Advanced Dynamics Notes

Jed Rembold
September 24, 2009

Contents
1 Survey of Elementary Particles
1.1 Mechanics of a Particle . . . . . . . . . . . . . . . . . . . .
1.2 Mechanics of a System of Particles . . . . . . . . . . . . .
1.3 Constraints . . . . . . . . . . . . . . . . . . . . . . . . . .
1.4 DAlemberts Principle and Lagranges Equations . . . . .
1.5 Velocity Dependent Potentials and Dissipation Functions
1.6 Simple Applications of Lagrangian Formalism . . . . . . .

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

5
5
7
11
13
15
15

2 Variational Principles and Lagranges Equations

2.1 Hamiltons Principle and Calculus of Variations . . . . . . . . . . . .
2.2 The 3 Classic Problems of the Calculus of Variations . . . . . . . . .
2.3 Calculus of Variations and Lagranges Equations . . . . . . . . . . .
2.4 Non-Holonomic Systems and Lagrange Multipliers . . . . . . . . . .
2.5 Conservation Theorems and Symmetry Properties . . . . . . . . . .
2.5.1 Energy Conservation and Time Homogeneity . . . . . . . . .
2.5.2 Momentum Conservation and Space Homogeneity . . . . . .
2.5.3 Angular Momentum Conservation and Rotational Invariance

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.

19
19
21
21
24
29
30
31
32

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

.
.
.
.
.
.

3 The Central Force Problem

35
3.1 2 body to 1 body problem and Reduced Mass . . . . . . . . . . . . . . . . . . . . . . 35

CONTENTS

Chapter 1

Survey of Elementary Particles

1.1

Mechanics of a Particle

We begin with a review of common expressions. Recall that

r = position vector of a particle from a given origin
v = particles vector velocity
We thus know that
v=

dr
dt

and p = mv

where
p = linear momentum of a particle
m = the mass of a particle
F = total force exerted on the particle
The mechanics or motion of the particle are given by Newtons 2nd Law of Motion:
F =

dp
= p
dt

For an inertial or Galilean reference frame, this can be expressed as

F =

d
(mv)
dt

For a particle with constant mass, this simplifies to:

F = ma = m

dv
d2 r
=m 2
dt
dt

then p equals a constant, and thus we

where a is the particles acceleration. Now, if F = 0 = p,
have Conservation of Linear Momentum of the particle.
5

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

We also know that

L = angular momentum about a point O
and that
L=rp

The change of the angular momentum with respect to time is given by

d
d
d
L = (r p) = (r mv)
dt
dt
dt
dr
d
=
mv +r (mv)
dt
|dt {z }
0

=rF

=N
which is called the moment of force or, more commonly, the torque. Note here that L and N
depend on O, the point about which the moments were taken. Now, if N = L = 0, then L is a
constant, and thus we have Conservation of Angular Momentum.

The Work done on a particle by force F is

WAB =

F ds

where s is the trajectory. Assuming that the mass is constant:

Z

dv
v dt
A dt
Z
m B d 2
(v )
=
2 A dt


m 2
2
=
vB vA
2
TB TA

WAB = m

where T = 12 mv 2 is the kinetic energy. Thus we see evidence of the Work-Energy Theorem, which
states that the work done is equal to the change in kinetic energy. Now, if WAB depends only on
the endpoints A and B (i.e. is path independent), then the force responsible is conservative.

1.2. MECHANICS OF A SYSTEM OF PARTICLES

Thus we see that, in a conservative system, the work around a closed path is
I
F ds = 0
Remark 1: Dissipative forces such as friction are not conservative since
F ds > 0
for work done by the particle.
For a conservative force (WAB path independent), we can write
F = V (r)
where V is the potential or potential energy. Thus we have
F ds = dV

Fs =

dV
ds

Remark 2: Note that

F = (V (r) + constant) = V (r)
and thus the zero of V (r) is arbitrary.
Remark 3: For a conservative system, WAB = VA VB . Thus we have that
VA VB = TB TA

VA + TA = VB + TB

which is our expression for the Conservation of Total Energy!

