Properties of Commutator
Properties of Commutator
Power Round
Introduction (0 pts)
The Algebra of Noncommutative Operators: In this power round, we will consider the algebra of noncommutative operators. We will define operators as any objects that satisfy the properties below. We will use
bold-faced, upper case letters, such as A, to denote operators and lower case letters to denote complex
numbers. We can add operators, multiply two operators, and multiply operators by numbers. Almost all
of the properties that we take for granted for real numbers hold for operators. The only exception is that
AB 6= BA where A, B are operators. Also you cannot add a number and an operator. In particular, the
following properties hold:
Suppose that a, b, ... are complex numbers and A, B, ... are operators. You may use the following properties without explicitly stating them:
(a) A + B = B + A
(b) (A + B) + C = A + (B + C).
(c) A(BC) = (AB)C.
(d) A(B + C) = AB + AC, (A + B)C = AC + BC.
(e) (a + b)C = aC + bC.
(f) a(B + C) = aB + aC
(g) a(bC) = (ab)C.
(h) A(bC) = b(AC) = (bA)C.
(i) There is a zero operator, often written as 0 such that 0 + A = A for all operators A.
(j) 0A = 0A = A0 = 0 (where 0 is the zero operator and 0 is the complex number zero).
(k) Given a operator, A, there exists another operator, denoted by A such that A + (A) = 0.
(l) There exists a multiplicative identity for operators called the identity operator, often written as
I. The identity operator is such that IA = AI = A for all operators A.
Commutator : As discussed above, it is not generally true that AB = BA for operators. Therefore, it
can be useful to consider the quantity, AB BA, which is not generally zero. This quantity is called the
commutator of A and B and is written as follows
[A, B] = AB BA
(1)
(2)
(3)
Solution:
(a) [A, A] = AA AA = 0 and [A, B] = AB BA = (BA AB) = [B, A].
(b) B[A, C] + [A, B]C = B(AC CA) + (AB BA)C = ABC BCA = [A, BC] and b[A, B] +
c[A, C] = b(AB BA) + c(AC CA) = A(bB + cC) (bB + cC)A = [A, bB + cC]
1
BMT 2013
Power Round
(c) [A, [B, C]] = ABC ACB BCA + CBA = (ABC BCA) + (CBA ACB). Summing
cyclically gives the result.
P2 (3 pts) Suppose that [A, B] = 0. If n is a positive integer, prove that [An , B] = 0.
Solution: Proceed by induction. The base case, n = 1, is given. Then using the inductive hypothesis,
we get
[An , B] = [An1 , B]A + An1 [A, B] = 0
(4)
For your later reference, it is also true that
[AB, C] = A[B, C] + [A, C]B
(5)
Angular Momentum Algebra: An ordered triple of three operators, (J1 , J2 , J3 ) = ~J, are said to form an
angular momenum algebra if
[J1 , J2 ] = iJ3
[J2 , J3 ] = iJ1
[J3 , J1 ] = iJ2
(6)
[J3 , J ] = J
[J+ , J ] = 2J3
(7)
(b) (3 pts) Define J2 = (J1 )2 + (J2 )2 + (J3 )2 . Prove that [J2 , Ji ] = 0 for i = 1, 2, 3.
Solution:
(a) [J3 , J ] = [J3 , J1 ] i[J3 , J2 ] = iJ2 i(1)iJ1 = (J1 iJ2 ) which proves the first two identities.
Next, [J+ , J ] = [J1 + iJ2 , J1 iJ2 ] = [J1 , J1 ] + (i)[J1 , J2 ] + (i)[J2 , J1 ] + [J2 , J2 ] = 2J3
(b) By cyclic symmetry, it suffices to consider i = 1. In this case, [J21 + J22 + J23 , J1 ] = 0 + [J22 , J1 ] +
[J23 , J1 ] = J2 [J2 , J1 ]+[J2 , J1 ]J2 +J3 [J3 , J1 ]+[J3 , J1 ]J3 = J2 (iJ3 )+(iJ3 )J2 +J3 (iJ2 )+iJ2 J3 =
0
P4 (4 pts) Suppose that
J2 A = A
J3 A = A
(8)
J3 (J A) = ( 1)(J A)
(9)
(10)
(11)
BMT 2013
Power Round
Operator Exponentials: Just as we can take a real number x, and then form a new real number, ex , we can
define the exponential of an operator using a similar formula:
ex = 1 + x +
k=0
k=0
X xk
X Ak
x2
x3
A2
A3
+
+ ... =
exp(A) I + A +
+
+ ... =
2!
