A calculator can return a beautifully precise decimal while answering a different mathematical question from the one a student intended. The interesting part of checking is therefore not pressing the same buttons again. It is finding another reason why the answer should have a particular sign, scale or range.
For an IB Mathematics: analysis and approaches exploration, check a numerical result against an independently justified bound, then investigate any disagreement. State the function, interval and assumptions, find what the result cannot exceed or fall below, and compare the numerical method with that restriction. An answer inside a broad range remains possible; it is not automatically established as correct.
This guide uses original calculus exercises for learners preparing for assessment in 2027 under the current AA course. It is not a completed exploration, a prescribed investigation or an assessment template. The aim is the independent judgement described in the Education parent guide: being able to explain why a result deserves confidence when the familiar worked example disappears.
Find your next learning step
Build independent checks for a numerical integral using bounds, exact calculations and changed conditions, with original IB Mathematics AA practice cases.
ROUTE 1 · CHAPTERS 1–3
Match the exploration and check
Identify the cohort, diagnose the difficulty and understand a bound.
ROUTE 2 · CHAPTERS 4–6
Build an independent comparison
Calculate a coarse bound, refine it and compare numerical methods.
ROUTE 3 · CHAPTERS 7–9
Interpret the mathematical result
Use exact checks, units and signed quantities carefully.
ROUTE 4 · CHAPTERS 10–12
Repair and transfer the reasoning
Inspect input errors, preserve authorship and change assumptions.
ROUTE 5 · CHAPTERS 13–15
Try the next independent check
Attempt fresh cases, resolve questions and consult primary sources.
Full chapter index · Independent learning check · Education parent guide
Full chapter index
Match the exploration and check · Chapters 1–3
Build an independent comparison · Chapters 4–6
Interpret the mathematical result · Chapters 7–9
Repair and transfer the reasoning · Chapters 10–12
Try the next independent check · Chapters 13–15
CHAPTER 1 OF 15 · Match the exploration and check
1. Match the exploration to the correct course
Begin with the exact course and assessment session. AA and AI are different IB Mathematics courses, each available at SL and HL. An article containing the letters IB does not establish which course its examples or assessment information concern. Keep the school-issued requirements for your own level and session beside your planning.
The IB's current Mathematics page distinguishes the courses whose final assessment is November 2028 from the revised courses taught from 2027 for first assessment in 2029. A learner assessed in 2027 should not import the revised 2029 exploration criteria simply because a newer web page appears first in a search. Teaching year and first assessment year are different dates.
The current AA subject brief describes mathematical argument, critical reflection and an internally assessed exploration. It places the exploration at 20% of the final grade at both levels. That context supports careful mathematical investigation, but it does not turn the exercise below into a required topic or promise that using numerical integration earns particular marks.
For the broader topic-selection and cohort distinction, use Bukit Timah Tutor's IB Mathematics exploration guide. Preserve the student's responsibility for choosing and developing the actual investigation. A useful exercise can teach a technique without deciding the student's assessed question, argument or conclusion.
Our narrow job is checking. We will compare several ways of understanding a defined integral and ask what each comparison establishes. The calculations are deliberately manageable so that attention can stay on the reasoning connecting them. A more complicated function is not automatically a more thoughtful investigation.
CHAPTER 2 OF 15 · Match the exploration and check
2. Diagnose which checking decision is missing
Imagine a fictional learner, Naomi, who obtains 46.667 for a small integral and accepts it because the calculator display contains several decimal places. Asking her to repeat the entry may reveal a typing error, but it does not yet show whether she understands the quantity being calculated. First ask what size of answer she expected.
If she cannot identify the interval, return to the question. If she knows the interval but cannot describe the function's values there, inspect its behaviour. If she can establish a sensible range but ignores it when using technology, the difficulty concerns connecting mathematical reasoning with the output. These are different next lessons.
Another learner, Amir, knows that an answer should be positive and less than ten. He obtains 4.8 and declares it correct. His check has value: it excludes negative values and very large mistakes. It does not distinguish 4.8 from several other plausible answers. He needs a tighter comparison or a genuinely different calculation.
