Advanced Quantum Mechanics: Fall 2017

Final Exam: Brief Solutions

NOTE: Sentences in italic fonts are questions to be answered.
Possibly Useful facts:



 +1, abc = xyz, or yzx, or zxy;
• abc ≡ abc
= bca = cab = −acb = −bac = −cba .

−1, abc = zyx, or xzy, or yxz; 


 0,

otherwise.

 1, a = b; P
abc cdf
δab ≡  0, a 6= b. c  = δad δbf − δaf δbd .

• Some Taylor expansions: exp(x) = ∞ xn x2 x3 4

P
n=0 n! = 1 + x + 2 + 6 + O(x ),
√
1 + x = 1 + 12 x − 81 x2 + O(x3 ), √1+x
1
= 1 − 21 x + 38 x2 + O(x3 ).
P+∞ 1
• Baker-Hausdorff formula: exp(Â) · B̂ · exp(−Â) = B̂ + n=1 n! [Â, [Â, . . . [Â, B̂] . . . ]].
| {z }
n-fold commutator

• Spin (angular momentum) operators satisfy [Ŝa , Ŝb ] = i abc Ŝc .

P
c

Ladder operators Ŝ± ≡ Ŝx ± iŜy , and [Ŝz , Ŝ± ] = ±Ŝ± , and
p
Ŝ± |S, Sz = mi = (S ∓ m)(S ± m + 1)|S, Sz = m ± 1i.

• Ŝ i · Ŝ j ≡ Ŝiz Ŝjz + Ŝix Ŝjx + Ŝiy Ŝjy = Ŝiz Ŝjz + 21 (Ŝi+ Ŝj− + Ŝi− Ŝj+ ).

• Spin-1/2: Ŝa = σa /2 under the Ŝz eigenbasis,


for a= x, y, z.
 

. {σa , σb } = 2δab 1.
0 1 0 −i 1 0
The Pauli matrices σa are σx =   , σy =   , σz =  
1 0 i 0 0 −1
exp(−iθn · σ) = cos(θ)1 − i sin(θ)(n · σ), for unit-length 3-component real vector n.
(here n · σ ≡ nx σx + ny σy + nz σz )
     
0 1 0 0 −i 0 1 0 0
     
• Spin-1: Ŝx = √1
2

1 0 1



, Ŝy = √1
2
i


0 −i


, Ŝz = 
0

0

0 

, under the Ŝz eigenbasis.
0 1 0 0 i 0 0 0 −1

• The D4 group: {(C4 )(n mod 4)

(σs )(m mod 2)
|C44 = σs2 = C4 σs C4 σs = 1}.
8 elements, 5 conjugacy classes,
1 2C4 C42 2σs 2σd
{1}, {C4 , C43 }, {C42 }, {σs , C42 σs }, and {C4 σs ≡ σd , C43 σs }.
Γ1 1 1 1 1 1
Character table for irreducible representations (irrep)
Γ2 1 1 1 −1 −1
Γ1,2,3,4,5 is given on the right,
Γ3 1 −1 1 1 −1
Γ4 1 −1 1 −1 1
Γ5 2 0 −2 0 0
1/13
 Problem 1. (20 points) Consider two spin-1 moments, Ŝ 1 and Ŝ 2 . They
= δi,j c i Ŝi,c (here a, b, c label x, y, z components), and
P abc
satisfy [Ŝi,a , Ŝj,b ]
2 2
Ŝ 1 = Ŝ 2 = 1 · (1 + 1) = 2. A complete orthonormal basis for the 9-dimensional
Hilbert space is the Ŝz basis, |s1 , s2 i. Here si = 1, 0, −1 are eigenvalues of Ŝi,z for i = 1, 2
respectively. The matrix elements of Ŝi,a for i = 1, 2 and a = x, y, z are given on page 1.

(a) (10pts) Write down the eigenvalues and normalized eigenstates (in terms of Ŝz
basis) of Ĥ0 = −J · Ŝ 1 · Ŝ 2 . Here J > 0. [Hint: Ĥ0 is related to (Ŝ 1 + Ŝ 2 )2 ]

(b) (10pts) The full Hamiltonian is Ĥ = Ĥ0 + D · [(Ŝ1,z )2 + (Ŝ2,z )2 ]. D is a real “small”
parameter. Solve the energy eigenvalue(s) of the ground state(s) of Ĥ to second order of
D. [Hint: the unperturbed ground states of Ĥ0 are degenerate, but you may not need to
use degenerate perturbation theory due to some symmetry]

Solution. This is similar to homework #6 Problem 1.

