CalcRef
How to Choose the Right Convergence Test
Ask students which part of second-semester calculus feels most like guesswork and infinite series wins comfortably. It's rarely the tests themselves — each one is a short statement with clear conditions. The struggle is selection: a problem hands you a series, five or six tests sit in your toolkit, and nothing announces which one applies. This guide turns that guess into a procedure: read the form of the general term, and let the form pick the test.
Step zero: the divergence check
Before any named test, look at the limit of the general term. If aₙ does not approach 0, the series diverges — done. But mind the one-way arrow: terms going to 0 proves nothing about convergence. The harmonic series Σ1/n has terms shrinking to zero and still diverges. Treating "terms → 0" as a proof of convergence is the single most common series error on exams.
Know your reference series
Two families anchor everything, because comparison-style tests need something to compare against:
- Geometric series Σarⁿ: converges exactly when |r| < 1, with sum a/(1−r). The only elementary series whose exact sum you're routinely expected to produce.
- p-series Σ1/nᵖ: converges exactly when p > 1. The harmonic series is the boundary case p = 1, and it diverges.
Reading the term: form → test
| General term looks like… | Reach for | Why / conditions |
|---|---|---|
| Factorials, or n in an exponent (n!, 2ⁿ, nⁿ) | Ratio test | Successive-term ratios collapse factorials and exponentials. Limit < 1 converges (absolutely), > 1 diverges, = 1 says nothing. |
| Entire term raised to the n-th power | Root test | The n-th root peels the exponent off cleanly. Same limit trichotomy as the ratio test, including the inconclusive = 1 case. |
| Rational function of n, or n times a log | Comparison / limit comparison with a p-series | Keep dominant terms: (n+3)/(n³+1) behaves like 1/n². Terms must be positive; limit comparison needs a finite positive limit. |
| Term is a decreasing, positive, continuous function you can integrate (1/(n ln n)) | Integral test | Series and improper integral share a fate. All three conditions — positive, continuous, decreasing — are required. |
| Alternating signs: (−1)ⁿbₙ | Alternating series test | Converges if bₙ decreases to 0. Bonus: the error after n terms is at most the first omitted term. |
The inconclusive cases are part of the skill
Both the ratio and root tests go silent when their limit equals 1 — and that happens for exactly the series students most want to throw them at: every p-series gives ratio limit 1. Applying the ratio test to Σ1/n² isn't wrong, just useless; the p-series criterion answers instantly. Knowing where each test is blind is as valuable as knowing where it works, and it's why "just always use the ratio test" fails as a strategy.
Absolute vs. conditional convergence
For series with mixed signs, ask the stronger question first: does Σ|aₙ| converge? If yes, the series converges absolutely and everything downstream is safer. If not, the alternating series test may still give convergence — but only the conditional kind. The alternating harmonic series Σ(−1)ⁿ⁺¹/n is the canonical example: convergent by the alternating series test, but not absolutely, since the absolute values form the divergent harmonic series. Exams love this distinction precisely because it can't be answered by pattern-matching a single test.
Why this all matters: power series
Convergence tests aren't an isolated topic — they are the machinery behind Taylor and Maclaurin series. The interval of convergence of a power series comes from running the ratio test on its terms, then checking both endpoints with the tools above (p-series facts, alternating series test). If you know the common Maclaurin expansions — eˣ, sin x, cos x, 1/(1−x) — you should also know each one's interval of convergence, because endpoint behavior is where the free-response points hide. For where this sits in exam prep overall, see the AP calculus formula review.
Drilling the selection skill
Because the hard part is choosing, practice should target the choice. Collect series prompts, and before solving anything, name the test and the reason ("factorial → ratio test"; "looks like 1/n³ → comparison"). Then verify the conditions out loud — positivity, monotonicity, the limit that must exist. This is classic retrieval practice applied to decisions rather than statements; the memorization guide explains why naming-before-solving builds exactly the reflex mixed exams demand. And keep the base facts crisp: comparison arguments die quietly when the p-series criterion or a basic derivative (for the integral test) is fuzzy.
How CalcRef helps
CalcRef ships a dedicated Convergence Tests reference table covering the ratio, root, integral, comparison, and alternating series tests, and a Maclaurin Series table listing the common expansions with their convergence intervals — the two references this guide leans on, one tap apart. The Sequences and Series topic (one of eight in the app) presents each formula with LaTeX-rendered notation, a plain-language description, conditions for use, and variable definitions, so the hypotheses that make each test valid are attached to the statement itself. Flashcard quiz rounds of 10 cards let you drill Sequences and Series on its own or mixed with all topics, with per-topic stats and history to show when the selection reflex is actually solid. Free and fully offline on iPhone, iPad, and Android.
Quick answers
Ratio test vs. root test — which should I learn first?
The ratio test: it handles factorials, which the root test does not do gracefully, and it's the standard tool for power series intervals. Add the root test for terms that are entirely raised to the n-th power.
Does the alternating series test prove absolute convergence?
No — it can only establish convergence of the alternating series itself. To claim absolute convergence you must separately show Σ|aₙ| converges, usually by comparison or the p-series criterion.