Free Cheatsheet · AI HL · SL 5.5 · AHL 5.11–5.13

IB Math AI HL Integration — Complete Cheatsheet

Every integration skill for IB Mathematics Applications & Interpretation HL — the reverse power rule and constant of integration, definite integrals, area under a curve, the trapezoidal rule and kinematics — with each exam trap flagged. Hand-built by an IBO-certified Singapore tutor, with a print-ready PDF to download.

Topic: Integration (Calculus) Syllabus: SL 5.5 · AHL 5.11–5.13 Read time: ~11 minutes Last updated: Jul 2026

Integration is differentiation run in reverse, and in the IB Mathematics Applications & Interpretation HL calculus strand it is the tool of accumulation: given a rate — a velocity, a marginal cost, a flow into a reservoir — integrating it recovers the total built up. AI treats it as a modelling skill rather than an exercise in algebra: a GDC will evaluate almost any definite integral for you, so the marks live in setting up the right integral, choosing between an exact area and a trapezoidal estimate, and reading the answer back into the context with the correct units.

This cheatsheet condenses the whole range — from the reverse power rule and the constant of integration in SL 5.5 to areas, volumes of revolution and kinematics in AHL 5.11–5.13 — onto one page, and flags the traps that quietly cost method marks. The print-ready PDF is at the bottom, free to download.

§1 — Integration as anti-differentiation SL 5.5

To integrate is to undo differentiation: you look for a function whose derivative is the one in front of you. Because the derivative of a constant is zero, an indefinite integral always carries an unknown constant of integration $+C$.

Reverse power rule:$\displaystyle\int ax^{n}\,dx=\frac{a}{n+1}x^{n+1}+C\quad(n\neq-1)$ — raise the power, then divide by the new power
Sum rule:integrate a polynomial term by term
Constant term:$\displaystyle\int a\,dx=ax+C$ — a plain number integrates to $ax$

For example $\displaystyle\int\left(6x^{2}-4x+5\right)dx=2x^{3}-2x^{2}+5x+C$: each power goes up by one and is divided by that new power. Integration measures accumulation — the total built up from a rate of change — which is why every application on this sheet begins by integrating a rate.

TrickIntegration and differentiation are inverses, so check any integral by differentiating it — if you don't get back the original integrand, you have slipped somewhere.
TrapDivide by the new power $n+1$, never the old one, and never drop the $+C$ on an indefinite integral. The rule fails only at $n=-1$, where $\int x^{-1}\,dx=\ln|x|+C$ instead (see §7).

§2 — The definite integral SL 5.5

Put limits on an integral and it stops being a function and becomes a number. Evaluate any antiderivative $F$ at the top limit and subtract its value at the bottom — the $+C$ cancels, so you never need it here.

Definite integral:$\displaystyle\int_{a}^{b}f(x)\,dx=F(b)-F(a)$, where $F'(x)=f(x)$
Polynomial:$\displaystyle\int_{a}^{b}\!\left(px^{2}+qx+r\right)dx=\left[\frac{p}{3}x^{3}+\frac{q}{2}x^{2}+rx\right]_{a}^{b}$

So $\displaystyle\int_{1}^{3}6x^{2}\,dx=\left[2x^{3}\right]_{1}^{3}=2(27)-2(1)=52$. Work out the antiderivative first, wrap it in square brackets with the limits, then substitute the upper limit and subtract the lower-limit value.

NoteOn Paper 2 the GDC evaluates $\int_{a}^{b}f\,dx$ directly, so use it to check — but still write down the integral you typed in, because the setup is what earns the method mark.
TrapKeep the polynomial terms lined up: the classic slip is forgetting to divide the $x^{2}$ term by $3$ or the $x$ term by $2$. Substitute the upper limit first, then subtract the lower — reversing them flips the sign of the answer.

§3 — The constant of integration: boundary conditions SL 5.5

An indefinite integral leaves $C$ unknown, so it really describes a whole family of parallel curves. One known point — a boundary condition — pins down exactly which curve you have.

Method:integrate $\dfrac{dy}{dx}$ to get $y=F(x)+C$, substitute the given point to solve for $C$, then evaluate $y$ where required

If $\dfrac{dy}{dx}=4x-3$ and the curve passes through $(2,\,5)$, then $y=2x^{2}-3x+C$; substituting gives $5=2(4)-3(2)+C=2+C$, so $C=3$ and $y=2x^{2}-3x+3$. This is how a marginal rate plus a single observed reading rebuilds an entire model — in economics that constant is the fixed cost, $C(0)$ (see §9).