3!
k!
2!
3!
k!
(12)
X
2
2
Y
1
I+
+
2
2!
Y
2
2
X
1
I+
+
2
2!
X
2
2
Y
1
I+
+
2
2!
(16)
(No proof necessary). Solution: Expanding out straightforwardly gives (a, b, c, d, e, f, g) = (1, 1, 1, 1/2, 3/4, 1/4, 1/2).
We call this finding the expansion of M2 to second order because we are finding the coefficients of all
terms with degree less than equal to degree 2.
P7 (7 pts) Suppose that n is a positive integer. Find, with proof, the expansions of Mn and exp(X + Y)
to second order and show that their difference goes to zero if we let n go to infinity.
Solution: First, consider Mn . There are 2n factors to expand out. Zeroth order is 1I. For first
2
order, there are n ways to choose one X
n so we get 1X. Same with 1Y. Let us consider X . There
n
(
)
2
are two ways to do this. First, we can pick two factors of X
n for a coefficient of n2 . The other way
n
is to choose one factor of X2 . This can be done in n ways to get n2!2 . This gives a total coefficient of
1 2
1 2
the jth factor where
2 X . The same argument gives 2 Y . For XY, suppose that we choose Y from
P
j = 1, 2, . . . n. Then there are j ways to choose a X. So the total coefficient is
n
j=1
n2
1
2
1
+ 2n
. Finally
1 Technically, polynomials have only a finite number of terms. The appropriate description of what we are doing is a formal
power series expansion, but we will not be concerned with this distincition.
BMT 2013
Power Round
n1
j=1
n2
1
2
1
2n .
XYYX
,
2n
P8 (7 pts) Find, with proof, the coefficient of X10 Y10 in the expansion of M2 .
Solution: Two cases. First, suppose that when we expand, we take at least one factor of Y from the
first factor. Then no factors of X can be taken from the second factor with Xs. Thus we can ignore
the Xs from the second factor. If we multiply exp(Y/2) exp(Y/2), we get exp(Y). The coefficient of
1
. The coefficient from X10 is 210110! . Second, We take all of our factors of Y from the second
Y10 is 10!
Y factor. Then all of the X factors come from the first two terms, and we use the same argument as
211 1
X10 Y 10
before. We double counted a term 10!2
.
10 10!210 It follows that the answer is
220 (10!)2
k
P9 (12 pts) In the polynomial expansion of Mn , find, with proof, the coefficient of (XY) = XYXY . . . XY
where k is a positive integer.
Solution: It is sufficent to consider the expansion of ((I + X)(I + Y))n because there are no powers
of X2 in the term that we want the coefficient of. In the end, we will multiply by n12k to account for
X and Y being divided by n. When we expand out the expression above to get the desired coefficient,
we choose which factor to pull out X and Y. Let i1 be the factor that the first Y comes from. i2 is the
number of factors after that factor to get the next factor of Y, etc, until ik . Given these placements of
Ys, there are i1 . . . lk ways to choose the Xs. Call ik+1 = n i1 ... ik 0 Thus the coefficient
we want is
k
Y
X
ij
(17)
Pk+1
j=1
ij =n j=1
(18)
This is the n kth coefficient of (1 x)(2k+1) , which is given by the binomal series:
n+k
nk (2k + 1)
nk (2k 1)(2k 2) . . . (2k 1 n + k + 1)
(1)
= (1)
=
(19)
nk1
(n k)!
2k
It follows that the final answer for the coefficient is
n+k
2k
n2k
In this section, we will show that we can rotate vectors about axes using our noncommutative operators.
First, let us explain the geometric problem: suppose that we are given the components of some vector ~a and
we rotate that vector about the n
axis an angle . Call the resulting vector ~b. What are the components of
~b?
