CHAPTER 12

Coupled Oscillations

12-1.
m1 = M m2 = M

k1 k12 k2
x1 x2

The equations of motion are

1 + (κ 1 + κ 12 ) x1 − κ 12 x2 = 0 
Mx
 (1)
Mx2 + (κ 2 + κ 12 ) x2 − κ 12 x1 = 0 


We attempt a solution of the form

x1 ( t ) = B1 e iωt 
 (2)
x2 ( t ) = B2 e iωt 

Substitution of (2) into (1) yields

(κ )
+ κ 12 − Mω 2 B1 − κ 12 B2 = 0 
1
 (3)

(
−κ 12 B1 + κ 2 + κ 12 − Mω 2 B2 = 0  )
In order for a non-trivial solution to exist, the determinant of coefficients of B1 and B2 must
vanish. This yields

 (  )  (
κ 1 + κ 12 − Mω 2  κ 2 + κ 12 − Mω 2  = κ 12
2
) (4)

from which we obtain

κ 1 + κ 2 + 2κ 12 1
ω2 = ± (κ 1 − κ 2 )2 + 4κ 122 (5)
2M 2M

This result reduces to ω 2 = (κ + κ 12 ± κ 12 ) M for the case κ 1 = κ 2 = κ (compare Eq. (12.7)].

397
398 CHAPTER 12

If m2 were held fixed, the frequency of oscillation of m1 would be

1
ω 01
2
= (κ 1 + κ 12 ) (6)
M
while in the reverse case, m2 would oscillate with the frequency
1
ω 02
2
= (κ 2 + κ 12 ) (7)
M
Comparing (6) and (7) with the two frequencies, ω + and ω − , given by (5), we find
1  
ω +2 = κ 1 + κ 2 + 2κ 12 + (κ 1 − κ 2 )2 + 4κ 122
2M  
1 1
> κ 1 + κ 2 + 2κ 12 + (κ 1 − κ 2 )  = (κ 1 + κ 12 ) = ω 021 (8)
2M M
so that
ω + > ω 01 (9)
Similarly,

ω −2 =
1 κ + κ + 2κ − (κ 1 − κ 2 )2 + 4κ 122 
2M  1 2 12

1 1
< κ 1 + κ 2 + 2κ 12 − (κ 1 − κ 2 )  = (κ 2 + κ 12 ) = ω 022 (10)
2M M
so that
ω − < ω 02 (11)
If κ 1 > κ 2 , then the ordering of the frequencies is
ω + > ω 01 > ω 02 > ω − (12)

12-2. From the preceding problem we find that for κ 12  κ 1 , κ 2

κ 1 + κ 12 κ 2 + κ 12
ω1 ≅ ; ω2 ≅ (1)
M M
If we use
κ1 κ2
ω 01 = ; ω 02 = (2)
M M
then the frequencies in (1) can be expressed as
COUPLED OSCILLATIONS 399

κ 12 
ω 1 = ω 01 1 + ≅ ω 01 (1 + ε 1 ) 
κ1 
 (3)
κ 
ω 2 = ω 02 1 + 12 ≅ ω 02 (1 + ε 2 ) 
κ2 
where
κ 12 κ
ε1 = ; ε 2 = 12 (4)
2κ 1 2κ 2
For the initial conditions [Eq. 12.22)],
x1 ( 0 ) = D, x2 ( 0 ) = 0, x 1 ( 0 ) = 0, x 2 ( 0 ) = 0 , (5)

the solution for x1 ( t ) is just Eq. (12.24):

ω + ω2  ω − ω2 
x1 ( t ) = D cos  1 t  cos  1 t (6)
 2   2 
Using (3), we can write
ω 1 + ω 2 = (ω 01 + ω 02 ) + ( ε1ω 01 + ε 2ω 02 )
(7)
≡ 2Ω + + 2 ε +

ω 1 − ω 2 = (ω 01 − ω 02 ) + ( ε 1ω 01 − ε 2ω 02 )
(8)
≡ 2Ω − + 2ε −
Then,
x1 ( t ) = D cos ( Ω + t + ε + t ) cos ( Ω − t + ε − t ) (9)

Similarly,
ω + ω2  ω − ω2 
x2 ( t ) = D sin  1 t  sin  1 t
 2   2 

= D sin ( Ω + t + ε + t ) sin ( Ω − t + ε − t ) (10)

Expanding the cosine and sine functions in (9) and (10) and taking account of the fact that ε +
and ε − are small quantities, we find, to first order in the ε’s,

x1 ( t ) ≅ D cos Ω + t cos Ω − t − ε + t sin Ω + t cos Ω − t − ε − t cos Ω + t sin Ω − t  (11)

x2 ( t ) ≅ D sin Ω + t sin Ω − t + ε + t cos Ω + t sin Ω − t + ε − t sin Ω + t cos Ω − t  (12)

When either x1 ( t ) or x2 ( t ) reaches a maximum, the other is at a minimum which is greater

than zero. Thus, the energy is never transferred completely to one of the oscillators.
400 CHAPTER 12

12-3. The equations of motion are

m 
x1 + x2 + ω 02 x1 = 0 
M
 (1)
m 
x2 + x2 + ω 02 x2 = 0 
M 
We try solutions of the form
x1 ( t ) = B1 e iωt ; x2 ( t ) = B2 e iωt (2)

We require a non-trivial solution (i.e., the determinant of the coefficients of B1 and B2 equal to
zero), and obtain
2

(ω 02 − ω 2 ) − ω 4  m  = 0
2
(3)
M
so that
m
ω 02 − ω 2 = ±ω 2 (4)
M
and then
ω 02
ω2 = (5)
m
1±
M
Therefore, the frequencies of the normal modes are

ω 02
ω1 =
m
1+
M
(6)
ω 2
ω2 = 0
m
1−
M
where ω 1 corresponds to the symmetric mode and ω 2 to the antisymmetric mode.
By inspection, one can see that the normal coordinates for this problem are the same as those for
the example of Section 12.2 [i.e., Eq. (12.11)].

