• Practice
  • About 22 min
  • 5 practice questions

Infinite series: convergence, binomial, exponential and logarithmic series

Understand when an infinite series has a sum, find sums of infinite GP and AGP series, expand (1 + x)ⁿ for a rational n, and use the exponential and logarithmic series.

If something looks wrong, please tell us.

By the end of this lesson you can

  • Explain convergence of a sequence and the sum of an infinite series through partial sums.
  • Find the sum of an infinite GP, an infinite AGP and a telescoping infinite series.
  • Expand (1 + x)ⁿ for a rational n when |x| < 1, and state where the expansion is valid.
  • Use the binomial series for approximations.
  • Write the exponential series for eˣ and the logarithmic series for log(1 + x).

Cut a cake in half and move one half to a second plate; cut what is left in half and move one piece again; keep going. The first plate holds 1/2, 1/4, 1/8, ... of the cake, which shrinks towards 0, and the second plate holds 1/2 + 1/4 + 1/8 + ..., which grows towards the whole cake. This lesson makes the idea of adding infinitely many numbers precise.

Convergence

A sequence (aₙ) converges to a number a, written lim aₙ = a as n → ∞, if for any small positive number you choose there is a stage after which every aₙ is closer to a than that number. The sequence 1, 1/10, 1/100, ... converges to 0. The sequence 1, 0, 1, 0, ... does not converge to anything. A convergent sequence has only one limit.

For an infinite series a₁ + a₂ + a₃ + ..., form the partial sums sₙ = a₁ + ... + aₙ. If (sₙ) converges to s, the series is convergent with sum s. For 1/2 + 1/4 + 1/8 + ..., sₙ = 1 − 1/2ⁿ, which tends to 1, so the sum is 1. The series 1 − 1 + 1 − 1 + ... has partial sums 1, 0, 1, 0, ..., so it has no sum; grouping its terms in different ways seems to give 0 or 1, which is a warning that rules for finite sums cannot be used blindly on infinite ones.

The Fibonacci sequence

The textbook also introduces the Fibonacci sequence, defined by x₁ = 1, x₂ = 1 and xₙ = xₙ₋₁ + xₙ₋₂ for n ≥ 3: 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, ... . It is named after Leonardo of Pisa (Fibonacci), whose book Liber Abaci of 1202 included it. A pattern to notice: every third term (2, 8, 34, 144, ...) is even, a multiple of x₃ = 2, and every fourth term (3, 21, 144, ...) is a multiple of x₄ = 3.

Infinite geometric and arithmetico-geometric series

For a GP with ratio r, sₙ = a(1 − rⁿ)/(1 − r). If |r| < 1, rⁿ → 0, so the series converges and a + ar + ar² + ... = a/(1 − r). If |r| ≥ 1 (and a ≠ 0), it does not converge. In particular 1/(1 − x) = 1 + x + x² + ... and 1/(1 + x) = 1 − x + x² − ... for |x| < 1. For an AGP with |r| < 1, letting n → ∞ in the finite formula gives S = a/(1 − r) + dr/(1 − r)².

Worked example 1: three infinite sums

(i) 0.333... = 3/10 + 3/100 + 3/1000 + ... is a GP with a = 3/10 and r = 1/10, so it equals (3/10)/(9/10) = 1/3. (ii) 1 + 3/2 + 5/4 + 7/8 + ... is an AGP with a = 1, d = 2, r = 1/2: S = 1/(1/2) + (2 × 1/2)/(1/2)² = 2 + 4 = 6. (iii) Σₙ₌₁^∞ 1/(n² + 3n + 2): since n² + 3n + 2 = (n + 1)(n + 2), the term is 1/(n + 1) − 1/(n + 2), so sₙ = 1/2 − 1/(n + 2), which tends to 1/2.

The binomial series for a rational index

For any rational n and |x| < 1: (1 + x)ⁿ = 1 + nx + n(n − 1)x²/2! + n(n − 1)(n − 2)x³/3! + ...

— Binomial theorem for a rational exponent

The textbook states this theorem without proof. When n is a positive integer the series stops after n + 1 terms; otherwise it goes on for ever and is valid only for |x| < 1. Four expansions are worth knowing by heart, all for |x| < 1: (1 + x)⁻¹ = 1 − x + x² − x³ + ...; (1 − x)⁻¹ = 1 + x + x² + x³ + ...; (1 − x)⁻² = 1 + 2x + 3x² + 4x³ + ...; (1 + x)⁻² = 1 − 2x + 3x² − 4x³ + ... .

Worked example 2: expansions and their ranges

(i) (1 + x)^(1/2) up to x³: n = 1/2, n(n − 1)/2! = (1/2)(−1/2)/2 = −1/8 and n(n − 1)(n − 2)/3! = (1/2)(−1/2)(−3/2)/6 = 1/16, so (1 + x)^(1/2) = 1 + x/2 − x²/8 + x³/16 − ..., |x| < 1. (ii) 1/(1 − 2x)² = (1 − y)⁻² with y = 2x: 1 + 2y + 3y² + 4y³ + ... = 1 + 4x + 12x² + 32x³ + ..., valid for |2x| < 1, that is |x| < 1/2. (iii) 1/(2 + x): first take out the 2, since the series needs 1 + (something small): 1/(2 + x) = (1/2)(1 + x/2)⁻¹ = (1/2)(1 − x/2 + x²/4 − x³/8 + ...) = 1/2 − x/4 + x²/8 − x³/16 + ..., valid for |x/2| < 1, that is |x| < 2.

