17 Lecture 11-4

17.1 Chapter 7 Lagranges Equations (con)

17.1.1 Examples of Lagranges Equations
We will now consider four examples of Lagranges equations. The rst two are
relatively simple and can be easily solved using Newtons second law. They are
included to provide some experience with using the Lagrangian approach. NTL,
even these simple cases show some advantages over the Newtonian formalism
in that they obviate any need to consider the forces of constraint. The last
two are suciently complex that the solution using the Newtonian approach
requires considerable ingenuity. By contrast the Lagrangian approach lets us
write down the equations of motion almost without thinking.
The Lagrangian formalism always (at least almost always) provides us with
a straightforward means of writing down the equations of motion. However,
they cannot guarantee that the resulting equations are easy to solve. But even
if an analytic solution is not possible, writing down the equations of motion is
an essential rst step to understanding the solutions. In a worst case scenario,
we can always solve the equations of motion numerically and usually solve for
the positions of equilibrium very quickly.
Atwoods Machine In an Atwood machine, see gure 7.3, the two masses,
Figure 7.3 Atwood machine with constraints. Because the length of the string
is xed, the position of the whole system can be specied by the single
variable r.
:
1
and :
2
are suspended by an inextensible string of length | which passes over
a pulley with frictionless bearings, a radius 1. and a momentum of inertia 1.
The kinetic energy for this system is
T =
1
2
:
1

r
2
+
1
2
:
2

n
2
+
1
2
1

c
2
. (1)
1
where

c is the rate of angular rotation of the pulley. Because the length of the
string is xed the heights of the two masses, r and n, cannot vary independently.
Rather r + n + 1 = |. From this we see that

r =

n. Additionally from a no
slip condition between the string and the pulley the angular frequency of the
pulley satises 1

c =

r. We can now write the kinetic energy for the Atwood
machine as
T =
1
2
:
1

r
2
+
1
2
:
2

r
2
+
1
2
1
1
2

r
2
. (2)
The potential energy (to within a constant) is
l = :
1
or :
2
on = (:
1
:
2
) or. (3)
Combining these expressions we nd the Lagrangian to be
L = T l =
1
2

:
1
+ :
2
+ 11
2

r
2
+ (:
1
:
2
) or. (4)
The Lagrange equation of motion is
0L
0r
=
d
dt
0L
0

r
!(:
1
:
2
) o =

:
1
+ :
2
+ 11
2

r. (5)
Solving for the acceleration we nd

r = (:
1
:
2
) o

:
1
+ :
2
+ 11
2

. (6)
By choosing :
1
and :
2
fairly close together, one can make this acceleration
much less than o. Hence, the Atwood machine gave an early and reasonably
accurate method for measuring o.
What is relevant here is that the corresponding Newtonian solution requires
three free body equations which involves the two constaint forces, the tensions
T
1
and T
2
. The result is three equations with three unknowns, the tensions
T
1
. T
2
. and the acceleration

r. We can eliminate both T
1
and T
2
which reduces
the three equations to the expression shown in equation (6). The Lagrangian
solution of the Atwood machine is too simple to truly appreciate the advantage
of this approach. NTL the Lagrangian approach did eliminate the need to reduce
three Newtonian equations to the one that see in equation (6).
Particle Conned to Move on a Cylinder Consider a particle of mass
: constrained to move on a frictionless cylinder of radius 1. see gure 7.4.
Besides the force of constraint (the normal force from the wall of the cylinder),
the particle experiences a force due to a spring anchored at the origin (j = . = 0)
given by
!
1 = /
!
r . Since the particles radial coordinate is xed at j = 1. we
can specify the position of the particle with just it height, ., and its angular
2
Figure 7.4 A mass : is conned to the surface of a cylinder, j = 1. and
subject to Hooks law,
!
1 = /
!
r .
coordinate c. Hence the kinetic energy of the particle is
T =
1
2
:
2
=
1
2
:

.
2
+ 1
2

c
2

. (7)
The potential energy of the spring is
l =
1
2
/r
2
=
1
2
/

j
2
+ .
2

=
1
2
/

1
2
+ .
2

. (8)
and the Lagrangian is
L =
1
2
:

.
2
+ 1
2

c
2

1
2
/

1
2
+ .
2

. (9)
Since the system has two degrees of freedom, there are two equations of
motion. The equation for the . coordinate is
0L
0.
=
d
dt
0L
0

.
!/. = :