TrapYou cannot skip $C$. Integrate first, use the given point to solve for $C$, and only then substitute the target value of $x$. Substituting the target before you have found $C$ defeats the whole method.

§4 — Area under a curve SL 5.5

The definite integral has a picture. When $f(x)\ge 0$ across $[a,b]$, $\displaystyle\int_{a}^{b}f(x)\,dx$ is exactly the area between the curve $y=f(x)$ and the $x$-axis — the meaning behind cross-sections, land areas and glass panels.

Area (above axis):$\displaystyle A=\int_{a}^{b}f(x)\,dx\quad\text{for } f(x)\ge 0 \text{ on } [a,b]$
On the GDC:graph $y=f(x)$, then use the f(x)dx (area/integral) tool between $x=a$ and $x=b$

Because AI always allows a calculator, the fastest route on Paper 2 is to graph the model and let the GDC return the area — but you must still state the integral you are evaluating to secure the method mark.

TrapArea needs integration, not substitution: the height $f(b)$ is never the area. Integrate first. The limits $a$ and $b$ are the $x$-values that bound the region — often the roots where the curve crosses the axis.
NoteFor a region trapped between two curves, the area is $\displaystyle\int_{a}^{b}\big(\text{top}-\text{bottom}\big)dx$ — subtract the lower function from the upper, then integrate across the interval where they enclose the region.

§5 — The trapezoidal rule SL 5.5

When you only have a table of readings — not a formula — you cannot integrate, so you estimate the area by slicing it into thin trapezoidal strips. This is the single most common integration task in AI, and it needs no calculus at all.

Trapezoidal rule:$\displaystyle\int_{a}^{b}\! y\,dx\approx\frac{h}{2}\Big(y_{0}+y_{n}+2\big(y_{1}+y_{2}+\dots+y_{n-1}\big)\Big)$
Strip width:$h=\dfrac{b-a}{n}$ — the equal spacing between readings ($n$ strips, $n+1$ ordinates)

The two end ordinates are counted once; every interior ordinate is doubled. With $h=2$ the front factor $\frac{h}{2}$ is simply $1$.

x y O y₀ y₁ y₂ y₃ y₄ y = f(x) h
The shaded region — the area $\int_a^b f(x)\,dx$ — approximated by four trapezoidal strips of equal width $h$. For this concave-up curve the straight tops (gold) sit above the curve, so the trapezoidal rule over-estimates the true area.
TrapOnly $y_{0}$ and $y_{n}$ are counted once — every value in between is doubled. Miscounting the interior ordinates, or using the number of readings as $n$ instead of the number of strips, are the usual errors.
NoteWhether the estimate is too big or too small depends on the curvature: where the graph is concave up (bending upwards) the straight strip-tops lie above the curve, so the rule over-estimates; where it is concave down the rule under-estimates.

§6 — Signed area & area below the axis AHL 5.12

A definite integral returns a signed area: any part of the region below the $x$-axis, where $f(x)<0$, counts as negative. So the raw integral gives the net area, which can differ from the total physical area.

Net (signed) area:$\displaystyle\int_{a}^{b}f(x)\,dx$ — parts below the axis subtract
Total area:$\displaystyle\int_{a}^{b}\big|f(x)\big|\,dx$ — split at the roots, or integrate the modulus on the GDC

If the curve crosses the axis inside $[a,b]$, find the roots, integrate over each piece separately, and add the absolute values — otherwise the positive and negative parts partly cancel and you understate the area.

TrapIf a question asks for the area and the curve dips below the axis, do not simply integrate across the whole interval — that gives the signed (net) value. Split at the roots and add the magnitudes, or evaluate $\int|f(x)|\,dx$ on the GDC.

§7 — Reverse chain rule & further integrals AHL 5.11

AHL widens the toolkit beyond powers. Two standard results come straight from reversing the derivatives in Differentiation, and one technique undoes the chain rule.

