Sie sind auf Seite 1von 9

lecture 3

Topics:
Where are we?
Consequences of Time Translation Invariance and Linearity
Uniform circular motion
Harmonic oscillation for more degrees of freedom
The double pendulum
The damped harmonic oscillator
Where are we?
Last time, we saw how initial conditions appeared in a number of different examples of force laws.
The last of these, the harmonic oscillator, is a particularly important system because it has two
general properties, time translation invariance and linearity, that appear in many many physical
systems.
Because linearity went by pretty quickly last time, let me briey review how it works. The
equation of motion, F = ma, for the harmonic oscillator is linear because there is a single x in
each term. It can be written as
m
d
2
x
dt
2
+ K x = 0 (1)
We can think of this as a single operator acting on x.
=
_
m
d
2
dt
2
+ K
_
x = 0 (2)
In this form, it may be more clear why you the solutions form a linear space. If you have two
solutions, x
1
(t) and x
2
(t), you can form an arbitrary linear combinations and still get solutions,
because the same operator acts on both, and multiplying by a constant doesnt affect the validity
of the solution,
_
m
d
2
dt
2
+ K
_
x
1
= 0
_
m
d
2
dt
2
+ K
_
x
2
= 0

_
m
d
2
dt
2
+ K
_
(a x
1
+ b x
2
) = 0
(3)
This fact has remarkable consequences, as we will see shortly.
Consequences of Time Translation Invariance and Linearity
Time translation invariance is an example of a symmetry. The physics of the harmonic oscillator
looks the same if all clocks are reset by the same amount. When a symmetry is combined with
the property of linearity, the result is an extremely powerful tool for studying the solutions of the
systems equation of motion. The reason is that because of linearity, the solutions of the equation
1
of motion form what mathematicians call a linear space. You can add them together and multiply
them by constants and you still have solutions. Because of this, we can use the tools of linear
algebra to understand them. In particular, we can choose a convenient set of basis solutions that
behave as simply as possible under time translations. For the symmetry of time translation, it is a
mathematical fact that the basis solutions are just exponentials. We can always nd solutions of
the form
1
z(t) = z(0) e
Ht
(4)
What is special about this form (and I am not going to discuss this in detail - I hope that you will
see this beautiful argument in more detail in Physics 15c) is that when you change the setting of
your clock by taking t t + a, the exponential (4) is the only function that just changes by a
multiplicative constant,
z(t + a) = z(0) e
H(t+a)
= z(0) e
Ht
e
Ha
= e
Ha
z(t) (5)
You can always use the linearity of the space of solutions to nd particularly convenient solutions
that behave in this simple way under time translations - and then the result has to be an exponential.
Once you realize that the solutions are just exponentials, you can nd the possible values of H
in a very simple way. If you do take (4) and put it into the equation of motion, time derivatives
acting on e
Ht
just bring down factors of H. This converts the differential equation into an algebraic
equation.
d
dt
H (6)
Then once we have the basis soltutions, we can form linear combinations to satisfy the initial
conditions, in the same way that we can write a general vector in three dimensional space as a sum
of coordinates times basis vectors
r = x x + y y + z z
This simple dependence of r on the coordinates x, y and z is the power of linearity at work. We will
see that it works in a similarly simple way for the initial conditions in a linear and time translation
invariant mechanical system.
Lets do this for the linear frictional force we talked about last time, of the form
mv (7)
with
0 (8)
We know how to solve this directly by integrating, as we did last time, but you will notice, I hope
that this force law is also time translation invariant and linear. There is just a single factor of x in
v =
dx
dt
(9)
and t enters only through derivatives. Therefore, we expect that the solution is a sum of exponen-
tials in time times coefcients that depend on the initial conditions. The equation of motion has
the form
m x = m x (10)
1
This is essentially equivalent to the statement that a linear differential equation with constant coefcients always
has an exponential solution see Morins text.
2
This allows us to solve the problem in a different way, by assuming that the solution is a linear
combination of exponentials. If we put in a trial solution of the form
x(t) = e
H t
(11)
each dot becomes a factor of H and we get
mH
2
e
H t
= mH e
H t
(12)
or
H
2
= H H = or H = 0 (13)
so there are two simple solutions,
x(t) = e
0
= 1 and x(t) = e
t
(14)
The general solution, because of linearity, is then a combination of these two solution with coef-
cients that depend on the initial conditions:
x(t) = A + B e
t
(15)
Comparing with the solution that we got by direct integration,
x(t) = x(t
0
) +
v(t
0
)

