4296

FIRST PUBLIC EXAMINATION

Trinity Term

Preliminary Examination in Physics

Paper CP4: MATHEMATICAL METHODS 2

also

Preliminary Examination in Physics and Philosophy

Thursday 6 June 2013, 2.30 pm – 5 pm

Time allowed: 2 21 hours

Answer all of Section A and three questions from Section B.

Start the answer to each question on a fresh page.

The use of calculators is not permitted.

A list of physical constants, mathematical formulae

and conversion factors accompanies this paper.

The numbers in the margin indicate the weight that the Moderators expect to
assign to each part of the question.

Do NOT turn over until told that you may do so.

1
 Section A

1. Let u = y/x, and differentiate first with respect to x, applying the chain rule:

∂w ∂w ∂u
=
∂x ∂u ∂x
and then do the same thing but with respect to y:
∂w ∂w ∂u
= .
∂y ∂u ∂y
Now multiply the first expression by x, the second by y and add them together:
∂w ∂w ∂w −y ∂w 1 ∂w y y
x +y =x · +y · = (− + ) = 0 .
∂x ∂y ∂u x2 ∂u x ∂u x x

2. (a)
tan(2x) tan(2x) tan(2x − π)
limπ π = 2 limπ = 2 limπ
x→ 2 x− 2 x→ 2 2x − π x→ 2 2x − π
since the tangent function is π-periodic. Setting X = 2x − π, and Taylor expanding
tan X around X = 0, one then gets:

tan(2x) tan X X
lim = 2 lim = 2 = 2.
x→ π2 x − π2 X→0 X X

(b) We have
lim f (x) = a + 2b
x→2+

and
lim f (x) = b − 4a
x→2−

so for the limit to exist and be equal to f (2), a and b must simultaneously satisfy
a + 2b = b − 4a = 3, i.e. a = −1/3 and b = 5/3.

3. (a) In that case, points (0, 0, 0) and (1, 1, 1) correspond to t = 0 and t = 1

respectively, so:
Z Z 1 Z 1
2 2 2 3 2 3 2 3
A dr = (3t +6t )d(t)−14t t d(t )+20t(t ) d(t ) = (9t2 −28t6 +60t9 )dt = 5.
C t=0 t=0

(b) The straight line joining (0, 0, 0) and (1, 1, 1) is given in parametric form by
x = t, y = t, z = t. Thus:
Z 1 Z 1
13
Z
A dr = (3t2 + 6t)d(t) − 14ttd(t) + 20tt2 d(t) = (6t − 11t2 + 20t3 )dt = .
C t=0 t=0 3

4296 2
 4. (a) Calculating the gradient of T , one gets:

2xi + 2yj
∇T =
x2 + y 2

which yields, at point P = (1, 2),

2i + 4j
(∇T )P =
5
√
so T is increasing most rapidly in the direction of (∇T )P which is (i + 2j)/ 5.
√
(b) The rate of increase of T in the direction (∇T )P is |(∇T )P | = 2/ √5. Calling
the distance to travel in this direction s, one gets for the increase in T , 2/ 5s = 0.2,
i.e. s ≈ 0.22.
(c) The increase in T in the direction of u = i + j reads:

dT 2i + 4j i + j 6
 
= (∇T )P · u = · √ = √
ds u 5 2 5 2
√
So, as in (b) we get 6/(5 2)s = 0.12, i.e. s ≈ 0.14.
√
(d) In the directions perpendicular to (∇T )P , i.e. ±(2i − j)/ 5.

