Optics 412 The Wigner-Eckart Theorem Study Guide

The theorem of Wigner and Eckart concerns the matrix elements of operators between states of denite
angular momentum. It shows that for a large and important class of operators, the eect of the operator on
the angular momentum of the state is to change the quantum numbers in a way just as if you had added
a new angular momentum to the system. The purpose of this guide is to enable you to state this theorem
precisely and to prove it. Basic ideas about angular momentum addition are reviewed rst, but the reader
is assumed to have read the relevant class notes, and if necessary to have read Cohen-Tannoudji Chapter X,
AX, BX.
Section 1. Review of addition of angular momentum. Clebsch-Gordan coecients.
The following review questions are about two systems whose individual total angular momenta are
j
1
= 4 and j
2
= 1.
1. system 2 has three magnetic substates. There are (how many?) magnetic substates of
system 1. (Write in the answer and then compare it with the answer given on the line below.)
1. nine.
2. Given that system 1 is in the state |j
1
, m
1
= |4, 2 and that system 2 is in state |j
2
, m
2
= |1, 1, then
the combined state is in the state |j
1
, j
2
, m
1
, m
2
= |4, 1, 2, 1. For the same j
1
and j
2
(4 and 1), how
many dierent states are there of the combined system, including |4, 1, 2, 1?
2. twenty-seven, namely:
|4, 1, 4, 1, |4, 1, 4, 0, |4, 1, 4, 1,
|4, 1, 3, 1, |4, 1, 3, 0, |4, 1, 3, 1,
|4, 1, 2, 1, |4, 1, 2, 0, |4, 1, 2, 1, etc. .
3. What are the possible values of J for the combined system, given that j
1
= 4 and j
2
= 1?
3. three, four, and ve.
The problem of addition of angular momenta is to take the 27 states |j
1
, j
2
, m
1
, m
2
and form linear
combinations of them in certain ways to make 27 new states which are all eigenstates of

J
2
and

J
z
with
eigenvalues J(J + 1) and M.
4. How many of the new states will have the eigenvalue J = 3?
4. seven.
5. How many will have J = 4?
5. nine
And there will be eleven states with J = 5. Notice that altogether that makes 7 + 9 + 11 = 27, the
same as the number of states |4, 1, m
1
, m
2
. This is no accident, since the states |J, M are to span the same
space as the states |4, 1, m
1
, m
2
and that space has 27 dimensions.
1
Now consult Appendix A in the back.
6. Exactly two of the twenty-seven states |4, 1, m
1
, m
2
, are already eigenstates of

J
2
and

J
z
. They are
and .
6. |4, 1, 4, 1 and |4, 1, 4, 1.
7. The coecients in this transformation from the set {|j
1
, j
2
, m
1
, m
2
} to the set {|J, M} are called
Clebsch-Gordan coecients and are written j
1
, j
2
, m
1
, m
2
| J, M. For example, how much is 4, 1, 2, 1| 5, 1?
(Appendix A.)
7.
_
6/45.
8. From the Appendix, you see that 4, 1, 2, 1| 5, 0 = 0. This is an illustration of the rule relating m
1
, m
2
and M, which is .
8. m
1
+ m
2
= M (for non-zero C.G. coecients.)
8.1 4, 1, 0, 1| 7, 1 is zero. Why?
8.1 Because J = 7 is not allowed from j
1
= 4, j
2
= 1.
8.2 4, 1, 1, 9| 5, 8 is zero. Why?
8.2 Because m
2
> j
2
. Or because M > J.
9. From inspection of the Appendix, you can ll in the summation indices and ranges of the summations
below:

=
j
1
, j
2
, m
1
, m
2
| J, M = 1.
9.

