This action might not be possible to undo. Are you sure you want to continue?
Solutions to selected problems
Chapter 1
1.1 Yes, because in this case even in the initial state some properties would not be deter
mined and therefore also the properties of future states could not be precisely deﬁned.
As a consequence, it would be also impossible to predict future properties.
1.2 Not necessarily. In principle, one can imagine the existence of a “discrete” world.
However, it would still be necessary to have some laws allowing to infer determinis
tically a future (or past) state from a given initial (or ﬁnal) state.
1.4 Consider the Jacobi identity (Eq. (1.10d)) and take h = H. The ﬁrst two terms of the
rhs are then identically zero and therefore
{H, { f , g}} = 0.
1.5 ℘(45
◦
) = 1. ℘(135
◦
) = 0. As a matter of fact, since the component states [ :) and
[ h) are equally wighted in the superposition (1.79), the latter represents the state
of a photon with polarization oriented at 45
◦
relative to the horizontal axis. On the
contrary, such a state will not pass the ﬁlter at 135
◦
as this orientation is orthogonal
to the polarization of the state (1.79).
1.6 Since [ j ) and [ k) are elements of an orthonormal basis, we have ! j [ k) for j ,= k
and 0 1 for j = k.
1.7
ˆ
P
k
[ψ) = c
k
[k).
1.9 The norm of the vector [ ϕ) = c
k
[k) is  ϕ = [c
k
[
2
. It is  ϕ ≤ 1 since
¸
N
j =1
[c
j
[
2
= 1, being  ϕ = 1 if and only if c
j
= δ
j k
. Physically this means that
the projector acts as a ﬁlter and therefore it selects a subensemble of the initial
ensemble.
1.10 Let us assume that [ ξ) is in an ndimensional Hilbert space and expand it as
[ ξ) =
¸
j
c
j
[ j ) ,
where [ j ) is an orthonormal basis for this space. It is then easy to show that
c
∗
1
c
∗
2
. . . c
∗
n
⎛
⎜
⎜
⎝
c
1
c
2
. . .
c
n
⎞
⎟
⎟
⎠
= 1,
since
¸
n
j =1
c
j
2
= 1 according to the normalization condition.
1.11 We can imagine of having N “boxes” (one for each oscillator) each of them with a
certain number of energy quanta c(ν). We are able to calculate n
E
, which becomes
9780521869638sol CUP/AUL November 6, 2008 18:21 Page2
2 Sol ut i ons to sel ected probl ems
t
the total number of different ways in which we can arrange the n
c
quanta among the
N oscillators (or boxes), i.e. the total number of permutations of the quanta n
c
plus
the N −1 partitions
1
n
E
=
(N −1 ÷n
c
)!
(N −1)! n
c
!
.
(N ÷n
c
)
N÷n
c
N
N
n
n
c
c
,
where in the last equality we have made use of the Stirling formula
m! . m
m
e
−m
√
2πm,
which is a very good approximation for large m. Consequently, the total entropy is
given by
S
N
B
= k
B
N
¸
1 ÷
n
c
N
ln (N ÷n
c
) −ln N −
n
c
N
ln n
c
¸
.
By making use of Eq. (1.62), we obtain the desired result.
1.12 In order to ﬁnd the solution, let us ﬁrst compute the differential
∂S
N
B
∂U
=
∂S
N
B
∂ N
¨
E
=
1
N
∂S
N
B
∂
¨
E
=
k
B
c
¸
ln
1 ÷
¨
E
c
−ln
¨
E
c
¸
=
k
B
c
ln
¨
E ÷c
¨
E
=
1
T
.
From the last equality it follows that
¨
E ÷c
¨
E
= e
c¡k
B
T
,
and, ﬁnally,
¨
E =
c
e
c¡k
B
T
−1
.
1.13 Applying the Carnot theorem to the upper triangle in Fig. 1.19 we obtain
(n:)
2
=
hν
s
c
2
÷
hν
i
c
2
−2
hν
s
c
hν
i
c
cos θ.
On the other hand energy conservation (Eq. (1.70)) yields
m:
2
= 2h (ν
i
−ν
s
).
In order to eliminate :, we multiply this last equation times m and equate the resulting
rhs to the rhs of the momentum conservation equation and obtain
ν
i
−ν
s
=
h
2mc
2
ν
2
s
÷ν
2
i
−2ν
s
ν
i
cos θ
.
1
See [Huang 1963, 181–82].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page3
3 Sol ut i ons to sel ected probl ems
t
Now, since the wavelength of the scattered photon is only slightly larger than the
original one, we have
ν
i
. ν
s
→ ν
2
i
. ν
2
s
. ν
i
ν
s
.
Consequently,
ν
i
−ν
s
=
h
mc
2
ν
i
ν
s
(1 −cos θ).
Replacing ν
i
=
c
λ
i
and ν
s
=
c
λ
s
we ﬁnally obtain
λ
s
−λ
i
=
h
mc
(1 −cos θ) =
2h
mc
sin
2
θ
2
.
Chapter 2
2.1 Take e.g.
ˆ
O
/
=
ˆ
I −
ˆ
O.
2.2
!h [ ψ)
!: [ ψ)
=
¸
!h [ b) !h [ b
⊥
)
!: [ b) !: [ b
⊥
)
¸
!b [ ψ)
!b
⊥
[ ψ)
=
!h [ b) !b [ ψ) ÷!h [ b
⊥
) !b
⊥
[ ψ)
!: [ b) !b [ ψ) ÷!: [ b
⊥
) !b
⊥
[ ψ)
=
!h [ ψ)
!: [ ψ)
.
The last passage is due to the wellknown property of projectors
[b ) !b[ ÷[b
⊥
) !b
⊥
[ =
ˆ
P
b
÷
ˆ
P
b
⊥
=
ˆ
I .
2.3 Let us write the transposed conjugate of
ˆ
U:
ˆ
U
†
=
¸
!h [ b)
∗
!: [ b)
∗
!h [ b
⊥
)
∗
!: [ b
⊥
)
∗
¸
=
¸
!b [ h) !b [ :)
!b
⊥
[ h) !b
⊥
[ :)
¸
.
Then we have
ˆ
U
ˆ
U
†
=
¸
!h [ b) !h [ b
⊥
)
!: [ b) !: [ b
⊥
)
¸ ¸
!b [ h) !b [ :)
!b
⊥
[ h) !b
⊥
[ :)
¸
=
¸
!h [ b) !b [ h) ÷!h [ b
⊥
) !b
⊥
[ h) !h [ b) !b [ :) ÷!h [ b
⊥
) !b
⊥
[ :)
!: [ b) !b [ h) ÷!: [ b
⊥
) !b
⊥
[ h) !: [ b) !b [ :) ÷!: [ b
⊥
) !b
⊥
[ :)
¸
=
¸
!h [ ([ b) !b[ ÷[ b
⊥
) !b
⊥
[) [ h) !h [ ([ b) !b[ ÷[ b
⊥
) !b
⊥
[) [ :)
!: [ ([ b) !b[ ÷[ b
⊥
) !b
⊥
[) [ h) !: [ ([ b) !b[ ÷[ b
⊥
) !b
⊥
[) [ :)
¸
=
¸
!h [ h) !h [ :)
!: [ h) !: [ :)
¸
=
¸
1 0
0 1
¸
=
ˆ
I ,
since ([ b) !b[ ÷[ b
⊥
) !b
⊥
[) =
ˆ
I . Similarly it can be shown that
ˆ
U
†
ˆ
U =
ˆ
I .
9780521869638sol CUP/AUL November 6, 2008 18:21 Page4
4 Sol ut i ons to sel ected probl ems
t
2.6 From the condition of unitarity
ˆ
O
ˆ
O
†
=
¸
a b
c d
¸ ¸
a
∗
c
∗
b
∗
d
∗
¸
=
¸
[a[
2
÷[b[
2
ac
∗
÷bd
∗
a
∗
c ÷b
∗
d [c[
2
÷[d[
2
¸
=
¸
1 0
0 1
¸
we derive the following relations
[a[
2
÷[b[
2
= 1,
[c[
2
÷[d[
2
= 1,
ac
∗
÷bd
∗
= 0.
Similarly, from the condition of Hermiticity
¸
a b
c d
¸
=
¸
a
∗
c
∗
b
∗
d
∗
¸
we obtain that a and d are real and that c = b
∗
. Collecting these results, we derive
that b(a ÷d) and a
2
= d
2
. From these relations it follows either that a = d, and in
this case b = 0 and
ˆ
O is a multiple of identity, or that a = −d, and in this case we
have
ˆ
O =
¸
a b
b
∗
−a
¸
,
2.7 The solution of the problem is
b
ˆ
O
P
b
=
!: [ c
∗
:
÷ !h [ c
∗
h
([ :) !: [ −[ h) !h [) (c
:
[ :) ÷c
h
[ h) )
=
!: [ c
∗
:
÷ !h [ c
∗
h
(c
:
[ :) −c
h
[ h) )
= [c
:
[
2
−[c
h
[
2
.
2.8 Let us take a generic ket [ ψ) . Then,
ˆ
O [ ψ) =
¸
j
ˆ
O
o
j
o
j
[ ψ
=
¸
j
o
j
o
j
o
j
[ ψ
=
¸
j
o
j
ˆ
P
j
[ ψ) ,
where the
o
j
are eigenkets of
ˆ
O. Since these equalities must be valid for any [ ψ) ,
it follows that
ˆ
O =
¸
j
o
j
ˆ
P
j
.
2.9 We proceed as follows:
9780521869638sol CUP/AUL November 6, 2008 18:21 Page5
5 Sol ut i ons to sel ected probl ems
t
(a) Let us take the usual basis
[ b
1
) =
1
0
, [ b
2
) =
0
1
We have to ﬁnd the vectors [ o
1
) and [ o
2
) and the (complex) numbers o
1
and o
2
such that
ˆ
O [ o
1
) = o
1
[ o
1
) ,
ˆ
O [ o
2
) = o
2
[ o
2
) .
The characteristic polynomial is given by
λ ı
−ı λ
= λ
2
−1,
whose zeros (eigenvalues) are λ
1,2
= ±1. Let us take the eigenvalues o
1
= ÷1
and o
2
= −1. Rewriting explicitely the ﬁrst eigenvalue equation, we must have
¸
0 −ı
ı 0
¸
o
1
1
o
2
1
=
o
1
1
o
2
1
that yields o
2
1
= ı o
1
1
. The normalized eigenvector [ o
1
) will then given by
[ o
1
) =
1
√
2
1
ı
.
With a similar procedure we ﬁnd
[ o
2
) =
1
√
2
1
−ı
.
Notice that
ˆ
O is Hermitian, [ o
1
) and [ o
2
) are orthogonal, i.e. !o
1
[ o
2
) = 0.
We have constructed the eigenvectors so that they are also normalized, that
is
o
j
[ o
k
= δ
j k
. In other words, {[ o
1
) , [ o
2
) } is an orthonormal basis on the
bidimensional Hilbert space.
(b) The diagonalizing matrix is simply given by
ˆ
U =
1
√
2
¸
1 1
ı −ı
¸
.
Finally, the diagonal form of
ˆ
O is
ˆ
O
/
=
ˆ
U
†
ˆ
O
ˆ
U
=
1
2
¸
1 −ı
1 ÷ı
¸ ¸
0 −ı
ı 0
¸ ¸
1 1
ı −ı
¸
=
1
2
¸
1 −ı
1 ÷ı
¸ ¸
1 −1
ı ı
¸
=
1
2
¸
2 0
0 −2
¸
=
¸
1 0
0 −1
¸
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page6
6 Sol ut i ons to sel ected probl ems
t
2.10 The result is
9
16
N.
2.15 The momentum eigenfunctions in the momentum representation are
˜ ϕ
p
0
( p
x
) =
1
√
2π
dxe
−
ı
¨ h
p
x
x
ϕ
p
0
(x)
=
1
√
2π
dxe
−
ı
¨ h
( p
x
−p
0
)x
= δ( p
x
− p
0
).
2.16 The position eigenfunctions in the momentum representation are
˜ ϕ
x
0
( p
x
) =
1
√
2π
dxe
−
ı
¨ h
p
x
x
ϕ
x
0
(x)
=
1
√
2π
dxe
−
ı
¨ h
p
x
x
δ(x − x
0
)
=
1
√
2π
e
−
ı
¨ h
p
x
x
0
.
In the general case we have
˜ ϕ
x
( p
x
) =
1
√
2π
e
−
ı
¨ h
p
x
x
.
2.17 We proceed as follows:
÷∞
−∞
dp
x
[
˜
ψ( p
x
)[
2
=
÷∞
−∞
dp
x
˜
ψ( p
x
)
˜
ψ
∗
( p
x
)
=
÷∞
−∞
dp
x
1
√
2π
÷∞
−∞
dxψ(x)e
−
ı
¨ h
p
x
x
1
√
2π
÷∞
−∞
dx
/
ψ
∗
(x
/
)e
ı
¨ h
p
x
x
/
=
÷∞
−∞
dx
÷∞
−∞
dx
/
ψ(x)ψ
∗
(x
/
)
1
2π
÷∞
−∞
dp
x
e
ı
¨ h
p
x
(x−x
/
)
=
÷∞
−∞
dx
÷∞
−∞
dx
/
ψ(x)ψ
∗
(x
/
)δ(x − x
/
)
=
÷∞
−∞
dx[ψ(x)[
2
.
2.18 First, calculate
xψ(x) =
1
√
2π
÷∞
−∞
dp
x
x
˜
ψ( p
x
)e
ı
¨ h
p
x
x
=
1
√
2π
˜
ψ( p
x
)
¨
h
ı
e
ı
¨ h
p
x
x
÷∞
−∞
−
1
√
2π
÷∞
−∞
dp
x
¨
h
ı
e
ı
¨ h
p
x
x
∂
˜
ψ( p
x
)
∂p
x
=
ı
¨
h
√
2π
÷∞
−∞
dp
x
e
ı
¨ h
p
x
x
∂
˜
ψ( p
x
)
∂p
x
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page7
7 Sol ut i ons to sel ected probl ems
t
where we have made use of the inverse Fourier transform (2.150a) and of integration
by parts. If we substitute this result into the deﬁnition of the expectation value of ˆ x,
we obtain
! ˆ x) =
÷∞
−∞
dxψ
∗
(x)
1
√
2π
÷∞
−∞
dp
x
ı
¨
he
ı
¨ h
p
x
x
∂
˜
ψ( p
x
)
∂p
x
=
÷∞
−∞
dp
x
¸
1
√
2π
÷∞
−∞
dxψ
∗
(x)e
ı
¨ h
p
x
x
¸
ı
¨
h
∂
˜
ψ( p
x
)
∂p
x
=
÷∞
−∞
dp
x
˜
ψ
∗
( p
x
)ı
¨
h
∂
˜
ψ( p
x
)
∂p
x
.
This has to be equal to the expectation value of ˆ x computed in the momentum
representation, i.e.
! ˆ x) =
÷∞
−∞
dp
x
˜
ψ
∗
( p
x
) ˆ x
˜
ψ( p
x
).
By comparison, we ﬁnd
ˆ x
˜
ψ( p
x
) = ı
¨
h
∂
˜
ψ( p
x
)
∂p
x
.
2.19 We have
ˆ
P(x) [ p
x
) = [ x) !x [ p
x
) = ϕ
p
(x) [ x),
where ϕ
p
(x) [ x) is the eigenfunction of the momentum operator corresponding to the
eigenvalue p
x
, that is
ϕ
p
(x) =
1
√
2π
e
ı
¨ h
p
x
x
.
This shows that the action of the position projector
ˆ
P(x) onto the momentum eigen
vector [ p
x
) (in which the position is completely undetermined) selects the position
eigenvector [ x) with a weight that is given by ϕ
p
(x).
2.22 The delta function can be obtained as a limit of a normalized Gaussian
f
a
(x) =
1
√
2πa
e
−
x
2
2a
2
,
with a > 0, that is
δ(x) = lim
a→0
f
a
(x).
Then, we have
δ
2
(x) = lim
a→0
f
2
a
(x)
= lim
a→0
1
2πa
2
e
−
x
2
a
2
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page8
8 Sol ut i ons to sel ected probl ems
t
from which we obtain
÷∞
−∞
dxδ
2
(x) =
1
2π
÷∞
−∞
dx lim
a→0
1
a
2
e
−
x
2
a
2
=
1
2π
lim
a→0
1
a
2
÷∞
−∞
dxe
−
x
2
a
2
=
1
2
√
π
lim
a→0
1
a
= ÷∞.
2.24 LxLE ≥
1
2
¨ h p
0
m
.
2.26 The solution is given by
¸
ˆ p
x
, f ( ˆ x)
¸
=
¨
h
ı
∂
∂x
f ( ˆ x) −
¨
h
ı
f ( ˆ x)
∂
∂x
=
¨
h
ı
f
/
( ˆ x) ÷
¨
h
ı
f ( ˆ x)
∂
∂x
−
¨
h
ı
f ( ˆ x)
∂
∂x
=
¨
h
ı
f
/
( ˆ x)
2.30 With T = 0 and R = 1 we have
[ f
1
) = −
1
√
2
e
ı φ
([ 3) ÷ı [ 4)),
which, up to the irrelevant global phase factor, corresponds to the state [ 1) . With
T = 1 and R = 0, we have
[ f
0
) =
1
√
2
(−[ 3) ÷ı [ 4)),
which corresponds to the state [ 2) . Finally, with T = R = 1¡
√
2, we have
f
1¡2
=
1
2
¸
−
1 ÷e
ı φ
[ 3) ÷ı
1 −e
ı φ
[ 4)
¸
=
1
√
2
¸
1
√
2
(−[ 3) ÷ı [ 4) ) −
1
√
2
e
ı φ
([ 3) ÷ı [ 4))
¸
=
1
√
2
[ f
0
) −e
ı φ
[ f
1
)
.
2.31 The examination of Sec. 2.4 shows that there are cases in which an event occurs
(a certain detector clicks, a state of affairs described by the proposition c) and not
withstanding the object system is neither in the state expressed by the proposition a
/
,
nor in the state expressed by the proposition a
//
. This means that, in these cases, it is
true that
¸
c ∧
a
/
∨ a
//
¸
∧ ¬
¸
c ∧ a
/
∨
c ∧ a
//
,
from which we easily derive that it is also true that ¬
¸
c ∧ a
/
∨
c ∧ a
//
¸
and,
therefore, that
¬
c ∧ a
/
and ¬
c ∧ a
//
9780521869638sol CUP/AUL November 6, 2008 18:21 Page9
9 Sol ut i ons to sel ected probl ems
t
are also true. This means that, in the ﬁrst case, either c or a
/
must be false, and, in
the second case either c or a
//
. However, we have assumed c to be true. Then, both
a
/
and a
//
are false. This means that we have a situation in which a
/
∨ a
//
is true (by
deﬁnition) but also both ¬a
/
and ¬a
//
are. By substituting a
/
to ¬a
/
as well as a
//
to
¬a
//
we obtain that also ¬
a
/
∧ a
//
is true, which is the desired result.
Chapter 3
3.2 It is straightforward to prove the result given the linearity of the Schrödinger equation.
3.3 ψ(x) =
1
√
2π
e
±ı kx
. These are the plane waves described in Subsecs. 2.2.4 and 2.2.6.
They are doubly degenerate: in fact, the plus and minus signs correspond to “waves”
moving from left to right and from right to left, respectively, both having the same
positive energy E =
¨ h
2
k
2
2m
.
3.4 At time t
0
= 0 and t we have
˜
ψ( p
x
, 0) =
¸
n
c
(0)
n
˜
ψ
n
( p
x
) and
˜
ψ( p
x
, t ) =
¸
n
e
−
ı
¨ h
E
n
t
c
(0)
n
˜
ψ
n
( p
x
)
where
˜
ψ
n
( p
x
) = ! p
x
[ ψ
n
) and
˜
ψ
n
( p
x
, t ) = ! p
x
[ ψ(t )).
3.5 Let us rewrite Eq. (3.26) as
ψ(x, t ) =
¸
n
c
n
(t )ψ
n
(x),
from which we obtain
dx [ψ(x, t )[
2
=
dx
¸
m
c
∗
m
(t )ψ
∗
m
(x)
¸
n
c
n
(t )ψ
n
(x)
=
¸
n,m
c
n
(t )c
∗
m
(t )
dxψ
∗
m
(x)ψ
n
(x)
=
¸
n
[c
n
(t )[
2
=
¸
n
e
−
ı
¨ h
E
n
t
c
n
(0)
2
=
¸
n
[c
n
(0)[
2
= 1.
where
dxψ
∗
m
(x)ψ
n
(x) = δ
n,m
.
3.7 We have
ψ(t ) =
dkc(k)e
ı (kx−ω
k
t )
, where ω
k
=
E
¨
h
=
¨
h
k
2
2m
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page10
10 Sol ut i ons to sel ected probl ems
t
3.10 First let us compute the mean value of the position
2
ˆ x
=
a
0
dxψ
∗
n
(x)xψ
n
(x)
=
2
a
a
0
dxx sin
2
nπ
a
x
=
2a
(nπ)
2
nπ
0
dyy sin
2
y
=
a
2
,
where we have made use of the fact that we are in the position representation and that
ψ
∗
n
(x) = ψ
n
(x), as well as the mean value of the square of the position
3
ˆ x
2
=
2
a
a
0
dxx
2
sin
2
nπ
a
x
=
2a
2
(nπ)
3
nπ
0
dyy
2
sin
2
y
=
a
2
3
−
1
2
a
nπ
2
.
Analogously, we calculate the mean value of the momentum
ˆ p
x
= −ı
¨
h
2
a
a
0
dx sin
nπ
a
x
∂
∂x
sin
nπ
a
x
=
2
¨
h
ı a
nπ
a
a
0
dx sin
nπ
a
x
cos
nπ
a
x
=
h
2ı a
¸
−
1
2π
cos
2nπ
a
x
¸
a
0
= 0,
and of the square of the momentum
ˆ p
2
x
= −
¨
h
2
2
a
a
0
dx sin
nπ
a
x
∂
2
∂x
2
sin
nπ
a
x
=
2
¨
h
2
a
n
2
π
2
a
2
a
0
dx sin
2
nπ
a
x
2
See [Gradstein/Ryshik 1981, 3.821].
3
See [Gradstein/Ryshik 1981, 2.631.2].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page11
11 Sol ut i ons to sel ected probl ems
t
=
h
2
n
2a
2
π
¸
1
2
nπ
a
x
−sin
nπ
a
x
cos
nπ
a
x
¸
nπ
0
=
h
2
n
2
4a
2
.
Then, we calculate the uncertainties of position and momentum
Lp
x
=
nh
2a
, Lx = a
1
12
−
1
2n
2
π
2
,
and ﬁnally obtain the uncertainty relation
LxLp
x
=
¨
h
2
⎡
⎣
n
2
π
2
3
−2
⎤
⎦
.
Since the square root is certainly a growing function of n for n ≥ 1, in order to verify
that the uncertainty relation is always satisﬁed it is sufﬁcient to prove that, for n = 1,
it is ≥1. A direct calculation shows that
π
2
3
−2 = 1.136.
3.11 The threedimensional stationary Schrödinger equation for the wave function in the
position representation reads as
¸
−
¨
h
2
2m
∂
2
∂x
2
÷
∂
2
∂y
2
÷
∂
2
∂z
2
÷ V(x, y, z)
ψ(x, y, z) = Eψ(x, y, z).
For a particle in a “cubic” box, we have V(x, y, z) = 0 inside the box, so that the
previous equation becomes
−
¨
h
2
2m
∂
2
∂x
2
÷
∂
2
∂y
2
÷
∂
2
∂z
2
ψ(x, y, z) = Eψ(x, y, z).
with the boundary conditions
ψ(x, y, z) = 0, for x  0, x > a ; y  0, y > b ; z  0, z > c.
This problem is separable, i.e.
ˆ
H =
ˆ
H
1
(x) ÷
ˆ
H
2
(y) ÷
ˆ
H
3
(z), where
ˆ
H
1
(x) = −
¨
h
2
2m
∂
2
∂x
2
,
and similar expressions for
ˆ
H
2
(y) and
ˆ
H
3
(z). In
ψ(x, y, z) = ψ
(1)
(x) · ψ
(2)
(y) · ψ
(3)
(z),
and
E = E
x
÷ E
y
÷ E
z
.
The required solution therefore reduces to
ψ
n
x
,n
y
,n
z
(x, y, z) = ψ
(1)
n
x
(x) · ψ
(2)
n
y
(y) · ψ
(3)
n
z
(z),
9780521869638sol CUP/AUL November 6, 2008 18:21 Page12
12 Sol ut i ons to sel ected probl ems
t
with
ψ
(1)
n
x
(x) =
2
a
sin
n
x
π
a
x
,
ψ
(2)
n
y
(y) =
2
b
sin
n
y
π
b
y
,
ψ
(3)
n
z
(z) =
2
c
sin
n
z
π
c
z
,
and
E
n
x
=
π
¨
h
2
2ma
2
n
2
x
,
E
n
y
=
π
¨
h
2
2mb
2
n
2
y
,
E
n
z
=
π
¨
h
2
2mc
2
n
2
z
.
3.12 Referring to Fig. 3.8(b), the transformation effected by an asymmetric beam splitter
should satisfy the following constraints:
ˆ
U
BS
1
0
∝
T
ı R,
ˆ
U
BS
0
1
∝
ı R
T
.
The requirement of unitarity leads us to the ﬁnal form
ˆ
U
BS
=
¸
T ı R
∗
ı R T
∗
¸
.
3.13 We have
ˆ
U
t
[ ψ) = e
−
ı
¨ h
ˆ
Ht
¸
j
c
j
ψ
j
=
¸
j
c
j
e
−
ı
¨ h
E
j
t
ψ
j
,
where use has been made of the eigenvalue equation
ˆ
H
ψ
j
= E
j
ψ
j
.
Now, it is clear that
c
j
(t ) = c
j
e
−
ı
¨ h
E
j
t
,= 0, if c
j
,= 0.
Furthermore, we also have that the probabilities of the energy eigenvalues ℘
j
(t ) =
c
j
(t )
2
= ℘
j
(0) are constant under time evolution.
Note that, in the case of beamsplitting (see Subsec. 3.5.2), the vectors
[ 1) =
1
0
and [ 2) =
0
1
9780521869638sol CUP/AUL November 6, 2008 18:21 Page13
13 Sol ut i ons to sel ected probl ems
t
are neither eigenvectors of the beamsplitter unitary transformation
ˆ
U
BS
=
1
√
2
¸
1 ı
ı 1
¸
.
nor of the Hermnitian operator
ˆ
O that is the generator of the transformation, i.e.
ˆ
U
BS
= e
ı
ˆ
O
. This is the reason why certain superposition states of the basis vectors
[ 1) and [ 2) may be transformed into [ 1) or [ 2) under the unitary transformation
ˆ
U
BS
. For instance, the state
1
√
2
([ 1) −ı [ 2) ) is transformed by
ˆ
U
BS
into [ 1) .
3.14 This result may be proved by taking into account the uniqueness of the unitary
transformation
ˆ
U
†
=
ˆ
U
−1
(its deterministic nature). If this connection had to be not
completely clear, see Sec. 15.2.
3.15 We have
ˆ
H
H
(t ) =
ˆ
U
†
t
ˆ
H
S
ˆ
U
t
=
ˆ
U
†
t
ˆ
U
t
ˆ
H
S
=
ˆ
H
S
,
since
ˆ
H
S
=
ˆ
H commutes with
ˆ
U
t
= e
−
ı
¨ h
ˆ
Ht
.
3.17 We have to show that
¸¸
ˆ
O,
ˆ
O
/
,
ˆ
H
¸
= 0.
The explicit calculation is
¸¸
ˆ
O,
ˆ
O
/
,
ˆ
H
¸
=
¸
ˆ
O
ˆ
O
/
,
ˆ
H
¸
−
¸
ˆ
O
/
ˆ
O,
ˆ
H
¸
=
ˆ
O
ˆ
O
/
ˆ
H −
ˆ
H
ˆ
O
ˆ
O
/
−
ˆ
O
/
ˆ
O
ˆ
H ÷
ˆ
H
ˆ
O
/
ˆ
O
=
ˆ
H
ˆ
O
ˆ
O
/
−
ˆ
H
ˆ
O
ˆ
O
/
−
ˆ
H
ˆ
O
/
ˆ
O ÷
ˆ
H
ˆ
O
/
ˆ
O
= 0,
since
ˆ
H commutes with both
ˆ
O and
ˆ
O
/
.
3.19 We can proceed as follows:
ı
¨
h
d
dt
[ ψ(t ))
I
= ı
¨
h
d
dt
e
ı
¨ h
ˆ
H
0
t
[ ψ(t ))
S
= ı
¨
h
¸
ı
¨
h
ˆ
H
0
e
ı
¨ h
ˆ
H
0
t
[ ψ(t ))
S
÷e
ı
¨ h
ˆ
H
0
t
d
dt
[ ψ(t ))
S
¸
=
¸
−
ˆ
H
0
ˆ
U
†
H
0
,t
÷
ˆ
U
†
H
0
,t
ˆ
H
¸
[ ψ(t ))
S
=
¸
−
ˆ
U
†
H
0
,t
ˆ
H
0
÷
ˆ
U
†
H
0
,t
ˆ
H
¸
ˆ
U
H
0
,t
[ ψ(t ))
I
=
ˆ
U
†
H
0
,t
¸
−
ˆ
H
0
÷
ˆ
H
¸
ˆ
U
H
0
,t
[ ψ(t ))
I
=
ˆ
U
†
H
0
,t
ˆ
H
I
ˆ
U
H
0
,t
[ ψ(t ))
I
=
ˆ
H
I
I
(t ) [ ψ(t ))
I
.
3.22 Given any two operators
ˆ
O and
ˆ
O
/
such that [
ˆ
O,
ˆ
O
/
] = −ı
¨
h, it is straightforward to
prove that (see Prob. 2.25)
¸
ˆ
O
n
,
ˆ
O
/
¸
= −nı
¨
h
ˆ
O
n−1
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page14
14 Sol ut i ons to sel ected probl ems
t
Using this result, we may compute the commutator
¸
e
ı αˆ t
,
ˆ
H
¸
=
⎡
⎣
∞
¸
j =0
(ı α)
j
j !
ˆ t
j
,
ˆ
H
⎤
⎦
=
∞
¸
j =0
(ı α)
j
j !
¸
ˆ t ,
ˆ
H
¸
= −
∞
¸
j =1
(ı α)
j
j !
j ı
¨
hˆ t
j −1
= α
¨
h
∞
¸
j =1
(ı α)
j −1
( j −1)!
ˆ t
j −1
= α
¨
h
∞
¸
n=0
(ı α)
n
(n)!
ˆ t
n
= α
¨
he
ı αˆ t
.
Therefore,
ˆ
He
ı αˆ t
= e
ı αˆ t
ˆ
H −α
¨
he
ı αˆ t
,
which yields the desired result.
Chapter 4
4.1 The energy levels are
E(i , j , k) = E
x
(i ) ÷ E
y
( j ) ÷ E
z
(k) =
π
¨
h
2
2m
i
2
a
2
÷
j
2
b
2
÷
k
2
c
2
,
and the wave functions are given by
ψ
i , j ,k
(r) = ψ
i
(x)ψ
j
(y)ψ
k
(z) =
8
abc
sin
i π
a
x
sin
j π
b
y
sin
kπ
c
z
.
4.3 (a) The normalization coefﬁcients N
÷
and N
−
may be derived from the continuity
of the wave function and its ﬁrst derivative at x = 0, that is,
ψ

(0) = ψ
>
(0),
ψ
/

(0) = ψ
/
>
(0),
from which we obtain the conditions
N
÷
= 1 ÷N
−
,
k
1
N
÷
= k
2
−k
2
N
−
.
A simple calculation yields the desired result
N
−
=
k
2
−k
1
k
1
÷k
2
,
N
÷
=
2k
2
k
1
÷k
2
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page15
15 Sol ut i ons to sel ected probl ems
t
(b) Equations (4.29), together with the previous result, yield
T
2
R
2
=
k
1
k
2
[N
÷
[
2
[N
−
[
2
=
4k
1
k
2
(k
2
−k
1
)
2
=
4
1 −
V
0
E
2 −
V
0
E
−2
1 −
V
0
E
,
from which we ﬁnally obtain that T
2
¡R
2
= 16¡15.
It should also be noted that, when E = V
0
, the ratio T
2
¡R
2
vanishes, and we
have a total reﬂection of the particle at the potential barrier. However, even in this
case (and also for E  V
0
), there is a nonzero probability of ﬁnding the particle
in the classically forbidden region. Nevertheless, this probability exponentially
vanishes with x.
4.4 (a) For E > V
0
, the wave function is
for x ≤ 0, ψ
I
(x) = e
ı k
1
x
÷ Ae
−i mat hk
1
x
,
for 0 ≤ x ≤ a, ψ
II
(x) = Be
ı k
2
x
÷ B
/
e
−ı k
2
x
,
for x ≥ a, ψ
III
(x) = Ce
ı k
1
x
,
where
k
1
=
1
¨
h
√
2mE, k
2
=
1
¨
h
2m(E − V
0
).
The constants A, B, B
/
, and C may be derived from the conditions
ψ
I
(0) = ψ
II
(0), ψ
/
I
(0) = ψ
/
II
(0),
ψ
II
(a) = ψ
III
(a), ψ
/
II
(a) = ψ
/
III
(a),
which yield
1 ÷ A = B ÷ B
/
,
k
1
−k
1
A = k
2
B −k
2
B
/
,
Be
ı k
2
a
÷ B
/
e
−ı k
2
a
= Ce
ı k
1
a
,
k
2
Be
ı k
2
a
−k
2
B
/
e
−ı k
2
a
= k
1
Ce
ı k
1
a
.
From these conditions we obtain
A =
ı
k
2
2
−k
2
1
k
1
k
2
sin(k
2
a)
2 cos(k
2
a) −ı
k
2
1
÷k
2
2
k
1
k
2
sin(k
2
a)
.
In this case, we have
R
2
= [ A[
2
and T
2
= k
1
[C[
2
k
1
= [C[
2
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page16
16 Sol ut i ons to sel ected probl ems
t
From T
2
÷R
2
= 1, we immediatly obtain
T
2
= [C[
2
= 1 −[ A[
2
=
4k
2
1
k
2
2
4k
2
1
k
2
2
÷
k
2
1
−k
2
2
2
sin
2
(k
2
a)
.
(b) For E  V
0
, k
2
becomes pure imaginary, which implies that the wave func
tion decreases exponentially in region II. The expression for the transmission
coefﬁcient may be obtained from the previous result if one replaces k
2
by ı κ
2
,
where
κ
2
=
1
¨
h
2m(V
0
− E).
4.6 We may use an inductive argument. First, we have to verify that Eq. (4.70) is satisﬁed
for n = 0 (see Eq. (4.69)). Then, in order to solve the problem, it sufﬁces to showthat,
if Eq. (4.70) is assumed to be valid for n, i.e.
x
n,n÷1
=
(n ÷1)
¨
h
2mω
,
then it is also satisﬁed for n ÷1. We rewrite Eq. (4.68) fo n ÷1
x
2
n÷1,n
ω
n,n÷1
÷ x
2
n÷1,n÷2
ω
n÷2,n÷1
=
¨
h
2m
,
from which, be using the relations ω
n÷2,n÷1
= ω and ω
n,n÷1
= −ω, we can derive
x
n÷1,n÷2
=
(n ÷2)
¨
h
2mω
,
that completes the argument.
4.7 There at least two simple ways to obtain this result, both of which do not involve
anything but straightforward calculations. The ﬁrst method is direct inspection of
Eq. (4.60). In fact, if k = n, we have that ω
nk
= 0 and therefore x
nk
= 0. The other
method starts from the deﬁnition of
x
nn
=
ˆ x
n
=
n
ˆ x
n
=
÷∞
−∞
dx ψ
∗
n
(x)xψ
n
(x) =
÷∞
−∞
dx ψ
2
n
(x)x,
where we recall that ψ
n
(x) = !x [ n) is the nth harmonic oscillator (real) eigenfunc
tion. These eigenfucntions are either odd or even functions of x. As a consequence,
the integrand of the last equality in the previous equation is necessarily odd, hence
the integral over the whole line is zero.
4.8 We have
n
ˆ p
x
n
=
dx !n [ x) !x [ p
x
[ n) =
dxψ
∗
n
(x)
¨
h
ı
∂
∂x
ψ
n
(x).
Without loss of generality, we may choose the eigenfunctions to be real. Then,
n
ˆ p
x
n
=
¨
h
ı
dxψ
n
(x)
∂
∂x
ψ
n
(x).