Listen to the explanation before deciding that either learner is careless. Can the student state what the result represents? Can they identify which assumption makes the bound valid? Can they explain what would change if the function decreased instead of increased? A specific response gives you a repairable mathematical difficulty.
Bukit Timah Tutor's guide to calculator input and independent checking develops the broader habit. Here we extend it into an exploration decision: choose a check capable of detecting the particular mistake or uncertainty that matters, then describe its limits honestly.
CHAPTER 3 OF 15 · Match the exploration and check
3. Understand what a bound actually says
A bound restricts the possible value of a quantity. Suppose a continuous function stays between a minimum m and a maximum M throughout an interval from a to b, with a less than b. Its definite integral lies between m multiplied by the interval width and M multiplied by that width. The restriction concerns every point in the interval.
The reason is accessible. A function that never rises above M cannot accumulate more signed area than the constant function M over the same interval. A corresponding comparison holds for m below it. We are comparing entire functions, not merely noticing that several plotted points happen to fit inside a rectangle.
This distinction becomes important with measured data. Five observations below a ceiling do not, by themselves, prove that an unknown continuous process stayed below it between measurements. For a fully defined mathematical function, we can investigate the behaviour between points. For a model fitted to observations, the conclusion additionally depends on the model and its assumptions.
There is also a difference between a permitted range and an answer. If the integral lies between two and ten, then four, five and nine all survive that initial check. Saying that a value is consistent with a bound is a smaller claim than saying that it is the correct value. Mathematical communication should preserve that difference.
Bounds need not be complicated to be useful. A coarse bound may expose a misplaced decimal immediately. Later, a refined bound may distinguish methods or establish a requested tolerance. Choose the strength of the check in response to the question, rather than adding increasingly elaborate calculations without a clear purpose.
CHAPTER 4 OF 15 · Build an independent comparison
4. Build a coarse check for a defined integral
Consider our original example: f(x) = x² + 1 on the interval from zero to two. Let I represent its definite integral over that interval. The function is increasing there, begins at one and ends at five. Its values therefore stay between one and five throughout the interval.
The interval has width two. Multiplying that width by the minimum and maximum gives 2 ≤ I ≤ 10. This is a broad restriction, but it already tells Naomi that 46.667 cannot be the requested integral. She has a reason to inspect her expression before discussing decimal places or rounding.
Notice what has been checked. We used the function's behaviour and the interval width, not the calculator's integration routine. Entering the same incorrect expression into another device could reproduce a wrong result. An independently justified restriction is valuable because its reasoning is less likely to share the same input mistake.
Amir's value of 4.8 passes this coarse test. We should say exactly that. It lies in the allowed interval, so the broad comparison does not reject it. We still need more information to determine whether it equals, approximates or incorrectly represents the integral. A successful first check is the beginning of verification.
Now change the interval in your imagination to zero through one. The width and maximum both change. Keeping the earlier bound would answer an earlier question. Before using any saved result, make sure that the expression, interval and variable definitions still describe the task now in front of you.
CHAPTER 5 OF 15 · Build an independent comparison
5. Tighten the range using rectangles
Divide zero through two into four equal subintervals. Each has width 0.5. Because our function increases, its value at the left endpoint of each subinterval is a lower height for that entire strip. Its value at the right endpoint is an upper height. Multiplying heights by widths and adding creates a lower and an upper sum.
The left endpoint heights are 1, 1.25, 2 and 3.25. Their sum is 7.5, giving a lower sum of 3.75 after multiplication by 0.5. The right endpoint heights are 1.25, 2, 3.25 and 5. Their sum is 11.5, giving an upper sum of 5.75. Consequently, 3.75 ≤ I ≤ 5.75.
| Comparison | Lower value | Upper value | What it establishes |
|---|---|---|---|
| One interval, minimum and maximum heights | 2 | 10 | A coarse restriction on the defined integral |
| Four equal strips, increasing function | 3.75 | 5.75 | A narrower enclosing interval |
| Eight equal strips, increasing function | 4.1875 | 5.1875 | A further refinement using the same justified relationship |
| Exact antiderivative calculation | 14/3 | 14/3 | The exact value for this particular polynomial |
The table compares our own calculations. It is not an IB marking grid. Each row answers a different checking question, and the numerical bounds depend on the function and interval stated above. The row for eight strips cannot be transferred unchanged to another function.
