Kevin Zhou Physics Olympiad Handouts

Waves III: Specific Waves

Sound waves are covered in chapter 19 of Halliday and Resnick, while light waves are covered in
chapters 39 and 44. For more about light waves, see chapter 9 of Purcell. For water waves, and many
other neat wave phenomena, see chapter I-51 of the Feynman lectures and chapter 7 of Crawford’s
Waves. For more about polarization, see chapter I-33 of the Feynman lectures, or for more detail,
chapter 8 of Hecht’s Optics. Basic geometrical optics is covered in chapter 40 of Halliday and
Resnick. There is a total of 77 points.

1 Sound and Longitudinal Waves

[4] Problem 1. In this problem, you’ll work through Newton’s slick derivation of the speed of sound.
This derivation avoids thinking about how individual parcels of gas move, which can be confusing,
by instead considering the motion of a piston at the end of the gas.

(a) Consider a cylinder of gas of length L and area A, closed on one end with a movable piston
on the other end. Suppose the gas exerts a force F on the piston when in equilibrium. We
may define an equivalent spring constant by K = −dF/dL. Show that for the gas,
dp
K = −A2 .
dV

(b) Argue that the speed v of longitudinal waves on a spring of mass M , spring constant K, and
length L obeys
KL2
v2 = .
M
(c) If we assume the sound waves are adiabatic, show that for the gas,
γp
v2 = .
ρ
Check this answer is reasonable by evaluating the result for air. If each gas molecule has mass
m, write the result in terms of γ, T , and m.

Next, we consider some limitations of this result.

(d) In an ideal gas, we assume the particles are noninteracting: they pass right through each
other. But for sound waves to propagate, adjacent packets of ideal gas must exert pressure
on each other. How is this possible? Use this observation to estimate the maximum possible
frequency of sound in a gas in terms of the number density n = N/V = P/kB T , the radius r
of a gas molecule, and the speed of sound v.

(e) Our analysis also breaks down if the pressure variations are no longer adiabatic. The rate of
heat conduction in a gas with thermal conductivity kt across a surface of area A is
dQ dT
= −Akt
dt dx
Taking a sinusoidal temperature variation, show that the adiabatic approximation holds when
ω ≪ pkB /mkt . Is this approximation good for audible sound in air, where kt ≈ 25 mW/m K?

1
 Kevin Zhou Physics Olympiad Handouts

(f) In a more traditional derivation of the speed of sound, such as the one in Halliday and Resnick,
one would show that
B
v2 =
ρ
where B is the bulk modulus, the pressure per fractional change in volume,
dP
B = −V .
dV
Check that this result is compatible to your result in (c) for adiabatic compression.

Liquids and solids typically have a much higher B and ρ than gases, and B is high enough so that
the speed of sound in liquids and solids is typically greater as well.

Remark
Phase shifts upon reflection for sound waves can be a bit tricky. Recall from W1 that a hard
boundary for a transverse string wave y(x, t) sets y to zero. As a result, upon reflection, y
flips sign, but vy = ∂y/∂t stays the same.

When a sound wave hits a hard wall, the wall sets the displacement ξ(x, t) to zero. Then
upon reflection, the displacement flips sign, while the pressure variation δP (x, t) ∝ ∂ξ/∂x
stays the same. In standing waves, a hard wall is thus a node for ξ and an antinode for δP .
Similarly, when sound waves in a tube reflect off an open end, the end sets δP to zero, so it
flips sign. An open end is thus a node for δP and an antinode for ξ.

The rule is always the same: whatever quantity gets fixed to zero by the boundary gets
flipped in sign upon reflection, and for a standing wave, that quantity has a node at the
boundary. But it’s confusing enough that several common high school textbooks get it wrong.
Some even state, in their confusion, that “hard boundaries flip transverse waves but not
longitudinal ones”, which is definitely not true in general.

[2] Problem 2 (HRK). Some conceptual questions about sound waves.