5. (a)
1.5

1.0

0.5

-2.0 -1.5 -1.0 -0.5 0.5 1.0 1.5

-0.5

-1.0

-1.5

(b) From the sketch, it is clear that we need the polar angles of the intersection
points of the cardioid and the circle. To find them we solve the two equations r = 3/2
and r = 1 − cos θ simultaneously. Eliminating r, we get 3/2 = 1 − cos θ, from which we
obtain θ = 2π/3 and θ = 4π/3. The area of the region is then given by:
4π/3
9√
Z 4π/3 Z 1−cos θ
sin(2θ) 3θ π
ZZ 
rdr dθ = rdr dθ = − sin θ − = 3 − ≈ 1.163
R 2π/3 3/2 8 8 2π/3 8 4

4296 3 [Turn over]

 6. Calculating the Jacobian for this change of coordinates, one obtains:

∂(u, v) 1 −3 ∂(x, y) 1
= = 7 and = .
∂(x, y) 2 1 ∂(u, v) 7

So the double integral becomes:

Z 7Z 4
x − 3y 1 u
ZZ
dx dy = dv du ,
D 2x + y 7 0 1 v
where the lower and upper limits of the integrals on u and v are obtained by plugging in
the equations of the boundary lines in the expressions for u and v: u = x − 3(x/3) = 0
and u = x − 3(x − 7)/3 = 7; v = 2x + (−2x + 1) = 1 and v = 2x + (−2x + 4) = 4.
Finally,
1 7 4u 1 7 1 49 ln 4
Z Z Z
dv du = u ln 4 du = = 7 ln 2 .
7 0 1 v 7 0 7 2

7. According to the wave propagation equation, the shape of the string is a super-
position of the incoming pulse and the reflected one. These pulses are drawn in dashes
below, the string is the solid curve.

a  
t=0   0   x  

b   b  

a  
t=b/(2cs)   x  
i.e.  cst=b/2  
b   b/2  

t=3b/(4cs)   a/2   x  
i.e.  cst=3b/4  
b   b/4  

t=b/cs   x  
i.e.  cst=b  

4296 4
 Section B

8. (a) Let us define a system of Cartesian coordinates with the origin at the pivot
point in the ceiling, an horizontal x-axis pointing towards the right hand side, and a
vertical y-axis pointing downwards. The positions of the two masses in this system of
coordinates are given by:
x1 = 2l sin θ1
y1 = 2l cos θ1
and
x2 = 2l sin θ1 + l sin θ2
y2 = 2l cos θ1 + l cos θ2
respectively. The potential energy U , which is due to gravity, then reads:

U = −mgy1 − mgy2 + U0 = −4mgl cos θ1 − mgl cos θ2 + U0

At equilibrium, θ1 = θ2 = 0, and U = 0 by convention so U0 = 5mgl and the potential

energy becomes:
U = mgl(5 − 4 cos θ1 − cos θ2 ) .

(b) To obtain the kinetic energy, we differentiate the positions of the masses with
respect to time:
ẋ1 = 2lθ̇1 cos θ1
ẏ1 = −2lθ̇1 sin θ1
and
ẋ2 = 2lθ̇1 cos θ1 + lθ̇2 cos θ2
ẏ2 = −2lθ̇1 sin θ1 − lθ̇2 sin θ2
So the kinetic energy K reads:

m 2 m ml2
K = K1 + K2 = (ẋ1 + ẏ12 ) + (ẋ22 + ẏ22 ) = 2ml2 θ̇12 + (4θ̇12 + θ̇22 + 4θ̇1 θ̇2 cos(θ1 − θ2 )) ,
2 2 2
i.e.
ml2
K= (8θ̇12 + θ̇22 + 4θ̇1 θ̇2 cos(θ1 − θ2 )) .
2

4296 5 [Turn over]

 (c) Using the small angle approximation, we can simplify the expressions for U
and K as follows:
mgl
U= (4θ12 + θ22 )
2
ml2
K= (8θ̇12 + θ̇22 + 4θ̇1 θ̇2 )
2
Using the vector notation: !
θ1
θ=
θ2
we can write, in the small angle approximation
! !
1 4mgl 0 θ1
U = (θ1 , θ2 )
2 0 mgl θ2

and ! !
1  8ml2 2ml2 θ̇1
K= θ̇1 , θ̇2 .
2 2ml2 ml2 θ̇2
and thus: !
4mgl 0
V=
0 mgl
and !
8ml2 2ml2
T=
2ml2 ml2

