You are on page 1of 8

Notes on Oscillations

Physics 103; Fall 2006


November 25, 2006

This week we will continue studying systems which are oscillating in response to a linear
restoring force. Such systems are described by differential equations; for example a mass m
on a spring with spring constant k constrained to move only along the x axis experiences
force
F = kx = ma = md2 x/dt2 .
(1)
We will explore the use of complex numbers in solving that equation and related equations.
For a further exploration of these topics, Professor Bill Bialek has prepared detailed notes
for the Integrated Science class this year. I have drawn from those notes to prepare these.

General Properties of the Problem. Let us begin with some mathematical properties
of differential equations. Equation 1 includes a second derivative, so it is called a secondorder differential equation, and its general solution will include two constants which will have
to be determined from the initial conditions of the problem. The equation is linear because
all its terms involving x are proportional to x.Because it is a linear differential equation, the
sum of any two solutions will also be a solution. (One last special thing about Equation 1
is that the coefficients in it do not depend on time.) An important theorem is that the
functions which satisfy linear differential equations are unique. That means that if we find a
function that satisfies a linear differential equation and the initial conditions of the problem,
then we are done; there is no other solution.
Last week we discussed the general form of the solution to Equation 1:
x(t) = A cos(t + ),

(2)

where A and are constants which must be determined from the initial conditions of the
problem. Any object undergoing motion of the form described by Equation 2 is said to be
undergoing simple harmonic motion, or SHM. We can find an expression for by taking the
second derivative of x(t),
 
d dx
d
d2 x
=
=
(A sin(t + )) = 2 A cos(t + ) = 2 x,
2
dt
dt dt
dt
1

and putting it into the original differential equation, Equation 1:


F = kx = m( 2 x).
We see that k = m 2 so =

k/m.

Notice that we may use the trigonometry identity cos(A + B) = cos A cos B sin A sin B to
show that an alternative way of expressing the SHM solution in Equation 2 is:
x(t) = [A cos ] cos t + [A sin ] sin t = B cos t + C sin t,

(3)

where B A cos and C A sin are now the two constants which must be determined
from the initial conditions of the problem. Another way of expressing the solution is x(t) =
A sin(t + ), and A and + /2 are the two constants determined by the initial
conditions of the problem.
Does the fact that there exists a variety of ways to express the solution x(t) contradict the
earlier assertion that the functions satisfying differential equations are unique? No. A second
order differential equation is satisfied by a family of solutions in which different members
of the family have different values for the two constants (which may be written as [A, ],
[B, C] or [A, ].) By picking the right member of the family, we can find a solution which is
consistent with the initial conditions of any problem involving SHM.
You have already seen the idea of a family of solutions in the case of kinematics in cases
where an object has no force on it. The position of such an object as a function of time is
x(t) = x0 + v0 t, where x0 and v0 are the initial conditions of the object: its position and
velocity at t = 0. In other words, the general solution is that the plot of x versus t is a
straight line, but the intercept and slope of the line are determined by x0 and v0 , respectively.
Thus, there is a family of lines, all of which satisfy the differential equation F = ma = 0,
but to pick one of them, we must know the initial position and velocity of the object.
We pause here to quote A. P. French1 : The [...] reason the mathematical one for the
profound importance of purely sinusoidal vibrations is to be found in a famous theorem
propounded by the french mathematician J. B. Fourier in 1807. According to Fouriers
theorem, any disturbance that repeats itself regularly with a period T can be built up from
(or is analyzable into) a set of pure sinusoidal vibrations of periods T , T /2, T /3, etc., with
appropriately chosen amplitudes i.e., an infinite series made up (to use musical teminology)
of a fundamental frequency and all its harmonics.

French, A. P., Vibrations and Waves, p.5, W. W. Norton & Company, Inc., New York, NY, 1971.

A cos

A sin

imaginary

(a, b)

x
|z|

= t +

|z|cos

Graphical representation of x(t) = Acos(t+)


as the x-coordinate of the position vector of a
particle undergoing uniform circular motion.

|z| sin
real

Graphical representation of z=a+ib as a


point in the complex plane.

