Beruflich Dokumente
Kultur Dokumente
Contents
1 Goldstein 8.3.
1.1 Form of G. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2 Test Case: Simple Harmonic Oscillator. . . . . . . . . . . . . . . . . . . . . . . . . .
2
2
3
4
4
4
5
5
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4 Goldstein 8.26.
4.1 Lagrangian and Hamiltonian. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Change of Coordinates. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
8
8
5 Goldstein 8.19.
10
Dylan J. Temples
Goldstein 8.3.
1.1
Form of G.
(1)
L
L
L
dqi +
dqi +
dt .
qi
qi
t
(2)
In order to properly transform q p, and get the Hamiltonian, H(q, p; t), there must be no dqi
dependence in the total differential. Therefore the variable being transformed from is multiplied
by the variable being transformed to, and the function that is being transformed is subtracted,
H = qi pi L .
(3)
H
H
L
L
L
H
dqi +
dpi +
dt = pi dqi + qi dpi
dqi
dqi
dt ,
qi
pi
t
qi
qi
t
(4)
after plugging in the Lagrangian. As stated before, the terms with dqi dependence must cancel, so
the following relationship for the transformation is obtained by equating the coefficients of these
terms,
L
pi =
.
(5)
qi
Hamiltons equations can be found by equating the coefficients of similar total differential terms:
qi =
H
pi
H
L
=
S
qi
qi
H
L
=
.
t
t
(6)
(7)
so S = pi and the correct Hamilton equation is obtained (as expected). Now consider a second Legendre transformation, qi S = pi . Following the procedure above, the terms with dqi dependence
must cancel in the total differential of the new double-transformed function,
L
L
G = qi pi H dG = qi dpi + pi dqi qi dpi
dqi
dt ,
(8)
qi
t
so the plus form must be correct (to get the terms to cancel), so the new function is
G = qi pi + H
dG = qi dpi + qi dpi
L
H
dt = qi dpi + qi dpi +
dt .
t
t
(9)
Page 2 of 11
Dylan J. Temples
To get the equivalent of Hamiltons equations for this double-Legendre transformed function, consider the definition of the total differential of G,
dG(pi , pi ; t) =
G
G
G
dpi +
dpi +
dt ,
pi
pi
t
(10)
and equate the coefficients of like terms from the previous expression for the total differential of G.
The resulting equations are
qi =
1.2
G
pi
qi =
G
pi
G
H
L
=
=
.
t
t
t
(11)
Consider the simple case of a one dimensional horizontal harmonic oscillator consisting of a mass
m attached to a spring of constant k. This can be solved with the function G as defined above. The
Lagrangian for the simple one dimensional oscillator is trivial to write down, but using Equations 7
and 5, it can be shown that
p = mq
p = m 2 q ,
(12)
p
where = k/m is the frequency of oscillation. Therefore,
p2
p2
m 2
G=
+
+
m 2 2m
2
2
p
p2
p2
.
m 2
2m 2m 2
(13)
q=
p
,
m 2
(14)
so by differentiating the left most equation, solving the right most equation for p,
and plugging
that into the expression for q, the equation
q = 2 q ,
(15)
is obtained, which is easily recognizable as the equation for a simple harmonic oscillator in one
dimension. Note that the potential term in G does not explicitly depend on velocity, so the Hamiltonian is the total energy. Additionally because the partial derivative with respect to time of L, H,
and G are all zero, the Hamiltonian is conserved, and therefore total energy is a constant of the
motion.
Page 3 of 11
Dylan J. Temples
Consider a full three dimensional central potential UE (r), using standard spherical polar coordinates {r, , } as the generalized coordinates. In terms of the spherical coordinates, the Cartesian
coordinates are
x = r sin cos x = r sin cos + r cos cos r sin sin
y = r sin sin y = r sin sin + r cos sin + r sin cos
sin .
z = r cos z = r cos r
(16)
(17)
(18)
2.1
(19)
The Hamiltonian.