j
1
m1=j1

j
2
m2=j2
etc.
See Appendix B, eqn. (1). Other properties of the transformation will be found in Appendix B. Note
especially the recursion relation. The recursion relations are occasionally useful in calculations:
10. From the recursion relations, and from the data in Appendix A, nd the value of
j
1
, j
2
, m
1
, m
2
| J, M = 4, 1, 1, 1| 5, 2 ,
using the data from the line giving |5, 3. Answer: .
10. From the recursion relation (Eq. 3 in Appendix B),
_
5 6 (3)(2) 4, 1, 1, 1| 5, 2 =
_
4 5 (1)(2)
_
28/45 + zero,
therefore
4, 1, 1, 1| 5, 2 =
_
21/45.
2
11. You can ll in the missing line in Appendix A either by using the recursion relations (dont) or by
noticing a symmetry relating |J, M and |J, M (do). Check your result below.
11.
|5, 2 = |4, 1, 1, 1
_
21/45 + |4, 1, 2, 0
_
21/45 + |4, 1, 3, 1
_
3/45.
We have left out two details: The rst is that J and M do not completely specify a state of a system
composed of subsystems having a given j
1
and j
2
. You can get a |J, M = |5, 3 not only from subsystems
with (j
1
, j
2
) = (4, 1), but also from those with (j
1
, j
2
) = (3, 2) for example. We should have been writing
states of the combined system as |j
1
, j
2
, J, M. then the transformation of basis we have been talking about
would be expressed as that from a set {|j
1
, j
2
, m
1
, m
2
, xedj
1
, j
2
} to another set {|j
1
, j
2
, J, M, xedj
1
, j
2
}.
The Clebsch-Gordan coecients are still abbreviated as j
1
, j
2
, m
1
, m
2
| J, M

instead of the redundant form

j
1
, j
2
, m
1
, m
2
| j
1
, j
2
, J, M

.
The last detail, a very important one, is this: An eigenstate of a hydrogen atom, for example, is not
completely specied by giving the values of j and m; what about the value of the principal quantum number
n? When we are discussing angular momentum, we shall label all the other quantum numbers of a system
by a single letter. A completely specied state would be called |a, j, m,

for example.
So what about the transformation from the set {|a, j
1
, j
2
, m
1
, m
2
xed a, j
1
, j
2
}, to the set {|a, j
1
, j
2
, J, M, xed a, j
1
, j
2
}?
The important and beautiful thing is that the coecients do not depend on a. That is why you nd tables
of Clebsch-Gordan coecients for arbitrary systems.
Section 2. Irreducible Tensor Operators Order 1 and 3
Operators sometimes come in sets. The Hamiltonian for an isolated system is a single operator; it
belongs to a set of order 1. The position operator of a system ( x, y, z) is an operator set of order three. Now
it so happens that most of the operators that one is interested in can be tted into the following description
by a suitable choice of basis:
Let
_

T
(k)
q
, q = k, k + 1, . . . , +k
_
be a set of 2k+1 operators which satisfy the following commutation
relations with angular momentum operators:
_

J
+
,

T
(k)
q
_
=
_
k(k + 1) q(q + 1)

T
(k)
q+1
_

,

T
(k)
q
_
=
_
k(k + 1) q(q 1)

T
(k)
q1
_

J
z
,

T
(k)
q
_
= q

T
(k)
q
.
Then the operators are the standard components of an irreducible tensor operator

T
(k)
of order 2k + 1.
1. A trivial example: The Hamiltonian of an isolated system commutes with

J
+
,

J

and

J
z
. The Hamilto-
nian operator is the sole member of a set of order . Is it a standard component? That is, does
it satisfy the above relations? Answer:
1. one. Yes.
2k + 1 = 1 k = 0 q = 0 all commutators = 0.
2. Are x, y and z the standard components of the position operator? To answer this one, consult Appendix
C. Answer .
3
2. No. Just look at [

J
z
, x] for example.
3. We can construct a set of standard components easily, however. Notice [

J
z
, z] = 0 z. Perhaps z is
the standard component having q = . (Check the commutation rules at the beginning of this
section.)
3. zero.
4. Suppose z is the q = 0 component. (Obviously, since we have three components, k = 1 and q =
1, 0, +1.) To nd the q = 1 component, just use the appropriate commutation rule. Which rule?
4.
_

J
+
,

T
(k)
q
_
.
5. Using Appendix C and the appropriate commutation rule, nd the q = +1 standard component of
position.
5. q = 0, k = 1,

T
(1)
0
= z.
Thus,
_

J
+
,

T
(k)
q
_
=
_
k(k + 1) q(q(+1)

T
(k)
q+1
=

2

T
(1)
+1
,
but,
_

J
+
,

T
(k)
q
_
=
_

J
+
, z]
_
= x i y from Appendix C.
Therefore, the q = 1 standard component is ( x i y)/

2.
6. Find the q = 1 standard component of position. Answer: .
6. ( x i y)/

2.
7. This result for the position operator can be generalized to the case of any vector operator: If

K
x
,

K
y
,

K
z
are the cartesian components of a vector operator, its standard components are

K
(1)
1
= ?

