Alternating Series and Error Control

Distinguish absolute from conditional convergence and bound a valid alternating remainder.

Builds on Ratio and Root Tests

The bigger question: Can infinitely many contributions have a finite total?

On this page

Cancellation can produce convergence

If bn≥0b_n\ge0 decreases eventually to zero, the alternating series ∑(−1)n−1bn\sum(-1)^{n-1}b_n converges. Pairing consecutive terms shows that even and odd partial sums approach a common limit from opposite sides. The eventual decrease and zero limit are both needed for this test.

Absolute convergence means ∑∣an∣\sum|a_n| converges and guarantees convergence of ∑an\sum a_n. Conditional convergence means the signed series converges but its absolute-value series diverges. This distinction affects operations such as rearranging terms: arbitrary rearrangements preserve absolutely convergent sums, but not generally conditionally convergent sums.

Visual guide

VISUAL GUIDEPartial sums bracket the limit
Alternating harmonic partial sums oscillate around ln 2. Even sums lie below and odd sums above. For decreasing term magnitudes tending to zero, the next omitted term bounds the error.00.430.57560.7590.925121.1terms Npartial sum
  • Limit ln 2
Alternating harmonic partial sums oscillate around ln 2. Even sums lie below and odd sums above. For decreasing term magnitudes tending to zero, the next omitted term bounds the error.

Worked example: alternating harmonic series

The magnitudes 1/n1/n decrease to zero, so 1−1/2+1/3−⋯1-1/2+1/3-\cdots converges. Its absolute-value series is the divergent harmonic series. Therefore convergence is conditional.

The alternating remainder bound is ∣S−SN∣≤bN+1|S-S_N|\le b_{N+1} when the alternating-test hypotheses apply from the relevant tail onward. To guarantee error at most 0.010.01, choosing N=100N=100 suffices because the next magnitude is 1/101<0.011/101<0.01. This bound is conservative and does not require knowing the sum.

PAUSE & THINKA quick check, not a grade

Try it yourself.

Choose an answer and explain your reasoning to yourself. Use a hint if you get stuck.

For an alternating series with decreasing positive terms tending to 0, how large is the error after N terms?

Hint 1 · Find a starting point

Use the alternating-series remainder estimate.

Hint 2 · Take the next step

The omitted terms partly cancel each other.

Show the reasoning

Answer: At most the next term’s magnitude

The error magnitude is bounded above by aₙ₊₁; equality is not generally true.

Before moving on: what would make one of the other answers wrong? Saying why is part of understanding.

Worked example: absolute convergence first

For ∑(−1)n−1/n2\sum(-1)^{n-1}/n^2, taking absolute values produces a convergent pp-series. Thus the original series converges absolutely. The alternating bound also applies and is sharper here than the positive-series integral bound: after NN terms, its error is at most 1/(N+1)21/(N+1)^2.

A sequence of small terms with changing signs is not automatically an alternating series. The theorem applies to a strict alternating tail with decreasing magnitudes. Check the pattern and hypotheses before quoting the next-term error bound.

Practice

  1. Classify ∑(−1)n/n\sum(-1)^n/\sqrt n.
  2. Bound the error after 1010 terms of ∑n=1∞(−1)n−1/n2\sum_{n=1}^\infty(-1)^{n-1}/n^2.
  3. Does ∑(−1)nn/(n+1)\sum(-1)^n n/(n+1) converge?
Show worked solutions
  1. Conditionally: the alternating test works, but the absolute series has p=1/2p=1/2 and diverges.
  2. At most 1/1211/121. The next term is positive, so the sum is above the even partial sum.
  3. No. Magnitudes approach 11, so the terms do not approach zero.
MAKE IT YOURS

Pause before the next idea.

Can you explain how the visualization connects to this lesson’s goal? If a step still feels uncertain, put this lesson on your review list and try the check again another day.

Optional marks, not a grade. Saved in this browser only. Open notebook →