(d) Writing the total energy, E, in matrix form:

1 1 T
E = U + K = θT Vθ + θ̇ Tθ̇
2 2
and enforcing energy conservation, i.e. dE/dt = 0 — which allows to simplify by θ̇ —
yields
Tθ̈ + Vθ = 0 .
Multiplying this last expression by
!
1/(4ml2 ) −1/(2ml2 )
T−1 =
−1/(2ml2 ) 2/(ml2 )

gives the familiar harmonic oscillator equations of motion:

θ̈ + Gθ = 0 ,

where !
g/l −g/(2l)
G= .
−2g/l 2g/l

4296 6
 (e) Calculating the roots of the determinant of G − ω 2 I:
3g 2 g 2
ω4 −ω + 2 = 0,
l l
we find the following normal mode frequencies of the system:
2 g √ √
ω± = (3 ± 5) = ω02 (3 ± 5)
2l
p
where ω0 = g/(2l) is the frequency of a single pendulum of length 2l in the small
angle limit. To obtain the normal modes, we now have to find the eigenvectors η± of
the matrix G, corresponding to each of the eigenvalues ω±2 . Substituting ω and η in
+ +
the equation  
G − ω2I η = 0 ,
√
we get η2+ = −(1 +√ 5)η1+ . In a similar vein, substituting ω− and η− in the equation
yields η2− = −(1 − 5)η1− . So the normal modes are given by:
!
1√
η± = C± ,
−(1 ± 5)
where C± are constants.

9. (a) Assume that M and N are continuous functions and have continuous partial
derivatives everywhere in the region R bounded by C, so that Green’s theorem in the
plane is applicable. Then
∂N ∂M
I ZZ  
M dx + N dy = − dx dy .
C R ∂x ∂y
If ∂N/∂x = ∂M/∂y in R, then clearly
I
M dx + N dy = 0 .
C
H
Conversely, suppose C M dx + N dy = 0 for all closed curves C. If ∂N/∂x −
∂M/∂y > 0 at a point P , then from the continuity of the partial derivatives, it follows
that ∂N/∂x − ∂M/∂y > 0 in some region A surrounding P . Let Γ be the boundary of
A. Then
∂N ∂M
I ZZ  
M dx + N dy = − dx dy > 0
Γ A ∂x ∂y
which contradicts the assumption that the line integral is zero around any closed curve.
Similarly, ∂N/∂x − ∂M/∂y < 0 leads to a contradiction. We thus conclude that
∂N/∂x − ∂M/∂y = 0 at all points.

(b) Let F = M i + N j, where we set M = −y/(x2 + y 2 ) and N = x/(x2 + y 2 ).

   
∂/∂x M
∂N ∂M
 
∇ × F =  ∂/∂y  ×  N  = − k=0
   
∂x ∂y
∂/∂z 0
in any region excluding (0,0), since ∂N/∂x = ∂M/∂y = (−x2 + y 2 )/(x2 + y 2 )2 in such
a region. So ∇ × F = 0 is equivalent to ∂N/∂x = ∂M/∂y.

4296 7 [Turn over]

 (c)

y   y  

P  

Φ  
Φ   Φ0  
x   x  
O   A   O  

Fig.  a   Fig.  b  

−ydx + xdy
I I
Fdr =
x2 + y 2
Let x = ρ cos φ and y = ρ sin φ where (ρ, φ) are polar coordinates. We then have
dx = −ρ sin φdφ + dρ cos φ and dy = ρ cos φdφ + dρ sin φ, so that
−ydx + xdy
= dφ .
x2 + y 2
As illustrated on Fig. a, for a closed curve surrounding the origin, φ = 0 at the
intersection of the curve and the x-axis, A, and φ = 2π after a complete circuit back to
A. Therefore the line integral
I Z 2π
Fdr = dφ = 2π
0

in that case. On the other hand, for a closed curve not surrounding the origin (Fig. b),
φ = φ0 at some point P on the curve, and φ = φ0 as well after a complete circuit back
to P (see figure). Therefore in that case the line integral
I Z φ0
Fdr = dφ = 0 .
φ0

4296 8
 (d) The result obtained in (c) does not contradict the theorem because in this
case, M and N do not have continuous derivatives throughout any region containing
the origin (0, 0). So the theorem does not apply to these regions.

