Sunteți pe pagina 1din 21

E.

Harikumar 28

6 Generalized Coordinates
For various reasons, it may be convenient to work with co-ordinate system
other than the Cartesian co-ordinates. An example is the planetary motion
which is easy to describe in (r, θ) co-ordinates. Here,

q1 = r = X12 + X22 (6.1)
� �
X2
q2 = θ = tan−1 (6.2)

T
X1
Note here r(X1 , X2 ), θ(X1 , X2 ); X1 (r, θ) = r cos θ, X2 (r, θ) = r sin θ. Thus,
in general we may need to work with coordinates.

Xi = Xi (q1 , ..., qn , t), i = 1, 2, 3, (6.3)


AF
and the inverse relations are

qn = qn (Xi , t) (6.4)

In the above, in general no.of values “n” can take need not be same as that
of “i”! An example where such a situation can be seen is the case of a
simple pendulum. The position of the bob of a simple pendulum is given
by X and Y co-ordinates. But just “θ” is enough to describe the same in
(r − θ)system.This is due to the constraint X 2 + Y 2 = r2 . Thus here we have
two dimensional Cartesian coordinate with one condition among them and
we say that the system has just one degree of freedom. Thus for a particle
DR

in “d” dimension with “k” number of constraints, the degrees of freedom is


(d−k). In general, we need to describe system with N coordinates and having
k no. of constraint relations among these N coordinates. Our description of
mechanics will be in terms of independent, generalised coordinates.
The new set of coordinates qm are called generalized coordinates and
dqm
dt
= q̇m are called generalized velocities. From Eqn.(6.3), we see
� ∂Xi
δXi = δqn . (6.5)
n
∂qn

This inhomogeneous, linear equation in δqn can be solved if,


� �
� ∂Xi �
� �
� ∂qn � �= 0. (6.6)

When we deal with systems of large number of particles, it is more conve-


nient to use generalized coordinates. Generalised coordinates can be used to
E. Harikumar 29

describe systems with constraints also one can use generalized coordinates.
In the case of such systems, these conditions/constraints between the coor-
dinates which can be expressed generically as Φ(X1 , X2 , ....Xn ) = 0 3 .
See the example of simple pendulum. Here we have the condition that
the length of the pendulum is fixed, that is, the coordinates of the bob should
satisfy the condition

X12 + X22 − l2 = 0. (6.7)

T
If we know X1 , then X2 is uniquely fixed. We can choose the generalized
coordinates q1 and q2 such that

q2 = (X12 + X22 − l2 ) = 0 (6.8)

and this reduces the problem to an one-dimensional problem.


AF
In general, among the qn generalized coordinates, we can choose, say the
last “k” of them to be the constraints φk (X1 , .....Xd , t) = 0 themselves. Then
(n − k) is the degrees of freedom of the system.
Next consider a system of N particles. To describe the mechanics of this
(α)
system, we need 3N number of coordinates-that is Xi , i = 1, 2, and 3 and
α = 1, 2, ....N . Here, what we want to know is the change in position of the
(α)
system in time. That is, we need to find Xi (t). Thus, the dynamics is
nothing but the description of the change of 3N variables -3N coordinates-
as time changes. We can think of this as the motion of a single particle
in 3N dimension. This 3N dimensional space is known as unconstrained
configuration space. If there are “k” number of constraints-relations among
DR

the coordinates of the system- then the degrees of freedom is (3N − k).
That is, there exist φs (X1 , .....X3N , t) = 0; s = 1, ..k relation between 3N
coordinates. Then we have Xiα = Xiα (q1 .....q3N −k , t) and this can be inverted
as

qi = qi (X1 , ......X3N −k , t), i = 1, .....3N − k (6.9)


qi+s = 0(= φs ), s = 1, 2, ....k. (6.10)

This (3N − k) dimensional space is called configuration space of the sys-


tem. Thus the problem of dynamics of N particles in 3-dimension with ‘k’
constraints in the coordinate space is equivalent to the problem of a sin-
gle particle in (3N − k) dimensional configuration space defined in terms of
qi , i = 1, 2, ....3N − k generalized coordinates.
3
There can be constraints which are inequalities, conditions involving derivatives of
coordinates, conditions that are time dependent.
E. Harikumar 30

We want a framework which will allow us to predict the mechanical state


of a system at a later time, if it is given at a particular instant of time. This
can be obtained by solving equations of motion. Equations of motion are the
relations between the accelerations, velocities and positions of the system.