(a) This is the same as homework #6 Problem 1(a).
Define “total spin” Ŝ 1+2 = Ŝ 1 + Ŝ 2 . It’s easy to check that different components of Ŝ 1+2
2
satisfy the commutation relations of angular momentum, and [Ŝ 1+2 , Ŝ1+2,z ] = 0. Define
ladder operators Ŝ1+2,± = Ŝ1,± + Ŝ2,± .
Ĥ0 = − J2 (Ŝ 1 + Ŝ 2 )2 + J2 (Ŝ 1 )2 + J2 (Ŝ 2 )2 = − J2 (Ŝ 1 + Ŝ 2 )2 + J
2
· 1 · (1 + 1) + J
2
· 1 · (1 + 1)
2
= − J2 (Ŝ 1 + Ŝ 2 )2 + 2J. So the eigenstates of Ĥ0 can be chosen as the eigenstates of Ŝ 1+2
and Ŝ1+2,z , |S1+2 , S1+2,z i, with S1+2 = 0 or 1 or 2, and S1+2,z = −S1+2 , −S1+2 + 1, . . . , S1+2 .
|S1+2 , S1+2,z i are linear combinations of |s1 , s2 i with s1 + s2 = S1+2,z .
For each S1+2 , first solve |S1+2 , S1+2,z = S1+2 i in terms of Ŝz basis by the fact that
Ŝ1+2,+ |S1+2 , S1+2,z = S1+2 i = 0, then use the lowering ladder operator to generate other
S1+2,z states.
S1+2 = 2 states are eigenstates of Ĥ0 with eigenvalue − J2 · 2 · 3 + 2J = −J,
|S1+2 = 2, S1+2,z = 2i = |1, 1i;
S1+2,−
|S1+2 = 2, S1+2,z = 1i = √
4
|S1+2 = 2, S1+2,z = 2i = √12 (|0, 1i + |1, 0i);
|S1+2 = 2, S1+2,z = 0i = S1+2,−
√
6
|S1+2 = 2, S1+2,z = 1i = √16 (| − 1, 1i + 2|0, 0i + |1, −1i);
|S1+2 = 2, S1+2,z = −1i = S1+2,−
√
6
|S1+2 = 2, S1+2,z = 0i = √12 (| − 1, 0i + |0, −1i);
|S1+2 = 2, S1+2,z = −2i = S1+2,−
√
4
|S1+2 = 2, S1+2,z = −1i = | − 1, −1i.

2/13
 Suppose |S1+2 = 1, S1+2,z = 1i = c1 |1, 0i + c2 |0, 1i, then by 0 = Ŝ1+2,+ |S1+2 = 1, S1+2,z =
√
1i = (Ŝ1,+ + Ŝ2,+ )(c1 |1, 0i + c2 |0, 1i) = 2(c1 + c2 )|1, 1i, we have c2 = −c1 . The normalized
state |S1+2 = 1, S1+2,z = 1i = √1 (|1, 0i − |0, 1i).
2
S1+2 = 1 states are eigenstates of Ĥ0 with eigenvalue − J2 · 1 · 2 + 2J = J,
|S1+2 = 1, S1+2,z = 1i = √1 (|1, 0i − |0, 1i);
2
S1+2,−
|S1+2 = 1, S1+2,z = 0i = √2 |S1+2 = 1, S1+2,z = 1i = √12 (|1, −1i − |0, −1i);
|S1+2 = 1, S1+2,z = −1i = S1+2,−
√
2
|S1+2 = 1, S1+2,z = 0i = √12 (|0, −1i − | − 1, 0i).

Suppose |S1+2 = 0, S1+2,z = 0i = c1 |1, −1i+c2 |0, 0i+c3 |−1, 0i, then by 0 = Ŝ1+2,+ |S1+2 =
√ √
0, S1+2,z = 0i = (Ŝ1,+ + Ŝ2,+ )(c1 |1, −1i + c2 |0, 0i + c3 | − 1, 0i) = 2(c1 + c2 )|1, 0i + 2(c2 +
c3 )|0, 1i, we have c2 = −c1 and c3 = −c2 . The normalized state |S1+2 = 0, S1+2,z = 0i =
√1 (|1, −1i − |0, 0i + | − 1, 1i).
3
S1+2 = 0 states are eigenstates of Ĥ0 with eigenvalue − J2 · 0 · 1 + 2J = 2J,
|S1+2 = 0, S1+2,z = 0i = √1 (|1, −1i − |0, 0i + | − 1, 1i).
3

(b). This is similar to homework #6 Problem 1(b).

2
The full Hamiltonian Ĥ commutes with Ŝ1+2,z = Ŝ1,z + Ŝ2,z . Because Ĥ0 = (− J2 Ŝ 1+2 +
2 2
constant) commutes with each component of Ŝ 1+2 , and the perturbation D · (Ŝ1,z + Ŝ2,z )
commutes with both Ŝ1,z and Ŝ2,z according to the commutation relation given in main text
of this problem.
The Ŝ1+2,z = 2 subspace has basis |Ŝ1+2 = 2, Ŝ1+2,z = 2i.
The Ŝ1+2,z = 1 subspace has basis (|Ŝ1+2 = 2, Ŝ1+2,z = 1i, |Ŝ1+2 = 1, Ŝ1+2,z = 1i).
The Ŝ1+2,z = 0 subspace has basis (|Ŝ1+2 = 2, Ŝ1+2,z = 0i, |Ŝ1+2 = 1, Ŝ1+2,z = 0i, |Ŝ1+2 =
0, Ŝ1+2,z = 0i).
The Ŝ1+2,z = −1 subspace has basis (|Ŝ1+2 = 2, Ŝ1+2,z = −1i, |Ŝ1+2 = 1, Ŝ1+2,z = −1i).
The Ŝ1+2,z = −2 subspace has basis |Ŝ1+2 = 2, Ŝ1+2,z = −2i.
Under these basis choices, Ĥ0 is diagonal with diagonal elements being the eigenvalues
in (a). The matrix elements of the perturbation term can be computed using the result of
2 2
(a) about the |S1+2 , S1+2,z i states. For example, D · (Ŝ1,z + Ŝ2,z )|S1+2 = 2, S1+2,z = 0i
= D · √16 ((12 +(−1)2 )·|1, −1i+(02 +02 )·|0, 0i+((−1)2 +12 )·|−1, 1i) = 2D
√ (|1, −1i+|−1, 1i)
6
√
2D 2 2D
= 3
|S1+2 = 2, S1+2,z = 0i + 3
|S1+2 = 0, S1+2,z = 0i

3/13
 The Hamiltonian matrices and (approximate) ground state energy in each subspace are
given in the following table,
S1+2,z Ĥ0 + perturbation ground state energy
2 (−J) + (2D) = −J + 2D
   
