22.51 Problem Set 1 (due Mon, Sept.

10)
1 Least-Action Principle (25 pt)
Question: Action has the unit of energytime, therefore a plausible expression for the action
of a photon would be,
A =
_
x
2
(t
2
)
x
1
(t
1
)
hdt, (1)
because h, the energy of a photon, is an apparent energy scale. It is well-known that the
speed of light in a material is v = c/n where n 1 is the refraction index. Therefore,
dt =
ds
v
=
nds
c
A =
h
c
_
x
2
(t
2
)
x
1
(t
1
)
nds, (2)
where ds is the dierential distance, and we have used the knowledge that remains constant.
n1 n2
1
2
(x1,y1)
(x2,y2)
y
x
Figure 1: Least-Action Principle.
Now suppose we have an interface between two materials of constant refraction indices n
1
,n
2
,
prove Snells law,
n
1
sin(
1
) = n
2
sin(
2
), (3)
that is, such a path would minimize A among all feasible paths {x(t)} that start at x(t
1
) =
(x
1
, y
1
) and end at x(t
2
) = (x
2
, y
2
).
Answer: Let x = x
i
be the interface. All feasible light paths will intersect the interface
at some point, say (x
i
, y
i
). Among light paths that crosses a certain (x
i
, y
i
), straight lines
linking (x
1
, y
1
) to (x
i
, y
i
), and then (x
i
, y
i
) to (x
2
, y
2
) are the shortest, the total action for
1
which would be (we omit the h/c constant factor for the sake of simplicity),
A = n
1
_
(x
i
x
1
)
2
+ (y
i
y
1
)
2
+ n
2
_
(x
i
x
2
)
2
+ (y
i
y
2
)
2
. (4)
y
x
n1 n2
(x1,y1)
(x2,y2)
(xi,yi)
Figure 2: Least-Action Principle.
We then minimize with respect to y
i
, the only degree of freedom left,
n
1
(y
i
y
1
)
_
(x
i
x
1
)
2
+ (y
i
y
1
)
2
+
n
2
(y
i
y
2
)
_
(x
i
x
2
)
2
+ (y
i
y
2
)
2
= 0. (5)
This is just the Snells law.
Bonus Question (10 pt): Suppose we have a graded material with n(x, y) = 1 + x. Let
x(t
1
) = (0, 0), x(t
2
) = (0.5, 0.3), calculate the light path y(x).
Answer: The action as a functional of y(x) is,
A[y] =
_
x
2
x
1
nds =
_
x
2
x
1
n(x)
_
dx
2
+ dy
2
=
_
x
2
x
1
(1 + x)
_
1 + y
2
dx, (6)
0 = A =
_
x
2
x
1
(1 + x)
y

1 + y
2
dx,
=
_
x
2
x
1
_
d
dx
_
(1 + x)
y

1 + y
2
_

d
dx
_
(1 + x)y

1 + y
2
_
y
_
dx. (7)
The rst term always vanishes since,
y|
x
1
= 0, y|
x
2
= 0. (8)
2
The second term gives zero for all feasible y if and only if,
(1 + x)y

