Sie sind auf Seite 1von 11

Non commutative algebra in presence of non-uniform magnetic eld

and Quantum mechanics


Shivraj Prajapat
Indian Institute of Science Education and Research-Pune
S. N. Bose National Centre for Basic Sciences,JD Block, Sector III, Salt Lake,
Kolkata-700098, India
February 17, 2014
1 Introduction
There has been a recent upsurge in research about non-commutative quantum mechanics as it is believed that
it can provide an insight into the nature of space time at the Planck scale. In their seminal paper, Doplicher
et al [1] argued from the considerations of both general relativity and quantum mechanics that the localization
of an event in spacetime with arbitrary accuracy is operationally impossible. The reason being given was that
to measure length at these scales, the probe wavelength has to be of the order of Compton wavelength which is
inversely proportional to the probe mass. To measure innitesimally small length scales thus requires innitely
high mass and hence this huge mass localized in a small volume would generate a black hole which would
make the measurement impossible. One natural way to avoid this is to postulate a non-vanishing uncertainty
relation for the simultaneous measurement of all the four space-time coordinates and this is captured by invoking
non-commuting operator- valued coordinates, the simplest model being
[ x

, x

] = i

(1)
All this comes as a direct consequence of the principle of gravitational stability against localization, which can
be stated as follows: The gravitational eld generated by the concentration of energy required by the Heisenberg
Uncertainty Principle to localise an event in space-time should not be so strong to hide the event itself to any
distant observer - distant compared to the Planck scale.
But on the other hand, coordinate non-commutativity may also arise in more physical situations involving
the motion of electric charges in a strong external magnetic eld. When a charged and massive particle moves
on a plane in the presence of a strong constant magnetic eld B pointing perpendicular to the plane, it has been
shown that the non-commutativity of space coordinates is of order of the inverse of the magnetic eld [2],[3].
The coordinate non-commutativity has also been generalised for non-uniform magnetic eld in the presence of
some other background potential [6],[7]. In this paper we explore how the momenta behave in such problems and
work out the general non-commutative algebra in presence of non-uniform magnetic eld, magnetic monopoles
in particular . First, we give a short review of the usual commutative algebra, then we move on to the planar
Landau problem and discuss the non-commutative algebra in that case and lastly we work out the general non-
commutative algebra in presence of an arbitrary magnetic eld and also discuss it in the light of the monopole
problem.
2 Short review of commutative( = 0) quantum mechanics
The usual (commutative) quantum mechanical description of the nature follows from classical mechanical pic-
ture by promoting dynamical variables like position(x), momenta(p) to the level of corresponding Hermitian
operators, x, p which are subjected to certain set of commutation relations. Dirac prescribed that these commuta-
tion relations can be obtained by multiplying the operator corresponding to the Poisson bracket of corresponding
dynamical variables with i.
[ x, p] = i

{x, p} = iI (2)
The physical systems of concern are themselves described in quantum mechanics by certain abstract states say

_
which correspond to a ray in a certain Hilbert space, and are obtained by superposing the simultaneous
1
eigen-states of maximal number of commuting observables. In the context of our simple example, this correspond
to either the position operator ( x) or the momentum operator ( p).

_
=
_
dx

x
_
x

_
dx(x)

x
_
; x

x
_
= x

x
_
, (3)
or

_
=
_
dp

p
_
p

_
dp

(p)

p
_
; p

p
_
= p

p
_
(4)
Further, the overlaps

_
(x) and

_


(p) are regarded respectively as coordinate space and
momentum space representations of the abstract state

_
respectively and are related by Fourier transformation
of each other.

(x) =
1

2
_
dpe
ix p

(p) (5)

(p) =
1

2
_
dxe
ix p

(x) (6)
Correspondingly, the operators x and p too have their respective representations.
In coordinate space representation, the coordinates x
i
acts directly on (x) by left multiplication and momentum
acts adjointly as:

X
i
: (x) x
i
(x) = x
i
(x) (7)

P
i
: (x) [ p
i
, (x)] = i

{p
i
, (x)}
PB
= i

x
i
, (8)
if there are more than one degree of freedom. On the other hand, in momentum space representation, it is the
momentum which acts on

(p) directly by the left multiplication and x
i
acts adjointly as:

X
i
:

(p) [ x
i
,

(p)] = i

{x
i
,

(p)}
PB
= i

p
i
(9)

