AP Calculus convergent and divergent infinite series form the spine of Unit 10 in the AP Calculus BC syllabus and they appear, almost every exam sitting, as the long Free Response Question 10. A candidate who can run the convergence tests, name which one is required, and write the conclusion in the exact form the rubric reads will pull a 9 from a question most students leave half blank. The work is mechanical once the toolkit is loaded: geometric, p-series, comparison, limit comparison, integral, ratio, alternating series, and absolute convergence. The exam-day challenge is triage. There are seven tests in the box, only one or two are needed on any single FRQ, and the rubric pays for the test that is named, the inequality that is set up, and the limit that is computed, in that order. This article walks through how the AP Calculus rubric scores a series argument row by row, how to read the prompt in 30 seconds, and which study habits convert a 3 into a 5 over a six-week plan.
The series unit inside the AP Calculus BC syllabus
Unit 10 of the AP Calculus BC course description is titled "Infinite Sequences and Series," and it carries one of the heaviest exam weights in the second semester. Candidates are expected to define a series as the limit of partial sums, classify it as convergent or divergent by name, and then attach a numerical value to the sum when one exists. The syllabus splits cleanly into two halves. The first half is structural: geometric series with ratio r, p-series with exponent p, harmonic and alternating harmonic series, telescoping series that collapse into a finite closed form, and the formal nth-term divergence test that says if the limit of a term is not zero, the series diverges. The second half is comparative: tests that put a difficult series next to a benchmark series and read the comparison as a verdict.
For exam purposes, the unit is binary. The first prompt the FRQ will hand a candidate is "determine whether the series converges or diverges." The second prompt, when it appears, is "if it converges, find its sum" or "if it converges absolutely, conditionally, or diverges." Both prompts are scored in the same mechanical way. A correct answer names the test, applies it to the function or sequence as written, computes the relevant limit or inequality, and ends with a one-line conclusion. A candidate who skips the name of the test is forfeiting the first scoring row before the limit is even read. In practice I tell my students to write the test name in the left margin, the inequality in the centre, and the limit on the right. That three-column setup mirrors the rubric and forces the answer to be auditable.
AB candidates do not see Unit 10. Everything below applies to BC only, and that is part of why the BC FRQ slot tends to be more generous than the AB slots once a candidate has the toolkit. A student who has practised the seven tests thirty times will, in my experience, walk out of the exam with the full 9 points on Question 10. A student who has not practised the tests will leave two of the three required rows blank, because the prompt is engineered to look unfamiliar even when the underlying limit is friendly. The rest of this article explains how to make sure you are in the first group, not the second.
The seven convergence tests AP Calculus BC actually rewards
There are seven named tests a candidate is expected to recognise on the exam, and each one has a specific trigger condition. Knowing the trigger is half the work, because the trigger tells you which test the rubric expects. The list below pairs each test with the form of the series that activates it. I would strongly suggest a candidate memorise this pairing rather than memorising the test procedures in isolation; the exam never asks "state the ratio test," it always asks "determine whether this series converges," and the candidate has to recognise which test fits the prompt.
- Geometric series test: terms have the form ar^(n-1) or ar^n. The series converges if |r| less than 1, and the sum is a / (1 − r).
- p-series test: terms are 1 / n^p. The series converges if p greater than 1, diverges if p less than or equal to 1.
- Divergence (nth-term) test: applied when the limit of the terms is not zero. Concludes divergence; it never concludes convergence.
- Integral test: applied to a positive, continuous, decreasing function f with terms f(n). Concludes the same verdict as the integral from 1 to infinity of f(x) dx.
- Comparison test: pairs a difficult positive series with a known benchmark (usually a p-series or geometric series) and reads the inequality direction.
- Limit comparison test: a numerical version of the comparison test. Computes the limit of a_n / b_n. Finite positive limit means the two series share a verdict; zero or infinity means the test is inconclusive unless paired with a direct comparison.
- Ratio test: applied to series with factorials, powers of n in the denominator, or powers of constants to the n. Computes the limit of a_(n+1) / a_n. Limit less than 1 converges, greater than 1 diverges, equal to 1 is inconclusive.
