Midterm Solution 2017
Midterm Solution 2017
Midterm Solution 2017
P+∞ 1
• Baker-Hausdorff formula: exp(Â) · B̂ · exp(−Â) = B̂ + n=1 n! [Â, [Â, . . . [Â, B̂] . . . ]].
| {z }
n-fold commutator
• If [Â, B̂] is a c-number, then exp(Â + B̂) = exp(Â) exp(B̂) exp(− 12 [Â, B̂]).
p̂2 mω 2 2 2
h̄ ∂ 2 mω 2 2
• 1D harmonic oscillator: Ĥ0 = 2m
+ 2
x̂ = − 2m ∂x2
+ 2
x.
Here x̂ is position operator, p̂ is momentum operator, [x̂, p̂] = ih̄, and in position
representation p̂ = −ih̄ ∂∂x . Define b̂ = (x̂ + i mω
p mω 1
p mω h̄ ∂
2h̄
p̂) = 2h̄
(x + mω ∂x
). Then
[b̂, b̂† ] = 1 and Ĥ0 = h̄ω (b̂† b̂ + 21 ). It has a unique ground state |0i with b̂|0i = 0, and
excited states |ni ≡ √1 (b̂† )n |0i with energy En = (n + 21 )h̄ω. These states |ni can be
n!
viewed as occupation basis of a single boson mode.
– Given complete orthonormal basis |ei i of single particle states, one set of com-
plete orthonormal basis for the Fock space is the occupation basis |n1 , n2 , . . . i =
† †
√1 (ê )n1 √1 (ê )n2
n1 ! 1 n2 ! 2
· · · |vaci. Here |vaci is the particle “vacuum”. ê†i are creation
operators for state |ei i. For bosons, [êi , ê†j ] = δi,j ; for fermions, {êi , ê†j } = δi,j .
– [ê†i êj , ê†k ] = δj,k ê†i , for both bosons and fermions.
(|ai, Â|bi) = ( i |iiai , j,k |ji Ajk bk ) = i,j,k a∗i Gij Ajk bk = ~a† · (G · A) · ~b.
P P P
Equivalent condition #1: If ha|Â|ai is real for any state |ai, then  is real. Here
ha|Â|ai = ~a† · (G · A) · ~a = ~a† · ( 37 σ0 − σ1 ) · ~a = 73 (|a1 |2 + |a2 |2 ) − (a∗1 a2 + a∗2 a1 ), is always real.
To check whether  is unitary, we need to check whether (Â|ii, Â|ji) equals to hi|ji for
any pair of basis states |ii and |ji.
Equivalent condition #1: If the inner product (Â|ai, Â|ai) = (|ai, |ai) for any
state |ai, then  is unitary. This condition is ~a† · (A† · G · A) · ~a = ~a† · G · ~a for any
complex vector ~a, or equivalently, A† · G · A = G. Here A† · G · A = 6σ0 − 23 σ1 6= G = σ0 − 23 σ1 .
(b) The method #2 of (a) already produces the answer for (b).
q
 has eigenvalue 4 with normalized eigenstate 32 (|1i + |2i);
q
3
and eigenvalue 2 with normalized eigenstate 10 (|1i − |2i).
The eigenvalues and un-normalized eigenvectors
can also be obtained from the eigenval-
3 1
ues and eigenvectors of the Aij matrix, , defined in method #1 of (a).
1 3
(b) (10pts) Denote the normalized ground state of Ĥ 0 by |00 i. Evaluate h00 |x̂|00 i and
h00 |p̂|00 i. [Hint: result of (a) may help.]
(c) (10pts) At t = 0, let the state |ψ(t = 0)i = |00 i, evolve this state under Ĥ0 , namely
|ψ(t)i = exp(− h̄i Ĥ0 · t)|ψ(t = 0)i. Evaluate hψ(t)|x̂|ψ(t)i and hψ(t)|p̂|ψ(t)i. [Hint: you
can use either Schrödinger or Heisenberg picture.]