P
i
:

(p) p
i

(p) = p
i

(p) (10)
These operator-valued maps

X
i
and

P
j
obey following Heisenberg commutation relations:
[

X
i
,

P
j
] = i
ij
and [

X
i
,

X
j
] = [

P
i
,

P
j
] = 0 (11)
3 Non-commutativity in presence of constant magnetic eld
We would like to extend this coordinate and momentum space representations in non-commutative quantum
mechanics in this note and for this it will be advantageous to begin by considering the planar Landau problem,
which provides the simplest example, where spatial non-commutativity is realized.
3.1 Review of planar Landau problem
We begin by considering a particle of mass () and charge (q) moving in a plane subjected to a perpendicular
magnetic eld B. The Langrangian describing the particles motion is given by(in units of, = c = 1):
L =
_
1
2

x
2
+ q

x

A(x)
_
(12)
So the Hamiltonian for the particle becomes:
H =
1
2
_
p q

A
_
2
(13)
where p is the momentum conjugate to x i.e
p
i
=
L
x
i
= x
i
+ qA
i
= v
i
+ qA
i
, (14)
such that the mechanical momentum is:

=
_
p q

A
_
= v, (15)
2
which is clearly gauge invariant. In terms of this the Hamiltonian can be re-written as:
H =

2
2
(16)
Now using the non-commutative structure of

at quantum mechanical level i.e [

i
,

j
] = iqB
ij
we obtain the
time evolution of velocity operator of the particle

v
i
=
1
i
[ v
i
,

H] =
qB


ij
v
j
(17)
Recasting above equation in matrix form we get:

V = T

V (18)
with

V =
_
v
1
v
2
_
and T =
qB

_
0 1
1 0
_
Integrating above equation with respect to time we get

V =

X = exp(Tt)

V
0
with

X =
_
x
1
x
2
_
(19)
A further integration yields

X
0
=

qB
_
0 1
1 0
_

V
0
(20)
Above equation can also be written simply as:
x
0i
= x
i
+

qB

ij
v
0j
(21)
Here x
0i
, are the operator-valued constants of integration and are known as guiding center coordinates, which
commutes with v
i
, and should not be identied with x
i
(t = 0). Now taking the coordinates x
i
to be commuting
i.e. [ x
i
, x
j
] = 0, we nd that the guiding center coordinates x
0i
becomes non-commuting:
[ x
0i
, x
0j
] =
i
qB

ij
= i
ij
, (22)
where we have introduced the NC parameter =
1
qB
3.2 Non-commutivity in classical picture
The Landau problem discussed above can also be studied classically by using Diracs constraint analysis. The
non-commutivity that arise here among the classically projected coordinates(which has been done by Diracs
prescription) has the same form as was obtained earlier.
3.2.1 Short Review of Diracs Constraint Analysis
If a Hamiltonian system has an even number of second class constraints

0( = 1, 2, ...2N) then for a


generic phase space function F(q
i
, p
i
) one can associate another phase space function F

(q
i
, p
i
) F(q
i
, p
i
)
(here the notation stands for weak equality i.e equal only on the constraint surface

0) such that it has


vanishing Poisson Brackets (at least weakly) with these constraints i.e,
F

= F {F,

}
PB
{
1
}

F s.t. {F

}
PB
0 (23)
Furthermore we see that
(F

= F

{F

}
PB
{
1
}

(24)
So F

may be regarded as a projected value of F on the constraint surface in the classical level. This is the
classical analogue of quantum mechanical projection of corresponding operator

F i.e P

FP. One can easily


verify that for the pair of phase space functions f(q
i
, p
i
) and g(q
i
, p
i
) following relation holds true:
{f

, g

}
PB
{f, g}
DB
, (25)
where the Dirac bracket is dened as:
{f, g}
DB
= {f, g}
PB
{f,

}
PB
{
1
}

, g}
PB
, (26)
with {}

= {

}
PB
being a matrix element obtained by taking Poisson bracket between a pair of
constraints

and

.
3
3.2.2 Landau problem in the light of constraint analysis
We now consider the Hamiltonian (13) in the limit of high magnetic eld or small mass(limits for the particle
to be conned on lowest Landau level), where we eectively have following pair of constraints:

1
= p
1
qA
1
0

2
= p
2
qA
2
0, (27)
so that using (23), the classically projected coordinate variable is:
x

i
= x
i
+
1
2

ij
v
j
where =
qB
2
(28)
Which matches with equation (21). Hence we conclude that above projected coordinate and guiding center
coordinate are same for our system.
Now using (25) we get following result:
{x
1
, x
2
}
DB
{x

1
, x

2
}
PB
=
1
qB
, (29)
which at the level of quantum mechanical operator becomes:
[ x
1
, x
2
]
DB
= i
ij
, (30)
which is the same commutation relation that we got earlier (22). Similarly we can nd the projected momenta
variables as follows:
p

i
= p
i
+

ij
2
x
j
, (31)
where projected momenta are scaled up by a factor of 2 i.e. p

i
= 2p

i
, and they can be shown to satisfy following
relation using (25).
{p

1
, p

2
}
DB
= {p

1
, p

2
}
PB
=
1

, (32)
which at the level of quantum mechanical operator becomes:
[ p

1
, p

2
]
DB
=
i

ij
(33)
Also these projected coordinates and momenta variables satisfy following relation:
p

i
=
1

ij
x

j
(34)
3.3 Non-commutativity from projection to the lowest Landau level
We now proceed ahead to construct lowest Landau level states and study them. By choosing a symmetric gauge

A =
B
2
(x
2
, x
1
, 0), the Hamiltonian describing the motion of the particle takes following form:

H =
1
2

p
2
+
1
2

x
2

L
z
, (35)
which has the form of a 2D harmonic oscillator, augmented by a Zeeman term(involving z component of angular
momentum). Now by using following creation and annihilation operators:

b
i
=
_

2
_
x
i
+
i p
i

_
,

i
=
_

2
_
x
i

i p
i

_
; i = 1, 2 (36)
our Hamiltonian becomes

H =
_

b
1
+

b
2
+ 1
_
i
_

b
1

b
2
_
(37)
We see that the Zeeman term prevents the Hamiltonian to be in diagonal form. In order to diagonalize it, we
introduce following circular basis:

b
+
=
1

2
_

b
1
i

b
2
_
,

+
=
1

2
_

1
+ i

2
_

=
1

2
_

b
1
+ i

b
2
_
,

=
1

2
_

1
i

2
_
(38)
4
so that we get the diagonalized Hamiltonian as:

H =
_

b
+
+

+ 1
_

b
+

_
= 2
_

+ 1
_
(39)
The energy spectrum can be read o immediately using a suitable eigen state

n
+
, n

_
to get
E
n

= 2
_
n

+
1
2
_
(40)
Since the spectrum obtained is independent of n
+
, each energy level is innitely degenerate. We re-label the
eigen state

n
+
, n

, l
_
, by introducing angular momentum l = n
+
n

. Then we move on to the lowest


Landau level i.e

= 0, l
_
. To conne our particle to the lowest Landau level, we require that the band gap(of
the spectrum) should be very large, which can be achieved by demanding that either the particles mass is very
small, or magnetic eld strength is very large, see equation (39) and (35).
We notice that any lowest Landau state is kicked o of the Hilbert space of the states in lowest Landau level
under a spatial translation say x
1
x

1
= x
1
+a for innitesimal a. Infact spatial translation gives a coherent
state e
ia p

= 0, l
_
, which is an eigen state of

b

e
ia p

= 0, l
_
=
a
2

e
ia p

= 0, l
_
(41)
Hence we need to introduce appropriate projection operator which brings back dynamical variables to the lowest
Landau level:
P =

l=0

= 0, l
_
n

= 0, l

l=0

l
_
l

(42)
so that we get the projected coordinates as follows:
PXP =
_
P x
1
P
P x
2
P
_
(43)
now using the fact that
x
1
=
1
2

b
+
+

+
+

_
x
2
=
i
2

b
+

+
+

_
One can show after a straightforward calculation that
PXP =
_
P x
1
P
P x
2
P
_
=
1
2

l + 1
_

l
_
l + 1

l + 1
_
l

i(

l
_
l + 1

l + 1
_
l

)
_
(44)
Similarly working out for momenta we see
p
1
=
i

2
_

b
+
+

_
p
2
=

2
_

+
+

b
+

_
So that our projected momenta becomes:
PPP =
_
P p
1
P
P p
2
P
_
=

l + 1
_
i
_

l
_
l + 1

l + 1
_
l

l
_
l + 1

l + 1
_
l

_
(45)
It is important to note here that projected momenta and projected coordinate are related in following order
(P p
i
P) =
ij
(P x
j
P) =
qB
2

