UNIT-9: Poisson brackets

Poisson brackets and other canonical invariants, equations of motion and conservation theorems
in Poisson bracket formulation, angular momentum.

9.0 Objectives

After studying this unit you will be able to

• Define Poisson brackets and find the Poisson brackets of any given pair of functions.

• Describe the properties of Poisson brackets.

• Deduce the equations of motion in terms of Poisson brackets.

• Demonstrate the canonical invariance of Poisson brackets.

• Describe angular momentum Poisson brackets.

9.1 Introduction

In mathematics and classical mechanics, the Poisson bracket is an important binary operation.
Especially Hamiltonian mechanics Poisson brackets play a central role. They govern the time
evolution of a Hamiltonian dynamical system. The Poisson bracket also distinguishes a certain
class of coordinate transformations, called canonical transformations, which map one set of
canonical variables on to another set of canonical variables. The Poisson brackets also have a
similar counterpart in quantum mechanics that are called commutators. They are closely related
with the uncertainty principle of quantum mechanics. Hence learning Poisson brackets takes us
one step ahead in understanding basic physics.

9.2 Poisson brackets

Hamilton’s equations of motion for q̇ and ṗ give the time evolution of the coordinates and mo-
menta of a system in phase space. Using these equations, we can find the equation of motion
MPDSC 1.1 KSOU

for any function F (q, p) in terms of what is known as Poisson brackets. The Poisson brackets of
any two functions F (q, p, t) and G (q, p, t) with respect to the canonical variables (q, p) is written
as [ F, G ]q,p and is defined as

 
∂F ∂G ∂F ∂G
[ F, G ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

9.2.1 Fundamental Poisson brackets

There are three fundamental Poisson brackets. That are [q j , qk ]q,p , [ p j , pk ]q,p and [q j , pk ]q,p .

1. [q j , qk ]q,p

 
∂q j ∂qk ∂q j ∂qk
[q j , qk ]q,p = ∑ − |
i
∂qi ∂pi ∂pi ∂qi

We know that the generalized coordinates and momenta are independent variables. Hence,
∂q
= 0 for any combination of q and p. Then,
∂p

∴ [q j , qk ]q,p = 0

2. [ p j , pk ]q,p

 
∂p j ∂pk ∂p j ∂pk
[ p j , pk ]q,p = ∑ − |
i
∂qi ∂pi ∂pi ∂qi

∂p
Again since the generalized coordinates and momenta are independent variables, =0
∂q
for all combination of q and p.

∴ [ p j , pk ]q,p

3. [q j , pk ]q,p

 
∂q j ∂pk ∂q j ∂pk
[ F, G ]q,p = ∑ − |
i
∂qi ∂pi ∂pi ∂qi

The second term in the right hand side of the above equation becomes zero as the general-
∂q j
ized coordinates and momenta are independent. However, in the first term is 0 if i 6= j
∂qi

143
MPDSC 1.1 KSOU

∂pk
and 1 if i = j. This is represented by a delta function δij . Similarly is represented by
∂pi
δki . On summation, all the terms for which i 6= j 6= k vanish leaving behind only one delta
function that δjk . Henc,

[q j , pk ]q,p = δjk

9.2.2 Properties of Poisson brackets

Let F = F (q, p, t), G = G (q, p, t) and S = S(q, p, t) be functions of canonical variables q, p

and time. When describing the following properties of the Poisson brackets we are omitting
the subscript q, p. It is understood that the Poisson brackets are defined with respect to these
canonical variables only.

1. [ F, F ] = 0

 
∂F ∂F ∂F ∂F
[ F, F ] = ∑ − =0
i
∂qi ∂pi ∂pi ∂qi

2. [ F, C ] = 0 The Poisson bracket of a function with a constant is zero.

3. [ F, G ] = −[ G, F ]

   
∂F ∂G ∂F ∂G ∂G ∂F ∂G ∂F
[ F, G ] = ∑ − = −∑ − = −[ G, F ]
i
∂qi ∂pi ∂pi ∂qi i
∂qi ∂pi ∂pi ∂qi

4. [ F, G + S] = [ F, G ] + [ F, S]

 
∂F ∂( G + S) ∂F ∂( G + S)
[ F, G + S] = ∑ −
i
∂qi ∂pi ∂pi ∂qi
   
∂F ∂G ∂F ∂G ∂F ∂S ∂F ∂S
=∑ − +∑ − = [ F, G ] + [ F, S]
i
∂qi ∂pi ∂pi ∂qi i
∂qi ∂pi ∂pi ∂qi

5. [ F, GS] = [ F, G ]S + G [ F, S]

 
∂F ∂( GS) ∂F ∂( GS)
[ F, GS] = ∑ −
i
∂qi ∂pi ∂pi ∂qi

Using product rule of differentiation we can write,

144
MPDSC 1.1 KSOU

 
∂F ∂S ∂F ∂G ∂F ∂S ∂F ∂G
=∑ G + S− G − S
i
∂qi ∂pi ∂qi ∂pi ∂pi ∂qi ∂pi ∂qi

   
∂F ∂S ∂F ∂S ∂F ∂G ∂F ∂G
= ∑G − +∑ − S = [ F, G ]S + G [ F, S]
i
∂qi ∂pi ∂pi ∂qi i
∂qi ∂pi ∂pi ∂qi

∴ [ F, GS] = [ F, G ]S + G [ F, S]

6. Jacobi identity: [ F, [ G, S]] + [ G, [S, F ]] + [S, [ F, G ]] = 0

 
∂F ∂[ G, S] ∂F ∂[ G, S]
[ F, [ G, S]] = ∑ −
i
∂qi ∂pi ∂pi ∂qi

" ( !) ( !)#
∂F ∂ ∂G ∂S ∂G ∂S ∂F ∂ ∂G ∂S ∂G ∂S
=∑ ∑ − − ∑ −
i
∂qi ∂pi j
∂q j ∂p j ∂p j ∂q j ∂pi ∂qi j
∂q j ∂p j ∂p j ∂q j

( !)
∂F ∂G ∂2 S ∂2 G ∂S ∂G ∂2 S ∂2 G ∂S
=∑ ∑ + − −
i
∂qi j
∂q j ∂pi ∂p j ∂pi ∂q j ∂p j ∂p j ∂pi ∂q j ∂pi ∂p j ∂q j
( !)
∂F ∂G ∂2 S ∂2 G ∂S ∂G ∂2 S ∂2 G ∂S
−∑ ∑ + − −
i
∂pi j
∂q j ∂qi ∂p j ∂qi ∂q j ∂p j ∂p j ∂qi ∂q j ∂qi ∂p j ∂q j

Similarly we can find [ G, [S, F ]] and [S, [ F, G ]] and add them to prove the Jacobi identity.
d dF dG
7. [ F, G ] = [ , G ] + [ F, ]
dt dt dt

9.3 Equations of motion in terms of Poisson brackets

In addition to the properties of Poisson brackets discussed in the previous section we shall see
some more properties related to the Hamiltonian and equations of motion.
We know from the definition of Poisson brackets

 
∂F ∂G ∂F ∂G
[ F, G ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

If G = qk ,

145
MPDSC 1.1 KSOU

 
∂F ∂qk ∂F ∂qk
[ F, qk ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

Since the generalized coordinates and momenta are independent, the first term of the above
equation becomes zero. In the second term all the terms except the term for i = k vanish. Hence
we will be left with,

∂F
[ F, qk ]q,p = −
∂pk
Similarly if G = pk ,

 
∂F ∂pk ∂F ∂pk
[ F, pk ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

With the similar argument we can see that the first term in the right hand side of the above
expression becomes zero and the after summation only one term for i = k will remain. Then

∂F
[ F, pk ]q,p =
∂qk
Now we have two very important results as ,

∂F ∂F
[ F, qk ]q,p = − and [ F, pk ]q,p =
∂pk ∂qk
Now if the function F is H, then,

∂H ∂H
[ H, qk ]q,p = − and [ H, pk ]q,p =
∂pk ∂qk
or

∂H ∂H
[qk , H ]q,p = and [ pk , H ]q,p = −
∂pk ∂qk
From Hamiltonian formulation we know the Hamiltonian equations of motion as

∂H ∂H
q̇k = and ṗk = −
∂pk ∂qk
Comparing the above two sets of equations, we can write,

146
MPDSC 1.1 KSOU

q̇k = [qk , H ]q,p and ṗk = [ pk , H ]q,p

These two are the Hamiltonian equations of motion in terms of Poisson brackets.
Further, let us consider the Poisson bracket of a general function F = F (q, p, t), with the
Hamiltonian.

 
∂F ∂H ∂F ∂H
[ F, H ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

Using the Hamiltonian equations of motion, we can write,

   
∂F ∂F ∂F dqi ∂F dpi
[ F, H ]q,p = ∑ q̇i + ṗ =∑ +
i
∂qi ∂pi i i
∂qi dt ∂pi dt

We know that

 
dF ∂F dqi ∂F dpi ∂F
dt
= ∑ ∂qi dt
+
∂pi dt
+
∂t
i

Hence, we can write the equation of motion of the function F as

dF ∂F
= [ F, H ]q,p +
dt ∂t
If the function F is not an explicit function of time, then,

dF
= [ F, H ]q,p
dt
dF
Interestingly if the Poisson bracket of F with the Hamiltonian becomes zero, then =
dt
0. In other words, all the functions whose Poisson bracket with Hamiltonian vanishes are all
constants of motion of the system. Thus, the Poisson bracket of all conserved quantities with
Hamiltonian turns out to be zero.

9.4 Canonical invariance of Poisson brackets

Probably the most important property of Poisson bracket is that it is invariant under canonical
transformation. This means that if (q, p) and ( Q, P) are two canonical conjugate sets, then

[ F, G ]q,p = [ F, G ]Q,P

147
MPDSC 1.1 KSOU

Where, F and G are any pair of functions of (q, p) or ( Q, P). The two sets (q, p) and ( Q, P)
are related through a canonical transformations given by

Qi = Qi (q, p, t) and Pi = Pi (q, p, t)

The Poisson bracket of the functions F and G with respect to (q, p) set is given by

 
∂F ∂G ∂F ∂G
[ F, G ]q,p = ∑ −
k
∂qk ∂pk ∂pk ∂qk

Consider the function G after transformation as G = G ( Q, P). Then,

   
∂G ∂G ∂Qi ∂G ∂Pi ∂G ∂G ∂Qi ∂G ∂Pi
∂pk
= ∑ ∂Qi ∂pk
+
∂Pi ∂pk
and
∂qk
= ∑ ∂Qi ∂qk
+
∂Pi ∂qk
i i

Substituting the above two expressions in the Poisson bracket we get,

"    #
∂F ∂G ∂Qi ∂G ∂Pi ∂F ∂G ∂Qi ∂G ∂Pi
[ F, G ]q,p = ∑ ∑ + − ∑ +
k
∂qk i
∂Qi ∂pk ∂Pi ∂pk ∂pk i
∂Qi ∂qk ∂Pi ∂qk

    
∂F ∂G ∂Qi ∂G ∂Pi ∂F ∂G ∂Qi ∂G ∂Pi
[ F, G ]q,p = ∑ ∑ + − +
k i
∂qk ∂Qi ∂pk ∂Pi ∂pk ∂pk ∂Qi ∂qk ∂Pi ∂qk

