FYST420 Advanced electrodynamics Final project

Olli Aleksanteri Koskivaara [email protected]

The Poisson bracket and magnetic

monopoles

Abstract:
In this work magnetic monopoles are studied using the Poisson bracket. Using
the Hamiltonian formulation of classical mechanics a system can be described
by it’s Hamiltonian. The Hamiltonian of a particle in an electromagnetic field is
introduced. The Poisson bracket and some of it’s properties are also outlined.
Assuming that a magnetic monopole exists, the behaviour of a particle under it’s
influence is studied using the machinery presented earlier. It is found that the
trajectory of the particle lies on a cone.
1 Introduction
Magnetic monopoles are a widely studied subject in physics, even though they
remain yet to be found in Nature. Every magnet we have observed so far has
both a north and a south pole. The interest in magnetic monopoles arises from
the fact that many modern theories in particle physics predict their existence.
In this work I study how a charged particle would behave under the influence
of a magnetic monopole, main goal being the trajectory of the particle. I start
with some basic Hamiltonian mechanics, which I assume the reader is familiar
with. I then move on to introduce the Poisson brackets, which will be my main
tool in deriving the particle’s trajectory.

2 Hamiltonian mechanics
2.1 Basics
In Lagrangian mechanics a system with n degrees of freedom is described by the
generalized position and velocity coordinates qi and q̇i , where i = 1, ..., n. The
Lagrangian L = L(qi , q̇i , t) is defined as

L = T − V, (1)

where T is the kinetic energy of the system and V is the potential energy of the
system. Once the Lagrangian of the system is known, the equations of motion
are directly obtained from the Euler–Lagrange equation
 
∂L d ∂L
− = 0, (2)
∂qi dt ∂q̇i
which follows from minimazing the action integral of the system.
The transition from Lagrangian mechanics to Hamiltonian mechanics is made
by performing a Legendre transformation. This gives rise to the Hamiltonian of
the system
n
H ( qi , pi , t ) = ∑ q˙j p j − L(qi , q̇i , t), (3)
j =1

where pi is the conjugate momentum defined by

∂L
pi = . (4)
∂q̇i
Now the equations of motion turn out to be the so-called Hamilton’s equations
∂H ∂H
q̇i = & ṗi = − . (5)
∂pi ∂qi

2
2.2 A charged particle in an electromagnetic field
Let’s take a look at a particle with charge q moving in an electromagnetic field.
The electric and magnetic fields E and B can be written in terms of a vector
potential A and a scalar potential φ as

1 ∂A
E = −∇φ − , B = ∇ × A. (6)
c ∂t
It turns out, that in this case the Lagrangian solving the Euler–Lagrange equation
(2) is of the form [1]

1 2 q
L= mṙ − qφ + ṙ · A, (7)
2 c
where r is the position vector of the particle and c is the speed of light. It is quite
straightforward to convince oneself that this is indeed the right Lagrangian by
substituting it into equation (2) which eventually yields the familiar Lorentz
force law.
We can now calculate the conjugate momentum pi using equations (4) and
(7):

∂L q
pi = = mṙi + Ai . (8)
∂ṙi c

We can immediately see that now the conjugate momentum is not just mass
times velocity; instead, there is an extra term given by the vector potential A.
Solving equation (8) for r˙i and substituting into equations (3) and (7) gives us
the following Hamiltonian:

1 3  q  1  q 2 q  q 
m i∑
H= p i − A i p i − p − A + qφ − p − A ·A
=1
c 2m c mc c
1 q 1  q 2 q q2 2
= p2 − p·A− p− A − p·A+ A + qφ
m mc 2m c mc mc2
1  q 2 1  q 2
= p− A − p − A + qφ
m c 2m c
1  q 2
= p − A + qφ. (9)
2m c
Substituting this into Hamilton’s equations one can further derive the equations
of motion for the particle, again corresponding to the Lorentz force law. I won’t
go through the calculations here, because I will only be needing the form of
the Hamiltonian. Instead, I move on to introduce my main tool for studying
magnetic monopoles.

