Chapter 7

Continuous Groups, Lie

Groups, and Lie Algebras
Zeno was concerned with three problems . . . These are the problem of
the innitesimal, the innite, and continuity . . .
Bertrand Russell
The groups we have considered so far have been, in all but a few cases,
discrete and nite. Most of the central theorems for these groups and
their representations have relied on carrying out sums over the group
elements, often in conjunction with the Rearrangement Theorem (The-
orem 2.1). These results provide the basis for the application of groups
and representations to physical problems through the construction and
manipulation of character tables and the associated computations that
require direct sums, direct products, orthogonality and decomposition.
But the notion of symmetry transformations that are based on con-
tinuous quantities also occur naturally in physical applications. For
example, the Hamiltonian of a system with spherical symmetry (e.g.,
atoms and, in particular, the hydrogen atom) is invariant under all
three-dimensional rotations. To address the consequences of this in-
variance within the framework of group theory necessitates confronting
several issues that arise from the continuum of rotation angles. These
include dening what we mean by a multiplication table, determin-
ing how summations over group elements are carried out, and deriving
107
108 Continuous Groups, Lie Groups, and Lie Algebras
the appropriate re-statement of the Rearrangement Theorem to enable
the Great Orthogonality Theorem and its consequences to be obtained
for continuous groups. More generally, the existence of a continuum
of group elements, when combined with the requirement of analyticity,
introduces new structures associated with constructing dierentials and
integrals of group elements. In eect, this represents an amalgamation
of group theory and analysis, so such groups are the natural objects
for describing the symmetry of analytic structures such as dierential
equations and those that arise in dierential geometry. In fact, the
introduction of analytic groups by Sophus Lie late in the 19th century
was motivated by the search for symmetries of dierential equations.
In this chapter we begin our discussion about the modications to
our development of groups and representations that are necessitated by
having a continuum of elements. We begin in the rst section with the
denition of a continuous group and specialize to the most common
type of continuous group, the Lie group. We then introduce the idea
of an innitesimal generator of a transformation, from which every
element can be obtained by repeated application. These generators
embody much of the structure of the group and, because there are a
nite number of these entities, are simpler to work with than the full
group. This leads naturally to the Lie algebra associated with a Lie
group. All of these concepts are illustrated with the groups of proper
rotations in two and three dimensions. The representation of these
groups, their character tables, and basis functions will be discussed in
the next chapter.
7.1 Continuous Groups
Consider a set of elements R that depend on a number of real continuous
parameters, R(a) R(a
1
, a
2
, . . . , a
r
). These elements are said to form
a continuous group if they fulll the requirements of a group (Section
2.1) and if there is some notion of proximity or continuity imposed
on the elements of the group in the sense that a small change in one
of the factors of a product produces a correspondingly small change in
their product. If the group elements depend on r parameters, this is
called an r-parameter continuous group.
Continuous Groups, Lie Groups, and Lie Algebras 109
In general terms, the requirements that a continuous set of elements
form a group are the same as those for discrete elements, namely, clo-
sure under multiplication, associativity, the existence of a unit, and
an inverse for every element. Consider rst the multiplication of two
elements R(a) and R(b) to yield the product R(c):
R(c) = R(a)R(b) .
Then c must be a continuous real function f of a and b:
c = f(a, b) .
This denes the structure of the group in the same way as the multi-
plication table does for discrete groups. The associativity of the com-
position law,
R(a)
_
R(b)R(c)
_
. .
R[f(b, c)]
= [R(a)R(b)
_
. .
R[f(b, c)]
R(c) ,
requires that
f[a, f(b, c)] = f[f(a, b), c] .
The existence of an identity element, which we denote by R(a
0
),
R(a
0
)R(a) = R(a)R(a
0
) = R(a) ,
is expressed in terms of f as
f(a
0
, a) = f(a, a
0
) = a .
The inverse of each element R(a), denoted by R(a

