AP Calc BC convergence tests: when to use which

You know all the tests. Ratio, root, alternating series, integral, comparison, limit comparison, p-series, geometric, divergence — you can recite the conditions for each one. And then a series shows up on a practice paper and your brain freezes. Which one do I use? Why this one and not that one?

That gap — between knowing the tests and knowing when to reach for them — is what this post is for. There's a decision order that works almost every time, a small set of visual cues that tell you which test fits which shape of series, and three traps that quietly cost students marks on the FRQ. Let's go through them.

Why there are seven tests in the first place

The reason BC has so many convergence tests isn't that the curriculum designers wanted to make the unit harder. It's that no single test handles every possible series. Each test is built for a specific structure — a specific shape that the series takes — and trying to force one test on the wrong kind of series is what makes the unit feel impossible.

Here's the practical list you actually use: the divergence test (a quick disqualifier), geometric series, p-series, the ratio test, the root test, comparison or limit comparison, the integral test, and the alternating series test. That's eight or nine depending on how you count, but in any given exam you only ever reach for around seven.

The whole post hinges on one shift. It's not really about knowing the tests. It's about recognizing the shape of the series in front of you and matching it to the test that handles that shape. Once that shift happens, BC-only content like series gets dramatically easier.

The 30-second decision order

When a series appears on the page, run through this order. Stop at the first test that applies.

1. Does the nth term go to zero? If no, the series diverges. Stop.
2. Is it a geometric series or a p-series? Apply the rule directly — no test needed.
3. Does it alternate? The alternating series test is usually the fastest path.
4. Are there factorials or n-th powers? Reach for the ratio test (or the root test if it's the entire term raised to the n-th).
5. Does it look like a known series with messy extras? Limit comparison.
6. Is it a clean positive series that resembles an integrable function? Integral test.
7. None of the above? Direct comparison, or rethink the algebra and try again.

Here's what the order looks like live. Take the series Σ (−1)ⁿ · (n+1) / (n² + 3n). It's a useful first example because it forces you through the order without resolving on the first or second step — exactly the kind of series that trips students up.

Step 1: Does the nth term go to zero? The magnitude (n+1)/(n²+3n) behaves like 1/n for large n, so yes, it goes to zero. Continue. Step 2: Geometric or p-series? No. Step 3: Does it alternate? Yes — there's a (−1)ⁿ out front. Try the alternating series test. The magnitudes are eventually decreasing and the limit is zero, so the series converges by AST.

But that's not the end. AP rubrics want absolute or conditional convergence specified. Check Σ |aₙ| = Σ (n+1)/(n²+3n). Use limit comparison with Σ 1/n: the limit of (n+1)/(n²+3n) divided by 1/n equals 1, and since Σ 1/n diverges, so does Σ |aₙ|. The original series converges, but only conditionally.

That last step — checking absolute vs conditional after AST gives convergence — is where students lose marks they shouldn't.

The visual cues that tell you which test to use

These are the patterns to train your eye on. Once you see the cue, the test almost picks itself.

Factorials in the term → ratio test. Examples: Σ n!/nⁿ, Σ (2n)!/(n!)² · 4ⁿ. Factorials simplify beautifully when you take aₙ₊₁/aₙ — they cancel down to a clean expression. Most other tests can't handle factorials at all.

The whole term is raised to the n-th power → root test. Examples: Σ (n/(n+1))ⁿ², Σ (3n+1)ⁿ / n²ⁿ. Taking the n-th root collapses the exponent and usually leaves a clean limit you can evaluate.

Alternating sign, (−1)ⁿ or (−1)ⁿ⁺¹ → alternating series test first. Examples: Σ (−1)ⁿ / n, Σ (−1)ⁿ · n / (n²+1), Σ (−1)ⁿ / ln(n). It's the fastest test when it applies. Confirm magnitudes are decreasing and the limit is zero, and you're done — except for the absolute-vs-conditional check.