K
(1)
0
= ?

K
(1)
1
= ?
7. (

K
x
i

K
y
)/

2,

K
z
, (

K
x
i

K
y
)/

2.
You can test yourself at this point to see that you can dene an irreducible tensor operator. If so, you
are ready to go on to a statement of the Wigner-Eckart Theorem.
Aside: Maybe you have noticed that while we have dened the order of the set as 2k +1, many authors
call k the degree or rank of the tensor.
4
Wigner-Eckart Theorem: If

T
(k)
q
is the qth standard component of an irreducible tensor operator

T
(k)
of rank k and order 2k + 1, then its matrix element between states of denite total angular momentum
a, J, M|

T
(k)
q
|b, J

, M

is equal to the product of the Clebsch-Gordan coecient

J

, k, M

, q| J, M
with a quantity independent of M, M

and q.
8. The quantity independent of M, M

and q can depend upon what remaining parameters? Answer:

.
8. a, J, b, and J

, (and of course, the tensor

T
(k)
).
9. this quantity is usually written
_
a, J||

T
(k)
||b, J

_
/

2J + 1.
Use this notation to write out as fully as you can the matrix element of the operator

T
(k)
q
a, J, M|

T
(k)
q
|b, J

, M

= ?
by the Wigner-Eckart theorem.
9.
_
a, J||

T
(k)
||b, J

_
J

, k, M

, q| J, M /

2J + 1.
10. Review: the position operator z is a standard component of an operator such that k = and q =
.
10. k = 1, q = 0.
11. Consider a system whose total angular momentum is j = 4 and whose z-component is m = 4. Write out
an expression for the matrix element of the position operator z from this state to a state |J, M for arbi-
trary J, and M. (Drop the other quantum numbers a and b, and assume that the quantity multiplying
the Clebsch-Gordan coecient is unity in this case.) Answer: .
11. J, M| z |4, 4 = 4, 1, 4, 0| J, M,
since the intial state is |4, 4 and since k = 1 and q = 0 in the case of the operator z.
12. Let us try that once more, just for practice. If the quantum mechanical system is such that by strange
coincidence,
a, J||position||b, J =

2J + 1,
then the matrix element of the position operator z from state (b arbitrary, j = 4, m = 3) to the state
(a arbitrary, j = J, m = M) is .
12. a, J, M| z |b, 4, 3 = 4, 1, 3, 0| J, M .
5
You can see that operating on |j, m with operator z is something like adding a certain amount of
angular momentum to the state, namely, adding an angular momentum having j = 1 and m = 0.
13. Loosely speaking, the Wigner-Eckart theorem says that the operation of

T
(k)
q
on |J

is something
like adding to the state an angular momentum having j = ? and m = ?
13. j = k and m = q.
14. Use whatever reference material that you need to nd the ratio
5, 2| z |4, 2
5, 1| z |4, 1
= ? (a number).
14.
Since each matrix element has the same value of initial and nal J-value (4 and 5, respectively) the
quantity multiplying the Clebsch-Gordan coecients is the same for both and cancels out. The answer is
4, 1, 2, 0| z |5, 2
4, 1, 1, 0| z |5, 1
=
_
21/24 = 0.935.
15. The utility of the Wigner-Eckart theorem is that the angular dependence (m-dependence) of matrix
elements of an operator between states of given angular momentum is completely given by the Clebsch-
Gordan coecients once you have expressed the operator in terms of its .
15. standard components.
16. How much is 5, 2| z |4, 1? Answer: .
16. Zero,
since the initial m = 1, the added m-equivalent to z is zero, but the nal m =2.
17. The operator ( x i y)/

2 is the q = +1 standard component of position. What are the only possible

nal states |J, M having non-zero matrix elements of ( xi y)/

2 from initial state |4, 0? (There are

three of them.) .
17. |5, 1, |4, 1, |3, 1.
18. What nal states have non-zero matrix elements of