(d) (5pts) Further evaluate hψ(t)|x̂2 |ψ(t)i and hψ(t)|p̂2 |ψ(t)i for |ψ(t)i defined in (c).
Check that the uncertainty relation for x̂ and p̂ is always satisfied. [Hint: it’ll be most
efficient to use the Schrödinger picture and knowledge about boson coherent states.]
(e) (10pts) Define two Hermitian operators: Ô1 = m2 ω 2 x̂2 − p̂2 , Ô2 = mω(x̂p̂ + p̂x̂).
Their Heisenberg picture under Ĥ0 are Ôi,H (t) = exp( h̄i Ĥ0 · t) · Ôi · exp(− h̄i Ĥ0 · t). Write
d
down the Heisenberg equations of motion, Ô (t)
dt i,H
= . . . for i = 1, 2. The right-hand side
of these equations should be expressed in terms of Ôj,H (t) with j = 1, 2.
(f) (5pts) Solve the equations in (e). Namely solve Ôi,H (t) in terms of Ôj,H (t = 0).
i
(d) According to the method #1 of (c), |ψ(t)i = e− h̄ E0 ·t |ze−iωt i is a boson coherent state,
q
−iωt f
b̂|ψ(t)i = ze |ψ(t)i with z = mω2 h̄ h̄mω
2
.
x̂2 = h̄
2mω
(b̂ + b̂† )2 = h̄
2mω
[b̂2 + (b̂† )2 + 2b̂† b̂ + 1].
p̂2 = − h̄mω
2
(b̂ − b̂† )2 = h̄mω
2
[−b̂2 − (b̂† )2 + 2b̂† b̂ + 1].
Finally
hψ(t)|x̂2 |ψ(t)i = h̄
2mω
[z 2 e−2iωt + (z ∗ )2 e2iωt + 2|z|2 + 1] = h̄
2mω
[(ze−iωt + z ∗ eiωt )2 + 1]
f 2 h̄
= [ mω 2 cos(ωt)] + 2mω
, and
hψ(t)|p̂2 |ψ(t)i = h̄mω
2
[−z 2 e−2iωt − (z ∗ )2 e2iωt + 2|z|2 + 1] = h̄mω
2
[−(ze−iωt − z ∗ eiωt )2 + 1]
= [ ωf sin(ωt)]2 + h̄mω
2
.
Combine these with the result of (c), the variance of x̂ and p̂ under state |ψ(t)i are
h̄ h̄mω
hx2 i − hxi2 = 2mω
and hp2 i − hpi2 = 2
, independent of time, and satisfy the uncertainty
2
h̄
relation (hx2 i − hxi2 )(hp2 i − hpi2 ) ≥ 4
.
d d
(e). Ô (t)
dt 1,H
= 2ω Ô2,H (t), and Ô (t)
dt 2,H
= −2ω Ô1,H (t).
Method #1: use the Heisenberg equations of motion, d
Ô (t)
dt H
= h̄i [Ĥ0,H (t), ÔH (t)], and
compute the commutators using [ÂB̂, Ĉ D̂] = Â[B̂, Ĉ]D̂ + [Â, Ĉ]B̂ D̂ + Ĉ Â[B̂, D̂] + Ĉ[Â, D̂]B̂
and [x̂H (t), p̂H (t)] = ih̄.
Method #2: use the Heisenberg equations of motion for x̂H and p̂H in method #2 of (c).
d 1 d
x̂
dt H
= p̂ ,
m H
and p̂
dt H
= −mω 2 x̂H .
For notation simplicity, the argument t for Heisenberg picture operators are omitted here.
Method #2: In fact these can be obtained without using the equations of motion in (e).