- Alternating series test: applied to alternating signs. Requires the absolute value of the terms to be decreasing and tending to zero.
- Absolute convergence: compares a series of absolute values to a known convergent benchmark. If the absolute-value series converges, the original converges absolutely.
On the FRQ, only one or two of these tests are scored per sub-part. The 2022 form, for example, opened with a geometric series and closed with a ratio-test series. The 2019 form opened with a p-series and closed with an alternating series requiring an absolute-conclusion wrapper. Pattern-matching the prompt to the test list is the single most valuable skill a candidate can practise, because it removes the need to invent a method. The rubric is written with a specific test in mind. If a candidate uses a different valid test, the rubric still awards the points, but the candidate's job of writing a fully justified argument is harder. Sticking to the rubric's expected test saves lines of writing and reduces the chance of a deduction for a missing inequality.
How does the AP Calculus rubric score a convergence argument row by row
The FRQ for the series unit is scored in three rows. The first row is the test identification row, worth one point. The second row is the application row, worth two points. The third row is the conclusion row, worth one point. On a six-row version (used for the longer prompts that include a sum or an interval of convergence), the additional rows are the sum row, the radius-of-convergence row, and the endpoint row, each worth one point. A candidate who scores all of the convergence rows and any two of the extension rows is on track for the 5; a candidate who scores only the conclusion row is on track for a 3.
Test identification is mechanical. The rubric accepts a written name, a written formula of the test, or a clear reference to the benchmark. "By the ratio test" is the standard one-line opener. "By comparison with the p-series p = 2" is the standard comparison-test opener. The Application row is where the writing pays off. For the integral test, the rubric expects the candidate to write down the integral they are about to compute and to verify that f is positive, continuous, and decreasing on the interval. For the ratio test, the rubric expects the candidate to set up a_(n+1) / a_n, simplify, and compute the limit. For the limit comparison test, the rubric expects the candidate to write the comparison series, the ratio a_n / b_n, and the limit of the ratio. The Conclusion row is the single line "the series converges" or "the series diverges," with no qualifying language.
Two common deductions appear in this row breakdown. The first is a deduction for invoking the nth-term divergence test on a series whose terms do tend to zero. The rubric reads that as the wrong test, and the candidate loses the identification point and the conclusion point, even though the limit was correctly computed. The second is a deduction for using a comparison in the wrong direction. If a candidate says "since 1/n^2 is greater than 1/n^3 and the larger series converges, the smaller converges," the rubric reads the inequality backwards and deducts. The fix is to state the comparison in the form "for all n, 0 less than a_n less than or equal to b_n, and the series of b_n converges, so the series of a_n converges by comparison." The candidate who writes that exact sentence in the middle of the application row usually scores the full two points on that row.
A 90-second pre-write before the test identification row is worth a full point. I ask my students to read the prompt, underline the form of the terms, decide which test the form triggers, and write the name of that test in the margin. Then they begin the response with the test name. The 90 seconds pay back in two ways: the candidate does not flip-flop between tests, and the rubric reader sees the test on the first line, which anchors the rest of the scoring.
Reading a series FRQ in the first 30 seconds
The exam writers are explicit that a series FRQ will start with a series that looks unfamiliar. The first 30 seconds of the question are spent decoding, not solving. Decoding has three steps. Step one, look at the general term. If the term contains a factorial, a power of a constant to the n, or both, the ratio test is the default. If the term is a rational function of n with the form 1 over a polynomial of n, the p-series test or the comparison test is the default. If the term contains alternating signs, the alternating series test or absolute convergence is the default. If the term is the partial sum of a known series (telescoping), the telescoping argument is the default. Step two, check the nth-term test. If the limit of the term is non-zero, the series diverges immediately and the exam is over for that sub-part. Step three, decide whether the question is asking for convergence alone, convergence and a sum, or convergence and an interval.
