Chapter 8

Irreducible Representations
of SO(2) and SO(3)
The shortest path between two truths in the real domain passes through
the complex domain.
Jacques Hadamard
1
Some of the most useful aspects of group theory for applications to
physical problems stem from the orthogonality relations of characters
of irreducible representations. The widespread impact of these relations
stems from their role in constructing and resolving new representations
from direct products of irreducible representations. Direct products are
especially important in applications involving continuous groups, with
the construction of higher dimensional irreducible representations, the
derivation of angular momentum coupling rules, and the characteriza-
tion of families of elementary particles all relying on the formation and
decomposition of direct products.
Although the notion of an irreducible representation can be carried
over directly from our development of discrete groups through Schurs
rst lemma, a transcription of Schurs second lemma and the Great
Orthogonality Theorem to the language of continuous groups requires
a separate discussion. This is because proving the latter two theorems
1
Quoted in The Mathematical Intelligencer 13(1), 1991.
125
126 Irreducible Representations of SO(2) and SO(3)
necessitates performing summations over group elements and invoking
the Rearrangement Theorem (Theorem 2.1). This theorem guarantees
the following equality

g
f(g) =

g
f(g

g) , (8.1)
where the summation is over elements g in a group G, g

is any other el-

ement in G, and f is some function of the group elements. The crucial
point is that the same quantities appear on both sides of the equa-
tion; the only dierence is the order of their appearance. To proceed
with the proofs of these theorems for continuous groups requires an
equality analogous to (8.1):
_
f(R) dR =
_
f(R

R) dR, (8.2)
where R and R

are the elements of a continuous group and f is some

function of these elements. To appreciate the issues involved, we write
the integral on the left-hand side of (8.2) as an integral over the pa-
rameters
_
f(R) dR =
_
f(R)g(R) da , (8.3)
where g(R) is the density of group elements in parameter space in the
neighborhood of R. The equality in (8.2) will hold provided that the
density of group elements is arranged so that the density of the points
R

R is the same as that of the points R. Our task is to nd the form of

g(R) which ensures this. A related concept that will arise is the notion
of the order of the continuous group as the volume of its elements in
the space dened by the parameters of the group.
This chapter is devoted to the characters and irreducible represen-
tations of SO(2) and SO(3). For SO(2), we will show that the density of
group elements is uniform across parameter space, so the density func-
tion reduces to a constant. But, for SO(3), we will need to carry out
the determination of the density function in (8.3) explicitly. This will
illustrate the general procedure which is applicable to any group. For
both SO(2) and SO(3), we will derive the basis functions for their irre-
ducible representations which will be used to obtain the corresponding
characters and to demonstrate their orthogonality
Irreducible Representations of SO(2) and SO(3) 127
8.1 Orthogonality of Characters for SO(2)
The structure of SO(2) is simple enough that many of the results ob-
tained for discrete groups can be taken over directly with little or no
modication. The basis of this claim is that the Rearrangement Theo-
rem for this group is, apart from the replacement of the sum by an in-
tegral, a direct transcription of that for discrete groups which, together
with this group being Abelian, renders the calculation of characters a
straightforward exercise.
8.1.1 The Rearrangement Theorem
We rst show that the rearrangement theorem for this group is
_
2
0
R(

)R() d =
_
2
0
R() d.
This implies that the weight function appearing in (8.3) is unity, i.e.,
the density of group elements is uniform in the space of the parameter
. Using the fact that R(

)R() = R(

+ ), we have
_
2
0
R(

)R() d =
_
2
0
R(

+ ) d.
We now introduce a new integration variable =

+ . Since

is
xed, we have that d = d. Then, making the appropriate changes
in the upper and lower limits of integration, and using the fact that
R( + 2) = R(), yields
_
2
0
R(

+ ) d =
_

+2

R() d
=
_
2

R() d +
_

+2
2
R() d
=
_
2

R() d +
_

0
R() d
=
_
2
0
R() d ,
which veries our assertion.
128 Irreducible Representations of SO(2) and SO(3)
8.1.2 Characters of Irreducible Representations
We can now use Schurs rst lemma for SO(2). Since SO(2) is an
Abelian group, this rst lemma requires all of the irreducible represen-
tations to be one-dimensional (cf. Problem 4, Problem Set 5). Thus,
every element is in a class by itself and the characters must satisfy the
same multiplication rules as the elements of the group:
()(