12-4. The total energy of the system is given by

E = T +U
(1)
=
1
2
( )1
2
( 1
)
M x 12 + x 22 + κ x12 + x22 + κ 12 ( x2 − x1 )
2
2

Therefore,
COUPLED OSCILLATIONS 401

dE
= M ( x 1 x1 + x 2 
x2 ) + κ ( x1 x 1 + x2 x 2 ) + κ 12 ( x 2 − x 1 )( x2 − x1 )
dt

1 + κ x1 − κ 12 ( x2 − x1 )   x 1 +  Mx2 + κ x2 + κ 12 ( x2 − x1 )   x 2

=  Mx

1 + (κ + κ 12 ) x1 − κ 12 x2  x 1 +  Mx2 − κ 12 x1 + (κ + κ 12 ) x2  x 2
=  Mx (2)

which exactly vanishes because the coefficients of x 1 and x 2 are the left-hand sides of Eqs.
(12.1a) and (12.1b).
An analogous result is obtained when T and U are expressed in terms of the generalized
coordinates η1 and η2 defined by Eq. (12.11):

T=
1
4
(
M η 12 + η 22 ) (3)

U=
1
4
( 1
)
κ η12 + η22 + κ 12η12
2
(4)

Therefore,
dE
2⋅ 1 + (κ + 2κ 12 ) η1  η 1 + [ Mη
=  Mη 2 + κη2 ] η 2 (5)
dt 
which exactly vanishes by virtue of Eqs. (12.14).
When expressed explicitly in terms of the generalized coordinates, it is evident that there is only
one term in the energy that has κ 12 as a coefficient (namely, κ 12η12 ), and through Eq. (12.15) we
see that this implies that such a term depends on the C1 ’s and ω 1 , but not on the C2 ’s and ω 2 .
To understand why this is so, it is sufficient to recall that η1 is associated with the
anitsymmetrical mode of oscillation, which obviously must have κ 12 as a parameter. On the
other hand, η2 is associated to the symmetric mode, x1 ( t ) = x2 ( t ) , x 1 ( t ) = x 2 ( t ) , in which both
masses move as if linked together with a rigid, massless rod. For this mode, therefore, if the
spring connecting the masses is changed, the motion is not affected.

12-5. We set κ 1 = κ 2 = κ 12 ≡ κ . Then, the equations of motion are

m1 x1 + 2κ x1 − κ x2 = 0 
 (1)
x2 + 2κ x2 − κ x1 = 0 
m2 

Assuming solutions of the form

x1 ( t ) = B1e iωt 
 (2)
iω t 
x2 ( t ) = B2 e 

we find that the equations in (1) become

402 CHAPTER 12

( 2κ − m ω ) B
1 − κ B2 = 0 
2
1
 (3)

(
−κ B1 + 2κ − m2ω 2 B2 = 0  )
which lead to the secular equation for ω 2 :

( 2κ − m ω )( 2κ − m ω ) = κ
1
2
2
2 2
(4)

Therefore,

κ  3µ 
ω2 = 1 ± 1 −  (5)
µ  m1 + m2 

where µ = m1 m2 ( m1 + m2 ) is the reduced mass of the system. Notice that (5) agrees with Eq.
(12.8) for the case m1 = m2 = M and κ 12 = κ . Notice also that ω 2 is always real and positive since
the maximum value of 3µ ( m1 + m2 ) is 3 4 . (Show this.)

Inserting the values for ω 1 and ω 2 into either of the equations in (3), we find
1

 m1  3µ 
2 − 1 + 1 −   a11 = a21 (6)
 µ  m1 + m2  

and
 m  3µ 
a12 =  2 − 1 1 − 1 −   a22 (7)
 µ  m1 + m2  

Using the orthonormality condition produces

1
a11 = (8)
D1

m1 + m2  3m1m2 
2− 1 + 1 − 
m2 
 ( m1 + m2 ) 2 
a21 = (9)
D1

where
m12
D1 ≡ 2 ( m2 − m1 ) + 2 +
2
m12 − m22 ( ) 1−
3m1m2
(10)
m2 m2 ( m1 + m2 ) 2
The second eigenvector has the components

m1 + m2  3m1m2 
2− 1 − 1 − 
m2 
 ( m1 + m2 ) 2 
a12 = (11)
D2

1
Recall that when we use ω = ω 1 , we call the coefficients β1 (ω = ω 1 ) = a11 and β 2 (ω = ω 1 ) = a21 , etc.
COUPLED OSCILLATIONS 403

1
a22 = (12)
D2

where
 m2   3m12   m12  3m1 m2
D2 ≡ 2m1 1 + 12  + m2 1 − 2  + 2m1 1 − 2  1 − (13)
 m2   m2   m2  ( m1 + m2 )2
The normal coordinates for the case in which q j ( 0 ) = 0 are

η1 ( t ) = ( m1 a11 x10 + m2 a21 x20 ) cos ω 1t

(14)
η2 ( t ) = ( m1 a12 x10 + m2 a22 x20 ) cos ω 2t

12-6.

k
m
k
m

x1 x2

If the frictional force acting on mass 1 due to mass 2 is

f = − β ( x 1 − x 2 ) (1)

then the equations of motion are

1 + β ( x 1 − x 2 ) + κ x1 = 0 
mx
 (2)
mx2 + β ( x2 − x1 ) + κ x2 = 0 
  

Since the system is not conservative, the eigenfrequencies will not be entirely real as in the
previous cases. Therefore, we attempt a solution of the form
x1 ( t ) = B1eα t ; x2 ( t ) = B2 eα t (3)
where α = λ + iω is a complex quantity to be determined. Substituting (3) into (1), we obtain the
following secular equation by setting the determinant of the coefficients of the B’s equal to zero:

( mα )
2
2
+ βα + κ = β 2α 2 (4)

from which we find the two solutions

κ κ 
α1 = ± i ; ω1 = ±
mm 
(5)

α2 =
1
m
(
−β ± β − mκ
2 
 )
The general solution is therefore
404 CHAPTER 12

x1 ( t ) = B11+ e
i κ mt
+ B11− e
i κ mt
(
+ e − βt m B12+ e β 2 − mκ t m
+ B12− e β 2 − mκ t m
) (6)

and similarly for x2 ( t ) .

The first two terms in the expression for x1 ( t ) are purely oscillatory, whereas the last two terms
contain the damping factor e − βt . (Notice that the term B12
+
exp ( )
β 2 − mκ t increases with time if
β 2 > mκ , but B12+ is not required to vanish in order to produce physically realizable motion
because the damping term, exp(–βt), decreases with time at a more rapid rate; that is
− β + β 2 − mκ < 0 .)

To what modes do α 1 and α 2 apply? In Mode 1 there is purely oscillating motion without
friction. This can happen only if the two masses have no relative motion. Thus, Mode 1 is the
symmetric mode in which the masses move in phase. Mode 2 is the antisymmetric mode in which
the masses move out of phase and produce frictional damping. If β 2 < mκ , the motion is one of
damped oscillations, whereas if β 2 > mκ , the motion proceeds monotonically to zero amplitude.