9780521869638sol CUP/AUL November 6, 2008 18:21 Page17
17 Sol ut i ons to sel ected probl ems
t
Integrating by parts the previous expression, we arrive at the desired result. The same
result nay be obtained taking into account the fact that the ﬁrst derivative of an even
(odd) function is odd (even).
4.9 We have
¸
ˆ a, ˆ a
†
¸
=
m
2
¨
hω
ω
2
¸
ˆ x, ˆ x
¸
÷ı ω
¸
ˆ
˙ x, ˆ x
¸
−ı ω
¸
ˆ x,
ˆ
˙ x
¸
÷
¸
ˆ
˙ x,
ˆ
˙ x
¸
= −
ı m
¨
h
¸
ˆ x,
ˆ
˙ x
¸
= −
ı
¨
h
¸
ˆ x, ˆ p
x
¸
=
ˆ
I ,
where we have made use of the fact that
¸
ˆ
O,
ˆ
O
¸
= 0, that
¸
ˆ
O,
ˆ
O
/
¸
= −
¸
ˆ
O
/
,
ˆ
O
¸
(see
Eqs. (2.94) and (2.97)), and that
¸
ˆ x, ˆ p
x
¸
= ı
¨
h.
4.10 Starting from Eqs. (4.73), we can write the product
ˆ aˆ a
†
=
m
2
¨
hω
ωˆ x ÷ı
ˆ
˙ x
ωˆ x −ı
ˆ
˙ x
=
1
2
¨
h
mωˆ x
2
÷
ˆ p
2
x
mω
÷ı
¸
ˆ p
x
, ˆ x
¸
=
1
¨
hω
1
2
mω
2
ˆ x
2
÷
ˆ p
2
x
2m
÷
1
2
,
where we have made use of the fact that
ˆ
˙ x =
ˆ p
x
m
and
¸
ˆ p
x
, ˆ x
¸
= −ı
¨
h.
From the above expression and the commutator (Eq. (4.74)) we can derive
ˆ
H =
¨
hω
ˆ aˆ a
†
−
1
2
and
ˆ
H =
¨
hω
2
ˆ aˆ a
†
÷ ˆ a
†
ˆ a
.
4.11 We have
¸
ˆ a,
ˆ
N
¸
=
¸
ˆ a, ˆ a
†
ˆ a
¸
= ˆ aˆ a
†
ˆ a − ˆ a
†
ˆ a ˆ a
=
¸
ˆ a, ˆ a
†
¸
ˆ a = ˆ a,
and
¸
ˆ a
†
,
ˆ
N
¸
=
¸
ˆ a
†
, ˆ a
†
ˆ a
¸
= ˆ a
†
ˆ a
†
ˆ a − ˆ a
†
ˆ aˆ a
†
= ˆ a
†
¸
ˆ a
†
, ˆ a
¸
= −ˆ a
†
.
4.12 We have
¸
ˆ a,
ˆ a
†
2
¸
= ˆ aˆ a
†
ˆ a
†
− ˆ a
†
ˆ a
†
ˆ a
= ˆ aˆ a
†
ˆ a
†
− ˆ a
†
ˆ aˆ a
†
÷ ˆ a
†
ˆ aˆ a
†
− ˆ a
†
ˆ a
†
ˆ a
=
¸
ˆ a, ˆ a
†
¸
ˆ a
†
÷ ˆ a
†
¸
ˆ a, ˆ a
†
¸
= 2ˆ a
†
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page18
18 Sol ut i ons to sel ected probl ems
t
and
¸
ˆ a
2
, ˆ a
†
¸
= ˆ a ˆ aˆ a
†
− ˆ a
†
ˆ a ˆ a
= ˆ a ˆ aˆ a
†
− ˆ aˆ a
†
ˆ a ÷ ˆ aˆ a
†
ˆ a − ˆ a
†
ˆ a ˆ a
= ˆ a
¸
ˆ a, ˆ a
†
¸
÷
¸
ˆ a, ˆ a
†
¸
ˆ a = 2ˆ a,
which can be both generalized by induction. We prove only the ﬁrst one. Assuming
that the relation
¸
ˆ a,
ˆ a
†
n
¸
= n(ˆ a
†
)
n−1
holds for a given n, we have to prove that it also holds for n ÷1, that is
¸
ˆ a,
ˆ a
†
n÷1
¸
= ˆ a
ˆ a
†
n
ˆ a
†
−
ˆ a
†
n
ˆ a
†
ˆ a
= ˆ a
ˆ a
†
n
ˆ a
†
−
ˆ a
†
n
ˆ aˆ a
†
÷
ˆ a
†
n
ˆ aˆ a
†
−
ˆ a
†
n
ˆ a
†
ˆ a
=
¸
ˆ a,
ˆ a
†
n
¸
ˆ a
†
÷
ˆ a
†
n
¸
ˆ a, ˆ a
†
¸
= n
ˆ a
†
n
÷
ˆ a
†
n
= (n ÷1)
ˆ a
†
n
.
4.13 Again, we prove this relation by induction. From the second of Eqs. (4.85) we
immediately obtain
[ 1) = ˆ a
†
[ 0) .
Assuming that the relation holds for a given n, we must prove that it holds for n ÷1
as well, that is,
[ n ÷1) =
ˆ a
†
n÷1
√
(n ÷1) !
[ 0) .
In fact, we have
ˆ a
†
n÷1
√
(n ÷1) !
[ 0) =
ˆ a
†
√
n ÷1
ˆ a
†
n
√
n!
[ 0)
=
ˆ a
†
√
n ÷1
[ n) = [ n ÷1) .
4.14 We must have
1 = [N[
2
÷∞
−∞
dxe
−
mω
¨ h
x
2
.
Using the formula
4
÷∞
−∞
dye
−ay
2
=
π
a
,
4
See [Gradstein/Ryshik 1981, 3.325].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page19
19 Sol ut i ons to sel ected probl ems
t
with
√
mω¡
¨
h = a, we obtain, taking N real for simplicity and without loss of
generality (see Property (iv) of Subsec. 3.2.2),
N =
mω
π
¨
h
1
4
.
4.16 (i) The ﬁrst six Hermite polynomials are
H
0
(ζ ) = 1, H
1
(ζ ) = 2ζ ,
H
2
(ζ ) = 4ζ
2
−2, H
3
(ζ ) = 8ζ
3
−12ζ ,
H
4
(ζ ) = 16ζ
4
−48ζ
2
÷12, H
5
(ζ ) = 32ζ
5
−160ζ
3
÷120ζ .
(iii) We limit ourselves to derive the third recursion relation. Let us start from direct
differentiation of the nth order Hermite polynomial, that is
d
dζ
H(ζ ) = (−1)
n
2ζ e
ζ
2 d
n
dζ
n
e
−ζ
2
÷(−1)
n
e
ζ
2 d
n÷1
dζ
n÷1
e
−ζ
2
= 2ζ (−1)
n
e
ζ
2 d
n
dζ
n
e
−ζ
2
−(−1)
n÷1
e
ζ
2 d
n÷1
dζ
n÷1
e
−ζ
2
= 2ζ H
n
(ζ ) −H
n÷1
(ζ ).
(iv) We start from
d
2
dζ
2
H
n
(ζ ) = 2n
d
dζ
H
n−1
(ζ ) = 4n (n −1) H
n−2
(ζ ),
where we have made use of the ﬁrst recursion relation in (iii) twice. Replacing
the ﬁrst and the second derivatives into the differential equation as above, we
obtain an identity.
4.17 We write the Schrödinger equation as
d
2
dx
2
ψ(x) ÷
2m
¨
h
2
E −
1
2
mω
2
x
2
ψ(x) = 0,
and make the change of variable ξ = x
mω
¨ h
. The resulting equation is
d
2
dξ
2
ψ(ξ) ÷
2E
¨
hω
−ξ
2
ψ(ξ) = 0.
It is now convenient to consider the asymptotic behavior of ψ(ξ): for large [ξ[
2
, we
may neglect 2E¡
¨
hω with respect to ξ
2
and the asymptotic solutions of
d
2
dξ
2
ψ(ξ) = ξ
2
ψ
are ψ(ξ) = e
±ξ
2
¡2
. Due to the ﬁniteness condition of ψ(ξ) for ξ → ±∞, we have to
discard the solution ψ(ξ) = e
ξ
2
¡2
. As a consequence, it is natural to make the ansatz
ψ(ξ) = e
−
ξ
2
2
ϕ(ξ),
9780521869638sol CUP/AUL November 6, 2008 18:21 Page20
20 Sol ut i ons to sel ected probl ems
t
and obtain for ϕ(ξ) the differential equation
¸
d
2
dξ
2
−2ξ
d
dξ
÷
2E
¨
hω
−1
¸
ϕ(ξ) = 0.
The solutions of the previous equation with the condition that ϕ(ξ) be ﬁnite for ﬁnite
values of ξ and that it grow at most as a power of ξ for ξ → ±∞exist only for inte
ger values of n = E¡
¨
hω −1¡2 ≥ 0 and are given by the Hermite polynomials (see
Prob. 4.16) up to a normalization factor. This gives the eigenvalues (Eq. (4.72)) and
eigenfunctions (Eq. (4.97)), and represents an alternative solution of the harmonic
oscillator problem.
5
4.19 It is always interesting to make a comparison between the quantum results and the
corresponding classical ones. In the case of the harmonic oscillator, this is enlightning
and relatively straightforward. As we shall see later (in Chs. 9 and 10), however, in
general the classical limit in quantum mechanics is far from being obvious.
The equation of motion for a onedimensional classical harmonic oscillator
6
is
x
cl
(t ) = A sin(ωt ÷φ),
from which
˙ x
cl
(t ) = Aωcos(ωt ÷φ).
The total energy of the system is then given by
E
cl
=
1
2
m
˙ x
2
cl
÷ω
2
x
2
cl
=
1
2
mω
2
A
2
,
that yields A =
2E
cl
¡(mω
2
). In order to obtain the classical mean values, instead
of averaging – as in the quantummechanical case – over the ensemble, we have to
average over a time period of the motion (τ = 2π¡ω). We have
x
cl
= 0, ˙ x
cl
= 0,
whereas
x
2
cl
=
A
2
τ
τ
0
dt sin
2
(ωt ÷φ)
=
A
2
τ
τ
0
dt
1 −cos[2(ωt ÷φ)]
2
=
A
2
2
=
E
cl
mω
2
,
and
( p
2
x
)
cl
= m
2
˙ x
2
cl
=
1
2
A
2
ω
2
m
2
= mE
cl
.
5
See [Landau/Lifshitz 1976b, Ch. 3].
6
See [Goldstein 1950, Chs. 8–9]
9780521869638sol CUP/AUL November 6, 2008 18:21 Page21
21 Sol ut i ons to sel ected probl ems
t
Next, we calculate the corresponding quantummechanical mean values. We have
(see Probs. 4.7 and 4.8)
! ˆ x)
ψ
n
= ! ˆ p
x
)
ψ
n
= 0,
and
! ˆ x
2
)
ψ
n
=
n
ˆ x
2
n
=
¸
k
n
ˆ x
k
k
ˆ x
n
=
n
ˆ x
n ÷1
n ÷1
ˆ x
n
÷
n
ˆ x
n −1
n −1
ˆ x
n
=
¨
h(n ÷1)
2mω
÷
¨
hn
2mω
=
E
n
mω
2
,
where we have made use of Eq. (4.70). Similarly,
! ˆ p
2
x
)
ψ
n
=
n
ˆ p
2
x
n
= m
2
n
ˆ
˙ x
2
n
= m
2
¸
k
n
ˆ
˙ x
k
k
ˆ
˙ x
n
= m
2
¸
k
ı ω
nk
x
nk
ı ω
kn
x
kn
=
1
2
mω
¨
h (n ÷1) ÷
1
2
mω
¨
hn = mE
n
.
The results above show a complete correspondence between the classical and the
quantum expectation values. This result is valid in this particular case but, due to the
Ehrenfest theorem (see Sec. 3.7), it cannot be generalized.
4.20 Using the results of the Prob. 4.19 and Eqs. (2.184), we immediatly have
L
ψ
n
x · L
ψ
n
p
x
=
E
n
ω
=
¨
h
n ÷
1
2
.
Here, it is clear that the harmonic oscillator is an exceptional case: the Gaussian wave
function of the ground state saturates the uncertainty relation (L
ψ
n
x · L
ψ
n
p
x
=
¨
h¡2),
as happens for all coherent states (the subject of Subsec. 13.4.2). This ensures that
the lower bound of the uncertainty product given by Heisenberg relation is the best
constraint attainable.
4.21 Under the harmonicoscillator Hamiltonian
ˆ
H =
¨
hω
ˆ
N ÷
1
2
,
any initial state
[ ψ(0)) =
¸
n
c
n
[ n)
evolves according to
[ ψ(t )) = e
−
ı
¨ h
ˆ
Ht
[ ψ(0)) .
9780521869638sol CUP/AUL November 6, 2008 18:21 Page22
22 Sol ut i ons to sel ected probl ems
t
After one period τ = 2π¡ω, we have
[ ψ(τ)) = e
−
ı
¨ h
ˆ
Hτ
[ ψ(0))
=
¸
n
c
n
e
−ı ω
ˆ
N÷
1
2
τ
[ n)
=
¸
n
c
n
e
−ı
n÷
1
2
2π
[ n)
= e
−ı π
[ ψ(0)) .
4.22 Let us explicitly derive the xcomponent of Eq. (4.127). Then, we have
∂L
∂ ˙ x
= m:
x
÷
e
c
A
x
,
with :
x
= ˙ x, while we have
∂L
∂x
= −e
∂U
∂x
÷
e
c
:
x
∂ A
x
∂x
÷:
y
∂ A
y
∂x
÷:
z
∂ A
z
∂x
,
from which we derive (Eq. (1.15))
m
d
dt
:
x
÷
e
c
d
dt
A
x
= −e
∂U
∂x
÷
e
c
:
x
∂ A
x
∂x
÷:
y
∂ A
y
∂x
÷:
z
∂ A
z
∂x
.
But, making use of Eq. (4.139), we have
d A
x
dt
=
∂ A
x
∂t
÷
:
x
∂ A
x
∂x
÷:
y
∂ A
x
∂y
÷:
z
∂ A
x
∂z
.
Collecting these results together, we obtain
m
d
dt
:
x
= e
¸
−
∂ A
x
∂t
−
∂U
∂x
÷:
x
∂ A
x
∂x
−:
x
∂ A
x
∂x
−:
y
∂ A
x
∂y
−:
z
∂ A
x
∂z
÷:
y
∂ A
y
∂x
÷:
z
∂ A
z
∂x
¸
= e
¸
−
∂ A
x
∂t
−
∂U
∂x
−:
y
∂ A
y
∂x
−
∂ A
x
∂y
÷:
z
∂ A
z
∂x
−
∂ A
x
∂z
¸
.
Notice that this result corresponds to the xcomponent of Eq. (4.127), i.e.
d
dt
p
x
= e
−
∂
∂x
U −
∂
∂t
A
x
÷
e
c
(v B)
x
,
since the explicit expression for v B is given by
v B = v (∇ A)
=
¸
:
y
∂ A
y
∂x
−
∂ A
x
∂y
−:
z
∂ A
x
∂z
−
∂ A
z
∂x
¸
ı
÷
¸
:
z
∂ A
z
∂y
−
∂ A
y
∂z
−:
x
∂ A
y
∂x
−
∂ A
x
∂y
¸
j
÷
¸
:
x
∂ A
x
∂z
−
∂ A
z
∂x
−:
y
∂ A
z
∂y
−
∂ A
y
∂z
¸
k.
One may proceed in a similar way to derive the expressions for the y and z
components.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page23
23 Sol ut i ons to sel ected probl ems
t
Chapter 5
5.2 Equation (2.117) states that
dxϕ
ξ
/ (x)ϕ
∗
ξ
(x) = δ(ξ −ξ
/
),
where ϕ
ξ
(x) = !x [ ξ). It follows that, for ξ
/
= ξ and ˆ ρ = [ ξ) !ξ [, we have
Tr
ˆ ρ
=
dx
x
ˆ ρ
x
= ∞.
5.5 It conﬁrms that there is no unitary transformation from a pure density matrix to
a mixtures, since unitary transformations preserve scalar products (and therefore
probabilities). This is a further formulation of the measurement problem.
5.6 Let us ﬁrst compute the density matrix
ˆ ρ = [ ψ) !ψ[ =
1
2
¸
1 1
1 1
¸
.
Its characteristic equation, Det
ˆ ρ −λ
ˆ
I
= 0, is
1
2
−λ
2
−
1
4
= 0,
from which we easily obtain the solutions
λ
1
= 1, λ
2
= 0.
We can now compute the eigenvectors, which are
[ ÷) =
1
√
2
([ h) ÷[ :) ),
[ −) =
1
√
2
([ h) −[ :) ).
The diagonalized form of ˆ ρ is then:
ˆ
U
−1
ˆ ρ
ˆ
U =
¸
1 0
0 0
¸
,
where
ˆ
U =
1
√
2
¸
1 1
1 −1
¸
.
If we indicate with
ˆ
P
1
the diagonalized form of ˆ ρ, it is easy to show that ˆ ρ
/
has the
form
ˆ ρ
/
=
1
2
ˆ
P
1
÷
1
2
ˆ
P
2
,
where
ˆ
P
2
is the projector complementary to
ˆ
P
1
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page24
24 Sol ut i ons to sel ected probl ems
t
5.7 It is possible to write [ +)
12
= [ ψ)
1
⊗[ ϕ)
2
, with
[ ψ)
1
=
1
√
2
([ h)
1
÷[ :)
1
) and
1
√
2
[ ϕ)
2
= ([ h)
2
÷[ :)
2
).
5.8 It is easy to verify that, if the entangled state [ +)
12
could be written as a product
state of [ ψ)
1
and [ ϕ)
2
, we should have
c
h:
= c
h
c
/
:
and c
:h
= c
:
c
/
h
.
Now, in order to obtain the entangled state [ +)
12
, we must also have c
h
c
/
h
=c
:
c
/
:
=0.
However, this would imply that either c
h
or c
/
h
and either c
:
or c
/
:
be equal to zero,
which contradicts the the fact that neither c
h:
nor c
:h
can be zero.
5.9 The state [ +)
12
can be rewritten as (see Prob. 5.7)
[ +)
12
=
1
√
2
([ h)
1
÷[ :)
1
) ⊗
1
√
2
([ h)
2
÷[ :)
2
).
The reduced density matrices are therefore given by
ˆ o
1
=
1
2
([ h)
1
÷[ :)
1
) (
1
!h [ ÷
1
!: [),
ˆ o
2
=
1
2
([ h)
2
÷[ :)
2
) (
2
!h [ ÷
2
!: [),
which both describe pure states.
5.10 Let us write the factorized state as
ˆ ρ
12
= [ ψ)
1
!ψ[ ⊗[ ϕ)
2
!ϕ [ .
The ﬁrst reduced density matrix is
ˆ o
1
= [ ψ)
1
!ψ[ ⊗
¸
j
2
! j [ ϕ)
2
!ϕ [ j )
2
= [ ψ)
1
!ψ[ ⊗
¸
j
2
! j [ ϕ)
2
2
= [ ψ)
1
!ψ[ ⊗
¸
j
c
j
2
= [ ψ)
1
!ψ[,
if [ ϕ)
2
is normalized and where {[ j ) } is an arbitrary basis in the Hilbert space of the
system.
5.11 The reduced density matrix is
ˆ o
2
=
¸
[c
00
[
2
÷[c
10
[
2
c
00
c
∗
01
÷c
10
c
∗
11
c
∗
00
c
01
÷c
∗
10
c
11
[c
01
[
2
÷[c
11
[
2
¸
.
It is given again by the sum of the diagonal elements of the matrix (5.45) when one
interchanges the second and third row and the second and the third column.
5.12 The matrix
ˆ
C is given by
ˆ
C =
1
√
2
¸
0 1
1 0
¸
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page25
25 Sol ut i ons to sel ected probl ems
t
and it is clear that
ˆ
C =
ˆ
C
†
. Let us now compute
ˆ
C
ˆ
C
†
=
ˆ
C
†
ˆ
C =
1
2
¸
1 0
0 1
¸
,
which is a multiple of the identity operator. Its eigenvalues are degenerate and both
1¡2. Since these are c
2
n
, we have that c
n
= ±1¡
√
2. Morevoer, we are free to choose
the eigenvectors of
ˆ
C
ˆ
C
†
and
ˆ
C
†
ˆ
C, since any vector is eigenvector of the identity. In
particular, we choose
[ :
α
) = [ n
α
) =
1
√
2
([ 0) ÷[ 1) ),
:
β
=
n
β
=
1
√
2
([ 0) −[ 1) ).
Therefore, we can write
[ +) =
β
¸
n=α
c
n
[ :
n
) [ n
n
)
=
1
√
2
([ α) [ α) −[ β) [ β) ),
with
[ α) [ α) =
1
2
([ 0) [ 0) ÷[ 1) [ 1) ÷[ 0) [ 1) ÷[ 1) [ 0) ),
[ β) [ β) =
1
2
([ 0) [ 0) ÷[ 1) [ 1) −[ 0) [ 1) −[ 1) [ 0) ).
It is easy to verify that from the two previous equations one obtains again
[ +) =
1
√
2
([ 0) [ 1) ÷[ 1) [ 0) ),
as expected. From our calculations, the two unitary matrices are easily derived (they
are the matrices whose column vectors are the two vectors [ :
n
) and [ n
n
) ), that is
ˆ
U =
ˆ
U
/
=
1
√
2
¸
1 1
1 −1
¸
.
It is ﬁnally straightforward to verify that, given these matrices, Eq. (5.51) holds.
5.14 The fact that the density operator corresponding to the center of a (n ÷1)–
dimensional (hyper)sphere is always given by
1
n
ˆ
I , where n is the number of
dimensions of the system, can be easily understood by considering that the trace
of a density matrix is always equal to 1 (see Property (1.41a)), and the fact that any
density matrix may be written as ˆ ρ =
¸
n
j =1
n
j
ˆ
P
j
, where again, if ˆ ρ is a mixture, n
are the dimensions of the system.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page26
26 Sol ut i ons to sel ected probl ems
t
Chapter 6
6.2 Writing
ˆ
L
2
=
ˆ
L
2
x
÷
ˆ
L
2
y
÷
ˆ
L
2
z
, we have
¸
ˆ
L
x
,
ˆ
L
2
¸
=
¸
ˆ
L
x
,
ˆ
L
2
y
÷
ˆ
L
2
z
¸
, since
¸
ˆ
L
x
,
ˆ
L
2
x
¸
=0.
Moreover,
¸
ˆ
L
x
,
ˆ
L
2
y
¸
=
ˆ
L
x
ˆ
L
2
y
−
ˆ
L
2
y
ˆ
L
x
=
ˆ
L
x
ˆ
L
y
ˆ
L
y
−
ˆ
L
y
ˆ
L
x
ˆ
L
y
÷
ˆ
L
y
ˆ
L
x
ˆ
L
y
−
ˆ
L
y
ˆ
L
y
ˆ
L
x
=
¸
ˆ
L
x
,
ˆ
L
y
¸
ˆ
L
y
÷
ˆ
L
y
¸
ˆ
L
x
,
ˆ
L
y
¸
= ı
¨
h
ˆ
L
z
ˆ
L
y
÷
ˆ
L
y
ˆ
L
z
= ı
¨
h
¸
ˆ
L
z
,
ˆ
L
y
¸
÷
.
On the other hand, proceeding in an analogous way, we have
¸
ˆ
L
x
,
ˆ
L
2
z
¸
= −ı
¨
h
ˆ
L
y
ˆ
L
z
÷
ˆ
L
z
ˆ
L
y
= ı
¨
h
¸
ˆ
L
y
,
ˆ
L
z
¸
÷
.
In conclusion
¸
ˆ
L
x
,
ˆ
L
2
¸
= 0. It is easy to verify that this result holds true also for
ˆ
L
y
and
ˆ
L
z
, which proves the desired result.
6.4 Making use of Properties (2.97) and (2.99), and of Eq. (2.174), we obtain
¸
ˆ
L
z
, ˆ x
¸
=
¸
ˆ x ˆ p
y
− ˆ y ˆ p
x
, ˆ x
¸
=
¸
ˆ x ˆ p
y
, ˆ x
¸
−
¸
ˆ y ˆ p
x
, ˆ x
¸
=
¸
ˆ x, ˆ x
¸
ˆ p
y
÷ ˆ x
¸
ˆ p
y
, ˆ x
¸
−
¸
ˆ y, ˆ x
¸
ˆ p
x
− ˆ y
¸
ˆ p
x
, ˆ x
¸
= ı
¨
h ˆ y,
and, similarly,
¸
ˆ
L
z
, ˆ p
x
¸
=
¸
ˆ x ˆ p
y
, ˆ p
x
¸
−
¸
ˆ y ˆ p
x
, ˆ p
x
¸
=
¸
ˆ x, ˆ p
x
¸
ˆ p
y
÷ ˆ x
¸
ˆ p
y
, ˆ p
x
¸
−
¸
ˆ y, ˆ p
x
¸
ˆ p
x
− ˆ y
¸
ˆ p
x
, ˆ p
x
¸
= ı
¨
h ˆ p
y
.
6.6 For any state vector [ ψ) we have (see also Eq. (4.63))
!ψ[
ˆ
l
2
x
[ ψ) =
!ψ[
ˆ
l
†
x
ˆ
l
x
[ ψ)
≥ 0.
Along the same lines, we also have !ψ[
ˆ
l
2
y
[ ψ) ≥ 0 which, together with the previous
equation, proves the result.
6.7 Making use of Eqs. (6.6) and (6.7), it is straightforward to obtain
¸
ˆ
l
z
,
ˆ
l
±
¸
=
¸
ˆ
l
z
,
ˆ
l
x
±ı
ˆ
l
y
¸
=
¸
ˆ
l
z
,
ˆ
l
x
¸
±ı
¸
ˆ
l
z
,
ˆ
l
y
¸
= ı
ˆ
l
y
±
ˆ
l
x
= ±
ˆ
l
±
,
¸
ˆ
l
÷
,
ˆ
l
−
¸
=
¸
ˆ
l
x
÷ı
ˆ
l
y
,
ˆ
l
x
−ı
ˆ
l
y
¸
= −ı
¸
ˆ
l
x
,
ˆ
l
y
¸
÷ı
¸
ˆ
l
y
,
ˆ
l
x
¸
= 2
ˆ
l
z
,
¸
ˆ
l
2
,
ˆ
l
±
¸
=
¸
ˆ
l
2
,
ˆ
l
x
±ı
ˆ
l
y
¸
=
¸
ˆ
l
2
,
ˆ
l
x
¸
±ı
¸
ˆ
l
2
,
ˆ
l
y
¸
= 0.
6.9 Starting from Eq. (6.30), we apply
ˆ
l
−
to both sides. Since
ˆ
l
−
[ l, l) = k [ l, l −1) , we
have
ˆ
l
−
ˆ
l
2
[ l, l) = l(l ÷1)k [ l, l −1) .
9780521869638sol CUP/AUL November 6, 2008 18:21 Page27
27 Sol ut i ons to sel ected probl ems
t
Given that
ˆ
l
−
and
ˆ
l
2
commute (see Eq. (6.24)), we ﬁnally obtain
ˆ
l
2
k [ l, l −1) = l(l ÷1)k [ l, l −1) ,
which proves the desired result. Successively applying the lowering operator
ˆ
l
−
as
above, we obtain Eq. (6.31) for any m
l
.
6.10 In the case of
ˆ
l
z
, we have
ˆ
l
z
[ 1, 1) = [ 1, 1) ,
ˆ
l
z
[ 1, 0) = 0,
ˆ
l
z
[ 1, −1) = −[ 1, −1) ,
from which the ﬁrst matrix follows. For
ˆ
l
÷
, we have
ˆ
l
÷
[ 1, 1) = 0,
ˆ
l
÷
[ 1, 0) =
√
2 [ 1, 1) ,
ˆ
l
÷
[ 1, −1) =
√
2 [ 1, 0) ,
from which the second matrix follows. Finally, for
ˆ
l
−
we have
ˆ
l
−
[ 1, 1) =
√
2 [ 1, 0) ,
ˆ
l
−
[ 1, 0) =
√
2 [ 1, −1) ,
ˆ
l
−
[ 1, −1) = 0,
from which the third matrix follows.
6.11 Making use of Eqs. (6.32)–(6.33), we have
∂(r, φ, θ)
∂(x, y, z)
=
⎡
⎢
⎣
cos φ sin θ sin φ sin θ cos θ
−
1
r
sin φ
sin θ
1
r
cos φ
sin θ
0
cos φ cos θ
r
sin φ cos θ
r
−
sin θ
r
⎤
⎥
⎦.
6.12 To solve this problem one could use the transformations (6.33) and the partial
derivatives (see Prob. 6.11) to transform by brute force the Laplacian
L =
∂
2
∂x
2
÷
∂
2
∂y
2
÷
∂
2
∂z
2
from Cartesian to spherical coordinates and the desired result then would follow
straightforwardly.
However, there is a more elegant and “physical” solution of the problem. Let us
start from the simple equation (see Eq. (2.134))
ˆ p
2
= ˆ p
2
x
÷ ˆ p
2
y
÷ ˆ p
2
z
= −
¨
h
2
∂
2
∂x
2
÷
∂
2
∂y
2
÷
∂
2
∂z
2
= −
¨
h
2
L.
On the other hand, we also have (in the following a summation over repeated indices
is understood) (see Eq. (6.3))
ˆ
L
2
=
¨
h
2
ˆ
l
2
=
ˆ r × ˆ p
2
= c
i j k
ˆ r
j
ˆ p
k
c
i ab
ˆ r
a
ˆ p
b
= ˆ r
j
ˆ p
k
ˆ r
a
ˆ p
b
c
i j k
c
i ab
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page28
28 Sol ut i ons to sel ected probl ems
t
Since c
i j k
c
i ab
= δ
j a
δ
kb
−δ
j b
δ
ka
, we then have
ˆ
L
2
= ˆ r
j
ˆ p
k
ˆ r
a
ˆ p
b
δ
j a
δ
kb
− ˆ r
j
ˆ p
k
ˆ r
a
ˆ p
b
δ
j b
δ
ka
= ˆ r
j
ˆ p
k
ˆ r
j
ˆ p
k
− ˆ r
j
ˆ p
k
ˆ r
k
ˆ p
j
= ˆ r
j
ˆ r
j
ˆ p
k
÷
¸
ˆ p
k
, ˆ r
j
¸
ˆ p
k
−
ˆ p
k
ˆ r
j
÷
¸
ˆ r
j
, ˆ p
k
¸
ˆ r
k
ˆ p
j
= ˆ r
j
ˆ r
j
ˆ p
k
−ı
¨
hδ
j k
ˆ p
k
−
ˆ p
k
ˆ r
j
÷ı
¨
hδ
j k
ˆ r
k
ˆ p
j
= ˆ r
j
ˆ r
j
ˆ p
k
ˆ p
k
−ı
¨
hˆ r
j
ˆ p
j
− ˆ p
k
ˆ r
k
ˆ r
j
ˆ p
j
−ı
¨
hˆ r
k
ˆ p
k
= ˆ r
2
ˆ p
2
−ı
¨
h ˆ r · ˆ p −
ˆ p · ˆ r
ˆ r · ˆ p
−ı
¨
h ˆ r · ˆ p.
Since ˆ pˆ r = ˆ r ˆ p −3ı
¨
h, we have
ˆ
L
2
= ˆ r
2
ˆ p
2
−2ı
¨
h ˆ r · ˆ p −
ˆ r · ˆ p −3ı
¨
h
ˆ r · ˆ p
= ˆ r
2
ˆ p
2
−
ˆ r · ˆ p
2
÷ı
¨
h ˆ r · ˆ p.
On the other hand,
ˆ r · ˆ p =
¨
h
ı
x
∂
∂x
÷ y
∂
∂y
÷ z
∂
∂z
,
and, from r = (x
2
÷ y
2
÷ z
2
)
1
2
,
∂
∂x
=
x
r
∂
∂r
,
∂
∂y
=
y
r
∂
∂r
,
∂
∂z
=
z
r
∂
∂r
.
Therefore,
ˆ r · ˆ p =
¨
h
ı
r
∂
∂r
.
From the previous equations, we obtain
ˆ p
2
=
1
r
2
¸
ˆ r · ˆ p
2
−ı
¨
hˆ r · ˆ p ÷
ˆ
L
2
¸
=
¨
h
2
−
∂
2
∂r
2
−
2
r
∂
∂r
÷
ˆ
l
2
r
2
.
Together with the ﬁrst equation, we ﬁnally obtain
L =
∂
2
∂r
2
÷
2
r
∂
∂r
−
ˆ
l
2
r
2
=
1
r
2
∂
∂r
r
2
∂
∂r
−
ˆ
l
2
r
2
,
which proves the desired result.
It is interesting to note that the equation
ˆ p
2
=
1
r
2
¸
ˆ r · ˆ p
2
−ı
¨
hˆ r · ˆ p ÷
ˆ
L
2
¸
bears an important physical meaning: dividing both sides of the equation by 2m,
we have that the total energy of a pointlike free quantum particle in the three
dimensional case ( ˆ p
2
¡2m) may be interpreted as the sum of three terms. The ﬁrst
term, i.e.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page29
29 Sol ut i ons to sel ected probl ems
t
ˆ r · ˆ p
2
2mr
2
,
represents an element of the radial part of the energy. The second term,
−
ı
¨
hˆ r · ˆ p
2mr
2
,
is a typical quantummechanical term that arises fromthe commutators between posi
tion and momentum in the derivation above,
7
and is the other element of the radial
part. The last term, i.e.
ˆ
L
2
2mr
2
,
is the angular part of the total energy.
6.14 We start from Eq. (6.59) and write, for m = l and m
/
= l
/
,
dOY
∗
l
/
l
/ (φ, θ)Y
ll
(φ, θ) = δ
ll
/ .
The lhs of the previous equation turns out to be
2π
0
dφ
e
ı (l−l
/
)φ
2π
π
0
dθ sin θO
∗
l
/
l
/ (θ)O
ll
(θ).
Now, since
2π
0
dφ
e
ı (l−l
/
)φ
2π
= δ
ll
/ ,
we must also have
I
ll
=
π
0
dθ sin θO
∗
ll
(θ)O
ll
(θ) = 1.
On the other hand,
I
ll
= [N[
2
π
0
dθ (sin θ)
2l÷1
= 2 [N[
2
π
2
0
dθ (sin θ)
2l÷1
.
Since
8
π
2
0
dθ (sin θ)
2l÷1
=
(2l) ! !
(2l ÷1) ! !
,
then
I
ll
= 2 [N[
2
(2l) ! !
(2l ÷1) ! !
,
from which it follows that
[N[ =
(2l ÷1) ! !
2 (2l) ! !
=
1
2
l
l!
(2l ÷1) !
2
.
7
This term vanishes in the classical limit ¨ h → 0 (see Pr. 2.3: p. 72).
8
See [Gradstein/Ryshik 1981, 3.621.4].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page30
30 Sol ut i ons to sel ected probl ems
t
The latter result may be proved by induction (it is trivially true for l = 1, and, if it is
true for l, it is also true for l ÷1).
6.16 A spherically symmetric Hamiltonian is of the type (6.84). On the other hand,
ˆ
l
2
is
the angular part of the Laplacian (see Eq. (6.55)). If one bears in mind the explicit
expression of the Laplacian in spherical coordinates (Eq. (6.255)), it is clear that
¸
ˆ p
2
,
ˆ
l
2
¸
=
¸
V(r),
ˆ
l
2
¸
= 0.
Concerning
ˆ
l
z
, we have
¸
ˆ p
2
,
ˆ
l
z
¸
=
¸
ˆ p
2
x
÷ ˆ p
2
y
÷ ˆ p
2
z
, ˆ x ˆ p
y
− ˆ y ˆ p
x
¸
=
¸
ˆ p
2
x
, ˆ x ˆ p
y
¸
−
¸
ˆ p
2
y
, ˆ y ˆ p
x
¸
=
¸
ˆ p
2
x
, ˆ x
¸
ˆ p
y
−
¸
ˆ p
2
y
, ˆ y
¸
ˆ p
x
= −2ı
¨
h ˆ p
x
ˆ p
y
÷2ı
¨
h ˆ p
y
ˆ p
x
= 0,
where we have made use of the result of Prob. 2.23. Moreover,
¸
ˆ r
2
,
ˆ
l
z
¸
=
¸
ˆ x
2
÷ ˆ y
2
÷ ˆ z
2
, ˆ x ˆ p
y
− ˆ y ˆ p
x
¸
= −
¸
ˆ x
2
, ˆ y ˆ p
x
¸
÷
¸
ˆ y
2
, ˆ x ˆ p
y
¸
= −ˆ y
¸
ˆ x
2
, ˆ p
x
¸
÷ ˆ x
¸
ˆ y
2
, ˆ p
y
¸
= −2ı
¨
h ˆ y ˆ x ÷2ı
¨
h ˆ x ˆ y = 0,
where we have made use of the result of Prob. 2.26, and that completes the argument.