7 Lagrange’s Equation
For a system of (3N −k) degrees of freedom(dof), we have Xiα (q1 , q2 .....q3N −k , t),

T
the generalized coordinates describing the system. Then ,
dXiα � ∂Xiα dqm ∂Xi α
= + , m = 1, 2, ....., 3N − k (7.1)
dt m
∂qm dt ∂t

Note
� ∂qm

and
AF
q̇m =
dqm =

dqm � ∂qm
=
i
∂Xi

Ẋi +
∂qm
dXi +
∂qm
∂t
,

= q̇(Xi , Ẋi , t).


(7.2)

(7.3)
dt i
∂X i ∂t
Note that q̇m is independent of qm . Thus we get from Eqn.(7.1)

∂ Ẋiα ∂Xiα
= (7.4)
∂ q̇m ∂qm
DR
E. Harikumar 31


If this system is subject to a force F� = α F� α , the work done by this force
in displacing the system is

W = Fiα · dXiα (7.5)
α,i

= Mα Ẍiα dXiα (7.6)
α,i
� � �
∂Xiα ∂Xiα
= Mα Ẍiα dqm + dt (7.7)

T
α,i,m
∂qm ∂t
� �
� d ∂ Ẋ α
d ∂X α
i
= Mα (Ẋ α ) − Ẋiα ( i ) dqm
α,i,m
dt i ∂ q̇m dt ∂qm
� (α)
(α) ∂Xi
+ Mα Ẍi dt (7.8)

=
α,i

α,i,m
AF
� � d ∂ � Mα
dt ∂ q̇m
∂t

α α
Ẋ Ẋ
2 i i



∂qm

Mα (α) (α)
Ẋ Ẋi
2 i
��
dqm

� (α)
(α) ∂Xi
+ Mα Ẍi dt (7.9)
α,i
∂t

Thus4
� � d ∂T ∂T
� � (α) ∂Xi
W = − dqm + Mα Ẍi dt (7.10)
m
dt ∂ q̇ m ∂q m α,i
∂t
DR

Alternatively from Eqn(13.7), we get


� �
� (α) ∂X (α) ∂X
(α)
i i
W = Fi dqm + dt (7.11)
α,i,m
∂q m ∂t

= Qm dqm + Qt dt (7.12)
m

where
� (α) ∂Xi
(α)
Qm = Fi (7.13)
α,i
∂qm
� (α)
(α) ∂Xi
Qt = Fi (7.14)
α,i
∂t

4 d ∂Xi
α
∂ ∂Xi
α � ∂
α
∂Xi ∂
α
dxi
In the above, we have used, dt ( ∂qm )= ∂t ( ∂qm )+ n ∂qn ( ∂qm )q̇n = ∂qm ( dt ).
E. Harikumar 32

are the components of generalized force Q. Equating coefficients of dqm and


dt in Eqn.(7.10) and Eqn.(7.12), we get the Lagrange’s Equations of motion,
� �
d ∂T ∂T
− = Qm (7.15)
dt ∂ q̇m ∂qm

(α)
Note here that the F�i appearing in the above equations are just the applied
forces(that is, the constraint forces are not involved in this equation). Thus,

T
even for systems subjected to constraints, we need NOT know the forces of
constraints. This is an advantage over the Newton’s equations.
If the applied forces are derivable from a potential, then the generalized
force can also be derived from the same potential(which is now expressed in
terms of generalized co-ordinates), that is, if
AF Fiα = −

then Qm =

∂Xi
(α)

Fi
,
∂V

α ∂Xi
(α)

∂qm
(7.16)

(7.17)
α,i
� ∂V ∂X (α) ∂V (qn , ...qn )
i
= − (α) ∂q
=− (7.18)
α,i ∂Xi m ∂qm

Using this in Eqn(7.15), we get,


d ∂T ∂T ∂V
DR

− + =0 (7.19)
dt ∂ q̇m ∂qm ∂qm
∂V
since =0 (7.20)
∂ q̇m
� �
d ∂ ∂
(T − V ) − (T − V ) = 0. (7.21)
dt ∂ q̇m ∂qm
Defining the Lagrangian
L=T −V (7.22)
we can re-express above Lagrangian equations of motion as
� �
d ∂L ∂L
− =0 (7.23)
dt ∂ q̇m ∂qm

Note that KE term can depend on generalized co-ordinates as well as on


generalized velocities.
E. Harikumar 33

Thus, it is enough to know L to find out the equations of motion. In


(α) ∂V
the above, we have assumed that the force can be written as Fi = − ∂X α.
i
But this assumption is not necessary. It is enough if we can express the
generalized force as
� �
d ∂M ∂M ∂V
Qm = − − (7.24)
dt ∂ q̇m ∂qm ∂qm

where M (q̇m , qm ). Then the Lagrangian is defined as L = T − V − M where

T
M is the generalized potential. Note that M depends on the generalised
velocities also. Thus the generalised force Q will also have dependence on
generalised velocities.