−J 0 D 0
1  +  = −J + D
0 J 0 D
   √ 
2D 2 2D
−J 0 0 0
   3 3  √ √
2D (2 2D/3)·( 2D/3) 2D 8D2
0  0 J 0  +  0 2D 0  ≈ −J + + = −J + −
   
3 (−J)−2J 3 27
   √ 
2 2D 4D
0 0 2J 3
0 3
   
−J 0 D 0
−1  +  = −J + D
0 J 0 D
−2 (−J) + (2D) = −J + 2D

The exact ground 

state energy in the Ŝ
1+2,z = 0 space can be obtained by diagonalization
√
2D 2 2D
−J + 3 3
√
2 2D
of the 2 × 2 matrix  √  = ( J + D)σ0 + σ1 + (− 3J − D
)σ3 , and is
2 2D 4D 2 3 2 3
3
2J + 3
q √ q
8D2 /9
( J2 + D) − ( 2 32D )2 + (− 3J
2
− D 2
3
) = ( J2 + D) − ( 3J
2
+ D
3
) · 1+ (3J/2+D/3)2
8D2 /9 8D2
≈ ( J2 + D) − ( 3J
2
+ D
3
) · (1 + 12 · 9J 2 /4
) ≈ −J + 2D
3
− 27J
.

Problem 2. (30 points). Consider two fermion modes with annihilation opera-
tors denoted by fˆi for i = 1, 2, satisfying {fˆi , fˆj† } = δi,j and {fˆi , fˆj } = 0. Denote
the normalized vacuum state by |vaci. The occupation basis states are |ψ0 i ≡ |vaci,
|ψ1 i ≡ fˆ1† |vaci, |ψ2 i ≡ fˆ2† |vaci, and |ψ3 i ≡ fˆ1† fˆ2† |vaci. Let Ĥ0 = E · (n̂1 + n̂2 ),
where E is a positive constant, n̂i ≡ fˆ† fˆ are occupation number operators. Consider
i i

the Hamiltonian Ĥ = Ĥ0 + ∆ · (fˆ1† fˆ2† + fˆ2 fˆ1 + fˆ1† fˆ2† fˆ2 fˆ1 ), where ∆ is a real “small” parameter,

(a) (15pts) At time t = 0, set the initial state to |ψ(t = 0)i = |ψ0 i = |vaci. Evolve this
state by Ĥ, namely ih̄ ddt |ψ(t)i = Ĥ|ψ(t)i. Solve the transition probability to |ψi i state
over time t, namely |hψi |ψ(t)i|2 , for i = 1, 2, 3, to lowest non-trivial order of ∆. [Hint: use
the interaction picture; note that Ĥ preserves particle number parity, decompose the Fock
space into even- and odd-particle-number subspaces]

4/13
 (b) (10pts) Solve all eigenvalues of Ĥ up to cubic order of ∆. [Hint: you don’t have to
use perturbation theory]

(c) (5pts) Solve the transition probabilities defined in (a) exactly. Expand to lowest
non-trivial order of ∆ and compare with (a). [Hint: some facts on page 1 may be useful]

Solution.
Rearrange the basis to (|ψ0 i, |ψ3 i, |ψ1 i, |ψ2 i).
   
0 0 0 0 0 ∆ 0 0
   
0 2E 0 0  ∆ ∆ 0 0
Ĥ is the following block-diagonal matrix,  + .
   
0 0 E 0   0 0 0 0 
   
0 0 0 E 0 0 0 0
Because it preserves total-particle-number parity, so does not have matrix element between
even-particle-number states (|ψ0,3 i) and odd-particle-number states (|ψ1,2 i).
The steps for computing these matrix elements are omitted here.
(a). Because Ĥ is independent of time, |ψ(t)i = exp(− h̄i Ĥ · t)|ψ(t = 0)i.
Use the interaction picture, define ÛI (t) = exp( h̄i Ĥ0 · t) · exp(− h̄i Ĥ · t). Then ih̄ dt
d
ÛI (t) =
V̂I (t)ÛI (t), where V̂I (t) ≡ exp( h̄i Ĥ0 · t) · V̂S · exp(− h̄i Ĥ0 · t), and the Schrödinger picture
perturbation operator V̂S is V̂S ≡ ∆ · (fˆ1† fˆ2† + fˆ2 fˆ1 + fˆ1† fˆ2† fˆ2 fˆ1 ). The Dyson series form of ÛI
is ÛI (t) = 1 + −h̄i 0 dt1 V̂I (t1 ) + ( −h̄i )2 0 dt1 0 1 dt2 V̂I (t1 )V̂I (t1 ) + . . . .
Rt Rt Rt

Note that |ψi i are eigenstates of Ĥ0 with eigenvalue Ei = 0, E, E, 2E for i = 1, 2, 3, 4

respectively. Then |hψi |ψ(t)i|2 = |hψi |e−iĤ0 ·t/h̄ ÛI (t)|ψ0 i|2 = |hψi |e−iEi ·t/h̄ ÛI (t)|ψ0 i|2 =
|hψi |ÛI (t)|ψ0 i|2 .
i
 
0 ∆e− h̄ 2E·t 0 0
 i 
∆e h̄ 2E·t ∆ 0 0
Under the (|ψ0 i, |ψ3 i, |ψ1 i, |ψ2 i) basis, V̂I (t) is  .
 
 0 0 0 0
 
0 0 0 0
Combine this with the Dyson series form of ÛI , it is easy to see that hψ1 |ÛI (t)|ψ0 i =
hψ2 |ÛI (t)|ψ0 i = 0 to all orders of perturbation, because hψ1,2 |V̂I (t) = 0.
Rt
To lowest non-trivial order, hψ3 |ÛI (t)|ψ0 i ≈ 0 + −h̄i 0 hψ3 |V̂I (t1 )|ψ0 i
· (e h̄ 2E·t − 1) = −iei h̄ ∆
Rt i i E·t
= 0 + −h̄i 0 dt1 ∆e h̄ 2E·t1 = − 2E
∆
E
sin( E·t
h̄
).
Finally |hψ1 |ψ(t)i|2 = 0, |hψ2 |ψ(t)i|2 = 0, |hψ3 |ψ(t)i|2 ≈ [ ∆
E
sin( E·t
h̄
)]2 .