12-7.

m x1

m
x2

We define the coordinates x1 and x2 as in the diagram. Including the constant downward
gravitational force on the masses results only in a displacement of the equilibrium positions and
does not affect the eigenfrequencies or the normal modes. Therefore, we write the equations of
motion without the gravitational terms:
1 + 2κ x1 − κ x2 = 0 
mx
 (1)
2 + κ x2 − κ x1 = 0 
mx

Assuming a harmonic time dependence for x1 ( t ) and x2 ( t ) in the usual way, we obtain

( 2κ − mω ) B− κ B2 = 0 
2
1
 (2)

(
−κ B1 + κ − mω 2 B2 = 0 )
Solving the secular equation, we find the eigenfrequencies to be
COUPLED OSCILLATIONS 405

3+ 5 κ
ω 12 =
2 m
(3)
3− 5 κ
ω 22 =
2 m
Substituting these frequencies into (2), we obtain for the eigenvector components

1− 5 
a11 = a21 
2  (4)

1+ 5
a12 = a22 
2 

For the initial conditions x 1 ( 0 ) = x 2 ( 0 ) = 0 , the normal coordinates are

 1− 5 
η1 ( t ) = ma11  x10 + x20  cos ω 1t
 2 
(5)
 1+ 5 
η2 ( t ) = ma12  x10 + x20  cos ω 2t
 2 

Therefore, when x10 = −1.6180 x20 , η2 ( t ) = 0 and the system oscillates in Mode 1, the
antisymmetrical mode. When x10 = 0.6180 x20 , η1 ( t ) = 0 and the system oscillates in Mode 2, the
symmetrical mode.
When mass 2 is held fixed, the equation of motion of mass 1 is
mx1 + 2κ x1 = 0 (6)
and the frequency of oscillation is
2κ
ω 10 = (7)
m
When mass 1 is held fixed, the equation of motion of mass 2 is
2 + κ x2 = 0
mx (8)
and the frequency of oscillation is
κ
ω 20 = (9)
m
Comparing these frequencies with ω 1 and ω 2 we find

3+ 5 2κ 2κ 
ω1 = = 1.1441 > ω 10 
4 m m 

3− 5 κ κ 
ω2 = = 0.6180 < ω 20 
4 m m 
406 CHAPTER 12

Thus, the coupling of the oscillators produces a shift of the frequencies away from the
uncoupled frequencies, in agreement with the discussion at the end of Section 12.2.

12-8. The kinetic and potential energies for the double pendulum are given in Problem 7-7. If
we specialize these results to the case of small oscillations, we have

T=
1
2
(
mA 2 2φ12 + φ22 + 2φ1 φ2 ) (1)

U=
1
2
(
mgA 2φ12 + φ 22 ) (2)

where φ1 refers to the angular displacement of the upper pendulum and φ 2 to the lower
pendulum, as in Problem 7-7. (We have also discarded the constant term in the expression for
the potential energy.)
Now, according to Eqs. (12.34),
1
T= ∑ mjk q j qk
2 j,k
(3)

1
U= ∑ Ajk qj qk
2 j,k
(4)

Therefore, identifying the elements of {m} and {A} , we find

 2 1
{m} = mA 2   (5)
 1 1
2 0
{A} = mgA   (6)
0 1
and the secular determinant is
g
2 − 2ω 2 −ω 2
A
=0 (7)
g
−ω 2
− ω2
A
or,
 g 2  g 2
 2 − 2ω   − ω  − ω = 0
4
(8)
A A 
Expanding, we find
2
g g
ω − 4 ω2 + 2   = 0
4
(9)
A A
which yields
COUPLED OSCILLATIONS 407

(
ω2 = 2 ± 2 ) gA (10)

and the eigenfrequencies are

g g
ω1 = 2 + 2 = 1.848
A A
(11)
g g
ω2 = 2 − 2 = 0.765
A A

To get the normal modes, we must solve

∑(A jk )
− ω r2 m jk a jr = 0
j

For k = 1, this becomes:

(A 11 ) (
− ω r2 m11 a1r + A21 − ω r2 m21 a2 r = 0)
For r = 1:

( g
)
2 g
(
 2mgA − 2 + 2 A 2mA  a11 − 2 + 2 A mA a21 = 0
2
)
Upon simplifying, the result is

a21 = − 2 a11

Similarly, for r = 2, the result is

a22 = 2 a12

The equations
x1 = a11 η1 + a12 η2

x2 = a21 η1 + a22 η2
can thus be written as
1
x1 = a11 η1 + a22 η2
2

x1 = − 2 a11 η1 + a22 η2

Solving for η1 and η2 :

2 x1 − x2 2 x1 + x2
η1 = ; η2 =
2 2 a11 2 a22
408 CHAPTER 12

x2
η1 occurs when η2 = 0; i.e. when x1 = −
2
x2
η2 occurs when η1 = 0; i.e. when x1 =
2
Mode 2 is therefore the symmetrical mode in which both pendula are always deflected in the
same direction; and Mode 1 is the antisymmetrical mode in which the pendula are always
deflected in opposite directions. Notice that Mode 1 (the antisymmetrical mode), has the higher
frequency, in agreement with the discussion in Section 12.2.

12-9. The general solutions for x1 ( t ) and x2 ( t ) are given by Eqs. (12.10). For the initial
conditions we choose oscillator 1 to be displaced a distance D from its equilibrium position,
while oscillator 2 is held at x2 = 0 , and both are released from rest:
x1 ( 0 ) = D, x2 ( 0 ) = 0, x 1 ( 0 ) = 0, x 2 ( 0 ) = 0 (1)

Substitution of (1) into Eq. (12.10) determines the constants, and we obtain
D
x1 ( t ) =
2
( cos ω1t + cos ω 2t ) (2)

D
x2 ( t ) =
2
( cos ω 2t − cos ω1t ) (3)

where
κ + 2κ 12 κ
ω1 > ω2 = (4)
M M
As an example, take ω 1 = 1.2 ω 2 ; x1 ( t ) vs. x2 ( t ) is plotted below for this case.

It is possible to find a rotation in configuration space such that the projection of the system
point onto each of the new axes is simple harmonic.
By inspection, from (2) and (3), the new coordinates must be
x1′ ≡ x1 − x2 = D cos ω 1t (5)

x2′ ≡ x1 + x2 = D cos ω 2t (6)

These new normal axes correspond to the description by the normal modes. They are
represented by dashed lines in the graph of the figure.
COUPLED OSCILLATIONS 409

x2(t)/D x2′

ω1 = 1.2 ω2
0.8 4π

5π 0.6
ω 2t =
2
0.4 ω2t = 2π
π
ω 2t =
2 π
0.2 ω 2t =
3 ω 2t = 0
x1(t)/D
1.0 0.2 0.4 0.6 0.8 1.0
ω 2t = π 7π
2
3π
ω 2t =
–0.5 2

3π
x1′

12-10. The equations of motion are

1 + bx 1 + (κ + κ 12 ) x1 − κ 12 x2 = F0 cos ω t 
mx
 (1)
2 + bx 2 + (κ + κ 12 ) x2 − κ 12 x1 = 0
mx 

The normal coordinates are the same as those for the undamped case [see Eqs. (12.11)]:
η1 = x1 − x2 ; η2 = x1 + x2 (2)
Expressed in terms of these coordinates, the equations of motion become

m (η 1 ) + b (η 2 + η 1 ) + (κ + κ 12 )(η2 + η1 ) − κ 12 (η2 − η1 ) = 2F0 cos ω t 

2 + η
 (3)
m (η2 − η1 ) + b (η2 − η1 ) + (κ + κ 12 )(η2 − η1 ) − κ 12 (η2 + η1 ) = 0
    

By adding and subtracting these equations, we obtain the uncoupled equations:
b κ + 2κ 12 F 
η1 + η 1 + η1 = 0 cos ω t 
m m m
 (4)
b κ F 
η2 + η 2 + η2 = 0 cos ω t 
m m m 
With the following definitions,
410 CHAPTER 12

b 
2β = 
m

κ + 2κ 12 
ω 12 = 
m 
 (5)
κ 
ω = 2
2 
m

F0 
A= 
m 
the equations become

η1 + 2βη 1 + ω 12η1 = A cos ω t

(6)
η2 + 2βη 2 + ω 22η2 = A cos ω t

Referring to Section 3.6, we see that the solutions for η1 ( t ) and η2 ( t ) are exactly the same as
that given for x(t) in Eq. (3.62). As a result η1 ( t ) exhibits a resonance at ω = ω 1 and η2 ( t )
exhibits a resonance at ω = ω 2 .

