Coupled oscillators

responsible for the coupling. To start our study of coupled oscillations, we will assume that the
forces involved are spring-like forces (the magnitude of the force is proportional to the
magnitude of the displacement from equilibrium).

Example: Two Coupled Harmonic Oscillators

Consider a system of two objects of mass M. The two objects are attached to two springs
with spring constants κ (see Figure 1). The interaction force between the masses is represented
by a third spring with spring constant κ12, which connects the two masses.

Figure 1. Two coupled harmonic oscillators.

We will assume that when the masses are in their equilibrium position, the springs are also in
their equilibrium positions. The force on the left mass is equal to

F1 = !" x1 + " 12 ( x2 ! x1 ) = ! (" + " 12 ) x1 + " 12 x2 = M!!

x1

The force on the right mass is equal to

F2 = !" x2 + " 12 ( x1 ! x2 ) = ! (" + " 12 ) x2 + " 12 x1 = M!!

x2

The equations of motion are thus

- 1 -
The equation of motion are:
𝑀𝑥̈ ! + 𝜅 + 𝜅!" 𝑥! − 𝜅!" 𝑥" = 0 Coupled 2nd Differential equations!!
𝑀𝑥̈ " + 𝜅 + 𝜅!" 𝑥" − 𝜅!" 𝑥! = 0

Since it is reasonable to assume that the resulting motion has an oscillatory behavior, we consider the
following trial functions:

𝑥! 𝑡 = 𝐵! 𝑒 #$%
𝑥" 𝑡 = 𝐵" 𝑒 #$%

Substituting these trial functions into the equations of motion we obtain the following conditions:

𝜅 + 𝜅!" − 𝑀𝜔" 𝐵! − 𝜅!" 𝐵" = 0

−𝜅!" 𝐵! + 𝜅 + 𝜅!" − 𝑀𝜔" 𝐵" = 0
Consider two simultaneous equations
Ax + By = 0
Cx + Dy = 0
where A,B,C,D are constants. We want to find the solutions (x,y) which satisfy two equations simultaneous.
From eqtn. 1 and 2, we have
B
x= y
A
D
x= y
C
For the solution which satisfy two equations, it requires

B D
= ! AD BC = 0
A C
It means that if 𝐴𝐷 − 𝐵𝐶 ≠ 0, we can’t find a solution which satisfies two equations simultaneous (except x=y=0).

A B
Determinant = = AD BC
C D
Substituting these trial functions into the equations of motion we obtain the following conditions:

(! + ! 12 )
" M # 2 B1 " ! 12 B2 = 0

(
!" 12 B1 + " + " 12 ! M # 2 B2 = 0 )
These equations only will have a non-trivial solution if

! + ! 12 " M # 2 "! 12
=0
"! 12 ! + ! 12 " M # 2

Note: the trivial solution is B1 = B2 = 0. The requirement for a non-trivial solution requires that
the angular frequency of the system is equal to one of the following two characteristic
frequencies (the so called eigen frequencies):

" + 2" 12
!1 = ±
M

"
!2 = ±
M

For each of these frequencies, we can now determine the amplitudes B1 and B2. Let us first
consider the eigen frequency ω1. For this frequency we obtain the following relations between
B1 and B2:

(! + ! 12 " (! + 2! 12 )) B1 " ! 12 B2 = "! 12 B1 " ! 12 B2 = "! 12 ( B1 + B2 ) = 0

For each of these frequencies, we can now determine the amplitudes 𝐵! and 𝐵" . Let us first consider the
eigenfrequency 𝜔! . For this frequency we obtain the following relations between 𝐵! and 𝐵" :
𝜅 + 𝜅!" − 𝜅 + 2𝜅!" 𝐵! − 𝜅!" 𝐵" = −𝜅!" 𝐵! − 𝜅!" 𝐵" = −𝜅!" 𝐵! + 𝐵" = 0
⇒ 𝐵! = −𝐵" ≡ 𝐶
We can do the same calculation for −𝜔! . Hence the solution is:

𝑥! 𝑡 = 𝐶𝑒 #$! % + 𝐷𝑒 &#$! %
𝑥" 𝑡 = −𝐶𝑒 #$" % − 𝐷𝑒 &#$! %

For the eigenfrequency 𝜔" , we obtain the following relations:

𝜅 + 𝜅!" − 𝜅 𝐵! − 𝜅!" 𝐵! − 𝜅!" 𝐵" = 𝜅!" 𝐵! − 𝐵" = 0
⇒ 𝐵! = 𝐵" = 𝐸
The solution has the form:

𝑥! 𝑡 = 𝐸𝑒 #$" % + 𝐹𝑒 &#$" %
𝑥" 𝑡 = 𝐸𝑒 #$" % + 𝐹𝑒 &#$" %
The general solution of the two-coupled oscillators is

𝑥! 𝑡 = 𝐶𝑒 #$! % + 𝐷𝑒 &#$! % + 𝐸𝑒 #$" % + 𝐹𝑒 &#$" %

𝑥" 𝑡 = −𝐶𝑒 #$" % − 𝐷𝑒 &#$! % + 𝐸𝑒 #$" % + 𝐹𝑒 &#$" %

How to specify the values of 4 constants (𝐶, 𝐷, 𝐸, 𝐹)?

We can use the initial position and velocity of two objects.

N-coupled oscillators
Foundations of Wave Phenomena, Version 8.2

m k

q1 q2
… qN
q1 = 0 q2 = 0 qN = 0

All Figure
the 7.masses and the elastic spring are identical
Linear chain of coupled oscillators. Each oscillator of mass m is
Equation
coupled to itsof motion
nearest (EOM):
neighbor with a spring with ( =
(𝜔spring 𝑘/𝑚)
constant k . As in the
case of the two-coupled oscillator problem, displacement from equilibrium qi
is restricted to be along the chain of oscillators, as illustrated.
d2 qj
! 2 (qj+1 2qj + qj 1) = 0, j = 1, 2, . . . , N
dt2
q =q =0
The equations of motion (4.1) are, mathematically speaking, a system of N coupled,
0 N +1
linear, homogeneous, ordinary di↵erential equations with constant coefficients. Note that
each oscillator is coupled only to its “nearest neighbors” (exercise). As it turns out, the
system of coupled oscillators described by (4.1) exhibits resonant frequencies and normal
N coupled
modes of vibration. To see2this linear
nd
we couldhomogeneous differential
set up (4.1) as a matrix equations
equation (see Problems)
and use the linear algebraic techniques discussed above. In particular, the generalization
of the matrix K from the last section will be symmetric and hence will admit N linearly
independent eigenvectors, which define the normal modes and whose eigenvalues define
the characteristic frequencies. While this is a perfectly reasonable way to proceed, partic-
ularly for relatively low values for N , for arbitrary values of N we can reduce the analysis
considerably by employing a shortcut. We shall explain this in a moment.
 We We
stillcan
must
solveenforce the condition
this Using
relationthis qspecifying
in (4.6)
by further gives
N +1 = 0, which
(exercise)
our is now ASuppose
trial solution. N +1 =A0. This condition
j is of the
means form:
Aj = + a sin(j ⌦ 2 sin(j ) = 2! 2 [1 cos( )](4.5)sin(j ).(4.9)
sin[(N 1) ] ),
= 0,
so1 that where
Guess:
is some real number and a can
q (t) = a sin(j
Givenq (4.3) )e i⌦tbe complex.† Note that this trial solution satisfies
and (4.5), 𝑎, Ω are 3 undetermined
we𝜙,naturally constants
assume that sin(j ) does not vanish
the boundaryjcondition 0 = 0, but we still have to take care of the condition qN +1 = 0
(N +
j. Thus the1)recursion
= n⇡, relation
n = 1, 2,(and
. . . , N. (4.10) (4.1
hence the equations of motion
— we shall do this below by specifying the parameter
Foundations of Wave. Phenomena,
We plug Version
(4.5) into
8.2 the recursion
EOM
In (4.10)relation
we takeimplies:
to get (4.3) andvalue
the(exercise)
maximum (4.5)forif and
n toonly
be Nif to avoid redundant solutions; if n > N
n o
then wesin[(j
obtain solutions
± 1) ] = 2 for A j that
2 were already found when n  N (see below and also
⌦ sin(j =!)±
a sin(j ) )cos( cos(j )1)
a sin[(j sin(
] ).2a 2sin[j ] +2 a sin[(j + 1) ]
⌦ = 2! [1 cos( )] = 4! sin ( /2), . 2 2
(4.6)
the homework problems). We exclude the solution corresponding to n = 0 because this
s in (4.6)
solution Simplify:
gives (exercise)
has that=a 0,
willwhich
thatforces
is, Aj = 0, i.e., this is the trivial solution qj (t) = 0 (for all
Note drop out of this condition, that is, a is not determined by (4.4)
values of j) ⌦of2 sin(j
the coupled oscillator
) = 2! 2 [1 cos( )]equations.
sin(j ). OR ⌦ = 2!| sin( /2)|.
Exercise: What property of the equations (4.1) and/or (4.4) guarantees that a will drop
3) and (4.5), out
we of Notethat
(4.6)? assume
naturally thatsin(j
we are adhering
) does to our
not vanish convention
identically for allthat ⌦ be non-negative.
Exercise: What
boundary property of (4.1) guarantees that qj = 0 is a solution?
conditions
the recursion relation (and hence the equations of motion (4.1)) are satisfied by
2 q0 (t) = qN +1 (t) = 0 We ! sin
still((N
must+enforce
1) ) =the 0! (N + 1)qN +1
condition = n⇡ for any
= 0, which integer
is now n =
AN +1
(4.5) if and only if
To analyze (4.6) we use the trigonometric identity (exercise),
To summarize thusmeans far, there are N distinct resonant frequencies, which we label by
We can⌦2obtain N different values 2 of 2Ω
an3 integer n, = 2! 2 [1
where 1, 2, )]. sin(↵
n =cos( .= N+. sin
. , 4! )=
They )/2),
(sin ↵ cos the
take + cos
form ↵ sin sin[(N(4.7) + 1) ] = 0,
* A recursion relation
so for ✓
a set of variables
that Aj◆
, , j = 1, 2, . . . n, is a sequence of equations
which allows one to⌦determine
n⇡
Ak from the set A
n = 2!| sin |,1 , A(N
n2 , .=.+
. 1,
, A2,
k .1 .. . , N.
1) = n⇡, n = 1, 2, . . . , (4.11)
N.
⌦ = 2!| sin( /2)|. 2N + not
† This form of the trial solution is certainly (4.8)
2 obvious. It can be motivated by studying
several special cases with N small. Alternatively, one can consider (4.4) for very large
t we Compare
are adhering
valuestoofwith
this our the In
which (4.10)
j, inconvention
case
case that
onetwo
of we take
⌦ coupled
can be thethat
maximum
non-negative.
pretend value
Aj is a function
oscillators, treated forthe
of n continuous
to bewhere
earlier, N to avoid
there redundant
variable were 2 s
j. One can then interpret the recursion relation as (approximately) saying that the second
resonant frequencies.
ill must enforce the condition
derivative then
qN +1 we
of this function obtain
which solutions
=is0,proportional
is now for Aj that
ANfunction
to the +1 = 0. This were
itself. Usingalready
condition found
A0 = 0 one when n  N (
arrives
at (4.5) (exercise). the homework problems). We exclude the solution corresponding to n
 Normal mode