.. (10)
The c equation is even simpler as the Lagrangian does not depend on c. and
the c equation is
d
dt
0L
0

c
= :1
2

c = 0. (11)
The . equation tells us that the particle undergoes simple harmonic motion in
the . direction, . = cos (.t c) . The c equation tells that the quantity :1
2

c
is constant, i.e. the angular momentum about the . axis is conserved. Since
there is no torque on the particle, this is a result that we should have expected.
The entire motion of the particle is that of moving around the cylinder at a
constant angular velocity while oscillating in the . direction about . = 0 at
angular frequency . =
p
/:.
3
Block sliding on a Wedge Consider the block and wedge shown below,
Figure 7.5 A block of mass : slides down an incline of mass ' that resides on
a frictionless horizontal table.
in gure 7.5. The block of mass : is free to slide on the wedge, and the wedge
of mass ' is free to slide on the horizontal table, both with negligible friction.
The block is released from the top of the wedge while both are initially at rest.
The wedge has an angle c and the length of its slope is |. and we are interested
in determining how long it takes the block to reach the bottom.
The system has two degrees of freedom, one for the block and one for the
wedge. A reasonable choice for the generalized coordinates is shown in gure
7.5, and they are the distance of the block from the top of the wedge, c
1
. and
the distance of the wedge from some xed point on the table, c
2
.
The kinetic energy of the wedge is simply T
M
=
1
2
'

c
2
2
. however the kinetic
energy of the block is a bit more complicated. The block has a velocity

c
1
down
the wedge, but that velocity is relative to the wedge not the table. Remembering
that we need to write the Lagrangian in a nonaccelerating frame requires that
we express the kinetic energy of the block in the inertial frame of the table. The
r and n components for the velocity of the block are

x
=

c
1
cos c +

c
2
and
y
=

c
1
sinc. (12)
Thus the kinetic energy of the block is
T
m
=
1
2
:

2
x
+
2
y

=
1
2
:

c
2
1
+

c
2
2
+ 2

c
1

c
2
cos c

. (13)
The potential energy of the wedge is constant while that of the block is :on,
where n = c
1
sinc. Therefore the potential energy of the system is simply
l = :oc
1
sinc. (14)
and the Lagrangian is
L =
1
2
'

c
2
2
+
1
2
:

c
2
1
+

c
2
2
+ 2

c
1

c
2
cos c

+ :oc
1
sinc. (15)
Once we have found the Lagrangian, all that is left is two write down the
two Lagrange equations, one for c
1
and the other for c
2
, and then of course solve
4
them. The c
2
equation is
0L
0c
2
=
d
dt
0L
0

c
2
. (16)
Since the Lagrangian is independent of c
2
, we know that the generalized mo-
mentum 0L0

c
2
is constant,
'

c
2
+ :

c
2
+

c
1
cos c

= co::t. (17)
This is the sum of the total momentum in the r direction and something you
could have written down without any help from the Lagrangian.
The c
1
equation
0L
0c
1
=
d
dt
0L
0

c
1
(18)
is more complicated, since neither derivative vanishes. Performing the partial
derivatives we nd
:o sinc =
d
dt
:

c
1
+

c
2
cos c

= :

c
1
+

c
2
cos c

. (19)
We now have two equations and wish to solve for

c
1
. Dierentiating equation
(17) and solving for

c
2
we nd
(' + :)

c
2
= :

c
1
cos c !

c
2
=
:
' + :

c
1
cos c. (20)
Substituting this result into equation (19) yields
:o sinc = :

1
:
' + :
cos
2
c

c
1
.

c
1
=
(' + :) o sinc
' + :sin
2
c
. (21)
Since the acceleration of c
1
is constant, we can integrate it immediately and nd
c
1
=
1
2
(' + :) o sinc
' + :sin
2
c
t
2
. (22)
so that the time, t, to reach the end of the wedge (c
1
= |) is
t =
q
2|

' + :sin
2
c

((' + :) o sinc). (23)

As a check we can determine if this expression agrees with several dierent
limits. In the limit of c = 2 (sinc = 1) . we have

c
1
= o. which is as it
should be. The next limit of interest is when ' ! 1. In that limit we have

c
1
= o sinc. which is again correct. The limit when ' !0 is left as an exercise
for the student, problem 7.19.
5
Bead on a Spinning Wire Hoop A bead of mass : is attached to a fric-
tionless wire hoop of radius 1. The hoop lies in a vertical plane, which is forced
to rotate about the hoops vertical diameter with a constant angular velocity,

c = .. as shown in gure 7.6. The bead position on the hoop is specied by the
angle 0 measured up from the vertical while the entire system is in a uniform
Figure 7.6 A bead is free to move around a frictionless wire hoop, which is
spinning at a xed rate ..
gravitational eld. The kinetic energy of the bead is T =
1
2
:1
2

