Absolute and conditional convergence sits at the centre of the AP Calculus BC series unit, and it is one of the few sub-topics where a candidate can write a numerically correct limit comparison and still lose a full rubric row. The College Board frames every convergence question on the BC exam around a two-stage decision: first, does the series of absolute values converge, and second, if it does not, does the original signed or alternating series still converge. The answer to those two questions is what determines whether a series is called absolutely convergent, conditionally convergent, or divergent. On a Free Response Question that judgement is worth one or two rubric lines, and the lines are scored independently, so an incorrect verdict on absolute convergence does not automatically zero out the conditional-convergence row.
The distinction matters because the AP exam tests it directly. Both the multiple-choice and the FRQ sections include items where the series has the form Σ aₙ with aₙ alternating in sign, oscillating with n, or carrying a factor of (−1)ⁿ. A candidate who treats the series as if all its terms were positive loses the alternating-series row, and a candidate who forgets to take the absolute value of the general term loses the absolute-convergence row. The remainder of this article walks through the two-row decision, the four tests that drive it, and the specific scoring behaviours the AP reader applies when a candidate writes each row.
The two-row decision the AP rubric actually scores
Every convergence FRQ on AP Calculus BC is, in effect, a two-line argument. The first line answers: does Σ |aₙ| converge? The second line answers: does Σ aₙ converge, given whatever the first line told you? The reader awards points for each line separately, and the verdicts are not coupled. A student who shows that Σ |aₙ| diverges by the limit comparison test with a p-series, and then shows that the original alternating series converges by the alternating-series test, earns both rows. A student who confuses the two and runs the alternating-series test on |aₙ| instead of aₙ earns neither.
In practice, the rubric is written so that the absolute-convergence row tests a single skill: take the absolute value of the general term, then apply a standard convergence test to a positive series. The conditional-convergence row tests a different skill: recognise that the original series is signed, and apply the alternating-series test, the Abel test, or Dirichlet's test to a series whose terms are not all positive. The two rows therefore sit on different sides of a single sign, and that is the source of most of the marks lost on this topic.
For most candidates reading this, the highest-leverage habit is to write the absolute-value step explicitly. The phrase 'consider Σ |aₙ|' is cheap, takes three seconds, and tells the reader that the candidate knows the two-row structure. Without that line, a reader has to infer whether Σ aₙ was tested directly, which costs clarity and can cost a point when the test applied is one that only works on positive series. With the line, the argument reads as a deliberate decision, and the rubric's two rows become visible to the scorer.
There is also a third case that the rubric scores as a separate verdict: divergence. If Σ |aₙ| diverges and Σ aₙ also diverges, the series is divergent, and a candidate who writes a correct nth-term or limit-comparison argument earns the divergence row. The conditional-convergence verdict only exists when the absolute-value series diverges and the original signed series converges. That asymmetry is what makes the question feel like a fork rather than a checklist.
Four tests the AP exam uses to decide the absolute-convergence row
The absolute-convergence row is scored as a positive-series test applied to |aₙ|. The four tests that appear most often on AP Calculus BC FRQs are the p-series test, the geometric-series test, the integral test, and the limit comparison test. Each one has a different algebraic signature, and the reader will accept any of them so long as the setup row, the comparison row, and the verdict row are all present.
The p-series test is the fastest. If |aₙ| = 1/nᵖ, the verdict is convergence for p > 1 and divergence for p ≤ 1. The AP reader scores this in under a line: write the form, identify p, state the verdict. There is no partial credit for identifying the form without naming p, and there is no point for naming p without stating the verdict.
The geometric-series test applies when |aₙ| = c·rⁿ or c·rⁿ⁻¹ for some ratio r. The verdict is convergence when |r| < 1, divergence when |r| ≥ 1. The rubric scores two things: that the student identified the ratio correctly, and that the student compared |r| to 1 rather than r to 1. A candidate who writes r < 1 instead of |r| < 1 loses a row, because the absolute-value comparison is precisely the absolute-convergence row.
The integral test applies when |aₙ| is positive, decreasing, and continuous on the relevant interval. The student writes the corresponding integral, evaluates it, and uses the result as the verdict. On the AP exam this test shows up most often when |aₙ| = 1/(n·ln n) or a similar logarithmic form. The reader scores the integral's value, not the work, so a clean evaluation matters more than the antiderivative method.
The limit comparison test is the workhorse. The student picks a comparison series bₙ (almost always a p-series or geometric series), computes L = lim (n→∞) |aₙ| / bₙ, and then uses 0 < L < ∞ to transfer the verdict. On the AP exam the comparison series is usually dictated by the dominant term in |aₙ|: if |aₙ| behaves like 1/n² for large n, the comparison is 1/n²; if it behaves like (2/3)ⁿ, the comparison is (2/3)ⁿ. The reader scores the chosen bₙ, the computed limit, and the inferred verdict. A limit of 0 or ∞ does not earn the verdict row on its own, because the limit comparison test requires 0 < L < ∞.