(a) What is larger for a sound wave, the relative density variations ∆ρ/ρ or the relative pressure
variations ∆P/P ? Or does it depend on the situation?

(b) What is larger, the velocity of a sound wave v or the amplitude of the velocity variations ∆u
of the underlying particles? Or does it depend on the situation?

[2] Problem 3. A rubber rope with unstretched length L0 is stretched to length L > L0 .

(a) Find the ratio of the speeds of transverse and longitudinal waves.

(b) Experimentally, it is found that the longitudinal waves are much more strongly damped. (You
can check this at home, by making such a rope by tying together cut rubber bands.) Can you
explain why, by considering the molecular structure of rubber?

2
Kevin Zhou Physics Olympiad Handouts

Idea 1: Doppler Effect

Working in one dimension with speed of sound c, if a source of sound at frequency f0 travels
at velocity vs while an observer to their right travels at velocity vo , the observed frequency is
c − vo
f= f0 .
c − vs

Example 1

A speaker is between two perfectly reflective walls and emits a sound of frequency f0 . If you
carry the speaker and walk with small speed v towards one of the walls, what do you hear?

Solution
Let’s work to lowest order in v/c everywhere. The wall you’re walking toward experiences a
sound of approximate frequency f0 (1 + v/c) by the Doppler effect, and this is the frequency
it reflects. Since you’re walking towards the wall, a second Doppler effect occurs, causing you
to hear frequency f0 (1 + 2v/c). (We also saw this “double Doppler shift” back in R1.) By
similar reasoning, you hear sound of frequency f0 (1 − 2v/c) from the wall behind you. Since
v/c is small, what this actually means is that you’ll perceive a single tone of frequency f0 ,
but with “beats” of frequency 4v/c.

Example 2

In my former college at Oxford, there is a long staircase that is said to “quack” when one
claps at it. What is the explanation of this phenomenon?

Solution
A diagram of the staircase is given below, courtesy of Felix Flicker, fellow of New College.

The key is that each clap reflects off a stair individually. When the echoes arrive back at the
listener, they arrive quickly enough to be heard as a pitch.

The width and height of the steps are w = 30 cm and h = 16 cm. Suppose one claps at a
distance L ≫ w, h. The path length differences for reflections off the bottom few steps are

3
 Kevin Zhou Physics Olympiad Handouts

approximately 2w, giving the frequency

v
f= = 570 Hz
2w
where we used v = 343 m/s. The quack then continues, due to reflections off higher and
higher stairs. Once the stairs are much further
√ away than L, path length differences for
subsequent reflections are approximately 2 w + h2 , giving frequency
2

v
f= √ = 500 Hz.
2 w2 + h2
Hence the quack consists of a pitch that starts high and then falls slightly lower as it fades
away. For further discussion, see the article How the Mound got its Quack .

[3] Problem 4. 01mƒ

USAPhO 1998, problem B1. (The official solution has a qualitatively incorrect

01W
answer for the final part of the problem; see Stefan Ivanov’s errata for the correct answer.)

[3] Problem 5. USAPhO 2016, problem A1.

[2] Problem 6. Some problems about sound waves in everyday life.

(a) Get a coffee cup with a handle and tap on the rim with a spoon. You will hear two distinct
pitches, e.g. if you tap directly above the handle, or 45◦ away from this point. Investigate
what happens for different angles. Can you explain why this happens?

(b) The octave key on an oboe forces the resonant mode from the fundamental to the first overtone,
doubling the frequency of the note produced. It does this by opening a small hole on the back
of the clarinet. Should this hole be placed at a pressure node or antinode for the fundamental,
or somewhere else entirely?

(c) According to introductory textbooks, the fundamental mode for a pipe of length L and radius
r ≪ L, closed at one end and open at the other, has wavelength 4L. In reality, it’s a little bit
different because the radius is nonzero. Is the wavelength actually higher or lower than 4L?

01hƒ
(d) Find a way to produce beats in real life.