1. The Lagrange equation are (3N − k), second order differential equa-
tions.
AF
2. Thus, their solutions will involve 2(3N − k) arbitrary constants. These
constants are fixed using boundary conditions.

8 Constants of Motion
EX. 1. E = ∂∂L q̇ − L is a constant of motion. That is to say Ė = 0.
q̇m m
To check this, we note
� �
dE d ∂L ∂L dL
= q̇m + q̈m − (8.1)
dt dt ∂ q̇m ∂ q̇m dt
DR

Using
dL ∂L ∂L ∂L
= q̈m + q̇m + (8.2)
dt ∂ q̇m ∂qm ∂t
in the above gives
� � � �
dE d ∂L ∂L ∂L
= − q̇m − (8.3)
dt dt ∂ q̇m ∂qm ∂t
∂L
= − (8.4)
∂t
where we have used the Lagrange’s equation in obtaining the last line in the
above.
Thus, for the system satisfying ∂L∂t
= 0, dE
dt
= 0 , and E is a constant.
We have L = T − V , and L contains, at the most quadratic terms in gen-
eralized velocities and these terms are homogeneous functions of generalized
E. Harikumar 34

velocities. Then writing,

L = L(2) + L(1) + L(0) (8.5)


∂L
q̇m = 2L(2) + L(1) (8.6)
∂ q̇m
Therefore E = L(2) − L(0) (8.7)

L(2) which is quadratic in velocities is nothing but KE = T . L(0) which de-


pends only on generalized co- ordinates is just the (−ve)potential. Therefore

T
we get E = T + V -the total mechanical energy of the system. Thus we see
that the total energy of the system is conserved.
∂L
Consider a system for which ∂q m
= 0. Then we get
� �
d ∂L ∂L
=0⇒ = constant (8.8)
AF dt ∂ q̇m

∂L
∂ q̇m
∂ q̇m

= Pm = constant

Pm is called conjugate momentum corresponding to qm . If qm do not appear


∂L
in L, ( then ∂q m
= 0) we call “qm ” as ignorable or cyclic coordinate. The
conjugate momentum corresponding to this ignorable or cyclic coordinate
is always conserved. This can be illustrated by the example of the two
dimensional simple harmonic oscillator. We have
1 m
T = m(Ẋ 2 + Ẏ 2 ) = (ṙ2 + r2 θ̇2 ) (8.9)
DR

2 2
2
1 kr
V = k(X 2 + Y 2 ) = (8.10)
2 2
2
m 2 kr
L=T −V = (ṙ + r2 θ̇2 ) − (8.11)
2 2
Note “θ” do not appear in L and Lagrangian equations sets
d ∂L
= 0. (8.12)
dt ∂ θ̇
Using this and Eqn.(8.11), we get

mr2 θ̇ = Pθ = constant (8.13)

i.e., Pθ − the angular momentum with respect to origin is conserved. Notice


that we have found that Pθ is conserved with out using Newton’s equations
E. Harikumar 35

and just by the inspection of the Lagrangian. Such conservations ( obtained


by inspecting the Lagrangian) are now true even for non-central force.
Why we have cyclic coordinates? Let us consider a system of N particles
(see Fig). Let us chose some origin and axis. Let θ be the angular position of
a particular particle with respect to this axis. Let us specify all other particles
in the system with respect to this particle. If the angular displacement of
this chosen particle is δθ, the entire system also get displaced about the same
axis by δθ. Since L describes the dynamics, θ̇ will be present in L. But since
θ is chosen arbitrarily, “θ” by itself will not be present in L. Thus ∂L = Pθ =

T
∂ θ̇
constant for this system. (Pθ is the total angular momentum).