6.17 We know from Prob. 6.16 that
ˆ
l
z
commutes with ˆ p
2
and ˆ r
2
. In turn, this means
¸
ˆ
l
z
, ˆ p
2
x
÷ ˆ p
2
y
¸
=
¸
ˆ
l
z
, ˆ p
2
− ˆ p
2
z
¸
= 0,
¸
ˆ
l
z
, ˆ x
2
÷ ˆ y
2
¸
=
¸
ˆ
l
z
, ˆ r
2
− ˆ z
2
¸
= 0,
since
ˆ
l
z
commutes with both ˆ p
z
and ˆ z (see Eq. (6.2c)). This completes the proof.
6.20 We explicitly derive the commutation relation
¸
ˆ σ
x
, ˆ σ
y
¸
= 2ı ˆ σ
z
and leave the remain
ing to the reader. We have
¸
ˆ σ
x
, ˆ σ
y
¸
=
¸
0 1
1 0
¸ ¸
0 −ı
ı 0
¸
−
¸
0 −ı
ı 0
¸ ¸
0 1
1 0
¸
= 2ı
¸
1 0
0 −1
¸
= 2ı ˆ σ
z
.
6.22 We prove the result for ˆ σ
x
and leave to the reader the calculation involving ˆ σ
y
. The
characteristic equation for ˆ σ
x
is
det
ˆ σ
x
−λ
ˆ
I
= λ
2
−1 = 0,
which gives the eigenvalues λ
1,2
= ±1. We write the eigenvectors as
[ ↑)
x
=
a
b
, and [ ↓)
x
=
c
d
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page31
31 Sol ut i ons to sel ected probl ems
t
The conditions (6.158a) imply a = b and d = −c, so that we ﬁnally arrive at the
normalized eigenvectors
[ ↑)
x
=
1
√
2
1
1
=
1
√
2
¸
1
0
÷
0
1
¸
=
1
√
2
[ ↑)
z
÷[ ↓)
z
,
[ ↓)
x
=
1
√
2
1
−1
=
1
√
2
¸
1
0
−
0
1
¸
=
1
√
2
[ ↑)
z
−[ ↓)
z
.
6.23 It is very easy to show that
ˆ σ
z
= [ ↑)
z
!↑[ −[ ↓)
z
!↓[ .
For calculating ˆ σ
x
and ˆ σ
y
, we need to use Eqs. (6.159), that is,
ˆ σ
x
=
1
2
[ ↑)
x
!↑[ −[ ↓)
x
!↓[
=
1
2
¸
[ ↑)
z
÷[ ↓)
z
z
!↑[ ÷
z
!↓[
−
[ ↑)
z
−[ ↓)
z
z
!↑[ −
z
!↓[
¸
= [ ↑)
z
!↓[ −[ ↑)
z
!↓[ ,
and
ˆ σ
y
=
1
2
[ ↑)
y
!↑[ −[ ↓)
y
!↓[
=
1
2
¸
[ ↑)
z
÷ı [ ↓)
z
z
!↑[ −ı
z
!↓[
−
[ ↑)
z
−ı [ ↓)
z
z
!↑[ ÷ı
z
!↓[
¸
= ı
−[ ↑)
z
!↓[ ÷[ ↑)
z
!↓[
.
A comparison with the matricial expressions of the vectors [ ↑)
z
and [ ↓)
z
gives
immediately the Pauli matrices.
6.24 That Pauli matrices are Hermitian can be immediatly recognized by inspection. This
property immediately shows that ˆ σ
2
j
=
ˆ
I for j = x, y, z.
6.25 We proceed as follows:
ˆ σ · f
ˆ σ · f
/
= ˆ σ
j
ˆ σ
k
f
j
f
/
k
=
¸
1
2
¸
ˆ σ
j
, ˆ σ
k
¸
÷
1
2
¸
ˆ σ
j
, ˆ σ
k
¸
÷
¸
f
j
f
/
k
=
¸
ı c
j kn
ˆ σ
n
÷
ˆ
I δ
j k
¸
f
j
f
/
k
= ı c
j kn
f
j
f
/
k
ˆ σ
n
÷
ˆ
I δ
j k
f
j
f
/
k
= ı
f f
/
ˆ σ ÷
f · f
/
ˆ
I .
6.27 It is sufﬁcient to consider the form
¸
1
2m
p
x
÷
e
c
By
2
−
¨
h
2
2m
∂
2
∂y
2
÷
p
2
z
2m
− ˜ μs
z
B − E
e
ı
¨ h
( p
x
x÷p
z
z)
ϕ(y) = 0.
6.28 For the electron we have g . −2 (see Eq. (6.169) and comments), and ˜ μ
(Eq. (6.173)) becomes
˜ μ = −
e
¨
h
mc
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page32
32 Sol ut i ons to sel ected probl ems
t
Therefore, the energy eigenvalues (6.179) may be rewritten as
E =
n ÷
1
2
÷s
z
¨
hω
B
. ÷
p
2
z
2m
.
6.29 Given the harmonicoscillator character of the Schrödinger equation (6.176), we may
take full advantage of the eigenfunctions (4.97), with a suitable change of notation.
We ﬁnally obtain
ϕ
n
(y) =
mω
B
π
¨
h
1
4
2
−
n
2
(n! )
−
1
2
e
−
(y−y
0
)
2
mω
B
2¨ h
H
n
¸
(y − y
0
)
mω
B
¨
h
¸
.
6.30 The ﬁrst part is trivial. For the second part notice that
ˆ
l
z
does not commute with
ˆ
l
1,2
.
6.31 We prove the results for the states (6.193c) and (6.194). The derivation for the other
two cases is straightforward. Since ˆ s
z
= ˆ s
1
z
÷ ˆ s
2
z
and
ˆ s
k
z
[ ↑)
k,z
=
1
2
[ ↑)
k,z
and ˆ s
k
z
[ ↓)
k,z
= −
1
2
[ ↓)
k,z
,
we have
ˆ s
z
[ 1, 0)
12
=
1
√
2
¸
1
2
−
1
2
[ ↑)
1
[ ↓)
2
÷
−
1
2
÷
1
2
[ ↓)
1
[ ↑)
2
¸
= 0,
ˆ s
z
[ 0, 0)
12
=
1
√
2
¸
1
2
−
1
2
[ ↑)
1
[ ↓)
2
−
−
1
2
÷
1
2
[ ↓)
1
[ ↑)
2
¸
= 0.
Concerning the total spin, we ﬁrst notice that
ˆ s
2
= ˆ s
2
1
÷ ˆ s
2
2
÷2ˆ s
1
· ˆ s
2
=
3
2
÷2ˆ s
1
z
ˆ s
2
z
÷ ˆ s
1
÷
ˆ s
2
−
÷ ˆ s
1
−
ˆ s
2
÷
,
which can be derived by explicit substitution of ˆ s
1
±
, ˆ s
2
±
, ˆ s
2
1
, and ˆ s
2
2
from Eqs. (6.149)
and (6.190). By direct application of the operator ˆ s
2
above onto the two desired states,
we obtain
ˆ s
2
[ 0, 0)
12
= 0,
ˆ s
2
[ 1, 0)
12
= 2,
that proves the requested result.
6.34 We know that
(Lφ)
2
=
φ
2
−!φ)
2
=
÷π
−π
dφψ
∗
(φ)φ
2
ψ(φ) −
⎡
⎣
÷π
−π
dφψ
∗
(φ)φψ(φ)
⎤
⎦
2
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page33
33 Sol ut i ons to sel ected probl ems
t
With the change of variable φ ÷η = ξ, we have
f (η) =
÷π
−π
dξψ
∗
(ξ)ξ
2
ψ(ξ) ÷η
2
÷π
−π
dξψ
∗
(ξ)ξψ(ξ) −2η
÷π
−π
dξψ
∗
(ξ)ξψ(ξ)
=
ξ
2
−2η !ξ) ÷η
2
.
As a consequence, f (η) represents a parabola as a function of η and its minimum
value (corresponding to the vertex of the parabola) is obtained for η = !ξ) and is
equal to
ξ
2
−!ξ)
2
, that is to (Lφ)
2
.
6.35 From the equation
L
ψ
O
·
L
ψ
O
/
≥
1
2
!ψ[
¸
ˆ
O,
ˆ
O
/
¸
[ ψ)
,
we derive for the uncertainty product of x and y components of the angular momen
tum, when we suppose to obtain the outcome
¨
hm
j
in a measurement of
ˆ
J
z
, the
formula
L
ψ
J
x
·
L
ψ
J
y
≥
1
2
!ψ[
¸
ˆ
J
x
,
ˆ
J
y
¸
[ ψ)
=
¨
h
2
ψ
ˆ
J
z
ψ
= m
j
¨
h
2
2
.
This result, together with the fact that the maximum value of m is j while the length
of the j vector is
√
j ( j ÷1), forces us to conclude that the angular momentum vector
can never point exactly in the zdirection. Stated in other terms, since the zdirection
is arbitrary, the orientation of the angular momentumis always intrinsically uncertain.
It is also possible to derive a ﬁner estimate of the
ˆ
J
x
and
ˆ
J
y
uncertainties.
9
In fact,
we have (see Eqs. (2.184))
L
2
ψ
ˆ ;
x
=
ψ
ˆ ;
x
−
ψ
ˆ ;
x
ψ
2
ψ
=
ˆ ;
2
x
ψ
−
ˆ ;
x
2
ψ
=
ˆ ;
2
x
ψ
,
and similarly L
2
ψ
ˆ ;
y
=
ˆ ;
2
y
ψ
. Therefore, we have
ˆ j
2
ψ
=
ˆ ;
2
x
ψ
÷
ˆ ;
2
y
ψ
÷
ˆ ;
2
z
ψ
= L
2
ψ
ˆ ;
x
÷L
2
ψ
ˆ ;
y
÷m
2
j
.
Since
ˆ j
2
ψ
= j ( j ÷1), we obtain
L
2
ψ
ˆ ;
x
÷L
2
ψ
ˆ ;
y
= j
2
÷ j −m
2
j
.
In the case in which m
j
= j we have that L
2
ψ
ˆ ;
x
÷L
2
ψ
ˆ ;
y
= j .
9
See [Edmonds 1957, 18–19].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page34
34 Sol ut i ons to sel ected probl ems
t
Chapter 7
7.1 Multiplying Eq. (7.14) on the left by
r
(1)
δ
r
(2)
η
r
(3)
ζ
, it follows that
ˆ
U
123
P
+(r
(1)
δ
, r
(2)
η
, r
(3)
ζ
) = +(r
(1)
η
r
(2)
ζ
r
(3)
δ
),
where +(r
(1)
α
r
(2)
β
r
(3)
γ
) is the wave function given by the scalar product
+(r
(1)
α
r
(2)
β
r
(3)
γ
) =
r
(1)
α
r
(2)
β
r
(3)
γ
[ +
.
7.3 It is impossible: it would be a violation of Pauli exclusion principle.
7.6 The position–momentum uncertainty relation (2.190) states that
Lp
x
Lx .
¨
h.
On the other hand,
Lp
x
.
p
2
x
.
√
2mE.
Substituting E = K
B
T (see Subsec. 1.5.1) in the previous equation, we ﬁnally obtain
Lx .
¨
h
√
2mk
B
T
,
which is the socalled thermal wavelength.
Chapter 8
8.2 The mean value of
ˆ
P calculated on the output state is given by
ψ
/
ˆ
P
ψ
/
=
1
2
( !1[ −ı !2[) [[ 1) !1[ −[ 2) !2[] ([ 1) −ı [ 2) )
=
1
2
(1 ÷ı ı ) = 0.
The mean value of
ˆ
P
/
calculated on the input state is given by
1
ˆ
P
/
1
= !1[
¸
1
2
([ 1) −ı [ 2) ) ( !1[ ÷ı !2[) −
1
2
(−ı [ 1) ÷[ 2) ) (ı !1[ ÷ !2[)
¸
[ 1)
=
¸
1
1
2
1
−
¸
1
(−ı )(ı )
1
2
1
= 0.
8.3 A generic twodimensional state [ ϕ) may be expanded as
[ ϕ) = cos
θ
2
[ 1) ÷e
ı φ
sin
θ
2
[ 2) ,
where θ and φ are the polar and azimuthal angles of the Poincaré sphere, respectively.
Let us consider the transformation (8.4). The input state [ 1) corresponds to the north
pole, i.e. to θ = 0 (φ is not deﬁned). On the contrary, the output state
ψ
/
lies on
the equator of the Poincaré sphere and corresponds to θ = π¡2, φ = π¡2. In a similar
way, we can show that the input state [ 2) (south pole, θ = π) is transformed into an
equator state (θ = π¡2). We may then conclude that the considered transformation
performs a π¡2 rotation on the polar angle of the Poincaré sphere.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page35
35 Sol ut i ons to sel ected probl ems
t
8.4 Consider a transformation
ˆ
U such that
ψ
/
=
ˆ
U [ ψ) . After some time we shall
have [ ψ) → [ ψ(t )) = e
−
ı
¨ h
ˆ
Ht
[ ψ) and
ψ
/
→
ψ
/
(t )
= e
−
ı
¨ h
ˆ
Ht
ψ
/
,
ˆ
H being the
Hamiltonian of the system. If the symmetry under the transformation
ˆ
U must be con
served, then
ψ
/
(t )
must be also equal to
ˆ
U [ ψ(t )) . Combining the previous equations
we obtain on one hand
ψ
/
(t )
= e
−
ı
¨ h
ˆ
Ht
ˆ
U [ ψ) ,
and, on the other hand,
ψ
/
(t )
=
ˆ
Ue
−
ı
¨ h
ˆ
Ht
[ ψ) .
In conclusion, for the symmetry to be conserved, the operator
ˆ
U must commute with
the Hamiltonian.
8.5 The hypothesis is that we have
¸
ˆ
O,
ˆ
O
/
¸
= ı
¨
h and that
ˆ
O and
ˆ
O
/
are unitarily
transformed. Then, we have
ˆ
U
ˆ
O
ˆ
U
†
ˆ
U
ˆ
O
/
ˆ
U
†
−
ˆ
U
ˆ
O
/
ˆ
U
†
ˆ
U
ˆ
O
ˆ
U
†
=
ˆ
U
ˆ
O
ˆ
O
/
ˆ
U
†
−
ˆ
U
ˆ
O
/
ˆ
O
ˆ
U
†
=
ˆ
U
¸
ˆ
O,
ˆ
O
/
¸
ˆ
U
†
=
ˆ
U
ˆ
U
†
ı
¨
h =
¸
ˆ
O,
ˆ
O
/
¸
.
You may have recognized that we have already solved this problem in Subsec. 3.5.1.
8.6 It is evident that, if
ˆ
U(a) represents a continuous transformation, a¡2 exists and
ˆ
U(a) =
ˆ
U(a¡2)
ˆ
U(a¡2). Now,
ˆ
U(a¡2) can be either unitary or antiunitary. However,
the square of both a unitary operator and an antiunitary operator must be a unitary oper
ator (see Properties (8.11)). Therefore,
ˆ
U(a) is unitary (see also the Stone theorem:
p. 122).
Chapter 9
9.1 After a unitary timeevolution
ˆ
U
t
, the density matrix ˆ ρ
0
is transformed into
ˆ ρ
t
=
ˆ
U
t
ˆ ρ
0
ˆ
U
†
t
,
such that
ˆ ρ
2
t
=
ˆ
U
t
ˆ ρ
0
ˆ
U
†
t
ˆ
U
t
ˆ ρ
0
ˆ
U
†
t
=
ˆ
U
t
ˆ ρ
2
0
ˆ
U
†
t
,
so that, if ˆ ρ
2
0
= ˆ ρ
0
, then we also have
ˆ ρ
2
t
= ˆ ρ
t
.
9.2 The part of
ˆ
H
SM
related to the system is already diagonal, that is, taking into account
the third Pauli’s matrix (6.154),
1 ÷ ˆ σ
S
z
=
¸
2 0
0 0
¸
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page36
36 Sol ut i ons to sel ected probl ems
t
Concerning ˆ σ
M
x
, we calculate the determinant of the ﬁrst Pauli’s matrix
ˆ σ
M
x
=
¸
0 1
1 0
¸
.
Then
det
ˆ σ
M
x
−λ
ˆ
I
=
−λ 1
1 −λ
= λ
2
−1,
i.e. λ
1,2
= ±1. In other words, the action of ˆ σ
M
x
on its eigenkets [ ↑)
M
x
, [ ↓)
M
x
is
ˆ σ
M
x
[ ↑)
M
x
= [ ↑)
M
x
, ˆ σ
M
x
[ ↓)
M
x
= −[ ↓)
M
x
.
It follows than that (see Eqs. (6.158a) and Prob. 6.22)
[ ↑)
M
x
=
1
√
2
1
1
, [ ↓)
M
x
=
1
√
2
1
−1
.
9.3 Since a density matrix is a Hermitian operator (see Sec. 5.2), its eigenstates form a
basis in the Hilbert space (see Th. 2.2). In this basis it is apparent that the offdiagonal
terms of the density matrix are exactly zero.
9.5 We have
[ ϕ(t )) =
1
√
2
[ψ
↑
(t )) [ ↑)
z
÷[ψ
↓
(t )) [ ↓)
z
.
In the limit of a perfect overlap between [ψ
↑
(t )) and [ψ
↓
(t )), we have
ψ
↑
(t ) [ ψ
↓
(t )
= 1, and it follows that
ˆ ρ(t ) =
1
2
[ψ(t ) ) !ψ(t ) [
1
1
1 1
,
where [ ψ(t )) = [ψ
↑
(t )) = [ψ
↓
(t )).
9.6 By making use of the relations (see Eq. (6.147), (4.87), and the ﬁrst of Eq. (4.85),
respectively)
1
2
1 ÷ ˆ σ
z
[ ↑) = [ ↑) ,
ˆ a
1
[ 0) = 0,
ˆ a
0
[ N) =
√
N [ N −1) ,
we derive
!↑[
N
ˆ
H
SA
N −1
[ ↑) = ε
/
SA
√
N,
which yields
ε
/
SA
√
Nτ
0
. 1.
9.7 (i) In order to diagonalize the Hamiltonian it sufﬁces to substitute the deﬁnitions
(9.215) into Eq. (9.61), so as to obtain
ˆ
H
/
SA
(
ˆ
b) =
1
2
ε
SA
√
N
1 ÷ ˆ σ
z
ˆ
b
†
0
ˆ
b
0
−
ˆ
b
†
1
ˆ
b
1
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page37
37 Sol ut i ons to sel ected probl ems
t
The eigenstates of
ˆ
H
/
SA
(
ˆ
b) will be given by
[ ν)
b
[ ↑) =
1
√
ν!
1
√
(N −ν) !
ˆ
b
†
0
ν
ˆ
b
†
1
N−ν
[ 0) [ ↑) ,
with
ˆ
H
/
SA
(
ˆ
b) [ ν)
b
[ ↑) =
ε
SA
√
N
(2ν − N) [ ν)
b
[ ↑) .
(ii) Taking advantage of the expression
[ N) =
1
√
N!
ˆ a
†
0
N
[ 0) ,
where
ˆ a
†
0
=
1
√
2
ˆ
b
†
0
−
ˆ
b
†
1
,
and
ˆ a
†
0
N
=
1
√
2
N
ˆ
b
†
0
−
ˆ
b
†
1
N
=
1
√
2
N
N
¸
ν=0
N
ν
ˆ
b
†
0
ν
−
ˆ
b
†
1
N−ν
,
the initial state can be written as
[ +
SA
(t
0
)) =
1
√
2
N
N
¸
ν=0
1
√
N!
N
ν
ˆ
b
†
0
ν
−
ˆ
b
†
1
N−ν
[ 0) [ S)
=
1
√
2
N
N
¸
ν=0
1
√
N!
N
ν
(−1)
N−ν
√
ν!
(N −ν)! [ ν)
b
[ S)
=
1
√
2
N
N
¸
ν=0
(−1)
N−ν
√
N!
√
ν! (N −ν) !
[ ν)
b
[ S) .
Inserting this expression into Eq. (9.67) and making use of the above eigenvalue
equation for
ˆ
H
/
SA
(
ˆ
b), we obtain
[ +
SA
(t )) = c
÷
¸
1
√
2
N
N
¸
ν=0
(−1)
N−ν
√
N!
√
ν! ( N −ν) !
e
−
ı
¨ h
ε
SA
√
N
(2ν−N)t
[ ν)
b
[ ↑)
÷c
−
[ N) [ ↓) .
In order to obtain the ﬁnal state (9.68), it is necessary to transform back to the
physical states [ a
0
) , [ a
1
) , i.e. from [ ν)
b
to [ n) . The transformation is standard
but rather cumbersome, and requires some relabeling.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page38
38 Sol ut i ons to sel ected probl ems
t
2>
1>
1>
1>
1>
1>
1>
2>
2>
2>
2> 2>
9.8 Stirling’s formula states that, for large m,
ln(m! ) . m ln m −m ÷
1
2
ln m.
Taking only the ﬁrst two (dominant) terms of this expansion, and substituting the
resulting approximation
m! . m
m
e
−m
into Eq. (9.73), one obtains ℘
!n)
. 1.
9.9 With reference to the above ﬁgure, we label by [ 1) the input state, that is trans
mitted at the ﬁrst and at each subsequent beam splitter, and by [ 2) the state that is
reﬂected at the ﬁrst beam splitter and then transmitted at each subsequent beam split
ter. Using the generic mapping derived in the solution of Prob. 3.12, we may write
the transformation at the ﬁrst beam splitter as
[ 1) .→ T[ 1) ÷ı R[ 2) ,
where, for the sake of simplicity, we have assumed R and T to be real. The subsequent
two mirrors and beam splitter induce the transformations
T[ 1) .→ ı T(T[ 1) ÷ı R[ 2)),
ı R[ 2) .→ −R(ı R[ 1) ÷T[ 2)).
Collecting these results, we have
T[ 1) ÷ı R[ 2) .→ ı
T
2
−R
2
[ 1) −2RT[ 2) .
This means that, after two beam splitters, the state has become
[ out)
2
= ı cos
π
N
[ 1) −sin
π
N
[ 2) ,
where R = cos π¡2N and T = sin π¡2N. It follows that, after N beam splitters (with
N even), we shall have
[ out)
N
∝ cos
π
2
[ 1) ÷sin
π
2
[ 2) = [ 2) .
9.10 In order to prove that ˆ ρ
j
is a density operator it is sufﬁcient to show that ˆ ρ
j
= ˆ ρ
†
j
and Tr( ˆ ρ
j
) = 1.
Since ˆ ρ is a density operator and therefore ˆ ρ = ˆ ρ
†
, the ﬁrst condition is immedi
ately veriﬁed. In order to verify the second condition we use the cyclic property of
the trace and the fact that
ˆ
P
2
j
=
ˆ
P
j
to obtain
9780521869638sol CUP/AUL November 6, 2008 18:21 Page39
39 Sol ut i ons to sel ected probl ems
t
Tr
ˆ ρ
j
=
Tr
ˆ
P
j
ˆ ρ
ˆ
P
j
Tr
ˆ ρ
ˆ
P
j
=
Tr
ˆ ρ
ˆ
P
2
j
Tr
ˆ ρ
ˆ
P
j
=
Tr
ˆ ρ
ˆ
P
j
Tr
ˆ ρ
ˆ
P
j
= 1.
9.11 In this case we have, using Eq (9.110),
T
ˆ ρ
i
= [ :) !: [ ˆ ρ
i
[ :) !: [ = sin
2
θ [ :) !: [.
Similarly, using the cyclic property of the trace (see Prob. 5.4), we obtain
Tr
¸
T
ˆ ρ
i
¸
= Tr
¸
ˆ
P
:
ˆ ρ
i
ˆ
P
:
¸
= Tr
¸
ˆ
P
:
ˆ ρ
i
¸
=
:
ˆ ρ
i
:
= sin
2
θ,
where we have used the obvious fact that
ˆ
P
2
:
=
ˆ
P
:
(see Eq (1.41b)). The desired
result immediately follows from the latter two equations.
9.12 The problem can be solved by making repeated use of the cyclic property of the trace
and the linearity of the trace operation, that is,
Tr
¸
ˆ ρT
∗
ˆ
O
¸
= Tr
¸
ˆ ρ
¸
k
ˆ
ϑ
†
k
ˆ
O
ˆ
ϑ
k
= Tr
¸
¸
k
ˆ
ϑ
†
k
ˆ
O
ˆ
ϑ
k
ˆ ρ
=
¸
k
Tr
¸
ˆ
ϑ
†
k
ˆ
O
ˆ
ϑ
k
ˆ ρ
¸
=
¸
k
Tr
¸
ˆ
O
ˆ
ϑ
k
ˆ ρ
ˆ
ϑ
†
k
¸
=
¸
k
Tr
¸
ˆ
ϑ
k
ˆ ρ
ˆ
ϑ
†
k
ˆ
O
¸
= Tr
¸
¸
k
ˆ
ϑ
k
ˆ ρ
ˆ
ϑ
†
k
ˆ
O.
9.13 From Eq. (9.122) and from the fact that Tr[ ˆ ρ
f
] = 1 it follows that
1
℘(x
m
)
Tr
¸
ˆ
ϑ(x
m
) ˆ ρ
i
ˆ
ϑ
†
(x
m
)
¸
= 1,
from which, by the cyclic property of the trace, it further follows that
℘(x
m
) = Tr
¸
ˆ
ϑ
†
(x
m
)
ˆ
ϑ(x
m
) ˆ ρ
i
¸
.
On the other hand, according to Eq. (9.120)
℘(x
m
) = Tr
¸
ˆ
E(x
m
) ˆ ρ
i
¸
,
and therefore we have
Tr
¸
ˆ
ϑ
†
(x
m
)
ˆ
ϑ(x
m
) ˆ ρ
i
¸
= Tr
¸
ˆ
E(x
m
) ˆ ρ
i
¸
.
Since the previous equation must hold for any ˆ ρ
i
, this implies that
ˆ
ϑ
†
(x
m
)
ˆ
ϑ(x
m
) =
ˆ
E(x
m
).
9780521869638sol CUP/AUL November 6, 2008 18:21 Page40
40 Sol ut i ons to sel ected probl ems
t
9.14 Using expressions (9.125) and (9.127b), we have
÷∞
−∞
dx
m
℘(x
m
)
x
S
ˆ ρ
/
(x
m
)
x
S
=
÷∞
−∞
dx
m
℘(x
m
) !x
S
[
1
℘(x
m
)
ˆ
E
1
2
(x
m
) ˆ ρ
i
ˆ
E
1
2
(x
m
) [ x
S
)
=
÷∞
−∞
dx
m
!x
S
[
÷∞
−∞
dx
/
S
¸
℘(x
m
[x
/
S
)
¸ 1
2
[x
/
S
)!x
/
S
[
ˆ ρ
i
÷∞
−∞
dx
//
S
¸
℘(x
m
[x
//
S
)
¸ 1
2
[x
//
S
)!x
//
S
[ x
S
) .
Given that !x
S
[ y) = δ(x
S
− y), where [ y) is an arbitrary eigenstate of ˆ x, we ﬁnally
obtain
÷∞
−∞
dx
m
℘(x
m
)
x
S
ˆ ρ
/
(x
m
)
x
S
=
÷∞
−∞
dx
m
℘(x
m
[x
S
)
x
S
ˆ ρ
i
x
S
=
x
S
ˆ ρ
i
x
S
.
9.15 The ﬁrst beam splitter, the mirror M1, and the phase shifter induce the following
transformation on the initial state [ 1) :
[ 1)
BS1,M1,PS
.→
1
√
2
e
ı φ
[ 1) −[ 2)
,
which, after the second mirror and BS3, becomes
M2,BS3
.→
1
√
2
¸
ı e
ı φ
√
η [ 1) ÷ı
1 −η [ 3)
−[ 2)
.
The ﬁnal state, after BS2, can be written as
[ f ) =
1
√
2
¸
ı e
ı φ
√
η
1
√
2
([ 1) ÷ı [ 2) ) −e
ı φ
1 −η [ 3) −
1
√
2
(ı [ 1) ÷[ 2) )
¸
=
ı
2
e
ı φ
√
η −1
[ 1) −
1
2
e
ı φ
√
η ÷1
[ 2) −
e
ı φ
√
1 −η
√
2
[ 3) .
The detection probabilities can then be calulated as follows:
℘
1
=
1
4
e
ı φ
√
η −1
e
−ı φ
√
η −1
=
1
4
(η ÷1) −
√
η cos φ
2
,
℘
2
=
1
4
(η ÷1) ÷
√
η cos φ
2
,
℘
3
=
1 −η
2
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page41
41 Sol ut i ons to sel ected probl ems
t
In the case in which η = 1, we have
℘
1
=
1
2
(1 −cos φ) = sin
2
φ
2
,
℘
2
=
1
2
(1 ÷cos φ) = cos
2
φ
2
,
℘
3
= 0,
whereas, when η = 0, we obtain
℘
1
= ℘
2
=
1
4
,
℘
3
=
1
2
.
9.18 It is easy to show that
ˆ
E
1
=
1
2
¸
ˆ
P
P
d
÷η
ˆ
P
P
u
−
√
η
ˆ
P
V
2
−
ˆ
P
V
1
¸
=
¸
1
4
(1 ÷η) −
√
η
2
cos φ
1
4
(1 −η) ÷ı
√
η
2
sin φ
1
4
(1 −η) −ı
√
η
2
sin φ
1
4
(1 ÷η) ÷
√
η
2
cos φ
,
ˆ
E
2
=
1
2
¸
ˆ
P
P
d
÷η
ˆ
P
P
u
÷
√
η
ˆ
P
V
2
−
ˆ
P
V
1
¸
=
¸
1
4
(1 ÷η) ÷
√
η
2
cos φ
1
4
(1 −η) −ı
√
η
2
sin φ
1
4
(1 −η) ÷ı
√
η
2
sin φ
1
4
(1 ÷η) −
√
η
2
cos φ
,
ˆ
E
3
= (1 −η)
ˆ
P
P
u
=
1 −η
2
¸
1 −1
−1 1
¸
.
Now, we need to calculate the three expectations for the three detectors, that is
℘
1
=
1 0
¸
1
4
(1 ÷η) −
√
η
2
cos φ
1
4
(1 −η) ÷ı
√
η
2
sin φ
1
4
(1 −η) −ı
√
η
2
sin φ
1
4
(1 ÷η) ÷
√
η
2
cos φ
1
0
=
1
4
(η ÷1) −
√
η cos φ
2
,
℘
2
=
1 0
¸
1
4
(1 ÷η) ÷
√
η
2
cos φ
1
4
(1 −η) −ı
√
η
2
sin φ
1
4
(1 −η) ÷ı
√
η
2
sin φ
1
4
(1 ÷η) −
√
η
2
cos φ
1
0
=
1
4
(η ÷1) ÷
√
η cos φ
2
,
and
℘
3
=
1 0
1 −η
2
¸
1 −1
−1 1
¸
1
0
=
1 −η
2
,
which ﬁt with the previous calculations.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page42
42 Sol ut i ons to sel ected probl ems
t
9.19 The proof of completeness is straightforward. Indeed,
ˆ
E
1
÷
ˆ
E
2
÷
ˆ
E
3
=
1
2
¸
ˆ
P
P
d
÷η
ˆ
P
P
u
−
√
η
ˆ
P
V
2
−
ˆ
P
V
1
¸
÷
1
2
¸
ˆ
P
P
d
÷η
ˆ
P
P
u
÷
√
η
ˆ
P
V
2
−
ˆ
P
V
1
¸
÷(1 −η)
ˆ
P
P
u
=
ˆ
P
P
d
÷η
ˆ
P
P
u
÷
ˆ
P
P
u
−η
ˆ
P
P
u
=
ˆ
I .
The proof of the noncommutability of the above effects is more cumbersome. We
here only consider the noncommutability between
ˆ
E
1
and
ˆ
E
2
. In order to simplify
the proof, we take η = 1¡2. Then,
¸
ˆ
E
1
,
ˆ
E
2
¸
=
1
4
¸
2
√
2
ˆ
P
V
1
−
ˆ
P
V
2
ˆ
P
P
d
−
2
√
2
ˆ
P
P
d
ˆ
P
V
1
−
ˆ
P
V
2
−
1
√
2
ˆ
P
P
u
ˆ
P
V
1
−
ˆ
P
V
2
÷
1
√
2
ˆ
P
V
1
−
ˆ
P
V
2
ˆ
P
P
u
¸
=
1
4
¸
2
√
2
2
ˆ
P
V
1
−
ˆ
I
ˆ
P
P
d
−
2
√
2
ˆ
P
P
d
2
ˆ
P
V
1
−
ˆ
I
−
1
√
2
ˆ
P
P
u
2
ˆ
P
V
1
−
ˆ
I
÷
1
√
2
2
ˆ
P
V
1
−
ˆ
I
ˆ
P
P
u
¸
=
1
2
√
2
ˆ
P
V
1
ˆ
P
P
d
−
ˆ
P
P
d
ˆ
P
V
1
,
which, as a direct calculation with the explicit expressions (9.133) and (9.137) of the
involved projectors shows, does not vanish (see also Eq. (2.90)). The formal reason
for this result lies in the fact that projectors
ˆ
P
V
1
and
ˆ
P
P
u
belong to different sets.
9.20 See the original paper [de Muynck et al. 1991].
9.21 We are looking for the N functions ℘
H
j
(D) which make the quantity
!C) =
+
n
d D
N
¸
j =1
R
j
(D)℘
H
j
(D)
as small as possible. These probability functions are evidently subject to the
conditions
0 ≤ ℘
H
j
(D) ≤ 1,
N
¸
j =1
℘
H
j
(D) = 1.
It is clear that the average cost will be minimum if the integrand in the rhs of the
ﬁrst equation is chosen to be as small as possible for each data point D ∈ IR
n
. This
procedure corresponds to the choice, at each data point D, of the hypothesis for which
the risk R
j
(D) is smallest. In practice, at each point D for which R
k
(D)  R
j
(D)
(∀j ,=k), we choose
℘
H
k
(D) = 1, ℘
H
j
(D) = 0.
This is the required solution.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page43
43 Sol ut i ons to sel ected probl ems
t
In order to render the solution more compact, and to facilitate the comparison with
the quantum case, we may also introduce the function
L(D) = min
j
R
j
(D),
so that we may rewrite our solution in the form
¸
R
j
(D) −L(D)
¸
℘
H
j
(D) = 0,
R
j
(D) −L(D) ≥ 0.
Finally, note that, by summing the ﬁrst of the latter two equations over j , the function
L(D) may be rewritten as
L(D) =
N
¸
j =1
R
j
(D)℘
H
j
(D),
so that the minimum average cost can be simply expressed as
!C)
min
=
IR
n
d DL(D).
9.23 In case of binary decision, detection operators commute, since
ˆ
E
H
1
÷
ˆ
E
H
0
=
ˆ
I . The
Lagrange operator (see Eq. (9.177)) is given by
ˆ
L =
ˆ
R
0
ˆ
E
H
0
÷
ˆ
R
1
ˆ
E
H
1
,
and we have
ˆ
R
0
−
ˆ
L =
ˆ
R
0
−
ˆ
R
0
ˆ
E
H
0
−
ˆ
R
1
ˆ
E
H
1
= (
ˆ
R
0
−
ˆ
R
1
)
ˆ
E
H
1
.
Eq. (9.178a) now becomes
ˆ
R
0
−
ˆ
R
1
ˆ
E
H
1
ˆ
E
H
0
= 0.
Since in general the operator
ˆ
R
0
−
ˆ
R
1
does not vanish, we again deal with projectors
because
ˆ
E
H
1
and
ˆ
E
H
0
must be orthogonal. In conclusion, in this case the optimal
POVM is given by a PVM.
9.24 In view of the symmetry of the expression (9.208), when it is integrated over the
hypersphere the terms with k ,= j vanish. Moreover, since
¸
n
k=1
[c
/
k
[
2
= 1, we have
S
2n
dS[c
/
k
[
2
=
1
n
S
2n
dS
n
¸
k=1
[c
/
k
[
2
=
A
2n
n
,
which proves the result.