1 + y
2
= C, (9)
for all x, where C is a constant. This means,
y

1 + y
2
=
C
1 + x
, (10)
1
1 + y
2
= 1
C
2
(1 + x)
2
, (11)
y

_
1
1
C
2
(1+x)
2
1 =
C
_
(1 + x)
2
C
2
, (12)
Using Maple, one obtains,
0.3 = y(0.5) y(0) =
_
0.5
0
C
_
(1 + x)
2
C
2
C = .6296364651. (13)
> int(C/sqrt((1+x)^2-C^2),x=0..0.5);
C ( ) ln + 1.5 2.25 1. C
2
1. ( ) ln + 1. 1. ( ) C 1. ( ) + C 1. C
> fsolve(C*ln(1.5+sqrt(2.25-1.*C^2))-1.*ln(1.+sqrt(-1.*(C-1.)*(C+1.)
))*C - 0.3, C);
.6296364651
> C := .6296364651;
:= C .6296364651
> with(plots):
> p:= dsolve({D(y)(x) = C/sqrt((1+x)^2-C^2), y(0)=0}, type=numeric,
range=0..0.5):
> odeplot(p,[x,y(x)],0..0.5,refine=3);
0
0.05
0.1
0.15
0.2
0.25
0.3
y
0.1 0.2 0.3 0.4 0.5
x
Figure 3: Least-Action Principle.
3
2 Double Pendulum (25 pt)
Question: Lagrangian mechanics is formally equivalent to Newtonian mechanics (the free
body diagram and torque analysis approach), but tends to be less susceptible to human
mistakes, especially when the system is complex or has constraints. Consider a double
pendulum shown in Fig. 4, composed of two uniform rods of lengths l
1
, l
2
and total masses
m
1
, m
2
(mass is uniformly distributed on each rod),
g
(m1,l1)
(m2,l2)
Figure 4: Double Pendulum.
(a). Choose variables {q
k
} which describes the systems degrees of freedom most naturally.
(b). The gravitational constant is g. Express the potential energy V in {q
k
}.
(c). Express the kinetic energy T in {q
k
, q
k
}.
(d). Write down L = T V . Derive the equation of motion.
(e). What are the small oscillation normal-mode frequencies? Describe those modes in
physical terms.
(f). Determine the conjugate momenta {p
k
} from L.
(g). Write down the Hamiltonian H =

k
q
k
p
k
L. Reexpress H in {q
k
, p
k
} instead of
{q
k
, q
k
}.
(h). Rederive the equations of motion using q
k
= H/p
k
, p
k
= H/q
k
, and check
whether they are equivalent to (d).
(i). Rationalize (rederive) the equations of motion using Newtonian mechanics.
Answer:
(a).
1
,
2
.
4
y
x
1
2
Figure 5: Double Pendulum.
(b). The center of mass of the rst rod is at,
(x
1
, y
1
) =
_
l
1
2
sin
1
,
l
1
2
cos
1
_
. (14)
The center of mass of the second rod is at,
(x
2
, y
2
) =
_
l
1
sin
1
+
l
2
2
sin
2
, l
1
cos
1

l
2
2
cos
2
_
. (15)
Therefore the total potential energy is,
V = m
1
gy
1
+ m
2
gy
2
=
(m
1
+ 2m
2
)gl
1
cos
1
2

m
2
gl
2
cos
2
2
. (16)
(c). From (b), we have,
( x
1
, y
1
) =
_
l
1
2
cos
1

1
,
l
1
2
sin
1

1
_
, (17)
( x
2
, y
2
) =
_
l
1
cos
1

1
+
l
2
2
cos
2

2
, l
1
sin
1

1
+
l
2
2
sin
2

2
_
. (18)
Therefore the center of mass kinetic energy is,
T
1
=
m
1
2
( x
2
1
+ y
2
1
) +
m
2
2
( x
2
2
+ y
2
2
)
=
m
1
2
_
l
2
1
4

2
1
_
+
m
2
2
_
l
2
1

2
1
+
l
2
2
4

2
2
+ l
1
l
2
cos(
1

2
)

2
_
. (19)
In addition to the center of mass translational kinetic energies, there are also rotational
5
kinetic energies. The sum is,
T
2
=
I
1
2

2
1
+
I
2
2

2
2
=
m
1
l
2
1

2
1
24
+
m
2
l
2
2

2
2
24
. (20)
Therefore the total kinetic energy is,
T =
(m
1
+ 3m
2
)l
2
1

2
1
6
+
m
2
l
2
2

2
2
6
+
m
2
l
1
l
2
cos(
1

2
)

2
2
, (21)
(d).
L(
1
,
2
,

1
,

2
) =
(m
1
+ 3m
2
)l
2
1

2
1
6
+
m
2
l
2
2

2
2
6
+
m
2
l
1
l
2
cos(
1

2
)

2
2
+
(m
1
+ 2m
2
)gl
1
cos
1
2
+
m
2
gl
2
cos
2
2
. (22)
L

1
=
(m
1
+ 3m
2
)l
2
1

1
3
+
m
2
l
1
l
2
cos(
1

2
)

2
2
, (23)
L

1
=
m
2
l
1
l
2
sin(
1

2
)