 
∂F ∂G ∂Qi ∂F ∂G ∂Pi ∂F ∂G ∂Qi ∂F ∂G ∂Pi
[ F, G ]q,p = ∑ ∑ + − −
k i
∂qk ∂Qi ∂pk ∂qk ∂Pi ∂pk ∂pk ∂Qi ∂qk ∂pk ∂Pi ∂qk

 
∂F ∂G ∂Qi ∂F ∂G ∂Qi ∂F ∂G ∂Pi ∂F ∂G ∂Pi
[ F, G ]q,p = ∑ ∑ − + −
k i
∂qk ∂Qi ∂pk ∂pk ∂Qi ∂qk ∂qk ∂Pi ∂pk ∂pk ∂Pi ∂qk

    
∂G ∂F ∂Qi ∂F ∂Qi ∂G ∂F ∂Pi ∂F ∂Pi
[ F, G ]q,p = ∑ ∑ − + −
k i
∂Qi ∂qk ∂pk ∂pk ∂qk ∂Pi ∂qk ∂pk ∂pk ∂qk

 
∂G ∂G
[ F, G ]q,p = ∑ [ F, Qi ] + [ F, Pi ]
i
∂Qi ∂Pi

We have the relation

∂F ∂F
[ F, Qi ] = − and [ F, Pi ] =
∂Pi ∂Qi
Then,

148
MPDSC 1.1 KSOU

     
∂G ∂F ∂G ∂F ∂F ∂G ∂F ∂G
[ F, G ]q,p = ∑ − + = ∑ − = [ F, G ]Q,P
i
∂Qi ∂Pi ∂Pi ∂Qi i
∂Qi ∂Pi ∂Pi ∂Qi

Note that we did not assume any restricted properties for F and G. Hence, the Poisson
bracket between F and G is same with respect to any set of canonical variables. In other words,
the Poisson brackets are invariant under canonical transformations. Based on this property we
can use the Poisson brackets to verify any transformation is canonical or not.

9.5 Angular momentum

Using the definition of linear and angular momentum a number of interesting and useful pois-
son bracket relations can be obtained. Poisson brackets relation between the components of P
and J will be highly useful in several analytical applications. Here we are using upper case let-
ters to represent the linear and angular momentum to avoid the confusion between the linear
momentum and generalized momenta.
According to the definition of angular momentum, we have,

   
~J = ~r × ~P = xî + y ĵ + zk̂ × Px î + Py ĵ + Px k̂

Or



J = yPz − zPy î + (zPx − xPz ) ĵ + xPy − yPx k̂

Therefore,


Jx = yPz − zPy , Jy = (zPx − xPz ) and Jz = ( xPy − yPx )

From the definition of Poisson bracket,

n  
∂F ∂G ∂F ∂G
[ F, G ] = ∑ −
∂qk ∂pk ∂pk ∂qk
k =1

In the above expression, q1 = x; q2 = y and q3 = z. Similarly p1 = Px ; p2 = Py and p3 = Pz .

 
[ Px , Px ] = Py , Py = [ Pz , Pz ] = 0

   
Px , Py = Py , Pz = [ Pz , Px ] = 0

149
MPDSC 1.1 KSOU

We know,

∂F ∂F
[ F, qi ] = − and [ F, pi ] =
∂pi ∂qi
The Poisson bracket between the angular momentum and linear momentum will be,

∂Jx ∂Jx ∂Jx

[ Jx , p x ] = = 0; [ Jx , p y ] = = pz ; [ Jx , p z ] = = − py
∂x ∂y ∂z
Similarly we can show

∂Jy ∂Jy ∂Jy

[ Jy , p x ] = = − pz ; [ Jy , py ] = = 0; [ Jy , pz ] = = px
∂x ∂y ∂z

∂Jz ∂Jz ∂Jz

[ Jz , p x ] = = py ; [ Jz , py ] = = − px ; [ Jz , pz ] = =0
∂x ∂y ∂z
We can summarize the above set of relation as
 
0 pz − py
 
[ Ji , p j ] = − py 0 px 
 
 
py − p x 0

Now the Poisson brackets of angular momentum with position coordinates can be written
as

∂Jx ∂Jx ∂Jx

[ Jx , x ] = − = 0; [ Jx , y ] = − = z; [ Jx , z ] = − = −y
∂p x ∂py ∂pz
Similarly,

∂Jy ∂Jy ∂Jy

[ Jy , x ] = − = −z; [ Jy , y] = − = 0; [ Jy , z] = − =x
∂p x ∂py ∂pz

∂Jz ∂Jz ∂Jz

[ Jz , x ] = − = y; [ Jz , y] = − = − x; [ Jz , z] = − =0
∂p x ∂py ∂pz
We can summarize the above set of relation as

150
MPDSC 1.1 KSOU

 
0 z −y
 
[ Ji , x j ] = −y 0 x 
 
 
y −x 0

Now consider the Poisson bracket relation among the components of angular momentum.

 
∂Jx ∂Jy ∂Jx ∂Jy
∑
 
Jx , Jy = −
k
∂qk ∂pk ∂pk ∂qk

  ∂Jx ∂Jy ∂Jx ∂Jy ∂Jx ∂Jy ∂Jx ∂Jy ∂Jx ∂Jy ∂Jx ∂Jy
Jx , Jy = − + − + −
∂x ∂p x ∂p x ∂x ∂y ∂py ∂py ∂y ∂z ∂pz ∂pz ∂z


= 0 − 0 + 0 − 0 + − py (− x ) − (y)( p x ) = xpy − yp x = Jz

Thus,

[ Jx , Jy ] = Jz

Similarly we can show,

[ Jy , Jz ] = Jx and [ Jz , Jx ] = Jy

In compact, we can write the above set of relations as

[ Ji , Jk ] = eijk Jk

Where eijk is Levi-Civita function. The value of Levi-Civita function is +1 for non repeating
clock wise values of i, j and k. That is e123 = e231 = e312 = +1. The value of the Levi-Civita
function is -1 for non repeating counter clock wise values of i, j and k. That is e321 = e132 =
e213 = −1. Its value is zero for all other combinations of repeating indices.

9.6 Check your progress

Check your progress by answering the questions below.

1. Define Poisson brackets.

2. Does Poisson brackets satisfy commutative property.

151
MPDSC 1.1 KSOU

3. What is canonical invariance of Poisson brackets?

4. Write the Poisson bracket relation among the components of angular momentum.

9.7 Keywords

• Poisson brackets

• Jacobin identity

• Canonical invariance

9.8 Worked examples

1. Find the Poisson bracket of F (q, p) = q2 + p2 and G (q, p) = 2qp.

Answer: We know the definition of Poisson bracket

Given,

F (q, p) = q2 + p2 and G (q, p) = 2qp

∂F ∂G
= 2q and = 2p
∂q ∂q

∂F ∂G
= 2p and = 2q
∂p ∂p

We have,

∂F ∂G ∂F ∂G
[ F, G ] = −
∂q ∂p ∂p ∂q

[ F, G ] = 2q × 2q − 2p × 2p = 4(q2 − p2 )

Hence the Poisson bracket between F (q, p) = q2 + p2 and G (q, p) = 2qp is 4(q2 − p2 ).

152
MPDSC 1.1 KSOU

2. Using Poisson bracket show that the transformation

√ √
q= 2P sin Q and p= 2P cos Q

is canonical.

Answer: From the definition of Poisson bracket, it is obvious that [ Q, Q] = 0, [ P, P] = 0

[ Q, P] = 1

From the given transformations we have

q
tan Q = and 2P = q2 + p2
p

∂Q 1 ∂Q cos2 Q
sec2 Q = =⇒ =
∂q p ∂q p

∂Q q ∂Q q cos2 Q
sec2 Q =− 2 =⇒ =−
∂p p ∂p p2

∂P ∂P
=q and =p
∂q ∂p

Now consider

∂Q ∂P ∂Q ∂P
[ Q, P] = −
∂q ∂p ∂p ∂q

cos2 Q q2 q2
[ Q, P] = p + 2 cos2 Q = cos2 Q(1 + 2 )
p p p

[ Q, P] = cosQ (1 + tan2 Q) = 1

Hence, the transformation is canonical.

3. Show that the Poisson bracket [ L2 , Li ] = 0.

Answer:

153
MPDSC 1.1 KSOU

Consider

[ L2 , L x ] = [ L2x + L2y + L2z , L x ] = [ L2x , L x ] + [ L2y , L x ] + [ L2z , L x ]

We know that [ F, GS] = [ F, G ]S + G [ F, S]

Then

[ L2 , L x ] = [ L x , L x ] L x + L x [ L x , L x ] + [ L y , L x ] L y + L y [ L y , L x ] + [ L z , L x ] L z + L z [ L z , L x ]

We also know that, [ Ji , Jk ] = eijk Jk . Hence, [ L x , L x ] = 0; [ Ly , L x ] = − Lz ; [ Lz , L x ] = Ly . Using

these relation in above expression we can write

[ L2 , L x ] = −2Ly Lz + 2Lz Ly = 0

Similarly we can show [ L2 , Ly ] = [ L2 , Lz ] = 0

9.9 Questions for self study

1. Define Poisson brackets and obtain the Poisson bracket between the function F = mω 2 pq
1
and G = ( q2 + p2 ).
mω 2
2. Explain the different properties of Poisson brackets.

3. Prove the Jacobin identity of Poisson brackets.

4. Obtain the Hamilton’s equations of motion in terms of Poisson brackets.

5. Show that Poisson brackets are invariant under canonical transformations.

6. Discuss angular momentum Poisson brackets.

9.10 Answers to check your progress

1. Poisson bracket is a binary operation between two functions of canonical variables.

154
MPDSC 1.1 KSOU

The Poisson brackets of any two functions F (q, p, t) and G (q, p, t) with respect to the canon-
ical variables (q, p) is written as [ F, G ]q,p and is defined as

 
∂F ∂G ∂F ∂G
[ F, G ]q,p = ∑ −
i
∂qi ∂pi ∂pi ∂qi

2. Poisson brackets do not satisfy commutative property. They are anti-commutative in na-
ture.

3. Poisson bracket between two functions does not change under canonical transformations.
This is known as canonical invariance.

4.
[ Ji , Jk ] = eijk Jk

. Where eijk is Levi-Civita function.

9.11 References

• Herbert Goldstein, Charles Poole, John Safko, (2010) Classical Mechanics. Pearson Educa-
tion.

• G. Aruldhas, (2008) Classical Mechanics. Prentice Hall of India pvt. Ltd.

• R. G. Takwale and P. S. Puranik (2006) Introduction to Classical Mechanics. Tata McGraw-

Hill Publishing Company Ltd.

• N. C. Rana and P. S. Joag (2005) Classical Mechanics. Tata McGraw-Hill Publishing Com-
pany Ltd.

155
 UNIT-10: Hamilton-Jacobi theory

Hamilton-Jacobi equation, harmonic oscillator using Hamilton-Jacobi method, Hamilton’s char-

acteristic function, ignorable coordinates and the Kepler problem.

10.0 Objectives

After studying this unit you will be able to

• Define Hamilton-Jacobi equation.

• Describe Hamilton’s characteristic function.

• Use the Hamilton-Jacobi method to obtain the equation of motion of a harmonic oscillator.

• Solve the Kepler’s problem using Hamilton-Jacobi method.

