maths.

org/step

STEP Support Programme

STEP 3 Differential Equations: Solutions

1 (i) Substituting u = x gives:

0 + a(x) + xb(x) = 0 .
Substituting u = e−x gives:

e−x − a(x)e−x + b(x)e−x = 0

1 − a(x) + b(x) = 0 since e−x 6= 0 .

Then substituting a(x) = 1 + b(x) into the first equation gives:

1 + b(x) + xb(x) = 0
−1
=⇒ b(x) =
1+x
−1
=⇒ a(x) = 1 +
1+x
x
=
1+x

The general solution of the equation is u = Ax + Be−x .

1 du
Differentiating y = gives:
3u dx
2
1 d2 u

dy 1 du
= 2
− 2
dx 3u dx 3u dx

Substituting into (∗):

du 2
 2   2
1 d2 u
 
1 1 du
 x 1 du 1
2
− 2  + 3  + × =
3u dx 3u dx
  3u
 dx 1 + x 3u dx 3(1 + x)

Then multiplying throughout by 3u gives:

d2 u x du 3
u
+ =
dx2 1 + x dx 3
(1 + x)
which is the required equation, which happens to be the first equation with the a(x)
and b(x) you found earlier. This is probably not a coincidence.
The general solution of this is u = Ax + Be−x (from before), so we have:
1 du
y=
3u dx
1 1
× A − Be−x


= × −x
3 Ax + Be
A − Be−x
=
3 (Ax + Be−x )
STEP 3 DEs: solutions 1
 maths.org/step

Using the initial condition y = 0 at x = 0 gives:

A−B
0=
3(B)

A − Ae−x 1 − e−x
so we have A = B and y = = .
3 (Ax + Ae−x ) 3 (x + e−x )

(ii) A similar sort of substitution is likely to be a good idea here, and the differences
between this equation and (∗) is the absence of “3”’s and a 1 − x appearing rather
than 1 + x.
1 du
Using a substitution of y = gives:
u dx
dy x 1
+ y2 + y =
dx 1−x 1−x
2  2  
1 d2 u du 2
 
1 du 
 1 x 1 du 1
− 2

 +  + =
u dx2  u dx u  dx 1 − x u dx 1−x

2
d u x du u
2
+ =
dx 1 − x dx 1−x
2
d u x du u
+ − =0
dx2 1 − x dx 1 − x
This now looks very similar to the previous part. If you substitute in u = x and u = ex
you will find that these both satisfy the equation in u, and so the general solution is
u = Cx + Dex . This gives:
C + Dex
y=
Cx + Dex
Using the initial condition y = 2 at x = 0 gives:
C +D
2=
D
2D = C + D
D=C

1 + ex
and so the solution is y = .
x + ex

STEP 3 DEs: solutions 2

 maths.org/step

d 1
2 One of the “required formulae” is sin−1 x = √ (See the STEP specifications for
dx 1 − x2
the complete list of required formulae). This means that you should either be able to recall
this result or be able to derive it with little trouble (in this case write sin y = x and use
implicit differentiation).

dy m
If y = cos(m arcsin x)1 we have = − sin(m arcsin x) × √ and
dx 1 − x2

d2 y m2
= − cos(m arcsin x) × − m sin(m arcsin x) × − 12 × −2x × (1 − x2 )−3/2
dx2 1 − x2

To simplify things, let S = sin(m arcsin x) and C = cos(m arcsin x). We then have:

y=C
dy mS
= −√
dx 1 − x2
2
d y m2 C mxS
2
= − 2
−
dx 1−x (1 − x2 )3/2

Substituting these into the given equation we have:

2 m2 C
   
2 d y dy 2 2 mxS mS
+ m2 C


1−x −x +m y = 1−x − − − x −√
dx2 dx 1 − x2 (1 − x2 )3/2 1 − x2
mxS xmS
= −m2 C − 2 1/2
+√ + m2 C
(1 − x ) 1−x 2

=0

Hence y = cos(m arcsin x) is a solution to the given differential equation.

