The integral test turns one family of series into a one-line decision: the p-series. Together with geometric series, p-series are the benchmarks you will compare almost everything else to, so it pays to know them cold.
The p-series test
A p-series has the form
n=1∑∞np1=1+2p1+3p1+4p1+⋯
where p is a constant. The harmonic series is the p-series with p=1.
To decide convergence, apply the integral test to f(x)=xp1. For p>0 this function is positive, continuous and decreasing on [1,∞). You already know from improper integrals that
∫1∞xpdx=⎩⎨⎧p−11∞if p>1,if p≤1.
(If p≤0, the terms don't even go to 0, so the nth term test settles those cases.)
The p-series test
n=1∑∞np1 converges if p>1 and diverges if p≤1.
That's it: find the exponent, compare with 1. Some examples:
Series
p
Result
∑n21
2
converges
∑n1
21
diverges
∑n1
1
diverges (harmonic)
∑n1.011
1.01
converges
∑nn1
23
converges
Why p=1 is the dividing line
Compare the partial sums of ∑n1 and ∑n21. Both have terms that go to zero. But the harmonic terms shrink too slowly: the partial sums keep creeping upward forever. The terms of n21 shrink fast enough that the partial sums level off near 1.645.
Upper dots: partial sums of Σ 1/n keep climbing. Lower dots: partial sums of Σ 1/n² level off below π²/6 ≈ 1.645 (dashed).Open in grapher →
Even p=1.001 converges, while p=1 does not. The borderline is sharp.
For p>1 the integral also gives a quick sense of size. The rectangle picture from the integral test shows that the sum is between p−11 and 1+p−11. For p=2 that means the sum is between 1 and 2, consistent with 6π2≈1.645. But notice what happens when p is just barely above 1: for p=1.01 the sum is more than 100. Convergent, yes, but only just.
Rewriting into p-series form
Often a series is a p-series in disguise. Simplify the terms first:
∑n2n=∑n3/21, so p=23: converges.
∑3n25=5∑n2/31, so p=32: diverges. A constant multiple never changes convergence.
∑n−e has p=e≈2.718: converges.
Common mistake
Don't confuse a p-series with a geometric series. In np1 the variable is in the base and the exponent is fixed. In (21)n the variable is in the exponent. So ∑n21 is a p-series (p=2), while ∑2n1 is geometric (r=21). Both converge, but for different reasons, and only the geometric one has an easy exact sum.
Worked examples
Worked example: Classifying p-series
Determine whether each series converges or diverges.
(c) n4n=n5/4, so the series is 4∑n5/41 with p=45>1: converges.
Worked example: Using a known sum
It is a famous fact that n=1∑∞n21=6π2. Use it to find n=1∑∞(2n)21, the sum over even squares only.
Solution. Factor out the constant:
n=1∑∞4n21=41n=1∑∞n21=41⋅6π2=24π2.
Worked example: A parameter in the exponent
For which values of k does n=1∑∞n3−k1 converge?
Solution. This is a p-series with p=3−k. It converges when 3−k>1, that is, when k<2.
Tip
When you see a sum of a rational or root expression in n, ask "what power does this behave like for large n?" If the answer is np1, you already have a strong guess about convergence. The next lesson shows how to turn that guess into a proof.
Practice
Practice 1
Which of the following series converges?
Practice 2
Does n=1∑∞n3nn2 converge or diverge?
Practice 3
For which values of p does n=1∑∞n2p−11 converge? Write an inequality in p.
Enter an inequality, e.g. x >= 4, -2 < x <= 3, or x < 1 or x > 5
Practice 4
Given n=1∑∞n21=6π2, find n=2∑∞n21. Give an exact answer.
Enter a number. Fractions like 3/4 and sqrt(2) are OK.
Practice 5
Given n=1∑∞n21=6π2, find n=1∑∞(2n)21. Give an exact answer.
Enter a number. Fractions like 3/4 and sqrt(2) are OK.
Practice 6
Using the previous two results, find the sum of the reciprocals of the odd squares, n=1∑∞(2n−1)21=1+91+251+⋯. Give an exact answer.
Enter a number. Fractions like 3/4 and sqrt(2) are OK.
Practice 7
Consider the series I. ∑nπ/31, II. ∑5n41, III. ∑nn. Which of them converge?