2
2

(m
1
+ 2m
2
)gl
1
sin
1
2
(24)
=
d
dt
_
L

1
_
=
L

1
. (25)
L

2
=
m
2
l
2
2

2
3
+
m
2
l
1
l
2
cos(
1

2
)

1
2
, (26)
L

2
=
m
2
l
1
l
2
sin(
1

2
)

2
2

m
2
gl
2
sin
2
2
. (27)
=
d
dt
_
L

2
_
=
L

2
. (28)
(e). When
1
,
2
0, keeping only the linear order terms,
(m
1
+ 3m
2
)l
2
1

1
3
+
m
2
l
1
l
2

2
2
=
(m
1
+ 2m
2
)gl
1

1
2
, (29)
m
2
l
2
2

2
3
+
m
2
l
1
l
2

1
2
=
m
2
gl
2

2
2
, (30)
6
Or,
(2m
1
+ 6m
2
)l
2
1

1
+ 3m
2
l
1
l
2

2
= (3m
1
+ 6m
2
)gl
1

1
, (31)
2m
2
l
2
2

2
+ 3m
2
l
1
l
2

1
= 3m
2
gl
2

2
, (32)
To determine the normal modes, let us multiply (32) by and add it onto (31),
[(2m
1
+6m
2
)l
2
1
+3m
2
l
1
l
2
]

1
+[3m
2
l
1
l
2
+2m
2
l
2
2
]

2
= (3m
1
+ 6m
2
)gl
1

1
3m
2
gl
2

2
. (33)
For properly chosen , the coecient ratio of

1
,

2
on LHS would equal to the coecient
ratio of
1
,
2
on RHS, allowing one to consolidate a single equation out of two coupled
equations,
(2m
1
+ 6m
2
)l
2
1
+ 3m
2
l
1
l
2

3m
2
l
1
l
2
+ 2m
2
l
2
2

=
(3m
1
+ 6m
2
)gl
1
3m
2
gl
2

, (34)
for which there are two roots,
+
> 0 and

< 0. And the vibrational frequencies of the

two normal modes would then be,

_
(3m
1
+ 6m
2
)gl
1
(2m
1
+ 6m
2
)l
2
1
+ 3m
2
l
1
l
2

+
,
+
=

_
(3m
1
+ 6m
2
)gl
1
(2m
1
+ 6m
2
)l
2
1
+ 3m
2
l
1
l
2

. (35)
The low frequency (

) mode (3m
1
+ 6m
2
)gl
1

1
+ 3m
2
gl
2

2
corresponds to the case of

1
,
2
swinging in phase. The high frequency (
+
) mode (3m
1
+ 6m
2
)gl
1

1
+ 3m
2
gl
2

2
corresponds to the case of
1
,
2
swinging anti-phase.
(f).
p
1

L

1
=
(m
1
+ 3m
2
)l
2
1

1
3
+
m
2
l
1
l
2
cos(
1

2
)

2
2
, (36)
p
2

L

2
=
m
2
l
2
2

2
3
+
m
2
l
1
l
2
cos(
1

2
)

1
2
, (37)
from which we can solve for

1
,

2
(using Maple) as,

1
=
12l
2
p
1
+ 18l
1
cos(
1

2
)p
2
l
2
l
2
1
(4m
1
12m
2
+ 9m
2
cos
2
(
1

2
))
(38)

2
=
12l
1
m
1
p
2
36l
1
m
2
p
2
+ 18p
1
m
2
l
2
cos(
1

2
)
m
2
l
2
2
l
1
(4m
1
12m
2
+ 9m
2
cos
2
(
1

2
))
(39)
(g).
H

1
p
1
+

2
p
2
L
7
=
(m
1
+ 3m
2
)l
2
1

2
1
6
+
m
2
l
2
2

2
2
6
+
m
2
l
1
l
2
cos(
1

2
)

2
2

(m
1
+ 2m
2
)gl
1
cos
1
2

m
2
gl
2
cos
2
2
. (40)
Plugging (38),(39) into above, and using Maple,
H =
6m
2
l
2
2
p
2
1
6(m
1
+ 3m
2
)l
2
1
p
2
2
+ 18m
2
l
1
l
2
cos(
1