1.2

Mechanics of a System of Particles

The goal here is to generalize Newtons 2nd law to a system of particles. Starting with the equation
of motion for the ith particle:
X
(e)
Fji + Fi = p i
j

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

where
Fji = the internal force on particle i by particle j and
(e)

Fi

= an external force

Now, assuming that Fij = Fji (Newtons 3rd Law, or the Weak Law of Action and Reaction),
then
X
X (e) X
d2 X
Fi = 2
mi ri =
Fi +
Fji
dt
i
i
i
i6=j
| {z }
0

We define R as:

P
P
mi ri
i mi ri
P
R=
= i
M
i mi
where R is the center of mass and M is the total mass. Thus we have that
X (e)
d2
M 2R =
Fi F (e)
dt
i

Consequentally, purely internal forces (forces on particles from other particles) vanish by Newtons
3rd law. The total linear momentum can thus be expressed as
X dri
dR
=M
P =
mi
dt
dt
i

Thus total linear momentum is conserved when the total external forces are equal to zero.

The total angular momentum is written as

X
X
L=
Li =
ri pi
i

Hence,

X d
dL
= L =
(ri pi )
dt
dt
i
X
=
ri p i
i

(e)

ri Fi

ri Fji

i6=j

but
X

ri Fji =

1
2

i6=j

[ri Fji + rj Fij ]

i6=j

1
2

X
i6=j

(ri rj ) Fji

1.2. MECHANICS OF A SYSTEM OF PARTICLES

If we define
rij = ri rj
then we can write
X

ri Fji =

i6=j

1
2

rij Fji

i6=j

Note that if Fji is parallel to rij , then rij Fji = 0, which is called the Strong Law of Action and
Reaction. If this is true, then
X
dL X
(e)
rij F ji = N (e)
=
ri Fi + 12
dt
i
i6=j
|
{z
}
0

Thus, if the applied external torque, N (e) , equals 0, then L is constant in time and we have
Conservation of Angular Momentum.
Remark 1: The Strong Law of Action and Reaction requires the internal forces to be central.
We can also describe the particle position with respect to the center of mass:

So we have
ri = R + ri
vi = V + vi
where
dR
= velocity of center of mass
dt
dr
vi = i = velocity about the center of mass
dt

V =

10

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

Then
L=

ri pi =

R + ri mi V + mi vi

R mi V +

X
i

= R MV +

R mi vi +
| {z }
0

X
i

ri mi vi

ri mi V +
| {z }
0

ri mi vi

So the total angular momentum is:

X
L = R MV +
ri mi vi
i

= angular momentum of the CM + angular momentum about the CM

Now,
WAB =

XZ

XZ

(e)
Fi

dsi +

XZ
i6=j

Fji si
A

mi v i vi dt

Fi dsi =

XZ

XZ

2
1
2 mi vi

= TB TA
P
with total kinetic energy T = 21 i mi vi2 . We can also write
X
X




T = 12
mi vi2 = 21
mi V + vi V + vi
i

2
1
2MV

mi vi V +
| {z }
i
0
X
2
= 21 M V 2 + 12
mi vi

1
2

mi vi

= KE of CM + KE about CM
We now go back to
WAB =

XZ
i

B
A

(e)

Fi

dsi +

XZ
i6=j

Fji si

For conservative external forces, we have that:

XZ
i

B
A

(e)
Fi

dsi =

XZ
i

i Vi dsi =
A

X
i

B


Vi 

A

11

1.3. CONSTRAINTS
For conservative internal forces:
Fji = j Vji
where Vji = Vji (|ri rj |) to satisfy the Strong Law of Action and Reaction
= i Vij
= Fij
where there is no implied sum over repeated indices above. Therefore
WAB

B


=
Vi 

i
A
B
X 
=
Vi 

i
A
B

X 
=
Vi 

i
A
B

X 
=
Vi 

X

X
i

1
2

XZ
i6=j

1
2

XZ
i6=j

1
2

XZ
i6=j

1
2

XZ
i6=j

B


Vi 


1
2

X
i6=j

(Fji dsi + Fij dsj )

A
B

Fji (dsi dsj )

A
B

Fji dsij
A
B

ji Vji dsij
A

B


Vji 

A

Thus the total potential energy is

V =

X
i

Vi

1
2

Vij

i6=j

Remark 2: T + V is conserved for T, V the total kinetic and potential energy.