12-11. Taking a time derivative of the equations gives ( q = I )

I
LI1 + 1 + MI2 = 0
C
I
LI2 + 2 + MI1 = 0
C
Assume I1 = B1e iω t , I 2 = B2 e iωt ; and substitute into the previous equations. The result is
1
−ω 2 LB1e iω t + B1 e iωt − Mω 2 B2 e iωt = 0
C
1
−ω 2 LB2 e iωt + B2 e iωt − Mω 2 B1e iωt = 0
C
These reduce to
1 
B1  − ω 2 L  + B2 − Mω 2 = 0
C 
( )
1 
( )
B1 − Mω 2 + B2  − ω 2 L  = 0
C 
This implies that the determinant of coefficients of B1 and B2 must vanish (for a non-trivial
solution). Thus
COUPLED OSCILLATIONS 411

1
− ω 2L − Mω 2
C
=0
1
− Mω 2 − ω 2L
C
2
1 2 
( )
2
 C − ω L  − Mω =0
2

1
− ω 2 L = ± Mω 2
C
or
1
ω2 =
C (L ± M)
Thus

1
ω1 =
C (L + M)

1
ω2 =
C (L − M)

12-12. From problem 12-11:

1
LI1 + I1 + MI2 = 0 (1)
C
1
LI2 + I 2 + MI1 = 0 (2)
C
Solving for 
I1 in (1) and substituting into (2) and similarly for 
I 2 , we have

 M 2   1 M 
 L −  I1 + I1 − I2 = 0 
L  C CL 
 (3)
 M 2   1 M 
 L − L  I 2 + C I 2 − CL I1 = 0 

If we identify


M2
m= L− 
L
M 
κ 12 =  (4)
LC 

1 M 
κ = 1 −  
C L 
412 CHAPTER 12

then the equations in (3) become

mI1 + (κ + κ 12 ) I1 − κ 12 I 2 = 0 
 (5)

mI 2 + (κ + κ 12 ) I 2 − κ 12 I1 = 0 


which are identical in form to Eqs. (12.1). Then, using Eqs. (12.8) for the characteristic
frequencies, we can write

M
1+
L 1
ω1 = =
 M  C (L − M)
2
CL − 
 L 
(6)
M
1−
L 1
ω2 = =
 M  C (L + M)
2
CL −
 L 

which agree with the results of the previous problem.

12-13.
q1 q2
C1 C2
L1 L12 L2
I1 I2

The Kirchhoff circuit equations are


L1 I1 +
q1
C1
(
+ L12 I1 − I2 = 0  )

 (1)
q
(
L2 I2 + 2 + L12 I2 − I1 = 0 
C2  )
Differentiating these equations using q = I , we can write

1 
( L1 + L12 ) I1 + I1 − L12
I2 = 0 
C1
 (2)

1
( L2 + L12 ) I 2 + I 2 − L12 I1 = 0 
 
C2 
As usual, we try solutions of the form
I1 ( t ) = B1 e iωt ; I 2 ( t ) = B2 e iωt (3)
which lead to
COUPLED OSCILLATIONS 413

 2 1 
ω ( L1 + L12 ) −  B1 − L12 ω B2 = 0 
2

 C1  
 (4)
 2 1 
− L12 ω B1 + ω ( L2 + L12 ) −  B2 = 0 
2

 C1  
Setting the determinant of the coefficients of the B’s equal to zero, we obtain
 2 1  2 1 
ω ( L1 + L12 ) −  ω ( L2 + L12 ) −  = ω L12
4 2
(5)
 C1  C 2

with the solution

( L1 + L12 ) C1 + ( L2 + L12 ) C2 ± ( L1 + L12 ) C1 − ( L2 + L12 ) C2 

2
+ 4 L212C1C2
ω 2
= (6)
2C1C2 ( L1 + L12 )( L2 + L12 ) − L212 

We observe that in the limit of weak coupling ( L12 → 0 ) and L1 = L2 = L , C1 = C2 = C , the

frequency reduces to
1
ω= (7)
LC
which is just the frequency of uncoupled oscillations [Eq. (3.78)].

12-14.
L1 L2

C1 C12 C2
I1 I2

The Kirchhoff circuit equations are (after differentiating and using q = I )

1 1  1 
L1
I1 +  +  I1 − I2 = 0 
 C1 C12  C12 
 (1)
 1 1  1 
L2
I2 +  +  I2 − I1 = 0 
 C2 C12  C12 

Using a harmonic time dependence for I1 ( t ) and I 2 ( t ) , the secular equation is found to be

 C1 + C12   C2 + C12  1
 L1ω −   L2ω − = 2
2 2
(2)
 C1 C12   C2 C12  C12

Solving for the frequency,

C1L1 ( C2 + C12 ) + C2 L2 ( C1 + C12 ) ± C1L1 ( C2 + C12 ) − C2 L2 ( C1 + C12 )  + 4C12C22 L1L2

2

ω =
2
(3)
2L1L2C1C2C12
414 CHAPTER 12

Because the characteristic frequencies are given by this complicated expression, we examine the
normal modes for the special case in which L1 = L2 = L and C1 = C2 = C . Then,

2C + C12
ω 12 =
LCC12
(4)
1
ω 22 =
LC
Observe that ω 2 corresponds to the case of uncoupled oscillations. The equations for this
simplified circuit can be set in the same form as Eq. (12.1), and consequently the normal modes
can be found in the same way as in Section 12.2. There will be two possible modes of oscillation:
(1) out of phase, with frequency ω 1 , and (2) in phase, with frequency ω 2 .
Mode 1 corresponds to the currents I1 and I 2 oscillating always out of phase:

;
I1 I2 I1 I2

Mode 2 corresponds to the currents I1 and I 2 oscillating always in phase:

;
I1 I2 I1 I2

(The analogy with two oscillators coupled by a spring can be seen by associating case 1 with
Fig. 12-2 for ω = ω 1 and case 2 with Fig. 12-2 for ω = ω 2 .) If we now let L1 ≠ L2 and C1 ≠ C2 , we
do not have pure symmetrical and antisymmetrical symmetrical modes, but we can associate
ω 2 with the mode of highest degree of symmetry and ω 1 with that of lowest degree of
symmetry.