Differentiating the given equation gives:

d3 y d2 y d2 y dy dy
1 − x2 3
− 2x 2
− x 2
− + m2 =0
dx dx dx dx dx

d3 y d2 y
dy
1 − x2 3
− 3x 2
+ m2 − 1 =0
dx dx dx
And then differentiating again gives:

d4 y d3 y d3 y d2 y
d2 y
1 − x2 − 2x − 3x − 3 + m2
− 1 =0
dx4 dx3 dx3 dx2 dx2

d4 y d3 y
d2 y
1 − x2 − 5x + m2
− 4 =0
dx4 dx3 dx2

Looking at the three differential equations (and also thinking about what happens as we
differentiate) we make the hypothesis that:

dn+2 y dn+1 y n
2 d y
1 − x2 − (2n + 1)x + (m2
− n ) =0
dxn+2 dxn+1 dxn
For 2019 onwards STEP will use the sin−1 x notation. I slightly prefer arcsin x, and old papers might use this.
1

It doesn’t matter which you use when answering questions!

STEP 3 DEs: solutions 3
 maths.org/step

Base case: we know that this is true when n = 0 (and also n = 1, 2).

Inductive step: assume that the hypothesis is true when n = k, i.e. we have:

dk+2 y dk+1 y k
2 d y
1 − x2 − (2k + 1)x + (m2
− k ) =0 (*)
dxk+2 dxk+1 dxk
Differentiating (∗) with respect to x gives:

dk+3 y dk+2 y dk+2 y dk+1 y 2 d
k+1 y
1 − x2 − 2x − (2k + 1)x − (2k + 1) + (m 2
− k ) =0
dxk+3 dxk+2 dxk+2 dxk+1 dxk+1

dk+3 y dk+2 y
dk+1 y
1 − x2 − (2x + (2k + 1)x) + m 2
− k 2
− (2k + 1) =0
dxk+3 dxk+2 dxk+1

dk+3 y dk+2 y
dk+1 y
1 − x2 − (2 + 2k + 1) x + m 2
− (k 2
+ 2k + 1) =0
dxk+3 dxk+2 dxk+1

dk+3 y dk+2 y
k+1 y
2 d
1 − x2 − (2(k + 1) + 1) x + m 2
− (k + 1) =0
dxk+3 dxk+2 dxk+1
and this is the same as (∗) with n = k + 1, Hence if it is true for n = k then it is true for
n = k + 1, and as it is true for n = 0 it is true for all integers n > 0.
The general Maclaurin’s Series expansion is:
x2 00
f(x) = f(0) + xf 0 (0) + f (0) + . . .
2!
Since y = cos(m arcsin x), when x = 0 we have y = 1. Using the expressions found for the
dy d2 y
first two derivatives of y found at the start of the question, we have = 0 and = −m2
dx dx2
when x = 0.
d3 y d2 y dy d3 y
Using (1 − x2 ) 3 − 3x 2 + (m2 − 1) = 0 with x = 0 gives = 0.
dx dx dx dx3
d4 y d3 y d2 y d4 y
Finally (1 − x2 ) 4 − 5x 3 + (m2 − 4) 2 = 0 with x = 0 gives = m2 (m2 − 4).
dx dx dx dx4
Hence the first three non-zero terms of the Maclaurin series for y are:
m2 2 m2 (m2 − 4) 4
y =1− x + x + ... .
2! 4!
Letting x = sin θ gives:
m2 m2 (m2 − 4)
y = cos(mθ) = 1 − sin2 θ + sin4 θ + . . .
2! 4!
By considering

dn+2 y dn+1 y n
2 d y
1 − x2 − (2n + 1)x + (m2
− n ) =0
dxn+2 dxn+1 dxn
and letting x = 0, we have:
dn+2 y 2
n
2 d y
= −(m − n )
dxn+2 dxn
This means that all the odd differentials are going to
be zero, and the even ones will be
given by (−1)k+1 m2 (m2 − 22 )(m2 − 42 ) . . . m2 − (2k)2 . This means that if m is even this
Maclaurin’s series for cos mθ will terminate since at some point all the derivatives will be
zero. Hence the series is a polynomial in sin θ and will have degree m (to convince yourself
of this final bit, try considering m = 2, m = 4 etc.).