Remark 3: For a rigid body, internal potential energy is constant, and thus internal forces due
no work.

1.3

Constraints

Constraints limit the motion of a system: ie, beads on a string, ball on a circular track, or gas
molecules in a container.

12

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

Holonomic Constraints: A Holonomic constraint is one where the constraint can be expressed
as an equation connecting the space and time coordinates of a particle with the form:
f (r1 , r2 , r3 , . . . , rN , t) = 0
2 = 0) and
Examples include rigid bodies (which have equations of the form (ri rj )2 Cij
particles constrained to move on a curve or surface.

Non-Holonomic Constraints: A Non-Holonomic Constraint is one where the constraint can not
be expressed as above. Examples include gas molecules in a box or a particle placed on a
sphere surface (but not stuck to it)

Rheonomous Constraints: A Rheonomous constraint is one which depends explicitly on time

Scleronomous Constraints: A Scleronomous constraint is one which has no explicit time dependence.
Remark 1: Constraints are equivalent to saying that there are forces that can not be specified
explicitly; only by their effect on the systems motion.
Now we will consider Holonomic constraints and introduce generalized coordinates. A system of
N particles has d N degrees of freedom (independent coordinates) in d spacial dimensions. If the
holonomic constraints are expressed in K equations, then we have d N K degrees of freedom
expressed in terms of the generalized coordinates q1 , q2 , q3 , . . . , qdN K .

Transformation Equations relate the original variables r1 , r2 , . . . , rN in terms to qi via

rj = rj (q1 , q2 , q3 , . . . , qdN K , t)
Examples would include a particle constrained to move on a sphere of fixed radius, a pendalum
with a sliding attach-point, or a double pendalum.

1.4. DALEMBERTS PRINCIPLE AND LAGRANGES EQUATIONS

1.4

13

DAlemberts Principle and Lagranges Equations

Let
Fi
|{z}

(a)

total force
on ith
particle

= Fi
+
|{z}
applied
forces

fi
|{z}

force of
constraint

By Newtons 2nd law:

Fi = p i Fi p i = 0
which gives
X
i

(Fi p i ) ri = 0 =
|{z}
infinitesimal
displacement

Consider the case where


X
X  (a)
(fi p i ) ri
Fi p i ri +
i

fi ri = 0. This is akin to

i.e. the forces of constraint do no work. This leaves us with:


X  (a)

Fi pi ri = 0 DAlemberts Principle

()

Note that the above equation contains no constraints, so well drop the (a) designation unambiguously in the future.

For ri = ri (q1 , q2 , q3 , . . . , qn , t), we will write () in terms of the generalized coordinates qi :

ri =

X ri
i

Fi ri =

where
Qj =

X
i,j

X
i

Fi

qj

Fi

qj

X
ri
qj
Qj qj
qj
j

ri
= the generalized force
qj

14

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

Working on the other half of the equation, we have:

X
X
ri
qj

p i ri =
mi v i
qj
i
i,j



X d 
d ri
ri
=
mi v i
qj
mi v i
dt
qj
dt qj
i,j

Using the fact that

v i =

X ri

dri
= ri +
qk
dt
t
qk

by chain rule

vi
ri
=
qj
qj

Thus:
X
i



vi
vi
mi v i
qj
mi v i
p i ri =
dt
qj
qj
i,j
( "
!#
!)
X d
X
X

2
2
1
1
=

qj
2 mi vi
2 mi vi
dt qj
qj
X d 
j

And DAlemberts principle becomes:

X

(Fi p i ) ri =

X
j

T
d T
+
Qj
dt qj
qj

qj = 0

with T = i 21 mi vi2 . Now if the system is Holonomic, then it is possible to find qj such that qj is
independent of qk for all j 6= k. Thus, the individual coefficients must vanish:
d T
T