[3] Problem 7. BPhO 2008, problem 2.

2 Polarization
Now we’ll introduce polarization for light waves, putting the results of E7 to work.

Idea 2
The polarization of a light wave refers to the direction of its electric field; the light waves
we saw in E7 were linearly polarized. For example, a light wave traveling along ẑ with its
polarization an angle θ from the x-axis has electric field

Ex (z, t) = (E0 cos θ) cos(kz − ωt), Ey (z, t) = (E0 sin θ) cos(kz − ωt).

4
 Kevin Zhou Physics Olympiad Handouts

A polarizer lets only light of a certain linear polarization through; if light with a linear
polarization an angle θ from this axis passes through it, then a fraction cos2 θ of the energy is
transmitted. Just as light can be incoherent, it can be unpolarized; unpolarized light hitting
a polarizer loses half its energy.

[3] Problem 8. A simple polarizer contains many very thin, closely spaced wires. If the wires are
vertical, they block vertical electric fields, allowing only horizontally polarized light to go through;
this is a horizontal polarizer. One can similarly make diagonal and vertical polarizers.

(a) Suppose that perfectly monochromatic, but unpolarized light is incident on a double slit. (In
this case, assume “unpolarized” means that at each instant in time, the polarization of the
light passing through each slit is the same, but over longer timescales that polarization can
vary.) What does the intensity pattern on the screen look like?

(b) Next, suppose a vertical polarizer is placed in front of one slit, and a horizontal polarizer is
placed in front of the other slit. Now what does the intensity pattern look like?

(c) Finally, we further modify the setup of part (b) by placing many diagonal polarizers, at 45◦
to the vertical and horizontal, right in front of the screen. What does the intensity pattern
on the screen look like now?

(d) On an unrelated note, suppose we wish to rotate the polarization of linearly polarized light by
using N ≫ 1 intermediate polarizers. What’s the best way to do this, and what’s the fraction
of light that passes through the stack?

Idea 3
For a plane wave propagating along the z-axis with general polarization, it’s useful to write
 
E(z, t) = Re E0 ei(kz−ωt)

where E0 is a complex two-component vector, describing both its amplitude and polarization.
For example, if E0 = E0 x̂, the light wave is horizontally polarized, if E0 = iE0 x̂, it’s
horizontally polarized with a phase shifted by π/2, if E0 = E0 ŷ it’s vertically polarized, and
if E0 = E0 (x̂ + ŷ)/2 it’s diagonally polarized.

When linear polarizations are combined with a relative phase, the result√is circular (or more
generally, elliptical) polarization. For example, when E0 = E0 (x̂ + iŷ)/ 2, we have

E0 E0
Ex (z, t) = √ cos(kz − ωt), Ey (z, t) = √ sin(kz − ωt)
2 2
which is a circularly polarized light wave; the electric field at a fixed point rotates in a circle
over time, and if one draws the electric field vectors in a line along k̂, they trace out a spiral.
Birefringent materials, which have different indices of refraction in different directions, cause
such phase shifts, and thus can convert linear polarizations into other polarizations.

5
 Kevin Zhou Physics Olympiad Handouts

Example 3

A plane wave with amplitude E0 = E0 x̂ enters a linear optical device, which does not absorb
or reflect any energy.
√ When the plane wave exits the device, it has circular polarization,
E0 = E0 (x̂ + iŷ)/ 2. What does the device do to light with vertical polarization?

Solution
Vertical polarized √
light has to exit with circular polarization of the other handedness, i.e. with
E0 = E0 (x̂ − iŷ)/ 2, because this is the only possibility consistent with energy conservation.

To see this, note that the energy of a light wave is proportional to the time-averaged value
of |E|2 , which is turn proportional to |E0 |2 . Since horizontal and vertical polarizations are
orthogonal, they don’t interfere, so sending in both a horizontal and vertical light wave of
amplitude E0 at the same time just doubles the input energy. This must also double the
output energy, and indeed, under the above ansatz we have
x̂ + iŷ x̂ − iŷ √
x̂ + ŷ → √ + √ = 2 x̂
2 2
which indeed has double the energy of one wave by itself.