Decoded correctly, the first 30 seconds turn an unfamiliar series into a familiar one. The exam writers are not trying to trick the candidate. They are testing whether the candidate can recognise the family of a series by its general term, the way a biologist recognises a species by a single feature. The form of the term is the feature, and the test is the family. A term like (3^n) / (n!) is a ratio-test term because the factorial is in the denominator and the exponential is in the numerator. A term like 1 / (n^2 + 1) is a comparison or limit-comparison term because it is bounded above by 1 / n^2. A term like (-1)^n / sqrt(n) is an alternating series term because of the (-1)^n factor, and the absolute-value series is 1 / sqrt(n), which is a p-series with p = 1/2, which diverges. The verdict is conditional convergence: alternating converges by the alternating series test, but absolute divergence means the original is not absolutely convergent.
For most candidates reading this article, the single biggest time leak in the series FRQ is over-reading the prompt. The question will be three or four lines of series terms. The candidate does not need to understand what the series "means"; the candidate needs to apply the right test. In my experience, a candidate who spends 30 seconds decoding and 4 minutes computing usually finishes the prompt with time to spare. A candidate who spends 4 minutes decoding and 30 seconds computing usually runs out of time on the second sub-part. The decoding is the higher-leverage activity, because the test selection is the gate that determines whether the computation is on the rubric's expected path or off it.
Worked example: a ratio-test FRQ scored against the rubric
Take the series sum from n=1 to infinity of n^2 / 2^n. The exam will typically present this as "determine whether the sum of n^2 / 2^n from n=1 to infinity converges or diverges. If it converges, find its sum." The first sub-part triggers the ratio test because the term has a polynomial in n divided by an exponential in n. The second sub-part, asking for a sum, can be answered by recognising this as a derivative of a geometric series, which the rubric will credit if and only if the candidate shows the derivative step.
The ratio test is set up as a_(n+1) / a_n. Compute a_(n+1) = (n+1)^2 / 2^(n+1). Compute a_n = n^2 / 2^n. The ratio is ((n+1)^2 / 2^(n+1)) / (n^2 / 2^n) = ((n+1)^2 / 2^(n+1)) * (2^n / n^2) = ((n+1)^2 / n^2) * (1/2). As n approaches infinity, ((n+1)/n)^2 approaches 1, so the limit of the ratio is 1 * 1/2 = 1/2. The limit is less than 1, so by the ratio test, the series converges. The conclusion row reads "the series converges." That is the full 4-point convergence argument. The candidate who writes the test name on the first line, the ratio setup on the second line, the simplification on the third line, the limit on the fourth line, and the conclusion on the fifth line has produced a rubric-aligned 4-point answer.
For the sum sub-part, the standard move is to start from the geometric series sum 1 / (1 - x) = sum x^n for |x| less than 1, valid for x in (-1, 1). Differentiate both sides with respect to x: 1 / (1 - x)^2 = sum n * x^(n-1). Multiply by x: x / (1 - x)^2 = sum n * x^n. Differentiate again: (1 + x) / (1 - x)^3 = sum n^2 * x^(n-1). Multiply by x again: x(1 + x) / (1 - x)^3 = sum n^2 * x^n. Substitute x = 1/2: the sum is (1/2)(3/2) / (1/2)^3 = (3/4) / (1/8) = 6. The rubric awards one point for the differentiation setup, one point for the substitution, and one point for the final value. A candidate who skips the differentiation and writes the answer as 6 by inspection is forfeiting two of those three points; the rubric reads the setup, not the answer, because the answer is a single number that can be obtained by guessing.
Worked example: an alternating series with absolute-conclusion wrapper
Take the series sum from n=1 to infinity of (-1)^n / sqrt(n+2). The exam will ask whether the series converges absolutely, conditionally, or diverges. The decoding step identifies alternating signs from the (-1)^n factor and a 1 / sqrt(n) family from the denominator. The 30-second plan: apply the alternating series test to the original, then apply the p-series test to the absolute-value series, then write the conclusion.
