EN TH
Worked Example

01006030 | Week 15 | Lecture Notes

Series Convergence Tests

The nth-term test (Week 14) can only prove divergence. This week builds the five tests that can actually prove convergence — each one reusing a tool from earlier in the course: improper integrals, limits, absolute values, and one-sided monotonic decrease.

There is no single universal convergence test; instead there is a toolkit, and part of the skill this week teaches is recognising which tool fits which series. Every test below answers the same question — does ∑a_n converge? — using a different piece of earlier machinery.

15.1 The Integral Test

If f(x) is continuous, positive, and decreasing for x ≥ 1, and an = f(n), then ∑n=1∞ an and ∫1∞ f(x) dx either both converge or both diverge. The area under the continuous curve and the sum of rectangle areas built from it (exactly Week 7's Riemann-sum picture) trap each other within a bounded difference, so one is finite exactly when the other is.

15.2 The Harmonic Series and p-Series

The harmonic series ∑ 1/n has terms → 0 (so the nth-term test is silent), yet it diverges — the Integral Test confirms this immediately, since ∫1∞ (1/x)dx diverges (Week 13, §13.10). More generally, the p-series:

∑n=1∞ 1np  converges if p > 1, diverges if p ≤ 1 (15.1)

— identical to the improper integral p-test of Eq. 13.3, by the Integral Test itself. The p-series family is the standard benchmark the comparison tests below measure other series against.

15.3 The Direct Comparison Test

If 0 ≤ an ≤ bn for all n beyond some point, then:

Example 15.1 — Direct comparison

Determine whether ∑n=1∞ 1/(n2+3) converges.

Step 1 — Find a larger, known series

1/(n2+3) < 1/n2 for every n, and ∑1/n2 is a p-series with p=2>1, so it converges (Eq. 15.1).

Step 2 — Apply the comparison

Since the given series is smaller termwise than a convergent series, it converges too.

15.4 The Limit Comparison Test

If an, bn > 0 and limn→∞ an/bn = c for some finite c > 0, then ∑an and ∑bn either both converge or both diverge. This avoids the sometimes-awkward inequality of §15.3, at the cost of computing a limit instead (Week 3 §3.9's degree-comparison shortcut is often the fastest route).

Example 15.2 — Limit comparison

Determine whether ∑n=1∞ (2n+1)/(n3+n) converges.

Step 1 — Choose a comparison series based on dominant terms

For large n, the terms behave like 2n/n3 = 2/n2, so compare with ∑1/n2 (a convergent p-series).

Step 2 — Compute the limit ratio

limn→∞ (2n+1)/(n3+n)1/n2 = limn→∞ 2n3+n2n3+n = 2 (Week 3, §3.9)

Step 3 — Apply the test

The limit is a finite positive number (2), and ∑1/n2 converges, so the given series converges too.

15.5 Absolute Convergence

A series ∑an converges absolutely if ∑|an| converges. Absolute convergence implies convergence (the series itself must converge too), but not conversely — a series that converges only because of cancellation between positive and negative terms is called conditionally convergent (§15.9).

15.6 The Ratio Test

Let ρ = limn→∞ |an+1/an|. Then:

The Ratio Test is the standard choice whenever an involves factorials or nth powers, since consecutive-term ratios cause massive cancellation there.

Example 15.3 — The Ratio Test with a factorial

Determine whether ∑n=1∞ n!/nn converges.

Step 1 — Set up the ratio

an+1an = (n+1)!/(n+1)n+1n!/nn = (n+1)·nn(n+1)n+1 = nn(n+1)n = 1(1+1/n)n

Step 2 — Take the limit (recognise Example 12.3's limit, generalised)

limn→∞ (1+1/n)n = e (a standard limit, provable the way Week 12 §12.4 handled indeterminate powers), so ρ = 1/e < 1.

Step 3 — Conclude

Since ρ = 1/e < 1, the series converges (absolutely).

15.7 The Root Test

Let ρ = limn→∞ ⁿ√|an|. The same three cases as the Ratio Test apply. The Root Test is the natural choice when an already involves an nth power throughout.

Example 15.4 — The Root Test

Determine whether ∑n=1∞ (2n/(3n+1))n converges.

Step 1 — Take the nth root, which cancels the outer power directly

ⁿ√|an| = 2n/(3n+1)

Step 2 — Take the limit (Week 3, §3.9)

limn→∞ 2n/(3n+1) = 2/3

Step 3 — Conclude

ρ=2/3<1, so the series converges.

15.8 Alternating Series and the Alternating Series Test

An alternating series has terms that strictly alternate sign, ∑(−1)nbn with bn > 0. The Alternating Series Test: if bn is eventually decreasing and bn → 0, the series converges.

Example 15.5 — The Alternating Series Test

Show that the alternating harmonic series ∑n=1∞ (−1)n+1/n converges.

Step 1 — Check b_n is decreasing

bn = 1/n is decreasing for every n ≥ 1.

Step 2 — Check b_n → 0

limn→∞ 1/n = 0.

Step 3 — Conclude

Both conditions hold, so the alternating harmonic series converges (to ln 2, though the test does not reveal the sum's value).

15.9 Conditional Convergence

The alternating harmonic series converges (§15.8), yet the series of its absolute values, ∑1/n, is the harmonic series, which diverges (§15.2). It is therefore conditionally convergent: convergent as written, but not absolutely convergent. This is why §15.5's Absolute Convergence and §15.8's Alternating Series Test must be checked separately and in that order — a series can pass one and fail the other.

15.10 Worked Capstone Example

Example 15.6 — Choosing the right test

Classify ∑n=1∞ (−1)nn/2n as absolutely convergent, conditionally convergent, or divergent.

Step 1 — Test for absolute convergence first (§15.5) using the Ratio Test (§15.6)

|an| = n/2n. |an+1||an| = (n+1)/2n+1n/2n = n+12n

Step 2 — Take the limit

ρ = limn→∞ (n+1)/(2n) = 1/2 < 1.

Step 3 — Conclude directly, without needing the Alternating Series Test at all

By the Ratio Test, ∑|an| converges, so the original series converges absolutely. Since absolute convergence already implies convergence, there is no need to separately verify the alternating series conditions of §15.8.

15.11 Nomenclature

SymbolMeaning
p-series∑1/n^p; converges iff p>1
ρThe limiting ratio or root used in the Ratio/Root Tests
absolutely convergent∑|a_n| converges
conditionally convergent∑a_n converges but ∑|a_n| diverges
alternating seriesA series whose terms strictly alternate in sign

15.12 Problems

  1. Integral Test. Use the Integral Test to determine whether ∑ ne−n2 converges.
  2. p-series. State, without further work, whether ∑1/n1.5 and ∑1/√n converge.
  3. Direct comparison. Determine whether ∑1/(n3+1) converges.
  4. Limit comparison. Determine whether ∑(n+1)/(n2√n) converges.
  5. Ratio Test. Determine whether ∑3n/n! converges.
  6. Root Test. Determine whether ∑(n/(2n+1))n converges.
  7. Alternating Series Test. Determine whether ∑(−1)n/√n converges.
  8. Absolute vs. conditional. Classify ∑(−1)n/n2 as absolutely convergent, conditionally convergent, or divergent.
  9. Choosing a test. Determine whether ∑(2n−1)/(n3+2n) converges, explaining why you chose the test you used.
  10. Combined. Classify ∑(−1)nn2/n! as absolutely convergent, conditionally convergent, or divergent.

15.13 References