ij
(P x
j
P) (46)
This is precisely the operatorial version of (27)in symmeteric gauge. Now working out the commutation relation
between these projected coordinates and momenta we see:
[P x
i
P, P x
j
P] =
i
qB

ij
P (47)
5
[P x
i
P, P p
j
P] =
i
2

ij
P (48)
[P p
i
P, P p
j
P] = i
qB
4

ij
P (49)
Writing above commutation relation on projected plane(where P becomes essentially an identity operator) and
using a non-canonical transformation p
i
p

i
= 2p
i
, to transform (48) to the standard Heisenberg form, we
get:
[ x
i
, x
j
] = i
ij
(50)
[ x
i
, p

j
] = i
ij
(51)
[ p

i
, p

j
] =
i

ij
, (52)
where
1
qB
. We see that the commutation relation (50) and (52) are same as (30) and (33). We can obtain
following result on the projected plane by using (46), (35) and the denition of :
p

i
=
1

ij
x
j
, (53)
which is same as equation (34) at the level of classical coordinate and momentum variables. This kind of linear
relation between x
i
and p
i
is a new feature in the present case and was absent in the commutative case discussed
in Section 2. Using this, we can correspondingly obtain coordinate and momentum space representation. As
we shall see subsequently that in the former case, this gives a representation of Heisenberg algebra where the
noncommutativity occurs in the coordinate operators, in contrast to the latter where it occurs between the
momenta operators.
3.3.1 Coordinate Representation
Like in the commutative case, here too the coordinate representation is obtained by direct left action of the
position operator on an abstract state vector |)

X
i
: |)

X
i
|) = x
i
|) (54)
and adjoint action of momentum:

P
i
: |)

P
i
|) = [ p
i
, |)] =
1

ij
[ x
j
, |)] (55)
where we have used (53), by dropping the prime over p
i
. By using above actions and Jacoby identity, one can
easily show that the commutation relations (50-52) yields the following commutators between the maps

X
i
and

P
i
:
[

X
i
,

X
j
] = i
ij
(56)
[

X
i
,

P
j
] = i
ij
(57)
[

P
i
,

P
j
] = 0, (58)
which are the usual non-commutative Heisenberg algebra [4]. We know that Moyal and Voros basis states of
the quantum Hilbert space H
q
, where these algebra (56 - 58) are realised, [4] are dened as follows:
|x)
V
=
_

2
_
d
2
p e

|p|
2
4
e
ipx
|p) (59)
|x)
M
=
_
d
2
p
2
e
ipx
|p), (60)
where the momentum eigen-states |p) are given by:
|p) =
_

2
e
ip x
,

P
i
|p) = p
i
|p) (61)
6
Further in H
q
one can dene operators

X
c
i
[5], which are the average of left (

X
L
i
) and right(

X
R
i
) action non-
commuting position operators

X
c
i
=
1
2
(

X
L
i
+

X
R
i
). These operators are mutually commuting i.e. [

X
c
i
,

X
c
j
] = 0
and admits the Moyal basis states (60) as the eigen-states of these operators:

X
c
i
|x)
M
= x
i
|x)
M
(62)
Expanding an arbitrary state |
m
), in momentum basis as,
|
m
) =
_
d
2
p
2
(p)|p) (63)
and then taking the overlap of Voros and Moyal basis states we get position-space wave function i.e
V/M
(x) :

V
(x) =
V
(x|
m
) =
_

2
_
d
2
p
2
e
|p|
2
4
e
ipx
(p) (64)

M
(x) =
M
(x|
m
) =
_
d
2
p
(2)
2
e
ipx
(p) (65)

V
(x) =
V
(x|
m
) =
_

2
e

2
x
4
_
d
2
p
2
e
ipx
(p) =
_

2
e

2
x
4

M
(x) (66)
3.3.2 Momentum space Representation
In this case the role of position and momentum operators (as discussed last time) are interchanged. We can
reverse equation (53) in following way to get(we are again dropping the prime deliberately):
x
i
=
ij
p
j
(67)
The action of these operators are given as follows:

P
i
= p
i
and

X
i
= [ x
i
, ] =
ij
[ p
j
, ] (68)
One can modify the commutation relations (50-52) by using above action of operators in following ways:
[

X
i
,

X
j
] = 0 (69)
[

X
i
,

P
j
] = i
ij
(70)
[

P
i
,

P
j
] =
i

ij
(71)
Now one can extend the analysis of Moyal and Voros basis states carried out in [4] to our momentum space
representation where in eect the role of x and p gets interchanged and we obtain following results:
|p)
V
=
1

2
_
d
2
x e

|x|
2
4
e
ixp
|x) (72)
|p)
M
=
_
d
2
x
2
e
ixp
|x), (73)
where position eigen-states |x) are given by:
|x) =
1

2
e
ix p
,

X
i
|x) = x
i
|x) (74)
Similar to the commuting position operators

X
c
i
in our last discussion, one can introduce commuting momentum
operator

P
c
i
here, which are again the average of left (

P
L
i
) and right (

P
R
i
) action momentum operators of H
q
.
The Moyal basis constructed in momentum conguration are again the simultaneous eigen-states of the operators

P
c
i
.

P
c
i
=

P
i
+
i
2

ij

X
j
,

P
c
i
|p)
M
= p
i
|p)
M
(75)
Also this time our abstract state |
p
) will expand in following manner:
|
p
) =
1

2
_
d
2
x
2
e
ix p
(x) =
_
d
2
x
2
(x)|x), (76)
7
Note that the arbitrary functions (x)(76) and (p)(63) are related by Fourier transform of each other.
Now in the similar fashion as was done earlier one can obtain momentum-space wave function i.e
V/M
(p),
corresponding to the abstract state |
p
) for both the basis i.e Moyal and Voros as follows:

V
(p) =
V
(p|
p
) =
1

2
_
d
2
x
2
e

|x|
2
4
e
ixp
(x) (77)

M
(p) =
M
(p|
p
) =
_
d
2
x
(2)
2
e
ixp
(x) (78)
Since the position operator acts adjointly, so here again we can replace |x|
2
with Laplacian operator (
2
p
) in
momentum space representation, such that above momentum-space wave function for Voros basis becomes:

V
(p) =
V
(p|
p
) =
1

2
e

2
p
4
_
d
2
x
2
e
ixp
(x) (79)
Now we see that position-space(65) and momentum-space(78) wave functions for Moyal basis are clearly related
by Fourier transform of each other, which is expected as Moyal basis states are the simultaneous eigen-states of
commuting operators

X
c
i
(62) and

P
c
i
(75), and hence are similar to the wave functions obtained in commuting
case (section 1).
While the position-space(66) and momentum-space(79) wave functions for Voros basis not exactly related by
usual Fourier transform, but what we now call as non-commutative generalization of Fourier transform, where
the parts relevant for the usual Fourier transform are related with the help of an extra factor of exponential of
their corresponding Laplacian operators(upto some irrelevant factor).
4 Non-commutativity in the presence of non-uniform magnetic eld
The Hamiltonian of a charged particle with mass m and electric charge e in any non-uniform magnetic eld B
and the external potential V is
H =

2
2m
+ V (x) =
(p
e
c
A)
2
2m
+ V (x) (80)
Where is the kinematic momentum and p is the canonical momentum an A denotes the vector potential of
the Maxwells theory. Here we are interested to study the system in the limit B , which is equivalent to
the limit m 0. In order to take m 0 limit in the Hamiltonian we must impose
i
= 0 as a constraint.
We have a system with some constraints. The system can be handled by Diracs method of dealing with
the constraints. The constraints are divided in to two category- primary constraints and secondary constraints.
Primary constraints are the constraints which we know or given already, while the secondary constraints are
obtained from the consistency condition. In the Diracs method constraints are divided in to another two
category- rst class constraints and second class constraints. Constraints which has vanishing Poisson bracket
with all the constraint is called rst class while the constraint which has non-vanishing Poisson bracket with
atleast one of the constraint is called the second class constraint. The second class constraints are actually the
useless degrees of freedom which has no importance in the dynamics of the system. Diracs method is actually
about the taking care of second class constraints. The
i
= 0 are rst class constraints. The advantage of
considering the particle in the external potential or the external force is that in the presence of this force the