BMT 2013
Power Round
P3
2. [Ji , Xj ] = i k=1 i,j,k Xk . Note the i in front of the sum is 1 and the subscript i is 1, 2, 3.
3. X1 , X2 , X3 are linearly independent.
In order to use the commutators, let us first associate the vectors ~a = (a1 , a2 , a3 ) and so on with operators:
~a = (a1 , a2 , a3 )
n
= (n1 , n2 , n3 )
~b = (b1 , b2 , b3 )
~ = a1 X1 + a2 X2 + a3 X3
~a X
n
~J = n1 J1 + n2 J2 + n3 J3
~b X
~ = b1 X1 + b2 X2 + b3 X3
(20)
Let us begin with some problems. Be sure to use the definition of the exponential of an operator.
P10 (7 pts) Prove that exp((s + t)A) = exp(sA) exp(tA).
Solution: We expand using the definition of the operator exponential
exp(sA) exp(tA) =
X
m=0
mA
m X
m!
n=0
nA
n!
m+n
m
(m + n)!
m=0 n=0
Am+n sm tn
(21)
k
l
k!
Ak sl tkl =
X
Ak
k=0
k!
(22)
BMT 2013
Power Round
(a) (7 pts) Suppose that B0 = B and that Bn+1 = [A, Bn ] for n 0. Prove the following lemma
for N 1.
N
X
N
[AN , B0 ] =
Bj AN j
(23)
j
j=1
Solution: Proceed by induction. N = 1 gives [A, B0 ] = B1 , which is clear. Now,
[AN +1 , B0 ] = A[AN , B0 ] + [A, B0 ]AN
(24)
N
N
X
X
N
N
N j
N
=A
Bj A
+ B1 A =
([A, Bj ] + Bj A)AN j + B1 AN
j
j
j=1
j=1
(25)
N
N
X
X
N
N
N j
Bj+1 A
+
Bj AN +1j + B1 AN
=
j
j
j=1
j=1
(26)
N
+1
X
j=2
N
X
N
N
Bj AN +1j +
Bj AN +1j + B1 AN
j1
j
j=1
N
X
N
N
+
Bj AN +1j + N B1 AN + B1 AN
= BN +1 +
j
1
j
j=2
(27)
(28)
Then using Pascals identity and collecting the other terms back into the sum gives the desired
result.
(b) (4 pts) Prove that2
[exp(A), B0 ] =
X
Bj
j=1
j!
exp(A)
(29)
N
X
X
1
1 X N
Bj AN j =
[exp(A), B0 ] =
Bj AN j
N ! j=1 j
(N
j)!j!
j=1
N =0
X
j=1
(30)
N =j
Bj
X
M =0
X Bj
1
AN =
exp(A)
M !j!
j!
j=1
(31)
X
Bj
j=0
j!
(32)
Solution: First, note that exp(A) exp(A) = exp(0) = I with s = 1, t = 1 from the identity
in the start of this section. Using the previous equation, we multiply by both sides on the right
by exp(A) and then
[exp(A), B0 ] exp(A) = exp(A)B exp(A) B0 exp(A) exp(A)
Moving B0 to the right gives us the result.
2 Ignore
issues of convergence.
(33)
BMT 2013
Power Round
(34)
where (a1 , a2 , a3 ) = (x1 cos x2 sin , x1 sin + x2 cos , x3 ). The following identities might be helpful:
cos = 1
2
4
6
+
+ ...
2!
4!
6!
sin =
3
5
7
+
+ ...
3!
5!
7!
(35)
X
Bk
1
2
(2 )2
= x3 X3 + (x1 X1 + x2 X2 )
+
+ . . . = x3 X3 + (x1 X1 + x2 X2 ) cos (36)
k!
0!
2!
4!
k=0,even
Likewise,
X
k=0,odd
Bk
= (x1 X2 x2 X1 )
k!
3
5
+
+ ...
1!
3!
5!
= (x1 X2 x2 X1 ) sin
(37)
X
k=2
~ak ~
X
k!