), produces
R(a

)R(a) = R(a)R(a

) = R(a
0
) .
Therefore,
f(a

, a) = f(a, a

) = a
0
.
If f is an analytic function, i.e., a function with a convergent Tay-
lor series expansion within the domain dened by the parameters, the
110 Continuous Groups, Lie Groups, and Lie Algebras
resulting group is called an r-parameter Lie group, named after Sophus
Lie, a Norwegian mathematician who provided the foundations for such
groups.
Our interest in physical applications centers around transformations
on d-dimensional spaces. Examples include Euclidean spaces, where the
variables are spatial coordinates, Minkowski spaces, where the variables
are space-time coordinates, and spaces associated with internal degrees
of freedom, such as spin or isospin. In all cases, these are mappings of
the space onto itself and have the general form
x

i
= f
i
(x
1
, x
2
, . . . , x
d
; a
1
, a
2
, . . . , a
r
), i = 1, 2, . . . , d .
If the f
i
are analytic, then this denes an r-parameter Lie group of
transformations.
Example 7.1 Consider the one-dimensional transformations
x

= ax (7.1)
where a is an non-zero real number. This transformation corresponds
to stretching the real line by a factor a. The product of two such
operations, x

= ax

and x

= bx is
x

= ax

= abx .
By writing x

= cx, we have that

c = ab , (7.2)
so the multiplication of two transformations is described by an analytic
function that yields another transformation of the form in (7.1). This
operation is clearly associative, as well as Abelian, since the product
transformation corresponds to the multiplication of real numbers. This
product can also be used to determine the inverse of these transfor-
mations. By setting c = 1 in (7.2), so that x

= x, the inverse of
(7.1) is seen to correspond to the transformation with a

= a
1
, which
explains the requirement that a = 0. Finally, the identity is deter-
mined from x

= x, which clearly corresponds to the transformation

Continuous Groups, Lie Groups, and Lie Algebras 111
with a = 1. Hence, the transformations dened in (7.1) form a one-
parameter Abelian Lie group.
Example 7.2 Now consider the one-dimensional transformations
x

= a
1
x + a
2
, (7.3)
where again a
1
is an non-zero real number. These transformations cor-
responds to the stretching of the real line by a factor a
1
, as in the
preceding Example, and a translation by a
2
. The product of two oper-
ations is
x

= a
1
x

+ a
2
= a
1
(b
1
x + b
2
) + a
2
= a
1
b
1
x + a
1
b
2
+ a
2
.
By writing x

= c
1
x + c
2
, we have that
c
1
= a
1
b
1
, c
2
= a
1
b
2
+ a
2
,
so the multiplication of two transformations is described by an analytic
function and yields another transformation of the form in (7.1). How-
ever, although this multiplication is associative, it is not Abelian, as
can be seen from the fact that the indices do not enter symmetrically
in c
2
. By setting, c
1
= c
2
= 1, the inverse of (7.3) is the transformation
x

=
x
a
1

a
2
a
1
.
The identity is again determined from x

= x, which requires that

a
1
= 1 and a
2
= 0. Hence, the transformations in (7.3) form a two-
parameter (non-Abelian) Lie group.
7.2 Linear Transformation Groups
An important class of transformations is the group of linear transfor-
mations in d dimensions. These can be represented by d d matrices.
For example, the most general such transformation in two dimensions
is x

