This action might not be possible to undo. Are you sure you want to continue?
https://www.scribd.com/doc/47808918/GroupTheory
11/13/2011
text
original
Ferdi Aryasetiawan
Dept. of Theoretical Physics,
University of Lund,
S¨olvegatan 14A, 223 62 Lund  Sweden
Contents
1 Abstract Group Theory 1
1.1 Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Abelian Group . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Subgroup . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.5 Multiplication Table . . . . . . . . . . . . . . . . . . . . . 3
1.6 Cyclic Groups . . . . . . . . . . . . . . . . . . . . . . . . 4
1.7 Order of an Element . . . . . . . . . . . . . . . . . . . . . 5
1.8 Properties of Finite Groups . . . . . . . . . . . . . . . . . 5
1.9 Cosets . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.10 Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.11 Isomorphism and Homomorphism . . . . . . . . . . . . . . 7
1.12 Invariant Subgroup . . . . . . . . . . . . . . . . . . . . . . 8
1.13 Direct Product Group . . . . . . . . . . . . . . . . . . . . 8
2 Theory of Group Representations 11
2.1 Deﬁnitions . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Equivalent Representations . . . . . . . . . . . . . . . . . 11
2.3 Reducible and Irreducible Representation . . . . . . . . . . 12
2.4 Equivalent Unitary Representation . . . . . . . . . . . . . . 12
2.5 Schur’s Lemma . . . . . . . . . . . . . . . . . . . . . . . . 14
2.6 Orthogonality Theorem . . . . . . . . . . . . . . . . . . . 17
2.7 The Characters of a Representation . . . . . . . . . . . . . 19
2.8 The Regular Representation . . . . . . . . . . . . . . . . . 20
2.9 The Character Table . . . . . . . . . . . . . . . . . . . . . 24
2.10 Properties of the Character Table . . . . . . . . . . . . . . 24
2.11 Class Multiplication . . . . . . . . . . . . . . . . . . . . . 26
i
ii CONTENTS
2.12 Direct Product Groups . . . . . . . . . . . . . . . . . . . . 27
2.13 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3 Group Theory in Quantum Mechanics 33
3.1 Linear Vector Space . . . . . . . . . . . . . . . . . . . . . 33
3.2 Symmetry Transformations . . . . . . . . . . . . . . . . . 35
3.3 Invariant Subspace and Its Generation . . . . . . . . . . . . 37
3.4 Basis Functions for a Representation . . . . . . . . . . . . 37
3.5 Projection Operators . . . . . . . . . . . . . . . . . . . . . 39
3.6 Orthogonality of Basis Functions . . . . . . . . . . . . . . 41
3.7 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.8 Relationship to Quantum Mechanics . . . . . . . . . . . . 47
3.8.1 The Group of the Schr¨odinger Equation . . . . . . . 48
3.8.2 Degeneracy and Invariant Subspace . . . . . . . . . 49
3.8.3 Partial Diagonalisation of
ˆ
H . . . . . . . . . . . . . 50
3.9 Irreducible Sets of Operators . . . . . . . . . . . . . . . . . 50
3.10 Direct Product Representations of a Group . . . . . . . . . 51
3.11 ClebschGordan Coeﬃcients . . . . . . . . . . . . . . . . . 53
3.12 The WignerEckart Theorem . . . . . . . . . . . . . . . . . 54
3.13 Applications . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.14 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4 Lie Groups 63
4.1 Deﬁnitions . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.2 Inﬁnitesimal Operators . . . . . . . . . . . . . . . . . . . . 64
4.3 Lie Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.4 Casimir Operators . . . . . . . . . . . . . . . . . . . . . . 68
4.5 Ladder Operators and Multiplets . . . . . . . . . . . . . . 69
4.6 Summation over group elements . . . . . . . . . . . . . . . 71
4.7 Rotation Group R
3
. . . . . . . . . . . . . . . . . . . . . . 72
4.7.1 Inﬁnitesimal operators . . . . . . . . . . . . . . . . 73
4.7.2 Commutation Relations . . . . . . . . . . . . . . . 74
4.7.3 Casimir Operator . . . . . . . . . . . . . . . . . . . 75
4.8 Irreducible Representations of R
3
. . . . . . . . . . . . . . 76
4.9 Characters of R
3
. . . . . . . . . . . . . . . . . . . . . . . 79
4.10 The VectorCoupling Theorem . . . . . . . . . . . . . . . . 81
4.11 ClebschGordan Coeﬃcients . . . . . . . . . . . . . . . . . 82
CONTENTS iii
4.12 Spherical Harmonics . . . . . . . . . . . . . . . . . . . . . 87
5 Atomic Physics 91
5.1
ˆ
H
0
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
5.2
ˆ
H
0
+
ˆ
H
1
. . . . . . . . . . . . . . . . . . . . . . . . . . . 93
5.2.1 Conﬁguration and Term . . . . . . . . . . . . . . . 94
5.2.2 Ladder Operator Method . . . . . . . . . . . . . . 95
5.2.3 Projection Operator Method . . . . . . . . . . . . . 99
5.2.4 Term Energies . . . . . . . . . . . . . . . . . . . . 100
5.3
ˆ
H =
ˆ
H
0
+
ˆ
H
1
+
ˆ
H
2
. . . . . . . . . . . . . . . . . . . . . 102
6 The Group SU
2
: Isospin 105
6.1 The Group SU
2
. . . . . . . . . . . . . . . . . . . . . . . 105
6.2 Relationship between SU
2
and R
3
. . . . . . . . . . . . . . 107
6.3 SU
2
in Nuclei . . . . . . . . . . . . . . . . . . . . . . . . 109
6.4 Tensor Operators . . . . . . . . . . . . . . . . . . . . . . . 111
7 The Point Groups 113
7.1 Crystal Symmetry . . . . . . . . . . . . . . . . . . . . . . 113
7.2 Schoenﬂies Notation . . . . . . . . . . . . . . . . . . . . . 113
7.3 Commuting Operations . . . . . . . . . . . . . . . . . . . 114
7.4 Enumeration of Point Groups . . . . . . . . . . . . . . . . 115
7.5 Improper Point Groups . . . . . . . . . . . . . . . . . . . 118
7.6 DoubleGroup Representations . . . . . . . . . . . . . . . 118
7.7 Space Group . . . . . . . . . . . . . . . . . . . . . . . . . 120
8 The Group SU
3
125
8.1 Inﬁnitesimal Operators and Lie Algebra . . . . . . . . . . . 125
8.2 Subgroups . . . . . . . . . . . . . . . . . . . . . . . . . . 127
8.3 Multiplet Structure . . . . . . . . . . . . . . . . . . . . . . 127
8.4 Example . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8.5 Product Representations . . . . . . . . . . . . . . . . . . . 132
8.6 SU
3
Multiplets and Hadrons . . . . . . . . . . . . . . . . . 134
8.7 Casimir Operators . . . . . . . . . . . . . . . . . . . . . . 135
Chapter 1
Abstract Group Theory
1.1 Group
A group is a set of elements that have the following properties:
1. Closure: if a and b are members of the group, c = ab is also a
member of the group.
2. Associativity: (ab)c = a(bc) for all a, b, c in the group.
3. Unit element: there is an element e such that ea = a for every
element a in the group.
4. Inverse: to every element a there is a corresponding inverse element
a
−1
in the group such that a
−1
a = e.
The following properties follow from the above deﬁnition:
1. Left cancellation: If ax = ay then x = y for all a in the group.
Proof: ax = ay → a
−1
(ax) = a
−1
(ay) → (a
−1
a)x = (a
−1
a)y →
ex = ey →x = y.
2. Unit element on the right: ae = a = ea.
Proof: a
−1
(ae) = (a
−1
a)e = ee = e = a
−1
a and using the left
cancellation law we have ae = a.
1
2 CHAPTER 1. ABSTRACT GROUP THEORY
3. Inverse element on the right: aa
−1
= e = a
−1
a.
Proof: a
−1
(aa
−1
) = (a
−1
a)a
−1
= ea
−1
= a
−1
= a
−1
e. Using the
left cancellation law, aa
−1
= e.
4. Right cancellation: If xa = ya then x = y for all a in the group.
Proof: xa = ya → (xa)a
−1
= (ya)a
−1
→ x(aa
−1
) = y(aa
−1
) →
xe = ye →x = y.
We note the importance of associativity in the above proofs.
The following identity is often useful:
(ab)
−1
= b
−1
a
−1
which follows from (ab)
−1
(ab) = 1 →(ab)
−1
a = b
−1
→(ab)
−1
= b
−1
a
−1
.
1.2 Abelian Group
If a group has a further property that ab = ba for all a, b in the group, the
group is called Abelian.
1.3 Subgroup
A subgroup is a set of elements within a group which forms a group by
itself. Evidently, the unit element forms a subgroup by itself.
We note that the deﬁnition of a group does not require all elements to
be distinct from one another. However, when two elements are identical,
then one of them is redundant so that we usually assume that all elements
are diﬀerent. A group with n identical elements evidently has at least n
identical subgroups.
1.4 Examples
1. Integers under addition. The unit element e = 0 and the inverse of
an element a is a
−1
= −a. This group is Abelian and inﬁnite.
1.5. MULTIPLICATION TABLE 3
2. A set of all n n unitary matrices U (U
†
= U
−1
) under matrix
multiplication. The unit element is the unit matrix and the inverse
of U is U
†
by deﬁnition. We have to show that a product of two
unitary matrices is unitary .
(U
1
U
2
)
†
= U
†
2
U
†
1
= U
−1
2
U
−1
1
= (U
1
U
2
)
−1
3. The set of permutation operations that take ABC into ABC, ACB,
BAC, CAB, BCA, CBA. The elements of the group are
e =
A B C
A B C
α =
A B C
B C A
β =
A B C
C A B
γ =
A B C
A C B
µ =
A B C
C B A
ν =
A B C
B A C
The operation αβ means: ﬁrst do permutation β and then permuta
tion α on the previous result. We show below that every permutation
has an inverse permutation and two successive permutations corre
spond to a single permutation i.e. the permutations form a group.
We see from the above examples that ”multiplication” can mean addi
tion, matrix multiplication etc. or simply that one operation is performed
on the result of the preceeding operation like in the example 3.
1.5 Multiplication Table
The group multiplication table is a matrix M
ab
= ab, where a and b are
elements of the group and the matrix element corresponding to row a and
column b is given by the group multiplication ab. We illustrate this deﬁni
tion by constructing the multiplication table of the permutation group in
the example 3 above.
4 CHAPTER 1. ABSTRACT GROUP THEORY
e α β γ µ ν
e e α β γ µ ν
α α β e µ ν γ
β β e α ν γ µ
γ γ ν µ e β α
µ µ γ ν α e β
ν ν µ γ β α e
Thus, for example, the inverse of α is β and vice versa. The group is
clearly not Abelian because e.g. αγ = γα. The elements (e, γ) form a
subgroup, so do (e, µ), (e, ν) and (e, α, β). Obviously, the unit element e
forms a subgroup by itself.
The group multiplication table mathematically characterises the group.
All groups with the same multiplication table are mathematically identical
with respect to their group properties. The multiplication table has the
following properties:
1. Rearrangement Theorem: In every row or column, each element must
appear once and once only and therefore each row (column) is dif
ferent from any other row (column).
Proof: Suppose element b appears twice in the row a. This means
that ac = b and ad = b where c = d. But this implies that
ac = ad → c = d which is a contradiction. Thus element b cannot
appear more than once and since the size of the row or column is the
same as the number of elements in the group, it follows that each
element must appear once and once only.
2. The multiplication table is symmetric across the diagonal when the
group is Abelian because the elements commute with one another.
1.6 Cyclic Groups
A group of n elements is said to be cyclic if it can be generated from one
element. The elements of the group must be
a, a
2
, a
3
, . . . , a
n
= e
1.7. ORDER OF AN ELEMENT 5
n is called the order of the cyclic group. A cyclic group is evidently Abelian
but an Abelian group is not necessarily cyclic. It is shown below that every
noncyclic group has at least a cyclic subgroup.
Examples of cyclic groups are the subgroups of the permutation group in
the example 3. The subgroup (e, α, β) is the same as (α, α
2
= β, α
3
= e).
1.7 Order of an Element
Let a = e be an element of a group. Form the products a
2
, a
3
, . . .. a
2
must be either e or a diﬀerent element from a because if a
2
= a →a = e.
If a
2
= e we continue forming a
3
. By a similar argument, a
3
must be
either e or a diﬀerent element from a and a
2
. If a, a
2
, . . . , a
n
are distinct
from each other and a
n
= e then n is called the order of element a. These
elements form a cyclic group. Thus every group must have at least one
cyclic subgroup. In the example 3 above, α and β are of order 3 and γ, µ,
and ν are of order 2.
1.8 Properties of Finite Groups
We summarise below the properties of ﬁnite groups.
1. Every element a has a ﬁnite order n such that a
n
= e.
2. Rearrangement Theorem: Multiplying all elements in a group by an
arbitrary element reproduces the group. This has been proven above
in the properties of the multiplication table.
3. Every noncyclic group has at least a cyclic subgroup.
1.9 Cosets
If S is a subgroup of G and a is an element of G not in S, then the sets
aS and Sa are called the left and right cosets of S respectively. a cannot
be a unit element and therefore a coset can never be a group because it
has no unit element. It is also evident that aS or Sa have no element in
common with S for otherwise, a should be included in S. Morevoer, if b is
6 CHAPTER 1. ABSTRACT GROUP THEORY
an element of G which is neither in S nor in aS, then bS has no element
in common with either S or aS.
Proof: Let x and y be in S such that ax = by. We have axy
−1
= b, but
xy
−1
is an element in S and therefore b is in aS which is a contradiction.
Consequently, a ﬁnite group can be factorised as
G = S + aS + bS +. . .
but the factorisation is not unique. It depends on the choice of S, a, b,
. . ., etc. The number of factors must be ﬁnite and the factorisation must
exhaust the group. Thus we have the following theorem:
Let h be the number of elements of a group G and g be the number of
elements of a subgroup S of G. Then
h
g
= integer
As a consequence, if h is prime then G must be cyclic and it has no
subgroup other than the trivial subgroup of e.
1.10 Class
Two elements a and b of a group are conjugate to one another if there is
an element g in the group such that
a = gbg
−1
(1.1)
(Since every element has an inverse, it also follows that b = g
ag
−1
where
g
= g
−1
). The transformation gbg
−1
is often called a similarity transfor
mation. If a and b are conjugate to each other and b and c are conjugate to
each other, then a and c are also conjugate to each other. This is because
a = g
1
bg
−1
1
and b = g
2
cg
−1
2
→a = g
1
g
2
cg
−1
2
g
−1
1
= g
1
g
2
c(g
1
g
2
)
−1
.
A class C is a set of elements which are conjugate to each other. The
unit element evidently forms a class by itself. If G is Abelian, each element
also forms a class by itself. It is clear that a group may be broken up into
classes
G = C
1
+C
2
+ . . .
and an element cannot belong to more than one class.
1.11. ISOMORPHISM AND HOMOMORPHISM 7
As an example, we work out one of the classes of the permutation
group. In working out the classes, it is obviously not necessary to consider
similarity transformations with either the unit element or the element itself.
We use the multiplication table that we have constructed previously:
βαβ
−1
= βαα = ββ = α
γαγ
−1
= γαγ = γµ = β
µαµ
−1
= µαµ = µν = β
ναν
−1
= ναν = νγ = β
Thus α and β belong to the same class. Similarly we can show that (γ, µ, ν)
also form a class.
Elements belonging to the same class usually have the same characteris
tic. In the above example, α and β both correspond to cyclic permutations
whereas γ, µ, and ν correspond to permutations with one member ﬁxed.
1.11 Isomorphism and Homomorphism
Two groups are said to be isomorphic if they have the same multiplication
table, by reordering the elements if necessary. If F is isomorphic to G then
there is a onetoone correspondence between the elements of F and G,
f
i
→ g
i
, such that if f
i
f
j
= f
k
then g
i
g
j
= g
k
. Two groups can only
be isomorphic if they have the same number of elements. As an example,
the permutation group in the example 3 is isomorphic to the symmetry
operations that take an equilateral triangle into itself. This isomorphism
can be seen by labelling the corners of the triangle with A, B, C.
Homomorphism is similar to isomorphism except that the relationship
is manytoone. Two groups G and G
are said to be homomorphic if each
element of G can be associated with some elements of G
: a → a
=
(a
1
, a
2
, . . .), b → b
= (b
1
, b
2
, . . .), c → c
= (c
1
, c
2
, . . .), such that if
ab = c then a
i
b
j
= c
k
i.e. for every i, j, c
k
lies in c
.
8 CHAPTER 1. ABSTRACT GROUP THEORY
1.12 Invariant Subgroup
We prove the following theorem:
Theorem: If S is a subgroup of G and a is an element of G then
S
= aSa
−1
is also a subgroup and it is isomorphic to S. We assume
that the elements of S are distinct.
Proof: We ﬁrst show that the elements of S
are distinct. Let as
i
a
−1
=
as
j
a
−1
and i = j. Then by the cancellation law, s
i
= s
j
, which is a
contradiction. Thus the elements of S
are distinct. We now show that if
s
i
s
j
= s
k
then s
i
s
j
= s
k
. Proof: s
i
s
j
= as
i
a
−1
as
j
a
−1
= as
i
s
j
a
−1
=
as
k
a
−1
= s
k
. Thus S and S
are isomorphic with the same unit element
e
= aea
−1
= e and the inverse of s
i
is s
i
−1
= (as
i
a
−1
)
−1
= as
−1
i
a
−1
=
(s
−1
i
)
.
If S
is the same as S for every a in the group then the subgroup S is
called an invariant subgroup. This implies that the left coset of S is the
same as its right coset, i.e.
aS = Sa
If we break a group into its cosets,
G = S + aS + bS, . . .
with an invariant subgroup S, then these cosets form a group with the unit
element equal to S. Thus aSbS = abSS = abS = cS if ab = c. This
group is called a factor group. It is an example of homomorphism where
the invariant subgroup S is associated with the elements of S and the coset
aS is associated with the elements of aS.
1.13 Direct Product Group
Let F be a group with elements f
i
, i = 1, . . . , h
F
, and G be a group with
elements g
i
, i = 1, . . . , h
G
, such that f
i
g
j
= g
j
f
i
for all i and j. The
direct product group F ⊗G is deﬁned to be the set of all distinct elements
f
i
g
j
. If F and G have no common element, apart from the identity, then
the order of the direct product group will be h
F
h
G
.
We show that F ⊗G is a group.
1.13. DIRECT PRODUCT GROUP 9
1. Closure: (f
i
g
j
)(f
k
g
l
) = (f
i
f
k
)(g
j
g
l
) = f
r
g
s
. Since f
r
is in F and g
s
is in G then by deﬁnition f
r
g
s
is in F ⊗G.
2. Unit element is e
F
e
G
.
3. Inverse of f
i
g
j
is (f
i
g
j
)
−1
= g
−1
j
f
−1
i
= f
−1
i
g
−1
j
.
4. Associativity: obvious.
We have the following theorem:
Theorem: The classes of the direct product group are given by the direct
products of the classes of the individual groups.
Proof: We label the elements of the product group a
ij
= f
i
g
j
. According
to the deﬁnition of a class, we have
C
ij
= a
−1
rs
a
ij
a
rs
, for all r, s
= f
−1
r
g
−1
s
f
i
g
j
f
r
g
s
= (f
−1
r
f
i
f
r
)(g
−1
s
g
j
g
s
)
10 CHAPTER 1. ABSTRACT GROUP THEORY
Chapter 2
Theory of Group
Representations
2.1 Deﬁnitions
A group ¦A, B, C, ...¦ may be represented by a set of square matrices
¦T(A), T(B), T(C), . . .¦. These matrices are said to form a representa
tion for the group if they satisfy the same group multiplication rule:
AB = C →T(A)T(B) = T(C) (2.1)
The identity element E is represented by a unit matrix and each matrix
must have an inverse. Some or all of the matrices may be the same. If
the matrices are diﬀerent, the representation is called faithful (isomorphic).
The order of the matrices is called the dimension of the representation.
2.2 Equivalent Representations
If T is a representation then
T
= S
−1
TS (2.2)
with an arbitrary but nonsingular S is also a representation because
T
(A)T
(B) = S
−1
T(A)SS
−1
T(B)S
11
12 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
= S
−1
T(A)T(B)S
= S
−1
T(AB)S
= T
(AB)
T and T
are equivalent and there are an inﬁnite number of equivalent
representations.
2.3 Reducible and Irreducible Represen
tation
A representation T is reducible if there exists a nonsingular matrix S such
that the equivalent representation T
= S
−1
TS has the form
T
=
¸
¸
¸
¸
¸
¸
T
1
T
2
T
3
.
.
¸
for all elements in the group. T
1
, T
2
, T
3
, . . . are square matrices and the
rest of the elements are zero. The representation is irreducible if it cannot
be reduced into the above form.
2.4 Equivalent Unitary Representation
There are an inﬁnite number of equivalent representations but it is possible
to ﬁnd one which is unitary.
Theorem 1:
Given an arbitrary representation T, there is always an equivalent unitary
representation.
Proof:
Construct a Hermitian matrix
H =
¸
G
T(G)T
†
(G)
2.4. EQUIVALENT UNITARY REPRESENTATION 13
According to matrix algebra, we can always diagonalise a Hermitian matrix
by a unitary transformation
D = U
−1
HU
=
¸
G
U
−1
T(G)T
†
(G)U
=
¸
G
U
−1
T(G)UU
−1
T
†
(G)U
=
¸
G
T
(G)T
†
(G)
U is made up of the eigenvectors of H and D is diagonal by deﬁnition and
its diagonal elements are real and positive.
D
jj
=
¸
G
¸
k
T
jk
(G)T
†
kj
(G)
=
¸
G
¸
k
T
jk
(G)T
∗
jk
(G)
=
¸
G
¸
k
[T
jk
(G)[
2
≥ 0
In fact, it can never be equal to zero because otherwise T
= 0. Since D is
diagonal, we may deﬁne a diagonal matrix D
−1/2
with diagonal elements
equal to D
−1/2
ii
so that
1 = D
−1/2
¸
G
T
(G)T
†
(G)D
−1/2
Using the above result, the matrix T
(G) = D
−1/2
T
(G)D
1/2
can be shown
to be unitary.
T
(G)T
†
(G) = D
−1/2
T
(G)D
1/2
[D
−1/2
¸
G
T
(G
)T
†
(G
)D
−1/2
]
D
1/2
T
†
(G)D
−1/2
= D
−1/2
¸
G
T
(G)T
(G
)T
†
(G
)T
†
(G)D
−1/2
= D
−1/2
¸
G
T
(G)T
(G
)[T
(G)T
(G
)]
†
D
−1/2
= D
−1/2
¸
G
T
(GG
)T
†
(GG
)D
−1/2
14 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
= D
−1/2
¸
G
T
(G
)T
†
(G
)D
−1/2
= 1
The second last step has been obtained by using the Rearrangement The
orem. Thus, given an arbitrary representation T, it is always possible to
construct a unitary representation by forming
T
= D
−1/2
U
−1
TUD
1/2
(2.3)
From now on we assume that we have a unitary representation.
2.5 Schur’s Lemma
Schur’s lemma is used in proving many of the theorems in group theory.
Theorem 2 (Schur’s lemma):
Given that [M, T] = MT −TM = 0 for all elements in the group, then
(a) if T is irreducible, M = cI where c is a constant and I is a unit matrix.
(b) if M = cI then T is reducible.
Proof: We ﬁrst show that it is suﬃcient to prove the theorem for a Her
mitian M.
TM = MT
(TM)
†
= (MT)
†
M
†
T
†
= T
†
M
†
T(M
†
T
†
)T = T(T
†
M
†
)T
TM
†
= M
†
T
since T
†
= T
−1
(unitary). Adding and substracting the ﬁrst and the last
equations, we get
T(M + M
†
) = (M + M
†
)T
and
T i(M −M
†
) = i(M −M
†
) T
Let H
1
= M + M
†
and H
2
= i(M −M
†
) so that [H
1
, T] = [H
2
, T] = 0.
H
1
and H
2
are Hermitian and M = 1/2(H
1
−iH
2
). Thus, if the theorem is
2.5. SCHUR’S LEMMA 15
true for a Hermitian matrix, it must be true for M. We can now assume M
to be Hermitian and perform a unitary transformation so that M becomes
diagonal, D = U
−1
MU, and deﬁne an equivalent representation T
=
U
−1
TU. Then
T
D = U
−1
TUU
−1
MU
= U
−1
TMU
= U
−1
MTU
= U
−1
MUU
−1
TU
= DT
We have to show that the diagonal elements of D are all identical if T, or
equivalently T
, is irreducible. Consider the ij, i = j, element,
T
ij
D
jj
= D
ii
T
ij
T
ij
(D
jj
−D
ii
) = 0 (2.4)
Let us order the diagonal elements of D such that D
ii
≤ D
jj
for i < j
which can always be done by rearranging the columns of U. Suppose
D
11
= D
nn
and D
11
< D
ii
where n ≥ 1 and i > n. Then it follows from
Eq. ( 2.4) that T
must have the form
T
=
X 0
0 Y
where X is an n n matrix, i.e. T
is reducible.
Let us summarise the conclusions: If M = cI then T must be reducible.
If M = cI then obviously it commutes with T whether it is reducible or
irreducible. Hence if T is irreducible then M must be equal to cI for oth
erwise T would be reducible.
Theorem 3:
Given T
α
M = MT
β
, where T
α
and T
β
are irreducible representations of
dimension l
α
and l
β
respectively and M is a rectangular l
α
l
β
matrix,
then
(a) if l
α
= l
β
, M = 0 or
(b) if l
α
= l
β
either M = 0 or [M[ = 0. In the latter case, M will have an
16 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
inverse so that T
α
and T
β
are equivalent.
Proof:
T
α
(G)M = MT
β
(G)
(T
α
(G)M)
†
= (MT
β
(G))
†
M
†
T
α†
(G) = T
β
†
(G)M
†
M
†
T
α−1
(G) = T
β
−1
(G)M
†
MM
†
T
α−1
(G) = MT
β
−1
(G)M
†
MM
†
T
α
(G
−1
) = MT
β
(G
−1
)M
†
MM
†
T
α
(G
−1
) = T
α
(G
−1
)MM
†
Since T
α
is irreducible, it follows from Schur’s lemma (Theorem 2) that
MM
†
= cI. Consider the case l
α
= l
β
.
[MM
†
[ = [cI[; [M[[M
†
[ = [[M[[
2
= c
l
α
If c = 0, then [M[ = 0 i.e. M
−1
exists. It follows that T
α
and T
β
are
equivalent since
M
−1
(T
α
(G)M) = M
−1
(MT
β
(G)) = T
β
(G)
If c = 0, then MM
†
= 0. Let us look at the diagonal elements of MM
†
:
¸
k
M
ik
M
†
ki
= 0
¸
k
M
ik
M
∗
ik
= 0
¸
k
[M
ik
[
2
= 0
Since this is positive deﬁnite, each term must vanish i.e. M = 0. Consider
now the case l
α
= l
β
, l
α
> l
β
. We enlarge M into a l
α
l
α
square matrix
N with the additional elements equal to zero. Then NN
†
= MM
†
= cI.
But [N[ = 0 because one of its columns is zero.
NN
†
= cI →[NN
†
[ = c
l
α
→[N[[N
†
[ = c
l
α
→c = 0
Hence NN
†
= 0 which implies N = 0. Since M is contained in N, then
M = 0.
2.6. ORTHOGONALITY THEOREM 17
2.6 Orthogonality Theorem
We are now in a position to prove the Orthogonality Theorem which is
central in representation theory.
Theorem 4: Orthogonality Theorem
¸
G
T
α
ij
(G)
∗
T
β
kl
(G) = (h/l
α
)δ
αβ
δ
ik
δ
jl
(2.5)
α and β label the irreducible representations, l
α
and l
β
are the dimensions
of these irreducible representations, and h is the number of elements in the
group.
Proof:
Deﬁne a matrix
M =
¸
G
T
α
(G)XT
β
(G
−1
)
where X is an arbitrary matrix and α = β. We want to show that this
matrix satisﬁes the postulates of Theorem 3. Consider
T
α
(A)M =
¸
G
T
α
(A)T
α
(G)XT
β
(G
−1
)
=
¸
G
T
α
(A)T
α
(G)XT
β
(G
−1
)T
β
(A
−1
)T
β
(A)
=
¸
G
T
α
(AG)XT
β
(G
−1
A
−1
)T
β
(A)
=
¸
G
T
α
(AG)XT
β
((AG)
−1
)T
β
(A)
=
¸
G
T
α
(AG)XT
β
−1
(AG)T
β
(A)
=
¸
G
[T
α
(G)XT
β
−1
(G)]T
β
(A)
= MT
β
(A)
We have again made use of the Rearrangement Theorem in the second last
step. According to Theorem 3, M = 0 since we are considering the case
of α = β so that
M
ij
= 0 =
¸
G
¸
kl
T
α
ik
(G)X
kl
T
β
lj
(G
−1
)
18 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
Since X is arbitrary, we may set X
kl
= δ
km
δ
ln
. Then
¸
G
T
α
im
(G)T
β
nj
(G
−1
) =
¸
G
T
α
im
(G)T
β
−1
nj
(G)
=
¸
G
T
α
im
(G)T
β
†
nj
(G)
=
¸
G
T
α
im
(G)T
β
∗
jn
(G)
= 0
When α = β, according to Schur’s lemma (Theorem 2)
M
ij
= c δ
ij
=
¸
G
¸
kl
T
α
ik
(G)X
kl
T
α
lj
(G
−1
)
=
¸
G
T
α
im
(G)T
α
nj
(G
−1
)
Putting i = j and summing over i yields
c l
α
=
¸
G
¸
i
T
α
im
(G)T
α
ni
(G
−1
)
=
¸
G
T
α
nm
(E)
= hδ
nm
Thus we have
¸
G
T
α
im
(G)T
α
nj
(G
−1
) = (h/l
α
)δ
nm
δ
ij
Since T is unitary,
¸
G
T
α
im
(G)T
α
jn
(G)
∗
= (h/l
α
)δ
nm
δ
ij
Eqn. ( 2.5) has the form of a dot product with T
α
ij
(G) as the G’th
component of a vector labelled by α, i and j, in the h dimensional space of
the group elements. This means that the number of distinct labels (α, i, j)
cannot exceed h, i.e.
¸
α
l
2
α
≤ h. It will be proven later that in fact
¸
α
l
2
α
= h
The result is very useful in working out the irreducible representations of a
group.
2.7. THE CHARACTERS OF A REPRESENTATION 19
2.7 The Characters of a Representation
We have seen that a representation is not unique. We recall that if T is a
representation, so is T
= S
−1
TS. Given T and T
, how can we then tell
that they are equivalent ? One way will be to ﬁgure out if there is a matrix
S, such that T
= S
−1
TS. But this is a complicated procedure. What
we are looking for are properties of a matrix which are invariant under a
similarity transformation. The eigenvalues are one of them but they are
too cumbersome to calculate. A simpler quantity is the trace of a matrix
or the sum of the diagonal elements:
χ
α
(G) = Tr T
α
(G) =
¸
i
T
α
ii
(G) (2.6)
We show that the trace or character of a matrix is invariant under a simi
larity transformation:
Tr T
=
¸
i
¸
jk
S
−1
ij
T
jk
S
ki
=
¸
jk
(
¸
i
S
ki
S
−1
ij
)T
jk
=
¸
jk
δ
kj
T
jk
= Tr T
The characters of a representation α is a set of h numbers ¦χ
α
(G)¦. It
will be shown later that if the representation is irreducible, its characters
are unique.
Theorem 5:
If elements A and B of a group belong to the same class, then the char
acters of their representations are the same.
Proof:
Since A and B belong to the same class, there is an element C such that
A = C
−1
BC. Consequently,
Tr T(A) = Tr [T(C
−1
) T(B) T(C)]
20 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
= Tr [T(C) T(C
−1
) T(B)]
= Tr [T(E) T(B)]
= Tr T(B)
Theorem 6: First Orthogonality Relation
¸
G
χ
α
(G)
∗
χ
β
(G) =
¸
k
χ
α
(C
k
)
∗
χ
β
(C
k
)N
k
= hδ
αβ
(2.7)
C
k
labels a class with N
k
elements.
Proof:
Setting i = j and k = l in Eqn. ( 2.5) (Orthogonality Theorem) and
summing over i and l we have
¸
G
χ
α
(G)
∗
χ
β
(G) = (h/l
α
)δ
αβ
¸
il
δ
il
δ
il
¸
k
χ
α
(C
k
)
∗
χ
β
(C
k
)N
k
= hδ
αβ
Eqn. ( 2.7) has the form of a weighted dot product with weight N
k
in a
space with dimension equal to the number of classes of the group. χ
α
(C
k
)
is the k’th component of vector α. It follows that the number of irreducible
representations must be less than or equal to the number of classes.
2.8 The Regular Representation
We prove the following theorem:
Theorem 7:
¸
α
l
2
α
= h (2.8)
Proof:
2.8. THE REGULAR REPRESENTATION 21
A proof of this theorem is provided by considering the so called regular
representation. The regular representation is formed in the following way.
We write the multiplication table as follows:
e α β γ . . .
e e
α
−1
e
β
−1
e
γ
−1
e
We form the regular representation of an element a by writing 1 wher
ever the element a occurs in the table and zero otherwise. Let the group
elements be a
i
, i = 1, 2, . . . , h, then the regular representation is given by
T
jk
(a
i
) = 1 if a
−1
j
a
k
= a
i
= 0 otherwise
It is clear from the deﬁnition that the representation is faithful. To show
that it is really a representation, we have to prove that
¸
k
T
jk
(a
i
)T
kl
(a
m
) = T
jl
(a
p
)
if and only if a
i
a
m
= a
p
. By deﬁnition, T
jk
(a
i
) = 0 only when k is such that
a
k
= a
j
a
i
and similarly T
kl
(a
m
) = 0 only when k is such that a
k
= a
l
a
−1
m
.
Therefore
¸
k
T
jk
(a
i
)T
kl
(a
m
) = 1 if and only if a
k
= a
j
a
i
= a
l
a
−1
m
and
zero otherwise. However, a
j
a
i
= a
l
a
−1
m
implies a
−1
j
a
l
= a
i
a
m
= a
p
.
The character is given by
χ(a
i
) =
h
¸
j=1
T
jj
(a
i
) = h if a
i
= e
= 0 otherwise
This is clear by inspection of the multiplication table. We can now prove
Theorem 7 by writing the character of the regular representation as a sum
over characters of the irreducible representations:
χ(a
i
) =
¸
α
m
α
χ
α
(a
i
)
22 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
From the orthogonality relation we have
m
α
=
1
h
¸
i
χ
α∗
(a
i
)χ(a
i
)
Since χ
α
(e) = l
α
, χ(e) = h, and χ(a
i
= e) = 0 it follows that m
α
=
l
α
. Each irreducible representation occurs in the regular representation a
number of times equal to the dimension of the irreducible representation.
On the other hand, h =
¸
α
m
α
l
α
so that
h =
¸
α
l
2
α
This result together with the orthogonality theorem show that there are
exactly h orthogonal vectors T
α
ij
.
Theorem 8: Second Orthogonality Relation
¸
α
χ
α
(C
k
)
∗
χ
α
(C
l
) = (h/N
k
)δ
kl
(2.9)
Proof:
We recall that the matrix representation T
α
ij
can be regarded as a set of or
thogonal vectors in the h dimensional space of the group elements. T
α
ij
(G)
is the component of the vector T
α
ij
along the ”Gaxis”. T
α
ij
themselves can
be regarded as a basis in the space of the group elements since the number
of α, i, j is precisely h. Any vector in this space can therefore be expanded
in T
α
ij
:
χ =
¸
αij
c
αij
T
α
ij
We sum the components of χ along Gaxes which belong to a given class
C
k
,
χ(C
k
) =
¸
G∈C
k
χ(G)
=
¸
G∈C
k
¸
αij
c
αij
T
α
ij
(G)
=
1
h
¸
G
1
¸
G∈C
k
¸
αij
c
αij
T
α
ij
(G
−1
1
GG
1
)
2.8. THE REGULAR REPRESENTATION 23
=
1
h
¸
G
1
¸
G∈C
k
¸
αij
¸
kl
c
αij
T
α
ik
(G
−1
1
)T
α
kl
(G)T
α
lj
(G
1
)
=
¸
G∈C
k
¸
αij
¸
kl
c
αij
δ
ij
δ
kl
T
α
kl
(G)/l
α
=
¸
G∈C
k
¸
αi
c
αii
χ
α
(G)/l
α
=
¸
α
a
α
χ
α
(C
k
)
where a
α
= (N
k
/l
α
)
¸
i
c
αii
. The third step uses the fact that G
−1
1
C
k
G
1
=
C
k
for every G
1
in the group, the ﬁfth step uses the orthogonality theorem,
and the last step uses the fact that the characters of elements belonging
to the same class are identical. Thus the set of vectors ¦χ(C
k
)¦ are lin
early independent because the group elements can be divided uniquely and
completely into distinct classes.
Multiplying χ(C
k
) by χ
β
∗
(C
k
)N
k
and summing over the classes k yields
¸
k
χ(C
k
)χ
β
∗
(C
k
)N
k
=
¸
α
a
α
¸
k
χ
α
(C
k
)χ
β
∗
(C
k
)N
k
=
¸
α
a
α
hδ
αβ
= ha
β
Consequently,
χ(C
l
) =
¸
β
a
β
χ
β
(C
l
)
= (1/h)
¸
β
¸
k
χ(C
k
)χ
β
∗
(C
k
)N
k
χ
β
(C
l
)
0 =
¸
k
χ(C
k
)
(N
k
/h)
¸
β
χ
β
∗
(C
k
)χ
β
(C
l
) −δ
kl
¸
¸
This is true for an arbitrary χ(C
k
) so that the quantity in the square bracket
must vanish which proves the theorem.
Eqn. ( 2.9) has the form of a dot product in a space with dimen
sion equal to the number of irreducible representations. It follows that the
number of classes must be less than or equal to the number of irreducible
representations. On the other hand by Theorem 6 the number of irreducible
24 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
representations must be less than or equal to the number of classes. Thus
we conclude the following corollary:
The number of irreducible representations is equal to the number of
classes.
Theorem 6 and 8 imply that the irreducible representations are uniquely
characterised by their characters and consequently two distinct irreducible
representations cannot have the same set of characters.
2.9 The Character Table
The character table is a square matrix with dimension equal to the number
of classes or irreducible representations:
Q
αk
= χ
α
(C
k
)
It has the form
C
1
= C
E
N
2
C
2
N
3
C
3
. . .
T
1
1 1 1
T
2
l
2
. . . . . .
T
3
l
3
. . . . . .
The row is labelled by the irreducible representations and the column by
the classes.
2.10 Properties of the Character Table
The properties listed below are useful for constructing the character table.
1. The number of irreducible representations is equal to the number of
classes.
2. The ﬁrst column contains the dimension of each representation and
the sum of its square is the total number of elements in the group
(Theorem 7):
¸
α
l
2
α
= h.
2.10. PROPERTIES OF THE CHARACTER TABLE 25
3. As a convention, the ﬁrst row is the identity or unit representation
consisting of 1.
4. Each row or column can be regarded as a vector. In this sense:
(a) The rows are orthogonal with weighting factor N
k
(Theorem 6)
and normalised to h.
(b) The columns are orthogonal and normalised to h/N
k
(Theorem
8).
5. If the character table of a factor group is known, then due to homo
morphism, the character of an element aS in the factor group may
be assigned to the classes which are contained in aS. We recall that
if a group has an invariant subgroup S, then we can form a group
S, aS, bS, . . . where a is not in S, b is not in S and aS etc. This
group is a called a factor group with S as the unit element and it
is homomorphic with the group itself. An invariant subgroup means
that x
−1
Sx = S for every element x in the group and by Theorem
9 discussed in the next section, S must consist wholly of classes.
The homomorphism between the factor group and the group itself
means that if T(aS) is a matrix representation of the element aS of
the factor group then the same matrix may be used to represent all
elements in the coset aS. Thus, the characters of the classes in aS
is just the same as the character of T(aS).
The above ﬁve rules are often suﬃcient to construct the character
table but the following rule, which is described in the next section,
can be of help.
6. The characters of the α representation are related by
N
i
χ
α
(C
i
)N
j
χ
α
(C
j
) = l
α
¸
k
m
ijk
N
k
χ
α
(C
k
) (2.10)
where m
ijk
is the coeﬃcients of class multiplication, C
i
C
j
=
¸
k
m
ijk
C
k
.
26 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
2.11 Class Multiplication
Let C be a collection of classes. From the deﬁnition of a class we have
x
−1
Cx = C
for every element x in the group. This is evident from the fact that all
elements in a class are conjugate to one another and if two elements a and
b are diﬀerent then x
−1
ax = x
−1
bx. Moreover, an element cannot belong
to more than one class. Thus the elements on the right hand side must be
identical to the elements in C.
Theorem 9:
A set of elements C obeying x
−1
Cx = C for all x in the group must be
composed wholly of classes.
Proof:
Suppose R is a set of elements in C which do not form a class. But x
−1
Rx
must be equal to R itself since (C−R) form classes and by deﬁnition they
will never be conjugate to elements in R. Thus R must be a class.
We now consider a product of two classes and here we do count all re
sulting elements even when some are the same.
C
i
C
j
= x
−1
C
i
xx
−1
C
j
x
= x
−1
C
i
C
j
x
By Theorem 9, it follows that C
i
C
j
must consist wholly of classes. There
fore it must be possible to write
C
i
C
j
=
¸
k
m
ijk
C
k
(2.11)
where m
ijk
are integers telling how often the class C
k
appears in the prod
uct C
i
C
j
.
Let us consider the class multiplication in terms of matrix representa
tions. If we let S(C
i
) be the sum of matrices of all elements in the class
C
i
and T(x) be a matrix representation of the element x then we have
T
−1
(x)S(C
i
)T(x) = S(C
i
) or S(C
i
)T(x) = T(x)S(C
i
). If the representa
tion T is irreducible, then it follows from Schur’s Lemma that S(C
i
) = c
i
I.
2.12. DIRECT PRODUCT GROUPS 27
Thus
c
i
c
j
=
¸
k
m
ijk
c
k
Taking the trace of S(C
i
) and assuming we are in the irreducible represen
tation α we get
Tr S(C
i
) = Tr c
i
I = c
i
l
α
On the other hand,
Tr S(C
i
) = N
i
χ
α
(C
i
)
Therefore
c
i
= N
i
χ
α
(C
i
)/l
α
and Eq. ( 2.10) follows.
2.12 Direct Product Groups
If a group is a direct product of two groups, then the irreducible represen
tations and the character table of the product group can be worked out
easily from those of the two individual groups with the help of the following
theorem.
Theorem 10:
If G = G
1
⊗G
2
then T
α⊗β
= T
α
1
⊗T
β
2
is an irreducible representation of
G. Moreover, for all T
α
1
and T
β
2
these are all possible irreducible represen
tations of the product group.
Proof:
We ﬁrst show that T is a representation of G. Let a
1
, a
2
, . . . and b
1
, b
2
, . . .
be the elements of G
1
and G
2
respectively and let c
ij
= a
i
b
j
be the
elements of the direct product group G. Suppose c
ij
c
kl
= c
mn
which
implies that a
i
b
j
a
k
b
l
= a
m
b
n
or a
i
a
k
b
j
b
l
= a
m
b
n
since the elements of
G
1
and G
2
commute. This means that a
i
a
k
= a
m
and b
j
b
l
= b
n
or
T
α
(a
i
)T
α
(a
k
) = T
α
(a
m
) and T
β
(b
j
)T
β
(b
l
) = T
β
(b
n
). Writing the last
two matrix equations in component form, multiplying them and using the
deﬁnition of direct product yield the required result.
Let M be a matrix which commutes with every T
α⊗β
. M may be
written in the form M = M
1
⊗ M
2
where M
1
and M
2
have the same
dimensions as T
α
1
and T
β
2
respectively. The commutivity of M with T
α⊗β
28 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
implies the commutivity of M
1
with T
α
1
and M
2
with T
β
2
. By Schur’s
lemma, M
1
= k
1
I
1
and M
2
= k
2
I
2
and hence M = k
1
k
2
I and no non
constant matrix M exists so that by Schur’s lemma T
α⊗β
is irreducible.
To show the last part of the theorem, we recall that
¸
α
l
2
1α
= h
1
and
¸
β
l
2
2β
= h
2
. Let l
αβ
= l
α
1
l
β
2
be the dimension of the irreducible
representation T
α⊗β
. We have
¸
αβ
l
2
αβ
=
¸
α
l
2
1α
¸
β
l
2
2β
= h
1
h
2
But h
1
h
2
is the order of the direct product group and therefore there are
no more possible irreducible representations other than T
α⊗β
for all T
α
1
and
T
β
2
.
2.13 Examples
1. We construct the character table of the permutation group of three
objects which we have considered previously. There are three classes
C
1
= e, C
2
= (α, β), and C
3
= (γ, µ, ν) and therefore by rule
1 there are three irreducible representations. Rule 2 tells us that
l
2
1
+ l
2
2
+ l
2
3
= 6. There is always an identity representation so we
may take l
1
= 1. Then l
2
= 1 and l
3
= 2 are the only possible
solution and the character table must look like as follows:
C
1
2C
2
3C
3
T
1
1 1 1
T
2
1 a b
T
3
2 c d
To determine a, b, c, and d we need to know one of them and
we can ﬁnd the others by orthogonality (rule 4). The subgroup
S = (e, α, β) = C
1
+ C
2
is an invariant subgroup and the corre
sponding factor group has the multiplication table
S γS
S S γS
γS γS S
2.13. EXAMPLES 29
where γS = µS = νS = (γ, µ, ν) = C
3
. This is a group of two
elements and there are evidently two classes, the ﬁrst class consists
of the unit element S and the second class consists of the element
γS. The number of irreducible representations is two and both must
have dimension one, by rule 2. The irreducible representations of
the factor group must be (1, 1) and (1, −1). Then from rule 5 we
may assign a = 1 and b = −1 (a = 1 and b = 1 gives the iden
tity representation). Note that we have taken a and b and not c
and d because the irreducible representations of the factor group are
one dimensional and T
2
is also a one dimensional representation.
The other two characters c and d can be easily obtained from the
orthogonality between columns.
1 1 + 1 a + 2 c = 0 →c = −1
1 1 + 1 b + 2 d = 0 →d = 0
Finally we have
C
1
2C
2
3C
3
T
1
1 1 1
T
2
1 1 −1
T
3
2 −1 0
2. Abelian groups: Every element in an Abelian group forms a class by
itself. Therefore the number of irreducible representations is simply
equal to the number of elements in the group. Moreover, rule 2
implies that all irreducible representations have dimension one. The
consequences are
(a) Unitarity of representations implies that the numbers represent
ing an Abelian group have modulus one.
30 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
(b) If the number X represents an element x then X
∗
represents
x
−1
.
(c) The character table serves also as a representation table.
As an example we consider a noncyclic group of fourth order with
the following multiplication table
e a b c
e e a b c
a a e c b
b b c e a
c c b a e
a
2
= b
2
= c
2
= e so a, b, and c are represented either by ±1.
If a is represented by 1 then b = c = ±1 because bc = a. If a = −1
then either b = 1 and c = −1 or b = −1 and c = 1. Thus we have
e a b c
T
1
1 1 1 1
T
2
1 1 −1 −1
T
3
1 −1 1 −1
T
4
1 −1 −1 1
3. Cyclic groups: These are Abelian groups which can be generated by
a single element: G = (a, a
2
, . . . , a
h
= e). If a is represented by a
number A
k
then a
2
is represented by A
2
k
etc. and A
h
k
= 1 since e
is always represented by 1. Thus, A
k
must be one of the h roots of
2.13. EXAMPLES 31
unity, i.e. A
k
= exp(i2πk/h) and the character table is given by
e a a
2
. . . a
h−1
T
1
1 1 1 . . . 1
T
2
1 A
1
A
2
1
. . . A
h−1
1
T
3
1 A
2
A
2
2
. . . A
h−1
2
. . . . . . . .
. . . . . . . .
T
h
1 A
h−1
A
2
h−1
. . . A
h−1
h−1
32 CHAPTER 2. THEORY OF GROUP REPRESENTATIONS
Chapter 3
Group Theory in Quantum
Mechanics
3.1 Linear Vector Space
A set ¦r
1
, r
2
, . . .¦ is said to form a linear vector space L if r
i
+ r
j
is
another member in the set for every i, j and cr
i
is also another member
in the set where c is a complex constant. The quantities ¦r
1
, r
2
, . . .¦ are
called vectors and there are an inﬁnite number of them.
A set of vectors ¦v
1
, v
2
, . . . , v
p
¦ is said to be linearly independent if
none of them can be expressed as a linear combination of the others. If a
set of coeﬃcients ¦c
k
¦ can be found such that
p
¸
k=1
c
k
v
k
= 0
then the vectors are said to be linearly dependent. The largest number of
vectors in L which form a linearly independent set is called the dimension
of L. This is the same as the smallest number of vectors needed to describe
every vector in L and these vectors are said to form a basis. We denote
the dimension of L by s and the basis vectors by ¦e
1
, e
2
, . . . , e
s
¦ so that
any vector v may be written as
v =
s
¸
i=1
v
i
e
i
33
34 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
In order to determine the coeﬃcients v
i
from a given vector v we in
troduce the concept of scalar or dot product between two vectors v
1
and
v
2
which we denote by (v
1
, v
2
). The particular deﬁnition of the scalar
product is arbitrary but it must satisfy the following general conditions:
1. (v
1
, v
2
) is a complex number
2. (v
1
, v
2
) = (v
2
, v
1
)
∗
3. (v
1
, cv
2
) = c(v
1
, v
2
) →(cv
1
, v
2
) = (v
2
, cv
1
)
∗
= c
∗
(v
1
, v
2
)
4. (v
1
+v
2
, v
3
) = (v
1
, v
3
) + (v
2
, v
3
)
5. (v
1
, v
1
) ≥ 0. It is only zero when v
1
= 0.
Examples:
1. Vectors in 3dimensional space. The dot product is deﬁned as
(v
1
, v
2
) = v
1
v
2
cosθ where v
1
and v
2
are the lengths of the vec
tors and θ is the angle between the two vectors.
2. Function spaces. The dot product is deﬁned as (ψ
i
, ψ
j
) =
d
3
r w(r)ψ
∗
i
(r)ψ
j
(r) where w is a weight function which is usu
ally real and positive deﬁnite so that the last condition is satisﬁed.
We can also deﬁne orthonormal basis functions ¦φ
i
¦ such that any
function can be expanded in this basis: ψ =
¸
∞
i=1
c
i
φ
i
The linear
space spanned by this basis is called the Hilbert space.
We deﬁne an operator
ˆ
T by
ˆ
Tv = v
where v is an arbitrary vector or function which is carried into another
vector or function v
. Both v and v
are in L. The operator is called linear
if
ˆ
T(v
1
+v
2
) =
ˆ
Tv
1
+
ˆ
Tv
2
ˆ
Tcv = c
ˆ
Tv
3.2. SYMMETRY TRANSFORMATIONS 35
We will consider only linear operators.
Since any vector in L can be expanded in the basis vectors, it is only
necessary to study the eﬀects of the operator on the basis vectors.
ˆ
Te
j
=
¸
i
e
i
T
ij
The matrix T
ij
is said to form a representation for the operator
ˆ
T in the
linear vector space L.
It is usually convenient to choose an orthonormal basis, by this we
mean (φ
i
, φ
j
) = δ
ij
. This can always be done by the GrammSchmidt
orthogonalisation procedure. We illustrate the method for three vectors.
The generalisation to an arbitrary number of vectors is straightforward. Let
φ
1
, φ
2
, φ
3
be three linearly independent functions which are not necessarily
orthogonal nor normalised. Let ψ
1
= φ
1
/
(φ
1
, φ
1
) and construct
˜
ψ
2
=
φ
2
−ψ
1
(ψ
1
, φ
2
) which is orthogonal to ψ
1
. We normalise
˜
ψ
2
to get ψ
2
=
˜
ψ
2
/
(
˜
ψ
2
,
˜
ψ
2
) and construct
˜
ψ
3
= φ
3
− ψ
2
(ψ
2
, φ
3
) − ψ
1
(ψ
1
, φ
3
) which
is orthogonal to both ψ
1
and ψ
2
and which is then normalised to give
ψ
3
=
˜
ψ
3
/
˜
ψ
3
,
˜
ψ
3
) In this way we have constructed an orthonormal set of
functions and we can see that for a general case we have
˜
ψ
n
= φ
n
−
n−1
¸
i=1
ψ
i
(ψ
i
, φ
n
)
3.2 Symmetry Transformations
A symmetry transformation G with respect to a given function f(x, y, z)
is a real linear coordinate transformation that preserves the length (real
unitary transformation)
r
= Gr →r
i
=
¸
j
G
ij
r
j
, r
i
= x, y, z
such that the function calculated with the new variables r
i
by the above
substitution of coordinates is the same as the function calculated with the
old variables: f(x
, y
, z
) = f
(x, y, z) = f(x, y, z) → f(Gr) = f(r). In
36 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
other words, the function looks the same in the old and the new coordinate
system. This also implies that
f(Gr) = f(r) = f(G
−1
r)
because f(r
) = f(r) = f(G
−1
r
) and r
is just a dummy variable.
A symmetry operator
ˆ
T(G) associated with a symmetry transformation
G is deﬁned by
ˆ
T(G)f(r) = f(G
−1
r)
It is important to note that the operator acts upon the coordinates r, and
not upon the argument of f. Thus we mean that
ˆ
T(G)f(Ar) = f(AG
−1
r) = f(G
−1
Ar)
We also have G
−1
instead of G in our deﬁnition. This is a natural choice
because two successive operations
ˆ
T(G
1
) and
ˆ
T(G
2
) on a function f(r)
correspond to the same order of coordinate transformation:
ˆ
T(G
2
)
ˆ
T(G
1
)f(r) =
ˆ
T(G
2
)f(G
−1
1
r) = f(G
−1
1
G
−1
2
r) = f([G
2
G
1
]
−1
r)
Theorem 1:
If ¦G¦ is a set of all symmetry transformations of a function, then the
associated symmetry operators ¦
ˆ
T(G)¦ form a group.
Proof:
The set ¦G¦ evidently form a group because if G is a symmetry transfor
mation, then so is G
−1
as discussed above. A succession of two symmetry
transformations is itself a symmetry transformation and the identity trans
formation is obviously a symmetry transformation. From the deﬁnition, we
have
ˆ
T(G
1
)
ˆ
T(G
2
)f(r) =
ˆ
T(G
1
)f(G
−1
2
r)
= f(G
−1
2
G
−1
1
r)
= f([G
1
G
2
]
−1
r)
=
ˆ
T(G
1
G
2
)f(r)
Hence
ˆ
T(G
1
)
ˆ
T(G
2
) =
ˆ
T(G
1
G
2
) and the symmetry operators form a group.