STEP 3 DEs: solutions 4

 maths.org/step

3 We have:
Q(x)R0 (x) − Q0 (x)R(x)
Z Z
P(x)

2 dx =
2 dx
Q(x) Q(x)
Z  
d R(x)
= dx
dx Q(x)
R(x)
= +k
Q(x)

(i) Let P(x) = 5x2 −4x−3 and let Q(x) = 1+2x+3x2 . We then want R(x) = a+bx+cx2
to satisfy:
P(x) = Q(x)R0 (x) − Q0 (x)R(x)
5x2 − 4x − 3 = 1 + 2x + 3x2 (b + 2cx) − (2 + 6x) a + bx + cx2

Note that the x3 terms on the RHS cancel out. Equating coefficients gives:
−3 = b − 2a
−4 = 2b + 2c − 6a − 2b
5 = 3b + 4c − 2c − 6b
Simplifying gives:
−3 = b − 2a (1)
−2 = c − 3a (2)
5 = 2c − 3b (3)

These three equations are not linearly independent, i.e. if we cancel a from the first
two (by considering 2 × (2) − 3 × (1)) we get:
(2 × −2) − (3 × −3) = 2c − 3b =⇒ 5 = 2c − 3b
Which is the same as (3) and hence the three equations do not have a unique solution.
This means that we can pick a arbitrarily. Letting a = 1 gives b = −1 and c = 1 and
hence the integral is equal to:
R(x) 1 − x + x2
+k = +k
Q(x) 1 + 2x + 3x2
If instead we took a = 2, we would have b = 1 and c = 4. This gives:
R(x) 2 + x + 4x2
+ k0 = + k0
Q(x) 1 + 2x + 3x2
We can re-write the second integral as follows:
2 + x + 4x2 0 (1 + 2x + 3x2 ) + 1 − x + x2
+ k = + k0
1 + 2x + 3x2 1 + 2x + 3x2
1 − x + x2
=1+ + k0
1 + 2x + 3x2
1 − x + x2
= +k
1 + 2x + 3x2
I.e. the only difference is in the arbitrary constant.
STEP 3 DEs: solutions 5
 maths.org/step

(ii) Start by rearranging the equation to get:

dy sin x − 2 cos x 5 − 3 cos x + 4 sin x
+ y=
dx 1 + cos x + 2 sin x 1 + cos x + 2 sin x
The integrating factor is:
sin x−2 cos x
R
dx
e 1+cos x+2 sin x = e− ln(1+cos x+2 sin x)
1
=
1 + cos x + 2 sin x
Multiplying by the integrating factor gives:
1 dy sin x − 2 cos x 5 − 3 cos x + 4 sin x
+ y=
1 + cos x + 2 sin x dx (1 + cos x + 2 sin x)2 (1 + cos x + 2 sin x)2
 
d 1 5 − 3 cos x + 4 sin x
×y =
dx 1 + cos x + 2 sin x (1 + cos x + 2 sin x)2
5 − 3 cos x + 4 sin x
Z
1
×y = dx
1 + cos x + 2 sin x (1 + cos x + 2 sin x)2

Now let P(x) = 5 − 3 cos x + 4 sin x, Q(x) = 1 + cos x + 2 sin x and R(x) = a + b cos x +
c sin x. Using P(x) = Q(x)R0 (x) − Q0 (x)R(x) gives:
5 − 3 cos x + 4 sin x = (1 + cos x + 2 sin x) (−b sin x + c cos x)
− (− sin x + 2 cos x) (a + b cos x + c sin x)
= −b sin x + c cos x − 2b sin2 x + c cos2 x + (2c − b) sin x cos x

− −a sin x + 2a cos x − c sin2 x + 2b cos2 x + (2c − b) sin x cos x

= (c − 2b) sin2 x + cos2 x + (c − 2a) cos x + (a − b) sin x

So we have:
5 = c − 2b (4)
−3 = c − 2a (5)
4=a−b (6)
Note that (4) − (5) gives 8 = −2b + 2a =⇒ 4 = aZ− b. The equations are linearly
P(x) R(x)
dependent. Picking a = 5, b = 1 gives c = 72 . Using
2 dx = + k gives:
Q(x) Q(x)
5 − 3 cos x + 4 sin x
Z
5 + cos x + 7 sin x
2
dx = +k
(1 + cos x + 2 sin x) 1 + cos x + 2 sin x
y 5 + cos x + 7 sin x
=⇒ = +k
1 + cos x + 2 sin x 1 + cos x + 2 sin x
=⇒ y = 5 + cos x + 7 sin x + (1 + cos x + 2 sin x)k
This answer is not unique, as we could rewrite it as
y = 4 + 5 sin x + (1 + cos x + 2 sin x) + (1 + cos x + 2 sin x)k
= 4 + 5 sin x + (1 + cos x + 2 sin x)(k + 1)
= 4 + 5 sin x + (1 + cos x + 2 sin x)k 0
etc.
2
Note that you could have picked any value of a you liked.
STEP 3 DEs: solutions 6
 maths.org/step