5/13
 (b) From the matrix form of Ĥ, it is easy to see that |ψ1 i and |ψ2 i are already eigenstates
of perturbed Hamiltonian Ĥ with eigenvalue E.    
0 0 0 ∆
In the subspace spanned by (|ψ0 i, |ψ3 i), Ĥ is a 2 × 2 matrix,  + 
0 2E ∆ ∆
∆
(E + 2
)σ0 + (−E − ∆2 )σ3 + ∆σ1 . Perturbation theory can be used, but the exact eigenvalues
can also be easily computed. The eigenvalues are [see homework #1 Problem 4(a)],
q q
∆2
(E + 2 ) ± (−E − ∆2 )2 + ∆2 = (E + ∆2 ) ± (E + ∆2 ) 1 + (E+∆/2)
∆
2

∆ 2
≈ (E + 2
± (E + ∆2 ) · [1 + 12 · (E+∆/2)
) ∆ 4
2 + O(∆ )]
h 2
i
≈ (E + ∆2 ) ± (E + ∆2 ) + 21 ∆E (1 − 2E ∆
) + O(∆4 ),
where O(∆4 ) means terms of 4th or higher orders.
√ 2 3
You can also use 1 + x ≈ 1 + x2 − x8 + x16 + O(x4 ) to do the expansion,
q q
2
(−E − ∆2 )2 + ∆2 = E · 1 + ( ∆ E
+ 54 ∆E2
)
h i
1 ∆ 5 ∆2 1 ∆ 5 ∆2 2 1 ∆ 5 ∆2 3
≈ E · 1 + 2 · ( E + 4 E 2 ) − 8 · ( E + 4 E 2 ) + 16 · ( E + 4 E 2 ) + O(∆4 )
h i
5 ∆2 1 ∆2 5 ∆3 1 ∆3
≈ E · 1 + ( 21 ∆
E
+ 8 E2
) − ( 8 E2
+ 16 E 3
) + 16 E 3
+ O(∆4 )
∆ ∆2 ∆3
=E+ 2
+ 2E
− 4E 2
+ O(∆4 ).
2 ∆3 ∆2 ∆3
Finally, the approximate eigenvalues are E, E, − ∆
2E
+ 4E 2
, and 2E + ∆ + 2E
− 4E 2
.

(c) The time-evolution operator exp(− h̄i Ĥ · t) can be computed exactly.

 Under
 the  previous basis,
 it has the following block diagonal form,
0 ∆
exp − i   · t 02×2 
h̄
∆ 2E + ∆ .
 

 
02×2 e− h̄i E·t
12×2
Then obviously hψ1,2 | exp(− h̄i Ĥ · t)|ψ0 i = 0.
Use
 the  facts on page
 1 [also in homework #1 Problem 4(b)],
0 ∆
exp − h̄i   · t = exp[−i t ((E + ∆ )σ0 + (−E − ∆ )σ3 + ∆σ1 )]
h̄ 2 2
∆ 2E + ∆
· [cos(ω · t)σ0 − i sin(ω · t)( −E−∆/2
(E+∆/2)
= e−i h̄
t
ω
σ3 + ∆ σ )],
ω 1
p p
where ω = (−E − ∆/2)2 + ∆2 = E 2 + E∆ + 5∆2 /4.
· i sin(ω · t) ·
(E+∆/2)
Therefore hψ3 | exp(− h̄i Ĥ · t)|ψ0 i = −e−i h̄
t ∆
ω
.
Finally the exact transition probabilities are
|hψ1 |ψ(t)i|2 = 0, |hψ2 |ψ(t)i|2 = 0, |hψ3 |ψ(t)i|2 = [ ∆
ω
sin(ω · t)]2 .
To lowest order approximation ω ≈ E, this reduces to the result of (a).

6/13
 Problem 4. (40 points) Consider four spin-1/2 moments, labeled by subscripts i with
i = 1, 2, 3, 4 respectively. Denote the tensor product of Ŝz eigenbasis by |s1 , s2 , s3 , s4 i, where
si = ± 21 is the eigenvalue of Ŝi,z . For simplicity, denote si = + 12 by ↑, and si = − 21 by ↓.
The commutation relation of the spin operators is [Ŝi,a , Ŝj,b ] = δi,j i c abc Ŝi,c .
P

(a). (4pts) Show that the following four hermitian operators mutually commute,
(Ŝ 1 + Ŝ 2 + Ŝ 3 + Ŝ 4 )2 , (Ŝ1,z + Ŝ2,z + Ŝ3,z + Ŝ4,z ), (Ŝ 1 + Ŝ 3 )2 , (Ŝ 2 + Ŝ 4 )2 .