) = ( +

) . (8.4)
The character corresponding to the unit element, (0), which must
map onto the identity for ordinary multiplication, is clearly unity for
all irreducible representations:
(0) = 1 . (8.5)
Finally, we require the irreducible representations to be single-valued,
i.e., an increase in the rotation angle by 2 does not change the eect
of the rotation. Thus,
( + 2) = () . (8.6)
The three conditions in (8.4), (8.5), and (8.6) are sucient to determine
the characters of all of the irreducible representations of SO(2).
We will proceed by writing Eq. (8.4) as a dierential equation and
using (8.5) as an initial condition and (8.6) as a boundary condi-
tion. In (8.4), we set

= d,
()(d) = ( + d) ,
and expand both sides of this equation to rst order in d:
()
_
(0) +
d
d

=0
d
_
= () +
d
d
d.
Then, using (8.5) and cancelling common terms, this equation reduces
to a rst-order ordinary dierential equation for ():
d
d
=

0
() ,
Irreducible Representations of SO(2) and SO(3) 129
where

0
=

(0) is to be determined. The general solution to this

equation is
() = Ae

,
where A is a constant of integration which is also to be determined.
In fact, by setting = 0 and invoking (8.5), we see that A = 1. The
requirement (8.6) of single-valuedness, when applied to this solution,
yields the condition that
e

0
(+2)
= e

,
or,
e
2

0
= 1 .
The most general solution of this equation is

0
= im, where i
2
= 1
and m is any integer. This produces an innite sequence of characters
of the irreducible representations of SO(2):

(m)
() = e
im
, m = . . . , 2, 1, 0, 1, 2, . . . (8.7)
The identical representation corresponds to m = 0. In contrast to the
case of nite groups, we see that SO(2) has an innite set of irreducible
representations, albeit one that is countably innite.
8.1.3 Orthogonality Relations
Having determined the characters for SO(2), we can now examine the
validity of the orthogonality theorems for characters which were dis-
cussed for discrete groups in Theorem 5.1. We proceed heuristically
and begin by observing that the exponential functions in (8.7) are or-
thogonal over the interval 0 < 2:
_
2
0
e
i(m

m)
d = 2
m,m
.
By writing this relation as
_
2
0

(m)
()
(m

)
() d = 2
m,m
, (8.8)
130 Irreducible Representations of SO(2) and SO(3)
we obtain an orthogonality relation of the form in Eq. (5.4), once we
identify the order of SO(2) as the quantity
_
2
0
d = 2 .
This is the volume of the group in the space of the parameter ,
which lies in the range 0 < 2, given that the density function is
unity, according to the discussion in thew preceding section. Note that
the integration over is eectively a sum over classes.
Example 8.1. Consider the representation of SO(2) derived in Section
7.2:
R() =
_
cos sin
sin cos
_
. (8.9)
Since SO(2) is an Abelian group, this representation must be reducible.
We can decompose this representation into its irreducible components
by using either the analogue of the Decomposition Theorem (Section
5.3) for continuous groups or, more directly, by using identities between
complex exponential and trigonometric functions:
() tr[R()]
= 2 cos
= e
i
+ e
i
.
A comparison with (8.7) yields
() =
(1)
() +
(1)
() ,
so the representation in (8.9) is a direct sum of the irreducible repre-
sentations corresponding to m = 1 and m = 1.
2
2
This example illustrates the importance of the eld used in the entries of the
matrices for SO(2). If we are restricted to real entries, then the representation in
(8.9) is irreducible. But, if the entries are complex, then this example shows that
this representation is reducible.
Irreducible Representations of SO(2) and SO(3) 131
8.2 Basis Functions for Irreducible Rep-
resentations
We were able to determine the characters for all of the irreducible rep-
resentations of SO(2) without any knowledge of the representations
themselves. But this is not the typical case for continuous groups. We
will see, for example, when determining the characters for SO(3) that
we will be required to construct explicit representations of rotations
corresponding to dierent classes. The action of these rotations on the
basis functions will determine the representation of that class and the
character will be calculated directly from this representation. As an
introduction to that discussion, in this section we will determine the
basis functions of the irreducible representations of SO(2).
We begin by calculating the eigenvalues of the matrix in (8.9) from
det(R I) = 0:

cos sin
sin cos

= (cos )
2
+ sin
2

=
2
2cos + 1 = 0 .
Solving for yields
= cos i sin = e
i
. (8.10)
The corresponding eigenvectors are proportional to x iy. Thus, op-
erating on these eigenvectors with R() (see below) generates the irre-
ducible representations corresponding to m = 1 and m = 1 in (8.7),
i.e., the characters
(1)
() and
(1)
().
Obtaining the basis functions for the other irreducible representa-
tions of SO(2) is now a matter of taking appropriate direct products,
since