i
s can be transformed to get the new set of constraints where two of them are second class. The transformed
set of constraints are
= B.
1
= f.
2
= B f. (81)
is a rst class constraint while

i
s are second class constraints. The second class constraints can be handled
by replacing the usual Poisson bracket by Diracs bracket which is dened to be
{x
i
, x
j
}
D
= {x
i
, x
j
} {x
i
,
k
}C
kl
{
l
, x
j
} (82)
where C
kl
is dened by
C
kl
{
l
,
i
} =
ik
(83)
The C matrix in this case is
e
cB
2
f
2
_
0 1
1 0
_
8
The Dirac bracket between coordinates turns out to be
{x
i
, x
j
}
D
=

ijk
B
k
B
2
(84)
The Dirac bracket involving the momentum is
{x
i
, p
j
}
D
=
ij

1
B
2
A
l
x
j

ilk
B
k
(85)
{p
i
, p
j
}
D
=
1
B
2
A
l
x
i
A
m
x
j

klm
B
k
(86)
It is clear from the expression that Dirac bracket involving momentum is gauge dependent.
second class constraints can be handled by another way also in which phase space quantities are projected to
the constraint surface. The projection of any phase space quantity f on the constraint surface is dened to be
f

= f {f,
k
}C
kl

l
(87)
Where f

is the projection of phase space quantity f. Now it can be shown that the poisson bracket between
two projected quantity is same as the Dirac bracket between those quantity. The projected coordinates and
momentums turns out to be
x

i
= x
i
+
1
B
2
f
2
(f
i

2
(B f)
i

1
) (88)
p

i
= p
i
+
1
B
2
f
2
(f
j
A
j
x
i

2
(B f)
j
A
j
x
i

1
) (89)
These projected coordinates and projected momentum satises linear relationship
p

i
p
i
=
A
j
x
i
(x

j
x
j
) (90)
For consistency, such a noncommutative algebra must satisfy the Jacobi identity:
{x
i
, {x
j
, x
k
}} +{x
k
, {x
i
, x
j
}} +{x
j
, {x
k
, x
i
}} = 0 (91)
{x
i
, {x
j
,
k
}} +{
k
, {x
i
, x
j
}} +{x
j
, {
k
, x
i
}} = 0 (92)
{x
i
, {
j
,
k
}} +{
k
, {x
i
,
j
}} +{
j
, {
k
, x
i
}} = 0 (93)
{
i
, {
j
,
k
}} +{
k
, {
i
,
j
}} +{
j
, {
k
,
i
}} = 0 (94)
It can be checked that indeed the commutation relation obtained above satises the Jacobi identity.Here in the
Jacobi identity we use kinematical momentum
i
and not the canonical momentum p
i
, because p
i
are gauge
dependent so the Jacobi identity involving p
i
will depend upon gauge and identity will be satised in a particular
choice of gauge,while
i
are gauge independent so the Jacobi identity involving
i
will be independent upon
choice of gauge and identity will be satised in all gauges.
Now we discuss about the rst class constraint = B..The rst class constraint are found to be the generator
of some symmetry transformation in the Diracs theory for handling the constraints.To check what kind of
symmetry transformation this rst class constraint generate,we check the Dirac bracket {x
i
, B.} and {p
i
, B.}
{x
i
, B.} = B
i
(95)
{p
i
, B.} = B
j
A
j
x
i
(96)
from {x
i
, B.} = B
i
it can be visualized that there is a translation symmetry along B
i
.
5 Magnetic monopole
Now we discuss a particular scenario, where the magnetic eld is produced by a magnetic monopole. The eld
produced by magnetic monopole is

B =
g
r
2
r (97)
9
Following Diracs prescription, one particular choice of gauge which produces this magnetic eld is given by [8]

A(r) =
g
4|r|
r

k
|r| r.