Worked example 3: an approximate square root

Find √26 approximately. Write √26 = √(25 + 1) = 5(1 + 1/25)^(1/2). With x = 1/25 = 0.04, (1 + x)^(1/2) ≈ 1 + x/2 − x²/8 = 1 + 0.02 − 0.0002 = 1.0198. So √26 ≈ 5 × 1.0198 = 5.099. (Squaring: 5.099² ≈ 25.9998.) The method works because x is small, so later terms are tiny.

The exponential series

The series Σ xⁿ/n! converges for every real x, and its sum is eˣ: eˣ = 1 + x/1! + x²/2! + x³/3! + ..., where e = 1 + 1/1! + 1/2! + 1/3! + ... ≈ 2.71828. Replacing x by −x gives e⁻ˣ = 1 − x/1! + x²/2! − x³/3! + ... . Adding and subtracting: (eˣ + e⁻ˣ)/2 = 1 + x²/2! + x⁴/4! + ... and (eˣ − e⁻ˣ)/2 = x/1! + x³/3! + x⁵/5! + ... . For example e^(3x) = 1 + 3x + 9x²/2 + 9x³/2 + ..., since 3²/2! = 9/2 and 3³/3! = 27/6 = 9/2.

The logarithmic series

Here log means the natural logarithm, to base e. For |x| < 1 (and also at x = 1): log(1 + x) = x − x²/2 + x³/3 − x⁴/4 + .... Replacing x by −x: log(1 − x) = −x − x²/2 − x³/3 − ..., |x| < 1. Subtracting: log[(1 + x)/(1 − x)] = 2(x + x³/3 + x⁵/5 + ...), |x| < 1. For a series such as log(1 + 3x), put y = 3x: log(1 + 3x) = 3x − 9x²/2 + 9x³ − ..., valid for |3x| < 1, that is |x| < 1/3.

Worked example 4: log 2 from a fast series

To compute log 2, solve (1 + x)/(1 − x) = 2, which gives x = 1/3. Then log 2 = 2(1/3 + 1/81 + 1/1215 + 1/15309 + ...) ≈ 2(0.33333 + 0.01235 + 0.00082 + 0.00007) = 2 × 0.34657 ≈ 0.6931. This series gives four correct decimal places with only four terms, because x = 1/3 is small.

Infinite series in this chapter

  • Geometric family Needs |r| < 1.
    • a/(1 − r)
    • AGP: a/(1 − r) + dr/(1 − r)²
  • Binomial series Needs |x| < 1 after writing 1 + (small).
    • (1 + x)ⁿ for rational n
    • Approximate roots
  • Exponential series Valid for every real x.
    • eˣ = Σ xⁿ/n!
    • e⁻ˣ, (eˣ ± e⁻ˣ)/2
  • Logarithmic series Needs |x| < 1.
    • log(1 + x), log(1 − x)
    • log[(1 + x)/(1 − x)]

Key takeaways: an infinite series has a sum only when its partial sums converge; a GP converges exactly when |r| < 1; the binomial series for a non-integer index needs |x| < 1, so first write the expression as a number times (1 + small); the exponential series works for every x; and the logarithmic series needs |x| < 1.

Check your understanding

  1. 1. What is the sum of the infinite series 4 + 2 + 1 + 1/2 + ...?

    1. 7
    2. 8
    3. The series has no sum
    4. 16
    Show answer

    Answer: B. This is a GP with a = 4 and r = 1/2. Since |r| < 1, the sum is a/(1 − r) = 4/(1/2) = 8.

  2. 2. For which values of x is the expansion (1 + 2x)⁻¹ = 1 − 2x + 4x² − 8x³ + ... valid?

    1. |x| < 2
    2. |x| < 1
    3. |x| < 1/2
    4. All real x
    Show answer

    Answer: C. (1 + y)⁻¹ = 1 − y + y² − ... needs |y| < 1. With y = 2x this means |2x| < 1, that is |x| < 1/2.

  3. 3. What is the coefficient of x² in the expansion of (1 + x)^(1/2)?

    1. 1/4
    2. −1/8
    3. 1/8
    4. −1/4
    Show answer

    Answer: B. The coefficient is n(n − 1)/2! with n = 1/2: (1/2)(−1/2)/2 = (−1/4)/2 = −1/8.

  4. 4. What is the sum of 1 − 1/1! + 1/2! − 1/3! + 1/4! − ...?

    1. e
    2. e − 1
    3. log 2
    4. 1/e
    Show answer

    Answer: D. This is e⁻ˣ = 1 − x/1! + x²/2! − ... with x = 1, so the sum is e⁻¹ = 1/e.

  5. 5. What is the coefficient of x⁴ in the series for log(1 + x)?

    1. −1/4
    2. 1/4
    3. 1/24
    4. −1/24
    Show answer

    Answer: A. log(1 + x) = x − x²/2 + x³/3 − x⁴/4 + ..., so the coefficient of x⁴ is −1/4. The values ±1/24 would come from 4! in the exponential series, which has no alternating 1/n pattern.

Privacy · Terms · Refunds · Delivery