Which Of The Following Series Is Conditionally Convergent

You're staring at a series on an exam or a problem set. You're pretty sure of that. But then the question asks: "Is it absolutely convergent, conditionally convergent, or divergent?It converges. " And suddenly you're not so sure anymore.

Been there. It's one of those concepts that sounds straightforward in lecture — "a series converges conditionally if it converges but the series of absolute values diverges" — but gets messy fast when you're actually looking at a specific sum.

Let's clear it up once and for all.

What Conditional Convergence Actually Means

Start with the basics. You have an infinite series:

∑ aₙ (from n=1 to ∞)

Three things can happen:

It diverges — the partial sums don't settle down to a finite limit.
It converges absolutely — the series of absolute values ∑ |aₙ| converges. This is the "strong" kind of convergence. If a series converges absolutely, it converges no matter how you rearrange the terms. It's stable.
It converges conditionally — the original series ∑ aₙ converges, but ∑ |aₙ| diverges.

That's the definition. But here's why it matters: conditional convergence is weird*. Practically speaking, riemann proved that if a series converges conditionally, you can rearrange its terms to make it sum to any real number you want*. So or diverge to infinity. Or oscillate. The sum literally depends on the order of the terms.

Absolutely convergent series don't do that. They're well-behaved. Conditionally convergent series are the rebels.

The Classic Example: Alternating Harmonic Series

∑ (-1)^(n+1) / n = 1 - 1/2 + 1/3 - 1/4 + 1/5 - ...

This converges to ln(2). You can prove it with the Alternating Series Test — terms decrease in magnitude and approach zero.

But the absolute value series? That's the harmonic series:

∑ 1/n = 1 + 1/2 + 1/3 + 1/4 + ...

Which famously diverges. So the alternating harmonic series is the textbook example of conditional convergence.

Why It Matters / Why People Care

You might wonder: if the series converges anyway, why does the distinction matter?

Rearrangement is the big one. In applied math, physics, and numerical computing, you sometimes reorder terms for efficiency or parallelization. If you're working with a conditionally convergent series and you reorder terms without thinking, you can change the answer. That's a bug waiting to happen.

Integration and differentiation of series. Term-by-term integration or differentiation of a power series is safe inside the radius of convergence — but at the boundary, conditional convergence shows up and you have to be careful.

Fourier series. Many Fourier series converge conditionally at points of discontinuity (Gibbs phenomenon territory). Understanding the difference between absolute and conditional convergence helps explain why Fourier series behave the way they do near jumps.

It's also a favorite exam trap. Professors love giving a series that looks* like it might converge absolutely, but doesn't. Or one that converges conditionally and asking you to prove it. Knowing the tests cold saves you points.

How to Test for Conditional Convergence

You don't guess. You follow a decision tree.

Step 1: Check the Original Series

Does ∑ aₙ converge? Use whatever test fits:

Alternating Series Test (Leibniz): if aₙ = (-1)ⁿ bₙ or (-1)^(n+1) bₙ with bₙ > 0, decreasing, and bₙ → 0, it converges.
Ratio Test: lim |aₙ₊₁/aₙ| < 1 → converges absolutely (stronger than needed here, but useful).
Root Test: lim sup |aₙ|^(1/n) < 1 → converges absolutely.
Comparison / Limit Comparison: compare to a known convergent series.
Integral Test: if aₙ = f(n) for positive, decreasing f, check ∫ f(x) dx.
p-series test: ∑ 1/n^p converges iff p > 1.

If the original series diverges, you're done. It's not conditionally convergent — it's just divergent.

Step 2: Check the Absolute Value Series

Now test ∑ |aₙ|.

If it converges → the original series converges absolutely.
If it diverges → the original series converges conditionally (assuming step 1 said it converges).

That's the whole algorithm. The trick is picking the right test for each step.

Example Walkthrough

Series: ∑ (-1)^n / √n

Step 1: Original series. Day to day, this is alternating with bₙ = 1/√n. Yes. Terms decrease? Yes. Limit zero? Alternating Series Test says converges.

Step 2: Absolute series. Which means ∑ 1/√n = ∑ n^(-1/2). Practically speaking, this is a p-series with p = 1/2 < 1. Diverges.

Conclusion: Conditionally convergent.

Another One

Series: ∑ (-1)^n / n²

Step 1: Alternating, bₙ = 1/n². Decreasing, limit zero. Converges.

Step 2: Absolute series: ∑ 1/n². Here's the thing — p-series with p = 2 > 1. Converges.

Conclusion: Absolutely convergent.

If you found this helpful, you might also enjoy what is the basic function of hydrostatic pressure or a state function is best described as.

A Tricky One

Series: ∑ (-1)^n (n/(n+1))

Step 1: Check the nth term. aₙ = (-1)^n (n/(n+1)). The absolute value |aₙ| = n/(n+1) → 1 ≠ 0.