Why does refinement help here? Each strip covers a smaller horizontal distance, so the increasing function varies less between its endpoints. The enclosing rectangles fit more closely. That explanation matters more than the slogan that more strips are better, because it identifies the behaviour supporting the improvement.
For this example, the difference between the right and left sums is the strip width multiplied by f(2) minus f(0). With n equal strips, that gap is 8/n. The intermediate endpoint terms cancel when the sums are subtracted. At n = 80, the enclosing interval has width 0.1. This bounds the bracket's width; it is not automatically a statement about correct decimal places in a chosen rounded output.
Keep each strip's width attached to its height
Equal strips make a calculation convenient because every rectangle has the same width. The comparison principle does not require equal strips, but an unequal subdivision needs the actual width of each strip. Losing that distinction can produce a tidy calculation that no longer represents the rectangles drawn.
For the same quadratic, split the interval at 0.5, giving widths 0.5 and 1.5. The lower sum is now 0.5 multiplied by 1, plus 1.5 multiplied by 1.25, which equals 2.375. The upper sum is 0.5 multiplied by 1.25, plus 1.5 multiplied by 5, which equals 8.125. These are valid but relatively loose bounds.
Why are they less informative than four equal strips? The long second strip covers a part of the function with substantial variation. Its two bounding rectangles leave considerable room between them. The issue is not that unequal strips are forbidden; it is how the chosen subdivision interacts with the function's behaviour.
A useful learner check is to draw each strip and write its width underneath. Then state which function value supplies its height and why. The calculation should follow that representation. Reusing one common width after changing the subdivision breaks the connection between the sum and the diagram.
CHAPTER 6 OF 15 · Build an independent comparison
6. Compare a numerical method with the bound
For four equal strips, the composite trapezoidal estimate is the average of the left and right sums: (3.75 + 5.75) divided by two, which gives 4.75. It lies inside the enclosing interval, as that construction suggests. We have a numerical estimate alongside a separately justified statement about where the exact integral lies.
A midpoint estimate uses the function at the centres of the four strips. At x values 0.25, 0.75, 1.25 and 1.75, the heights are 1.0625, 1.5625, 2.5625 and 4.0625. Their sum, multiplied by 0.5, gives 4.625. The two methods have produced different approximations without either becoming a different exact integral.
Do not choose the more impressive-looking decimal as the better result. Ask how each estimate was constructed and what is known about its error. For our convex quadratic, the trapezoids sit above the graph, while midpoint rectangles give the smaller result calculated here. Later exact comparison will quantify the difference for this function.
NIST's Digital Library of Mathematical Functions section on quadrature provides formal trapezoidal rules and error terms with stated conditions. Those conditions are part of the mathematics. You should not import a formula from a reference and apply it without understanding the differentiability or other assumptions it requires.
The exploration opportunity is to explain relationships. Why do these two methods disagree? What evidence would justify trusting one to a specified accuracy? What changes when the function changes? A useful investigation grows from such questions, while the actual selection and development remain the student's own work.
Separate an enclosing interval from an error guarantee
Suppose an unknown exact value lies between L and U. The centre of that interval is no farther than half its width from any value inside it. Therefore, choosing the centre provides an absolute-error guarantee of at most (U − L)/2, before any additional rounding. This is a logical consequence of enclosure.
For our equal-strip left and right sums, their centre is the trapezoidal estimate. Since the gap is 8/n, this particular enclosure guarantees an error of at most 4/n. At eighty strips, the guarantee is 0.05. That is different from saying that the displayed result has eighty-strip accuracy without defining an error measure.
The guarantee may be conservative. The actual error for the quadratic can be much smaller than the largest error allowed by the enclosing interval. A loose but valid restriction and a precise actual-error calculation serve different purposes. Comparing them can help a student understand why an estimate and a proven tolerance are different claims.