10.1 Introduction

In the previous units we have learned that canonical transformations are one the very useful
tools in classical mechanics. The usefulness of the canonical transformation lies in increasing
the number of cyclic coordinates in the Hamiltonian. For a given Hamiltonian H (q, p, t) if we
define a set of transformations for both generalized coordinates and conjugate momenta, so
that all the generalized coordinates become cyclic in the new Hamiltonian K ( Q, P, t). Then the
corresponding conjugate momenta becomes conserved and hence constants of motion.
The Hamilton-Jacobi formalism is even a step further. This approach is to effect a canonical
transformation such that the new Hamiltonian K ( Q, P, t) is zero. Then each Q̇i and Ṗi becomes
zero. It means that all Q’s and P’s will be constants of motion. Further, the transformation
equations of such canonical transformation itself becomes the equations of motion.

10.2 Hamilton-Jacobi equation

Consider a dynamical system with Hamiltonian H (q, p, t). If a canonical transformation is made
from the (q, p, t) set to ( Q, P, t) set with the transformed Hamiltonian K = 0, Hamilton’s equa-
MPDSC 1.1 KSOU

tions will be

∂K ∂K
Q̇i = =0 and Ṗi = − =0
∂Pi ∂Qi
We know the the relation between the old and new Hamiltonian as

∂F
K= H+
∂t
If K = 0,

∂F
H+ =0
∂t
Where F is the generating function of the transformation. It is convenient to take F as a func-
tion of the original coordinates and the new constant momenta Pi and time t which corresponds
to F2 (q, P, t). Then,

∂F2
pi =
∂qi
With this value of pi , the above equation becomes

 
∂F2 ∂F2 ∂F2 ∂F2
H q1 , q2 , ...qn , , , ..., ,t + =0
∂q1 ∂q2 ∂qn ∂t
This equation is known as Hamilton-Jacobi equation. Usually the solution of the above
equation is denoted by S. This solution is also called Hamilton’s principle function. Then the
Hamilton-Jacobi equation in terms of Hamilton’s principle function becomes

 
∂S ∂S ∂S ∂S
H q1 , q2 , ...qn , , , ..., ,t + =0
∂q1 ∂q2 ∂qn ∂t
This is a first order differential equation in the (n + 1) variables. Hence, the solution will
have (n + 1) independent constants of integration. In Hamilton-Jacobi equation, only partial
derivatives of the type (∂S/∂qi ) and (∂S/∂t) will appear. S as such does not appear in the
equation. Therefore, if S is the solution, S + α, where α a constant, is also a solution. From
the (n + 1) constants α1 , α2 , ... αn+1 we may chose αn+1 = α as an additive constant. Hence a
complete solution of Hamilton-Jacobi equation can be written as

157
MPDSC 1.1 KSOU

S = S(q1 , q2 , ...qn , α1 , α2 , ...αn , t)

Comparing the above equation to the Hamiltonian function and also as we know that the
new Hamiltonian is zero and hence, all coordinates are cyclic and all momenta are constants,
we can choose the constants of integration as the new momenta.

Pi = αi

The 2n transformation equations now take the form

∂S(q, α, t) ∂S(q, α, t)
pi = and Qi = β i =
∂qi ∂αi
These equations are also the equations of motion.

10.3 Hamilton’s characteristic function

In systems where the time dependent part of Hamilton’s principle function S could be separated
out, the integration of the H-J equation is straight forward. Such a separation of variables is
always possible whenever the original Hamiltonian does not depend on time explicitly. In such
cases, the H-J equation reduces to

∂S ∂S ∂S ∂S
H (q1 , q2 , ...qn , , , ..., )+
∂q1 ∂q2 ∂qn ∂t
The first term involves only the q’s whereas the second term depends on time. Hence the
time variation can be separated by assuming a solution for S of the type

S(q, α, t) = W (q, α) − α1 t
∂S ∂W ∂S
Since = and = −α1 , substituting the trial solution we get,
∂qi ∂qi ∂t
 
∂W
H q, = α1
∂q
This equation does not involve time. By virtue of above equation, the constant of integration
α1 appearing in S is equal to the constant value of H, which is the total energy E if H does
not depend on time explicitly. The function W (q, α) is now called Hamilton’s characteristics

158
MPDSC 1.1 KSOU

function. Since, W does not involve time, the original and new Hamiltonians are equal and
hence K = α1 . Replacing α1 by E in the above equation we get,

 
∂W
H q, =E
∂q
In addition, we know that


∂W
∂S

 for i 6= 1
Qi = β i = = ∂W ∂α i
∂αi 
 − t for i = 1
∂αi
Thus Qi is the only coordinate that is not a constant of motion.
The physical significance of the Hamilton’s characteristic function can easily be obtained.
We have,

dS ∂S ∂S ∂S
dt
= ∑ ∂qi q̇i + ∑ ∂α̇i + ∂t
i i

∂S
We know, = pi and since αi are constants α̇i = 0. Then the above equation becomes
∂qi

dS ∂S
dt
= ∑ pi q̇i + ∂t
i

∂S ∂S
Since, H + = 0 we can replace by − H. Then,
∂t ∂t

dS
dt
= ∑ pi q̇i − H = L
i

Integrating on either side we have,

Z
S= Ldt + constant
R
We just know Ldt is nothing but the action integral of the Hamilton’s principle of least
action.

10.4 Harmonic Oscillator in the H-J method

Consider a one dimensional harmonic oscillator. The Hamiltonian of the system is

159
MPDSC 1.1 KSOU

p2 kq2
H= +
2m 2
We know the H-J equation,

∂S ∂S
H+ with p=
∂t ∂q
Then the H-J equation for harmonic oscillator will be

 2
1 ∂S 1 ∂S
+ kq2 + =0
2m ∂q 2 ∂t
Since the function S does not depend on time explicitly, we can write the solution of the
above equation as

S(q, α, t) = W (q, α) − αt

Substituting the above solution in the H-J equation gives us,

 2
1 ∂W 1
+ kq2 = α
2m ∂q 2
Since the left hand side of this equation is Hamiltonian, the constant α is the total energy.
Using this fact and rearranging the above equation we can write as,
s
∂W
q √ kq2
= 2mE − mkq2 = 2mE 1−
∂t 2E

s s
q √ kq2
Z q √ kq2
dW = 2mE − mkq2 = 2mE 1− dt =⇒ W = 2mE − mkq2 = 2mE 1− dt
2E 2E

Then the solution of H-J equation will be

s
Z q √ kq2
S= 2mE − mkq2 = 2mE 1− dt − αt
2E
Then,

160
MPDSC 1.1 KSOU

s
kq2 (kq2 /2α)
r r
m m
Z Z
∂S
β= = 1− dq + p dq − t
∂α 2α 2α 2α 1 − (kq2 /2α)

m dq
Z
β= p −t
2α 1 − (kq2 /2α)
The solution of above integral can be taken as

r r !
m k
β= sin−1 q −t
k 2α
=⇒
r r !
2α k
q= sin ( β + t)
k m
r
k
We know that = ω and α = E. Using these in the above, we can write the equation of
m
motion as,

r
2E
q= sin (ω (t + β))
mω 2

10.5 Kepler problem

Keplet’s problem is a very general one, unlike linear harmonic oscillator considered in the pre-
vious section. In the Kepler problem a planet of mass m moves around the sun in space. In
spherical polar coordinates, the Hamiltonian is

1 2 p2θ p2φ
H= ( p + 2 + 2 2 ) + V (r )
2m r r r sin θ
In Kepler problem, the gravitational force involved in inverse square in nature. Then the
k
potential will be V = − . Then the Hamiltonian will become,
r
2
!
p 2 p
1 φ k
H= p2r + 2θ + 2
−
2m r r2 sin θ r

Note that the coordinate φ is cyclic. Hence, pφ is a constant. Then the Hamilton-Jacobi
equation of the system becomes,

161
MPDSC 1.1 KSOU

" 2  2  2 #
1 ∂W 1 ∂W 1 ∂W k
+ 2 + 2 2 − =E
2m ∂r r ∂θ r sin θ ∂φ r

Assuming the solution of above equation can be obtained with variable separable method,
we can propose a solution as

W (r, θ, φ) = W1 (r ) + W2 (θ ) + W3 (φ)

On substituting the proposed solution to the H-J equation we can obtain three equations as,

 2
∂W3
= α2φ = constant
∂φ

α2φ
 2
∂W2
+ 2
= α2θ = constant
∂θ sin θ

2
α2θ
  
∂W1 k
+ 2 = 2m E +
∂r r r
Integrating the above three equations gives the generating functions W3 , W2 and W1 . The
sum of them will give the generating function W.
We shall restrict out discussion to bound orbits, that is, those for which energy E is negative.
In such case, the motion will be periodic in (r, θ, φ) coordinates. Next we shall find the action-
angle variables of the system. We have,

dW1 dW2 dW3

pr = pθ = pφ =
dr dθ dφ
Using the values of pφ , pθ and pr in above equations we can write,

I I
Jφ = pφ dφ = αφ dφ = 2παφ = 2π pφ

s
I I α2φ
Jθ = pθ dθ = α2θ − dθ
sin2 θ
s
2mk α2θ
I I
Jr = pr dr = 2mE + − 2 dr
r r
We know the Hamiltonian as Legendre transform of Lagrangian and the expression for ki-

162
MPDSC 1.1 KSOU

netic energy as

H= ∑( pi q̇i − L) 2T = ∑ pi q̇i = pr ṙ + pθ θ̇ + pφ φ̇
i i

Taking the plane in which the planet is moving as the plane for a plane polar coordinate
system

2T = pr ṙ + pψ ψ̇

Evaluating the two expressions for 2T

pθ θ̇ = pψ ψ̇ − pφ φ̇ =⇒ pθ dθ = pψ dψ − pφ dφ

Consequently,

I I I
J−θ = pθ dθ = pψ dψ − pφ dφ =⇒ Jθ = 2π pψ − 2π pφ

Then the Hamiltonian of the system in (r, ψ) coordinates will be

p2ψ p2φ
!
1
H= p2r + withp2ψ = p2θ +
2m r2 sin2 θ

The right hand side of the second equation given above is α2θ and therefore p2ψ = α2θ . On
substituting this value of pψ gives us

Jθ J Jφ
Jθ = 2π (αθ − αφ ) and αθ = + αφ = θ +
2π 2π 2π
With these values of αθ we can write,
s
2mk ( Jθ + Jφ )2
I
Jr = 2mE + − dr
r 4π 2 r2
This integral can be evaluated by complex integration. Its value is given by

r
2m
Jr = −( Jθ + Jφ ) + πk −
E
Solving for the energy of the system in the above expression we can write,

163
MPDSC 1.1 KSOU

2π 2 k2 m
E=−
( Jr + Jθ + Jφ )2
As expected with an inverse square law force, energy is negative for bound orbits.

10.6 Check your progress

Check your progress by answering the questions below.