The general solution of the oscillators are:

XN ✓ ◆
n⇡j
qj (t) = an sin cos(⌦n t + ↵n ), j = 1, 2, . . . , N
n=1
N + 1
✓ ◆
where n⇡
⌦n = 2!| sin |
2(N + 1)
There are 2N undetermined constants 𝑎) , 𝛼) !!
Fit by the 2N initial conditions (position and velocity) of N oscillators!!
 0.2
0 10 20 30 40 50

INTERMEDIATE FREQUENCY NORMAL MODE

0.2

DISPLACEMENT
0
0

Normal modes for 50 coupled oscillators (N=50)

0.2
0 10 20 30 40 50
Foundations of Wave Phenomena, Version 8.2

LOWEST FREQUENCY NORMAL MODE SECOND HIGHEST FREQUENCY NORMAL MODE

0.2 0.2

DISPLACEMENT
DISPLACEMENT

0
0
0.1

0.2
0 10 20 30 40 50
0
0 10 20 30 40 50

HIGHEST FREQUENCY NORMAL MODE

SECOND LOWEST FREQUENCY NORMAL MODE 0.2
0.2

DISPLACEMENT
DISPLACEMENT

0
0 0
0

0.2
0 10 20 30 40 50
0.2
0 10 20 30 40 50
MASS INDEX

INTERMEDIATE FREQUENCY NORMAL MODE

0.2 Figure 8. Selected normal modes for an N = 50 linear chain of coupled
oscillators.
DISPLACEMENT

0
0 34 c C. G. Torre

0.2
0 10 20 30 40 50

SECOND HIGHEST FREQUENCY NORMAL MODE

0.2
LACEMENT

0
0