Problems for QCAPM

Questions and solutions compiled by Ka Man (Ambrose) Yim

1 Physical Analysis
1. ‘Lumpy String’. A massless string of tension T joins N masses distance l apart each having mass m and
is fixed at both ends. Each mass mp is allowed to move perpendicular to the string with displacement yp .
T
(a) Show that the expression ω02 = ml has dimension frequency squared
th
(b) Consider the p mass along the string and show that
d2
yp = ω02 (yp+1 − 2yp + yp−1 )
dt2
assuming small angle approximations are valid to the first order i.e. tan θ ≈ sin θ ≈ θ
(c) We want to investigate the case where the masses are vibrating at the same frequency ω. By considering
the ansatz, or trial solution, yp = −ω 2 yp show that
2ω02 − ω 2 yp+1 + yp−1
=
ω02 yp

(d) The left hand side of this equation is a constant independent of time. Try the ansatz yp = sin(pθ) and
find an expression for sin θ. By considering the boundary conditions y0 = yN +1 = 0, show that
nπ
θ=
N +1
 
kπ
ω = 2ω0 sin
2(N + 1)
where k = 1, 2, 3, · · · , N.
(e) For large N show that the solutions to ω is consistent with the picture of standing waves on a string
with fixed ends.

Resolve the forces on mass yp vertically:

T T
mÿp = (yp+1 − yp ) + (yp−1 − yp )
l l
T
ÿp = (yp+1 − 2yp + yp−1 )
ml
ÿp = ω02 (yp+1 − 2yp + yp−1 )

We want to obtain standing wave solutions where all masses are oscillating at the same angular frequency
ω = 2πf . Try the oscillatory ansatz ÿp = −ω 2 yp

−ω 2 yp = ω02 (yp+1 − 2yp + yp−1 )

2ω02 − ω 2 yp+1 + yp−1
=
ω02 yp

Try ansatz y = sin(pθ):

2ω02 − ω 2 sin(pθ) + sin(p − 1)θ
= = 2 cos θ
ω02 sin(pθ)

1
Applying the double angle formula:
ω
2 θ
2− = 2 − 4 sin2
ω0 2
θ ω
sin =
2 2ω0
θ
ω = 2ω0 sin
2
Boundary conditions:
y0 = 0 yN +1 = sin(N + 1)θ = 0
Therefore we obtain an expression for θ
nπ
θ=
N +1
where n = 1, 2, · · · , N . Thus the ‘normal mode’ frequencies of the string is
nπ
ωn = 2ω0 sin
2(N + 1)

For large N the lumpy string approximates a continuous string and we take small angle approximations
to first order:
nπ nπ nπ
ωn = 2ω0 sin ≈ 2ω0 = ω0
2(N + 1) 2(N + 1) N +1
Recall: r
T
ω0 =
ml
In a continuous string m can be expressed in terms of the mass per unit length ρ:
s
T
ω0 =
(ρl)l
q
T
But from the solution to the wave equation ρ is the speed of the wave c:

c
ω0 =
l
Hence
nπ nπ
ωn = c =c
l(N + 1) L
where L is the total length of the string. This is consistent with the picture of standing waves on a
string.

2
2. A metal chain of mass per unit length λ and length l is fixed horizontally at both ends and hangs freely
in equilibrium. The string makes an angle φ to the horizontal at the end points. Show that the equation of
the hanging chain satisfies the equation
s  2
dy
1+ = αy + constant
dx
Where α is a real constant. Furthermore, with a judicious choice of the coordinates of the midpoint of the
chain show that
1 1
y= (exp(αx) + exp(−αx)) = cosh(αx)
2α α
satisfies the equation. What happens to the centre of mass of the chain when a force is applied to it?

Consider a section of the string length s starting from the the midpoint where (x, y) = (0, 1). The
tension at the midpoint T0 acts horizontally. Let T be the tension at the other end. Resolving the forces
on this section both horizontally and vertically,

T cos φ = T0
T sin φ = gλs

Dividing the two equations,

dy gλs
tan φ = = = αs
dx T0
Differentiating both sides w.r.t. x gives

d2 y ds
=α
dx2 dx
s  2
d2 y dy
2
= α 1 +
dx dx

Spot the integral solution

s  2
dy d2 y dy dy
2
/ 1+ =α
dx dx dx dx
s  2
d dy dy
1+ =α
dx dx dx
s  2
dy
1+ = αy + constant
dx
dy
Fix the constant to be 0 such that at the midpoint, x = 0, y = 1, dx = 0. Thus
s  2
dy
1+ = αy
dx
 2
dy
= α2 y 2 − 1
dx
Since the chain is hanging in equilibrium, it must be at its lowest potential energy, i.e. its centre of
mass is as low as possible. Any distortion of the shape of the chain would displace it from its position
of lowest potential energy thus elevating its centre of mass.