Eulers Approach. Recall that another place in which the function x(t) = A cos(t + )
arises is in the position vector r(t) = x(t)i + y(t)j of an object undergoing circular motion
of radius A about the origin of an xy coordinate system, with constant speed v = A, if the
object is at x0 = A cos and y0 = A sin at time t = 0. See the left side of Figure for a
depiction.
The observation that SHM has a geometrical solution leads to a very convenient method for
working with the differential equation of particles experiencing linear restoring forces. We
will consider motion in the complex plane, rather than the xy plane, so we begin with some
discussion of complex numbers.
A complex number can be written as a sum of a purely real part and a purely imaginary
part: z = a+ib, where i is the imaginary number defined by i2 = 1, and the coefficient
b is a real number.
Each complex number has a complex conjugate which is denoted z ; if z = a + ib, then
z = a ib.
The real part of a complex number z is often written as Re[z], while the imaginary
part is described as Im[z]. Note that if z = a + ib, then Re[z] a and Im[z] b.
The magnitude of a complex number z = a + ib is denoted by |z| and is found as
p

|z| = zz = (a + ib)(a ib) = a2 i2 b2 = a2 + b2


.
A complex number z = a + ib can be written as z = |z|ei using Eulers formula, which
states that
ei = cos + i sin .
3

The complex number z can be thought of as describing a point in the complex plane.
The vertical axis in the complex plane is the imaginary axis, while the horizontal axis
is the real axis. Thus the point z = 1 + 2i is located one unit to the right of the origin,
and two units up. The angle is the angle between the vector to the point z, and the
real (horizontal) axis. From this geometric picture, which is shown in the right side of
Figure , we see that tan = b/a.
The sum of two complex numbers z1 = a1 + ib1 and z2 = a2 + ib2 is found by adding
the real parts and the imaginary parts:
z1 + z2 = (a1 + a2 ) + i(b1 + b2 ).
The product of two complex numbers z1 = a1 +ib1 and z2 = a2 +ib2 is found as follows:
z1 z2 = (a1 + ib1 )(a2 + ib2 ) = a1 a2 + i2 b1 b2 + ia2 b1 + ib1 a2 = (a1 a2 b1 b2 ) + i(a2 b1 + b1 a2 ).
Note that if instead you have expressed the complex numbers as z1 = |z1 |ei1 and
z2 = |z2 |ei2 then you can use the rules for multiplication of exponentials to find the
product as:
z1 z2 = |z1 ||z2 |ei(1 +2 ) .
Dividing one complex number by another can be done quickly if the two numbers are
expressed as z1 = |z1 |ei1 and z2 = |z2 |ei2 ; then one finds:
|z1 | i(1 2 )
z1
e
.
=
z2
|z2 |
If one begins with the two numbers expressed as z1 = a1 + ib1 and z2 = a2 + ib2 , then
it is useful to use the following trick to clear the denominator of imaginary parts:
z1
z1 z2
(a1 + ib1 ) (a2 ib2 )
a1 a2 + b1 b2 + i(b1 a2 b2 a1 )
=
=
.
=

z2
z2 z2
(a2 + ib2 ) (a2 ib2 )
a22 + b22

Using Complex Numbers to Solve Physics Problems. We saw above that x(t) =
A cos(t + ) describes the x coordinate of a particle undergoing uniform circular motion
about the origin of the xy plane. We could also say that it describes the real part of a
complex vector z(t) = A exp(i(t + )) which is rotating about the origin of the complex
plane with angular velocity . The beauty of this approach is that it is very easy to take
multiple derivatives of eit ; each successive derivative brings another factor of i to multiply
the exponential.
Let us consider the SHM differential equation of Equation 1. We can find the complex
function z(t) which solves Equation 1 and then take as our solution only the real part,
Re[z(t)] x(t). Since we already know the solution should have Re[z(t)] x(t) = A cos(t+
), we expect the solution to be z(t) = A exp(i(t + )) to agree with what we have already
4