3.3. INVARIANT SUBSPACE AND ITS GENERATION 37
3.3 Invariant Subspace and Its Genera
tion
A set of functions or vectors ¦φ
i
¦ is said to span an invariant subspace V
S
under a given set of symmetry operators ¦
ˆ
T(G)¦ if
ˆ
T(G)φ
i
is also in V
S
for all G and i.
An invariant subspace can be generated from an arbitrary function f and
a set of symmetry operators ¦
ˆ
T(G)¦ by applying each of the operators on
f. The set of functions ¦
ˆ
T(G)f¦ form an invariant subspace under ¦
ˆ
T(G)¦
because
ˆ
T(G
)[
ˆ
T(G)f] =
ˆ
T(G
G)f, which must lie in the subspace since
ˆ
T(G
G) is just another member of ¦
ˆ
T(G)¦. We note that the number of
functions that span the subspace is not necessarily equal to the number of
G. Some of the functions may be the same or can be expressed as linear
combinations of the others. The largest number of linearly independent
functions in V
S
is called the dimension of the subspace.
It may happen that we can construct a function f
in V
S
such that the
invariant subspace ¦
ˆ
T(G)f
¦ has a smaller dimension than that of V
S
. In
this case the subspace V
S
is said to be reducible. When such a function
cannot be found than V
S
is said to be irreducible.
3.4 Basis Functions for a Representation
Let ¦φ
i
¦ be a set of orthonormal functions that span an invariant subspace
V
S
. Then
ˆ
T(G)φ
i
(r) = φ
i
(G
−1
r) is in V
S
by deﬁnition and it can therefore
be expanded as a linear combination of ¦φ
i
¦:
ˆ
T(G)φ
i
=
¸
j
φ
j
T
ji
(G) (3.1)
If the basis functions are orthonormal, then T(G) is unitary because
(
ˆ
T(G)φ
i
,
ˆ
T(G)φ
j
) = δ
ij
= (
¸
k
φ
k
T
ki
(G),
¸
l
φ
l
T
lj
(G))
=
¸
kl
T
∗
ki
(G)T
lj
(G)(φ
k
, φ
l
)
=
¸
k
T
†
ik
(G)T
kj
(G)
38 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
Theorem 2:
The matrices ¦T(G)¦ form a representation for ¦
ˆ
T(G)¦
Proof:
ˆ
T(G
1
)
ˆ
T(G
2
)φ
i
=
¸
j
ˆ
T(G
1
)φ
j
T
ji
(G
2
)
ˆ
T(G
1
G
2
)φ
i
=
¸
jk
φ
k
T
kj
(G
1
)T
ji
(G
2
)
¸
k
φ
k
T
ki
(G
1
G
2
) =
¸
k
φ
k
T
kj
(G
1
)T
ji
(G
2
)
Thus T(G
1
)T(G
2
) = T(G
1
G
2
). The functions ¦φ
i
¦ are said to form a ba
sis for a representation, which may be reducible. Evidently, if the subspace
is irreducible under ¦
ˆ
T(G)¦, then the representation is also irreducible.
This simply follows from the fact that one always generates the entire sub
space starting from any function in the subspace. If the representation
were reducible, then it would be possible to start from a function that does
not generate the entire subspace, in which case the subspace is reducible.
We now see the relevance of matrix representation theory studied in the
previous chapter. Since the theorems were developed for arbitrary matrices,
they are also applicable here.
Theorem 3:
A change of basis vectors ψ
j
=
¸
i
φ
i
S
ij
corresponds to a similarity trans
formation of the original representation, which is an equivalent representa
tion. S is assumed to be unitary so that (ψ
i
, ψ
j
) = δ
ij
.
Proof:
ˆ
T(G)φ
i
=
¸
j
φ
j
T
ji
(G)
ˆ
T(G)
¸
j
ψ
j
S
−1
ji
=
¸
j
¸
k
ψ
k
S
−1
kj
T
ji
(G)
3.5. PROJECTION OPERATORS 39
Multiplying both sides by S
im
and summing over i yields
ˆ
T(G)ψ
m
=
¸
k
ψ
k
T
km
(G)
where T
km
(G) =
¸
ji
S
−1
kj
T
ji
(G)S
im
(G) or T
= S
−1
TS which is a simi
larity transformation. It is clear that the reverse is also true, i.e. a similarity
transformation of the representation with a unitary matrix S corresponds
to a unitary transformation of the basis vectors.
3.5 Projection Operators
It is possible to construct a projection operator
ˆ
P
β
such that given an arbi
trary function ψ,
ˆ
P
β
ψ is a component of ψ in the irreducible subspace β,
i.e.
ˆ
P
β
ψ transforms according to the irreducible representation β. This is a
very useful tool. In the following, we construct these projection operators.
Suppose ¦ψ
α
i
¦ span an irreducible subspace transforming according to
the irreducible representation α of ¦
ˆ
T(G)¦. That is
ˆ
T(G)ψ
α
i
=
¸
j
ψ
α
j
T
α
ji
(G)
Multiplying by T
β∗
kl
(G) and summing over G yields
¸
G
T
β∗
kl
(G)
ˆ
T(G)ψ
α
i
=
¸
G
¸
j
ψ
α
j
T
α
ji
(G)T
β∗
kl
(G)
Using the Orthogonality Theorem on the right side, we get
¸
G
T
β∗
kl
(G)
ˆ
T(G)ψ
α
i
=
¸
j
ψ
α
j
(h/l
β
)δ
jk
δ
il
δ
αβ
= (h/l
β
)δ
il
δ
αβ
ψ
α
k
We deﬁne
ˆ
P
β
kl
≡ (l
β
/h)
¸
G
T
β
∗
kl
(G)
ˆ
T(G)
so that
ˆ
P
β
kl
ψ
α
i
= δ
il
δ
αβ
ψ
β
k
→
ˆ
P
α
ki
ψ
α
i
= ψ
α
k
(3.2)
40 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
Starting from a given function that belongs to the ith row of a given
irreducible representation α, the projection operator can be used to gener
ate the other ”partner functions” corresponding to the other rows. These
functions form a basis for the irreducible representation α.
Moreover, we can also deﬁne
ˆ
P
β
≡
¸
i
ˆ
P
β
ii
≡ (l
β
/h)
¸
G
χ
β∗
(G)
ˆ
T(G)
so that
ˆ
P
β
ψ
α
i
= δ
βα
ψ
α
i
(3.3)
The projection operators allow us to decompose the space, in which the
symmetry operators are deﬁned, into irreducible subspaces. Any function
ψ can therefore be written as follows
ψ =
¸
α
¸
i
ψ
α
i
(3.4)
where ψ
α
i
belongs to the ith row of the irreducible representation α. We
note also that if φ
α
i
and ψ
α
i
belong to the ith row of the irreducible repre
sentation α, so does any linear combination aφ
α
i
+bψ
α
i
.
Suppose that a given representation T is reducible. Then there is a uni
tary matrix S such that T
= S
−1
TS =
¸
α
⊕m
α
T
α
has the block form
and each block cannot be reduced any further. α labels the irreducible rep
resentation and m
α
tells us how many times the irreducible representation
α appears in the reduction. We note that when m
α
> 1 the irreducible
representations T
α
are assumed to be the same. This can always be done
by means of similarity transformations in each subspace which transforms
according to the same irreducible representation α. The new basis vectors
according to Theorem 3 are given by ψ
j
=
¸
i
φ
i
S
ij
and they may be la
belled according to the irreducible subspaces to which they belong. Any
vector in V
S
can therefore be expanded as follows:
v =
¸
α,t,i
ψ
αt
i
c
αt
i
=
¸
α,i
v
α
i
(3.5)
3.6. ORTHOGONALITY OF BASIS FUNCTIONS 41
where t labels the diﬀerent possible subspaces transforming according to
the same irreducible representation α, which is necessary when m
α
> 1,
and i labels the basis functions in the subspace α, t. We have also deﬁned
v
α
i
=
¸
t
ψ
αt
i
c
αt
i
. However,
ˆ
T(G)v
α
i
is not necessarily equal to
¸
j
v
α
j
T
ji
(G)
unless there is only one t.
3.6 Orthogonality of Basis Functions
Theorem 4:
Two vectors belonging to two inequivalent irreducible representations must
be orthogonal. If they transform according to the same irreducible repre
sentation, they are still orthogonal if they belong to diﬀerent rows.
Proof:
Let φ
α
i
and ψ
β
j
belong to the irreducible representations T
α
and T
β
respec
tively. Then
ˆ
T(G)φ
α
i
=
¸
k
φ
α
k
T
α
ki
(G)
ˆ
T(G)ψ
β
j
=
¸
l
ψ
β
l
T
β
lj
(G)
Recalling that T is or may always be chosen to be unitary, we have
(φ
α
i
, ψ
β
j
) = (
ˆ
T(G)φ
α
i
,
ˆ
T(G)ψ
β
j
)
=
¸
kl
T
α∗
ki
(G)T
β
lj
(G)(φ
α
k
, ψ
β
l
)
= (1/h)
¸
G
¸
kl
T
α∗
ki
(G)T
β
lj
(G)(φ
α
k
, φ
β
l
)
= (1/l
α
)δ
αβ
δ
ij
¸
k
(φ
α
k
, ψ
β
k
)
3.7 Examples
As an example, we consider the group D
3
which is the group of permu
tations of three objects. The elements are e = unit element, c
1
, c
2
=
anticlockwise rotations about the z−axis by 2π/3 and 4π/3 respectively,
and d
1
, d
2
, d
3
= rotations by π about axes lebelled 1, 2, 3 in Figure 3.1.
42 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
The classes are C
e
= e, C
c
= (c
1
, c
2
), and C
d
= (d
1
, d
2
, d
3
). We choose
a set of basis vectors or functions in order to form a representation for the
group elements. The choice is arbitrary.
1. The ﬁrst obvious choice is the three Cartesian unit vectors e
x
, e
y
, e
z
as shown in the ﬁgure. Let us see how they transform under the
group operations.
e
x
e
y
e
z
ˆ
T(e) e
x
e
y
e
z
ˆ
T(c
1
) −
1
2
e
x
+
√
3
2
e
y
−
√
3
2
e
x
−
1
2
e
y
e
z
ˆ
T(c
2
) −
1
2
e
x
−
√
3
2
e
y
√
3
2
e
x
−
1
2
e
y
e
z
ˆ
T(d
1
) −e
x
e
y
−e
z
ˆ
T(d
2
)
1
2
e
x
+
√
3
2
e
y
√
3
2
e
x
−
1
2
e
y
−e
z
ˆ
T(d
3
)
1
2
e
x
−
√
3
2
e
y
−
√
3
2
e
x
−
1
2
e
y
−e
z
We see that applying the group operations on e
x
or e
y
generate
an invariant subspace (e
x
, e
y
) whereas e
z
forms an invariant sub
space by itself. The set of Cartesian unit vectors (e
x
, e
y
, e
z
) do form
a basis for the representation of the group. The matrices can be
calculated by using the deﬁnition T
ij
(G) = (φ
i
,
ˆ
T(G)φ
j
):
T(e) =
¸
¸
1 0 0
0 1 0
0 0 1
¸
T(c
1
) =
¸
¸
−
1
2
−
√
3
2
0
√
3
2
−
1
2
0
0 0 1
¸
T(c
2
) =
¸
¸
−
1
2
√
3
2
0
−
√
3
2
−
1
2
0
0 0 1
¸
T(d
1
) =
¸
¸
−1 0 0
0 1 0
0 0 −1
¸
T(d
2
) =
¸
¸
1
2
√
3
2
0
√
3
2
−
1
2
0
0 0 −1
¸
T(d
3
) =
¸
¸
1
2
−
√
3
2
0
−
√
3
2
−
1
2
0
0 0 −1
¸
The characters are χ(C
e
) = 3, χ(C
c
) = 0, and χ(C
d
) = −1. As can
be seen, the 33 representation is reducible into a 22 and a 11
3.7. EXAMPLES 43
representations. The 11 representation is obviously irreducible, but
it may be possible to further reduce the 2 2 representation. We
recall that the character of a given representation can be written as
χ =
¸
α
m
α
χ
α
and by using the ﬁrst orthonormality relation between
characters the coeﬃcients m
α
can be calculated as follows:
m
α
=
1
h
¸
k
N
k
χ
α∗
(C
k
)χ(C
k
)
For convenience, we reproduce the character table of the group D
3
.
C
e
2C
c
3C
d
T
1
1 1 1
T
2
1 1 −1
T
3
2 −1 0
Thus
m
1
=
1
6
[1 1 3 + 2 1 0 + 3 1 (−1)] = 0
m
2
=
1
6
[1 1 3 + 2 1 0 + 3 (−1) (−1) = 1
m
3
=
1
6
[1 2 3 + 2 (−1) 0 + 3 0 (−1) = 1
Hence the representation breaks up into T = T
2
⊕T
3
and therefore
(e
x
, e
y
) do form an irreducible subspace for the representation T
3
.
We could have arrived at this conclusion simply by calculating the
characters of the 22 representation which are χ(C
e
) = 2, χ(C
c
) =
−1, and χ(C
d
) = 0 and which agree with the characters of the T
3
representation.
2. As a second example, we consider the same group but we choose a
basis which consists of functions rather than the Cartesian vectors.
We can take an arbitrary function, say x
2
, and generate an invariant
subspace from it. We construct an orthonormal basis that spans this
subspace and we form the representation for the group. We then use
44 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
the projection operator method to ﬁgure out which functions belong
to a given irreducible subspace.
What we need to know is how x transforms under the group op
erations. We have deﬁned
ˆ
T(G)f(r) = f(G
−1
r), i.e.
ˆ
T(G)x =
ˆ
T(G)(e
x
r) = (e
x
G
−1
r) = (Ge
x
r). Thus x, y and z transform
like the Cartesian vectors e
x
, e
y
, and e
z
and we have
ˆ
T(e)x
2
= x
2
ˆ
T(c
1
)x
2
= (−
1
2
x +
√
3
2
y)
2
=
1
4
x
2
+
3
4
y
2
−
√
3
2
xy
ˆ
T(c
2
)x
2
= (−
1
2
x −
√
3
2
y)
2
=
1
4
x
2
+
3
4
y
2
+
√
3
2
xy
ˆ
T(d
1
)x
2
= (−x)
2
= x
2
ˆ
T(d
2
)x
2
= (
1
2
x +
√
3
2
y)
2
=
1
4
x
2
+
3
4
y
2
+
√
3
2
xy
ˆ
T(d
3
)x
2
= (
1
2
x −
√
3
2
y)
2
=
1
4
x
2
+
3
4
y
2
−
√
3
2
xy
These six functions form an invariant subspace for D
3
but all of them
can be written as linear combinations of only three distinct functions
χ
1
= x
2
, χ
2
= y
2
, and χ
3
= xy. Evidently none of these three
functions can be written as a linear combination of the other two so
that they are linearly independent and form a basis for the invariant
subspace, although they are neither orthogonal nor normalised.
We now use the GrammSchmidt orthogonalisation procedure to con
struct a set of orthonormal basis needed to construct a unitary repre
sentation for the group. We can start with any of the three functions
or any linear combination of them. Let us start with χ
1
= x
2
which
we normalise inside a unit sphere so that
[A[
2
(χ
1
, χ
1
) =
1
0
dr r
2
π
0
dθsin θ
2π
0
dφ χ
∗
1
χ
1
= 1
which gives A =
35/4π. The normalised function is then
φ
1
= Ax
2
3.7. EXAMPLES 45
The second function is constructed by using
˜
φ
2
= χ
2
− φ
1
(φ
1
, χ
2
)
which yields after normalisation
φ
2
= A
1
2
√
2
(3y
2
−x
2
)
The function xy is already orthogonal to both x
2
and y
2
due to
antysymmetry so that the third orthonormalised function is
φ
3
= A
√
3xy
To construct the representation matrices, we need to know how these
basis functions transform into one another under the group opera
tions.
φ
1
φ
2
φ
3
ˆ
T(e) φ
1
φ
2
φ
3
ˆ
T(c
1
)
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
1
√
2
(φ
1
+ φ
3
)
1
2
φ
1
−
1
√
2
φ
2
−
1
2
φ
3
ˆ
T(c
2
)
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
1
√
2
(φ
1
−φ
3
) −
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
ˆ
T(d
1
) φ
1
φ
2
−φ
3
ˆ
T(d
2
)
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
1
√
2
(φ
1
−φ
3
)
1
2
φ
1
−
1
√
2
φ
2
+
1
2
φ
3
ˆ
T(d
3
)
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
1
√
2
(φ
1
+ φ
3
) −
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
The matrix representations can be easily obtained and these are
T(e) =
¸
¸
1 0 0
0 1 0
0 0 1
¸
T(c
1
) =
¸
¸
¸
1
2
1
√
2
1
2
1
√
2
0 −
1
√
2
−
1
2
1
√
2
−
1
2
¸
T(c
2
) =
¸
¸
¸
1
2
1
√
2
−
1
2
1
√
2
0
1
√
2
1
2
−
1
√
2
−
1
2
¸
T(d
1
) =
¸
¸
1 0 0
0 1 0
0 0 −1
¸
T(d
2
) =
¸
¸
¸
1
2
1
√
2
1
2
1
√
2
0 −
1
√
2
1
2
−
1
√
2
1
2
¸
T(d
3
) =
¸
¸
¸
1
2
1
√
2
−
1
2
1
√
2
0
1
√
2
−
1
2
1
√
2
1
2
¸
46 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
Unlike the previous example, these matrices are not in block form but
we know that they are reducible because there is no three dimensional
representation of D
3
. Let us ﬁgure out which irreducible representations
are contained in these matrices by using the same formula as in the pre
vious example. The reducible characters are χ(C
e
) = 3, χ(C
c
) = 0, and
χ(C
d
) = 1. We get
m
1
=
1
6
[1 1 3 + 2 1 0 + 3 1 1] = 1
m
2
=
1
6
[1 1 3 + 2 1 0 + 3 (−1) 1] = 0
m
3
=
1
6
[1 2 3 + 2 (−1) 0 + 3 0 1] = 1
Hence the representation breaks up into T = T
1
⊕T
3
.
Since the representation contains an identity representation, it must be
possible to construct a linear combination of φ
1
, φ
2
, and φ
3
such that it
is invariant under the group operations. We do this in a systematic way
using the projection operator method. The projection operator is given by
ˆ
P
α
=
l
α
h
¸
G
χ
α∗
(G)
ˆ
T(G)
which when applied to an arbitrary function gives the component of the
function in the irreducible subspace α. Let us apply
ˆ
P
1
to φ
1
, φ
2
, and φ
3
:
ˆ
P
1
φ
1
=
1
6
¸
φ
1
+ (
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
) + (
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
)
+φ
1
+ (
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
) + (
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
)
¸
=
√
2
3
(
√
2φ
1
+φ
2
)
ˆ
P
1
φ
2
=
1
6
¸
φ
2
+ (
1
√
2
(φ
1
+ φ
3
)) + (
1
√
2
(φ
1
−φ
3
))
+φ
2
+ (
1
√
2
(φ
1
−φ
3
)) + (
1
√
2
(φ
1
+φ
3
))
¸
3.8. RELATIONSHIP TO QUANTUM MECHANICS 47
=
1
3
(
√
2φ
1
+φ
2
)
ˆ
P
1
φ
3
= 0
The linear combination (
√
2φ
1
+φ
2
)/
√
3 transforms according to the iden
tity representation. The component of φ
1
in the third irreducible subspace
is given by
ˆ
P
3
φ
1
=
2
6
¸
2φ
1
−(
1
2
φ
1
+
1
√
2
φ
2
−
1
2
φ
3
) −(
1
2
φ
1
+
1
√
2
φ
2
+
1
2
φ
3
)
¸
=
1
3
(φ
1
−
√
2φ
2
)
Thus we may choose a new basis
ψ
1
=
1
√
3
(
√
2φ
1
+ φ
2
)
ψ
2
=
1
√
3
(φ
1
−
√
2φ
2
)
ψ
3
= φ
3
which brings the representation into block form with ψ
1
transforming ac
cording to the identity representation T
1
and (ψ
2
, ψ
3
) transforming accord
ing to the irreducible representation T
3
. The new basis may be written as
ψ
i
=
¸
j
φ
j
S
ji
where S is the unitary matrix that reduces the representa
tion into block form which is given by
S =
¸
¸
¸
2
3
1
3
0
1
3
−
2
3
0
0 0 1
¸
3.8 Relationship to Quantum Mechanics
A basic problem in quantum mechanics is ﬁnding the eigenvectors and
eigenvalues of a Hamiltonian
ˆ
H. A general procedure is to calculate the
Hamiltonian matrix in an arbitrary basis and diagonalise it, but in most
cases, this procedure is too diﬃcult to carry out. Instead one tries to ﬁnd
48 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
a basis in which the oﬀ diagonal elements of H are small or zero so that
they can be neglected or treated perturbatively. It is here that the study
of the symmetry of the Hamiltonian is of great value because group theory
can tell us which matrix elements are zero and how the matrix elements
are related to one another. The results are exact and depend only on the
symmetry of the systems and not on the particular form of interactions or
potentials.
3.8.1 The Group of the Schr¨ odinger Equation
If ¦G¦ is a set of all symmetry transformations of a Hamiltonian,
ˆ
H, then
according to Theorem 1 the associated symmetry operators ¦
ˆ
T(G)¦ form
a group which is called the group of the Schr¨odinger equation or in short
the Schr¨odinger group.
Theorem 5:
Each element of the Schr¨odinger group commutes with the Hamiltonian.
Moreover, if ψ(r) is an eigenvector of
ˆ
H, then
ˆ
T(G)ψ(r) = ψ(G
−1
r) is
also an eigenvector of
ˆ
H with the same eigenvalue.
Proof:
By deﬁnition
ˆ
T(G)
ˆ
H(r) =
ˆ
H(G
−1
r) =
ˆ
H(r)
Let us investigate the consequences of this symmetry by considering the
eﬀect of
ˆ
T(G) on
ˆ
Hψ(r) where ψ is an arbitrary function .
ˆ
T(G)
ˆ
H(r)ψ(r) =
ˆ
H(G
−1
r)ψ(G
−1
r)
=
ˆ
H(r)
ˆ
T(G)ψ(r)
ˆ
T(G)
ˆ
H(r) −
ˆ
H(r)
ˆ
T(G)
ψ(r) = 0
Since ψ(r) is arbitrary, [
ˆ
T,
ˆ
H] = 0. The second part of the theorem follows
directly from this result:
ˆ
H(r)
ˆ
T(G)ψ(r) =
ˆ
T(G)
ˆ
H(r)ψ(r) =
ˆ
T(G)ψ(r)
3.8. RELATIONSHIP TO QUANTUM MECHANICS 49
3.8.2 Degeneracy and Invariant Subspace
The implication of Theorem 5 is that a degenerate set of eigenfunctions
of
ˆ
H form an invariant subspace under the symmetry operations of the
Schr¨odinger group and therefore form a basis for a representation of the
group. Let ¦ψ
i
¦ be a set of degenerate eigenfunctions with energy . Since
ˆ
T(G)ψ
i
(r) = ψ
i
(G
−1
r) is also an eigenfunction with the same energy , it
must be possible to expand
ˆ
T(G)ψ
i
(r) as a linear combination of ¦ψ
i
¦:
ˆ
T(G)ψ
i
=
¸
j
ψ
j
T
ji
(G)
The matrices ¦T(G)¦ form a representation for the Schr¨odinger group.
The representation is normally irreducible but if it is reducible, then
there are two possibilities:
1. There is an accidental degeneracy so that the subspace can be broken
up into two or more irreducible subspaces. A vector in one irreducible
subspace evidently cannot be brought into another irreducible sub
space by the symmetry operators.
2. The group is a subgroup of a larger group.
In practice, accidental degeneracies my be lifted up by varying parameters
in the Hamiltonian without changing the symmetry of it. If the subspace
is irreducible, the degeneracy is termed ”normal”.
Assuming that there is no accidental degeneracy, each degenerate set
of eigenfunctions of
ˆ
H form a basis for an irreducible representation of
the Schr¨odinger group. We may label each eigenfunction by the row and
the irreducible representation to which it belongs. In this sense, the row
index and the irreducible representation provide ”quantum numbers”. The
degeneracy is simply given by the dimensionality of the irreducible repre
sentation. Thus by knowing the symmetry of the Hamiltonian, we can tell
the degrees of degeneracy possible in any problem. It follows that these de
generacies can only be broken by a perturbation if it has a lower symmetry
than the unperturbed Hamiltonian. Moreover, by knowing the symmetry
of the perturbation, we can also tell how the degeneracies are split.
50 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
3.8.3 Partial Diagonalisation of
ˆ
H
Theorem 6:
The matrix element (φ
α
i
,
ˆ
Hψ
β
j
) is zero unless α = β and i = j.
Proof:
(φ
α
i
,
ˆ
Hψ
β
j
) = (
ˆ
T(G)φ
α
i
,
ˆ
T(G)
ˆ
Hψ
β
j
)
= (
ˆ
T(G)φ
α
i
,
ˆ
H
ˆ
T(G)ψ
β
j
)
=
¸
kl
T
α∗
ki
(G)T
β
lj
(G)(φ
α
k
,
ˆ
Hψ
β
l
)
=
1
h
¸
G
¸
kl
T
α∗
ki
(G)T
β
lj
(G)(φ
α
k
,
ˆ
Hψ
β
l
)
=
1
l
α
¸
kl
δ
kl
δ
ij
δ
αβ
(φ
α
k
,
ˆ
Hψ
β
l
)
=
1
l
α
δ
αβ
δ
ij
¸
k
(φ
α
k
,
ˆ
Hψ
β
k
)
This result is very useful because by choosing basis functions with the
appropriate symmetry, the Hamiltonian is partially diagonalised with very
little eﬀort. Also, (φ
α
i
,
ˆ
Hψ
β
i
) is independent of i. This is a special case of
the so called WignerEckart theorem discussed later.
3.9 Irreducible Sets of Operators
We have studied the concept of irreducible subspace formed by a set of
functions which transform among themselves under the group operations.
A similar concept may be extended to operators and we deﬁne an irreducible
set of operators
ˆ
O
α
i
by the closure property
ˆ
O
α
i
=
ˆ
T(G)
ˆ
O
α
i
ˆ
T
−1
(G) =
¸
j
ˆ
O
α
j
T
α
ji
(G) (3.6)
Thus the irreducible set of operators
ˆ
O
α
i
transform among themselves under
the irreducible representation T
α
. The number of operators in such a set
is equal to the dimension l
α
of the irreducible representation. An example
of an irreducible set of operators is given by the momentum operators p
x
,
p
y
, and p
z
under rotation.
3.10. DIRECT PRODUCT REPRESENTATIONS OF A GROUP 51
3.10 Direct Product Representations of a
Group
The direct product of an n n matrix A and and m m matrix B is
deﬁned by
(A ⊗B)
ii
,jj
= A
ij
B
i
j
(3.7)
(A⊗B) is an (mn mn) matrix and it is not the same as a matrix mul
tiplication of A and B.
Theorem 7:
If T
α
and T
β
are two representations of a group, then T
α⊗β
= T
α
⊗T
β
is
also a representation of the group.
Proof:
[T
α⊗β
(G
1
)T
α⊗β
(G
2
)]ii
, jj
=
¸
kk
T
α⊗β
ii
,kk
(G
1
)T
α⊗β
kk
,jj
(G
2
)
=
¸
kk
T
α
ik
(G
1
)T
β
i
k
(G
1
)T
α
kj
(G
2
)T
β
k
j
(G
2
)
= T
α
ij
(G
1
G
2
)T
β
i
j
(G
1
G
2
)
= T
α⊗β
ii
,jj
(G
1
G
2
)
Theorem 8:
The character of a direct product representation T
α⊗β
is the product of
the characters of T
α
and T
β
.
Proof:
χ
α⊗β
=
¸
ii
T
α⊗β
ii
,ii
=
¸
ii
T
α
ii
T
β
i
i
= χ
α
χ
β
52 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
Theorem 9:
If ¦ψ
α
i
¦ transform according to the representation T
α
and ¦ψ
β
j
¦ transform
according to the representation T
β
, then ¦ψ
α
i
ψ
β
j
¦ transform according to
the direct product representation T
α⊗β
.
Proof:
ˆ
T(G)(ψ
α
i
ψ
β
j
) =
¸
k
¸
l
ψ
α
k
T
α
ki
(G)ψ
β
l
T
β
lj
(G)
=
¸
kl
(ψ
α
k
ψ
β
l
)T
α⊗β
kl,ij
In general, the direct product representation is reducible even when the
two constituent representations are irreducible. There is a unitary matrix
U, independent of the group elements, which transforms T
α⊗β
into a block
form:
¸
ij,i
j
U
γmk,ij
T
α⊗β
ij,i
j
U
†
ij,γ
m
k
= δ
γγ
δ
mm
T
γ
kk
(3.8)
The label m distinguishes blocks with the same irreducible representation.
Thus, we may write
T
α⊗β
=
¸
γ
⊕m
γ
T
γ
(3.9)
γ labels the irreducible representations and the coeﬃcients m
γ
, which tells
us the number of times the irreducible representation γ occurs in the de
composition, can be calculated using the ﬁrst orthogonality relation of the
characters
χ
α⊗β
=
¸
γ
m
γ
χ
γ
χ
α
χ
β
=
¸
γ
m
γ
χ
γ
m
γ
= (1/h)
¸
k
N
k
χ
α
(C
k
)χ
β
(C
k
)χ
γ∗
(C
k
)
3.11. CLEBSCHGORDAN COEFFICIENTS 53
3.11 ClebschGordan Coeﬃcients
Since the set of product functions ψ
α
i
ψ
β
j
transform according to the prod
uct representation T
α⊗β
which is in general reducible, evidently the space
formed by the product functions are also reducible and the product func
tions may then be projected into these irreducible subspaces:
ψ
α
i
ψ
β
j
=
¸
γm
F
γ
ij
(m)
where F
γ
ij
(m) belongs to the irreducible subspace γ. The label m distin
guishes functions with the same γ which is necessary only when m
γ
> 1
in the decomposition of the product representation. In other words, there
may be more than one subspace which transform according to the same
irreducible representation. F
γ
ij
(m) itself may be expanded in an orthogonal
basis which spans the irreducible subspace γ:
F
γ
ij
(m) =
¸
k
f
γ
k
(m)C
γ
k,ij
(m)
and therefore
ψ
α
i
ψ
β
j
=
¸
γm
¸
k
f
γ
k
(m)C
γ
k,ij
(m) (3.10)
The coeﬃcients C are called the ClebschGordan coeﬃcients and they
depend on α and β.
Apart from a possible phase factor, the ClebschGordan coeﬃcients are
actually identical to the unitary matrix U deﬁned in the previous section. To
see this, we consider how the state Ψ
γ
k
(m) =
¸
ij
ψ
α
i
ψ
β
j
U
†
ij,γmk
transform
under a group symmetry operation:
ˆ
T(G)Ψ
γ
k
(m) =
¸
ij,i
j
ψ
α
i
ψ
β
j
T
α⊗β
i
j
,ij
U
†
ij,γmk
=
¸
i
j
k
ψ
α
i
ψ
β
j
U
†
i
j
,γmk
T
γ
k
k
=
¸
k
Ψ
γ
k
(m)T
γ
k
k
We have used Eq. ( 3.8) in the second step. Thus the states Ψ
γ
k
(m)
transform according to the irreducible representation γ. We may also write
ψ
α
i
ψ
β
j
=
¸
kγm
Ψ
γ
k
(m)U
γmk,ij
(3.11)
54 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
which shows that the ClebschGordan coeﬃcients are identical to the uni
tary matrix U, apart from a phase factor. By convention, they are nor
malised as follows:
¸
ij
[C
γ
ijk
(m)[
2
= 1 (3.12)
Since C is unitary, it satisﬁes the following relations
¸
ij
C
γ
k,ij
(m)C
∗γ
k
,ij
(m
) = δ
γγ
δ
mm
δ
kk
¸
γkm
C
γ
k,ij
(m)C
∗γ
k,i
j
(m) = δ
ii
δ
jj
3.12 The WignerEckart Theorem
In physics, we are often faced with a problem of calculating matrix elements
of an operator
ˆ
O between two states ψ and φ: (ψ,
ˆ
Oφ). In practice, the
calculation of matrix elements can be very complicated and we would like
to study if group theory can help us in simplifying this problem.
First of all, from previous sections, we know that an arbitrary function
belonging to a linear space L which transforms into itself under group
operations can be projected into the diﬀerent irreducible components of
the group. Consequently, we need only consider matrix elements of the
type:
(f
γ
k
,
ˆ
O
α
i
f
β
j
)
We study the transformation of the function ψ
ij
=
ˆ
O
α
i
f
β
j
under a group
symmetry operation:
ˆ
T(G)ψ
ij
=
ˆ
T(G)
ˆ
O
α
i
f
β
j
= [
ˆ
T(G)
ˆ
O
α
i
ˆ
T
−1
(G)]
ˆ
T(G)f
β
j
=
¸
k
ˆ
O
α
k
T
α
ki
(G)
¸
m
f
β
m
T
β
mj
(G)
=
¸
km
ψ
km
T
α⊗β
km,ij
(G)
The new function ψ
ij
may then be completely expanded into its irreducible
components as in Eq. ( 3.11):
ψ
ij
=
¸
γ
k
m
Ψ
γ
k
(m)C
γ
k
,ij
(m)
3.13. APPLICATIONS 55
The matrix element is then given by
(f
γ
k
,
ˆ
O
α
i
f
β
j
) =
¸
γ
k
m
(f
γ
k
, Ψ
γ
k
(m))C
γ
k
,ij
(m)
=
¸
m
(f
γ
k
, Ψ
γ
k
(m))C
γ
k,ij
(m)
The signiﬁcance of the above result is that the ClebschGordan coeﬃcients
C
γ
k,ij
(m) are obtained entirely from symmetry and they do not depend on
the details of the potential of the system. Information about the details
of the system is contained in the matrix elements (f
γ
k
, Ψ
γ
k
(m)) which do
not depend on i and j. In fact, it does not depend on k as shown in
section 3.6. These matrix elements are often referred to as reduced matrix
elements and they are written as
(f
γ
k
, Ψ
γ
k
(m)) = 'f
γ
[[
ˆ
O
α
[[f
β
`
m
emphasising the fact that the reduced matrix elements are independent of
i, j, and k. Thus we have
(f
γ
k
,
ˆ
O
α
i
f
β
j
) =
¸
m
'f
γ
[[
ˆ
O
α
[[f
β
`
m
C
γ
k,ij
(m) (3.13)
This is known as the WignerEckart theorem which tells us that the matrix
element (f
γ
k
,
ˆ
O
α
i
f
β
j
) is zero unless the product representation T
α⊗β
contains
the irreducible representation γ. This is a general form of selection rules.
The ClebschGordan coeﬃcients relate the matrix elements for a given α,
β, and γ but with diﬀerent i, j, and k to one another.
We can see that theorem 6 follows immediately from the WignerEckart
theorem because the Hamiltonian is an invariant under the group symmetry
operations, i.e.
ˆ
T(G)
ˆ
H
ˆ
T
−1
(G) =
ˆ
H, so that
ˆ
H transforms according to
the identity representation.
3.13 Applications
We list a few of the common applications of group theory.
1. Selection rules: As an example we consider optical transitions in a
molecule which has a D
3
symmetry. This could correspond to a
56 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
molecule with three identical atoms at the corners of a triangle. Ac
tually, the full symmetry is higher than D
3
but as an illustration it
is suﬃcient to consider the D
3
symmetry. Optical transitions are
governed by dipole operators x, y, and z which transform like the
three Cartesian unit vectors. The groundstate of the molecule may
be assumed to be nondegenerate and symmetric so that it trans
forms according to the identity representation. From the previous
example, the dipole operators transform according to the represen
tation T = T
2
⊕T
3
and therefore the ﬁnal states can only be those
which transform according to (T
2
⊕T
3
) ⊗T
1
= T
2
⊕T
3
.
2. Symmetry breaking: In the above example, we know from symmetry
alone that the eigenfunctions can only be at most doubly degener
ate. If we replace one of the atoms by a diﬀerent atom so that the
symmetry becomes C
2
, then this degeneracy, if it exists, must split
into two nondegenerate levels because C
2
is an Abelian group and
the irreducible representations are one dimensional. The actual re
duction can be ﬁgured out straightforwardly using the orthogonality
relation between the characters. The ordering of the energy levels,
however, cannot be determined by group theory alone.
3. Variational principle: We are often interested in ﬁnding the ground
state of a given Hamiltonian which usually cannot be solved analyti
cally. One method of obtaining an approximate groundstate is based
on variational principle. We start with some basis functions ¦φ
i
¦ and
approximate the true groundstate Ψ as a linear combination of these
functions.
Ψ =
¸
i
c
i
φ
i
From the variational principle, the energy expectation value < E >=
(Ψ,
ˆ
HΨ) ≥ E
0
, where E
0
is the true groundstate energy. The co
eﬃcients ¦c
i
¦ are determined by minimising < E > subject to the
condition (Ψ, Ψ) = 1:
∂
∂c
∗
i
[< E > −λ(Ψ, Ψ)] =
∂
∂c
∗
i
¸
jk
c
∗
j
(φ
j
,
ˆ
Hφ
k
)c
k
−λ
¸
j
c
∗
j
c
j
¸
¸
=
¸
k
H
ik
c
k
−λc
i
= 0
3.14. EXAMPLES 57
which is just an eigenvalue problem.
From Theorem 6, it is clear that we should sort out the basis func
tions, using the projection operator method for example, according
to the rows and irreducible representations ¦φ
αm
i
¦ where m labels
diﬀerent possible subspaces transforming according to the same ir
reducible representation α. It is only necessary to express the trial
function as follows: ψ =
¸
m
c
αm
i
φ
αm
i
. When m = 1, the Hamilto
nian is already diagonal and when m > 1, we need only diagonalise
an mm matrix instead of an l l matrix where l =
¸
αmi
1, which
is usually much larger than m.
4. Degenerate perturbation theory: In many cases, the Hamiltonian
may be broken up into two terms
ˆ
H =
ˆ
H
0
+
ˆ
H
1
such that
ˆ
H
1
may
be treated as a small perturbation to
ˆ
H
0
.
ˆ
H
0
usually has a higher
symmetry than
ˆ
H
1
. A typical example of this is an atom placed in a
crystal where the crystal ﬁeld symmetry is lower than the rotational
symmetry of the atom. If the eigenfunctions of
ˆ
H
0
are degenerate, we
may use the variational principle to calculate the energy shifts. The
eigenfunction of
ˆ
H is approximated as a linear combination of the
degenerate eigenfunctions of
ˆ
H
0
. This leads to an eigenvalue problem
for
ˆ
H
1
and the eigenvalues give the ﬁrst order shifts. As before, we
should sort out the degenerate eigenfunctions of
ˆ
H
0
according to the
rows and irreducible representations of the symmetry group of
ˆ
H
1
in
order to minimise the size of the matrix to be diagonalised.
3.14 Examples
1. Bloch Theorem: A lattice is a parallelepiped formed by three linearly
independent vectors a
1
, a
2
, a
3
. A set of vectors t = n
1
a
1
+ n
2
a
2
+
n
3
a
3
for all possible integers n
1
, n
2
, n
3
generate lattice points which
together with the associated lattices form a crystal. The set t = ¦t¦
clearly form a group under vector addition and the group is Abelian.
In fact, it is a direct product of three groups
t = t
1
⊗t
2
⊗t
3
58 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
where t
1
= ¦n
1
a
1
¦, t
2
= ¦n
2
a
2
¦, and t
3
= ¦n
3
a
3
¦. It is customary
to have a periodic boundary condition such that
ˆ
T(N
1
a
1
)ψ(r) = ψ(r + N
1
a
1
) = ψ(r) →
ˆ
T(N
1
a
1
) =
ˆ
T(0) = 1
Similarly for the other two groups. N
1
is taken to be inﬁnite at the
end. The groups t
1
, t
2
, and t
3
become cyclic and the irreducible
representations are given by e
i2πk
1
/N
1
, k
1
= 1, 2, . . . , N
1
, and their
powers. It is useful at this stage to introduce the concept of reciprocal
space deﬁned by
g
1
= 2π
a
2
a
3
a
1
a
2
a
3
g
2
= 2π
a
3
a
1
a
1
a
2
a
3
g
3
= 2π
a
1
a
2
a
1
a
2
a
3
so that
g
i
a
j
= 2πδ
ij
With this deﬁnition, the direct product representations can be written
T
k
(t) = e
ik·t
where k = (k
1
/N
1
)g
1
+ (k
2
/N
2
)g
2
+ (k
3
/N
3
)g
3
. It follows that the
eigenfunctions of a Hamiltonian with translational lattice symmetry
form basis functions for the irreducible representations of the group
of translations and they may be labelled by the irreducible represen
tations k. This means
ˆ
T(t)ψ
k
(r) = ψ
k
(r +t) = ψ
k
(r)e
ik·t
which is the Bloch theorem.
2. Normal modes (chapter 6 of Elliot and Dawber): When the atoms
in a molecule are displaced from their equilibrium positions, the po
tential energy will increase. If the atoms are let free, there will be
an interchange between potential and kinetic energies resulting in
vibrations. These vibrations may be thought of as being composed
3.14. EXAMPLES 59
of eigen modes or normal modes, each with a well deﬁned frequency,
in the same way that a wavefunction can be decomposed into a
sum of eigenstates with well deﬁned energies. The eigen modes may
be classiﬁed according to the irreducible representations of the sym
metry group of the molecule, just like the the eigenstates of the
Hamiltonian.
The potential energy for arbitrary displacements of N atoms in a
molecule is given by
V = V
0
+
¸
i
∂V
∂q
i
q
i
+
¸
ij
∂
2
V
∂q
i
∂q
j
q
i
q
j
The sums run from 1 to 3N. At equilibrium, ∂V/∂q
i
= 0 so that
the Hamiltonian for molecular vibrations is given by
H =
¸
i
1
2
M
i
˙ q
2
i
+
¸
ij
1
2
B
ij
q
i
q
j
where B
ij
= ∂
2
V/∂q
i
∂q
j
. It is conventional to absorb the masses
into the displacement coordinates by deﬁning
α
i
=
M
i
q
i
, D
ij
= B
ij
/
M
i
M
j
The Hamiltonian becomes
H =
¸
i
1
2
˙ α
2
i
+
¸
ij
1
2
D
ij
α
i
α
j
The matrix D is known as the dynamical matrix.
We can now introduce a 3N dimensional abstract space with or
thonormal basis vectors ¦e
i
¦. The 3N displacements of atoms in
real space may be mapped into a vector in the 3N dimensional ab
stract space:
(q
1
, q
2
, . . . , q
3N
) →q =
¸
i
e
i
α
i
The unit vector e
i
corresponds to a displacement q
i
= 1/
√
M
i
in real
space but we note that e
i
itself is not a displacement in real space
60 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
but rather a displacement in the abstract space. We may deﬁne an
operator
ˆ
D in the abstract space so that
V (q) =
1
2
(q,
ˆ
Dq)
=
1
2
¸
ij
α
i
α
j
(e
i
,
ˆ
De
j
)
which implies
D
ij
= (e
i
,
ˆ
De
j
) →
ˆ
De
j
=
¸
i
e
i
D
ij
We would like to transform the Hamiltonian into the form H =
¸
i
h
i
so that the eigenvenctors of H can be written as a product of the
eigenvectors of h. This can be done by a unitary transformation
u
k
=
¸
i
e
i
a
ik
An arbitrary vector q may be written as
q =
¸
i
e
i
α
i
=
¸
k
u
k
Q
k
=
¸
ki
e
i
a
ik
Q
k
→α
i
=
¸
k
a
ik
Q
k
By choosing u
k
to be an eigenvector of
ˆ
D, i.e.
ˆ
Du
k
= ω
k
u
k
or
¸
j
D
ij
a
jk
= ω
k
a
ik
, the Hamiltonian becomes
H =
¸
k
¸
1
2
˙
Q
2
k
+
1
2
ω
k
Q
2
k
which is the desired form. Quantum mechanically, we replace
˙
Q
2
k
with −∂
2
/∂Q
2
k
and the solutions to h = −∂
2
/∂Q
2
k
+
1
2
ω
k
Q
2
k
are
ψ
n(k)
= AH
n(k)
(
√
ω
k
Q
k
)exp(−ω
k
Q
k
/2)
n(k)
= (n(k) +
1
2
)ω
k
We have set the Planck constant h/2π = 1.
So far we have not made any use of group theory and in principle we
can calculate the dynamical matrix D and diagonalise it to obtain
3.14. EXAMPLES 61
the normal modes and the corresponding eigen frequencies. The
diagonalisation of D is greatly simpliﬁed by using group theory. The
eigen frequencies, however, cannot be obtained from group theory
alone because they depend on the strength of the potential. By
deﬁnition, we have
[
ˆ
T(G),
ˆ
D] = 0
for all G and it immediately follows from Theorem 5 that the degen
erate eigenvectors of
ˆ
D (normal modes) form basis vectors for the
irreducible representations of the group.
The 3N dimensional space forms a 3N 3N representation which
is in general reducible:
χ
3N
=
¸
γ
m
γ
χ
γ
To calculate χ
3N
, we ﬁrst write the basis vector as e
i
(R), i = x, y, z,
R = 1, 2, . . . , N, which corresponds to a displacement of atom R
along the iaxis. The eﬀect of the symmetry operation on the basis
vector is given by
ˆ
T(G)e
i
(R) =
¸
R
j
e
j
(R
)T
R
R
ji
(G)
The character is given by
χ
3N
=
¸
Ri
T
RR
ii
= N
R
χ
V
where N
R
is the number of atoms unmoved by the symmetry oper
ation
ˆ
T(G) and χ
V
is the character of the three dimensional vector
representation. This result can be understood by observing that when
ˆ
T(G) moves atom R to atom R
,
¸
i
T
RR
ii
= 0. When atom R is
not moved, then
¸
i
T
RR
ii
= χ
V
.
As an example, we consider a molecule with D
3
symmetry (three
atoms at the corners of an equilateral triangle). We have already
worked out the characters for the vector representation χ
V
and they
are given by χ
V
(C
e
) = 3, χ
V
(C
c
) = 0, and χ
V
(C
d
) = −1. Thus
χ
3N
(C
e
) = 3 3 = 9, χ
3N
(C
c
) = 0, and χ
3N
(C
d
) = 1 (−1) =
62 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS
−1. It is then straightforward to decompose χ
3N
into the diﬀerent
irreducible components using the character table of D
3
. We get
m
1
=
1
6
(1 1 9 + 2 1 0 + 3 1 (−1)) = 1
m
2
=
1
6
(1 1 9 + 2 1 0 + 3 (−1) (−1)) = 2
m
3
=
1
6
(1 2 9 + 2 (−1) 0 + 3 0 (−1)) = 3
There are three nondegenerate modes and three doubly degenerate
modes. Actually, six of them correspond to rigid motions, i.e. there is
no relative motion between the atoms. The signiﬁcance of classifying
the vibrations into normal modes is that when calculating absorption
or emission spectra it is possible to ﬁgure out which modes are al
lowed in the absorption process. The simplest absorption processes
are caused by electric dipole transitions from the groundstate to some
vibrational excited states. The groundstate usually tranforms accord
ing to the identity representation and the dipole operators transform
according to the vector representation. Thus the modes allowed are
given by those corresponding to the irreducible representations oc
curing in the decomposition of the vector representation. In more
complicated absorption processes, there are intermediate states in
volved. In this case the dominant operators are of the form xy, yz,
zx, x
2
, y
2
, and z
2
(Raman spectra). These are in fact products of
x, y, and z, i.e. they transform according to the product of the
vector representations which can be decomposed into the irreducible
components. Care, however, has to be taken due to the fact that
the product functions are not linearly independent. See chapter 6 of
Elliot and Dawber for more examples and details.
Chapter 4
Lie Groups
4.1 Deﬁnitions
So far we have considered groups with a ﬁnite number of elements. In
this Chapter, we will study the so called continuous groups in which the
elements G are deﬁned uniquely by a set of n independent continuous
real parameters a = (a
1
, a
2
, . . . , a
n
) so that G(a) = G(b) unless a = b.
The number n is called the dimension of the group. Such groups have an
inﬁnite number of elements. The elements of a continuous group satisfy
the multiplication law:
G(a)G(b) = G(c) (4.1)
This means that operating a group element G(b) followed by G(a) has
the same eﬀect as operating a group element G(c) represented by a set of
parameters c. c must be a function of the paramaters a and b:
c = F(a, b) (4.2)
and we may write the multiplication law as:
G(a)G(b) = G(F(a, b)) (4.3)
As a convention, a = 0 is chosen to represent the unit element:
G(0) = 1 (4.4)
With this deﬁnition, it follows from Eqs. ( 4.1) and ( 4.2) that
F(a, 0) = a ; F(0, b) = b ; F(0, 0) = 0 (4.5)
63
64 CHAPTER 4. LIE GROUPS
∂F(0, 0)
∂a
=
∂F(0, 0)
∂b
= 1 (4.6)
In addition, every element must have an inverse i.e. for a given set of
parameters a, there is a set of parameters q such that
G(a)G(q) = G(q)G(a) = G(0) = 1 (4.7)
The functions F(a,b) are assumed to be analytic functions of the pa
rameters a and b, and this kind of group is called a Liegroup.
4.2 Inﬁnitesimal Operators
An important concept in continuous groups is the concept of inﬁnitesimal
operators which are obtained by considering group elements near the unit
element i.e. elements with small a:
G(a) = 1 +X a +. . . (4.8)
X
i
=
¸
∂G
∂a
i
¸
a=0
(4.9)
The set of operators X = (X
1
, X
2
, . . . , X
n
) are called the inﬁnitesimal
operators, or generators, which are independent of the parameters a. It
will be shown below that every group element may be expressed in terms
of the inﬁnitesimal operators so that the study of continuous groups may
be restricted in most cases to the study of these inﬁnitesimal operators.
For small a and b we have
F(a, b) = F(a, 0) +
∂F(0, 0)
∂b
b
= a +b
using Eq. ( 4.6). As a convention, a dot denotes a dot product or a
summation over the lower and upper vectors. This implies that
G(a)G(b) = G(F(a, b)) = G(a +b)
G(b)G(a) = G(F(b, a)) = G(b +a)
4.2. INFINITESIMAL OPERATORS 65
i.e. the group elements are commutative in the vicinity of the unit element.
To relate the group elements and the inﬁnitesimal operators, we study
the product G(a)G(b) for an arbitrary a but small b. To simplify the
writing, we deﬁne a matrix
M(a) ≡
∂F(a, 0)
∂b
→M
ij
(a) ≡
∂F
i
(a, 0)
∂b
j
(4.10)
and a convention that a dot on the left of M means a summation over the
rows whereas a dot on the right means a summation over the columns.
G(a)G(b) = G(F(a, b))
= G(F(a, 0) +M(a) b)
= G(a +M(a) b)
= G(a) +
∂G(a)
∂a
M(a) b
On the other hand
G(a)G(b) = G(a)[1 +X b]
so that
G(a)X b =
∂G(a)
∂a
M(a) b
Since b are arbitrary and independent parameters, it must be that
G(a)X =
∂G(a)
∂a
M(a) (4.11)
This equation relates the group element and its derivative at a to the
inﬁnitesimal operators X.
We now try to ﬁnd an explicit expression relating G(a) to X. Let
u = (u
1
, u
2
, . . . , u
n
) be a unit vector (u u = 1) in the parameter space
and let us project Eq. ( 4.11) along this unit vector:
G(a)X u =
∂G(a)
∂a
M(a) u (4.12)
We deﬁne a curve in the parameter space by the following equation
da(t)
dt
= M(a) u (4.13)
66 CHAPTER 4. LIE GROUPS
with boundary condition a(0) = 0. The direction of this curve at t = 0 is
along the unit vector u because M(0) =
∂F(0,0)
∂b
= 1. Using Eq. ( 4.13)
in ( 4.12) yields
G(a)X u =
∂G(a)
∂a
da
dt
=
dG(a)
dt
which can be formally solved to give
G(a(t)) = e
tX·u
(4.14)
Thus, the group elements at ﬁnite parameters a can be obtained from
the inﬁnitesimal operators X. The global properties of Lie groups are
determined by the inﬁnitesimal operators X which are independent of the
parameters a. We have made an assumption, however, that every point a
in the parameter space can be reached by the curve deﬁned in Eq. ( 4.13),
by starting along an appropriate direction u from the origin.
4.3 Lie Algebra
The inﬁnitesimal operators satisfy commutation relations and to ﬁnd these
relations, we diﬀerentiate Eq. ( 4.11) with respect to a parameter a
i
for
each component j:
∂G(a)
∂a
i
X
j
=
¸
k
¸
∂
2
G(a)
∂a
i
∂a
k
∂F
k
(a, 0)
∂b
j
+
∂G(a)
∂a
k
∂
2
F
k
(a, 0)
∂a
i
∂b
j
¸
Evaluating this expression for a = 0, we get
X
i
X
j
=
∂
2
G(0)
∂a
i
∂a
j
+
¸
k
X
k
∂
2
F
k
(0, 0)
∂a
i
∂b
j
Interchanging i ↔j and taking the diﬀerence, we get
[X
i
, X
j
] =
¸
k
X
k
c
k
ij
(4.15)
4.3. LIE ALGEBRA 67
where
c
k
ij
=
∂
2
F
k
(0, 0)
∂a
i
∂b
j
−
∂
2
F
k
(0, 0)
∂a
j
∂b
i
(4.16)
We see that the inﬁnitesimal operators possess a closure property under
commutation. Operators possessing this property are said to form an al
gebra and for Lie groups it is called Lie algebra. For every Lie group there
is a corresponding Lie algebra.
The coeﬃcients c
k
ij
are called the structure constants and they are
purely determined by the functions F(a,b) which means that they are
independent of any particular representation of the group.
The structure constants are antisymmetric in the lower indices:
c
k
ij
= −c
k
ji
(4.17)
We can construct symmetric and antisymmetric quantities from the struc
ture constants. By summing over products of two structure constants, we
obtain symmetric matrices
g
ij
=
¸
kl
c
k
il
c
l
jk
= g
ji
(4.18)
Summing over products of these matrices and the structure constants yields
a
ijk
=
¸
m
g
im
c
m
jk
(4.19)
which are antisymmetric with respect to an interchange of two indices.
This is because we have the following relation
¸
l
c
l
ij
c
m
lk
+ c
l
jk
c
m
li
+c
l
ki
c
m
lj
= 0 (4.20)
which follows from the Jacobi identity
[[A, B], C] + [[B, C], A] + [[C, A], B] = 0
In this way, we can continue forming symmetric and antisymmetric quan
tities from the structure constants.
68 CHAPTER 4. LIE GROUPS
4.4 Casimir Operators
To construct the irreducible representations of a continuous group, it will
be useful to have some operators which commute with all of the inﬁnitesi
mal operators. The degenerate eigenfunctions of these operators may then
be used as basis functions for the irreducible representations because they
transform among themselves under group operations. We recall from Chap
ter 3 that when [
ˆ
T(G),
ˆ
H] = 0, the sets of degenerate eigenfunctions of
ˆ
H
form basis sets for the irreducible representations of the group (assuming
we have the full group, not a subgroup of a larger group).