The more general principle here is that, since x̂ and ŷ were orthogonal to each other, they
must be mapped to two other unit vectors which are still orthogonal, as complex vectors.
That is indeed true, because

(x̂ + iŷ)† (x̂ − iŷ) = x̂ · x̂ + i2 ŷ · ŷ = 0.

By the way, now that we know what the device does to both horizontal and vertically polarized
light, we also know what it does to any polarization of light, by superposition.

[2] Problem 9 (MPPP 127). A birefringent material is placed between two orthogonal polarizers. The
material has thickness d, and has an index of refraction of n1 for light linearly polarized along the
axis e, and n2 for light polarized about an orthogonal axis.

If the system is illuminated with light of wavelength λ, give a value for d and orientation of e that
maximizes the transmitted light.

[2] Problem 10 (MPPP 128). In the first 3D movies, spectators would wear glasses with one eye
tinted blue and the other tinted red. This was quickly abandoned in favor of a system that used

6
 Kevin Zhou Physics Olympiad Handouts

the polarization of light.

(a) If you wear an old pair of 3D movie glasses, close one eye, and look in the mirror, then
you can only see the open eye. Explain how these glasses employ light polarization. What
disadvantages might this system have?

(b) If you wear a new pair of 3D movie glasses and do the same, then you can only see the closed
eye. Explain why.
[2] Problem 11 (HRK). A quarter-wave plate is a birefringent plate that causes a π/2 phase shift
between light polarized along e and perpendicular to e. Similarly, a half-wave plate causes a π
phase shift. Suppose you are given an object, which may be a quarter-wave plate, a half-wave plate,
a linear polarizer, or just a semi-opaque disk of glass. How can you identify the object? You can use
an unpolarized light source, and any number of polarizers and quarter-wave and half-wave plates.
[2] Problem 12 (HRK). A polarizer and a quarter-wave plate are glued together so that, if the
combination is placed with face A against a shiny coin, the face of the coin can be seen when
illuminated by light of appropriate wavelength. When the combination is placed with face A away
from the coin, the coin cannot be seen. Which component is on face A and what is the relative

01T†
orientation of the components?

[4] Problem 13. IZhO 2021, problem 3. A problem on the propagation of light through a
waveguide, unifying material from E7 and W1.

3 Water Waves
Water waves are the most familiar examples of waves in everyday life, but you won’t find them
mentioned often in introductory textbooks, because they’re far more complicated than any other
kind of wave we’ll consider. In all the problems below, we will completely neglect viscosity, surface
tension, and compressibility of the water. Despite this, our results will still only be approximate.
[4] Problem 14. Consider water with depth d and density ρ. A shallow water wave (i.e. one with
wavelength much greater than d) travels along the x-direction with height h(x, t) ≪ d relative to
the water level. It turns out that the water has a horizontal velocity v(x, t) which is approximately
independent of height, and negligible vertical velocity. Even though the water is moving, the
hydrostatic pressure formula still works because the water’s vertical acceleration is negligible.

(a) Find a relation between the derivatives of h(x, t) and v(x, t) using conservation of mass. Using
this result, show that the phase velocity vw of the water waves obeys v ≪ vw .

(b) Find a relation between the derivatives of h(x, t) and v(x, t) using force and momentum.

(c) Combining these results, find the phase velocity vw of shallow water waves.

Now let’s consider what happens when a shallow water wave created at sea approaches the shore,
and the depth d slowly decreases.

(d) Explain why waves always arrive at the shore moving perpendicular to the shoreline.

(e) If the depth is gradually halved, by what factor is the height of the wave multiplied? This
phenomenon is known as shoaling.