AF
9 Charged Particle in Electromagnetic Field
DR

For the systems where the force is not derivable from a potential, we have
stated that if we can still express the generalized force as (see Eqn.(7.24)
� �
d ∂M ∂M ∂V
Qm = − − , M (q, q̇, t),
dt ∂ q̇m ∂qm ∂qm

and then the Lagrangian can be defined as L = T − V − M . An ex-


ample for such a system is the motion of a charged particle in an external
electromagnetic field. Next we will discuss this problem.
Magnetic and electric fields can be expressed in terms of vector potential

A and scalar potential φ as

� =∇
B � × A,
� E � − 1 ∂A
� = −∇φ (9.1)
c ∂t
and the force experienced by an electrically charged particle moving in an
E. Harikumar 36

external electromagnetic field is given by the Lorentz force equation,



� + V × B).
F� = e(E � (9.2)
c
We re-express the above equation as

d � �
� − 1 ∂ A + 1 (V� × ∇
mV = e[−∇φ � × A].
� (9.3)
dt c ∂t c

T
Using
�×B
(A � × C)
� = B(
� A� · C)
� − C(
� A� · B),
� (9.4)
which in the component form is
�×B
(A � × C)
� i = �ijk Aj �klm Bl Cm = (δil δjm − δim δjl )Aj Bl Cm , (9.5)

d �
mV
AF
we re-write Eqn.(9.3)

� −
= e −∇φ
1 ∂ A� 1
+ ∇(� V� · A) 1
� − (V� · ∇)
� A�

dt c ∂t c c
� �
1 ∂ �
A �

� − (
= e −∇φ + (V� · ∇)
� A)
� + (V� · A)

c ∂t c
� � � �
( � · A)
V � 1 d �
A
� φ−
= e −∇ − (9.6)
c c dt
DR

Re-arranging the terms, we get


� �
d � � e �� � φ−
� ·A
V �
mV + A = −e∇ (9.7)
dt c c

This equation is in the form of


� �
d ∂L ∂L
= . (9.8)
dt ∂ q̇m ∂qm
∂L V ·A
Equating ∂Vi
= mVi + ec Ai and ∂L
∂Xi
= −e∇i (φ − c
), we get

1 V ·A
L = mV 2 − eφ + e . (9.9)
2 c
E. Harikumar 37

10 Hamilton’s Principle
10.1 Introduction
Consider a function f (qi ), i = 1, .......N, which is continuous, and its first
and second partial derivatives exists. Under a change in the coordinates
qn → qn� = qn + δqn , first variation of f is
N �
� ��
∂f ��

T
δf = δqn , (10.1)
n=1
∂qn � 0

where δqn is the variation of qn . If δf = 0, f is said to have a stationary


value at q. If each δqn are independent (and reversible), then

∂f ��
AF δf = 0 ⇒
∂qn �0
= 0, n = 1, 2, ......N.

(For static equilibrium, in Newtonian frame work we had F� = 0 = −∇U


Above equation is in the same footing).
� .
(10.2)

Second variation of f is
�N � 2
���
1 ∂ f �
δ2f = � δqn δqm , (10.3)
2 n,m ∂qn ∂qm �
0

which can be written as


1�
DR

δ2f = kmn δqn δqm (10.4)


2 m,n

where Kmn is a N × M matrix.


If each δqn are independent (and reversible), then q is a local mini-
mum/maximum if k is ±. If k is indefinite, q is a saddle point.
We assume δqn to be reversible to exclude the possibility of q being a point
on the boundary of its allowed range. It can be possible to have situations
when f has extreme value at a point in boundary even when δf �= 0.

10.2 Derivation of Equations of Motion using Hamil-


ton’s Principle
We have derived Lagrange’s equation by starting with the definition of
work done in small displacements of system, that is, using a differential
approach. In this approach we found second order differential equation for qn .
E. Harikumar 38

We need to specify initial values of each qn and q̇n to solve the equation. We
can equivalently specify the motion of the system completely by specifying
the coordinates qn , at two different times, say

• Point 1: q(t1 ) = {q1 (t1 ), q2 (t1 ).........qn (t1 )} at t1 and

• Point 2: q(t2 ) = {q1 (t2 ), q2 (t2 )......qn (t2 )} at t2

Now as the system develops in time, it moves from q(t1 ) to q(t2 ). Of all

T
possible paths between these two points, system takes only one path. How
this path is chosen?.

AF
DR

Along each of the allowed paths(see FIG), one can find qn and q̇n and thus
write down corresponding Lagrangian. Since the time dependence of qs are
different in different paths, L will also have different time dependence. We
now define “action functional” as
� t2
S(C) = L(q, q̇, t)dt (10.5)
t1

over the path C . Note: The LHS is just a number depending only on the
path. S(C) will be different for different paths.