The simplest operator that commute with the inﬁnitesimal operators
X must be at least of second order in these operators because a linear
combination of X
i
clearly will not in general commute with X due to Eq.
( 4.15). Thus we form the most general linear combination of second order
operators
C =
¸
ij
g
ij
X
i
X
j
(4.21)
The matrix g
ij
must be symmetric since every matrix can be written as a
sum of symmetric and antisymmetric matrices and from Eq. ( 4.15) the
antisymmetric part reduces to a linear combination of X
i
. We evaluate
the commutation of this operator with an arbitrary X
k
:
[C, X
k
] =
¸
ij
g
ij
[X
i
X
j
, X
k
]
=
¸
ij
g
ij
¦X
i
[X
j
, X
k
] + [X
i
, X
k
]X
j
¦
=
¸
ij
g
ij
¸
l
X
i
X
l
c
l
jk
+ X
l
c
l
ik
X
j
¸
=
¸
ijl
g
ij
c
l
jk
¦X
i
X
l
+ X
l
X
i
¦
The last step makes use of the fact that g
ij
is symmetric. If we can ﬁnd g
ij
such that
¸
j
g
ij
c
l
jk
is antisymmetric with respect to the interchange i ↔l
then C is the required operator. We have in Eq. ( 4.19) an antisymmetric
quantity with three indices but it involves no upper indices in the structure
constant. We may, however, write
c
l
jk
=
¸
ml
g
lm
g
ml
c
l
jk
4.5. LADDER OPERATORS AND MULTIPLETS 69
=
¸
m
g
lm
a
mjk
where we have deﬁned the inverse of g
ij
¸
m
g
lm
g
ml
= δ
ll
and
a
mjk
=
¸
l
g
ml
c
l
jk
as in Eq. ( 4.19) Continuing the evaluation of [C, X
k
] we get
[C, X
k
] =
¸
ijl
g
ij
¸
m
g
lm
a
mjk
¦X
i
X
l
+ X
l
X
i
¦
=
1
2
¸
ijlm
[g
ij
g
lm
+ g
lj
g
im
]a
mjk
¦X
i
X
l
+ X
l
X
i
¦
=
1
4
¸
ijlm
[g
ij
g
lm
+ g
lj
g
im
][a
mjk
+ a
jmk
] ¦X
i
X
l
+ X
l
X
i
¦
Since a
mjk
= −a
jmk
we obtain [C, X
k
] = 0.
In general, it may be possible to form higher order Casimir operators,
which also commute with all of the inﬁnitesimal operators. The irreducible
representations of the group may be labelled by the eigenvalues of these
Casimir operators. In the rotation group discussed later, we need only one
Casimir operator to label the irreducible representations.
4.5 Ladder Operators and Multiplets
For a given Lie group, some of the inﬁnitesimal operators may commute
among themselves. The largest number of mutually commuting inﬁnites
imal operators is called the rank of the group. Let ¦
ˆ
H
i
¦ be such a set
and ¦
ˆ
E
α
¦ be the rest of the inﬁnitesimal operators. Let also ¦
ˆ
C
i
¦ be a
set of Casimir operators. Since ¦
ˆ
C
i
¦ commute with ¦
ˆ
H
i
¦ it is possible
to construct simultaneous eigenstates of ¦
ˆ
C
i
¦ and ¦
ˆ
H
i
¦ with eigenvalues
¦C
i
¦ and ¦H
i
¦:
ˆ
C
i
[C
i
, H
i
` = C
i
[C
i
, H
i
`
ˆ
H
i
[C
i
, H
i
` = H
i
[C
i
, H
i
`
70 CHAPTER 4. LIE GROUPS
The remaining set of operators ¦
ˆ
E
α
¦ can be expressed as ladder oper
ators:
[
ˆ
H
i
,
ˆ
E
α
] =
iα
ˆ
E
α
(4.22)
This can be done by forming a linear combination
ˆ
E
α
=
¸
i
ˆ
X
i
a
iα
and
requiring that
[
ˆ
H
i
,
¸
j
ˆ
X
j
a
jα
] =
iα
¸
j
ˆ
X
j
a
jα
¸
jk
ˆ
X
k
c
k
ij
a
jα
=
iα
¸
j
ˆ
X
j
a
jα
¸
k
ˆ
X
k
¸
j
[c
k
ij
−
iα
δ
jk
]a
jα
= 0 →
¸
j
c
k
ij
a
jα
=
iα
a
kα
The coeﬃcients a
iα
are the eigenvectors of the structure constants and
iα
are called the root vectors. Eq. ( 4.22) implies that
[
ˆ
H
i
,
ˆ
E
α
][C
i
, H
i
` =
iα
ˆ
E
α
[C
i
, H
i
`
(
ˆ
H
i
−H
i
)
ˆ
E
α
[C
i
, H
i
` =
iα
ˆ
E
α
[C
i
, H
i
`
ˆ
H
i
(
ˆ
E
α
[C
i
, H
i
`) = (H
i
+
iα
)(
ˆ
E
α
[C
i
, H
i
`)
i.e.
ˆ
E
α
[C
i
, H
i
` is also an eigenstate of
ˆ
H
i
but with an eigenvalue (H
i
+
iα
).
This means
ˆ
E
α
[C
i
, H
i
` = A
α
(C
i
, H
i
)[C
i
, H
i
+
iα
` (4.23)
The constants A
α
can be determined from the Lie algebra. The states
generated by the ladder operators are the degenerate eigenstates of the
Casimir operators. This is because the Casimir operators commute with
all the inﬁnitesimal operators and therefore the degenerate eigenstates of
the Casimir operators form an invariant subspace under the inﬁnitesimal
operators. Acting
ˆ
X
i
on [C
i
, H
i
` results in a linear combination of states
with the same eigenvalues ¦C
i
¦. Such a set of degenerate states are called
a multiplet and they form an invariant subspace under the group operations.
Assuming that the multiplet spans a ﬁnite space, there must be a state
[ψ
+
` and [ψ
−
` with the largest and smallest values of ¦H
i
¦ such that
ˆ
E
α
[ψ
+
` = 0,
ˆ
E
β
[ψ
−
` = 0
if
ˆ
E
α
and
ˆ
E
β
increases and decreases the eigenvalues.
4.6. SUMMATION OVER GROUP ELEMENTS 71
We also have
ˆ
H
i
, [
ˆ
E
α
,
ˆ
E
β
]
=
[
ˆ
H
i
,
ˆ
E
α
],
ˆ
E
β
]
+
ˆ
E
α
, [
ˆ
H
i
,
ˆ
E
β
]
= (
iα
+
iβ
)[
ˆ
E
α
,
ˆ
E
β
]
so that the commutator [
ˆ
E
α
,
ˆ
E
β
] is itself a ladder operator with a root
(
iα
+
iβ
). If
iα
+
iβ
= 0, then [
ˆ
E
α
,
ˆ
E
β
] is a linear combination of ¦
ˆ
H
i
¦.
It is clear from the above discussion that the number of Casimir oper
ators needed to specify a multiplet is equal to the rank of the Lie group.
4.6 Summation over group elements
Many of the theorems that we have encountered involve summation over
group elements. In this section, we will study how to perform this sum
mation for continuous groups. We have to replace the summation by an
integral over the parameters in the following way:
¸
G
f(G) →
da w(a)f(G(a)) (4.24)
w(a) is a weight function describing the ”density” of elements in the pa
rameter space.
In proving some theorems for ﬁnite groups, we make use of the rear
rangement theorem. For these theorems to remain valid in the case of
continuous groups, we must require
da w(a)f(G(a)G(b)) =
da w(a)f(G(c)) =
dc w(c)f(G(c))
(4.25)
From Eq. ( 4.3), G(a)G(b) = G(c) = G(F(a, b)), and for a ﬁxed b, c is
a function of a.
da w(a)f(G(a)) =
dc w(c)f(G(c))
=
da
∂c
∂a
w(c)f(G(c))
Comparison with Eq. ( 4.25) gives
∂c
∂a
w(c) = w(a) (4.26)
72 CHAPTER 4. LIE GROUPS
where ∂c/∂a is the Jacobian.
To relate a to c we consider a small b:
c = F(a, b)
= F(a, 0) +
∂F(a, 0)
∂b
b
= a +M(a) b
where the matrix M is deﬁned as in Eq. ( 4.10). To ﬁrst order in b, the
Jacobian is given by
∂c
∂a
= 1 +
∂
∂a
M(a) b
Expanding w(c) about a to ﬁrst order in b yields
w(c) = w(a +M(a) b)
= w(a) +
∂w(a)
∂a
M(a) b
Substituting the above expressions into Eq. ( 4.26) gives
¸
1 +
∂
∂a
M(a) b
¸ ¸
w(a) +
∂w(a)
∂a
M(a) b
¸
= w(a)
Since b is arbitrary, we thus have
1
w(a)
∂w(a)
∂a
M(a) = −
∂
∂a
M(a)
or
∂ ln w(a)
∂a
= −
∂
∂a
M(a) M
−1
(a) (4.27)
4.7 Rotation Group R
3
As an example, we consider the rotation group in three dimensions, R
3
,
which often appears in many applications in physics, in particular in the
theory of angular momentum. The technique used in constructing the
irreducible representations of the rotation group is actually quite common
in Lie groups.
4.7. ROTATION GROUP R
3
73
4.7.1 Inﬁnitesimal operators
A rotation in three dimensions may be denoted by R(αa) where a is the
axis of rotation and α is the angle of anticlockwise rotation. It is easy to
show that
R(αa)r = cos α r + (1 −cos α)(r a)a + sin α (a r) (4.28)
For small α we have to ﬁrst order
R(a, α)r = (1 + α a)r
= (1 + α
¸
i
a
i
e
i
)r
= (1 +
¸
i
α
i
e
i
)r (4.29)
We identify
X
x
= e
x
, X
y
= e
y
, X
z
= e
z
,
as the inﬁnitesimal operators of the rotation group associated with the
parameters α
x
= αa
x
, α
y
= αa
y
and α
z
= αa
z
corresponding to rotations
about x−, y− and z− axes respectively. In other words, when α is small,
the rotation can be regarded as a vector sum of rotations about x−, y−
and z− axes.
In the Cartesian basis, we have explicitly
X
x
=
¸
¸
0 0 0
0 0 −1
0 1 0
¸
X
y
=
¸
¸
0 0 1
0 0 0
−1 0 0
¸
X
z
=
¸
¸
0 −1 0
1 0 0
0 0 0
¸
74 CHAPTER 4. LIE GROUPS
4.7.2 Commutation Relations
In Cartesian coordinates, the commutation relations for the inﬁnitesimal
operators are
[X
x
, X
y
] = X
z
, [X
y
, X
z
] = X
x
, [X
z
, X
x
] = X
y
(4.30)
The inﬁnitesimal operators are skew Hermitian (this is true in general if the
representation is unitary) but it is possible to deﬁne equivalent inﬁnitesimal
operators J
k
= iX
k
which are Hermitian. The commutation relations
become
[J
x
, J
y
] = iJ
z
, [J
y
, J
z
] = iJ
x
, [J
z
, J
x
] = iJ
y
(4.31)
We will show below that the commutation relations also follow from the
general results derived in the previous sections.
We consider a rotation R(βb) followed by R(αa) which may be aﬀected
by a rotation R(γc):
R(γc) = R(αa)R(βb)
The main quantities to calculate are the three functions F(αa, βb) =
γc. What we have to calculate are the structure constants deﬁned in Eq.
( 4.16). Since the derivatives of F(αa, βb) are taken at αa = 0 and
βb = 0, it is suﬃcient to consider small α and β. For small α, R(αa)
takes the form:
R(αa) =
¸
¸
1 −α
z
α
y
α
z
1 −α
x
−α
y
α
x
1
¸
and similarly for R(βb). Thus we have
R(γc) = R(αa)R(βb) =
¸
¸
1 −α
y
β
y
−α
z
β
z
−(α
z
+ β
z
−α
y
β
x
) α
y
+ β
y
+ α
z
β
x
α
z
+ β
z
+ α
x
β
y
1 −α
x
β
x
−α
z
β
z
−(α
x
+ β
x
−α
z
β
y
)
−(α
y
+ β
y
+ α
x
β
z
) α
x
+ β
x
−α
y
β
z
1 −α
x
β
x
−α
y
β
y
¸
The trace of R(γc) is (1 + 2cos γ), independent of the basis. Equating
this with the trace of R(αa)R(βb) gives
1 + 2cos γ = 3 −2(α
x
β
x
+α
y
β
y
+ α
z
β
z
)
1 + 2(1 −γ
2
/2) = 3 −2(α
x
β
x
+α
y
β
y
+ α
z
β
z
)
γ
2
= 2(α
x
β
x
+ α
y
β
y
+ α
z
β
z
)
4.7. ROTATION GROUP R
3
75
We also know that R(γc) is unitary and it does not alter γc so that
R(γc)γc = R
†
(γc)γc = γc or
[R(γc) −R
†
(γc)]γc = 0
This gives us the following equations:
−Cγ
y
+ Bγ
z
= 0
Cγ
x
−Aγ
z
= 0
−Bγ
x
+ Aγ
y
= 0
A = 2(α
x
+ β
x
) + α
y
β
z
−α
z
β
y
B = 2(α
y
+ β
y
) + α
z
β
x
−α
x
β
z
C = 2(α
z
+ β
z
) + α
x
β
y
−α
y
β
x
With γ
2
= γ
2
x
+ γ
2
y
+ γ
2
z
, the solution to the above equations is
γ
x
= pA, γ
y
= pB, γ
z
= pC
where
p =
¸
2(α
x
β
x
+ α
y
β
y
+ α
z
β
z
)
A
2
+ B
2
+ C
2
¸
1/2
A straightforward algebra shows that p = 1/2 to linear order in α and β.
The nonzero structure constants are then
c
z
xy
= c
y
zx
= c
x
yz
= 1 (4.32)
which gives the commutation relations in Eq. ( 4.30).
4.7.3 Casimir Operator
The Casimir operator can also be obtained directly from the results of the
previous section. We calculate the matrix g
ij
deﬁned in Eq. ( 4.18):
g
xx
=
¸
kl
c
k
xl
c
l
xk
= c
z
xy
c
y
xz
+ c
y
xz
c
z
xy
= −2
76 CHAPTER 4. LIE GROUPS
The same results are obtained for g
yy
and g
zz
whereas the other matrix
elements are zero. Thus we get
g
ij
= −2δ
ij
, g
ij
= −
1
2
δ
ij
and the Casimir operator is therefore
C =
¸
ij
g
ij
X
i
X
j
= −
1
2
(X
2
x
+ X
2
y
+ X
2
z
)
= −
1
2
X
2
=
1
2
J
2
(4.33)
4.8 Irreducible Representations of R
3
The method of constructing the irreducible representations for the rotation
group R
3
is quite general since it is based on the commutation relations
in Eqn. ( 4.30) alone. The method may therefore be applied to other
continuous groups. Since every rotation can be built up from the inﬁnites
imal operators, an irreducible subspace under these inﬁnitesimal operators
is also an irreducible subspace under the rotation group. The steps for
constructing the irreducible representations are as follows:
1. The sets of degenerate eigenvectors of the Casimir operator
ˆ
J
2
form
irreducible subspaces under the inﬁnitesimal operators
ˆ
J
x
,
ˆ
J
y
, and
ˆ
J
z
.
The rank of R
3
is one because there are no two mutually commuting
operators. Since
ˆ
J
2
commutes with the inﬁnitesimal operators, the
eigenvectors of
ˆ
J
2
may be chosen simultaneously to be eigenvectors
of one of the inﬁnitesimal operators. As a convention, we choose
ˆ
J
z
.
We construct the following ladder operators
ˆ
J
±
=
ˆ
J
x
±i
ˆ
J
y
(4.34)
which may be obtained by solving the eigenvectors and eigenvalues
of c
i
zj
. Clearly, the set of operators
ˆ
J
±
,
ˆ
J
z
are equivalent to the set
4.8. IRREDUCIBLE REPRESENTATIONS OF R
3
77
of operators
ˆ
J
x
,
ˆ
J
y
,
ˆ
J
z
. We may therefore work with the former set.
The commutation relations of these operators are
[
ˆ
J
z
,
ˆ
J
±
] = ±
ˆ
J
±
, [
ˆ
J
+
,
ˆ
J
−
] = 2
ˆ
J
z
(4.35)
The usefulness of these operators may be seen by applying
ˆ
J
z
to
ˆ
J
±
ψ
m
where ψ
m
is an eigenfunction of
ˆ
J
z
with an eigenvalue m.
ˆ
J
z
(
ˆ
J
±
ψ
m
) = (
ˆ
J
±
ˆ
J
z
±
ˆ
J
±
)ψ
m
=
ˆ
J
±
(
ˆ
J
z
±1)ψ
m
=
ˆ
J
±
(m±1)ψ
m
= (m±1)(
ˆ
J
±
ψ
m
)
Thus,
ˆ
J
±
ψ
m
must be proportional to ψ
m±1
.
2. We assume that the irreducible representations of R
3
have a ﬁnite
dimension. We choose basis functions which are eigenfunctions of
ˆ
J
z
. Let j be the largest eigenvalue of
ˆ
J
z
in this basis. Then
ˆ
J
+
ψ
j
= 0 (4.36)
There may be more than one function with the largest eigenvalue.
We take one of these and operate
ˆ
J
−
repeatedly on this function:
ˆ
J
−
ψ
j
= c
j−1
ψ
j−1
ˆ
J
−
ψ
j−1
= c
j−2
ψ
j−2
etc.
The c’s are normalisation constants. Since we have assumed that
the dimension of the irreducible representations is ﬁnite, it must be
that for some integer n we have
ˆ
J
−
ψ
j−n
= 0
The problem is to ﬁgure out what n is.
3. We show that ψ
m
is an eigenfunction of
ˆ
J
2
irrespective of the value
of m. This is to be expected because a set of degenerate eigenvectors
78 CHAPTER 4. LIE GROUPS
of
ˆ
J
2
form a basis for an irreducible representation.
ˆ
J
2
=
ˆ
J
2
x
+
ˆ
J
2
y
+
ˆ
J
2
z
=
1
2
(
ˆ
J
+
ˆ
J
−
+
ˆ
J
−
ˆ
J
+
) +
ˆ
J
2
z
=
ˆ
J
+
ˆ
J
−
+
ˆ
J
2
z
−
ˆ
J
z
=
ˆ
J
−
ˆ
J
+
+
ˆ
J
2
z
+
ˆ
J
z
The last two relations have been obtained by using the the second
commutation relation in Eq. ( 4.35). Operating
ˆ
J
2
on ψ
j
we get
ˆ
J
2
ψ
j
= (
ˆ
J
−
ˆ
J
+
+
ˆ
J
2
z
+
ˆ
J
z
)ψ
j
= j(j + 1)ψ
j
Operating
ˆ
J
2
ˆ
J
−
on ψ
j
yields
ˆ
J
2
ˆ
J
−
ψ
j
=
ˆ
J
−
ˆ
J
2
ψ
j
= j(j + 1)
ˆ
J
−
ψ
j−1
Thus ψ
j−1
is also an eigenfunction of
ˆ
J
2
with eigenvalue j(j + 1).
We can continue the same procedure starting from ψ
j−1
so that it
follows that ψ
m
, m = j, j − 1, . . . , j − n is an eigenfunction of
ˆ
J
2
with eigenvalue j(j + 1). We now operate
ˆ
J
2
on ψ
j−n
:
ˆ
J
2
ˆ
ψ
j−n
= (
ˆ
J
+
ˆ
J
−
+
ˆ
J
2
z
−
ˆ
J
z
)ψ
j−n
= [(j −n)
2
−(j −n)]ψ
j−n
= (j −n)(j −n −1)ψ
j−n
It follows that (j − n)(j − n − 1) = j(j + 1), which implies that
n = 2j and therefore j can be an integer or a halfinteger.
4. Finally we show that the 2j + 1 functions ψ
j
, ψ
j−1
, . . . , ψ
−j
form
an irreducible subspace (transform among themselves or invariant)
under the operators
ˆ
J
+
,
ˆ
J
−
,
ˆ
J
z
. It is clear that the functions trans
form among themselves under
ˆ
J
−
and
ˆ
J
z
. We show that they also
transform among themselves under
ˆ
J
+
:
ˆ
J
+
c
m
ψ
m
=
ˆ
J
+
ˆ
J
−
ψ
m+1
4.9. CHARACTERS OF R
3
79
= (
ˆ
J
2
−
ˆ
J
2
z
+
ˆ
J
z
)ψ
m+1
= [j(j + 1) −(m + 1)
2
+ (m + 1)]ψ
m+1
= (j + m + 1)(j −m)ψ
m+1
To calculate the normalisation constants c
m
we take the dot product
c
m
(ψ
m+1
,
ˆ
J
+
ψ
m
) = (j+m+1)(j−m). Using the fact that J
−
= J
†
+
,
we have
(ψ
m+1
,
ˆ
J
+
ψ
m
) = (J
†
+
ψ
m+1
, ψ
m
)
= (J
−
ψ
m+1
, ψ
m
)
= (c
m
ψ
m
, ψ
m
) = c
∗
m
Thus [c
m
[
2
= (j +m+1)(j −m). There is an arbitrary phase factor
in c
m
and the usual convention due to Condon and Shortley is to
choose c
m
to be real and positive so that
c
m
=
(j + m + 1)(j −m) (4.37)
ˆ
J
+
ψ
m
=
(j +m + 1)(j −m)ψ
m+1
(4.38)
ˆ
J
−
ψ
m
=
(j +m)(j −m + 1)ψ
m−1
(4.39)
To summarise, the irreducible representations of R
3
are labelled by R
j
,
with j = 0, ±1/2, ±1, ±3/2, ±2 . . ., which transform the (2j + 1)
functions ψ
j
, ψ
j−1
, . . . , ψ
−j
among themselves.
4.9 Characters of R
3
Rotations by the same angle,
ˆ
R(a, φ) and
ˆ
R(b, φ), belong to the same
class irrespective of the axes a and b. This must be true because there is
always a rotation
ˆ
R(c, φ
) that brings the axis a to b or vice versa, i.e.
ˆ
R(a, φ) =
ˆ
R
−1
(c, φ
)
ˆ
R(b, φ)
ˆ
R(c, φ
)
Geometrically,
ˆ
R(c, φ
) rotates a to b,
ˆ
R(b, φ) does a rotation by φ about
b and ﬁnally
ˆ
R
−1
(c, φ
) rotates b back to a and the net result is the same
as a rotation by φ about a.
80 CHAPTER 4. LIE GROUPS
Since the characters of representations belonging to the same class are
the same, it is suﬃcient to consider the characters of R
z
. We consider
the subgroup R
z
consisting of rotations about the zaxis. This subgroup is
Abelian and the irreducible representations are therefore one dimensional.
We require R(z, φ)R(z, δφ) = R(z, φ + δφ). Expanding both sides in δφ
we get
R(z, φ)(1 + δφR
(z, 0)) = R(z, φ) + δφR
(z, φ)
R
(z, 0)R(z, φ) = R
(z, φ)
The solution is R(z, φ) = c exp(R
(z, 0)φ). For a unitary representation,
we may choose c = 1 and from the boundary condition R(z, φ) = R(z, φ+
2π), we deduce that R
(z, 0) = −im where m is a positive or negative
integer. Thus the irreducible representations of R
z
are
R
m
z
(φ) = exp(−imφ) (4.40)
labelled by m = 0, ±1, ±2, . . . and the angle φ. These are also the eigen
functions of the inﬁnitesimal operator
ˆ
J
z
with eigenvalues m, as may be
easily veriﬁed.
It is clear that representation R
j
(a, φ) is reducible under the group R
z
and by choosing the zaxis along a, it may be decomposed as
R
j
(a, φ) = R
−j
z
(φ) ⊕R
−j+1
z
(φ) ⊕. . . ⊕R
j
z
(φ)
=
m=j
¸
m=−j
⊕R
m
z
(φ)
Thus the character is given by
χ
j
(φ) =
m=j
¸
m=−j
exp(−imφ)
=
sin (j + 1/2)φ
sin φ/2
(4.41)
which is easily obtained from the geometric series formula S
n
=
¸
k=n−1
k=0
ar
k
= a(1 − r
n
)/(1 − r) with n = (2j + 1), a = exp(ijφ),
and r = exp(iφ).
4.10. THE VECTORCOUPLING THEOREM 81
4.10 The VectorCoupling Theorem
The direct product of two irreducible representations of R
3
referring to the
same rotation φ about a given axis a is given by
R
j
1
(a, φ) ⊗R
j
2
(a, φ) =
¸
j
⊕c
j
R
j
(a, φ) (4.42)
To ﬁgure out the coeﬃcient c
j
we use the fact that the character of a direct
product representation is the product of the characters of the individual
representations:
χ
j
1
(a, φ)χ
j
2
(a, φ) =
¸
j
⊕c
j
χ
j
(a, φ)
The left hand side is given by
χ
j
1
χ
j
2
=
j
1
¸
m
1
=−j
1
j
2
¸
m
2
=−j
2
e
−im
1
φ
e
−im
2
φ
= [e
i(j
1
+j
2
)φ
+ e
i(j
1
+j
2
−1)φ
+. . . +e
−i(j
1
+j
2
)φ
]
+ [e
i(j
1
+j
2
−1)φ
+ e
i(j
1
+j
2
−2)φ
+ . . . + e
−i(j
1
+j
2
−1)φ
]
+ . . .
+ [e
i(j
1
−j
2
)φ
+ e
i(j
1
−j
2
−1)φ
+. . . +e
−i(j
1
−j
2
)φ
]
= χ
j
1
+j
2
+ χ
j
1
+j
2
−1
+ . . . + χ
j
1
−j
2
assuming that j
1
≥ j
2
without loss of generality. Therefore j
1
−j
2
becomes
[j
1
−j
2
[ in the general case. Comparison with the previous equation shows
that c
j
= 1 for j = j
1
+ j
2
, j
1
+ j
2
− 1, . . . + [j
1
− j
2
[ which yields the
vectorcoupling theorem:
R
j
1
(a, φ) ⊗R
j
2
(a, φ) =
j
1
+j
2
¸
j=j
1
−j
2

⊕R
j
(a, φ) (4.43)
The rotation group is called simply reducible because m
γ
= 1.
82 CHAPTER 4. LIE GROUPS
4.11 ClebschGordan Coeﬃcients
As discussed in section 3.11, a product of two states ψ
j
1
m
1
ψ
j
2
m
2
transforming
according to the product representation R
j
1
⊗j
2
of the rotation group may
be expanded as follows:
ψ(m
1
, m
2
) = ψ
j
1
m
1
ψ
j
2
m
2
=
¸
JM
Ψ
J
M
C
J∗
M
(m
1
, m
2
)
where the C’s are the ClebschGordan coeﬃcients which form a matrix
that brings the product representation R
j
1
⊗j
2
into the block diagonal form
(section 3.4, Theorem 3):
R
j
1
m
1
m
1
R
j
2
m
2
m
2
=
j
1
+j
2
¸
J=j
1
−j
2

¸
MM
C
J
M
(m
1
, m
2
)R
J
M
M
C
J∗
M
(m
1
, m
2
) (4.44)
This expansion is known as the ClebschGordan series. No additional label
is needed in C because the rotation group is simply reducible. We have
dropped the dependence on j
1
and j
2
for clarity. We may also write, using
the unitarity of C
Ψ
J
M
=
¸
m
1
,m
2
C
J
M
(m
1
, m
2
)ψ(m
1
, m
2
) (4.45)
Thus
C
J
M
(m
1
, m
2
) = (ψ(m
1
, m
2
), Ψ
J
M
) (4.46)
Since the new functions Ψ
J
M
are orthonormal, we have
¸
m
1
m
2
[C
J
M
(m
1
, m
2
)[
2
= 1 (4.47)
By applying
ˆ
J
z
to both sides of Eq. ( 4.45), it follows that C
J
M
(m
1
, m
2
) = 0
unless m = m
1
+ m
2
and from the vectorcoupling theorem we also have
[j
1
−j
2
[ ≤ J ≤ j
1
+ j
2
:
C
J
M
(m
1
, m
2
) = 0 unless
M = m
1
+ m
2
or
[j
1
−j
2
[ ≤ J ≤ j
1
+ j
2
(4.48)
4.11. CLEBSCHGORDAN COEFFICIENTS 83
With the Condon and Shortley phase convention
ˆ
J
+
ψ
j
m
= A
j,+m
ψ
j
m+1
ˆ
J
−
ψ
j
m
= A
j,−m
ψ
j
m−1
where
A
j,m
=
(j −m)(j + m + 1), A
j,j
= A
j,m>j
= 0 (4.49)
the ClebschGordan coeﬃcients become real.
The ClebschGordan coeﬃcients for ﬁxed j
1
and j
2
and a chosen j can
be calculated from recursion relations obtained by applying
ˆ
J
−
and
ˆ
J
+
to
both sides of Eq. ( 4.45):
A
J,−M
Ψ
J
M−1
=
¸
n
1
n
2
C
J
M
(n
1
, n
2
) [ A
j
1
,−n
1
ψ(n
1
−1, n
2
) + A
j
2
,−n
2
ψ(n
1
, n
2
−1) ]
A
J,+M
Ψ
J
M+1
=
¸
n
1
n
2
C
J
M
(n
1
, n
2
) [ A
j
1
,+n
1
ψ(n
1
+ 1, n
2
) + A
j
2
,+n
2
ψ(n
1
, n
2
+ 1) ]
Taking the dot product with ψ(m
1
, m
2
) and setting M → M + 1 and
M →M −1 in the ﬁrst and second equation above we get the recursion
relations:
A
J,+M
C
M
(m
1
, m
2
) = A
j
1
,+m
1
C
M+1
(m
1
+ 1, m
2
)
+ A
j
2
,+m
2
C
M+1
(m
1
, m
2
+ 1) (4.50)
A
J,−M
C
M
(m
1
, m
2
) = A
j
1
,−m
1
C
M−1
(m
1
−1, m
2
)
+ A
j
2
,−m
2
C
M−1
(m
1
, m
2
−1) (4.51)
To simplify the notation, we have dropped the superscript J since it is
ﬁxed. There is an ambiguity in the sign and this is ﬁxed by choosing the
convention that
C
J
(j
1
, J −j
1
) > 0 (4.52)
84 CHAPTER 4. LIE GROUPS
As an example we evaluate the ClebschGordan coeﬃcients for a given
j
2
= 1/2 which gives possible values of J = j
1
− 1/2, j
1
+ 1/2. To
simplify the notation further, we drop the second argument in C since
it can be obtained by substracting the ﬁrst argument from the subscript.
We consider the case J = j
1
− 1/2. With m
2
= ±1/2 we get from the
recursion relations
A
J,M
C
M
(M −1/2) = A
j
1
,M−1/2
C
M+1
(M + 1/2)
A
J,−M−1
C
M+1
(M + 1/2) = A
j
1
,−M−1/2
C
M
(M −1/2)
+ A
j
2
,−1/2
C
M+1
(M + 1/2)
A
J,M−1
C
M−1
(M −1/2) = A
j
1
,M−1/2
C
M
(M + 1/2)
+ A
j
2
,−1/2
C
M
(M −1/2)
A
J,−M
C
M
(M + 1/2) = A
j
1
,−M−1/2
C
M−1
(M −1/2)
These are four linear equations with four unknowns but since the equations
are homogeneous, the solutions can only be expressed as ratios. However,
the additional condition in Eq. ( 4.47) determines the absolute magnitudes
and condition 4.52 ﬁxes the sign. Similarly we can generate the recursion
relations for J = j
2
+ 1/2. The solutions in complete notation are
C
j
1
±1/2
M
(M −1/2, 1/2) = ±
j
1
±M + 1/2
2j
1
+ 1
C
j
1
±1/2
M
(M + 1/2, −1/2) =
j
1
∓M + 1/2
2j
1
+ 1
The ClebschGordan coeﬃcients can be also calculated by using the
projection operator:
ˆ
P
J
=
¸
K=J
ˆ
P
J
K
=
¸
K=J
¸
ˆ
J
2
−K(K + 1)
J(J + 1) −K(K + 1)
¸
(4.53)
which when applied to ψ
m
1
,m
2
will annihilate all components except
C
J
M
(m
1
, m
2
)Ψ
J
M
where M = m
1
+ m
2
:
ˆ
P
J
ψ
m
1
,m
2
= C
J
M
(m
1
, m
2
)Ψ
J
M
⇒ C
J
M
(m
1
, m
2
) C
J
M
(m
1
, m
2
) = (ψ
m
1
,m
2
,
ˆ
P
J
ψ
m
1
,m
2
)
4.11. CLEBSCHGORDAN COEFFICIENTS 85
We evaluate
ˆ
P
J
K
ψ
m
1
,m
2
, using the identity
ˆ
J
2
=
ˆ
J
−
ˆ
J
+
+
ˆ
J
z
(
ˆ
J
z
+ 1):
ˆ
P
J
K
ψ
m
1
,m
2
= [J(J + 1) −K(K + 1)]
−1
[ ¦ A
j
1
,m
1
A
j
1
,−m
1
−1
+ A
j
2
,m
2
A
j
2
,−m
2
−1
+M(M + 1) −K(K + 1) ¦ ψ(m
1
, m
2
)
+A
j
1
,m
1
A
j
2
,−m
2
ψ(m
1
+ 1, m
2
−1)
+A
j
2
,m
2
A
j
1
,−m
1
ψ(m
1
−1, m
2
+ 1) ]
= [J(J + 1) −K(K + 1)]
−1
[ ¦ (j
1
−m
1
)(j
1
+ m
1
+ 1) + (j
2
−m
2
)(j
2
+ m
2
+ 1)
+M(M + 1) −K(K + 1) ¦ ψ(m
1
, m
2
)
+
(j
1
−m
1
)(j
1
+ m
1
+ 1)(j
2
+ m
2
)(j
2
−m
2
+ 1)
ψ(m
1
+ 1, m
2
−1)
+
(j
2
−m
2
)(j
2
+ m
2
+ 1)(j
1
+ m
1
)(j
1
−m
1
+ 1)
ψ(m
1
−1, m
2
+ 1) ]
We evaluate the previous example using the projection operator
method. Consider the case J = j
1
−1/2, m
2
= −1/2 ⇒m
1
= M +1/2.
We have
ˆ
P
J
=
ˆ
P
J
j
1
+1/2
since there are only two possible values of
K = j
1
±1/2. Thus,
[C
j
1
−1/2
M
(M + 1/2, −1/2)[
2
= (ψ(M + 1/2, −1/2),
ˆ
P
j
1
−1/2
j
1
+1/2
ψ(M + 1/2, −1/2))
=
j
1
+M + 1/2
2j
1
+ 1
Using the sign convention in ( 4.52) we get
C
j
1
−1/2
M
(M + 1/2, −1/2) =
j
1
+ M + 1/2
2j
1
+ 1
because it must be positive when M = j
1
−1/2. We also have
C
j
1
−1/2
M
(M −1/2, 1/2) C
j
1
−1/2
M
(M + 1/2, −1/2)
=
(j
1
+ M + 1/2)(j
1
−M + 1/2)
−(2j
1
+ 1)
86 CHAPTER 4. LIE GROUPS
so that
C
j
1
−1/2
M
(M −1/2, 1/2 = −
j
1
−M + 1/2
2j
1
+ 1
in agreement with the previous results.
The general formula for the ClebschGordan coeﬃcients has been eval
uated by Wigner (Gruppentheorie, ViewegVerlag, Brunswick, 1931 or in
the translated version, Group Theory, Academic Press, New York, 1959).
Wigner’s closed expression is
C
J
M
(m
1
, m
2
) = δ
M,m
1
+m
2
√
2J + 1
(J + j
1
−j
2
)!(J −j
1
+j
2
)!(j
1
+ j
2
−J)!(J +M)!(J −M)!
(j
1
+ j
2
+J + 1)!(j
1
−m
1
)!(j
1
+m
1
)!(j
2
−m
2
)!(j
2
+ m
2
)!
¸
ν
(−1)
ν+j
2
+m
2
ν!
(j
2
+ J + m
1
−ν)!(j
1
−m
1
+ ν)!
(J −j
1
+ j
2
−ν)!(J +M −ν)!(ν + j
1
−j
2
−M)!
(4.54)
Racah (G. Racah, Phys. Rev. 62, 438 (1942)) derived a more symmetrical
form:
C
J
M
(m
1
, m
2
) = δ
M,m
1
+m
2
√
2J + 1
(j
1
+ j
2
−J)!(J + j
1
−j
2
)!(J +j
2
−j
1
)!
(j
1
+ j
2
+ J + 1)!
(j
1
+m
1
)!(j
1
−m
1
)!(j
2
+ m
2
)!(j
2
−m
2
)!(J + M)!(J −M)!
¸
ν
(−1)
ν
ν!
1
(j
1
+j
2
−J −ν)!(j
1
−m
1
−ν)!(j
2
+ m
2
−ν)!
1
(J −j
2
+ m
1
+ν)!(J −j
1
−m
2
+ ν)!
¸
(4.55)
The most frequently used ClebschGordan coeﬃcients are tabulated in
many books on atomic physics.
4.12. SPHERICAL HARMONICS 87
4.12 Spherical Harmonics
We wish to ﬁnd explicit basis functions for the irreducible representations of
the rotation group. To do this, we need to know the form of the inﬁnitesi
mal operators in function space. We form an operator representation of the
group elements, R →
ˆ
T(R), and study its eﬀect on an arbitrary function
ψ(x, y, z). First we consider a small rotation about the zaxis:
ˆ
T(R(γz))ψ(x, y, z) = ψ(R
−1
(γz)(x, y, z))
= ψ(x + γy, y −γx, z)
=
¸
1 + γ
y
∂
∂x
−x
∂
∂y
¸
ψ(x, y, z)
The same procedure can be applied to the x and yaxes. The inﬁnitesimal
operators in function space are therefore
ˆ
J
z
= i
y
∂
∂x
−x
∂
∂y
(4.56)
ˆ
J
y
= i
x
∂
∂z
−z
∂
∂x
(4.57)
ˆ
J
x
= i
z
∂
∂y
−y
∂
∂z
(4.58)
These are identical to the angular momentum operators in quantum me
chanics.
The procedure for constructing the irreducible representations of R
3
tells us that we should start with an eigenfunction of
ˆ
J
z
and
ˆ
J
2
and then
we apply the ladder operators repeatedly in order to span the irreducible
subspace corresponding to a set of degenerate eigenvectors of
ˆ
J
2
. The sim
plest function, other than a constant, that one can think of is x −iy which
is an eigenfunction of
ˆ
J
z
with eigenvalue −1 and it is also an eigenfunction
of
ˆ
J
2
. Furthermore,
ˆ
J
−
= (
ˆ
J
x
−iJ
y
)(x −iy) = 0
which implies that (x −iy) has the smallest value of m among the degen
erate eigenfunctions of
ˆ
J
2
. To generate the irreducible subspace, we apply
88 CHAPTER 4. LIE GROUPS
the operator
ˆ
J
+
repeatedly on (x −iy) until it vanishes:
ˆ
J
+
(x −iy) = 2z
ˆ
J
+
z = −(x + iy)
ˆ
J
+
(x + iy) = 0
(x−iy), z, and −(x+iy) thus form an irreducible subspace with eigenvalues
of
ˆ
J
z
equal to m = −1, 0, 1 respectively. We immediately recognise that
this set corresponds to the j = 1 representation with eigenvalue of
ˆ
J
2
equal
to j(j +1) = 2, as may be easily veriﬁed. These functions are proportional
to the spherical harmonics for l = 1. Since r
2
= x
2
+ y
2
+ z
2
is invariant
under rotation, we may choose r = 1. In polar coordinates, we then have
z = cos θ, y = sin θ sin φ, and x = sin θ cos φ, so that
Y
1
1
= −
3
8π
(x +iy) = −
3
8π
sin θ e
iφ
Y
1
0
=
3
4π
z =
3
4π
cos θ
Y
1
−1
=
3
8π
(x −iy) =
3
8π
sin θ e
−iφ
which are the spherical harmonics Y
l
m
for l = 1. The normalisation con
stants are determined by integrating the square of each function over the
solid angle and requiring the integral to be unity. Higher l spherical har
monics may be generated in the same fashion by applying the operator
ˆ
J
+
on (x −iy)
l
repeatedly and normalising the resulting functions as just
mentioned above. There is an ambiguity in the sign. As a convention, the
spherical harmonics are deﬁned such that
Y
l
−l
=
(2l)!
2
l
l!
2l + 1
4π
(x −iy)
l
(4.59)
which ﬁxes the sign.
The spherical harmonics for each l form an irreducible subspace under
rotation and therefore provide an irreducible representation of the rotation
group R
3
. Under rotation, the spherical harmonics transform among them
selves with the same l. Two spherical harmonics with diﬀerent l values will
never transform to each other under rotation because they belong to two
4.12. SPHERICAL HARMONICS 89
distinct irreducibe subspaces of the rotation group. This means that we can
decompose an arbitrary function f(r, θ, φ) into the irreducible subspaces
formed by the spherical harmonics as follows:
f(r, θ, φ) =
¸
lm
f
lm
(r)Y
l
m
(θ, φ)
From the deﬁnition of the spherical harmonics it is easy to show that:
Y
l
m
(0, φ) =
2l + 1
4π
δ
m,0
(4.60)
Y
l
0
(θ, φ) =
2l + 1
4π
P
l
(cos θ) (4.61)
The last equation may be thought of as the deﬁnition of the Legendre
polynomials.
We can now derive some useful identities but ﬁrst we write down the
transformation of spherical harmonics under an arbitrary rotation R in the
following way:
ˆ
T(R)Y
l
m
(θ, φ) = Y
l
m
(θ
, φ
) =
¸
m
Y
l
m
(θ, φ)R
l
m
m
(4.62)
or
Y
l
m
(θ, φ) =
l
¸
m
=−l
R
l∗
mm
Y
l
m
(θ
, φ
) (4.63)
where R
l
m
m
is the irreducible representation of a rotation R in the spher
ical harmonics basis. We take the point of view that (θ, φ) and (θ
, φ
)
correspond to the same physical point, i.e. it is the coordinate system
which is rotated. Considering a point (θ, φ) = (θ
z
, φ
z
) on the rotated z
axis corresponding to θ
= 0 we get
R
l
m0
=
4π
2l + 1
Y
l∗
m
(θ
z
, φ
z
) (4.64)
It follows immediately by substituting this into Eq. ( 4.62) for m = 0 that
Y
l
0
(θ
, φ
) =
4π
2l + 1
l
¸
m=−l
Y
l
m
(θ, φ)Y
l∗
m
(θ
z
, φ
z
) (4.65)
90 CHAPTER 4. LIE GROUPS
By Eq. ( 4.61) we get the addition theorem
P
l
(cos θ
) =
4π
2l + 1
l
¸
m=−l
Y
l∗
m
(θ
z
, φ
z
)Y
l
m
(θ, φ) (4.66)
For the special case of θ = θ
z
and φ = φ
z
we have θ
= 0 so that
l
¸
m=−l
Y
l∗
m
(θ, φ)Y
l
m
(θ, φ) =
2l + 1
4π
Thus the quantity on the left hand side is invariant under rotation since
the (θ, φ) is arbitrary.
A product of two spherical harmonics transforms according to the direct
product representation which is in general reducible. Using Eq. ( 4.64) in
Eq. ( 4.44) we obtain
Y
l
1
m
1
(θ, φ)Y
l
2
m
2
(θ, φ) =
l
1
+l
2
¸
L=l
1
−l
2

(2l
1
+ 1)(2l
2
+ 1)
4π(2L + 1)
C
L
0
(0, 0)C
L
M
(m
1
, m
2
)Y
L
M
((θ, φ)
(4.67)
where M = m
1
+m
2
. From this we can evaluate the frequently encountered
integral of three spherical harmonics:
dΩ Y
L∗
M
Y
l
1
m
1
Y
l
2
m
2
=
(2l
1
+ 1)(2l
2
+ 1)
4π(2L + 1)
C
L
0
(0, 0)C
L
M
(m
1
, m
2
)
(4.68)
Chapter 5
Atomic Physics
The Hamiltonian
ˆ
H of an atom in the absence of external ﬁelds is given by
ˆ
H =
¸
i
[t
i
−Z/r
i
] + 1/2
¸
i=j
1/r
ij
+
¸
i
ξ(r
i
)l
i
s
i
(5.1)
It is useful to break up
ˆ
H as follows:
ˆ
H =
ˆ
H
0
+
ˆ
H
1
+
ˆ
H
2
(5.2)
where
ˆ
H
0
=
¸
i
ˆ
h
0
(r
i
) =
¸
i
[t
i
−Z/r
i
+ V
eff
(r
i
)] (5.3)
ˆ
H
1
= 1/2
¸
i=j
1/r
ij
−
¸
i
V
eff
(r
i
) (5.4)
ˆ
H
2
=
¸
i
ξ(r
i
)l
i
s
i
(5.5)
V
eff
is a spherically symmetric singleparticle potential. The reason for the
above division of
ˆ
H is that
ˆ
H
0
is a singleparticle operator which can be
solved easily. By choosing V
eff
appropriately,
ˆ
H
1
may be made small so
that it can be treated as a perturbation.
ˆ
H
2
is usually small and can also
be treated as a perturbation and we assume that
ˆ
H
2
<<
ˆ
H
1
.
The objective of the present chapter is to set up the Hamiltonian ma
trix and exploit the symmetry properties of the Hamiltonian by means of
group theory in order to simplify our task. We consider the terms in the
Hamiltonian one by one starting with
ˆ
H
0
.
91
92 CHAPTER 5. ATOMIC PHYSICS
5.1
ˆ
H
0
ˆ
H
0
has two kinds of symmetry. It is invariant under permutations of the
electron coordinates (indistinguishability) and under separate rotations of
the electron coordinates. The eigenfunctions of
ˆ
H
0
must therefore form
basis functions for the irreducible representations of the permutation and
rotation groups. For a permutation group of N elements, there is always a
onedimensional irreducible representation whose basis functions are totally
antisymmetric, meaning that the functions change sign whenever two co
ordinates are interchanged. It does not follow from group theory that the
eigenfunctions of the Hamiltonian should be totally antisymmetric but for
fermions, these are the only ones found in nature (Pauli principle). These
antisymmetric functions can be written most conveniently as Slater de
terminants which are linear combinations of
¸
i
ψ
i
(r
i
), where
ˆ
h
0
ψ
i
=
i
ψ
i
,
with the coordinates permuted:
Ψ(1, 2, . . . , Z) =
1
√
Z
ψ
1
(1) ψ
2
(1) . . . ψ
Z
(1)
ψ
1
(2) ψ
2
(2) . . . ψ
Z
(2)
. . . . . . . . . . . .
. . . . . . . . . . . .
. . . . . . . . . . . .
ψ
1
(Z) ψ
2
(Z) . . . ψ
Z
(Z)
(5.6)
We have labeled the coordinates r
i
by i for simplicity. Take for example
the simplest nontrivial case of two electrons. The Slater determinant is
Ψ
12
(1, 2) =
1
√
2
[ψ
1
(1)ψ
2
(2) −ψ
1
(2)ψ
2
(1)]
which changes sign when the coordinates or the quantum labels 1 and 2
are interchanged Moreover, it becomes zero when the coordinates or the
quantum labels 1 = 2, which is a general property of a determinant that
it is zero whenever two rows or columns are the same. This is simply the
Pauli exclusion principle. Since
ˆ
H
0
is a sum of singleparticle operators, its
solution is just a single Slater determinant. From now on we assume that
basis functions needed to set up the secular equation for
ˆ
H must be Slater
determinants.
We now consider the rotational symmetry which implies that
[
ˆ
T(R),
ˆ
h
0
] = 0
5.2.
ˆ
H
0
+
ˆ
H
1
93
It follows that
[
ˆ
h
0
,
ˆ
L
x
] = [
ˆ
h
0
,
ˆ
L
y
] = [
ˆ
h
0
,
ˆ
L
z
] = [
ˆ
h
0
,
ˆ
L
2
] = 0
and the eigenstates of
ˆ
h
0
may be chosen to be simultaneous eigenstates
of
ˆ
L
2
and
ˆ
L
z
. Each energy level of
ˆ
h
0
is (2l + 1) degenerate since the
irreducible representations of R
3
are labelled by R
l
with dimensions (2l+1).
Neglecting spin, the eigenstates must have the following form:
ψ
nlm
(r) = u
nl
(r)Y
l
m
(θ, φ)
where Y
l
m
is an eigenfunction of
ˆ
L
z
and
ˆ
L
2
. Operating
ˆ
h
0
on ψ
nlm
shows
that u
nl
(r) must be a solution to the radial Schr¨odinger equation corre
sponding to
ˆ
h
0
, which is a familiar result.
We now take into account spin and label the spin states by χ
m
s
(σ)
where the spin coordinate σ can only take two discrete values σ = ±1:
χ
+
(1) = 1, χ
+
(−1) = 0, χ
−
(1) = 0, χ
−
(−1) = 1,
Physical observation shows that the spin behaves like angular momentum
so that
ˆ
T(R)χ
m
s
(σ) =
¸
m
s
χ
m
s
(σ)T
1/2
m
s
m
s
(R)
The spin states form a basis for the two dimensional irreducible representa
tion of R
3
corresponding to j = 1/2. The eigenstate of
ˆ
h
0
including spin
is
ψ
nlmm
s
(r, σ) = u
nl
(r)Y
l
m
(θ, φ)χ
m
s
(σ)
To summarise, the eigenfunctions of
ˆ
H
0
are simply single Slater deter
minants with the orbitals given by the eigenfunctions of
ˆ
h
0
. In determining
the eigenstates of (
ˆ
H
0
+
ˆ
H
1
) we will use these eigenfunctions as basis
functions.
5.2
ˆ
H
0
+
ˆ
H
1
We now consider (
ˆ
H
0
+
ˆ
H
1
) which is invariant under simultaneous rotations
of the position coordinates of the electrons. It is also implicitly invariant
under simultaneous rotations in the spin coordinates. Although (
ˆ
H
0
+
94 CHAPTER 5. ATOMIC PHYSICS
ˆ
H
1
) has no explicit spin dependence, it is assumed that each electron
carries a spin and the spin dependence enters implicitly through the Pauli
principle. The degenerate eigenfunctions of (
ˆ
H
0
+
ˆ
H
1
) must therefore
form an invariant subspace under simultaneous rotations of the space or
spin coordinates.
The inﬁnitesimal operators for the simultaneous rotations in space or
spin coordinates are given by the sum of the inﬁnitesimal operators corre
sponding to each electron coordinate. Thus
ˆ
L
x
=
¸
i
ˆ
L
x
(i),
ˆ
L
y
=
¸
i
ˆ
L
y
(i),
ˆ
L
z
=
¸
i
ˆ
L
z
(i)
where
ˆ
L
z
(i) acts on the coordinate r
i
. Similarly for the inﬁnitesimal op
erators in the spin space
ˆ
S
x
,
ˆ
S
y
, and
ˆ
S
z
. These six inﬁnitesimal operators
correspond to a direct product R
L
3
⊗R
S
3
since the orbital and spin operators
commute. (
ˆ
H
0
+
ˆ
H
1
) commutes with
ˆ
L
x
,
ˆ
L
y
,
ˆ
L
z
, and
ˆ
L
2
and with the spin
counterpart. As before we may choose the eigenfunctions of (
ˆ
H
0
+
ˆ
H
1
) to
be simultaneous eigenfunctions of
ˆ
L
z
and
ˆ
L
2
and of
ˆ
S
z
and
ˆ
S
2
. (
ˆ
H
0
+
ˆ
H
1
)
no longer commutes with the angular momentum or spin operators of each
electron. The eigenfunctions of
ˆ
H
0
are still eigenfunctions of
ˆ
L
z
and
ˆ
S
z
but not of
ˆ
L
2
or
ˆ
S
2
.
5.2.1 Conﬁguration and Term
In atoms, we need only consider the valence electrons because the electrons
forming a closed shell act eﬀectively like a spherically symmetric potential
which can be combined with V
eff
. We deﬁne a conﬁguration by a sym
bol such as s
1
p
1
meaning that there is one electron in an sshell and one
electron in a pshell. Thus the superscript indicates the number of elec
trons. We use the spectroscopic notations s, p, d, f, . . . corresponding to
l = 0, 1, 2, 3, . . . respectively.
When there are more than one electron, the representation for simul
taneous rotations of the electron coordinates is given by a direct product
of each rotation. According to the vectorcoupling theorem, the direct
product representation is in general reducible. Since (
ˆ
H
0
+
ˆ
H
1
) com
mute with rotation operators in space and spin coordinates separately,
the energy levels of (
ˆ
H
0
+
ˆ
H
1
) can be labelled with L and S and each
energy level is (2L + 1)(2S + 1) degenerate. The corresponding states
5.2.
ˆ
H
0
+
ˆ
H
1
95
are conventionally written as
2S+1
L and they are called ”term” (multi
plet). To illustrate the concept of term, consider two p electrons. Ac
cording to the vectorcoupling theorem, the irreducible representations for
simultaneous rotations of the two electron space coordinates correspond to
L = l
1
+l
2
, l
1
+l
2
−1, . . . , [l
1
−l
2
[ = 2, 1, 0 since l
1
= l
2
= 1. Similarly, the
irreducible representations for simultaneous rotations of the two electron
spin coordinates correspond to S = 1, 0 since s
1
= s
2
= 1/2. The possible
terms are then given by
3
D,
3
P,
3
S,
1
D,
1
P,
1
S. Some of these terms
may not be allowed by the Pauli exclusion principle when the two electrons
are equivalent as we shall see below.
5.2.2 Ladder Operator Method
As just discussed above, the eigenfunctions of
ˆ
H
0
are still eigenfunctions
of
ˆ
L
z
and
ˆ
S
z
but not of
ˆ
L
2
or
ˆ
S
2
. However, we know that we can al
ways generate an invariant subspace by applying the symmetry operators
repeatedly on an arbitrary function in that subspace. If we know one of the
eigenfunctions of
ˆ
L
2
and
ˆ
S
2
then we can construct the other degenerate
eigenfunctions by using the ladder operators. Let us illustrate this method
by considering a speciﬁc example since the generalisation to an arbitrary
case is quite clear.
Consider a conﬁguration p
2
and let us use the notation [ + 1
+
`[ −1
−
`
to represent a Slater determinant where one electron has m
l
= +1 and
m
s
= +1/2 and the other electron has m
l
= −1 and m
s
= −1/2. m
l
and m
s
are respectively the eigenvalues of
ˆ
L
z
and
ˆ
S
z
of the individual elec
trons. Let us list the possible Slater determinants corresponding to this
conﬁguration starting with the largest possible value of M
L
=
¸
i
m
l
(i).
