Integration
Revise Integration for Further Pure Mathematics 3 WFM03 (AS Level) — revision notes and instant AI marking. Free to start.
Integration
Every integral in this chapter is really the same three moves in disguise: spot a pattern you already know, force the integrand into that shape, then execute — whether that's a clean substitution, a "reverse chain rule" guess, or building a reduction formula for when the power just won't quit.
📎 Builds on: Chapter 1 (Hyperbolic Functions) — you'll lean heavily on sinh, cosh, their derivatives, and the log forms of arsinh/arcosh/artanh. Chapter 3's derivatives of inverse functions also feed straight into the standard substitutions here.
Summary
- Recognise a standard function — some integrals are just derivatives you already know, run backwards.
- Using formulae to change the integrand — trig/hyperbolic identities (like double angle formulae) turn an unintegrable expression into one you can do term by term.
- Reverse chain rule — spot the "function of a function times its derivative" pattern, guess the antiderivative shape, differentiate to fix the constant.
- Standard substitutions — four go-to substitutions (trig or hyperbolic) for integrals containing a² ± b²x², with or without a square root.
- Integrating inverse functions and ln x — integration by parts with dv/dx = 1, a trick that works because you can differentiate the inverse function even though you can't easily integrate it.
- Reduction formulae — for integrals with a power n, express Iₙ in terms of Iₙ₋₁ or Iₙ₋₂ instead of integrating from scratch every time.
- Arc length — using Pythagoras on an infinitesimal triangle to find the length of a curve, in Cartesian or parametric form.
- Area of a surface of revolution — spinning a curve around the x-axis and adding up the surface area of infinitely many thin cylindrical bands.
Standard Techniques
Before reaching for anything clever, always ask: "Have I seen this exact shape before, as a derivative?" This section builds up your toolkit in the order you should actually try it in an exam, from laziest to most involved.
1. Recognise a standard function
This is the "if it's staring you in the face, just write the answer down" category. If you can look at the integrand and immediately think "that's the derivative of something," you're done — no working required beyond stating the antiderivative and checking it by differentiating.
These work because d(sec x)/dx = sec x tan x and d(sech x)/dx = −sech x tanh x — you memorised these derivatives in Chapter 3, and integration is just running the film backwards.
Recall d(cosech x)/dx = −cosech x coth x.
So ∫ cosech x coth x dx = −cosech x + c
2. Using formulae to change the integrand
Sometimes there's no standard derivative that matches — yet. But an identity (Pythagorean, double angle, Osborne's rule) can rewrite the integrand into a form that does match. This is the "disguise removal" technique.
tan²x on its own isn't a standard integral. But 1 + tan²x = sec²x is.
⇒ tan²x = sec²x − 1
⇒ ∫ tan²x dx = ∫ sec²x − 1 dx = tan x − x + c
Use cosh 2x = 1 + 2sinh²x ⇒ sinh²x = ½(cosh 2x − 1)
∫ sinh²x dx = ½ ∫ cosh 2x − 1 dx = ½(½ sinh 2x − x) + c
3. Reverse chain rule
This is the technique for integrands that look like f'(g(x)) · g'(x) — a function of a function, multiplied by the derivative of the "inner" bit. The method: notice the pattern uⁿ · du/dx, guess that the answer involves un+1, differentiate your guess, and see what constant you need to divide by to make it match.
Looks like u² du/dx with u = cos x (since d(cos x)/dx = −sin x is "nearly" the sin x factor). Try u³ ⇔ cos³x.
d(cos³x)/dx = 3cos²x(−sin x) = −3cos²x sin x
So we need to divide by −3:
⇒ ∫ cos²x sin x dx = −⅓ cos³x + c
Looks like u⁴ du/dx with u = 2x³ − 7. Try u⁵ ⇔ (2x³ − 7)⁵.
d[(2x³−7)⁵]/dx = 5(2x³−7)⁴ × 6x² = 30x²(2x³−7)⁴
Divide by 30:
⇒ ∫ x² (2x³ − 7)⁴ dx = 1/30 (2x³ − 7)⁵ + c
Rewrite: ∫ sech³x (sech x tanh x) dx — looks like u³ du/dx with u = sech x. Try u⁴ ⇔ sech⁴x.