(38)
~ means v1 X1 + v2 X2 + v3 X3 .
where ~ak = n
~ak1 and ~a0 = ~a and the dot product ~v X
(b) (6 pts) Prove that
X
ak
k~
|~a|2 e||
k!
(39)
k=2
where |~v | =
Solution:
(a) Using the Hadamard Lemma gives a series where
~b X
~ = ~a X
~ + [i
~ + ...
n ~J, ~a X]
(40)
j=1
i=1 j=1
i=1 j=1
(41)
k=1
~ =
~ In
Evaluating the sum and using the definition of i,j,k gives us that [i
n ~J, aX]
n ~a X.
other words, taking that commutator is the same as taking the cross product with
n. Plugging
this into the Hadamard lemma gives the result.
7
BMT 2013
Power Round
(b) Since n
is a unit vector, we have
|~ak | = |
n ~ak1 | |
n||~ak1 | |~ak1 |
(42)
Repeating this inductively gives |~ak | |~a|. Now we use the triangle inequality to prove that the
LHS is less than or equal to
X
||k2
(43)
|~a|2
k!
k=2
X
||k2
k=2
k!
X
k=0
X ||k
||k
= e||
(k + 2)!
k!
(44)
k=0
[Ai , Yj ] = 0
[Xi , Bj ] = 0
[Xi , Yj ] = 0
(45)
Define the total angular momentum Ji = Ai + Bi for i = 1, 2, 3. We will consider how J generates (not
necessarily vector) rotations on the nine objects Ti,j = Xi Yj where i, j {1, 2, 3}. To study this question,
we will look at [Ji , Tj,k ]. In a sense, we will recombine these nine objects and then divide them into three
types. One type that doesnt change at all under a rotation, one that behaves like a vector under rotation,
and one with a more general behavior.
P15 Suppose that Ji = Ai + Bi for i = 1, 2, 3.
P3
(a) (3 pts) Prove that [Ji , Jj ] = i k=1 i,j,k Jk .
(b) (3 pts) Prove that J1 , J2 , J3 are linearly independent using the fact that the As and the Bs are
linearly independent (see the definition of linearly independent in section 2).
Solution:
8
BMT 2013
Power Round
~ and B
~ are angular momentum algebras and the given commutation relations,
(a) Since A
[Ji , Jj ] = [Ai + Bi , Aj + Bj ] = [Ai , Aj ] + [Ai , Bj ] + [Bi , Aj ] + [Bi , Bj ]
i
3
X
i,j,k Ak + 0 + 0 + i
k=1
3
X
i,j,k Bk = i
k=1
3
X
i,j,k Jk
(46)
(47)
k=1
(48)
Taking the commutator of both sides with A1 gives us c2 A3 c3 A2 = 0. Therefore by the linear
independence of the As, we get c2 = c3 = 0. Repeating with say A2 gives c1 = 0 as well.
Transforming like a rank k Spherical Tensor : A collection of 2k + 1 operators Tk ,Tk+1 ,...,Tk is said to
transform like a rank k spherical tensor under ~J if the following are true:
1. (J1 , J2 , J3 ) forms an angular momentum algebra.
2. [J3 , Tq ] = qTq for q = k, k + 1, ..., k.
p
3. [J , Tq ] = (k q)(k q + 1)Tq1 for q = k, k + 1, ..., k (take all upper signs or all lower signs).3
4. Tk ,Tk+1 ,...,Tk are linearly independent.
where J = J1 iJ2 .
P16 (4 pts) Prove that
~ Y
~ X1 Y1 + X2 Y2 + X3 Y3
T0 = X
(49)
transforms like a rank 0 tensor under ~J. In other words, prove that [J , T0 ] = 0 where is replaced
with 3, +, .
Solution: Observe that it is equivalent to prove this with = 1, 2, 3. Then compute:
3
3
3
3 X
3
X
X
X
X
Ji ,
Xj Yj =
[Ji , Xj Yj ] =
Xj [Ji , Yj ]+[Ji , Xj ] Yj = i
i,j,k (Xj Yk +Xk Yj ) (50)
j=1
j=1
j=1
j=1 k=1
Now we note that i,j,k = i,k,j , split up the sums, and interchanging the dummy variables j, k in the
second sum.