1. What is Hamilton-Jacobi equation.

2. Define Hamilton characteristic function.

10.7 Keywords

• Hamilton-Jacobi equation

• Hamilton characteristic function

10.8 Worked examples

1. Constitute Hamilton Jacobi equation for the motion of an electron in a hydrogen atom.

Answer: Let m be the mass of the electron and e be the charge on nucleus and −e be the
charge of electron. Then the kinetic and potential energies of the electron will be

1 2 1 h 2 2 2 2 2 2
i
T = mv = m ṙ + r θ̇ + r sin θ φ̇
2 2

p2φ
!
1 p2
T= p2r + 2θ +
2m r r2 sin2 θ

1 e2
V=−
4πeo r

Then the Hamiltonian of the system will be

H = T+V

164
MPDSC 1.1 KSOU

p2φ
!
1 2 p2θ 1 e2
H= pr + 2 + −
2m r r2 sin2 θ 4πeo r

Let S be the Hamilton characteristic function. Then,

∂S ∂S ∂S
pr = ; pθ = ; pφ =
∂r ∂θ ∂φ

Then we can write the Hamilton Jacobi equation as

" 2 2 2 #
1 e2 ∂S
 
1 ∂S 1 ∂S 1 ∂S
+ 2 + 2 2 − + =0
2m ∂r r ∂θ r sin θ ∂φ 4πeo r ∂t

This is the Hamilton-Jacobi equation of an electron in hydrogen atom.

2. Apply Hamilton-Jacobi method to study the motion of a freely falling body.

Answer: Let the vertical direction be the z-axis and the zero of the potential energy be the
ground level. The system has only one degree of freedom and the coordinate is the value
of z. The Hamiltonian of the system is written as

p2
H= + mgz = E
2m

∂S
Let p =
∂z
Then the Hamilton Jacobi equation will be

 2
1 ∂S ∂S
H= + mgz + =0
2m ∂z ∂t

Let the solution to the above equation be

S(z, α, t) = W (z, α) − αt

∂S ∂W ∂S
Since = and = −α, the above equation will take the form,
∂z ∂z ∂t
 2
1 ∂W
+ mgz − α = 0
2m ∂z

165
MPDSC 1.1 KSOU

=⇒

q Z q
∂W
= 2m(α − mgz) =⇒ W= 2m(α − mgz)dz + C
∂z

Then the Hamilton characteristic function

Z q
S= 2m(α − mgz)dz − αt + C

√
2m dz
Z
∂S
= √ −t = β
∂α 2 α − mgz

√ √ r √
2m 2 α − mgz 2 α − mgz
β+t = =−
2 −mg m g

Squaring and rearranging we get,

α ( β + t )2 g
z= −
mg 2

Using initial condition that when t = 0, z = zo and p = 0, we can show α = mgzo and
β = 0. Then the equation of motion of the freely falling body becomes,

1
z = zo − gt2
2

3. Consider a particle of mass m moving in a potential V (r ). Write the Hamilton-Jacobi equa-

p2φ
2
tion in spherical polar coordinates and show that pθ + is a constant.
sin2 θ
Answer:

From Kepler problem problem we can write,

W (r, θ, φ) = W1 (r ) + W2 (θ ) + W3 (φ)

On substituting the proposed solution to the H-J equation we can obtain three equations
as,

166
MPDSC 1.1 KSOU

 2
∂W3
= α2φ = constant
∂φ

α2φ
 2
∂W2
+ 2
= α2θ = constant
∂θ sin θ

2
α2θ
  
∂W1 k
+ 2 = 2m E +
∂r r r

dW2 dW2
In above equations, = pθ and = pφ . Then we can write
dθ dφ

p2φ
p2θ + = constant
sin2 θ

10.9 Questions for self study

1. Define Hamilton-Jacobi equation and explain its importance.

2. Describe Hamilton’s characteristic function.

3. Obtain the solution of harmonic oscillator using Hamilton-Jacobi method.

4. Discuss Kepler’s problem using Hamilton-Jacobi method.

10.10 Answers to check your progress

1. The equation  
∂F2 ∂F2 ∂F2 ∂F2
H q1 , q2 , ...qn , , , ..., ,t + =0
∂q1 ∂q2 ∂qn ∂t
is known as Hamilton-Jacobi equation .

2. The solution of Hamilton-Jacobi equation is known as Hamilton characteristic function.

10.11 References

• Herbert Goldstein, Charles Poole, John Safko, (2010) Classical Mechanics. Pearson Educa-
tion.

• G. Aruldhas, (2008) Classical Mechanics. Prentice Hall of India pvt. Ltd.

167
MPDSC 1.1 KSOU

• R. G. Takwale and P. S. Puranik (2006) Introduction to Classical Mechanics. Tata McGraw-

Hill Publishing Company Ltd.

• N. C. Rana and P. S. Joag (2005) Classical Mechanics. Tata McGraw-Hill Publishing Com-
pany Ltd.

168
 UNIT-11: Rigid body dynamics-1

Degrees of freedom, center of mass, orthogonal transformations, Euler angles, Euler’s theorem,
infinitesimal rotations, Coriolis effect.

11.0 Objectives

After studying this unit you will be able to

• Describe a rigid body and its dynamics.

• Define center of mass and explain its properties.

• Explain the space fixed and body fixed coordinate systems using to describe the motion of
rigid body.

• Describe orthogonal transformations.

• Define Euler angles and Euler theorem.

• Explain Coriolis force.

11.1 Introduction

The body which does not deform under the influence of external forces is known as Rigid body.
In other words, a rigid body is defined as a system of particles in which the distance between
any two particles remains fixed throughout the motion. Thus, a system of N particles forming a
solid to be rigid body if it is subjected to holonomic constraints of the form, rij = cij , a constant.
Where rij is distance between ith and jth particle.
Before moving to study the dynamics of rigid bodies let us familiarize certain basic terms.
These terms have already been defined and discussed in the previous sections. It is going to be
just revision here.
MPDSC 1.1 KSOU

11.2 Degrees of freedom

We know the minimum number of independent variables or co-ordinates required to specify

the position of a dynamical system, consisting of one or more particles is called number of
degrees of freedom of the system. In general, a system of N number of particles in 3D space,
we require 3N number of co-ordinates to specify the position of each and every particle. This
set of 3N number of variables configuration of the system. A 3N dimensional space where the
configuration of the system can be represented by a single point is called configuration space of
a system. Even though rigid body as a very large number of particles, its degrees of freedom is
just 6 due to the presence of the constraints between every pair of particles to maintain distance
between them constant. We can show this with the following discussion.
As the rigid body consists of N number of particles, let us consider one particle to start
with. The degrees of freedom of a single particle is 3 as it can move in any combination of three
mutually perpendicular directions in three dimensional space.
If there are two such particles, the total degrees of freedom will be 6 (three for each particle).
Let us introduce a constraint saying the distance between the two particle is fixed. Then number
of constraint is 1. Then the degree of freedom of this two particle system will be 3N − k =
3 × 2 − 1 = 5.
Now let us introduce another particle with a constraint that the distance of the new particle
from the previous two particles are fixed. Hence the total number of particles is 3 and total
number of constraints is 3. Then the degree of freedom would be 3N − k = 3 × 3 − 3 = 6.
Let us introduce the fourth particle with constraints that its distance from the other three
particles are fixed. Hence the total number of constraints will be 6. Then the degrees of freedom
would be 3N − k = 3 × 4 − 6 = 6
For any new particle introduced, the number of constraints also increase proportionately
that keeps the degrees of freedom of rigid body at 6.

11.3 Centre of mass

Centre of mass is the point at which the distribution of mass is equal in all directions.
We define the center of mass of a system of particles in order to predict the possible motion
of the system. It is the point that moves as though all of the system’s mass were concentrated
there and all external forces were applied there.

170
MPDSC 1.1 KSOU

If we have a system of N number of particles, of mass m1 , m2 , m3 ,...m N at ~r1 , ~r2 , ~r3 ,...~r N
respectively. Then the center of mass of the system of particles is defined as a point with position
vector given by the following equation.

~R = ∑i mi~ri = m1~r1 + m2~r2 + ... + m N~r N

∑i mi m1 + m2 + ... + m N
Rigid bodies have continuous distribution of matter rather than collection of discrete parti-
cles. In such systems, the position of center of mass is defined by considering a small volume
element of mass dm at ~r and integrating over the entire system as,

R
~R = R~rdm = 1 ~rdm
Z

dm M
When there is no external force acting on the system, then the center of mass moves with a
constant velocity. If there is an external force acts on the system, then the acceleration of center
of mass will be the ratio of the force to the total mass of the system.
The center of mass of the system can be chosen as the origin of the coordinate system used
to describe the system. In such cases, for a system having discrete particles,

∑ mi~ri = 0 =⇒ ∑ mi xi = ∑ mi yi = ∑ mi zi = 0
i i i i

For systems with continuous distribution of matter,

Z Z Z Z
~rdm = 0 =⇒ x dm = y dm = z dm = 0

11.4 Coordinates of rigid body

A rigid body in space needs six independent generalized coordinates (as it has six degrees of
freedom) to specify its configuration, no matter how many particles it may contain. Of course,
there may be additional constraints on the body besides the constraint of rigidity. For example,
the body may be constrained to move on a surface or with one point fixed. In such case, the
additional constraints will further reduce the number of degrees of freedom, and hence the
number of independent coordinates.
Note that the set of configuration of a rigid body is completely specified by locating a Carte-
sian set of coordinates fixed in the rigid body (shown as primed axes in the figure below) relative

171
MPDSC 1.1 KSOU

y y0

x0

z0
x

Figure 11.19: Unprimed axes represent external reference set of axes and primed axes are fixed
to the rigid body.

to the coordinate axes of the external space. Clearly three of the coordinates are needed to spec-
ify the coordinates of the origin of this body set of axes. The remaining three coordinates must
then specify the orientation of the primed axes relative to a coordinate system parallel to the
external axes, but with the same origin as the primed axes.
There are many ways of specifying the orientation of a Cartesian set of axes relative to an-
other set with common origin. One fruitful procedure is to state the direction cosines of the
primed axes relative to the unprimed. Thus the x 0 axis could be specified by its three direction
cosines α1 , α2 and α3 with respect to the x, y and z axes respectively. If as customary, î, ĵ and k̂
are three unit vectors along x, y and z. Similarly, iˆ0 , jˆ0 and kˆ0 are three unit vectors along x 0 , y0
and z0 axes. Then the direction cosines are defined such that

iˆ0 = cos θ11 î + cos θ12 ĵ + cos θ13 ĵ

jˆ0 = cos θ21 î + cos θ22 ĵ + cos θ23 ĵ

kˆ0 = cos θ31 î + cos θ32 ĵ + cos θ33 ĵ

These sets of nine direction cosines completely specify the orientation of the primed axes
relative to the unprimed axes hence specifying the orientation of the rigid body.

172
MPDSC 1.1 KSOU

Using these direction cosines, one can transform between the two set of axes as

x 0 = ~r · iˆ0 = cos θ11 x + cos θ12 y + cos θ13 z

y0 = ~r · jˆ0 = cos θ21 x + cos θ22 y + cos θ23 z

z0 = ~r · kˆ0 = cos θ31 x + cos θ32 y + cos θ33 z

11.5 Orthogonal transformations

To study the properties of the nine direction cosines with greater ease, it is convenient to change
the notation and denote all the coordinates by x, distinguishing the exes by the subscripts as,

x → x1 ; y → x2 ; z → x3

Also change the notation for the direction cosines to

aij = cos θij

Then the transformation equations from ( x1 , x2 , x3 ) to ( x10 , x20 , x30 ) that form a linear transfor-
mations are given as

x10 = a11 x1 + a12 x2 + a13 x3

x20 = a21 x1 + a22 x2 + a23 x3

x30 = a31 x1 + a32 x2 + a33 x3

In matrix form we can write them

    
x 0 a a a x
 1   11 12 13   1 
xi0 = ∑ aij x j
 0 
 x2  =  a21 a22 a23   x2  =⇒
 
j
    
x30 a31 a32 a33 x3

Sometimes we ignore the summation symbol and write the transformation as xi0 = aij x j . Be-
cause the index j is present only in the right hand side, it is understood that there is a summation
with respect to j.