d(sech⁴x)/dx = 4sech³x × (−sech x tanh x) = −4sech³x sech x tanh x
Divide by −4:
⇒ ∫ sech⁴x tanh x dx = −¼ sech⁴x + c
4. Standard substitutions
This is the big one — four "template" substitutions to memorise, each designed to knock out a square root or a quadratic denominator by turning it into a hyperbolic or trig identity (like cosh²u − sinh²u = 1). The choice depends on two things: is there a square root, and is it a sum or a difference?
| Integral | Substitution | Why |
|---|---|---|
| ∫ 1/(a² + b²x²) dx | bx = a tan u | No root present — use tan (avoids messy hyperbolic forms) |
| ∫ 1/√(a² + b²x²) dx | bx = a sinh u | Root of a sum — sinh, since 1 + sinh²u = cosh²u |
| ∫ 1/(a² − b²x²) dx | bx = a tanh u (or partial fractions) | No root, difference — tanh |
| ∫ 1/√(b²x² − a²) dx | bx = a cosh u | Root of a difference, x² term first — cosh, since cosh²u − 1 = sinh²u |
Square root present? Use sinh (sum under the root) or cosh (difference under the root, x² term positive and first).
4x² − 8x − 5 = 4(x² − 2x + 1) − 9 = 4(x − 1)² − 9
So the integral is ∫ 1/√[4(x−1)² − 9] dx — a "root of a difference," x² term first → use cosh.
Substitute 2(x − 1) = 3 cosh u ⇒ 2 dx = 3 sinh u du
= ∫ [1/√(9(cosh²u − 1))] × (3 sinh u / 2) du
= ∫ [1/(3 sinh u)] × (3 sinh u / 2) du = ½ ∫ du = ½u + c
⇒ ½ arcosh[(2x − 2)/3] + c
Let u = 2x − 3 ⇒ dx = ½ du
I = ∫ 1/√(u² + 5²) × ½ du = ½ arsinh(u/5) + c
⇒ I = ½ arsinh[(2x − 3)/5] + c
5. Integrating inverse functions and ln x
You can't easily write down an antiderivative of arctan x or ln x by inspection — but you can differentiate them (Chapter 3 gave you all these derivatives). The trick: treat the whole function as "u" in integration by parts, and treat dv/dx = 1, so v = x. That converts an integral you can't do into one you can.
| f(x) | f'(x) |
|---|---|
| arcsin x | 1/√(1−x²) |
| arccos x | −1/√(1−x²) |
| arctan x | 1/(1+x²) |
| arsinh x | 1/√(1+x²) |
| arcosh x | 1/√(x²−1) |
| artanh x | 1/(1−x²) |
Let u = arctan x ⇒ du/dx = 1/(1+x²), and dv/dx = 1 ⇒ v = x
∫ arctan x dx = x arctan x − ∫ x × 1/(1+x²) dx
The remaining integral is a reverse chain rule: ∫ x/(1+x²) dx = ½ ln(1+x²)
⇒ ∫ arctan x dx = x arctan x − ½ ln(1 + x²) + c
Let u = arcosh x ⇒ du/dx = 1/√(x²−1), and dv/dx = 1 ⇒ v = x
∫ arcosh x dx = x arcosh x − ∫ x/√(x²−1) dx
The remaining integral is reverse chain rule: ∫ x(x²−1)^(−½) dx = √(x²−1)
⇒ ∫ arcosh x dx = x arcosh x − √(x² − 1) + c
Reduction Formulae
Some integrals have a power n in them, and computing them from scratch for large n would be brutal. A reduction formula expresses Iₙ in terms of a lower-power version — usually Iₙ₋₁ or Iₙ₋₂ — so you can "step down" to a base case you can integrate directly, then climb back up.
