Skip to content
Tutorials Center
My learning

Topics

Lesson 17 of 37 · Series and Convergence

Alternating Series and Absolute Convergence

Learn the alternating series test with error bounds, absolute versus conditional convergence, and why absolutely convergent series are the well-behaved ones.

So far the terms were positive. When signs vary, cancellation can make a series converge even if the sizes alone would not. This lesson separates two kinds of convergence: absolute, which is robust, and conditional, which is delicate. It also gives the alternating series test, with an error bound that is handy for computing.

Absolute convergence

A series Σ a_k converges absolutely if Σ |a_k| converges.

Theorem. Absolute convergence implies convergence.

Proof. Use the Cauchy criterion. For m > n ≥ N, |a_{n+1} + ... + a_m| ≤ |a_{n+1}| + ... + |a_m|, and the right side is below ε for large n because Σ|a_k| converges (so it is Cauchy). ∎

So you can test a series with sign changes by applying the positive-term tests to |a_k|. For example, Σ sin(k)/k² converges absolutely since |sin(k)/k²| ≤ 1/k².

A series that converges but not absolutely is conditionally convergent.

The alternating series test

Theorem (Leibniz). If b_k ≥ 0, b_k is decreasing, and b_k → 0, then the alternating series Σ (−1)^{k+1} b_k = b_1 − b_2 + b_3 − ... converges.

Proof idea. The even partial sums s_{2n} = (b_1 − b_2) + ... + (b_{2n−1} − b_{2n}) increase (each bracket is ≥ 0). The odd partial sums s_{2n+1} = s_{2n} + b_{2n+1} decrease, because s_{2n+1} − s_{2n−1} = b_{2n+1} − b_{2n} ≤ 0. Also s_{2n} ≤ s_{2n+1} ≤ b_1. So the even sums converge by MCT to some S, and s_{2n+1} = s_{2n} + b_{2n+1} → S + 0 = S. Both subsequences converge to S, so s_n → S. ∎

Error bound. The sum lies between any two consecutive partial sums, so |S − s_n| ≤ b_{n+1}: the error is at most the first omitted term.

Example: the alternating harmonic series 1 − 1/2 + 1/3 − 1/4 + ... converges (its sum is ln 2) although Σ 1/k diverges. So it is conditionally convergent.

Python Alternating harmonic series and its error bound

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

The decreasing condition matters. The series 1 − 1/4 + 1/2 − 1/16 + 1/3 − 1/36 + ... alternates and its terms tend to 0, but they are not decreasing, and it diverges: the positive terms 1/n sum to infinity while the negative terms 1/(2n)² have a finite sum. Always check all three hypotheses.

Conditional versus absolute: the table

| Series | Σ |a_k| | Σ a_k | Type | |---|---|---|---| | Σ (−1)^{k+1}/k² | converges | converges | absolute | | Σ (−1)^{k+1}/k | diverges | converges | conditional | | Σ (−1)^k | diverges | diverges | divergent |

Computing with an error bound

Suppose you want Σ (−1)^{k+1}/k² correct to three decimal places. The error after n terms is at most 1/(n+1)². We need 1/(n+1)² < 0.0005, i.e. n + 1 > 44.7, so n = 44 terms suffice. (The exact sum is π²/12 ≈ 0.822467.)

Python Choose n from the error bound

Press Run (or Ctrl + Enter) to see the result here.

Program input · what you'd type while the program runs

Type the answers before you press Run, one per line. Each time the program reads a line (with input()), it gets the next line from this box.

Why the distinction matters

  • Absolutely convergent series can be regrouped, split and (as the next lesson shows) rearranged freely without changing the sum.
  • Conditionally convergent series cannot: rearranging terms can change the sum, even make it diverge.
  • Products and many manipulations (such as the Cauchy product) need absolute convergence.

A useful comparison: Σ a_k is like a bank account where deposits and withdrawals alternate. If the total size of all transactions is finite (absolute), order does not matter. If the sizes are infinite but cancel in the usual order (conditional), reordering the transactions can leave you with any balance you like.

flowchart TD
  A["Series Σ a_k"] --> B{"Σ |a_k| converges?"}
  B -- "yes" --> C["Absolutely convergent<br/>(safe to rearrange)"]
  B -- "no" --> D{"Σ a_k converges?"}
  D -- "yes" --> E["Conditionally convergent<br/>(delicate)"]
  D -- "no" --> F["Divergent"]

A decision path for a general series.

Check your understanding

Quick check · 4 questions

of 4 answered You scored / 4 · perfect!

  1. 1. What does absolute convergence of Σ a_k mean?

    Absolute convergence is convergence of the series of absolute values.
  2. 2. Which hypotheses does the alternating series test require?

    All three conditions are needed: nonnegative, decreasing and tending to 0.
  3. 3. The alternating harmonic series is

    It converges (to ln 2) but Σ 1/k diverges.
  4. 4. After n terms of a convergent alternating series satisfying the test, the error is at most

    The sum lies between consecutive partial sums, so the first omitted term bounds the error.

Summary

  • Absolute convergence implies convergence (Cauchy criterion).
  • Alternating series test: b_k ≥ 0 decreasing to 0 gives convergence, with error at most the first omitted term.
  • Conditionally convergent means convergent but not absolutely.
  • Absolute convergence is the safe kind; conditional convergence is fragile.

Next: Rearrangements and the Cauchy Condensation Test