=i
3 X
3
X
i,j,k Xj Yk + i
j=1 k=1
3 X
3
X
i,j,k Xk Yj = i
j=1 k=1
=i
3 X
3
X
3 X
3
X
j=1 k=1
i,j,k Xj Yk + i
j=1 k=1
3 X
3
X
i,j,k Xj Yk + i
3 X
3
X
i,k,j Xk Yj
(51)
j=1 k=1
i,j,k Xj Yk = 0
(52)
j=1 k=1
Span: Let M1 , M2 , .., Mk be operators. Then the span of M1 , . . . , Mk is the set of all elements of the form
c1 M1 + c2 M2 + . . . + ck Mk where c1 , . . . , ck are complex numbers.
P17 (a) (4 pts) Find three operators T1 , T0 , T1 in Span(X1 , X2 , X3 ) such that both of the following
are true:
i. Tq where q = 1, 0, 1 transforms like a spherical tensor of rank 1
p
for + and q = k, we have [J+ , Tk ] = (k k)(k + k + 1)Tk+1 = 0 so we dont have to worry about Tk+1 technically
not being defined and we just say that [J+ , Tk ] = 0. Similarly with and q = k.
3 Note
BMT 2013
Power Round
X1 iX2
T0 = X3
(54)
To finish the question, we just need to verify the first and second property for q = 1, k = 1.
Lastly, we need to verify linear independence.
c1 (1)
X1 + iX2
X1 iX2
+ c0 X3 + c1
=0
2
2
(55)
(56)
(57)
(58)
(59)
Correspondingly, we compute
i,j,k Zk = i,j,k k,l,m Xl Ym = (i,l j,m i,m j,l )Xl Ym = Xi Yj Xj Yi
(60)
(61)
P18 (a) (10 pts) Suppose that we have an operator, A such that [J3 , A] = kA and [J+ , A] = 0 where
2k is a non-negative integer. For the sake of notation, write Tk = A, and define Tq1 =
1
Tk1 = 0. Prove that the 2k + 1 matrices Tq for q = k, ..., k form a spherical tensor of rank
k. Hint: Use the Jacobi Identity to calculate commutators like [J3 , [J , Tq ]].
10
BMT 2013
Power Round
(b) (6 pts) Let V be the span of the nine operators Xi Yj where i, j {1, 2, 3}. Find five operators
in V that transform like a spherical tensor of rank 2.
Solution:
(a) The part of the relations with the J are satisfied by construction. Now lets look at the J3
identities. Let us prove by induction that [J3 , Tq ] = qTq going from q = k down to q = k. The
1
base case q = k is given. Now notice that by the Jacobi identity, for =
(k+q)(kq+1)
(62)
(64)
(66)
p
Then noting that (k q)(k + q + 1) + 2q = (k + q)(k q + 1) =
(k + q)(k q 1) =
p
(k (q 1))(k + (q 1) + 1), we get the result.
Pk
Lastly, we must prove linear independence. Suppose that q=k cq Tq = 0. Suppose that Q is
the largest integer such that cQ 6= 0. Now take commutators of both sides with J Q + k times.
This gives us cQ (constants)Tk = 0. But it is assumed that Tk 6= 0, so cQ = 0, a contradiction.
Thus all cq = 0.
(b) Use the construction in the previous problem to take (X1 , X2 , X3 ) to a spherical tensor of rank
~ Now we use the first part of this question, and note
1: X0q for q = 1, 0, 1. Do the same with Y.
that
[J3 , X01 Y10 ] = [A3 , X01 ]Y10 + X01 [A3 , Y10 ] + [B3 , X01 ]Y10 + X01 [B3 , Y10 ] = 2X01 Y10
(67)
Thus our A = X01 Y10 . To finish, we just need to repeatedly apply J . Then we get
1
0
T0 = (X01 Y1
+ 2X00 Y00 + X01 Y10 )
6
1
T1 = (X01 Y00 + X00 Y10 )
2
1
0
T1 = (X01 Y00 + X00 Y1
)
2
0
T2 = X01 Y1
(68)
(69)
Then the final answers follow from plugging in the following relations (and with X Y).