We note also that the two electrons are equivalent because they occupy
the same p shell. There are (2 3)!/2!(2 3 −2)! = 15 possible ways of
placing two identical electrons in a pshell:
96 CHAPTER 5. ATOMIC PHYSICS
M
L
M
S
Ψ
1
= [1
+
`[1
−
` 2 0
Ψ
2
= [1
+
`[0
+
` 1 1
Ψ
3
= [1
+
`[0
−
` 1 0
Ψ
4
= [1
−
`[0
+
` 1 0
Ψ
5
= [1
−
`[0
−
` 1 −1
Ψ
6
= [1
+
`[ −1
+
` 0 1
Ψ
7
= [1
+
`[ −1
−
` 0 0
Ψ
8
= [1
−
`[ −1
+
` 0 0
Ψ
9
= [1
−
`[ −1
−
` 0 −1
Ψ
10
= [0
+
`[0
−
` 0 0
Ψ
11
= [0
+
`[ −1
+
` −1 1
Ψ
12
= [0
+
`[ −1
−
` −1 0
Ψ
13
= [0
−
`[ −1
+
` −1 0
Ψ
14
= [0
−
`[ −1
−
` −1 −1
Ψ
15
= [ −1
+
`[ −1
−
` −2 0
We keep in mind that Ψ
i
is a Slater determinant so that e.g. [1
+
`[1
−
` =
−[1
−
`[1
+
`. It is natural when using the ladder operator method to start
from a Slater determinant with the largest values of M
L
and/or M
S
since
it is more likely that it is an eigenstate of
ˆ
L
2
and/or
ˆ
S
2
. Ψ
1
has the largest
value of M
L
and it is the only one. Moreover, it has M
S
= 0 and therefore
it must be an eigenfunction belonging to singlet
1
D. We check that it is
an eigenfunction of
ˆ
L
2
and
ˆ
S
2
:
ˆ
L
+
Ψ
1
= [
ˆ
L
+
(1) +
ˆ
L
+
(2)][ + 1
+
`[ + 1
−
` = 0
ˆ
L
z
Ψ
1
= (
ˆ
L
z
(1) +
ˆ
L
z
(2))[ + 1
+
`[ + 1
−
`
= (1 + 1)[ + 1
+
+ 1
−
`
= 2Ψ
1
ˆ
L
2
z
Ψ
1
= 4Ψ
1
ˆ
L
2
Ψ
1
= 6Ψ
1
ˆ
S
+
Ψ
1
= [
ˆ
S
+
(1) +
ˆ
S
+
(2)][ + 1
+
`[ + 1
−
` = 0
ˆ
S
z
Ψ
1
= [
ˆ
S
z
(1) +
ˆ
S
z
(2)][ + 1
+
`[ + 1
−
` = 0
ˆ
S
2
Ψ
1
= 0
5.2.
ˆ
H
0
+
ˆ
H
1
97
We generate the invariant subspace formed by the degenerate eigenfunc
tions of
ˆ
L
2
corresponding to the term
1
D:
ˆ
L
−
Ψ
1
= [
ˆ
L
−
(1) +
ˆ
L
−
(2)][ + 1
+
`[ + 1
−
`
=
(1 + 1)(1 −1 + 1)[0
+
`[ + 1
−
`
+
(1 + 1)(1 −1 + 1)[ + 1
+
`[0
−
`
=
√
2(Ψ
3
−Ψ
4
)
We continue applying the lowering operator:
ˆ
L
−
(Ψ
3
−Ψ
4
) =
√
2(2Ψ
10
+ Ψ
7
−Ψ
8
)
ˆ
L
−
(2Ψ
10
+ Ψ
7
−Ψ
8
) = 3
√
2(Ψ
12
−Ψ
13
)
ˆ
L
−
(Ψ
12
−Ψ
13
) = 2
√
2Ψ
15
ˆ
L
−
Ψ
15
= 0
The functions Ψ
1
, (Ψ
3
−Ψ
4
), (Ψ
7
−Ψ
8
+ 2Ψ
10
), (Ψ
12
−Ψ
13
), and Ψ
15
are degenerate eigenfunctions of
ˆ
L
2
and
ˆ
S
2
corresponding to the term
1
D.
Ψ
2
must be an eigenfunction belonging to
3
P because it is the only one
with M
L
= 1 and M
S
= 1 and the term
3
D does not exist. We operate
ˆ
S
−
on Ψ
2
to obtain the triplets with M
L
= 1.
ˆ
S
−
Ψ
2
=
ˆ
S
−
[1
+
`[0
+
`
= [1
−
`[0
+
` +[1
+
`[0
−
`
= Ψ
4
+ Ψ
3
ˆ
S
−
(Ψ
4
+ Ψ
3
) = 2Ψ
5
ˆ
S
−
Ψ
5
= 0
Operating
ˆ
L
+
on Ψ
2
yields a state [ + 1
+
`[ + 1
+
` which is forbidden
by the Pauli exclusion principle. This state would be allowed if the two
electrons occupied diﬀerent p shells. We operate
ˆ
L
−
on Ψ
2
and then use
ˆ
S
−
to generate the triplets with M
L
= 0:
ˆ
L
−
Ψ
2
=
√
2Ψ
6
ˆ
S
−
Ψ
6
= Ψ
8
+ Ψ
7
ˆ
S
−
(Ψ
8
+ Ψ
7
) = 2Ψ
9
ˆ
S
−
Ψ
9
= 0
98 CHAPTER 5. ATOMIC PHYSICS
Continuing along the same fashion we generate the triplets with M
L
=
−1
ˆ
L
−
Ψ
6
=
√
2Ψ
11
ˆ
S
−
Ψ
11
= Ψ
13
+ Ψ
12
ˆ
S
−
(Ψ
13
+ Ψ
12
) = 2Ψ
14
ˆ
S
−
Ψ
14
= 0
We have generated 14 eigenfunctions. The only one left must belong
to the term
1
S. There are three Slater determinants, Ψ
7
, Ψ
8
, Ψ
10
, with
M
L
= 0 and M
S
= 0. We have already found two eigenfunctions Ψ
7
−Ψ
8
+
2Ψ
10
and Ψ
7
+Ψ
8
belonging to
1
D and
3
P respectively. The eigenfunction
belonging to
1
S must be Ψ
7
−Ψ
8
−Ψ
10
since it is orthogonal to the previous
two eigenfunctions.
Below we summarise our results, where the eigenfunctions have been
normalised:
Term M
L
M
S
Eigenfunction
1
D 2 0 Ψ
1
1 0 (Ψ
3
−Ψ
4
)/
√
2
0 0 (Ψ
7
−Ψ
8
+ 2Ψ
10
)/
√
6
1 0 (Ψ
12
−Ψ
13
)/
√
2
2 0 Ψ
15
3
P 1 1 Ψ
2
0 (Ψ
3
+ Ψ
4
)/
√
2
1 Ψ
5
0 1 Ψ
6
0 (Ψ
7
+ Ψ
8
)/
√
2
1 Ψ
9
1 1 Ψ
11
0 (Ψ
12
+ Ψ
13
)/
√
2
1 Ψ
14
1
S 0 0 (Ψ
7
−Ψ
8
−Ψ
10
)/
√
3
The method that we have used to construct the eigenfunctions of
ˆ
L
2
and
ˆ
S
2
and hence of (
ˆ
H
0
+
ˆ
H
1
) is systematic but it becomes tedious as
5.2.
ˆ
H
0
+
ˆ
H
1
99
the number of Slater determinants increases. We discuss below a diﬀerent
method which is more suitable for the latter case.
5.2.3 Projection Operator Method
Let us illustrate the method by using the example we have just considered.
Each of the Slater determinant Ψ
1
, Ψ
2
, . . . must be a linear combination
of eigenfunctions belonging to
1
D,
3
P, and
1
S. It is best to start with a
Slater determinant with M
L
= 0 and M
S
= 0 since it is contained in all
the terms. Let us take Ψ
7
and suppose that we are interested in ﬁnding
the component of
1
D. We ﬁrst project out the component of
3
P out of
Ψ
7
and then we project out the comoponent of
1
S:
ˆ
L
2
−1(1 + 1)
Ψ
7
= 2Ψ
10
ˆ
L
2
−0(0 + 1)
Ψ
10
= 2Ψ
10
+ Ψ
7
−Ψ
8
This is proportional to the eigenfunction of
1
D with M
L
= 0 and M
S
= 0
which is in agreement with the result obtained by using the ladder operator
method.
We can arrive at the same result if we ﬁrst ﬁnd the eigenfunction of
ˆ
S
2
with S = 0 by projecting out the component with S = 1:
[
ˆ
S
2
−1(1 + 1)]Ψ
7
= Ψ
7
+ Ψ
8
−2Ψ
7
= Ψ
8
−Ψ
7
Thus (Ψ
8
−Ψ
7
) is an eigenfunction of
ˆ
S
2
corresponding to S = 0 (singlet).
Now we project out the component of
1
S. It is not necessary to project
out the component of
3
P because (Ψ
8
− Ψ
7
) is a singlet and therefore it
cannot contain eigenfunctions of
3
P.
[
ˆ
L
2
−0(0 + 1)](Ψ
8
−Ψ
7
) = 2(−Ψ
10
+ Ψ
8
) −2(Ψ
10
+ Ψ
7
)
= −2(Ψ
7
−Ψ
8
+ 2Ψ
10
)
which is proportional to the previous result.
From a given Slater determinant, the projection operator method allows
us to arrive at a particular eigenfunction of a term by projecting out all the
other terms one by one.
100 CHAPTER 5. ATOMIC PHYSICS
5.2.4 Term Energies
In this section we discuss the calculations of term energies. To calculate
the term energies, we need matrix elements of the type
'Ψ
i
[
ˆ
H
0
+
ˆ
H
1
[Ψ
j
`
It is convenient to calculate them in the occupation number representation
in which (
ˆ
H
0
+
ˆ
H
1
) is given by
ˆ
H
0
+
ˆ
H
1
=
¸
ij
ˆ c
i
'i[
ˆ
h
0
[j`ˆ a
j
+
1
2
¸
ijkl
ˆ c
i
ˆ c
j
'ij[ˆ v[kl`ˆ a
l
ˆ a
k
(5.7)
where
'i[
ˆ
h
0
[j` = δ(s
i
, s
j
)
d1 ψ
∗
i
(1)h
0
(1)ψ
j
(1), 1 ≡ r
1
'ij[ˆ v[kl` = δ(s
i
, s
k
)δ(s
j
, s
l
)
d1d2
ψ
∗
i
(1)ψ
∗
j
(2)ψ
k
(1)ψ
l
(2)
[1 −2[
ˆ c
i
= ˆ a
†
i
and ˆ a
i
= ˆ c
†
i
are creation and annihilation operators respectively
which obey the commutation relations
[ˆ c
i
, ˆ c
j
]
+
= 0, [ˆ a
i
, ˆ a
j
]
+
= 0, [ˆ c
i
, ˆ a
j
]
+
= δ
ij
(5.8)
We write the Slater determinants Ψ
i
in the occupation number repre
sentation and order the singleparticle states ψ
i
as follows
[1
+
, 1
−
, 0
+
, 0
−
, −1
+
, −1
−
` = ˆ c
1
+ ˆ c
1
− ˆ c
0
+ ˆ c
0
− ˆ c
−1
+ ˆ c
−1
−[0`
where [0` is a state with no electrons in the p shell. We note that for
fermions the ordering of the states is very important. Thus we have
Ψ
1
= ˆ c
1
+ ˆ c
1
−[0`
Ψ
2
= ˆ c
1
+ ˆ c
0
+[0`
.
.
Ψ
15
= ˆ c
−1
+ ˆ c
−1
−[0`
5.2.
ˆ
H
0
+
ˆ
H
1
101
The matrix elements of
ˆ
H
1
between Ψ = ˆ c
p
ˆ c
q
[0` and Ψ
= ˆ c
r
ˆ c
s
[0` will
be of the following form:
'Ψ[
ˆ
H
1
[Ψ
` =
1
2
¸
ijkl
'ij[ˆ v[kl`'0[ˆ a
q
ˆ a
p
ˆ c
i
ˆ c
j
ˆ a
l
ˆ a
k
ˆ c
r
ˆ c
s
[0`
= 'pq[ˆ v[rs` −'pq[ˆ v[sr`
The ﬁrst term is the direct term and the second term is the so called
exchange term.
The matrix elements of
ˆ
H
0
between Ψ = ˆ c
p
ˆ c
q
[0` and Ψ
= ˆ c
r
ˆ c
s
[0` are
quite simple:
'Ψ[
ˆ
H
0
[Ψ
` =
¸
ij
'i[
ˆ
h
0
[j`'0[ˆ a
q
ˆ a
p
ˆ c
i
ˆ a
j
ˆ c
r
ˆ c
s
[0`
= ['p[
ˆ
h
0
[p` +'q[
ˆ
h
0
[q`](δ
pr
δ
qs
−δ
ps
δ
qr
)
We have made use of the fact that ψ
i
are eigenfunctions of
ˆ
h
0
so that
'i[
ˆ
h
0
[j` = δ
ij
.
ˆ
H
0
is diagonal because the Slater determinants are eigen
functions of
ˆ
H
0
and they are degenerate. Therefore
ˆ
H
0
only contributes
the same constant term to all the terms.
We must also evaluate the integral 'pq[ˆ v[rs` before we can calculate the
term energies. To do this, we use the multipole expansion of the Coulomb
potential:
1
[r
1
−r
2
[
=
¸
lm
4π
2l + 1
r
l
<
r
l+1
>
Y
l∗
m
(Ω
1
)Y
l
m
(Ω
2
) (5.9)
where r
>
and r
<
are the greater and the smaller of r
1
and r
2
. We recall
that ψ
i
= φ
l
(r)Y
l
m
(Ω) where φ
l
(r) is the solution to the radial Schr¨odinger
equation.
'ψ
1
ψ
2
[ˆ v[ψ
3
ψ
4
` =
d
3
r
1
d
3
r
2
¸
lm
4π
2l + 1
r
l
<
r
l+1
>
φ
l
1
(r
1
)φ
l
2
(r
2
)φ
l
3
(r
1
)φ
l
4
(r
2
)
dΩ
1
Y
l∗
m
(Ω
1
)Y
l
1
∗
m
1
(Ω
1
)Y
l
3
m
3
(Ω
1
)
dΩ
2
Y
l
m
(Ω
2
)Y
l
2
∗
m
2
(Ω
2
)Y
l
4
m
4
(Ω
2
)
The integrals over three spherical harmonics are given by Eq. ( 4.68) and
the ClebschGordan coeﬃcients are tabulated extensively in many books
on atomic physics.
102 CHAPTER 5. ATOMIC PHYSICS
Let us calculate the energy of the term
1
D. Since all the eigenfunctions
of
1
D are degenerate we can take the state Ψ
1
:
'Ψ
1
[
ˆ
H
1
[Ψ
1
` = 'pq[ˆ v[rs` −'pq[ˆ v[sr`
=
d
3
r
1
d
3
r
2
¸
lm
4π
2l + 1
r
l
<
r
l+1
>
φ
2
1
(r
1
)φ
2
1
(r
2
)
dΩ
1
Y
l∗
m
(Ω
1
)Y
1∗
1
(Ω
1
)Y
1
1
(Ω
1
)
dΩ
2
Y
l
m
(Ω
2
)Y
1∗
1
(Ω
2
)Y
1
1
(Ω
2
)
The second term (exchange term) vanishes because of the orthogonality of
the spin states.
5.3
ˆ
H =
ˆ
H
0
+
ˆ
H
1
+
ˆ
H
2
Finally we include the spinorbit term
ˆ
H
2
which is invariant under simulta
neous rotations in space and spin coordinates. This implies
[
ˆ
H,
ˆ
J
x
] = [
ˆ
H,
ˆ
J
y
] = [
ˆ
H,
ˆ
J
z
] = [
ˆ
H,
ˆ
J
2
] = 0
where
ˆ
J
x
=
¸
i
[
ˆ
L
x
(i) +
ˆ
S
x
(i)] etc.
ˆ
H does not commute with
ˆ
L
i
or
ˆ
S
i
and the group is now reduced to R
3
.
This means that the (2L + 1)(2S + 1) degeneracy of each term will be in
general split and each energy level of
ˆ
H is now (2J + 1) degenerate with
J = L+S, L+S −1, . . . , [L−S[ which follows from the vectorcoupling
rule R
L
⊗R
S
=
¸
L+S
J=L−S
R
J
. The J dependence of the splitting for given
L, S may be evaluated by using the WignerEckart theorem. We assume
that
ˆ
H
2
is small enough so that it causes no mixing between the diﬀerent
terms. Let ψ
JM
be an eigenstate of
ˆ
J
2
which is also an eigenstate of
ˆ
L
2
and
ˆ
S
2
since it is formed as a linear combination of the eigenstates of a
given term. We have
∆
J
= (ψ
JM
,
¸
i
ξ(r
i
)l
i
s
i
ψ
JM
)
= A(ψ
JM
, L S ψ
JM
)
=
1
2
A[J(J + 1) −L(L + 1) −S(S + 1)]
5.3.
ˆ
H =
ˆ
H
0
+
ˆ
H
1
+
ˆ
H
2
103
The second step follows from the fact that L S = 1/2(
ˆ
J
2
−
ˆ
L
2
−
ˆ
S
2
)
transforms exactly like
¸
i
ξ(r
i
)l
i
s
i
and the fact that R
3
is simply reducible
i.e. m = 1 in the WignerEckart theorem. The actual magnitude of the
splitting depends on the details of the wavefunctions which are contained
in A.
104 CHAPTER 5. ATOMIC PHYSICS
Chapter 6
The Group SU
2
: Isospin
It is found experimentally that the strong nuclear force that binds nucleons
(protons and neutrons) is the same within 1 % between protonproton,
protonneutron and neutronneutron. We may regard a proton and a neu
tron as two diﬀerent states of a nucleon. The Hamiltonian describing the
nucleus is invariant under the interchange of proton and neutron states.
This symmetry is called isospin. The group corresponding to isospin is
called SU
2
, which is homomorphic with the rotation group R
3
and the
mathematical description is identical to that of the spin 1/2. This means
that the results we have derived for the group R
3
may be readily used here.
6.1 The Group SU
2
A 2 2 unitary matrix has four independent parameters and it may be
written as
U = exp(iθ)
α β
γ δ
Since UU
†
= 1, we have mod(det U) = 1 and we may choose αδ−βγ = 1.
A little algebra shows that δ = α
∗
and γ = −β
∗
so that
U = exp(iθ)
α β
−β
∗
α
∗
105
106 CHAPTER 6. THE GROUP SU
2
: ISOSPIN
The set of matrices U with θ = 0 and determinant = 1 form a subgroup
of U
2
, called SU
2
(special unitary). Thus, SU
2
is a group with elements
U =
α β
−β
∗
α
∗
(6.1)
and the condition [α[
2
+[β[
2
= 1. We should note that these matrices are
themselves the elements of the group SU
2
and they may be regarded as
representations of rotation operators in a twodimensional Hilbert space.
Since U are unitary, the corresponding inﬁnitesimal matrices must be
skew Hermitian which can be made Hermitian by including the factor i:
1 =
1 0
0 1
σ
x
=
0 1
1 0
σ
y
=
0 −i
i 0
σ
z
=
1 0
0 −1
Since the determinant of SU
2
is +1, its inﬁnitesimal matrices must have
zero trace and the unit matrix is therefore excluded. The remaining three
matrices are just the Pauli spin matrices which are also the inﬁnitesimal
matrices of the rotation group R
3
corresponding to j = 1/2. However,
the relationship between the two group elements of SU
2
and R
3
is not
necessarily onetoone (isomorphic).
Physically, we may generate the group SU
2
by considering two nucleon
states,
[p` = c
+
p
[0`, [n` = c
+
n
[0`
corresponding to proton and neutron. The bilinear products of the annihi
lation and creation operators that conserve particle number are
c
+
p
c
p
, c
+
n
c
n
, c
+
p
c
n
, c
+
n
c
p
We deﬁne a new set of operators
ˆ
B = c
+
p
c
p
+ c
+
n
c
n
ˆ τ
+
= c
+
p
c
n
ˆ τ
−
= c
+
n
c
p
ˆ τ
z
=
1
2
(c
+
p
c
p
−c
+
n
c
n
) =
ˆ
Q−
1
2
ˆ
B
which have the following Lie algebra
[
ˆ
B, ˆ τ
+
] = [
ˆ
B, ˆ τ
−
] = [
ˆ
B, ˆ τ
z
] = 0
6.2. RELATIONSHIP BETWEEN SU
2
AND R
3
107
[ˆ τ
z
, τ
+
] = τ
+
, [ˆ τ
z
, τ
−
] = τ
−
, [ˆ τ
+
, ˆ τ
−
] = 2ˆ τ
z
The operator
ˆ
B may be interpreted as the Baryon number operator and
it is an invariant reﬂecting the conservation of Baryon number. The three
operators ˆ τ
+
, ˆ τ
−
, and ˆ τ
z
form a Lie algebra by themselves and it is identical
to the Lie algebra of the rotation group R
3
. The two dimensional repre
sentations of these operators are obtained by evaluating matrix elements
such as (τ
+
)
12
= 'p[c
+
p
c
n
[n` = 1 and this gives:
B = 1 τ
+
=
1
2
(σ
x
+iσ
y
), τ
−
=
1
2
(σ
x
−iσ
y
), τ
z
=
1
2
σ
z
In physical applications, it is usually the Lie algebra which is important
rather than the Lie group. Two diﬀerent Lie groups may have the same
Lie algebra. As we have seen, the rotation group R
3
is not the same as
the group SU
2
although they have the same Lie algebra.
6.2 Relationship between SU
2
and R
3
To relate the elements of R
3
to those of SU
2
we consider explicitly the
twodimensional representation of R
3
corresponding to j = 1/2. A rotation
about an axis a by an angle φ is represented by
D
1/2
(a, φ) = exp(φˆ a X) (6.2)
with X = −iσ/2. The anticommutation relations of the inﬁnitesimal
matrices X are given by
¦X
i
, X
j
¦ = −
1
2
δ
ij
(6.3)
which implies
(a X)
2
=
¸
ij
a
i
a
j
X
i
X
j
= −
1
4
¸
i
a
2
i
E = −
1
4
E
sine ˆ a is a unit vector. We can now easily work out the representation
matrix D
1/2
(a, φ)
D
1/2
(a, φ) = exp(φˆ a X)
108 CHAPTER 6. THE GROUP SU
2
: ISOSPIN
= 1 + φ(ˆ a X) +
1
2
φ
2
(ˆ a X)
2
+
1
6
φ
3
(ˆ a X)
3
+
1
24
φ
4
(ˆ a X)
4
+ . . .
= 1 +
1
2
φ
2
(ˆ a X)
2
+
1
24
φ
4
(ˆ a X)
4
+. . .
+(ˆ a X)[φ +
1
6
φ
3
(ˆ a X)
2
+ . . .]
= 1 −
1
2
(φ/2)
2
+
1
4
(φ/2)
4
+ . . .
+2(ˆ a X)[φ/2 −
1
3
(φ/2)
3
+ . . .]
= cos φ/2 + 2(ˆ a X)sin φ/2
=
cos φ/2 −ia
z
sin φ/2 −(a
y
+ ia
x
)sin φ/2
(a
y
−ia
x
)sin φ/2 cos φ/2 + ia
z
sin φ/2
These matrices are the matrices of SU
2
in Eq. ( 6.1) if we make the
identiﬁcation α = cos φ/2−ia
z
sin φ/2 and β = −(a
y
+ia
x
)sin φ/2 with
0 ≤ φ < 2π and ˆ a pointing in all possible directions. The representation
D
1/2
for the rotation group is however doublevalued because there are two
matrices corresponding to the same rotation. This can be seen by noting
that, to produce all rotations, the angle φ need only cover either 0 to π,
or π to 2π since
R(a, φ) = R(−a, −φ) = R(−a, 2π −φ)
The D
1/2
representations corresponding to this same physical rotation are
diﬀerent because
D
1/2
(−a, 2π −φ) = −D
1/2
(a, φ)
The relationship between SU
2
and R
3
is thus homomorphic because there
are two distinct SU
2
matrices corresponding to the same physical rotation.
The representation D
1/2
for the rotation group can be made singlevalued
by extending the physical rotation angle artiﬁcially to 4π (similar to Rie
mannian sheets in complex plane). A rotation R(a, 2π + φ) is regarded
as a new rotation diﬀerent from R(a, φ) whereas R(a, 4π +φ) = R(a, φ)
since D
1/2
(a, 4π + φ) = D
1/2
(a, φ). A rotation such as R(−a, 2π −φ)
is to be regarded as equivalent to a rotation R(a, 2π + φ) because
R(−a, 2π −φ) = R(a, −(2π −φ)) = R(a, 4π −(2π −φ)) = R(a, 2π + φ).
6.3. SU
2
IN NUCLEI 109
6.3 SU
2
in Nuclei
As we have discussed in the previous section, the inﬁnitesimal operators
of the group SU
2
have the same commutation relations as those of the
rotation group R
3
. Since the irreducible representations of R
3
were deduced
entirely from the commutation relations of the inﬁnitesimal operators, it
follows that we have the same irreducible representations in SU
2
which we
may label with D
T
, T = 0, 1/2, 1, 3/2 . . ., as in R
3
. T is called the isospin,
similar to L for angular momentum in R
3
.
We recall that when a Hamiltonian is invariant under R
3
, the eigen
functions transforming according to the irreducible representations D
L
are
degenerate labelled by the z components of angular momentum M
L
=
L, L − 1, . . . , −L. Similarly, when a Hamiltonian is invariant under SU
2
,
the degenerate eigenfunctions tranforming according to the irreducible rep
resentation D
T
will have zcomponents of isospin M
T
= T, T −1, . . . , −T.
These multiplets are interpreted as diﬀerent possible states of nuclei. For
example, the twodimensional space of proton and neutron, [p` and [n`,
tranforms according to D
1/2
and consequently the eigenfunctions of a sin
gle nucleon will have M
T
= 1/2, −1/2. Conventionally, M
T
= 1/2 corre
sponds to [p` and M
T
= −1/2 to [n`.
Physically, the operator
ˆ
Q =
1
2
ˆ
B+
ˆ
T
z
is regarded as a charge operator.
A given nucleus with a ﬁxed neutron number N and proton number Z
corresponds to a state with M
T
= (Z − N)/2 since 'Q` = Z. This
state may correspond to a state with T = [Z −N[/2 or higher. It is found
experimentally that the ground state of a nucleus has the minimum value of
the total isospin T whereas in an atom, the ground state has the maximum
value of the total spin S (Hund’s rule). This is because the forces between
the nucleons are attractive whereas the forces between the electrons are
repulsive. The physical interpretation of this result is that a nucleus with
a ﬁxed mass number A = N +Z can exist in diﬀerent nucleon states with
zcomponents of isospin M
T
= (Z −N)/2, where Z and N may take any
values constrained by the mass number. For a given nucleus with ﬁxed N
and Z, states with M
T
= (Z − N)/2 but T > [Z − N[/2 correspond to
excited states.
For the general case of several nucleons, we consider simultaneous SU
2
transformation in every nucleon. To ﬁnd the irreducible representations of
two nucleons, we use the same vectorcoupling theory as in the angular
110 CHAPTER 6. THE GROUP SU
2
: ISOSPIN
momentum theory:
D
T
1
⊗D
T
2
=
T
1
+T
2
¸
T
1
−T
2

⊕D
T
which may be extended to an arbitrary number of nucleons.
As a ﬁrst example, we consider two nucleons. From the vectorcoupling
theory, we have states with T = 1 and T = 0. The ground state will
correspond to a singlet with T = 0 and the excited states to a triplet with
T = 1 with M
T
= 1, 0, −1. To work out these states, let us write down
explicitly the four vectors that form the space of two nucleons:
[1` = [p
1
p
2
` [2` = [p
1
n
2
` [3` = [n
1
p
2
` [4` = [n
1
n
2
`
The vectors [1` and [4` have M
T
= 1 and −1 respectively and therefore
belong to T = 1 states. To ﬁnd the other states, we apply the lowering
operator
ˆ
T
−
=
ˆ
t
−
(1) +
ˆ
t
−
(2)
on the vector [1` which conserves the value of T but lowers the value of
M
T
by one:
ˆ
T
−
[1` = [n
1
p
2
` +[p
1
n
2
`
= [2` +[3`
The resulting vector has M
T
= 0 and we conclude that the T = 1 states
must be
[1`, 1/
√
2([2` +[3`), [4`
The remaining vector
1/
√
2([2` −[3`)
must have T = 0 as may be checked by considering
ˆ
T
2
([2` −[3`) = (
ˆ
T
+
ˆ
T
−
+
ˆ
T
2
z
−
ˆ
T
z
)([n
1
p
2
` −[p
1
n
2
`)
= 0
Another example of isospin labelling is provided by nuclei with A = 13.
The lowest value of T = 1/2 must be given by the nuclei with Z = 6, N =
7, M
T
= −1/2 (C
13
) and Z = 7, N = 6, M
T
= 1/2 (N
13
). Since
A = 13 is odd, it is not possible to have states with T = 1. The next higher
6.4. TENSOR OPERATORS 111
states must be formed by the nuclei with Z = 5, N = 8, M
T
= −3/2
and Z = 8, N = 5, M
T
= 3/2 corresponding to T = 3/2. The remaining
two states with M
T
= ±1/2 must correspond to excited states of C
13
and
N
13
.
6.4 Tensor Operators
Tensor operators are irreducible sets of operators associated in particular
with the rotation group R
3
. The general deﬁnition of an irreducible set of
operators ¦
ˆ
O
α
i
¦ is
ˆ
T(G)
ˆ
O
α
i
ˆ
T
−1
(G) =
¸
k
O
α
k
T
α
ki
(G)
In Lie groups,
ˆ
T(a) = 1 +
¸
q
a
q
ˆ
X
q
for small a so that
ˆ
T(a)
ˆ
O
α
i
ˆ
T
−1
(a) = (1 +
¸
q
a
q
ˆ
X
q
)
ˆ
O
α
i
(1 −
¸
q
a
q
ˆ
X
q
)
=
ˆ
O
α
i
+
¸
q
a
q
[
ˆ
X
q
,
ˆ
O
α
i
]
Since
¸
k
O
α
k
T
α
ki
(a) =
¸
k
O
α
k
¸
δ
ki
+
¸
q
a
q
(X
α
q
)
ki
¸
we have
[
ˆ
X
q
,
ˆ
O
α
i
] =
¸
k
O
α
k
(X
α
q
)
ki
(6.4)
which is the equivalent deﬁnition of irreducible set of operators in Lie
groups.
The relations deﬁning tensor operators associated with the rotation
group are
[
ˆ
J
z
,
ˆ
O
j
m
] =
¸
m
ˆ
O
j
m
(ψ
j
m
,
ˆ
J
z
ψ
j
m
) = m
ˆ
O
j
m
[
ˆ
J
±
,
ˆ
O
j
m
] =
¸
m
ˆ
O
j
m
(ψ
j
m
,
ˆ
J
±
ψ
j
m
) =
(j ±m + 1)(j ∓m)O
j
m±1
112 CHAPTER 6. THE GROUP SU
2
: ISOSPIN
j is called the rank of the tensor and the number of operators in a set is
(2j +1) with m = −j, −j +1, . . . , j. For example, the angular momentum
operators themselves form a tensor operator of rank 1 (vector operator):
ˆ
O
1
1
= −
ˆ
J
+
√
2,
ˆ
O
1
0
=
ˆ
J
z
,
ˆ
O
1
−1
=
ˆ
J
−
√
2
Chapter 7
The Point Groups
7.1 Crystal Symmetry
A lattice is deﬁned by three linearly independent vectors a
1
, a
2
, and a
3
and a crystal is a regular array of lattices such that it is invariant under
translations t = n
1
a
1
+ n
2
a
2
+ n
3
a
3
. Apart from translations, there are
also symmetry operations which are performed with a point ﬁxed (origin)
and which form a point group. The fundamental covering operations of
point groups are
1. Rotations about axes through the origin.
2. Inversion
The complete set of symmetry operations including translations is called
the space group.
7.2 Schoenﬂies Notation
There are several notations used in describing point groups. We follow the
notation due to Schoenﬂies.
E = identity
C
n
= rotation through 2π/n.
σ = reﬂection in a plane.
113
114 CHAPTER 7. THE POINT GROUPS
σ
h
= reﬂection in the plane through the origin and perpendicular to the
axis of highest rotation symmetry.
σ
v
= reﬂection in the plane through the axis of highest rotation symmetry.
σ
d
= reﬂection in the plane through the axis of highest rotation symmetry
and bisecting the angle between the twofold axes perpendicular to the
symmetry axis. This is a special kind of σ
v
.
S
n
= improper rotation through 2π/n. An improper rotation is a rotation
followed by a reﬂection in a plane perpendicular to the axis of rotation.
i = S
2
= inversion.
7.3 Commuting Operations
We list pairs of operations which commute. They are useful for working
out group multiplication tables.
1. Two rotations about the same axis.
2. Two reﬂections in perpendicular planes
3. Two rotations by π about perpendicular axes.
4. A rotation and a reﬂection in a plane perpendicular to the axis of
rotation
5. The inversion and any rotation or reﬂection
The following rules are useful in determining the complete symmetry of
the problem.
1. The intersection of two reﬂection planes is a symmetry axis. The
axis is n−fold if the angle between the planes is π/n. If a reﬂection
plane contains an n−fold axis, there must be n −1 other reﬂection
planes at angles of π/n.
2. If two 2−fold axes make an angle π/n, then there must be a per
pendicular n−fold axis. A twofold axis and an n−fold axis implies
the existence of n − 1 additional twofold axes separated by angles
of π/n.
7.4. ENUMERATION OF POINT GROUPS 115
3. Any two of the following: an evenfold axis, a reﬂection plane per
pendicular to it, and an inversion centre implies the existence of the
other.
7.4 Enumeration of Point Groups
The point groups may be divided into two categories:
1. Simple rotation groups which have one symmetry axis of highest
symmetry.
2. Higher symmetry groups which have no unique axis of highest sym
metry, but which have more than one n−fold axis where n > 2.
It is convenient to visualise the point group operations by means of
stereographic projections as shown in Figure 7.1. These are formed by
projecting on to the XY plane a general point on a unit sphere which is
subjected to the symmetry operations. The labels +/◦ mean that the
point is above/below the plane respectively. The nfold rotation axes are
indicated by the symbols at the centre of the circle. Solid radial lines
indicate vertical reﬂection planes and dashed radial lines indicate rotation
axes. Horizontal reﬂection planes are indicated by unit circles with solid
lines rather than broken lines.
There are thirty two point groups and we list them in order of increasing
complexity.
Simple rotation groups:
C
n
• These are groups of rotations about n−fold axes. They are cyclic
groups of order n. It can be shown that n can only be 1, 2, 3, 4, and
6. Other rotations are not consistent with translational symmetry. The
groups are therefore C
1
, C
2
, C
3
, C
4
, and C
6
.
Proof: If t is a translational vector and t
= Rt, then the translational
vector t
− t must be perpendicular to the rotation axis. Let s be the
shortest translational vector perpendicular to the axis of rotation. Apply
116 CHAPTER 7. THE POINT GROUPS
ing C
n
on s generates a new vector s’ and we must have
[s
−s[ = 2ssin π/n ≥ s
since s is deﬁned to be the smallest translational vector perpendicular to
the axis of rotation. It follows immediately that n ≤ 6 and geometrical
consideration shows that n = 5 is not possible.
C
nh
• These groups have a σ
h
reﬂection plane in addition to the C
n
axis. When n is even, the group includes the inversion element so that
C
2nh
= C
2n
⊗I. The groups are C
1h
, C
2h
= C
2
⊗I, C
3h
, C
4h
= C
4
⊗I,
and C
6h
= C
6
⊗I.
C
nv
• These groups have a σ
v
reﬂection plane in addition to the C
n
axis.
According to rule 1 there must also be n reﬂection planes separated by an
angle π/n around the C
n
axis. C
1v
is identical to C
1h
. The groups are
C
2v
, C
3v
, C
4v
, and C
6v
.
S
2n
• These groups have an 2nfold axis for improper rotations and each
has C
n
as a subgroup. The groups are S
2
, S
4
, and S
6
= C
3
⊗I.
D
n
• These groups have n twofold axes perpendicular to the C
n
axis.
The groups are D
2
, D
3
, D
4
, D
6
.
D
nh
• Each of these groups has D
n
as a subgroup and in addition it
contains the horizontal reﬂection plane σ
h
. Since σ
h
commutes with rota
tions, D
nh
= D
n
⊗I.
D
nd
• Each of these groups has D
n
as a subgroup and in addition it con
tains diagonal reﬂection planes σ
d
bisecting the angles between the twofold
axes perpendicular to the principal rotation axis. The groups are D
2d
and
D
3d
.
Higher symmetry groups:
T • This group consists of 12 proper rotations that take a regular tetrahe
dron into itself. This is shown in Figure 7.2. The elements are E, 3 C
2
’s
about the X, Y, and Z axes and 8 C
3
’s about the body diagonals which
7.4. ENUMERATION OF POINT GROUPS 117
consist of two classes, clockwise and anticlockwise rotations by 2π/3.
T
d
• This group has T as a subgroup and in addition it contains 6 diagonal
reﬂection planes σ
d
normal to a cube face and containing a tetrahedral
edge such as ac or ab. The presence of 6 σ
d
’s implies the existence of 6
S
4
’s about the X, Y and Z axes. The total number of elements is 24.
T
h
• This group has T as a subgroup and in addition it contains inver
sion so that T
h
= T ⊗I.
O • This is one of the most important point groups, shown in Figures
7.3a and 7.3b. This group consists of 24 proper rotations that take an
octahedron into itself. These are E, 8 C
3
’s about the body diagonals, 3
C
2
’s and 6 C
4
’s about the X, Y, and Z axes, and 6 C
2
’s about axes through
thr origin parallel to face diagonals. All the C
3
’s are in the same class but
the two kinds of C
2
operations form two distinct classes.
O
h
• This group has O as a subgroup and in addition it contains inversion
so that O
h
= O ⊗I.
The following point groups are useful to know but they are not consis
tent with translational symmetry because they contain ﬁvefold axes.
Y • This group consists of 60 proper rotations that take an icosahedron
or a dodecahedron into itself. This is illustrated in Figures 7.4a and 7.4b.
Y
h
• This group has Y as a subgroup and in addition it contains inversion
so that Y
h
= Y ⊗I.
We list also in Table 7.1 the classiﬁcation of the 32 point groups according
to the crystal systems to which they belong. The unit cell required for
compatibility with the pointgroup symmtery is speciﬁed by three trans
lations a, b, and c and three angles α, β, and γ as illustrated in Figure
7.5.
118 CHAPTER 7. THE POINT GROUPS
7.5 Improper Point Groups
The product of two improper elements is a proper element since an improper
element has a determinant −1. It follows also that a product of a proper
and an improper element is an improper element. Let G be an improper
group. We may write
G = P + Q
wher P and Q contain proper and improper elements respectively. Let i
be any improper element. Then from the rearrangement theorem we have
G = iP + iQ
and therefore iP = Q and iQ = P so that
G = P(1 + i)
If i commutes with all elements in the group, then
G = P ⊗I
where I = (1, i). For example, i can be the inversion or horizontal reﬂec
tion. If the inversion is not contained in the group, it is isomorphic with the
proper group G
= P + iQ. This isomorphism can be seen by noting that
iQ cannot overlap with P since i is not contained in G. Thus we associate
the element iQ
k
of G
with the element Q
k
of G while P is common to
both groups.
7.6 DoubleGroup Representations
We have seen that the halfinteger representations of the rotation group R
3
are doublevalued. Each rotation is represented by two distinct matrices.
The halfinteger representations arise for example when the basis functions
are the spin functions χ
σ
(m
s
) which give the representation D
1/2
.
Since point groups are subgroups of the rotations group, we also expect
to have doublevalued representations when the basis functions transform
according to the halfinteger representations, which are often the case in
physical applications.
7.6. DOUBLEGROUP REPRESENTATIONS 119
Consider a point group g consisiting of proper rotations and the corre
sponding group G of matrices taken from the representation D
1/2
, say. G
contains twice as many elements as g does and the group G is called the
double group of g. The two groups G and g are homomorphic. To work
out the character table of the double group, we can proceed as usual but
the following rules simplify the procedure:
1. Let T(R) and T(
¯
R) be two matrices of G representing the same
rotation R. T(
¯
R) =
¯
ET(R) where
¯
E is minus a unit matrix repre
senting a rotation by 2π and
¯
E
2
= E represents a rotation by 4π.
Evidently,
¯
E commutes with all elements of the double group. The
characters of T(R) and T(
¯
R) must therefore have opposite sign:
χ(
¯
R) = −χ(R)
except when R is a rotation by π, in which case χ(
¯
R) = χ(R) = 0.
2. Neglecting for the moment the special case of rotation by π, it follows
that if a set of rotations R form a class of g, then the matrices T(R)
and T(
¯
R) form two classes of G, such that if T(R) belongs to one
than T(
¯
R) belongs to the other.
3. If R is rotation by π, then T(R) and T(
¯
R) may or may not belong
to the same class. They belong to the same class if and only if the
axis of rotation is bilateral, i.e. there is an element that reverses
the direction of the axis. This is because for bilateral axis, R(±θ)
belong to the same class so that in the double group R(π) and
R(−π) = R(3π), or T(R) and T(
¯
R), belong to the same class.
As an example, we consider the double group of D
3
. The classes of
D
3
are C
e
, C
c
, and C
d
. The classes of the double group can be worked
out by using the rules discussed above. ¯ e is in a class by itself because it
commutes with all elements. There is no bilateral axis so that the character
table is given by
120 CHAPTER 7. THE POINT GROUPS
C
e
¯
C
e
2C
c
2
¯
C
c
3C
d
3
¯
C
d
T
1
1 1 1 1 1 1
T
2
1 1 1 1 −1 −1
T
3
2 2 −1 −1 0 0
T
4
1 −1 1 −1 i −i
T
5
1 −1 1 −1 −i i
T
6
2 −2 −1 1 0 0
The ﬁrst three are the singlevalued representations of D
3
and the last three
are the new representations arising from the double group. The characters
χ(
¯
C
d
) can be obtained by observing that d
2
1
= ¯ e so that χ
2
(
¯
C
d
) = −1 and
therefore χ(
¯
C
d
) = ±i. The rest of the table can be ﬁlled in by using the
orthogonality between the characters.
¯
E commutes with all elements and
¯
E
2
= E so that (E,
¯
E) form a
subgroup. The double group, however, is not a direct product of this
subgroup and the original group because from the point of view of the
double group, the original group g is no longer a group.
As a simple example of physical applications of the double groups we
consider an atom with three selectrons. The product of three spin func
tions tranform according to the product representation D
1/2
⊗ D
1/2
⊗
D
1/2
= D
3/2
⊕ 2D
1/2
. Let us suppose that we break the rotational sym
metry by putting the atom in a potential with D
3
symmetry. The product
functions transform according to the double group of D
3
and not accord
ing to the group D
3
. The representations D
3/2
and D
1/2
may be reduced
by using the usual formula χ =
¸
γ
m
γ
χ
γ
where γ labels the irreducible
representations of the double group.
7.7 Space Group
The space group, which is the full symmetry of a crystal, contains
translationalsymmetry operators, rotationalsymmetry operators from the
point group, and the combinations of the two. The elements of the space
group are conventionally represented by ¦R[t¦ meaning
¦R[t¦r = Rr +t (7.1)
7.7. SPACE GROUP 121
The multiplication rule for the elements of the space group is
¦R[t¦¦R
[t
¦r = ¦R[t¦(R
r +t
)
= R(R
r +t
) +t
= ¦RR
[Rt
+t¦r (7.2)
If ¦R
[t
¦ = ¦R[t¦
−1
then ¦RR
[Rt
+t¦ = ¦E[0¦ so that
R
= R
−1
, t
= −R
−1
t (7.3)
Thus the inverse element is
¦R[t¦
−1
= ¦R
−1
[ −R
−1
t¦ (7.4)
We have already shown in example 1 (3.14) that the eigenfunctions
of a Hamiltonian with a lattice translational symmetry must obey Bloch
theorem:
ψ
k
(r +t) = e
ik·t
ψ
k
(r) (7.5)
Functions having this property are called Bloch functions. This means that
we can write
ψ
k
(r) = u
k
(r)e
ik·r
(7.6)
where u
k
is periodic, i.e.
u
k
(r +t) = u
k
(r)
We consider here only simple space groups (symmorphic). By this we
mean that all the point group elements are members of the space group or
in other words, the point group is a subgroup of the space group. This is
not true in general.
The form of the Bloch functions explicitly takes into account the lattice
translational symmetry. Now we consider the eﬀects of the point group
operations on these functions.
ˆ
T(R)ψ
k
(r) =
ˆ
T(R)u
k
(r)e
ik·r
= u
k
(R
−1
r)e
ik·R
−1
r
= u
k
(R
−1
r)e
iRk·r
(7.7)
122 CHAPTER 7. THE POINT GROUPS
The function u
k
(R
−1
r) is also periodic because
u
k
(R
−1
(r +t)) = u
k
(R
−1
r +t
) = u
k
(R
−1
r)
where t
= R
−1
t is just another translational vector.
If we apply all the pointgroup operations on a wave vector k we obtain
what is called the star of k. If k lies on a symmetry point, some of the
operations will take k into itself. These operations that take k into itself
clearly form a subgroup because R
1
R
2
k = R
3
k = k where R
1
R
2
= R
3
.
This subgroup is called the group of the wave vector. Under the operations
of this group, we have
ˆ
T(R)ψ
k
(r) = u
k
(R
−1
r)e
ik·r
(7.8)
Thus we see that ¦u
k
(R
−1
r) transform among themselves under the group
of k according to an irreducible representation of the group which is called
the small representation.
The eigenfunctions of the Hamiltonian of a crystal are then speciﬁed
by two quantities: the k vector (Bloch theorem) and the irreducible repre
sentation of the group of k.
Formally, we have (in atomic units ¯ h = m = e = 1)
¸
−
1
2
∇
2
+ V (r)
Ψ
k
= E
k
Ψ
k
(7.9)
Using the Bloch form, we get
−
1
2
∇
2
+ 2ik ∇−k
2
u
k
= E
k
u
k
(7.10)
which may be rewritten
¸
−
1
2
∇
2
+ V (r) +k p
u
k
= E
k
u
k
(7.11)
where E
= E −k
2
/2. This equation shows the signiﬁcance of the group
of k. We can think of the last term as a perturbation which breaks the
symmetry of the lattice potential V (r).
Consider ﬁrst the case of k = 0. Then u
k
will transform according to
an irreducible representation of the full point group. The degeneracy at this
7.7. SPACE GROUP 123
point is given by the dimension of the irreducible representation as usual.
Let us now consider the case k = 0 with the corresponding k vector group
which is a subgroup of the full point group. In this case, the degenerate
states at k = 0 still form an invariant subspace for the subgroup but they
are in general reducible. We expect that the degeneracy will be split. In
this way we can ﬁgure out the degeneracies at any k point. As a simple
example, consider a two dimensional crystal with a square primitive lattice.
The reciprocal space is also a square. Let us assume that the point group is
D
4
. If we take a point along the xaxis, then the k vector group is (e, a
x
)
(using the same notation as in the exercise). This is an Abelian group
and the irreducible representations are one dimensional. Thus the states at
this point cannot be degenerate. If we consider the point k = 0, we might
have states which are doubly degenerate because there is a two dimensional
irreducible representation of D
4
. Along the xaxis, these degenerate states
will in general be split.
124 CHAPTER 7. THE POINT GROUPS
Chapter 8
The Group SU
3
8.1 Inﬁnitesimal Operators and Lie Alge
bra
The group SU
3
is an extension of the group SU
2
. We derive the Lie
algebra from a physical point of view by considering three states which
may be associated with the proton, neutron and lambda particle states:
[p` = c
+
p
[0`, [n` = c
+
n
[0`, [Λ` = c
+
Λ
[0`
The bilinear products of annihilation and creation operators that conserve
particle number are:
c
+
p
c
p
, c
+
n
c
n
, c
+
Λ
c
Λ
,
c
+
p
c
n
, c
+
n
c
Λ
, c
+
Λ
c
p
,
c
+
p
c
Λ
, c
+
n
c
p
, c
+
Λ
c
n
These operators form a Lie algebra. We deﬁne a new set of inﬁnitesimal
operators as follows:
ˆ
B = c
+
p
c
p
+ c
+
n
c
n
+c
+
Λ
c
Λ
,
ˆ
T
z
=
1
2
(c
+
p
c
p
−c
+
n
c
n
),
ˆ
Y =
1
3
(c
+
p
c
p
+ c
+
n
c
n
−2c
+
Λ
c
Λ
),
125
126 CHAPTER 8. THE GROUP SU
3
ˆ
T
+
= c
+
p
c
n
,
ˆ
U
+
= c
+
n
c
Λ
,
ˆ
V
+
= c
+
Λ
c
p
,
ˆ
T
−
= c
+
n
c
p
,
ˆ
U
−
= c
+
Λ
c
n
,
ˆ
V
−
= c
+
p
c
Λ
We could have chosen
ˆ
U
z
=
1
2
(c
+
n
c
n
−c
+
Λ
c
Λ
) =
3
4
ˆ
Y −
1
2
ˆ
T
z
(8.1)
or
ˆ
V
z
=
1
2
(c
+
Λ
c
Λ
−c
+
p
c
p
) = −
3
4
ˆ
Y −
1
2
ˆ
T
z
(8.2)
instead of
ˆ
T
z
. The operator
ˆ
B commutes with all other operators and
it may be interpreted as the (baryon) number operator. The rest of the
operators form a Lie algebra by themselves and there are at most two
operators that mutually commute,
ˆ
T
z
and
ˆ
Y , so that the corresponding
Lie group is of rank 2. The other operators
ˆ
T
±
,
ˆ
U
±
, and
ˆ
V
±
are the ladder
operators, as the notation already suggests. Thus the desired Lie algebra
is
ˆ
T
z
,
ˆ
Y
= 0 (8.3)
ˆ
T
z
,
ˆ
J
±
= ±
1
2
(3δ
JT
−1)
ˆ
J
±
(8.4)
ˆ
Y ,
ˆ
J
±
= ±(δ
JU
−δ
JV
)
ˆ
J
±
(8.5)
where J = T, U, V . The Lie algebras for U and V can be obtained from
Eqs. ( 8.1), ( 8.2), ( 8.3), ( 8.4) and ( 8.5). In additions, we have
commutation relations among the ladder operators which are useful for
generating the multiplets:
ˆ
J
±
,
ˆ
J
∓
= 2δ
JJ
ˆ
J
z
(8.6)
ˆ
T
±
,
ˆ
U
±
= ±
ˆ
V
∓
and cyclic permutations (8.7)
We observe also that
[
ˆ
J
z
,
ˆ
J
±
] = ±
ˆ
J
±
, [
ˆ
J
+
,
ˆ
J
−
] = 2
ˆ
J
z
(8.8)
which is the Lie algebra of SU
2
.
8.2. SUBGROUPS 127
The representation of the inﬁnitesimal operators in the three dimen
sional space of [p`, [n`, and [Λ` is given by
T
z
=
¸
¸
1
2
0 0
0 −
1
2
0
0 0 0
¸
Y =
¸
¸
1
3
0 0
0
1
3
0
0 0 −
2
3
¸
T
+
=
¸
¸
0 1 0
0 0 0
0 0 0
¸
T
−
=
¸
¸
0 0 0
1 0 0
0 0 0
¸
U
+
=
¸
¸
0 0 0
0 0 1
0 0 0
¸
U
−
=
¸
¸
0 0 0
0 0 0
0 1 0
¸
V
+
=
¸
¸
0 0 0
0 0 0
1 0 0
¸
V
−
=
¸
¸
0 0 1
0 0 0
0 0 0
¸
These are in fact the generators of the group SU
3
.
8.2 Subgroups
Knowledge of the subgroups is useful for working out the multiplet struc
ture. Eq. ( 8.8) shows that there are three SU
2
subgroups.
ˆ
Y is a generator
of an Abelian group as indicated by the diagonal form of Y , just like
ˆ
T
z
is a generator of an Abelian group R
2
. The generators (
ˆ
Y ,
ˆ
T
z
,
ˆ
T
±
) form a
direct product group SU
2
⊗ U
1
since
ˆ
Y commutes with
ˆ
T
z
and
ˆ
T
±
. The
Abelian group U
1
can be shown to be isomorphic with the group R
2
. There
are other subgroups but they are not important for the purpose of working
out the multiplet structure.
8.3 Multiplet Structure
There are two mutually commuting operators,
ˆ
T
z
and
ˆ
Y , so that a multiplet
may be labelled by two numbers related to the eigenvalues of two Casimir
operators. Each state in a multiplet may be labelled by the eigenvalues
of
ˆ
T
z
and
ˆ
Y . However, due to the presence of SU
2
subgroups, the states
128 CHAPTER 8. THE GROUP SU
3
may be degenerate and additional labels are required to distinguish these
degenerate states. Thus the states in a given multiplet can be written as
ψ(M
T
, Y, α)
where M
T
and Y are the eigenvalues of
ˆ
T
z
and
ˆ
Y respectively and α is
a set of additional labels required to distinguish states with the same M
T
and Y . Since
ˆ
Y corresponds to an Abelian group, its representation is one
dimensional and uniquely labelled by Y . The label M
T
on the other hand
may arise from several diﬀerent values of T. To label each state uniquely,
we may then choose α = T.
From Eq. ( 8.8), the operators (
ˆ
T
z
,
ˆ
T
+
,
ˆ
T
−
) are the inﬁnitesimal op
erators of the group SU
2
and therefore M
T
takes the values M
T
=
0, ±1/2, ±1, . . .. Similarly for (
ˆ
U
z
,
ˆ
U
+
,
ˆ
U
−
) and (
ˆ
V
z
,
ˆ
V
+
,
ˆ
V
−
). It fol
lows from Eq. ( 8.1) or ( 8.2) that the eigenvalues of
ˆ
Y are Y =
0, ±1/3, ±2/3, ±1, . . .. Each state in a given multiplet may then be repre
sented by a point in (M
T
, Y )plane as illustrated in Fig. (8.1), keeping in
mind that each point may represent more than one state. Each state is not
only an eigenstate of
ˆ
T
z
and
ˆ
Y but it is also an eigenstate of
ˆ
U
z
and
ˆ
V
z
as
follows from Eqs. ( 8.1) and ( 8.2) but
ˆ
T
z
,
ˆ
U
z
, and
ˆ
V
z
are interrelated so
that one of them determines the other two.