(m)
()
(m

)
() =
(m+m

)
() .
In particular, the m-fold products (xiy)
m
generate irreducible repre-
sentations for the m-fold direct product, as discussed in Sec. 6.5. This
132 Irreducible Representations of SO(2) and SO(3)
can be veried directly from the transformation (8.9) applied to x and
y:
x

= x cos y sin ,
y

= x sin + y cos .
Then,
(x

iy

)
m
=
_
x cos y sin i(x sin + y cos )
_
m
=
_
x(cos i sin ) iy(cos i sin )
_
m
=
_
(x iy) e
i
_
m
= (x iy)
m
e
im
.
Therefore, we can now complete the character table for SO(2), including
the basis functions which generate the irreducible representations:
SO(2) E R()

m
: (x iy)
m
1 e
im
We note for future reference that the basis functions (xiy)
m
could
have been derived in a completely dierent manner. Consider Laplaces
equation in two dimensions:

2
u
x
2
+

2
u
y
2
= 0 .
This equation is invariant under all the elements of SO(2), as can be
easily veried. The general solution to this equation is
u(x, y) = F(x + iy) + G(x iy) ,
where F and G are arbitrary functions. Thus, if we are interested
in solutions which are homogeneous polynomials of degree m, we can
Irreducible Representations of SO(2) and SO(3) 133
choose in turn solutions with F(s) = s
m
and G(s) = 0 and then with
F(s) = 0 and G(s) = s
m
. We thereby obtain the expressions
u(x, y) = (x iy)
m
(8.11)
as solutions of Laplaces equations which are also the basis functions of
the irreducible representations of SO(2). These functions are the ana-
logues in two dimensions of spherical harmonics, which are the solutions
of Laplaces equations in three dimensions. These will be discussed later
in this chapter.
8.3 AxisAngle Representation of Proper
Rotations in Three Dimensions
The three most common parametrizations of proper rotations were dis-
cussed in Section 7.4. For the purposes of obtaining the orthogonal-
ity relations for the characters of SO(3), the representation in terms
of a xed axis about which a rotation is carried outthe axisangle
representationis the most convenient. We begin this section by show-
ing how this representation emerges naturally from the basic properties
of orthogonal matrices.
8.3.1 Eigenvalues of Orthogonal Matrices
Let A be any proper rotation matrix in three dimensions. Denoting the
eigenvalues of A by
1
,
2
, and
3
, and the corresponding eigenvalues
by u
1
, u
2
, and u
3
, we then have
Au
i
=
i
u
i
for i = 1, 2, 3. We can also form the adjoint of each equation:
u

i
A
t
=

i
.
These eigenvalue equations imply
u

i
u
i
= u

i
A
t
Au
i
= |
2
i
|u

i
u
i
,
134 Irreducible Representations of SO(2) and SO(3)
which shows that |
2
i
| = 1, i.e., that the modulus of every eigenvalue
of an orthogonal matrix is unity [cf. (8.10)]. The most general form
of such a quantity is a complex number of the form e
i
for some angle
. But these eigenvalues are also the roots of the characteristic equa-
tion det(A I) = 0 so, according to the Fundamental Theorem of
Algebra,
3
if they are complex, they must occur in complex conjugate
pairs (because the coecients of this polynomial, which are obtained
from the entries of A, are real). Hence, the most general form of the
eigenvalues of an orthogonal matrix in three dimensions is