done. As an exercise, try putting that form in for z(t) and then taking the real part at the
end.
Here, we will try a slightly more general approach. Let us guess that the solution will be an
exponential, z(t) = A exp(t). Note that if is real and positive, the function z(t) grows
exponentially with time, but if is imaginary, the function oscillates in time. Here we insert
that form for z(t) into Equation 1:
kz = m

d2 z
d
= m ( exp(t)) = m2 exp(t) = m2 z.
2
dt
dt

p
In order for this equation to be satisfied, we must have 2 = k/m, so that = i k/m;
in other words, we have found to be imaginary, so the solution function oscillates in time,
as we expected! In terms of our previous terminology, = i. Now we need to discuss what
is to be done with the in the exponent. Either sign gives a valid solution, so the general
form of the solution is z(t) = A+ exp(t) + A exp(t) = A+ exp(it) + A exp(it),
where we have replaced the arbitrary (complex) constant A with two arbitrary (complex)
constants A+ and A . We will expand each of these into its real and imaginary parts below,
to make the solution look more familiar:
z(t) = A+ (cos(t) + i sin((t)) + A (cos(t) i sin((t)) ,
= (Re[A+ ] + Re[A ]) cos(t) + (iIm[A+ ] iIm[A ]) (i sin(t))
+i [(Re[A+ ] Re[A ]) sin(t) + (Im[A+ ] + Im[A ]) (cos(t))]
= B cos(t) + C sin(t) + i [D sin(t) + E cos(t)] ,
where we have written the result in the last line in terms of four real constants, B
Re[A+ ] + Re[A ] and C Im[A ] Im[A+ ], and similar combinations for D and E, which
will not concern us. Note that we made use of i2 = 1 between the second and fourth lines.
We now assert x(t) = Re[z(t)] to find
x(t) = B cos(t) + C sin(t),
p
with B and C arbitrary real constants and = k/m, which is one of the formulations we
have found before!

Damped Harmonic Oscillator. In order to see the power of this method, now consider
the case of damped motion, in which, for example, the entire mass + spring system is
submerged in viscous honey, so that the mass m experiences not only the linear restoring
force from the spring with spring constant k, but also a drag force which we take to be
propotional to the masss velocity, Fdrag = bv = b(dx/dt). Then we apply Newtons law
to show:
dx
d2 x
Fnet = kx b
= ma = m 2 ,
(4)
dt
dt
which is the differential equation for a damped harmonic oscillator. Let us again try an
exponential solution for the complex function z(t) of the form z(t) = A exp t. As before,
5

we will take the real part of z(t) as the answer when we are done. We substitute z(t) into
Equation 4, noting that dz/dt = A exp(t) = z:
kz bz = m2 z;

m2 + b + k z = 0.
We require the coefficient of z to itself equal 0 so that the equation will be true at all times
t, which results in a quadratic equation in with solution
s
r
2

1
k

b
b
b
2

=
02 ,
=
b2 4mk =
2m 2m
2m
2m
m
2
4
wherepwe have written b/m and 02 = k/m to shorten the notation. The quantity
0 = k/m is often called the natural frequency of the spring/mass system. (Note that
Knight defines the time constant = m/b = 1/.) As before, the in the exponent
means the general solution will be a sum with two arbitrary complex coefficients: z(t) =
A+ exp(+ t) + A exp( t). Notice that we can distinguish three interesting cases depending
on whether the argument of the square root is positive, negative, or zero.