12-15.
C1 C2

L1 R L2
I1 I2

Setting up the Kirchhoff circuit equations, differentiating, and using q = I , we find

L1 (
I1 + R I1 − I2 +
1
C1
)
I1 = 0 

 (1)

L2 (
I 2 + R I2 − I1 +
1
C2
)
I2 = 0 


Using a harmonic time dependence for I1 ( t ) and I 2 ( t ) , the secular equation is

 2 1  2 1 
 ω L1 − C − iω R  ω L2 − C − iω R + ω R = 0
2 2
(2)
1 2
COUPLED OSCILLATIONS 415

From this expression it is clear that the oscillations will be damped because ω will have an
imaginary part. (The resistor in the circuit dissipates energy.) In order to simplify the analysis,
we choose the special case in which L1 = L2 = L and C1 = C2 = C . Then, (2) reduces to
2
 2 1 
 ω L − − iω R + ω R = 0
2 2
(3)
C
which can be solved as in Problem 12-6. We find
1
ω1 = ±
LC
(4)
i  L 
ω2 = R ± R − 
2

L  C 

The general solution for I1 ( t ) is

I1 ( t ) = B11 + e − Rt L B12 
+ − − i 1 LC t + R2 − L C t L − − R2 − L C t L
+ B11 + B12
i 1 LC t
e e e e (5)
 

and similarly for I 2 ( t ) . The implications of these results follow closely the arguments presented
in Problem 12-6.
Mode 1 is purely oscillatory with no damping. Since there is a resistor in the circuit, this means
that I1 and I 2 flow in opposite senses in the two parts of the circuit and cancel in R. Mode 2 is the
mode in which both currents flow in the same direction through R and energy is dissipated. If
R2 < L C , there will be damped oscillations of I1 and I 2 , whereas if R2 > L C , the currents will
decrease monotonically without oscillation.

12-16.
y

O x

R
θ
P
R
φ
Q(x,y)
Mg

Mg

Let O be the fixed point on the hoop and the origin of the coordinate system. P is the center of
mass of the hoop and Q(x,y) is the position of the mass M. The coordinates of Q are

x = R ( sin θ + sin φ ) 
 (1)

y = − R ( cos θ + cos φ ) 

The rotational inertia of the hoop through O is

416 CHAPTER 12

I O = I CM + MR 2 = 2 MR 2 (2)
The potential energy of the system is
U = U hoop + U mass
(3)
= − MgR ( 2 cos θ + cos φ )

Since θ and φ are small angles, we can use cos x ≅ 1 − x 2 2 . Then, discarding the constant term in
U, we have

U=
1
2
(
MgR 2θ 2 + φ 2 ) (4)

The kinetic energy of the system is

T = Thoop + Tmass

=
1 2 1
2 2
(
I Oθ + M x 2 + y 2 ) (5)

1
= MR 2θ 2 + MR 2 θ 2 + φ 2 + 2θφ
 

2
where we have again used the small-angle approximations for θ and φ. Thus,
1
T= MR 2  3θ 2 + φ 2 + 2θφ
 
 (6)
2
Using Eqs. (12.34),
1
T= ∑ mjk q j qk
2 j,k
(7)

1
U= ∑ Ajk qj qk
2 j,k
(8)

we identify the elements of {m} and {A} :

 3 1
{m} = MR2   (9)
 1 1
2 0
{A} = MgR   (10)
0 1 
The secular determinant is
g
2 − 3ω 2 −ω 2
R
=0 (11)
g
−ω 2
− ω2
R
from which
COUPLED OSCILLATIONS 417

 g 2  g 2
 2 − 3ω   − ω  − ω = 0
4
(12)
R R
Solving for the eigenfrequencies, we find

g
ω1 = 2
R
(13)
2 g
ω2 =
2 R

To get the normal modes, we must solve:

∑(A jk )
− ω r2 m jk a jr = 0
j

For k = r = 1, this becomes:

 g 2 g
 2mgR − 2 3mR  a11 − 2 mR a21 = 0
2

R R
or
a21 = −2a11
For k = 1, r = 2, the result is
a12 = a22
Thus the equations
x1 = a11 η1 + a12 η2

x2 = a21 η1 + a22 η2

can be written as
x1 = a11 η1 + a22 η2

x2 = −2 a11 η1 + a22 η2
Solving for η1 , η2
x1 − x2 2x1 + x2
η1 = ; η1 =
3a11 3a22

1
η1 occurs when the initial conditions are such that η2 = 0 ; i.e., x10 = − x20
2
This is the antisymmetrical mode in which the CM of the hoop and the mass are on opposite
sides of the vertical through the pivot point.
η2 occurs when the initial conditions are such that η1 = 0 ; i.e., x10 = x20
418 CHAPTER 12

This is the symmetrical mode in which the pivot point, the CM of the hoop, and the mass
always lie on a straight line.

12-17.
k m k m k m k

x1 x2 x3

Following the procedure outlined in section 12.6:

1 1 1
T= mx 12 + mx 22 + mx 32
2 2 2
1 2 1 1 1
kx1 + k ( x2 − x1 ) + k ( x3 − x2 ) + kx32
2 2
U=
2 2 2 2

= k  x12 + x22 + x32 − x1 x2 − x2 x3 

Thus
m 0 0 
 
m= 0 m 0
 0 0 m 

 2k − k 0 
 
A =  − k 2k − k 
 0 − k 2k 

Thus we must solve

2k − ω 2 m −k 0
−k 2k − ω m 2
−k =0
0 −k 2k − ω m 2

This reduces to

( 2 k − ω m) ( )
3
2
− 2k 2 2k − ω 2 m = 0

or

( 2k − ω m) ( 2k − ω m) − 2k 2  = 0
2 2 2


If the first term is zero, then we have

2k
ω1 =
m
If the second term is zero, then

2k − ω 2 m = ± 2 k
COUPLED OSCILLATIONS 419

which leads to

ω2 =
(2 + 2 ) k ; ω3 =
(2 − 2) k
m m
To get the normal modes, we must solve

∑(A jk )
− ω r2 m jk a jr = 0
j

For k = 1 this gives:

( 2 k − ω m) a 2
r 1r + ( − k ) a2 r = 0

Substituting for each value of r gives

r = 1: ( 2k − 2k ) a11 − ka21 = 0 → a21 = 0
r = 2: (− )
2 k a12 − ka22 = 0 → a22 = − 2 a12

r = 3: ( )
2 k a13 − ka23 = 0 → a23 = 2 a13

Doing the same for k = 2 and 3 yields

a11 = − a31 a21 = 0

a12 = a32 a22 = − 2 a32

a13 = a33 a23 = 2 a33

The equations
x1 = a11 η1 + a12 η2 + a13 η3

x2 = a21 η1 + a22 η2 + a23 η3

x3 = a31 η1 + a32 η2 + a33 η3

can thus be written as

1
x1 = a11 η1 − a22 η2 + a33 η3
2

x2 = a22 η2 + 2 a33 η3

1
x3 = − a11 η1 − a22 η2 + a33 η3
2
We get the normal modes by solving these three equations for η1 , η2 , η3 :
420 CHAPTER 12

x1 − x3
η1 =
2a11

− x1 + 2 x2 − x3
η2 =
2 2 a22
and
x1 + 2 x2 + x3
η3 =
4 a33
The normal mode motion is as follows
η1 : • → • ← • x1 = − x3