T
(2)
1
, whatever that is, from initial state |4, 4.
(There are two of them.) .
18. |5, 5, |6, 5.
You cannot have |4, 5 or |3, 5 since J cannot be less than M.
19. Complete the following: The Wigner-Eckart theorem: If

T
(k)
q
is the of
an tensor operator

T
(k)
, then a, J, M|

T
(k)
q
|b, J

, M

= f(a, b, J, J

,

T
(k)
) .
19. standard component, irreducible, J

, k, M

, q| J, M.
6
This is the end of Section 2. if you stop here, try to state the Wigner-Eckart theorem without looking
before you state again.
Section 3. Proof of the Wigner-Eckart Theorem for Scalar Operators.
Let

S be a scalar operator, invariant under rotation; that is

S commutes with

J
+
,

J

,

J
z
.
1. Thus

S is an irreducible tensor having k = and q = .
1. 0,0.
2. The Wigner-Eckart theorem implies that a, J, M|

S |a

, J

, M

= 0 unless J = and M = .
2. J

, M

.
3. Furthermore, the Wigner-Eckart theorem says that A, J, M|

S |a

, J, M is independent of .
3. M.
To prove this, rst show that

S|a

, J

, M

is a state of denite angular momentum with eigenvalues J

and M

, thus allowing you to write

S|a

, J

= |a

, J

, M

.
4. I think that you can do it without help. Give it a try.
4.

J
2

S|a

, J

, M

=

S

J
2
|a

, J

, M

= J

(J

+ 1)|a

, J

, M

J
s

S|a

, J

, M

=

S

J
z
|a

, J

, M

= M

S|a

, J

, M

.
Second, you clam that a, J, M| a

, J

, M

= 0 unless J

= J and M

= M (obviously). And last, you claim

that a, J, M| a

, J, M is independent of M, as we shall see next.

5. Prove the fact that a, J, J| a

, J, J equals a, J, J 1| a

, J, J 1 from the fact that

|b, J, J =
_
J(J + 1) J(J 1)|b, J, J 1,
and the fact that

J
+

J

|b, J, J = 2J|b, J, J,
(which you can easily derive.)
5.

|a

, J, J =

2J|a

, J, J 1
|a

, J, J 1 =

J

|a

, J, J

2J
Thus
a, J, J 1| a

, J, J 1 =
a, J, J|

J

|a

, J, J
2J
= a, J, J| a

, J, J .
By the same method, you can easily show that a, J, J| a

, J, J = a, J, M| a

, J, M for any allowed value

of M.
7
So, the eect of

S is to leave the quantum numbers J and M unchanged, although other quantum
numbers may be aected (a a

). And the matrix elements of

S are independent of M.
Now test yourself on a separate sheet of paper, state and prove the Wigner-Eckart theorem for scalar
operators. you can check your proof either from the preceding pages or by consulting a textbook.
Section 4. Proof of the Wigner-Eckart Theorem.
Here are three facts that you need.
Fact 1.

J
+
|a, J, M = .

|a, J, M = .

J
z
|a, J, M = .
Fact 2. If

T
(k)
q
is a standard component of an irreducible tensor operator, then
_

J
+
,

T
(k)
q
_
= .
_

,

T
(k)
q
_
= .
_

J
z
,

T
(k)
q
_
= .
Fact 3. The following mess simplies as shown:
_
j
1
(j
1
+ 1) m
1
(m
1
1) j
1
, j
2
, m
1
1, m
2
| j
3
, m
3

+
_
j
2
(j
2
+ 1) m
2
(m
2
1) j
1
, j
2
, m
1
, m
2
1| j
3
, m
3
=?
Facts 2 and 3 you can nd in the Appendices. You may remember Fact 1, especially if I remind you
that the coecients are similar to those in Fact 2. So, go ahead and ll in the blank spaces above.
How do we start the proof? For the case of scalar operator, the rst thing we did was show that

S|a, J, M is an eigenstate of

J
2
and

J
z
. But

T
(k)
z
|a, J, M is not necessarily such an eigenstate. We start
by showing that the state
F =

,q

T
(k)
q
|a

, J

, M

, k, M

, q| J

, M

is an eigenstate of

J
2
and

J
z
.
1. Instead of proving this using

J
2
and

J
z
directly, it will be better to use

J
+
and

J

, and

J
z
. Use Facts 1
and 2 to express

J
+
F as a sum involving

T
(k)
q
|a

, J

, M

as a factor in the summand, with a coecient

which cries out longingly for use of Fact 3.
1.
This answer is interrupted by lines where you can stop reading, use the hints you have picked up, and
complete your own proof.

