Gregory Friedman The Greg Wide Web on the name →

The Drill

A question a day

The GRE Mathematics Subject Test is 66 questions in 170 minutes, roughly half calculus, a quarter algebra and a quarter everything else. This is a drawer of questions to meet one of them a day, a rapid-fire run built to the same proportions when you want the whole shape at once, and a record of which ones went badly, kept on the machine you study on or kept nowhere at all.

Whose tool this is

Mine. I am sitting the mathematics subject test, and this is the drawer I study out of, put on a public site because the questions are worth reading and the design of the thing is worth arguing about. That matters for what it does with your answers, so here it is plainly.

There is no account, because there is no server: this site is static files, and there is nowhere for a login to go. Nothing you do here is sent anywhere, and nothing about you is stored at all unless you press one button, at which point every attempt is written to this browser and to nowhere else. No streaks, no reminders, no countdown to test day, and no percentage presented as a mark. The panel further down keeps coverage and evidence: how much of a topic you have seen, how many you had first time, which way the last few went.

Today's question is not chosen by a nightly job. It is computed from your own calendar date, which is why it works from a cached copy, works on a plane, and cannot be wrong because a build failed on a Tuesday. The drawer currently holds 780 days before it comes back around, which is both an honest disclosure and a standing argument for writing more of them.

Applications of integrals · difficulty 3 of 3 · about 2.6 min

Revolve the curve y=1xy = \dfrac{1}{x}, for x≥1x \ge 1, about the xx-axis. Which of the following is true of the resulting infinite solid?

  1. A Its volume and its surface area are both finite.
  2. B Its volume and its surface area are both infinite.
  3. C Its volume is finite and its surface area is infinite.
  4. D Its volume is infinite and its surface area is finite.
  5. E Its volume and its surface area are both equal to π\pi.
The solution

The answer is C

Volume. By disks,

V=π∫1∞(1x)2dx=π∫1∞x−2dx=π.V = \pi\int_{1}^{\infty}\left(\frac{1}{x}\right)^{2}dx = \pi\int_{1}^{\infty}x^{-2}dx = \pi.

Finite, by the pp-test with p=2>1p = 2 > 1.

Surface area.

S=2π∫1∞1x1+1x4 dx ≥ 2π∫1∞dxx=∞,S = 2\pi\int_{1}^{\infty}\frac{1}{x}\sqrt{1 + \frac{1}{x^{4}}}\,dx \ \ge\ 2\pi\int_{1}^{\infty}\frac{dx}{x} = \infty,

since the radical is at least 11. Infinite, by comparison with the harmonic integral.

So the solid holds π\pi cubic units of paint but its wall has infinite area. The apparent paradox (fill it and you have painted the inside surface with finitely much paint) dissolves once you notice that the layer of paint would have to thin to zero thickness, whereas painting a surface means a coat of fixed thickness.

The mathematical content is that squaring the radius changes the exponent from 11 to 22, and p=1p = 1 is precisely the borderline at which ∫1∞x−pdx\int_{1}^{\infty}x^{-p}dx turns from divergent to convergent. This solid is known as Torricelli's trumpet or Gabriel's horn.

Adapted from APEX Calculus, Section 7.4 (Arc Length and Surface Area) and Section 6.8 (Improper Integration) by Gregory Hartman, with Jennifer Bowen, Troy Siemers, Brian Heinold and Dimplekumar Chalishajar (APEX Calculus); PreTeXt edition hosted by the University of Lethbridge · CC BY-NC 4.0 · retrieved 2026-08-13. Changes: Adapted at the section level; the section's surface-of-revolution formula was combined with the section on improper integrals to build the classical finite-volume, infinite-area comparison. Written as five choices covering every combination of finite and infinite.

This is the question for 18 November 2026, the day the page was last built. With JavaScript on, the drill hands you the question for your own today instead, and the choices become buttons that mark themselves.

Plate I The question for today, computed from your own date; or a set dealt from the same drawer.

Rapid fire

The sheet above is a sheet: everything at once, in the test's proportions, to work through or print. Rapid fire is the same composition served one question at a time, with the clock running and the response time kept on every attempt.