Polynomial over polynomial → limit comparison with a p-series. Examples: Σ (3n²+5)/(n⁴+2n+1) compares to Σ 1/n². Σ (n+1)/(n³−n) compares to Σ 1/n². Σ √n/(n²+1) compares to Σ 1/n^(3/2). The limit comparison strips the constants and lower-order terms; what's left is a p-series whose convergence you already know.

Clean 1/nᵖ shape → p-series rule, no test needed. Σ 1/n² converges (p=2). Σ 1/√n diverges (p=1/2). Σ 1/n^(1.0001) converges, by a hair. The whole rule: p > 1 converges, p ≤ 1 diverges.

Geometric shape, terms involve rⁿ → geometric series rule. Σ (2/3)ⁿ, Σ 5 · (−1/4)ⁿ, Σ (e/π)ⁿ. If |r| < 1, converges to a/(1−r). If |r| ≥ 1, diverges.

Positive, decreasing, looks integrable → integral test. Σ 1/(n · ln n), Σ n · e⁻ⁿ², Σ 1/(n² + 1). When the antiderivative is clean, the integral test gives a definitive answer where comparison gets fiddly.

The nth term doesn't go to zero → divergence test, and stop. Σ n/(n+1) — limit is 1, diverges. Σ cos(1/n) — limit is 1, diverges. Σ (−1)ⁿ — limit doesn't exist, diverges. This is your free five-second check before any other work.

The three traps students fall into

Trap 1: Treating the divergence test as a "passing" test. The divergence test only ever proves divergence. If the nth term goes to zero, the test tells you nothing — the series might converge or diverge. Plenty of students write "limit is zero, so the series converges" on FRQs. That sentence is wrong, and graders mark it.

Trap 2: Forgetting that the alternating series test has two conditions. Magnitudes must be eventually decreasing AND the limit must be zero. Both. Skipping the decreasing-magnitude check is the more common error, especially under exam pressure when the limit-to-zero part feels obvious.

Trap 3: Stopping at "converges" without checking absolute vs conditional. A test gives convergence; the student writes "converges" and moves on. But for an alternating series, the rubric usually wants you to specify absolutely or conditionally. That's a separate check on Σ |aₙ|. It's almost always worth one or two marks, and it's almost always the difference between a 4 and a 5 on a series FRQ.

There's a fourth trap worth naming, even though it's procedural rather than conceptual: skipping the endpoint tests on intervals of convergence. The ratio test gives you the open interval. The endpoints need to be tested separately, one at a time, and they're often graded as separate marks. The next worked example shows why that matters.

Worked example — AP Calc BC 2018, FRQ 6 (b)

The 2018 BC exam released question 6 publicly, and part (b) is one of the cleanest demonstrations of the decision order on a real FRQ. It asks for the interval of convergence of the Maclaurin series for f(x) = x · ln(1 + x/3).

Start by writing the series. The Maclaurin series for ln(1 + x) is x − x²/2 + x³/3 − x⁴/4 + … with general term (−1)ⁿ⁺¹ · xⁿ / n. Substitute x/3 in for x and multiply the whole thing by x. The general term for f's Maclaurin series becomes:

aₙ = (−1)ⁿ⁺¹ · xⁿ⁺¹ / (n · 3ⁿ), for n ≥ 1.

Now find the radius of convergence. The cue here is the n in the exponent and the n in the denominator — that's a ratio-test setup.

|aₙ₊₁ / aₙ| = |xⁿ⁺² / ((n+1) · 3ⁿ⁺¹)| · |(n · 3ⁿ) / xⁿ⁺¹| = (n / (n+1)) · (|x| / 3)

As n → ∞, this limit is |x|/3. The ratio test gives convergence when |x|/3 < 1, so the open interval is (−3, 3). The radius of convergence is 3.

But that's only the open interval. The endpoints x = 3 and x = −3 need separate tests — and this is where students leave marks behind.