1
= 1,
2
= e
i
,
3
= e
i
. (8.12)
The eigenvector corresponding to
1
= 1, which is unaected by the
action of A, thereby denes the axis about which the rotation is taken.
The quantity appearing in
2
and
3
denes the angle of rotation
about this axis.
8.3.2 The Axis and Angle of an Orthogonal Matrix
In this section, we show how the axis and angle of an orthogonal matrix
can be determined from its matrix elements. We take the axis of the
rotation to be a unit vector n, which is the eigenvector corresponding
to the eigenvalue of unity:
An = n. (8.13)
This equation and the orthogonality of A (AA
t
= A
t
A = 1) enables us
to write
A
t
n = A
t
An = n. (8.14)
Subtracting (8.14) from (8.13) yields
(A A
t
)n = 0 .
3
K. Homan and R. Kunze, Linear Algebra 2nd edn (PrenticeHall, Englewood
Clis, NJ, 1971), p. 138.
Irreducible Representations of SO(2) and SO(3) 135
In terms of the matrix elements a
ij
of A and the components n
i
of n,
we then have
(a
12
a
21
)n
2
+ (a
13
a
31
)n
3
= 0 ,
(a
21
a
12
)n
1
+ (a
23
a
32
)n
3
= 0 ,
(a
31
a
13
)n
1
+ (a
32
a
23
)n
2
= 0 .
Notice that these equations involve only the o-diagonal elements of A.
The solution of these equations yield the relations
n
2
n
1
=
a
31
a
13
a
23
a
32
,
n
3
n
1
=
a
12
a
21
a
23
a
32
, (8.15)
which, when combined with the normalization condition
nn = n
2
1
+ n
2
2
+ n
2
3
= 1
determines n uniquely.
The angle of the rotation can be determined from the invariance of
the trace of A under similarity transformations. Noting that the trace
is the sum of the eigenvalues, and using (8.12), we have
a
11
+ a
22
+ a
33
= 1 + e
i
+ e
i
= 1 + 2 cos , (8.16)
so is determined only by the diagonal elements of A.
8.3.3 Normal Form of an Orthogonal Matrix
We conclude this section by deriving the form of a rotation matrix in an
orthogonal coordinate system which naturally manifests the axis and
angle. The diagonal form of a rotation matrix is clearly given by
=
_
_
_
_
1 0 0
0 e
i
0
0 0 e
i
_
_
_
_
.
The eigenvector n corresponding to
1
= 1 is the axis of the rotation
and can always be chosen to be real. However, the eigenvectors of
2
=
136 Irreducible Representations of SO(2) and SO(3)
e
i
and
3
= e
i
are inherently complex quantities. An orthonormal
set can be chosen as
n
2
=
1
2

2(0, 1, i), n
2
=
1
2

2(0, 1, i) ,
respectively. Since we are interested in transformations of real coor-
dinates, we must perform a unitary transformation from this complex
basis to a real orthogonal basis, in which case our rotation matrix will
no longer be diagonal. The required unitary matrix which accomplishes
this is
U =
_
_
_
_
_
1 0 0
0
1
2

2
1
2
i

2
0
1
2

2
1
2
i

2
_
_
_
_
_
.
Thus,
R = U
1
U =
_
_
_
_
1 0 0
0 cos sin
0 sin cos
_
_
_
_
. (8.17)
When expressed in this basis, the rotation matrix clearly displays the
axis of rotation through the entry R
11
= 1, and the angle of rotation
through a 22 rotational submatrix in a plane perpendicular to this
axis.
8.3.4 Parameter Space for SO(3)
The axis-angle representation of three-dimensional rotations provides
a convenient parametrization of all elements of SO(3). We have seen
that every element of SO(3) can be represented by a unit vector corre-
sponding to the rotation axis and a scalar corresponding to the rotation
angle. Thus, consider the space dened by the three quantities
(n
1
, n
2
, n
3
) , (8.18)
where n
2
1
+ n
2
2
+ n
2
3
= 1. Every direction is represented by a point on
the unit sphere. Thus, dening an azimuthal angle and a polar angle
Irreducible Representations of SO(2) and SO(3) 137
A
B
C
D
E
O
Figure 8.1: Two-dimensional representation of the parameter space of SO(3)
as the interior of a sphere of radius . The point A represents a rotation
whose axis is along the direction OA and whose angle is the length of OA.
The points at A, B and C correspond to rotations with the same angle but
about axis along dierent directions. This denes the classes of SO(3). The
diametrically opposite points at D and E correspond to the same operation.
according to the usual conventions in spherical polar coordinates, the
parameter space of SO(3) can be represented as
(cos sin , sin sin , cos ) , (8.19)
where
0 , 0 2, 0 .
We can now see directly that this parameter space corresponds to the
interior of a sphere of radius (Fig. 8.1). For every point within the
sphere, there is a unique assignment to an element of SO(3): the direc-
tion from the radius to the point corresponds to the direction of the
rotation axis and the distance from the point to the origin represents
the rotation angle. Two diametrically opposed points on the surface of
the sphere ( = ) correspond to the same rotation, since a rotation
by about n is the same as a rotation by about this axis which,
in turn, is the same as a rotation by about n (whatever the sense
of rotation).
138 Irreducible Representations of SO(2) and SO(3)
Another useful feature of the axis-angle parametrization is the rep-
resentation of classes of SO(3). Consider two elements of SO(3) which
have the same angle of rotation but about dierent axes n and n