3
3. Consider two concentric spherical conducting shells of radii a and b respectively, where b > a. Show that
the capacitance of this system is
ab
C = 4π0
b−a
A third spherical shell of radius c > b > a now encloses the two shells, also in a concentric configuration.
The inner (r = a) and outer (r = c) shells are raised to a potential V and the middle shell (r = b) is earthed.
Using the previous result, write down the capacitance of the new system.

Take the centre of the spheres to be the origin. In the region between the small sphere and the big
sphere (a ≤ r ≤ b), the charge on the big sphere contributes nothing to the electric field (Gauss’ Law)
and the electric field generated by the small sphere is identical to that of a point charge sitting at the
origin.
Q
E(r) = a≤r≤b
4π0 r2
The potential difference across the two spheres is simply
ˆ b
Q 1 1
Q b − a

Vab = − E(r)dr = − =
a 4π0 a b 4π0 ab

Recall that the definition of the capacitance is

Q
C=
V
Then the capacitance of this system is simply
ab
C = 4π0
b−a
If you think carefully about it, the three sphere construction is identical to a battery connected to two
bi-spherical capacitors Cab and Cbc in parallel. So the overall capacitance is simply the sum of the
bi-spherical capacitances.

ab bc b2 (c − a)
C = 4π0 + 4π0 = 4π0
b−a c−b (b − a)(c − b)

4
 22
4. ‘Statistical Signficance’. Consider an experiment measuring the rate of gamma decays of Na. Gamma
ray photons are collected and counted by a semiconductor detector.

Assume photons arrive at the detector at a constant rate dN dt = λ. The number of counts N after time t
λN e−λ
p
obeys Poisson statistics: probability of obtaining N counts given rate λ is p(N ) = N ! ; so σ(N ) = hN i
where σ(N ) is the standard deviation of N and hN i is the expectation of N (optional: prove this).

a)Taking σ(N ) to be our uncertainty in N , find an expression for the percentage uncertainty of N .

b)Background radiation also contributes to the number of counts. We could estimate the background
count B by running the experiment for the same duration with the gamma source 22 Na removed. Write
down an expression for our estimate of the signal count S in terms of N and B, assuming background
radiation and 22 Na gamma decays are independent. (optional: why might this assumption be invalid
for some nuclei?)

c)Find an expression for σ(S), assuming background radiation also follows Poisson statistics. [Hint: For
C = A + B, the rule for combining uncertainties is σ 2 (C) = σ 2 (A) + σ 2 (B).]

A signal can only be ‘seen’ if it sits above the background noise; if the signal count is smaller than the
fluctuation in background radiation, the signal cannot be reasonably distinguished from background noise.
We define the ‘significance’ of the signal to be
S
s= (1)
σ(B)

By considering the dependence of S and σ(B) on time, suggest how the significance of a signal could be
increased in the experiment. If s > 3 the signal is deemed to be statistically significant. Suppose the signal
to noise ratio is
S
f= (2)
B
Briefly justify why f is time independent. For values of f = 2.0, 1.0, 0.5, 0.1, 10−2 , calculate the threshold
count Nthresh needed to establish statistical significance i.e. s > 3.

p
[Derivation of σ(N ) = hN i is detailed in any standard statistics textbook.]
√
Percentage uncertainty is σ(N )/N = 1/ N . So the percentage uncertainty is reduced as the amount
of data collected N is increased.

Signal = Total count - background, so S = N − B.

[Certain types of nuclei might capture background radiation and go into excited states which then
decay, introducing a correlation between decay events and background radiation. These events are
usually quite unlikely and can be safely neglected in the case of 22 Na. ]
√
Uncertainty of S: σ 2 (S) = σ 2 (N ) + σ 2 (B) so σ(S) = N +B

5
 √
Statistical significance: s √
= S/σ(B) = S/ B. Photons arrive at a constant rate so S and B are both
√ to t, so s ∼ t - the statistical significance increases with time as more data is collected.
proportional
However, t increase far slower than t, so you are stuck with a law of diminishing returns when you try
to improve s by running the experiment for a longer duration.

[What is all this business about ‘5 sigma discoveries’ you hear about when a new particle is dis-
covered at the LHC? In experimental physics, it is important to distinguish between fluctuations
in background events and actual signals that are of interest to the experiment e.g. gamma ray
photons from 22 Na vs background radiation. A small peak might actually be a weird jump in random
background noise. Where do we draw the line? How much confidence do we place on a perceived ‘signal’ ?