4 There are various ways you can find the differential equation in y. The following is one
possible approach.

Firstly, we have ẏ = −2y + 2z. This gives:

ẏ
z=+y
2
ÿ
=⇒ ż = + ẏ
2
and so

ż = −ẏ − 3z
 
ÿ ẏ
=⇒ + ẏ = −ẏ − 3 +y
2 2
ÿ + 2ẏ = −2ẏ − 3 (ẏ + 2y)
ÿ + 7ẏ + 6y = 0

This gives the characteristic polynomial/ auxiliary equation r2 + 7r + 6 = 0 which has roots
r = −1, −6. The general equation is therefore

y = Ae−t + Be−6t =⇒
ẏ
z = +y
2
= − 21 Ae−t − 3Be−6t + Ae−t + Be−6t
= 21 Ae−t − 2Be−6t

(i) Using z(0) = 0 and y(0) = 5 gives us:

0 = 12 A − 2B and 5=A+B

Solving these simultaneously gives A = 4 and B = 1, and so:

y1 (t) = 4e−t + e−6t and z1 (t) = 2e−t − 2e−6t

(ii) z(0) = z(1) = c gives:

−1
1
2A − 2B = c and 1
2 Ae − 2Be−6 = c

Hence:
1
2A − 2B = c (7)
− 2Be−5 = c × e
1
2A (8)
(2) − (1) =⇒ 2B 1 − e−5 = c(e − 1)


(9)

and so
c e6 − e5


c(e − 1)
B= =
2 (1 − e−5 ) 2 (e5 − 1)
STEP 3 DEs: solutions 7
 maths.org/step

and

A = 2c + 4B
2c e6 − e5


= 2c +
(e5 − 1)

!
e6 − e5
= 2c 1 + 5
(e − 1)
 5
e − 1 + e 6 − e5

= 2c
e5 − 1
 6 
e −1
= 2c 5
e −1

We therefore have:

e6 − 1 −t c e6 − e5 −6t
 

y2 (t) = 2c 5 e + e and
e −1 2 (e5 − 1)
e − 1 −t c e6 − e5 −6t
 6 

z2 (t) = c 5 e − e
e −1 (e5 − 1)

(iii) We have z1 (t − n) = 2e−(t−n) − 2e−6(t−n) = 2e−t en − 2e−6t e6n . This means we have:
0
X 0
X 0
X
z1 (t − n) = 2e−t en − 2e−6t e6n
n=−∞ n=−∞ n=−∞
X∞ ∞
X
−t −i −6t −6i
= 2e e − 2e e
i=0 i=0

The last step is taken by letting the index n = −i (and reversing the sum). The first
sum is the infinite sum of a Geometric Progression with common ratio e−1 and the
second is a GP with common ratio e−6 . We therefore have:
0
X 1 1
z1 (t − n) = 2e−t × −1
− 2e−6t ×
n=−∞
1−e 1 − e−6
2e −t 2e6 −6t
= e − 6 e („)
e−1 e −1
Comparing the coefficient of e−t of this to the coefficient of e−t from z2 (t) we need:
 6 
e −1 2e
c 5 =
e −1 e−1
e5 − 1


2e
=⇒ c = × 6
e−1 (e − 1)

STEP 3 DEs: solutions 8

 maths.org/step

This means that the coefficient of e−6t in z2 (t) is:

c e6 − e5 e5 − 1 e6 − e5




2e
− =− × 6 × 5
(e5 − 1) e−1 (e − 1) (e − 1)
5


2e e− 1 e5 −
(e 1)
  
=− × × 5 


(e
 − 1)
 (e6 − 1) e− 1
2e6
=−
e6 − 1

which is the same as the coefficient of e−6t in (†) and hence we have
0
X
z2 (t) = z1 (t − n)
n=−∞

when
e5 − 1


2e
c= × 6 .
e−1 (e − 1)

STEP 3 DEs: solutions 9