9.25 We have
ϕ
R
j
−
ˆ
L
ϕ
= A
−1
2n
ϕ
ˆ
I − ˆ ρ
ϕ
= A
−1
2n
!ϕ [ ϕ) −
ϕ
ˆ ρ
ϕ
≥ 0,
because, for any state [ ϕ) ,
ϕ
ˆ ρ
ϕ
= [!ϕ [ ψ)[
2
≤ !ϕ [ ϕ).
9780521869638sol CUP/AUL November 6, 2008 18:21 Page44
44 Sol ut i ons to sel ected probl ems
t
Chapter 10
10.1 If the two values of E
1
in Eq. (10.27) are equal, we must go to the second order for
removing the degeneracy. By multiplying from the left both sides of Eq. (10.9c) by
ψ
q
and !ψ
l
[, we obtain
/
¸
n
c
(1)
n
ψ
q
ˆ
H
/
ψ
n
−d
q
E
(2)
= 0,
/
¸
n
c
(1)
n
ψ
l
ˆ
H
/
ψ
n
−d
l
E
(2)
= 0,
where the prime on the summation symbol denotes omission of the two terms n = q
and n = l. Substitution of c
(1)
n
from Eq. (10.30) into the previous equations yields
⎛
⎜
⎝
/
¸
n
ψ
q
ˆ
H
/
ψ
n
2
E
(0)
− E
(0)
n
− E
(2)
⎞
⎟
⎠d
q
÷
/
¸
n
ψ
q
ˆ
H
/
ψ
n
ψ
n
ˆ
H
/
ψ
l
E
(0)
− E
(0)
n
d
l
= 0,
/
¸
n
ψ
l
ˆ
H
/
ψ
n
ψ
n
ˆ
H
/
ψ
q
E
(0)
− E
(0)
n
d
q
÷
⎛
⎜
⎝
/
¸
n
ψ
l
ˆ
H
/
ψ
n
2
E
(0)
− E
(0)
n
− E
(2)
⎞
⎟
⎠d
l
= 0 .
Making use of the argument employed after Eq. (10.27), the analogues of
Eqs. (10.28) are
/
¸
n
ψ
q
ˆ
H
/
ψ
n
2
E
(0)
− E
(0)
n
=
/
¸
n
ψ
l
ˆ
H
/
ψ
n
2
E
(0)
− E
(0)
n
/
¸
n
ψ
q
ˆ
H
/
ψ
n
ψ
n
ˆ
H
/
ψ
l
E
(0)
− E
(0)
n
= 0.
Unless both of these conditions are satisﬁed, the degeneracy is removed in second
order. A generalization to higher orders and to the cases in which the “ground” state
is more than doubly degenerate is straightforward.
10.2 From Eqs. (4.73), we obtain
ˆ x =
¨
h
2mω
ˆ a
†
÷ ˆ a
,
and
ˆ x
4
=
¨
h
2mω
2
ˆ a
4
÷4ˆ a
†
ˆ a
3
÷6ˆ a
2
÷6
ˆ a
†
2
ˆ a
2
÷12ˆ a
†
ˆ a ÷4
ˆ a
†
3
ˆ a ÷6
ˆ a
†
2
÷
ˆ a
†
4
÷3
,
where we have made use of the commutation relation ˆ aˆ a
†
= 1 ÷ ˆ a
†
ˆ a in order to
write the annihilation operators to the right of the creation operators.
10
From the
10
This way of ordering terms involving annihilation and creation operators is called normal ordering.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page45
45 Sol ut i ons to sel ected probl ems
t
previous equation we immediately infer that
n
ˆ x
4
m
is different from zero only
when m = n −4, n −2, n, n ÷2, n ÷4. In particular, we easily obtain
n
ˆ x
4
n
=
¨
h
2mω
2
(6n
2
÷6n ÷3),
n
ˆ x
4
n ÷2
=
¨
h
2mω
2
(4n ÷6))
(n ÷1)(n ÷2),
n
ˆ x
4
n ÷4
=
¨
h
2mω
2
(n ÷1)(n ÷2)(n ÷3)(n ÷4).
In deriving the previous equations we have taken into account that, for example, in
the last one only the ˆ a
4
term survives.
10.3 Starting from Eq. (10.58) and using Eq. (10.62), we have
c
(1)
k
(t ≥ t
∞
) =
2
ψ
k
ˆ
H
/
i
ψ
l
ı
¨
h
t
−∞
dt
/
e
ı (ω
kl
t
/
sin ωt
/
= −
ψ
k
ˆ
H
/
i
ψ
l
ı
¸
t
∞
0
dt
/
e
ı (ω
kl
÷ω)t
/
−
t
∞
0
dt
/
e
ı (ω
kl
−ω)t
/
¸
= −
ψ
k
ˆ
H
/
i
ψ
l
ı
¨
h
¸
e
ı (ω
kl
÷ω)t
∞
−1
ω
kl
÷ω
−
e
ı (ω
km
−ω)t
∞
−1
ω
kl
−ω
¸
.
10.4 The problem is solved by considering that
e
ı xt
−1
2
=
e
ı xt
−1
e
−ı xt
−1
= 2 −
e
ı xt
÷e
−ı xt
= 4 sin
2
xt
2
.
10.5 Hamilton equations (1.7) give
˙ x =
p
x
m
, ˙ p
x
= −V
/
(x),
where V(x) is the potential energy of the onedimensional Hamiltonian H =
p
2
x
¡2m ÷ V(x). Then, we have
´ x =
˙ p
x
m
= −
V
/
(x)
m
=
f (x)
m
,
which is Newton’s law. Finally, we obtain
...
x=
˙
f (x)
m
=
1
m
d f (x)
dx
dx
dt
=
p
x
m
2
f
/
(x).
10.6 Quantummechanically, Eqs. (3.126) and (3.128) give
ˆ
˙ x =
ˆ p
x
m
,
ˆ
˙ p
x
= −V
/
( ˆ x),
from which it follows that
ˆ
´ x = −
1
m
V
/
( ˆ x) =
f ( ˆ x)
m
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page46
46 Sol ut i ons to sel ected probl ems
t
Making use of the Heisenberg equation (3.108), we have
ı
¨
h
ˆ
...
x =
¸
ˆ
´ x,
ˆ
H
¸
=
1
m
¸
f ( ˆ x),
ˆ
H
¸
=
1
2m
2
¸
f ( ˆ x)
m
, ˆ p
2
x
¸
.
Now,
¸
f ( ˆ x), ˆ p
2
x
¸
= f ( ˆ x) ˆ p
2
x
− ˆ p
2
x
f ( ˆ x) ÷ ˆ p
x
f ( ˆ x) ˆ p
x
− ˆ p
x
f ( ˆ x) ˆ p
x
=
¸
f ( ˆ x), ˆ p
x
¸
ˆ p
x
÷ ˆ p
x
¸
f ( ˆ x), ˆ p
x
¸
= ı
¨
h
¸
f
/
( ˆ x) ˆ p
x
÷ ˆ p
x
f
/
( ˆ x)
,
where we have taken advantage of the result given in Prob. 2.26. This proves the
ﬁnal result.
10.8 If V
///
( ˆ x) is small, according to Eqs. (10.117), one has
η =
ˆ
H
− E
c
=
ˆ p
2
x
−
ˆ p
x
2
2m
÷
ˆ
V
− V
c
=
1
2m
σ
2
p
÷mσ
2
x
V
//
c
= const.
Applying Eq. (3.107) to the operator ˆ x
2
−
ˆ x
2
, one has
d
dt
σ
2
x
=
d
dt
ˆ x
2
−
ˆ x
2
=
¸
d
dt
ˆ x
2
−
ˆ x
2
=
1
m
ˆ p
x
ˆ x ÷ ˆ x ˆ p
x
−2
ˆ x
ˆ p
x
.
Proceeding in a similar way, with (d¡dt )σ
2
x
, one obtains
d
2
dt
2
σ
2
x
=
2
m
2
σ
2
p
−
1
m
ˆ
V
/
ˆ x ÷ ˆ x
ˆ
V
/
−2
ˆ x
ˆ
V
/
,
where
σ
2
p
=
ˆ p
2
x
−
ˆ p
x
2
is the square deviation of ˆ p
x
from its mean. If we replace V
/
in the last equation
with the ﬁrst two terms of the expansion (10.116b), we arrive at
d
2
dt
2
σ
2
x
.
2
m
2
σ
2
p
−mV
//
c
σ
2
x
,
or, by taking into account that η is constant,
d
2
dt
2
σ
2
x
.
4
m
η − V
//
c
σ
2
x
.
This last equation, within the approximation V
///
( ˆ x) . 0, is rather general.
In the case of the free particle,
ˆ x
performs a uniform rectilinear motion with
velocity
ˆ p
x
¡m, and the momentum square deviation remains exactly constant, i.e.
σ
2
p
(t ) = σ
2
p
(t
0
) = 2mη.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page47
47 Sol ut i ons to sel ected probl ems
t
Morevoer, since
d
2
dt
2
σ
2
x
=
2σ
2
p
(t
0
)
m
2
,
one rigorously obtains Eq. (10.119).
Finally, if the free wave packet is taken to be minimum at time t
0
, i.e.
σ
2
x
(t
0
) · σ
2
p
(t
0
) =
¨
h
4
,
then ˙ σ
2
x
(t
0
) = 0 and
Lx(t ) =
¸
[Lx(t
0
)]
2
÷
¸
Lp
x
(t
0
)
m
(t −t
0
)
¸
2
¸1
2
.
The second “spreading” term of the rhs of the previous equation allows a classi
cal interpretation of the free wave packet: a bunch of pointlike particlesinitally
contained within a small interval Lx(t
0
) about the average value
ˆ x(t
0
)
. Since the
velocities of these particles are dispersed over an interval
L:
x
=
Lp
x
(t
0
)
m
about the group velocity of the packet
:
x
=
1
m
ˆ p
x
(t
0
)
,
then particles initially located around the same point become uniformly distributed
over a band L:
x
· t at time t , and the width of the band increases indeﬁnitely.
10.9 The Hamiltonian of the harmonic oscillator is given by Eq. (4.48). Then,
Eqs. (10.118) may be rewritten as
d
dt
ˆ x
=
ˆ p
x
m
and
d
dt
ˆ p
x
= −mω
2
ˆ x
,
which imply
d
2
dt
2
ˆ x
= −ω
2
ˆ x
and
d
2
dt
2
ˆ p
x
= −ω
2
ˆ p
x
,
i.e.
ˆ x
and
ˆ p
x
carry out sinusoidal oscillations of frequency ω¡2π about the origin.
Moreover, we have V
//
c
= mω
2
, i.e.
η =
1
2m
σ
2
p
÷m
2
ω
2
σ
2
x
,
d
2
dt
2
σ
2
x
=
4
m
η −mω
2
σ
2
x
,
d
2
dt
2
σ
2
p
= 4mω
2
η −
σ
2
p
m
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page48
48 Sol ut i ons to sel ected probl ems
t
which show that σ
2
x
and σ
2
p
oscillate sinusoidally with frequency ω¡π about
σ
2
x
=
η
mω
2
and σ
2
p
= mη,
respectively.
The conditions
d
2
dt
2
σ
2
x
= 0 and
d
2
dt
2
σ
2
p
= 0
require
σ
2
x
=
σ
2
p
m
2
ω
2
,
or
η =
σ
2
p
m
= mω
2
σ
2
x
.
As we see in Ch. 13, this is equivalent to the condition for a state to be coherent (see
Subsec. 13.4.2).
10.10 Consider the potential depicted in Fig. 10.4. When x  x
1
, we have V(x) = V
0
=
const. Instead, when x > x
1
, V(x) is a positive function decreasing monotonically
from the positive value V(x
1
) to V(∞) = 0. The point of discontinuity x
1
and the
turning point x
2
divide the xaxis in regions I, II, and III. In order to ﬁnd the trans
mission coefﬁcient, we must construct the solution of the Schrödinger equation
whose asymptotic form in region III represents a purely transmitted wave (in the
direction of increasing x). In that region, the WKB approximation will have the
form (10.134). The condition we impose upon its asymptotic form determines that
solution (to within a constant), that is, for x · x
2
,
ψ
III
= λ
1
2
¸
cos
x
x
2
dx
1
λ
−ı
π
4
÷ı sin
x
x
2
dx
1
λ
−
π
4
¸
,
where the phase π¡4 has been added for the sake of computation. According to
Eqs. (10.139), this solution extends to region II (where x
1
 x < x
2
) in the form
ψ
II
= −ı λ
1
2
q
exp
x
2
x
dx
1
λ
q
= −ı λ
1
2
q
e
τ
exp
−
x
x
1
dx
1
λ
q
,
where
τ =
x
2
x
1
dx
1
λ
q
.
If we deﬁne
λ
q
(x
1
) =
¨
h
√
2m [V(x
1
) − E]
and k =
√
2m [E − V
0
]
¨
h
,
in region I the solution of the Schrödinger equation may be written
ψ
I
= C sin [k(x − x
1
) ÷δ),
9780521869638sol CUP/AUL November 6, 2008 18:21 Page49
49 Sol ut i ons to sel ected probl ems
t
where the constants C and δ are obtained applying the continuity conditions to the
wave function and its logarithmic derivative at the point x
1
. One then ﬁnds that
k cot δ = −
1
λ
q
(x
1
)
and C sin δ = −ı
λ
q
(x
1
)e
τ
.
Given these results, one can calculate the transmission probabilities (see also Sub
sec. 4.2.1). Finally, it should be noted that the present calculation is correct if V(x)
varies sufﬁciently slowly in regions II and III where the WKB approximation has
been made. This in turn requires that the barrier be at least several wavelengths thick
and that the trasmission probability be extremely small ( 10
−5
).
10.11 Even though following result can be cast in general terms, we shall limit ourselves
to a particle moving in one dimension. Let us start from Eq. (3.99) for the one
dimensional case, i.e.
ψ(x
/
, t
/
) =
dxG(x
/
, t
/
; x, t )ψ(x, t ),
where we have omitted the ı factor (see footnote 16, p. 390). In order to derive the
desired differential equation, we consider the case in which the time t
/
is only an
inﬁnitesimal interval c after t , i.e. t
/
= t ÷c. Taking into account Eq. (10.191) and
the fact that for a small time interval c the action is approximately equal to c times
the Lagrangian, we have
ψ(x, t ÷c) =
1
N
÷∞
−∞
dx
/
e
c
ı
¨ h
L
x÷x
/
2
,
x−x
/
c
ψ(x
/
, t ).
In the onedimensional case with a potential V(x, t ),
L(x, ˙ x) =
1
2
m˙ x
2
− V(x, t ),
and
ψ(x, t ÷c) =
1
N
÷∞
−∞
dx
/
e
ı
¨ h
m(x−x
/
)
2
2c
e
−
ı
¨ h
cV
x÷x
/
2
,t
ψ(x
/
, t ).
Consider the ﬁrst exponential in the previous equation: if x
/
is appreciably different
from x, this factor oscillates very rapidly, making the integral over x
/
vanish. Only
for x
/
values that are close to x do we obtain signiﬁcant contributions. We therefore
make the substitution x
/
= x ÷η, where we expect important contributions only for
small values of η. This yields
ψ(x, t ÷c) =
1
N
÷∞
−∞
dηe
ı
¨ h
mη
2
2c
e
−
ı
¨ h
cV(x÷
η
2
,t )
ψ(x ÷η, t ),
where the main contribution to the integral will come from values of η of the order
√
c. Expanding in power series the lhs up to ﬁrst order in c and the rhs up to second
order in η, we obtain
9780521869638sol CUP/AUL November 6, 2008 18:21 Page50
50 Sol ut i ons to sel ected probl ems
t
ψ(x, t ) ÷c
∂ψ
∂t
=
1
N
÷∞
−∞
dηe
ı
¨ h
mη
2
2c
¸
1 −
ı c
¨
h
V(x, t )
¸ ¸
ψ(x, t ) ÷η
∂ψ
∂x
÷
1
2
η
2
∂
2
ψ
∂x
2
¸
,
where we have replaced cV(x ÷η¡2, t ) by cV(x, t ), since the error is of order
higher than c. Let us consider the leading terms in both sides of the previous
equation. We have
ψ(x, t ) = ψ(x, t )
1
N
÷∞
−∞
dηe
ı
¨ h
mη
2
2c
.
In the limit as c approaches zero, the normalization factor N must be chosen so that
the equality holds. Since
11
÷∞
−∞
dxe
ax
2
=
−
π
a
,
where a is a complex number, we have
N =
2πı
¨
hc
m
1
2
.
By making use of the integrals
÷∞
−∞
dηe
ı
¨ h
mη
2
2c
η = 0
and
12
÷∞
−∞
dηe
ı
¨ h
mη
2
2c
η
2
=
ı
¨
hc
m
,
we ﬁnally obtain the equality
ψ(x, t ) ÷c
∂ψ
∂t
= ψ(x, t ) −
ı c
¨
h
V(x, t )ψ(x, t ) −
¨
hc
2ı m
∂
2
ψ
∂x
2
.
This equality holds to order c if ψ(x, t ) satisﬁes the differential equation
ı
∂ψ
∂t
= −
¨
h
2
2m
∂
2
ψ
∂x
2
÷ V(x, t )ψ,
that is precisely the Schrödinger equation for a onedimensional system. In a similar
way the same result can be obtained in more complicated contexts.
11
[Gradstein/Ryshik 1981, 3.322.2].
12
[Gradstein/Ryshik 1981, 3.462.8].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page51
51 Sol ut i ons to sel ected probl ems
t
x
1
x
2
Chapter 11
11.4 We have:
lim
x→÷∞
y(x) = 0, lim
x→0
÷
y(x) = ÷∞.
Furthermore, studying the ﬁrst and the second derivatives, we obtain a minimum at
(see ﬁgure above)
x
1
≡
2a
b
; −
b
2
4a
,
and a ﬂex at
x
2
≡
3a
b
; −
2b
2
9a
.
11.5 The reduced mass is given by
m =
m
e
Zm
p
m
e
÷ Zm
p
= m
e
1 ÷
m
e
Zm
p
−1
. m
e
1 −
m
e
Zm
p
,
from which we derive
m
e
−m
m
e
=
Lm
m
e
.
m
e
Zm
p
.
In the case of the hydrogen atom (Z = 1), we have
Lm
m
e
. 0.05%.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page52
52 Sol ut i ons to sel ected probl ems
t
11.6 Starting from Eqs. (11.13c) and (11.18), and making use of substitutions (11.21),
(11.23a), and (11.24), we have
ˆ
H
I
− E
r
E
0
= −
1
E
0
¸
E
r
÷
¨
h
2
2m
∂
2
∂r
2
−
¨
h
2
2m
l(l ÷1)
r
2
÷
Ze
2
r
= −
E
r
E
0
−
¨
h
4
2m
2
e
4
∂
2
∂r
2
÷
¨
h
4
2m
2
e
4
l(l ÷1)
r
2
−
¨
h
2
me
2
Z
r
= −
˜
E −
r
2
0
2
∂
2
∂r
2
÷
r
2
0
2
l(l ÷1)
r
2
−r
0
Z
r
= −
˜
E −
1
2
∂
2
∂ ˜ r
2
÷
1
2
l(l ÷1)
˜ r
2
−
Z
˜ r
,
from which, for Z = 1, the desired result can be derived.
11.7 Substituting Eq. (11.27b), i.e.
ξ
//
(˜ r) =
¸
−
l ÷1
˜ r
2
÷
W
//
W
−
W
/
2
W
2
÷
(l ÷1)
2
˜ r
2
÷
1
n
2
÷
W
/
2
W
2
−
2(l ÷1)
n˜ r
÷
2(l ÷1)W
/
˜ r W
−
2W
/
nW
¸
˜ r
l÷1
e
−
˜ r
n
W,
and Eq. (11.26), i.e.
˜ r
l÷1
e
−
˜ r
n
W(˜ r) = ξ(˜ r),
into Eq. (11.23b), we obtain
0 =
¸
−
l ÷1
˜ r
2
÷
W
//
W
−
W
/
2
W
2
÷
(l ÷1)
2
˜ r
2
÷
1
n
2
÷
W
/
2
W
2
−
2(l ÷1)
n˜ r
÷
2(l ÷1)W
/
˜ r W
−
2W
/
nW
−
1
n
2
÷
2
˜ r
−
l(l ÷1)
˜ r
2
¸
ξ
=
¸
W
//
W
÷
2(l ÷1)
˜ r
−
2
n
W
/
W
÷
2
˜ r
1 −
l ÷1
n
¸
ξ.
By multiplying by W˜ r, we obtain ﬁnally the desired result.
11.8 Consider Eq. (11.30) with z = −y, that is,
−y
d
2
d(−y)
2
f (−y) ÷(γ ÷ y)
d
d(−y)
f (−y) −αf (−y) = 0.
It is easy to derive
y
d
2
dy
2
g(y) ÷(γ ÷ y)
d
d(y)
g(y) ÷αg(y) = 0,
where g(y) = f (−y), which shows that the solution of this equation has the form
g(y) = F(α; γ ; −y). For y = η we obtain the result (11.39).
9780521869638sol CUP/AUL November 6, 2008 18:21 Page53
53 Sol ut i ons to sel ected probl ems
t
11.9 For each value of n, the possible values of l range from 0 to n −1; moreover, for
each value of l there are 2l ÷1 sublevels. The total number of degenerate states is
then given by
n−1
¸
l=0
(2l ÷1) =
2n(n −1)
2
÷n = n
2
,
since
m
¸
k=0
k =
m(m ÷1)
2
,
and
m
¸
k=0
1 = m.
11.10 We ﬁnd R
10
(˜ r) and leave the following ones to the reader. In this case, we have
n = 1, l = 0. As a consequence (see Eq. (11.45)), we have c
0
= 1, c
1
= 0, and,
for any j > 1, c
j
= 0. This implies that (see Eq. (11.47)), W(η) = 1, and (see
Eq. (11.26)) ξ(˜ r) = ˜ re
−˜ r
. Finally (see Eq. (11.19)), we obtain
R
10
(˜ r) ∝ e
−˜ r
.
11.11 Again, we ﬁnd the normalization factor for R
10
(r) and leave the others to the reader.
Denoting by N the normalization factor for R
10
(r), we must have
N
2
÷∞
0
dre
−2
r
r
0
r
2
= 1.
Integrating by parts, we obtain
÷∞
0
dxe
−x
x
2
= 2,
which implies
2
r
0
2
3
N
2
= 1,
or
N = 2r
−
3
2
0
.
11.12 Again, we ﬁnd the radial probability density corresponding to R
10
(r) and leave the
others to the reader. We have
℘
10
(r) =
4r
2
r
3
0
e
−2
r
r
0
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page54
54 Sol ut i ons to sel ected probl ems
t
11.13 Using the explicit form of the radial wave functions (Eq. (11.51)), we ﬁnd
r
r
0
2
¸
=
1
2
¸
3n
2
−l(l ÷1)
,
¸
r
r
0
= n
2
¸
5n
2
÷1 −3l(l ÷1)
,
r
r
0
−1
¸
=
1
n
2
r
r
0
−2
¸
=
1
n
3
l ÷
1
2
,
r
r
0
−3
¸
=
1
n
3
l
l ÷
1
2
(l ÷1)
.
Notice that the latter case is divergent for l = 0, but ﬁnite for l > 0. This is due to
the fact that R(0) ,= 0 only for l = 0, and in this case the radial integral is divergent.
The previous results may be used to obtain an interesting consequence. We know
that, for a hydrogenoid atom,
ˆ
H =
ˆ p
2
2m
−
Ze
2
r
,
and
!
ˆ
V) =
¸
−
Ze
2
r
= −
Ze
2
r
0
Z
r
r
0
−1
¸
= −
Ze
2
n
2
mZe
2
¨
h
2
= 2E
n
,
where, for a hydrogenoid atom,
E
n
= −
1
2
mc
2
Zα
n
2
,
c is the speed of light, and
α =
e
2
¨
hc
.
1
137
is the ﬁnestructure constant.
Moreover, the mean value of the kinetic energy becomes
!
ˆ
T) = !
ˆ
H) −!
ˆ
V) = E
n
−2E
n
=
1
2
m
:
2
,
so that
¸
:
2
c
2
=
2!
ˆ
T)
mc
2
=
Zα
n
2
.
For most hydrogenoid atoms this ratio is small because Z is much smaller than 100.
11.14 Impose that the condition
÷∞
0
drr
2
π
0
dθ sin θ
2π
0
dφ [ϕ
nlm
(r, θ, φ)[
2
= 1
is satisﬁed for ϕ
100
, ϕ
200
, ϕ
300
.
11.15 Make use of Eqs. (6.68)–(6.69) and Eqs. (11.51) and impose the condition of
Prob. 11.14.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page55
55 Sol ut i ons to sel ected probl ems
t
11.16 The slevels (l = 0) are obviously unaffected by the spin–orbit interaction, since
ˆ
l · ˆ s = 0 identically.
For the plevels (l = 1), we have either j = l ÷1¡2 = 3¡2 or j = l −1¡2 =
1¡2. In the former case, the energy correction is equal to κ¡2, whereas in the latter
it is simply given by −κ, where κ = κ
l=1
(n).
Similarly, for the dlevels (l = 2), the correction is equal to κ
l=2
(n) for j = 5¡2
and to −3κ
l=2
(n)¡2 for j = 3¡2.
Finally, for the f levels (l = 3), the correction is equal to 3κ
l=3
(n)¡2 for j = 7¡2
and to −2κ
l=3
(n) for j = 5¡2.
It should be noted that the proportionality constant κ grows as Z
4
. As a con
sequence, the spin–orbit correction is particularly important in the case of heavy
atoms. For example, the spin–orbit correction to the 6p level of thallium (Z =
81, A = 204) gives rise to a shift of about 1500 ˆ ρAin the wavelength of the radiation
emitted in the corresponding transition.
11.21 From the deﬁnitions (11.112), we immediatly infer that the unit of length is given by
r
0
. Moreover, the unit of mass is simply given by the electron mass m
e
, and the unit
of electron charge is equal to the opposite e of the electron charge. Finally, since the
physical dimensions of energy are given by
[E] = [m][l]
2
[t ]
−2
,
we obtain
m
e
¨
h
4
m
2
e
e
4
t
−2
0
=
m
e
e
4
¨
h
2
,
where t
0
is the atomic unit of time. From this it follows that
t
0
=
¨
h
3
m
e
e
4
.
11.22 We have
E
(1)
1
=
Z
6
π
2
dr
1
dr
2
e
−2Zr
1
e
−2Zr
2
1
r
12
=
Z
6
π
2
dr
1
e
−2Zr
1
dr
2
d cos θ
2πr
2
2
e
−2Zr
2
r
2
1
÷r
2
2
−2r
1
r
2
cos θ
=
Z
6
π
2
2π
dr
1
e
−2Zr
1
dr
2
r
2
2
e
−2Zr
2
1
2r
1
r
2
2
r
2
1
÷r
2
2
−2r
1
r
2
cos θ
cos θ=−1
cos θ=1
=
2Z
6
π
dr
1
e
−2Zr
1
dr
2
r
2
e
−2Zr
2
1
r
1
[(r
1
÷r
2
) −[r
1
−r
2
[]
=
2Z
6
π
dr
1
e
−2Zr
1
r
1
¸
r
1
0
dr
2
e
−2Zr
2
2r
2
2
÷
∞
r
1
dr
2
r
2
e
−2Zr
2
2r
1
¸
,
where θ is the polar angle. From the previous equation, given that, if r
2
 r
1
, the
second integral in the square brackets is zero, we ﬁnally obtain
9780521869638sol CUP/AUL November 6, 2008 18:21 Page56
56 Sol ut i ons to sel ected probl ems
t
E
(1)
1
=
2Z
6
π
16π
∞
0
dr
1
r
1
e
−2Zr
1
r
1
r
1
0
dr
2
e
−2Zr
2
r
2
2
,
where the extra factor 2 comes from the contribution of all the symmetric conﬁgu
rations in which r
2
> r
1
. Now, we compute the previous integral by making use of
the expression
13
u
0
dxx
n
e
−μx
=
n!
μ
n÷1
−e
−uμ
n
¸
k=0
n!
k!
u
k
μ
n−k÷1
,
obtaining
E
(1)
1
= 32Z
6
dr
1
e
−2Zr
1
¸
2
(2Z)
3
−e
−2Zr
1
2
(2Z)
3
÷
2r
1
(2Z)
2
÷
r
2
1
2Z
=
5
8
Z.
Chapter 12
12.2 The solution of the problem can be found by making use of the mathematical
identity
∇
2
( f g) = ∇
¸
∇( f g)
¸
= ∇
¸
g∇ f ÷ f ∇g
¸
= g∇
2
f ÷ f ∇
2
g ÷2∇ f ∇g,
for any pairs of functions f = f (x, y, z) and g = g(x, y, z).
12.3 Given the simpliﬁcations (12.16)–(12.21), by making use of the deﬁnition (12.10)
and of Eqs. (12.11) and (12.13), we obtain
¸
−
N
n
¸
k=1
¨
h
2
2m
k
∇
2
k
÷
ˆ
V
n
÷ E
e
ψϕ
r
n
k
= ϕ
r
n
k
¸
−
N
n
¸
k=1
¨
h
2
2m
k
∇
2
k
÷
ˆ
V
n
÷ E
e
ψ,
from which the result is easily obtained, since the equality
ϕ
r
n
k
¸
−
N
n
¸
k=1
¨
h
2
2m
k
∇
2
k
÷
ˆ
V
n
÷ E
e
ψ = Eψϕ
r
n
k
must hold for any ϕ
r
n
k
.
12.4 The change of variable (Eq. (12.27)) is equivalent to the following six changes of
coordinates:
x
c
=
m
a
x
n
a
÷m
b
x
n
b
m
a
÷m
b
, y
c
=
m
a
y
n
a
÷m
b
y
n
b
m
a
÷m
b
, z
c
=
m
a
z
n
a
÷m
b
z
n
b
m
a
÷m
b
,
x
0
= x
n
b
− x
n
a
, y
0
= y
n
b
− y
n
a
, z
0
= z
n
b
− z
n
a
,
given r
n
a
= (x
n
a
, y
n
a
, z
n
a
) and r
n
b
= (x
n
b
, y
n
b
, z
n
b
). Limiting ourselves to the x
coordinate, we have
13
See [Gradstein/Ryshik 1981, 3.351].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page57
57 Sol ut i ons to sel ected probl ems
t
∂
∂x
a
=
∂x
c
∂x
a
∂
∂x
c
÷
∂x
0
∂x
a
∂
∂x
0
=
m
a
m
a
÷m
b
∂
∂x
c
−
∂
∂x
0
,
∂
∂x
b
=
∂x
c
∂x
b
∂
∂x
c
÷
∂x
0
∂x
b
∂
∂x
0
=
m
b
m
a
÷m
b
∂
∂x
c
÷
∂
∂x
0
,
from which we obtain
∂
2
∂x
2
a
=
m
2
a
(m
a
÷m
b
)
2
∂
2
∂x
2
c
÷
∂
2
∂x
2
0
−
2m
a
m
a
÷m
b
∂
2
∂x
c
∂x
0
,
∂
2
∂x
2
b
=
m
2
b
(m
a
÷m
b
)
2
∂
2
∂x
2
c
÷
∂
2
∂x
2
0
÷
2m
b
m
a
÷m
b
∂
2
∂x
c
∂x
0
.
Taking into account only the xcoordinate terms in the Laplacians of Eq. (12.24),
we have
−
¨
h
2
2m
a
∂
2
∂x
2
a
−
¨
h
2
2m
b
∂
2
∂x
2
b
= −
¨
h
2
2(m
a
÷m
b
)
∂
2
∂x
2
c
−
¨
h
2
2m
∂
2
∂x
2
0
.
Adding the similar terms for the y and zcoordinates, we ﬁnally obtain the desired
result.
12.5 We may write
ˆ
H
a
÷
ˆ
H
b
ϕ
a
(r
a
)ϕ
B
(r
B
) = Eϕ
a
(r
a
)ϕ
B
(r
B
),
i.e.
ϕ
b
(r
b
)
ˆ
H
a
ϕ
a
(r
a
) ÷ϕ
a
(r
a
)
ˆ
H
b
ϕ
b
(r
b
) = Eϕ
a
(r
a
)ϕ
b
(r
b
).
This last equation leads to
E = E
a
÷ E
b
,
provided that we have
ˆ
H
a
ϕ
a
(r
a
) = E
a
ϕ
a
(r
a
),
ˆ
H
b
ϕ
b
(r
b
) = E
b
ϕ
b
(r
b
).
12.8 See the ﬁgure below, which represents the Morse potential for a diatomic molecule.
V(r
ab
)
–D
r
ab
r
ab
_
9780521869638sol CUP/AUL November 6, 2008 18:21 Page58
58 Sol ut i ons to sel ected probl ems
t
12.9 Just make use of Eq. (12.48) to derive
d
dr
ab
=
d ˜ r
dr
ab
d
d ˜ r
= −a˜ r
d
d ˜ r
,
and, consequently,
d
2
dr
2
ab
=
d
dr
ab
d
dr
ab
= a
2
˜ r
d
d ˜ r
÷ ˜ r
d
2
d ˜ r
2
.
12.11 In order to solve the problem, we need to make use of deﬁnitions (12.50), which
implies
−2mE
vib
n
a
2
¨
h
2
=
¸
−
n ÷
1
2
÷
√
2mD
a
¨
h
2
,
from which Eq. (12.55) is easily obtained.
12.12 Let us ﬁrst rewrite Eq. (12.55) as
E
vib
n
= −D ÷
n ÷
1
2
¨
ha
2D
m
−
n ÷
1
2
2
a
2
¨
h
2
2m
,
where the last term represents the anharmonic correction of the Morse potential.
Comparing the ﬁrst two terms with Eq. (12.42), we have
ω
0
= a
2D
m
.
Chapter 13
13.2 Inserting Eqs. (13.3b), (13.9), and (13.8) into Eq. (13.1d), we obtain
∇ (∇ A) = −
1
c
2
∂
2
∂t
2
A.
Using the mathematical identity
∇ (∇ V) = ∇(∇ · V) −∇
2
V,
which holds for any vector V, and Eq. (13.7), we arrive at
∇
2
A =
1
c
2
∂
2
∂t
2
A,
which is the desired result.
13.4 The expression (13.20a) for E(r, t ) is calculated using Eq. (13.9) and making use of
the expansion
ˆ
A =
¸
k
c
k
l
−
3
2
¸¸
ˆ a
k,1
e
ı k·r
e
−ı ω
k
t
÷ ˆ a
†
k,1
e
−ı k·r
e
ı ω
k
t
¸
e
1
÷
¸
ˆ a
k,2
e
ı k·r
e
−ı ω
k
t
÷ ˆ a
†
k,2
e
−ı k·r
e
ı ω
k
t
¸
e
2
¸
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page59
59 Sol ut i ons to sel ected probl ems
t
from which we obtain easily the desired expression for the electric ﬁeld. The
expression (13.20b) for B(r, t ) is a bit more cumbersome to derive. We recall the
mathematical expression
∇ V =
∂V
z
∂y
−
∂V
y
∂z
ı ÷
∂V
x
∂z
−
∂V
z
∂x
j ÷
∂V
y
∂x
−
∂V
x
∂y
k
for any vector V, from which, taking into account that
e
ı k·r
= e
ı (k
x
x÷k
y
y÷k
z
z)
,
we have
∇
e
ı k·r
e
1
= ı k
z
e
ı k·r
j −ı k
y
e
ı k·r
k,
∇
e
−ı k·r
e
1
= −ı k
z
e
−ı k·r
j ÷ı k
y
e
−ı k·r
k,
∇
e
ı k·r
e
2
= −ı k
z
e
ı k·r
ı ÷ı k
x
e
ı k·r
k,
∇
e
−ı k·r
e
2
= ÷ı k
z
e
−ı k·r
ı −ı k
x
e
−ı k·r
k.
Collecting these results together, we have
∇
ˆ
A = ı
¸
k
c
/
k
¸¸
ˆ a
k,1
e
ı (k·r−ω
k
t )
− ˆ a
†
k,1
e
−ı (k·r−ω
k
t )
¸
e
2
−
¸
ˆ a
k,2
e
ı (k·r−ω
k
t )
− ˆ a
†
k,2
e
−ı (k·r−ω
k
t )
¸
e
1
¸
= ı
¸
k
c
/
k
¸
ˆ a
k,1
e
ı (k·r−ω
k
t )
− ˆ a
†
k,1
e
−ı (k·r−ω
k
t )
b
1
÷
ˆ a
k,2
e
ı (k·r−ω
k
t )
− ˆ a
†
k,2
e
−ı (k·r−ω
k
t )
b
2
,
since k
x
= k
y
= 0, b
1
= e
2
and b
2
= −e
1
(see Eq. (13.21)), and where
c
/
k
= k
z
¨
hω
k
2c
0
l
3
1
2
=
¨
hk
2cl
3
c
0
1
2
.