. We
denote these operations by R(n, ) and R(n

, ). Let U(n, n

) denote
the rotation of n into n

. The inverse of this operation then rotates n

into n. The relationship between R(n, ), R(n

, ), and U(n, n

) is,
therefore,
R(n, ) =
_
U(n, n

)
_
1
R(n

, )U(n, n

) ,
i.e., R(n, ) and R(n

, ) are related by a similarity transformation

and, therefore, belong to the same equivalence class. Referring to
Fig. 8.1, equivalence classes of SO(3) correspond to operations which
lie on the same radius. Thus, a summation over the classes of SO(3)
is equivalent to a sum over spherical shells.
8.4 Orthogonality Relations for SO(3)
The axis-angle representation of rotations provides, in addition to a
conceptual simplicity of elements of SO(3) in parameter space, a natural
framework within which to discuss the integration over the elements
of SO(3) and thereby to obtain the Rearrangement Theorem for this
group. In this section, we derive the density function g in (8.3) for
this group and then use this to identify the appropriate form of the
orthogonality relations for characters
8.4.1 The Density Function
As discussed in the introduction, one of the basic quantities of interest
for continuous groups is the density of group elements as a function
of position in parameter space. To determine this function for SO(3),
we rst consider the elements in the neighborhood of the identity and
then examine the behavior of these points under an arbitrary element
of SO(3). Referring to the discussion in Section 7.4.2, these elements
correspond to rotations by innitesimal angles
1
,
2
, and
3
about
each of the three coordinate axes. The rotation matrix associated with
Irreducible Representations of SO(2) and SO(3) 139
this transformation is
R =
_
_
_
_
1
3

2

3
1
1

2

1
1
_
_
_
_
.
The identity of SO(3) corresponds to the origin in the three-dimensional
parameter space,
1
=
2
=
3
= 0, and is indicated by the point
O in Fig. 8.1. For innitesimal rotation angles, the parameter space
spanned by R is associated with an innitesimal volume element in
the neighborhood of the origin.
We now follow the innitesimal transformation R by a nite trans-
formation R(n, ), i.e., we form the product RR. This generates a
volume element in the neighborhood of R and the product RR can be
viewed as transformation of the volume near the origin to that near R.
The Jacobian of this transformation is the relative change of volume
near the origin to that near R or, equivalently, is the relative change
of the density of operations near the origin to that near R. According
to the discussion in the introduction, this is the information required
from the density function for SO(3).
We have already seen that equivalence classes of SO(3) are com-
prised of all rotations with the same rotation angle, regardless of the
direction of the rotation axis. Thus, the density function is expected
to depend only on . Referring to Fig. 8.1, this means that the density
of elements depends only on the radial distance from the origin, not
on the direction, so we can choose R in accordance with this at our
convenience. Therefore, in constructing the matrix RR, we will use
for R a matrix of the form in (8.17). Thus,
RR =
_
_
_
_
1 0 0
0 cos sin
0 sin cos
_
_
_
_
_
_
_
_
1
3

2

3
1
1

2

1
1
_
_
_
_
=
_
_
_
_
1
3

2

3
cos +
2
sin cos
1
sin
1
cos sin

3
sin
2
cos sin
1
cos
1
sin + cos
_
_
_
_
.
140 Irreducible Representations of SO(2) and SO(3)
We can now use (8.15) and (8.16) to determine the axis n

and angle

of this product. The angle is determined from

1 + 2 cos

= 1 + 2 cos 2
1
sin ,
which, upon cancelling common factors, becomes
cos

= cos
1
sin .
Using the standard trigonometric formula for the cosine of a sum, we
nd, to rst order in
1
, that

= +
1
.
The unnormalized components of n

are determined from (8.15) to be

n

1
= 2
1
cos 2 sin ,
n

2
=
3
sin
2
(1 + cos ) ,
n

3
=
2
sin
3
(1 + cos ) .
To normalize the axis, we rst determine the length based on these
components. To rst order in the
i
, we nd
|n

| = 2
1
cos + 2 sin .
Thus, the components of the normalized rotation axis of RR are
n