1
Use x̂H = x̂ cos(ωt) + mω
p̂ sin(ωt), and p̂H = p̂ cos(ωt) − mωx̂ sin(ωt). Then
1
Ô1,H = m2 ω 2 x̂2H − p̂2H = m2 ω 2 [x̂ cos(ωt) + mω
p̂ sin(ωt)]2 − [p̂ cos(ωt) − mωx̂ sin(ωt)]2
= (m2 ω 2 x̂2 − p̂2 ) · [cos(ωt)2 − sin(ωt)2 ] + mω(x̂p̂ + p̂x̂) · 2 cos(ωt) sin(ωt), and
n
1
Ô2,H = mω(x̂H p̂H + p̂H x̂H ) = mω · [x̂ cos(ωt) + mω p̂ sin(ωt)] · [p̂ cos(ωt) − mωx̂ sin(ωt)]
o
1
+[p̂ cos(ωt) − mωx̂ sin(ωt)] · [x̂ cos(ωt) + mω p̂ sin(ωt)]
= (m2 ω 2 x̂2 − p̂2 ) · [−2 cos(ωt) sin(ωt)] + mω(x̂p̂ + p̂x̂) · [cos(ωt)2 − sin(ωt)2 ].
(a) (5pts) Write down a complete orthonormal basis for the entire Fock space.
(b) (5pts) Define four “Majorana fermion” operators, η̂1 ≡ (fˆ1 + fˆ1† ), η̂2 ≡ −i(fˆ1 − fˆ1† ),
η̂3 ≡ (fˆ2 + fˆ2† ), η̂4 ≡ −i(fˆ2 − fˆ2† ). They
are obviously hermitian. Check the anti-
0, if i 6= j;
commutation relations {η̂i , η̂j } = 2δi,j = [Hint: use the multi-linearity of
2, if i = j.
anti-commutators. Only i ≤ j cases need to be checked.]
(c) (10pts) Under the basis in (a), write down the matrix representations for each η̂i
†
operator. [Hint: first compute the matrices for fˆ1,2 , then fˆ1,2 are just hermitan conjugates;
for later convenience it may help to write these results as tensor products of Pauli matrices.]
(d) (5pts) Define hermitian operators L̂x = iη̂3 η̂4 , L̂y = iη̂4 η̂2 , and L̂z = iη̂2 η̂3 . Compute
their commutators [L̂a , L̂b ] for a, b = x, y, z and a 6= b, express the results in terms of L̂x,y,z .
Check that L̂2x = L̂2y = L̂2z = 1. [Hint: use the anti-commutation relations in (b). Only
three combinations of a, b need to be considered.]
(e) (5pts) Solve the eigenvalues and normalized eigenstates of L̂x in the entire Fock space.
(f) (5pts) Solve the eigenvalues and normalized eigenstates of L̂z in the entire Fock space.
(g) (5pts) Compute exp(iθL̂z ) · (c1 L̂x + c2 L̂y + c3 L̂z ) · exp(−iθL̂z ), where θ and c1,2,3 are
c-numbers. The result should be a finite degree polynomial of L̂x,y,z . [Hint: either use the
Baker-Hausdorff formula and the result of (d), or expand and compute exp(iθL̂z ) explicitly.]
Solution:
(a) Complete orthonormal basis can be chosen as the occupation basis,
|vaci ≡ |n1 = 0, n2 = 0i, fˆ1† |vaci ≡ |n1 = 1, n2 = 0i, fˆ2† |vaci ≡ |n1 = 0, n2 = 1i, fˆ1† fˆ2† |vaci ≡
|n1 = 1, n2 = 1i.
(c) Denote the complete orthonormal basis in (a) as |ei i with i = 1, 2, 3, 4, the matrix
P
element Oji ≡ hej |Ô|ei i of an operator Ô can be obtained by Ô|ei i = j |ej i · Oji .
0 1 0 0
0 0 0 0
fˆ1 is , because fˆ1 |0, 0i = 0, fˆ1 |1, 0i = |0, 0i, fˆ1 |0, 1i = 0, fˆ1 |1, 1i = |0, 1i.