0
2
+ sin
2
0.
2

.
The gravitational potential energy is (as measured from the bottom of the hoop)
is identical to that for the pendulum and is l = :o1(1 cos 0) . Hence the
Lagrangian is given by
L =
1
2
:1
2

0
2
+ sin
2
0.
2

:o1(1 cos 0) . (24)

There is only one generalized coordinate and therefore only one Lagrange equa-
tion,
0L
00
=
d
dt
0L
0

0
!:1
2
sin0 cos 0.
2
:o1sin0 = :1
2

0. (25)
Dividing through by :1
2
. we nd that the angular acceleration of 0 is

0 =

.
2
cos 0 o1

sin0. (26)
Now this expression cannot be solved analytically in terms of elementary
functions. NTL it can tell us a lot about the systems behavior. For starters we
can determine the equilibrium position, 0 = 0
o
. of the bead by setting

0 to zero
(and

0 if it appeared) in equation (26). This results in

.
2
cos 0
o
o1

sin0
o
= 0 (27)
6
This equation is satised when either of the two factors is zero. The factor
sin0
o
= 0 when 0
o
= 0 or . Thus the bead can remain at the bottom or top of
the hoop. The rst factor vanishes when
cos 0
o
= o.
2
1. (28)
Since the cosine function is less than or equal to one this expression can only be
satised if .
2
o1. When this condition is satised then there are two more
positions of equilibrium (although they are equivalent positions for a rotating
wire) whose positions are given by
0
o
= cos
1

o.
2
1

. (29)
We can then conclude that when the hoop is rotating slowly (.
2
< o1) there
are only two equilibrium positions, at the top and bottom of the hoop. However,
when the hoop is rotating fast enough (.
2
o1) there are two more, sym-
metrically placed on either side of the bottom of the ring as given by equation
(29).
So we see that at the top or the bottom of the hoop, the bead is on the
axis of rotation and j = 0. where j is the distance from the axis of rotation.
Therefore the centripetal force :.
2
j is zero. Also, at those locations the force
of gravity is normal to the hoop, so there is no force tending to move the bead
along the wire and the bead remains at rest. For the position o the axis the
force of gravity has a component that is pulling the bead inward along the wire
(as long as j0j 2) Meanwhile the centrifugal force is pushing the bead
outward along the wire. At the points given by equation (29) these forces are
balanced and the bead remains at rest.
All of this being said, the equilibrium points are not especially interesting
unless it is a position of stable equilibrium. Using the equation of motion for
0. equation (26), we can easily address this issue. First we will start with the
equilibrium at 0 = 0. For 0 near zero the equation reduces to

0 =

.
2
o1

0. (30)
If the hoop is rotating slowly, .
2
< o1, then the equation of motion near 0 = 0
is of the form

0 = /0 (/ is a positive constant). This expression is analogous
to the spring equation where any displacement away from equilibrium induces
a restoring force that pushes the object back towards equilibrium. Clearly, for
this range of frequency, this is a position of stable equilibrium. However if we
speed up the hoop so that .
2
o1. then the equation of motion is of the
form

0 = +/0 (again / is a positive constant). Any nite value of 0 induces an
acceleration toward larger 0. hence this position is now unstable. Thus as we
increase . this equilibrium position goes from being stable to unstable.
For obvious reasons the position at the top of the hoop is unstable for any
rate of rotation. If the bead uctuates just a bit away from 0 = . then the grav-
itational force and the centrifugal both push the bead away from this position.
This is easily seen in equation (26) by letting 0 = + c0 where c0 << 1.
7
The other two equilibrium positions only exist when .
2
o1. To determine
whether or not the position at 0
o
= +cos
1

o.
2
1

is stable, we will consider

small deviations away from this equilibrium position. To do this we let 0 =
0
o
+c0 = cos
1

o.
2
1

+c0. This is most easily done by expanding the function

1 (0) =

.
2
cos 0 o1

sin0 in a Taylors series about 0

o
. Since 1 (0
o
) = 0 we
nd
d1 (0
o
)
d0
=

.
2
sin0
o

sin0
o
= .
2

1 cos
2
0
o

d1 (0
o
)
d0
=

.
2
o
2
.
2
1
2

. (31)
Hence to rst order in c0 the equation of motion becomes
c

0 =

.
2
o
2
.
2
1
2

c0. (32)
We see that for small oscillations about the equilibrium position, what was
once a very nonlinear dierential equation is now linear. This makes it easy
to analyze. Since .
2
o1 this expression reduces to the form