Hamilton’s principle says that the actual path fol-


lowed by the system in going from q(t1 ) to q(t2 ) is the
one for which S(C) is an extream.
E. Harikumar 39

10.3 Finding the actual path using the Hamilton’s Prin-


ciple
Consider a family of paths between q(t1 ) and q(t2 ), parameterized by
a parameter “�”. Let � = 0 gives the actual path taken by the system.
Each path in this family is a function of “t” and is labelled by “�”, i.e., the
coordinate is specified as q(t, �) for fixed �.
For example: q(t, �) = q(t, 0) + �η(α) represents the family of paths. As �
changes, we go from one path to another in the family. We also assume that

T
q(t, �) is differentiable with respect to � for fixed “t”. Since the end points of
paths are fixed, that is,

q(t1 , �) = q(t1 , 0) = q(t1 ) (10.6)


q(t2 , �) = q(t2 , 0) = q(t2 ). (10.7)
AF
At the end points, we also have,

∂q(t1 , �)
∂�
=0=
∂q(t2 , �)
∂�
(∀qα , α = 1, 2, .....N ). (10.8)

Action functional has different value for different path, that is,
� t2
S(�) = dtL(q(t, �), q̇(t, �), t). (10.9)
t1

Action principle imply that the variation of S with respect to � is zero for
the actual path taken by the system, i.e.,
DR

� � t2 �
dS �� d �
� =0
= dtL (10.10)
d� ��=0 d� = 0 t1 �

(See eqn.(10.2.) Since � t2


dS ∂L
= dt (10.11)
d� t1 ∂�
we get, � �
� t2 � ∂L ∂qα � ∂L ∂ q̇α
dt + = 0. (10.12)
t1 α
∂q α ∂� α
∂ q̇α ∂�

(Our notations are as follows. For F (q, q̇, t), ∂F


∂t
is the time derivative with
q and q̇ held fixed (we allow � to vary) and Ḟ = dF dt
is the same with � held
fixed.)
E. Harikumar 40

� �
In the above q̇ = dq
dt
(t, �). Therefore ∂ q̇
∂�
= d ∂q
dt ∂�
. Using this, the second
term of Eqn.(10.12) is rewritten as
� � � � � �
∂L d ∂qα d ∂L ∂qα d ∂L ∂qα
= − (10.13)
∂ q̇α dt ∂� dt ∂ q̇α ∂� dt ∂ q̇α ∂�
Using this, we find that Eqn.(10.12) becomes
� t 2 �� � �� �t 2
∂L d ∂L ∂qα � ∂L ∂qα ��
dt − + � =0 (10.14)

T
t1 α
∂qα dt ∂ q̇α ∂� α
∂ q̇α ∂� �
t1

But note that we have assumed ∂q(t∂�1 ,�) = 0 = ∂q(t∂�


2 ,�),�
and hence the last term
vanishes.Thus we get,
∂L d ∂L
− = 0, (10.15)
∂qα dt ∂ q̇α
AF
since ∂q∂�α are arbitrary functions of time.
Thus we see that the least action principle gives the equations of mo-
tion(EOM). This equations are known as Euler-Lagrange equations of motion(E-
L eqn).

11 Symmetries and Conservation Laws


Action principle also allows us to study symmetries and their relation to
conservation laws. Conservation laws allow us to integrate EOM , converting
a second order differential equation to a first order equation.
DR

In many cases, we would know the conservation laws of nature from ex-
periments and observations. For example, in particle physics, many a times
we would not know the interactions/forces involved, but experimentally ob-
served conservation laws will give us information about the symmetries of the
interaction. This will allow us to model the interaction through the action.

11.1 Symmetries
Earlier we have considered a generic transformation of the coordinates,
q(t, �) = q(t, 0) + �η(α). Here, as � changes q(t) changes to, say q � (t). Thus
∂q
∂�
= η(α) = q � − q = δq. Consider another infinitesimal transformation,

qk → qk� = qk + δ̄qk = qk + λφk , λ << (11.1)

In the above, δ̄qk denotes a specific transformation (in contrast to the arbi-
trary transformation we considered in deriving E-L Eqn.).
E. Harikumar 41

If L(qk + λqk , q̇k + λq̇k , t) − L(qk , q̇k , t) = O(λ2 ), then we say


that the system is invariant under the above infinites-
imal transformation.
When we apply an infinitesimal transformation to a Lagrangian- L, de-
fined over a specific path “C” in the configuration space, we get the action
for a neighboring path (infinitesimally close to ‘C’). Then, we get

δS[C] = dt (L(qk� , q̇k� , t) − L(qk , q̇k , t) (11.2)