173
MPDSC 1.1 KSOU

Transformations represented using direction cosines is only a special case of the above gen-
eral linear transformation since the direction cosines are not all independent. Since the actual
vector remains unchanged no matter which coordinate system is used, the magnitude of the vec-
tor must be the same in both systems. In symbols, we can state this invariance of the magnitude
as

xi0 xi0 = xi xi =⇒ aij aik x j xk = xi xi

The above equation will be valid if and only if,

aij aik = 1 for j = k

aij aik = 0 for j 6= k

In other words, aij aik = δjk . This is known as orthogonality condition. Any linear transfor-
mation that satisfies this condition is known as orthogonal transformation. Thus the transition
from coordinates fixed in space to coordinate fixed in the rigid body is accomplished by means
of an orthogonal transformation.
A simple example can be given to such a transformation as the rotation by an angle φ with
respect to z axis. The transformation matrix for such a transformation is given by
   
a a a cos φ sin φ 0
 11 12 13   
A =  a21 a22 a23  = − sin φ cos φ 0
   
   
a31 a32 a33 0 0 1

11.6 Euler angles

To specify the position of a rigid body, six coordinates must be specified. Invariably, three of
these are taken to be the coordinates of the center of mass of the body. The other three coordi-
nates are taken to be the angles that describe the orientation of the body axes with respect to the
space-fixed axes. Though several choices are available, Euler’s angles are the most commonly
used ones.
Let the coordinate system that is fixed to the rigid body called as primed coordinates and
represented with ( x 0 , y0 , z0 ). The coordinate system fixed in space be called unprimed and rep-
resented with ( x, y, z). Euler’s angles are the three successive angles of rotations involved when
we go from the unprimed to primed system. These are illustrated in the figure 11.20.

174
MPDSC 1.1 KSOU

Figure 11.20: Euler angles

Step-1: Rotate ( x, y, z) axes in positive direction (anticlockwise) through an angle φ about the
z-axis. Let the resulting coordinate system be (ξ, η, z). Because the rotation is carried about z-
axis, the ξη plane will be same as xy plane. The angle φ by which it is rotated is called precession
angle. The transformation matrix of this rotation will be
 
cos φ sin φ 0
 
Rφ = − sin φ cos φ 0
 
 
0 0 1

Step-2: Now rotate the (ξ, η, z) system through an angle θ about ξ axis. This results in
(ξ 0 , η 0 , ζ 0 ) coordinate system. Note that ξ and ξ 0 are one and the same. This angle θ is called
nutation angle. The corresponding transformation matrix will be given by,
 
1 0 0
 
Rθ = 0 cos θ sin θ 
 
 
0 − sin θ cos θ

Step-3: Now rotate (ξ 0 , η 0 , ζ 0 ) system through and angle ψ with respect to ξ 0 axis. This results
in the final primed coordinate system ( x 0 , y0 , z0 ). Note that ξ 0 -axis and z0 -axis are one and the
same. The angle ψ is called body angle. The corresponding transformation matrix is given by,
 
cos ψ sin ψ 0
 
Rψ = − sin ψ cos ψ 0
 
 
0 0 1

These three angles φ, θ and ψ are called Euler’s angles.

175
MPDSC 1.1 KSOU

Rφ R Rψ
( x, y, z) −→ (ξ, η, z) −→
θ
(ξ 0 , η 0 , ζ 0 ) −→ ( x 0 , y0 , z0 )

Combining the rotations we can write the transformation equation from unprimed system
to primed system as

Rφ · Rθ · Rψ
( x, y, z) −−−−−→ ( x 0 , y0 , z0 )

∴ ( x 0 , y0 , z0 ) = Rψ (ξ 0 , η 0 , ζ 0 ) = R psi · Rθ (ξ, η, z) = R psi · Rθ · Rφ ( x, y, z)

Let R denote this total effective transformation. Hence, the transformation matrix for the
entire transformation will be

R = Rψ · Rθ · Rφ

     
cos ψ sin ψ 0 1 0 0 cos φ sin φ 0
     
R = − sin ψ cos ψ 0 · 0 cos θ sin θ  · − sin φ cos φ 0
     
     
0 0 1 0 − sin θ cos θ 0 0 1

On multiplying the above matrices we get,

 
cos ψ cos φ − cos θ sin ψ sin φ cos ψ sin φ + sin ψ cos θ cos φ sin ψ sin θ
 
R = − sin ψ cos φ − cos ψ cos θ sin φ − sin ψ sin φ + cos θ cos φ cos ψ cos ψ sin θ 
 
 
sin θ sin φ − sin θ cos φ cos θ

The line formed by the intersection of xy and x 0 y0 planes is called the line of nodes. The
range of Euler angles is

0 < φ ≤ 2π 0≤θ≤π 0 < ψ ≤ 2π

The angular velocity ω is a vector pointing along the axis of rotation. The general infinites-
~ can be thought of consisting of the three successive in-
imal rotation associated with vector ω
finitesimal rotations with angular velocities

176
MPDSC 1.1 KSOU

ωφ = φ̇ ωθ = θ̇ ωψ = ψ̇

Among these, φ̇ is directed along z-axis, θ̇ is directed along the line of nodes and ψ̇ is directed
along z0 axis. If ω1 , ω2 and ω3 are the components of ω
~ about x10 = x 0 , x20 = y0 and x30 = z0 axes.
Then we can write

ω1 = ωx0 = φ̇ sin θ sin ψ + θ̇ cos ψ

ω2 = ωy0 = φ̇ sin θ cos ψ − θ̇ sin ψ

ω3 = ωz0 = φ̇ cos θ + ψ̇

11.7 Euler’s theorem

Euler’s theorem describes the rotational motion of a rigid body. It states that The general displace-
ment of a rigid body with one point fixed is a rotation about some axis. The theorem means that for
every such rotation it always possible to find an axis through the fixed point oriented at partic-
ular angles θ and φ such that a rotation by the particular angle ψ about this axis duplicates the
general rotation. Thus three parameters characterize the general rotation. It is also possible to
find three Euler angles to produce the same rotation.
If the fixed point is taken as the origin of the body set of axes, then the displacement of
the rigid body involves no translation of the body axes; the only change is in orientation. The
theorem then states that the body set of axes at any time t can always be obtained by a single
rotation of the initial set of axes. In other words, the operation implied in the matrix R describing
the physical motion of the rigid body is a rotation. Now it is characteristic of the rotation that
one direction, namely, the axis of rotation, is left unaffected by the operation. Thus any vector
lying along the axis of rotation must have the same components in both the initial and final axes.
The other necessary condition for a rotation, that the magnitude of the vectors be unaffected,
is automatically provided by the orthogonality conditions. Hence, Euler’s theorem will prove if
~ having the same components in both systems. Then
it can be shown that there exists a vector A
we can write

A0 = RA = R

177
MPDSC 1.1 KSOU

This equation constitutes a special case of the more general equation

A0 = RA = λA =⇒ ( R − λI ) A = 0

Where, λ is a constant, which may be complex. The values of λ for which the above equation
is soluble are known as characteristic values or eigenvalues of the transformation matrix. The
problem of finding vectors that satisfy the above equation is therefore called eigenvalue problem
and the above equation is referred to as the eigenvalue equation. Now the Euler’s theorem can
be restated as The real orthogonal matrix specifying the physical motion of a rigid body with one point
fixed has the eigenvalue +1.

11.8 Infinitesimal rotations

While discussing Euler’s angles, we associated vectors with infinitesimal rotations. We will now
justify it by considering how vectors behave under rotation. Consider the change in the radius
vector ~r of the point M produced by an infinitesimal anticlockwise rotation through and angle
dφ about the axis of rotation. This is illustrated in the figure 11.21.

n̂dφ = dΩ

N dφ
L
M

~r ~r + d~r

Figure 11.21: An infinitesimal anticlockwise rotation of a vector.

The distance ML represent the magnitude of the change in vector

ML = |dr | = r sin θdφ

178
MPDSC 1.1 KSOU

The direction of the vector d~r is along ML which is perpendicular to both ~r and dΩ = n̂dφ.
Where n̂ is the unit vector along the axis of rotation. The direction of the vectors d~r is the
direction in which the right had screw advances as vector dΩ
~ is turned into vector ~r. Hence,

d~r = dΩ
~ ×~r

From the figure, we can state that the infinitesimal displacement d~r is equivalent to an in-
finitesimal rotation dφ which can be represented by a vector dΩ = n̂dφ pointing along the
instantaneous axis or rotation. Dividing the above equation by dt we get,

d~r dΩ
~
= ×~r = ω
~ ×~r
dt dt
~ we can write,
Generalize for a vector A

~
dA ~
~ ×A
=ω
dt

11.9 Rate of change of a vector

Rotational motion of rigid bodies is generally formulated in a body-fixed coordinate system. To

convert the results from a body-fixed system to a space fixed system and vice versa, we should
know how the time derivative of a vector in one system changes to the time derivative in the
other system. Let x 0 y0 z0 be the coordinate system fixed to the rotating body and xyz be the
space-fixed coordinate system with common origin. Consider the time derivative of vector

~r = iˆ0 x 0 + jˆ0 y0 + kˆ0 z0 and ~r = îx + ĵy + k̂z

be the position vector of a mass point with respect to the body fixed system and space fixed
system respectively.
The rate of change of the vector with respect to the body frame can be written as


d~r
 0
dx dy0 dz 0
= iˆ0 + jˆ0 + kˆ0
dt r dt dt dt
Similarly the rate of change of the vector with respect to the stationary frame or the space
frame would be

179
MPDSC 1.1 KSOU

!
dx 0 dy0 dz0 diˆ0 d jˆ0 dkˆ0
   
d~r
= iˆ0 + jˆ0 + kˆ0 + x+ y+ z
dt s dt dt dt dt dt dt

!
diˆ0 d jˆ0 dkˆ0
   
d~r d~r
= + x+ y+ z
dt s dt r dt dt dt

From the discussion of the previous section we can write,

diˆ0 d jˆ0 dkˆ0

~ × iˆ0 x
=ω ~ × jˆ0 y
=ω ~ × kˆ0 z
=ω
dt dt dt
Then we can write

   
d~r d~r  
= ~ × îx + ĵy + k̂z
+ω
dt s dt r

   
d~r d~r
= ~ ×~r
+ω =⇒ ~ ×~r
~vs = ~vr + ω
dt s dt r

Generalizing the above result with respect to a general vector we can write,
! !
~
dA ~
dA
= +ω ~
~ ×A
dt dt
s r

We can just extract the operator from the above relation and write as

   
d d
= ~×
+ω
dt s dt r

11.10 Coriolis force

The result obtained in the previous section can e used to obtain the relation connecting the
inertial acceleration of the particle of mass m and its acceleration relative to the rotating frame.
The time rate of change of velocity of the particle ~vs can be hence written as

   
d~vs d~vs
= ~ × ~vs
+ω
dt s dt r

~ ×~r. Using this in the above expression we get,

We also know ~vs = ~vr + ω

~ ×~r )
   
d~vs d(~vr + ω
= ~ × (~vr + ω
+ω ~ ×~r )
dt s dt r

180
MPDSC 1.1 KSOU

~ ×~r )
   
d~vs d~vr dω
= + ~ × ~vr + ω
+ω ~ × (~
ω ×~r )
dt s dt dt r

     
d~vs d~vr ~
dω
= + ×~r ~ × ~vr + ω
+ω ~ × ~vr + ω
~ × (~
ω ×~r )
dt s dt r dt r

~
dω
When the angular momentum is constant, = 0. Then we will have the expression for the
dt
acceleration,

ω × ~vr ) + ω
~as = ~ar + 2(~ ~ × (~
ω ×~r )

The corresponding force will be,

~F = m~as

=⇒

~F − 2m(~
ω × ~vr ) − mω
~ × (~
ω ×~r ) = m~ar

Therefore for an observer in the rotating frame of reference, it appears as if the particle is
moving under the influence of an effective force given by

~Feff = ~F − 2m(~
ω × ~vr ) − mω
~ × (~
ω ×~r )

In the above expression the term −mω

~ × (~
ω ×~r ) is called the centrifugal force. The term
−2m(~
ω × ~vr ) is called Coriolis force. Both these forces are not real force but fictitious force.
Both these forces do not do the work as these forces acts in a direction perpendicular to the
velocity.