(b). (16pts) Suppose the four operators in (a) have the following simultaneous eigen-
values, S(S + 1), Sz , S1+3 (S1+3 + 1), S2+4 (S2+4 + 1), respectively. Find the possible
combinations of these “quantum numbers”, and the normalized eigenstates |S, Sz , S1+3 , S2+4 i
in terms of Ŝz basis. [Hint: add the spins ‘1’ and ‘3’ first, solve |S1+3 , S1+3,z i in terms
of |s1 , s3 i; add the spins ‘2’ and ‘4’ in similar fashion; then make linear combinations of
|S1+3 , S1+3,z i|S2+4 , S2+4,z i, the results of some previous problem may help]

(c). (4pts) Solve the eigenvalues of Ĥ = Ŝ 1 · Ŝ 2 + Ŝ 2 · Ŝ 3 + Ŝ 3 · Ŝ 4 + Ŝ 4 · Ŝ 1 . [Hint:

try to relate Ĥ with the operators in (a), then use the result of (b)]

(d). (16pts) Consider the D4 group (see page 1) generated by

C4 : |s1 , s2 , s3 , s4 i 7→ |s4 , s1 , s2 , s3 i; and σs : |s1 , s2 , s3 , s4 i 7→ |s1 , s4 , s3 , s2 i.
The D4 group elements commute with the total spin operators 4i=1 Ŝ i . We can construct
P

complete orthonormal basis states of D4 irreducible representations(irrep), labeled by

|S, Sz , (Γi , j)i. S and Sz are defined in (b). i = 1, 2, 3, 4, 5 label the type of irrep. j = 1, 2
for the 2-dimensional Γ5 irrep, label the two basis states of a Γ5 irrep; and j = 1 (can be
omitted) for other 1-dimensional irrep. Under the action of group generators, these states
transform as C4 : |S, Sz , (Γi , j)i 7→ j 0 |S, Sz , (Γi , j 0 )i · [RΓi (C4 )]j 0 ,j ; and
P

σs : |S, Sz , (Γi , j)i 7→ j 0 |S, Sz , (Γi , j 0 )i · [RΓi (σs )]j 0 ,j . Here RΓi (g) is the representation
P

matrix (1 × 1 for Γ1,2,3,4 , 2 × 2 for Γ5 ) for group element g. Construct these |S, Sz , (Γi , j)i
states in terms of Ŝz eigenbasis. And write down the representation matrices RΓi (C4 ) and
RΓi (σs ) under these basis. [Hint:
method #1: use the basis in (b), figure out how they transform under the group
generators, and (if necessary) make linear combinations to form irreducible represen-

7/13
 P P
tations, |S, Sz , (Γi , j)i = S1+3 S2+4 |S, Sz , S1+3 , S2+4 ihS, Sz , S1+3 , S2+4 |S, Sz , (Γi , j)i,
because the D4 group elements commute with ladder operators, the coefficients
hS, Sz , S1+3 , S2+4 |S, Sz , (Γi , j)i is independent of Sz , so you only need to work out
these coefficients for highest Sz = S by e.g. “projection operator”;
method #2: use the “projection operator” on the |s1 , s2 , s3 , s4 i basis directly in each total
Sz subspace, and compare the results to (b) to figure out the S quantum number, you may
need to make linear combinations of states with the same Sz and the same irrep]

Solution.
(a). Define Ŝ 1+3 = Ŝ 1 + Ŝ 3 , Ŝ 2+4 = Ŝ 2 + Ŝ 4 , and Ŝ = Ŝ 1 + Ŝ 2 + Ŝ 3 + Ŝ 4 = Ŝ 1+3 + Ŝ 2+4 .
The fact that these four operators commute has been used in the “addition of angular
momentum”.
Components of Ŝ satisfy the commutation relation of angular momentum, [Ŝa , Ŝb ] =
2
i c  Ŝc . Then [Ŝ , Ŝz ] = [Ŝx2 +Ŝy2 , Ŝz ]+0 = [Ŝx , Ŝz ]Ŝx +Ŝx [Ŝx , Ŝz ]+[Ŝy , Ŝz ]Ŝy +Ŝy [Ŝy , Ŝz ] =
P abc

−iŜy · Ŝx + Ŝx · (−iŜy ) + iŜx · Ŝy + Ŝy · iŜx = 0.

2 2
Similarly we have [Ŝ 1+3 , Ŝ1+3,a ] = 0 and [Ŝ 2+4 , Ŝ2+4,a ] = 0, for a = x, y, z.
2 2 2 2
It is obvious that [Ŝ 1+3 , Ŝ 2+4 ] = 0, and [Ŝ 2+4 , Ŝ1+3,a ] = 0 and [Ŝ 1+3 , Ŝ2+4,a ] = 0 for
a = x, y, z. Because the two operators in the commutator do not share spin operators of the
same spin.
2 2 2 2
Then [Ŝ 2+4 , Ŝz ] = [Ŝ 2+4 , Ŝ1+3,z + Ŝ2+4,z ] = 0, and [Ŝ 1+3 , Ŝz ] = [Ŝ 1+3 , Ŝ1+3,z + Ŝ2+4,z ] = 0.
P 2 2
Finally Ŝ 1+3 · Ŝ 2+4 = a Ŝ1+3,a · Ŝ2+4,a commutes with Ŝ 1+3 and Ŝ 2+4 , be-
2 2
cause the factors Ŝ1+3,a and Ŝ2+4,a commute with Ŝ 1+3 and Ŝ 2+4 . And consider
2 2 2 2 2 2 2
Ŝ = Ŝ 1+3 + Ŝ 2+4 + 2Ŝ 1+3 · Ŝ 2+4 . We have [Ŝ , Ŝ 1+3 ] = 0, and [Ŝ , Ŝ 2+4 ] = 0.

(b) Compose |S1+3 , S1+3,z i states in terms of |s1 , s3 i. This is the same as homework #5
1 1
Problem 1(a). S1+3 can be 1 or 0, 2
⊗ 2
= 1 ⊕ 0.
|S1+3 = 1, S1+3,z = 1i = | ↑↑i,
|S1+3 = 1, S1+3,z = 0i = √1 (| ↑↓i + | ↓↑i),
2
|S1+3 = 1, S1+3,z = −1i = | ↓↓i.
|S1+3 = 0, S1+3,z = 0i = √1 (| ↑↓i − | ↓↑i).
2
The |S2+4 , S2+4,z i states in terms of |s2 , s4 i have exactly the same form. Finally compose
|S, Sz , S1+3 , S2+4 i in terms of |S1+3 , S1+3,z i|S2+4 , S2+4,z i. Note |s1 , s3 i|s2 , s4 i = |s1 , s2 , s3 , s4 i.