= Ax or, in matrix form,

_
x

_
=
_
a
11
a
12
a
21
a
22
__
x
y
_
, (7.4)
112 Continuous Groups, Lie Groups, and Lie Algebras
where det(A) = a
11
a
22
a
12
a
21
= 0 (Example 2.4). With no further
restriction, and with the composition of two elements given by the usual
rules of matrix multiplication, these matrices form a four-parameter Lie
group. This Lie group is called the general linear group in two dimensions
and is denoted by GL(2,R), where the R signies that the entries
are real; the corresponding group with complex entries is denoted by
GL(2,C). In n dimensions, these transformation groups are denoted by
GL(n,R), or, with complex entries, by GL(n,C).
7.2.1 Orthogonal Groups
Many transformations in physical applications are required to preserve
length in the appropriate space. If that space is ordinary Euclidean
n-dimensional space, the restriction that lengths be preserved means
that
x
2
1
+ x
2
2
+ + x
2
n
= x
2
1
+ x
2
2
+ + x
2
n
. (7.5)
The corresponding groups, which are subgroups of the general linear
group, are called orthogonal, and are denoted by O(n).
Consider the orthogonal group in two-dimensions, i.e., O(2), where
the coordinates are x and y. By substituting the general transformation
(7.4) into (7.5), we require that
x
2
+ y
2
= (a
11
x + a
12
y)
2
+ (a
21
x + a
22
y)
2
= (a
2
11
+ a
2
21
)x
2
+ 2(a
11
a
12
+ a
21
a
22
)xy + (a
2
12
+ a
2
22
)y
2
.
For the right-hand side of this equation to be equal to x
2
+y
2
for all x
and y, we must set
a
2
11
+ a
2
21
= 1, a
11
a
12
+ a
21
a
22
= 0, a
2
12
+ a
2
22
= 1 .
Thus, we have three conditions imposed on four parameters, leaving one
free parameter. These conditions can be used to establish the following
relation:
(a
11
a
22
a
12
a
21
)
2
= 1 .
Continuous Groups, Lie Groups, and Lie Algebras 113
Recognizing the quantity in parentheses as the determinant of the trans-
formation, this condition implies that
det(A) = 1 .
If det(A) = 1, then the parity of the coordinate system is not changed
by the transformation; this corresponds to a proper rotation. If det(A) =
1, then the parity of the coordinate system is changed by the transfor-
mation; this corresponds to an improper rotation. As we have already
seen, both types of transformations are important in physical applica-
tions, but we will rst examine the proper rotations in two-dimensions.
This group is called the special orthogonal group in two dimensions and
is denoted by SO(2), where special signies the restriction to proper
rotations. The parametrization of this group that we will use is
R() =
_
cos sin
sin cos
_
, (7.6)
where , the single parameter in this Lie group, is the rotation angle of
the transformation. As can easily be checked using the trigonometric
identities for the sum of two angles,
R(
1
+
2
) = R(
1
)R(
2
) , (7.7)
so this group is clearly Abelian.
7.3 Innitesimal Generators
A construction of immense utility in the study of Lie groups, which was
introduced and extensively studied by Lie, is the innitesimal generator.
The idea behind this is that instead of having to consider the group as
a whole, for many purposes it is sucient to consider an innitesimal
transformation around the identity. Any nite transformation can then
be constructed by the repeated application, or integration, of this
innitesimal transformation.
114 Continuous Groups, Lie Groups, and Lie Algebras
7.3.1 Matrix Form of Generators
For SO(2), we rst expand R() in a Taylor series around the identity
( = 0):
R() = R(0) +
dR
d

=0
+
1
2
d
2
R
d
2

=0

2
+ . (7.8)
The coecients in this series can be determined directly from (7.6), but
a more elegant solution may be found by rst dierentiating (7.7) with
respect to
1
,
d
d
1
R(
1
+
2
) =
dR(
1
)
d
1
R(
2
) , (7.9)
then setting
1
= 0. Using the chain rule, the left-hand side of this
equation is
_
dR(
1
+
2
)
d(
1
+
2
)
d(
1
+
2
)
d
1
_

1
=0
=
dR(
2
)
d
2
,
so Eq. (7.9) becomes
dR()
d
= XR() , (7.10)
where
dR(
1
)
d
1

1
=0
=
_
0 1
1 0
_
X . (7.11)
Equations (7.10) and (7.11) allow us to determine all of the expan-
sions coecients in (7.9). By setting = 0 in (7.10) and observing
that R(0) = I, where I is the 2 2 unit matrix,
I =
_
1 0
0 1
_
,
we obtain
dR()
d

=0
= X . (7.12)
Continuous Groups, Lie Groups, and Lie Algebras 115
To determine the higher-order derivatives of R, we dierentiate (7.10)
n times, and set = 0:
d
n
R()
d
n

=0
= X
d
n1
R()
d
n1

=0
.
This yields, in conjunction with (7.12),
d
n
R()
d
n

=0
= X
n
.
Substituting this expression into the Taylor series in (7.8) allows us to
write
R() = I + X +
1
2
X
2