J
+
F =

,q

J
+

T
(k)
q
|a

, J

, M

, k, M

, q| J

, M

,q

T
(k)
q+1
|a

, J

, M

_
k(k + 1) q(q + 1) J

, k, M

q| J

, M

,Q

T
(k)
q

J
+
|a

, J

, M

, k, M

, q| J

, M

.
8

J
+
F =

,q

T
(k)
q+1
|a

, J

, M

_
k(k + 1) q(q + 1) J

, k, M

, q| J

, M

,q

T
(k)
q
|a

, J

, M

+ 1
_
J

(J

+ 1) M

(M

+ 1) J

, k, M

q| J

, M

.
In the rst sum replace q by q 1.
In the second sum replace M

by M

1.

J
+
F =

,q

T
(k)
q
|a

, J

, M

_
_
k(k + 1) q(q 1) J

, k, M

, q 1| J

, M

+
_
J

(J

+ 1) M

(M

1) J

, k, M

1, q| J

, M

_
.
Thus,

J
+
F =
_

M

,q

T
(k)
q
|a

, J

, M

, k, M

, q| J

, M

+ 1
__
J

(J

+ 1) M

(M

+ 1).
2. The last line of the argument above says that

J
+
F equals the quantity
_
J

(J

+ 1) M

(M

+ 1) times an operator which, from its expression as a summation, one can see
is the one called except that M

is replaced by .
2. M

+ 1
This suggests that the state F is an eigenstate of denite angular momentum having quantum numbers
J

and M

. A complete proof would require us to show that

J

and

J
z
have the proper eect upon F as
well. You can check this yourself.
An outline of the proof of the Wigner-Eckart Theorem is given in Appendix D. Turn now to this
appendix.
3. How far have we gotten on Appendix D with the argument presented so far? Answer: Step .
3. (A)
4. Step (B) relies on a property of the Clebsch-Gordan coecients which is stated in Appendix
under the title .
4. B, orthogonality relations.
At this point we can see the entire strategy of the proof. We are interested in the matrix elements of

T
(k)
q
between states of denite angular momentum. We rst let

T
(k)
q
operate on a state of denite angular
momentum, so we need an expansion of the result in terms of states of denite angular momentum.
5. The required expansion is labeled in Appendix D.
5. (3).
6. What happened to the summation in going from Eq. 3 to Eq. 4? (Justify the step.)
6. a, J, M

| a

, J

, M

= 0 unless J = J

and M = M

.
9
Lets review the basic ideas of this proof. We are concerned with the matrix elements of an (irreducible
tensor) operator between states of denite angular momentum: A, J, M|

T
(k)
q
|a

, J

, M

. It would be natural
to begin by considering

T
(k)
q
|a

, J

, M

. We need next the product of this state with the nal state |a, J, M.
What easier way is there to evaluate this product than rst to expand the expression above in terms of a
basis of states having denite angular momentum. It was a little tricky of us to start out with the state
we called F and invert its expansion to get the expansion we wanted by use of the orthogonality relations
governing the expansion coecients. (We had rst, of course, to show that F as we dened it was a state
of denite angular momentum.) After that the trick worked and we had our expansion of

T
(k)
q
|a

, J

, M

.
The product with |a, J, M was then trivially done. The nal step was to show that a, J, M| a

, J, M is
independent of M.
If you want to prepare yourself to reproduce this proof, now is the time to try to write it out. This is
about as far as a programmed text can conveniently go.
10
Appendix A Addition of Angular Momenta.
|J, M =
j
1