The ladder operators may be conveniently represented by arrows as
shown in Fig. (8.2). For example, Eqs. ( 8.4) and ( 8.5) tell us that
ˆ
V
−
ψ(M
T
, Y, T) =
¸
T
c(T
) ψ(M
T
+
1
2
, Y + 1, T
)
To generate a multiplet, we use the same procedure as in the case of
the rotation group. We start from a state ψ with the highest value of Y
and for this Y the highest value of M
T
. Or alternatively, we could start
from the highest value of M
T
and for this M
T
the highest value of Y . The
ﬁrst alternative implies:
ˆ
T
+
ψ =
ˆ
U
+
ψ =
ˆ
V
−
ψ = 0
Although
ˆ
U
−
increases M
T
,
ˆ
U
−
ψ is not necessarily zero because
ˆ
U
−
lowers
the value of Y . The above equations imply that ψ is an eigenfunction of
ˆ
T
2
,
ˆ
U
2
, and
ˆ
V
2
which may be veriﬁed by operating
ˆ
T
2
=
ˆ
T
−
ˆ
T
+
+
ˆ
T
2
z
+
ˆ
T
z
8.3. MULTIPLET STRUCTURE 129
on ψ, and similarly for
ˆ
U
2
and
ˆ
V
2
. In other words it has deﬁnite T, U,
and V . Conventionally, ψ is labelled by two numbers λ = 2T and µ = 2U,
which implies from Eqs. ( 8.1) and ( 8.2) that λ + µ = 2V . Or
T =
1
2
λ, U =
1
2
µ, V =
1
2
(λ + µ) (8.9)
The two numbers λ and µ may also be used to label the multiplet: D
λµ
.
The state ψ has the following values of M
T
and Y :
M
T
=
1
2
λ, Y =
1
3
(λ + 2µ)
and let us denote the state by the point A = [
1
2
λ,
1
3
(λ + 2µ)] as shown in
the Fig. (8.3). The steps to generate the multiplet is as follows:
1. We apply
ˆ
T
−
repeatedly on ψ(A) until we reach point F at which
ˆ
T
−
ψ(F) = 0. The coordinate of point F is F = [−
1
2
λ,
1
3
(λ + 2µ)]
2. We apply
ˆ
U
−
repeatedly on ψ(A) until we reach point B at which
ˆ
U
−
ψ(B) = 0. To ﬁgure out the coordinate of point B, we observe
that
ˆ
U
−
increases M
T
by 1/2 and decreases Y by 1 and
ˆ
U
−
has been
applied µ times from A to B. Thus B = [
1
2
(λ+µ),
1
3
(λ−µ)]. There
cannot be states to the right of AB because from Eq. ( 8.6) we have
ˆ
T
+
ˆ
U
n
−
ψ =
ˆ
V
−
ˆ
U
n
−
ψ = 0 Thus the states along AB still have deﬁnite
but diﬀerent values of T and V and the same value of U.
3. The states along BC can be obtained by applying
ˆ
V
+
repeatedly on
ψ(B) or we can argue from symmetry that the states must be sym
metrical about the diagonal bisecting AB. The rest of the hexagon
can be obtained by symmetry about the diagonal bisecting BC. We
have shown that the states in a given multiplet must lie on or inside
a hexagon. In general, it can also be a triangle which is the case
when λ = 0 or µ = 0. States on the outermost hexagon are unique
and they have deﬁnite T, U, and V .
4. We can now generate states inside the hexagon and there may be
more than one independent state at each point. Consider the state
130 CHAPTER 8. THE GROUP SU
3
labelled by point G. There are three basic ways of reaching point G
from A which generate the states:
φ
1
=
ˆ
T
−
ˆ
U
−
ψ φ
2
=
ˆ
U
−
ˆ
T
−
ψ φ
3
=
ˆ
V
+
ψ,
The states φ
1
and φ
2
are linearly independent since
ˆ
T
−
and
ˆ
U
−
do
not commute but φ
3
is not independent of the other two because
[
ˆ
T
−
,
ˆ
U
−
]ψ = −
ˆ
V
+
ψ →φ
1
−φ
2
= −φ
3
The state φ
1
has a deﬁnite T =
1
2
(λ + 1) and it is possible to form
a linear combination of φ
2
and φ
3
so that it has a deﬁnite T:
ˆ
T
+
(φ
2
+ cφ
3
) =
ˆ
T
+
ˆ
U
−
ˆ
T
−
ψ + c
ˆ
T
+
ˆ
V
+
ψ
=
ˆ
U
−
ˆ
T
+
ˆ
T
−
ψ + c(
ˆ
V
+
ˆ
T
+
−
ˆ
U
−
)ψ
=
ˆ
U
−
(
ˆ
T
−
ˆ
T
+
+ 2
ˆ
T
z
)ψ −c
ˆ
U
−
ψ
= (λ −c)
ˆ
U
−
ψ
using the fact that
ˆ
T
+
ψ = 0. By choosing c = λ the linear combi
nation
˜
φ
2
= φ
2
+ λφ
3
has a deﬁnite T =
1
2
(λ − 1) since
ˆ
T
+
˜
φ
2
= 0
and
ˆ
T
z
˜
φ
2
=
1
2
(λ −1)
˜
φ
2
. It is of course possible to choose the states
to have deﬁnite U or V . Applying
ˆ
T
−
on ψ(G) generate two sets of
states parallel to AF. Continuing along the same fashion, it is not
diﬃcult to show that the next line parallel to AF consists of three
sets of states with T =
1
2
(λ − 2),
1
2
λ,
1
2
(λ + 2). For example, the
state
ˆ
U
2
−
ψ has T =
1
2
(λ+2) while the state
ˆ
U
−
˜
φ
2
has T =
1
2
λ since
ˆ
T
+
ˆ
U
−
˜
φ
2
=
ˆ
U
−
ˆ
T
+
˜
φ
2
= 0. The third state may be formed by a similar
linear combination as the one described above.
The multiplet structure consists of hexagons of decreasing size with
the degeneracy increased by one as we move inside. Eventually we
may arrive at a triangle instead of a hexagon at which stage the
degeneracy remains constant as we move further. The reason is that
with a triangle we cannot generate an new state such as
˜
φ
2
. This
implies that when λ = 0 or µ = 0 each point corresponds to one state
because we start with a triangle and no extra states are generated
as we move inside. This is illustrated by the representation D
30
in
8.4. EXAMPLE 131
Fig. (8.4) where the point O can be reached from A by
ˆ
V
+
ˆ
T
−
ψ(A)
or
ˆ
T
−
ˆ
V
+
ψ(A) but [
ˆ
V
+
,
ˆ
T
−
] = 0 so that the two paths are equivalent.
The multiplet structure also allows us to work out the dimension
of the irreducible representation. Thus the outermost hexagon
contributes 3(λ + µ) states and the next one inside contributes
2 3(λ +µ −2) states and so on taking care of the special case of
a triangle. The general formula is given by
d(λ, µ) =
1
2
(λ + 1)(µ + 1)(λ + µ + 2) (8.10)
which is symmteric with respect to interchange of λ and µ. Given
a representation D
λµ
it is straightforward to map the states in the
(M
T
, Y )−plane. Starting from the point A = [
1
2
λ,
1
3
(λ + 2µ)], we
form the line AF by moving λ steps along
ˆ
T
−
and µ steps along
ˆ
U
−
.
The rest of the hexagon and the inner hexagons can be formed from
the rules discussed above.
8.4 Example
As an example, let us construct explicitly the multiplet D
21
as shown in
Fig. (8.5). We start with the state
ψ(A) = ψ(1,
4
3
, 1)
which has deﬁnite values of T = 1, U =
1
2
, and V =
3
2
. Applying
ˆ
T
−
twice
generates
ψ(0,
4
3
, 1), ψ(−1,
4
3
, 1)
Applying
ˆ
U
−
on ψ(A) gives
ψ(B) = ψ(
3
2
,
1
3
,
3
2
)
The T =
3
2
multiplet is generated by repeated applications of
ˆ
T
−
on ψ(B).
The T =
1
2
multiplet is generated from the linear combination
ψ(
1
2
,
1
3
,
1
2
) = c(
ˆ
U
−
ˆ
T
−
+ 2
ˆ
V
+
)ψ(A)
132 CHAPTER 8. THE GROUP SU
3
The normalisation constant is obtained by requiring
[c[
2
(
ˆ
U
−
ˆ
T
−
+ 2
ˆ
V
+
)ψ(A), (
ˆ
U
−
ˆ
T
−
+ 2
ˆ
V
+
)ψ(A)
= 1
We have
(
ˆ
U
−
ˆ
T
−
ψ(A),
ˆ
U
−
ˆ
T
−
ψ(A)) = (ψ(A),
ˆ
T
+
ˆ
U
+
ˆ
U
−
ˆ
T
−
ψ(A))
= (ψ(A),
ˆ
T
+
(
ˆ
U
−
ˆ
U
+
+ 2
ˆ
U
z
)
ˆ
T
−
ψ(A))
= 2(ψ(A),
ˆ
T
+
(
ˆ
T
−
ˆ
U
z
+
1
2
ˆ
T
−
ψ(A))
= 2(ψ(A),
ˆ
T
+
ˆ
T
−
ψ(A))
= 4
(
ˆ
U
−
ˆ
T
−
ψ(A),
ˆ
V
+
ψ(A)) = (ψ(A),
ˆ
T
+
ˆ
U
+
ˆ
V
+
ψ(A))
= (ψ(A),
ˆ
T
+
(
ˆ
V
+
ˆ
U
+
+
ˆ
T
−
ψ(A))
= 2
(
ˆ
V
+
ψ(A),
ˆ
V
+
ψ(A)) = 3
Thus the normalisation constant is given by c = 1/
√
24. The states at C
are given by
ψ(0, −
2
3
, 1) = c
1
ˆ
T
−
ˆ
V
+
ψ(B), ψ(0, −
2
3
, 0) = c
2
ˆ
U
−
ψ(
1
2
,
1
3
,
1
2
)
The normalisation constants can be calculated in the same way as above.
8.5 Product Representations
In physical applications such as analysing the possible ﬁnal states in a colli
sion between two particles, we have to decompose product representations:
D
λ
1
µ
1
⊗D
λ
2
µ
2
=
¸
λµ
⊕ m
λµ
D
λµ
(8.11)
The general formula for the coeﬃcients m
λµ
is complicated but it possible
to obtain them directly by the method of reduction. We illustrate this
8.5. PRODUCT REPRESENTATIONS 133
method by considering the product D
11
⊗D
11
. First we list all possible T
and Y multiplets contained in D
11
and label them by (T, Y ):
D
11
= (
1
2
, 1) ⊕(1, 0) ⊕(0, 0) ⊕(
1
2
, −1)
We form products of (T, Y )
(T
1
, Y
1
) ⊗(T
2
, Y
2
) =
T
1
+T
2
¸
T=T
1
−T
2

⊕ (T, Y
1
+ Y
2
)
To determine T we have used the usual vector coupling theorem and to
determine Y we note that it represents an Abelian group so that the repre
sentation is of the form exp(icY ) and therefore the product of two represen
tations Y
1
and Y
2
is exp[ic(Y
1
+Y
2
)]. Forming all products and substract
ing all possible representations contained in the product representation we
have:
Y = 2 1 0 −1 −2
D
11
⊗D
11
T = 1, 0 2(
3
2
), 4(
1
2
) 2, 5(1), 4(0) 2(
3
2
), 4(
1
2
) 1, 0
D
22
T = 1
3
2
,
1
2
2, 1, 0
3
2
,
1
2
1
T = 0
3
2
, 3(
1
2
) 4(1), 3(0)
3
2
, 3(
1
2
) 0
D
03
T = 0
1
2
1
3
2
T =
3
2
, 2(
1
2
) 3(1), 3(0) 3(
1
2
) 0
D
30
T =
3
2
1
1
2
0
T = 2(
1
2
) 2(1), 3(0) 2(
1
2
)
2D
11
T = 2(
1
2
) 2(1), 2(0) 2(
1
2
)
T = 0
D
00
T = 0
The ﬁrst row is a list of all possible Y that arise from the product rep
resentation. The second row is a list of all possible T and the number
preceeding each bracket is the number of times the T multiplets arise for
a given Y . The product D
11
⊗ D
11
must contain a representation with
T = 1, Y = 2 which is the representation D
22
. We substract the T values
of D
22
listed in the third row from the product representation which yields
the fourth row. Following the same argument, there must be a represen
tation with T = 0, Y = 2 which is the representation D
03
. We substract
134 CHAPTER 8. THE GROUP SU
3
the T values of D
03
from the remaining T’s and continue along the same
fashion. Thus we conclude ﬁnally
D
11
⊗D
11
= D
22
⊕D
03
⊕D
30
⊕2D
11
⊕D
00
8.6 SU
3
Multiplets and Hadrons
Hadrons are strongly interacting particles with decay time < 10
−8
sec, ex
cept proton and neutron. By strong we mean that the energy involved in a
reaction is of the order of MeV or greater. Apart from, mass, spin, charge,
and parity, hadrons may be assigned the following intrinsic quantities:
1. Isospin T: Among an isospin multiplet, hadrons have (almost) the
same masses. For example, proton and neutrons which correspond to
T = 1/2 or the three πparticles which correspond to T = 1. Strong
interactions are found to be invariant under isospin tranformation,
i.e. states in a given isospin multiplets transform among themselves
under isospin transformation such as transformation among (p, n) or
(π
+
, π
−
, π
0
) (c.f. angular momentum states).
2. Hypercharge Y: The quantity Y in the group SU
3
is related to the
charge and the zcomponent of isospin by
Y = 2(Q−M
T
)
This relation is a physical interpretation which does not follow from
group theory. It is known experimentally that the hypercharge is
conserved in strong interaction. Hadrons can be divided into baryons
with halfinteger spins and mesons with integer spins, the former are
heavier than the latter.
3. Baryon number B: It is conserved in all reactions, including weak
ones. The assignment of baryon number if as follows: B = +1 for
baryons, B = −1 for antibaryons, and B = 0 for mesons.
With the above assignments of intrinsic quantities, hadrons can be
beautifully classiﬁed according to SU
3
multiplets. Fig. (8.6) (ﬁgure 11.8 of
Elliot and Dawber) shows some of these classiﬁcations. The corresponding
8.7. CASIMIR OPERATORS 135
particles are listed in Table (8.1) (from table 11.1 of Elliot and Dawber).
The representation D
11
and D
30
consist of baryons with spin 1/2 and 3/2
respectively while the other D
11
representation corresponds to the mesons.
Particles with short lifetime, which arise as resonances in collisions, may
often be regarded as excited states of the more stable ones.
It has been suggested that baryons are composed of three more funda
mental spin 1/2 particles known as quarks which are labelled by u, d, and
s. These three quarks are supposed to correspond to the representation
D
10
so that
u →T =
1
2
, M
T
= +
1
2
, Y = +
1
3
, Q = +
2
3
d →T =
1
2
, M
T
= −
1
2
, Y = +
1
3
, Q = −
1
3
s →T = 0, M
T
= 0, Y = −
2
3
, Q = −
1
3
Since the largest spin is S = 3/2, it is then proposed that the baryons can
be constructed from the triple product
D
10
⊗D
10
⊗D
10
= D
30
⊕2D
11
⊕D
00
A more complicated model involving nine quarks has also been proposed
which introduces, apart from spin, additional intrinsic properties known as
colours.
8.7 Casimir Operators
We have not made any explicit use of the Casimir operators in working out
the structure of the multiplets. To construct the second order Casimir op
erator, it is possible to use the general formula derived before and a similar
formula for higher order Casimir operators can also be derived although
they become more complicated. For the group SU
3
there is a simple way
of working out the Casimir operators which is based on the observation
that the inﬁnitesimal operators may be written in terms of the matrices
deﬁned by:
(A
j
i
)
kl
= δ
ik
δ
jl
−
1
3
δ
ij
δ
kl
(8.12)
136 CHAPTER 8. THE GROUP SU
3
We have
T
z
= (A
1
1
−A
2
2
)/2, Y = −A
3
3
, T
+
= A
2
1
, T
−
= A
1
2
U
+
= A
3
2
, U
−
= A
2
3
, V
+
= A
1
3
, V
−
= A
3
1
The commutation relations for the A
j
i
are
[A
j
i
, A
l
k
] = δ
jk
A
l
i
−δ
il
A
j
k
It is then easy to show that the operators
C
2
=
¸
ij
A
j
i
A
i
j
, C
3
=
¸
ijk
A
k
j
A
j
i
A
i
k
commute with all of the inﬁnitesimal operators. Instead of C
3
it is more
convenient to deﬁne
C
3
= C
3
+
3
2
C
2
so that
ˆ
C
2
ψ =
¸
2
3
(λ
2
+ µ
2
+λµ) + 2(λ + µ)
ψ
ˆ
C
3
ψ =
1
9
(λ −µ)(2λ + µ + 3)(2µ + λ + 3) ψ
which may be easily veriﬁed by using the Lie algebra.
Contents
1 Abstract Group Theory 1.1 Group . . . . . . . . . . . . . . 1.2 Abelian Group . . . . . . . . . . 1.3 Subgroup . . . . . . . . . . . . . 1.4 Examples . . . . . . . . . . . . . 1.5 Multiplication Table . . . . . . . 1.6 Cyclic Groups . . . . . . . . . . 1.7 Order of an Element . . . . . . . 1.8 Properties of Finite Groups . . . 1.9 Cosets . . . . . . . . . . . . . . 1.10 Class . . . . . . . . . . . . . . . 1.11 Isomorphism and Homomorphism 1.12 Invariant Subgroup . . . . . . . . 1.13 Direct Product Group . . . . . . 1 1 2 2 2 3 4 5 5 5 6 7 8 8 11 11 11 12 12 14 17 19 20 24 24 26
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . .
2 Theory of Group Representations 2.1 Deﬁnitions . . . . . . . . . . . . . . . . 2.2 Equivalent Representations . . . . . . . 2.3 Reducible and Irreducible Representation 2.4 Equivalent Unitary Representation . . . . 2.5 Schur’s Lemma . . . . . . . . . . . . . . 2.6 Orthogonality Theorem . . . . . . . . . 2.7 The Characters of a Representation . . . 2.8 The Regular Representation . . . . . . . 2.9 The Character Table . . . . . . . . . . . 2.10 Properties of the Character Table . . . . 2.11 Class Multiplication . . . . . . . . . . . i
ii
CONTENTS 2.12 Direct Product Groups . . . . . . . . . . . . . . . . . . . . 27 2.13 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3 Group Theory in Quantum Mechanics 3.1 Linear Vector Space . . . . . . . . . . . . . . 3.2 Symmetry Transformations . . . . . . . . . . 3.3 Invariant Subspace and Its Generation . . . . . 3.4 Basis Functions for a Representation . . . . . 3.5 Projection Operators . . . . . . . . . . . . . . 3.6 Orthogonality of Basis Functions . . . . . . . 3.7 Examples . . . . . . . . . . . . . . . . . . . . 3.8 Relationship to Quantum Mechanics . . . . . 3.8.1 The Group of the Schr¨dinger Equation o 3.8.2 Degeneracy and Invariant Subspace . . ˆ 3.8.3 Partial Diagonalisation of H . . . . . . 3.9 Irreducible Sets of Operators . . . . . . . . . . 3.10 Direct Product Representations of a Group . . 3.11 ClebschGordan Coeﬃcients . . . . . . . . . . 3.12 The WignerEckart Theorem . . . . . . . . . . 3.13 Applications . . . . . . . . . . . . . . . . . . 3.14 Examples . . . . . . . . . . . . . . . . . . . . 4 Lie 4.1 4.2 4.3 4.4 4.5 4.6 4.7 Groups Deﬁnitions . . . . . . . . . . . . Inﬁnitesimal Operators . . . . . . Lie Algebra . . . . . . . . . . . . Casimir Operators . . . . . . . . Ladder Operators and Multiplets Summation over group elements . Rotation Group R3 . . . . . . . . 4.7.1 Inﬁnitesimal operators . . 4.7.2 Commutation Relations . 4.7.3 Casimir Operator . . . . . 4.8 Irreducible Representations of R3 4.9 Characters of R3 . . . . . . . . . 4.10 The VectorCoupling Theorem . . 4.11 ClebschGordan Coeﬃcients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
33 33 35 37 37 39 41 41 47 48 49 50 50 51 53 54 55 57 63 63 64 66 68 69 71 72 73 74 75 76 79 81 82
. . . . . . . . . . . . . . . . . . . .CONTENTS iii 4. . . . . .2. . . . . . . . Enumeration of Point Groups Improper Point Groups . . . . 91 92 93 94 95 99 100 102 105 . . . . . . . .4 7 The 7. . . . 113 . . . . . . . . . . . . . . .12 Spherical Harmonics . . . . . . . . . . .4 7.2 7. . . . . . . . . . . . . . . . . .7 8 The 8. . . . . .6 7. . . . . . . . . . . . . . . . . . . . . . . . . . Tensor Operators . . . . . . . . . . . . . . . . . Product Representations . . . . . . . . . . .3 Projection Operator Method 5. . . . . . . . . . 5. . . . . . . .2. Schoenﬂies Notation . . . . .5 8. .2 H0 + H1 . . . .3 8. . . . .1 6. . . . . . .2 8. . . . . . . . . . . . . . . . . . . . . . 113 .4 Term Energies . . . . . . . 120 . . . . . .1 H0 . . . . . . . . . . . . . . . . . . . . . . . . 109 . . . . . . . . . . .5 7. . . . . . . . . . DoubleGroup Representations Space Group . .4 8. . . .1 7. R3 . . . . . . . . . . . . . . . . . . . .2 6. . . .1 Conﬁguration and Term . . . . . . . . . SU3 Multiplets and Hadrons . . . . . . . Commuting Operations . . . . . . . . . . . . Casimir Operators . . . .2. . . . ˆ ˆ ˆ ˆ 5. . . . . . . . . . . . . . . . . . . 5. . . . . . . . . . . . . . . . . . . 114 . . . . . . Multiplet Structure . . . . . . . . . . . . 6 The 6. . . . 125 125 127 127 131 132 134 135 Group SU3 Inﬁnitesimal Operators and Lie Algebra Subgroups . . . . . . . 118 . . . . . . . . . . .7 Group SU2 : Isospin The Group SU2 . . . . . . . . . . . Point Groups Crystal Symmetry . Example . . . . . . . . . . . . . . . . .3 6. . . . . . Relationship between SU2 and SU2 in Nuclei .6 8. .3 7.3 H = H0 + H1 + H2 . 118 . 111 113 . 87 5 Atomic Physics ˆ 5. . . . . . . . . ˆ ˆ 5. . . . . . . 105 . .2. . . . . . . . . . . . . . . . . . 107 . . . . . . . . . . . . . . . 115 . . . . . . . . .2 Ladder Operator Method . . . . . . . . . . . . . . . .1 8. . . . . . . . . . . . 5. . . .
c in the group. Proof: a−1 (ae) = (a−1 a)e = ee = e = a−1 a and using the left cancellation law we have ae = a. b. c = ab is also a member of the group. Unit element on the right: ae = a = ea. The following properties follow from the above deﬁnition: 1. Closure: if a and b are members of the group. Left cancellation: If ax = ay then x = y for all a in the group. 2. 2.1 Group A group is a set of elements that have the following properties: 1. 4.Chapter 1 Abstract Group Theory 1. 1 . Associativity: (ab)c = a(bc) for all a. Inverse: to every element a there is a corresponding inverse element a−1 in the group such that a−1 a = e. Proof: ax = ay → a−1 (ax) = a−1 (ay) → (a−1 a)x = (a−1 a)y → ex = ey → x = y. Unit element: there is an element e such that ea = a for every element a in the group. 3.
Evidently. We note the importance of associativity in the above proofs. the unit element forms a subgroup by itself. . A group with n identical elements evidently has at least n identical subgroups. then one of them is redundant so that we usually assume that all elements are diﬀerent.2 CHAPTER 1. Proof: xa = ya → (xa)a−1 = (ya)a−1 → x(aa−1 ) = y(aa−1 ) → xe = ye → x = y.4 Examples 1. This group is Abelian and inﬁnite. Proof: a−1 (aa−1 ) = (a−1 a)a−1 = ea−1 = a−1 = a−1 e. However. the group is called Abelian. Using the left cancellation law. Right cancellation: If xa = ya then x = y for all a in the group. 1. 4. The following identity is often useful: (ab)−1 = b−1 a−1 which follows from (ab)−1 (ab) = 1 → (ab)−1 a = b−1 → (ab)−1 = b−1 a−1 . The unit element e = 0 and the inverse of an element a is a−1 = −a. Inverse element on the right: aa−1 = e = a−1 a. aa−1 = e. when two elements are identical. 1. Integers under addition.2 Abelian Group If a group has a further property that ab = ba for all a.3 Subgroup A subgroup is a set of elements within a group which forms a group by itself. ABSTRACT GROUP THEORY 3. 1. We note that the deﬁnition of a group does not require all elements to be distinct from one another. b in the group.
MULTIPLICATION TABLE 3 2.5 Multiplication Table The group multiplication table is a matrix Mab = ab. where a and b are elements of the group and the matrix element corresponding to row a and column b is given by the group multiplication ab.1.5. The elements of the group are e = γ = A B A B A B A C C C C B α= µ= A B B C C A C A β= ν= A B C A A B B A C B C C A B C B The operation αβ means: ﬁrst do permutation β and then permutation α on the previous result. The unit element is the unit matrix and the inverse of U is U † by deﬁnition.e. matrix multiplication etc. the permutations form a group. ACB. A set of all n × n unitary matrices U (U † = U −1 ) under matrix multiplication. † † (U1 U2 )† = U2 U1 −1 −1 = U2 U1 = (U1 U2 )−1 3. or simply that one operation is performed on the result of the preceeding operation like in the example 3. 1. The set of permutation operations that take ABC into ABC. BCA. We show below that every permutation has an inverse permutation and two successive permutations correspond to a single permutation i. . CBA. We illustrate this deﬁnition by constructing the multiplication table of the permutation group in the example 3 above. CAB. We see from the above examples that ”multiplication” can mean addition. We have to show that a product of two unitary matrices is unitary . BAC.
a3 . for example. γ) form a subgroup. αγ = γα. Rearrangement Theorem: In every row or column. the unit element e forms a subgroup by itself. Obviously. The multiplication table has the following properties: 1.4 CHAPTER 1. an = e . The elements (e. 2. Thus element b cannot appear more than once and since the size of the row or column is the same as the number of elements in the group. α. ABSTRACT GROUP THEORY e α β γ µ ν e e α β γ µ ν α α β e ν γ µ β β e α µ ν γ γ γ µ ν e α β µ µ ν γ β e α ν ν γ µ α β e Thus. . the inverse of α is β and vice versa. a2 . β). . The group multiplication table mathematically characterises the group. so do (e. µ). The multiplication table is symmetric across the diagonal when the group is Abelian because the elements commute with one another. each element must appear once and once only and therefore each row (column) is different from any other row (column). But this implies that ac = ad → c = d which is a contradiction.g. (e. The group is clearly not Abelian because e. . . The elements of the group must be a. This means that ac = b and ad = b where c = d. it follows that each element must appear once and once only. 1.6 Cyclic Groups A group of n elements is said to be cyclic if it can be generated from one element. Proof: Suppose element b appears twice in the row a. ν) and (e. All groups with the same multiplication table are mathematically identical with respect to their group properties.
It is also evident that aS or Sa have no element in common with S for otherwise. Form the products a2 . If a2 = e we continue forming a3 . Every noncyclic group has at least a cyclic subgroup. ORDER OF AN ELEMENT 5 n is called the order of the cyclic group. Examples of cyclic groups are the subgroups of the permutation group in the example 3. The subgroup (e.7. a2 . . . . an are distinct from each other and an = e then n is called the order of element a.7 Order of an Element Let a = e be an element of a group. and ν are of order 2. Thus every group must have at least one cyclic subgroup. α2 = β. .1. a cannot be a unit element and therefore a coset can never be a group because it has no unit element. Every element a has a ﬁnite order n such that an = e. a2 must be either e or a diﬀerent element from a because if a2 = a → a = e. By a similar argument. 1. a should be included in S. . These elements form a cyclic group. In the example 3 above. β) is the same as (α. α. 3. . α and β are of order 3 and γ.9 Cosets If S is a subgroup of G and a is an element of G not in S. Morevoer. Rearrangement Theorem: Multiplying all elements in a group by an arbitrary element reproduces the group. α3 = e). . 1. a3 . then the sets aS and Sa are called the left and right cosets of S respectively. 1. µ. This has been proven above in the properties of the multiplication table. if b is . 2. 1. a3 must be either e or a diﬀerent element from a and a2 . A cyclic group is evidently Abelian but an Abelian group is not necessarily cyclic. If a.8 Properties of Finite Groups We summarise below the properties of ﬁnite groups. It is shown below that every noncyclic group has at least a cyclic subgroup..
If G is Abelian.6 CHAPTER 1. . then a and c are also conjugate to each other. . Proof: Let x and y be in S such that ax = by. but the factorisation is not unique.1) (Since every element has an inverse. If a and b are conjugate to each other and b and c are conjugate to each other. a. . A class C is a set of elements which are conjugate to each other. . Then h = integer g As a consequence. It is clear that a group may be broken up into classes G = C1 + C2 + . then bS has no element in common with either S or aS. and an element cannot belong to more than one class. if h is prime then G must be cyclic and it has no subgroup other than the trivial subgroup of e. It depends on the choice of S..10 Class Two elements a and b of a group are conjugate to one another if there is an element g in the group such that a = gbg −1 (1. . Thus we have the following theorem: Let h be the number of elements of a group G and g be the number of elements of a subgroup S of G. ABSTRACT GROUP THEORY an element of G which is neither in S nor in aS. b. This is because −1 −1 −1 −1 a = g1 bg1 and b = g2 cg2 → a = g1 g2 cg2 g1 = g1 g2 c(g1 g2 )−1 . etc. The number of factors must be ﬁnite and the factorisation must exhaust the group. . The transformation gbg −1 is often called a similarity transformation. each element also forms a class by itself. . it also follows that b = g ag −1 where g = g −1 ). a ﬁnite group can be factorised as G = S + aS + bS + . 1. . Consequently. The unit element evidently forms a class by itself. but xy −1 is an element in S and therefore b is in aS which is a contradiction. We have axy −1 = b.
such that if ab = c then a i b j = c k i. a 2 . fi → gi . 1. α and β both correspond to cyclic permutations whereas γ. Similarly we can show that (γ.11 Isomorphism and Homomorphism Two groups are said to be isomorphic if they have the same multiplication table. and ν correspond to permutations with one member ﬁxed. This isomorphism can be seen by labelling the corners of the triangle with A. Homomorphism is similar to isomorphism except that the relationship is manytoone. c k lies in c . . Elements belonging to the same class usually have the same characteristic. c 2 . by reordering the elements if necessary. We use the multiplication table that we have constructed previously: βαβ −1 γαγ −1 µαµ−1 ναν −1 = = = = βαα = ββ = α γαγ = γµ = β µαµ = µν = β ναν = νγ = β Thus α and β belong to the same class. µ. . c → c = (c 1 . . Two groups G and G are said to be homomorphic if each element of G can be associated with some elements of G : a → a = (a 1 .1. b → b = (b 1 .). B. . .11. µ.e. it is obviously not necessary to consider similarity transformations with either the unit element or the element itself.). If F is isomorphic to G then there is a onetoone correspondence between the elements of F and G. . ISOMORPHISM AND HOMOMORPHISM 7 As an example. In the above example.). such that if fi fj = fk then gi gj = gk . . b 2 . As an example. C. ν) also form a class. j. we work out one of the classes of the permutation group. . for every i. In working out the classes. . Two groups can only be isomorphic if they have the same number of elements. . the permutation group in the example 3 is isomorphic to the symmetry operations that take an equilateral triangle into itself.
This group is called a factor group.12 Invariant Subgroup We prove the following theorem: Theorem: If S is a subgroup of G and a is an element of G then S = aSa−1 is also a subgroup and it is isomorphic to S. . Proof: s i s j = asi a−1 asj a−1 = asi sj a−1 = ask a−1 = s k . Let asi a−1 = asj a−1 and i = j. . .13 Direct Product Group Let F be a group with elements fi . . and G be a group with elements gi . It is an example of homomorphism where the invariant subgroup S is associated with the elements of S and the coset aS is associated with the elements of aS. such that fi gj = gj fi for all i and j. which is a contradiction. If F and G have no common element. . i = 1. aS = Sa If we break a group into its cosets. Proof: We ﬁrst show that the elements of S are distinct. hG . si = sj .e. ABSTRACT GROUP THEORY 1. i = 1. i If S is the same as S for every a in the group then the subgroup S is called an invariant subgroup. . Thus S and S are isomorphic with the same unit element e = aea−1 = e and the inverse of si is si −1 = (asi a−1 )−1 = as−1 a−1 = i (s−1 ) . . i. . This implies that the left coset of S is the same as its right coset. We now show that if si sj = sk then s i s j = s k . hF . with an invariant subgroup S. .8 CHAPTER 1. Thus the elements of S are distinct. We show that F ⊗ G is a group. then these cosets form a group with the unit element equal to S. G = S + aS + bS. Then by the cancellation law. 1. Thus aSbS = abSS = abS = cS if ab = c. The direct product group F ⊗ G is deﬁned to be the set of all distinct elements fi gj . We assume that the elements of S are distinct. then the order of the direct product group will be hF hG . . apart from the identity. . .
−1 −1 3. 4. 2. Inverse of fi gj is (fi gj )−1 = gj fi−1 = fi−1 gj . Closure: (fi gj )(fk gl ) = (fi fk )(gj gl ) = fr gs . According to the deﬁnition of a class.1. s rs −1 −1 = fr gs fi gj fr gs −1 −1 = (fr fi fr )(gs gj gs ) . Unit element is eF eG . we have Cij = a−1 aij ars .13. Associativity: obvious. Since fr is in F and gs is in G then by deﬁnition fr gs is in F ⊗ G. DIRECT PRODUCT GROUP 9 1. Proof: We label the elements of the product group aij = fi gj . We have the following theorem: Theorem: The classes of the direct product group are given by the direct products of the classes of the individual groups. f or all r.
10 CHAPTER 1. ABSTRACT GROUP THEORY .
. C. Some or all of the matrices may be the same.1) The identity element E is represented by a unit matrix and each matrix must have an inverse. .} may be represented by a set of square matrices {T (A).1 Deﬁnitions A group {A.2 Equivalent Representations T = S −1 T S (2. The order of the matrices is called the dimension of the representation. . T (C).2) If T is a representation then with an arbitrary but nonsingular S is also a representation because T (A)T (B) = S −1 T (A)SS −1 T (B)S 11 .}. the representation is called faithful (isomorphic).Chapter 2 Theory of Group Representations 2.. . B. T (B). These matrices are said to form a representation for the group if they satisfy the same group multiplication rule: AB = C → T (A)T (B) = T (C) (2. 2.. If the matrices are diﬀerent.
Proof: Construct a Hermitian matrix H= G T (G)T † (G) . . THEORY OF GROUP REPRESENTATIONS = S −1 T (A)T (B)S = S −1 T (AB)S = T (AB) T and T are equivalent and there are an inﬁnite number of equivalent representations. .12 CHAPTER 2.4 Equivalent Unitary Representation There are an inﬁnite number of equivalent representations but it is possible to ﬁnd one which is unitary. Theorem 1: Given an arbitrary representation T . The representation is irreducible if it cannot be reduced into the above form. .3 Reducible and Irreducible Representation A representation T is reducible if there exists a nonsingular matrix S such that the equivalent representation T = S −1 T S has the form T = T1 T2 T3 . there is always an equivalent unitary representation. . T2 . are square matrices and the rest of the elements are zero. T3 . 2. for all elements in the group. T1 . 2.
we may deﬁne a diagonal matrix D−1/2 with diagonal elements −1/2 equal to Dii so that 1 = D−1/2 G T (G)T (G)D−1/2 † Using the above result.2. Since D is diagonal. the matrix T (G) = D−1/2 T (G)D1/2 can be shown to be unitary. we can always diagonalise a Hermitian matrix by a unitary transformation D = U −1 HU = U −1 T (G)T † (G)U G = G U −1 T (G)U U −1 T † (G)U T (G)T (G) G † = U is made up of the eigenvectors of H and D is diagonal by deﬁnition and its diagonal elements are real and positive. it can never be equal to zero because otherwise T = 0. T (G)T (G) = D−1/2 T (G)D1/2 [D−1/2 G † T (G )T (G )D−1/2 ] † × D 1/2 T (G)D G † −1/2 † † = D−1/2 = D = D −1/2 G −1/2 G T (G)T (G )T (G )T (G)D−1/2 T (G)T (G )[T (G)T (G )]† D−1/2 T (GG )T (GG )D−1/2 † .4. Djj = G k T T G k † jk (G)T kj (G) ∗ jk (G)T jk (G) jk (G) 2 = = G k T ≥0 In fact. EQUIVALENT UNITARY REPRESENTATION 13 According to matrix algebra.
T ] = M T − T M = 0 for all elements in the group. if the theorem is . given an arbitrary representation T . it is always possible to construct a unitary representation by forming T = D−1/2 U −1 T U D1/2 From now on we assume that we have a unitary representation. Theorem 2 (Schur’s lemma): Given that [M. then (a) if T is irreducible. Adding and substracting the ﬁrst and the last equations.3) 2.14 CHAPTER 2. Thus. (2. T ] = [H2 . M = cI where c is a constant and I is a unit matrix. Thus.5 Schur’s Lemma Schur’s lemma is used in proving many of the theorems in group theory. T ] = 0. TM (T M )† M †T † T (M † T † )T TM† = = = = = MT (M T )† T †M † T (T † M † )T M †T since T † = T −1 (unitary). Proof: We ﬁrst show that it is suﬃcient to prove the theorem for a Hermitian M . we get T (M + M † ) = (M + M † )T and T i(M − M † ) = i(M − M † ) T Let H1 = M + M † and H2 = i(M − M † ) so that [H1 . (b) if M = cI then T is reducible. THEORY OF GROUP REPRESENTATIONS = D−1/2 G T (G )T (G )D−1/2 † = 1 The second last step has been obtained by using the Rearrangement Theorem. H1 and H2 are Hermitian and M = 1/2(H1 −iH2 ).
it must be true for M . or equivalently T . i. M will have an . is irreducible. D = U −1 M U . SCHUR’S LEMMA 15 true for a Hermitian matrix. Let us summarise the conclusions: If M = cI then T must be reducible. Hence if T is irreducible then M must be equal to cI for otherwise T would be reducible. element. then (a) if lα = lβ . In the latter case. Consider the ij. Theorem 3: Given T α M = M T β .4) Let us order the diagonal elements of D such that Dii ≤ Djj for i < j which can always be done by rearranging the columns of U .5. where T α and T β are irreducible representations of dimension lα and lβ respectively and M is a rectangular lα × lβ matrix. T is reducible.4) that T must have the form T = X 0 0 Y where X is an n × n matrix. M = 0 or (b) if lα = lβ either M = 0 or M  = 0.e. T ij Djj = Dii T T ij (Djj − Dii ) = 0 ij (2.2. ( 2. We can now assume M to be Hermitian and perform a unitary transformation so that M becomes diagonal. Suppose D11 = Dnn and D11 < Dii where n ≥ 1 and i > n. i = j. If M = cI then obviously it commutes with T whether it is reducible or irreducible. and deﬁne an equivalent representation T = U −1 T U . Then TD = = = = = U −1 T U U −1 M U U −1 T M U U −1 M T U U −1 M U U −1 T U DT We have to show that the diagonal elements of D are all identical if T . Then it follows from Eq.
lα > lβ . each term must vanish i. M M †  = M 2 = clα If c = 0. Let us look at the diagonal elements of M M † : † Mik Mki = 0 k ∗ Mik Mik = 0 k Mik 2 = 0 k Since this is positive deﬁnite. Since M is contained in N . then M M † = 0. then M  = 0 i. . M M †  = cI. But N  = 0 because one of its columns is zero.e. Consider now the case lα = lβ .16 CHAPTER 2. N N † = cI → N N †  = clα → N N †  = clα → c = 0 Hence N N † = 0 which implies N = 0. Proof: T α (G)M = M T β (G) (T α (G)M )† = (M T β (G))† M † T α† (G) = T β (G)M † M † T α−1 (G) = T β −1 † (G)M † −1 M M † T α−1 (G) = M T β (G)M † M M † T α (G−1 ) = M T β (G−1 )M † M M † T α (G−1 ) = T α (G−1 )M M † Since T α is irreducible. It follows that T α and T β are equivalent since M −1 (T α (G)M ) = M −1 (M T β (G)) = T β (G) If c = 0. THEORY OF GROUP REPRESENTATIONS inverse so that T α and T β are equivalent. Then N N † = M M † = cI. We enlarge M into a lα × lα square matrix N with the additional elements equal to zero. M −1 exists. then M = 0.e. it follows from Schur’s lemma (Theorem 2) that M M † = cI. M = 0. Consider the case lα = lβ .
6 Orthogonality Theorem We are now in a position to prove the Orthogonality Theorem which is central in representation theory.2.5) α and β label the irreducible representations. Consider T α (A)M = G T α (A)T α (G)XT β (G−1 ) T α (A)T α (G)XT β (G−1 )T β (A−1 )T β (A) G = = G T α (AG)XT β (G−1 A−1 )T β (A) T α (AG)XT β ((AG)−1 )T β (A) G = = G T α (AG)XT β [T α (G)XT β G −1 (AG)T β (A) = −1 (G)]T β (A) = M T β (A) We have again made use of the Rearrangement Theorem in the second last step.6. According to Theorem 3. lα and lβ are the dimensions of these irreducible representations. ORTHOGONALITY THEOREM 17 2. and h is the number of elements in the group. Proof: Deﬁne a matrix M= G T α (G)XT β (G−1 ) where X is an arbitrary matrix and α = β. We want to show that this matrix satisﬁes the postulates of Theorem 3. M = 0 since we are considering the case of α = β so that Mij = 0 = G kl β α Tik (G)Xkl Tlj (G−1 ) . Theorem 4: Orthogonality Theorem β α Tij (G)∗ Tkl (G) = (h/lα )δαβ δik δjl G (2.
Then β α Tim (G)Tnj (G−1 ) = G G α Tim (G)T β nj (G) α Tim (G)T β nj (G) G α Tim (G)T β jn (G) G ∗ † −1 = = = 0 When α = β. j) 2 cannot exceed h. according to Schur’s lemma (Theorem 2) Mij = c δij = G α α Tik (G)Xkl Tlj (G−1 ) kl α α Tim (G)Tnj (G−1 ) = G Putting i = j and summing over i yields c lα = G α α Tim (G)Tni (G−1 ) i α Tnm (E) = G = hδnm Thus we have α α Tim (G)Tnj (G−1 ) = (h/lα )δnm δij G Since T is unitary. α lα ≤ h. THEORY OF GROUP REPRESENTATIONS Since X is arbitrary.e. i.18 CHAPTER 2. ( 2. i. in the h dimensional space of the group elements. i and j. .5) has the form of a dot product with Tij (G) as the G’th component of a vector labelled by α. It will be proven later that in fact 2 lα = h α The result is very useful in working out the irreducible representations of a group. we may set Xkl = δkm δln . This means that the number of distinct labels (α. α α Tim (G)Tjn (G)∗ = (h/lα )δnm δij G α Eqn.
2. The eigenvalues are one of them but they are too cumbersome to calculate. T r T (A) = T r [T (C −1 ) T (B) T (C)] . so is T = S −1 T S. Given T and T . how can we then tell that they are equivalent ? One way will be to ﬁgure out if there is a matrix S.6) We show that the trace or character of a matrix is invariant under a similarity transformation: Tr T = i jk −1 Sij Tjk Ski = jk ( i −1 Ski Sij )Tjk = jk δkj Tjk = Tr T The characters of a representation α is a set of h numbers {χα (G)}. there is an element C such that A = C −1 BC. We recall that if T is a representation.7.7 The Characters of a Representation We have seen that a representation is not unique. What we are looking for are properties of a matrix which are invariant under a similarity transformation. Proof: Since A and B belong to the same class. Consequently. A simpler quantity is the trace of a matrix or the sum of the diagonal elements: χα (G) = T r T α (G) = i α Tii (G) (2. Theorem 5: If elements A and B of a group belong to the same class. But this is a complicated procedure. then the characters of their representations are the same. its characters are unique. THE CHARACTERS OF A REPRESENTATION 19 2. such that T = S −1 T S. It will be shown later that if the representation is irreducible.
7) has the form of a weighted dot product with weight Nk in a space with dimension equal to the number of classes of the group. 2.5) (Orthogonality Theorem) and summing over i and l we have χα (G)∗ χβ (G) = (h/lα )δαβ χ (Ck )∗ χβ (Ck )Nk = hδαβ k G α il δil δil Eqn. χα (Ck ) is the k’th component of vector α. It follows that the number of irreducible representations must be less than or equal to the number of classes.20 CHAPTER 2. ( 2.8 The Regular Representation We prove the following theorem: Theorem 7: 2 lα = h (2.7) Ck labels a class with Nk elements. ( 2.8) α Proof: . THEORY OF GROUP REPRESENTATIONS = T r [T (C) T (C −1 ) T (B)] = T r [T (E) T (B)] = T r T (B) Theorem 6: First Orthogonality Relation χα (G)∗ χβ (G) = G k χα (Ck )∗ χβ (Ck )Nk = hδαβ (2. Proof: Setting i = j and k = l in Eqn.
8. . We can now prove Theorem 7 by writing the character of the regular representation as a sum over characters of the irreducible representations: χ(ai ) = α mα χα (ai ) . . To show that it is really a representation. The regular representation is formed in the following way. e α−1 β −1 γ −1 We form the regular representation of an element a by writing 1 wherever the element a occurs in the table and zero otherwise. i = 1. then the regular representation is given by Tjk (ai ) = 1 if a−1 ak = ai j = 0 otherwise It is clear from the deﬁnition that the representation is faithful. we have to prove that Tjk (ai )Tkl (am ) = Tjl (ap ) k if and only if ai am = ap . 2.. aj ai = al a−1 implies a−1 al = ai am = ap . j m The character is given by h χ(ai ) = j=1 Tjj (ai ) = h if ai = e = 0 otherwise This is clear by inspection of the multiplication table. h.. THE REGULAR REPRESENTATION 21 A proof of this theorem is provided by considering the so called regular representation. We write the multiplication table as follows: e e α e e e β γ . Let the group elements be ai . However.2. . . By deﬁnition. m Therefore k Tjk (ai )Tkl (am ) = 1 if and only if ak = aj ai = al a−1 and m zero otherwise. Tjk (ai ) = 0 only when k is such that ak = aj ai and similarly Tkl (am ) = 0 only when k is such that ak = al a−1 .
χ(Ck ) = G∈Ck χ(G) α cαij Tij (G) G∈Ck αij = = 1 h α cαij Tij (G−1 GG1 ) 1 G1 G∈Ck αij .22 CHAPTER 2.9) Proof: α We recall that the matrix representation Tij can be regarded as a set of orα thogonal vectors in the h. Each irreducible representation occurs in the regular representation a number of times equal to the dimension of the irreducible representation. h = α mα lα so that h= α 2 lα This result together with the orthogonality theorem show that there are α exactly h orthogonal vectors Tij . Tij themselves can be regarded as a basis in the space of the group elements since the number of α. χ(e) = h. Tij (G) α α is the component of the vector Tij along the ”Gaxis”. i.dimensional space of the group elements. j is precisely h. Theorem 8: Second Orthogonality Relation χα (Ck )∗ χα (Cl ) = (h/Nk )δkl α (2. and χ(ai = e) = 0 it follows that mα = lα . Any vector in this space can therefore be expanded α in Tij : α χ= cαij Tij αij We sum the components of χ along Gaxes which belong to a given class Ck . THEORY OF GROUP REPRESENTATIONS From the orthogonality relation we have mα = α 1 h χα∗ (ai )χ(ai ) i Since χ (e) = lα . On the other hand.
The third step uses the fact that G−1 Ck G1 = 1 Ck for every G1 in the group. χ(Cl ) = β aβ χβ (Cl ) χ(Ck )χβ (Ck )Nk χβ (Cl ) β ∗ = (1/h) k 0 = k χ(Ck ) (Nk /h) β χβ (Ck )χβ (Cl ) − δkl ∗ This is true for an arbitrary χ(Ck ) so that the quantity in the square bracket must vanish which proves the theorem.2. ( 2. the ﬁfth step uses the orthogonality theorem. It follows that the number of classes must be less than or equal to the number of irreducible representations.9) has the form of a dot product in a space with dimension equal to the number of irreducible representations. and the last step uses the fact that the characters of elements belonging to the same class are identical. Eqn. THE REGULAR REPRESENTATION = = G∈Ck αij kl 23 1 h α α α cαij Tik (G−1 )Tkl (G)Tlj (G1 ) 1 G1 G∈Ck αij kl α cαij δij δkl Tkl (G)/lα = G∈Ck αi cαii χα (G)/lα aα χα (Ck ) α = where aα = (Nk /lα ) i cαii . Thus the set of vectors {χ(Ck )} are linearly independent because the group elements can be divided uniquely and completely into distinct classes. ∗ Multiplying χ(Ck ) by χβ (Ck )Nk and summing over the classes k yields χ(Ck )χβ (Ck )Nk = k α ∗ aα k α χα (Ck )χβ (Ck )Nk ∗ = aα hδαβ = haβ Consequently.8. On the other hand by Theorem 6 the number of irreducible .
THEORY OF GROUP REPRESENTATIONS representations must be less than or equal to the number of classes.9 The Character Table The character table is a square matrix with dimension equal to the number of classes or irreducible representations: Qαk = χα (Ck ) It has the form T T2 T3 1 C1 = CE 1 l2 l3 N2 C2 1 . .. . 2..24 CHAPTER 2.. 1.. Theorem 6 and 8 imply that the irreducible representations are uniquely characterised by their characters and consequently two distinct irreducible representations cannot have the same set of characters. 2. 2. . . The ﬁrst column contains the dimension of each representation and the sum of its square is the total number of elements in the group 2 (Theorem 7): α lα = h. The row is labelled by the irreducible representations and the column by the classes.. N3 C3 1 .. Thus we conclude the following corollary: The number of irreducible representations is equal to the number of classes.10 Properties of the Character Table The properties listed below are useful for constructing the character table. The number of irreducible representations is equal to the number of classes.....
This group is a called a factor group with S as the unit element and it is homomorphic with the group itself. Ci Cj = k mijk Ck . then we can form a group S. the character of an element aS in the factor group may be assigned to the classes which are contained in aS. the ﬁrst row is the identity or unit representation consisting of 1. Thus.2. bS. 5. Each row or column can be regarded as a vector.10. The homomorphism between the factor group and the group itself means that if T (aS) is a matrix representation of the element aS of the factor group then the same matrix may be used to represent all elements in the coset aS. 4. . b is not in S and aS etc. the characters of the classes in aS is just the same as the character of T (aS). . The above ﬁve rules are often suﬃcient to construct the character table but the following rule. PROPERTIES OF THE CHARACTER TABLE 25 3. where a is not in S. The characters of the α representation are related by Ni χα (Ci )Nj χα (Cj ) = lα k mijk Nk χα (Ck ) (2. aS. then due to homomorphism. As a convention. . which is described in the next section. S must consist wholly of classes. 6. can be of help. We recall that if a group has an invariant subgroup S. An invariant subgroup means that x−1 Sx = S for every element x in the group and by Theorem 9 discussed in the next section.10) where mijk is the coeﬃcients of class multiplication. If the character table of a factor group is known. In this sense: (a) The rows are orthogonal with weighting factor Nk (Theorem 6) and normalised to h. (b) The columns are orthogonal and normalised to h/Nk (Theorem 8). .
If the representation T is irreducible. Theorem 9: A set of elements C obeying x−1 Cx = C for all x in the group must be composed wholly of classes. an element cannot belong to more than one class. it follows that Ci Cj must consist wholly of classes. Thus the elements on the right hand side must be identical to the elements in C. . This is evident from the fact that all elements in a class are conjugate to one another and if two elements a and b are diﬀerent then x−1 ax = x−1 bx. Proof: Suppose R is a set of elements in C which do not form a class. Moreover. Therefore it must be possible to write Ci Cj = k mijk Ck (2. Thus R must be a class. We now consider a product of two classes and here we do count all resulting elements even when some are the same. THEORY OF GROUP REPRESENTATIONS 2. If we let S(Ci ) be the sum of matrices of all elements in the class Ci and T (x) be a matrix representation of the element x then we have T −1 (x)S(Ci )T (x) = S(Ci ) or S(Ci )T (x) = T (x)S(Ci ). Ci Cj = x−1 Ci xx−1 Cj x = x−1 Ci Cj x By Theorem 9. From the deﬁnition of a class we have for every element x in the group. Let us consider the class multiplication in terms of matrix representations. But x−1 Rx must be equal to R itself since (C − R) form classes and by deﬁnition they will never be conjugate to elements in R.11 Class Multiplication x−1 Cx = C Let C be a collection of classes.11) where mijk are integers telling how often the class Ck appears in the product Ci Cj .26 CHAPTER 2. then it follows from Schur’s Lemma that S(Ci ) = ci I.
M may be written in the form M = M1 ⊗ M2 where M1 and M2 have the same β α dimensions as T1 and T2 respectively.12. then the irreducible representations and the character table of the product group can be worked out easily from those of the two individual groups with the help of the following theorem. for all T1 and T2 these are all possible irreducible representations of the product group. multiplying them and using the deﬁnition of direct product yield the required result. a2 . This means that ai ak = am and bj bl = bn or T α (ai )T α (ak ) = T α (am ) and T β (bj )T β (bl ) = T β (bn ). b2 . and b1 . Suppose cij ckl = cmn which implies that ai bj ak bl = am bn or ai ak bj bl = am bn since the elements of G1 and G2 commute. . ( 2.10) follows. . The commutivity of M with T α⊗β . .2. Theorem 10: β α If G = G1 ⊗ G2 then T α⊗β = T1 ⊗ T2 is an irreducible representation of β α G. Proof: We ﬁrst show that T is a representation of G. Let M be a matrix which commutes with every T α⊗β .12 Direct Product Groups If a group is a direct product of two groups. Moreover. . . DIRECT PRODUCT GROUPS Thus ci cj = k 27 mijk ck Taking the trace of S(Ci ) and assuming we are in the irreducible representation α we get T r S(Ci ) = T r ci I = ci lα On the other hand. 2. T r S(Ci ) = Ni χα (Ci ) Therefore ci = Ni χα (Ci )/lα and Eq. . be the elements of G1 and G2 respectively and let cij = ai bj be the elements of the direct product group G. Let a1 . Writing the last two matrix equations in component form.
Then l2 = 1 and l3 = 2 are the only possible solution and the character table must look like as follows: C1 1 1 2 2C2 1 a c 3C3 1 b d T1 T2 T3 To determine a. β) = C1 + C2 is an invariant subgroup and the corresponding factor group has the multiplication table S γS S S γS γS γS S . µ. c. ν) and therefore by rule 1 there are three irreducible representations. 2 To show the last part of the theorem. By Schur’s lemma. and d we need to know one of them and we can ﬁnd the others by orthogonality (rule 4). THEORY OF GROUP REPRESENTATIONS β α implies the commutivity of M1 with T1 and M2 with T2 . b. β). There are three classes C1 = e. The subgroup S = (e. C2 = (α. M1 = k1 I1 and M2 = k2 I2 and hence M = k1 k2 I and no nonconstant matrix M exists so that by Schur’s lemma T α⊗β is irreducible. We have 2 lαβ = αβ 2 l1α β 2 l2β = h1 h2 α But h1 h2 is the order of the direct product group and therefore there are α no more possible irreducible representations other than T α⊗β for all T1 and β T2 .28 CHAPTER 2. 2. We construct the character table of the permutation group of three objects which we have considered previously. Rule 2 tells us that 2 2 2 l1 + l2 + l3 = 6. Let lαβ = l1 l2 be the dimension of the irreducible representation T α⊗β . α. and C3 = (γ.13 Examples 1. we recall that α l1α = h1 2 α β and β l2β = h2 . There is always an identity representation so we may take l1 = 1.