11.11 Check your progress

Check your progress by answering the questions below.

1. What is a rigid body?

2. Define center of mass of a rigid body.

3. Define orthogonal transformations.

181
MPDSC 1.1 KSOU

4. Define Euler’s theorem.

5. Define Coriolis force.

11.12 Keywords

• Rigid body

• Center of mass

• Euler’s angles

• Euler’s theorem

• Coriolis force

11.13 Worked examples

1. A coordinate transformation is represented by a matrix

 
2 −2 1
1 
1 2 2 
 
3 
2 1 −2

Check whether the transformation is orthogonal transformation or not.

Answer:

We know that the condition for orthogonal transformation is that aij aik = δjk . In matrix
e = I.
representation, if a matrix T represents an orthogonal transformation then, TT

Given    
2 −2 1 2 1 2
1 
e= 1
 
T= 1 2 2  =⇒ T  −2 2 1 
  
3  3 
2 1 −2 1 2 −2

Consider

182
MPDSC 1.1 KSOU

    
2 1 2 2 −2 1 4 + 1 + 4 −4 + 2 + 2 2 + 2 − 4
1  
e =  −2 2 1  1 2
 1  
TT 2  =  −4 + 2 + 2 4 + 4 + 1 −2 + 4 − 2
     
9   9 
1 2 −2 2 1 −2 2 + 2 − 4 −2 + 4 − 2 1 + 4 + 4

   
9 0 0 1 0 0
1   
TT = 0 9 0 = 0 1 0 = I
e    
9   
0 0 9 0 0 1

Thus the given transformation is an orthogonal transformation.

2. A rotation matrix is given by

 
3 −6 2
1 
2 3 6 
 
7 
6 2 −3

Determine the Euler angles corresponding to the above rotation matrix.

Answer:

The given rotation matrix is

     
3 −6 2 3/7 −6/7 2/7 0.42857 −0.85714 0.28571
1     
R = 2 3 6  = 2/7 3/7 6/7  = 0.28571 0.42857 0.85714 
     
7     
6 2 −3 6/7 2/7 −3/7 0.85714 0.28571 −0.42857

We know the general rotation matrix in terms of Euler angles

 
cos ψ cos φ − cos θ sin ψ sin φ cos ψ sin φ + sin ψ cos θ cos φ sin ψ sin θ
 
R = − sin ψ cos φ − cos ψ cos θ sin φ − sin ψ sin φ + cos θ cos φ cos ψ cos ψ sin θ 
 
 
sin θ sin φ − sin θ cos φ cos θ

Comparing the above two matrices we can write,

183
MPDSC 1.1 KSOU

cos θ = −0.42857 =⇒ θ = cos−1 (−0.42857) = 115.377o

 
−1 0.85714
sin θ sin φ = 0.85714 =⇒ φ = sin = 71.564o
sin(115.377)
 
−1 0.28571
sin ψ sin φ = 0.28571 =⇒ ψ = sin = 18.435o
sin(115.377)

3. A particle of mass 500 g is at rest in an frame rotating with an angular velocity of 10πrad/s
about its z-axis. If the particle is situated at ~r 0 = 9î0 − 12 ĵ0 + 4k̂0 with respect to the rotating
frame, find the centrifugal force acting on the particle. (The distances are in cm)

If the particle starts to move with a velocity ~v = 3î0 + 4 ĵ0 − 2k̂0 , determine the Coriolis force
acting on the particle. (The velocity is in cm/s)

Answer: Given m = 500g and,

~ = 10π k̂0
ω

~r 0 = 9î0 − 12 ĵ0 + 4k̂0

~v = 3î0 + 4 ĵ0 − 2k̂0

We know that the centrifugal force is

~F = −mω
~ × (~ω ×~r )
   
0 0 0  0 0 0
î ĵ k̂ î ĵ k̂ 
 
 
   
~F = −mω 0 0 0
~ ×  0 0 10π  = −m[(10π k̂ ) × (120π î + 90π ĵ )] =  0 0 10π 
   
   
 9 −12 4 
  
120π 90π 0 


~F = 0.5(−900π 2 î − 1200π 2 ĵ) × 10−2 N

The magnitude of the force will be 7.5π 2 N.

We know the expression for Coriolis force

 
0 0 0
î ĵ k̂ 

 
F = −2m(~ω × ~vr ) = −2 × 0.5 ×  0 0 10π  × 10−1 N
 
 
 3 4 −2 
 

ω × ~vr ) = −2 × 0.5 × (40π î − 30π ĵ) × 10−2 N

F = −2m(~

F = −0.4π î + 0.3π ĵ N

The magnitude of Coriolis force is 0.5 π N.

184
MPDSC 1.1 KSOU

11.14 Questions for self study

1. Define center of mass of a rigid body.

2. Explain orthogonal transformations.

3. Explain how Euler angles are defined.

4. State and explain Euler’s theorem.

5. Describe how rate of change of a vector is connected between the space fixed and body
fixed reference frames.

6. Deduce an expression for Coriolis force.

11.15 Answers to check your progress

1. A solid body in which the distance between any two particles is always constant is called
a rigid body.

2. A point where the entire mass of the rigid body can be assumed to be concentrated is
called center of mass.

3. A transformation that satisfies the condition aij aik = δjk is called orthogonal transforma-
tion.

4. Euler’s theorem states that the general displacement of a rigid body with one point fixed
is a rotation motion about some axis..

5. The non inertial force that exclusively acts on a moving object in rotating frame is called
Coriolis force.

11.16 References

• Herbert Goldstein, Charles Poole, John Safko, (2010) Classical Mechanics. Pearson Educa-
tion.

• G. Aruldhas, (2008) Classical Mechanics. Prentice Hall of India pvt. Ltd.

185
MPDSC 1.1 KSOU

• R. G. Takwale and P. S. Puranik (2006) Introduction to Classical Mechanics. Tata McGraw-

Hill Publishing Company Ltd.

• N. C. Rana and P. S. Joag (2005) Classical Mechanics. Tata McGraw-Hill Publishing Com-
pany Ltd.

186
 UNIT-12: Rigid body dynamics-2

The angular momentum and kinetic energy of motion about a point, inertia tensor and prin-
cipal axis transformations, Euler equations of motion, torque free motion of rigid body, heavy
symmetrical top with one point fixed.

12.0 Objectives

After studying this unit you will be able to

• Describe angular momentum of rigid body.

• Obtain an expression for kinetic energy of a rigid body.

• Describe inertia tensor and explain its properties.

• Explain principal axes of a rigid body.

• Deduce Euler equations of motion of rigid body.

• Describe torque free motion of a symmetric top.

• Describe the motion of heavy symmetric top with one point fixed.

12.1 Introduction

In the previous unit, we discussed several kinematical aspects of rigid body dynamics. In this
unit we shall study several more interesting concepts.

12.2 Angular momentum

Consider a rigid body consisting of n particles of mass mi with i = 1, 2, 3, ...n. Let the body rotate
~ about an axis passing through its center of mass. Let
with an instantaneous angular velocity ω
the radius vector of a particle of mass mi be ~ri . Then its instantaneous translation velocity ~vi
would be
MPDSC 1.1 KSOU

~ ×~ri
~vi = ω

Let ωx , ωy and ωz be the components of angular velocity. Then the angular momentum of
the particle about origin is defined as

~l = ~ri × ~pi

Then the angular momentum of the whole rigid body will be

n
~L = ∑ ~ri × ~pi = ∑ mi~ri × ~vi
i =1 i

~L = ∑ mi~ri × (~
ω ×~ri ) = ∑ mi [ri2 ω
~ −~ri (~ri · ω
~ )]
i i

We can expand the above expression for the angular momentum in terms of the component
as,

~L = ∑ mi [( xi2 + y2i + z2i )(îωx + ĵωy + k̂ωz ) − (îx + ĵy + k̂z)( xi ωx + yi ωy + zi ωz )]

i

h  
~L = ∑ îmi y2i ωx + z2i ωx − xi yi ωy − xi zi ωz
i
 
+ ĵmi xi2 ωy + z2i ωy − xi yi ωx − yi zi ωz
 i
+k̂mi xi2 ωz + y2i ωz − xi zi ωx − yi zi ωy

We can expand the angular momentum also as ~L = L x î + Ly ĵ + Lz k̂. Then we can extract the
expressions for components of angular momentum from above equation as

 
Lx = ∑ i i
m y 2
ω x + z 2
i ω x − x y
i i ω y − x z
i i ω z
i
 
Ly = ∑ mi xi2 ωy + z2i ωy − xi yi ω x − yi zi ωz
i
 
Lz = ∑ i i z i z ii x ii y
m x 2
ω + y 2
ω − x z ω − y z ω
i

188
MPDSC 1.1 KSOU

Because y2i + z2i = ri2 − xi2 ; xi2 + z2i = ri2 − y2i and xi2 + y2i = ri2 − z2i we can rewrite the above
set of equations as

 
Lx = ∑ mi (ri2 − xi2 )ωx − xi yi ωy − xi zi ωz
i
 
Ly = ∑ i i i y i i x ii z
m ( r 2
− y 2
) ω − x y ω − y z ω
i
 
Lz = ∑ i i i z ii x ii y
m ( r 2
− z 2
) ω − x z ω − y z ω
i

With a small rearrangement, the above equations are written as,

 
Lx = ∑ i i i x i i y ii z
m ( r 2
− x 2
) ω − x y ω − x z ω
i
 
Ly = ∑ i
m − x y
i i ω x + ( r 2
i − y 2
i ) ω y − y z
i i ω z
i
 
Lz = ∑ mi − xi zi ωx − yi zi ωy + (ri2 − z2i )ωz
i

The above equations can be written in matrix form as

    
Lx
   ∑i mi (ri2 − xi2 ) − ∑i mi xi yi − ∑i mi xi zi
 
ωx
 Ly  =  − ∑i mi xi yi ∑i mi (ri2 − y2i ) − ∑i mi yi zi   ωy 
    
    
− ∑i mi xi zi − ∑i mi yi zi 2 − z2 )
Lz ∑i i i
m ( r i ωz

The above equation gives the expression for angular momentum of the rigid body. Com-
paring the above expression for the angular momentum with the expression ~L = I ω
~ we can
write
 
 ∑i mi (ri2 − xi2 ) − ∑i mi xi yi − ∑i mi xi zi

I= − ∑i mi xi yi ∑i mi (ri2 − y2i ) − ∑i mi yi zi
 

 
− ∑i mi xi zi − ∑i mi yi zi ∑i mi (ri2 − z2i )

This is known as moment of inertia tensor or simply inertia tensor. Thus we can conclude
that the moment of inertia is not a scalar but a second rank tensor with the components given
as below.