η2 : ← • • → ← • x2 = − 2 x1 = − 2 x3

η3 : • → • → • → x2 = 2 x1 = 2 x3

12-18.
x

θ b

m
y1
x1

x1 = x + b sin θ ; x 1 = x + bθ cos θ

y1 = b − b cos θ ; y 1 = bθ sin θ

Thus

T=
1
2
1
(
Mx 2 + m x 12 + y 12
2
)
=
1
2
1
(
Mx 2 + m x 2 + b 2 θ 2 + 2b xθ cos θ
2
)
U = mgy1 = mgb (1 − cos θ )

θ2
For small θ, cos θ  1 − . Substituting and neglecting the term of order θ 2θ gives
2
COUPLED OSCILLATIONS 421

T=
1
2
1
(
( M + m) x 2 + m b 2θ 2 + 2b xθ
2
)
mgb 2
U= θ
2
Thus
 M + m mb 
m=
 mb mb 2 

0 0 
A= 
0 mgb 
We must solve

−ω 2 ( M + m) −ω 2 mb
=0
−ω 2 mb mgb − ω 2 mb 2

which gives

ω 2 ( M + m) (ω 2 mb 2 − mgb ) − ω 4 m2 b 2 = 0

ω 2 ω 2 Mb 2 − mgb ( m + M )  = 0
Thus
ω1 = 0

g
ω2 = ( M + m)
mb

∑(A jk )
− ω r2 m jk a jr = 0
j

Substituting into this equation gives

a21 = 0 ( k = 2, r = 1)
bm
a12 = − a
( m + M ) 22
( k = 2, r = 2)
Thus the equations
x = a11 η1 + a12 η2

θ = a21 η1 + a22 η2
become
mb
x = a11 η1 − a η
( m + M ) 22 2
422 CHAPTER 12

θ = a22 η2
Solving for η1 , η2 :
θ
η2 =
a22

bm
x+ θ
(m + M)
n1 =
a11

n1 occurs when n2 = 0; or θ = 0

bm
n2 occurs when n1 = 0; or x = − θ
(m + M)

12-19. With the given expression for U, we see that { A} has the form

 1 − ε 12 − ε 13 
{A} =  −ε12 1

− ε 23  (1)
 − ε 13 − ε 23 1 

The kinetic energy is

T=
2
(
1 2 2 2
θ1 + θ 2 + θ 3 ) (2)

so that {m} is

1 0 0
{m} = 0 1 0  (3)
 0 0 1 

The secular determinant is

1− ω2 − ε 12 − ε 13
− ε 12 1−ω 2
− ε 23 = 0 (4)
− ε 13 − ε 23 1−ω 2

Thus,

(1 − ω ) − (1 − ω )( ε
2 3 2 2
12 + ε 13
2
+ ε 23
2
)
− 2ε 12 ε 13 ε 23 = 0 (5)

This equation is of the form (with 1 − ω 2 ≡ x )

x 3 − 3α 2 x − 2β 2 = 0 (6)
which has a double root if and only if

(α )
2 32
= β2 (7)
COUPLED OSCILLATIONS 423

Therefore, (5) will have a double root if and only if

32
 ε 12
2
+ ε 13
2
+ ε 23
2

  = ε 12 ε 13 ε 23 (8)
 3 
This equation is satisfied only if

ε 12 = ε13 = ε 23 (9)

Consequently, there will be no degeneracy unless the three coupling coefficients are identical.

12-20. If we require a11 = 2a21 , then Eq. (12.122) gives a31 = −3a21 , and from Eq. (12.126) we
obtain a21 = 1 14 . Therefore,

 2 1 3 
a1 =  , ,− (1)
 14 14 14 

The components of a2 can be readily found by substituting the components of a1 above into Eq.
(12.125) and using Eqs. (12.123) and (12.127):

 4 −5 1 
a2 =  , , (2)
 42 42 42 

These eigenvectors correspond to the following cases:

a1 a2

12-21. The tensors {A} and {m} are:

 1 
 κ1 2
κ3 0 
 
1 1 
{A} =  κ 3 κ2 κ3 (1)
2 2 
 1 
 0 κ3 κ1 
 2 

m 0 0 
{m} =  0 m 0  (2)
 0 0 m 

thus, the secular determinant is

424 CHAPTER 12

1
κ 1 − mω 2 κ3 0
2
1 1
κ3 κ 2 − mω 2 κ3 =0 (3)
2 2
1
0 κ3 κ 1 − mω 2
2
from which

(κ ) (κ 1
) ( )
2
1 − mω 2 2 − mω 2 − κ 32 κ 1 − mω 2 = 0 (4)
2
In order to find the roots of this equation, we first set (1 2)κ 32 = κ 1κ 2 and then factor:

(κ 1 )( )(
− mω 2  κ 1 − mω 2 κ 2 − mω 2 − κ 1κ 2  = 0 )
(κ 1 )
− mω 2  m2ω 4 − (κ 1 + κ 2 ) mω 2  = 0 (5)

(κ 1 )
− mω 2 mω 2  mω 2 − (κ 1 + κ 2 )  = 0

Therefore, the roots are

κ1
ω1 =
m

κ1 +κ 2
ω2 = (6)
m
ω3 = 0

Consider the case ω 3 = 0 . The equation of motion is

η3 + ω 32η3 = 0 (7)

so that
η3 = 0 (8)
with the solution
η3 ( t ) = at + b (9)
That is, the zero-frequency mode corresponds to a translation of the system with oscillation.
COUPLED OSCILLATIONS 425

12-22. The equilibrium configuration is shown in diagram (a) below, and the non-
equilibrium configurations are shown in diagrams (b) and (c).
(a)
x3

4 3

O x2
2B
2A
1 x1 2

x3
(b)
2
1
O′ A θ }Aθ

A  x3

O x2

x3
(c)
4
1
Bφ { B O′
φ

x3 B
 x2
O

The kinetic energy of the system is

1 1 1
T= Mx 32 + I1θ 2 + I 2φ 2 (1)
2 2 2
where I1 = (1 3) MA2 and I 2 = (1 3) MB2 .