7
 Kevin Zhou Physics Olympiad Handouts

Remark
Textbooks commonly say that liquids can’t support transverse waves, because they don’t
support shear stresses. But the waves considered in problem 14 are clearly transverse. This
is possible because the textbook statement only applies to the internal forces of water alone.
At the surface of the water, gravity provides the transverse restoring force; that’s why these
waves are also commonly called “gravity waves”.

[2] Problem 15 (Japan). In the above problem, we considered shallow water waves, d ≪ λ, in which
case the motion of the water is approximately horizontal. But in general, it turns out that the
motion of the water will depend on height, and also that the individual wave molecules move in
ellipses. (This happens even when the wave has small amplitude, which we will always assume; the
motion for large amplitude is even more complicated.) In the limit of deep water waves, d ≫ λ, the
water molecules move in circles. Using this fact, we can quickly derive the wave speed.
Assume the water molecules at the surface of the wave move in uniform circular motion with
radius a and angular velocity ω, as shown.

(a) Consider the frame of reference moving to the right with velocity v. In this frame, the surface
of the water is completely stationary, while molecules travel along the surface. Consider a
small parcel of water which travels from a valley to a peak. By applying conservation of
energy, derive a relationship between v, ω, and g.

(b) Find the phase and group velocity of the wave, in terms of g and the wavenumber k. In
addition, find the condition on a and k for this derivation to make sense.

Showing that circular motion actually occurs takes more work, and involves solving partial differential
equations; you can find a complete derivation here or here.
As you can see, water waves are quite complex. A diagram of the speeds of nine different limiting
cases of water waves can be found in section 8.4 of The Art of Insight.

4 Reflection and Refraction

Now we’ll introduce reflection and refraction with some real-world applications.

8
 Kevin Zhou Physics Olympiad Handouts

Idea 4
If a wave hits an interface, while traveling at an angle θ1 to the normal to the interface,
then it will generically both reflect and refract. The angle of the reflected ray is θ2 = θ1 , and
the angle of the refracted ray obeys n1 sin θ1 = n2 sin θ2 . If there is no solution for θ2 in the
latter equation, then only reflection occurs.

These results follow directly from Huygens’ principle, so they are very general, applying to
light waves, sound waves, water waves, and so on, as long as the index of refraction ni is
always defined to be inversely proportional to the wave speed in each medium.

[2] Problem 16. Some conceptual questions about reflection and refraction.
(a) Does the index of refraction determine the phase velocity or the group velocity?
(b) Does a light beam of finite width get wider or narrower upon passing from air to water?
Assume the light enters at an angle to the normal.
(c) Three mutually perpendicular mirrors intersect so as to form an internal right-angled corner.
If a light ray strikes all three mirrors, show that it ends up traveling exactly opposite to its
original direction. Can you think of a practical application of such a “corner reflector”?

Example 4

Let the index of refraction at height h above the Earth’s surface be n(h). In terms of n(0)
and the Earth’s radius R, what should dn/dh be at the surface so that light rays orbit in
circles around the Earth, with constant height?

Solution
First, let’s ignore the curvature of the Earth. Consider a light ray moving slightly upward,
at a small angle θ to the horizontal, experiencing index of refraction n. Over a horizontal
distance L, it goes up by a height Lθ. At this point, it will have a different angle θ′ to the
horizontal, and experience index of refraction n + Lθ dn/dh. Snell’s law says
 
dn
n cos θ = n + Lθ cos θ′
dh
and expanding to lowest order in the small angles θ and θ′ gives
n ′2 dn
(θ − θ2 ) = Lθ .
2 dh
Approximating again to lowest order gives
L dn
θ − θ′ ≈ − .
n dh
Thus, the light ray turns through an angle of (1/n) dn/dh per unit horizontal distance. For
the light ray to stay at a constant height over the curved Earth, this must equal 1/R, giving
dn n(0)
=− .
dh R

9
 Kevin Zhou Physics Olympiad Handouts

More generally, this calculation shows that light bends towards the direction with higher n.
In the case of air, where n − 1 ≪ 1, we can rewrite this as

d(n − 1) 1
≈−
dh R
which can plausibly occur on Earth, due to the nice coincidence that n − 1 and H/R (where
H is the typical scale height of the atmosphere) are both of order 10−3 .