The clock counts up and nothing expires. The real test allows 170 minutes for 66 questions, so the pace figure sits beside the elapsed time as a reference; it is not a limit, nothing advances on its own, and no question is ever taken away mid-sentence. A stopwatch that reports is a study instrument. A timer that can fail you is an examination, and this page does not sit examinations.

The real test is 66 questions in 170 minutes, which is 2 min 35 sec a question. A run here is sampled to the same composition and the clock counts up rather than down: nothing expires, nothing advances on its own, and the pace figure is a reference to read yourself against rather than a limit to fail.

Every topic of the drill, the share of the real test it carries, how many questions a full-length run of 66 would draw from it, and how many the drawer holds.
Topic Share of the test In a run of 66 In the drawer
Limits and continuity Calculus 3.5% 2 49
Derivatives Calculus 7.5% 5 38
Applications of derivatives Calculus 7.0% 4 37
Integrals Calculus 7.5% 5 37
Applications of integrals Calculus 5.5% 4 36
Sequences and series Calculus 6.0% 4 25
Multivariable calculus Calculus 5.5% 3 37
Multiple integrals Calculus 3.0% 2 37
Vector calculus Calculus 2.0% 1 38
Differential equations Calculus 2.5% 2 39
Elementary algebra Algebra 4.5% 3 32
Trigonometry Algebra 2.5% 2 28
Linear algebra Algebra 8.0% 5 40
Group theory Algebra 4.5% 3 37
Rings and fields Algebra 3.0% 2 34
Number theory Algebra 2.5% 2 33
Real analysis Additional topics 7.0% 4 27
Complex analysis Additional topics 2.5% 2 18
Point-set topology Additional topics 3.0% 2 31
Probability Additional topics 3.5% 2 17
Statistics Additional topics 1.5% 1 23
Combinatorics Additional topics 3.0% 2 16
Graph theory Additional topics 1.5% 1 17
Numerical analysis Additional topics 1.0% 1 18
Algorithms Additional topics 1.0% 1 17
Logic and set theory Additional topics 1.0% 1 19
A full-length run 100% 66 780
Plate II Rapid fire: a run sampled to the real test's composition, timed but never cut off.

The record

Nothing below is a grade. It is coverage and evidence: what you have met, what you had on the first attempt, and how the last twenty attempts in a topic compare with the whole record, which is the only way improvement shows up rather than drowning in a running average. The confidence chip has three states and no fourth, because three is what this much evidence supports.

The drawer holds 780 questions across 26 of 26 topics. Everything in this table is the shape of the bank, not of anybody's reading of it.

Every topic in the drill: how many questions the bank holds, what share of the real test the topic is, and, once a record is kept on this device, how much of it has been seen.
Topic In the drawer Seen First try Last 20 Trend Median time Pace Confidence
Limits and continuity Calculus 49
Derivatives Calculus 38
Applications of derivatives Calculus 37
Integrals Calculus 37
Applications of integrals Calculus 36
Sequences and series Calculus 25
Multivariable calculus Calculus 37
Multiple integrals Calculus 37
Vector calculus Calculus 38
Differential equations Calculus 39
Elementary algebra Algebra 32
Trigonometry Algebra 28
Linear algebra Algebra 40
Group theory Algebra 37
Rings and fields Algebra 34
Number theory Algebra 33
Real analysis Additional topics 27
Complex analysis Additional topics 18
Point-set topology Additional topics 31
Probability Additional topics 17
Statistics Additional topics 23
Combinatorics Additional topics 16
Graph theory Additional topics 17
Numerical analysis Additional topics 18
Algorithms Additional topics 17
Logic and set theory Additional topics 19
Plate III Coverage, evidence, pace and trend, per topic. Kept on this device only, and only if you ask.

Which way it is going

The table above is the record in total. This one is the record across time, which is where improvement actually lives: accuracy and response time on the same buckets, the topics worth working on next with the strength of the evidence attached to each, and what you are practising against what the test asks for. A topic seen twice is not weak, it is unmeasured, and the panel keeps the two apart rather than ranking them together.

Four measures, each with the rule it uses written down, because a study panel that will not say how it computed a number is asking to be believed rather than read.