= Qj
dt qj
qj
If Fi = U (conservative force), then
Qj =

X
i



X
ri
ri
dU
dU
d dU
Fi
=
i U
=
=
+
qj
qj
dqj
dqj
dt dqj
i
| {z }
0

where the last term is 0 since we are taking U to be velocity independent. Thus we have

d
(T U )
(T U ) = 0
dt qj
qj
Define the Lagrangian to be L = T U . Thus Lagranges equations become:
L
d L

=0
dt qj
qj
Note that the Lagrangian is not unique! (See HW)

1.5. VELOCITY DEPENDENT POTENTIALS AND DISSIPATION FUNCTIONS

1.5

15

Velocity Dependent Potentials and Dissipation Functions

If U = U (qj , qj ), then


U
d
Qj =
+
qj
dt

U
qj

and

L=T U

An example would be a charged particle in an EM field (See HW). If frictional forces are present,
then well have


L
d L
= Qj = forces not arising from a potential

dt qj
qj
Example: Consider Rayleighs dissipation function:
F=

1
2

2
2
2
kx vix
+ ky viy
+ kz viz

And
Ff x =

F
;
vx

Ff = v F

where Ff x is the x component of the frictional force and the sum over i is over all particles.
j (the generalized force due to friction) is
Thus we have that Q
j =
Q

F fi

X
i

1.6

X
ri
ri
=
v F
qj
qj

d
dt

vi
F
v F
=
qj
qj

L
qj

F
L
+
=0
qj
qj

Simple Applications of Lagrangian Formalism

Example 1: Consider a single free particle in space:

T = 21 m x 21 + x 22 + . . . + x 2d
U =0
X
rj
Qi =
Fj
qi
j

in d-dimensional space

16

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

where, in cartesian coordinates,

xj
qi

= ij

L=T

d L
L

=0
dt xi xi

m
xi = Fxi = 0

Example 2: Consider a particle constrained on a sphere. Using polar coordinates, we have

x1 = R sin cos
x2 = R sin sin
x3 = R cos
Calculate x 1 , x 2 , x 3 and substitute into T it get the kinetic energy for the system whose
natural coordinates are spherical.
Example 3: Here we will consider a pendalum with a sliding pivot:

So we have that
x = x + l sin
y = l l cos

1.6. SIMPLE APPLICATIONS OF LAGRANGIAN FORMALISM

17

Thus we have:


T = 12 m x 2 + y 2

2 
2 

1
= 2 m x + l cos + l sin
 2

= 12 m x + 2l cos x + l2 cos2 2 + l2 sin2 2
 2

= 12 m x + 2l cos x + l2 2
U = mgy = mg (l l cos )
 2

L = 12 m x + 2l cos x + l2 2 mgl (1 cos )
Remark 1: Generalized coordinates are not necessarily orthogonal!
d L
L

dt qi qi

L
d 
d L
0=0
m
x

+
ml
cos

=
dt x x
dt

d L L
d 
ml cos x + ml2 + ml sin x + mgl sin = 0
=

dt

dt
To find the equilibrium points:
U = mgl mgl cos
U
= mgl sin = 0

2U
= mgl cos |=0,
2

= 0,
(
mgl > 0,
=0
=
mgl < 0, =

So we have a stable equilibrium at = 0 and an unstable equilibrium at = (as wed

expect).

18

CHAPTER 1. SURVEY OF ELEMENTARY PARTICLES

Chapter 2

Variational Principles and Lagranges

Equations
2.1

Hamiltons Principle and Calculus of Variations

Consider a system with n generalized coordinates: q1 , q2 , . . . , qn . These coordinates make up an

n-dimensional configuration space with each q corresponding to an axis in the space.

1. t is a parameter of curve C with t [t , t+ ] = interval

2. q = q(t ),
3.

d
dt q(t)

q+ = q(t+ )

q(t)
is the tangent vector

4. If L(q(t), q(t),

t) is a function of the qs and their tangents, then we define a number that

characterizes the path:
t+
X
L(q(t), q(t),

t) dt
S(C) =
t

Note that this is more general than t = time and L = Lagrangian. The above is true of any
function of q, q and t.