_
1 e
(tt
0
)
_
(16)
we see that it is equivalent to (15) with
A = x(t
0
) +
v(t
0
)

(17)
and
B =
v(t
0
)

e
t
0
(18)
I hope you agree that this calculation was quite a bit easier and quicker than integrating. Using
these general principles of time translation invariance and linearity is not only cool, it saves you
work. As you will see in more complicated examples in a few minutes, it often saves you a LOT
of work!
Before going on to the harmonic oscillator, lets describe this process in the form of two simple
steps for dealing with systems with time translation invariance and linearity:
2
1. Put a trial solution of the form e
Ht
into your equation of motion. The derivatives become
powers of H and you can nd the values of H = h
j
that work by solving an algebraic
equation. This gives you your basis solutions e
h
j
t
2. Write a general solution by making a general linear combination of your basis solutions
x(t) =

j
A
j
e
h
j
t
(19)
and nd the constants A
j
by imposing initial conditions, as usual.
2
The only time this procedure doesnt work is when the algebraic equation you get has degenerate roots. We will
see what to do in this special case later.
3
Now lets apply these two steps for the mass on the spring. Step 1 is straightforward.
m z(t) = K z(t) (20)
z(t) e
Ht
(21)
m
d
2
dt
2
e
Ht
= K e
Ht
(22)
mH
2
e
Ht
= K e
Ht
(23)
_
mH
2
+ K
_
e
Ht
= 0 (24)
mH
2
+ K = 0 (25)
H = i for i =

1 and =

K
M
(26)
So our two basis solutions are
e
it
and e
it
(27)
The calculation in step 1 was very simple. The only curious thing here is that we are led to
complex numbers. While time translation invariance tells us that there are solutions of the form
e
Ht
, it doesnt tell us that H is real, and for oscillations, it isnt.
Now in step 2, we form the general solution by forming a general linear combination of our
basis solutions. Thus the most general solution for the harmonic oscillator looks like this:
x(t) = c e
it
+ d e
it
(28)
Now this is a little peculiar. Unlike the situation with the frictional force, in (16), this doesnt look
the same as the cosine and sine that we got by solving the differential equation. But in fact, it is the
same. The connection with sines and cosines is Eulers formula, one of the more amusing relations
in mathematics:
e
i
= cos + i sin (29)
There are many ways of seeing this. Lets just use the most important formula in physics again
Taylors expansion
e
i
= 1 + (i) +
1
2
(i)
2
+
1
3!
(i)
3
+
1
4!
(i)
4
+ (30)
=
_
1
1
2

2
+
1
4!
()
4
+
_
+ i
_

1
3!

3
+
_
(31)
Because of Eulers formula, you see that the general solution in terms of cosine and sine is
completely equivalent to the general solution in terms of complex exponentials.
x(t) = a cos t + b sin t
is equivalent to
x(t) = c e
it
+ d e
it
= c (cos t + i sin t) + d (cos t i sin t)
a = c + d b = i(c d)
(32)
4
Notice again how linearity is at work here. Linearity is what guarantees that a linear combination
of two possible trajectories is another possible trajectory. This is what allows us to write the most
general solution as a combination of the two complex exponential solutions times constants:
x(t) = c e
it
+ d e
it
(33)
It is the fact that the initial conditions appear in this extremely simple way as the coefcients of
simple basis solutions that makes all of this work.
If you havent seen this before, and perhaps even if you have, this probably looks really strange.
And you might also be asking yourself, if this is equivalent to the familiar solution in terms of co-
sine and sine, what is the advantage of using these unfamiliar complex exponential. The right
answer, I think, is that once you get used to using complex exponentials, they will simplify your
life a lot. We will see this in a few minutes when we discuss damped oscillators, but the message
is really more general. We dont have to use complex exponential. We can do everything using
cosines and sines, using a combination of trigonometry and algebra. But with complex exponen-
tials, all we need is algebra!
In fact, Eulers formula is the connection between algebra and trigonometry! You can dene
the trigonometric functions this way:
cos
e
i
+ e
i
2
(34)
sin
e
i
e
i
2i
(35)
Now you can derive all trigonometric identities just using algebra, and you never have to do
trigonometry again.
Uniform circular motion
One very evocative way to think about these complex solutions is in what is called the complex
plane. Because a complex number has two real components, its real and its imaginary part, we
can think of a complex number as a real vector in a two dimensional space in which the real part
is the x component of a two dimensional vector and the imaginary part is the y component. This
two dimensional space is the complex plane. Eulers formula, (29), tells us that the basis solution
e
it
has real part cos t and imaginary part sin t, so its counterpart in the complex plane is the
two dimensional vector, (cos t, sin t),
e
it
= cos t + i sin t (cos t, sin t) (36)
But this is a unit vector an angle t from the x axis. Thus as t increases, e
it
executes uniform
circular motion in the complex plane. You can see this in the ANIMATION circular motion.exe
More generally, a complex number z = x+iy can be written equivalently as a positive number
R times a complex exponential e
i
. Note the connection of this with the relation between Cartesian
and Polar coordinates in the complex plane.
z = x + iy = Re
i
(x, y)
Cartesian
(R, )
Polar
(37)
5
R = |z| =
_
x
2
+ y
2
(38)
= arg(z) (39)
=
_