Remark: Mirages

There are two classes of mirages.

• When dn/dh < 0, light rays bend down. If there is a distant object at the horizon, its
image will appear above the horizon. This is called a “superior” mirage.

• When dn/dh > 0, light rays bend up. Then a distant object at the horizon will appear
below the horizon, forming an “inferior” mirage. This also applies to the sky near the
horizon, producing the illusion of water on the ground sometimes seen in deserts.

In air, the refractive index is close to 1, and n − 1 ∝ ρ ∝ P/T , where ρ is the air density
and the second step used the ideal gas law. Usually we have dρ/dh < 0, since dP/dh < 0 in
hydrostatic equilibrium, but it depends on the value of dT /dh.

• In normal conditions, the Sun warms the ground and the hot air rises and adiabatically
mixes the atmosphere (as discussed in T1), so that dT /dh < 0. This partially cancels the
effect of the pressure variation, so that dn/dh is still negative but has small magnitude,
so that mirage effects aren’t apparent.

• In rare “thermal inversion” conditions, we have dT /dh > 0, so that dn/dh is negative
with large magnitude, leading to strong superior mirage effects. If dn/dh is negative
enough, it can match the value computed in example 4, allowing an observer to see
arbitrarily far along the horizon despite the curvature of the Earth. This was the reason
the famous Bedford Level experiment concluded the Earth was flat.

• In hot deserts, the air near the ground is very hot, so that dT /dh < 0 with a large
magnitude. (A strongly negative dT /dh also occurs in cold days above water, since the
water stays warmer than the air above it.) Here the temperature gradient overpowers
the pressure gradient, so that dn/dh > 0 and inferior mirages can occur.

Proponents of the flat Earth hypothesis claim that the Earth only seems curved due to
atmospheric refraction. But they have it backwards: in almost all conditions dn/dh < 0,
which makes the Earth look less curved than it actually is.

[4] Problem 17. 01T† IPhO 1995, problem 2. Refraction in the presence of a linearly varying wave
speed. (This is a classic setup with a neat solution, also featured in IPhO 1974, problem 2.)

01hƒ
[3] Problem 18. INPhO 2019, problem 1. Another exercise on refraction, with an uglier solution.

[3] Problem 19. IPhO 2003, problem 3B. An exercise on refraction and radiation pressure.

10
 Kevin Zhou Physics Olympiad Handouts

[4] Problem 20. 01T† IPhO 1993, problem 2. Another exercise on the same theme.

5 Ray Tracing
Idea 5
A pointlike object emits light rays in all directions. When those light rays subsequently
converge at some other point, that point is the object’s real image. If they don’t actually
converge, but all propagate outward with a common center, that point is the object’s virtual
image. In general, if we’re given that an image exists, we can find its location by following
the paths of selected rays from the object and looking for intersections.

[2] Problem 21. A pinhole camera is a simplified camera with no lens. It simply consists of a box
with a small hole (the “aperture”). An image of the outside appears on the inside of the box (the
“screen”), opposite the hole.

(a) Explain how the pinhole camera works by ray tracing.

(b) What are the disadvantages of having a larger or smaller aperture?

(c) Assuming the object being photographed is very bright, estimate the optimal aperture size
for taking a clear picture with a pinhole camera, for a box of side length L.

The next three problems will exercise your intuition with real-world examples.

[2] Problem 22. AuPhO 2020, section C.

[2] Problem 23. AuPhO 2013, problem 11. Write your answers on the official answer booklet.

[3] Problem 24. AuPhO 2019, problem 12. Write your answers on the official answer booklet.

[3] Problem 25. 01^‚ IZhO 2020, problem 1.3. A test of your intuition for 3D ray tracing.