8/13
 If S1+3 = 0 and S2+4 = 0, then S must be 0,
|S = 0, Sz = 0, S1+3 = 0, S2+4 = 0i = |S1+3 = 0, S1+3,z = 0i|S2+4 = 0, S2+4,z = 0i
= 21 (| ↑↑↓↓i − | ↑↓↓↑i − | ↓↑↑↓i + | ↓↓↑↑i).
If S1+3 = 1 and S2+4 = 0, then S must be 1,
|S = 1, Sz = 1, S1+3 = 1, S2+4 = 0i = |S1+3 = 1, S1+3,z = 1i|S2+4 = 0, S2+4,z = 0i
= √1 (| ↑↑↑↓i − | ↑↓↑↑i),
2
|S = 1, Sz = 0, S1+3 = 1, S2+4 = 0i = |S1+3 = 1, S1+3,z = 0i|S2+4 = 0, S2+4,z = 0i
= 21 (| ↑↑↓↓i − | ↑↓↓↑i + | ↓↑↑↓i − | ↓↓↑↑i),
|S = 1, Sz = −1, S1+3 = 1, S2+4 = 0i = |S1+3 = 1, S1+3,z = −1i|S2+4 = 0, S2+4,z = 0i
= √1 (| ↓↑↓↓i − | ↓↓↓↑i).
2
If S1+3 = 0 and S2+4 = 1, then S must be 1,
|S = 1, Sz = 1, S1+3 = 0, S2+4 = 1i = |S1+3 = 0, S1+3,z = 0i|S2+4 = 1, S2+4,z = 1i
= √1 (| ↑↑↓↑i − | ↓↑↑↑i),
2
|S = 1, Sz = 0, S1+3 = 0, S2+4 = 1i = |S1+3 = 0, S1+3,z = 0i|S2+4 = 1, S2+4,z = 0i
= 21 (| ↑↑↓↓i + | ↑↓↓↑i − | ↓↑↑↓i − | ↓↓↑↑i),
|S = 1, Sz = −1, S1+3 = 0, S2+4 = 1i = |S1+3 = 0, S1+3,z = 0i|S2+4 = 1, S2+4,z = −1i
= √1 (| ↑↓↓↓i − | ↓↓↑↓i).
2
If Ŝ1+3 = 1 and S2+4 = 1, then S can be 2 or 1 or 0, this part is exactly the same as
problem 1(a), just replace the |s1 , s2 i basis there by |S1+3 , S1+3,z = s1 i|S2+4 , S2+4,z = s2 i,
|S = 2, Sz = 2, S1+3 = 1, S2+4 = 1i = | ↑↑↑↑i,
|S = 2, Sz = 1, S1+3 = 1, S2+4 = 1i = 12 (| ↑↑↓↑i + | ↓↑↑↑i + | ↑↑↑↓i + | ↑↓↑↑i),
|S = 2, Sz = 0, S1+3 = 1, S2+4 = 1i
= √1 (| ↓↑↓↑i + | ↑↑↓↓i + | ↑↓↓↑i + | ↓↑↑↓i + | ↓↓↑↑i + | ↑↓↑↓i),
6
|S = 2, Sz = 1, S1+3 = −1, S2+4 = 1i = 12 (| ↓↑↓↓i + | ↓↓↓↑i + | ↑↓↓↓i + | ↓↓↑↓i),
|S = 2, Sz = −2, S1+3 = 1, S2+4 = 1i = | ↓↓↓↓i.

|S = 1, Sz = 1, S1+3 = 1, S2+4 = 1i = 21 (| ↑↑↑↓i + | ↑↓↑↑i − | ↑↑↓↑i − | ↓↑↑↑i),

|S = 1, Sz = 0, S1+3 = 1, S2+4 = 1i = √1 (| ↑↓↑↓i − | ↓↑↓↑i),
2
|S = 1, Sz = −1, S1+3 = 1, S2+4 = 1i = 12 (| ↑↓↓↓i + | ↓↓↑↓i − | ↓↑↓↓i − | ↓↓↓↑i),

|S = 0, Sz = 0, S1+3 = 1, S2+4 = 1i
1
= √
2 3
(2| ↑↓↑↓i − | ↑↑↓↓i − | ↑↓↓↑i − | ↓↑↑↓i − | ↓↓↑↑i + 2| ↓↑↓↑i).

9/13
 (c) Ĥ = (Ŝ 1 + Ŝ 3 ) · (Ŝ 2 + Ŝ 4 ) = 21 [(Ŝ 1 + Ŝ 2 + Ŝ 3 + Ŝ 4 )2 − (Ŝ 1 + Ŝ 3 )2 − (Ŝ 2 + Ŝ 4 )2 ].
Therefore the |S, Sz , S1+3 , S2+4 i states in (b) are eigenstates of Ĥ with eigenvalues
1
2
[S(S + 1) − S1+3 (S1+3 + 1) − S2+4 (S2+4 + 1)]. These are summarized in the following table,
S Sz S1+3 S2+4 H eigenvalue
0 0, 0 0 0
1 1, 0, −1, 1 0 0
1 1, 0, −1, 0 1 0
2 2, 1, 0, −1, −2, 1 1 1
1 1, 0, −1, 1 1 −1
0 0, 1 1 −2

(d) This is similar to homework #5 Problem 1(c).