Underdamped Oscillations. Consider the case where the damping term is small, as
defined by 2 /4 < 02 , so that the argument of the square root is negative. In that case, we
can identify 12 = 02 2 /4 and = /2 i1 . Then we find (with a bit of algebraic
manipulation of the two arbitrary complex constants) that
x(t) = Re[z(t)] = et/2 [B cos(1 t) + C sin(1 t)]
= Aet/2 cos(1 t + ),
where in the last line we used the two arbitrary real constants A and instead of the equally
valid set B and C, in analogy to what was
p done in Equation 3. As Knight notes, in many
cases, 2 /4 << 02 so that 1 0 = k/m. In those cases, it is easy to picture the
solution: x(t) looks like an oscillating wave with nearly the same period as it would have
had in the absence of damping, with slowly decaying amplitude, as in Figure 14.22 of Knight.
Note that as long as the damping coefficient b is constant, 1 is constant, so the period of
the oscillating part of the motion is constant.

Overdamped Oscillations Consider the case where the damping term is large, as
defined by 2 /4 > 02 , so that the argument of the square root is positive. In that case, we
find are completely real, so that for Re[z(t)] x(t) we need only take the real parts of
the two arbitrary coefficients and can write the general solution as

x(t) = et/2 Bet + Cet ,
p
where 2 /4 02 is a real number. We have two arbitrary constants B and C with
which to match the initial conditions of the problem (x(t = 0) and v(t = 0)).
6

Critically Damped Oscillations The special case of 2 /4 = 02 is called critical damping.


In that case, the argument of the square root vanishes, so that = + and now the solution
only has one arbtirary constant. You will learn in your differential equations class that
the solution is therefore not sufficient. It turns out that the full solution in that case is
x(t) = (A + Bt) exp t/2, which you may prove to yourself by substituting that form back
into Equation 4 if you like.

Forced Oscillations With these methods, we can also solve the case of a mass on a spring
with damping which is
pbeing subjected to an external force F = F0 cos(t), where might
not be equal to 0 = k/m. Then Newtons law shows:
Fnet = F0 cos(t) kx b

dx
d2 x
= ma = m 2 .
dt
dt

This is a different type of differential equation than we have yet discussed; it is inhomogeneous
in that there is a constant term in it which is not proportional to x or its derivatives. We
will see that the most important aspect of its solution is the steady state solution, which
in this case, as you might expect, is the state in which the mass is moving with the same
angular frequency as the driving force.
In order to use our technique of generalizing the differential equation to describe a complex
function, we will generalize the force also, and take F = F0 exp it. Then, after some
regrouping of terms, we find
dz
F0 it d2 z
e = 2 + + 02 z,
m
dt
dt

(5)

where = b/m and 02 = k/m as before. Here since we expect the motion to be governed
by the driving force, a natural choice for a solution is z(t) = Aei exp(it) = A exp i(t + ),
with A and real constants. That is, we explicitly allow for the possibility that the motion
might not be in phase with the driving force, but require that it will have the same period.
This is very easy to plug into Equation 5:
F0 it
e = (i)2 Aei eit + iAei eit + 02 Aei eit ;
m

F0 i
e = 2 + i + 02 A,
m
where we have eliminated the common exponential in the second line. Now we can solve for
the real constants and A by writing ei = cos() + i sin and then requiring the real parts
of the left side of the equation above equal the real parts of the right side, and similarly for
the imaginary terms:
F0
cos = (02 2 )A;
m
F0
sin = A.
m
7

(6)
(7)

The ratio of Equation 7 to Equation 6 gives:


tan =

,
02

while squaring and adding the two equations gives:


A2 =
A() =

(F0 /m)2
;
( 2 02 )2 + 2 2
F0 /m
[( 2 02 )2 + 2 2 ]

1/2

Then the motion of the mass is


x(t) = Re[z(t)] = A() cos(t + ()),
and we see that the amplitude of the motion depends on the driving frequency and that
the amplitude reaches a maximum when = 0 , which is the condition known as resonance.

Summary We have considered here only the case of the mass on the spring. In your
studies this week, you will come across several other situations where you find a differential
equation with the general form:
d2 x
+ 02 x = 0,
dt2
though the equation might be for a variable other than the position; it might be for an angle
(t) in the case of the pendulum, for example. Next semester you will study the differential
equation of Equation 5 in great detail in the context of electronic alternating current (AC)
circuits.
Suzanne Staggs; 25 Nov 2006.

You might also like