13.5 Let us ﬁrst compute the squares of the electric and magnetic ﬁelds
ˆ
E
2
=
¸
k,λ
¨
hω
k
2c
0
l
3
¸
−ˆ a
2
k,λ
e
2ı k·r
−
ˆ a
†
k,λ
2
e
−2ı k·r
÷ ˆ a
k,λ
ˆ a
†
k,λ
÷ ˆ a
†
k,λ
ˆ a
k,λ
¸
,
ˆ
B
2
=
¸
k,λ
¨
hk
2c
0
l
3
¸
−ˆ a
2
k,λ
e
2ı k·r
−
ˆ a
†
k,λ
2
e
−2ı k·r
÷ ˆ a
k,λ
ˆ a
†
k,λ
÷ ˆ a
†
k,λ
ˆ a
k,λ
¸
.
Now, due to the periodic boundary conditions, we have
l
3
dre
2ı k·r
= 0,
since
l
dxe
ı k
x
x
=
1
ı k
x
e
ı
2πnx
l
x
l
0
= 0,
and similarly for the other components.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page60
60 Sol ut i ons to sel ected probl ems
t
Substituting these results into Eq. (13.22) and using Eq. (13.2), we obtain
Eq. (13.23).
13.6 In order to evaluate the trace in Eq. (13.32) it is easier to work in the energy
(number) representation, where
Z(β) =
∞
¸
n=0
n
e
−β
ˆ
H
n
=
∞
¸
n=0
e
−β ¨ hω
n÷
1
2
= e
−
1
2
β ¨ hω
∞
¸
n=0
e
−β ¨ hω
n
=
e
−
1
2
β ¨ hω
1 −e
−β ¨ hω
.
13.7 (a) We have
ln Z = −
1
2
β
¨
hω −ln
1 −e
−β ¨ hω
,
and, therefore,
!E) = −
∂
∂β
ln Z =
1
2
¨
hω ÷
¨
hω
e
β ¨ hω
−1
.
(b) The same result can be obtained by making use of the number distribution. In
fact, since the energy is diagonal in the number basis, we have
!E) =
∞
¸
n=0
E
n
ρ
nn
=
∞
¸
n=0
n ÷
1
2
¨
hω
e
−β
n÷
1
2
¨ hω
Z(β)
=
¨
hω
2Z(β)
∞
¸
n=0
e
−β
n÷
1
2
¨ hω
÷
¨
hω
Z(β)
∞
¸
n=0
ne
−β
n÷
1
2
¨ hω
=
¨
hω
2
÷
¨
hω
1 −e
−β ¨ hω
∞
¸
n=0
ne
−βn¨ hω
=
¨
hω
2
÷
¨
hω
e
β ¨ hω
−1
,
where we have made use of the mathematical relation
∞
¸
n=0
nx
n−1
=
d
dx
∞
¸
n=0
x
n
=
1
(1 − x)
2
.
13.11 Take a generic matrix element of the operators
¨
e
ı φ
and
¨
e
−ı φ
in the number basis.
Then, using Eq. (13.39), we have
n
¨
e
ı φ
m
=
¸
!n [ m −1) = δ
n,m−1
for m ,= 0
0 for m = 0
and
n
¨
e
−ı φ
m
= !n [ m ÷1)
= δ
n,m÷1
.
Since for a Hermitian operator
ˆ
O we must have O
mn
= O
∗
nm
, we have also proven
that
¨
e
ı φ
and
¨
e
−ı φ
are not Hermitian.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page61
61 Sol ut i ons to sel ected probl ems
t
13.12 Making use of Eqs. (13.37) and (13.40), we have
¸
¨ cosφ,
¨
sinφ
¸
=
1
4ı
¸
ˆ
N ÷1
−
1
2
ˆ a ÷ ˆ a
†
ˆ
N ÷1
−
1
2
¸ ¸
ˆ
N ÷1
−
1
2
ˆ a − ˆ a
†
ˆ
N ÷1
−
1
2
¸
−
1
4ı
¸
ˆ
N ÷1
−
1
2
ˆ a − ˆ a
†
ˆ
N ÷1
−
1
2
¸ ¸
ˆ
N ÷1
−
1
2
ˆ a ÷ ˆ a
†
ˆ
N ÷1
−
1
2
¸
=
1
2ı
¸
−
ˆ
N ÷1
−
1
2
ˆ aˆ a
†
ˆ
N ÷1
−
1
2
÷ ˆ a
†
ˆ
N ÷1
−1
ˆ a
¸
=
1
2ı
¸
ˆ a
†
ˆ
N ÷1
−
1
2
,
ˆ
N ÷1
−
1
2
ˆ a
¸
=
1
2ı
¸
¨
e
−ı φ
,
¨
e
ı φ
¸
.
13.14 In an eigenstate of the number operator we obviously have L
ˆ
N = 0. We also have
!n [ ¨ cosφ[ n) = 0,
n
¨
sinφ
n
= 0.
However,
n
¨ cos
2
φ
n
=
1
4
n
¨
e
ı φ ¨
e
ı φ
÷
¨
e
ı φ ¨
e
−ı φ
÷
¨
e
−ı φ ¨
e
ı φ
÷
¨
e
−ı φ ¨
e
−ı φ
n
=
¸
1
2
if n ,= 0
1
4
if n = 0
.
The same values can be derived for
n
¨
sin
2
φ
n
. This means that both L¨ cosφ and
L
¨
sinφ for n ,= 0 are equal to 1¡
√
2, which conﬁrms that the phase is completely
undetermined for an eigenstate of the number operator (i.e. it corresponds to a
uniform phase distribution between 0 and 2π).
13.15 By using Eqs. (13.20) and (13.17), one has
ˆ
E = c
B
¸¸
ˆ ae
ı (k·r−ωt )
− ˆ a
†
e
−ı (k·r−ωt )
¸
e
1
÷
¸
ˆ ae
ı (k·r−ωt )
− ˆ a
†
e
−ı (k·r−ωt )
¸
e
2
¸
,
and
ˆ
B = c
B
¸¸
ˆ ae
ı (k·r−ωt )
− ˆ a
†
e
−ı (k·r−ωt )
¸
[k[e
2
−
¸
ˆ ae
ı (k·r−ωt )
− ˆ a
†
e
−ı (k·r−ωt )
¸
[k[e
1
¸
,
where
c
E
= ı
¨
hω
2c
0
l
3
1
2
and c
B
=
¨
hk
2cc
0
l
3
1
2
,
and we have made use of the fact that
b
1
= k e
1
= [k[e
2
and b
2
= k e
2
= −[k[e
1
.
We also have
¸
ˆ
E,
ˆ
B
¸
=
¸
ˆ
E
x
,
ˆ
B
x
¸
÷
¸
ˆ
E
y
,
ˆ
B
y
¸
÷
¸
ˆ
E
z
,
ˆ
B
z
¸
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page62
62 Sol ut i ons to sel ected probl ems
t
By recalling that the vectors e
1
and e
2
are parallel to the x and ydirections,
respectively, and that k is in the direction of z, the ﬁrst commutator reads
¸
ˆ
E
x
,
ˆ
B
x
¸
=
¨
hω
2
2cc
0
l
3
¸
−
ˆ a − ˆ a
†
2
¸
,
and similarly for the other commutators.
13.16 The ﬁrst equation is trivial, since, by using Eq. (13.49a), we immediately obtain
α
ˆ a
†
ˆ a
α
= [α[
2
.
Analogously, we have
ˆ
N
2
α
=
α
ˆ a
†
ˆ aˆ a
†
ˆ a
α
= [α[
2
α
ˆ a
†
ˆ a ÷1
α
= [α[
2
1 ÷[α[
2
,
from which we obtain
L
α
n =
[α[
2
1 ÷[α[
2
−[α[
4
=
[α[
2
= [α[.
13.17 Inverting Eqs. (13.61), we obtain
ˆ a =
1
√
2
(
ˆ
X
1
÷ı
ˆ
X
2
),
ˆ a
†
=
1
√
2
(
ˆ
X
1
−ı
ˆ
X
2
).
Comparing with Eqs. (4.73) we may identify
ˆ
X
1
and
ˆ
X
2
as
ˆ
X
1
=
mω
¨
h
1
2
ˆ x,
ˆ
X
2
=
1
mω
¨
h
1
2
ˆ p
x
.
13.18 Let us ﬁrst write an explicit expression for the uncertainties of the quadratures in
the coherent state [ α) . Then we have
L
α
ˆ
X
1
=
α
ˆ
X
2
1
α
−
α
ˆ
X
1
α
2
=
1
2
¸
α
ˆ a
2
÷ ˆ aˆ a
†
÷ ˆ a
†
ˆ a ÷(ˆ a
†
)
2
α
−
α
ˆ a ÷ ˆ a
†
α
2
¸
,
and
L
α
ˆ
X
2
=
α
ˆ
X
2
2
α
−
α
ˆ
X
2
α
2
=
−
1
2
¸
α
ˆ a
2
− ˆ aˆ a
†
− ˆ a
†
ˆ a ÷(ˆ a
†
)
2
α
−
α
ˆ a
†
− ˆ a
α
2
¸
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page63
63 Sol ut i ons to sel ected probl ems
t
where we have made use of Eqs. (13.61). Now we know or may easily derive that
α
ˆ a
α
= α,
α
ˆ a
†
α
= α
∗
,
α
ˆ a
2
α
= α
2
,
α
(ˆ a
†
)
2
α
= (α
∗
)
2
,
α
ˆ a
†
ˆ a
α
= [α[
2
,
α
ˆ aˆ a
†
α
= 1 ÷[α[
2
.
Making use of these relations, we ﬁnally obtain
L
α
ˆ
X
1
=
1
2
¸
α
2
÷1 ÷[α[
2
÷[α[
2
÷(α
∗
)
2
−(α ÷α
∗
)
2
¸
=
1
√
2
L
α
ˆ
X
2
=
−
1
2
¸
α
2
−1 −[α[
2
−[α[
2
÷(α
∗
)
2
−(α
∗
−α)
2
¸
=
1
√
2
.
13.19 To the ﬁrst order in η we have
e
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
=
1 −η
ˆ
ξ
ˆ π
1 ÷η
ˆ
ξ
= ˆ π −η
¸
ˆ
ξ, ˆ π
¸
.
Similarly, to the second order in η, we may write
e
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
=
1 −η
ˆ
ξ ÷η
2
ˆ
ξ
2
2
ˆ π
1 ÷η
ˆ
ξ ÷η
2
ˆ
ξ
2
2
= ˆ π −η
¸
ˆ
ξ, ˆ π
¸
÷η
2
ˆ
ξ
2
2
ˆ π −η
2
ˆ
ξ ˆ π
ˆ
ξ ÷η
2
ˆ π
ˆ
ξ
2
2
.
Now, we have that
¸
ˆ
ξ,
¸
ˆ
ξ, ˆ π
¸¸
=
¸
ˆ
ξ,
ˆ
ξ ˆ π − ˆ π
ˆ
ξ
¸
=
ˆ
ξ
2
ˆ π −2
ˆ
ξ ˆ π
ˆ
ξ ÷ ˆ π
ˆ
ξ
2
,
from which Eq. (13.232) follows to the second order in η.
For the proof of the general result, let us ﬁrst compute the derivative of the lhs of
Eq. (13.232) with respect to η , that is,
d
dη
e
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
= −
ˆ
ξe
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
÷e
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
ˆ
ξ.
Let us make use of the substitution
ˆ
O ≡ e
−η
ˆ
ξ
ˆ πe
η
ˆ
ξ
. Then, the previous equation
reads as
d
ˆ
O
dη
=
¸
ˆ
O,
ˆ
ξ
¸
.
If we denote the rhs of Eq. (13.232) as
ˆ
O
/
, we easily see that, by computing again
the derivative with respect to η,
d
ˆ
O
/
dη
=
¸
ˆ
O
/
,
ˆ
ξ
¸
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page64
64 Sol ut i ons to sel ected probl ems
t
Now, the last two equations show that
ˆ
O and
ˆ
O
/
satisfy the same differential equa
tion. As a consequence, for any boundary condition
ˆ
O ≡
ˆ
O
/
and the lhs and the rhs
of Eq. (13.232) are equal.
13.20 Let us ﬁrst write
14
ˆ
ξ(η) = e
η
ˆ
O
e
η
ˆ
O
/
,
and compute its derivative with respect to η:
d
dη
ˆ
ξ(η) =
ˆ
Oe
η
ˆ
O
e
η
ˆ
O
/
÷e
η
ˆ
O
ˆ
O
/
e
η
ˆ
O
/
=
ˆ
O ÷e
η
ˆ
O
ˆ
O
/
e
−η
ˆ
O
ˆ
ξ(η).
However, by making use of the results of Prob. 13.19 and of the fact that
¸
ˆ
O,
¸
ˆ
O,
ˆ
O
/
¸¸
=
¸
ˆ
O
/
,
¸
ˆ
O,
ˆ
O
/
¸¸
= 0,
we obtain
e
η
ˆ
O
ˆ
O
/
e
−η
ˆ
O
=
ˆ
O
/
−η
¸
ˆ
O
/
,
ˆ
O
,
which implies
d
dη
ˆ
ξ(η) =
ˆ
O ÷
ˆ
O
/
÷η
¸
ˆ
O,
ˆ
O
/
¸
ˆ
ξ(η).
ˆ
ξ(η) is the solution of this differential equation for which
ˆ
ξ(0) = 1. Since the com
mutator
¸
ˆ
O
/
,
ˆ
O
¸
commutes with
ˆ
O ÷
ˆ
O
/
, by integrating the previous equation,
we ﬁnally obtain
ˆ
ξ(η) = e
η
ˆ
O÷
ˆ
O
/
e
η
2
2
[
ˆ
O,
ˆ
O
/
]
,
which, for η = 1, proves the desired result.
13.23 Let us consider an initial coherent state [ α) . Its time evolution is given by
[ α, t ) = e
−
ı
¨ h
ˆ
Ht
[ α) ,
where
ˆ
H =
¨
hω
ˆ a
†
ˆ a ÷
1
2
is the singlemode freeﬁeld harmonicoscillator Hamiltonian. We want now to
verify that [ α, t ) is an eigenstate of the annihilation operator. We have
ˆ a [ α, t ) = ˆ ae
−
ı
¨ h
ˆ
Ht
[ α) = e
−
ı
¨ h
ˆ
Ht
e
÷
ı
¨ h
ˆ
Ht
ˆ ae
−
ı
¨ h
ˆ
Ht
[ α) .
To evaluate e
÷
ı
¨ h
ˆ
Ht
ˆ ae
−
ı
¨ h
ˆ
Ht
we use the formula (13.234),
e
−
ˆ
O
ˆ
O
/
e
ˆ
O
=
¸
n
(−1)
n
n!
¸
ˆ
O,
ˆ
O
/
¸
n
.
14
See [Gardiner 1991, 138–39] and [Messiah 1958, 442].
9780521869638sol CUP/AUL November 6, 2008 18:21 Page65
65 Sol ut i ons to sel ected probl ems
t
The evaluation of the ﬁrst two terms of the inﬁnite sum in the formula above with
ˆ
O = −ı ω
ˆ a
†
ˆ a ÷
1
2
t and
ˆ
O
/
= ˆ a allows us to write
e
÷
ı
¨ h
ˆ
Ht
ˆ ae
−
ı
¨ h
ˆ
Ht
=
¸
n
(−ı ωt )
n
n!
ˆ a = ˆ ae
−ı ωt
,
where we have also made use of the result of Prob. 4.11. Using this result, we obtain
ˆ a [ α, t ) = e
−
ı
¨ h
ˆ
Ht
e
−ı ωt
ˆ a [ α) = αe
−ı ωt
e
−
ı
¨ h
ˆ
Ht
[ α)
= αe
−ı ωt
[ α, t ) ,
which shows that the initial coherent state [ α) under time evolution remains a
coherent state, i.e. the state [ α, t ) = [ α(t )) with eigenvalue α(t ) = αe
−ı ωt
.
13.24 Using the completeness relation (13.71) we have, for any state [ ψ
F
) of the radiation
ﬁeld,
[ ψ
F
) =
1
π
d
2
α [ α) !α [ ψ
F
) .
In particular, for a generic coherent state [ β) we have
[ β) =
1
π
d
2
α !α [ β) [ α)
=
1
π
e
−
1
2
[β[
2
d
2
αe
α
∗
β−
1
2
[α[
2
[ α) ,
where we have made use of expression (13.69).
13.27 Given the most general expression of the density matrix ˆ ρ =
¸
j
n
j
ψ
j
ψ
j
with
¸
j
n
j
= 1 (see Eq. (5.20)) and {
ψ
j
} representing an orthonormal basis, we may
write
Q(α, α
∗
) =
1
π
¸
j
n
j
[
α [ ψ
j
[
2
≤
1
π
¸
j
n
j
=
1
π
,
since [
α [ ψ
j
[
2
≤ 1.
13.28 Any density matrix may be written as ˆ ρ =
¸
j
n
j
ψ
j
ψ
j
, {
ψ
j
} representing
an orthonormal basis. On the other hand we have
¸
j
n
j
= 1. Then, we may write
[
n
ˆ ρ
m
[ =
¸
j
n
j
n [ ψ
j
ψ
j
[ m
≤
¸
j
n
j
n [ ψ
j
ψ
j
[ m
.
Making use of the fact that, for any real x and y, 2xy ≤ x
2
÷ y
2
, we have
¸
j
n
j
n [ ψ
j
ψ
j
[ m
≤
¸
j
1
2
n
j
¸
n [ ψ
j
2
÷
ψ
j
[ m
2
¸
=
1
2
⎡
⎣
¸
j
n
j
n [ ψ
j
2
÷
¸
j
n
j
ψ
j
[ m
2
⎤
⎦
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page66
66 Sol ut i ons to sel ected probl ems
t
Since
n [ ψ
j
2
≤ 1 and
ψ
j
[ m
2
≤ 1,
we ﬁnally have
1
2
⎡
⎣
¸
j
n
j
n [ ψ
j
2
÷
¸
j
n
j
ψ
j
[ m
2
⎤
⎦
≤
1
2
⎛
⎝
2
¸
j
n
j
⎞
⎠
= 1.
13.30 Using the identity
1
2π
÷∞
−∞
dxe
ı xy
= δ(y),
we have
÷∞
−∞
dp
x
W(x, p
x
) =
1
πh
÷∞
−∞
dx
/
x ÷ x
/
ˆ ρ
x − x
/
÷∞
−∞
dp
x
e
2ı
px x
/
h
=
x
ˆ ρ
x
= ℘(x).
Chapter 14
14.1 Let us write [ R) =
¸
j
a
j
r
j
, [ R
0
) =
¸
j
a
0
j
r
j
, and [ R
1
) =
¸
j
a
1
j
r
j
.
Then, we have
ˆ ρ
SR
= [c
0
[
2
[ 0) !0[
¸
n, j
a
0
n
a
0
j
∗
r
n
)
r
j
÷[c
1
[
2
[ 1) !1[
¸
n, j
a
1
n
a
1
j
∗
r
n
)
r
j
÷c
0
c
∗
1
[0 ) !1[
¸
n, j
a
0
n
a
1
j
∗
r
n
)
r
j
÷c
∗
0
c
1
[1 ) !0[
¸
n, j
a
1
n
a
0
j
∗
r
n
)
r
j
.
Now, by tracing the reservoir out, we obtain the reduced density matrix
ˆ
˜ ρ
S
=
¸
k
r
k
ˆ ρ
SR
r
k
= [c
0
[
2
[ 0) !0[
¸
n
a
0
n
2
÷[c
1
[
2
[ 1) !1[
¸
n
a
1
n
2
÷c
0
c
∗
1
[0 ) !1[
¸
n
a
0
n
a
1
n
∗
÷c
∗
0
c
1
[1 ) !0[
¸
n
a
1
n
a
0
n
∗
.
The sums
¸
n
a
0
n
2
and
¸
n
a
1
n
2
are the traces of
ˆ
P
0
and
ˆ
P
1
, respectively, where
ˆ
P
0
= [ R
0
) !R
0
[ =
¸
n, j
a
0
n
a
0
j
∗
r
n
)
r
j
,
ˆ
P
1
= [ R
1
) !R
1
[ =
¸
n, j
a
1
n
a
1
j
∗
r
n
)
r
j
,
and, by the normalization condition, both are equal to 1. The terms
¸
n
a
0
n
a
1
n
∗
and
¸
n
a
1
n
a
0
n
∗
are equal to !R
1
[ R
0
) and !R
0
[ R
1
), respectively. In fact,
!R
1
[ R
0
) =
¸
n, j
a
0
n
a
1
j
∗
r
j
[ r
n
=
¸
n
a
0
n
a
1
n
∗
,
!R
0
[ R
1
) =
¸
n, j
a
1
n
a
0
j
∗
r
j
[ r
n
=
¸
n
a
1
n
a
0
n
∗
.
Collecting these results and writing them in matrix form, we obtain Eq. (14.5).
9780521869638sol CUP/AUL November 6, 2008 18:21 Page67
67 Sol ut i ons to sel ected probl ems
t
14.2 It is easy to see that
¸
ˆ
O
ˆ
O
/
¸
=
¸
l,m, j ,k
O
∗
l,m
O
/
j ,k
{l, m[ j , k} = O
∗
l,m
O
/
j ,k
δ
l, j
δ
k,m
=
¸
l,m
O
∗
l,m
O
/
l,m
=
¸
l,m
O
†
ml
O
/
l,m
= Tr
ˆ
O
†
ˆ
O
/
,
and
{ j , k[
ˆ
O} = { j , k[
¸
l,m
O
l,m
[l, m}
=
¸
l,m
O
l,m
δ
l, j
δ
k,m
= O
j ,k
.
14.5 Let us write the density matrix ˆ ρ in terms of the expansion (13.92), that is,
ˆ ρ = π
¸
n,m
Q
n,m
ˆ a
†
n
ˆ a
m
.
Then, we may write
ˆ ρˆ a
†
= π
¸
n,m
Q
n,m
ˆ a
†
n
ˆ a
m
ˆ a
†
= ˆ a
†
π
¸
n,m
Q
n,m
ˆ a
†
n
ˆ a
m
÷π
¸
n,m
mQ
n,m
ˆ a
†
n
ˆ a
m−1
= ˆ a
†
ˆ ρ ÷
d ˆ ρ
d ˆ a
,
where we have taken into account the result
ˆ a
m
ˆ a
†
= ˆ a
†
ˆ a
m
÷mˆ a
m−1
of Prob. 4.12. The ﬁnal equality on the rhs of Eq. (14.53) may be derived by taking
into account the expansion (13.91) that yields
¸
α
d ˆ ρ
d ˆ a
α
= π
∂Q
∂α
.
14.6 We have that
α
ˆ a ˆ ρˆ a
†
α
=
¸
α
ˆ a
†
ˆ ρ ˆ a ÷ ˆ a
†
d ˆ ρ
d ˆ a
†
÷ ˆ ρ ÷
d ˆ ρ
d ˆ a
ˆ a ÷
d
2
ˆ ρ
d ˆ ad ˆ a
†
α
,
since
ˆ ρˆ a
†
= ˆ a
†
ˆ ρ ÷
d ˆ ρ
d ˆ a
and ˆ a ˆ ρ = ˆ ρ ˆ a ÷
d ˆ ρ
d ˆ a
†
.
By making use of the previous results we obtain the desired solution.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page68
68 Sol ut i ons to sel ected probl ems
t
14.8 We have that
s
2
= s
2
x
÷s
2
y
÷s
2
z
= ρ
2
eg
÷ρ
2
ge
÷2ρ
eg
ρ
ge
−ρ
2
eg
−ρ
2
ge
÷2ρ
eg
ρ
ge
÷ρ
2
ee
÷ρ
2
gg
−2ρ
ee
ρ
gg
= ρ
2
ee
÷ρ
2
gg
÷4ρ
eg
ρ
ge
−4ρ
ee
ρ
gg
÷2ρ
ee
ρ
gg
= 1 ÷4
ρ
eg
ρ
ge
−ρ
ee
ρ
gg
,
where we have made use of the fact that, since ρ
ee
÷ρ
gg
= 1 (they are the diagonal
elements of the density matrix), also ρ
2
ee
÷ρ
2
gg
÷2ρ
ee
ρ
gg
= 1. Now it is easy to
show that 1 ÷4
ρ
eg
ρ
ge
−ρ
ee
ρ
gg
is equal to one in the case of pure states and
strictly smaller than 1 in the case of mixtures. In fact, if ˆ ρ is a pure state, we must
have that Tr( ˆ ρ
2
) = 1. This means that the sum of the diagonal elements of ˆ ρ
2
must
be equal to 1, i.e.
ρ
2
gg
÷ρ
ge
ρ
eg
÷ρ
eg
ρ
ge
÷ρ
2
ee
= 1,
from which it follows that
ρ
ge
ρ
eg
= ρ
ee
ρ
gg
,
which immediately gives the desired result.
In the case in which ˆ ρ is a mixture, we have that Tr( ˆ ρ
2
)  1, and, therefore, we also
must have that
ρ
2
gg
÷ρ
ge
ρ
eg
÷ρ
eg
ρ
ge
÷ρ
2
ee
 1,
from which it follows that
ρ
ge
ρ
eg
 ρ
ee
ρ
gg
,
from which it follows that 1 ÷4
ρ
eg
ρ
ge
−ρ
ee
ρ
gg
 1.
14.9 We recall that
ˆ σ
−
[ e) = [ g) , ˆ σ
−
[ g) = 0,
ˆ σ
÷
[ e) = 0, ˆ σ
÷
[ g) = [ e) .
Then, we may calculate the time derivatives of the elements of ˆ ρ:
˙ ρ
ee
=
e
˙
ˆ ρ
e
= −γ
e
ˆ σ
÷
ˆ σ
−
ˆ ρ
e
÷
e
ˆ ρ ˆ σ
÷
ˆ σ
−
e
= −γ
g
ˆ σ
−
ˆ ρ
e
÷
e
ˆ ρ ˆ σ
÷
g
= −γ
e
ˆ ρ
e
÷
e
ˆ ρ
e
= −2γρ
ee
.
Analogously, we have for ˙ ρ
gg
:
˙ ρ
gg
=
g
˙
ˆ ρ
g
= γ
2
g
ˆ σ
−
ˆ ρ ˆ σ
÷
g
−
g
ˆ σ
÷
ˆ σ
−
ˆ ρ
g
−
g
ˆ ρ ˆ σ
÷
ˆ σ
−
g
= 2γ
e
ˆ ρ
e
= 2γρ
ee
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page69
69 Sol ut i ons to sel ected probl ems
t
For the time derivative of the matrix element ρ
eg
, we have
˙ ρ
eg
=
e
˙
ˆ ρ
g
= γ
2
e
ˆ σ
−
ˆ ρ ˆ σ
÷
g
−
e
ˆ σ
÷
ˆ σ
−
ˆ ρ
g
−
e
ˆ ρ ˆ σ
÷
ˆ σ
−
g
= −γ
g
ˆ σ
−
ˆ ρ
g
= −γ
e
ˆ ρ
g
= −γρ
eg
.
The solutions of these differential equations are then given by
ρ
ee
(t ) = e
−2γ t
ρ
ee
(0),
ρ
gg
(t ) = 1 ÷e
−2γ t
¸
ρ
gg
(0) −1
,
ρ
eg
(t ) = e
−γ t
ρ
eg
(0),
where, for deriving the second solution, we have used the following procedure:
ρ
gg
(t ) = 1 −ρ
ee
(t ) = 1 −e
−2γ t
ρ
ee
(0)
= 1 −e
−2γ t
¸
1 −ρ
gg
(0)
.
From these results, it follows that the time derivatives of the Bloch vector’s
components are
˙ s
x
= ˙ ρ
eg
÷ ˙ ρ
ge
= −γ
ρ
eg
÷ρ
ge
= −γ s
x
,
˙ s
y
= ı
˙ ρ
eg
− ˙ ρ
ge
= −ı γ
ρ
eg
−ρ
ge
= −γ s
y
,
˙ s
z
= ˙ ρ
ee
− ˙ ρ
gg
= −2γρ
ee
−2γρ
ee
== −4γρ
ee
= −2γ (1 ÷s
z
).
The solution of these differential equations are ﬁnally
s
x
(t ) = e
−γ t
s
x
(0),
s
y
(t ) = ı e
−γ t
¸
ρ
eg
(0) −ρ
ge
(0)
¸
= e
−γ t
s
y
(0),
s
z
(t ) = ρ
ee
(t ) −ρ
gg
(t ) = e
−2γ t
ρ
ee
(0) −1 −e
−2γ t
ρ
gg
(0) ÷e
−2γ t
= e
−2γ t
s
z
(0) ÷e
−2γ t
−1,
which are in agreement with Eqs. (14.83).
14.10 This equation plays an important role and it can be proved in many ways. Let us call
L(ξ, η) the lefthand side and R(ξ, η) the righthand side.
A ﬁrst possibility is to verify that:
• Both terms are equal to the identity at ξ = 0.
• Both terms satisfy the same differential equation
∂L(ξ, η)
∂ξ
= (
ˆ
O ÷η
ˆ
O
/
)L(ξ, η),
∂ R(ξ, η)
∂ξ
= (
ˆ
O ÷η
ˆ
O
/
)R(ξ, η).
The proof of the ﬁrst equation is trivial (it is essentially the deﬁnition of the expo
nential), while the proof of the second equation can be obtained by inspecting the
different terms.
Both the functions L(ξ, η) and R(ξ, η) satisfy the same ﬁrstorder differential
equation with the same boundary condition and therefore they must be equal.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page70
70 Sol ut i ons to sel ected probl ems
t
A second proof can be obtained by performing the following steps:
• One veriﬁes that, for small ξ,
R(ξ, η) = 1 ÷(
ˆ
O ÷η
ˆ
O
/
)ξ ÷ O(ξ
2
).
• The function R(ξ, η) satisﬁes the semigroup property (see Sec. 8.4)
R(ξ
1
, η)R(ξ
2
, η) = R(ξ
1
÷ξ
2
, η),
as can be proved by combining the different terms present in the lhs.
The function L(ξ, η) trivially satisﬁes the same equations. The two properties
identify in a unique way the function and therefore the two functions must coincide.
However the most instructive (and constructing) way to verify Eq. (14.91) is to
check that the lhs and the rhs do coincide term by term in the Taylor expansion
around η = 0. The Taylor expansion of the rhs is trivial, so we have to compute the
Taylor expansion of the lhs. To this end it is convenient to notice that, for any real
δ ,= 0, we can write
L(ξ, η) = (e
(
ˆ
O÷η
ˆ
O
/
)δ
)
ξ
δ
.
In particular, we can write
L(ξ, η) = lim
δ→0
(e
(
ˆ
O÷η
ˆ
O
/
)δ
)
ξ
δ
= lim
δ→0
(1 ÷δ(
ˆ
O ÷η
ˆ
O
/
))
ξ
δ
.
Without loss of generality we can evaluate the limit δ → 0 by restricting to the
sequence of δ where ζ = ξ¡δ is an integer. Using this deﬁnition of ζ we have
L(ξ, η) = lim
δ→0
¸
j =1,ζ
(1 ÷δ(
ˆ
O ÷η
ˆ
O
/
)).
We can now easily compute the expansion in powers of η of the previous formula:
the rhs is the product of terms linear in η. Let us compute the coefﬁcient of the order
η
2
. The term η
2
may come from both the j th and the lth terms of the product
(where all the other factors give a contribution equal to 1¡δ
ˆ
O). We thus ﬁnd that the
coefﬁcient of η
2
is just given by
δ
2
¸
j ,l=1,ζ ; j l
⎛
⎝
⎛
⎝
¸
a=1, j −1
(1 ÷δ
ˆ
O)
⎞
⎠ ˆ
O
/
⎛
⎝
¸
b=j ÷1,l−1
(1 ÷δ
ˆ
O)
⎞
⎠ ˆ
O
/
⎛
⎝
¸
c=l÷1,ζ
(1 ÷δ
ˆ
O)
⎞
⎠
⎞
⎠
= δ
2
¸
j ,l=1,ζ ; j l
(1 ÷δ
ˆ
O)
j −1
ˆ
O
/
(1 ÷δ
ˆ
O)
l−j −2
ˆ
O
/
(1 ÷δ
ˆ
O)
ζ −l−1
.
In the limit δ → 0 each individual term in the sum over j and l is irrelevant and we
can assume that both j and l are of order δ
−1
. We can thus substitute the sums with
integrals; neglecting terms going to zero with δ we obtain
ξ
0
dξ
1
ξ
ξ
1
dξ
2
(1 ÷δ
ˆ
O)
ξ
1
δ ˆ
O
/
(1 ÷δ
ˆ
O)
ξ
2
−ξ
1
δ ˆ
O
/
(1 ÷δ
ˆ
O)
ξ−ξ
2
δ
.
We can now perform the limit δ → 0 and obtain
ξ
0
dξ
1
ξ
ξ
1
dξ
2
e
ξ
1
ˆ
O
ˆ
O
/
e
(ξ
2
−ξ
1
)
ˆ
O
ˆ
O
/
e
(ξ−ξ
2
)
ˆ
O
,
9780521869638sol CUP/AUL November 6, 2008 18:21 Page71
71 Sol ut i ons to sel ected probl ems
t
which is essential the result stated in Eq. (14.91) for k = 2, apart from a redeﬁnition
of the integration variables. It should be clear to the reader how to generalize the
result to higher (and lower) values of k.
The same result could also be obtained starting from path integral representation
(see Sec. 10.8) for the function L(ξ, η), but we shall not show this further derivation.
We should note that all the proofs presented here have a rather formal character: we
are implicitly assuming that the rhs of Eq. (14.91) is a convergent series (and that
the lhs exists). If not (the perturbative expansion is often not convergent), the rhs
should be interpreted as an asymptotic expansion (see also Secs. 10.1–10.2).
14.11 Let us start from the identity
ˆ ρ = [ ψ) !ψ[ .
Time derivation leads to
˙
ˆ ρ =
˙
ψ ) !ψ
÷
ψ )
˙
ψ
,
where
˙
ψ
=
1
ı
¨
h
˜
ˆ
H [ ψ) ,
˙
ψ
=
ı
¨
h
!ψ[
˜
ˆ
H
†
.
Then, we have
˙
ˆ ρ =
1
ı
¨
h
˜
ˆ
H ˆ ρ −
1
ı
¨
h
ˆ ρ
˜
ˆ
H
†
=
1
ı
¨
h
ˆ
H
0
÷
ˆ
H
/
ˆ ρ −
1
ı
¨
h
ˆ ρ
ˆ
H
0
÷
ˆ
H
/
=
1
ı
¨
h
¸
ˆ
H
0
, ˆ ρ
¸
÷
1
ı
¨
h
¸
ˆ
H
/
, ˆ ρ
¸
÷
.
14.13 First, the reader may verify that the exponential e
−ı
π
2
t j
2
is equal to 1 for even values
of j ( j = 2k) and to −ı for odd values of j ( j = 2k ÷1). Then, we have
e
−
[α[
2
2
∞
¸
j =0
α
j
e
−ı
π
2
t j
2
√
j !
[ j ) =
e
−
[α[
2
2
2
¸
∞
¸
k=0
2α
2k
√
2k!
[ 2k) ÷
∞
¸
k=0
(−2ı )
α
2k÷1
√
(2k ÷1)!
[ 2k ÷1)
= e
−
[α[
2
2
¸
(1 ÷ı )
∞
¸
k=0
α
2k
√
2k!
[ 2k) ÷(1 −ı )
∞
¸
k=0
α
2k
√
(2k)!
[ 2k)
÷(1 −ı )
∞
¸
k=0
α
2k÷1
√
(2k ÷1)!
[ 2k ÷1) −(1 ÷ı )
∞
¸
k=0
α
2k÷1
√
(2k ÷1)!
[ 2k ÷1)
,
where we have made use of the identities 2 = 1 ÷ı ÷1 −ı and −2ı = 1 −ı −
(1 ÷ı ). Now we group the ﬁrst and last terms, and the second and third terms of the
rhs, so as to obtain
9780521869638sol CUP/AUL November 6, 2008 18:21 Page72
72 Sol ut i ons to sel ected probl ems
t
e
−
[α[
2
2
∞
¸
j =0
α
j
e
−ı
π
2
t j
2
√
j !
[ j ) =
1
√
2
¸
(1 −ı )
√
2
e
−
[α[
2
2
∞
¸
k=0
α
k
√
k!