At x = 3: the terms become (−1)ⁿ⁺¹ · 3ⁿ⁺¹ / (n · 3ⁿ), which simplifies to 3 · (−1)ⁿ⁺¹ / n. That's three times the alternating harmonic series. Apply the alternating series test: magnitudes 3/n are decreasing, limit is zero. Converges. (Note: only conditionally, since Σ 3/n diverges. The interval-of-convergence question doesn't ask for absolute vs conditional, but the underlying check is what tells you the endpoint is included.)

At x = −3: the terms become (−1)ⁿ⁺¹ · (−3)ⁿ⁺¹ / (n · 3ⁿ). The two (−1) factors combine: (−1)ⁿ⁺¹ · (−1)ⁿ⁺¹ = 1. Simplifies to 3/n. That's three times the harmonic series. Diverges by the p-series rule (p=1).

Interval of convergence: (−3, 3].

Notice what happened across one problem. The ratio test gave the open interval. The alternating series test handled one endpoint. The p-series rule handled the other. Three different tests, picked by reading the structure of the series at each step. That's the decision order doing its job.

Part (c) of this same FRQ uses the alternating series error bound — applying the alternating series test in a different way, to bound the error of a Taylor polynomial approximation. It's worth working through after this one, because it shows the same test serving a different purpose. For the broader pattern of how these procedural moves get rewarded on the FRQ rubric, more on AP Calc FRQ strategy is worth reading separately.

How to practice this so it sticks

Don't drill the tests one at a time. That's how you end up knowing each test perfectly and still freezing when you have to pick one. The skill the exam tests is recognition, not recall.

What works better: build mixed sets. Take ten series from past BC papers, scramble them, no labels, and force yourself to decide which test in 30 seconds before working any of them. Then check yourself. The first few rounds will feel uncomfortable. By the fourth or fifth, the cues start coming automatically.

The College Board has released FRQs going back many years on their site, and series questions appear regularly. Working through five or six of them, with the decision order in mind, is worth more than another twenty mechanical practice problems.

One last note on the formula sheet. Most of the convergence-test conditions are not on the AP-provided sheet. The conditions for AST, the ratio test, the integral test — these have to live in your head. That's worth a deliberate memorization pass in the week before the exam, separate from problem-solving practice. (You can grab the AP Calculus formula sheet to see exactly what's provided and, just as importantly, what isn't.)

Frequently asked questions

How many convergence tests do I really need to know for AP Calc BC?

Realistically seven: divergence, geometric, p-series, ratio, root, alternating series, and one of comparison or limit comparison. The integral test is useful but rarely the only path. Knowing all of them isn't the bottleneck — knowing when to reach for which is.

Is the ratio test always the safest first try?

No. It's the right tool when factorials or n-th powers are involved, but it's wasted effort on simple polynomial-over-polynomial series, and it gives an inconclusive answer (limit equals 1) often enough that defaulting to it costs you time. Match the test to the shape of the series instead.

What's the difference between the ratio test and the root test?

The ratio test uses the limit of aₙ₊₁/aₙ; the root test uses the limit of the n-th root of aₙ. Ratio is better when terms involve factorials or products that simplify when you divide. Root is better when the entire term is raised to the n-th power.

Do I need to memorize the conditions for the alternating series test?

Yes — both of them. Terms must be decreasing in magnitude (eventually), and the limit must be zero. Forgetting the second condition is one of the most common AP exam errors and costs at least one mark on FRQ rubrics.

How much of the BC exam is series questions?

Series and Taylor polynomials together make up a meaningful share of the BC exam, and at least one FRQ almost always involves series. It's high-yield to drill — but more importantly, it's the unit students lose easy marks on through procedural errors, not conceptual ones.

The convergence tests aren't the hard part of BC. The decision of which test to use is. Train pattern recognition on series shapes, run the decision order under time pressure, and remember that absolute-vs-conditional check at the end. Those three habits are most of what separates a 4 from a 5 on the series FRQ.

If you'd like to drill the decision order with someone watching your reasoning out loud — that's exactly the kind of work AP Calculus BC tutoring is for. The BC course at Tangible Learning runs 65 hours, with 15 of those dedicated to the series unit specifically — enough time to build the recognition habit properly. Book a free trial lesson to start.