3
2.3 The Poisson bracket
Let f (qi , pi ) and g(qi , pi ) be two functions of the coordinates and the conjugate
momenta on our phase space. The Poisson bracket is defined as
n  
∂ f ∂g ∂ f ∂g
{ f , g} = ∑ − . (10)
i =1
∂qi ∂pi ∂pi ∂qi
Some properties following straight from the definition are
i) {f, f} = 0
ii) { f , g} = −{ g, f }
iii) {α f + βg, h} = α{ f , h} + β{ g, h}, α, β ∈ R
iv) { f g, h} = f { g, h} + g{ f , h}
v) { f , { g, h}} + {h, { f , g}} + { g, {h, f }} = 0.
The three last ones are the familiar linearity, Leibniz rule and Jacobi identity. The
Poisson brackets seem to behave a lot like matrix commutators. People familiar
with differential geometry may also notice the analogy with the Lie bracket or
Lie derivative of two vector fields.
Maybe the most important application of the Poisson bracket is related to
the so-called canonical transformations. A transformation in a phase space is said
to be canonical if it leaves the Hamilton’s equations invariant. It turns out, that
the Poisson bracket can be used to check whether a coordinate transformation is
canonical or not. [2]
The Poisson bracket has also an important historical role in physics. When
developing quantum mechanics, people related Poisson brackets to commuta-
tors and classical quantities to operators in order to quantize the classical sys-
tem. Indeed, as noted above, the Poisson bracket has a clear resemblance to
commutators. [3]
The Poisson bracket has one property that will be especially useful when
studying magnetic monopoles. Let f = f (qi , pi , t) be a function such that qi and
pi are solutions to the Hamilton’s equations. Then using chain rule we have
n n n n
df ∂ f dqi ∂ f dpi ∂f ∂ f ∂H ∂ f ∂H ∂ f
=∑ +∑ + =∑ −∑ +
dt i =1
∂qi dt i =1
∂pi dt ∂t i =1
∂qi ∂pi i=1 ∂pi ∂qi ∂t
∂f
= { f , H} +,
∂t
where on the first line we used the Hamilton’s equations (3). If we further
df
assume that dt = 0, i.e., f is a constant of motion, then we obtain the so-called
Liouville equation
∂f
{ f , H} + = 0. (11)
∂t
Often it is also possible to assume that f = f (qi , pi ) does not depend explicitly
on time. Then f is a constant of motion if { f , H } = 0.

4
3 Magnetic monopoles
So far we haven’t really done any calculations with magnetic monopoles. Let’s
now assume that such a point-like magnetic charge would exist. It would be
described by a magnetic field singular at the origin
r
B=ξ , (12)
| r |3
where ξ is a constant determining the strength of the field [4]. We see immedi-
ately that the field is problematic. One of Maxwell’s equations tells us that the
divergence of the magnetic field should vanish, ∇ · B = 0, but in this case the
condition cannot be satisfied because of the infinities in the origin. Furthermore,
we do not have any vector potential A to play with.
However, we can put these problems aside and start to work with the Poisson
bracket. When calculating the Hamiltonian for a particle in an electromagnetic
field (3) we saw that

1  q 
ṙi = pi − Ai .
m c
Using this we can calculate the following Poisson bracket:
n q q o
{mṙi , mṙ j } = pi − Ai , p j − A j
c c
q q q2
= { p i , p j } − { p i , A j } − { A i , p j } + 2 { A i , A j }.
c c c
Straight from the definition of the Poisson bracket we see that the first term
vanishes. The last term vanishes too, since the vector potential A depends only
on the position vector r. For the two remaining terms we can use the fact that
for any function f (r) depending only on the position vector r we have
n  
∂pi ∂ f (r) ∂p ∂ f (r) ∂ f (r)
{ pi , f (r)} = ∑ ∂r a ∂p a
− i
∂p a ∂r a
=−
∂ri
a =1 | {z } |{z}
=0 = δia

Using this result our Poisson bracket becomes

 
q ∂A j ∂Ai q
{mṙi , mṙ j } = − = eijk Bk , (13)
c ∂ri ∂r j c

where in the last step we used the Maxwell’s equation B = ∇ × A and the
expression of the cross product in terms of the Levi-Civita symbol eijk .
We see that using the Poisson bracket we got rid of the vector potential A,
even though it appeared in our derivation. In a similiar fashion we can calculate

5
another Poisson bracket:
n q o q
{ri , mṙ j } = ri , p j − A j = {ri , p j } − {ri , A j }
c c
q 3
 
∂ri ∂A j ∂ri ∂A j
= −{ p j , ri } − ∑ −
c a=1 ∂r a ∂p a ∂p a ∂r a
|{z} |{z}
=0 =0
∂r
= i = δij . (14)
∂r j

Now we can use these Poisson brackets (13) and (14) to calculate the Poisson
bracket of the angular momentum L = r × mṙ with r and mṙ. After some index
milling one should end up with the following Poisson bracket structure:

