📚 Infinite Series: Convergence and Summation | 无穷级数的收敛与求和
Infinite series appear throughout A-Level mathematics, from geometric progressions to binomial expansions. The central questions are simple: does the sum approach a finite value, and if so, what is it? This article builds a complete toolkit for answering both questions.
无穷级数贯穿 A-Level 数学的始终，从等比数列到二项展开，核心问题只有两个：级数是否趋近于某个有限值？若收敛，其和是多少？本文将为你搭建一套完整的解题工具。
1. Sequences and Partial Sums | 数列与部分和
Given an infinite sequence a₁, a₂, a₃, …, we form the partial sums Sₙ = a₁ + a₂ + … + aₙ. The infinite series Σ aₙ (from n=1 to ∞) is defined as the limit of the sequence of partial sums: S = limₙ→∞ Sₙ. If this limit exists and is finite, the series converges and S is its sum; otherwise it diverges.
给定无穷数列 a₁, a₂, a₃, …，我们构造部分和 Sₙ = a₁ + a₂ + … + aₙ。无穷级数 Σ aₙ（n 从 1 到 ∞）定义为部分和序列的极限：S = limₙ→∞ Sₙ。若该极限存在且有限，则称级数收敛，S 即为级数的和；否则称级数发散。
'Converges' means the partial sums settle on a fixed number; it does not mean the individual terms shrink to zero quickly. The journey of infinitely many additions can end at a finite destination.
“收敛”意味着部分和稳定在一个固定数值，并不意味着各项迅速趋于零。无限次相加的旅程也可以到达有限的终点。
A divergent series may grow without bound (like 1 + 2 + 3 + …), oscillate forever (like 1 − 1 + 1 − 1 + …), or simply fail to approach any limit.
发散级数可能无限增大（如 1 + 2 + 3 + …），可能永远振荡（如 1 − 1 + 1 − 1 + …），也可能根本不趋近任何极限。
2. The Geometric Series | 等比级数
The geometric series a + ar + ar² + ar³ + … is the most important series in the course. Its partial sum is Sₙ = a(1 − rⁿ)/(1 − r) when r ≠ 1. Letting n → ∞, the term rⁿ tends to 0 if |r| < 1 and fails to settle otherwise. Hence the series converges if and only if |r| < 1, with sum
等比级数 a + ar + ar² + ar³ + … 是本课程最重要的级数。当 r ≠ 1 时，其部分和为 Sₙ = a(1 − rⁿ)/(1 − r)。令 n → ∞：当 |r| < 1 时 rⁿ 趋于 0，否则不趋于固定值。因此级数收敛当且仅当 |r| < 1，其和为
S = a / (1 − r)
If |r| ≥ 1, the series diverges. Common pitfalls: forgetting the first term a, confusing rⁿ with rⁿ⁻¹, or applying the formula when checking a finite geometric sum.
若 |r| ≥ 1，级数发散。常见陷阱：忘记首项 a，混淆 rⁿ 与 rⁿ⁻¹，或对有限等比和误用无穷公式。
Example: 5 + 5(0.2) + 5(0.2)² + … has a = 5, r = 0.2, |r| < 1, so S = 5/(1 − 0.2) = 6.25.
例：5 + 5(0.2) + 5(0.2)² + … 中 a = 5，r = 0.2，|r| < 1，故 S = 5/(1 − 0.2) = 6.25。
3. The nth Term Test for Divergence | 第 n 项判别法
A necessary condition for convergence is that the terms shrink to zero: if aₙ does not tend to 0 as n → ∞, then the series Σ aₙ must diverge. This test is cheap and should be applied first in every problem.
收敛的必要条件是通项趋于零：若 n → ∞ 时 aₙ 不趋于 0，则级数 Σ aₙ 必定发散。该判别法成本极低，每道题都应首先使用。
Warning: the converse is false. The harmonic series Σ 1/n has terms tending to zero, yet it diverges. Passing the nth term test proves nothing; more refined tests are required.