0 0 0 1
0 0 0 0
0 0 1 0
0 0 0 −1
fˆ2 is , because fˆ2 |0, 0i = 0, fˆ2 |1, 0i = 0, fˆ2 |0, 1i = |0, 0i, fˆ2 |1, 1i = −|1, 0i.
0 0 0 0
0 0 0 0
Be careful about the minus sign here.
0 −i 0 0
0 1 0 0
i 0 0 0
1 0 0 0
η̂1 is = σ0 ⊗ σ1 , η̂2 is = σ0 ⊗ σ2 ,
0 0 0 1 0 0 0 − i
0 0 1 0 0 0 i 0
0 0 −i 0
0 0 1 0
0 0 0 i
0 0 0 −1
η̂3 is = σ1 ⊗ σ3 , η̂4 is = σ2 ⊗ σ3 .
1 0 0 0 i 0 0 0
0 −1 0 0 0 −i 0 0
(e) From the result of (c), L̂x is (−σ3 ⊗ σ0 ) under the basis of (a), which is already
diagonal.
In fact L̂x = 2fˆ2† fˆ2 − 1, according to the definition of η̂i .
The occupation basis are normalized eigenstates of L̂x .
L̂x has
eigenvalue +1 for |n1 = 0, n2 = 1i and |n1 = 1, n2 = 1i;
and eigenvalue −1 for |n1 = 0, n2 = 0i and |n1 = 1, n2 = 0i.
(f) From the result of (c), L̂z is (−σ1 ⊗ σ1 ) under the basis of (a).
This is similar to the operator in Homework #1 Problem 5(a), with an additional minus
sign.
σ1 has eigenvalue +1 with normalized eigenvector √1 (1, 1)T ; and eigenvalue −1 with
2
normalized eigenvector √1 (1, −1).
2
Therefore L̂z has
eigenvalue +1 for normalized eigenvectors √1 (1, 1)T ⊗ √1 (1, −1)T = 1
(1, −1, 1, −1)T and
2 2 2
√1 (1, −1)T ⊗ √1 (1, 1)T = 21 (1, 1, −1, −1)T ; and
2 2
eigenvalue −1 for normalized eigenvectors √1 (1, 1)T ⊗ √1 (1, 1)T = 1
(1, 1, 1, 1)T and
2 2 2
Further use L̂z L̂x = −L̂x L̂z = −iL̂y , and L̂z L̂y = −L̂y L̂z = iL̂x , we have
exp(iθL̂z ) · L̂x · exp(−iθL̂z ) = (cos θ + i sin θL̂z ) · L̂x · (cos θ − i sin θL̂z )
= (cos θL̂x + sin θL̂y ) · (cos θ − i sin θL̂z ) = (cos2 θ − sin2 θ)L̂x + 2 cos θ sin θL̂y
= cos(2θ)L̂x + sin(2θ)L̂y , and
exp(iθL̂z ) · L̂y · exp(−iθL̂z ) = (cos θ + i sin θL̂z ) · L̂y · (cos θ − i sin θL̂z )
= (cos θL̂y − sin θL̂y ) · (cos θ − i sin θL̂z ) = (cos2 θ − sin2 θ)L̂y − 2 cos θ sin θL̂x
= cos(2θ)L̂y − sin(2θ)L̂x , and obviously
exp(iθL̂z ) · L̂z · exp(−iθL̂z ) = L̂z .
These lead to exp(iθL̂z ) · (c1 L̂x + c2 L̂y + c3 L̂z ) · exp(−iθL̂z )
= c1 · [cos(2θ)L̂x + sin(2θ)L̂y ] + c2 · [cos(2θ)L̂y − sin(2θ)L̂x ] + c3 L̂z
= c3 L̂z + L̂x [c1 cos(2θ) − c2 sin(2θ)] + L̂y [c2 cos(2θ) + c1 sin(2θ)]