We could decide whether a given peak with count N and known background B is a false positive using
the method of hypothesis testing. Suppose there is no radioactive source present, only background (the
null hypothesis). Given a probability distribution of random background fluctuations (e.g. a normal
distribution), the likelihood of background fluctuations producing the same count as the peak N can be
computed. Appealing to intuition, you already know that the further away N is from the background
count B, the less likely it is a background fluctuation. It is useful to consider how √ many standard
deviations N is from B; the ‘number of sigmas’ √ is simply s = (N − B)/σ(B) = S/ B. If N is 3
sigma away from the mean i.e. |N − B| > 3 B, the likelihood of the peak being a false positive is
only 0.03%. Scientists usually reject the null hypothesis (i.e. reject the notion that the result
√ is due
to background events) when N is five sigmas away from B - the chances
√ of |N − B| > 5 B being a
background event is 1 in 1744278! Signals with counts of |N −B| > 5 B are declared as as ‘discoveries’. ]

Since S and B are both proportional to time, their ratio is constant. Calculating Nthresh :

S = fB
S √
s= √ =f B>3
B
N = S + B = (f + 1)B > 9(f + 1)/f 2 = Nthresh

Note that Nthresh ∼ s2 . Tabulating values of Nthresh for different values of f :

f 2 1 0.5 0.1 0.01

Nthresh 5 18 54 900 9.09 × 104

For smaller f i.e. weaker signal compared to the background, more data is needed to establish statistical
significance.

Moral of the story: More data = more confidence in your science.

6
5. Planet Z is an Earth-like planet with radius R. At height h (h  R) from the planet’s surface, the
refractive index of the atmosphere n is modelled to be a linear function

n(h) = 1.01 − αh

where α is a constant. The intrepid X-man Cyclops arrives at the surface of Z and fires a laser beam from his
eyes, level to the surface of the planet. Unfortunately, Cyclops fails to take atmospheric effects into account.
The laser beam curves along the surface of the planet in a circle of radius R and hits Cyclops in the back of
his head. Show that
1.01
α=
R
by considering internal reflections along the path of laser.

Suppose the beam reaches a boundary at an infinitesimally small height h at which the refractive index
changes from n(0) = 1.01 to n(h) = 1.01 − αh and light is refracted critically i.e. along the boundary.
Then by Snell’s Law
1.01 cos θ = 1.01 − αh
where θ is the angle subtended between the boundary and the ray. From trigonometry,
R
cos θ =
R+h
h
Expand the fraction as an infinite series in ( R ):

h h h
cos θ = 1 − + O(( )2 ) ≈ 1 −
R R R
Substitute back into Snell’s Law:
h 1.01
1.01 − 1.01 = 1.01 = αh ⇒ α=
R R

7
2 Mathematical logic
1. The surface of a football is a tessellation of regular pentagons and hexagons. Prove that it must have 12
pentagons. Given that no pentagon is adjacent to another pentagon, deduce the minimum number of vertices
and hence the number of hexagons in that configuration. Hint: recall Euler’s formula for a polyhedron with
F faces, E edges and V vertices
V −E+F =2

Let H and P be the number of hexagons and pentagons respectively. Observe that each vertex is shared
by three faces and each edge is shared by two faces in the tessellation pattern. Thus deduce

F =H +P
6H + 5P 5P
V = = 2H +
3 3
6H + 5P 5P
E= = 3H +
2 2
Substitution into Euler’s formula gives P = 12 and all the H terms cancel i.e. The number of hexagons
can be arbitrary in the tessellation. If no pentagon is adjacent to another pentagon, then all the vertices
and edges are shared between pentagons and hexagons in the case of minimum vertices. Given we have
twelve pentagons, there will be minimum of 60 vertices. Substituting this into
5P
V = 2H +
3
gives 20 hexagons.

8
2. At what values of x is sin x zero? Given that the series expansion of sin x is

x3 x5 x2n+1
sin x = x − + − · · · + (−1)n−1 + ···
3! 5! (2n + 1)!