(a) Reduction formula: u = xⁿ ⇒ du/dx = nx^(n−1); dv/dx = eˣ ⇒ v = eˣ
Iₙ = xⁿeˣ − ∫ nx^(n−1)eˣ dx = xⁿeˣ − n Iₙ₋₁
(b) I₀ = ∫ eˣ dx = eˣ + c
(c) Climb back up:
I₄ = x⁴eˣ − 4I₃ = x⁴eˣ − 4(x³eˣ − 3I₂) = x⁴eˣ − 4x³eˣ + 12(x²eˣ − 2I₁)
= x⁴eˣ − 4x³eˣ + 12x²eˣ − 24(xeˣ − I₀)
⇒ I₄ = x⁴eˣ − 4x³eˣ + 12x²eˣ − 24xeˣ + 24eˣ + c
Iₙ = ∫ tan^(n−2)x tan²x dx = ∫ tan^(n−2)x (sec²x − 1) dx
= ∫ tan^(n−2)x sec²x dx − ∫ tan^(n−2)x dx
The first integral is reverse chain rule (u = tan x): gives (1/(n−1)) tan^(n−1)x.
⇒ Iₙ = (1/(n−1)) tan^(n−1)x − Iₙ₋₂
Split sinⁿx = sin^(n−1)x sin x, take u = sin^(n−1)x, dv/dx = sin x ⇒ v = −cos x
Iₙ = [−cos x sin^(n−1)x]₀^(π/2) + (n−1)∫₀^(π/2) cos²x sin^(n−2)x dx
= 0 + (n−1)∫₀^(π/2)(1 − sin²x) sin^(n−2)x dx = (n−1)Iₙ₋₂ − (n−1)Iₙ
⇒ nIₙ = (n−1)Iₙ₋₂ ⇒ Iₙ = [(n−1)/n] Iₙ₋₂
Using it: I₆ = (5/6)I₄ = (5/6)(3/4)I₂ = (5/6)(3/4)(1/2)I₀
I₀ = ∫₀^(π/2) 1 dx = π/2
⇒ I₆ = 5π/32
First pass: u = xⁿ, dv/dx = cos x ⇒ v = sin x
Iₙ = [xⁿ sin x]₀^(π/2) − n∫₀^(π/2) x^(n−1) sin x dx = (π/2)ⁿ − n∫₀^(π/2) x^(n−1) sin x dx
Second pass on the remaining integral: u = x^(n−1), dv/dx = sin x ⇒ v = −cos x
∫₀^(π/2) x^(n−1) sin x dx = [−x^(n−1)cos x]₀^(π/2) + (n−1)∫₀^(π/2) x^(n−2)cos x dx = 0 + (n−1)Iₙ₋₂
⇒ Iₙ = (π/2)ⁿ − n(n−1) Iₙ₋₂
u = xⁿ ⇒ du/dx = nx^(n−1); dv/dx = (1+x)² ⇒ v = ⅓(1+x)³
Iₙ = [xⁿ·⅓(1+x)³]₋₁⁰ − ∫₋₁⁰ nx^(n−1)·⅓(1+x)³ dx = 0 − (n/3)∫₋₁⁰ x^(n−1)(1+x)³ dx
Write (1+x)³ = (1+x)²(1+x) to split into two integrals in terms of Iₙ₋₁ and Iₙ:
Iₙ = −(n/3)Iₙ₋₁ − (n/3)Iₙ ⇒ [(n+3)/3]Iₙ = −(n/3)Iₙ₋₁
⇒ Iₙ = −[n/(n+3)] Iₙ₋₁
Write sin nx = sin[(n−2)x + 2x] and expand using addition formulae, then use double angle formulae for cos 2x and sin 2x. After simplifying:
Iₙ = ∫ [sin(n−2)x/sin x] dx + 2∫ cos(n−1)x dx = Iₙ₋₂ + [2/(n−1)] sin(n−1)x
⇒ Iₙ = Iₙ₋₂ + [2/(n−1)] sin(n−1)x
Arc Length
Imagine zooming so far into a curve that a tiny piece of it looks like a straight line segment, forming the hypotenuse of a tiny right-angled triangle with legs δx and δy. Pythagoras gives (δs)² ≈ (δx)² + (δy)². Divide through and let δx → 0, and you get the arc length formula.