IntegralResult
$\displaystyle\int e^{x}\,dx$$e^{x}+C$ — unchanged
$\displaystyle\int \frac{1}{x}\,dx$$\ln|x|+C$ — the $n=-1$ case the power rule cannot reach
$\displaystyle\int \sin x\,dx$$-\cos x+C$
$\displaystyle\int \cos x\,dx$$\sin x+C$

When an integrand is a function raised to a power, multiplied by a multiple of the derivative of its inside, the reverse chain rule (substitution $u=g(x)$) undoes it:

Reverse chain rule:$\displaystyle\int g'(x)\big(g(x)\big)^{n}\,dx=\frac{\big(g(x)\big)^{n+1}}{n+1}+C$

For instance $\dfrac{d}{dx}(x^{2}+1)=2x$, exactly the factor out front, so $\displaystyle\int 2x\,(x^{2}+1)^{3}\,dx=\frac{(x^{2}+1)^{4}}{4}+C$.

TrapBefore substituting, check the factor in front is a multiple of the derivative of the inside bracket. If it is out by a constant you can adjust for it; if it is the wrong power of $x$ altogether, the reverse chain rule does not apply.

§8 — Volumes of revolution AHL 5.12

Spin the region under $y=f(x)$ right around the $x$-axis and it sweeps out a solid of revolution. Picture slicing it into thin discs, each a circle of radius $y$ and area $\pi y^{2}$; stacking them gives the volume.

About the $x$-axis:$\displaystyle V=\pi\int_{a}^{b} y^{2}\,dx$

For $y=\sqrt{kx}$ the square is simply $y^{2}=kx$, which turns the integral into a plain power. The two marks students lose here are forgetting to square $y$ and forgetting the factor $\pi$.

TrapSquare $y$ before integrating, and keep the $\pi$ outside the whole integral. $\pi\int y^{2}\,dx$ is not the same as $\left(\pi\int y\,dx\right)^{2}$, nor may you drop the $\pi$.

§9 — Kinematics & rate-to-total applications AHL 5.13

This is where AI integration pays off: whenever you know a rate, integrating it over an interval gives the total change. The headline case is motion — velocity is the derivative of displacement, so displacement is the integral of velocity.

Displacement:$\displaystyle s=\int_{t_{1}}^{t_{2}} v\,dt$ — the signed area under the $v$–$t$ graph
Distance travelled:$\displaystyle \int_{t_{1}}^{t_{2}}\big|v\big|\,dt$ — split where $v=0$ if the object reverses direction
Velocity from acceleration:$\displaystyle v=\int a\,dt$ — integrate acceleration, then use a known velocity to fix $C$

The same "rate → total" logic runs through the other AI applications. In economics the marginal cost $C'(x)$ integrates to the total variable cost, with the constant of integration equal to the fixed set-up cost, $C(0)$. And when the rate arrives as a data table rather than a formula, first fit a model — a quadratic regression $R(t)=at^{2}+bt+c$ on the GDC — then integrate it.

TrickThe AI synthesis question is two clean steps: run the regression to turn the table into $R(t)$, then the total over $[t_{1},t_{2}]$ is $\int_{t_{1}}^{t_{2}} R\,dt$ — never the single reading $R(t_{2})$.
TrapDisplacement is a signed integral of $v$: if the object turns around, integrate $|v|$ (splitting where $v=0$) for the total distance. And never just substitute $t$ into $v$ — a velocity value is not a displacement.

§10 — Exam attack plan All sections

Question cueWhat to doWatch for
"Find $\int f(x)\,dx$" (indefinite)Reverse power rule term by term; add $+C$Divide by the new power; never drop $+C$
"Evaluate $\int_a^b f\,dx$"Antiderivative in brackets, upper $-$ lower (or GDC)$+C$ cancels; substitute the upper limit first
"Curve through a point, given $\frac{dy}{dx}$"Integrate, sub the point to find $C$, then evaluateSolve for $C$ before the target $x$
"Area between the curve and the $x$-axis"$\int_a^b f\,dx$; if it dips below, use $\int|f|\,dx$Integrate, don't substitute; split at the roots
"Estimate the area from a table"Trapezoidal rule $\frac{h}{2}\big(y_0+y_n+2\sum\text{interior}\big)$Interior ordinates doubled; concavity sets over/under
"Volume when rotated about the $x$-axis"$V=\pi\int_a^b y^2\,dx$Square $y$ first; keep the $\pi$ outside
"Displacement / distance from velocity"$s=\int v\,dt$; distance $=\int|v|\,dt$Signed vs total; split where $v=0$
"Total from a rate or marginal"Fit a model if given data, then integrate the rateFixed cost $=C(0)$; not the final reading