2
)p
1
p
2
m
2
l
2
2
l
2
1
(4m
1
12m
2
+ 9m
2
cos
2
(
1

2
))

(m
1
+ 2m
2
)gl
1
cos
1
2

m
2
gl
2
cos
2
2
. (41)
(h). It is straightforward to verify that

1
= H/p
1
,

2
= H/p
2
indeed agrees with (38),
(39). Similarly, one can verify that p
1
= H/
1
, p
2
= H/
2
indeed agrees with (25),
(28).
(i). That is pretty tricky!
3 Collision Cross-Section of Hard Spheres (25 pt)
Question: This course is mainly concerned with probing materials by radiation. Assume
both the probe and the target are hard spheres of mass m and radius a, and assume they
interact by frictionless elastic collision only.
m
m
m

probe out
probe in
target
Figure 6: Collision cross-section.
A central concept of the course is the double dierential cross-section d
2
/ddE, which has
the following interpretation: imagine immersing a single target sphere in a uniform ux f
of probing spheres, where f is the number of incoming probes per unit cross-sectional area
8
per unit time, and suppose the incoming probes all have the same kinetic energy E
0
, then
probabilistically two changes may occur to the probes after they leave the target area,
1. A probe could be deected by angle from the original incoming direction.
2. A probe could acquire a dierent energy E = E
0
.
Thus, per unit time, there is a certain number of incoming probes dN that are deected into
certain (, +d) that corresponds to solid angle d = 2d cos (we use d instead of dd
because there is axial symmetry) and with outgoing kinetic energies between (E, E + dE),
dN = f
d
2

ddE
ddE, (42)
and this can be taken as the denition of d
2
/ddE.
a. Assume the target is transxed (constrained to be immobile), calculate d
2
/ddE. You
are allowed to use the -function.
b. Assume the target is free and at rest before collision, calculate d
2
/ddE. You are
allowed to use the -function.
c. Verify that in both cases,
_ _
d
2

ddE
ddE = 4a
2
. (43)
Explain why this should be obvious.
Answer:
a. When the target is transxed, there is no energy transfer so E = E
0
. As for the momentum
transfer, there is a simple relationship between and the contact point characterized by the
contact angle ,
=

2
. (44)
At the contact, the Cartesian coordinate of the probes center is,
(x, y) = (2a cos , 2a sin ), (45)
so there is,
y = 2a sin
_

2
_
, (46)
9
y
x

target

y
Figure 7: Collision cross-section.
and therefore,
|dy| =

a cos
_

2
_
d

, (47)
which says that if the incoming probe happens to be inside the ring (y, y + dy) of thickness
dy, the outgoing angle will be between (, + d). That ring has dierential cross-section,
d = |2ydy| , (48)
to intercept the probe ux, and the corresponding outgoing solid angle is,
d = 2 sin d, (49)
therefore,
d
d
=
|2ydy|
|2 sin d|
=

2a
2
sin
_

2
_
cos
_

2
_
sin

= a
2
. (50)
The double dierential cross-section d
2
/ddE is formally,
d
ddE
= a
2
(E E
0
), (51)
and the total scattering cross-section of hard spheres is simply,
_ _
d
ddE
ddE = 4a
2
. (52)
b. When the target is free, there will be both momentum and energy transfer, and the
relationship between and is not straightforward. One way of solving () is to write
10
down the energy and momentum conservation equations with an additional constraint that
the momentum transfer is in the direction of contact, . A simpler method, however, is
by switching into a dierent inertial frame comoving with speed v
f
= v/2, where v is the
incoming probe speed in rest frame: E
0
mv
2
/2. In the comoving frame, the probe comes
in with speed v/2, and exits with speed v/2 in the direction

. The target does exactly the

opposite. And it is clear that in the comoving frame,
=

2
,

= 2, (53)
is still valid. The exit velocity of the probe in the comoving frame is therefore,
(v

x
, v

y
) = (v cos

/2, v sin

/2), (54)
and transforming back to the rest frame, it is,
(v
x
, v
y
) = (v
x
+ v
f
, v
y
) = (v(1 + cos