[ k) ÷
(1 ÷ı )
√
2
e
−
[α[
2
2
∞
¸
k=0
(−α)
k
√
k!
[ k)
=
1
√
2
e
−ı
π
4
[ α) ÷e
ı
π
4
[ −α)
,
where we have made use of the mathematical identity
(1 ±ı )
√
2
= e
±ı
π
4
and of the fact that (see Eq. (13.58))
e
−
[α[
2
2
∞
¸
k=0
(±α)
k
√
k!
[ k) = [ ±α) .
14.15 Any distance between two elements a, b has to satisfy four wellknown prop
erties. The property d(a, a) = 0 is easily satisﬁed by the Fubini–Study distance,
since [!ψ
1
[ ψ
1
)[
2
= 1. Similarly, the Fubini–Study distance is always positive
when the two states are different except the trivial case in which the difference
is only given by the global phase, since [!ψ
1
[ ψ
2
)[
2
 1. Also the property that
d(a, b) = d(b, a), since [!ψ
1
[ ψ
2
)[
2
= [!ψ
2
[ ψ
1
)[
2
, Finally, the triangular property
d(a, b) ÷d(b, c) ≥ d(a, c) is also satisﬁed. Indeed we have
1 −[!ψ
1
[ ψ
2
)[
2
÷
1 −[!ψ
2
[ ψ
3
)[
2
≥
1 −[!ψ
1
[ ψ
3
)[
2
.
To prove this, it sufﬁces to assume, without loss of generality, that the involved state
vectors are normalized and real. Then, consider that
[ [ ψ
1
) −[ ψ
2
) [
2
= 2 −2 !ψ
1
[ ψ
2
) ,
from which it follows that it sufﬁces to multiply both the lhs and the rhs of the above
inequality so as to obtain
[ [ ψ
1
) −[ ψ
2
) [ ÷[ [ ψ
2
) −[ ψ
3
) [ ≥ [ [ ψ
1
) −[ ψ
3
) [,
which is satisﬁed, since it is an instance of the triangular inequality (see also
Subsec. 2.3.2)
[a[ ÷[b[ ≥ [a ÷b[.
When applying the deﬁnition of the Fubini–Study distance to the distance between
the coherent states in Box 14.1, we must make use of the square modulus
(Eq. (13.70)), so that
d
f s
=
1 −e
−[αe
ı φ
−αe
−ı φ
[
2
=
1 −e
−4[α[
2
sin
2
φ
.
Expanding in power series to the ﬁrstorder in sin φ for φ < 1, we obtain
d
f s
. 2[α[ sin φ = d.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page73
73 Sol ut i ons to sel ected probl ems
t
Chapter 15
15.1 If you do not succeed, please refer to the quoted specialized literature.
15.7 It is easy to prove that (see also Prob. 13.21)
ˆ
H
xp
ˆ
D
xp
[ψ
n
) =
ˆ
D
xp
ˆ
H
ˆ
D
−1
xp
ˆ
D
xp
[ψ
n
)
=
ˆ
D
xp
ˆ
H[ψ
n
)
=
ˆ
D
xp
(−1)
n
[ψ
n
)
= (−1)
n
[ψ
n
xp
),
from which the solution follows.
15.9 First of all, we note that the exponential factor e
−ξα
∗
÷ξ
∗
α
may be rewritten as
e
2ı (ξ
r
α
i
−ξ
i
α
r
)
. We start from Eq. (15.59), which then turns to
W(α
r
, α
i
) =
1
π
2
÷∞
−∞
dξ
r
dξ
i
e
2ı (ξ
r
α
i
−ξ
i
α
r
)
χ(ξ
r
, ξ
i
).
Changing the variables according to
ξ
r
= −
1
√
2
ζ sin θ and ξ
i
=
1
√
2
ζ cos θ,
we also have dξ
r
dξ
i
=
1
2
[ζ [dζ dθ, and, using Eq. (15.64), we may rewrite the W
function as
W(α
r
, α
i
) =
1
2π
2
÷∞
−∞
dζ
π
0
dθ[ζ [e
−ı
√
2ζ (α
i
sin θ÷α
r
cos θ)
χ
℘
(ζ , θ).
Making the inverse Fourier transform of Eq. (15.61), that is,
χ
℘
(ζ , θ) =
d X℘(X, θ)e
ı ζ X
,
and inserting this into the previous equation, we ﬁnally obtain
W(α
r
, α
i
) =
1
2π
2
÷∞
−∞
d X
÷∞
−∞
dζ
π
0
dθ[ζ [℘(X, θ)e
ı ζ
¸
X−
√
2(α
r
cos θ÷α
i
sin θ)
¸
.
15.11 We sketch the main steps of the derivation. First, since all the basis states in
Eq. (15.71) are orthogonal, perform the square modulus of Eq. (15.72), so as to
obtain (see Sec. 9.9)
℘
e
(t , φ) =
∞
¸
n=0
ψ
e,n
(t )
2
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page74
74 Sol ut i ons to sel ected probl ems
t
Then, make use of the following facts:
∞
¸
n=0
[c
n÷1
[
2
= 1 −[c
0
[
2
,
ı
z − z
∗
= −2`(z), z ∈
(
C,
2 sin θ cos θ = sin 2θ,
cos
2
θ =
1 ÷cos 2θ
2
.
Chapter 16
16.1 From Eqs. (16.3)–(16.5) we have
+(x
1
, x
2
) =
1
2π
÷∞
−∞
dpe
ı
¨ h
(x
1
−x
2
÷x
0
) p
=
h
√
2π
δ(x
1
− x
2
÷ x
0
) =
h
√
2π
÷∞
−∞
dxδ(x
1
− x)δ(x − x
2
÷ x
0
)
=
1
2π
÷∞
−∞
dxδ(x
1
− x)
÷∞
−∞
dpe
−
ı
¨ h
(x−x
2
÷x
0
) p
=
1
√
2π
÷∞
−∞
dxψ
x
(x
2
)ϕ
x
(x
1
).
16.4 Let us take [ ς) = [ ς
k
) as an element of a complete set of orthogonal vectors {
ς
j
}.
Then, by Eq. (16.27), and by the fact that each
ˆ
P
ς
j
ϕ
cannot be negative, we
immediately have that for any j ,= k all
ˆ
P
ς
j
ϕ
vanish.
16.6 Since
ˆ
P
ς
1
ϕ
=
ˆ
P
ς
2
ϕ
= 0, then the rhs of Eq. (16.29) vanishes, which in turn
implies that
ˆ
P
ψ
1
ϕ
=
ˆ
P
ψ
2
ϕ
= 0. Writing, e.g., [ ψ
1
) as the linear combination
[ ψ
1
) = c
1
[ ς
1
) ÷c
2
[ ς
2
) ,
with arbitrary c
1
, c
2
, we obtain the desired result.
16.8 The problem is solved when considering that (see Eqs. (6.154))
ˆ σ
1
· a = a
x
¸
0 1
1 0
¸
1
÷a
y
¸
0 −ı
ı 0
¸
1
÷a
z
¸
1 0
0 −1
¸
1
and
ˆ σ
2
· b = b
x
¸
0 1
1 0
¸
2
÷b
y
¸
0 −ı
ı 0
¸
2
÷b
z
¸
1 0
0 −1
¸
2
,
The expectation value on the singlet state (16.12) of the product of these two scalar
products gives nine terms, of which the ﬁrst three have the form
9780521869638sol CUP/AUL November 6, 2008 18:21 Page75
75 Sol ut i ons to sel ected probl ems
t
+
0
a
x
b
x
ˆ σ
1x
ˆ σ
2x
+
0
= !+
0
[
a
x
b
x
√
2
¸
0 1
1 0
¸
1
¸
0 1
1 0
¸
2
¸
1
0
1
⊗
0
1
2
−
0
1
1
⊗
1
0
2
= !+
0
[
a
x
b
x
√
2
¸
0
1
1
⊗
1
0
2
−
1
0
1
⊗
0
1
2
= !+
0
[−a
x
b
x
[ +
0
) = −a
x
b
x
.
Similar calculations show that we also have
+
0
a
y
b
y
ˆ σ
1y
ˆ σ
2y
+
0
= −a
y
b
y
and
+
0
a
z
b
z
ˆ σ
1z
ˆ σ
2z
+
0
= −a
z
b
z
.
The six cross terms are instead all zero. Indeed, we have
+
0
a
x
b
y
ˆ σ
1x
ˆ σ
2y
+
0
= !+
0
[
a
x
b
y
√
2
¸
0
1
1
⊗
−ı
0
2
−
1
0
1
⊗
0
ı
2
=
a
x
b
y
2
¸
1 0
1
⊗
0 1
2
−
0 1
1
⊗
1 0
2
¸
¸
0
1
1
⊗
−ı
0
2
−
1
0
1
⊗
0
ı
2
=
a
x
b
y
2
[−(1 · ı ) −(1 · (−ı ))] = 0,
and similarly for all other cross terms, so that we may ﬁnally conclude that
!+
0
ˆ σ
1
·a
ˆ σ
2
·b
+
0
) = −(a
x
b
x
÷a
y
b
y
÷a
z
b
z
) = −a · b.
16.9 The solution is obtained when one considers that, for any integrable function f (x),
we have
dx f (x)
≤
dx [ f (x)[ ,
and [ A
b
(λ)A
c
(λ) −1[ = 1 − A
b
(λ)A
c
(λ)
16.10 We have that
a
/
, b
=
A
dλρ(λ)A
a
/ (λ)B
b
(λ) = 1 −δ
=
A
÷
dλρ(λ) −
A
−
dλρ(λ).
Since
A
÷
dλρ(λ) ÷
A
−
dλρ(λ) = 1,
we immediately obtain the desired result.
16.11 Making use of Eq. (16.65), we have that
A
dλρ(λ)B
b
(λ)B
b
/ (λ) =
A
÷
dλρ(λ)A
a
/ (λ)B
b
/ (λ) −
A
−
dλρ(λ)A
a
/ (λ)B
b
/ (λ),
9780521869638sol CUP/AUL November 6, 2008 18:21 Page76
76 Sol ut i ons to sel ected probl ems
t
which together with the fact that
A
÷
dλρ(λ)B
b
(λ)B
b
/ (λ) =
A
dλρ(λ)A
a
/ (λ)B
b
/ (λ) −
A
−
dλρ(λ)A
a
/ (λ)B
b
/ (λ),
allows us to derive
A
dλρ(λ)B
b
(λ)B
b
/ (λ) =
A
dλρ(λ)A
a
/ (λ)B
b
/ (λ) −2
A
−
dλρ(λ)A
a
/ (λ)B
b
/ (λ).
Now, we also have that
A
dλρ(λ)A
a
/ (λ)B
b
/ (λ) −2
A
−
dλρ(λ)A
a
/ (λ)B
b
/ (λ)
≥
A
dλρ(λ)A
a
/ (λ)B
b
/ (λ) −2
A
−
dλρ(λ)[ A
a
/ (λ)B
b
/ (λ)[.
Since A
a
/ (λ)B
b
/ (λ) = ±1, these results, together with Eqs. (16.66) and (16.62), give
the desired solution.
16.12 Let us do the substitution a
/
= b
/
in inequality (16.69) in its formulation:
[ !a, b) −
a, b
/
[ ≤ 2 ÷
a
/
, b
/
÷
a
/
, b
.
Since
b
/
, b
/
= −1,
the result is easily obtained.
16.13 By making use of Eq. (16.53), let us ﬁrst rewrite Eq. (16.55) as
!a, b) −
a, b
/
=
dλρ(λ) [A
a
(λ)B
b
(λ) (1 ± A
a
/ (λ)B
b
/ (λ))]
−
dλρ(λ) [A
a
(λ)B
b
/ (λ) (1 ± A
a
/ (λ)B
b
(λ))).
Since we have that
[A
a
(λ)]
2
= 1 ≥ A
a
(λ)B
b
(λ), A
a
(λ)B
b
/ (λ),
which implies both
dλρ(λ) [A
a
(λ)B
b
(λ) (1 ± A
a
/ (λ)B
b
/ (λ))] ≤
dλρ(λ) (1 ± A
a
/ (λ)B
b
/ (λ))
dλρ(λ) [A
a
(λ)B
b
/ (λ) (1 ± A
a
/ (λ)B
b
(λ))] ≤
dλρ(λ) (1 ± A
a
/ (λ)B
b
(λ)),
we also have
[ !a, b) −
a, b
/
[ ≤
dλρ(λ) (1 ± A
a
/ (λ)B
b
/ (λ)) ÷
dλρ(λ) (1 ± A
a
/ (λ)B
b
(λ)),
which amounts to
[ !a, b) −
a, b
/
[ ≤ 2 ±
a
/
, b
/
±
a
/
, b
,
from which the CHSH inequality follows.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page77
77 Sol ut i ons to sel ected probl ems
t
16.14 From Eqs. (16.73), (16.75), and (16.172) we obtain
−1 ≤ π
12
(λ, a, b) −π
12
(λ, a, b
/
) ÷π
12
(λ, a
/
, b) ÷π
12
(λ, a
/
, b
/
)
−π
1
(λ, a
/
) −π
2
(λ, b) ≤ 0,
where in Eqs. (16.171) we have taken
x
k
= π
k
(λ, j), with j = {a, a
/
},
y
k
= π
k
(λ, l), with l = {b, b
/
},
and X = Y = 1. Integrating this inequality over A with the probability distribution
ρ
λ
, and using Eqs. (16.74), yields
−1 ≤ ℘
12
(a, b) −℘
12
(a, b
/
) ÷℘
12
(a
/
, b) ÷℘
12
(a
/
, b
/
) −℘
1
(a
/
) −℘
2
(b) ≤ 0,
which may be rewritten as Eq. (16.76).
16.19 The threeparticle interference in the GHSZ state (Eq. (16.143)) is evidenced by the
sine term in Eq. (16.144a), which represents the probability that the three particles
are detected at D1, D2, and D3. In order to test if the twoparticle interference
is present, we have to consider the probability that two detectors (say D2 and
D3) click independently of what happens in the third arm of the inteferometer. In
this case, for instance, we have to sum Eqs. (16.144a) and (16.144b) and obtain
℘
+
D
2
D
3
(φ
1
, φ
2
, φ
3
) = 1¡4, which is independent of the phases φ
1
, φ
2
, and φ
3
.
Chapter 17
17.2 In the case of pure states, we have ˆ ρ =
ˆ
P
k
= [ b
k
) !b
k
[, which means that in
Eq. (17.4) we have n
k
δ
j k
, with the consequence that the diagonal matrix has only
an element different from zero, namely n
k
= 1. This proves the result because the
logarithmic function of 1 is 0.
17.8 Applying one of the unitary operators (17.73) to the corresponding state of particle
3 as represented in Eq. (17.71), we recover the information contained in particle 1,
which may be rewritten in the form
c
1
0
1
÷c
/
0
1
1
=
c
c
/
1
.
Indeed,
¸
−1 0
0 −1
¸
3
−c
−c
/
3
¸
−1 0
0 1
¸
3
−c
c
/
3
¸
0 1
1 0
¸
3
c
/
c
3
¸
0 1
−1 0
¸
3
−c
/
c
3
⎫
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎬
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎭
=
c
c
/
3
.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page78
78 Sol ut i ons to sel ected probl ems
t
In the case of the ﬁrst measurement outcome (the singlet one), the state of particle
3 is the same as of particle 1 except for an irrelevant phase factor, so that Bob
need do nothing further to reproduce the state of 1. In the three other cases, Bob
must apply one of the unitary operators
ˆ
U
2
,
ˆ
U
3
,
ˆ
U
4
– corresponding, respectively,
to 180
◦
rotations around the z, x, yaxes – in order to convert the state of 3 into the
state of 1. What Bob must do, obviously depends on the (classical) communication
of Alice’s result.
17.10 The most direct algorithm that can be used to factorize an integer number of l dig
its consists of checking whether the integer is divisible by each number from 2 to
√
M =
√
10
l
. Therefore, this algorithm requires a number of steps given by
N .
√
M =
10
l
,
which shows that the growth is exponential. For example, if l us equal to 100 dec
imal digits, even if the time required for a division is very small (say, 10
−20
s), the
total time required to factorize the original number is
τ . 10
10
10
−20
= 10
30
s,
i.e. 10
13
times the age of the universe. Of course, in the example above we have not
chosen the most efﬁcient algorithm. As a matter of fact, the best known classical
algorithm takes a time of the order of 10
l
1¡3
(i.e. it is subexponential).
17.11 The state (17.109) on which we want to evaluate the Boolean function may be
rewritten as
1
2
([ 0) ÷[ 1) ) ([ 0) −[ 1) ) =
1
2
([ 00) −[ 01) ÷[ 10) −[ 11) ).
Now, we evaluate each of the four components of the state above for each of the
four possible Boolean functions f
j
( j = 1, . . . , 4), i.e.
f
1
: [ 00) −[ 01) ÷[ 10) −[ 11) =
[ 0) Ê ÷[ 1)
([ 0) −[ 1) ),
f
2
: [ 01) −[ 00) ÷[ 11) −[ 10) =
[ 0) Ê ÷[ 1)
([ 1) −[ 0) ),
f
3
: [ 00) −[ 01) ÷[ 11) −[ 10) =
[ 0) Ê −[ 1)
([ 0) −[ 1) ),
f
4
: [ 01) −[ 00) ÷[ 10) −[ 11) =
[ 0) Ê −[ 1)
([ 1) −[ 0) ).
This leads to the desired solution.
17.12 The distance between ˆ ρ
1
and ˆ ρ
2
is given by
d( ˆ ρ
1
, ˆ ρ
2
) =
1
2
Tr[ ˆ ρ
1
− ˆ ρ
2
[
=
1
4
Tr[ (r −s) · ˆ σ[
=
1
2
[r −s[,
since (r −s) · ˆ σ has eigenvalues ±[r −s[, so that we obtain
Tr[ (r −s) · ˆ σ[ = 2[r −s[.
9780521869638sol CUP/AUL November 6, 2008 18:21 Page79
79 Sol ut i ons to sel ected probl ems
t
In other words, we have the important result that the distance between two single
qubit states is equal to one half the Euclidean distance between them on the Bloch
sphere. Since rotation on the Bloch sphere leaves this distance unaffected, unitary
transformations preserve it, i.e.
d( ˆ ρ
1
, ˆ ρ
2
) = d(
ˆ
U ˆ ρ
1
ˆ
U
†
,
ˆ
U ˆ ρ
2
ˆ
U
†
).
9780521869638sol
CUP/AUL
November 6, 2008
18:21
Page2
t
2
S o l u t i o n s to se l e c te d p ro b l e m s
the total number of different ways in which we can arrange the n quanta among the N oscillators (or boxes), i.e. the total number of permutations of the quanta n plus the N − 1 partitions1 wE = (N − 1 + n )! (N − 1)! n ! (N + n ) N +n , N N nn
where in the last equality we have made use of the Stirling formula √ m! m m e−m 2π m, which is a very good approximation for large m. Consequently, the total entropy is given by
N SB = kB N
1+
n N
ln (N + n ) − ln N −
n ln n . N
By making use of Eq. (1.62), we obtain the desired result. 1.12 In order to ﬁnd the solution, let us ﬁrst compute the differential
N N N ∂ SB ∂ SB 1 ∂ SB = = ¯ ¯ ∂U N ∂E ∂NE ¯ ¯ E E kB ln 1 + − ln =
=
kB
ln
¯ E+ ¯ E
=
1 . T
From the last equality it follows that ¯ E+ ¯ E and, ﬁnally, ¯ E= e
/kB T
=e
/k B T
,
−1
2
.
1.13 Applying the Carnot theorem to the upper triangle in Fig. 1.19 we obtain (nv)2 = hνs c
2
+
hνi c
−2
hνs hνi cos θ . c c
On the other hand energy conservation (Eq. (1.70)) yields mv 2 = 2h (νi − νs ). In order to eliminate v, we multiply this last equation times m and equate the resulting rhs to the rhs of the momentum conservation equation and obtain νi − νs = h ν 2 + νi2 − 2νs νi cos θ . 2mc2 s
1 See [Huang 1963, 181–82].
9780521869638sol
CUP/AUL
November 6, 2008
18:21
Page3
t
3
S o l u t i o n s to se l e c te d p ro b l e m s
Now, since the wavelength of the scattered photon is only slightly larger than the original one, we have νi Consequently, νi − νs = Replacing νi =
c λi
νs → νi2
2 νs
νi νs .
h νi νs (1 − cos θ ). mc2
and νs =
c λs
we ﬁnally obtain θ h 2h sin2 . (1 − cos θ ) = mc mc 2
λs − λi =
Chapter 2
ˆ ˆ 2.1 Take e.g. O = Iˆ − O. 2.2 h ψ v ψ = = = h b v b h  b⊥ v  b⊥ b ψ b⊥  ψ
h  b b  ψ + h  b⊥ b⊥  ψ v  b b  ψ + v  b⊥ b⊥  ψ h ψ v ψ .
The last passage is due to the wellknown property of projectors b b  + b⊥ ˆ ˆ b⊥  = Pb + Pb⊥ = Iˆ.
ˆ 2.3 Let us write the transposed conjugate of U : ˆ U† = Then we have ˆ ˆ UU† = = = = = Iˆ, since ( b b  +  b⊥ ˆ ˆ b⊥ ) = Iˆ. Similarly it can be shown that U † U = Iˆ. h b v b h  b⊥ v  b⊥ b h b⊥  h b v b⊥  v h  b b  v + h  b⊥ b⊥  v v  b b  v + v  b⊥ b⊥  v h  ( b v  ( b b  +  b⊥ b  +  b⊥ b⊥ )  v b⊥ )  v h b ∗ h  b⊥ ∗ v b ∗ v  b⊥ ∗ = b h b⊥  h b v b⊥  v .
h  b b  h + h  b⊥ b⊥  h v  b b  h + v  b⊥ b⊥  h h  ( b v  ( b h h v h b  +  b⊥ b  +  b⊥ h v v v = b⊥ )  h b⊥ )  h 1 0 0 1
2. ˆ where the o j are eigenkets of O. we derive that b(a + d) and a 2 = d 2 . and in this case we have ˆ O= 2.9780521869638sol CUP/AUL November 6.8 Let us take a generic ket  ψ . Collecting these results. from the condition of Hermiticity a c b d = a∗ b∗ c∗ d∗ a c b d a∗ b∗ c∗ d∗ a2 + b2 a ∗ c + b∗ d 1 0 0 1 ac∗ + bd ∗ c2 + d2 we obtain that a and d are real and that c = b∗ . Similarly. ˆ O ψ = j ˆ O oj oj  ψ oj oj oj  ψ j = = j ˆ o j Pj  ψ . and in ˆ this case b = 0 and O is a multiple of identity. or that a = −d.9 We proceed as follows: . From these relations it follows either that a = d. Then. Since these equalities must be valid for any  ψ .6 From the condition of unitarity ˆ ˆ O O† = = = we derive the following relations a2 + b2 = 1. ac∗ + bd ∗ = 0. = cv 2 − ch 2 . 2. ˆ ˆ it follows that O = j o j P j .7 The solution of the problem is ˆ b OP b = = ∗ ∗ v  cv + h  ch ( v v  −  h h ) (cv  v + ch  h ) ∗ ∗ v  cv + h  ch (cv  v − ch  h ) a b∗ b −a . 2008 18:21 Page4 t 4 S o l u t i o n s to se l e c te d p ro b l e m s 2. c2 + d2 = 1.
that is o j  ok = δ jk . i.  b2 = 0 1 We have to ﬁnd the vectors  o1 and  o2 and the (complex) numbers o1 and o2 such that ˆ O  o1 = o1  o1 . Let us take the eigenvalues o1 = +1 and o2 = −1. { o1 .9780521869638sol CUP/AUL November 6. the diagonal form of O is ˆ ˆ ˆ ˆ O = U † OU 1 1 −ı = 2 1 +ı = = 1 2 1 2 1 1 2 0 −ı +ı 0 −2 = 0 −ı ı 0 1 −1 ı ı 1 0 0 −1 . We have constructed the eigenvectors so that they are also normalized. . ˆ Notice that O is Hermitian.e. o1  o2 = 0.2 = ±1. The normalized eigenvector  o1 will then given by 1  o1 = √ 2 With a similar procedure we ﬁnd 1  o2 = √ 2 1 ı .  o2 } is an orthonormal basis on the bidimensional Hilbert space. we must have 0 −ı ı 0 1 o1 2 o1 = 1 o1 2 o1 2 1 that yields o1 = ıo1 . ˆ O  o2 = o2  o2 . 1 1 ı −ı 1 1 ı −ı . Rewriting explicitely the ﬁrst eigenvalue equation. The characteristic polynomial is given by λ −ı ı λ = λ2 − 1. In other words. whose zeros (eigenvalues) are λ1. 2008 18:21 Page5 t 5 S o l u t i o n s to se l e c te d p ro b l e m s (a) Let us take the usual basis  b1 = 1 0 . (b) The diagonalizing matrix is simply given by 1 ˆ U=√ 2 ˆ Finally.  o1 and  o2 are orthogonal. 1 −ı .
2008 18:21 Page6 t 6 S o l u t i o n s to se l e c te d p ro b l e m s 2. 16 2.10 The result is 9 N. 2. ˜ 2π 2. calculate +∞ ı 1 ˜ ¯ xψ(x) = √ dpx x ψ( px )e h px x 2π −∞ +∞ 1 h ı 1 ¯ ¯ ˜ = √ ψ( px ) e h px x −√ ı 2π 2π −∞ +∞ ˜ ı ∂ ψ( px ) ıh ¯ ¯ =√ dpx e h px x .16 The position eigenfunctions in the momentum representation are ı 1 ¯ ϕx0 ( px ) = √ ˜ d xe− h px x ϕx0 (x) 2π ı 1 ¯ =√ d xe− h px x δ(x − x0 ) 2π ı 1 ¯ = √ e− h px x 0 . 2.9780521869638sol CUP/AUL November 6.18 First.17 We proceed as follows: +∞ −∞ ˜ dpx ψ( px )2 = 1 dpx √ 2π +∞ +∞ ı ı 1 ¯ ¯ d xψ(x)e− h px x √ d x ψ ∗ (x )e h px x 2π −∞ −∞ −∞ +∞ +∞ +∞ ı 1 ¯ = dx d x ψ(x)ψ ∗ (x ) dpx e h px (x−x ) 2π −∞ −∞ −∞ = +∞ −∞ +∞ ˜ ˜ dpx ψ( px )ψ ∗ ( px ) = = +∞ +∞ −∞ +∞ −∞ dx −∞ d x ψ(x)ψ ∗ (x )δ(x − x ) d xψ(x)2 . ∂ px 2π −∞ +∞ −∞ ˜ ∂ ψ( px ) h ı ¯ ¯ dpx e h px x ı ∂ px . 2π In the general case we have ı 1 ¯ ϕx ( p x ) = √ e− h px x .15 The momentum eigenfunctions in the momentum representation are ı 1 ¯ ϕ p0 ( p x ) = √ ˜ d xe− h px x ϕ p0 (x) 2π ı 1 ¯ =√ d xe− h ( px − p0 )x 2π = δ( px − p0 ).
¯ ∂ px −∞ +∞ This has to be equal to the expectation value of x computed in the momentum ˆ representation. 2π ˆ This shows that the action of the position projector P(x) onto the momentum eigenvector  px (in which the position is completely undetermined) selects the position eigenvector  x with a weight that is given by ϕ p (x).19 We have ˆ P(x)  px =  x x  px = ϕ p (x)  x . ˜ ∂ ψ( px ) . ∂ px +∞ −∞ ˜ dpx ψ ∗ ( px )x ψ( px ). we ﬁnd x ψ( px ) = ı h ˆ˜ ¯ 2.e. 2008 18:21 Page7 t 7 S o l u t i o n s to se l e c te d p ro b l e m s where we have made use of the inverse Fourier transform (2. ˆ we obtain x = ˆ +∞ ˜ ı ∂ ψ( px ) 1 d xψ ∗ (x) √ dpx ı h e h px x ¯ ¯ ∂ px 2π −∞ −∞ +∞ +∞ ˜ ı 1 ∂ ψ( px ) ¯ = dpx √ d xψ ∗ (x)e h px x ı h ¯ ∂ px 2π −∞ −∞ +∞ ˜ ∂ ψ( px ) ˜ = dpx ψ ∗ ( px )ı h . a→0 2πa 2 . we have 2 δ 2 (x) = lim f a (x) a→0 = lim 2 1 − x2 e a . that is δ(x) = lim f a (x). Then. 2. If we substitute this result into the deﬁnition of the expectation value of x. that is ı 1 ¯ ϕ p (x) = √ e h px x . a→0 1 2πa e − x2 2a 2 .9780521869638sol CUP/AUL November 6. i. ˆ˜ where ϕ p (x)  x is the eigenfunction of the momentum operator corresponding to the eigenvalue px .150a) and of integration by parts.22 The delta function can be obtained as a limit of a normalized Gaussian f a (x) = √ with a > 0. x = ˆ By comparison.
2. corresponds to the state  1 . 2 √ which corresponds to the state  2 .4 shows that there are cases in which an event occurs (a certain detector clicks.24 x E ≥ 1 hm0 . nor in the state expressed by the proposition a .26 The solution is given by ˆ px . we have f 1/2 = 1 − 1 + eıφ  3 + ı 1 − eıφ  4 2 1 1 1 = √ √ (−  3 + ı  4 ) − √ eıφ ( 3 + ı  4 ) 2 2 2 1 = √  f 0 − eıφ  f 1 . 2 2.31 The examination of Sec.30 With T = 0 and R = 1 we have 1  f 1 = − √ eıφ ( 3 + ı  4 ). we have 1  f 0 = √ (−  3 + ı  4 ). 2 π a→0 a ¯p 2. up to the irrelevant global phase factor. With T = 1 and R = 0. it is true that c∧ a ∨a ∧¬ c∧a ∨ c∧a . in these cases. 2008 18:21 Page8 t 8 S o l u t i o n s to se l e c te d p ro b l e m s from which we obtain +∞ −∞ d xδ 2 (x) = +∞ 1 − x2 1 d x lim 2 e a2 a→0 a 2π −∞ 2 +∞ 1 1 −x lim 2 = d xe a2 2π a→0 a −∞ 1 1 = √ lim = +∞. a state of affairs described by the proposition c) and not withstanding the object system is neither in the state expressed by the proposition a . with T = R = 1/ 2. and. 2 2. f (x) = ˆ h ∂ h ∂ ¯ ¯ f (x) − f (x) ˆ ˆ ı ∂x ı ∂x h h ∂ h ∂ ¯ ¯ ¯ ˆ = f (x) + f (x) ˆ − f (x) ˆ ı ı ∂x ı ∂x h ¯ ˆ = f (x) ı 2. that ¬ c∧a and ¬ c∧a . from which we easily derive that it is also true that ¬ c ∧ a ∨ c ∧ a therefore. Finally. 2 which.9780521869638sol CUP/AUL November 6. This means that.
cn (t)ψn (x). t)2 = = n. both having the same ¯ 2k2 positive energy E = h2m . we have assumed c to be true. However. respectively.3 ψ(x) = √1 e±ıkx .26) as ψ(x. where ωk = E k2 .m dx m ∗ ∗ cm (t)ψm (x) n cn (t)ψn (x) ∗ cn (t)cm (t) ∗ d xψm (x)ψn (x) ¯ e− h E n t cn (0) ı = n cn (t)2 = n 2 = n cn (0) = 1. t) = n (0) ˜ ¯ e− h E n t cn ψn ( px ) ı where ˜ ψn ( px ) = px  ψn 3.6.2. This means that we have a situation in which a ∨ a is true (by deﬁnition) but also both ¬a and ¬a are. 2 where ∗ d xψm (x)ψn (x) = δn. =h ¯ 2m h ¯ . 3.2 It is straightforward to prove the result given the linearity of the Schrödinger equation. both a and a are false. By substituting a to ¬a as well as a to ¬a we obtain that also ¬ a ∧ a is true. and. t) = px  ψ(t) .5 Let us rewrite Eq. from which we obtain d x ψ(x. either c or a must be false. 3.9780521869638sol CUP/AUL November 6.4 At time t0 = 0 and t we have ˜ ψ( px .m . the plus and minus signs correspond to “waves” moving from left to right and from right to left.4 and 2. in the ﬁrst case. 2008 18:21 Page9 t 9 S o l u t i o n s to se l e c te d p ro b l e m s are also true. 3.2. These are the plane waves described in Subsecs. 0) = n (0) ˜ cn ψn ( px ) and ˜ ψ( px . which is the desired result.7 We have ψ(t) = dkc(k)eı(kx−ωk t) . t) = n and ˜ ψn ( px . Then. 2π They are doubly degenerate: in fact. 2. Chapter 3 3. in the second case either c or a . (3. This means that.
10 First let us compute the mean value of the position2 a x = ˆ 0 ∗ d xψn (x)xψn (x) a 2 = a 0 d x x sin2 nπ nπ x a 2a = (nπ )2 0 dyy sin2 y a = . and of the square of the momentum a px ˆ2 = −h ¯ 22 a 0 d x sin a ∂2 nπ nπ x x sin 2 a a ∂x nπ x a 2h 2 n 2 π 2 ¯ = a a2 0 d x sin2 2 See [Gradstein/Ryshik 1981. 2.9780521869638sol CUP/AUL November 6. as well as the mean value of the square of the position3 a x ˆ 2 2 = a 0 d x x 2 sin2 nπ nπ x a = = 2a 2 (nπ )3 a2 − dyy 2 sin2 y 0 1 a 2 .2]. 3. we calculate the mean value of the momentum a px ˆ 2 = −ı h ¯ a 0 d x sin a nπ ∂ nπ x sin x a ∂x a nπ nπ x cos x a a a 0 2h nπ ¯ = ıa a 0 d x sin 1 2nπ h − cos x = 2ıa 2π a = 0.631.821]. 2 where we have made use of the fact that we are in the position representation and that ∗ ψn (x) = ψn (x). 3 2 nπ Analogously. 3 See [Gradstein/Ryshik 1981. 2008 18:21 Page10 t 10 S o l u t i o n s to se l e c te d p ro b l e m s 3. .
for n = 1. 12 2n π and ﬁnally obtain the uncertainty relation ⎤ ⎡ h ⎣ n2π 2 ¯ − 2 ⎦. in order to verify that the uncertainty relation is always satisﬁed it is sufﬁcient to prove that. ˆ ˆ ˆ ˆ This problem is separable. where h2 ∂ 2 ¯ ˆ H1 (x) = − . y. 4a 2 Then. The required solution therefore reduces to (1) (2) (3) ψn x .136.n y . z) = ψ (1) (x) · ψ (2) (y) · ψ (3) (z). z) = Eψ(x. z). for x < 0. y. y > b . i. For a particle in a “cubic” box. we calculate the uncertainties of position and momentum px = nh . 2a x =a 1 1 − 2 2. z < 0. . z) = 0. H = H1 (x) + H2 (y) + H3 (z). A direct calculation shows that π2 − 2 = 1.9780521869638sol CUP/AUL November 6. y. z). z) = ψn x (x) · ψn y (y) · ψn z (z). y.11 The threedimensional stationary Schrödinger equation for the wave function in the position representation reads as − h2 ¯ 2m ∂2 ∂2 ∂2 + 2+ 2 2 ∂x ∂y ∂z + V (x.n z (x. and E = Ex + E y + Ez . so that the previous equation becomes − h2 ¯ 2m ∂2 ∂2 ∂2 + 2+ 2 ∂x2 ∂y ∂z ψ(x. y. y. z) = 0 inside the box. z > c. x > a . we have V (x. it is ≥1.e. with the boundary conditions ψ(x. x px = 2 3 Since the square root is certainly a growing function of n for n ≥ 1. z) = Eψ(x. z) ψ(x. 2m ∂ x 2 ˆ ˆ and similar expressions for H2 (y) and H3 (z). y < 0. y. y. 2008 18:21 Page11 t 11 S o l u t i o n s to se l e c te d p ro b l e m s = h2n 2a 2 π nπ nπ 1 nπ x − sin x cos x 2 a a a nπ 0 h2n2 = . In ψ(x. y. 3 3.