189
MPDSC 1.1 KSOU

Ixx = ∑i mi (ri2 − xi2 ) Ixy = − ∑i mi xi yi Ixz = − ∑i mi xi zi

Iyx = − ∑i mi xi yi Iyy = ∑i mi (ri2 − y2i ) Iyz = − ∑i mi yi zi
Izx = − ∑i mi xi zi Izy = − ∑i mi yi zi Izz = ∑i mi (ri2 − z2i )

Note that the inertia tensor is a symmetric tensor with just six independent components.
Then the angular momentum of the rigid body will be

~L = î (ωx Ixx + ωy Ixy + ωz Ixz ) + ĵ(ωx Iyx + ωy Iyy + ωz Iyz ) + k̂(ωx Izx + ωy Izy + ωz Izz )

12.3 Kinetic energy

We now derive a general expression for the rotational kinetic energy of a rigid body. Consider a
~ . A particle of mass mi
rigid body rotating through a fixed point in it with an angular velocity ω
at a distance ~ri has a velocity ~vi . The kinetic energy of the whole body will be

n
1 1
T= ∑ 2 mi v2i = 2 ∑(~ω ×~ri ) · (mi~vi )
i =1 i

The above expression for kinetic energy is making use of scalar triple product of three vec-
tors. We know that in scalar triple product of three vectors, the dot and cross are interchange-
~ × ~B) · C
able. i.e. ( A ~ = A
~ · (~B × C
~ ). Then we can rewrite the expression for the kinetic energy
will be

1 1
T= ∑
2 i
~ · (~ri × mi~vi ) = ω
ω
2
~ · ∑~ri × mi~vi
i

Interestingly, ~ri × mi~vi = ~L. hence the expression for kinetic energy will become

1 ~
T= ~ ·L
ω
2
~ and ~L from the previous section, we can further write
Using the expanded expressions for ω
the expression for kinetic energy as

1 2
T= (ωx Ixx + ωy2 Iyy + ωz2 Izz ) + ωx ωy Ixy + ωy ωz Iyz + ωz ωx Izx
2
In more compact form

190
MPDSC 1.1 KSOU

1
2∑
T= ωα ω β Iαβ
β

Further, we can observe that

∂T ∂T ∂T
Lx = Ly = Lz =
∂ωx ∂ωy ∂ωz

12.4 Inertia tensor

As we have discussed in the previous sections, the moment of inertia or simply inertia is no more
a scalar, but a second rank tensor. The inertia tensor is a symmetric tensor with six independent
components. The components of the inertia tensor are given by

Ixx = ∑i mi (ri2 − xi2 ) Ixy = − ∑i mi xi yi Ixz = − ∑i mi xi zi

Iyx = − ∑i mi xi yi Iyy = ∑i mi (ri2 − y2i ) Iyz = − ∑i mi yi zi
Izx = − ∑i mi xi zi Izy = − ∑i mi yi zi Izz = ∑i mi (ri2 − z2i )

However, one can argue that a rigid body is not a discrete collection of particles rather than
a continuous distribution of matter. Hence the integration would be more suitable than sum-
mation. Hence the components of inertia tensor can we rewritten as

Z Z
2 2
Ixx = ρ(r − x )dV Ixy = Iyx = − ρxydV
V V
Z Z
2 2
Iyy = ρ(r − y )dV Iyz = Izy = − ρyzdV
V V
Z Z
Izz = ρ(r2 − z2 )dV Izx = Ixz = − ρxzdV
V V

In a compact form the components of the inertia tensor is written as

Z
Ijk = ρ(r )(r2 δjk − x j xk )dV
V

The expression for the angular momentum as we have already discussed in terms of inertia
tensor will be

~L = I · ω
~

191
MPDSC 1.1 KSOU

    
L I I I ω
 x   xx xy xz   x 
 Ly  =  Iyx Iyy Iyz   ωy 
    
    
Lz Izx Izy Izz ωz

Similarly the expression for the kinetic energy would be

1˜ ~ 1˜
T= ~ ·L= ω
ω ~ ·I·ω
~
2 2

  
Ixx Ixy Ixz ω
1
   x
T= ωx ωy ωz  Iyx Iyy Iyz   ωy 
  
2   
Izx Izy Izz ωz

12.5 Principal axes

The inertia tensor we defined is with respect to a coordinate system which is fixed to a point
in the body. We can simplify the mathematical calculations considerably if we choose the co-
ordinate axes in such a way that the off-diagonal elements vanish. The axes of this coordinate
system are known as the principal axes of the body. The origin of the principle axes system is
called the principal point. The three coordinate planes each of which passes through two princi-
pal axes are called principal planes at the origin. IN this system, the inertia tensor is a diagonal
and the three elements of the inertia are called principal moments of inertia. It is the practice
to use a single subscript for the principal moments to distinguish them from the moments of
inertia about arbitrary axes. The principal moments of inertia are thus denoted by I1 , I2 and I3 .
Since the principal axes are attached to the rigid body, I1 , I2 and I3 do not change with time.
Therefore they may be treated as constants. It is for this reason that moving axes attached to the
body are employed. In the principal axes system,

~L = I1 ω1 eˆ1 + I2 ω2 eˆ2 + I3 ω3 eˆ3

Where eˆ1 , eˆ2 and eˆ3 are the unit vectors along the three principal axes. The expression for the
kinetic energy in principal axes system will be

1 1 1
T= I1 ω12 + I2 ω22 + I3 ω32
2 2 2

192
MPDSC 1.1 KSOU

Now let us see how to find the principal axes. Sometimes we may be able to fix up the
principal axes by examining the symmetry of the body. In the general case, suppose we have
the moments and products of inertia of a body with respect to an arbitrary set of x, y and z-axes.
Then we can write

I11 ω1 + I12 ω2 + I13 ω3 = Iω1

I21 ω1 + I22 ω2 + I23 ω3 = Iω2

I31 ω1 + I32 ω2 + I33 ω3 = Iω3

Rearranging the above equations we can write,

( I11 − I )ω1 + I12 ω2 + I13 ω3 = 0

I21 ω1 + ( I22 − I )ω2 + I23 ω3 = 0

I31 ω1 + I32 ω2 + ( I33 − I )ω3 = 0

For a solution to exist for the above set of homogeneous simultaneous equations the deter-
minant of the coefficient must become zero.
 
 I11 − I I12 I13 
 
 
 I21 I22 − I I23  = 0
 
 
 I31 I32 I33 − I 
 

On solving the above determinant for I we will get three values of I that form the principal
moments of inertia. The above determinant is nothing but the characteristic equation of the
inertia tensor. Hence, the eigenvalues of the inertia tensor are the principal moments of inertia.
The corresponding eigenvectors provide the direction of principal axes.
Hence to find the principal axes, one can start with the known inertia tensor and find its
eigenvalues and eigenvectors. The eigenvalues provide the values of principal moments of
inertia and the eigenvectors provide the direction of the principal axes.
Based on the values of principal moments of inertia rigid bodies are classified into three
categories.
Spherical top: If I1 = I2 = I3 for a rigid body then it is known as spherical top. Then the
inertia tensor will be a identity tensor. For spherical top, any three mutually perpendicular axes

193
MPDSC 1.1 KSOU

can be chosen as principal axes.

Symmetric top: If I1 = I2 < I3 or I1 < I2 = I3 for a rigid body, then it is known as symmetric
top. For such a rigid body, two principal moments of inertia are equal. The rigid bodies that
satisfy the first condition are known as oblate symmetric top and those satisfy the second con-
dition are called prolate symmetric top. For symmetric top, one of the principle axes get fixed
along the direction that exhibits rotational symmetry about that axis. The other two mutually
perpendicular axes can be chosen to be perpendicular to the fixed in any manner.
Asymmetric top: These are general shaped rigid bodies for which all the three principal
moments of inertia are different.

12.6 Euler equations of motion

The inertial tensor I with respect to a body fixed coordinate system is a constant. In a principal
axes system also it is diagonal. Consider the rotation of a rigid body with one point fixed,
which is taken as the origin of a body fixed coordinate system. The rotational analogue of
Newton’s second law gives the rate of change of angular momentum with respect to a space
~ acting on the body
fixed coordinate. The torque N
!
~ = d~L
N
dt
s

In rotating frame,
! !
d~L d~L
= ~ × ~L
+ω
dt dt
s r
!
~ = d~L
∴N ~ × ~L
+ω
dt
r

d~L
we know ~L = I ω
~ . Hence ~˙ . Then the above equation will become
= Iω
dt

N ~˙ + ω
~ = Iω ~ × ~L

If we select the principal axes of the body as the body axes, the above equation can be ex-
panded in terms of the components as

194
MPDSC 1.1 KSOU

N1 = I1 ω̇1 + ω2 L3 − ω3 L2 = I1 ω̇1 + ω2 ω3 ( I3 − I2 )

=⇒

N1 + ω2 ω3 ( I2 − I3 ) = I1 ω̇1

Similarly we can write the equations for all the three components as

I1 ω̇1 = ω2 ω3 ( I2 − I3 ) + N1

I2 ω̇2 = ω3 ω1 ( I3 − I1 ) + N2

I3 ω̇3 = ω1 ω2 ( I1 − I2 ) + N3

The above three equations are known as Euler’s equations of motion of rigid body with one
point fixed. In the absence of external torques, the Euler’s equations reduce to simpler forms.

I1 ω̇1 = ω2 ω3 ( I2 − I3 )

I2 ω̇2 = ω3 ω1 ( I3 − I1 )

I3 ω̇3 = ω1 ω2 ( I1 − I2 )

The external torques acting on the earth are so weak but the rotational motion can be consid-
ered as torque free in the first approximation. Then the angular velocity and angular momentum
are not parallel vectors. Though angular momentum is a conserved quantity, angular velocity
is not. In general, the angular velocity will precess around teh angular momentum vector and
the angle between them varies in time. This is known as nutation.

12.7 Torque free motion of a symmetric top

As an example of Euler’s equation of motion, we shall consider a special case in which the torque
~ = 0 and the body is symmetrical top. A symmetrical top possesses an axis of symmetry and
N
therefore two of the principal moments of inertia are equal. If the axis of symmetry is taken as
z-axis, I1 = I2 = I 6= I3 . For force free or torque free motion, the center of mass is either at
rest or in uniform motion relative to the spaced fixed inertial system. Therefore, we can take

195
MPDSC 1.1 KSOU

the center of mass as the origin of the body-fixed coordinate system. In such a case the angular
momentum arises only from the rotation about the center of mass. For such a symmetric body
the Euler equations reduces to

I ω̇1 = ω2 ω3 ( I − I3 )

I ω̇2 = ω3 ω1 ( I3 − f I )

I3 ω̇3 = 0

The last equation implies ω3 is a constant in time.