(20)
r = Pr /m
(21)
= P /mr2
(22)
= P /mr2 sin2 .
(23)
P2
mr2
P2
mr2 sin2
(24)
Pr2
P2
+
+
m
2m 2mr2
P2
P2
P2
= r +
+
+ UE = Etot ,
2m 2mr2 2mr2 sin2
=
P2
2mr2 sin2
+ UE
(25)
(26)
2.2
Constants of Motion.
From this Hamiltonian, it is easy to determine some constants of motion. Noting that the Hamiltonian is not explicitly dependent on time means it is conserved, and therefore the total energy is
as well. The conjugate momentum to is also conserved because from Hamiltons equations,
H
P =
=0
P = cst.
(27)
Page 4 of 11
Dylan J. Temples
2.3
Hamiltons Equations.
H
= Pr /m
r =
Pr
H
P
=
=
P
mr2
P
H
=
=
2
P
mr sin2
+
=
P2
mr3 sin2
UE
r
P2 cos
mr2 sin3
m
r = Pr
d
= P
[mr2 ]
dt
d
= P ,
[mr2 sin2 ]
dt
(28)
(29)
(30)
(31)
(32)
after differentiating the second set of equations with respect to time. Now the first three equations
can be equated to the second three using the expressions on the right hand side of the equations
above. Taking the indicated time derivatives and setting them equal to the corresponding momentum time derivative yields
2
P
P2
UE
+
(33)
(34)
(35)
From here the definitions of the conjugate momenta (Equations 72 - 23) can be substituted in, and
the expressions solved for the second time derivative terms,
UE
r
2
2
2
(36)
(37)
(38)
2.4
These equations of motion can now be compared to the equations of motion found by using the
Lagrangian method. The Euler-Lagrange equations are
d L
L
=
.
dt q
q
(39)
Page 5 of 11
Dylan J. Temples
Applying this to the Lagrangian given in Equation 20 gives the three Lagrangian equations of
motion,
d
Ue
[mr]
= mr2 + mr sin2 2
dt
r
d
2
2
[mr ] = mr sin cos 2
dt
d
=0
[mr2 sin2 ]
dt
Ue
m
r = mr2 + mr sin2 2
r
2
2
2
mr = mr sin cos 2mrr
(40)
(41)
(42)
which are exactly the same equations of motion found through the Hamiltonian approach.
Page 6 of 11
Dylan J. Temples
Consider a particle of mass m subject to a central force F = kr, where r is a vector from the origin
to the particle. The particle is constrained to move on a cylinder centered on the z-axis, defined
by x2 + y 2 = R2 . The transformation from cylindrical coordinates on this surface to Cartesian
coordinates is given by
(43)
x = R sin
y = R cos
z = z ,
(45)
x = R cos
y = R sin
z=z
(44)
(46)
1
2
2 mv .
(47)
P =
(48)
= P /mR2
(49)
z = Pz /m.
(50)
P
mR2
2
1
m
2
Pz
m
2
1
1
+ kR2 + kz 2
2
2
(51)
(52)
H
H
P =
=0
Pz =
= kz
(53)
z
H
P
H
Pz
=
=
z =
=
.
(54)
2
P
mR
Pz
m
From this, it is easy to see P is a constant of the motion. By differentiating the expression for z
with respect to time and substituting in the expression for Pz , the Hamiltonian equations of motion
are
k
z = z
(55)
m
` = mR2 ,
(56)
where ` is the angular momentum. So the motion is a harmonic oscillator in the z direction
(about the origin) while the particle moves around the z-axis in a circle, with conserved angular
momentum.
Page 7 of 11
Dylan J. Temples
Goldstein 8.26.
4.1
When the mass is a distance x from the left wall, the left spring is compressed/extended to a length
x while the right spring is compressed/expanded to a length a x, therefore the potential energy
of the system, using the generalized coordinate q, is
1
1
1
1
U = k1 q 2 + k2 (a q)2 = (k1 + k2 )q 2 k2 aq + k2 a2 .
