Thanks to visit codestin.com
Credit goes to www.scribd.com

0% found this document useful (0 votes)
64 views10 pages

MATH2018 - 2019 Exam T1 2020 Solutions

This document contains solutions to exam questions in mathematics. It includes worked solutions to multi-part questions involving linear algebra, differential equations, Fourier series, and separation of variables for the Laplace equation.

Uploaded by

Norris wong
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
64 views10 pages

MATH2018 - 2019 Exam T1 2020 Solutions

This document contains solutions to exam questions in mathematics. It includes worked solutions to multi-part questions involving linear algebra, differential equations, Fourier series, and separation of variables for the Laplace equation.

Uploaded by

Norris wong
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Math2018/2019 Exam Solutions

School of Mathematics, UNSW


Term 1, 2020

1
4

3. [20 marks]

i) a) [6 marks] We note that


   
2 2 > 3 4 x
3x + 8xy − 3y = ~x A~x where A = and ~x = .
4 −3 y

To find Q, we must diagonalise the symmetric matrix A by finding its eigenvalues


and eigenvectors. Since

(λ − 3) −4
det(λI − A) = = (λ − 3)(λ + 3) − (−4)2
−4 (λ + 3)
= λ2 − 9 − 16 = λ2 − 52 = (λ + 5)(λ − 5),

the eigenvalues are λ1 = 5 and λ2 = −5. Row operations give


     
2 −4 1 −2 2
λ1 I − A = ∼ so ~v1 = ,
−4 8 0 0 1

and      
−8 −4 2 1 −1
λ2 I − A = ∼ so ~v1 = .
−4 −2 0 0 2

Since k~v1 k = 5 = k~v2 k, the normalised eigenvectors are
   
1 2 1 −1
v̂1 = √ and v̂2 = √ ,
5 1 5 2
and therefore the desired orthogonal matrix is
 
1 2 −1
Q= √ .
5 1 2
b) [2 marks] We have

~ > A(QX)
~x> A~x = (QX) ~ =X
~ > (Q> AQ)X
~ =X
~ > DX
~ = λ1 X 2 + λ2 Y 2

so 5X 2 − 5Y 2 = 20, which is an hyperbola:

X2 Y 2
− 2 = 1.
22 2
c) [2 marks] Hence, the points (X, Y ) = (2, 0) and (X, Y ) = (−2, 0) are closest
to the origin in the XY -plane. In the original coordinates, the corresponding
points are       
x 1 2 −1 ±2 1 ±4
=√ =√ ,
y 5 1 2 0 5 ±2
√ √ √ √
that is, (x, y) = (4/ 5, 2/ 5) and (x, y) = (−4/ 5, −2/ 5).
ii) a) [2 marks] Since f (t) = 2 − 2u(t − 3), the table of Laplace transforms gives

2 2e−3s 2(1 − e−3s )


F (s) = − = .
s s s
5

b) [3 marks] Under Laplace transformation, the ODE becomes

s2 X(s) − sx(0) − ẋ(0) + 4[sX(s) − x(0)] + 3X(s) = F (s),

which, after applying the initial conditions x(0) = ẋ(0) = 0, simplifies to

(s2 + 4s + 3)X(s) = F (s).

Thus,
F (s) 2(1 − e−3s )
X(s) = = .
s2 + 4s + 3 s(s + 1)(s + 3)
c) [2 marks] We therefore seek a partial fraction expansion,

2 A B C
= + + ,
s(s + 1)(s + 3) s s+1 s+3

and from 2 = A(s + 1)(s + 3) + Bs(s + 3) + Cs(s + 1) it follows by taking s = 0,


−1 and −3 that 2 = 3A, 2 = −2B and 2 = 6C, so
2 1
A= , B = −1, C= .
3 3
Hence,
 
−s 2 1 1
X(s) = (1 − e )G(s) where G(s) = − + .
3s s + 1 3(s + 3)

d) [3 marks] Using the table, we find that


2 1
g(t) = L−1 {G(s)} = − e−t + e−3t
3 3
and
L−1 {e−3s G(s)} = g(t − 3)u(t − 3),
so
(
g(t), 0 < t < 3,
x(t) = g(t) − g(t − 3)u(t − 3) = .
g(t) − g(t − 3), 3 < t < ∞
6