T
� � � ∂L ∂L ∂L

= dt δqk + δ q̇k + (11.3)
k
∂q k ∂ q̇ k ∂t
� � ∂L� � ��
d ∂L
= λ dt − φk (t)
k
∂qk dt ∂ q̇k
� �t 2

( ∂L
∂t
AF + λ Pk φk (t)�

t1

where we have assumed that the Lagrangian has no explicit time dependence
= 0). If ‘C’ is a classical trajectory, then the L do satisfy E-L Eqns. and
(11.4)

thus the first term will be zero. But if this transformation is to be a symmetry,
δS[C] = 0(since only O(λ2 ) and higher terms can be non-vanishing). This
condition , δS[C] = 0, imply
� �
Pk φk (t1 ) = Pk φk (t2 ) (11.5)
k k
that is F (t1 ) = F (t2 ) (11.6)
DR

Thus we see that F is independent of time! That is dF dt


= 0, and F is
conserved.
Many a times, from the form of the Lagrangian, it is easy to guess which
geometrical transformation is a symmetry. Such finite transformations can
be build up from infinitesimal transformations and these symmetries will
imply conservation laws.
If L� − L = λ dΩ
dt
, then we say that L is quasi invariant.
In this case, we get
� � � ∂L d

∂L
��
λΩ(t)|tt21 = λ dt − φk (11.7)
k
∂qk dt ∂ q̇k

+ λ Pk φk (t)|tt21
k

⇒ Pk φk (t) − Ω(t) = G(t) (11.8)
k
E. Harikumar 42

is a conserved quantity.

11.2 Noether Theorem


Noether Theorem states that corresponding to every
continuous symmetry of the system, there exist a con-
served quantity.

Consider a system of N particles interacting through a central potential

T
V (|ri − rj |). Find symmetries and corresponding conservation laws.
Let us consider a system described the Lagrangian
N N
1� 2

L(X, Ẋ) = mα Ẋα − V (|Xα − Xβ |) (11.9)
2 α=1 α<β=1

11.2.1
AF
and see what are the symmetries of this system.

Translation
�α → X
Consider X �� = X
� α + �a. Under this transformation, we have
α

|Xα − Xβ | → |Xα� − Xβ� | = |Xα − Xβ | (11.10)


d�a
Ẋα� = Ẋα since =0 (11.11)
dt
Therefore, we see that the L(X � , Ẋ � ) = L(X, Ẋ), that is L is invariant. This
DR

symmetry leads to the conserved quantity,


� ∂L � � · �a
δqα = Mα Ẋ α (11.12)
α
∂ q̇ α
�� �
˙
= �a · �
Mα Xα = �a · (P� ) (11.13)

Thus we see
d dP� dP�
(�a · P ) = �a · =0⇒ =0 (11.14)
dt dt dt
where we have used the fact that �a is arbitrary. Thus, we see that transla-
tional invariance leads to the conservation of linear momentum P� .
E. Harikumar 43

11.2.2 Rotations
Consider rotations in two dimension(through a fixed angle θ, i.e., θ̇ = 0)
� (α) (α) (α)
X1 = X1 cos θ + X2 sin θ (11.15)
� (α) (α) (α)
X2 = −X1 sin θ + X2 cos θ (11.16)
� (α) (α)
X3 = X3 (11.17)

Notice

T
� (α) � (α) � (α) � (α) � (α) � (α) (α) (α) (α) (α) (α) (α)
X1 ·X1 +X2 · X2 +X3 ·X3 = X1 ·X1 +X2 ·X2 +X3 ·X3
(11.18)
which can be expressed as
� � (α) � (α) � (α) (α)

We also have
AF
Xi · Xi =
i


Xi · Xi .

Ẋi
� (α)
· Ẋi
� (α)
=
i

� (α)
Ẋi Ẋi .
(α)
(11.19)

(11.20)
i i

Similarly, we have
� �
�α − X
|X � β |2 = | X
�α − X
� β |2 (11.21)
� (α) (α)
also Ẋ3 = Ẋ3 . Using these, we find
� Mα ˙
L= X� 2 + V (|X
�α − X
� β |) = L(X � , Ẋ � ) (11.22)
2 α α α
DR

For infinitesimal rotations θ <<, Eqns. (11.15,11.16)reduces to


� (α) (α) (α)
X1 = X1 + X2 θ (11.23)
� (α) (α) (α)
X2 = −X1 θ + X2 (11.24)

Therefore
(α) (α) (α) (α)
δX1 = θX2 δX2 = −θX1 (11.25)
which can be written as
(α)
δXi ≡ θ�ij Xj (11.26)
(α)
δX3 = 0 (11.27)