13.2. Therefore the number of irreducible representations is simply equal to the number of elements in the group. The irreducible representations of the factor group must be (1. This is a group of two elements and there are evidently two classes. rule 2 implies that all irreducible representations have dimension one. Moreover. the ﬁrst class consists of the unit element S and the second class consists of the element γS. 1) and (1. Abelian groups: Every element in an Abelian group forms a class by itself. by rule 2. The other two characters c and d can be easily obtained from the orthogonality between columns. ν) = C3 . 1 · 1 + 1 · a + 2 · c = 0 → c = −1 1·1+1·b+2·d = 0→d=0 Finally we have C1 1 1 2 2C2 3C3 1 1 1 −1 −1 0 T T2 T3 1 2. −1). Then from rule 5 we may assign a = 1 and b = −1 (a = 1 and b = 1 gives the identity representation). Note that we have taken a and b and not c and d because the irreducible representations of the factor group are one dimensional and T 2 is also a one dimensional representation. The number of irreducible representations is two and both must have dimension one. EXAMPLES 29 where γS = µS = νS = (γ. µ. . The consequences are (a) Unitarity of representations implies that the numbers representing an Abelian group have modulus one.
and Ah = 1 since e k k is always represented by 1. a2 . (c) The character table serves also as a representation table. b. . Cyclic groups: These are Abelian groups which can be generated by a single element: G = (a. Thus. ah = e). Thus we have e 1 1 1 1 a 1 1 −1 −1 b c 1 1 −1 −1 1 −1 −1 1 T1 T2 T3 T4 3. Ak must be one of the h roots of . If a is represented by a number Ak then a2 is represented by A2 etc. If a = −1 then either b = 1 and c = −1 or b = −1 and c = 1. . If a is represented by 1 then b = c = ±1 because bc = a. and c are represented either by ±1. . . THEORY OF GROUP REPRESENTATIONS (b) If the number X represents an element x then X ∗ represents x−1 . As an example we consider a noncyclic group of fourth order with the following multiplication table e a b c e a e a a e b c c b b c b c c b e a a e a2 = b2 = c2 = e so a.30 CHAPTER 2.
. . Ak = exp(i2πk/h) and the character table is given by T1 T2 T3 . . 1 a 1 A1 A2 . Ah−1 a2 1 A2 1 A2 2 . .....e. . . ah−1 1 h−1 A1 h−1 A2 .2. ... EXAMPLES unity.13. A2 h−1 . .... .. . . .. i... Th e 1 1 1 . h−1 Ah−1 31 .
THEORY OF GROUP REPRESENTATIONS .32 CHAPTER 2.
. A set of vectors {v1 . r2 . .1 Linear Vector Space A set {r1 . .} is said to form a linear vector space L if ri + rj is another member in the set for every i. vp } is said to be linearly independent if none of them can be expressed as a linear combination of the others. If a set of coeﬃcients {ck } can be found such that p ck vk = 0 k=1 then the vectors are said to be linearly dependent. The largest number of vectors in L which form a linearly independent set is called the dimension of L. We denote the dimension of L by s and the basis vectors by {e1 .Chapter 3 Group Theory in Quantum Mechanics 3.} are called vectors and there are an inﬁnite number of them. j and cri is also another member in the set where c is a complex constant. The quantities {r1 . . . . . . . This is the same as the smallest number of vectors needed to describe every vector in L and these vectors are said to form a basis. . . e2 . v2 . . . es } so that any vector v may be written as s v= i=1 vi ei 33 . r2 . .
cv2 ) = c(v1 . GROUP THEORY IN QUANTUM MECHANICS In order to determine the coeﬃcients vi from a given vector v we introduce the concept of scalar or dot product between two vectors v1 and v2 which we denote by (v1 . v1 ) ≥ 0. It is only zero when v1 = 0. 2. The particular deﬁnition of the scalar product is arbitrary but it must satisfy the following general conditions: 1. v2 ) → (cv1 . v2 ) = v1 v2 cosθ where v1 and v2 are the lengths of the vectors and θ is the angle between the two vectors. ˆ We deﬁne an operator T by ˆ Tv = v where v is an arbitrary vector or function which is carried into another vector or function v . v2 ). v3 ) + (v2 . The operator is called linear if ˆ ˆ ˆ T (v1 + v2 ) = T v1 + T v2 ˆ ˆ T cv = cT v . v2 ) is a complex number 2. Both v and v are in L. (v1 . The dot product is deﬁned as (v1 . cv1 )∗ = c∗ (v1 . (v1 . v2 ) 4. (v1 + v2 . Vectors in 3dimensional space.34 CHAPTER 3. v3 ) = (v1 . v2 ) = (v2 . v3 ) 5. v1 )∗ 3. ψj ) = ∗ d3 r w(r)ψi (r)ψj (r) where w is a weight function which is usually real and positive deﬁnite so that the last condition is satisﬁed. (v1 . Examples: 1. v2 ) = (v2 . The dot product is deﬁned as (ψi . Function spaces. (v1 . We can also deﬁne orthonormal basis functions {φi } such that any function can be expanded in this basis: ψ = ∞ ci φi The linear i=1 space spanned by this basis is called the Hilbert space.
y. In . by this we mean (φi . Let φ1 . y. We illustrate the method for three vectors.2.3. Let ψ1 = φ1 / (φ1 . ψ2 ) and construct ψ3 = φ3 − ψ2 (ψ2 . z) is a real linear coordinate transformation that preserves the length (real unitary transformation) r = Gr → ri = j Gij rj . z such that the function calculated with the new variables ri by the above substitution of coordinates is the same as the function calculated with the old variables: f (x . This can always be done by the GrammSchmidt orthogonalisation procedure. y.2 Symmetry Transformations A symmetry transformation G with respect to a given function f (x. φj ) = δij . φ1 ) and construct ψ2 = ˜ φ2 − ψ1 (ψ1 . y . ˆ T ej = i ei Tij ˆ The matrix Tij is said to form a representation for the operator T in the linear vector space L. The generalisation to an arbitrary number of vectors is straightforward. φ2 . φ3 ) which is orthogonal to both ψ1 and ψ2 and which is then normalised to give ˜ ˜ ˜ ψ3 = ψ3 / ψ3 . ψ3 ) In this way we have constructed an orthonormal set of functions and we can see that for a general case we have ˜ ψn = φn − n−1 ψi (ψi . We normalise ψ2 to get ψ2 = ˜ ˜ ˜ ˜ ψ2 / (ψ2 . y. z) → f (Gr) = f (r). z) = f (x. φ2 ) which is orthogonal to ψ1 . it is only necessary to study the eﬀects of the operator on the basis vectors. It is usually convenient to choose an orthonormal basis. z ) = f (x. φn ) i=1 3. ri = x. SYMMETRY TRANSFORMATIONS 35 We will consider only linear operators. φ3 ) − ψ1 (ψ1 . φ3 be three linearly independent functions which are not necessarily ˜ orthogonal nor normalised. Since any vector in L can be expanded in the basis vectors.
GROUP THEORY IN QUANTUM MECHANICS other words. Thus we mean that ˆ T (G)f (Ar) = f (AG−1 r) = f (G−1 Ar) We also have G−1 instead of G in our deﬁnition. and not upon the argument of f . then the ˆ associated symmetry operators {T (G)} form a group. From the deﬁnition. This is a natural choice ˆ ˆ because two successive operations T (G1 ) and T (G2 ) on a function f (r) correspond to the same order of coordinate transformation: ˆ ˆ ˆ T (G2 )T (G1 )f (r) = T (G2 )f (G−1 r) = f (G−1 G−1 r) = f ([G2 G1 ]−1 r) 1 1 2 Theorem 1: If {G} is a set of all symmetry transformations of a function. This also implies that f (Gr) = f (r) = f (G−1 r) because f (r ) = f (r) = f (G−1 r ) and r is just a dummy variable. ˆ A symmetry operator T (G) associated with a symmetry transformation G is deﬁned by ˆ T (G)f (r) = f (G−1 r) It is important to note that the operator acts upon the coordinates r. .36 CHAPTER 3. then so is G−1 as discussed above. the function looks the same in the old and the new coordinate system. we have ˆ ˆ T (G1 )T (G2 )f (r) = = = = ˆ T (G1 )f (G−1 r) 2 −1 −1 f (G2 G1 r) f ([G1 G2 ]−1 r) ˆ T (G1 G2 )f (r) ˆ ˆ ˆ Hence T (G1 )T (G2 ) = T (G1 G2 ) and the symmetry operators form a group. Proof: The set {G} evidently form a group because if G is a symmetry transformation. A succession of two symmetry transformations is itself a symmetry transformation and the identity transformation is obviously a symmetry transformation.
The set of functions {T (G)f } form an invariant subspace under {T (G)} ˆ ˆ ˆ because T (G )[T (G)f ] = T (G G)f .4 Basis Functions for a Representation Let {φi } be a set of orthonormal functions that span an invariant subspace ˆ VS . T (G)φj ) = δij = ( k φk Tki (G).3. then T (G) is unitary because ˆ ˆ (T (G)φi . It may happen that we can construct a function f in VS such that the ˆ invariant subspace {T (G)f } has a smaller dimension than that of VS . We note that the number of ˆ T functions that span the subspace is not necessarily equal to the number of G. l φl Tlj (G)) = kl ∗ Tki (G)Tlj (G)(φk . The largest number of linearly independent functions in VS is called the dimension of the subspace.1) If the basis functions are orthonormal. Then T (G)φi (r) = φi (G−1 r) is in VS by deﬁnition and it can therefore be expanded as a linear combination of {φi }: ˆ T (G)φi = j φj Tji (G) (3. In this case the subspace VS is said to be reducible. φl ) † Tik (G)Tkj (G) k = . which must lie in the subspace since ˆ(G G) is just another member of {T (G)}. An invariant subspace can be generated from an arbitrary function f and ˆ a set of symmetry operators {T (G)} by applying each of the operators on ˆ ˆ f . 3.3.3 Invariant Subspace and Its Generation A set of functions or vectors {φi } is said to span an invariant subspace VS ˆ ˆ under a given set of symmetry operators {T (G)} if T (G)φi is also in VS for all G and i. INVARIANT SUBSPACE AND ITS GENERATION 37 3. When such a function cannot be found than VS is said to be irreducible. Some of the functions may be the same or can be expressed as linear combinations of the others.
in which case the subspace is reducible. then the representation is also irreducible. If the representation were reducible. then it would be possible to start from a function that does not generate the entire subspace. which may be reducible. they are also applicable here. Proof: ˆ T (G)φi = j φj Tji (G) −1 ψk Skj Tji (G) j k ˆ T (G) j −1 ψj Sji = . which is an equivalent representation. GROUP THEORY IN QUANTUM MECHANICS Theorem 2: ˆ The matrices {T (G)} form a representation for {T (G)} Proof: ˆ ˆ T (G1 )T (G2 )φi = j ˆ T (G1 )φj Tji (G2 ) φk Tkj (G1 )Tji (G2 ) jk ˆ T (G1 G2 )φi = φk Tki (G1 G2 ) = k k φk Tkj (G1 )Tji (G2 ) Thus T (G1 )T (G2 ) = T (G1 G2 ). Since the theorems were developed for arbitrary matrices. The functions {φi } are said to form a basis for a representation. Evidently. if the subspace ˆ is irreducible under {T (G)}. S is assumed to be unitary so that (ψi . Theorem 3: A change of basis vectors ψj = i φi Sij corresponds to a similarity transformation of the original representation.38 CHAPTER 3. ψj ) = δij . This simply follows from the fact that one always generates the entire subspace starting from any function in the subspace. We now see the relevance of matrix representation theory studied in the previous chapter.
ˆ β ψ transforms according to the irreducible representation β. we get β∗ α ˆ Tkl (G)T (G)ψi = G j α = (h/lβ )δil δαβ ψk α ψj (h/lβ )δjk δil δαβ We deﬁne ˆβ Pkl ≡ (lβ /h) G ˆ T β kl (G)T (G) (3.e.e.5 Projection Operators ˆ It is possible to construct a projection operator P β such that given an arbiβ ˆ trary function ψ.3. i.2) ∗ so that β α ˆα α ˆβ α Pkl ψi = δil δαβ ψk → Pki ψi = ψk . It is clear that the reverse is also true. we construct these projection operators. P ψ is a component of ψ in the irreducible subspace β. P very useful tool. a similarity transformation of the representation with a unitary matrix S corresponds to a unitary transformation of the basis vectors. PROJECTION OPERATORS Multiplying both sides by Sim and summing over i yields ˆ T (G)ψm = k 39 ψk T km (G) −1 where T km (G) = ji Skj Tji (G)Sim (G) or T = S −1 T S which is a similarity transformation. That is α ˆ T (G)ψi = j β∗ Multiplying by Tkl (G) and summing over G yields β∗ α ˆ Tkl (G)T (G)ψi = G G j β∗ α α ψj Tji (G)Tkl (G) α α ψj Tji (G) Using the Orthogonality Theorem on the right side. 3. In the following. This is a i.5. α Suppose {ψi } span an irreducible subspace transforming according to ˆ the irreducible representation α of {T (G)}.
into irreducible subspaces. We α note also that if φα and ψi belong to the ith row of the irreducible reprei α sentation α. Then there is a unitary matrix S such that T = S −1 T S = α ⊕mα T α has the block form and each block cannot be reduced any further. α labels the irreducible representation and mα tells us how many times the irreducible representation α appears in the reduction. The new basis vectors according to Theorem 3 are given by ψj = i φi Sij and they may be labelled according to the irreducible subspaces to which they belong. in which the symmetry operators are deﬁned. so does any linear combination aφα + bψi .5) . we can also deﬁne ˆ Pβ ≡ i ˆβ Pii ˆ χβ∗ (G)T (G) G ≡ (lβ /h) so that α ˆ α P β ψi = δβα ψi (3.i = (3. These functions form a basis for the irreducible representation α. GROUP THEORY IN QUANTUM MECHANICS Starting from a given function that belongs to the ith row of a given irreducible representation α. This can always be done by means of similarity transformations in each subspace which transforms according to the same irreducible representation α. Any vector in VS can therefore be expanded as follows: v = α.40 CHAPTER 3.3) The projection operators allow us to decompose the space.t. i Suppose that a given representation T is reducible.i αt ψi cαt i α vi α. We note that when mα > 1 the irreducible representations T α are assumed to be the same. the projection operator can be used to generate the other ”partner functions” corresponding to the other rows. Moreover. Any function ψ can therefore be written as follows ψ= α i α ψi (3.4) α where ψi belongs to the ith row of the irreducible representation α.
d2 .3. and i labels the basis functions in the subspace α.6. We have also deﬁned α αt α α ˆ vi = t ψi cαt . 3. we have β β ˆ ˆ (φα . ORTHOGONALITY OF BASIS FUNCTIONS 41 where t labels the diﬀerent possible subspaces transforming according to the same irreducible representation α. 2. φβ ) k l G kl β (φα . t. 3 in Figure 3. Then α ˆ T (G)φα = φα Tki (G) i k k β ˆ T (G)ψj = l β ψlβ Tlj (G) Recalling that T is or may always be chosen to be unitary. If they transform according to the same irreducible representation. The elements are e = unit element. d3 = rotations by π about axes lebelled 1.7 Examples As an example. ψk ) k k = (1/h) = (1/lα )δαβ δij 3. c1 . and d1 . which is necessary when mα > 1. they are still orthogonal if they belong to diﬀerent rows. ψlβ ) k β α∗ Tki (G)Tlj (G)(φα . However. T (G)ψj ) i i = kl β α∗ Tki (G)Tlj (G)(φα .1. we consider the group D3 which is the group of permutations of three objects. c2 = anticlockwise rotations about the z−axis by 2π/3 and 4π/3 respectively. ψj ) = (T (G)φα . Proof: β Let φα and ψj belong to the irreducible representations T α and T β respeci tively.6 Orthogonality of Basis Functions Theorem 4: Two vectors belonging to two inequivalent irreducible representations must be orthogonal. . T (G)vi is not necessarily equal to j vj Tji (G) i unless there is only one t.
d3 ). χ(Cc ) = 0. the 3 × 3 representation is reducible into a 2 × 2 and a 1 × 1 . ey ) whereas ez forms an invariant subspace by itself. GROUP THEORY IN QUANTUM MECHANICS The classes are Ce = e. and Cd = (d1 . d2 .42 CHAPTER 3. c2 ). The choice is arbitrary. ez as shown in the ﬁgure. ex ex √ 1 − 2 ex + √3 ey 2 − 1 ex − 23 ey 2 −e√ x 1 ex + √3 ey 2 2 1 ex − 23 ey 2 ey ey √ 3 − 2 ex − 1 ey 2 √ 3 ex − 1 ey 2 2 ey √ 3 e − 1 ey 2 x 2 √ − 23 ex − 1 ey 2 ez ez ez ez −ez −ez −ez ˆ T (e) ˆ T (c1 ) ˆ T (c2 ) ˆ T (d1 ) ˆ T (d2 ) ˆ T (d3 ) We see that applying the group operations on ex or ey generate an invariant subspace (ex . The ﬁrst obvious choice is the three Cartesian unit vectors ex . T (G)φj ): 1 −2 1 0 0 √ T (e) = 0 1 0 T (c1 ) = 3 2 0 0 1 0 1 −2 √3 T (c2 ) = − 2 0 − 23 −1 2 0 √ 0 0 1 √ 3 2 −1 2 0 0 −1 0 0 0 T (d1 ) = 0 1 0 0 0 −1 1 1 0 √ 23 0 T (d3 ) = − 2 −1 0 T (d2 ) = 1 2 √3 0 2 3 2 −1 2 √ 0 − 23 −1 2 0 √ 0 0 −1 The characters are χ(Ce ) = 3. As can be seen. The set of Cartesian unit vectors (ex . The matrices can be ˆ calculated by using the deﬁnition Tij (G) = (φi . ez ) do form a basis for the representation of the group. Let us see how they transform under the group operations. We choose a set of basis vectors or functions in order to form a representation for the group elements. 1. Cc = (c1 . ey . and χ(Cd ) = −1. ey .
2. χ(Cc ) = −1. we reproduce the character table of the group D3 . We construct an orthonormal basis that spans this subspace and we form the representation for the group. The 1×1 representation is obviously irreducible. We could have arrived at this conclusion simply by calculating the characters of the 2 × 2 representation which are χ(Ce ) = 2. we consider the same group but we choose a basis which consists of functions rather than the Cartesian vectors. We then use . We recall that the character of a given representation can be written as χ = α mα χα and by using the ﬁrst orthonormality relation between characters the coeﬃcients mα can be calculated as follows: mα = 1 h Nk χα∗ (Ck )χ(Ck ) k For convenience.7. and generate an invariant subspace from it. Ce 1 1 2 2Cc 3Cd 1 1 1 −1 −1 0 T T2 T3 Thus m1 = m2 m3 1 1 [1 · 1 · 3 + 2 · 1 · 0 + 3 · 1 · (−1)] = 0 6 1 = [1 · 1 · 3 + 2 · 1 · 0 + 3 · (−1) · (−1) = 1 6 1 = [1 · 2 · 3 + 2 · (−1) · 0 + 3 · 0 · (−1) = 1 6 Hence the representation breaks up into T = T 2 ⊕ T 3 and therefore (ex . say x2 . We can take an arbitrary function. EXAMPLES 43 representations. but it may be possible to further reduce the 2 × 2 representation.3. ey ) do form an irreducible subspace for the representation T 3 . As a second example. and χ(Cd ) = 0 and which agree with the characters of the T 3 representation.
T (G)x = −1 ˆ T (G)(ex · r) = (ex · G r) = (Gex · r). and ez and we have ˆ T (e)x2 = x2 ˆ T (c1 )x 2 = ˆ T (c2 )x2 = ˆ T (d1 )x2 = ˆ T (d2 )x2 = ˆ T (d3 )x2 = √ √ 1 3 2 1 2 3 2 3 (− x + y) = x + y − xy 2 2 4 4 2 √ √ 1 3 2 1 2 3 2 3 (− x − y) = x + y + xy 2 2 4 4 2 (−x)2 = x2 √ √ 1 3 2 1 2 3 2 3 ( x+ y) = x + y + xy 2 2 4 4 2 √ √ 3 2 1 2 3 2 3 1 ( x− y) = x + y − xy 2 2 4 4 2 These six functions form an invariant subspace for D3 but all of them can be written as linear combinations of only three distinct functions χ1 = x2 . ey . Let us start with χ1 = x2 which we normalise inside a unit sphere so that A2 (χ1 . Thus x. The normalised function is then φ1 = Ax2 . What we need to know is how x transforms under the group opˆ ˆ erations. χ2 = y 2 . We now use the GrammSchmidt orthogonalisation procedure to construct a set of orthonormal basis needed to construct a unitary representation for the group. χ1 ) = which gives A = 1 0 dr r2 π 0 dθsin θ 2π 0 dφ χ∗ χ1 = 1 1 35/4π.e. We have deﬁned T (G)f (r) = f (G−1 r).44 CHAPTER 3. Evidently none of these three functions can be written as a linear combination of the other two so that they are linearly independent and form a basis for the invariant subspace. y and z transform like the Cartesian vectors ex . i. We can start with any of the three functions or any linear combination of them. GROUP THEORY IN QUANTUM MECHANICS the projection operator method to ﬁgure out which functions belong to a given irreducible subspace. and χ3 = xy. although they are neither orthogonal nor normalised.
χ2 ) which yields after normalisation 1 φ2 = A √ (3y 2 − x2 ) 2 2 The function xy is already orthogonal to both x2 and y 2 due to antysymmetry so that the third orthonormalised function is √ φ3 = A 3xy To construct the representation matrices. we need to know how these basis functions transform into one another under the group operations.7. φ1 φ1 1 √ φ2 − 2 1 √ φ2 + 2 φ1 1 √ φ2 + 2 1 √ φ2 − 2 φ2 φ3 φ2 φ3 1 1 1 √ (φ1 + φ3 ) √ φ2 − 1 φ3 φ − 2 2 1 2 2 1 1 1 √ (φ1 − φ3 ) − φ1 + √ φ2 − 1 φ3 2 2 2 2 φ2 −φ3 1 1 √ (φ1 − φ3 ) √ φ2 + 1 φ3 φ − 12 2 1 2 2 1 1 1 √ (φ1 + φ3 ) − 2 φ1 + √2 φ2 + 1 φ3 2 2 ˆ T (e) ˆ T (c1 ) ˆ T (c2 ) ˆ T (d1 ) ˆ T (d2 ) ˆ T (d3 ) 1 φ 2 1 1 φ 2 1 1 φ 2 1 1 φ 2 1 + + + + 1 φ 2 3 1 φ 2 3 1 φ 2 3 1 φ 2 3 The matrix representations can be easily obtained and these are 1 0 0 2 1 T (e) = 0 1 0 T (c1 ) = √2 0 0 1 −1 2 T (c2 ) = 1 2 1 √ 2 1 2 1 √ 2 1 2 1 2 1 √ 2 1 1 √ 2 0 1 √ 2 −1 2 −1 2 0 1 − √2 1 √ 2 1 1 √ T (d ) = 0 1 2 1 2 1 − √2 1 −2 0 0 1 0 0 0 −1 1 √ 2 1 −2 1 √ 2 1 2 T (d2 ) = 0 1 − √2 1 2 1 − √2 1 2 1 2 1 T (d3 ) = √ 2 0 1 √ 2 1 −2 . EXAMPLES 45 ˜ The second function is constructed by using φ2 = χ2 − φ1 (φ1 .3.
Let us apply P 1 to φ1 . Since the representation contains an identity representation. these matrices are not in block form but we know that they are reducible because there is no three dimensional representation of D3 . GROUP THEORY IN QUANTUM MECHANICS Unlike the previous example. and χ(Cd ) = 1. it must be possible to construct a linear combination of φ1 . Let us ﬁgure out which irreducible representations are contained in these matrices by using the same formula as in the previous example. χ(Cc ) = 0. We do this in a systematic way using the projection operator method.46 CHAPTER 3. and φ3 such that it is invariant under the group operations. We get m1 = m2 m3 1 [1 · 1 · 3 + 2 · 1 · 0 + 3 · 1 · 1] = 1 6 1 = [1 · 1 · 3 + 2 · 1 · 0 + 3 · (−1) · 1] = 0 6 1 = [1 · 2 · 3 + 2 · (−1) · 0 + 3 · 0 · 1] = 1 6 Hence the representation breaks up into T = T 1 ⊕ T 3 . φ2 . The projection operator is given by lα ˆ Pα = h ˆ χα∗ (G)T (G) G which when applied to an arbitrary function gives the component of the ˆ function in the irreducible subspace α. The reducible characters are χ(Ce ) = 3. φ2 . and φ3 : 1 1 1 1 1 1 1 ˆ P 1 φ1 = φ1 + ( φ1 + √ φ2 − φ3 ) + ( φ1 + √ φ2 + φ3 ) 6 2 2 2 2 2 2 1 1 1 1 1 1 +φ1 + ( φ1 + √ φ2 + φ3 ) + ( φ1 + √ φ2 − φ3 ) 2 2 2 2 2 2 √ 2 √ = ( 2φ1 + φ2 ) 3 1 1 1 ˆ φ2 + ( √ (φ1 + φ3 )) + ( √ (φ1 − φ3 )) P 1 φ2 = 6 2 2 1 1 +φ2 + ( √ (φ1 − φ3 )) + ( √ (φ1 + φ3 )) 2 2 .
Instead one tries to ﬁnd . The new basis may be written as ψi = j φj Sji where S is the unitary matrix that reduces the representation into block form which is given by S= 0 2 3 1 3 1 3 0 2 3 − 0 0 1 3.8 Relationship to Quantum Mechanics A basic problem in quantum mechanics is ﬁnding the eigenvectors and ˆ eigenvalues of a Hamiltonian H. but in most cases.3. RELATIONSHIP TO QUANTUM MECHANICS = 1 √ ( 2φ1 + φ2 ) 3 47 ˆ P 1 φ3 = 0 √ √ The linear combination ( 2φ1 + φ2 )/ 3 transforms according to the identity representation. this procedure is too diﬃcult to carry out.8. ψ3 ) transforming according to the irreducible representation T 3 . A general procedure is to calculate the Hamiltonian matrix in an arbitrary basis and diagonalise it. The component of φ1 in the third irreducible subspace is given by 2 1 1 1 1 1 1 ˆ P 3 φ1 = 2φ1 − ( φ1 + √ φ2 − φ3 ) − ( φ1 + √ φ2 + φ3 ) 6 2 2 2 2 2 2 √ 1 = (φ1 − 2φ2 ) 3 Thus we may choose a new basis 1 √ ψ1 = √ ( 2φ1 + φ2 ) 3 √ 1 ψ2 = √ (φ1 − 2φ2 ) 3 ψ3 = φ3 which brings the representation into block form with ψ1 transforming according to the identity representation T 1 and (ψ2 .
also an eigenvector of H Proof: By deﬁnition ˆ ˆ ˆ ˆ T (G)H(r) = H(G−1 r) = H(r) Let us investigate the consequences of this symmetry by considering the ˆ ˆ eﬀect of T (G) on Hψ(r) where ψ is an arbitrary function . o ˆ ˆ Moreover. H] = 0.48 CHAPTER 3. The second part of the theorem follows directly from this result: ˆ ˆ ˆ ˆ ˆ H(r)T (G)ψ(r) = T (G)H(r)ψ(r) = T (G)ψ(r) . 3. if ψ(r) is an eigenvector of H. [T . It is here that the study of the symmetry of the Hamiltonian is of great value because group theory can tell us which matrix elements are zero and how the matrix elements are related to one another. GROUP THEORY IN QUANTUM MECHANICS a basis in which the oﬀ diagonal elements of H are small or zero so that they can be neglected or treated perturbatively. H.8.1 The Group of the Schr¨dinger Equation o ˆ If {G} is a set of all symmetry transformations of a Hamiltonian. The results are exact and depend only on the symmetry of the systems and not on the particular form of interactions or potentials. o Theorem 5: Each element of the Schr¨dinger group commutes with the Hamiltonian. ˆ ˆ ˆ T (G)H(r)ψ(r) = H(G−1 r)ψ(G−1 r) ˆ ˆ = H(r)T (G)ψ(r) ˆ ˆ ˆ ˆ T (G)H(r) − H(r)T (G) ψ(r) = 0 ˆ ˆ Since ψ(r) is arbitrary. then T (G)ψ(r) = ψ(G−1 r) is ˆ with the same eigenvalue. then ˆ according to Theorem 1 the associated symmetry operators {T (G)} form a group which is called the group of the Schr¨dinger equation or in short o the Schr¨dinger group.
Moreover. We may label each eigenfunction by the row and o the irreducible representation to which it belongs.3. The degeneracy is simply given by the dimensionality of the irreducible representation. Assuming that there is no accidental degeneracy. by knowing the symmetry of the perturbation.2 Degeneracy and Invariant Subspace The implication of Theorem 5 is that a degenerate set of eigenfunctions ˆ of H form an invariant subspace under the symmetry operations of the Schr¨dinger group and therefore form a basis for a representation of the o group. . Thus by knowing the symmetry of the Hamiltonian. 2. o The representation is normally irreducible but if it is reducible. each degenerate set ˆ of eigenfunctions of H form a basis for an irreducible representation of the Schr¨dinger group. Let {ψi } be a set of degenerate eigenfunctions with energy . Since ˆ T (G)ψi (r) = ψi (G−1 r) is also an eigenfunction with the same energy . In this sense. we can also tell how the degeneracies are split. It follows that these degeneracies can only be broken by a perturbation if it has a lower symmetry than the unperturbed Hamiltonian. accidental degeneracies my be lifted up by varying parameters in the Hamiltonian without changing the symmetry of it. the degeneracy is termed ”normal”. RELATIONSHIP TO QUANTUM MECHANICS 49 3. There is an accidental degeneracy so that the subspace can be broken up into two or more irreducible subspaces.8. it ˆ must be possible to expand T (G)ψi (r) as a linear combination of {ψi }: ˆ T (G)ψi = j ψj Tji (G) The matrices {T (G)} form a representation for the Schr¨dinger group. the row index and the irreducible representation provide ”quantum numbers”. we can tell the degrees of degeneracy possible in any problem. If the subspace is irreducible. In practice. then there are two possibilities: 1. The group is a subgroup of a larger group.8. A vector in one irreducible subspace evidently cannot be brought into another irreducible subspace by the symmetry operators.
This is a special case of i the so called WignerEckart theorem discussed later.9 Irreducible Sets of Operators We have studied the concept of irreducible subspace formed by a set of functions which transform among themselves under the group operations.3 ˆ Partial Diagonalisation of H Theorem 6: ˆ β The matrix element (φα . py . Hψlβ ) ki k h G kl 1 ˆ = δkl δij δαβ (φα . . Hψlβ ) ki k 1 β ˆ = T α∗ (G)Tlj (G)(φα . and pz under rotation. An example of an irreducible set of operators is given by the momentum operators px .8. T (G)Hψj ) i i β ˆ ˆˆ = (T (G)φα . i Proof: ˆ β ˆ ˆ ˆ β (φα . A similar concept may be extended to operators and we deﬁne an irreducible ˆα set of operators Oi by the closure property ˆα α ˆα ˆ ˆα ˆ (3. The number of operators in such a set is equal to the dimension lα of the irreducible representation. GROUP THEORY IN QUANTUM MECHANICS 3. Also.50 CHAPTER 3. Hψk ) k lα k This result is very useful because by choosing basis functions with the appropriate symmetry. 3. H T (G)ψj ) i = kl β ˆ T α∗ (G)Tlj (G)(φα .6) Oj Tji (G) Oi = T (G)Oi T −1 (G) = j ˆα Thus the irreducible set of operators Oi transform among themselves under the irreducible representation T α . Hψj ) = (T (G)φα . (φα . Hψi ) is independent of i. the Hamiltonian is partially diagonalised with very ˆ β little eﬀort. Hψj ) is zero unless α = β and i = j. Hψlβ ) k lα kl 1 ˆ β = δαβ δij (φα .
jj (G2 ) β α α Tik (G1 )Tiβk (G1 )Tkj (G2 )Tk j (G2 ) kk α Tij (G1 G2 )Tiβj α⊗β Tii .7) (A ⊗ B)ii . then T α⊗β = T α ⊗ T β is also a representation of the group.jj (G1 G2 ) = = = (G1 G2 ) Theorem 8: The character of a direct product representation T α⊗β is the product of the characters of T α and T β .10. jj = kk α⊗β α⊗β Tii .10 Direct Product Representations of a Group The direct product of an n × n matrix A and and m × m matrix B is deﬁned by (3.ii α Tii Tiβi ii α β = = χ χ .3.kk (G1 )Tkk .jj = Aij Bi j (A ⊗ B) is an (mn × mn) matrix and it is not the same as a matrix multiplication of A and B. Proof: [T α⊗β (G1 )T α⊗β (G2 )]ii . DIRECT PRODUCT REPRESENTATIONS OF A GROUP 51 3. Theorem 7: If T α and T β are two representations of a group. Proof: χα⊗β = ii α⊗β Tii .
There is a unitary matrix U . which transforms T α⊗β into a block form: α⊗β † γ Uγmk. which tells us the number of times the irreducible representation γ occurs in the decomposition. Proof: α β ˆ T (G)(ψi ψj ) = k l α⊗β α (ψk ψlβ )Tkl.i j (3.ij kl β α α ψk Tki (G)ψlβ Tlj (G) = In general.γ m k = δγγ δmm Tkk ij. the direct product representation is reducible even when the two constituent representations are irreducible.9) γ labels the irreducible representations and the coeﬃcients mγ .ij Tij. Thus. then {ψi ψj } transform according to the direct product representation T α⊗β . we may write T α⊗β = γ ⊕mγ T γ (3.i j Uij. GROUP THEORY IN QUANTUM MECHANICS Theorem 9: β α If {ψi } transform according to the representation T α and {ψj } transform α β according to the representation T β .8) The label m distinguishes blocks with the same irreducible representation. can be calculated using the ﬁrst orthogonality relation of the characters χα⊗β = γ mγ χγ mγ χγ γ χ χ α β = mγ = (1/h) k Nk χα (Ck )χβ (Ck )χγ∗ (Ck ) .52 CHAPTER 3. independent of the group elements.
( 3. We may also write α β ψi ψj = kγm Ψγ (m)Uγmk.ij (m) (3.ij (m) and therefore α β ψi ψj = γm k γ γ fk (m)Ck.γmk Tk k ijk γ Ψγ (m)Tk k k k = = We have used Eq.γmk transform k under a group symmetry operation: ˆ T (G)Ψγ (m) = k ij. Thus the states Ψγ (m) k transform according to the irreducible representation γ.ij k (3. The label m distinguishes functions with the same γ which is necessary only when mγ > 1 in the decomposition of the product representation.i j † α β ψi ψj Tiα⊗β Uij.11.11 ClebschGordan Coeﬃcients α β Since the set of product functions ψi ψj transform according to the product representation T α⊗β which is in general reducible.8) in the second step. the ClebschGordan coeﬃcients are actually identical to the unitary matrix U deﬁned in the previous section. evidently the space formed by the product functions are also reducible and the product functions may then be projected into these irreducible subspaces: α β ψi ψj = γ Fij (m) γm γ where Fij (m) belongs to the irreducible subspace γ. In other words. To α β † see this.γmk j . there may be more than one subspace which transform according to the same γ irreducible representation. we consider how the state Ψγ (m) = ij ψi ψj Uij. CLEBSCHGORDAN COEFFICIENTS 53 3.ij γ α β ψi ψj Ui† j .11) . Fij (m) itself may be expanded in an orthogonal basis which spans the irreducible subspace γ: γ Fij (m) = k γ γ fk (m)Ck. Apart from a possible phase factor.10) The coeﬃcients C are called the ClebschGordan coeﬃcients and they depend on α and β.3.
apart from a phase factor.ij (m)Ck. Oi fjβ ) ˆα We study the transformation of the function ψij = Oi fjβ under a group symmetry operation: ˆ ˆ ˆ T (G)ψij = T (G)Oα fjβ i ˆ ˆ ˆα ˆ = [T (G)Oi T −1 (G)] T (G)fjβ β β ˆα α = Ok Tki (G) fm Tmj (G) k = km m α⊗β ψkm Tkm. Oφ). In practice. they are normalised as follows: γ Cijk (m)2 = 1 (3.ij (m) k .12) ij Since C is unitary.11): ψij = γkm γ Ψγ (m)Ck .ij (m)Ck . GROUP THEORY IN QUANTUM MECHANICS which shows that the ClebschGordan coeﬃcients are identical to the unitary matrix U . we are often faced with a problem of calculating matrix elements ˆ ˆ of an operator O between two states ψ and φ: (ψ. By convention.ij (m ) = δγγ δmm δkk ij γ ∗γ Ck. the calculation of matrix elements can be very complicated and we would like to study if group theory can help us in simplifying this problem. from previous sections. we need only consider matrix elements of the type: γ ˆα (fk . ( 3.i j (m) = δii δjj γkm 3.12 The WignerEckart Theorem In physics. it satisﬁes the following relations γ ∗γ Ck. Consequently. First of all.54 CHAPTER 3.ij (G) The new function ψij may then be completely expanded into its irreducible components as in Eq. we know that an arbitrary function belonging to a linear space L which transforms into itself under group operations can be projected into the diﬀerent irreducible components of the group.
13) This is known as the WignerEckart theorem which tells us that the matrix γ ˆα element (fk . β. j. and k to one another. The ClebschGordan coeﬃcients relate the matrix elements for a given α. i. These matrix elements are often referred to as reduced matrix elements and they are written as γ ˆ (fk . Ψγ (m))Ck . T (G)H T −1 (G) = H. j. Thus we have γ ˆα (fk . 1. In fact.13 Applications We list a few of the common applications of group theory.ij (m) k γkm γ γ (fk . Ψγ (m)) = f γ Oα f β k m emphasising the fact that the reduced matrix elements are independent of i. We can see that theorem 6 follows immediately from the WignerEckart theorem because the Hamiltonian is an invariant under the group symmetry ˆ ˆˆ ˆ ˆ operations. it does not depend on k as shown in section 3. Ψγ (m)) which do k not depend on i and j. This is a general form of selection rules.ij (m) are obtained entirely from symmetry and they do not depend on the details of the potential of the system.3. Ψγ (m))Ck. and γ but with diﬀerent i. APPLICATIONS The matrix element is then given by γ ˆα (fk . Information about the details γ of the system is contained in the matrix elements (fk . so that H transforms according to the identity representation. Selection rules: As an example we consider optical transitions in a molecule which has a D3 symmetry. This could correspond to a .13. 3. and k. Oi fjβ ) = γ γ (fk .e.ij (m) (3. Oi fjβ ) is zero unless the product representation T α⊗β contains the irreducible representation γ. Oi fjβ ) = ˆ f γ Oα f β m γ m Ck.ij (m) k 55 = m The signiﬁcance of the above result is that the ClebschGordan coeﬃcients γ Ck.6.
The groundstate of the molecule may be assumed to be nondegenerate and symmetric so that it transforms according to the identity representation. and z which transform like the three Cartesian unit vectors. must split into two nondegenerate levels because C2 is an Abelian group and the irreducible representations are one dimensional.56 CHAPTER 3. HΨ) ≥ E0 . however. Ψ) = 1: ∂ ∂ ˆ [< E > −λ(Ψ. The coeﬃcients {ci } are determined by minimising < E > subject to the condition (Ψ. From the previous example. 3. then this degeneracy. the dipole operators transform according to the representation T = T 2 ⊕ T 3 and therefore the ﬁnal states can only be those which transform according to (T 2 ⊕ T 3 ) ⊗ T 1 = T 2 ⊕ T 3 . Hφk )ck − λ ∗ ∂ci ∂c∗ jk j i = k c∗ cj j j Hik ck − λci = 0 . GROUP THEORY IN QUANTUM MECHANICS molecule with three identical atoms at the corners of a triangle. Optical transitions are governed by dipole operators x. if it exists. The ordering of the energy levels. If we replace one of the atoms by a diﬀerent atom so that the symmetry becomes C2 . we know from symmetry alone that the eigenfunctions can only be at most doubly degenerate. the full symmetry is higher than D3 but as an illustration it is suﬃcient to consider the D3 symmetry. We start with some basis functions {φi } and approximate the true groundstate Ψ as a linear combination of these functions. Ψ)] = c∗ (φj . the energy expectation value < E >= ˆ (Ψ. The actual reduction can be ﬁgured out straightforwardly using the orthogonality relation between the characters. where E0 is the true groundstate energy. Ψ= ci φi i From the variational principle. One method of obtaining an approximate groundstate is based on variational principle. Symmetry breaking: In the above example. cannot be determined by group theory alone. Variational principle: We are often interested in ﬁnding the groundstate of a given Hamiltonian which usually cannot be solved analytically. Actually. 2. y.
3.14. EXAMPLES which is just an eigenvalue problem.
57
From Theorem 6, it is clear that we should sort out the basis functions, using the projection operator method for example, according to the rows and irreducible representations {φαm } where m labels i diﬀerent possible subspaces transforming according to the same irreducible representation α. It is only necessary to express the trial function as follows: ψ = m cαm φαm . When m = 1, the Hamiltoi i nian is already diagonal and when m > 1, we need only diagonalise an m × m matrix instead of an l × l matrix where l = αmi 1, which is usually much larger than m. 4. Degenerate perturbation theory: In many cases, the Hamiltonian ˆ ˆ ˆ ˆ may be broken up into two terms H = H0 + H1 such that H1 may ˆ ˆ be treated as a small perturbation to H0 . H0 usually has a higher ˆ symmetry than H1 . A typical example of this is an atom placed in a crystal where the crystal ﬁeld symmetry is lower than the rotational ˆ symmetry of the atom. If the eigenfunctions of H0 are degenerate, we may use the variational principle to calculate the energy shifts. The ˆ eigenfunction of H is approximated as a linear combination of the ˆ degenerate eigenfunctions of H0 . This leads to an eigenvalue problem ˆ 1 and the eigenvalues give the ﬁrst order shifts. As before, we for H ˆ should sort out the degenerate eigenfunctions of H0 according to the ˆ rows and irreducible representations of the symmetry group of H1 in order to minimise the size of the matrix to be diagonalised.
3.14
Examples
1. Bloch Theorem: A lattice is a parallelepiped formed by three linearly independent vectors a1 , a2 , a3 . A set of vectors t = n1 a1 + n2 a2 + n3 a3 for all possible integers n1 , n2 , n3 generate lattice points which together with the associated lattices form a crystal. The set t = {t} clearly form a group under vector addition and the group is Abelian. In fact, it is a direct product of three groups t = t1 ⊗ t2 ⊗ t3
58 CHAPTER 3. GROUP THEORY IN QUANTUM MECHANICS where t1 = {n1 a1 }, t2 = {n2 a2 }, and t3 = {n3 a3 }. It is customary to have a periodic boundary condition such that ˆ ˆ ˆ T (N1 a1 )ψ(r) = ψ(r + N1 a1 ) = ψ(r) → T (N1 a1 ) = T (0) = 1 Similarly for the other two groups. N1 is taken to be inﬁnite at the end. The groups t1 , t2 , and t3 become cyclic and the irreducible representations are given by ei2πk1 /N1 , k1 = 1, 2, . . . , N1 , and their powers. It is useful at this stage to introduce the concept of reciprocal space deﬁned by g1 = 2π g2 g3 so that gi · aj = 2πδij With this deﬁnition, the direct product representations can be written T k (t) = eik·t where k = (k1 /N1 )g1 + (k2 /N2 )g2 + (k3 /N3 )g3 . It follows that the eigenfunctions of a Hamiltonian with translational lattice symmetry form basis functions for the irreducible representations of the group of translations and they may be labelled by the irreducible representations k. This means ˆ T (t)ψk (r) = ψk (r + t) = ψk (r)eik·t which is the Bloch theorem. 2. Normal modes (chapter 6 of Elliot and Dawber): When the atoms in a molecule are displaced from their equilibrium positions, the potential energy will increase. If the atoms are let free, there will be an interchange between potential and kinetic energies resulting in vibrations. These vibrations may be thought of as being composed a2 × a3 a1 · a2 × a3 a3 × a1 = 2π a1 · a2 × a3 a1 × a2 = 2π a1 · a2 × a3
3.14. EXAMPLES
59
of eigen modes or normal modes, each with a well deﬁned frequency, in the same way that a wavefunction can be decomposed into a sum of eigenstates with well deﬁned energies. The eigen modes may be classiﬁed according to the irreducible representations of the symmetry group of the molecule, just like the the eigenstates of the Hamiltonian. The potential energy for arbitrary displacements of N atoms in a molecule is given by V = V0 +
i
∂V qi + ∂qi
ij
∂ 2V qi qj ∂qi ∂qj
The sums run from 1 to 3N . At equilibrium, ∂V /∂qi = 0 so that the Hamiltonian for molecular vibrations is given by H=
i
1 Mi qi + ˙2 2
ij
1 Bij qi qj 2
where Bij = ∂ 2 V /∂qi ∂qj . It is conventional to absorb the masses into the displacement coordinates by deﬁning αi = Mi qi , Dij = Bij / Mi Mj
The Hamiltonian becomes H=
i
1 2 α + ˙ 2 i
ij
1 Dij αi αj 2
The matrix D is known as the dynamical matrix. We can now introduce a 3N dimensional abstract space with orthonormal basis vectors {ei }. The 3N displacements of atoms in real space may be mapped into a vector in the 3N dimensional abstract space: (q1 , q2 , . . . , q3N ) → q = ei αi √ The unit vector ei corresponds to a displacement qi = 1/ Mi in real space but we note that ei itself is not a displacement in real space
i
So far we have not made any use of group theory and in principle we can calculate the dynamical matrix D and diagonalise it to obtain . we replace Q2 1 2 2 2 2 2 with −∂ /∂Qk and the solutions to h = −∂ /∂Qk + 2 ωk Qk are √ ψn(k) = AHn(k) ( ωk Qk )exp(−ωk Qk /2) 1 n(k) = (n(k) + )ωk 2 We have set the Planck constant h/2π = 1.e. Dej ) → Dej = i ei Dij We would like to transform the Hamiltonian into the form H = i hi so that the eigenvenctors of H can be written as a product of the eigenvectors of h. This can be done by a unitary transformation uk = i ei aik An arbitrary vector q may be written as q= i ei αi = k uk Qk = ki ei aik Qk → αi = k aik Qk ˆ ˆ By choosing uk to be an eigenvector of D. Dej ) 2 ij which implies ˆ ˆ Dij = (ei . Dq) 2 1 ˆ = αi αj (ei . Duk = ωk uk or j Dij ajk = ωk aik . GROUP THEORY IN QUANTUM MECHANICS but rather a displacement in the abstract space. i. Quantum mechanically. the Hamiltonian becomes H= k 1 ˙2 1 Qk + ωk Q2 k 2 2 ˙k which is the desired form.60 CHAPTER 3. We may deﬁne an ˆ operator D in the abstract space so that V (q) = 1 ˆ (q.
As an example. The diagonalisation of D is greatly simpliﬁed by using group theory. χ3N (Cc ) = 0. i Tii = 0. This result can be understood by observing that when RR ˆ T (G) moves atom R to atom R . . D] = 0 for all G and it immediately follows from Theorem 5 that the degenˆ erate eigenvectors of D (normal modes) form basis vectors for the irreducible representations of the group. The eﬀect of the symmetry operation on the basis vector is given by ˆ T (G)ei (R) = Rj R ej (R )Tji R (G) The character is given by χ3N = Ri RR Tii = NR χV where NR is the number of atoms unmoved by the symmetry operˆ ation T (G) and χV is the character of the three dimensional vector representation. and χV (Cd ) = −1. The eigen frequencies. we consider a molecule with D3 symmetry (three atoms at the corners of an equilateral triangle). i = x. then i Tii = χ .3. χV (Cc ) = 0. cannot be obtained from group theory alone because they depend on the strength of the potential. EXAMPLES 61 the normal modes and the corresponding eigen frequencies. however.14. and χ3N (Cd ) = 1 · (−1) = . y. we ﬁrst write the basis vector as ei (R). 2. The 3N dimensional space forms a 3N × 3N representation which is in general reducible: χ3N = γ mγ χγ To calculate χ3N . . we have ˆ ˆ [T (G). By deﬁnition. . z. N . When atom R is V RR not moved. Thus χ3N (Ce ) = 3 · 3 = 9. which corresponds to a displacement of atom R along the iaxis. . We have already worked out the characters for the vector representation χV and they are given by χV (Ce ) = 3. R = 1.
In more complicated absorption processes. yz. i. and z 2 (Raman spectra). It is then straightforward to decompose χ3N into the diﬀerent irreducible components using the character table of D3 . y.62 CHAPTER 3. six of them correspond to rigid motions. See chapter 6 of Elliot and Dawber for more examples and details. The simplest absorption processes are caused by electric dipole transitions from the groundstate to some vibrational excited states.e. . there is no relative motion between the atoms. i. The groundstate usually tranforms according to the identity representation and the dipole operators transform according to the vector representation. These are in fact products of x. Care. they transform according to the product of the vector representations which can be decomposed into the irreducible components.e. zx. We get m1 = m2 m3 1 (1 · 1 · 9 + 2 · 1 · 0 + 3 · 1 · (−1)) = 1 6 1 = (1 · 1 · 9 + 2 · 1 · 0 + 3 · (−1) · (−1)) = 2 6 1 = (1 · 2 · 9 + 2 · (−1) · 0 + 3 · 0 · (−1)) = 3 6 There are three nondegenerate modes and three doubly degenerate modes. y 2 . has to be taken due to the fact that the product functions are not linearly independent. and z. x2 . The signiﬁcance of classifying the vibrations into normal modes is that when calculating absorption or emission spectra it is possible to ﬁgure out which modes are allowed in the absorption process. In this case the dominant operators are of the form xy. there are intermediate states involved. Thus the modes allowed are given by those corresponding to the irreducible representations occuring in the decomposition of the vector representation. however. GROUP THEORY IN QUANTUM MECHANICS −1. Actually.
Such groups have an inﬁnite number of elements. it follows from Eqs. .5) (4. a2 .1 Deﬁnitions So far we have considered groups with a ﬁnite number of elements. . c must be a function of the paramaters a and b: c = F(a.Chapter 4 Lie Groups 4. 0) = 0 63 (4. ( 4.2) . an ) so that G(a) = G(b) unless a = b. . F(0. a = 0 is chosen to represent the unit element: G(0) = 1 With this deﬁnition.1) and ( 4. b) = b . b)) As a convention.4) (4. In this Chapter. F(0.3) (4. .1) This means that operating a group element G(b) followed by G(a) has the same eﬀect as operating a group element G(c) represented by a set of parameters c. b) and we may write the multiplication law as: G(a)G(b) = G(F(a. The number n is called the dimension of the group. The elements of a continuous group satisfy the multiplication law: G(a)G(b) = G(c) (4.2) that F(a. we will study the so called continuous groups in which the elements G are deﬁned uniquely by a set of n independent continuous real parameters a = (a1 . 0) = a .
and this kind of group is called a Liegroup. LIE GROUPS ∂F(0.2 Inﬁnitesimal Operators An important concept in continuous groups is the concept of inﬁnitesimal operators which are obtained by considering group elements near the unit element i. every element must have an inverse i.6). Xn ) are called the inﬁnitesimal operators. 0) = =1 (4.b) are assumed to be analytic functions of the parameters a and b. or generators. . .9) ∂ai a=0 The set of operators X = (X1 . a dot denotes a dot product or a summation over the lower and upper vectors. It will be shown below that every group element may be expressed in terms of the inﬁnitesimal operators so that the study of continuous groups may be restricted in most cases to the study of these inﬁnitesimal operators. 4. Xi = (4. which are independent of the parameters a. 0) + = a+b ∂F(0. 0) ∂F(0. This implies that G(a)G(b) = G(F(a. there is a set of parameters q such that G(a)G(q) = G(q)G(a) = G(0) = 1 (4. . As a convention. . a)) = G(b + a) .e.e.64 CHAPTER 4. ( 4.6) ∂a ∂b In addition.8) ∂G (4. for a given set of parameters a. b)) = G(a + b) G(b)G(a) = G(F(b. b) = F(a. For small a and b we have F(a. . X2 . .7) The functions F(a. 0) ·b ∂b using Eq. elements with small a: G(a) = 1 + X · a + .
b)) = G(F(a.11) along this unit vector: G(a)X · u = ∂G(a) · M(a) · u ∂a (4. 0) + M(a) · b) = G(a + M(a) · b) ∂G(a) = G(a) + · M(a) · b ∂a On the other hand G(a)G(b) = G(a)[1 + X · b] ∂G(a) · M(a) · b ∂a Since b are arbitrary and independent parameters. we deﬁne a matrix M(a) ≡ ∂F(a.10) and a convention that a dot on the left of M means a summation over the rows whereas a dot on the right means a summation over the columns. the group elements are commutative in the vicinity of the unit element. G(a)G(b) = G(F(a.13) . . we study the product G(a)G(b) for an arbitrary a but small b. un ) be a unit vector (u · u = 1) in the parameter space and let us project Eq.4.12) We deﬁne a curve in the parameter space by the following equation da(t) = M(a) · u dt (4. INFINITESIMAL OPERATORS 65 i. it must be that G(a)X · b = G(a)X = ∂G(a) · M(a) ∂a (4. 0) → Mij (a) ≡ ∂b ∂bj (4. To relate the group elements and the inﬁnitesimal operators. . u2 . .11) so that This equation relates the group element and its derivative at a to the inﬁnitesimal operators X. .2. We now try to ﬁnd an explicit expression relating G(a) to X. 0) ∂Fi (a. Let u = (u1 . To simplify the writing. ( 4.e.
0) ∂ai ∂bj Interchanging i ↔ j and taking the diﬀerence. The direction of this curve at t = 0 is (0 along the unit vector u because M(0) = ∂ F∂ b.13) in ( 4. we diﬀerentiate Eq. ( 4.14) Thus.3 Lie Algebra The inﬁnitesimal operators satisfy commutation relations and to ﬁnd these relations. 0) + ∂ai ∂ak ∂bj ∂ak ∂ai ∂bj k Evaluating this expression for a = 0. 4.0) = 1.15) . We have made an assumption. we get Xi Xj = ∂ 2 G(0) + ∂ai ∂aj Xk k ∂ 2 Fk (0. Xj ] = k Xk ck ij (4. The global properties of Lie groups are determined by the inﬁnitesimal operators X which are independent of the parameters a. we get [Xi .66 CHAPTER 4.11) with respect to a parameter ai for each component j: ∂G(a) Xj = ∂ai ∂ 2 G(a) ∂Fk (a. 0) ∂G(a) ∂ 2 Fk (a. by starting along an appropriate direction u from the origin. LIE GROUPS with boundary condition a(0) = 0. that every point a in the parameter space can be reached by the curve deﬁned in Eq. ( 4. ( 4. however.13). the group elements at ﬁnite parameters a can be obtained from the inﬁnitesimal operators X. Using Eq.12) yields G(a)X · u = ∂G(a) da · ∂a dt dG(a) = dt which can be formally solved to give G(a(t)) = etX·u (4.