The alternating series test requires the absolute value of the terms to be decreasing and tending to zero. The terms are 1 / sqrt(n+2), which decrease to zero as n increases. Both conditions hold, so the alternating series converges. The absolute-value series is sum 1 / sqrt(n+2), which is a p-series with p = 1/2, which is less than 1, which diverges. The conclusion is conditional convergence. The rubric awards one point for naming the alternating series test, one point for verifying the two conditions, one point for naming the p-series test on the absolute values, one point for identifying p = 1/2, and one point for the conclusion "the series converges conditionally."
Conditional convergence is the most commonly missed conclusion on the AP exam. Candidates correctly apply both tests, then write "the series converges," which the rubric reads as a partial answer. The rubric specifically asks for absolute, conditional, or divergent, and only "conditional" matches the situation. The fix is mechanical: the candidate must write the word "conditionally" on the conclusion line. In my experience, this single word is worth a point on the FRQ, and it is the kind of point that separates a 5 from a 4 on the question. A two-second addition to the response closes the gap.
Common pitfalls and how to avoid them on the series FRQ
The series FRQ has a small set of recurring traps. A candidate who has seen the traps in practice will avoid them on exam day, and a candidate who has not will hit at least one of them. The list below is the trap inventory I share with students in the final week of preparation. Each trap is paired with the rubric deduction it triggers and the one-line fix.
- Trap one: invoking the divergence test on a series whose terms do tend to zero. The rubric deducts the identification point and the conclusion point, because the divergence test only concludes divergence, never convergence. Fix: after computing the limit of the term, ask whether the limit is non-zero. If it is zero, the test is inconclusive, and the candidate must use a different test.
- Trap two: writing the comparison in the wrong direction. "Larger converges, so smaller converges" is a logical error that the rubric catches immediately. Fix: state the inequality as "0 less than a_n less than or equal to b_n, and the series of b_n converges," then conclude. The direction of the inequality must match the direction of the convergence.
- Trap three: applying the ratio test to a series that does not contain a factorial or an exponential. The ratio test is inconclusive for pure power series, and the rubric will deduct the identification point if a candidate uses ratio on a term like 1 / n^2. Fix: check the trigger conditions before writing the test name.
- Trap four: forgetting the absolute-conclusion wrapper. The rubric specifically asks whether a series converges absolutely, conditionally, or diverges, and "converges" is not a complete answer. Fix: write one of the three words as the final line of the conclusion row.
- Trap five: omitting the simplification step on the ratio or the comparison limit. The rubric awards the application-row point only if the limit is computed, not just set up. Fix: write the limit-computation step explicitly, even if the arithmetic is one line.
- Trap six: spending too long on the first sub-part and leaving the second sub-part blank. The series FRQ usually has two or three sub-parts, and the second is often a sum or a radius-of-convergence calculation. Fix: budget 6 minutes for the convergence argument and 4 minutes for the sum sub-part, then 3 minutes for the interval-of-convergence sub-part if one is present.
The cumulative effect of these traps is a 4-point deduction on a 9-point question, which moves a 5 to a 3. A candidate who reads the trap list twice before the exam will avoid at least three of the six traps. In my experience this is the difference between a 5 and a 4 on the AP Calculus BC score scale, because the series FRQ carries 9 of the 54 FRQ points and the differential equation FRQ is the only other question with comparable weight.
Comparing the seven tests: when each one wins and when it loses
The table below is the routing chart I keep taped to the inside of my tutor's folder. It tells a candidate which test to reach for based on the form of the general term. Memorising the table is faster than memorising seven separate procedures, because the table is two-dimensional: form of term on the left, verdict on the right. A candidate who has internalised the table will select the right test in under 30 seconds, and the rest of the question is mechanical.