Three tests the AP exam uses to decide the conditional-convergence row
Once the absolute-convergence row has been scored, the conditional-convergence row is only relevant if the absolute-value series diverges. The three tests that appear on AP Calculus BC FRQs for the conditional case are the alternating-series test, the Abel test, and Dirichlet's test. Of these, the alternating-series test is by far the most common, and it is the one a candidate should default to whenever aₙ contains a factor of (−1)ⁿ or (−1)ⁿ⁺¹.
The alternating-series test has two preconditions: the absolute values |aₙ| must decrease monotonically to 0, and the limit of aₙ as n → ∞ must be 0. The rubric scores each precondition as its own line. A candidate who writes 'by the alternating-series test' without checking both preconditions loses the setup row, even if the final verdict is correct. A candidate who checks the preconditions but writes the wrong verdict loses the verdict row, and a candidate who skips directly to 'therefore the series converges' loses both rows.
The Abel test applies when the partial sums of one factor are bounded and the other factor decreases monotonically to 0. On the AP exam this usually takes the form Σ aₙ bₙ where aₙ has bounded partial sums (often alternating ±1) and bₙ is a positive decreasing sequence with limit 0. The reader scores the boundedness claim, the monotone decrease claim, and the limit-0 claim. This test is rare on the FRQ but appears in multiple-choice items, and recognising its structure is worth a few seconds of saved time.
Dirichlet's test is a generalisation of Abel. The partial sums of one factor are bounded, the other factor decreases monotonically to 0. It is functionally similar to Abel on the exam, and the reader accepts either test as a valid argument. In practice, a candidate should write the simpler Abel form unless the problem specifically signals Dirichlet, because the AP reader does not award extra credit for a more general test and the simpler form is faster to write.
Worked example: a series that is conditionally, not absolutely, convergent
Consider the series Σ (−1)ⁿ⁺¹ / √n. The first row is the absolute-convergence row. Take the absolute value: Σ 1/√n = Σ 1/n¹ᐟ². This is a p-series with p = 1/2, and because p ≤ 1, the series of absolute values diverges. The reader awards the absolute-convergence row to any candidate who writes the absolute-value step, identifies the p-series, names p, and states that divergence follows.
The second row is the conditional-convergence row. The original series is alternating, so apply the alternating-series test. The absolute values |aₙ| = 1/√n decrease monotonically: for n ≥ 1, 1/√(n+1) < 1/√n, and the sequence is strictly decreasing. The limit of aₙ as n → ∞ is 0, since 1/√n → 0. Both preconditions are satisfied, so the alternating-series test applies and the series converges. The reader awards the conditional-convergence row, and the final verdict is conditional convergence.
Now consider Σ (−1)ⁿ⁺¹ / n². The absolute-value series is Σ 1/n², a p-series with p = 2 > 1, so Σ |aₙ| converges. The series is therefore absolutely convergent. Notice that the alternating-series test still applies, and it still gives convergence, but the verdict on the absolute-convergence row is the one the reader scores first, and absolute convergence is a stronger statement than conditional convergence. A candidate who writes 'conditionally convergent' here loses a point, because the rubric scores the strongest correct verdict.
Worked example: a series where both rows diverge
Consider Σ (−1)ⁿ⁺¹ · (n / (n+1)). The absolute-value series is Σ n/(n+1), whose general term tends to 1, not 0, so the absolute-value series diverges by the nth-term test. The original alternating series has the same general term with an alternating sign, so its general term does not tend to 0 either, and the alternating-series test cannot apply. Both rows diverge, and the series is divergent. The reader awards the divergence row to any candidate who correctly applies the nth-term test to either the original or the absolute-value series and notes that the limit of aₙ is not 0.
This second example shows a common error: a candidate runs the alternating-series test without checking the limit-of-aₙ precondition, and writes 'converges by the alternating-series test'. The reader scores the setup row against the candidate, finds that the limit-precondition was never checked, and awards zero for the conditional-convergence row. The habit of writing the limit check explicitly, even when it is obvious, costs nothing and protects the row.
Common pitfalls and how to avoid them
The four errors that cost the most marks on absolute versus conditional convergence are: forgetting the absolute-value step, running a positive-series test on a signed series, mis-stating the alternating-series test's preconditions, and confusing the absolute-value comparison for the geometric-series test.
Forgetting the absolute-value step. The single most common error is to apply a p-series or limit comparison test directly to aₙ rather than |aₙ|. The fix is mechanical: before any test, write the line 'consider Σ |aₙ|' in the solution. The line takes three seconds, makes the two-row structure visible, and protects the absolute-convergence row.
Running a positive-series test on a signed series. The p-series test, geometric-series test, and integral test only apply to positive series. A candidate who applies them to Σ (−1)ⁿ / n is using the wrong tool. The fix is to check the sign of aₙ before choosing a test. If aₙ changes sign, the candidate's first move should be the alternating-series test, not a comparison test.
Mis-stating the alternating-series test's preconditions. The two preconditions are monotonic decrease of |aₙ| and limit of aₙ equal to 0. The reader scores each one separately. The fix is to write both checks as labelled lines, for example 'monotone decrease: |aₙ₊₁| < |aₙ| for n ≥ N' and 'limit: lim aₙ = 0', and only then write 'by the alternating-series test, the series converges'.