0 = /0
(/ is a positive constant). For the reasons stated above this is a position of
stable equilibrium. Since the equilibrium position at 0
o
= cos
1

o.
2
1

is
an equivalent position, as you might expect, you arrive at the same conclusion
for uctuations about this equilibrium position as well.
With this result, we arrive at the following interesting story: When the
hoop is rotating slowly, there is only one position of stable equilibrium, 0 = 0.
If we speed up the rotation, then as . passes the critical value . =
p
o1. this
original equilibrium becomes unstable, but two new stable equilibrium points
appear. They emerge from 0 = 0 and move out to the right and left as we
continue to increase .. This phenomenon the disappearance of one stable
equilibrium and the simultaneous appearance of two others diverging from the
same point is called a bifurcation and will be one of the principle topics in
chapter 12 on chaos theory.
This example illustrates another strength of the Lagrangian method in that
the generalized coordinates can be coordinates in a noninertial frame, as long
as the Lagrangian itself is written in an inertial frame. In this example 0 is the
polar angle of the bead written in a noninertial rotating frame of the hoop. But
the Lagrangian, L = T l. was evaluated in inertial frame in which the hoops
rotates.
It may be of interest to note that a device of this example was used by
James Watt as a governor for his steam engines. The device rotated with the
engine, and as the engine sped up the bead rose on the hoop. When the angular
velocity . reached some maximum allowable value, the bead reached a height
that caused the supply of steam to be shut o.
Oscillations of the Bead near Equilibrium With our analysis, we are
well placed to determine the frequency of oscillations about equilibrium for our
8
example of a bead on the spinning hoop. First lets consider the frequency of
oscillations about the equilibrium position at 0 = 0. From equation (30) we
found that it was a position of stable equilibrium only when .
2
< o1. In this
frequency range we can write equation (30) as

0 =

o1 .
2

0 =
2
0 = 0. (33)
where =
p
o1 .
2
. Since this equation is that of a simple harmonic oscil-
lator, we know that the bead oscillates with a frequency about 0 = 0 as long
as .
2
< o1.
To nd the frequency of oscillations about the equilibrium about 0
o
=
cos
1

o.
2
1

. we merely have to examine equation (??). As long as .

2
o1.
which is required for equilibrium to exist at 0
o
. equation (??) can also be written
in the form of a simple harmonic oscillator as
c

0 +
02
c0 = 0. (34)
where
02
= .
2
o
2
.
2
1
2
. Therefore c0 oscillates about zero, which means
that the bead oscillates about the equilibrium position 0
o
with frequency
0
.
So our story is now a bit more complete. When the hoop is stationary the
bead is in equilibrium at 0 = 0 and for small amplitude oscillations, it oscil-
lates at a frequency of =
p
o1. As the hoop begins to spin, the frequency
of oscillations begins to slow via the expression =
p
o1 .
2
. This slowing
continues until the rate of spinning reaches the threshold .
h
=
p
o1. at which
point the bead no longer oscillates at all. Unless the bead was stationary at
the moment this threshold was reached (at a maximum in its oscillations), it
then continues to move slowly away from the bottom of the hoop at a uniform
rate. As the rate of spinning continues to increase the bead comes to equilib-
rium at 0
o
= cos
1

o.
2
1

. and as mentioned there are two positions of

equilibrium. For frequencies just greater than the threshold .
h
these angles
are close to zero, i.e. near the bottom of the hoop. The frequency of these
oscillations are also quite small. Now as the spinning rate continues to increase
both equilibrium positions move away from the bottom of the hoop and and
the frequency of oscillations about equilibrium increases. For large spinning
rates, .
2
o1. the equilibrium positions approach 2 and the frequency
of oscillations approaches the spinning frequency ..
When the equation of motion has no analytic solution in terms elementary
functions, in almost all cases it is still possible to examine the properties of the
solutions about positions of equilibrium using the technique that we demon-
strated in this example. First you nd the positions of equilibrium by setting
the accelerations to zero. Then you consider small uctuations about these po-
sitions of equilibrium. This can be done via a Taylors series expansion, or some
equivalent procedure as we did here, to rst order about equilibrium. If you
only include rst order corrections away from equilibrium, then you will have
succeeded in linearizing the equation, almost always in the form of a simple
harmonic oscillator.
9