Over time
Attempts bucketed by sitting (a stretch of practice with no two-hour gap in it) or by calendar week, in your own local time. Accuracy is correct out of attempted in that bucket; pace is the median response time, a median rather than a mean because one question you walked away from moves a mean by minutes and a median by nothing. A bucket you did not practise is absent rather than drawn as zero.
Where to work next
Topics ranked by first-try correctness with every topic starting from two imaginary right answers and two imaginary wrong ones, so that real evidence washes the prior out at the rate it arrives and a topic with nothing on it sits in the middle rather than at either end. None of 2 therefore ranks well above none of 20, which is the ordering anybody means when they say one of those is a problem and the other is a small sample. Under 5 attempts a topic is not weak at all, it is unmeasured, and it is listed separately under that name. The score itself is never shown: it decides an order, and an order is all the evidence supports.
Realized mix
Your attempts per topic against the number a run built to the real test's proportions would have handed you over the same total, so a topic you keep skipping shows up as a number rather than as a feeling. The target shares are estimates and the taxonomy says so.
Response time
Milliseconds from a question appearing to its answer being committed, stored on the attempt and never on a clock that can run out. The reference line is the real test's own pace, 2 min 35 sec a question. Fewer than 4 timed attempts and no median is reported at all.
Plate IV Accuracy and pace over time, where to work next, and the realized mix against the target.

What still needs an answer

The bank is being filled by retrieving questions first and pairing answers to them afterwards, so a question with no key on file is a normal state of this drawer rather than a fault. The drill shows those anyway: it puts the question up, lets you think, and asks you to mark how it went, so the attempt is still on the record and the item is still cataloged as one that needs a key. The table is the standing to-do list.

Of 780 questions, 779 have a checked answer, 1 have one that has not been checked, and 0 have none yet.

Every topic of the drill, with the questions the bank holds and how many of them have a checked answer, an unchecked answer, or no answer at all.
Topic Questions Checked Unchecked No answer
Limits and continuity Calculus 49 49
Derivatives Calculus 38 38
Applications of derivatives Calculus 37 37
Integrals Calculus 37 37
Applications of integrals Calculus 36 36
Sequences and series Calculus 25 25
Multivariable calculus Calculus 37 37
Multiple integrals Calculus 37 37
Vector calculus Calculus 38 38
Differential equations Calculus 39 39
Elementary algebra Algebra 32 32
Trigonometry Algebra 28 28
Linear algebra Algebra 40 40
Group theory Algebra 37 37
Rings and fields Algebra 34 34
Number theory Algebra 33 33
Real analysis Additional topics 27 27
Complex analysis Additional topics 18 18
Point-set topology Additional topics 31 31
Probability Additional topics 17 16 1
Statistics Additional topics 23 23
Combinatorics Additional topics 16 16
Graph theory Additional topics 17 17
Numerical analysis Additional topics 18 18
Algorithms Additional topics 17 17
Logic and set theory Additional topics 19 19
All topics 780 779 1 0
Plate V The drawer, counted at build time: what is in it and what it is still missing.

Where the questions come from

The bank is original work plus openly licensed sources, and no ETS material appears in it in any form: not reproduced, not re-typeset, not adapted, not paraphrased. The individual mathematical fact in a problem is nobody's property, but the wording of a stem, the choice and ordering of its distractors and the selection of what to test are, and a bank rebuilt out of a published practice book would be a derivative of that book's compilation however thoroughly each sentence was rewritten. "Adapted from ETS" names an infringement more precisely; it does not cure one.

What is here instead: questions I wrote, and questions built on sources whose licence permits it, each carrying its attribution under the solution where you can see it. Attribution the reader cannot see is not attribution. The test specification quoted at the top of this page is a fact about the test rather than the test, which is why it can be stated at all.

Everything on this page is under the site's usual terms: the code MIT, the writing CC BY-NC 4.0, and any borrowed item under whatever its own source says.

What this cannot do

Said plainly, because the alternative is a tool that quietly disappoints. Your laptop and your phone keep separate records and nothing merges them. Clearing site data erases the record, and so does a new browser profile, and so does Safari deciding it has not seen this site in a week. There are no aggregate statistics, because there is nowhere to compute them. And nothing can serve you a question that was not in the build, since adding a question is an edit to a repository rather than a row in a table.

All of that is the cost of there being no account and no server, and it is a price worth paying for a study tool with one serious reader.

Keys on this page

A figure's keys are its own: they answer once the figure has focus, so tab to it or click it first. Esc closes this sheet.