5. S is called a functional (a functional is an animal that eats a function and spits out a number).
6. If C C + C, then S S + S continuously
19

20

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

Example 1: Let n = 2 and well work in 2-d Euclidean space. We have that q = q(x1 , x2 ) and C
is a curve with parameter t.

We can find the length of C by:

l(C) = lim

= lim

t2

t1

N q
X

(q(ti+1 ) q(ti ))2

i=1

N
X
i=1

s

q(ti+1 ) q(ti )
ti+1 ti

2

(ti+1 ti )

2 dt
(q(t))

L(q, q,
t) = L(q)
=
S=l

q2

Example 2: Now well look at a point moving on curve C with velocity v(q) that takes time T to
go from q1 to q2 .
T = lim

t2
t1

n
X
i=1

1
v(q(ti ))

(q(ti+1 ) q(ti ))2

1 p 2
(q)

v(q(t))
p

L(q, q,
t) = L(q, q)
=

q2
v(q)

and

S=T

Example 3: Again, we have a moving particle with time= t and position= q. We can write the
kinetic and potential energy as:
T =

q2
,
2m

U (q)

Thus:
L(q, q,
t) = L(q, q)
= T (q)
U (q)

2.2. THE 3 CLASSIC PROBLEMS OF THE CALCULUS OF VARIATIONS

S=

21

t2

L(q(t), q(t),

t) dt
t1

So S is the Action. Hamiltons principle says that S has a stationary value for the actual
path of motion:
Z t2
L(q, q,
t) dt = 0
S =
t1

2.2

The 3 Classic Problems of the Calculus of Variations

The Brachistochrone Problem: Brachistochrone means short time, so these type of problems
are attempting to minimize the time. A massive particle moves from A to B under the force

of gravity along path C. Which C gives the shortest travel time?

The Geodesics Problem: A ship is traveling from Portland, OR to Hawaii along the surface of
a sphere. Which route is the shortest?
The Isoperimetric Problem (Didos Problem): For a curve C with given length l, which form
gives the maximum area?

2.3

Calculus of Variations, Hamiltons Principle, and Lagranges

Equations

Consider a curve C in configuration space.

C :

q ( ) = q (q1 (, ), . . . , qn (, ))

Some things to note:

1. q ( ) is a path for every fixed .
2. q=0 ( ) = q( )

22

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

3.
q ( ) = q( )
q (+ ) = q(+ )

for all

4. q ( ) are continuously differentiable with respect to (fixed ).

Consider a function L and functional S:
SL (C) =

L(q( ), q(
), ) d

Then, under variations of the path C:

1. S(C) is minimal if S(C ) > S(C) for all in the neighborhood of C
2. S(C) is maximal if S(C ) < S(C) for all in the neighborhood of C
3. S(C) is stationary, and C is an extremal of S if
1
[S(C ) S(C)] = 0 C in nbhd
0
= variation of functional S

S = lim

Remark 1: We find the extremals to solve classical problems.

1
[S(C ) S(C)]
0
Z +
1
= lim
[L(q , q , ) L(q, q,
)] d
0

S = lim

We then Taylor expand S(C ) for small . Let q ( ) = q( ) + ( ). Then

#

Z + "
n 
X
L q,i
1
L q
2
S = lim
L(q, q,
) +
+
+ O( ) L(q, q,
) d
0

q,i
q

i=1

2.3. CALCULUS OF VARIATIONS AND LAGRANGES EQUATIONS

23

Note that, since q ( ) = q( ) + ( ), we have

q
= ( )

L
L
L q
=
=
q
q q
q
|{z}

Thus

" n 
#

1 X L
L
S = lim
i +
i + O(2 ) d
0

qi
qi

i=1

+ Z +
Z + X
n
n
X
d L
L
L 

i d +
i 
d
i
=

q
q
dt

i
i

i=1
i=1 | i{z }
Z

n 
X
L
i=1

qi

d L
i ( ) d
dt qi

= 0 since S is stationary.
Now, applying the fundamental lemma of the calculus of variations:
Lemma: Let f ( ) be continuous for [ , + ] and
Z +
f ( )( ) d = 0

which are 2 times differentiable and obey

(+ ) = ( ) = 0
Then
f ( ) = 0
since i ( ) is arbitrary and vanishes at the endpoints.