{ Li , r j } = eijk rk (15)
q q
{ Li , mṙ j } = eijk mṙk + δij (r · B) − Bi r j . (16)
c c
It is easy to see that these indeed work by explicitely calculating the brackets for
a few components of the angular momentum L. If we now substitute the field
generated by the magnetic monopole (12) to the right hand side of equation (16)
we get
 
qξ r qξ ri
{ Li , mṙ j } = eijk mṙk + δij r· − rj
c | r |3 c | r |3
ri r j
 
qξ
= eijk mṙk + δij −
c|r| |r| |r|
 
qξ ri
= eijk mṙk + , mṙ j , (17)
c |r|

where one can confirm the last step by opening the Poisson bracket and using
the results and properties derived earlier.
From the previous chapter we remember that if the Poisson bracket of a
given function with the Hamiltonian vanishes, then the function is a constant
of motion. Using the Poisson bracket structure obtained above one can see that
the Poisson bracket of the angular momentum and the Hamiltonian does not
necessarily vanish in the case of the magnetic monopole. However, equation (17)
suggests that we should introduce a new quantity of the form

qξ r
J = L− . (18)
c |r|

This new quantity J is called the general or generalized angular momentum [4].
Consisting of the ordinary angular momentum and an additional part it looks
a lot like the total angular momentum familiar from quantum mechanics. One

6
also sees that if the strength parameter of the magnetic monopole ξ goes to zero,
then J reduces back to the ordinary angular momentum L.
For J we get the following Poisson bracket structure:
   
qξ ri qξ ri
{ Ji , r j } = Li − , rj = { Li , r j } − , r = eijk mrk (19)
c |r| c |r| j
| {z }
=0
   
qξ ri qξ ri
{ Ji , mṙ j } = Li − , mṙ j = { Li , mṙ j } − , mṙ j
c |r| c |r|
   
qξ ri qξ ri
= eijk mṙk + , mṙ j − , mṙ j = eijk mṙk . (20)
c |r| c |r|
This looks nice and simple as expected after the way we defined J. Now we can
see what we get if we take the Poisson bracket of the general angular momentum
J with our Hamiltonian H = 21 mṙ2 . For the first component J1 we have
1 1
{ H, J1 } = {mṙ2 , J1 } = {mṙ12 + mṙ22 + mṙ32 , J1 }
2 2
1 1 1
= {mṙ12 , J1 } + {mṙ22 , J1 } + {mṙ32 , J1 }
2 2 2
1 1
= [ṙ1 {mṙ1 , J1 } + ṙ1 {mṙ1 , J1 }] + [ṙ2 {mṙ2 , J1 } + ṙ2 {mṙ2 , J1 }]
2 2
1
+ [ṙ3 {mṙ3 , J1 } + ṙ3 {mṙ3 , J1 }]
2
= ṙ1 {mṙ1 , J1 } + ṙ2 {mṙ2 , J1 } + ṙ3 {mṙ3 , J1 }
= −ṙ1 e11k mṙk − ṙ2 e123 mṙ3 − ṙ3 e132 ṙ2
|{z} |{z} |{z}
=0 =1 = −1
= mṙ3 ṙ2 − mṙ2 ṙ3 = 0.
In exactly the same way we can calculate the Poisson bracket of the Hamiltonian
with J2 and J3 to find out that these vanish too.
The result, { H, Ji } = 0, tells us that every component of J and thus J itself is a
constant of motion. This means that the vector J, describing the general angular
momentum of a particle moving under the influence of a magnetic monopole,
points always in the same direction with a constant magnitude. If we now take
the scalar product between the unit vector |rr| pointing at the position of the
particle and the general angular momentum J we find out that
 
r r qξ r r qξ r r qξ
·J = · L− = · (r × mṙ) − · =− .
|r| |r| c |r| |r| c |r| |r| c
| {z } | {z }
= 0, since r ⊥ ṙ×r =1

On the other hand

r
· J = |J| cos θ,
|r|

7
where θ is the angle between J and the position vector of the particle. Thus
 
qξ qξ
cos θ = − ⇐⇒ θ = arccos − ,
c|J| c|J|

which tells us that the angle between the particle’s position vector and J is a
constant! This is the same result we obtained on the course when studying
magnetic monopoles in one of the exercises, even though the process was quite
different.
We have arrived at the main point of this work, i.e., the trajectory of the
particle moving near the magnetic monopole. The fact that θ stays constant
means that the trajectory of the particle lies on a cone. The cone has it’s apex on
the magnetic monopole, and it has the vector J as it’s axis.  The
 cone opens to
qξ
the opposite direction from J, and it has an angle of arccos c|J|
. Figure 1 below
may clarify the situation.

Figure 1: Particle under the influence of a magnetic monopole.

8
References
[1] P. Goddard and D. I. Olive, Magnetic monopoles in gauge field theories, Reports
on Progress in Physics 41 (1978), doi: 10.1088/0034-4885/41/9/001.

[2] H. Goldstein, C. Poole and J. Safko, Classical Mechanics, 3rd Edition, Addison
Wesley, San Francisco, 2002.

[3] P. A. M. Dirac, The Principles of Quantum Mechanics, 4th Edition, Snowball

Publishing, 2012.

[4] Y. M. Shnir, Magnetic Monopoles, Springer-Verlag, The Netherlands, 2005.