2
+
=

n=0
1
n!
(X)
n
e
X
,
where X
0
= I and the exponential of a matrix is dened by the Taylor
series expansion of the exponential. Thus, every rotation by a nite
angle can be obtain from the exponentiation of the matrix X, which
is called the innitesimal generator of rotations. Since X
2
= I, it is a
straightforward matter to show directly from the Taylor series of the
exponential (Problem 4, Problem Set 9) that
e
X
= I cos + X sin =
_
cos sin
sin cos
_
.
7.3.2 Operator Form of Generators
An alternative way of representing innitesimal generators through
which connections with quantum mechanics can be directly made is
in terms of dierential operators. To derive the operator associated
with innitesimal rotations, we expand (7.6) to rst order in d to
obtain the transformation
x

= x cos y sin = x y d,
y

= x sin + y cos = x d + y .
116 Continuous Groups, Lie Groups, and Lie Algebras
An arbitrary dierentiable function F(x, y) then transforms as
F(x

, y

) = F(x y d, x d + y) .
Retaining terms to rst order in d on the right-hand side of this equa-
tion yields
F(x

, y

) = F(x, y) +
_
y
F
x
+ x
F
y
_
d.
Since F is an arbitrary function, we can associate innitesimal rotations
with the operator
X = x

y
y

x
.
As we will see in the next section, this operator is proportional to the
z-component of the angular momentum operator.
The group SO(2) is simple enough that the full benets of an in-
nitesimal generator are not readily apparent. We will see in the next
section, where we discuss SO(3), that the innitesimal generators em-
body much of the structure of the full group.
7.4 SO(3)
The orthogonal group in three dimensions is comprised of the trans-
formations that leave the quantity x
2
+ y
2
+ z
2
invariant. The group
GL(3,R) has 9 parameters, but the invariance of the length produces six
independent conditions, leaving three free parameters, so O(3) forms a
three-parameter Lie group. If we restrict ourselves to transformations
with unit determinant, we obtain the group of proper rotations in three
dimensions, SO(3).
There are three common ways to parametrize these rotations:
Successive rotations about three mutually orthogonal xed axes.
Successive about the z-axis, about the new y-axis, and then about
the new z-axis. These are called Euler angles.
Continuous Groups, Lie Groups, and Lie Algebras 117
The axis-angle representation, dened in terms of an axis whose
direction is specied by a unit vector (two parameters) and a
rotation about that axis (one parameter).
In this section, we will use the rst of these parametrizations to
demonstrate some of the properties of SO(3). In the next chapter,
where we will develop the orthogonality relations for this group, the
axis-angle representation will prove more convenient.
7.4.1 Rotation Matrices
Consider rst rotations about the z-axis by an angle
3
:
R
3
(
3
) =
_
_
_
_
cos
3
sin
3
0
sin
3
cos
3
0
0 0 1
_
_
_
_
.
The corresponding innitesimal generator is calculated as in (7.11):
X
3
=
dR
3
d
3

3
=0
=
_
_
_
_
0 1 0
1 0 0
0 0 0
_
_
_
_
.
These results are essentially identical to those for SO(2). However,
for SO(3), we have rotations about two other axes to consider. For
rotations about the x-axes by an angle
1
, the rotation matrix is
R
1
(
1
) =
_
_
_
_
1 0 0
0 cos
1
sin
1
0 sin
1
cos
1
_
_
_
_
and the corresponding generator is
X
1
=
dR
1
d
1

1
=0
=
_
_
_
_
0 0 0
0 0 1
0 1 0
_
_
_
_
118 Continuous Groups, Lie Groups, and Lie Algebras
Finally, for rotations about the y-axis by an angle
2
, we have
R
2
(
2
) =
_
_
_
_
cos
2
0 sin
2
0 1 0
sin
2
0 cos
2
_
_
_
_
and the generator is
X
2
=
dR
2
d
2