Let us define k as

I3 − I
k= ω3
I
we can rewrite the first two Euler equations as

ω̇1 = −kω2 and ω̇2 = kω1

Differentiating the first equation with respect to time and replacing ω̇2 by kω1 we can write

ω̈1 = −k ω̇2 = −k2 ω1

Note that ω1 (with the similar procedure ω2 also) satisfies simple harmonic motion equation.
hence its solution can be taken as

ω1 = A cos(kt + θo ) and ω2 = A sin(kt + θo ) and ω12 + ω22 = A2 , a constant

Since ω3 is constant and ω12 + ω22 is also constant |~

ω |2 = ω12 + ω22 + ω32 is also a constant.
Hence, we can say that the magnitude of the angular velocity is a constant but the angular
velocity as a vector is a function of time. Specifically a harmonic function of time. The angular
momentum vector hence precesses the z-axis and form a cone. This cone is known as body cone.
The frequency of this harmonic motion would be

k ( I − I ) ω3
fp = = 3
2π 2πI

196
MPDSC 1.1 KSOU

And the half angle of the body cone is given by

q
ω12 + ω22
tan α =
ω3
An interesting example of the force free motion of a symmetric body is provided by the
earth’s rotation. The earth rotates freely about its polar axis. Its axis of rotation depends slightly
from its symmetry axis. From astronomical measurements ( I3 − I )/I is found to be 0.00329.
Then the period of precession of z-axis will be

2π 2πI 1 day
Tp = = = = 304 days
k ω3 ( I3 − I ) 0.00329
Recent measurements give Tp = 433 days. The discrepancy is probably due to the fact that
the earth is not perfectly rigid. Thus the earth’s rotation axis precesses about the north pole in a
circle of radius about 10 m with a period of about 433 days.

12.8 Heavy symmetric top with one point fixed

Figure 12.22: Rotating heavy symmetric top

As a second example of rigid body dynamics, we consider the motion of a heavy symmetrical
top spinning freely about its symmetry axis under the influence of a torque produced by its own
weight. The symmetry axis of the body is one of its principal axis and we choose it as the z-axis
of the coordinate system fixed in the body. The body is fixed at the point O which is on the

197
MPDSC 1.1 KSOU

symmetry axis but does not coincide with the center of gravity. Point O is taken as the origin of
the space fixed and body fixed coordinate system. As the symmetry axis is selected as the z-axis
of the body fixed system,

I1 = I2 = I 6= I3

As the translational kinetic energy is zero, the kinetic energy is purely rotational that can be
taken as

1
T= [ I (ω12 + ω22 ) + I3 ω32 ]
2
According to Euler’s geometrical equations,

ω1 = φ̇ sin θ sin ψ + θ̇ cos ψ

ω2 = φ̇ sin θ cos ψ − θ̇ sin ψ

ω3 = φ̇ cos θ + ψ̇

Substituting these in the expression for the kinetic energy, we get,

1 2 1
T= I (θ̇ + φ̇2 sin2 θ ) + I3 (ψ̇ + φ̇ cos θ )2
2 2
The height of the center of gravity from the point of support is l cos θ and therefore the
potential energy will be

V = Mgl cos θ

Then the Lagrangian of the top will be

1 2 1
L = T−V = I (θ̇ + φ̇2 sin2 θ ) + I3 (ψ̇ + φ̇ cos θ )2 − Mgl cos θ
2 2
Note that the Lagrangian does not contain φ and ψ explicitly. Hence, they are cyclic coordi-
nates. Therefore, the corresponding conjugate momenta are constants of motion. That is

∂L
pφ = = I φ̇ sin2 θ + I3 (ψ̇ + φ̇ cos θ ) = constant
∂φ̇

198
MPDSC 1.1 KSOU

∂L
pψ = = I3 (ψ̇ + φ̇ cos θ ) = constant
∂ψ̇
Here, pφ is the angular momentum due to the angular rotation of φ about z-axis. These two
are the first integrals of motion. Another first integral is the total energy E that is given as,

1 2 1
E= I (θ̇ + φ̇2 sin2 θ ) + I3 (ψ̇ + φ̇ cos θ )2 + Mgl cos θ
2 2
Using the first integrals of the momentum we can extract the expressions for the velocities
in terms of momentum and use them in the expression for energy. This gives us

1 2 ( pφ − pψ cos θ )2 p2ψ
E = I θ̇ + + + Mgl cos θ
2 2I sin2 θ 2I3
p2ψ ( pφ − pψ cos θ )2
Let us redefine the energy and effective potential as E0 = E− and V 0 (θ ) = +
2I3 2I sin2 θ
Mgl cos θ. Then the total effective energy will be

1 2
E0 = I θ̇ + V 0 (θ )
2
=⇒
r
2( E 0 − V 0 )
θ̇ =
I
Integrating the above equation we can write the equation of motion as

dθ
Z
t(θ ) = p
2/I ( E0 − V 0 )
A plot of effective potential V 0 (θ ) versus θ for the physically acceptable range of 0 ≤ θ ≤ π
is given in the figure above. It is obvious that the motion will be limited to the case E0 > V 0 .
For any energy value E0 = E10 the motion is limited between two extreme values θ1 and θ2 .
This implies that the angle that the symmetry axis can make with the vertical is m=limited
between these two angles. In other words, symmetry axis of the top will be bobbing back and
forth between two right circular cones of half angles θ1 and θ2 while precessing with an angular
velocity φ̇. Such a bobbing back and forth motion is called nutation. Different possibilities of
nutation are given in the figure .

199
MPDSC 1.1 KSOU

E10
V 0 (θ ) →

Eo0

O
θ1 θo θ2 θ→

Figure 12.23: Effective potential V 0 (θ ) versus θ for a heavy symmetric top

(a) (b) (c)

Figure 12.24: The motion of symmetry axis between θ1 ≤ θ ≤ θ2 projected on a unit sphere fixed
system: (a) φ̇ never changes sign. (b)φ̇ changes sign at θ > θ1 ; (c) φ̇ vanishes at θ = θ1 .

200
MPDSC 1.1 KSOU

12.9 Check your progress

Check your progress by answering the questions below.

1. Define moment of inertia tensor.

2. What are principal axes?

3. How inertia tensor looks like in principal axes system.

4. What is a symmetric top?

12.10 Keywords

• Inertia tensor

• Euler’s theorem

• Principal axes

• Principal moments of inertia

12.11 Worked examples

1. Find the moments of inertia of a homogeneous cube of side a for an origin at one corner,
with axes directed along the edges

Answer:

We know

Z Z a Z a Z a 
2 2 2 2
Ixx = ρ(y + z )dV = ρ dx (y + z )dy dz
V 0 0 0

Z a 3
a4 a4
  
a 2 2
Ixx = ρa + az dz = ρa + = ρa5
0 3 3 3 3

2 2
Ixx = ρa3 a2 = Ma2
3 3

Where M is the mass of the cube. Then by similar arguments we can write

201
MPDSC 1.1 KSOU

2
Iyy = Izz = Ma2
3

Now consider

Z Z a Z a Z a
Ixy = − ρxydV = −ρ dz xdx ydy
V 0 0 0

1
On integration we can show that Ixy = − Ma2
4
By symmetry we can argue that Ixy = Iyz = Izx . Then the inertia tensor will be

2 1 1
 
Ma2 − Ma2 − Ma2
 3 4 4 
 1 2 1 
I= 2 2 2
− 4 Ma 3
Ma − Ma 
4 
 1 1 2 
2
− Ma − Ma 2 Ma 2
4 4 3

2. A body can rotate freely about the principal axis corresponding to the principal moment
of inertia I3 . If it is given a small displacement, show that the rotation will be oscillatory if
I3 is either the largest or the smallest of the three principal moments of inertia.

Answer: Since the displacement is small, we may take ω1 and ω2 as small and the product
ω1 ω2 may be neglected. Then from third Euler equation of motion we can write,

ω̇3 = 0 =⇒ ω3 = constant

From the first Euler equation of motion,

ω3 ( I2 − I3 )
ω̇1 = ω̇2
I1

Substituting this in the second equation,

( I3 − I2 )( I1 − I3 ) 2
 
ω̈1 = ω3 ω1
I1 I2

Because everything inside the square brackets is constant, we can write

ω̈3 = k2 ω1

202
MPDSC 1.1 KSOU

( I3 − I2 )( I1 − I3 ) 2
 
Where, k2 = ω3
I1 I2
Thus we can observe the variation of ω1 will be oscillatory only if k2 is negative. This is
possible only if I3 is the highest or lowest of all the three principal moments of inertia.

3. A body moves about a point O under no force. The principal moments of inertia at O
being 3A, 5A and 6A. Initially the angular velocity has components ω1 = ω, ω2 = 0 and
ω3 = ω about the corresponding principal axes. Show that at time t

3ω ωt
ω2 = √ tan √
5 5

Answer: In the torque free case, the Euler’s equation are

I1 ω̇1 = ω2 ω3 ( I2 − I3 )

I2 ω̇2 = ω1 ω3 ( I3 − I1 )

I3 ω̇3 = ω1 ω2 ( I1 − I2 )

Replacing the principal moments of inertia by 3A, 5A and 6A respectively,

3ω̇1 = −ω2 ω3 5ω̇2 = 3ω3 ω1 6ω̇3 = −2ω1 ω2

Multiplying the first equation by 3ω1 and second equation by ω3 = 2 and adding the two
we get,

9ω1 ω̇1 + 5ω2 ω̇2 = 0

Integrating the above equation with the inital conditions we can write

9ω12 + 5ω22 = 9ω 2

Similarly we can show

ω12 = ω32

203
MPDSC 1.1 KSOU

Combining these results we can write,

5ω22 9ω 2 − 5ω22
5ω̇2 = 3ω12 2
= 3ω − =⇒ ω̇2 =
3 15

Integrating

dω2 dω2
Z Z
t = 15 =3
9ω − 5ω22
2 (9/5)ω − ω22

√ √ !
5 5ω2
t= tanh−1
ω 3ω

 
3ω ωt
∴ ω2 = √ tanh √
5 5

12.12 Questions for self study

1. Describe angular momentum of a rigid body.

2. Obtain an expression for kinetic energy of a rigid body.

3. Describe inertia tensor and explain its properties.

4. What are principal axes of a rigid body.

5. Explain the inertia tensor in principal axes system.

6. Deduce the Euler’s equations of motion of a rigid body.

7. Describe the torque free motion of a symmetric top.

8. Discuss the motion of heavy symmetric top with one point fixed rotating under the action
of gravity.

12.13 Answers to check your progress

1. Inertia tensor is defined as

204
MPDSC 1.1 KSOU

 
 ∑i mi (ri2 − xi2 ) − ∑i mi xi yi − ∑i mi xi zi

I= − ∑i mi xi yi ∑i mi (ri2 − y2i ) − ∑i mi yi zi
 

 
− ∑i mi xi zi − ∑i mi yi zi ∑i mi (ri2 − z2i )

2. The coordinate axes set fixed to the rigid body in which the inertia tensor will be a diagonal
tensor are called principal axes.

3. Diagonal tensor with all off diagonal elements zero.

4. A rigid body in which any two of its principal moments of inertia are equal to each other
and the third one being different is a symmetric top.

12.14 References

• Herbert Goldstein, Charles Poole, John Safko, (2010) Classical Mechanics. Pearson Educa-
tion.

• G. Aruldhas, (2008) Classical Mechanics. Prentice Hall of India pvt. Ltd.

• R. G. Takwale and P. S. Puranik (2006) Introduction to Classical Mechanics. Tata McGraw-

Hill Publishing Company Ltd.

• N. C. Rana and P. S. Joag (2005) Classical Mechanics. Tata McGraw-Hill Publishing Com-
pany Ltd.

205