(57)
2
2
2
2
Note the last term is a constant, so the zero point of potential enery can be set so this term is zero
(which will be done). The kinetic energy is the usual T = 12 mq2 . Therefore the Lagrangian and,
using the conjugate momentum to q: p = mq,
the Hamiltonian are given by
1
1
L = mq2 (k1 + k2 )q 2 + k2 aq
(58)
2
2
p2
1
H=
+ (k1 + k2 )q 2 k2 aq .
(59)
2m 2
Note that there is no explicit time dependence in the Hamiltonian or Lagrangian, so the Hamiltonian
is conserved. Additionally, the Hamilontian is easily recognizable as the total energy of the system,
which is also conserved.
4.2
Change of Coordinates.
k2 a
.
(60)
k1 + k2
For the Lagrangian to be written in this coordinate, it can be manipulated starting with Equation 58,
!
1 2 1
k
a
1
1
2
q = mq2 (k1 + k2 ) q 2 + 2bq ,
(61)
L = mq (k1 + k2 ) q 2 + 1
2
2
2
2
2 (k1 + k2 )
Q = q b sin t with b =
noting that the last term in parenthesis can be written as (q b)2 b2 by completing the square.
Again, the zero point of potential energy will be set so the constant term does not contribute.
Therefore the Lagrangian, using the definition of Q can be written
1
1
L = m(Q + b cos t)2 (k1 + k2 )(Q + b sin t b)2
2
2
1 2 2
2
cos t + Q 2 1 (k1 + k2 )(Q + b sin t b)2 .
= m b cos t + 2bQ
2
2
(62)
(63)
Page 8 of 11
Dylan J. Temples
L
= mQ + mb cos t
Q
P
Q =
b cos t ,
m
(64)
L=
H = QP
(65)
(66)
Notice now the Hamiltonian has explicit time dependence so it is not conserved. In these coordinates, the total energy is given by
1
P2 1
1
Figure 2: Depiction of the pendulum system and coordinates used in problem #5.
Page 9 of 11
Dylan J. Temples
Goldstein 8.19.
The point of suspension of a simple pendulum of length ` and mass m is constrained to move
on a parabola defined by z = ax2 in the vertical plane, in a uniform downwards gravitational
acceleration g. The position of the pivot point in the x z plane is given by (x, ax2 ). Now, let
a coordinate be the angle the pendulum makes with the vertical. The position of the mass is
then (x + ` sin , ax2 ` cos ). So the generalized coordinates for this system are x and , and the
kinetic energy is
2
2
1
m
2
2
(68)
T = m(x + z ) =
x + ` cos + 2axx + ` sin
2
2
i
mh
=
(1 + 4a2 x2 )x 2 + 2`(cos + 2ax sin )x + `2 2 .
(69)
2
Let the zero potential be defined such that when x = = 0, U = 0. Therefore the zero potential
surface is the plane defined by z = `, so the potential energy is
U = mg(` + ax2 ` cos ) .
(70)
(72)
(73)
(74)
x =
(75)
(76)
px =
H
.
x
(79)
Page 10 of 11
Dylan J. Temples
(81)
and
px =
2a
1
g`2 m2 x(2ax cos sin )3 `p px (2ax sin(2) + cos(2) + 3)
`2 m(sin 2ax cos )3
2
2 2
2
+ p (2ax sin + cos ) + ` px cos , (82)
after simplifying,
2a
px = 2agmx + 2
` m(sin 2ax cos )3
1
`p px [2ax sin(2) + cos(2) + 3]
2
p2 (2ax sin
+ cos ) +
`2 p2x cos
, (83)
The equations of motion can then be found by differentiating Equations 75 and 76 and plugging in
the above results, then substituting in the expressions for p and px .
Page 11 of 11