dy/dx = √x
s = ∫₃⁸ √(1 + x) dx = [⅔(1+x)^(3/2)]₃⁸ = ⅔(9)^(3/2) − ⅔(4)^(3/2)
= ⅔(27) − ⅔(8) = 18 − 16/3
⇒ s = 12⅔
dx/dt = a(1 − cos t), dy/dt = a sin t
ẋ² + ẏ² = a²(1 − 2cos t + cos²t + sin²t) = 2a²(1 − cos t)
Use 1 − cos t = 2sin²(t/2): √(ẋ²+ẏ²) = a√[4sin²(t/2)] = 2a sin(t/2) (for 0 ≤ t ≤ 2π, sin(t/2) ≥ 0)
s = ∫₀^(2π) 2a sin(t/2) dt = [−4a cos(t/2)]₀^(2π) = −4a − (−4a)
⇒ s = 8a
Area of a Surface of Revolution
Spin a curve around the x-axis and it sweeps out a 3D surface — think of a lampshade or a vase. To find its surface area, chop the curve into tiny arc pieces of length δs, each sitting a distance y from the axis. When rotated, each tiny piece traces out (approximately) a thin cylindrical band of circumference 2πy and width δs, so its area is ≈ 2πy·δs. Add them all up and take the limit — that's the integral.
where you plug in ds/dx = √[1+(dy/dx)²] or ds/dt = √[(dx/dt)²+(dy/dt)²] from the arc length section above — this formula is literally "2πy" glued onto the arc length integrand.
x² + y² = r² ⇒ 2x + 2y(dy/dx) = 0 ⇒ dy/dx = −x/y
ds/dx = √[1 + x²/y²]
A = ∫ₐᵇ 2πy √(1 + x²/y²) dx = ∫ₐᵇ 2π√(y² + x²) dx = ∫ₐᵇ 2πr dx (since x²+y²=r²)
⇒ A = [2πrx]ₐᵇ = 2πr(b − a)
Taking the whole sphere (a = −r to b = r) gives the famous 4πr² result.
dx/dt = 2at, dy/dt = 2a
ds/dt = √[(2at)² + (2a)²] = 2a√(t²+1)
A = ∫₀² 2π(2at) × 2a√(t²+1) dt = 8πa² ∫₀² t√(t²+1) dt
Let u = t²+1: A = 8πa² × [⅓(t²+1)^(3/2)]₀² = 8πa²[⅓(5^(3/2)) − ⅓]
⇒ A = (8πa²/3)(5√5 − 1)
What to Memorise
Standard substitutions
| ∫1/(a²+b²x²) dx | bx = a tan u |
| ∫1/√(a²+b²x²) dx | bx = a sinh u |
| ∫1/(a²−b²x²) dx | bx = a tanh u (or partial fractions) |
| ∫1/√(b²x²−a²) dx | bx = a cosh u |
Derivatives of inverse functions (needed for "parts with dv/dx=1")
| arcsin x | 1/√(1−x²) |
| arccos x | −1/√(1−x²) |
| arctan x | 1/(1+x²) |
| arsinh x | 1/√(1+x²) |
| arcosh x | 1/√(x²−1) |
| artanh x | 1/(1−x²) |
Core formulae
- Arc length (Cartesian): s = ∫√[1+(dy/dx)²] dx
- Arc length (parametric): s = ∫√(ẋ²+ẏ²) dt
- Surface of revolution: A = ∫2πy (ds/dx) dx or ∫2πy (ds/dt) dt
- Reduction formula strategy: integrate by parts to peel off one power at a time, relating Iₙ to Iₙ₋₁ or Iₙ₋₂
- Odd power under a root → substitute u = a² ± x² directly; even power → use the full trig/hyperbolic substitution
Concepts Checklist
Exam Tips
Read the full Integration notes free
That's the preview — create a free account to read the rest, plus flashcards and practice questions with instant AI marking. No credit card.
Unlock the full notes free →