1
= 1 ,
n

2
=
1
2

3
+
1
2

2
1 + cos
sin
,
n

3
=
1
2

2
+
1
2

3
1 + cos
sin
.
Expressed in terms of the parametrization in (8.18), RR is given by
(n

, n

, n

) =
_
+
1
,
1
2

3
+
2
1 + cos
sin
_
,
1
2

2
+
3
1 + cos
sin
__
.
Irreducible Representations of SO(2) and SO(3) 141
This denes the transformation from the neighborhood of the origin to
the neighborhood near RR. The Jacobian J of this transformation,
obtained from
J = det

(n

, (8.20)
determines how the density of elements of SO(3) near the origin is
transformed to the density of points near R. By taking the derivatives
in (8.20) to obtain the entries (i, j) in the Jacobian matrix, we obtain
J =

1 0 0
0
1 + cos
2 sin

1
2

0
1
2

1 + cos
2 sin

=

2
2(1 cos )
.
Notice that
lim
0
J = 1 ,
so that the normalization of the volume in parameter space is such
that the volume near the unit element is unity. Hence, the density
of elements in parameter space is the reciprocal of J, so the density
function g in (8.3) is
g() =
2

2
(1 cos ) . (8.21)
8.4.2 Integrals in Parameter Space
The density function in (8.21) now permits us to carry out integral over
the group. Thus, for a function F(, ), where denotes the angular
variables in the parametrization in (8.19), we have
__
g()F(, )
2
dd,
where we have used the usual volume element for spherical polar coor-
dinates. Using the density function in (8.21), this integral becomes
__
2(1 cos )F(, )dd.
142 Irreducible Representations of SO(2) and SO(3)
We can now establish the orthogonality relation for characters. If
we denote the characters for two irreducible representations of SO(3)
by

() and

(), then we have

__
2(1 cos )

()

()dd =
,
__
2(1 cos )dd.
The integral on the right-hand side of this equation, which has the
value 8
2
, corresponds to the volume of SO(3) in parameter space. The
integral over the angular variables on the left-hand side yields 2 4,
so cancelling common factors, we obtain
_

0
(1 cos )

()

() d =
,
. (8.22)
This is the orthogonality relation for characters of SO(3).
8.5 Irreducible Representations and Char-
acters for SO(3)
For SO(2), we were able to determine the characters of the irreducible
representations directly, i.e., without having to determine the basis
functions of these representations. The structure of SO(3), however,
does not allow for such a simple procedure, so we must determine the
basis functions from the outset.
8.5.1 Spherical Harmonics
We proceed as in Section 8.2 by determining the homogeneous polyno-
mial solutions of Laplaces equation, now in three dimensions:

2
u
x
2
+

2
u
y
2
+

2
u
z
2
= 0 .
We seek solutions of the form
u(x, y, z) =

a,b
c
ab
(x + iy)
a
(x iy)
b
z
ab
,
Irreducible Representations of SO(2) and SO(3) 143
which are homogeneous polynomials of degree . In spherical polar
coordinates,
x = r cos sin , y = r sin sin , z = r cos ,
where 0 < 2 and 0 , these polynomial solutions transform
to
u(r, , ) =

a,b
c
ab
r

sin
a+b
cos
ab
e
i(ab)
. (8.23)
Alternatively, Laplaces equation in spherical polar coordinates is
1
r
2

r
_
r
2
u
r
_
+
1
r
2
sin

_
sin
u

_
+
1
r
2
sin
2

2
u

2
= 0 .
When the method of separation of variables is used to nd solutions
of this equation of the form u(r, , ) = R(r)()(), the stipulation
that the solution be single-valued with respect changes in by 2,
u(r, , + 2) = u(r, , ) ,
requires that
() e
im
,
where m is an integer. Comparing this expression with the correspond-
ing factor in (8.23), we see that a b = m. Since the ranges of both
a and b are between 0 and , we see that there are 2 + 1 values of
m consistent with homogeneous polynomial solutions of degree . The
corresponding values of m are m . The 2 + 1 independent
homogeneous polynomials of degree are called the spherical harmonics
and denoted by Y
m
(, ). Their functional form is
Y
m
(, ) P

m
() e
im
, (8.24)
where P

m
() is a Legendre function. In the following discussion, we will
utilize only the exponential factor in the spherical harmonics.
144 Irreducible Representations of SO(2) and SO(3)
8.5.2 Characters of Irreducible Representations
The Y
m
(, ) form a (2 + 1)-dimensional representation of SO(3).
Thus, for a general rotation R, we have
RY
m
(, ) =