注意：逆命题不成立。调和级数 Σ 1/n 的通项趋于 0，但级数发散。通过第 n 项判别法什么都证明不了，需要更精细的判别法。
Example: Σ n/(n + 1) diverges because aₙ → 1 ≠ 0. Example: Σ (−1)ⁿ diverges because the terms oscillate between −1 and 1 and never approach 0.
例：Σ n/(n + 1) 发散，因为 aₙ → 1 ≠ 0。例：Σ (−1)ⁿ 发散，因为通项在 −1 与 1 之间振荡，从不趋于 0。
4. The Ratio Test | 比值判别法
For a series with positive terms, define L = limₙ→∞ |aₙ₊₁ / aₙ|. If L < 1 the series converges absolutely; if L > 1 it diverges; if L = 1 the test is inconclusive and another method must be used.
对于正项级数，定义 L = limₙ→∞ |aₙ₊₁ / aₙ|。若 L < 1，则级数绝对收敛；若 L > 1，则级数发散；若 L = 1，判别法失效，须改用其他方法。
The ratio test excels with factorials, exponentials, and products. Example: Σ n/2ⁿ has ratio |aₙ₊₁/aₙ| = (n+1)/(2n) → 1/2 < 1, so it converges.
比值判别法对含阶乘、指数和连乘积的级数特别有效。例：Σ n/2ⁿ 的比值 |aₙ₊₁/aₙ| = (n+1)/(2n) → 1/2 < 1，故收敛。
Example: Σ n!/10ⁿ gives ratio (n+1)/10 → ∞ > 1, so the series diverges despite the early terms being small.
例：Σ n!/10ⁿ 的比值 (n+1)/10 → ∞ > 1，尽管前几项很小，级数仍发散。
When L = 1, try the comparison tests or the integral test. The ratio test never decides the borderline cases such as Σ 1/n or Σ 1/n².
当 L = 1 时，可尝试比较判别法或积分判别法。比值判别法永远无法判断边界情形，如 Σ 1/n 或 Σ 1/n²。
5. Comparison and Limit Comparison Tests | 比较判别法与极限比较判别法
Direct comparison applies to series of non-negative terms. If aₙ ≤ bₙ for all large n and Σ bₙ converges, then Σ aₙ converges; if aₙ ≥ bₙ and Σ bₙ diverges, then Σ aₙ diverges. Choose a known benchmark: geometric series or p-series Σ 1/nᵖ.
直接比较适用于非负项级数。若对所有足够大的 n 有 aₙ ≤ bₙ 且 Σ bₙ 收敛，则 Σ aₙ 收敛；若 aₙ ≥ bₙ 且 Σ bₙ 发散，则 Σ aₙ 发散。基准级数通常选等比级数或 p 级数 Σ 1/nᵖ。
p-series converge for p > 1 and diverge for p ≤ 1. Thus Σ 1/n² converges while Σ 1/√n diverges because p = ½ ≤ 1.
p 级数在 p > 1 时收敛，在 p ≤ 1 时发散。因此 Σ 1/n² 收敛，而 Σ 1/√n 因 p = ½ ≤ 1 而发散。
Limit comparison is even more flexible: if limₙ→∞ aₙ/bₙ = c with 0 < c < ∞, then Σ aₙ and Σ bₙ share the same fate. This handles messy fractions like Σ (n² + 1)/(n⁴ − 3n), which behaves like Σ 1/n² and hence converges.
极限比较更加灵活：若 limₙ→∞ aₙ/bₙ = c 且 0 < c < ∞，则 Σ aₙ 与 Σ bₙ 同敛散。这能处理诸如 Σ (n² + 1)/(n⁴ − 3n) 的复杂分式，它等价于 Σ 1/n²，故收敛。
6. The Alternating Series Test | 交错级数判别法
An alternating series has the form Σ (−1)ⁿ bₙ or Σ (−1)ⁿ⁻¹ bₙ with bₙ > 0. Leibniz's test states: if bₙ is decreasing and bₙ → 0, then the series converges. Moreover, the error after N terms is bounded by the first neglected term: |Rₙ| ≤ bₙ₊₁.
