差 = 绝对误差 / |真实值|。在许多数值问题中，当相邻两次近似值之差小于指定容差，或区间宽度（二分法）足够小时，我们便停止迭代。
For bracketing methods, the final midpoint can be taken as the root estimate, and the error bound is half the last interval width. For iterative methods without brackets, an error estimate can be obtained from |xₙ₊₁ − xₙ|, but this is not a rigorous bound.
对于括根法，可取最后区间中点为根的估计值，误差界为最终区间宽度的一半。对于无区间的迭代法，可从 |xₙ₊₁ − xₙ| 得到误差估计，但这并非严格的误差界。
Common exam pitfalls: Forgetting to check continuity before applying sign change; using the wrong formula for Newton-Raphson or Simpson’s rule; misapplying the convergence condition for fixed-point iteration; and rounding too early, which can propagate errors. Always work to a higher precision during iterations and round only at the final answer.
常见考试失分点：应用符号变化前忘记检查连续性；用错牛顿−拉夫森或辛普森法则公式；误用不动点迭代的收敛条件；提前舍入导致误差传播。迭代过程中应始终以较高精度计算，仅在最终答案处进行舍入。
When answering ‘show that’ questions, clearly document your steps, state the method being used, and comment on why it works. If a graph or diagram is requested, use it to illustrate cobweb or staircase behaviour. Practice with past papers will build speed and confidence in these highly structured, mark-heavy topics.
在回答“证明”类问题时，清晰记录步骤，说明所用方法，并解释其原理。若要求作图，用图形说明蛛网或阶梯行为。通过历年真题练习，你将在这些结构性强、分值高的专题上提升速度与信心。
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