Confusing the absolute-value comparison for the geometric-series test. For a geometric series with ratio r, the absolute-convergence comparison is |r| < 1. A candidate who writes r < 1 loses the absolute-convergence row because the comparison is missing the absolute value. The fix is to write |r| explicitly, even when r is positive, because the rubric reads for that specific character.
How the AP reader scores a partially correct answer
On a typical AP Calculus BC FRQ worth 9 points, the absolute-convergence row is worth 1 point, the conditional-convergence row is worth 1 point, and the sum or radius-of-convergence row is worth 1 point, with the remaining points distributed across setup, justification, and the use of correct mathematical language. The reader scores each row independently, so a candidate who earns the absolute-convergence row but misses the conditional-convergence row still scores 1 out of 2 on the convergence portion of the question.
Partial credit is also available within a single row. On the absolute-convergence row, a candidate who writes the correct comparison series but computes the limit incorrectly earns the setup row but not the verdict row. On the conditional-convergence row, a candidate who checks one of the two alternating-series preconditions correctly earns a partial point. The reader is instructed to award what is earned and to read charitably when a candidate's work is ambiguous but pointed in the right direction.
One practical consequence of this scoring is that a candidate should always write something. A blank row scores 0, but a partially correct row scores at least a partial point. The habit of writing the absolute-value step, the chosen test, the setup, and the verdict — even when the candidate is unsure — protects at least one of the two rows. In my experience, candidates who blank a row because they cannot decide which test to use lose the most points, and candidates who write a wrong test on a signed series lose fewer, because the sign of aₙ is at least visible to the reader.
Preparation strategy: the 90-minute drill that maps to the FRQ
For most candidates preparing for the absolute-convergence and conditional-convergence rows, the highest-leverage drill is a 90-minute timed set of FRQ-style prompts. Take six series prompts, set a 15-minute timer for each, and write a full two-row argument for every series. The prompts should cover at least one p-series, one geometric series, one alternating series with a logarithmic or polynomial denominator, one alternating series with a non-zero limit, and one series that needs a limit comparison. Mark each row as earned or not, and identify which test the candidate is most often skipping.
After the first 90-minute drill, a clear pattern emerges. Most candidates are solid on the absolute-convergence row for p-series and geometric series, weaker on the limit comparison row, and weakest on the conditional-convergence row when the alternating-series test's preconditions need to be checked explicitly. The drill exposes this, and the second 90-minute drill — same prompts, but with the focus shifted to the weak row — usually closes the gap.
A second preparation habit is to read the AP Calculus BC Course and Exam Description's series section with a highlighter. The description lists the four convergence tests that are scored, the two-row structure of every convergence question, and the explicit requirement that a candidate check both preconditions of the alternating-series test. Memorising this list is faster than memorising a textbook chapter, and it maps directly to the rubric.
Question-type triage: how to recognise the verdict in under 60 seconds
The fastest way to triage an absolute-versus-conditional-convergence question is to read the general term, identify the sign pattern, and decide which row is the binding constraint. The table below summarises the four most common shapes and the verdict each one implies.
| Shape of aₙ | Absolute-value verdict | Original-series verdict | Final classification |
|---|---|---|---|
| Positive and a p-series with p > 1 | Converges | Converges | Absolutely convergent |
| Positive and a p-series with p ≤ 1 | Diverges | Diverges (no sign change to exploit) | Divergent |
| Alternating and 1/nᵖ with p > 0 | Σ 1/nᵖ diverges for p ≤ 1 | Converges by alternating-series test | Conditionally convergent when p ≤ 1; absolutely convergent when p > 1 |
| Alternating and aₙ → c ≠ 0 | Diverges by nth-term test | Diverges by nth-term test | Divergent |
Reading this table, the most important triage step is to ask whether aₙ has a sign. If aₙ is positive, the absolute-convergence row is the only one that matters, because the absolute-value series equals the original series. If aₙ has a sign, the absolute-convergence row is the first decision, and the conditional-convergence row is the second. A candidate who triages on this single question — positive or signed — picks the right test family in under 60 seconds and saves the remaining time for the verdict row.
Conclusion and next steps
Absolute and conditional convergence is a two-row decision, and the AP Calculus BC rubric scores each row independently. The absolute-convergence row is a positive-series test applied to |aₙ|, and the conditional-convergence row is a signed-series test applied to aₙ — most often the alternating-series test. A candidate who writes the absolute-value step explicitly, applies the correct test family to each row, and checks both preconditions of the alternating-series test earns both rows in nearly every FRQ prompt. The habit that costs the most marks is skipping the absolute-value step; the habit that protects the most marks is writing both preconditions of the alternating-series test as labelled lines.
AP Courses' one-to-one AP Calculus BC programme analyses each student's free-response convergence arguments row by row, maps the missed rows back to the specific test family, and turns the conditional-convergence verdict into a concrete, prompt-by-prompt preparation plan.