交错级数形如 Σ (−1)ⁿ bₙ 或 Σ (−1)ⁿ⁻¹ bₙ，其中 bₙ > 0。莱布尼茨判别法指出：若 bₙ 单调递减且 bₙ → 0，则级数收敛。更进一步，截断 N 项后的误差不超过第一个被舍去的项：|Rₙ| ≤ bₙ₊₁。
Example: Σ (−1)ⁿ⁺¹ /n is the alternating harmonic series. Since 1/n decreases to 0, it converges — even though the ordinary harmonic series diverges. Its sum is ln 2 ≈ 0.693.
例：Σ (−1)ⁿ⁺¹ /n 即为交错调和级数。由于 1/n 递减趋于 0，它收敛——尽管普通调和级数发散。其和为 ln 2 ≈ 0.693。
The error estimate is a favourite exam question: for Σ (−1)ⁿ/n!, summing to n = 3 gives an error less than 1/24, so the approximation is accurate to 0.0417.
误差估计是考试常客：对 Σ (−1)ⁿ/n!，求和至 n = 3 时误差小于 1/24，即近似精确到 0.0417。
7. Telescoping Series | 裂项相消级数
Many series can be evaluated exactly by partial fractions. When aₙ = f(n) − f(n + 1), the partial sum collapses to Sₙ = f(1) − f(n + 1), and the infinite sum is limₙ→∞ Sₙ.
许多级数可通过裂项精确求和。若 aₙ = f(n) − f(n + 1)，部分和将坍缩为 Sₙ = f(1) − f(n + 1)，无穷和为 limₙ→∞ Sₙ。
The classic example: Σ 1/[n(n+1)] from n=1 to ∞. Write 1/[n(n+1)] = 1/n − 1/(n+1). Then Sₙ = 1 − 1/(n+1) → 1. The infinite sum is exactly 1.
经典例子：n 从 1 到 ∞ 的 Σ 1/[n(n+1)]。将 1/[n(n+1)] 拆成 1/n − 1/(n+1)，则 Sₙ = 1 − 1/(n+1) → 1，无穷和恰为 1。
Another pattern: 1/(4n² − 1) = ½(1/(2n−1) − 1/(2n+1)), so Σ 1/(4n² − 1) = ½. Spotting such factorisations turns a fearful infinite sum into two lines of cancellation.
另一模式：1/(4n² − 1) = ½(1/(2n−1) − 1/(2n+1))，故 Σ 1/(4n² − 1) = ½。识别此类分解能把可怕的无穷和化为两行相消。
8. Arithmetic-Geometric Series | 等差-等比混合级数
A sequence of the form (a + nd)rⁿ combines an arithmetic factor with a geometric one. A-Level questions often ask for the sum to infinity of Σ (a + nd)rⁿ, which converges when |r| < 1. The method is to multiply S by r and subtract:
形如 (a + nd)rⁿ 的数列同时含等差因子与等比因子。A-Level 常考 Σ (a + nd)rⁿ 的无穷和，当 |r| < 1 时收敛。方法是先写出 S，乘以 r 再相减：
S = a + (a+d)r + (a+2d)r² + …
rS = ar + (a+d)r² + …
(1 − r)S = a + dr + dr² + dr³ + …
Hence (1 − r)S = a + dr/(1 − r), giving S = a/(1 − r) + dr/(1 − r)². This formula is well worth memorising: it already contains the geometric series as the special case d = 0.