Idea 6
Conic sections have some simple properties under reflection.

• Light rays emitted from one focus of an ellipse will all be reflected to its other focus.

• Light rays emitted from one focus on a hyperbola will all be reflected so that the resulting
rays all travel radially outward from the other focus.

• Parallel light rays entering a parabola along its symmetry axis (i.e. the axis perpendicular
to the directrix) will all be reflected to its focus.

In the language of idea 5, if the foci of an ellipse/hyperbola are called F1 and F2 , then
an object at F1 produces a real/virtual image at F2 . Note that a parabola is simply an
ellipse in the limit where F1 becomes very far away, so that rays coming in from F1 become
approximately parallel.

11
 Kevin Zhou Physics Olympiad Handouts

[2] Problem 26 (Povey). The mirascope is a toy consisting of two parabolic mirrors, pointing toward
each other, so that the focus of each one is at the vertex of the other.

(a) When an object is placed at the bottom vertex, a real image appears at the top vertex. Why?

(b) How is the image oriented relative to the object?

The real image made by this setup is very convincing. There’s a Michelin starred restaurant that
uses it in a course: when you reach for what looks like the food, your hand just passes through air.

Idea 7: Paraxial Approximation

If a light ray hits a thin lens of focal length f at a shallow angle, and at a distance y ≪ f
above the lens’s center, then it will exit the lens bent vertically by an angle ±y/f , where
the sign depends on whether the lens is converging or diverging. (For example, any light ray
going straight through the lens’s center isn’t bent at all.) This is the paraxial approximation,
which only holds for light rays incident at shallow angles near the center of the lens.

Conversely, if you don’t know the focal length of a system, you can use this idea to find it.
For example, the lensmaker’s equation, giving the focal length of a lens of radii of curvature
R1 and R2 and thickness d, is
 
1 1 1 (n − 1) d
= (n − 1) − +
f R1 R2 nR1 R2

and can be derived by computing the bending of the light ray at each interface.

Example 5

An object is placed a distance o behind a thin converging lens with focal length f .

An image is formed a distance i in front of the lens. How are o, i, and f related?

12
 Kevin Zhou Physics Olympiad Handouts

Solution
The horizontal light ray goes straight through, so let’s consider another light ray which
emerges at a small angle θ1 to the horizontal. Then we read off
y y
θ1 ≈ , θ2 ≈
o i
but their sum is the deflection y/f , from which we conclude
1 1 1
+ = .
o i f
This is the familiar thin lens equation.

[2] Problem 27. Some basic questions about ray tracing.

(a) Show that if you have height h, you can see your entire body in a vertical mirror of length
h/2 whose top is at eye level, no matter how close or far you stand from it. Nonetheless, when
people want to look at their whole body in the mirror, they typically stand back. Why?

(b) A candle is placed behind a converging lens. An image is formed on a screen on the other side
of the lens. Now suppose that the top half of the lens is covered with a black cloth. Describe

01^‚
how the image changes.

[3] Problem 28. USAPhO 2024, problem A3. A series of optics exercises relevant for real cameras.

Idea 8: Fermat’s Principle

For fixed starting and ending points, light always takes the path of least time. This implies
that if light from point P is all focused at point P ′ , then all the relevant paths from P to
P ′ take the same time. This principle is completely equivalent to the laws of reflection and
refraction above, but may be more useful in certain situations.

[2] Problem 29. Parallel light rays coming in along the +x̂ direction enter a lens of index of refraction
n, whose left edge is at x = 0 and whose right edge is described by the function x(y). If all the light
beams are to be focused at x = f , as shown at left below, what kind of curve does x(y) have to be?

You should find that x(y) is not an arc of a circle, which implies that a spherical lens will fail to
focus all incoming horizontal light to a point. Instead, we will get spherical aberration, as shown at
right above. However, most lenses are spherical because it’s easier to make them that way.

13