· H2 Mathematics (9758) — Ten-Year Series, worked ← All 12 questions

2024 A-Level · Paper 1 · Question 8

A series built from a relation, not from the function

Maclaurin series 9 marks 9758/01 Implicit differentiation · Maclaurin · standard series

Differentiating cos⁡(1−e2x)\cos\left(1-e^{2x}\right) three times directly is punishing. Part (a) hands you a relation between yy, y′y' and y′′y'' instead — and once you have that, every further derivative is one differentiation of a short expression rather than one of an ever-growing one. Beside the solution, the function and its approximations, drawn together.

The approximationsbuilding the series

Look this far from the origin0.60
The function The Maclaurin approximation The binomial series

Part (a)

3 marks

Given y=cos⁡(1−e2x)y=\cos\left(1-e^{2x}\right), show that d2ydx2=k(dydx−2ye4x)\dfrac{d^{2}y}{dx^{2}}=k\left(\dfrac{dy}{dx}-2ye^{4x}\right), finding kk.

Approach

Differentiate twice with the chain rule, then look at the result and hunt for the pieces you already have names for. The whole art of this part is not expanding everything out: dydx\frac{dy}{dx} and yy are already sitting inside d2ydx2\frac{d^{2}y}{dx^{2}}, and the target tells you they are what you are meant to find.

Let u=1−e2xu=1-e^{2x}, so y=cos⁡uy=\cos u and dudx=−2e2x\frac{du}{dx}=-2e^{2x}. Then

dydx=−sin⁡u⋅dudx=−sin⁡u(−2e2x)=2e2xsin⁡u\frac{dy}{dx}=-\sin u\cdot\frac{du}{dx}=-\sin u\left(-2e^{2x}\right)=2e^{2x}\sin u

Differentiating again, by the product rule on 2e2x2e^{2x} and sin⁡u\sin u:

d2ydx2=4e2xsin⁡u+2e2xcos⁡u⋅dudx=4e2xsin⁡u+2e2xcos⁡u(−2e2x)=4e2xsin⁡u−4e4xcos⁡u\begin{aligned}\frac{d^{2}y}{dx^{2}}&=4e^{2x}\sin u+2e^{2x}\cos u\cdot\frac{du}{dx}\\[2pt]&=4e^{2x}\sin u+2e^{2x}\cos u\left(-2e^{2x}\right)\\[2pt]&=4e^{2x}\sin u-4e^{4x}\cos u\end{aligned}

Now recognise the pieces. 4e2xsin⁡u=2(2e2xsin⁡u)=2dydx4e^{2x}\sin u=2\left(2e^{2x}\sin u\right)=2\frac{dy}{dx}, and cos⁡u=y\cos u=y, so

d2ydx2=2dydx−4ye4x=2(dydx−2ye4x)(shown)\frac{d^{2}y}{dx^{2}}=2\frac{dy}{dx}-4ye^{4x}=2\left(\frac{dy}{dx}-2ye^{4x}\right)\quad\textbf{(shown)}

To check your answer:

At x=0x=0: u=1−1=0u=1-1=0, so y=cos⁡0=1y=\cos0=1 and dydx=2(1)sin⁡0=0\frac{dy}{dx}=2(1)\sin0=0. The relation then predicts d2ydx2=2(0−2(1)(1))=−4\frac{d^{2}y}{dx^{2}}=2\left(0-2(1)(1)\right)=-4, which agrees with computing 4e0sin⁡0−4e0cos⁡0=−44e^{0}\sin0-4e^{0}\cos0=-4 directly ✓.

∴ k=2\therefore\ k=2.

Where the marks go

Two chain rules and one product rule, and the inner derivative −2e2x-2e^{2x} is used twice — once with a minus that cancels the one from dducos⁡u\frac{d}{du}\cos u, and once that does not. Writing uu explicitly and keeping it as uu until the last line is what stops the signs drifting; substituting 1−e2x1-e^{2x} back in at every step buries the structure the question wants you to see.

Part (b)

4 marks

By differentiating the result in part (a), find the first three non-zero terms of the Maclaurin expansion of cos⁡(1−e2x)\cos\left(1-e^{2x}\right).

Approach

Maclaurin needs the derivatives at x=0x=0, not as functions. So evaluate what you already have at x=0x=0, differentiate the relation once more to reach the third derivative, evaluate that too, and assemble. "By differentiating the result in part (a)" is an instruction, not a suggestion — it is the cheap route to y′′′y'''.

At x=0x=0, using u=1−e0=0u=1-e^{0}=0:

y(0)=cos⁡0=1y′(0)=2e0sin⁡0=0y′′(0)=2(y′(0)−2y(0)e0)=2(0−2(1)(1))=−4\begin{aligned}y(0)&=\cos0=1\\y'(0)&=2e^{0}\sin0=0\\y''(0)&=2\left(y'(0)-2y(0)e^{0}\right)=2\left(0-2(1)(1)\right)=-4\end{aligned}

Differentiating the relation y′′=2(y′−2ye4x)y''=2\left(y'-2ye^{4x}\right), with the product rule on ye4xye^{4x}:

y′′′=2(y′′−2[y′e4x+4ye4x])y'''=2\left(y''-2\left[y'e^{4x}+4ye^{4x}\right]\right)

and at x=0x=0:

y′′′(0)=2(−4−2[0+4(1)(1)])=2(−4−8)=−24y'''(0)=2\left(-4-2\left[0+4(1)(1)\right]\right)=2(-4-8)=-24