2
=0
=
_
_
_
_
0 0 1
0 0 0
1 0 0
_
_
_
_
As can be easily veried, the matrices R
i
(
i
) do not commute, nor
do the X
i
. However, the X
i
have an additional useful property, namely
closure under commutation. As an example, consider the products
X
1
X
2
and X
2
X
1
:
X
1
X
2
=
_
_
_
_
0 0 0
0 0 1
0 1 0
_
_
_
_
_
_
_
_
0 0 1
0 0 0
1 0 0
_
_
_
_
=
_
_
_
_
0 0 0
1 0 0
0 0 0
_
_
_
_
X
2
X
1
=
_
_
_
_
0 0 1
0 0 0
1 0 0
_
_
_
_
_
_
_
_
0 0 0
0 0 1
0 1 0
_
_
_
_
=
_
_
_
_
0 1 0
0 0 0
0 0 0
_
_
_
_
Thus, the commutator of X
1
and X
2
, denoted by [X
1
, X
2
] is given by
[X
1
, X
2
] X
1
X
2
X
2
X
1
=
_
_
_
_
0 1 0
1 0 0
0 0 0
_
_
_
_
= X
3
Similarly, we have
[X
2
, X
3
] = X
1
, [X
3
, X
1
] = X
2
The commutation relations among all of the X
i
can be succinctly sum-
marized by introducing the anti-symmetric symbol
ijk
, which takes the
Continuous Groups, Lie Groups, and Lie Algebras 119
value 1 for a symmetric permutation of distinct i, j, and k, the value
1 for an antisymmetric permutation, and is zero otherwise (i.e., if two
or more of i, j and k are equal). We can then write
[X
i
, X
j
] =
ijk
X
k
(7.13)
We will discuss the physical interpretation of these generators once we
obtain their operator form in the next section.
7.4.2 Operators for Innitesimal Rotations
As was the case in Section 7.3, an alternative to the matrix represen-
tation of innitesimal generators is in terms of dierential operators.
Proceeding as in that section, we rst write the general rotation as an
expansion to rst order in each of the
i
about the identity. This yields
the transformation matrix
_
_
_
_
x

_
_
_
_
=
_
_
_
_
1
3

2

3
1
1

2

1
1
_
_
_
_
_
_
_
_
x
y
z
_
_
_
_
Substituting this coordinate transformation into a dierentiable func-
tion F(x, y, z),
F(x

, y

, z

) = F(x
3
y +
2
z, y +
3
x
1
z, z
2
x +
1
y)
and expanding the right-hand side to rst order in the
i
yields the
following expression:
F(x

, y

, z

) = F(x, y, z)
+
_
F
z
y
F
y
z
_

1
+
_
F
x
z
F
z
x
_

2
+
_
F
y
x
F
x
y
_

3
Since F is an arbitrary dierentiable function, we can identify the gen-
erators X
i
of rotations about the coordinate axes from the coecients
of the
i
, i.e., with the dierential operators
X
1
= y

z
z

y
120 Continuous Groups, Lie Groups, and Lie Algebras
X
2
= z

x
x

z
(7.14)
X
3
= x

y
y

x
Notice that X
3
is the operator obtained for SO(2) in Section 7.3. We
can now assign a physical interpretation to these operators by compar-
ing them with the vectors components of the angular operators in the
coordinate representation, obtained from the denition
L = r p = r (i h)
Carrying out the cross-product yields the standard expressions
L
1
= i h
_
y

z
z

y
_
L
2
= i h
_
z

x
x

z
_
(7.15)
L
3
= i h
_
x

y
y

x
_
for the x, y, and z components of L, respectively. Thus, L
i
= i hX
i
,
for i = 1, 2, 3, and (7.13) becomes
[L
i
, L
j
] = i h
ijk
L
k
which are the usual angular momentum commutation relations. There-
fore, we can associate the vector components of the angular momentum
operator with the generators of innitesimal rotations about the cor-
responding axes. An analogous association exists between the vector
components of the coordinate representation of the linear momentum
operator and dierential translation operations along the corresponding
directions.
7.4.3 The Algebra of Innitesimal Generators
The commutation relations in (7.13) dene a product of two gener-
ators which yields the third generator. Thus, the set of generators is
Continuous Groups, Lie Groups, and Lie Algebras 121
closed under this operation. Triple products, which determine whether
or not this composition law is associative, can be written in a concise
form using only the denition of the commutator, i.e., in the form of
an identity, without any explicit reference to the quantities involved.
Beginning with the triple product
_
A, [B, C]
_
= A[B, C] [B, C]A
= ABC ACB BCA + CBA
We now add and subtract the quantities BAC and CAB on the right-
hand side of this equation and rearrange the resulting expression into
commutators to obtain
_
A, [B, C]
_
= ABC ACB BCA + CBA
+BAC BAC + CAB CAB
= C(AB BA) + (AB BA)C
+B(AC CA) (AC CA)B
=
_
[A, B], C
_
+
_
[C, A], B
_
A simple rearrangement yields the Jacobi identity:
_
A, [B, C]
_
+
_
B, [C, A]
_
+
_
C, [A, B]
_
= 0
Notice that this identity has been obtained using only the denition of
the commutator.
For the innitesimal generators of the rotation group, with the com-
mutator in (7.13), each of the terms in the Jacobi identity vanishes.
Thus,
_
A, [B, C]
_
=
_
[A, B], C]
_
so the product of these generators is associative. In the more general
case, however, products of quantities dened in terms of a commutator
are not associative. The Lie algebra associated with the Lie group from
122 Continuous Groups, Lie Groups, and Lie Algebras
which the generators are obtained consists of quantities A, B, C, . . .
dened by
A =
3