)/2, v sin

/2), (55)
therefore,
cos =
1 + cos

_
(1 + cos

)
2
+ sin
2

=
1 cos 2
_
(1 cos 2)
2
+ sin
2
2
= sin . (56)
As varies from 0 to /2, so does . There is,
d
d
=
|dy
2
|
|2d cos |
=

4a
2
d sin
2

2d cos

= 4a
2
cos , (0, /2). (57)
The exit energy is,
E =
m
2
(v
2
x
+ v
2
y
) =
1 + cos

2
E
0
= cos
2
E
0
, (58)
therefore,
d
ddE
= 4a
2
cos (E cos
2
E
0
), (59)
and,
_ _
d
ddE
ddE =
_
/2
0
2d cos (4a
2
cos ) = 4a
2
. (60)
c. Whenever a probe is within the cylinder of radius 2a to the target, its trajectory will be
inuenced. Therefore the total cross-section must be (2a)
2
= 4a
2
.
11
4 Poisson Bracket (25 pt)
Question:
a. Verify Jacobis identity:
{f, {g, h}} + {g, {h, f}} + {h, {f, g}} = 0. (61)
b. Use the above to show why if f and g are constants of motion, then {f, g} must also be.
c. Using (2.144), prove that,
{L
x
, L
y
} = L
z
, {L
y
, L
z
} = L
x
, {L
z
, L
x
} = L
y
. (62)
Answer: a. Using shorthand notations for partial derivative and index summation,
{g, h} = g
q
k
h
p
k
h
q
k
g
p
k
, {h, f} = h
q
k
f
p
k
f
q
k
h
p
k
, {f, g} = f
q
k
g
p
k
g
q
k
f
p
k
, (63)
{f, {g, h}} = f
q
l
(g
q
k
p
l
h
p
k
+g
q
k
h
p
k
p
l
h
q
k
p
l
g
p
k
h
q
k
g
p
k
p
l
)f
p
l
(g
q
k
q
l
h
p
k
+g
q
k
h
p
k
q
l
h
q
k
q
l
g
p
k
h
q
k
g
p
k
q
l
),
{g, {h, f}} = g
q
l
(h
q
k
p
l
f
p
k
+h
q
k
f
p
k
p
l
f
q
k
p
l
h
p
k
f
q
k
h
p
k
p
l
)g
p
l
(h
q
k
q
l
f
p
k
+h
q
k
f
p
k
q
l
f
q
k
q
l
h
p
k
f
q
k
h
p
k
q
l
),
{h, {f, g}} = h
q
l
(f
q
k
p
l
g
p
k
+f
q
k
g
p
k
p
l
g
q
k
p
l
f
p
k
g
q
k
f
p
k
p
l
)h
p
l
(f
q
k
q
l
g
p
k
+f
q
k
g
p
k
q
l
g
q
k
q
l
f
p
k
g
q
k
f
p
k
q
l
),
They indeed sum up to zero.
b. If f and g are constants of motion, then
{g, H} = 0, {H, f} = 0, (64)
and from (a) there is,
{f, {g, H}} + {g, {H, f}} + {H, {f, g}} = 0, (65)
therefore,
{H, {f, g}} = 0, (66)
which means {f, g} is also a constant of motion.
c. In Levi-Cevita notation,
L
i

ijk
r
j
p
k
, (67)
12
So,
{L
i
, L
i
} = {
ijk
r
j
p
k
,
i

k
r
j
p
k
}
=
ijk

k
{r
j
p
k
, r
j
p
k
}
=
ijk

k
(r
j
{p
k
, r
j
}p
k
+{r
j
, p
k
}p
k
r
j
)
=
ijk

k
(r
j

kj
p
k
+
jk
p
k
r
j
)
=
ijk

kk
r
j
p
k
+
ijk

j
p
k
r
j

= (
ii

jk

ik

i

j
)r
j
p
k
(
ii

kj

ij

ki
)p
k
r
j

=
ii
r
j
p
j
r
i
p
i

ii
p
k
r
k
+ p
i
r
i
= r
i
p
i
r
i
p
i
. (68)
For example, if i = x, j = y, then,
{L
x
, L
y
} = r
x
p
y
r
y
p
x
= L
z
. (69)
13