The Maclaurin series is y=y(0)+y′(0)x+y′′(0)2!x2+y′′′(0)3!x3+⋯y=y(0)+y'(0)x+\frac{y''(0)}{2!}x^{2}+\frac{y'''(0)}{3!}x^{3}+\cdots, so

y=1+0⋅x−42x2−246x3+⋯=1−2x2−4x3+⋯y=1+0\cdot x-\frac{4}{2}x^{2}-\frac{24}{6}x^{3}+\cdots=1-2x^{2}-4x^{3}+\cdots

To check your answer:

Expand from standard series instead. e2x=1+2x+2x2+43x3+⋯e^{2x}=1+2x+2x^{2}+\frac{4}{3}x^{3}+\cdots, so u=1−e2x=−2x−2x2−43x3−⋯u=1-e^{2x}=-2x-2x^{2}-\frac{4}{3}x^{3}-\cdots, giving u2=4x2+8x3+⋯u^{2}=4x^{2}+8x^{3}+\cdots. Then cos⁡u≈1−u22=1−2x2−4x3+⋯\cos u\approx1-\frac{u^{2}}{2}=1-2x^{2}-4x^{3}+\cdots ✓ — the same three terms by a completely different route.

∴ cos⁡(1−e2x)≈1−2x2−4x3\therefore\ \cos\left(1-e^{2x}\right)\approx1-2x^{2}-4x^{3}.

Where the marks go

"First three non-zero terms" — and the xx term is zero, so you must go as far as x3x^{3}. Stopping at 1−2x21-2x^{2} because that is three terms written down loses two marks. The other loss is dividing by the wrong factorial: the coefficient of x3x^{3} is y′′′(0)3!=−246\frac{y'''(0)}{3!}=\frac{-24}{6}, not −243\frac{-24}{3}.

Part (c)

2 marks

The first two non-zero terms of the expansion equal the first two non-zero terms of the series for 1a+bx2\dfrac{1}{\sqrt{a+bx^{2}}}. Using standard series from MF26, find aa and bb.

Approach

MF26 gives (1+z)n(1+z)^{n}, which needs a leading 1 — so factor aa out of the square root first. Then expand, take two terms, and match against 1−2x21-2x^{2}.

Factoring aa out:

1a+bx2=1a(1+bax2)−1/2\frac{1}{\sqrt{a+bx^{2}}}=\frac{1}{\sqrt{a}}\left(1+\frac{b}{a}x^{2}\right)^{-1/2}

and expanding with n=−12n=-\frac{1}{2}, z=bax2z=\frac{b}{a}x^{2}:

=1a[1−12⋅bax2+⋯ ]=1a−b2a3/2x2+⋯=\frac{1}{\sqrt{a}}\left[1-\frac{1}{2}\cdot\frac{b}{a}x^{2}+\cdots\right]=\frac{1}{\sqrt{a}}-\frac{b}{2a^{3/2}}x^{2}+\cdots

Matching the first two non-zero terms with 1−2x21-2x^{2}:

1a=1 ⟹ a=1\frac{1}{\sqrt{a}}=1\ \Longrightarrow\ a=1
−b2a3/2=−2 ⟹ −b2=−2 ⟹ b=4-\frac{b}{2a^{3/2}}=-2\ \Longrightarrow\ -\frac{b}{2}=-2\ \Longrightarrow\ b=4

To check your answer:

Expanding the answer back out:

11+4x2=(1+4x2)−1/2=1−12(4x2)+⋯=1−2x2+⋯\frac{1}{\sqrt{1+4x^{2}}}=\left(1+4x^{2}\right)^{-1/2}=1-\frac{1}{2}\left(4x^{2}\right)+\cdots=1-2x^{2}+\cdots

✓, matching the first two non-zero terms of part (b). The two curves are drawn together in the model — they agree to the eye near the origin and part company as ∣x∣\left|x\right| grows, which is exactly what "first two terms" promises and no more.

∴ a=1\therefore\ a=1 and b=4b=4.

Note — why the third terms need not agree

They do not: the expansion of (1+4x2)−1/2\left(1+4x^{2}\right)^{-1/2} is even, so its next term is in x4x^{4}, while cos⁡(1−e2x)\cos\left(1-e^{2x}\right) has a −4x3-4x^{3}. The question only ever claims agreement to two terms, and the model shows the two curves separating precisely where that claim runs out.

What this question was really testing

  1. Recognising your own expressions. Part (a) is two routine differentiations followed by the real step: spotting dydx\frac{dy}{dx} and yy inside the result instead of expanding them away.
  2. Why a relation beats a formula. Differentiating y′′=2(y′−2ye4x)y''=2\left(y'-2ye^{4x}\right) to get y′′′y''' is a two-line job; differentiating cos⁡(1−e2x)\cos\left(1-e^{2x}\right) three times is not.
  3. Non-zero terms. The xx coefficient vanishes, so "first three non-zero terms" reaches x3x^{3}. Read the words, count the surviving terms.
  4. Factoring before a binomial expansion. MF26's (1+z)n(1+z)^{n} needs a leading 1, so a+bx2\sqrt{a+bx^{2}} must become a1+bax2\sqrt{a}\sqrt{1+\frac{b}{a}x^{2}} first.

This is how we mark, and how we teach.

Every Ten-Year Series question, worked in full — with the reasoning written out, the traps named, and the curves you can actually take hold of.

Find out more