X01 =
X1 iX2
X0 = X3
Operator Space: We will call V a operator space if the following are true:
11
(70)
BMT 2013
Power Round
This section was not included in the original contest. Note that an explicit example of operators
are n by n matrices. This section has one do some computations.
Hint: Begin by computing A2 = AA, A3 = AAA, etc.
1. (2 pts) Find exp(A) where
0
0
A=
0
0
1
0
0
0
0
0
1
0
0
1
0
0
(71)
Note: if you found that A2 = A, you are not multiplying the operators correctly and you should see
the appendix to make sure that you understand how to multiply operators.
Solution: Computing powers of A, we see that A4
terms of the summation, which give us
1
0
exp(A) =
0
0
1
1
0
0
1
1
0
1
6
1
2
1
1
(72)
1
0
B=
0
0
0
1
0
0
0
0
1
0
1
0
0
1
(73)
0
1
0
0
0
0
1
0
n
0
0
1
(74)
1
0
n
B =
0
0
Now we sum. The diagonal entries, we get
e. Thus the answer is
1
k=0 k!
e
0
exp(B) =
0
0
12
0
e
0
0
0
0
e
0
e
0
0
e
k
k=0 k!
k
k=1 k!
(75)
BMT 2013
Power Round
1
0
C=
0
0
1
1
0
0
0
0
1
1
0
1
1
0
(76)
Solution: One can observe a pattern and use induction, however, here is a solution that offers a bit
more insight. Observe that C = I + A where A4 = 0. Thus
n
C = (I + A) =
n
X
n
j=0
where
n
j
3
X
n
A =
j=0
Aj
(77)
X
3 j
X
n A
n=0 j=0
n!
3
X
j=0
X
n=0
n
j
n!
(78)
Now,
X
n=0
n
j
n!
X
n=j
1
e
=
j! (n j)!
j!
(79)
e
0
exp(C) =
0
0
e
e
0
0
e/2
e
e
0
e/6
e/2
e
e
(80)
This section also did not appear in the actual contest. Here is a specific realization of an
angular momentum algebra in terms of matrices.
1. So far, we have considered angular momentum algebras to be abstract objects that we can manipulate
without refering to particular operators that satisfy the relations in Eq. (??). Now we will consider
an example. Define the operators
1
1
1
0 1
0 i
1 0
A1 =
A2 =
A3 =
(81)
1 0
i 0
0 1
2
2
2
2. (3 pts) Show that the operators A1 , A2 , A3 satisfy Eq. (??).
3. (3 pts) Find all 2 by 1 operators, v, with corresponding complex numbers 1 , 2 such that A2 v = 1 v
and A3 v = 2 v.
P3
4. Write exp(A)B exp(A) as i=1 ci Ai for the following cases:
(a) (2 pts) A = ixA3 , B = A1 .
(b) (4 pts) A = ixA1 , B = A2 .
13
BMT 2013
Power Round
This section was not included in the original contest. The focus of this section is that the
eigenvalues of the operators of an angular momentum algebra can only take on discrete values!
1. Matrix eigenvalues Suppose that we have a matrix, M . A column vector, v, is called an eigenvector of
M if and only if
v 6= 0
(82)
and there exists a complex number, , which is called the eigenvalue of v with respect to M , such that
M v = v
(83)
2. Introduction: Since J 2 and J3 commute, it can be shown that they have a simultaneous eigenvector.
Let va,b denote a simultaneous eigenvector of J 2 and J3 that is normalized. In particular, va,b 6= 0 is
a column vector that satisfies the following:
(va,b ) va,b = [1]
J 2 va,b = a va,b
J3 va,b = b va,b
(84)
(85)
7. (10 pts) Begin with the existence of va,b as in Eqs. (84) and Eq. (85). Then argue that there must
be a bmax and bmin such that the following are true:
J+ va,bmax = J va,bmin = 0
(86)
bmax = bmin + 2j
(87)
(88)
b {j, j + 1, . . . , j 1, j}
(89)
a = j(j + 1)
(90)
14