_
arctan(b/a) for a 0 ,
arctan(b/a) + for a < 0 .
(40)
Harmonic oscillation for more degrees of freedom
Because our analysis of the harmonic oscillator is very general, relying only on the general princi-
ples of linearity and time translation symmetry, the result of (4)
z(t) = z(0) e
Ht
(4)
applies to any system satisfying these two principles. For example, it is not necessary to restrict
ourselves to a single degree of freedom. With more degrees of freedom, z(t) becomes a vector
with number of components equal to the number of degrees of freedom, as does z(0) in (4). Thus
(4) implies that there are special solutions in which all the components of z move in lockstep, with
the same angular frequency. Such a motion is called a normal mode. The same two steps sufce
to solve these more complicated problems, but now there are more basis solutions (because there
are more degrees of freedom) and each of the basis solutions describes a motion of ALL the parts
of the system.
The double pendulum
Here is a very simple example of normal modes that I hope will make the idea clear. Consider the
double pendulum, which looks like this:
.........................................................................................................................................................................................................
.........................................................................................................................................................................................................
1 2



(41)
Two identical pendulums are constrained to move in the plane of the paper and coupled together
by a massless spring with spring constant K. In this case the conguration can be labeled by two
numbers, x
1
and x
2
the displacements of block 1 and block 2 from equilibrium. Thus this is a
system with two degrees of freedom. The vector q(t) that describes the conguration is just
q(t) =
_
x
1
(t)
x
2
(t)
_
(42)
6
Step 1 is now more complicated because we have to nd the normal modes. Without the spring,
the two pendulums would oscillate independently. For small oscillation, the oscillation of a single
pendulum is harmonic with angular frequency =
_
g/. The spring couples these motions
together. However, the normal modes are still harmonic. There are two basis solutions for each
normal mode.
The normal modes look like this:
.......................................................................................................................................................................................................
.......................................................................................................................................................................................................
1 2



(43)
There is one normal mode in which the blocks move together. In this mode, the spring in the
middle is never stretched from its equilibrium length. The angular frequency of this mode is just
the same as the angular frequency of a single pendulum, which (for small oscillations for which
the system is linear) is
1
=
_
g/ where is the distance from the pivot to the mass. The basis
solutions that describe this normal mode are
z(t) =
_
1
1
_
e
i
1
t
and z(t) =
_
1
1
_
e
i
1
t
(44)
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
........................................................................................................................................................................................................
1 2