By considering the zeros of sin x and its series expansion , prove that

π2 1 1
= 1 + + + ···
6 4 9

The zeros of sin x are the integer multiples of π:

sin x = 0 ⇐⇒ x = 0, ±π, ±2π, ±3π, · · ·

We expand sin x as a product of linear terms:

sin x = λx(x − π)(x + π) · · · (x − nπ)(x + nπ) · · · = λx(x2 − π 2 ) · · · (x2 − n2 π 2 ) · · ·

We want to choose λ to ‘normalise’ the expression such that it is convergent and agrees with the series
expansion of sin x. If we choose λ to be a finite number, say 1, then the series does not converge for any
finite value of x. We choose
1
= (π)2 (2π)2 · · · (nπ)2 · · ·
λ
Substitute this into our infinite product:

sin x x2
x2

= 2 − 1 ··· − 1 ···
x π (nπ)2

To test the convergence of this product, we use the following mathematical theorem.
Theorem. If 0 < an < 1, then
∞
Y ∞
X
(1 − an ) converges ⇐⇒ an converges
n=1 n=1

x2
You can easily check that our infinite product converges by letting an = π 2 n2 . Consider
∞ ∞ ∞
X 1 X 1 X 1
2
= 1 + 2
< 1 +
n=1
n n=2
n n=2
n(n − 1)

We can ‘telescope’ the sum:

∞ ∞
X 1 X 1 1

= −
n=2
n(n − 1) n=1 n − 1 n
 1 1 1 1 1
= lim 1 − + − + + · · · −
n→∞ 2 2 3 3 n
1

= lim 1 − =1
n→∞ n
P∞ Q∞ 2
1
hence n=1 n2 < 2 and n=1 ( πx2 n2 − 1) converges.

9
If we multiply out the products and collect the terms, then we notice that the constant term agrees with
the series expansion of sinx x . From this we deduce we have chosen the right λ. The x2 coefficient from
expanding the brackets is simply
1 1 1

− 2 1 + + + ···
π 4 9
sin x
Equating it with the x2 coefficient of our series expansion of x we obtain

π2 1 1
= 1 + + + ···
6 4 9

10
3. ‘Keeping it up’. Suppose I play a game where each round I try to bounce a ping pong ball off a ping
pong bat and try to keep the ball up in the air for as many rounds as possible. In a simple model of my
performance, I could either have a good, bad or missed bounce in each round. If the previous round had a
good bounce, then the probabilities of having a good bounce, bad bounce or a missed bounce in the next
round are 2/3, 1/6, 1/6 respectively; however, if the previous round had a bad bounce, then the probabilities
of having a good bounce, bad bounce or a missed bounce in the next round are 1/6, 1/2, 1/3. Given the 0th
round always has a good bounce, what is the probability that the ball is still in the air in the N th round?
What happens when N tends to infinity? Hint: you might find it helpful to diagonalise a matrix.

Suppose I conduct Q number of trials. Let’s define a vector n(i) = (gi , bi , mi ) where (gi , bi , mi ) are the
numbers of games in the ith round with good, bad and missed bounces respectively, such that
 
Q
n(0) =  0 
0

After the 0th round, given the probabilities in the instructions we have
 
2/3
n(1) =  1/6  Q
1/6

and so forth.

We make a general observation that suppose there are gi games that have a good bounce in the ith
round, 23 of those games will be passed to gi+1 . Similarly 16 bi is also passed to gi+1 . We can write down

2 1
gi+1 = gi + bi
3 6
Similarly for the bi+1 and mi+1 ,
1 1
bi+1 = gi + bi
6 2
and
1 1
mi+1 = gi + bi + mi
6 3
in vector form, this is written as a matrix equation
    
gi+1 2/3 1/6 0 gi
 bi+1  =  1/6 1/2 0   bi  = M n(i)
mi+1 1/6 1/3 1 mi
th
Or more compactly, n(i + 1) = M n(i). [More formally: The P probability of
P being in the k state (i.e.
good, bad or missed) in the (i + 1)th round is n(i + 1)k = j Mkj n(i)j = j p(k|j)n(i)j ; Mkj = p(k|j)
is the matrix of conditional probabilities. ]
Extrapolating this relation to n(0), we have

n(N ) = M N n(0)

The matrix M can be diagonalised into M = P DP −1 where

11
 0 √1 − √12
 
2
P =  0 −1 − √1 −1 + √12 
2
1 1 1

1 1 1
 

P −1 =  − 21 + √1
2
− 12 0 
− 21 − √1
2
− 12 0
 
1 0√ 0
D= 0 1
− 62 0√ 
 
2
0 0 1
2 + 62

Observe that M N = (P DP −1 )(P DP −1 ) · · · (P DP −1 )(P DP −1 ) = P DN P −1 . Thus we could calculate

n(N ) and thus obtain mN :
√ √
1 1 1 2 N 1 1 1 2 N
mN /Q = 1 + (− + √ )( − ) − ( + √ )( + )
2 2 2 6 2 2 2 6

The probability that the ball is still in the air in the N th round is
√ √
1 1 1 2 N 1 1 1 2 N
pN = 1 − mN /Q = ( + √ )( + ) + ( − √ )( − )
2 2 2 6 2 2 2 6

As N → ∞, pN → 0.

12