k=1
a
k
X
k
, B =
3

k=1
b
k
X
k
, C =
3

k=1
c
k
X
k
, etc.
where the a
k
, b
k
, c
k
, . . . are real coecients and from which linear com-
binations A + B with real and can be formed. The product is
given by
[A, B] = [B, A]
and the Jacobi identity is, of course, satised.
The formal denition of a Lie algebra, which is an abstraction of
the properties just discussed, is as follows.
Denition. A Lie algebra is a vector space L over some eld F
1
(typically the real or complex numbers) together with a binary opera-
tion [, ] : L L L, called the Lie bracket, which has the following
properties:
1. Bilinearity.
[ax + by, z] = a[x, z] + b[y, z]
[z, ax + by] = a[z, x] + b[z, y]
for all a and b in F and x, y, and z in L.
2. Jacobi identity.
_
[x, y], z
_
+
_
[z, x], y] +
_
[y, z], x
_
= 0
for all x, y, and z in L.
1
A eld is an algebraic system of elements in which the operations of addition,
subtraction, multiplication, and division (except by zero) may be performed without
leaving the system (closure) and the associative, commutative, and distributive
rules, familiar from the arithmetic of ordinary numbers, hold. Examples of elds
are the rational numbers, the real numbers, and the complex numbers. The smallest
eld has only two elements: {0, 1}. The concept of a eld is useful for dening vectors
and matrices, whose components can be elements of any eld.
Continuous Groups, Lie Groups, and Lie Algebras 123
3. Antisymmetry.
[x, y] = [y, x]
for all x and y in L.
7.5 Summary
In this chapter, we have described the properties of Lie groups in
terms of specic examples, especially SO(2) and SO(3). With this
background, we can generalize our discussion to any Lie group. An
r-parameter Lie group of transformations on an n-dimensional space is
x

i
= f
i
(x
1
, x
2
, . . . , x
n
; a
1
, a
2
, . . . , a
r
)
where i = 1, 2, . . . , n. If only one of the r parameters a
i
is changed
from zero, while all the other parameters are held xed, we obtain the
innitesimal transformations X
i
associated with this Lie group. These
can be expressed as dierential operators by examining the eect of
these innitesimal coordinate transformations on an arbitrary dieren-
tiable function F:
dF =
n

j=1
F
x
j
dx
j
=
n

j=1
F
x
j
_
r

i=1
f
j
a
i

a=0
da
i
_
=
r

i=1
da
i
_
n

j=1
f
j
a
i

a=0

x
j
_
F
We identify the dierential operators X
i
as the coecient of da
i
in this
dierential:
X
i
=
n

j=1
f
j
a
i

a=0

x
j
for r = 1, 2, . . . , r. These operators satisfy commutation relations of
the form
[X
i
, X
j
] = c
k
ij
X
k
124 Continuous Groups, Lie Groups, and Lie Algebras
where the c
k
ij
are called structure constants and are a property of the
group. The commutator satises the Jacobi identity,
_
X
i
, [X
j
, X
k
]
_
+
_
X
j
, [X
k
, X
i
]
_
+
_
X
k
, [X
i
, X
j
]
_
= 0
which places a constraint on the structure constants. The commutator
and the Jacobi identity, together with the ability to form real linear
combinations of the X
i
endows these generators with the structure of
an algebra, called the Lie algebra associated with the Lie group.