Use the method #1 in the Hint, and the following hint given during the exam.
Because the D4 group elements commute with Ŝ operators, and in particular commute
with the ladder operators. The action of group elements do not change S and Sz quantum
numbers, but may change S1+3 and S2+4 .
If the action of group element g on highest Sz = S states is
0 0
P
|S, Sz = S, S1+3 , S2+4 i 7→ S 0 ,S 0 |S, Sz = S, S1+3 , S2+4 i · [R(g)](S1+3
0 0
,S2+4 ),(S1+3 ,S2+4 ) ,
1+3 2+4
0 0
where R(g) is the representation matrix the combination (S1+3 , S2+4 ) is the row index and
(S1+3 , S2+4 ) is the column index, then this representation matrix is independent of Sz ,
0 0
P
|S, Sz , S1+3 , S2+4 i 7→ S 0 ,S 0 |S, Sz , S1+3 , S2+4 i · [R(g)](S1+3
0 0
,S2+4 ),(S1+3 ,S2+4 ) .
1+3 2+4

Because Sz can be changed by application of lowering ladder operator Ŝ− = 4i=1 Ŝi,− .
P

So we only need to work out the R(g) matrices for each subspace with certain S and
Sz = S. This can be done by the definition of C4 and σs in the |s1 , s2 , s3 , s4 i basis, and the
result of (b). The results are
basis R(C4 ) R(σs )
|S = 0, Sz = 0, S1+3 = 0, S2+4 = 0i (−1) (−1)
  
(|S = 1, Sz = 1, S1+3 = 1, S2+4 = 0i, 0 1 −1 0
  
|S = 1, Sz = 1, S1+3 = 0, S2+4 = 1i) −1 0 0 1
|S = 0, Sz = 0, S1+3 = 1, S2+4 = 1i (1) (1)
|S = 1, Sz = 1, S1+3 = 1, S2+4 = 1i (−1) (1)
|S = 2, Sz = 2, S1+3 = 1, S2+4 = 1i (1) (1)

10/13
 These are already irreducible representations. So the |S, Sz , (Γi , j)i basis are
|S, Sz , (Γi , j)i states in terms of |S, Sz , S1+3 , S2+4 i R(C4 ) R(σs )
|S = 0, Sz , (Γ4 , j = 1)i |S = 0, Sz , S1+3 = 0, S2+4 = 0i (−1) (−1)
   
|S = 1, Sz , (Γ5 , j = 1)i |S = 1, Sz , S1+3 = 1, S2+4 = 0i 0 1 −1 0
   
|S = 1, Sz , (Γ5 , j = 2)i |S = 1, Sz , S1+3 = 0, S2+4 = 1i −1 0 0 1
|S = 0, Sz , (Γ1 , j = 1)i |S = 0, Sz , S1+3 = 1, S2+4 = 1i (1) (1)
|S = 1, Sz , (Γ3 , j = 1)i |S = 1, Sz , S1+3 = 1, S2+4 = 1i (−1) (1)
|S = 2, Sz , (Γ1 , j = 1)i |S = 2, Sz , S1+3 = 1, S2+4 = 1i (1) (1)
Here Sz can be −S, −S + 1, . . . , S.
The choice of Γ5 basis are of course not unique.
If you use method #2 in the Hint, you will eventually get the same result.
But the “projection operator” results for Γ1 irrep in the Sz = 0 subspace are
1
(| ↑↑↓↓i
2 q
+ | ↑↓↓↑i + | ↓↑↑↓i + | ↓↓↑↑i)
q
2 1
= 3
|S = 2, Sz = 0, S1+3 = 1, S2+4 = 1i − 3
|S = 0, Sz = 0, S1+3 = 1, S2+4 = 1i, and
√1 (| ↑↓↑↓i + | ↓↑↓↑i)
2q q
= 13 |S = 2, Sz = 0, S1+3 = 1, S2+4 = 1i + 2
3
|S = 0, Sz = 0, S1+3 = 1, S2+4 = 1i.
You need to make linear combinations of these to get total-S eigenstates.

Problem 5. (10 points) (“Heisenberg chain”) Consider N spin-1/2 moments labeled by

subscripts i with i = 0, 1, . . . , (N − 1). Here N is a large integer. The 1D Heisenberg model
PN −1
Hamiltonian is Ĥ = −J i=0 (Ŝ i · Ŝ i+1 − 41 ). Here we assume periodic boundary condition,
so Ŝ N is actually Ŝ 0 . From Ŝ i · Ŝ i+1 = Ŝi,z Ŝi+1,z + 12 (Ŝi,+ Ŝi+1,− + Ŝi,− Ŝi+1,+ ), it is easy to see
that the fully polarized state, | ↓↓ . . . ↓i, is an eigenstate of Ĥ with eigenvalue (0). Label
the other Ŝz eigenstates by the positions of ↑, for example |xi ≡ Ŝx,+ | ↓↓ . . . ↓i has one ↑ at
position x = 0, 1, . . . , (N − 1).
The lattice translation, T̂ : |s1 , s2 , . . . , sN i 7→ |sN , s1 , . . . , sN −2 i, Ŝ i 7→ Ŝ i+1 , is a
symmetry of Ĥ, [Ĥ, T̂ ] = 0. Note that T̂ N = 1̂, so the irreps of translation group are
2π
Rk (T̂ ) = e−i N k for integer k (modulo N ). For the |xi states, T̂ |xi = |(x + 1) mod N i.