m
1
=j
1
j
2

m
2
=j
2
|j
1
, j
2
, m
1
, m
2
j
1
, j
2
, m
1
, m
2
| J, M .
We will take j
1
= 4, j
2
= 1, J = 5. We will not display the cases of J = 4 and J = 3.
|5, 5 = + |4, 1, 4, 1
|5, 4 = +
_
1
5
|4, 1, 4, 0 +
_
4
5
|4, 1, 3, 1
|5, 3 =
_
1
45
|4, 1, 4, 1 +
_
16
45
|4, 1, 3, 0 +
_
28
45
|4, 1, 3, 1
|5, 2 =
_
3
45
|4, 1, 3, 1 +
_
21
45
|4, 1, 2, 0 +
_
21
45
|4, 1, 1, 1
|5, 1 =
_
6
45
|4, 1, 2, 1 +
_
24
45
|4, 1, 1, 0 +
_
15
45
|4, 1, 0, 1
|5, 0 =
_
10
45
|4, 1, 1, 1 +
_
25
45
|4, 1, 0, 0 +
_
10
45
|4, 1, 1, 1
|5, 1 =
_
15
45
|4, 1, 0, 1 +
_
24
45
|4, 1, 1, 0 +
_
6
45
|4, 1, 2, 1
|5, 2 =
_
21
45
|4, 1, 1, 1 +
_
21
45
|4, 1, 2, 0 +
_
3
45
|4, 1, 3, 1
|5, 3 =
_
28
45
|4, 1, 2, 1 +
_
16
45
|4, 1, 3, 0 +
_
1
45
|4, 1, 4, 1
|5, 4 =
_
4
5
|4, 1, 3, 1 +
_
1
5
|4, 1, 4, 0
|5, 5 = |4, 1, 4, 1
Appendix B. Orthogonality and Recursion Relations.
Orthogonality Relations:

m1,m2
j
1
, j
2
, m
1
, m
2
| J, M j
1
, j
2
, m
1
, m
2
| J

, M

=
J,J

M,M
. (1)

J,M
j
1
, j
2
, m
1
, m
2
| J, M j
1
, j
2
, m

1
, m

2
| J, M =
m1,m

m2,m

2
. (2)
Recursion Relations:
_
J(J + 1) M(M + 1) j
1
, j
2
, m
1
, m
2
| J, M + 1
=
_
j
1
(j
1
+ 1) m
1
(m
1
1) j
1
, j
2
, m
1
1, m
2
| J, M
+
_
j
2
(j
2
+ 1) m
2
(m
2
1) j
1
, j
2
, m
1
, m
2
1| J, M (3)
_
J(J + 1) M(M 1) j
1
, j
2
, m
1
, m
2
| J, M 1
=
_
j
1
(j
1
+ 1) m
1
(m
1
+ 1) j
1
, j
2
, m
1
+ 1, m
2
| J, M
+
_
j
2
(j
2
+ 1) m
2
(m
2
+ 1) j
1
, j
2
, m
1
, m
2
+ 1| J, M (4)
Appendix C. Commutation of position and angular momentum in cartesian coordinates.
11
We will take h = 1 for simplicity.
From

J
x
= i
_
y

z
z

y
_
,

J
y
= i
_
z

x
x

z
_
,

J
z
= i
_
x

y
y

x
_
,
and

J
+
=

J
x
+ i

J
y
,

J

=

J
x
i

J
y
,
it is easy to show that
[

J
+
, x] = z [

J

, x] = z [

J
z
, x] = i y
[

J
+
, y] = i z [

J

, y] = i z [

J
z
, y] = i x
[

J
+
, z] = x i y [

J

, z] = x i y [

J
z
, z] = 0
Appendix D. Proof of the Wigner-Eckart Theorem Outline.
Dene
F =

,q

T
(k)
q
|a

, J

, M

, k, M

, q| J

, M

. (1)
(A) The state F is an eigenstate of

J
2
and

J
z
with eigenvalues J

(J

+ 1) and M

. Change notation,
F |a

, J

, M

.
(B) Since the Clebsch-Gordan coecients dene an orthogonal transformation
|a

, J

, M

,q

T
(k)
q
|a

, J

, M

, k, M

, q| J

, M

(2)
implies that

T
(k)
q
|a

, J

, M

,M

|a

, J

, M

, k, M

, q| J

, M

, (3)
(C) Thus,
a, J, M|

T
(k)
q
|a

, J

, M

= a, J, M| a

, J, M J

, k, M

, q| J, M . (4)
(D) But,
a, J, M| a

, J, M = a, J, J| a

, J, J , (5)
for all allowed values of M.
Wigner-Eckart Theorem
a, J, M|

T
(k)
q
|a

, J

, M

= a, J, J| a

, J, J
. .
independent of
M,M

orq
J

, k, M

, q| J, M . (6)
12