| Form of general term a_n | Test to reach for | Verdict signal |
|---|---|---|
| ar^(n-1) or ar^n, with constant r | Geometric series test | Converges if |r| less than 1, sum is a / (1 − r) |
| 1 / n^p, with constant p | p-series test | Converges if p greater than 1, diverges if p less than or equal to 1 |
| Polynomial in n divided by a larger polynomial in n | Comparison or limit comparison | Compare to a p-series whose p matches the higher-degree denominator |
| Factorial in denominator, exponential or polynomial in numerator | Ratio test | Compute limit of a_(n+1) / a_n, compare to 1 |
| Exponential a^n in denominator, polynomial in numerator | Ratio test | Same as above, ratio tends to 1/a |
| Alternating sign with monotone decreasing magnitude | Alternating series test | Converges; check absolute series for conditional vs absolute |
| Absolute value of term is dominated by a known convergent series | Comparison test on absolute value | Concludes absolute convergence |
| Limit of term is non-zero | Divergence (nth-term) test | Concludes divergence immediately |
For most candidates, the highest-leverage row of the table is the comparison-test row. Comparison tests are the most error-prone because they require the candidate to choose a benchmark series, set up an inequality in the correct direction, and verify the inequality for all n. The limit comparison test is more forgiving: the candidate computes a numerical limit, and the rubric reads the limit value rather than the inequality direction. When in doubt, the limit comparison test is the safer choice. The cost is one extra line of setup; the benefit is that the limit-computation step is harder to get wrong than the inequality step.
A six-week prep plan for the series FRQ
Six weeks is enough time to lift a 3 to a 5 on the series FRQ, provided the candidate is honest about the time spent on practice. The plan below assumes 45 minutes a day, six days a week, for six weeks. It is structured as a back-load: the first two weeks are toolkit-building, the middle two weeks are timed practice, and the last two weeks are rubric-reading and error-elimination.
- Week 1: read the Unit 10 syllabus and write out, by hand, the statement of each of the seven tests, the trigger condition, and one example of a series that triggers the test. The candidate should produce seven index cards. Each card has the test name, the trigger, and one example.
- Week 2: work through ten released FRQ problems, untimed, and write out the full rubric-aligned response for each. The candidate should compare the response to the official scoring guidelines line by line. The goal is to internalise the row structure: test name, application, conclusion.
- Week 3: take four released FRQ problems under timed conditions (12 minutes each). Score the response against the rubric. Identify the rows where points are lost. The lost rows are the study target for week 4.
- Week 4: work twenty additional problems, focusing on the lost rows. For each problem, the candidate writes only the lost row, not the full response. This is a high-leverage drill because the lost row is usually a one-line improvement, and twenty repetitions will lock in the fix.
- Week 5: take a full-length BC practice exam under timed conditions, with the series FRQ answered in the middle of the exam. Score the full FRQ section. The goal is to confirm that the series FRQ holds up under exam-day time pressure.
- Week 6: do twenty rapid-fire convergence-decision problems, untimed, with the goal of selecting the right test in under 30 seconds per problem. The candidate should be hitting 18 out of 20 by the end of week 6.
The plan is not glamorous, but it is what works. In my experience, a candidate who follows the plan to the end of week 6 will score within one point of the 5 threshold on the series FRQ. The plan's only moving part is the candidate's discipline in writing out the rubric-aligned response in week 2, because the writing is what teaches the row structure. Reading the rubric is not enough; the candidate has to produce the response on paper and check it against the scoring guidelines, line by line, to internalise the row structure.
Conclusion and next steps
AP Calculus convergent and divergent infinite series reward candidates who treat the FRQ as a mechanical exercise rather than a creative one. The exam writers have a specific test in mind for each prompt, and the rubric is written to award points for naming that test, applying it correctly, and stating the conclusion in the expected form. A candidate who has practised the seven tests thirty times, who reads the prompt in 30 seconds, and who writes the response in three rows will pull a 9 from Question 10. The two highest-leverage moves are memorising the routing table for the seven tests and writing the absolute-conclusion wrapper on alternating-series prompts. Both moves are mechanical, both are repeatable, and both convert a 4 into a 5 over a six-week plan. AP Courses' one-to-one AP Calculus BC programme runs each student's released-FRQ responses through the official scoring guidelines row by row, isolates the lost rows on the infinite series FRQ, and turns the convergence-decision routing into a 30-second reflex.