When a task specifies a tolerance, ask whether the chosen evidence establishes it. If the bound is too wide, refinement or another justified argument may help. Do not replace that question with the number of decimal places a device happens to show.
CHAPTER 7 OF 15 · Interpret the mathematical result
7. Use an exact result as another kind of check
For this polynomial we can calculate an antiderivative: x³/3 + x. Evaluating it at two and subtracting its value at zero gives 8/3 + 2 = 14/3. The exact integral is therefore approximately 4.666667. It fits all the bounds already established.
The trapezoidal estimate of 4.75 exceeds the exact value by 1/12. The midpoint estimate of 4.625 is below it by 1/24. These are exact error comparisons for the defined exercise. They do not establish that the midpoint rule always has half the error of the trapezoidal rule for every possible integrand.
We can now assess Amir's 4.8 more carefully. It passed a coarse bound, but it is not the exact value. Whether it would be an adequate approximation depends on the required accuracy and the method or instructions governing the task. The number of digits displayed does not settle that question.
A useful written explanation separates the exact quantity, the approximation and the rounded presentation. For example, 14/3 is exact here, 4.75 is a four-strip trapezoidal estimate, and 4.67 is the exact value rounded to two decimal places. Calling all three simply the answer conceals information the reader needs.
Not every exploration produces an elementary antiderivative. That does not make checking impossible. Bounds, convergence comparisons, known special cases, symmetry and other mathematically justified relationships can still be informative. The purpose of this easy polynomial is to make those roles visible before the calculations become less transparent.
CHAPTER 8 OF 15 · Interpret the mathematical result
8. Keep the units and the model in view
An integral combines a function's units with the units of its independent variable. If a deliberately defined rate is measured in litres per minute and time is in minutes, its accumulated quantity is in litres. A bare decimal loses part of its meaning when those definitions disappear.
Suppose a student changes a time axis from minutes to seconds while leaving the rate values and formula unchanged. The graph may look familiar, yet the mathematical representation no longer describes the same quantity. A unit check can expose the inconsistency even if a numerical routine executes perfectly.
Distinguish a mathematical definition from an empirical claim. Our f(x) is exactly the polynomial we stated. If someone uses that polynomial to model a real flow, temperature or environmental process, they must justify the connection separately. An exact integral of a model is not an exact measurement of the world.
The Punggol guide to IB Mathematics AA, models and the assessment-year distinction develops the wider modelling context. Here the numerical bounds belong to a defined function. Applying them to observations requires a defensible model, suitable units and an appropriate interval.
A strong explanation can therefore contain two different uncertainties. There may be numerical error in approximating an integral, and there may be model error in representing a real process. Reducing the numerical error does not automatically repair the model. Keep each question attached to the evidence capable of answering it.
CHAPTER 9 OF 15 · Interpret the mathematical result
9. Reconsider the check when values become negative
Now define g(x) = x − 1 on zero through two. Its definite integral is zero: the signed contribution below the horizontal axis cancels the equal contribution above it. The total unsigned area between the graph and the axis is one. These quantities answer different questions.
A learner who expects every integral to be a positive geometric area may reject the correct zero. The repair is not to memorise that this particular answer is zero. It is to examine whether the task concerns signed accumulation or an unsigned total, and how the chosen representation expresses that distinction.
The minimum and maximum comparison still works for signed integrals. Here g stays between minus one and one over an interval of width two, giving −2 ≤ the integral ≤ 2. The broad bound permits zero, although it does not establish the cancellation by itself. Symmetry or an antiderivative provides another argument.
Consider another changed function, 1 − x² on zero through two. It decreases from one to minus three. Left endpoint rectangles now provide the upper sums, not the lower sums used for our increasing function. The relationship changed because a supporting assumption changed.
This is a useful transfer test. Can the learner explain which rectangles are above or below the graph without relying on a remembered label? If they can, they have kept the mathematical comparison. If they can only repeat that left means lower, return to the function's behaviour within each strip.
CHAPTER 10 OF 15 · Repair and transfer the reasoning
10. Investigate disagreement without guessing at the cause
When an output violates a justified bound, inspect the mathematical expression before changing the result. Check the function, parentheses, limits, variable and units. Confirm that the operation was integration over the intended interval, rather than evaluation at an endpoint or calculation of a different quantity.