By summing over products of two structure constants. 0) − ∂ai ∂bj ∂aj ∂bi (4.17) We can construct symmetric and antisymmetric quantities from the structure constants. The coeﬃcients ck are called the structure constants and they are ij purely determined by the functions F(a.20) which follows from the Jacobi identity [[A. . A] + [[C.4. The structure constants are antisymmetric in the lower indices: ck = −ck ij ji (4.16) We see that the inﬁnitesimal operators possess a closure property under commutation. This is because we have the following relation cl cm + cl cm + cl cm = 0 ij lk jk li ki lj l (4. B]. 0) ∂ 2 Fk (0. LIE ALGEBRA where ck = ij 67 ∂ 2 Fk (0.b) which means that they are independent of any particular representation of the group. A].18) Summing over products of these matrices and the structure constants yields aijk = m gim cm jk (4. B] = 0 In this way. we obtain symmetric matrices gij = kl ck cl = gji il jk (4.19) which are antisymmetric with respect to an interchange of two indices. C].3. Operators possessing this property are said to form an algebra and for Lie groups it is called Lie algebra. we can continue forming symmetric and antisymmetric quantities from the structure constants. C] + [[B. For every Lie group there is a corresponding Lie algebra.
( 4. it will be useful to have some operators which commute with all of the inﬁnitesimal operators. write cl = jk ml g lm gml cl jk . Xk ]Xj } ij = = ij g ij l Xi Xl cl + Xl cl Xj jk ik {Xi Xl + Xl Xi } = ijl g ij cl jk The last step makes use of the fact that g ij is symmetric. ( 4. ( 4. Xk ] g ij {Xi [Xj . Xk ] = ij g ij [Xi Xj . the sets of degenerate eigenfunctions of H form basis sets for the irreducible representations of the group (assuming we have the full group. We may.15) the antisymmetric part reduces to a linear combination of Xi .19) an antisymmetric quantity with three indices but it involves no upper indices in the structure constant.68 CHAPTER 4. The degenerate eigenfunctions of these operators may then be used as basis functions for the irreducible representations because they transform among themselves under group operations.4 Casimir Operators To construct the irreducible representations of a continuous group.21) ij The matrix g ij must be symmetric since every matrix can be written as a sum of symmetric and antisymmetric matrices and from Eq. If we can ﬁnd g ij such that j g ij cl is antisymmetric with respect to the interchange i ↔ l jk then C is the required operator. however. Thus we form the most general linear combination of second order operators C= g ij Xi Xj (4.15). We recall from Chapˆ ˆ ˆ ter 3 that when [T (G). not a subgroup of a larger group). H] = 0. We evaluate the commutation of this operator with an arbitrary Xk : [C. Xk ] + [Xi . LIE GROUPS 4. The simplest operator that commute with the inﬁnitesimal operators X must be at least of second order in these operators because a linear combination of Xi clearly will not in general commute with X due to Eq. We have in Eq.
Hi = Ci Ci . Hi = Hi Ci . we need only one Casimir operator to label the irreducible representations. The largest number of mutually commuting inﬁnitesˆ imal operators is called the rank of the group. Since {C ˆ ˆ to construct simultaneous eigenstates of {Ci } and {Hi } with eigenvalues {Ci } and {Hi }: ˆ Ci Ci . Hi . Xk ] we get [C.4.5 Ladder Operators and Multiplets For a given Lie group. Hi ˆ Hi Ci .19) Continuing the evaluation of [C. LADDER OPERATORS AND MULTIPLETS = m 69 g lm amjk where we have deﬁned the inverse of g ij g lm gml = δll m and amjk = l gml cl jk as in Eq. In general. ( 4. Xk ] = 0. some of the inﬁnitesimal operators may commute among themselves. Let {Hi } be such a set ˆ ˆ and {Eα } be the rest of the inﬁnitesimal operators. it may be possible to form higher order Casimir operators.5. Xk ] = ijl g ij m g lm amjk {Xi Xl + Xl Xi } = = 1 [g ij g lm + g lj g im ]amjk {Xi Xl + Xl Xi } 2 ijlm 1 [g ij g lm + g lj g im ][amjk + ajmk ] {Xi Xl + Xl Xi } 4 ijlm Since amjk = −ajmk we obtain [C. 4. which also commute with all of the inﬁnitesimal operators. In the rotation group discussed later. The irreducible representations of the group may be labelled by the eigenvalues of these Casimir operators. Let also {Ci } be a ˆi } commute with {Hi } it is possible ˆ set of Casimir operators.
( 4. Eα ] = iα Eα (4. Hi )Ci . LIE GROUPS ˆ The remaining set of operators {Eα } can be expressed as ladder operators: ˆ ˆ ˆ [Hi . Eα Ci . Acting Xi on Ci . Eq. Hi = iα Eα Ci . Eα ]Ci . there must be a state ψ+ and ψ− with the largest and smallest values of {Hi } such that ˆ ˆ Eα ψ+ = 0. Hi = Aα (Ci . Hi ˆ ˆ ˆ (Hi − Hi )Eα Ci . This means ˆ Eα Ci . Assuming that the multiplet spans a ﬁnite space. The states generated by the ladder operators are the degenerate eigenstates of the Casimir operators. Hi ) = (Hi + iα )(Eα Ci .22) implies that ˆ ˆ ˆ [Hi . Hi = iα Eα Ci . Hi results in a linear combination of states with the same eigenvalues {Ci }.70 CHAPTER 4. Hi ˆ ˆ ˆ Hi (Eα Ci .23) The constants Aα can be determined from the Lie algebra. Hi is also an eigenstate of Hi but with an eigenvalue (Hi + iα ).22) ˆ ˆ This can be done by forming a linear combination Eα = i Xi aiα and requiring that ˆ [Hi . Hi ) iα ˆ ˆ i. This is because the Casimir operators commute with all the inﬁnitesimal operators and therefore the degenerate eigenstates of the Casimir operators form an invariant subspace under the inﬁnitesimal ˆ operators. Hi + iα (4.e. j ˆ Xj ajα ] = ˆ Xk ck ajα ij = iα j iα j ˆ Xj ajα ˆ Xj ajα ck ajα = ij iα akα jk ˆ Xk k j [ck ij − iα δjk ]ajα =0 → j The coeﬃcients aiα are the eigenvectors of the structure constants and are called the root vectors. . Such a set of degenerate states are called a multiplet and they form an invariant subspace under the group operations. Eβ ψ− = 0 ˆ ˆ if Eα and Eβ increases and decreases the eigenvalues.
G(a)G(b) = G(c) = G(F(a.25) gives ∂c w(c) = w(a) ∂a (4. Eβ ] = [Hi .25) From Eq. ( 4. c is a function of a. we will study how to perform this summation for continuous groups. If iα + iβ = 0. b)). Eβ ] + Eα . Eβ ] is itself a ladder operator with a root ˆ ˆ ˆ ( iα + iβ ). For these theorems to remain valid in the case of continuous groups. Eβ ] ˆ ˆ ˆ ˆ so that the commutator [Eα . [Hi . In this section.24) w(a) is a weight function describing the ”density” of elements in the parameter space. then [Eα . We have to replace the summation by an integral over the parameters in the following way: f (G) → G da w(a)f (G(a)) (4.6 Summation over group elements Many of the theorems that we have encountered involve summation over group elements. da w(a)f (G(a)) = = Comparison with Eq. [Eα .6. Eα ]. ( 4. we must require da w(a)f (G(a)G(b)) = da w(a)f (G(c)) = dc w(c)f (G(c)) (4. 4.3). SUMMATION OVER GROUP ELEMENTS We also have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ Hi .26) dc w(c)f (G(c)) da ∂c w(c)f (G(c)) ∂a . In proving some theorems for ﬁnite groups.4. and for a ﬁxed b. Eβ ] is a linear combination of {Hi }. we make use of the rearrangement theorem. It is clear from the above discussion that the number of Casimir operators needed to specify a multiplet is equal to the rank of the Lie group. Eβ ] = ( iα 71 + iβ )[Eα .
. b) ∂F(a.27) 4.72 CHAPTER 4.26) gives 1+ ∂ · M(a) · b ∂a w(a) + ∂w(a) · M(a) · b = w(a) ∂a Since b is arbitrary.10). LIE GROUPS where ∂c/∂a is the Jacobian. The technique used in constructing the irreducible representations of the rotation group is actually quite common in Lie groups. in particular in the theory of angular momentum.7 Rotation Group R3 As an example. which often appears in many applications in physics. 0) + where the matrix M is deﬁned as in Eq. we consider the rotation group in three dimensions. ( 4. R3 . To ﬁrst order in b. To relate a to c we consider a small b: c = F(a. the Jacobian is given by ∂ ∂c =1+ · M(a) · b ∂a ∂a Expanding w(c) about a to ﬁrst order in b yields w(c) = w (a + M(a) · b) ∂w(a) = w(a) + · M(a) · b ∂a Substituting the above expressions into Eq. ( 4. 0) ·b ∂b = a + M(a) · b = F(a. we thus have 1 ∂w(a) ∂ · M(a) = − · M(a) w(a) ∂a ∂a or ∂ ln w(a) ∂ =− · M(a) · M−1 (a) ∂a ∂a (4.
In other words. α)r = (1 + α a×)r ai ei ×)r = (1 + α i (4.4. we have explicitly 0 0 0 Xx = 0 0 −1 0 1 0 0 0 1 Xy = 0 0 0 −1 0 0 0 −1 0 Xz = 1 0 0 0 0 0 .7. Xz = ez ×. the rotation can be regarded as a vector sum of rotations about x−. y− and z− axes respectively. It is easy to show that R(αa)r = cos α r + (1 − cos α)(r · a)a + sin α (a × r) For small α we have to ﬁrst order R(a. In the Cartesian basis.7. y− and z− axes.29) We identify Xx = ex ×. αy = αay and αz = αaz corresponding to rotations about x−. as the inﬁnitesimal operators of the rotation group associated with the parameters αx = αax . ROTATION GROUP R3 73 4. Xy = ey ×.28) = (1 + i αi ei ×)r (4. when α is small.1 Inﬁnitesimal operators A rotation in three dimensions may be denoted by R(αa) where a is the axis of rotation and α is the angle of anticlockwise rotation.
2 Commutation Relations In Cartesian coordinates. Xz ] = Xx . βb) = γc. independent of the basis.74 CHAPTER 4. Jy ] = iJz . [Xz . ( 4.16). βb) are taken at αa = 0 and βb = 0. For small α. Jz ] = iJx . The commutation relations become [Jx . it is suﬃcient to consider small α and β. Xx ] = Xy (4. Xy ] = Xz . We consider a rotation R(βb) followed by R(αa) which may be aﬀected by a rotation R(γc): R(γc) = R(αa)R(βb) The main quantities to calculate are the three functions F(αa. Thus we have R(γc) = R(αa)R(βb) = 1 − αy βy − αz βz −(αz + βz − αy βx ) αy + βy + αz βx 1 − αx βx − αz βz −(αx + βx − αz βy ) αz + βz + αx βy −(αy + βy + αx βz ) αx + βx − αy βz 1 − αx βx − αy βy The trace of R(γc) is (1 + 2cos γ). R(αa) takes the form: 1 −αz αy R(αa) = αz 1 −αx −αy αx 1 and similarly for R(βb). LIE GROUPS 4. Jx ] = iJy (4. Equating this with the trace of R(αa)R(βb) gives 1 + 2cos γ = 3 − 2(αx βx + αy βy + αz βz ) 1 + 2(1 − γ 2 /2) = 3 − 2(αx βx + αy βy + αz βz ) γ 2 = 2(αx βx + αy βy + αz βz ) . [Xy . the commutation relations for the inﬁnitesimal operators are [Xx .7. [Jy . Since the derivatives of F(αa.30) The inﬁnitesimal operators are skew Hermitian (this is true in general if the representation is unitary) but it is possible to deﬁne equivalent inﬁnitesimal operators Jk = iXk which are Hermitian. What we have to calculate are the structure constants deﬁned in Eq. [Jz .31) We will show below that the commutation relations also follow from the general results derived in the previous sections.
the solution to the above equations is γx = pA.32) 4.3 Casimir Operator The Casimir operator can also be obtained directly from the results of the previous section. ( 4. γy = pB.4.7. ROTATION GROUP R3 75 We also know that R(γc) is unitary and it does not alter γc so that R(γc)γc = R† (γc)γc = γc or [R(γc) − R† (γc)]γc = 0 This gives us the following equations: −Cγy + Bγz = 0 Cγx − Aγz = 0 −Bγx + Aγy = 0 A = 2(αx + βx ) + αy βz − αz βy B = 2(αy + βy ) + αz βx − αx βz C = 2(αz + βz ) + αx βy − αy βx 2 2 2 With γ 2 = γx + γy + γz .7. (4.18): gxx = kl ck cl xl xk = cz cy + cy cz xz xy xy xz = −2 . We calculate the matrix gij deﬁned in Eq. The nonzero structure constants are then cz = cy = cx = 1 xy zx yz which gives the commutation relations in Eq. γz = pC where 2(αx βx + αy βy + αz βz ) p= A2 + B 2 + C 2 1/2 A straightforward algebra shows that p = 1/2 to linear order in α and β.30). ( 4.
and Jz . Since J 2 commutes with the inﬁnitesimal operators. Jy . ˆ ˆ irreducible subspaces under the inﬁnitesimal operators J The rank of R3 is one because there are no two mutually commuting ˆ operators.30) alone. an irreducible subspace under these inﬁnitesimal operators is also an irreducible subspace under the rotation group. ( 4. LIE GROUPS The same results are obtained for gyy and gzz whereas the other matrix elements are zero. g ij = − δij 2 and the Casimir operator is therefore C = ij g ij Xi Xj 1 2 2 2 = − (Xx + Xy + Xz ) 2 1 1 = − X2 = J2 2 2 (4. the set of operators J± .34) which may be obtained by solving the eigenvectors and eigenvalues ˆ ˆ of ci . The sets of degenerate eigenvectors of the Casimir operator J 2 form ˆx . We construct the following ladder operators ˆ ˆ ˆ J± = Jx ± iJy (4.8 Irreducible Representations of R3 The method of constructing the irreducible representations for the rotation group R3 is quite general since it is based on the commutation relations in Eqn. The method may therefore be applied to other continuous groups. Thus we get 1 gij = −2δij .33) 4. Jz are equivalent to the set zj .76 CHAPTER 4. Since every rotation can be built up from the inﬁnitesimal operators. As a convention. The steps for constructing the irreducible representations are as follows: ˆ 1. Clearly. we choose Jz . the ˆ2 may be chosen simultaneously to be eigenvectors eigenvectors of J ˆ of one of the inﬁnitesimal operators.
Jy . We assume that the irreducible representations of R3 have a ﬁnite dimension. We show that ψm is an eigenfunction of J 2 irrespective of the value of m. ˆ ˆ Jz (J± ψm ) = = = = ˆ ˆ ˆ (J± Jz ± J± )ψm ˆ ˆ J± (Jz ± 1)ψm ˆ J± (m ± 1)ψm ˆ (m ± 1)(J± ψm ) ˆ Thus.4. [J+ . 2. ˆ 3. Let j be the largest eigenvalue of Jz in this basis.8. IRREDUCIBLE REPRESENTATIONS OF R3 77 ˆ ˆ ˆ of operators Jx . Jz .35) ˆ The usefulness of these operators may be seen by applying Jz to ˆ ˆ J± ψm where ψm is an eigenfunction of Jz with an eigenvalue m. This is to be expected because a set of degenerate eigenvectors . it must be that for some integer n we have ˆ J− ψj−n = 0 The problem is to ﬁgure out what n is. We may therefore work with the former set. Since we have assumed that the dimension of the irreducible representations is ﬁnite. The commutation relations of these operators are ˆ ˆ ˆ ˆ ˆ ˆ [Jz . We choose basis functions which are eigenfunctions of ˆ ˆ Jz . ˆ We take one of these and operate J− repeatedly on this function: ˆ J− ψj = cj−1 ψj−1 ˆ J− ψj−1 = cj−2 ψj−2 etc.36) There may be more than one function with the largest eigenvalue. J± ] = ±J± . Then ˆ J+ ψj = 0 (4. The c’s are normalisation constants. J− ] = 2Jz (4. J± ψm must be proportional to ψm±1 .
. . . j − 1. m = j. . . Operating J 2 on ψj we get ˆ ˆ ˆ ˆ2 ˆ J 2 ψj = (J− J+ + Jz + Jz )ψj = j(j + 1)ψj ˆ ˆ Operating J 2 J− on ψj yields ˆ ˆ ˆ ˆ J 2 J− ψj = J− J 2 ψj ˆ = j(j + 1)J− ψj−1 ˆ Thus ψj−1 is also an eigenfunction of J 2 with eigenvalue j(j + 1).35). ψj−1 . ˆ ˆ2 ˆ2 ˆ2 J 2 = Jx + Jy + Jz 1 ˆ ˆ ˆ ˆ ˆ2 = (J+ J− + J− J+ ) + Jz 2 ˆ ˆ ˆ2 ˆ = J+ J− + Jz − Jz ˆ ˆ ˆ2 ˆ = J− J+ + Jz + Jz The last two relations have been obtained by using the the second ˆ commutation relation in Eq. ψ−j form an irreducible subspace (transform among themselves or invariant) ˆ ˆ ˆ under the operators J+ . LIE GROUPS ˆ of J 2 form a basis for an irreducible representation. We show that they also ˆ transform among themselves under J+ : ˆ ˆ ˆ J+ cm ψm = J+ J− ψm+1 . which implies that n = 2j and therefore j can be an integer or a halfinteger. j − n is an eigenfunction ofJ 2 ˆ2 on ψj−n : with eigenvalue j(j + 1). J− . Finally we show that the 2j + 1 functions ψj .78 CHAPTER 4. . It is clear that the functions transˆ ˆ form among themselves under J− and Jz . Jz . . 4. We now operate J ˆ ˆ ˆ2 ˆ ˆ ˆ J 2 ψj−n = (J+ J− + Jz − Jz )ψj−n = [(j − n)2 − (j − n)]ψj−n = (j − n)(j − n − 1)ψj−n It follows that (j − n)(j − n − 1) = j(j + 1). . ( 4. We can continue the same procedure starting from ψj−1 so that it ˆ follows that ψm .
Using the fact that J− = J+ . ±2 . φ ) ˆ ˆ Geometrically. belong to the same class irrespective of the axes a and b. J+ ψm ) = (J+ ψm+1 . . ψm ) = c∗ m Thus cm 2 = (j + m + 1)(j − m). . φ).38) (4.e. which transform the (2j + 1) functions ψj .39) To summarise. . φ) does a rotation by φ about ˆ −1 (c. . with j = 0. φ ) rotates b back to a and the net result is the same b and ﬁnally R as a rotation by φ about a. This must be true because there is ˆ always a rotation R(c. ψm ) = (J− ψm+1 . ±1/2. ψm ) = (cm ψm . R(b. i.37) (4.. ±3/2.9. φ ) rotates a to b. . R(c. φ )R(b. φ)R(c. . ψ−j among themselves. we have † ˆ (ψm+1 . ˆ ˆ ˆ ˆ R(a. ψj−1 . 4. φ ) that brings the axis a to b or vice versa. CHARACTERS OF R3 ˆ ˆ2 ˆ = (J 2 − Jz + Jz )ψm+1 = [j(j + 1) − (m + 1)2 + (m + 1)]ψm+1 = (j + m + 1)(j − m)ψm+1 79 To calculate the normalisation constants cm we take the dot product † ˆ cm (ψm+1 .9 Characters of R3 ˆ ˆ Rotations by the same angle. ±1.4. There is an arbitrary phase factor in cm and the usual convention due to Condon and Shortley is to choose cm to be real and positive so that cm = ˆ J+ ψm = ˆ J− ψm = (j + m + 1)(j − m) (j + m + 1)(j − m)ψm+1 (j + m)(j − m + 1)ψm−1 (4. J+ ψm ) = (j+m+1)(j−m). φ) = R−1 (c. the irreducible representations of R3 are labelled by Rj . φ) and R(b. . R(a.
80
CHAPTER 4. LIE GROUPS
Since the characters of representations belonging to the same class are the same, it is suﬃcient to consider the characters of Rz . We consider the subgroup Rz consisting of rotations about the zaxis. This subgroup is Abelian and the irreducible representations are therefore one dimensional. We require R(z, φ)R(z, δφ) = R(z, φ + δφ). Expanding both sides in δφ we get R(z, φ)(1 + δφR (z, 0)) = R(z, φ) + δφR (z, φ) R (z, 0)R(z, φ) = R (z, φ) The solution is R(z, φ) = c exp(R (z, 0)φ). For a unitary representation, we may choose c = 1 and from the boundary condition R(z, φ) = R(z, φ+ 2π), we deduce that R (z, 0) = −im where m is a positive or negative integer. Thus the irreducible representations of Rz are
m Rz (φ) = exp(−imφ)
(4.40)
labelled by m = 0, ±1, ±2, . . . and the angle φ. These are also the eigenˆ functions of the inﬁnitesimal operator Jz with eigenvalues m, as may be easily veriﬁed. It is clear that representation Rj (a, φ) is reducible under the group Rz and by choosing the zaxis along a, it may be decomposed as
−j −j+1 j Rj (a, φ) = Rz (φ) ⊕ Rz (φ) ⊕ . . . ⊕ Rz (φ) m=j
=
m=−j
m ⊕Rz (φ)
Thus the character is given by
m=j
χj (φ) =
m=−j
exp(−imφ) sin (j + 1/2)φ sin φ/2 (4.41)
=
which is easily obtained from the geometric series formula Sn = k=n−1 ark = a(1 − rn )/(1 − r) with n = (2j + 1), a = exp(ijφ), k=0 and r = exp(iφ).
4.10. THE VECTORCOUPLING THEOREM
81
4.10
The VectorCoupling Theorem
The direct product of two irreducible representations of R3 referring to the same rotation φ about a given axis a is given by Rj1 (a, φ) ⊗ Rj2 (a, φ) =
j
⊕cj Rj (a, φ)
(4.42)
To ﬁgure out the coeﬃcient cj we use the fact that the character of a direct product representation is the product of the characters of the individual representations: χj1 (a, φ)χj2 (a, φ) =
j
⊕cj χj (a, φ)
The left hand side is given by χ χ
j1 j2 j1 j2
=
m1 =−j1 m2 =−j2
e−im1 φ e−im2 φ
= + + + =
[ei(j1 +j2 )φ + ei(j1 +j2 −1)φ + . . . + e−i(j1 +j2 )φ ] [ei(j1 +j2 −1)φ + ei(j1 +j2 −2)φ + . . . + e−i(j1 +j2 −1)φ ] ... [ei(j1 −j2 )φ + ei(j1 −j2 −1)φ + . . . + e−i(j1 −j2 )φ ] χj1 +j2 + χj1 +j2 −1 + . . . + χj1 −j2
assuming that j1 ≥ j2 without loss of generality. Therefore j1 −j2 becomes j1 − j2  in the general case. Comparison with the previous equation shows that cj = 1 for j = j1 + j2 , j1 + j2 − 1, . . . + j1 − j2  which yields the vectorcoupling theorem: Rj1 (a, φ) ⊗ Rj2 (a, φ) =
j1 +j2 j=j1 −j2 
⊕Rj (a, φ)
(4.43)
The rotation group is called simply reducible because mγ = 1.
82
CHAPTER 4. LIE GROUPS
4.11
ClebschGordan Coeﬃcients
j1 j2 As discussed in section 3.11, a product of two states ψm1 ψm2 transforming according to the product representation Rj1 ⊗j2 of the rotation group may be expanded as follows: j1 j2 ψ(m1 , m2 ) = ψm1 ψm2 = JM J∗ ΨJ CM (m1 , m2 ) M
where the C’s are the ClebschGordan coeﬃcients which form a matrix that brings the product representation Rj1 ⊗j2 into the block diagonal form (section 3.4, Theorem 3):
j1 Rm
1 m1
j2 Rm
2 m2
j1 +j2
=
J=j1 −j2  M M
J J J∗ CM (m1 , m2 )RM M CM (m1 , m2 )
(4.44)
This expansion is known as the ClebschGordan series. No additional label is needed in C because the rotation group is simply reducible. We have dropped the dependence on j1 and j2 for clarity. We may also write, using the unitarity of C ΨJ = M Thus
J CM (m1 , m2 ) = (ψ(m1 , m2 ), ΨJ ) M m1 ,m2 J CM (m1 , m2 )ψ(m1 , m2 )
(4.45)
(4.46)
Since the new functions ΨJ are orthonormal, we have M
m1 m2 J CM (m1 , m2 )2 = 1
(4.47)
J ˆ By applying Jz to both sides of Eq. ( 4.45), it follows that CM (m1 , m2 ) = 0 unless m = m1 + m2 and from the vectorcoupling theorem we also have j1 − j2  ≤ J ≤ j1 + j2 : J CM (m1 , m2 ) = 0 unless
M = m1 + m2 or j1 − j2  ≤ J ≤ j1 + j2
(4.48)
( 4.+m2 CM +1 (m1 . m2 ) Aj2 .−M ΨJ −1 M = n1 n2 J CM (n1 . m2 ) and setting M → M + 1 and M → M − 1 in the ﬁrst and second equation above we get the recursion relations: AJ. CLEBSCHGORDAN COEFFICIENTS With the Condon and Shortley phase convention j ˆ j J+ ψm = Aj.+m ψm+1 j ˆ j J− ψm = Aj.−m2 CM −1 (m1 .m = (j − m)(j + m + 1).+m1 CM +1 (m1 + 1.52) . we have dropped the superscript J since it is ﬁxed. n2 ) [ Aj1 . m2 − 1) (4.4.50) (4.11.−M CM (m1 .45): AJ.+M ΨJ +1 M = n1 n2 J CM (n1 .49) the ClebschGordan coeﬃcients become real. Aj. n2 − 1) ] AJ. J − j1 ) > 0 (4.−m1 CM −1 (m1 − 1. n2 ) + Aj2 .−n1 ψ(n1 − 1.51) To simplify the notation.−m ψm−1 83 where Aj. n2 ) [ Aj1 . n2 ) + Aj2 . m2 ) = + Aj1 . m2 + 1) Aj1 . There is an ambiguity in the sign and this is ﬁxed by choosing the convention that CJ (j1 .+n2 ψ(n1 . The ClebschGordan coeﬃcients for ﬁxed j1 and j2 and a chosen j can ˆ ˆ be calculated from recursion relations obtained by applying J− and J+ to both sides of Eq. m2 ) = + AJ.m>j = 0 (4. m2 ) Aj2 .+M CM (m1 .+n1 ψ(n1 + 1.j = Aj. n2 + 1) ] Taking the dot product with ψ(m1 .−n2 ψ(n1 .
m2 will annihilate all components except J CM (m1 .−M −1/2 CM −1 (M − 1/2) These are four linear equations with four unknowns but since the equations are homogeneous. ( 4.M CM (M − 1/2) = Aj1 . 1/2) = ± j1 ± M + 1/2 2j1 + 1 j1 M + 1/2 2j1 + 1 1 CM j ±1/2 (M + 1/2. the solutions can only be expressed as ratios. the additional condition in Eq.53) which when applied to ψm1 . Similarly we can generate the recursion relations for J = j2 + 1/2.m2 ) .M −1 CM −1 (M − 1/2) = Aj1 . LIE GROUPS As an example we evaluate the ClebschGordan coeﬃcients for a given j2 = 1/2 which gives possible values of J = j1 − 1/2. With m2 = ±1/2 we get from the recursion relations AJ. m2 )ΨJ M J J ˆ ⇒ CM (m1 .−1/2 CM +1 (M + 1/2) AJ.M −1/2 CM +1 (M + 1/2) AJ. The solutions in complete notation are 1 CM j ±1/2 (M − 1/2.−M −1/2 CM (M − 1/2) + Aj2 .52 ﬁxes the sign. However. j1 + 1/2.84 CHAPTER 4.−M CM (M + 1/2) = Aj1 . We consider the case J = j1 − 1/2.m2 .M −1/2 CM (M + 1/2) + Aj2 . −1/2) = The ClebschGordan coeﬃcients can be also calculated by using the projection operator: ˆ PJ = K=J ˆJ PK = K=J ˆ J 2 − K(K + 1) J(J + 1) − K(K + 1) (4.−M −1 CM +1 (M + 1/2) = Aj1 . To simplify the notation further. m2 ) CM (m1 .m2 = CM (m1 .−1/2 CM (M − 1/2) AJ. P J ψm1 .47) determines the absolute magnitudes and condition 4. m2 ) = (ψm1 . m2 )ΨJ where M = m1 + m2 : M J ˆ P J ψm1 . we drop the second argument in C since it can be obtained by substracting the ﬁrst argument from the subscript.
m2 − 1) +Aj2 . −1/2) = (j1 + M + 1/2)(j1 − M + 1/2) −(2j1 + 1) .m2 Aj2 . −1/2) = because it must be positive when M = j1 − 1/2.m2 .m2 = [J(J + 1) − K(K + 1)]−1 × [ { Aj1 . We also have 1 CM j −1/2 1 (M − 1/2. m2 ) + + (j1 − m1 )(j1 + m1 + 1)(j2 + m2 )(j2 − m2 + 1) ×ψ(m1 + 1. m2 = −1/2 ⇒ m1 = M + 1/2.−m1 −1 + Aj2 . m2 − 1) (j2 − m2 )(j2 + m2 + 1)(j1 + m1 )(j1 − m1 + 1) ×ψ(m1 − 1. m2 ) +Aj1 . CLEBSCHGORDAN COEFFICIENTS ˆJ ˆ ˆ ˆ ˆ ˆ We evaluate PK ψm1 . ˆ ˆ We have P J = PjJ +1/2 since there are only two possible values of 1 K = j1 ± 1/2.m1 Aj1 .m1 Aj2 .−m2 ψ(m1 + 1. 1 CM j −1/2 (M + 1/2. −1/2)2 ˆ j −1/2 = (ψ(M + 1/2. Pj11+1/2 ψ(M + 1/2. −1/2). −1/2)) = j1 + M + 1/2 2j1 + 1 j1 + M + 1/2 2j1 + 1 j −1/2 Using the sign convention in ( 4.−m2 −1 +M (M + 1) − K(K + 1) } ψ(m1 . m2 + 1) ] We evaluate the previous example using the projection operator method. Consider the case J = j1 − 1/2.52) we get 1 CM j −1/2 (M + 1/2.m2 Aj1 . 1/2) CM (M + 1/2. Thus.11.4. using the identity J 2 = J− J+ + Jz (Jz + 1): ˆJ PK ψm1 .−m1 ψ(m1 − 1. m2 + 1) ] 85 = [J(J + 1) − K(K + 1)]−1 × [ { (j1 − m1 )(j1 + m1 + 1) + (j2 − m2 )(j2 + m2 + 1) +M (M + 1) − K(K + 1) } ψ(m1 .
ViewegVerlag. 62. Racah. LIE GROUPS j −1/2 (M − 1/2. The general formula for the ClebschGordan coeﬃcients has been evaluated by Wigner (Gruppentheorie. m2 ) = δM. 1931 or in the translated version.m1 +m2 2J + 1 × × × ν (j1 + j2 − J)!(J + j1 − j2 )!(J + j2 − j1 )! (j1 + j2 + J + 1)! (j1 + m1 )!(j1 − m1 )!(j2 + m2 )!(j2 − m2 )!(J + M )!(J − M )! (−1)ν ν! 1 (j1 + j2 − J − ν)!(j1 − m1 − ν)!(j2 + m2 − ν)! 1 × (J − j2 + m1 + ν)!(J − j1 − m2 + ν)! (4. 1/2 = − j1 − M + 1/2 2j1 + 1 in agreement with the previous results. . Phys.m1 +m2 2J + 1 × × ν (J + j1 − j2 )!(J − j1 + j2 )!(j1 + j2 − J)!(J + M )!(J − M )! (j1 + j2 + J + 1)!(j1 − m1 )!(j1 + m1 )!(j2 − m2 )!(j2 + m2 )! (j2 + J + m1 − ν)!(j1 − m1 + ν)! (−1)ν+j2 +m2 ν! (J − j1 + j2 − ν)!(J + M − ν)!(ν + j1 − j2 − M )! (4. Wigner’s closed expression is √ J CM (m1 .86 so that 1 CM CHAPTER 4. Group Theory. Rev. New York. m2 ) = δM. Brunswick. Academic Press.54) Racah (G.55) The most frequently used ClebschGordan coeﬃcients are tabulated in many books on atomic physics. 1959). 438 (1942)) derived a more symmetrical form: √ J CM (m1 .
z). z)) = ψ(x + γy.56) (4. y − γx. To generate the irreducible subspace.and yaxes.12. y. The procedure for constructing the irreducible representations of R3 ˆ ˆ tells us that we should start with an eigenfunction of Jz and J 2 and then we apply the ladder operators repeatedly in order to span the irreducible ˆ subspace corresponding to a set of degenerate eigenvectors of J 2 . we need to know the form of the inﬁnitesimal operators in function space. SPHERICAL HARMONICS 87 4. To do this. and study its eﬀect on an arbitrary function ψ(x.58) These are identical to the angular momentum operators in quantum mechanics. First we consider a small rotation about the zaxis: ˆ T (R(γz))ψ(x.12 Spherical Harmonics We wish to ﬁnd explicit basis functions for the irreducible representations of the rotation group. The inﬁnitesimal operators in function space are therefore ∂ ∂ ˆ Jz = i y −x ∂x ∂y ∂ ∂ ˆ Jy = i x − z ∂z ∂x ∂ ∂ ˆ Jx = i z −y ∂y ∂z (4. The simplest function. We form an operator representation of the ˆ group elements. R → T (R). Furthermore.4. y. z) ∂ ∂ = 1+γ y −x ∂x ∂y ψ(x. ˆ ˆ J− = (Jx − iJy )(x − iy) = 0 which implies that (x − iy) has the smallest value of m among the degenˆ erate eigenfunctions of J 2 . z) The same procedure can be applied to the x. y. z) = ψ(R−1 (γz)(x. that one can think of is x − iy which ˆ is an eigenfunction of Jz with eigenvalue −1 and it is also an eigenfunction ˆ of J 2 .57) (4. we apply . other than a constant. y.
Higher l spherical harmonics may be generated in the same fashion by applying the operator ˆ J+ on (x − iy)l repeatedly and normalising the resulting functions as just mentioned above. we then have z = cos θ. The spherical harmonics for each l form an irreducible subspace under rotation and therefore provide an irreducible representation of the rotation group R3 . the spherical harmonics are deﬁned such that l Y−l = (2l)! 2l l! 2l + 1 (x − iy)l 4π (4. Since r2 = x2 + y 2 + z 2 is invariant under rotation. 1 respectively. Two spherical harmonics with diﬀerent l values will never transform to each other under rotation because they belong to two . The normalisation constants are determined by integrating the square of each function over the solid angle and requiring the integral to be unity. As a convention. the spherical harmonics transform among themselves with the same l. and x = sin θ cos φ. and −(x+iy) thus form an irreducible subspace with eigenvalues ˆ of Jz equal to m = −1. In polar coordinates. LIE GROUPS ˆ the operator J+ repeatedly on (x − iy) until it vanishes: ˆ J+ (x − iy) = 2z ˆ J+ z = −(x + iy) ˆ J+ (x + iy) = 0 (x−iy). These functions are proportional to the spherical harmonics for l = 1.88 CHAPTER 4. 0. we may choose r = 1. as may be easily veriﬁed.59) which ﬁxes the sign. so that Y11 Y01 = − = 3 8π 3 8π (x + iy) = − 3 4π 3 8π 3 4π 3 8π sin θ eiφ cos θ z = = 1 Y−1 = (x − iy) sin θ e−iφ l which are the spherical harmonics Ym for l = 1. y = sin θ sin φ. Under rotation. z. We immediately recognise that ˆ this set corresponds to the j = 1 representation with eigenvalue of J 2 equal to j(j + 1) = 2. There is an ambiguity in the sign.
φ) = (θz . φ)Rm m (4. φ ) = l 4π l∗ Y l (θ.e.60) (4. θ. φ) into the irreducible subspaces formed by the spherical harmonics as follows: f (r.61) Y0l (θ. φz ) on the rotated z axis corresponding to θ = 0 we get l Rm0 = l∗ l Rmm Ym (θ .12.62) for m = 0 that Y0l (θ . φ) = The last equation may be thought of as the deﬁnition of the Legendre polynomials. φ) and (θ .63) 4π Y l∗ (θz . φ ) = m l l Ym (θ. φ) = Ym (θ .4. φ) = lm l flm (r)Ym (θ.64) It follows immediately by substituting this into Eq. φz ) 2l + 1 m=−l m (4. φ) From the deﬁnition of the spherical harmonics it is easy to show that: l Ym (0. θ.0 4π 2l + 1 Pl (cos θ) 4π (4. We take the point of view that (θ. φ) = l m =−l l where Rm m is the irreducible representation of a rotation R in the spherical harmonics basis. it is the coordinate system which is rotated. φ ) correspond to the same physical point. ( 4. φ) = 2l + 1 δm. SPHERICAL HARMONICS 89 distinct irreducibe subspaces of the rotation group. Considering a point (θ. φ ) (4. φz ) 2l + 1 m (4. φ)Ym (θz . This means that we can decompose an arbitrary function f (r. We can now derive some useful identities but ﬁrst we write down the transformation of spherical harmonics under an arbitrary rotation R in the following way: l l ˆ T (R)Ym (θ.62) or l Ym (θ. i.65) .
LIE GROUPS By Eq. 0)CM (m1 . A product of two spherical harmonics transforms according to the direct product representation which is in general reducible.68) .64) in Eq. m2 ) 4π(2L + 1) (4. φ) (4. φ)Ym (θ. φz )Ym (θ. φ) is arbitrary. φ)Ym22 (θ. Using Eq. φ) = l1 +l2 L=l1 −l2  (2l1 + 1)(2l2 + 1) 4π(2L + 1) L L L × C0 (0.90 CHAPTER 4. φ) 2l + 1 m=−l (4.66) For the special case of θ = θz and φ = φz we have θ = 0 so that l m=−l l∗ l Ym (θ.44) we obtain l l Ym11 (θ. φ) = 2l + 1 4π Thus the quantity on the left hand side is invariant under rotation since the (θ. ( 4. ( 4.61) we get the addition theorem Pl (cos θ ) = l 4π l∗ l Ym (θz . ( 4. 0)CM (m1 .67) where M = m1 +m2 . From this we can evaluate the frequently encountered integral of three spherical harmonics: L∗ l l dΩ YM Ym11 Ym22 = (2l1 + 1)(2l2 + 1) L L C0 (0. m2 )YM ((θ.
The reason for the ˆ ˆ above division of H is that H0 is a singleparticle operator which can be ˆ solved easily.3) (5. H2 is usually small and can also ˆ ˆ be treated as a perturbation and we assume that H2 << H1 .Chapter 5 Atomic Physics ˆ The Hamiltonian H of an atom in the absence of external ﬁelds is given by ˆ H= i [ti − Z/ri ] + 1/2 i=j 1/rij + i ξ(ri )li · si (5.2) ˆ h0 (ri ) = i [ti − Z/ri + Vef f (ri )] Vef f (ri ) i (5. The objective of the present chapter is to set up the Hamiltonian matrix and exploit the symmetry properties of the Hamiltonian by means of group theory in order to simplify our task.1) ˆ It is useful to break up H as follows: ˆ ˆ ˆ ˆ H = H0 + H1 + H2 where ˆ H0 = i (5.5) ˆ H1 = 1/2 i=j 1/rij − ˆ H2 = i ξ(ri )li · si Vef f is a spherically symmetric singleparticle potential. By choosing Vef f appropriately. We consider the terms in the ˆ Hamiltonian one by one starting with H0 . 91 . H1 may be made small so ˆ that it can be treated as a perturbation.4) (5.
. with the coordinates permuted: ψ1 (1) ψ2 (1) ψ1 (2) ψ2 (2) 1 . From now on we assume that ˆ basis functions needed to set up the secular equation for H must be Slater determinants. these are the only ones found in nature (Pauli principle). . .. ψ1 (Z) ψ2 (Z) .. Since H0 is a sum of singleparticle operators.92 CHAPTER 5. . This is simply the ˆ Pauli exclusion principle. 2) = √ [ψ1 (1)ψ2 (2) − ψ1 (2)ψ2 (1)] 2 which changes sign when the coordinates or the quantum labels 1 and 2 are interchanged Moreover. ... Z) = √ . . which is a general property of a determinant that it is zero whenever two rows or columns are the same. its solution is just a single Slater determinant. .. . . ψZ (2) .. ψZ (Z) (5.. there is always a onedimensional irreducible representation whose basis functions are totally antisymmetric.. . h0 ] = 0 . where h0 ψi = i ψi . ψZ (1) ... .. Z ... .. It does not follow from group theory that the eigenfunctions of the Hamiltonian should be totally antisymmetric but for fermions.6) We have labeled the coordinates ri by i for simplicity.. Ψ(1..... . .1 ˆ H0 ˆ H0 has two kinds of symmetry. For a permutation group of N elements. The Slater determinant is 1 Ψ12 (1. It is invariant under permutations of the electron coordinates (indistinguishability) and under separate rotations of ˆ the electron coordinates.. . . Take for example the simplest nontrivial case of two electrons. . The eigenfunctions of H0 must therefore form basis functions for the irreducible representations of the permutation and rotation groups.. . 2. meaning that the functions change sign whenever two coordinates are interchanged. .. These antisymmetric functions can be written most conveniently as Slater deˆ terminants which are linear combinations of i ψi (ri ). ATOMIC PHYSICS 5.. We now consider the rotational symmetry which implies that ˆ ˆ [T (R). it becomes zero when the coordinates or the quantum labels 1 = 2..
Ly ] = [h0 . φ)χms (σ) ˆ To summarise. Each energy level of h0 is (2l + 1) degenerate since the irreducible representations of R3 are labelled by Rl with dimensions (2l+1).ˆ ˆ 5. sponding to h We now take into account spin and label the spin states by χms (σ) where the spin coordinate σ can only take two discrete values σ = ±1: χ+ (1) = 1. χ− (−1) = 1. Lz ] = [h0 .2 ˆ ˆ H0 + H1 ˆ ˆ We now consider (H0 +H1 ) which is invariant under simultaneous rotations of the position coordinates of the electrons. H0 + H1 It follows that ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ [h0 . Operating h0 on ψnlm shows that unl (r) must be a solution to the radial Schr¨dinger equation correo ˆ 0 . 5. The eigenstate of h0 including spin is l ψnlmms (r. Lx ] = [h0 . In determining ˆ 0 + H1 ) we will use these eigenfunctions as basis ˆ the eigenstates of (H functions. It is also implicitly invariant ˆ under simultaneous rotations in the spin coordinates. L2 ] = 0 93 ˆ and the eigenstates of h0 may be chosen to be simultaneous eigenstates 2 ˆ ˆ ˆ of L and Lz . χ+ (−1) = 0. φ) l ˆ ˆ ˆ where Ym is an eigenfunction of Lz and L2 . the eigenfunctions of H0 are simply single Slater deterˆ minants with the orbitals given by the eigenfunctions of h0 . the eigenstates must have the following form: l ψnlm (r) = unl (r)Ym (θ. Although (H0 + .2. σ) = unl (r)Ym (θ. which is a familiar result. χ− (1) = 0. Neglecting spin. Physical observation shows that the spin behaves like angular momentum so that 1/2 ˆ T (R)χms (σ) = χms (σ)Tms ms (R) ms The spin states form a basis for the two dimensional irreducible representaˆ tion of R3 corresponding to j = 1/2.
. Lz = i ˆ Lz (i) ˆ where Lz (i) acts on the coordinate ri . the representation for simultaneous rotations of the electron coordinates is given by a direct product of each rotation. Lz . Thus the superscript indicates the number of electrons. p. ˆ ˆ the energy levels of (H0 + H1 ) can be labelled with L and S and each energy level is (2L + 1)(2S + 1) degenerate. 5. 1. The eigenfunctions of H0 are still eigenfunctions of Lz and Sz 2 2 ˆ ˆ but not of L or S . . . (H0 + H1 ) no longer commutes with the angular momentum or spin operators of each ˆ ˆ ˆ electron. d. (H0 + H1 ) commutes with Lx . ATOMIC PHYSICS ˆ H1 ) has no explicit spin dependence. 2. We deﬁne a conﬁguration by a symbol such as s1 p1 meaning that there is one electron in an sshell and one electron in a pshell. Similarly for the inﬁnitesimal opˆ ˆ ˆ erators in the spin space Sx . corresponding to l = 0. Ly . As before we may choose the eigenfunctions of (H0 + H1 ) to 2 2 ˆ ˆ ˆ ˆ ˆ ˆ be simultaneous eigenfunctions of Lz and L and of Sz and S . the direct ˆ ˆ product representation is in general reducible. Sy . When there are more than one electron.2. The inﬁnitesimal operators for the simultaneous rotations in space or spin coordinates are given by the sum of the inﬁnitesimal operators corresponding to each electron coordinate. We use the spectroscopic notations s. According to the vectorcoupling theorem. Thus ˆ Lx = i ˆ ˆ Lx (i). . Ly = i ˆ ˆ Ly (i). Since (H0 + H1 ) commute with rotation operators in space and spin coordinates separately. it is assumed that each electron carries a spin and the spin dependence enters implicitly through the Pauli ˆ ˆ principle. These six inﬁnitesimal operators L S correspond to a direct product R3 ⊗R3 since the orbital and spin operators ˆ ˆ ˆ ˆ ˆ ˆ commute. we need only consider the valence electrons because the electrons forming a closed shell act eﬀectively like a spherically symmetric potential which can be combined with Vef f . 3.1 Conﬁguration and Term In atoms.94 CHAPTER 5. f. The degenerate eigenfunctions of (H0 + H1 ) must therefore form an invariant subspace under simultaneous rotations of the space or spin coordinates. . The corresponding states . respectively. and Sz . and L2 and with the spin ˆ ˆ counterpart. .
1 P. To illustrate the concept of term. ml ˆ ˆ and ms are respectively the eigenvalues of Lz and Sz of the individual electrons. we know that we can always generate an invariant subspace by applying the symmetry operators repeatedly on an arbitrary function in that subspace. However. 1 S. According to the vectorcoupling theorem. the eigenfunctions of H0 are still eigenfunctions ˆ ˆ ˆ ˆ of Lz and Sz but not of L2 or S 2 . Let us list the possible Slater determinants corresponding to this conﬁguration starting with the largest possible value of ML = i ml (i). Consider a conﬁguration p2 and let us use the notation  + 1+  − 1− to represent a Slater determinant where one electron has ml = +1 and ms = +1/2 and the other electron has ml = −1 and ms = −1/2. 3 P. . We note also that the two electrons are equivalent because they occupy the same p shell. Some of these terms may not be allowed by the Pauli exclusion principle when the two electrons are equivalent as we shall see below. 5. . the irreducible representations for simultaneous rotations of the two electron spin coordinates correspond to S = 1. 0 since l1 = l2 = 1. Let us illustrate this method by considering a speciﬁc example since the generalisation to an arbitrary case is quite clear. l1 −l2  = 2. 1 D.2 Ladder Operator Method ˆ As just discussed above. the irreducible representations for simultaneous rotations of the two electron space coordinates correspond to L = l1 +l2 .2. 0 since s1 = s2 = 1/2.ˆ ˆ 5. . l1 +l2 −1. Similarly. 1. consider two p electrons. There are (2 × 3)!/2!(2 × 3 − 2)! = 15 possible ways of placing two identical electrons in a pshell: .2. If we know one of the ˆ ˆ eigenfunctions of L2 and S 2 then we can construct the other degenerate eigenfunctions by using the ladder operators. . H0 + H1 95 are conventionally written as 2S+1 L and they are called ”term” (multiplet). 3 S. The possible terms are then given by 3 D.
g. ATOMIC PHYSICS Ψ1 = 1 1 Ψ2 = 1+ 0+ Ψ3 = 1+ 0− Ψ4 = 1− 0+ Ψ5 = 1− 0− Ψ6 = 1+  − 1+ Ψ7 = 1+  − 1− Ψ8 = 1−  − 1+ Ψ9 = 1−  − 1− Ψ10 = 0+ 0− Ψ11 = 0+  − 1+ Ψ12 = 0+  − 1− Ψ13 = 0−  − 1+ Ψ14 = 0−  − 1− Ψ15 =  − 1+  − 1− + − We keep in mind that Ψi is a Slater determinant so that e. Ψ1 has the largest value of ML and it is the only one. We check that it is ˆ ˆ an eigenfunction of L2 and S 2 : ˆ ˆ ˆ L+ Ψ1 = [L+ (1) + L+ (2)] + 1+  + 1− = 0 ˆ ˆ ˆ Lz Ψ1 = (Lz (1) + Lz (2)) + 1+  + 1− = (1 + 1) + 1+ + 1− = 2Ψ1 ˆ 2 Ψ1 = 4Ψ1 Lz ˆ L2 Ψ1 = 6Ψ1 ˆ ˆ ˆ S+ Ψ1 = [S+ (1) + S+ (2)] + 1+  + 1− = 0 ˆ ˆ ˆ Sz Ψ1 = [Sz (1) + Sz (2)] + 1+  + 1− = 0 2 ˆ S Ψ1 = 0 .96 ML MS 2 0 1 1 1 0 1 0 1 −1 0 1 0 0 0 0 0 −1 0 0 −1 1 −1 0 −1 0 −1 −1 −2 0 CHAPTER 5. 1+ 1− = −1− 1+ . Moreover. it has MS = 0 and therefore it must be an eigenfunction belonging to singlet 1 D. It is natural when using the ladder operator method to start from a Slater determinant with the largest values of ML and/or MS since ˆ ˆ it is more likely that it is an eigenstate of L2 and/or S 2 .
We operate ˆ S− on Ψ2 to obtain the triplets with ML = 1. and Ψ15 ˆ ˆ are degenerate eigenfunctions of L2 and S 2 corresponding to the term 1 D. ˆ S− Ψ2 = = = ˆ S− (Ψ4 + Ψ3 ) = ˆ S− Ψ5 = ˆ S− 1+ 0+ 1− 0+ + 1+ 0− Ψ4 + Ψ3 2Ψ5 0 ˆ Operating L+ on Ψ2 yields a state  + 1+  + 1+ which is forbidden by the Pauli exclusion principle. This state would be allowed if the two ˆ electrons occupied diﬀerent p shells. We operate L− on Ψ2 and then use ˆ S− to generate the triplets with ML = 0: √ ˆ 2Ψ6 L− Ψ2 = ˆ− Ψ6 = Ψ8 + Ψ7 S ˆ S− (Ψ8 + Ψ7 ) = 2Ψ9 ˆ S− Ψ9 = 0 . Ψ2 must be an eigenfunction belonging to 3 P because it is the only one with ML = 1 and MS = 1 and the term 3 D does not exist. (Ψ12 − Ψ13 ). (Ψ7 − Ψ8 + 2Ψ10 ).2. H0 + H1 97 We generate the invariant subspace formed by the degenerate eigenfuncˆ tions of L2 corresponding to the term 1 D: ˆ ˆ ˆ L− Ψ1 = [L− (1) + L− (2)] + 1+  + 1− = (1 + 1)(1 − 1 + 1)0+  + 1− + (1 + 1)(1 − 1 + 1) + 1+ 0− √ = 2(Ψ3 − Ψ4 ) We continue applying the lowering operator: √ ˆ L− (Ψ3 − Ψ4 ) = 2(2Ψ10 + Ψ7 − Ψ8 ) √ ˆ − (2Ψ10 + Ψ7 − Ψ8 ) = 3 2(Ψ12 − Ψ13 ) L √ ˆ L− (Ψ12 − Ψ13 ) = 2 2Ψ15 ˆ L− Ψ15 = 0 The functions Ψ1 .ˆ ˆ 5. (Ψ3 − Ψ4 ).
Ψ10 . where the eigenfunctions have been normalised: Term ML MS 1 D 2 0 1 0 0 0 1 0 2 0 3 P 1 1 0 1 0 1 0 1 1 1 0 1 1 S 0 0 Eigenfunction Ψ1 √ (Ψ3 − Ψ4 )/ 2 √ (Ψ7 − Ψ8 + 2Ψ10 )/ 6 √ (Ψ12 − Ψ13 )/ 2 Ψ15 Ψ2 √ (Ψ3 + Ψ4 )/ 2 Ψ5 Ψ6 √ (Ψ7 + Ψ8 )/ 2 Ψ9 Ψ11 √ (Ψ12 + Ψ13 )/ 2 Ψ14 √ (Ψ7 − Ψ8 − Ψ10 )/ 3 ˆ The method that we have used to construct the eigenfunctions of L2 ˆ ˆ ˆ and S 2 and hence of (H0 + H1 ) is systematic but it becomes tedious as . Below we summarise our results. with ML = 0 and MS = 0. The eigenfunction belonging to 1 S must be Ψ7 −Ψ8 −Ψ10 since it is orthogonal to the previous two eigenfunctions. We have already found two eigenfunctions Ψ7 −Ψ8 + 2Ψ10 and Ψ7 + Ψ8 belonging to 1 D and 3 P respectively. There are three Slater determinants. ATOMIC PHYSICS Continuing along the same fashion we generate the triplets with ML = −1 √ ˆ L− Ψ6 = 2Ψ11 ˆ S− Ψ11 = Ψ13 + Ψ12 ˆ S− (Ψ13 + Ψ12 ) = 2Ψ14 ˆ S− Ψ14 = 0 We have generated 14 eigenfunctions.98 CHAPTER 5. Ψ8 . Ψ7 . The only one left must belong to the term 1 S.