4. i) a) [2 marks] The function h(x) is plotted below.


h(x)

π 2 /4

−π π 2π
x

−π 2 /4

b) [6 marks] Since h(x) is an odd function with period 2L = 2π (so L = π), the
Fourier coefficients are an = 0 for n = 0, 1, 2, . . . and
1 π 2 π
Z Z
bn = h(x) sin nx dx = h(x) sin nx dx
π −π π 0
2 π
Z
= h(x) d(−n−1 cos nx)
π 0
for n = 1, 2, . . . . Noting that h(0) = 0 = h(π), integration by parts gives
Z π Z π
2 0 2
bn = h (x) cos nx dx = h0 (x) d(n−1 sin nx),
nπ 0 πn 0
and since sin 0 = 0 = sin nπ with h00 (x) = −2, integrating by parts a second
time gives
Z π Z π
2 00 4 4  −1 π
bn = − 2 h (x) sin nx dx = sin nx dx = −n cos nx 0
.
πn 0 πn2 0 πn2
Since cos nπ = (−1)n , we have

 8 , n = 1, 3, 5, . . . ,

4 n
bn = [1 − (−1) ] = πn3
πn3 
0, n = 2, 4, 6, . . . ,

and therefore
∞  
X 8 sin 3x sin 5x
h(x) = bn sin nx = sin x + + + ··· .
n=1
π 33 53

ii) a) [2 marks] If u(x, y) = F (x)G(y) then

∂ 2u ∂ 2u
2
= F 00 (x)G(y) and 2
= F (x)G00 (y),
∂x ∂y
so u satisfies the Laplace equation iff F 00 (x)G(y) + F (x)G00 (y) = 0. Thus,
F 00 (x) G00 (y)
− = =k
F (x) G(y)
7

for some separation constant k, implying


π
F 00 (x) + kF (x) = 0 for 0 < x < ,
2
and
G00 (y) − kG(y) = 0 for 0 < y < ∞.
b) [5 marks] Since

∂u
u(0, y) = F (0)G(y) and (π/2, y) = F 0 (π/2)G(y) for 0 < y < ∞,
∂x
the boundary conditions at x = 0 and x = π/2 are satisfied iff F (0)G(y) = 0 =
F 0 (π/2)G(y) for all y. For any non-trivial G, we must therefore ensure that

F (0) = 0 and F 0 (π/2) = 0.

Assuming k = ω 2 > 0, so that F 00 + ω 2 F = 0, we conclude that

F (x) = A cos ωx + B sin ωx.

Thus, F (0) = A implying A = 0 and F 0 (x) = Bω cos ωx, so F (π/2) = Bω cos(ωπ/2)


and we obtain a non-trivial F iff cos(ωπ/2) = 0, which is the case iff ω = n for
n = 1, 3, 5, . . . . Therefore, the non-trivial solutions are

F (x) = B sin nx for n = 1, 3, 5, . . . .

c) [2 marks] With k = ω 2 = n2 so that G00 − n2 G = 0, we see that G(y) =


Ceny + De−ny , and to satisfy u → 0 as y → ∞, we must have C = 0. Thus,
G(y) = De−ny and so

un (x, y) = e−ny sin nx for n = 1, 3, 5, . . . .

d) [2 marks] By superposition, we obtain a solution

u(x, y) = b1 u1 (x, y) + b3 u3 (x, y) + b5 u5 (x, y) + · · · ,

where the coefficients must be chosen so that u(x, 0) = h(x), that is

h(x) = b1 u1 (x, 0) + b3 u3 (x, 0) + b5 u5 (x, 0) + · · ·


= b1 sin x + b3 sin 3x + b5 sin 5x + · · · .

Thus, if h(x) = 3 sin x − 2 sin 3x then we choose b1 = 3, b3 = −2 and b5 = b7 =


· · · = 0, to obtain
u(x, y) = 3e−y sin x − 2e−3y sin 3x.
e) [1 marks] The Fourier expansion from part i) b) shows that

e−3y sin 3x e−5y sin 5x


 
8 −y
u(x, y) = e sin x + + + ··· .
π 33 53

You might also like