Write down the expression you intended and compare it with the expression actually entered. A missing pair of parentheses can change the order of operations. An incorrect upper limit changes the domain. Rounding a parameter prematurely can change the calculation. Each has a different repair, so a general instruction to be more careful is insufficient.
If two methods disagree but both satisfy a coarse bound, refine the evidence. Compare a smaller step size, an exact special case or another valid relationship. Repeating an identical procedure with the same settings may confirm reproducibility while leaving the original mathematical error untouched.
Sengkang's distinction between what was observed, inferred and remains unknown is helpful. We observed a disagreement. We may infer an input error, but we should inspect the entry before asserting that explanation. The mathematical check tells us something is inconsistent; it does not always identify the cause.
Record a correction with its reason. Instead of writing that the calculator was wrong, explain which expression or assumption changed and why the corrected calculation answers the intended question. That record makes the learning usable when a later discrepancy has a different cause.
CHAPTER 11 OF 15 · Repair and transfer the reasoning
11. Practise independence without taking over the exploration
A teacher can begin with a short question: what must be true about the answer before you calculate it? Allow time for a diagram, an inequality or a verbal explanation. The point is to make the student's expectation visible, not to reward the first person who remembers a formal error formula.
Parents can ask the learner to explain one restriction in ordinary language. Why cannot the integral exceed the area of that rectangle? What does a negative answer mean? Which assumption makes the comparison work? They do not need to design the assessed investigation or supply polished mathematical writing.
A tutor can use unrelated functions to teach bounds, approximation and interpretation. The Singapore guide to choosing IB Mathematics support discusses the broader authorship boundary. Guidance should build the learner's mathematical capability while preserving their responsibility for the assessed exploration.
Reduce prompts over time. Initially, the teacher may point to monotonicity. Later, ask the student to identify whether monotonicity is relevant. Finally, present a different function and ask them to choose a useful check independently. The change in responsibility is a more informative learning signal than repeated success on the original example.
Keep practice manageable. One carefully analysed discrepancy can expose more than many outputs copied from a device. End the session by asking what the student would inspect next time and why. A useful plan names a mathematical decision, rather than promising that no future mistakes will occur.
CHAPTER 12 OF 15 · Repair and transfer the reasoning
12. Change a parameter and test the reasoning
Replace our function by f(x) = a x² + 1 on zero through two, initially with a nonnegative. The exact integral becomes 8a/3 + 2. When a is zero, the function is the constant one, and every correctly constructed rectangle or trapezoid calculation gives two.
That constant case is a useful diagnostic. A routine producing another value deserves inspection. The check does not require the student to trust a complicated output: it relies on the familiar accumulation of a constant over an interval of width two. Special cases can provide powerful independent evidence.
When a is positive, the function increases. Its minimum is one and its maximum is 4a + 1, giving the coarse bound 2 ≤ I ≤ 8a + 2. The quantities change with a, while the comparison principle remains the same. Explain that continuity before calculating several examples.
If a becomes negative, the function decreases and the endpoint roles reverse. Do not retain the earlier interpretation of left and right sums automatically. The useful transfer is recognising the changed condition and rebuilding the appropriate comparison, not inserting a negative number into a sentence whose assumptions no longer hold.
An investigation can also change its purpose. A broad range may be enough to reject an impossible result, while a tolerance question needs a tighter bound. Ask what the new question requires before choosing the next method. More precision is useful when it addresses a real mathematical need.
The AP Computer Science A lesson on ArrayList removal makes another type of check visible: a loop preserves a statement about processed and unprocessed work. Compare that with a numerical bound, which restricts possible values. Both demand a reason the check holds, while each qualification retains its own requirements.
CHAPTER 13 OF 15 · Try the next independent check
13. Try a fresh independent check
Use the original practice function h(x) = 2x + 3 on zero through one. Before integrating, identify its smallest and largest values on the interval. Explain why the integral must lie between three and five. Then calculate the exact result and decide whether your first restriction was informative enough.