Nth term test for divergence: if lim aₙ ≠ 0, the series diverges. Done. Not conditionally convergent. Not anything convergent.

This is the most common trap — forgetting to check the nth term first.

Common Mistakes / What Most People Get Wrong

Mistake 1: Confusing "converges" with "converges absolutely"

Students see a series converge and assume it's absolute. They skip step 2. Always check the absolute series separately.

Mistake 2: Using the Ratio Test on the alternating series directly

The Ratio Test checks absolute convergence. Inconclusive. Consider this: that doesn't mean the series diverges — it means the Ratio Test can't decide. If you apply it to ∑ (-1)^n/n, you get lim |aₙ₊₁/aₙ| = lim (n/(n+1)) = 1. You need the Alternating Series Test for the original, and p-series for the absolute. And that's really what it comes down to.

Mistake 3: Thinking conditional convergence means "barely converges"

It doesn't mean the sum is small or the convergence is slow. That's not "barely" — it's a perfectly normal number. 693. It means the convergence depends on cancellation* between positive and negative terms. The alternating harmonic series converges to ln(2) ≈ 0.But strip the signs, and the sum explodes.

The Rearrangement Theorem: Why Conditional Convergence Is Weird

There’s a deeper reason absolute vs. conditional convergence matters. It’s not just a classification exercise — it changes what you’re allowed to do with the series.

Riemann Rearrangement Theorem: If a series converges conditionally, you can rearrange its terms to sum to any real number* — or to diverge to ±∞, or to oscillate forever. That's the part that actually makes a difference.

Let that sink in. The alternating harmonic series:

$ 1 - \frac{1}{2} + \frac{1}{3} - \frac{1}{4} + \frac{1}{5} - \frac{1}{6} + \cdots = \ln 2 $

But if you reorder it as:

$ 1 + \frac{1}{3} - \frac{1}{2} + \frac{1}{5} + \frac{1}{7} - \frac{1}{4} + \cdots $

(two positives, one negative, repeat), the sum becomes $\frac{3}{2}\ln 2$. Different order, different sum.

Absolutely convergent series are immune to this. Rearrange them however you want — the sum stays exactly the same. This is why absolute convergence is the "safe" kind: it behaves like a finite sum. Conditional convergence is fragile; its value depends on the order* of terms, not just the set of terms.

This isn't a pathology. Fourier series, power series on the boundary of convergence, and many physical series are conditionally convergent. You must* respect the given ordering.

Quick Reference: Which Test When?

Series Form	First Try	Backup
$\sum \frac{1}{n^p}$ or similar	p-series	Limit Comparison with $1/n^p$
$\sum \frac{1}{a^n}$ or factorials	Ratio Test	Root Test
$\sum (a_n)^n$	Root Test	Ratio Test
Alternating $\sum (-1)^n b_n$	Alternating Series Test	Check $b_n \downarrow 0$
Rational function of $n$	Limit Comparison with leading term	Direct Comparison
$\sum f(n)$ with nice integral	Integral Test	Comparison
$\sum \frac{1}{n(\ln n)^p}$	Integral Test	Cauchy Condensation

Pro tip: The Ratio and Root Tests only* test absolute convergence. If they give $L=1$, they say nothing about conditional convergence. You must* switch tools.

The "Nuclear Option": Cauchy Condensation Test

For series of the form $\sum a_n$ where $a_n$ is positive and decreasing, there’s a heavy hammer:

$ \sum_{n=1}^\infty a_n \text{ converges } \iff \sum_{k=0}^\infty 2^k a_{2^k} \text{ converges} $

This turns $\sum \frac{1}{n (\ln n)^p}$ into $\sum \frac{2^k}{2^k (k \ln 2)^p} = \frac{1}{(\ln 2)^p} \sum \frac{1}{k^p}$, a p-series. It’s the cleanest way to handle log-heavy denominators.

Summary Checklist

Before you write "converges" on an exam or paper, run this mental script:

Nth term test: $\lim a_n \neq 0$? → Diverges. Stop.
Alternating? Try Alternating Series Test. Converges?
- Yes → Go to step 3.
- No → Test for absolute convergence directly (Ratio, Root, Comparison, Integral). If absolute converges → Absolutely convergent. If absolute diverges → Diverges.
Test $\sum |a_n|$:
- Converges → Absolutely convergent.
- Diverges → Conditionally convergent.

That’s it. No magic. Just discipline.

Conclusion

The distinction between absolute and conditional convergence isn't academic gatekeeping — it's the line between series you can manipulate freely and series that bite back when rearranged. Even so, absolute convergence gives you the full algebra of finite sums: reorder, regroup, multiply term-by-term (Cauchy product), integrate or differentiate term-by-term (within radius of convergence). Conditional convergence gives you only* the given order.

Master the two-step algorithm. Respect the nth term test. And never trust a Ratio Test result of $L=1$. The rest is just practice.