PN −1
(a) (5pts) Show that the “ferromagnetic spin-wave” states |ψp i = x=0 eip·x |xi is an
2π
eigenstate of Ĥ, where p = N
k and k is an integer. Solve the eigenvalues as a function of p

11/13
(“the spin-wave dispersion”). [Hint: first figure out the action result of Ĥ on the |xi basis,
then apply Ĥ on |ψp i, check that the result is proportional to |ψp i ]

(b) (5pts) (DIFFICULT) (“Bethe ansatz”) For the space with two ↑s, the basis are
|x, yi ≡ Ŝx,+ Ŝy,+ | ↓↓ . . . ↓i for 0 ≤ x < y ≤ (N − 1). Consider a special case of the “Bethe
ansatz”, |ψp,−p i = x,y,x<y (eip·x ei(−p)·y + eiθ ei(−p)·x eip·y )|x, yi. Solve the real parameters p
P

and θ for this state to be an eigenstate of Ĥ, and find the energy eigenvalue. [Hint: apply
Ĥ on this state, be careful about the “collision” case y = x + 1; and be careful about the
“boundary condition” at x = 0 or y = (N − 1), which is equivalent to the fact that |ψp,−p i
is an eigenstate of T̂ ]

Solution.
(a) Consider the action of (Ŝ i · Ŝ i+1 − 14 ) = (Ŝi,z Ŝi+1,z − 14 ) + 12 (Ŝi,+ Ŝi+1,− + Ŝi,− Ŝi+1,+ )
on the |xi basis:
if i 6= x and (i + 1) 6= x mod N , the result is ((− 12 ) · (− 21 ) − 14 )|xi + 0 + 0 = 0;
if i = x, the result is (( 12 ) · (− 12 ) − 41 )|xi + 0 + 21 |x + 1i = 12 (−|xi + |x + 1i);
if i + 1 = x mod N , the result is ((− 12 ) · ( 12 ) − 14 )|xi + 12 |x − 1i + 0 = 21 (−|xi + |x − 1i).
Therefore Ĥ|xi = J2 (2|xi − |x − 1i − |x + 1i).
Here the positions (x − 1), (x + 1) should be understood with implicit modulo N .
If we view the ↑ as a particle, Ĥ produces “hoppings” of this particle to neighboring sites.
PN −1 ip·x J
Ĥ|ψp i = x=0 e 2 (2|xi − |(x − 1) mod N i − |(x + 1) mod N i)
−1 ip·(x0 +1 mod N ) 0 P −1 ip·(x00 −1 mod N ) 00
= J|ψp i − J2 xN0 =0 |x i − J2 N
P
e x00 =0 e |x i,
here the dummy variables x0 = (x − 1) mod N , x00 = (x + 1) mod N . Because p = 2π
N
k
with integer k, eip·(x mod N )
= eip·x . Therefore Ĥ|ψp i = (J − J2 eip − J2 e−ip )|ψp i.
The energy eigenvalues is Ep = J · [1 − cos(p)].

(b) Similar to (a), we have the following action result of Ĥ on |x, yi states:
if x 6= y − 1 and x 6= (y + 1) mod N , Ĥ|x, yi
= 2J|x, yi − J2 (|x − 1, yi + |x + 1, yi + |x, y − 1i + |x, y + 1i);
if x = y − 1, Ĥ|x, yi = J|x, yi − J2 (|x − 1, yi + |x, y + 1i);
if x = y + 1 mod N , Ĥ|x, yi = J|x, yi − J2 (|x + 1, yi + |x, y − 1i).
This is two “hardcore” particles’ hopping (they cannot occupy the same site).

12/13
 Define ψ(x, y) = eipx e−ipy + eiθ e−ipx eipy for 0 ≤ x < y ≤ N − 1, and ψ(y, x) = ψ(x, y),
P
then |ψp,−p i = x,y,x<y ψ(x, y)|x, yi.
For the case of 0 < x < x + 1 < y < N − 1, we have
hx, y|Ĥ|ψp,−p i = 2Jψ(x, y) + J2 [ψ(x − 1, y) + ψ(x + 1, y) + ψ(x, y − 1) + ψ(x, y + 1)],
plug in the formula of ψ(x, y), this is
= [2J − J · (e−ip + eip )] · ψ(x, y).
Therefore, if |ψp,−p i is an eigenstate of Ĥ, the energy eigenvalue must be 2J · [1 − cos(p)].
Consider the case of 0 < x < x + 1 = y < N − 1,
hx, x + 1|Ĥ|ψp,−p i = Jψ(x, x + 1) − J2 [ψ(x − 1, x + 1) + ψ(x, x + 2)]
= J · (e−ip + eiθ eip ) − J
2
· (e−2ip + eiθ e2ip ) · 2,
this should be 2J · [1 − cos(p)] · ψ(x, x + 1) = 2J · [1 − cos(p)] · (e−ip + eiθ eip ).
From this we can solve eiθ = e−ip . So we can choose θ = −p.
Consider the case of 0 = x < x + 1 < y < N − 1,
h0, y|Ĥ|ψp,−p i = 2Jψ(0, y) − J2 [ψ(y, N − 1) + ψ(1, y) + ψ(0, y − 1) + ψ(0, y + 1)],
for this to be equal to 2J · [1 − cos(p)] · ψ(0, y)
= 2Jψ(0, y) − J2 [ψ(−1, y) + ψ(1, y) + ψ(0, y − 1) + ψ(0, y + 1)],
we must have ψ(y, N − 1) = eipy e−ip(N −1) + eiθ e−ipy eip(N −1)
= ψ(−1, y) = e−ip e−ipy + eiθ eip eipy , for all 1 < y < N − 1.
Therefore eiθ eipN = 1, together with θ = −p we have eip(N −1) = 1.
Finally,
Ep,−p = 2J · [1 − cos(p)],
θ = −p,
2π
p= N −1
k with integer k.

Note: as a check, consider N = 4 case:

− EJp + 1 = cos( 2π
4
k) = 1 or 0 or −1, and
− Ep,−p
J
+ 1 = 2 cos( 2π
3
k) − 1 = 1 or −2.
These are some of the eigenvalues in problem 3(c).

13/13