10. (a) c has the dimensions of a speed, and a the dimensions of a frequency squared.

(b) We inject u = Aei(kx−ωt+θ0 ) into the equation to obtain:

−ω 2 u + c2 k 2 u + au = 0

i.e. p
ω(k) = ± a + c2 k 2 .
ω is not a linear function of k, so the relation is dispersive.

(c) The solution u represent a monochromatic sinusoidal traveling wave with am-
plitude
√ A and phase θ. The phase speed is defined by θ = constant, i.e. cp = ω/k =
a + c2 k 2 /k.
√
(d) The group speed is cg = dω/dk = c2 k/ a + c2 k 2 .

(e) ω̄ and k̄ must be slowly varying functions of both space and time, i.e. they
must change very little over a few wavelengths or a few wave periods. However, they
may change considerably over many wavelengths or wave periods.

(f) Assuming slowly varying functions, the number of waves per unit time is given
by 1/τ = ω̄/(2π) and the number of waves per unit length by 1/λ = k̄/(2π). Writing
that waves are conserved (in the same way one would write mass conservation for a
fluid):
∂ k̄ ∂ ω̄
+ = 0.
2π∂t 2π∂x

(g) Substituting the expression obtained in (b) for ω̄ in the conservation equation
and using the chain rule yields:

∂ k̄ ∂ k̄
+ cg (k̄) = 0.
∂t ∂x
It then becomes clear that changes in k̄ (and ω̄) travel at the group speed. The group
speed is also the speed at which the wave energy propagates.

4296 9 [Turn over]

 11. (a) Let r = xi + yj + zk be the position vector to any point P (x, y, z) on the
surface φ. Then dr = dxi + dyj + dzk lies in the tangent plane to the surface at P .
Since φ(x, y, z) = c, we know that

∂φ ∂φ ∂φ ∂φ ∂φ ∂φ
dφ = dx + dy + dz = ( i + j+ k) · (dxi + dyj + dzk) = 0 ,
∂x ∂y ∂z ∂x ∂y ∂z
i.e. ∇φ · dr = 0, so that ∇φ is perpendicular to dr and therefore to the surface φ.

(b) Let rA = ai + bj + bk and rP = xi + yj + zk be the position vectors of A and

P respectively. We then have R = rP − rA = (x − a)i + (y − b)j + (z − c)k, so that
q
R= (x − a)2 + (y − b)2 + (z − c)2 .

It therefore follows that:

h i (x − a)i + (y − b)j + (z − c)k R
∇R = ∇ (x − a)2 + (y − b)2 + (z − c)2 = p = ,
(x − a)2 + (y − b)2 + (z − c)2 R

i.e. ∇R is a unit vector in the direction of R.

(c) Let R1 = AP and R2 = BP denote vectors drawn respectively from foci A

and B to any point P on the ellipse. Let T be a unit vector tangent to the ellipse at P ,
as shown on the figure. Since an ellipse is the locus of all points P such that the sum of
their distances to two fixed points A and B is a constant c, one can write the equation
of the ellipse as R1 + R2 = c. As proven in (a), ∇(R1 + R2 ) is normal to the ellipse,
hence we have:
[∇(R1 + R2 )] · T = 0
which can be rewritten as (∇R2 ) · T = −(∇R1 ) · T. Since we also know from (b) that
∇R1 and ∇R2 are unit vectors in the directions of R1 and R2 respectively, the cosine
of the angle between ∇R2 and T must be equal to the cosine of the angle between ∇R1
and −T. Hence the angles are themselves equal.

(d) If a source is emitting sound waves at focus A, these will be reflected from the
ellipse to focus B since the angles between AP and BP and the tangent to any point
P of the ellipse are identical.

4296 10 [LAST PAGE]