..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
..............................
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . .
(45)
There is another normal mode in which the two pendulums move in precisely opposite directions.
The frequency of this mode is slightly higher than
_
g/, because now the spring contributes to
the restoring force that produces the oscillation. In fact, in this case,
2
=
_
2K + g/. The basis
solutions that describe this normal mode are
z(t) =
_
1
1
_
e
i
2
t
and z(t) =
_
1
1
_
e
i
2
t
(46)
7
The thing about a normal mode is that the ratios of the displacements of all the parts of the
system are xed throughout the motion. In the double pendulum, in the motion (43), x
1
(t)/x
2
(t) =
1 throughout the motion, while in (45), x
1
(t)/x
2
(t) = 1.
You should have read a bit more about these in Morins book. In general, it is not easy to nd
normal modes. In a case like this, you can guess them just from the symmetry of the system. The
details of nding them in general involves heavy-duty linear algebra, and this can wait for Physics
15c. What I care about in this course is that you know what these things mean and how to use them
if someone tells you what they are.
The point is that step 2 is the same as before. The general solution is always obtained by taking
a general linear combination of the basis solutions:
_
x
1
(t)
x
2
(t)
_
= C
1
_
1
1
_
e
i
1
t
+ D
1
_
1
1
_
e
i
1
t
+ C
2
_
1
1
_
e
i
2
t
+ D
2
_
1
1
_
e
i
2
t
(47)
This is short-hand notation for two equations,
x
1
(t) = C
1
e
i
1
t
+ D
1
e
i
1
t
+ C
2
e
i
2
t
+ D
2
e
i
2
t
x
2
(t) = C
1
e
i
1
t
+ D
1
e
i
1
t
C
2
e
i
2
t
D
2
e
i
2
t
(48)
As usual, we can use Eulers equation to rewrite this in terms of sines and cosines.
_
x
1
(t)
x
2
(t)
_
= A
1
_
1
1
_
cos
1
t + B
1
_
1
1
_
sin
1
t + A
2
_
1
1
_
cos
2
t + B
2
_
1
1
_
sin
2
t (49)
Again, this is short-hand notation for two equations,
x
1
(t) = A
1
cos
1
t + B
1
sin
1
t + A
2
cos
2
t + B
2
sin
2
t
x
2
(t) = A
1
cos
1
t + B
1
sin
1
t A
2
cos
2
t B
2
sin
2
t
(50)
The damped harmonic oscillator
Lets add the linear friction we talked about earlier to our harmonic oscillator. With such a frictional
term, the equation of motion for the mass on a spring becomes
m
d
2
dt
2
x(t) = m
d
dt
x(t) K x(t) (51)
d
2
dt
2
x(t) +
d
dt
x(t) +
2
0
x(t) = 0 (52)
where
0
=
_
K/m. This still satises the conditions of time translation invariance and linearity.
Therefore we still expect the solutions to be an exponentials and we can still use our two steps to
construct the general solution.
Step 1 is the usual. Because of time translation invariance and linearity, we can look for expo-
nential basis solutions:
z(t) = e
Ht
(53)
8
d
2
dt
2
e
Ht
+
d
dt
e
Ht
+
2
0
e
Ht
= 0 (54)
As usual, derivatives with respect to t just bring down factors of H, so we can convert this to an
algebraic equation:
(H
2
+ H +
2
0
) e
Ht
= 0 (55)
Notice that everything has gone exactly the same way here as in our two previous example. Indeed
the only difference here is that the algebraic equation, (55), is a little more complicated that the
previous examples. We need to use the quadratic formula to nd the two solutions:
H =

2
4

2
0

(56)
where
> 0 Re H < 0 Re

> 0 (57)
z(t) oscillates if
2
0
>
2
/4 and it just dies out if
2
<
2
/4
z(t) e

t
=
with t
..
e
t/2

if /2>
0
..
e
t

2
/4
2
0
. .
if /2<
0
(58)
Step 2 is also the usual one. The general solution is a linear combination of the simple expo-
nential basis solutions:
x(t) = b
+
e

+
t
+ b

t
(59)
where

are dened in (56). The constants b

contain the information about the initial conditions,


just as in the undamped harmonic oscillator.
The nature of the trajectories described by (59) depends on the relative size of the two parame-
ters, and
0
. If /2 >
0
, the damping is large (this is called overdamped). In this case, both

+
and

are real and positive, and the trajectory is a sum of decaying exponentials.
If /2 <
0
, the damping is small (this is called underdamped). In this case, both
+
and

have a positive real part and an imaginary part (with opposites signs). In this case, the trajectory
oscillates (or circles in the complex plane), but also dies out with time exponentially in t the
ANIMATION damped.exe shows the oscillating case.
There is a slightly peculiar special case between these two if /2 =
0
, the system is
critically damped. In this case, the general solution is
x(t) = (A + B t) e
t/2
(60)
It is useful to look at a couple of limits of (59) to make contact with things we have already
done.
In the limit
0
0, the damped harmonic oscillator reduces to a system with the linear fric-
tional force, (7). This is automatically an overdamped situation, because there is no oscillation at
all. Indeed, in this case, (59) reduces to (15), as it must.
Similarly (though perhaps more obviously), in the limit 0, the damped harmonic oscillator
is underdamped and goes smoothly into the ordinary harmonic oscillator as the damping goes away.
9

Das könnte Ihnen auch gefallen