3 P .3 Projection Operator Method Let us illustrate the method by using the example we have just considered. From a given Slater determinant.ˆ ˆ 5. . We ﬁrst project out the component of 3 P out of Ψ7 and then we project out the comoponent of 1 S: ˆ L2 − 1(1 + 1) Ψ7 = 2Ψ10 ˆ L2 − 0(0 + 1) Ψ10 = 2Ψ10 + Ψ7 − Ψ8 This is proportional to the eigenfunction of 1 D with ML = 0 and MS = 0 which is in agreement with the result obtained by using the ladder operator method. We discuss below a diﬀerent method which is more suitable for the latter case. ˆ We can arrive at the same result if we ﬁrst ﬁnd the eigenfunction of S 2 with S = 0 by projecting out the component with S = 1: ˆ [S 2 − 1(1 + 1)]Ψ7 = Ψ7 + Ψ8 − 2Ψ7 = Ψ8 − Ψ7 ˆ Thus (Ψ8 −Ψ7 ) is an eigenfunction of S 2 corresponding to S = 0 (singlet). Now we project out the component of 1 S. . and 1 S. 5. ˆ [L2 − 0(0 + 1)](Ψ8 − Ψ7 ) = 2(−Ψ10 + Ψ8 ) − 2(Ψ10 + Ψ7 ) = −2(Ψ7 − Ψ8 + 2Ψ10 ) which is proportional to the previous result. It is not necessary to project out the component of 3 P because (Ψ8 − Ψ7 ) is a singlet and therefore it cannot contain eigenfunctions of 3 P . must be a linear combination of eigenfunctions belonging to 1 D. Let us take Ψ7 and suppose that we are interested in ﬁnding the component of 1 D. . Ψ2 .2. the projection operator method allows us to arrive at a particular eigenfunction of a term by projecting out all the other terms one by one. It is best to start with a Slater determinant with ML = 0 and MS = 0 since it is contained in all the terms.2. Each of the Slater determinant Ψ1 . . H0 + H1 99 the number of Slater determinants increases.
4 Term Energies In this section we discuss the calculations of term energies. cj ]+ = 0. we need matrix elements of the type ˆ ˆ Ψi H0 + H1 Ψj It is convenient to calculate them in the occupation number representation ˆ ˆ in which (H0 + H1 ) is given by ˆ ˆ H0 + H1 = ij ci ih0 j aj + ˆ ˆ ˆ 1 ci cj ijˆkl al ak ˆˆ v ˆˆ 2 ijkl (5. .8) We write the Slater determinants Ψi in the occupation number representation and order the singleparticle states ψi as follows 1+ . To calculate the term energies. aj ]+ = 0. −1+ . 0− . aj ]+ = δij c ˆ a ˆ c ˆ (5. −1− = c1+ c1− c0+ c0− c−1+ c−1− 0 ˆ ˆ ˆ ˆ ˆ ˆ where 0 is a state with no electrons in the p shell. Thus we have Ψ1 = c1+ c1− 0 ˆ ˆ Ψ2 = c1+ c0+ 0 ˆ ˆ . 1 ≡ r1 = δ(si . We note that for fermions the ordering of the states is very important.2. 0+ . ATOMIC PHYSICS 5. [ˆi . Ψ15 = c−1+ c−1− 0 ˆ ˆ . sk )δ(sj .100 CHAPTER 5. sj ) ∗ d1 ψi (1)h0 (1)ψj (1). sl ) d1d2 ∗ ∗ ψi (1)ψj (2)ψk (1)ψl (2) 1 − 2 ci = a† and ai = c† are creation and annihilation operators respectively ˆ ˆi ˆ ˆi which obey the commutation relations [ˆi . [ˆi . 1− .7) where ˆ ih0 j ijˆkl v = δ(si .
2. ˆ The matrix elements of H0 between Ψ = cp cq 0 and Ψ = cr cs 0 are ˆˆ ˆˆ quite simple: ˆ ΨH0 Ψ = ij ˆ ih0 j 0ˆq ap ci aj cr cs 0 a ˆ ˆˆ ˆ ˆ ˆ ˆ = [ ph0 p + qh0 q ](δpr δqs − δps δqr ) ˆ We have made use of the fact that ψi are eigenfunctions of h0 so that ˆ ˆ ih0 j = δij .9) l+1 Y (Ω1 )Ym (Ω2 ) r1 − r2  2l + 1 r> m lm where r> and r< are the greater and the smaller of r1 and r2 . . ( 4.68) and the ClebschGordan coeﬃcients are tabulated extensively in many books on atomic physics. ψ1 ψ2 ˆψ3 ψ4 v = × × d3 r1 d3 r2 lm l 4π r< l+1 φl1 (r1 )φl2 (r2 )φl3 (r1 )φl4 (r2 ) 2l + 1 r> l l l∗ dΩ1 Ym (Ω1 )Ym11∗ (Ω1 )Ym33 (Ω1 ) l l l dΩ2 Ym (Ω2 )Ym22∗ (Ω2 )Ym44 (Ω2 ) The integrals over three spherical harmonics are given by Eq. we use the multipole expansion of the Coulomb potential: l 4π r< l∗ 1 l = (5. Therefore H0 only contributes ˆ functions of H the same constant term to all the terms. To do this. H0 is diagonal because the Slater determinants are eigenˆ 0 and they are degenerate.ˆ ˆ 5. H0 + H1 101 ˆ The matrix elements of H1 between Ψ = cp cq 0 and Ψ = cr cs 0 will ˆˆ ˆˆ be of the following form: ˆ ΨH1 Ψ = = 1 ijˆkl 0ˆq ap ci cj al ak cr cs 0 v a ˆ ˆˆ ˆ ˆ ˆ ˆ 2 ijkl pqˆrs − pqˆsr v v The ﬁrst term is the direct term and the second term is the so called exchange term. We recall l that ψi = φl (r)Ym (Ω) where φl (r) is the solution to the radial Schr¨dinger o equation. We must also evaluate the integral pqˆrs before we can calculate the v term energies.
3 ˆ ˆ ˆ ˆ H = H0 + H1 + H2 ˆ Finally we include the spinorbit term H2 which is invariant under simultaneous rotations in space and spin coordinates. . L + S − 1. The J dependence of the splitting for given L. We assume ˆ that H2 is small enough so that it causes no mixing between the diﬀerent ˆ ˆ terms. S may be evaluated by using the WignerEckart theorem. This means that the (2L + 1)(2S + 1) degeneracy of each term will be in ˆ general split and each energy level of H is now (2J + 1) degenerate with J = L + S. J 2 ] = 0 where ˆ Jx = i ˆ ˆ [Lx (i) + Sx (i)] etc. . . Let ψJM be an eigenstate of J 2 which is also an eigenstate of L2 ˆ2 since it is formed as a linear combination of the eigenstates of a and S given term. We have ∆J = (ψJM . L − S which follows from the vectorcoupling J rule RL ⊗ RS = L+S J=L−S R . This implies ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ [H. L · S ψJM ) 1 = A[J(J + 1) − L(L + 1) − S(S + 1)] 2 . Jx ] = [H. i ξ(ri )li · si ψJM ) = A(ψJM . Jz ] = [H. ˆ ˆ ˆ H does not commute with Li or Si and the group is now reduced to R3 . ATOMIC PHYSICS Let us calculate the energy of the term 1 D. Jy ] = [H. Since all the eigenfunctions of D are degenerate we can take the state Ψ1 : ˆ Ψ1 H1 Ψ1 = pqˆrs − pqˆsr v v 1 = × × d3 r1 d3 r2 lm l 4π r< 2 2 l+1 φ (r1 )φ1 (r2 ) 2l + 1 r> 1 l∗ dΩ1 Ym (Ω1 )Y11∗ (Ω1 )Y11 (Ω1 ) l dΩ2 Ym (Ω2 )Y11∗ (Ω2 )Y11 (Ω2 ) The second term (exchange term) vanishes because of the orthogonality of the spin states. 5. .102 CHAPTER 5.
. The actual magnitude of the splitting depends on the details of the wavefunctions which are contained in A. m = 1 in the WignerEckart theorem.3. H = H0 + H1 + H2 103 ˆ ˆ ˆ The second step follows from the fact that L · S = 1/2(J 2 − L2 − S 2 ) transforms exactly like i ξ(ri )li ·si and the fact that R3 is simply reducible i.e.ˆ ˆ ˆ ˆ 5.
104
CHAPTER 5. ATOMIC PHYSICS
Chapter 6 The Group SU2: Isospin
It is found experimentally that the strong nuclear force that binds nucleons (protons and neutrons) is the same within 1 % between protonproton, protonneutron and neutronneutron. We may regard a proton and a neutron as two diﬀerent states of a nucleon. The Hamiltonian describing the nucleus is invariant under the interchange of proton and neutron states. This symmetry is called isospin. The group corresponding to isospin is called SU2 , which is homomorphic with the rotation group R3 and the mathematical description is identical to that of the spin 1/2. This means that the results we have derived for the group R3 may be readily used here.
6.1
The Group SU2
A 2 × 2 unitary matrix has four independent parameters and it may be written as α β U = exp(iθ) γ δ Since U U † = 1, we have mod(det U ) = 1 and we may choose αδ−βγ = 1. A little algebra shows that δ = α∗ and γ = −β ∗ so that U = exp(iθ) 105 α −β ∗ β α∗
106
CHAPTER 6. THE GROUP SU2 : ISOSPIN
The set of matrices U with θ = 0 and determinant = 1 form a subgroup of U2 , called SU2 (special unitary). Thus, SU2 is a group with elements U= α −β ∗ β α∗ (6.1)
and the condition α2 + β2 = 1. We should note that these matrices are themselves the elements of the group SU2 and they may be regarded as representations of rotation operators in a twodimensional Hilbert space. Since U are unitary, the corresponding inﬁnitesimal matrices must be skew Hermitian which can be made Hermitian by including the factor i: 1= 1 0 0 1 σx = 0 1 1 0 σy = 0 −i i 0 σz = 1 0 0 −1
Since the determinant of SU2 is +1, its inﬁnitesimal matrices must have zero trace and the unit matrix is therefore excluded. The remaining three matrices are just the Pauli spin matrices which are also the inﬁnitesimal matrices of the rotation group R3 corresponding to j = 1/2. However, the relationship between the two group elements of SU2 and R3 is not necessarily onetoone (isomorphic). Physically, we may generate the group SU2 by considering two nucleon states, p = c+ 0 , n = c+ 0 p n corresponding to proton and neutron. The bilinear products of the annihilation and creation operators that conserve particle number are c+ cp , c+ cn , c+ cn , c+ cp p n p n We deﬁne a new set of operators ˆ B = c+ cp + c+ cn p n + τ+ = cp cn ˆ τ− = c+ cp ˆ n 1 + ˆ 1ˆ (cp cp − c+ cn ) = Q − B τz = ˆ n 2 2 which have the following Lie algebra ˆ ˆ ˆ ˆ ˆ ˆ [B, τ+ ] = [B, τ− ] = [B, τz ] = 0
RELATIONSHIP BETWEEN SU2 AND R3 107 [ˆz . As we have seen. τ− ] = 2ˆz τ τ τ ˆ τ ˆ The operator B may be interpreted as the Baryon number operator and it is an invariant reﬂecting the conservation of Baryon number. the rotation group R3 is not the same as the group SU2 although they have the same Lie algebra. The two dimensional representations of these operators are obtained by evaluating matrix elements such as (τ+ )12 = pc+ cn n = 1 and this gives: p 1 1 1 B = 1 τ+ = (σx + iσy ). φ) D1/2 (a. Two diﬀerent Lie groups may have the same Lie algebra. The three operators τ+ .2. and τz form a Lie algebra by themselves and it is identical ˆ ˆ ˆ to the Lie algebra of the rotation group R3 . φ) = exp(φˆ · X) a (6. [ˆ+ . 6. [ˆz .2 Relationship between SU2 and R3 To relate the elements of R3 to those of SU2 we consider explicitly the twodimensional representation of R3 corresponding to j = 1/2. τ+ ] = τ+ . τ− = (σx − iσy ). The anticommutation relations of the inﬁnitesimal matrices X are given by 1 {Xi . φ) = exp(φˆ · X) a . Xj } = − δij 2 which implies (a · X)2 = ij (6. it is usually the Lie algebra which is important rather than the Lie group. τ− ] = τ− .6. A rotation about an axis a by an angle φ is represented by D1/2 (a. We can now easily work out the representation matrix D1/2 (a.3) ai aj Xi Xj = − 1 4 i 1 a2 E = − E i 4 ˆ sine a is a unit vector. τ− .2) with X = −iσ/2. τz = σz 2 2 2 In physical applications.
. . . The representation D1/2 for the rotation group can be made singlevalued by extending the physical rotation angle artiﬁcially to 4π (similar to Riemannian sheets in complex plane).108 CHAPTER 6. 4π + φ) = D1/2 (a. . φ) whereas R(a. . ( 6. 2π + φ) because R(−a. 2π − φ) is to be regarded as equivalent to a rotation R(a. A rotation such as R(−a. 2π − φ) = R(a.] a a 6 1 1 = 1 − (φ/2)2 + (φ/2)4 + . 2 4 1 +2(ˆ · X)[φ/2 − (φ/2)3 + .1) if we make the identiﬁcation α = cos φ/2 − iaz sin φ/2 and β = −(ay + iax )sin φ/2 with ˆ 0 ≤ φ < 2π and a pointing in all possible directions. φ) since D1/2 (a. This can be seen by noting that. 4π + φ) = R(a. The representation 1/2 D for the rotation group is however doublevalued because there are two matrices corresponding to the same rotation. a a a a 2 6 24 1 1 = 1 + φ2 (ˆ · X)2 + φ4 (ˆ · X)4 + . . . 2π − φ) The D1/2 representations corresponding to this same physical rotation are diﬀerent because D1/2 (−a. . φ) = R(−a. −φ) = R(−a. φ). A rotation R(a. the angle φ need only cover either 0 to π. or π to 2π since R(a. 4π − (2π − φ)) = R(a. φ) The relationship between SU2 and R3 is thus homomorphic because there are two distinct SU2 matrices corresponding to the same physical rotation. 2π + φ). 2π + φ) is regarded as a new rotation diﬀerent from R(a.] a 3 = cos φ/2 + 2(ˆ · X)sin φ/2 a = cos φ/2 − iaz sin φ/2 −(ay + iax )sin φ/2 (ay − iax )sin φ/2 cos φ/2 + iaz sin φ/2 These matrices are the matrices of SU2 in Eq. 2π − φ) = −D1/2 (a. . to produce all rotations. −(2π − φ)) = R(a. a a 2 24 1 +(ˆ · X)[φ + φ3 (ˆ · X)2 + . . . THE GROUP SU2 : ISOSPIN 1 1 1 = 1 + φ(ˆ · X) + φ2 (ˆ · X)2 + φ3 (ˆ · X)3 + φ4 (ˆ · X)4 + .
. Since the irreducible representations of R3 were deduced entirely from the commutation relations of the inﬁnitesimal operators. . where Z and N may take any values constrained by the mass number. 2 A given nucleus with a ﬁxed neutron number N and proton number Z corresponds to a state with MT = (Z − N )/2 since Q = Z. Conventionally. ˆ ˆ ˆ Physically. the operator Q = 1 B + Tz is regarded as a charge operator. when a Hamiltonian is invariant under SU2 . p and n . . 1/2. we use the same vectorcoupling theory as in the angular . −1/2. Similarly. . .3 SU2 in Nuclei As we have discussed in the previous section. For the general case of several nucleons. similar to L for angular momentum in R3 . . We recall that when a Hamiltonian is invariant under R3 . the eigenfunctions transforming according to the irreducible representations DL are degenerate labelled by the z components of angular momentum ML = L. This is because the forces between the nucleons are attractive whereas the forces between the electrons are repulsive. as in R3 . For a given nucleus with ﬁxed N and Z. tranforms according to D1/2 and consequently the eigenfunctions of a single nucleon will have MT = 1/2. To ﬁnd the irreducible representations of two nucleons. It is found experimentally that the ground state of a nucleus has the minimum value of the total isospin T whereas in an atom. . .3. −L. These multiplets are interpreted as diﬀerent possible states of nuclei. the degenerate eigenfunctions tranforming according to the irreducible representation DT will have zcomponents of isospin MT = T. the twodimensional space of proton and neutron. states with MT = (Z − N )/2 but T > Z − N /2 correspond to excited states. T = 0. 3/2 . For example. T is called the isospin. MT = 1/2 corresponds to p and MT = −1/2 to n . the inﬁnitesimal operators of the group SU2 have the same commutation relations as those of the rotation group R3 .6. 1. the ground state has the maximum value of the total spin S (Hund’s rule). −T . L − 1. T −1. we consider simultaneous SU2 transformation in every nucleon. This state may correspond to a state with T = Z − N /2 or higher. . it follows that we have the same irreducible representations in SU2 which we may label with DT . . . The physical interpretation of this result is that a nucleus with a ﬁxed mass number A = N + Z can exist in diﬀerent nucleon states with zcomponents of isospin MT = (Z − N )/2. SU2 IN NUCLEI 109 6.
N = 7. 0. −1. Since A = 13 is odd. it is not possible to have states with T = 1. From the vectorcoupling theory. 1/ 2(2 + 3 ). The next higher . To work out these states. let us write down explicitly the four vectors that form the space of two nucleons: 1 = p1 p2 2 = p1 n2 3 = n1 p2 4 = n1 n2 The vectors 1 and 4 have MT = 1 and −1 respectively and therefore belong to T = 1 states. The ground state will correspond to a singlet with T = 0 and the excited states to a triplet with T = 1 with MT = 1. 4 The remaining vector √ 1/ 2(2 − 3 ) must have T = 0 as may be checked by considering ˆ ˆ ˆ ˆ ˆ T 2 (2 − 3 ) = (T+ T− + Tz2 − Tz )(n1 p2 − p1 n2 ) = 0 Another example of isospin labelling is provided by nuclei with A = 13. we apply the lowering operator ˆ ˆ ˆ T− = t− (1) + t− (2) on the vector 1 which conserves the value of T but lowers the value of MT by one: ˆ T− 1 = n1 p2 + p1 n2 = 2 + 3 The resulting vector has MT = 0 and we conclude that the T = 1 states must be √ 1 . MT = −1/2 (C 13 ) and Z = 7. MT = 1/2 (N 13 ). The lowest value of T = 1/2 must be given by the nuclei with Z = 6. To ﬁnd the other states. we have states with T = 1 and T = 0. we consider two nucleons. THE GROUP SU2 : ISOSPIN DT1 ⊗ DT2 = T1 +T2 T1 −T2  ⊕DT which may be extended to an arbitrary number of nucleons.110 momentum theory: CHAPTER 6. As a ﬁrst example. N = 6.
6. Oi ] = k α α Ok (Xq )ki (6.4 Tensor Operators Tensor operators are irreducible sets of operators associated in particular with the rotation group R3 . T (a) = 1 + q ˆ aq Xq for small a so that q ˆ ˆα ˆ T (a)Oi T −1 (a) = (1 + ˆα = Oi + Since α α Ok Tki (a) = k k ˆ ˆα aq Xq )Oi (1 − q ˆ aq Xq ) q ˆ ˆα aq [Xq .4) which is the equivalent deﬁnition of irreducible set of operators in Lie groups. The general deﬁnition of an irreducible set of ˆα operators {Oi } is ˆ ˆα ˆ T (G)Oi T −1 (G) = k α α Ok Tki (G) ˆ In Lie groups. Oi ] α Ok δki + q α aq (Xq )ki we have ˆ ˆα [Xq . Jz ψm ) = mOm ˆ ˆj [J± . MT = −3/2 and Z = 8.4. The remaining two states with MT = ±1/2 must correspond to excited states of C 13 and N 13 . Om ] = m j ˆj ˆ j Om (ψm . Om ] = m j ˆj ˆj ˆ j Om (ψm . The relations deﬁning tensor operators associated with the rotation group are ˆ ˆj [Jz . 6. J± ψm ) = (j ± m + 1)(j j m)Om±1 . MT = 3/2 corresponding to T = 3/2. N = 5. TENSOR OPERATORS 111 states must be formed by the nuclei with Z = 5. N = 8.
. . O−1 = J− 2 . THE GROUP SU2 : ISOSPIN j is called the rank of the tensor and the number of operators in a set is (2j +1) with m = −j.112 CHAPTER 6. O0 = Jz . j. For example. . the angular momentum operators themselves form a tensor operator of rank 1 (vector operator): √ √ ˆ1 ˆ ˆ1 ˆ ˆ1 ˆ O1 = −J+ 2. . −j +1.
7. σ = reﬂection in a plane. a2 .Chapter 7 The Point Groups 7. 113 .2 Schoenﬂies Notation There are several notations used in describing point groups. 2. Rotations about axes through the origin. E = identity Cn = rotation through 2π/n. there are also symmetry operations which are performed with a point ﬁxed (origin) and which form a point group. Apart from translations. The fundamental covering operations of point groups are 1. and a3 and a crystal is a regular array of lattices such that it is invariant under translations t = n1 a1 + n2 a2 + n3 a3 . We follow the notation due to Schoenﬂies.1 Crystal Symmetry A lattice is deﬁned by three linearly independent vectors a1 . Inversion The complete set of symmetry operations including translations is called the space group.
2. then there must be a perpendicular n−fold axis. σv = reﬂection in the plane through the axis of highest rotation symmetry. The intersection of two reﬂection planes is a symmetry axis. 4. . A rotation and a reﬂection in a plane perpendicular to the axis of rotation 5. Two reﬂections in perpendicular planes 3. They are useful for working out group multiplication tables. 2. THE POINT GROUPS σh = reﬂection in the plane through the origin and perpendicular to the axis of highest rotation symmetry.114 CHAPTER 7.3 Commuting Operations We list pairs of operations which commute. Sn = improper rotation through 2π/n. A twofold axis and an n−fold axis implies the existence of n − 1 additional twofold axes separated by angles of π/n. 7. Two rotations by π about perpendicular axes. there must be n − 1 other reﬂection planes at angles of π/n. This is a special kind of σv . 1. The axis is n−fold if the angle between the planes is π/n. The inversion and any rotation or reﬂection The following rules are useful in determining the complete symmetry of the problem. If a reﬂection plane contains an n−fold axis. Two rotations about the same axis. 1. σd = reﬂection in the plane through the axis of highest rotation symmetry and bisecting the angle between the twofold axes perpendicular to the symmetry axis. i = S2 = inversion. If two 2−fold axes make an angle π/n. An improper rotation is a rotation followed by a reﬂection in a plane perpendicular to the axis of rotation.
2. Solid radial lines indicate vertical reﬂection planes and dashed radial lines indicate rotation axes. There are thirty two point groups and we list them in order of increasing complexity. and an inversion centre implies the existence of the other. then the translational vector t − t must be perpendicular to the rotation axis. 7. C2 . and 6. They are cyclic groups of order n. These are formed by projecting on to the XY plane a general point on a unit sphere which is subjected to the symmetry operations. Apply . The nfold rotation axes are indicated by the symbols at the centre of the circle. Simple rotation groups which have one symmetry axis of highest symmetry. Any two of the following: an evenfold axis.4 Enumeration of Point Groups The point groups may be divided into two categories: 1. ENUMERATION OF POINT GROUPS 115 3. Simple rotation groups: Cn • These are groups of rotations about n−fold axes. Horizontal reﬂection planes are indicated by unit circles with solid lines rather than broken lines. Let s be the shortest translational vector perpendicular to the axis of rotation. C4 . The groups are therefore C1 .1. Proof: If t is a translational vector and t = Rt. It is convenient to visualise the point group operations by means of stereographic projections as shown in Figure 7. 3. and C6 . C3 . Other rotations are not consistent with translational symmetry. a reﬂection plane perpendicular to it. It can be shown that n can only be 1. but which have more than one n−fold axis where n > 2. 2. 4.4. Higher symmetry groups which have no unique axis of highest symmetry.7. The labels +/◦ mean that the point is above/below the plane respectively.
D3 . Dnh • Each of these groups has Dn as a subgroup and in addition it contains the horizontal reﬂection plane σh . the group includes the inversion element so that C2nh = C2n ⊗ I. and S6 = C3 ⊗ I. Dnd • Each of these groups has Dn as a subgroup and in addition it contains diagonal reﬂection planes σd bisecting the angles between the twofold axes perpendicular to the principal rotation axis. D6 . The groups are D2 . Cnh • These groups have a σh reﬂection plane in addition to the Cn axis. The elements are E. C3h . This is shown in Figure 7. S4 . THE POINT GROUPS ing Cn on s generates a new vector s’ and we must have s − s = 2ssin π/n ≥ s since s is deﬁned to be the smallest translational vector perpendicular to the axis of rotation. and Z axes and 8 C3 ’s about the body diagonals which . C2h = C2 ⊗ I. D4 . Higher symmetry groups: T • This group consists of 12 proper rotations that take a regular tetrahedron into itself. C4h = C4 ⊗ I. It follows immediately that n ≤ 6 and geometrical consideration shows that n = 5 is not possible. The groups are D2d and D3d . and C6v .116 CHAPTER 7. C1v is identical to C1h . Cnv • These groups have a σv reﬂection plane in addition to the Cn axis. Dn • These groups have n twofold axes perpendicular to the Cn axis. Y. C3v . The groups are C1h . When n is even. 3 C2 ’s about the X. The groups are S2 . Dnh = Dn ⊗ I. S2n • These groups have an 2nfold axis for improper rotations and each has Cn as a subgroup. and C6h = C6 ⊗ I. C4v . The groups are C2v .2. Since σh commutes with rotations. According to rule 1 there must also be n reﬂection planes separated by an angle π/n around the Cn axis.
The unit cell required for compatibility with the pointgroup symmtery is speciﬁed by three translations a. All the C3 ’s are in the same class but the two kinds of C2 operations form two distinct classes. and 6 C2 ’s about axes through thr origin parallel to face diagonals. O • This is one of the most important point groups. and c and three angles α.7. . Y • This group consists of 60 proper rotations that take an icosahedron or a dodecahedron into itself.3b. The total number of elements is 24. Th • This group has T as a subgroup and in addition it contains inversion so that Th = T ⊗ I.4. 3 C2 ’s and 6 C4 ’s about the X. Yh • This group has Y as a subgroup and in addition it contains inversion so that Yh = Y ⊗ I. Oh • This group has O as a subgroup and in addition it contains inversion so that Oh = O ⊗ I. The presence of 6 σd ’s implies the existence of 6 S4 ’s about the X. shown in Figures 7. These are E.4b. b. 117 Td • This group has T as a subgroup and in addition it contains 6 diagonal reﬂection planes σd normal to a cube face and containing a tetrahedral edge such as ac or ab. We list also in Table 7.3a and 7. ENUMERATION OF POINT GROUPS consist of two classes. The following point groups are useful to know but they are not consistent with translational symmetry because they contain ﬁvefold axes.5. and γ as illustrated in Figure 7. clockwise and anticlockwise rotations by 2π/3.4a and 7. Y and Z axes. This is illustrated in Figures 7. β. 8 C3 ’s about the body diagonals. This group consists of 24 proper rotations that take an octahedron into itself.1 the classiﬁcation of the 32 point groups according to the crystal systems to which they belong. and Z axes. Y.
This isomorphism can be seen by noting that iQ cannot overlap with P since i is not contained in G. We may write G=P +Q wher P and Q contain proper and improper elements respectively. THE POINT GROUPS 7. which are often the case in physical applications. Then from the rearrangement theorem we have G = iP + iQ and therefore iP = Q and iQ = P so that G = P (1 + i) If i commutes with all elements in the group. Thus we associate the element iQk of G with the element Qk of G while P is common to both groups.118 CHAPTER 7. If the inversion is not contained in the group. The halfinteger representations arise for example when the basis functions are the spin functions χσ (ms ) which give the representation D1/2 . Since point groups are subgroups of the rotations group. i can be the inversion or horizontal reﬂection. . then G=P ⊗I where I = (1. it is isomorphic with the proper group G = P + iQ. i). Let G be an improper group. It follows also that a product of a proper and an improper element is an improper element. we also expect to have doublevalued representations when the basis functions transform according to the halfinteger representations. Each rotation is represented by two distinct matrices.5 Improper Point Groups The product of two improper elements is a proper element since an improper element has a determinant −1. Let i be any improper element. For example.6 DoubleGroup Representations We have seen that the halfinteger representations of the rotation group R3 are doublevalued. 7.
or T (R) and T (R).6. The classes of the double group can be worked out by using the rules discussed above.7. Let T (R) and T (R) be two matrices of G representing the same ¯ = ET (R) where E is minus a unit matrix repre¯ ¯ rotation R. T (R) 2 ¯ senting a rotation by 2π and E = E represents a rotation by 4π. E ¯ characters of T (R) and T (R) must therefore have opposite sign: ¯ χ(R) = −χ(R) ¯ except when R is a rotation by π. ¯ 3. then the matrices T (R) ¯ and T (R) form two classes of G. The classes of D3 are Ce . This is because for bilateral axis. there is an element that reverses the direction of the axis. say. we consider the double group of D3 . then T (R) and T (R) may or may not belong to the same class. The Evidently. i. The two groups G and g are homomorphic. in which case χ(R) = χ(R) = 0. R(±θ) belong to the same class so that in the double group R(π) and ¯ R(−π) = R(3π). As an example. If R is rotation by π. G contains twice as many elements as g does and the group G is called the double group of g. ¯ commutes with all elements of the double group. we can proceed as usual but the following rules simplify the procedure: ¯ 1. Neglecting for the moment the special case of rotation by π. e is in a class by itself because it ¯ commutes with all elements.e. DOUBLEGROUP REPRESENTATIONS 119 Consider a point group g consisiting of proper rotations and the corresponding group G of matrices taken from the representation D1/2 . Cc . belong to the same class. such that if T (R) belongs to one ¯ than T (R) belongs to the other. They belong to the same class if and only if the axis of rotation is bilateral. and Cd . To work out the character table of the double group. it follows that if a set of rotations R form a class of g. There is no bilateral axis so that the character table is given by . 2.
Let us suppose that we break the rotational symmetry by putting the atom in a potential with D3 symmetry. The double group. however. ¯ ¯ ¯ E commutes with all elements and E 2 = E so that (E. The product of three spin functions tranform according to the product representation D1/2 ⊗ D1/2 ⊗ D1/2 = D3/2 ⊕ 2D1/2 . The characters ¯ ¯ χ(Cd ) can be obtained by observing that d2 = e so that χ2 (Cd ) = −1 and ¯ 1 ¯ therefore χ(Cd ) = ±i. The representations D3/2 and D1/2 may be reduced by using the usual formula χ = γ mγ χγ where γ labels the irreducible representations of the double group. the original group g is no longer a group. which is the full symmetry of a crystal. The rest of the table can be ﬁlled in by using the orthogonality between the characters. rotationalsymmetry operators from the point group. contains translationalsymmetry operators. THE POINT GROUPS ¯ ¯ ¯ Ce 2Cc 2Cc 3Cd 3Cd 1 1 1 1 1 1 1 1 −1 −1 2 −1 −1 0 0 −1 1 −1 i −i −1 1 −1 −i i −2 −1 1 0 0 T T2 T3 T4 T5 T6 1 Ce 1 1 2 1 1 2 The ﬁrst three are the singlevalued representations of D3 and the last three are the new representations arising from the double group. E) form a subgroup.120 CHAPTER 7. is not a direct product of this subgroup and the original group because from the point of view of the double group. 7. The elements of the space group are conventionally represented by {Rt} meaning {Rt}r = Rr + t (7. and the combinations of the two. As a simple example of physical applications of the double groups we consider an atom with three selectrons.7 Space Group The space group. The product functions transform according to the double group of D3 and not according to the group D3 .1) .
7.e.14) that the eigenfunctions of a Hamiltonian with a lattice translational symmetry must obey Bloch theorem: ψk (r + t) = eik·t ψk (r) (7.5) Functions having this property are called Bloch functions. SPACE GROUP The multiplication rule for the elements of the space group is {Rt}{R t }r = {Rt}(R r + t ) = R(R r + t ) + t = {RR Rt + t}r If {R t } = {Rt}−1 then {RR Rt + t} = {E0} so that R = R−1 . The form of the Bloch functions explicitly takes into account the lattice translational symmetry. i. the point group is a subgroup of the space group.3) (7. ˆ ˆ T (R)ψk (r) = T (R)uk (r)eik·r −1 = uk (R−1 r)eik·R r = uk (R−1 r)eiRk·r (7. Now we consider the eﬀects of the point group operations on these functions.7.2) (7. t = −R−1 t Thus the inverse element is {Rt}−1 = {R−1  − R−1 t} 121 (7. uk (r + t) = uk (r) We consider here only simple space groups (symmorphic).4) We have already shown in example 1 (3.7) . This means that we can write ψk (r) = uk (r)eik·r (7. By this we mean that all the point group elements are members of the space group or in other words. This is not true in general.6) where uk is periodic.
Then uk will transform according to an irreducible representation of the full point group. The degeneracy at this . If k lies on a symmetry point. This equation shows the signiﬁcance of the group of k. Under the operations of this group. These operations that take k into itself clearly form a subgroup because R1 R2 k = R3 k = k where R1 R2 = R3 . we have (in atomic units h = m = e = 1) ¯ − 1 2 2 + V (r) Ψk = Ek Ψk (7. Formally. we get − 1 2 2 + 2ik · − k 2 uk = E k uk (7. some of the operations will take k into itself.8) Thus we see that {uk (R−1 r) transform among themselves under the group of k according to an irreducible representation of the group which is called the small representation. Consider ﬁrst the case of k = 0.10) which may be rewritten − 1 2 2 + V (r) + k · p uk = Ek uk (7. we have ˆ T (R)ψk (r) = uk (R−1 r)eik·r (7. We can think of the last term as a perturbation which breaks the symmetry of the lattice potential V (r). If we apply all the pointgroup operations on a wave vector k we obtain what is called the star of k.9) Using the Bloch form.122 CHAPTER 7.11) where E = E − k 2 /2. The eigenfunctions of the Hamiltonian of a crystal are then speciﬁed by two quantities: the k vector (Bloch theorem) and the irreducible representation of the group of k. THE POINT GROUPS The function uk (R−1 r) is also periodic because uk (R−1 (r + t)) = uk (R−1 r + t ) = uk (R−1 r) where t = R−1 t is just another translational vector. This subgroup is called the group of the wave vector.
SPACE GROUP 123 point is given by the dimension of the irreducible representation as usual.7. In this way we can ﬁgure out the degeneracies at any k point. Along the xaxis. these degenerate states will in general be split. In this case. Let us assume that the point group is D4 . Thus the states at this point cannot be degenerate. consider a two dimensional crystal with a square primitive lattice. This is an Abelian group and the irreducible representations are one dimensional. . If we consider the point k = 0. the degenerate states at k = 0 still form an invariant subspace for the subgroup but they are in general reducible. ax ) (using the same notation as in the exercise). We expect that the degeneracy will be split. If we take a point along the xaxis. As a simple example. then the k vector group is (e. The reciprocal space is also a square. Let us now consider the case k = 0 with the corresponding k vector group which is a subgroup of the full point group.7. we might have states which are doubly degenerate because there is a two dimensional irreducible representation of D4 .
THE POINT GROUPS .124 CHAPTER 7.
n 2 p 1 ˆ Y = (c+ cp + c+ cn − 2c+ cΛ ). Λ + c Λ cn These operators form a Lie algebra. Λ = c+ 0 p n Λ The bilinear products of annihilation and creation operators that conserve particle number are: c+ cp . Λ c+ cp . n = c+ 0 .1 Inﬁnitesimal Operators and Lie Algebra The group SU3 is an extension of the group SU2 .Chapter 8 The Group SU3 8. n c+ cΛ . We deﬁne a new set of inﬁnitesimal operators as follows: ˆ B = c+ cp + c+ cn + c+ cΛ . Λ n p 1 ˆ Tz = (c+ cp − c+ cn ). p c+ cΛ . neutron and lambda particle states: p = c+ 0 . n c + cp . We derive the Lie algebra from a physical point of view by considering three states which may be associated with the proton. p c+ cn . p c + cn . n Λ 3 p 125 . n c+ cΛ .
U± .1). Λ ˆ V+ = c+ cp .8) ˆ = 2δJJ Jz ˆ = ±V and cyclic permutations (8. The other operators T ˆ operators. J− ] = 2Jz (8. ( 8. V . p + ˆ T− = cn cp . Tz and Y . J ˆ ˆ T± . U.4) and ( 8. so that the corresponding ˆ± . ( 8. as the notation already suggests.3).7) . we have commutation relations among the ladder operators which are useful for generating the multiplets: ˆ ˆ J± . In additions. Thus the desired Lie algebra is ˆ ˆ Tz . Y ˆ ˆ Tz . ( 8.3) (8.4) (8. U ± We observe also that ˆ ˆ ˆ [Jz .1) 3ˆ 1ˆ 1 ˆ Vz = (c+ cΛ − c+ cp ) = − Y − Tz (8. n + ˆ U− = c cn . ˆ ˆ ˆ [J+ . which is the Lie algebra of SU2 . We could have chosen CHAPTER 8. The operator B commutes with all other operators and it may be interpreted as the (baryon) number operator. J± ] = ±J± . THE GROUP SU3 ˆ U+ = c + c Λ . J± = 0 1 ˆ = ± (3δJT − 1)J± 2 ˆ = ±(δJU − δJV )J± (8.6) (8. J± ˆ ˆ Y . and V± are the ladder ˆ Lie group is of rank 2. ( 8.2).5). The rest of the operators form a Lie algebra by themselves and there are at most two ˆ ˆ operators that mutually commute. The Lie algebras for U and V can be obtained from Eqs.5) where J = T.2) p Λ 2 4 2 ˆ ˆ instead of Tz .126 ˆ T+ = c+ cn . Λ + ˆ V− = c cΛ p 1 3ˆ 1ˆ ˆ Uz = (c+ cn − c+ cΛ ) = Y − Tz n Λ 2 4 2 or (8.
Tz .2 Subgroups Knowledge of the subgroups is useful for working out the multiplet strucˆ ture. ( 8. However.3 Multiplet Structure ˆ ˆ There are two mutually commuting operators. There are other subgroups but they are not important for the purpose of working out the multiplet structure. The ˆ ˆ direct product group SU2 ⊗ U1 since Y Abelian group U1 can be shown to be isomorphic with the group R2 . and Λ is given by 1 2 Tz = 0 0 0 −1 2 0 0 0 0 1 3 Y = 0 0 0 1 3 0 0 0 2 −3 0 1 0 T+ = 0 0 0 0 0 0 0 0 0 U+ = 0 0 1 0 0 0 0 0 0 V+ = 0 0 0 1 0 0 0 0 0 T− = 1 0 0 0 0 0 0 0 0 U− = 0 0 0 0 1 0 0 0 1 V− = 0 0 0 0 0 0 These are in fact the generators of the group SU3 . the states .2. so that a multiplet may be labelled by two numbers related to the eigenvalues of two Casimir operators. The generators (Y . just like Tz ˆ ˆ ˆ is a generator of an Abelian group R2 . Each state in a multiplet may be labelled by the eigenvalues ˆ ˆ of Tz and Y .8. SUBGROUPS 127 The representation of the inﬁnitesimal operators in the three dimensional space of p . Tz and Y . 8. 8. T± ) form a ˆ commutes with Tz and T± .8) shows that there are three SU2 subgroups. due to the presence of SU2 subgroups. Y is a generator ˆ of an Abelian group as indicated by the diagonal form of Y . Eq. n .
. and Vz are interrelated so ˆ ˆ follows from Eqs. Or alternatively. ±1/3. Uz . Y + 1. (8.1). It folˆ lows from Eq. . THE GROUP SU3 may be degenerate and additional labels are required to distinguish these degenerate states. Similarly for (Uz . Since Y corresponds to an Abelian group. T+ .8). U− ) and (Vz . Each state in a given multiplet may then be represented by a point in (MT . Eqs.2) that the eigenvalues of Y are Y = 0. The ladder operators may be conveniently represented by arrows as shown in Fig.2).1) and ( 8. α) ˆ ˆ where MT and Y are the eigenvalues of Tz and Y respectively and α is a set of additional labels required to distinguish states with the same MT ˆ and Y . The above equations imply that ψ is an eigenfunction of ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ T 2 . T ) 2 To generate a multiplet. ±1. T ) = T 1 c(T ) ψ(MT + . the operators (Tz . We start from a state ψ with the highest value of Y and for this Y the highest value of MT . ±2/3. . Y. For example.1) or ( 8. and V 2 which may be veriﬁed by operating T 2 = T− T+ + Tz2 + Tz . ( 8. U− ψ is not necessarily zero because U− lowers the value of Y . T− ) are the inﬁnitesimal operators of the group SU2 and therefore MT takes the values MT = ˆ ˆ ˆ ˆ ˆ ˆ 0. (8.2) but T that one of them determines the other two. ( 8.5) tell us that ˆ V− ψ(MT . its representation is one dimensional and uniquely labelled by Y .. The ﬁrst alternative implies: ˆ ˆ ˆ T+ ψ = U + ψ = V − ψ = 0 ˆ ˆ ˆ Although U− increases MT . U 2 . Y )plane as illustrated in Fig. we could start from the highest value of MT and for this MT the highest value of Y . Thus the states in a given multiplet can be written as ψ(MT . Each state is not ˆ ˆ ˆ ˆ only an eigenstate of Tz and Y but it is also an eigenstate of Uz and Vz as ˆz ..4) and ( 8. V− ). we use the same procedure as in the case of the rotation group. we may then choose α = T . ±1/2.128 CHAPTER 8. keeping in mind that each point may represent more than one state. The label MT on the other hand may arise from several diﬀerent values of T . ( 8. ±1. . . U+ . To label each state uniquely. . Y. ( 8. ˆ ˆ ˆ From Eq. V+ .
3. ( 8. (8. In other words it has deﬁnite T . and similarly for U 2 and V 2 . The coordinate of point F is F = [− 1 λ. 1 (λ + 2µ)] as shown in 2 3 the Fig. Consider the state . The states along BC can be obtained by applying V+ repeatedly on ψ(B) or we can argue from symmetry that the states must be symmetrical about the diagonal bisecting AB. MULTIPLET STRUCTURE 129 ˆ ˆ on ψ.8. We can now generate states inside the hexagon and there may be more than one independent state at each point. States on the outermost hexagon are unique and they have deﬁnite T. 3 (λ − µ)]. The state ψ has the following values of MT and Y : 1 MT = λ. ˆ 3. and V . 1 (λ + 2µ)] 2 3 ˆ 2. The rest of the hexagon can be obtained by symmetry about the diagonal bisecting BC. We apply T− repeatedly on ψ(A) until we reach point F at which ˆ T− ψ(F ) = 0. We apply U− repeatedly on ψ(A) until we reach point B at which ˆ U− ψ(B) = 0. Conventionally. In general. 2 1 V = (λ + µ) 2 (8.1) and ( 8. The steps to generate the multiplet is as follows: ˆ 1. Or 1 T = λ. There cannot be states to the right of AB because from Eq. U. To ﬁgure out the coordinate of point B. it can also be a triangle which is the case when λ = 0 or µ = 0. and V . U . Thus B = [ 2 (λ + µ).2) that λ + µ = 2V . we observe ˆ ˆ that U− increases MT by 1/2 and decreases Y by 1 and U− has been 1 1 applied µ times from A to B. which implies from Eqs.3). ( 8. ψ is labelled by two numbers λ = 2T and µ = 2U . We have shown that the states in a given multiplet must lie on or inside a hexagon.9) The two numbers λ and µ may also be used to label the multiplet: Dλµ .6) we have ˆ ˆn ˆ ˆn T+ U− ψ = V− U− ψ = 0 Thus the states along AB still have deﬁnite but diﬀerent values of T and V and the same value of U . 2 1 U = µ. 2 1 Y = (λ + 2µ) 3 and let us denote the state by the point A = [ 1 λ. 4.
Continuing along the same fashion. THE GROUP SU3 labelled by point G. The reason is that ˜ with a triangle we cannot generate an new state such as φ2 . By choosing c = λ the linear combi˜ ˆ ˜ nation φ2 = φ2 + λφ3 has a deﬁnite T = 1 (λ − 1) since T+ φ2 = 0 2 1 ˜ ˆ ˜ and Tz φ2 = 2 (λ − 1)φ2 . ˆ ˆ The states φ1 and φ2 are linearly independent since T− and U− do not commute but φ3 is not independent of the other two because ˆ ˆ ˆ [T− . This is illustrated by the representation D30 in . Applying T− on ψ(G) generate two sets of states parallel to AF . The third state may be formed by a similar linear combination as the one described above. It is of course possible to choose the states ˆ to have deﬁnite U or V . For example. it is not diﬃcult to show that the next line parallel to AF consists of three 1 1 sets of states with T = 2 (λ − 2).130 CHAPTER 8. 2 λ. There are three basic ways of reaching point G from A which generate the states: ˆ ˆ φ 1 = T− U − ψ ˆ ˆ φ2 = U− T− ψ ˆ φ3 = V+ ψ. Eventually we may arrive at a triangle instead of a hexagon at which stage the degeneracy remains constant as we move further. the 2 ˆ 2 ψ has T = 1 (λ + 2) while the state U− φ2 has T = 1 λ since ˆ ˜ state U− 2 2 ˆ ˆ ˜ ˆ ˆ ˜ T+ U− φ2 = U− T+ φ2 = 0. U− ]ψ = −V+ ψ → φ1 − φ2 = −φ3 1 The state φ1 has a deﬁnite T = 2 (λ + 1) and it is possible to form a linear combination of φ2 and φ3 so that it has a deﬁnite T : ˆ T+ (φ2 + cφ3 ) = = = = ˆ ˆ ˆ ˆ ˆ T+ U− T− ψ + cT+ V+ ψ ˆ ˆ ˆ ˆ ˆ ˆ U− T+ T− ψ + c(V+ T+ − U− )ψ ˆ ˆ ˆ ˆ ˆ U− (T− T+ + 2Tz )ψ − cU− ψ ˆ (λ − c)U− ψ ˆ using the fact that T+ ψ = 0. This implies that when λ = 0 or µ = 0 each point corresponds to one state because we start with a triangle and no extra states are generated as we move inside. The multiplet structure consists of hexagons of decreasing size with the degeneracy increased by one as we move inside. 1 (λ + 2).
We start with the state 4 ψ(A) = ψ(1. . ) = c(U− T− + 2V+ )ψ(A) 2 3 2 . 8. The rest of the hexagon and the inner hexagons can be formed from the rules discussed above. (8. .8. µ) = (λ + 1)(µ + 1)(λ + µ + 2) (8. 1 (λ + 2µ)]. 1) 3 1 3 ˆ which has deﬁnite values of T = 1. Applying T− twice generates 4 4 ψ(0. and V = 2 . . we 2 3 ˆ ˆ form the line AF by moving λ steps along T− and µ steps along U− . EXAMPLE 131 ˆ ˆ Fig.4) where the point O can be reached from A by V+ T− ψ(A) ˆ− V+ ψ(A) but [V+ . multiplet is generated from the linear combination 1 1 1 ˆ ˆ ˆ ψ( . . 1). let us construct explicitly the multiplet D21 as shown in Fig. (8. Thus the outermost hexagon contributes 3(λ + µ) states and the next one inside contributes 2 × 3(λ + µ − 2) states and so on taking care of the special case of a triangle. Y )−plane.5).4. .4 Example As an example.10) 2 which is symmteric with respect to interchange of λ and µ. 1) 3 3 ˆ Applying U− on ψ(A) gives 3 1 3 ψ(B) = ψ( . Given a representation Dλµ it is straightforward to map the states in the (MT . T− ] = 0 so that the two paths are equivalent. The general formula is given by 1 d(λ. U = 2 . ψ(−1. ) 2 3 2 The T = The T = 3 2 1 2 ˆ multiplet is generated by repeated applications of T− on ψ(B). ˆ ˆ ˆ or T The multiplet structure also allows us to work out the dimension of the irreducible representation. Starting from the point A = [ 1 λ.
T+ U+ U− T− ψ(A)) ˆ ˆ ˆ ˆ ˆ = (ψ(A). 3 2 1 1 1 ˆ ψ(0.5 Product Representations In physical applications such as analysing the possible ﬁnal states in a collision between two particles. − . 1) = c1 T− V+ ψ(B). V+ ψ(A)) = 3 √ Thus the normalisation constant is given by c = 1/ 24. We illustrate this . 0) = c2 U− ψ( .11) The general formula for the coeﬃcients mλµ is complicated but it possible to obtain them directly by the method of reduction. THE GROUP SU3 The normalisation constant is obtained by requiring ˆ ˆ ˆ ˆ ˆ ˆ c2 (U− T− + 2V+ )ψ(A). (U− T− + 2V+ )ψ(A) = 1 We have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ (U− T− ψ(A). T+ (T− Uz + T− ψ(A)) 2 ˆ+ T− ψ(A)) ˆ = 2(ψ(A). T+ U+ V+ ψ(A)) ˆ ˆ ˆ ˆ = (ψ(A). T+ (U− U+ + 2Uz )T− ψ(A)) 1ˆ ˆ ˆ ˆ = 2(ψ(A). T = 4 ˆ ˆ ˆ ˆ ˆ ˆ (U− T− ψ(A). − . ) 3 2 3 2 The normalisation constants can be calculated in the same way as above. 8. T+ (V+ U+ + T− ψ(A)) = 2 ˆ ˆ (V+ ψ(A). The states at C are given by 2 ˆ ˆ ψ(0. U− T− ψ(A)) = (ψ(A).132 CHAPTER 8. V+ ψ(A)) = (ψ(A). we have to decompose product representations: Dλ1 µ1 ⊗ Dλ2 µ2 = λµ ⊕ mλµ Dλµ (8. .
4( 2 ) 1. 4(0) 2 3 1 . 1 2 2 3 1 . Y = 2 which is the representation D22 . We substract the T values of D22 listed in the third row from the product representation which yields the fourth row. 0 = 1 = 0 = 0 = = = = = = 1 0 3 2( 2 ). 1) ⊕ (1. 3( 2 ) 4(1). 3(0) 2 3 1 2 1 2(1). Y1 ) ⊗ (T2 . Y = 2 which is the representation D03 . 3(0) 2( 2 ) 1 2( 2 ) 2(1). Y ): 1 1 D11 = ( .5. The product D11 ⊗ D11 must contain a representation with T = 1. −1) 2 2 We form products of (T. Y2 ) = T =T1 −T2  ⊕ (T. PRODUCT REPRESENTATIONS 133 method by considering the product D11 ⊗ D11 . there must be a representation with T = 0. Forming all products and substracting all possible representations contained in the product representation we have: D11 ⊗ D11 D22 D03 D 30 2D11 D00 Y T T T T T T T T T T = 2 = 1. First we list all possible T and Y multiplets contained in D11 and label them by (T. Following the same argument. 3( 2 ) 0 2 3 2 3( 1 ) 2 1 2 2( 1 ) 2 2( 1 ) 2 0 0 The ﬁrst row is a list of all possible Y that arise from the product representation. 5(1). 3(0) 2 1 1 2 3 1 . 2( 2 ) 3(1). The second row is a list of all possible T and the number preceeding each bracket is the number of times the T multiplets arise for a given Y . 1. Y ) T1 +T2 (T1 . Y1 + Y2 ) To determine T we have used the usual vector coupling theorem and to determine Y we note that it represents an Abelian group so that the representation is of the form exp(icY ) and therefore the product of two representations Y1 and Y2 is exp[ic(Y1 + Y2 )]. 0) ⊕ (0. 0 2 3 1 . 0 2 1 3 .8. 2(0) 0 0 −1 −2 1 2( 3 ).2 2. We substract . 0) ⊕ ( . 4( 1 ) 2.
8 of Elliot and Dawber) shows some of these classiﬁcations. n) or (π + . Strong interactions are found to be invariant under isospin tranformation. (8. hadrons may be assigned the following intrinsic quantities: 1. Hypercharge Y: The quantity Y in the group SU3 is related to the charge and the zcomponent of isospin by Y = 2(Q − MT ) This relation is a physical interpretation which does not follow from group theory. It is known experimentally that the hypercharge is conserved in strong interaction. The corresponding .f. states in a given isospin multiplets transform among themselves under isospin transformation such as transformation among (p. mass. With the above assignments of intrinsic quantities. charge.134 CHAPTER 8. except proton and neutron. 3. Hadrons can be divided into baryons with halfinteger spins and mesons with integer spins.6 SU3 Multiplets and Hadrons Hadrons are strongly interacting particles with decay time < 10−8 sec. 2. π 0 ) (c.e. hadrons can be beautifully classiﬁed according to SU3 multiplets. B = −1 for antibaryons. Thus we conclude ﬁnally D11 ⊗ D11 = D22 ⊕ D03 ⊕ D30 ⊕ 2D11 ⊕ D00 8. angular momentum states). proton and neutrons which correspond to T = 1/2 or the three πparticles which correspond to T = 1. spin.6) (ﬁgure 11. and B = 0 for mesons. the former are heavier than the latter. Fig. hadrons have (almost) the same masses. THE GROUP SU3 the T values of D03 from the remaining T ’s and continue along the same fashion. The assignment of baryon number if as follows: B = +1 for baryons. and parity. Apart from. i. π − . Isospin T : Among an isospin multiplet. including weak ones. By strong we mean that the energy involved in a reaction is of the order of M eV or greater. For example. Baryon number B: It is conserved in all reactions.
Y = + . For the group SU3 there is a simple way of working out the Casimir operators which is based on the observation that the inﬁnitesimal operators may be written in terms of the matrices deﬁned by: 1 (8. These three quarks are supposed to correspond to the representation D10 so that 1 1 1 2 u → T = . Y = + . 8. it is possible to use the general formula derived before and a similar formula for higher order Casimir operators can also be derived although they become more complicated. MT = 0. To construct the second order Casimir operator. Q = + 2 2 3 3 1 1 1 1 d → T = . Q = − 2 2 3 3 2 1 s → T = 0. it is then proposed that the baryons can be constructed from the triple product D10 ⊗ D10 ⊗ D10 = D30 ⊕ 2D11 ⊕ D00 A more complicated model involving nine quarks has also been proposed which introduces. Q=− 3 3 Since the largest spin is S = 3/2. d. It has been suggested that baryons are composed of three more fundamental spin 1/2 particles known as quarks which are labelled by u. MT = − . and s.12) (Aj )kl = δik δjl − δij δkl i 3 .1 of Elliot and Dawber). which arise as resonances in collisions.1) (from table 11.7. apart from spin. additional intrinsic properties known as colours. Y =− . MT = + .7 Casimir Operators We have not made any explicit use of the Casimir operators in working out the structure of the multiplets. may often be regarded as excited states of the more stable ones.8. Particles with short lifetime. The representation D11 and D30 consist of baryons with spin 1/2 and 3/2 respectively while the other D11 representation corresponds to the mesons. CASIMIR OPERATORS 135 particles are listed in Table (8.
1 2 3 1 U+ = A3 . 3 3 T− = A1 2 V− = A 3 1 The commutation relations for the Aj are i [Aj . i j C3 = ijk Ak Aj Ai j i k commute with all of the inﬁnitesimal operators. Y = −A3 . 2 U− = A2 . Instead of C3 it is more convenient to deﬁne 3 C3 = C3 + C2 2 so that ˆ C2 ψ = 2 2 (λ + µ2 + λµ) + 2(λ + µ) ψ 3 1 ˆ C3 ψ = (λ − µ)(2λ + µ + 3)(2µ + λ + 3) ψ 9 which may be easily veriﬁed by using the Lie algebra. T+ = A2 . Al ] = δjk Al − δil Aj i k i k It is then easy to show that the operators C2 = ij Aj Ai .136 We have CHAPTER 8. V+ = A1 . THE GROUP SU3 Tz = (A1 − A2 )/2. .
This action might not be possible to undo. Are you sure you want to continue?