L
d L

= 0 = Euler Equations
qi dt qi

()

Remark 2: When L = T U in (), these are the Euler-Lagrange equations.

R
Remark 3: The functional S = L(q( ), q(
), ) d is stationary only if the functional L obeys
the Euler-Lagrange equations.

24

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

Remark 4: () implies that

L X

qi
qk
k

X
k

and
Lik =

L
qi



X  L 
L
qk
qk
=0
qk qi
t qi
k

X
L
2L
Lik qk =

qi qi
k

2L
qi qk

2L
qk qi

qi

()

symmetric nn matrix

So the Euler-Lagrange equations () are equivalent to n-coupled ODEs of second order for q( ).
Therefore, the initial point of q( ) and initial tangent q(
) completely determine the path.
Remark 5: () has the structure of Newtons Second Law.
Remark 6: Lagranges equations of motion follow naturally from Hamiltons Principle.
Example: Geodesics in d-dimensional Euclidean space: Recall that the length of curve C is
given by

1/2
Z + p
d
X
p
) = q2 = (qj ( ))2
l(C) =
(q(
))2 d = L(q, q,
i

j=1

() then implies that

L
d
d L

=
dt qi qi
dt

q = a = constant

1 1
p 2qi
2 q2

= 0 i

q
pi = constant
q2

q( ) = a + b = straight lines

So the geodesics must run thru 2 points to specify a and b, implying that we require a 2-d
condition.

2.4

Extensions of Hamiltons Principle to Non-Holonomic Systems (Lagrange Multipliers)

S =

t+

L(q, q,
t) dt = 0 =


X
d L

d qi = 0
qi dt qi
i

2.4. NON-HOLONOMIC SYSTEMS AND LAGRANGE MULTIPLIERS

25

Non-Holonomic constraints imply that the generalized coordinates are not independent. This
implies that displacements of the path may or may not satisfy the constraints. If displacements
satisfy constraints, then the constraints are holonomic. If the displacements do not satisfy the
constraints, then we want to eliminate the constraints by means of Lagrange Multipliers. Lagrange
multipliers work for semi-holonomic constraints, which can be put in the form:
f (q1 , . . . , qn , q1 , . . . , qn ) = 0

()

where = 1, 2, . . . , n.
Remark 1: Semi-holonomic differs from holonomic in that the latter can be expressed in terms of
one constraint equation (function of generalized coordinates only), whereas the former con
be more than one (function of tangents as well).
Remark 2: In terms of path displacements, the semi-holonomic constraints can be expressed:
X
aik dqk + ait dt = 0
where i = 1, . . . , m
k

This is more restrictive than ().

Remark 3: () implies that:

m
X

f = 0

=1

where

= (q1 , . . . , qn , q1 , . . . , qn , t)
are undetermined functions.
Recall Hamiltons Principle:

which implies

t2

t1

t2

L dt = 0
t1

X  L
k

d L

qk dt qk

dt qk = 0

And qk are no longer independent (if we have non-holonomic constraints). But if the constraints
are semi-holonomic, then:
!
Z t2
m
X
f dt = 0
L+

t1

=1

Changing notation slightly, let

L = L1 ,

m
X

=1

f = L2
{z

26

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

where () in the piece that will make the qk s independent. Then

L3 = L1 + L2

t2

(L1 + L2 ) dt =

t2

t1

And L3 obeys Lagranges equations:

t2

L3 dt = 0

t1

t1

X  L3
k

d L3

qi
dt qi

dt qi = 0

d L3 L3

=0
dt qi
qi
Also
d

d L3 L3

=
(L1 + L2 )
(L1 + L2 )
dt qi
qi
dt qi
qi
L1
d

d
L1
+
(L2 )
(L2 )
=
dt qi
qi
dt qi
qi
=0
Now,

d (L2 )
d L1 L1

=
(L2 )
Qi
dt qi
qi
qi
dt qi

d L1 L1

= Qi = forces of constraint
dt qi
qi

with
Qi =

d
(L2 )
(L2 )
qi
dt qi

Remark 4: Lagrange multipliers allow us to map a semi-holonomic system of n-generalized coordinates and m-constraints to a holonomic one with n + m generalized coordinates.
Example:
Constraint:

L = 21 m x 2 + y 2 U (x, y) L1

f (x,
y,
y) = x y + ky = 0 L2
L3 = L1 + L2

and

where k = const

(1)

d L3 L3

=0
dt qi
qi

d
d L3 L3

=
(mx + y)
+
dt x
x
dt
d
d L3 L3

=
(my + x)
+
dt y
y
dt

U
=0
x
U
k = 0
y

(2)
(3)

27

2.4. NON-HOLONOMIC SYSTEMS AND LAGRANGE MULTIPLIERS

Thus we have 3 unknowns: x, y, , and 3 equations: (1), (2), (3).
Remark 5: Physically, the s represent forces of constraints.
Remark 6: Forces of constraint do no work in (virtual) displacements (See 2.4 of Goldstein).
Remark 7: Lagrange multipliers can be used for holonomic constraints ( = 1) as well.

Example: Hoop Rolling without Slipping Down an Inclined Plane: We are using the

generalized coordinates x, . Our constraint in the rolling constraint:

r = x

r = x

r x = 0

(3)

T = 12 mx 2 + 21 mr 2 2
U = mg(l x) sin

L1 = 21 mx 2 + 12 mr 2 2 mg(l x) sin
L2 = r x
L3 = L1 + L2
= 12 mx 2 + 12 mr 2 2 mg(l x) sin + (r x)
d L3 L3

= m
x mg sin + = 0
dt x
x
d L3 L3

= mr 2 r = 0
dt

m
x = mg sin
= mg sin mr
= mg sin m
x

from (2)
by (3)

(1)
(2)

28

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

2m
x = mg sin

x
= 12 g sin

So rolling has half the acceleration as the hoop sliding on a frictionless plane!
g
sin
=
2r
=

mg
sin frictional force of constraint
2

Example: Bead on a Wire Loop:

The normal way wed solve this problem would be:

T = 21 mR2 2 ,

U = mgy = mgR sin

L = 21 mR2 2 mgR sin

d L L
= mR2 + mgR cos = 0

dt

g
= cos
R

Now, using Lagrange multipliers:

T = 12 m(x 2 + y 2 ),

U = mgy

Our constraint is:

x2 + y 2 = R2

L1 = 12 m x 2 + y 2 mgy

x2 + y 2 R 2 = 0 = L 2

which implies

L3 = L1 L2 = 21 m x 2 + y 2 mgy + (x2 + y 2 R2 )
d L3 L3

= m
x 2x = 0
dt x
x
d L3 L3

= m
y + mg 2y = 0
dt y
y

2.5. CONSERVATION THEOREMS AND SYMMETRY PROPERTIES

29

So our 3 equations are:

m
x = 2x
m
y = mg + 2y
R2 = x2 + y 2
Letting
x = R cos
y = R sin

x = R sin
y = R cos
2
x
= R cos R sin y = R sin 2 + R cos
Taking m
x = 2x and plugging in x
implies:


m R cos 2 R sin = 2R cos

m 2 m sin

2
2 cos

Taking m
y = mg + 2y:



 m
m
m R sin 2 + R cos = mg + 2 2 tan R sin
2
2

2.5

(mR cos + mR tan sin ) = mg

g
= cos
R

Conservation Theorems and Symmetry Properties

Consider a function f (q, q,

t). If
d
(f (q, q,
t)) = 0
dt
then
f (q, q,
t) = const
where () is called the
first integral of the equations of motion
integrals of motion
constant of motion
Remark 1: Conservation laws are an example of ().
N
others Theorem: Symmetries in a system are related to conservation laws.