The potential energy is

1
U = κ ( x3 − Aθ − Bφ ) + ( x3 + Aθ − Bφ ) + ( x3 + Aθ + Bφ ) + ( x3 − Aθ + Bφ ) 
2 2 2 2

2  

1
(
= κ 4 x32 + 4 A2θ 2 + 4B2φ 2
2
) (2)

Therefore, the tensors {m} and {A} are

 
M 0 
0
 
1
{m} =  0 MA2 0  (3)
 3 
 1 
0 0 MB2 
 3 
 4κ 0 0 
{A} =  0 4κ A 2
0 

(4)
 0 0 4κ B2 

The secular equation is

426 CHAPTER 12

( 4κ − Mω )  4κ A
2 2
−
1
3


1
3

MA2ω 2   4κ B2 − MB2ω 2  = 0

(5)

Hence, the characteristic frequencies are

κ
ω1 = 2
M

3κ
ω2 = 2 (6)
M

3κ
ω3 = 2 = ω2
M
We see that ω 2 = ω 3 , so the system is degenerate.
The eigenvector components are found from the equation

∑(A jk )
− ω r2 m jk a jr = 0 (7)
j

Setting a32 = 0 to remove the indeterminacy, we find

1 M   0   0 
     
a1 =  0  ; a 2 =  3 MA2  ; a 3 =  0  (8)
 0     
  0  3 MB2 
 

The normal coordinates are (for x 3 ( 0 ) = θ ( 0 ) = φ ( 0 ) = 0 )

η1 ( t ) = x30 M cos ω 1t

θ0 A M
η2 ( t ) = cos ω 2t (9)
3

φ0 B M
η3 ( t ) = cos ω 3t
3
Mode 1 corresponds to the simple vertical oscillations of the plate (without tipping). Mode 2
corresponds to rotational oscillations around the x1 axis, and Mode 3 corresponds to rotational
oscillations around the x2 -axis.
The degeneracy of the system can be removed if the symmetry is broken. For example, if we
place a bar of mass m and length 2A along the x2 -axis of the plate, then the moment of inertia
around the x1 -axis is changed:
1
I1′ = ( M + m) A2 (10)
3
The new eigenfrequencies are
COUPLED OSCILLATIONS 427

κ
ω1 = 2
M

3κ
ω2 = 2 (11)
M+m

3κ
ω3 = 2
M
and there is no longer any degeneracy.

12-23. The total energy of the r-th normal mode is

Er = Tr + U r

1 2 1 2 2 (1)
= η r + ω r ηr
2 2
where
ηr = β r e iωr t (2)
Thus,
η r = iω r β r e iωr t (3)

In order to calculate Tr and U r , we must take the squares of the real parts of η r and ηr :

η r2 = ( Re η r ) =  Re i ω r ( µ r + iν r ) ( cos ω r t + i sin ω r t ) 
2 2

2
= −ω r ν r cos ω r t − ω r µ r sin ω r t  (4)

so that
1 2 2
Tr = ω r ν r cos ω r t + µr sin ω r t  (5)
2
Also

ηr2 = ( Re ηr ) =  Re ( µ r + iν r ) ( cos ω r t + i sin ω r t ) 

2 2

2
= µr cos ω r t − ν r sin ω r t  (6)

so that
1 2 2
Ur = ω r µ r cos ω r t − ν r xin ω r t  (7)
2
Expanding the squares in Tr and U r , and then adding, we find
428 CHAPTER 12

Er = Tr + U r

=
1 2 2
2
(
ω r µ r + ν r2 )
Thus,

1 2
ω r βr
2
Er = (8)
2
So that the total energy associated with each normal mode is separately conserved.
For the case of Example 12.3, we have for Mode 1
M
η1 =
2
( x10 − x20 ) cos ω1t (9)

Thus,
M
η 1 = −ω 1
2
( x10 − x20 ) sin ω1t (10)

Therefore,
1 2 1 2 2
E1 = η 1 + ω 1 η1 (11)
2 2
But
κ + 2κ 12
ω 12 = (12)
M
so that
1 κ + 2κ 12 M 1 κ + 2κ M
E1 =
2 M 2
( x10 − x20 ) 2 sin 2 ω1t + 2 M 12 2 ( x10 − x20 ) 2 cos2 ω1t
1
= (κ + 2κ 12 ) ( x10 − x20 ) 2 (13)
4
which is recognized as the value of the potential energy at t = 0. [At t = 0, x 1 = x 2 = 0 , so that the
total energy is U1 ( t = 0 ) .]

12-24. Refer to Fig. 12-9. If the particles move along the line of the string, the equation of
motion of the j-th particle is

( ) (
j = −κ x j − x j −1 − κ x j − x j+ 1
mx ) (1)

Rearranging, we find
κ
xj =

m
(x j −1 − 2x j + x j +1 ) (2)

which is just Eq. (12.131) if we identify τ md with κ m .

COUPLED OSCILLATIONS 429

12-25. The initial conditions are

q1 ( 0 ) = q2 ( 0 ) = q3 ( 0 ) = a 
 (1)
q1 ( 0 ) = q 2 ( 0 ) = q 3 ( 0 ) = 0 

Since the initial velocities are zero, all of the ν r [see Eq. (12.161b)] vanish, and the µr are given
by [see Eq. (12.161a)]
a  πr πr 3π r 
µr = sin 4 + sin 2 + sin 4  (2)
2
so that

2 +1 
µ1 = a
2 

µ2 = 0  (3)

2 −1 
µ3 = a
2 
The quantities sin  jrπ ( n + 1)  are the same as in Example 12.7 and are given in Eq. (12.165).
The displacements of the particles are

1 2
q1 ( t ) = a ( cos ω 1t + cos ω 3t ) + a ( cos ω 1t − cos ω 3t )
2 4

2 1
q2 ( t ) = a ( cos ω 1t − cos ω 3t ) + a ( cos ω 1t + cos ω 3t ) (4)
2 2

1 2
q3 ( t ) = a ( cos ω 1t − cos ω 3t ) + a ( cos ω 1t + cos ω 3t )
2 4
where the characteristic frequencies are [see Eq. (12.152)]

τ  rπ 
ωr = 2 sin   , r = 1, 2, 3 (5)
md 8
Because all three particles were initially displaced, there can exist no normal modes in which
any one of the particles is located at a node. For three particles on a string, there is only one
normal mode in which a particle is located at a node. This is the mode ω = ω 2 (see Figure 12-11)
and so this mode is absent.

 mb 2 
 
12-26. Kinetic energy T =
2
(
mb 2  2  2  2
θ1 + θ 2 + θ 3 ) ⇒ [ m] =  mb 2

 + mb 2 

Potential energy
430 CHAPTER 12

k
U = mgb (1 − cos θ 1 ) + (1 − cos θ 2 ) + (1 − cos θ 3 )  + b 2 ( sin θ 2 − sin θ 1 ) + ( sin θ 3 − sin θ 2 ) 
2 2

2  

kb 2 2
≈
mgb 2
2
(
θ1 + θ 22 + θ 32 +
2
) (
θ 1 + 2θ 22 + θ 32 − 2θ 1θ 2 − 2θ 2θ 3 )
 mgb + kb 2 − kb 2 0 
 
⇒ [ A] =  − kb 2 mgb + kb 2
− kb 
2

 0 − kb 2 mgb + kb 2 

The proper frequencies are solutions of the equation


(
 mgb + kb 2 − mb 2ω 2 ) − kb 2 0 

( )
0 = Det [ A] − ω 2 [ m] = D et  − kb 2 ( mgb + 2kb 2
− mb 2ω 2 ) − kb 2 
 
 0 − kb 2 ( )
mgb + kb 2 − mb 2ω 2 

We obtain 3 different proper frequencies

mg + kb mg + kb
ω 12 = ⇒ ω1 = = 4.64 rad/s
mb mb

mg + 3kb mg + 3kb
ω 22 = ⇒ ω2 = = 4.81 rad/s
mb mb

mg g
ω 32 = ⇒ ω3 = = 4.57 rad/s
mb f

Actually those values are very close to one another, because k is very small.