2ma 2 x π h2 2 ¯ = n .9780521869638sol CUP/AUL November 6. a a nyπ 2 sin y . the transformation effected by an asymmetric beam splitter should satisfy the following constraints: ˆ UBS ˆ UBS 1 0 0 1 ∝ ∝ T ıR. b b nz π 2 sin z . in the case of beamsplitting (see Subsec. c c π h2 2 ¯ n . 2mb2 y π h2 2 ¯ = n . Furthermore. it is clear that ¯ c j (t) = c j e− h E j t = 0.12 Referring to Fig. 3. ı where use has been made of the eigenvalue equation ˆ H ψj = E j ψj . ıR T . 3.2). cj ψj = j j ¯ c j e− h E j t ψ j . ı if c j = 0. 2mc2 z and Enx = En y Enz 3. The requirement of unitarity leads us to the ﬁnal form ˆ UBS = 3.8(b). the vectors 1 = 1 0 and 2 = 0 1 . we also have that the probabilities of the energy eigenvalues ℘ j (t) = 2 c j (t) = ℘ j (0) are constant under time evolution. 2008 18:21 Page12 t 12 S o l u t i o n s to se l e c te d p ro b l e m s with (1) ψn x (x) = (2) ψn y (y) = (3) ψn z (z) = nx π 2 sin x .5. Now.13 We have ˆ ˆ ¯ Ut  ψ = e − h H t ı T ıR∗ ıR T∗ . Note that.
2008 18:21 Page13 t 13 S o l u t i o n s to se l e c te d p ro b l e m s are neither eigenvectors of the beamsplitter unitary transformation 1 ˆ UBS = √ 2 1 ı ı 1 .t  ψ(t) ˆ† ˆ ˆ = U H0 . H = 0. If this connection had to be not completely clear. 3.t H U H0 .t HI U H0 . O .t  ψ(t) = ˆ HII (t)  ψ(t) I .17 We have to show that ˆ ˆ ˆ O. O = −nı h O n−1 .25) ˆ ˆ On. it is straightforward to ¯ prove that (see Prob. i.t − H0 + H U H0 . ˆ ˆ ˆ ¯ since H S = H commutes with Ut = e− h H t . 3.19 We can proceed as follows: ıh ¯ d  ψ(t) dt I ı ˆ ˆ O ˆ O d ı H0 t ˆ ¯  ψ(t) S eh dt ı ˆ ı H0 t ˆ ¯  ψ(t) = ıh H0 e h ¯ h ¯ = ıh ¯ S ¯ + e h H0 t ı ˆ d  ψ(t) dt S ˆ ˆ† ˆ† ˆ = − H0 U H0 . H = O O . For instance.e.t H0 + U H0 . the state √ ( 1 − ı  2 ) is transformed by UBS into  1 .2. ˆ ˆ ˆ since H commutes with both O and O .22 Given any two operators O and O such that [ O.9780521869638sol CUP/AUL November 6.14 This result may be proved by taking into account the uniqueness of the unitary ˆ ˆ transformation U † = U −1 (its deterministic nature). H ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = OO H − HOO − O OH + HO ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = HOO − HOO − HO O + HO = 0. H − O O.t H  ψ(t) S I ˆ† ˆ ˆ† ˆ ˆ = −U H0 . The explicit calculation is ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ O.t  ψ(t) ˆ ˆ† ˆ ˆ = U H0 .t + U H0 . This is the reason why certain superposition states of the basis vectors U  1 and  2 may be transformed into  1 or  2 under the unitary transformation 1 ˆ ˆ UBS . ¯ ˆ . see Sec. 3. ˆ ˆ BS = eı O .15 We have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ H H (t) = Ut† H S Ut = Ut† Ut H S = H S . ˆ nor of the Hermnitian operator O that is the generator of the transformation. 2 3. O . 15. 2. O ] = −ı h . I I ˆ ˆ ˆ ˆ 3.
¯ which yields the desired result. we may compute the commutator ⎡ ⎤ ∞ j (ıα) j ˆ ⎦ ˆ ˆ eıα tˆ.H = j! j=0 ∞ ∞ j=0 (ıα) j ˆ ˆ t.1 The energy levels are E(i. H j! =− j=1 ∞ (ıα) j j! jı h t j−1 ¯ˆ = αh ¯ j=1 ∞ (ıα) j−1 j−1 ˆ t ( j − 1)! (ıα)n n ˆ t (n)! = αh ¯ = αhe ¯ Therefore.3 (a) The normalization coefﬁcients N + and N− may be derived from the continuity of the wave function and its ﬁrst derivative at x = 0. ψ< (0) = ψ> (0). 2008 18:21 Page14 t 14 S o l u t i o n s to se l e c te d p ro b l e m s Using this result. that is.k (r) = ψi (x)ψ j (y)ψk (z) = iπ 8 sin x sin abc a jπ kπ y sin z . j. b c π h2 ¯ 2m i2 j2 k2 + 2 + 2 a2 b c . A simple calculation yields the desired result k2 − k1 . k) = E x (i) + E y ( j) + E z (k) = and the wave functions are given by ψi. from which we obtain the conditions N + = 1 + N− . N+ = k1 + k2 N− = . ψ< (0) = ψ> (0).9780521869638sol CUP/AUL November 6. H = ⎣ t . Chapter 4 4. k1 + k2 2k2 . j. n=0 ˆ ıα t . k 1 N+ = k 2 − k 2 N− . ˆ ˆ H eıα tˆ = eıα tˆ H − α h eıα tˆ. 4.
2 cos(k2 a) − ı sin(k2 a) . h ¯ k2 = 1 2m(E − V0 ).29). and we have a total reﬂection of the particle at the potential barrier. It should also be noted that. for x ≥ a. B. ψII (a) = ψIII (a). The constants A. −2 1− from which we ﬁnally obtain that T2 /R2 = 16/15. together with the previous result. 2008 18:21 Page15 t 15 S o l u t i o n s to se l e c te d p ro b l e m s (b) Equations (4. there is a nonzero probability of ﬁnding the particle in the classically forbidden region. the ratio T2 /R2 vanishes. ψII (x) = Beık2 x + B e−ık2 x . However. h ¯ ψI (x) = eık1 x + Ae−imathk1 x . for 0 ≤ x ≤ a. which yield 1+ A = B+ B . ψIII (x) = Ceık1 x . we have R2 = A2 and T2 = k1 C2 = C2 . Nevertheless. 4. where k1 = 1√ 2m E. + B e−ık2 a = Ceık1 a . B . this probability exponentially vanishes with x.9780521869638sol CUP/AUL November 6. and C may be derived from the conditions ψI (0) = ψII (0). yield k1 2 4k1 k2 T2 k2 N+  = = 2 R2 (k2 − k1 )2 N−  = 4 1− 2− V0 E V0 E V0 E . the wave function is for x ≤ 0. k1 ı 2 2 k2 −k1 k1 k2 sin(k2 a) 2 2 k1 +k2 k1 k2 . when E = V0 . k2 Beık2 a − k2 B e−ık2 a = k1 Ceık1 a . From these conditions we obtain A= In this case. Be ık2 a ψI (0) = ψII (0). k1 − k1 A = k2 B − k2 B . even in this case (and also for E < V0 ).4 (a) For E > V0 . ψII (a) = ψIII (a).
n+2 = (n + 2)h ¯ .70) is satisﬁed for n = 0 (see Eq. if Eq. we immediatly obtain T2 = C2 = 1 − A2 2 2 4k1 k2 = 2 2 2 2 4k1 k2 + k1 − k2 2 sin2 (k2 a) .n+1 = (n + 1)h ¯ . ∂x . we may choose the eigenfunctions to be real.70) is assumed to be valid for n. 2mω h ¯ . 4. (4. 2m = ω and ωn. (b) For E < V0 . The other method starts from the deﬁnition of +∞ +∞ 2 d x ψn (x)x. if k = n.e.n ωn.n+1 = −ω. k2 becomes pure imaginary. (4. The expression for the transmission coefﬁcient may be obtained from the previous result if one replaces k2 by ıκ2 . which implies that the wave function decreases exponentially in region II. in order to solve the problem.9780521869638sol CUP/AUL November 6. i. The ﬁrst method is direct inspection of Eq. (4. we can derive then it is also satisﬁed for n + 1. (4.n+1 + xn+1.7 There at least two simple ways to obtain this result.60). we have to verify that Eq.68) fo n + 1 2 2 xn+1. Then. both of which do not involve anything but straightforward calculations. Then. where 1 2m(V0 − E).n+1 xn+1. −∞ ˆ xnn = x n ˆ = n x n = −∞ ∗ d x ψn (x)xψn (x) = where we recall that ψn (x) = x  n is the nth harmonic oscillator (real) eigenfunction. 2mω that completes the argument. These eigenfucntions are either odd or even functions of x.69)).n+2 ωn+2. the integrand of the last equality in the previous equation is necessarily odd. xn. 2008 18:21 Page16 t 16 S o l u t i o n s to se l e c te d p ro b l e m s From T2 + R2 = 1. 4. First. As a consequence. κ2 = h ¯ 4.6 We may use an inductive argument. it sufﬁces to show that. we have that ωnk = 0 and therefore xnk = 0. be using the relations ωn+2.n+1 = from which.8 We have h ∂ ¯ ∗ ψn (x). hence the integral over the whole line is zero. (4. ˆ n px n = d x n  x x  px  n = d xψn (x) ı ∂x Without loss of generality. In fact. We rewrite Eq. ˆ n px n = h ¯ ı d xψn (x) ∂ ψn (x).
12 We have a.97)). a† ˆ ˆ 2 and hω ¯ ˆ H= a a† + a† a .73). we arrive at the desired result. (2. a† a = a† a† a − a† a a† ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆˆ = a† a† .9 We have m ˆ ˆ ˆ ˆ a. we can write the product a a† = ˆˆ m ˆ ˆ ω x + ıx ω x − ıx ˆ ˙ ˆ ˙ 2h ω ¯ 1 p2 ˆ = mω x 2 + x + ı px . ˆ ˆ ¯ 4. x + ıω x. x = −ı h .74)) we can derive 1 ˆ H = h ω a a† − ˆˆ ¯ 2 4.11 We have a. a† a = a. N = a. O (see where we have made use of the fact that px ˆ ˆ x= ˙ m and px . px = ı h . ˆ ˆ ˆ a 4. O = 0.9780521869638sol CUP/AUL November 6. a† = ˆ ˆ ω2 x. and that x. x + x. a† = 2ˆ † .10 Starting from Eqs. N = a† .94) and (2. a† a = a a† a − a† a a ˆ ˆ ˆ ˆ ˆ ˆˆ ˆ ˆ ˆˆ ˆ = a. px = Iˆ. a = −ˆ † . 2008 18:21 Page17 t 17 S o l u t i o n s to se l e c te d p ro b l e m s Integrating by parts the previous expression. 4. 2 ˆ ˆ = − O . ˆˆ ˆ ˆ 2 = a a† a† − a† a† a ˆˆ ˆ ˆ ˆ ˆ = a a† a† − a† a a† + a† a a† − a† a† a ˆˆ ˆ ˆ ˆˆ ˆ ˆˆ ˆ ˆ ˆ a ˆ ˆ ˆ = a. x ˆ ˆ ˙ ˆ ˆ ˙ ˙ ˙ 2h ω ¯ ım x. ˆ ˆ ˆ . (4. x ˆ ˆ ˙ =− h ¯ ı = − x. a† a† + a† a. (4. that O. O Eqs. The same result nay be obtained taking into account the fact that the ﬁrst derivative of an even (odd) function is odd (even). ˆ ˆ h ¯ ˆ ˆ ˆ ˆ where we have made use of the fact that O. x − ıω x. ˆ ˆ ˆ and a† . x ˆ ˆ ˆ 2h mω ¯ = 1 hω ¯ 1 p2 ˆ mω2 x 2 + x ˆ 2 2m 1 + . ˆ ˆ ¯ From the above expression and the commutator (Eq.
85) we immediately obtain  1 = a†  0 . a† ˆ ˆ n + a† ˆ = (n + 1) a† ˆ . dye−ay = 2 π . a† ˆ n+1 n d xe− mω 2 h x ¯ . we have a† ˆ 0 0 = √ √ √ (n + 1) ! n + 1 n! a† ˆ n = n + 1 .325].9780521869638sol CUP/AUL November 6. a† ˆ ˆ n = n(ˆ † )n−1 a holds for a given n.13 Again. a† ˆ ˆ = n a† ˆ n a† + a† ˆ ˆ n a. 3. From the second of Eqs. a 4 See [Gradstein/Ryshik 1981. =√ n+1 a† ˆ 4. 2008 18:21 Page18 t 18 S o l u t i o n s to se l e c te d p ro b l e m s and a 2 . a† a = 2a. 4.14 We must have 1 = N 2 Using the formula4 +∞ −∞ +∞ −∞ a† ˆ n+1 (n + 1) ! 0 . we have to prove that it also holds for n + 1. ˆ ˆ ˆ which can be both generalized by induction. a† + a. that is a. ˆ Assuming that the relation holds for a given n. a† ˆ ˆ n+1 = a a† ˆ ˆ = a a† ˆ ˆ n n a† − a† ˆ ˆ a† − a† ˆ ˆ n n n a† a ˆ ˆ a a† + a † ˆˆ ˆ n n a a† − a† ˆˆ ˆ n a† a ˆ ˆ = a. . n + 1 = √ In fact. Assuming that the relation a. we prove this relation by induction. a† = a a a† − a† a a ˆ ˆ ˆˆˆ ˆ ˆˆ = a a a† − a a† a + a a† a − a† a a ˆˆˆ ˆˆ ˆ ˆˆ ˆ ˆ ˆˆ ˆ ˆ ˆ ˆ = a a. (4. We prove only the ﬁrst one. that is. we must prove that it holds for n + 1 as well.
we obtain an identity. 2 mω h . we have to 2 discard the solution ψ(ξ ) = eξ /2 . taking N real for simplicity and without loss of ¯ generality (see Property (iv) of Subsec. (i) The ﬁrst six Hermite polynomials are H0 (ζ ) = 1. that is n n+1 d 2 d 2 2 d 2 H(ζ ) = (−1)n 2ζ eζ e−ζ + (−1)n eζ e−ζ dζ dζ n dζ n+1 n n+1 2 d 2 2 d 2 = 2ζ (−1)n eζ e−ζ − (−1)n+1 eζ e−ζ n n+1 dζ dζ = 2ζ Hn (ζ ) − Hn+1 (ζ ). we may neglect 2E/h ω with respect to ξ 2 and the asymptotic solutions of ¯ d2 ψ(ξ ) = ξ 2 ψ dξ 2 are ψ(ξ ) = e±ξ /2 . ¯ and make the change of variable ξ = x d2 ψ(ξ ) + dξ 2 The resulting equation is 2E − ξ 2 ψ(ξ ) = 0. As a consequence. 4. H4 (ζ ) = 16ζ 4 − 48ζ 2 + 12. 2 H1 (ζ ) = 2ζ .17 We write the Schrödinger equation as 2m d2 ψ(x) + 2 2 dx h ¯ 1 E − mω2 x 2 ψ(x) = 0. 2008 18:21 Page19 t 19 S o l u t i o n s to se l e c te d p ro b l e m s √ with mω/h = a. ξ2 . H3 (ζ ) = 8ζ 3 − 12ζ .2).2. Replacing the ﬁrst and the second derivatives into the differential equation as above. dζ dζ 2 where we have made use of the ﬁrst recursion relation in (iii) twice. H5 (ζ ) = 32ζ 5 − 160ζ 3 + 120ζ . H2 (ζ ) = 4ζ − 2. N = 4.9780521869638sol CUP/AUL November 6.16 mω πh ¯ 1 4 . we obtain. hω ¯ It is now convenient to consider the asymptotic behavior of ψ(ξ ): for large ξ 2 . Let us start from direct differentiation of the nth order Hermite polynomial. (iv) We start from d2 d Hn (ζ ) = 2n Hn−1 (ζ ) = 4n (n − 1) Hn−2 (ζ ). (iii) We limit ourselves to derive the third recursion relation. 3. it is natural to make the ansatz 2 ψ(ξ ) = e− 2 ϕ(ξ ). Due to the ﬁniteness condition of ψ(ξ ) for ξ → ±∞.
˙ dt sin2 (ωt + φ) 0 τ A2 τ dt 0 A2 E cl 1 − cos[2(ωt + φ)] = = .9780521869638sol CUP/AUL November 6. we have to average over a time period of the motion (τ = 2π/ω). whereas x 2 cl A2 = τ = and 2 ( px )cl = m 2 x cl = ˙ 2 τ x cl = 0. 9 and 10). The solutions of the previous equation with the condition that ϕ(ξ ) be ﬁnite for ﬁnite values of ξ and that it grow at most as a power of ξ for ξ → ±∞ exist only for integer values of n = E/h ω − 1/2 ≥ 0 and are given by the Hermite polynomials (see ¯ Prob. ˙ The total energy of the system is then given by E cl = 1 1 2 m xcl + ω2 xcl = mω2 A2 . ˙2 2 2 that yields A = 2E cl /(mω2 ). 8–9] . As we shall see later (in Chs. We have x cl = 0. 4.16) up to a normalization factor. The equation of motion for a onedimensional classical harmonic oscillator6 is xcl (t) = A sin(ωt + φ). In the case of the harmonic oscillator.5 4. (4. This gives the eigenvalues (Eq.97)).19 It is always interesting to make a comparison between the quantum results and the corresponding classical ones. from which xcl (t) = Aω cos(ωt + φ). 2008 18:21 Page20 t 20 S o l u t i o n s to se l e c te d p ro b l e m s and obtain for ϕ(ξ ) the differential equation d2 d + − 2ξ 2 dξ dξ 2E −1 hω ¯ ϕ(ξ ) = 0.72)) and eigenfunctions (Eq. however. (4. in general the classical limit in quantum mechanics is far from being obvious. 6 See [Goldstein 1950. Ch. Chs. this is enlightning and relatively straightforward. and represents an alternative solution of the harmonic oscillator problem. 2 5 See [Landau/Lifshitz 1976b. 2 2 mω2 1 2 2 2 A ω m = m E cl . instead of averaging – as in the quantummechanical case – over the ensemble. 3]. In order to obtain the classical mean values.
px ˆ2 ψn ˆ ˙ ˆ2 = n px n = m 2 n x 2 n = m2 k ˆ ˆ ˙ ˙ n x k k x n ıωnk xnk ıωkn xkn k = m2 1 1 = mωh (n + 1) + mωh n = m E n . This result is valid in this particular case but. we calculate the corresponding quantummechanical mean values. We have (see Probs. 4.21 Under the harmonicoscillator Hamiltonian 1 ˆ ˆ . ¯ as happens for all coherent states (the subject of Subsec.4.7 and 4. 4. Similarly. we immediatly have ψn x · ψn px = 1 En =h n+ .7). due to the Ehrenfest theorem (see Sec. (2. ˆ = n x2 n = k ˆ ˆ n x k k x n ˆ ˆ ˆ ˆ = n x n+1 n+1 x n + n x n−1 n−1 x n h (n + 1) hn En ¯ ¯ = . ¯ ¯ 2 2 The results above show a complete correspondence between the classical and the quantum expectation values. This ensures that the lower bound of the uncertainty product given by Heisenberg relation is the best constraint attainable.8) x ˆ and x2 ˆ ψn ψn = px ˆ ψn = 0. H = hω N + ¯ 2 any initial state  ψ(0) = n cn  n evolves according to ¯  ψ(t) = e− h H t  ψ(0) .20 Using the results of the Prob. ı ˆ . 4.2). 4. + = 2mω 2mω mω2 where we have made use of Eq.184).19 and Eqs. ¯ ω 2 Here.70). 3.9780521869638sol CUP/AUL November 6. 2008 18:21 Page21 t 21 S o l u t i o n s to se l e c te d p ro b l e m s Next. it cannot be generalized. 13. (4. it is clear that the harmonic oscillator is an exceptional case: the Gaussian wave function of the ground state saturates the uncertainty relation ( ψn x · ψn px = h /2).
.139). (4.15)) m ∂ Ay d ∂ Ax ∂ Az e d ∂U e vx + A x = −e + vx + vy + vz .e. Then. Notice that this result corresponds to the xcomponent of Eq. One may proceed in a similar way to derive the expressions for the y and z components. + e (v × B)x . 4. we have ¯  ψ(τ ) = e− h H τ  ψ(0) ı ˆ = n cn e cn e n −ıπ ˆ −ıω N + 1 τ 2 n n = =e 1 −ı n+ 2 2π  ψ(0) .127). making use of Eq.22 Let us explicitly derive the xcomponent of Eq. c ı j k. i. dt c dt ∂x c ∂x ∂x ∂x d Ax ∂ Ax ∂ Ax ∂ Ax ∂ Ax = + vx + vy + vz . dt ∂t ∂x ∂y ∂z Collecting these results together. while we have ∂ Ay ∂L ∂ Ax ∂ Az ∂U e vx . = −e + + vy + vz ∂x ∂x c ∂x ∂x ∂x from which we derive (Eq.9780521869638sol CUP/AUL November 6.127). (4. we have ∂ Az ∂ Ax − ∂x ∂z . we obtain m ∂U ∂ Ax ∂ Ax d ∂ Ax vx = e − − + vx − vx dt ∂t ∂x ∂x ∂x ∂ Ay ∂ Ax ∂ Ax ∂ Az − vz + vy + vz −v y ∂y ∂z ∂x ∂x ∂ Ay ∂ Ax ∂U ∂ Ax =e − − − vy − + vz ∂t ∂x ∂x ∂y ∂ d ∂ px = e − U − A x dt ∂x ∂t since the explicit expression for v × B is given by v×B= v × (∇ × A) ∂ Ay ∂ Ax ∂ Ax ∂ Az = vy − − vz − ∂x ∂y ∂z ∂x ∂ Ay ∂ Ay ∂ Az ∂ Ax − − vx − + vz ∂y ∂z ∂x ∂y ∂ Ay ∂ Ax ∂ Az ∂ Az − − vy − + vx ∂z ∂x ∂y ∂z But. 2008 18:21 Page22 t 22 S o l u t i o n s to se l e c te d p ro b l e m s After one period τ = 2π/ω. ∂x ˙ c ˙ with vx = x. we have ∂L e = mvx + A x . (4. (1.
λ2 = 0. which are 1  + = √ ( h +  v ). is ˆ 1 −λ 2 2 − 1 = 0. We can now compute the eigenvectors. 2008 18:21 Page23 t 23 S o l u t i o n s to se l e c te d p ro b l e m s Chapter 5 5.2 Equation (2. since unitary transformations preserve scalar products (and therefore probabilities). 2 The diagonalized form of ρ is then: ˆ ˆ ˆˆ U −1 ρ U = where 1 ˆ U=√ 2 1 1 1 −1 .6 Let us ﬁrst compute the density matrix ρ = ψ ψ  = ˆ 1 2 1 1 1 1 . Det ρ − λ Iˆ = 0. we have 5. ˆ where ϕξ (x) = x  ξ . 2 2 ˆ ˆ where P2 is the projector complementary to P1 . 4 from which we easily obtain the solutions λ1 = 1. 5. ξ . it is easy to show that ρ has the form ρ = ˆ 1 ˆ 1 ˆ P1 + P2 . It follows that. This is a further formulation of the measurement problem.117) states that ∗ d xϕξ (x)ϕξ (x) = δ(ξ − ξ ).5 It conﬁrms that there is no unitary transformation from a pure density matrix to a mixtures. 1 0 0 0 .9780521869638sol CUP/AUL November 6. Its characteristic equation. for ξ = ξ and ρ =  ξ Tr ρ = ˆ ˆ d x x ρ x = ∞. 2 1  − = √ ( h −  v ). ˆ ˆ ˆ If we indicate with P1 the diagonalized form of ρ. .
if  ϕ 2 is normalized and where { j } is an arbitrary basis in the Hilbert space of the system.10 Let us write the factorized state as ρ 12 =  ψ ˆ The ﬁrst reduced density matrix is ˆ1 = ψ = ψ = ψ 1 1 ψ  ⊗ ϕ 2 ϕ. It is given again by the sum of the diagonal elements of the matrix (5. with and 1 √ ϕ 2 12 2 1 = √ ( h 2 +  v 1) = ( h 2 +  v 2 ). 5. we must also have ch ch = cv cv = 0. ˆ 5. . this would imply that either ch or ch and either cv or cv be equal to zero. which both describe pure states. The reduced density matrices are therefore given by ˆ1 = 1 ( h 2 1 ˆ 2 = ( h 2 1 2 +  v 1 ) (1 h  + 1 v ).8 It is easy to verify that. ψ⊗ j 2 jϕ jϕ 2 2 ϕ j 2 2 1 ψ⊗ j 2 2 1 ψ⊗ j cj = ψ 1 ψ .7 It is possible to write  ψ 1 12 = ψ 1 1 ⊗  ϕ 2 .7)  12 1 = √ ( h 2 1 1 +  v 1 ) ⊗ √ ( h 2 2 +  v 2 ). in order to obtain the entangled state  12 . However. +  v 2 ) (2 h  + 2 v ). 5.45) when one interchanges the second and third row and the second and the third column. 5.12 The matrix C is given by 1 ˆ C=√ 2 0 1 1 0 .11 The reduced density matrix is ˆ2 = c00 2 + c10 2 ∗ ∗ c00 c01 + c10 c11 ∗ ∗ c00 c01 + c10 c11 2 + c 2 c01  11 . Now. we should have chv = ch cv and could be written as a product cvh = cv ch . which contradicts the the fact that neither chv nor cvh can be zero.9780521869638sol CUP/AUL November 6. 2008 18:21 Page24 t 24 S o l u t i o n s to se l e c te d p ro b l e m s 5.9 The state  12 can be rewritten as (see Prob. 5. 5. if the entangled state  state of  ψ 1 and  ϕ 2 .
where n is the number of dimensions of the system. can be easily understood by considering that the trace of a density matrix is always equal to 1 (see Property (1. 2008 18:21 Page25 t 25 S o l u t i o n s to se l e c te d p ro b l e m s ˆ ˆ and it is clear that C = C † . 2 as expected. we can write β  = n=α cn  vn  wn 1 = √ ( α  α −  β  β ). we choose 1  vα =  wα = √ ( 0 +  1 ). (5. n ˆ j=1 are the dimensions of the system. given these matrices. 2 α α = It is easy to verify that from the two previous equations one obtains again  1 = √ ( 0  1 +  1  0 ). the two unitary matrices are easily derived (they are the matrices whose column vectors are the two vectors  vn and  wn ). Since these are cn . 2 1  β  β = ( 0  0 +  1  1 −  0  1 −  1  0 ). 5. we have that cn = ±1/ 2.51) holds.9780521869638sol CUP/AUL November 6. It is ﬁnally straightforward to verify that. Eq. Let us now compute 1 ˆ ˆ ˆ ˆ C C † = C †C = 2 1 0 0 1 . Its √ eigenvalues are degenerate and both 2 1/2.41a)). 2 1 vβ = wβ = √ ( 0 −  1 ). we are free to choose ˆ ˆ ˆ ˆ the eigenvectors of C C † and C † C. . 2 Therefore. From our calculations. In particular.14 The fact that the density operator corresponding to the center of a (n + 1)– 1 dimensional (hyper)sphere is always given by n Iˆ. which is a multiple of the identity operator. where again. since any vector is eigenvector of the identity. and the fact that any ˆ ˆ density matrix may be written as ρ = n w j P j . that is 1 ˆ ˆ U =U = √ 2 1 1 1 −1 . Morevoer. 2 with 1 ( 0  0 +  1  1 +  0  1 +  1  0 ). if ρ is a mixture.
we have L x . l = k l. we also have ψ  lˆy  ψ ≥ 0 which. Since lˆ− l. similarly. ¯ˆ 6.99). lˆx ± ı ˆ2 . px px − y px . px p y + x p y . px − y . since L x . lˆx ± ı lˆy = lˆz . x − y . x = x p y − y px .97) and (2. x − y px . x ˆˆ ˆ = x p y . px = x p y . lˆ− = lˆx + ı lˆy . lˆy = ı lˆy ± lˆx = ±lˆ± . Ly Ly + Ly Lx. lˆ± = ˆ2 . l = l(l + 1)k l. L z ¯ ˆ ˆ ¯ ˆ ˆ + + . ˆ ˆ ˆˆ ˆ ˆˆ ˆ L z . Moreover. (6. x ˆ ˆ ˆ = ı h y. L y ¯ ˆ ˆ ¯ ˆ ˆ On the other hand. it is straightforward to obtain lˆz . ¯ˆ and. L2 = Lx L2 − L2 Lx ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = Lx Ly Ly − Ly Lx Ly + Ly Lx Ly − Ly Ly Lx ˆ ˆ ˆ ˆ ˆ ˆ = Lx. l − 1 . which proves the desired result. proves the result. ˆ ˆy ˆ ˆy ˆy ˆ Lx . l . px − y px . together with the previous equation. x px − y px .9 Starting from Eq. px = x. we obtain ˆ ˆ ˆˆ ˆˆ ˆ L z . x ˆˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = x.7).6 For any state vector  ψ we have (see also Eq.4 Making use of Properties (2. lˆx ± ı lˆy = ˆ2 . we apply lˆ− to both sides. ˆ ˆ ˆ2 In conclusion L x . 2 Along the same lines.9780521869638sol CUP/AUL November 6. lˆx − ı lˆy = −ı lˆx . L 2 = −ı h L y L z + L z L y = ı h L y . L 2 = 0. L2 = L x . (6. l − 1 . lˆy + ı lˆy . L 2 + L 2 . 6. lˆx = 2lˆz .7 Making use of Eqs. . (2. lˆy = 0. (4. we have ˆ ˆz ˆ ˆ L x . L = 0. Ly ˆ ˆ = ıh Lz L y + L y Lz = ıh Lz.2 Writing L2 = L 2 + L 2 + L 2 . lˆx ± ı lˆz .6) and (6. lˆ+ .30). ˆ2 . 6. x p y + x p y . It is easy to verify that this result holds true also for L y ˆ and L z . 2008 18:21 Page26 t 26 S o l u t i o n s to se l e c te d p ro b l e m s Chapter 6 ˆ ˆ ˆy ˆz ˆ ˆx ˆx ˆy ˆz ˆ ˆ 6. we have lˆ−ˆ2 l. and of Eq. px ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = ı h py . lˆ± = lˆz .63)) 2 ψ  lˆx  ψ = † ψ  lˆx lˆx  ψ ≥ 0.174). proceeding in an analogous way. l l l l 6.
6.32)–(6. from which the third matrix follows. we have lˆ+  1.11 Making use of Eqs.9780521869638sol CUP/AUL November 6.24)). from which the second matrix follows. 2008 18:21 Page27 t 27 S o l u t i o n s to se l e c te d p ro b l e m s Given that lˆ− and ˆ2 commute (see Eq. (6. Finally. φ. we have lˆz  1. lˆz  1. However. −1 = 0. we ﬁnally obtain l ˆ2 k l. 0 = 2  1. For lˆ+ . (6. Let us start from the simple equation (see Eq. for lˆ− we have √ lˆ−  1. 1 =  1. l − 1 . −1 = 2  1. √ lˆ+  1. Successively applying the lowering operator lˆ− as above. 0 . l − 1 = l(l + 1)k l. √ lˆ+  1. −1 = −  1. from which the ﬁrst matrix follows.31) for any m l . θ ) ⎢ 1 sin φ = ⎣ − r sin θ ∂(x. z) cos φ cos θ r sin φ sin θ 1 cos φ r sin θ sin φ cos θ r ⎤ cos θ 0 ⎥. we have ⎡ cos φ sin θ ∂(r . 1 = 0. 1 . ¯ On the other hand.3)) ˆ ˆ ˆ L2 = h 2ˆ2 = r × p ¯ l = r j pk ra pb i jk ˆ ˆ ˆ ˆ 2 = . (6.12 To solve this problem one could use the transformations (6.33). −1 . √ lˆ−  1. (2.33) and the partial derivatives (see Prob. lˆz  1. 1 .134)) ˆ ˆ2 ˆy ˆ2 p2 = px + p 2 + pz = −h 2 ¯ ∂2 ∂2 ∂2 + 2+ 2 ∂x2 ∂y ∂z = −h 2 . ˆ ˆ ˆ ˆ i jk r j pk iab ra pb iab . −1 . 0 . (6. 0 = 0. there is a more elegant and “physical” solution of the problem. 0 = 2  1.11) to transform by brute force the Laplacian = ∂2 ∂2 ∂2 + 2+ 2 ∂x2 ∂y ∂z from Cartesian to spherical coordinates and the desired result then would follow straightforwardly. 1 = 2  1. lˆ−  1. 6.10 In the case of lˆz . 6. we obtain Eq. l which proves the desired result. ⎦ sin θ − r 6. y. we also have (in the following a summation over repeated indices is understood) (see Eq.
∂z r ∂r ∂ y ∂ = .9780521869638sol CUP/AUL November 6. ¯ˆ ˆ h ¯ ı x ∂ ∂ ∂ +y +z . ∂x r ∂r Therefore. 2008 18:21 Page28 t 28 S o l u t i o n s to se l e c te d p ro b l e m s Since i jk iab = δ ja δkb − δ jb δka . ¯ˆ ˆ ˆˆ ˆˆ Since p r = r p − 3ı h . ı ∂r ˆ − ı h r · p + L2 ¯ˆ ˆ = h2 − ¯ ˆ2 l 2 ∂ ∂2 + 2 . ˆ ˆ r· p= and. r2 Together with the ﬁrst equation. r j ˆ ˆ ˆ ˆ ˆ pk − pk r j + r j . pk ˆ ˆ ˆ ˆ ˆ rk p j ˆ ˆ ˆ ˆ ˆ = r j r j pk − ı h δ jk pk − pk r j + ı h δ jk rk p j ˆ ˆ ˆ ˆ ˆ ¯ ¯ = r j r j pk pk − ı h r j p j − pk r k r j p j − ı h r k pk ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ¯ˆ ˆ ¯ˆ ˆ ˆ ˆ ˆ ˆ = r2 p2 − ı h r · p − p · r ¯ˆ ˆ ˆ ˆ r · p − ı h r · p. from r = (x 2 + y 2 + z 2 ) 2 . i. x ∂ ∂ = . − 2 r ∂r ∂r r ˆ2 l r2 − ˆ2 l . ˆ p2 = . ∂x ∂y ∂z ∂ z ∂ = . ∂y r ∂r h ∂ ¯ r .e. we ﬁnally obtain 2 ∂ ∂2 − + 2 r ∂r ∂r ∂ 1 ∂ r2 = 2 ∂r r ∂r = which proves the desired result. ˆ ˆ r·p= From the previous equations. The ﬁrst term. we then have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ L2 = r j pk ra pb δ ja δkb − r j pk ra pb δ jb δka = r j pk r j pk − r j pk r k p j ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = r j r j pk + pk . we obtain ˆ p2 = 1 r2 ˆ ˆ r·p 2 1 2 + ı h r · p. It is interesting to note that the equation 1 2 ˆ ˆ ˆ r · p − ı h r · p + L2 ¯ˆ ˆ r2 bears an important physical meaning: dividing both sides of the equation by 2m. we have that the total energy of a pointlike free quantum particle in the threeˆ dimensional case (p2 /2m) may be interpreted as the sum of three terms. we have ¯ ˆ ˆ ˆ ˆ ˆ L2 = r2 p2 − 2ı h r · p − r · p − 3ı h r · p ¯ˆ ˆ ¯ ˆ ˆ ˆ ˆ ˆ ˆ = r2 p2 − r · p On the other hand.
¯ 8 See [Gradstein/Ryshik 1981. On the other hand.3: p. θ ) = δll . i.e. 2 (2l) ! ! . since dφ 0 eı (l−l )φ = δll . θ )Yll (φ. ı hr · p ¯ˆ ˆ . 2008 18:21 Page29 t 29 S o l u t i o n s to se l e c te d p ro b l e m s ˆ ˆ r·p . 6. The lhs of the previous equation turns out to be 2π 2 dφ 0 eı (l−l )φ 2π 2π π 0 dθ sin θ ∗ ll (θ ) ll (θ ). 2π ∗ ll (θ ) we must also have Ill = 0 π dθ sin θ ll (θ ) = 1.14 We start from Eq. (2l + 1) ! ! 7 This term vanishes in the classical limit h → 0 (see Pr. 2mr 2 is a typical quantummechanical term that arises from the commutators between position and momentum in the derivation above. (2l + 1) ! ! then Ill = 2 N 2 from which it follows that N  = 1 (2l + 1) ! ! = l 2 (2l) ! ! 2 l! (2l + 1) ! . d Yl∗l (φ. for m = l and m = l . 2mr 2 is the angular part of the total energy. 72). (6. − ˆ L2 .59) and write. The last term. The second term. Ill = N  Since8 π 2 2 0 π dθ (sin θ ) 2l+1 = 2 N  2 0 π 2 dθ (sin θ )2l+1 .9780521869638sol CUP/AUL November 6.621.4]. Now. 3.7 and is the other element of the radial part. . 2mr 2 represents an element of the radial part of the energy. 2. dθ (sin θ )2l+1 = 0 (2l) ! ! .