()

30

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

2.5.1

Energy Conservation and Time Homogeneity

Consider the Lagrangian:

L = L(q( ), q(
), )
Time homogeneity (time-translational invariance) means that L does not depend explicitly on time.
L = L(q( ), q(
))
Then

X L
X L
d
L
L=
qi +
qi +
dt
qi
qi
t
|{z}
i

Recall Lagranges equations of motion:

d L
L
=
dt qj
qj
These imply
X  d L 
X L
d
L=
qj
qj +
dt
dt qj
qj
i
i
X d  L 
qj
=
dt
qj
i

"
#
d X L
qj
L =0
dt
qj

qj

L
LH =E
qj

Then H = E = const is the energy and energy is conserved in a system whose

Lagrangian has no explicit time dependence.
Example:
L = T (q)
U (q),

T =

1
2

mj qj2

H=

qi mi qi

1
2

1
2

mi qi2 + U (q)

mi qi2

+ U (q) = T (q)
+ U (q) = total energy

Remark 1: Here we have assumed that U = U (q), so there does not exist dissipative
forces (which dissipate energy).

2.5. CONSERVATION THEOREMS AND SYMMETRY PROPERTIES

31

Remark 2: If L is an explicit function of time:

dH
L
=
dt
t
Remark 3: H is sometimes called the energy function, and its form coincides with
the Hamiltonian.
Remark 4: For frictional forces that are derivable from a dissipative function F:


d L
F
L
From 1.5:
+
=0

dt qj
qj
qj
Then
X L
X L
L
d
L=
qi +
qi +
dt
qi
qi
t
i
i


X
X L
d L F
L
=
qi
+
qi +
+
dt qi
qi
qi
t
i
i


X F L
X d
L
+
+
qi
qi
=
dt
qi
qi
t
i
| i {z }
=2F by 1.5

L
dH
=
2F
dt
t
for L = L(q, q)
and H = E then this gives the dissipative rate.

2.5.2

Momentum Conservation and Space Homogeneity

This will correspond to space translational invariance. This implies that the Lagrangian is invariant under ri ri + (every particle moves by some same displacement). For now, well only consider shifts in coordinates, not velocities.
L =

X L
X L
ri = 0 =
qi
ri
qi |{z}
i

where () is arbitrary, and only for holonomic constraints. This implies that
d L
L
=0=
qi
dt qi

L
= const
qi

32

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

If U = U (q) only, then
L
= mqi = pi
qi

pi

L
= canonical (or conjugate) momentum
qi

p also generalizes to include U = U (q, q).

Now consider L = T (q)
U (q):
d T
U
+
=0
dt qi
qi

2.5.3

d T
U
= p i =
Qi
dt qi
qi

pi = Qi = generalized force

Angular Momentum Conservation and Rotational Invariance

Let be an infinitesimal rotation and

= n

L =

X L
X L
ri +
vi = 0
ri
vi
i

Note that

ri = ri
vi = vi
So
L =


L
( ri ) +
( vi )
ri
vi

X  L
i

but

assuming = 0

pi =

L
vi

and

L
d
= pi
ri
dt

2.5. CONSERVATION THEOREMS AND SYMMETRY PROPERTIES

X d

L =

dt

33


pi ( ri ) + pi ( vi )

Now, recall that a (b c) = b (c a) = c (a b).

X
L =
[ (ri p i ) + (vi pi )]
i

X d
(ri pi ) = 0
dt
i

But is arbitrary, so
d X
(ri pi ) = 0
dt
i

Now Mi = ri pi = angular momentum, so we have that

X
Mi = const
i

Remark 1: The components of angular momentum along any axis (e.g. the z-axis)
are given by
X L
Mz =
where = z
i

i
Example: Well take a look at how this works in cylindrical coordinates:
X
X
Mz =
(ri pi )z =
mi (xi y i yi x i )
i

x = r cos
y = r sin
Mz =

X
i

x = r cos r sin
y = r sin + r cos

i



h
mi ri cos i ri sin i + ri cos i i + ri sin i ri cos i ri sin i i

mi ri2 i

Comparing with

i

X h
X L
1
mi ri2 + ri2 2i + z 2 U (ri , , z)
=
2
i
i
i
i
X
=
mi ri2 i
i

34

CHAPTER 2. VARIATIONAL PRINCIPLES AND LAGRANGES EQUATIONS

Chapter 3

The Central Force Problem

3.1

2 Body to 1 Body Problem and Reduced Mass

The two body problem:

35