=
Y
m
(, ) .

m
(R)
To determine the character of this representation, it is convenient to
again invoke the fact that the classes of SO(3) are determined only by
the rotation angle, not by the direction of the rotation axis. Thus, we
can choose a rotation axis at our convenience and we therefore focus on
rotations through an angle about the z-axis. In this case, the form
of (8.24) allows us to write
R
z
()Y
m
(, ) = Y
m
(, ) = e
im
Y
m
(, ) .
Thus, the corresponding transformation matrix is given by

[R
z
()] =
_
_
_
_
_
_
_
_
e
i
0 0
0 e
i(1)
0
.
.
.
.
.
.
.
.
.
.
.
.
0 0 e
i
_
_
_
_
_
_
_
_
. (8.25)
The character
()
() of this class is obtained by taking the trace of
this matrix:

()
() = e
i
+ e
i(1)
+ + e
i
= e
i
_
1 + e
i
+ e
2i
+ + e
2i
_
= e
i
1 e
(2+1)i
1 e
i
=
e
(+1/2)i
e
(+1/2)i
e
i/2
e
i/2
=
sin [( +
1
2
)]
sin (
1
2
)
.
Irreducible Representations of SO(2) and SO(3) 145
The orthogonality integral for these characters takes the form
_

0
(1 cos )
sin [( +
1
2
)] sin [(

+
1
2
)]
sin
2
(
1
2
)
d.
Using the trigonometric identity
2 sin
2
(
1
2
) = 1 cos
enables us to write the orthogonality integral as
_

0
sin [( +
1
2
)] sin [(

+
1
2
)] d =
1
2

,
,
where the right-hand side of this equation follows either from (8.22) or
from the orthogonality of the sine functions over (0, ).
It is possible to show directly, using Schurs rst lemma, that the
spherical harmonics form a basis for (2 + 1)-dimensional irreducible
representations of SO(3). However, this requires invoking properties of
the Legendre functions in (8.24). If we conne ourselves to the matrices
in (8.25) then we can show that a matrix that commutes with all such
rotation matrices must reduce to a diagonal matrix. If we then consider
rotations about any other direction, which requires some knowledge of
the Legendre functions, we can then show that this constant matrix
must, in fact, be a constant multiple of the unit matrix. Hence, accord-
ing to Schurs rst lemma, these representations are irreducible. We can
now construct the character table for SO(3) with the basis functions
which generate the irreducible representations:
SO(3) E R()

: Y
m
(, ) 1
sin [( +
1
2
)]
sin (
1
2
)
8.6 Summary
In this chapter, we have shown how the orthogonality relations devel-
oped for nite groups must be adapted for continuous groups, using
146 Irreducible Representations of SO(2) and SO(3)
SO(2) and SO(3) as examples. For SO(2), which is a one-parameter
Abelian group, this proved to be a straightforward matter. However,
the corresponding calculations for SO(3) required us to determine ex-
plicitly the density function to produce the appropriate form of the
orthogonality relations. We found that the there are an innite se-
quence of irreducible representations of dimensionality 2 + 1, where
0. Because of the connection between SO(3) and angular mo-
mentum, the structure of these irreducible representations has several
physical consequences:
For systems that possess spherical symmetry, the energy eigen-
states have degeneracies of 2+1. The fact that there is a greater
degeneracy for the hydrogen atom is due to a hidden SO(4)
symmetry.
4
The formation and decomposition of direct products of the irre-
ducible representations of SO(3) forms the basis of angular mo-
mentum coupling rules (Problem 6, Problem Sets 10) and the
classication of atomic spectra.
5
When atoms are placed within crystals, the original spherical
symmetry is lowered to the symmetry of the crystal. This causes
levels which were degenerate in the spherically-symmetric envi-
ronment to split. Such crystal-eld eects are important for
many aspects for electrons in crystalline solids.
6
4
H.F. Jones, Groups, Representations and Physics (Institute of Physics, Bristol,
1998), pp. 124127.
5
E.P. Wigner, Group Theory and its Application to the Quantum Mechanics of
Atomic Spectra (Academic, New York, 1959), pp. 177194.
6
M. Tinkham, Group Theory and Quantum Mechanics (McGrawHill, New York,
1964), pp. 6580.