因此 (1 − r)S = a + dr/(1 − r)，得 S = a/(1 − r) + dr/(1 − r)²。此公式值得牢记：当 d = 0 时它正好退化为等比级数。
Example: Σ (2n + 1)(1/3)ⁿ from n=0 to ∞. Here a = 1, d = 2, r = 1/3, so S = 1/(2/3) + 2·(1/3)/(2/3)² = 1.5 + 1.5 = 3.
例：n 从 0 到 ∞ 的 Σ (2n + 1)(1/3)ⁿ。此时 a = 1，d = 2，r = 1/3，故 S = 1/(2/3) + 2·(1/3)/(2/3)² = 1.5 + 1.5 = 3。
9. Power Series and Maclaurin Expansions | 幂级数与麦克劳林展开
Power series are infinite polynomials Σ cₙxⁿ. The binomial series (1 + x)ᵖ = 1 + px + p(p−1)x²/2! + … converges absolutely for |x| < 1 and provides the standard expansion of fractional and negative powers; it is the key tool when α = −1 or a fraction.
幂级数是无穷多项式 Σ cₙxⁿ。二项级数 (1 + x)ᵖ = 1 + px + p(p−1)x²/2! + … 在 |x| < 1 时绝对收敛，为分数次幂与负次幂提供标准展开；当指数为 −1 或分数时它是核心工具。
The Maclaurin series for eˣ, ln(1 + x), sin x, cos x are expected knowledge. Each has a radius of convergence: eˣ and sin x converge for all x, while ln(1 + x) converges only for −1 < x ≤ 1 (at x = 1 it becomes the alternating harmonic series).
eˣ、ln(1 + x)、sin x、cos x 的麦克劳林级数属必备知识。各有收敛半径：eˣ 与 sin x 对所有 x 收敛，而 ln(1 + x) 仅在 −1 < x ≤ 1 收敛（在 x = 1 处恰为交错调和级数）。
Identities such as 1/(1 + x²) = Σ (−1)ⁿx²ⁿ for |x| < 1 arise from substituting x² into the geometric series. Recognising a power series as a disguised geometric series is a reliable route to both convergence intervals and exact sums.
恒等式 1/(1 + x²) = Σ (−1)ⁿx²ⁿ（|x| < 1）来自将 x² 代入等比级数。将幂级数识别为“伪装”的等比级数，是求收敛区间与精确和值的可靠途径。
10. Strategy Table and Final Advice | 策略表与最终建议
With so many tests, choose in order: glance at aₙ first, then use the table below.
面对众多判别法，按序选择：先瞄一眼 aₙ，再参考下表。
| Condition you notice | Test to use |
| aₙ does not → 0 | nth term test → divergence |
| Geometric pattern a rⁿ | |r| < 1 → sum a/(1−r) |
| Factorials or powers like xⁿ/n! | Ratio test |
| Rational functions, e.g. (n²+1)/(n⁴−3n) | Limit comparison with 1/nᵖ |
| Telescoping form f(n) − f(n+1) | Partial fractions and collapse |
| Alternating signs with decreasing terms | Alternating series test + error bound |
| (a + nd)rⁿ pattern | S − rS method or formula |
Finally, read exam questions with precision: 'show that the series converges' demands a justification, while 'find the sum to infinity' demands exact algebra, not a decimal approximation. Write out the test condition, state the limit clearly, and conclude with a full sentence.
最后，精确审题：“证明级数收敛”需要给出依据，而“求无穷和”需要精确代数运算而非小数近似。写清判别条件、明确极限值，并以完整句子下结论。
Mastery comes from classifying each series by its form. Memorise the geometric and p-series benchmarks, practise partial fractions, and the convergence of every A-Level series will become a routine decision.
掌握的关键在于按形式给级数分类。牢记等比级数与 p 级数基准，勤练裂项，那么每道 A-Level 级数题的敛散性判断都将成为常规操作。
Published by TutorHao | Mathematics Revision Series | aleveler.com
更多咨询请联系16621398022（同微信）
或截图后用「扫一扫 → 相册」选择本图