Worked Example — Displacement, a Boundary Condition & the Trapezoidal Rule

Question (AI HL Paper 2 style — 9 marks)

A drone rises so that its vertical velocity is modelled by $v(t)=3t^{2}-6t+4$ (in m s$^{-1}$), for $0\le t\le 4$ seconds. At $t=0$ the drone is at a height of $s=5$ m.
(a) Find the drone's displacement during the first $4$ seconds.
(b) Find an expression for its height $s(t)$, and hence its height when $t=4$.
(c) The onboard sensor logs the velocity once per second, giving $v=4,\,1,\,4,\,13,\,28$ at $t=0,1,2,3,4$. Use the trapezoidal rule to estimate the displacement, and state, with a reason, whether it over- or under-estimates the true value.

Solution

  1. (a) Displacement is the integral of velocity: $s=\displaystyle\int_{0}^{4}\left(3t^{2}-6t+4\right)dt=\left[t^{3}-3t^{2}+4t\right]_{0}^{4}$. (M1)(A1) Evaluating, $=\left(64-48+16\right)-0=32$ m. (A1)
  2. (b) Integrating with a constant, $s(t)=t^{3}-3t^{2}+4t+C$. The boundary condition $s(0)=5$ gives $C=5$. (M1) So $s(t)=t^{3}-3t^{2}+4t+5$ (A1), and $s(4)=64-48+16+5=37$ m. (A1) (Check: the start height $5$ plus the displacement $32$ gives $37$.)
  3. (c) With $h=1$: $\;s\approx\dfrac{1}{2}\big(4+28+2(1+4+13)\big)=\dfrac{1}{2}(32+36)=34$ m. (M1)(A1) Since $v''(t)=6>0$ the velocity model is concave up, so the straight strip-tops lie above the curve and the rule over-estimates (true value $32$ m). (R1)

Examiner's note: in AI you would normally get part (a) straight from the GDC's $\int$ tool — but write the integral down first, as that earns the method mark. Keep the units (m) and give $3$ s.f. unless told otherwise. Note the wording: this $v$ stays positive, so displacement and distance agree here; if the model dipped below zero you would integrate $|v|$ for the total distance. The trapezoidal estimate ($34$) sitting just above the exact area ($32$) is exactly what a concave-up curve predicts.

Common Student Questions

Do I add $+C$ to a definite integral?
No. The $+C$ only belongs on an indefinite integral, where the answer is a family of functions. In a definite integral $\int_a^b f\,dx=F(b)-F(a)$ the constant appears in both $F(b)$ and $F(a)$ and cancels, so it never affects the number. Add $+C$ only when there are no limits — for example when you are finding a curve from its gradient and need a boundary condition to pin the constant down.
How do I know whether the trapezoidal rule over- or under-estimates?
Look at the curvature of the graph over the interval. Where the curve is concave up (bending upwards, like exponential growth) the straight tops of the trapezoids sit above the curve, so the rule over-estimates. Where it is concave down (bending downwards, like the top of a hill) the tops fall below the curve, so it under-estimates. Using more strips (a smaller $h$) always brings the estimate closer to the true value.
What is the difference between displacement and distance travelled?
Displacement is the signed integral $\int v\,dt$ — the net change in position, so motion backwards subtracts. Distance travelled is $\int |v|\,dt$ — it counts every part of the journey as positive. If the velocity never changes sign they are equal; if the object reverses, find the times where $v=0$, integrate over each piece, and add the magnitudes for the total distance.
Can I just use my GDC for the area or the definite integral?
On Paper 2, yes — the numerical-integration and graph-area tools give the value instantly, and you should use them to check. But you still have to state the integral you are evaluating to earn the method mark. Two cautions: for an area where the curve dips below the axis, the plain $\int f\,dx$ returns the signed value, so integrate $|f(x)|$ for the true area; and on Paper 1 (no GDC) you must integrate by hand throughout.
Why can't I skip the constant of integration?
Because integrating a rate only recovers the original quantity up to a constant — every vertical shift of the curve has the same gradient, so the rate alone cannot tell them apart. A single known value (the boundary condition) fixes which one you have. This is exactly why, in a cost model, the constant is the fixed set-up cost $C(0)$: integrating the marginal cost rebuilds the variable cost, and the boundary condition adds the fixed cost back on.

Get the printable PDF cheatsheet — free

The same cheatsheet on one A4 page — formulas, methods and traps. No sign-in needed.