12-27. The coordinates of the system are given in the figure:

θ1
L1

m1

L2
θ2
m2

Kinetic energy:

T=
1 2 2 2 1
2
(
m1 θ 1 L1 + m2 L21 θ12 + L22 θ22 − 2L1L2θ1θ2 cos (θ 1 − θ 2 )
2
)
1
2
( 1
2
) 1
≈ θ12 m1L21 + m2 L21 + m2 L22θ22 − m2 L1L2θ1θ2 = ∑ m jk θ j θk
2 jk
COUPLED OSCILLATIONS 431

( m + m2 ) L21 − m2 L1L2 
⇒  m jk  =  1 
 − m2 L1 L2 m2 L22 

Potential energy:
U = m1 gL1 (1 − cos θ 1 ) + m2 g  L1 (1 − cos θ 1 ) + L2 (1 − cos θ 2 ) 

θ12 θ 22 1
≈ ( m1 + m2 ) gL1 + m2 gL2 = ∑ Ajkθ jθ k
2 2 2 jk

( m + m2 ) gL1 0 
⇒  Ajk  =  1 
 0 m2 gL2 

Proper oscillation frequencies are solutions of the equation

(
Det [ A] − ω 2 [ m] = 0 )
( m1 + m2 ) g ( L1 + L2 ) + ( m1 + m2 ) g 2  m1 ( L1 − L2 )2 + m2 ( L1 + L2 ) 2 
⇒ ω 1,2 =
2m1 L1 L2

 a11 
The eigenstate corresponding to ω 1 is   where
 a21 

 
a21
( m + m2 ) L1  1 −
= 1
2 gm1 L2  ×a
  11
m1L2  ( m1 + m2 ) g ( L1 + L2 ) + ( m1 + m2 ) g 2  m1 ( L1 − L2 ) + m2 ( L1 + L2 )  
2 2

 a12 
The eigenstate corresponding to ω 2 is   where
 a22 

 
a22
( m + m2 ) L1  1 −
= 1
2 gm1 L2  ×a
  12
m1L2  ( m1 + m2 ) g ( L1 + L2 ) − ( m1 + m2 ) g 2  m1 ( L1 − L2 ) + m2 ( L1 + L2 )  
2 2

These expressions are rather complicated; we just need to note that a11 and a21 have the same
a  a 
sign  11 > 0 while a12 and a22 have opposite sign  11 < 0 .
 a21   a21 

The relationship between coordinates (θ 1 , θ 2 ) and normal coordinates η1 , η2 are

 a12
θ1 = a11 η1 + a12 η2  η1 ~ θ 1 − a θ 2

⇔
22
 
θ 2 = a21 η1 + a22 η2  a
η2 ~ θ 1 − 11 θ 2
 a21
432 CHAPTER 12

To visualize the normal coordinate η1 , let η2 = 0 . Then to visualize the normal coordinate η2 ,
a a
we let η1 = 0 . Because 11 > 0 and 12 < 0 , we see that these normal coordinates describe two
a21 a22
oscillation modes. In the first one, the two bobs move in opposite directions and in the second,
the two bobs move in the same direction.

1 1  m2 b 2 0 
12-28. Kinetic energy: T= m1b 2θ12 + m2 b 2θ22 ⇒ [ m] =  
2 2  0 m2 b 2 

k
Potential energy: U = m1 gb (1 − cos θ 1 ) + m2 gb (1 − cos θ 2 ) + ( b sin θ1 − b sin θ 2 )
2

 m1 gb + kb 2 − kb 2 
⇒ [ A] ≈  
 − kb 2
m2 gb + kb  2

( )
Solving the equation, Det [ A] − ω 2 [ m] = 0 , gives us the proper frequencies of oscillation,

g g k k
ω 12 = = 25 (rad/s)2 ω 22 = + + = 25.11 (rad/s)2
b b m1 m2

 a11 
The eigenstate corresponding to ω 1 is   with a21 = 7.44 a11
 a22 

 a12 
The eigenstate corresponding to ω 2 is   with a22 = 8.55 a12
 a22 
From the solution of problem 12-27 above, we see that the normal coordinates are
a12
η1 ~ θ1 − θ 2 = θ 1 + 0.12 θ 2
a22

a11
η2 ~ θ 1 − θ 2 = θ 1 + 0.13 θ 2
a21
Evidently η1 then characterizes the in-phase oscillation of two bobs, and η2 characterizes the
out-of-phase oscillation of two bobs.
Now to incorporate the initial conditions, let us write the most general oscillation form:

θ1 = Re (α a11 e iω1t − iδ1 + α a12 e iω2t − iδ 2 )

θ 2 = Re (α a21 e iω1t − iδ1 + α a22 e iω 2t − iδ 2 )

(
= Re 7.44α a11 e iω1t − iδ1 − 8.35α a12 e iω 2t − iδ 2 )
where α is a real normalization constant. The initial conditions helps to determine parameters
α’s, a’s, δ’s.
COUPLED OSCILLATIONS 433

(
 Re θ 1 ( t = 0 ) = −7 ° ⇒ α a11 cos δ 1 + α a12 cos δ 2 = −0.122 rad )

 Re (θ 2 ( t = 0 ) = 0° ⇒ 7.44α a11 cos δ 1 + 8.35α a12 cos δ 2 = 0 )
⇒ sin δ 1 = sin δ 1 = 0 . Then

θ1 = α a11 cos ω 1t + α a12 cos ω 2t = −0.065 cos ω 1t − 0.057 cos ω 2

θ 2 = 7.44α a11 cos ω 1t + 8.35α a12 cos ω 2t = 0.48 ( cos ω 2t − cos ω 1t )

where ω 1 = 5.03 rad/s , ω 2 = 4.98 rad/s (found earlier)
Approximately, the maximum angle θ 2 is 0.096 rad and it happens when

 cos ω 2t = 1

cos ω 1t = −1
which gives
ω 2 t = 2n π  ω 1 2k + 1
 ⇒ =
ω 1t = ( 2k + 1) π  ω2 2n

ω 1 101 100π
because = we finally find k = n = 50 and t = = 63 s .
ω 2 100 ω2
Note: θ 2 max = 0.96 rad and at this value the small-angle approximation breaks down, and

the value θ 2max we found is just a rough estimate.

434 CHAPTER 12