In general, we have
� = −θ(k̂ × X).
δX � (11.28)
E. Harikumar 44

The conserved quantity corresponding to this symmetry is


� ∂L � � �
(α) (α) (α) (α) (α)
(α)
δX i = θ M α Ẋ 1 X 2 − Ẋ 2 X 1 (11.29)
α,i ∂ Ẋi α
� � (α) (α) (α) (α)

= θ P1 X 2 − P 2 X 1 (11.30)
α

= θ �ij Pi Xj = θ(P� × X)
� 3 (11.31)
α

T
�3
= −θL (11.32)

That is L� 3 = � (X � α × P α )3 is conserved. That is, rotational invariance


α
leads to the conservation of angular momentum. In general, a vector A � under
rotation about an arbitrary unit vector n̂, through an angle “δθ” changes by

11.2.3
AF
Time translations
� = −(δθ)n̂ × A.
δA

Consider the time translation t → t� = t + a. Then


d
= dt� d
=
(11.33)

d
Let
dt dt dt� dt�
us check whether the quantity
� �
dE d � ∂L
= q̇i − L (11.34)
dt dt i
∂ q̇i
� ∂L �
is conserved or not. Since dL dt
= ∂L
∂t
+ i ∂q i
q̇i + i ∂∂L q̈ we get
q̇i i
� � � �
DR

d � ∂L d � ∂L � ∂L � ∂L ∂L
q̇i − L = q̇i − q̈i − q̇i − = 0.
dt i
∂ q̇i dt i
∂ q̇i i
∂ q̇i i
∂qi ∂t
(11.35)
Using
� �
� ∂L d � ∂L � d � ∂L �
q̈i = q̇i − q̇i (11.36)
i
∂ q̇ i dt i
∂ q̇ i i
dt ∂ q̇ i

in the above, we get


� ∂L � d � ∂L � ∂L
− q̇i + q̇i = + . (11.37)
i
∂qi i
dt ∂ q̇i ∂t

Thus we write � �
dE d � ∂L ∂L
≡ q̇i − L =+ (11.38)
dt dt i
∂ q̇i ∂t
E. Harikumar 45

It is easy to see that the Lagrangian “L” is invariant under time translations
if L do not have any explicit time-dependence, that is if ∂L
∂t
= 0. But if this
is so, from the above, it is clear that E is a constant of motion. Thus we
see that the invariance under time transitions leads to the conservation of
Energy.

11.3 Galilean Invariance


Consider two frames S and S � moving with respect to each other with

T
constant velocity V0 . Since both are inertial frames, Newtons laws must be
true in both of them, that is

d2�rα� d2�rα
Mα = F α = M α (11.39)
dt2 dt2
implying

Thus we get
AF d2
dt2
(�rα − �rα� ) = 0. (11.40)

�rα − �rα� = V�0 t. (11.41)


Thus �rα� = �rα − V�0 t. Since Vα2 = |Vα − V0 |2 , we find,

M �2 M 2 M 2
V = V + V0 − M V�α · V�0 . (11.42)
2 α 2 α 2
Thus, we see
DR

�1 �2
� � �


L = M Vα − U |�rα − �rβ | (11.43)
α
2
�1 M 2
= V0 − M V�α · V�0 − U
M Vα2 + (11.44)
α
2 2
� �
d 1
= L+ M V0 t − M V�0 · �rα
2
(11.45)
dt 2

Hence
� �
d M 2 �
δL = V t − M V0 · �rα (11.46)
dt 2 0
d
= Ω (11.47)
dt
E. Harikumar 46

showing that the system is quasi-invariant under Galilean transformations.


Neglecting V02 term (see V0 plays the role of λ), we find the conserved quantity
to be
∂L � ∂L �
δq − Ω = δX α
i + � α · V�0
Mα X (11.48)
∂ q̇ i,α
∂ Ẋ α
i α
� ∂L � �
α�
α
δX α
i − Ω = − M α Ẋ i V 0 t + Mα Xiα V�0 (11.49)
i,α
∂ Ẋ i i,α α
� �

T
� �
= −V�0 Mα Ẋiα t − Mα Xiα (11.50)
α,i α,i
� �
= −V�0 P� t − M R
� (11.51)