The antiderivative is x² + 3x, giving four over the stated interval. A result of forty cannot satisfy the bound. A result of 4.1 does satisfy it but is not the exact integral. Explain that distinction without referring back to Naomi's or Amir's case.
Next consider h on zero through two. Both the interval width and the maximum change. The coarse bound is now six through fourteen, and the exact integral is ten. A learner who keeps the earlier three-through-five restriction has reused the answer without preserving its mathematical conditions.
Now suppose only three measured values of an unknown process are supplied. They happen to equal h at zero, one half and one. Can you conclude that the process is the straight line between them? No. A linear model might be proposed, but those observations alone do not prove the behaviour at every intervening point.
Finish by choosing a different check for a different error. How would you notice a unit mismatch? How would you investigate a sign error? What would help if a plausible decimal still differed from an exact calculation? The ability to select an appropriate check is the capability this article is trying to develop.
Explain a plausible wrong answer
In the fresh linear example, five is the value of h at the upper endpoint one. It also happens to lie at the top of the coarse interval. That makes it a useful diagnostic wrong answer: a learner may have evaluated the function at an endpoint instead of integrating it.
Ask what operation would produce five and what operation the question requests. The two outputs represent different mathematical quantities even though both can be calculated correctly. This is why checking the question and the units is essential before deciding that a plausible number is an acceptable answer.
Then change the function while keeping the interval. Can the learner identify the endpoint value, an integral bound and the exact integral as three separate objects? The exercise tests a specific distinction, rather than asking them to repeat the original correction from memory.
CHAPTER 14 OF 15 · Try the next independent check
14. Questions learners and parents often ask
Does passing a bound prove the answer is right?
No. It shows consistency with that restriction. A wide interval may contain many incorrect values. Use a tighter bound, an exact calculation or another justified comparison when the task requires stronger evidence. Explain the strength of the conclusion you have actually established.
Is numerical integration less mathematical than an exact method?
It is a mathematical method with its own construction, assumptions and error questions. Its value depends on how well it answers the problem. An unexplained exact expression can be poorly understood, while a carefully analysed approximation can involve substantial reasoning.
Must an exploration use all the techniques shown here?
No. These are independent teaching exercises. The actual exploration should develop a coherent question using suitable mathematics and the requirements for the student's course and assessment session. Adding every available technique can obscure the investigation's purpose.
Can a small numerical error establish that the real-world model is accurate?
No. It concerns the numerical treatment of the stated model. The connection between that model and observations requires separate evidence. Distinguish numerical approximation, measurement uncertainty and the limitations of the representation you have chosen.
Are left rectangles always lower bounds?
No. The comparison depends on the function's behaviour. For an increasing function they are lower; for a decreasing function they are upper. A function that changes direction needs further analysis or subdivision before the same statement can be justified.
Should a 2027 candidate use the new 2029 criteria?
Use the requirements for the actual assessment session. The IB distinguishes the current courses ending assessment in November 2028 from first assessment of the revised courses in 2029. Confirm the applicable guide through school rather than mixing two cohorts' expectations.
May I submit these calculations as my exploration?
They are published teaching examples, not your independent investigation. Learn the reasoning, acknowledge sources where relevant and develop your own assessed work within school and IB rules. You should be able to explain every mathematical decision you submit.
CHAPTER 15 OF 15 · Try the next independent check
15. Sources and the next mathematical question
Primary references checked on 24 October 2026 are the IB's current Mathematics curriculum page, the AA subject brief for the course first assessed in 2021, and NIST's quadrature reference. The IB page identifies the transition to first assessment of revised courses in 2029. Follow your school's full current requirements for your own session.
Naomi and Amir are fictional learners. All functions, calculations, comparisons and practice tasks here are original teaching examples. They are not official IB questions, an assessment rubric or evidence of measured student outcomes. The exact results concern the stated functions and intervals; real-world applications require additional modelling evidence.
For the next practice, choose one numerical result and one independent restriction. Explain why the restriction holds, what it excludes and what it leaves unresolved. Then change a condition. When the learner can rebuild the check and explain its limits, the calculator display becomes evidence to interpret rather than a conclusion to accept automatically.