k
(98)
where

k denotes the direction of the Dirac string. Now if one chooses the Dirac string to be in the negative Z-
direction then the vector potential can be given in the Cartesian coordinates as
A
x
=
g
r
y
r + z
A
y
=
g
r
y
r + z
A
z
= 0 (99)
where we have r = |r|. So for this problem, the Dirac brackets between the coordinates will be
{x
i
, x
j
}
D
=
c
eB

ijk
x
k
(100)
So one can see that the coordinates satisfy an SU(2) algebra. But the Dirac bracket involving the momentum
are trickier. One can easily nd the brackets between position and momentum components to be of the form
{z, p
z
}
D
= 1 r
r z
r + z
(101)
{x, p
x
}
D
= {y, p
y
}
D
= 1 r
z
r + z
(102)
{x, p
y
}
D
= {y, p
x
}
D
= 0 (103)
{x, p
z
}
D
= r
zx
(r + z)
2
(104)
{y, p
z
}
D
= r
zy
(r + z)
2
(105)
Now if one considers the north pole of the unit sphere (r=1) and scales the p
x
and p
y
components of the
momentum to 2p
x
and 2p
y
, then one has the familiar Dirac brackets as obtained in the 2D Landau problem:
{x
i
, p
j
}
D
=
ij
(106)
Similarly one can easily verify that the Dirac brackets for the momentum operators satisfy
{p
x
, p
y
}
D
=
eB
c
r
2
z
(r + z)
2
; {p
y
, p
z
}
D
=
eB
c
r
2
zx
(r + z)
3
; {p
z
, p
x
}
D
=
eB
c
r
2
zy
(r + z)
3
(107)
So scaling p
x
and p
y
components of the momentum by 2 as before, the brackets at the north pole take the usual
form as in the planar Landau problem
{p
x
, p
y
}
D
=
eB
c
; {p
y
, p
z
}
D
= {p
z
, p
x
}
D
= 0 (108)
It is not hard to gure out why these relations at the north pole mimic those as in the planar Landau case.
Since the magnetic eld in the monopole conguration is in the radial direction, so the magnetic eld at the
north pole is perpendicular to the tangent plane at the north pole. Not only that, if one looks at the north pole
from a point directly above it, one will see the gauge elds as concentric circles which resembles the symmetric
gauge for the planar Landau problem. So one can well expect that the commutation relations at this particular
point will be of the same form as that obtained in planar Landau problem.
As already noted, the commutation relations between the position coordinates are gauge invariant but those
between the momentum components and between position and momentum are gauge dependent. It so happens
that the choice of a particular gauge (viz. with the Dirac string in the negative Z-axis) and also the choice
of a particular point (viz. the north pole) on the unit sphere give us back the familiar commutation relations
as in the planar Landau case. There is no global gauge choice that can give us these familiar commutation
relations at every point on the manifold. But there exists suitable local gauge choice such that one can cast
the commutation relations in the familiar form at an arbitrary point on the sphere. Lets see how this can be
achieved.
Consider a point P on the sphere. Now due to the cylindrical symmetry of the vector potential, one can rotate
the sphere about the Z-axis so as to make P lie on the YZ plane without any change in
10
mono.png
physics. Now, consider a new gauge marked by the new direction of
the Dirac string along the

k
2
= cos z + sin y vector i.e. the new
conguration (

A
2
) is achieved just by rotating the previous cong-
uration (

A
1
) by an angle in the YZ plane. Now if we rotate the
coordinates about the X axis such that the point P lies on the new Z
axis and denote the new set of axes by (X

, Y

, Z

), then all the com-


mutation relations at the point P in terms of the new coordinates will give the same form as those calculated
before at the north pole 100,106,108. It can be easily veried that the new gauge diers from the old one by
the gradient of a scalar function. For example at the north pole (Q)

A
2
(Q)

A
1
(Q) = |
Q
(109)
where
=
g
4|r|
xtan

2
(110)
References
[1] S.Doplicher, K.Fredenhagen and J.E.Roberts, Comm.Math.Phys. 172,187 (1995).
[2] Dunne, Gerald, and Roman Jackiw, Nuclear Physics B-Proceedings Supplements 33.3 (1993): 114-118.
[3] Magro, Gabrielle. arXiv preprint quant-ph/0302001 (2003).
[4] P. Basu, B.Chakraborty and F.G. Scholtz, J. Physics. A 44, 285204 (2011).
[5] Balachandran A P, Pinzul A, Qureshi B A, Vaidya, S Phys. Rev. D. 76, 105025 (2007).
[6] Frenkel, J., and S. H. Pereira, Physical Review D 69.12 (2004): 127702.
[7] Govaerts, Jan, and Se

An Murray, Journal of High Energy Physics 2010.1 (2010): 1-8.


[8] Rajantie, Arttu, Contemporary Physics 53.3 (2012): 195-211.
11

Das könnte Ihnen auch gefallen