That is (M R � − P� t) is a conserved quantity. Here P� is the total linear mo-


mentum, and R
AF
� is position vector of CM. Thus for any closed system, we
have ten conserved quantities, namely, E, Pi , Li , and (M Ri − Pi t).
Let us assume now that V0 is not a constant, that is, dtd V0 �= 0. Then
L� = M2 V � · V � − U will lead to E-LEqns.
� �
d ∂L� ∂L�
− = 0. (11.52)
dt ∂V � ∂r�
˙
That is, M V� � + ∇U
� = 0, that is,M (�v̇ − �v̇ 0 ) + ∇U
� = 0 which is different from
the equation obtained from L. The extra term, −M v̇0 thus violate Newtons
second law as S � is a non-inertial frame.
DR

12 Calculus of Variations
Ex.1. Find the shortest distance between two points in a plane.
We have ds2 = dx2 + dy 2

� dy
ds = dx2 + dy 2 = 1 + ( )2 dx (12.1)
� dx
= 1 + y �2 dx = f dx. (12.2)
Starting from � x2 �
I= 1 + y �2 dx, (12.3)
x1
using � �
∂f d ∂f
− = 0,
∂y dx ∂y �
E. Harikumar 47


where f = 1 + y �2 , we get
y� 1
� =√ (12.4)
1+y 2 a+1
dy
where a is a constant. Thus, we find y �2 = a−1 , and hence dx = m. Therefore,
y = mx + c, which is the equation for a straight line.
Curves that are the shortest distance path between any two points on a
surface is called Geodesic of that surface.

T
1. HW. Find the geodesic for ds2 = dr2 + r2 dθ2

Example.2
Consider a curve passing between (x1 , y1 ) and (x2 , y2 ). When this curve is
revolved about Z-axis, it traces out a surface. Find the curve between these
two points such that the area of surface of revolution is minimum.
AF �
da = 2πxds = 2πx 1 + y �2 dx = f dx
� �
I = 2π x 1 + y �2 dx
(12.5)
(12.6)
� �
∂f d ∂f
= 0, =0 (12.7)
∂y dx ∂y �
Thus, we get
x x
y = a cos h−1 ( ) + b = a arc cos h( ) + b (12.8)
a a
DR

Example.3
Find the curve between two points along which a particle falling from rest
under the action of gravity takes minimum time.
� 2 �
ds
, ds = 1 + y �2 dx (12.9)
1 v

How to find “t”-the time e of flight? we have 12 mv 2 = mgy ⇒ v = 2gy.
We have
� 2�
1 + y �2
I= dx, (12.10)
1 2gy
and
� now proceed as earlier and find solution. We can also write ds =
1 + ( dx
dy
)2 dy. Then
� 2�
1 + x�2
I= dy
1 2gy
E. Harikumar 48

EL equation following from this action is


� �
∂ ∂f
= 0. (12.11)
∂y ∂x�
∂f ∂f x� √1
Since ∂x
= 0, ∂x�
=√ = 2a
=constant.
y(1+x�2 )

dx� y
=
dy 2a − y
�y �

T
y
Therefore x = 0
dy 2a−y
. Solving this, give

x = a(1 − cos θ),


y = a(θ − sin θ)
Example 4

� 2 � 2√ 2
AF
Consider a particle moving in the spherically symmetric potential V (r).
Find the path of shortest time of flight?
Starting with the action
� 2 � �2 � 2
ds dr + r2 dθ2 (r + r2 )
I= = � = dθ = f (r, r� , θ)dθ
(12.12)
1 v 1 2(E − V ) 1 2(E − V ) 1
� ∂f �
The E-L Equation- ∂f ∂r
− ∂
∂v ∂r�
= 0 becomes
� �
�� 2 v� v�
r − + r�2 − r − r2 = 0. (12.13)
r 2, (E − V ) 2(E − V )
DR

This is a second order differential equation. But there is an easy way to find
r(θ). Note that f do not have any explicit dependence on θ5 . Therefore
∂f �
r − f = Const (12.14)
∂r�
∂f 1 1 r� √ r

Using ∂r�
= (r�2 + r2 )− 2 (2(E − V )) 2 , and √ = �2
r +r 2
in the above
2(E−V )
dr
and setting dθ
= 0 at r = r0 , we get
kr4
r�2 = − r2
2(E − V )
with k = 2(E−V
r02
0)
=constant. By taking explicit form of v(r), we can solve
this equation. � �
r2
HW. Given V = 2R −3 + R2 , r ≤ R, E = − Gm
Gm
R
. Find r(θ)
5 ∂L
∂ q̇ q̇ − L = E and E is conserved. Also note that in the above θ plays the role of t.

S-ar putea să vă placă și