(6. σ y = ˆ ˆ = 2ı = 2ı σz . Concerning lˆz . and ↓ x = c d .55)). This completes the proof. r2 − z 2 = 0. . y px ˆ2 ˆ ˆ ˆy ˆ ˆ ˆ2 ˆ ˆ ˆy ˆ ˆ = −2ı h px p y + 2ı h p y px = 0. ˆ2 ˆy ˆ2 ˆ lˆz . (6. ¯ˆ ˆ ¯ˆ ˆ where we have made use of the result of Prob.2c)).9780521869638sol CUP/AUL November 6. p2 − pz = 0. ˆ ˆ ˆ 6. and that completes the argument. Moreover. ˆ ˆ 6.16 A spherically symmetric Hamiltonian is of the type (6. 6. ˆ which gives the eigenvalues λ1. x p y − y px ˆ ˆ ˆ ˆˆ ˆˆ ˆ ˆˆ ˆ ˆ ˆ ˆ ˆ ˆ = − x 2 . y px = px . lˆz = px + p 2 + pz . px + x y 2 . In turn. If one bears in mind the explicit expression of the Laplacian in spherical coordinates (Eq. The ˆ ˆ characteristic equation for σx is ˆ det σx − λ Iˆ = λ2 − 1 = 0. it is clear that ˆ l l p2 .20 We explicitly derive the commutation relation σx . (6. 2. ˆ 6. On the other hand. We have σx . ˆ ˆ ˆ since lˆz commutes with both pz and z (see Eq. ˆ2 = V (r ). ˆ2 = 0. p y ˆ ˆˆ = −2ı h y x + 2ı h x y = 0. if it is true for l. x p y − p 2 . x 2 + y 2 = lˆz . and.26.17 We know from Prob. x p y − p 2 . ˆ r2 .255)). 2008 18:21 Page30 t 30 S o l u t i o n s to se l e c te d p ro b l e m s The latter result may be proved by induction (it is trivially true for l = 1.22 We prove the result for σx and leave to the reader the calculation involving σ y .84). px + p 2 = lˆz . 2. it is also true for l + 1).2 = ±1. this means ˆ lˆz .23. σ y = 2ı σz and leave the remainˆ ˆ ing to the reader.16 that lˆz commutes with p2 and r2 . ¯ˆˆ ¯ˆˆ where we have made use of the result of Prob. we have ˆ p2 . y px + y 2 . x p y = − y x 2 . ˆ2 is l the angular part of the Laplacian (see Eq. lˆz = x 2 + y 2 + z 2 . 6. x p y − y px ˆ2 ˆy ˆ2 ˆ ˆ ˆˆ = px . We write the eigenvectors as ↑ x 0 1 1 0 1 0 0 −1 0 −ı ı 0 − 0 −ı ı 0 0 1 1 0 = a b .
6. that is. we need to use Eqs. so that we ﬁnally arrive at the normalized eigenvectors ↑ ↓ x x 1 =√ 2 1 =√ 2 1 =√ 2 1 1 =√ −1 2 1 1 1 0 1 0 + − 0 1 0 1 1 = √ ↑ z + ↓ z .23 It is very easy to show that σz =  ↑ ˆ z ↑ − ↓ z ↓ . z ↓ − ↑ z − ı ↓ z z ↑ + ı z ↓ A comparison with the matricial expressions of the vectors  ↑ z and  ↓ z gives immediately the Pauli matrices. (6. and μ ˜ μ=− ˜ eh ¯ . σk ˆ ˆ 2 2 f j fk + f j fk + Iˆδ jk + Iˆδ jk f j f k ˆ = ı f × f σ + f · f Iˆ.24 That Pauli matrices are Hermitian can be immediatly recognized by inspection. 2 6. 2008 18:21 Page31 t 31 S o l u t i o n s to se l e c te d p ro b l e m s The conditions (6.169) and comments). z − ↓ z z ↑ −z ↓ and σy = ˆ 1 ↑ y ↑ − ↓ y ↓ 2 1 ↑ z + ı ↓ z z ↑ − ı = 2 = ı − ↑ z ↓ + ↑ z ↓ . 2 1 = √ ↑ z − ↓ z .159). σk + ˆ ˆ σ j . For calculating σx and σ y .158a) imply a = b and d = −c.28 For the electron we have g (Eq.27 It is sufﬁcient to consider the form 1 2m px + e By c 2 − ı p2 h2 ∂ 2 ¯ ¯ ˜ + z − μsz B − E e h ( px x+ pz z) ϕ(y) = 0. ˆ ˆ σx = ˆ 1 ↑ x ↑ − ↓ x ↓ 2 1 ↑ z + ↓ z z ↑ +z ↓ − ↑ = 2 = ↑ z ↓ − ↑ z ↓ . (6. This property immediately shows that σ j = Iˆ for j = x.173)) becomes −2 (see Eq. y. mc . ˆ jkn f j f k σn 6. 2m ∂ y 2 2m 6.9780521869638sol CUP/AUL November 6. z. (6.25 We proceed as follows: ˆ ˆ ˆ ˆ σ · f σ · f = σ j σk f j f k = = ı =ı ˆ jkn σn 1 1 σ j . ˆ2 6.
that proves the requested result.31 We prove the results for the states (6. = 2.2 .z = 1 ↑ 2 k.9780521869638sol CUP/AUL November 6. 6. 0 ˆ s  1. ˆ ˆ ˆ ˆ 2 which can be derived by explicit substitution of s1± .193c) and (6.97). 0 ˆ sz  0. We ﬁnally obtain ϕn (y) = mω B πh ¯ 1 4 2− 2 (n! )− 2 e− n 1 (y−y 0 )2 mω B 2h ¯ Hn (y − y0 ) mω B . the energy eigenvalues (6.149) ˆ ˆ ˆ2 ˆ2 2 above onto the two desired states.190).z .30 The ﬁrst part is trivial. ˆ and (6. 2008 18:21 Page32 t 32 S o l u t i o n s to se l e c te d p ro b l e m s Therefore. Since sz = s1z + s2z and ˆ sk z  ↑ ˆ we have sz  1. we ﬁrst notice that ˆ1 ˆ2 ˆ s ˆ s2 = s2 + s2 + 2ˆ1 · s2 3 s ˆ = + 2ˆ1z s2z + s1+ s2− + s1− s2+ . By direct application of the operator s we obtain ˆ s2  0. and s2 from Eqs.194). ¯ 2 2m 6. + z . = 0. 0 2 12 12 = 0. 6.34 We know that ( φ)2 = φ 2 − φ +π 2 ⎡ +π dφψ (φ)φ ψ(φ) − ⎣ ∗ 2 ∗ −π ⎤2 dφψ (φ)φψ(φ)⎦ .z 1 = − ↓ 2 k. For the second part notice that lˆz does not commute with ˆ1. Concerning the total spin. s1 .29 Given the harmonicoscillator character of the Schrödinger equation (6. s2± . The derivation for the other ˆ ˆ two cases is straightforward. h ¯ l 6.176). (6.179) may be rewritten as E = n+ p2 1 + sz h ω B . we may take full advantage of the eigenfunctions (4. 12 1 =√ 2 1 =√ 2 1 1 ↑ − 2 2 1 1 ↑ − 2 2 1 + − + 2 1 1 ↓ 2 − − + 2 1 ↓ 2 1 ↓ 2 1 ↓ 2 1 ↑ 2 1 ↑ 2 = 0. = −π .z and sk z  ↓ ˆ k. with a suitable change of notation. 0 ˆ 12 k.
when we suppose to obtain the outcome h m j in a measurement of Jz . 18–19]. we have +π +π +π f (η) = −π dξ ψ (ξ )ξ ψ(ξ ) + η 2 2 ∗ 2 −π dξ ψ (ξ )ξ ψ(ξ ) − 2η −π ∗ dξ ψ ∗ (ξ )ξ ψ(ξ ) = ξ − 2η ξ + η . Stated in other terms. the ¯ formula ψ Jx · ψ Jy ≥ 1 2 ˆ ˆ ψ  Jx . we have (see Eqs. 2 2 This result. − jx ˆ 2 ψ = jx ˆ2 ψ and similarly 2j ψ ˆy = jy ˆ2 ψ . . we derive for the uncertainty product of x and y components of the angular momenˆ tum.35 From the equation ψO · ψO ≥ 1 2 ˆ ˆ ψ  O. Therefore. f (η) represents a parabola as a function of η and its minimum value (corresponding to the vertex of the parabola) is obtained for η = ξ and is equal to ξ 2 − ξ 2 .9780521869638sol CUP/AUL November 6. since the zdirection is arbitrary.184)) 2 ˆ ψ jx ˆ ˆ = ψ jx − ψ jx ψ = jx ˆ2 ψ 2 ψ . we obtain 2 ˆ ψ jx + 2 ˆ ψ jy = j 2 + j − m2. ˆ ˆ It is also possible to derive a ﬁner estimate of the Jx and Jy uncertainties. 2 As a consequence. together with the fact that the maximum value of m is j while the length √ of the jvector is j( j + 1). 2008 18:21 Page33 t 33 S o l u t i o n s to se l e c te d p ro b l e m s With the change of variable φ + η = ξ . the orientation of the angular momentum is always intrinsically uncertain. forces us to conclude that the angular momentum vector can never point exactly in the zdirection. we have = jx ˆ2 = + jy ˆ2 + + jz2 ˆ 2 ˆ j2 ψ ψ ψ ψ 2 ˆ ψ jx 2 ˆ ψ jy + mj . 9 See [Edmonds 1957. J y  ψ = h h2 ¯ ¯ ˆ ψ Jz ψ = m j . (2. ˆ Since j 2 ψ = j( j + 1). 6. j 2j ψ ˆx In the case in which m j = j we have that + 2j ψ ˆy = j. O  ψ .9 In fact. that is to ( φ)2 .
9780521869638sol CUP/AUL November 6.1 Multiplying Eq.1) in the previous equation. 1. r(2) . i. 7. θ = π ) is transformed into an equator state (θ = π/2). η (2) (3) (rα rβ rγ ) is the wave function given by the scalar product (r(1) rβ r(3) ) = r(1) rβ r(3)  α γ α γ (2) (2) . Substituting E = K B T (see Subsec. = 1 2 2 8.190) states that px x On the other hand. respectively. we ﬁnally obtain x h ¯ .3 It is impossible: it would be a violation of Pauli exclusion principle.4). φ = π/2.6 The position–momentum uncertainty relation (2. We may then conclude that the considered transformation performs a π/2 rotation on the polar angle of the Poincaré sphere. (7. ¯ √ 2m E.  ϕ = cos .5. px 2 px h. r(3) ) = UP η δ ζ where (1) (2) (3) (1) (2) (3) .14) on the left by rδ rη rζ ˆ 123 (r(1) . 2 2 where θ and φ are the polar and azimuthal angles of the Poincaré sphere. 2008 18:21 Page34 t 34 S o l u t i o n s to se l e c te d p ro b l e m s Chapter 7 7. √ 2mkB T which is the socalled thermal wavelength. The input state  1 corresponds to the north pole. 2 ˆ The mean value of P calculated on the input state is given by 1 1 ˆ 1 P 1 = 1 ( 1 − ı  2 ) ( 1  + ı 2 ) − (−ı  1 +  2 ) (ı 1  + 2 )  1 2 2 1 1 1 − 1 (−ı)(ı) 1 = 0.e. to θ = 0 (φ is not deﬁned).3 A generic twodimensional state  ϕ may be expanded as θ θ  1 + eıφ sin  2 . In a similar way. Chapter 8 ˆ 8. 7. we can show that the input state  2 (south pole. On the contrary.2 The mean value of P calculated on the output state is given by 1 ˆ ψ P ψ = ( 1  − ı 2 ) [ 1 1  −  2 2 ] ( 1 − ı  2 ) 2 1 = (1 + ıı) = 0. it follows that (r(1) rζ rδ ). Let us consider the transformation (8. the output state ψ lies on the equator of the Poincaré sphere and corresponds to θ = π/2.
After some time we shall ı ˆ ˆ ¯ → ψ (t) = e− h H t ψ . if ρ0 = ρ0 .11)). taking into account the third Pauli’s matrix (6. . ˆ 8. that is. O U † ˆ ˆ ¯ ˆ ˆ = U U † ı h = O. 2008 18:21 Page35 t 35 S o l u t i o n s to se l e c te d p ro b l e m s ˆ 8. and. if U (a) represents a continuous transformation. on the other hand. then we also have ˆ2 ˆ ˆ ρt2 = ρt . Then. the density matrix ρ 0 is transformed into ˆ ˆ ˆ ˆ ρ t = Ut ρ 0 Ut† .5.4 Consider a transformation U such that ψ have  ψ →  ψ(t) = e ı −h ¯ ˆ = U  ψ .6 It is evident that. 1 + σzS = ˆ 2 0 0 0 .154). If the symmetry under the transformation U must be conˆ  ψ(t) . O = ı h and that O and O are unitarily ¯ transformed. Chapter 9 ˆ 9. the operator U must commute with the Hamiltonian. ˆ such that ˆ ˆ ˆ ˆ ˆ ˆ ρt2 = Ut ρ0 Ut† Ut ρ0 Ut† ˆ ˆ ˆ2 ˆ = Ut ρ0 Ut† .9780521869638sol CUP/AUL November 6. 122). for the symmetry to be conserved. However. ı ı ˆ ˆ In conclusion. 3. Therefore. Now.5 The hypothesis is that we have O. we have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ U O U †U O U † − U O U †U O U † = U O O U † − U O O U † ˆ ˆ ˆ ˆ = U O. then ψ (t) must be also equal to U ˆ Ht  ψ and ψ we obtain on one hand ˆ ¯ ψ (t) = e− h H t U  ψ . a/2 exists and ˆ (a) = U (a/2)U (a/2). ˆ ˆ ˆ ˆ 8. so that. H being the ˆ Hamiltonian of the system. Combining the previous equations served. O . ˆ ¯ ˆ ψ (t) = Ue− h H t  ψ . U (a/2) can be either unitary or antiunitary. You may have recognized that we have already solved this problem in Subsec.1 After a unitary timeevolution Ut .1. ˆ ˆ ˆ U the square of both a unitary operator and an antiunitary operator must be a unitary operˆ ator (see Properties (8.2 The part of HSM related to the system is already diagonal. U (a) is unitary (see also the Stone theorem: p. ˆ ˆ 9.
respectively) 1 1 + σz  ↑ =  ↑ . λ1. 6. 1 ↓ M = √ x 2 1 −1 .7 (i) In order to diagonalize the Hamiltonian it sufﬁces to substitute the deﬁnitions (9. (4. ˆ √ a0  N = N  N − 1 .147). 2.2 = ±1. ˆ 2 a1  0 = 0.85). . In this basis it is apparent that the offdiagonal terms of the density matrix are exactly zero.61).e. we have 1 1 . 2 In the limit of a perfect overlap between ψ↑ (t) ψ↑ (t)  ψ↓ (t) = 1.5 We have 1  ϕ(t) = √ ψ↑ (t)  ↑ z + ψ↓ (t)  ↓ z . where  ψ(t) = ψ↑ (t) = ψ↓ (t) .215) into Eq. i. and it follows that ρ(t) = ˆ 1 ψ(t) 2 ψ(t)  1 1 and ψ↓ (t) .2). 0 1 1 0 . (9.87). In other words. the action of σx on its eigenkets  ↑ M . and the ﬁrst of Eq. 9. 2008 18:21 Page36 t 36 S o l u t i o n s to se l e c te d p ro b l e m s Concerning σx . (6.22) 1 ↑ M = √ x 2 1 1 . 5. 1. ˆM ˆM x x x x It follows than that (see Eqs. we calculate the determinant of the ﬁrst Pauli’s matrix ˆM σx = ˆM Then det σx − λ Iˆ = ˆM −λ 1 1 −λ = λ2 − 1. (4.6 By making use of the relations (see Eq. (6. 9.158a) and Prob. 9.2). so as to obtain 1 εSA ˆ ˆ HSA (b) = √ 1 + σz ˆ 2 N ˆ† ˆ ˆ† ˆ b0 b0 − b1 b1 .9780521869638sol CUP/AUL November 6. its eigenstates form a basis in the Hilbert space (see Th. σx  ↓ M = −  ↓ M .3 Since a density matrix is a Hermitian operator (see Sec. 9. ˆ we derive ˆ ↑  N HSA N − 1  ↑ = εSA which yields εSA √ N τ0 √ N.  ↓ M is ˆM x x σx  ↑ M =  ↑ M .
i. εSA  ↑ = √ (2ν − N )  ν N b ↑ . it is necessary to transform back to the physical states  a0 .67) and making use of the above eigenvalue ˆ ˆ equation for HSA (b). (9. In order to obtain the ﬁnal state (9. ˆ0 2 and a† ˆ0 N N 0 . the initial state can be written as  SA (t0 ) = = = 1 √ 2 1 √ 2 1 √ 2 N N ν=0 N N ν=0 N N ν=0 1 √ N! 1 √ N! (−1) N −ν N ν N ν √ ˆ† b0 ν ˆ† −b1 √ N −ν 0 S S (−1) N −ν √ ν! (N − ν)!  ν S .68).  a1 . The transformation is standard but rather cumbersome.e. (ii) Taking advantage of the expression 1 N = √ a† ˆ0 N! where 1 ˆ† ˆ† a† = √ b0 − b1 . 2008 18:21 Page37 t 37 S o l u t i o n s to se l e c te d p ro b l e m s ˆ ˆ The eigenstates of HSA (b) will be given by ν with ˆ ˆ HSA (b)  ν b b 1 1 ˆ† ↑ = √ √ b0 ν! (N − ν) ! ν ˆ† b1 N −ν 0 ↑ . and requires some relabeling. . b N! ν ν! (N − ν) ! b Inserting this expression into Eq. = = 1 √ 2 1 √ 2 N ˆ† ˆ† b0 − b1 N ν N N N ν=0 ˆ† b0 ν ˆ† −b1 N −ν . from  ν b to  n .9780521869638sol CUP/AUL November 6. we obtain  SA (t) = c+ 1 √ 2 N N (−1) ν=0 N −ν √ εSA N! − ı √ (2ν−N )t ¯ ν e h N ν!N − ν) ! ( √ b ↑ + c−  N  ↓ .
the state has become  out 2 = ı cos π π  1 − sin  2 . ˆ Since ρ is a density operator and therefore ρ = ρ † .9780521869638sol CUP/AUL November 6. 3. It follows that.8 Stirling’s formula states that. for large m. where. we have assumed R and T to be real. 9. after two beam splitters. that is transmitted at the ﬁrst and at each subsequent beam splitter. Collecting these results. The subsequent two mirrors and beam splitter induce the transformations T  1 → ıT (T  1 + ıR  2 ). and by  2 the state that is reﬂected at the ﬁrst beam splitter and then transmitted at each subsequent beam splitter. 2 2 9. we have T  1 + ıR  2 → ı T2 − R2  1 − 2RT  2 .9 With reference to the above ﬁgure. N N where R = cos π/2N and T = sin π/2N . ln(m! ) m ln m − m + 1 ln m. we label by  1 the input state. Using the generic mapping derived in the solution of Prob. In order to verify the second condition we use the cyclic property of ˆ ˆ the trace and the fact that P j2 = P j to obtain . we shall have  out N ∝ cos π π  1 + sin  2 =  2 . for the sake of simplicity. 2008 18:21 Page38 t 38 S o l u t i o n s to se l e c te d p ro b l e m s 1> 1> 2> 2> 1> 1> 2> 2> 1> 1> 2> 2> 9. (9. we may write the transformation at the ﬁrst beam splitter as  1 → T  1 + ıR  2 . ıR  2 → −R (ıR  1 + T  2 ). one obtains ℘ n 1. after N beam splitters (with N even). the ﬁrst condition is immediˆ ˆ ˆ ately veriﬁed.10 In order to prove that ρ j is a density operator it is sufﬁcient to show that ρ j = ρ † ˆ ˆ ˆj and Tr(ρ j ) = 1.12. and substituting the resulting approximation m! m m e−m into Eq. This means that. 2 Taking only the ﬁrst two (dominant) terms of this expansion.73).
according to Eq. ˆ Tr ρT ∗ O ˆ = Tr ρ ˆ k ˆ† ˆ ˆ ϑk O ϑk ˆ† ˆ ˆ ˆ ϑk O ϑk ρ = Tr k = k ˆ† ˆ ˆ ˆ Tr ϑk O ϑk ρ ˆ ˆ ˆ ˆ† Tr O ϑk ρ ϑk k = = k ˆ ˆ ˆ† ˆ Tr ϑk ρ ϑk O ˆ ˆ ˆ† ˆ ϑk ρ ϑk O. by the cyclic property of the trace. this implies that ϑ † (xm )ϑ(xm ) = ˆ ˆ E(xm ). (9. ˆ ˆ ℘(xm ) from which. ˆ ˆ ˆ ˆ Since the previous equation must hold for any ρ i .4). ˆ ˆ Similarly. it further follows that ˆ ˆ ˆ ℘(xm ) = Tr ϑ † (xm )ϑ(xm )ρi .122) and from the fact that Tr[ρ f ] = 1 it follows that ˆ 1 ˆ Tr ϑ(xm )ρi ϑ † (xm ) = 1.13 From Eq. we obtain Tr T ρ i ˆ ˆ ˆ ˆ ˆ ˆ ˆ = Tr Pv ρ i Pv = Tr Pv ρ i = v ρ i v = sin2 θ . (9.12 The problem can be solved by making repeated use of the cyclic property of the trace and the linearity of the trace operation. using the cyclic property of the trace (see Prob. 2008 18:21 Page39 t 39 S o l u t i o n s to se l e c te d p ro b l e m s Tr ρ j = ˆ ˆ ˆˆ Tr P j ρ P j Tr ρ P j ˆˆ = Tr ρ P j2 ˆˆ Tr ρ P j ˆˆ = Tr ρ P j ˆˆ Tr ρ P j ˆˆ = 1.120) ˆ ℘(xm ) = Tr E(xm )ρi .9780521869638sol CUP/AUL November 6. ˆ2 ˆ where we have used the obvious fact that Pv = Pv (see Eq (1. that is. 9. ˆ and therefore we have ˆ ˆ ˆ Tr ϑ † (xm )ϑ(xm )ρi = Tr E(xm )ρi . .41b)). 5. using Eq (9. On the other hand. 9.11 In this case we have.110). T ρ i =  v v  ρ i  v v  = sin2 θ  v v . k = Tr 9. The desired result immediately follows from the latter two equations.
we ﬁnally ˆ obtain +∞ +∞ ˆ d xm ℘(xm ) xS ρ (xm ) xS = −∞ −∞ ˆ d xm ℘(xm xS ) xS ρ i xS ˆ = xS ρ i xS .14 Using expressions (9. 2 ℘1 = . after BS2. = (η + 1) − 4 √ 2 η cos φ 1 ℘2 = (η + 1) + . the mirror M1. 9. where  y is an arbitrary eigenstate of x. becomes M2. 4 2 1−η ℘3 = .M1. 2008 18:21 Page40 t 40 S o l u t i o n s to se l e c te d p ro b l e m s 9.127b).15 The ﬁrst beam splitter. after the second mirror and BS3.9780521869638sol CUP/AUL November 6. e e η + 1 2 − η − 1 1 − = √ 2 2 2 The detection probabilities can then be calulated as follows: 1 ıφ √ √ e η − 1 e−ıφ η − 1 4 √ η cos φ 1 .125) and (9. 2 which.BS3 → 1 √ η 1 + ı 1 − η 3 √ ıeıφ 2 − 2 . we have +∞ +∞ ˆ d xm ℘(xm ) xS ρ (xm ) xS = −∞ d xm ℘(xm ) xS  −∞ +∞ +∞ 1 1 ˆ1 ˆ ˆ E 2 (xm )ρ i E 2 (xm )  xS ℘(xm ) = −∞ d xm xS  −∞ +∞ d xS ℘(xm xS ) 1 2 1 2 xS xS  × ρi ˆ −∞ d xS ℘(xm xS ) xS xS  xS . Given that xS  y = δ(xS − y). The ﬁnal state.PS → 1 √ eıφ  1 −  2 . and the phase shifter induce the following transformation on the initial state  1 : 1 BS1. can be written as 1 1 √ 1  f = √ ıeıφ η √ ( 1 + ı  2 ) − eıφ 1 − η  3 − √ (ı  1 +  2 ) 2 2 2 √ eıφ 1 − η 1 ıφ √ ı ıφ √ 3 .
2 √ η 1 4 (1 − η) − ı√2 sin φ η 1 4 (1 + η) − 2 cos φ 1 0 √ η 1 4 (1 + η) + √ cos φ 2 η 1 4 (1 − η) + ı 2 sin φ 1 = (η + 1) + 4 and √ 1 0 η cos φ .18 It is easy to show that 1 ˆP ˆ ˆP √ ˆV ˆV P + η Pu − η P2 − P1 E1 = 2 d √ = η 1 4 (1 + η) − √ cos φ 2 η 1 4 (1 − η) − ı 2 sin φ √ η 1 (1 − η) + ı√2 sin φ 4 η 1 4 (1 + η) + 2 cos φ . whereas. 2 2 1 2 φ ℘2 = (1 + cos φ) = cos . 1−η ˆP ˆ E 3 = (1 − η) Pu = 2 1 −1 −1 1 . when η = 0. 4 1 ℘3 = . we need to calculate the three expectations for the three detectors. Now. 2 9. that is ℘1 = 1 0 √ η 1 4 (1 + η) − √ cos φ 2 η 1 4 (1 − η) − ı 2 sin φ √ η 1 4 (1 − η) + ı√2 sin φ η 1 4 (1 + η) + 2 cos φ 1 = (η + 1) − 4 ℘2 = √ 1 0 η cos φ . 2008 18:21 Page41 t 41 S o l u t i o n s to se l e c te d p ro b l e m s In the case in which η = 1. we obtain ℘1 = ℘ 2 = 1 .9780521869638sol CUP/AUL November 6. 1 ˆP ˆP √ ˆV ˆV ˆ Pd + η Pu + η P2 − P1 E2 = 2 √ = η 1 4 (1 + η) + √ cos φ 2 η 1 4 (1 − η) + ı 2 sin φ √ η 1 4 (1 − η) − ı√2 sin φ η 1 4 (1 + η) − 2 cos φ . 2 2 ℘3 = 0. . we have ℘1 = φ 1 (1 − cos φ) = sin2 . 2 which ﬁt with the previous calculations. 2 1−η 2 1 −1 −1 1 1 0 ℘3 = = 1 0 1−η .
Then. at each data point D. These probability functions are evidently subject to the conditions N 0 ≤ ℘H j (D) ≤ 1. This is the required solution.137) of the involved projectors shows. . 1991]. In practice. The proof of the noncommutability of the above effects is more cumbersome.90)). we choose ℘Hk (D) = 1. E2 = √ P1 − P2 Pd − √ Pd P1 − P2 4 2 2 1 ˆP ˆV 1 V ˆV ˆ ˆV ˆP − √ Pu P1 − P2 + √ P1 − P2 Pu 2 2 1 2 2 ˆP V P ˆV ˆ ˆ = √ 2 P1 − Iˆ Pd − √ Pd 2 P1 − Iˆ 4 2 2 1 1 ˆP V ˆ ˆV ˆP − √ Pu 2 P1 − Iˆ + √ 2 P1 − Iˆ Pu 2 2 1 V ˆP P ˆV ˆ ˆ = √ P1 Pd − Pd P1 . does not vanish (see also Eq. 2 2 which.19 The proof of completeness is straightforward.133) and (9. 1 ˆP ˆ ˆ ˆP √ ˆV ˆV ˆ Pd + η Pu − η P2 − P1 E1 + E2 + E3 = 2 1 ˆP ˆP √ ˆV ˆV Pd + η Pu + η P2 − P1 + 2 ˆP ˆP ˆP ˆP = Pd + η Pu + Pu − η Pu = Iˆ.9780521869638sol CUP/AUL November 6. 9. of the hypothesis for which the risk R j (D) is smallest.20 See the original paper [de Muynck et al. 9. Indeed. This procedure corresponds to the choice. j=1 ℘H j (D) = 1. The formal reason ˆV ˆP for this result lies in the fact that projectors P1 and Pu belong to different sets. (2. at each point D for which Rk (D) < R j (D) (∀ j = k). We ˆ ˆ here only consider the noncommutability between E 1 and E 2 .21 We are looking for the N functions ℘H j (D) which make the quantity N ˆP + (1 − η) Pu C = dD n R j (D)℘H j (D) j=1 as small as possible. In order to simplify the proof. 2008 18:21 Page42 t 42 S o l u t i o n s to se l e c te d p ro b l e m s 9. 1 2 2 ˆP ˆV ˆ ˆ ˆV ˆV ˆP ˆV E1. we take η = 1/2. ℘H j (D) = 0. It is clear that the average cost will be minimum if the integrand in the rhs of the ﬁrst equation is chosen to be as small as possible for each data point D ∈ IRn . as a direct calculation with the explicit expressions (9.
we have k=1 dSck 2 = 1 n n dS S2n k=1 S2n ck 2 = A2n .24 In view of the symmetry of the expression (9. in this case the optimal POVM is given by a PVM. and we have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ R0 − L = R0 − R0 E H0 − R1 E H1 = (R0 − R1 ) E H1 . we may also introduce the function L(D) = min R j (D). Eq. 9. Finally. by summing the ﬁrst of the latter two equations over j. for any state  ϕ . j so that we may rewrite our solution in the form R j (D) − L(D) ℘H j (D) = 0. (9. we again deal with projectors ˆ ˆ because E H1 and E H0 must be orthogonal. 9. when it is integrated over the hypersphere the terms with k = j vanish. the function L(D) may be rewritten as N L(D) = j=1 R j (D)℘H j (D). R j (D) − L(D) ≥ 0. .178a) now becomes ˆ ˆ ˆ ˆ R0 − R1 E H1 E H0 = 0. detection operators commute. note that. since n ck 2 = 1. Moreover.25 We have ˆ ˆ ϕ R j − L ϕ = A−1 ϕ Iˆ − ρ ϕ 2n = A−1 ϕ  ϕ − ϕ ρ ϕ ≥ 0. ˆ ˆ Since in general the operator R0 − R1 does not vanish. n which proves the result. so that the minimum average cost can be simply expressed as C min = d DL(D). (9.208). The Lagrange operator (see Eq. IRn ˆ ˆ 9.23 In case of binary decision. In conclusion.177)) is given by ˆ ˆ ˆ ˆ ˆ L = R0 E H0 + R1 E H1 . 2008 18:21 Page43 t 43 S o l u t i o n s to se l e c te d p ro b l e m s In order to render the solution more compact. ˆ 2n because. and to facilitate the comparison with the quantum case. ˆ ϕ ρ ϕ =  ϕ  ψ 2 ≤ ϕ  ϕ .9780521869638sol CUP/AUL November 6. since E H1 + E H0 = Iˆ.
n n (1) ˆ cn ψl H ψn − dl E (2) = 0. (10.1 If the two values of E 1 in Eq.9c) by ψq and ψl . (10. (10. (4. Substitution of cn from Eq.30) into the previous equations yields ⎛ ⎞ 2 ˆ ˆ ˆ ψq H ψn ψq H ψn ψn H ψl ⎜ ⎟ − E (2) ⎠ dq + dl = 0. where we have made use of the commutation relation a a† = 1 + a† a in order to ˆˆ ˆ ˆ write the annihilation operators to the right of the creation operators. (0) − E (0) (0) − E (0) E E n n n n Making use of the argument employed after Eq. we obtain (1) ˆ cn ψq H ψn − dq E (2) = 0. A generalization to higher orders and to the cases in which the “ground” state is more than doubly degenerate is straightforward. ⎝ (0) (0) E (0) − E n E (0) − E n n n ⎞ ⎛ 2 ˆ ˆ ˆ ψl H ψn ψl H ψn ψn H ψq ⎟ ⎜ dq + ⎝ − E (2) ⎠ dl = 0 . where the prime on the summation symbol denotes omission of the two terms n = q (1) and n = l.27). we obtain x= ˆ and x4 = ˆ h ¯ 2mω + a† ˆ 2 h ¯ a† + a .27) are equal. 2008 18:21 Page44 t 44 S o l u t i o n s to se l e c te d p ro b l e m s Chapter 10 10. (10. . we must go to the second order for removing the degeneracy.10 From the 10 This way of ordering terms involving annihilation and creation operators is called normal ordering.9780521869638sol CUP/AUL November 6.73). 10. the degeneracy is removed in second order. ˆ ˆ 2mω a 4 + 4ˆ † a 3 + 6a 2 + 6 a† ˆ a ˆ ˆ ˆ 4 2 a 2 + 12ˆ † a + 4 a† ˆ a ˆ ˆ 3 a + 6 a† ˆ ˆ 2 +3 . Unless both of these conditions are satisﬁed. the analogues of Eqs.2 From Eqs. By multiplying from the left both sides of Eq.28) are ˆ ψq H ψn n 2 E (0) − E n (0) (0) = n ˆ ψl H ψn E (0) − E n (0) 2 ˆ ˆ ψq H ψn ψn H ψl n E (0) − E n = 0. (10.
2 (n + 1)(n + 2)(n + 3)(n + 4). we have (1) ck (t ≥ t∞ ) = ˆ 2 ψk Hi ψl ıh ¯ ˆ ψk Hi ψl ı ψk ˆ Hi ψl ıh ¯ eı xt − 1 2 t −∞ dt eı(ωkl t sin ωt dt eı(ωkl +ω)t − 0 t∞ =− =− t∞ 0 dt eı(ωkl −ω)t eı(ωkl +ω)t∞ − 1 eı(ωkm −ω)t∞ − 1 − . we obtain x= ¨ 1 d f (x) d x px f˙(x) = = 2 f (x).9780521869638sol CUP/AUL November 6. m m m which is Newton’s law.3 Starting from Eq.. n + 2. n. (3.126) and (3. In deriving the previous equations we have taken into account that.5 Hamilton equations (1.4 The problem is solved by considering that 10. Then. n + 4. Eqs.62). for example. In particular. we easily obtain ˆ n x4 n = ˆ n x4 n + 2 = ˆ n x4 n + 4 = h ¯ 2mω h ¯ 2mω h ¯ 2mω 2 (6n 2 + 6n + 3). n − 2. m m d x dt m 10. ˆ 10. px = −V (x).7) give px . ωkl + ω ωkl − ω = eı xt − 1 e−ı xt − 1 = 2 − eı xt + e−ı xt xt = 4 sin2 . 2 10. we have x= ˙ V (x) f (x) px ˙ =− = . ˙ m where V (x) is the potential energy of the onedimensional Hamiltonian H = 2 px /2m + V (x).128) give .58) and using Eq. ˙ ˆ f (x) ˆ 1 ˆ ˆ x = − V (x) = ¨ . 2 (4n + 6)) (n + 1)(n + 2). x= px ˆ ˆ . m m .6 Quantummechanically. (10. 2008 18:21 Page45 t 45 S o l u t i o n s to se l e c te d p ro b l e m s ˆ previous equation we immediately infer that n x 4 m is different from zero only when m = n − 4. (10. Finally.. in the last one only the a 4 term survives. x= ˙ m from which it follows that ˆ px = −V (x).
107) to the operator x 2 − x . one has ˆ η = H − Ec = = ˆ px − px ˆ2 2m 2 f (x).e. px px + px f (x). according to Eqs. ˆ m ˆ + V − Vc 1 2 2 σ p + mσx Vc 2m = const. 2008 18:21 Page46 t 46 S o l u t i o n s to se l e c te d p ro b l e m s Making use of the Heisenberg equation (3. d2 2 σ dt 2 x 4 2 η − Vc σx . H = ¨ ˆ ¯ m Now. H = ˆ ˆ 1 2m 2 f (x) 2 ˆ . i. within the approximation V (x) 0.116b). and the momentum square deviation remains exactly constant. ˆ ˆ ˆˆ = ˆ ˆ m 2 Proceeding in a similar way.26. with (d/dt)σx . px = f (x) px − px f (x) + px f (x) px − px f (x) px ˆ ˆ2 ˆ2 ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ = f (x). ˆ In the case of the free particle. ˆ ˆ2 f (x). m2 or.. x performs a uniform rectilinear motion with ˆ velocity px /m. ˆ ˆ ı h x = x. we arrive at d2 2 σ dt 2 x 2 2 2 σ p − mVc σx . ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ ¯ where we have taken advantage of the result given in Prob. one has ˆ ˆ d d 2 d 2 2 σx = x − x = ˆ x2 − x ˆ ˆ ˆ dt dt dt 1 px x + x px − 2 x px . by taking into account that η is constant. we have 1 . 2 Applying Eq. 2 ˆ ˆ2 σ p = px − px 2 is the square deviation of px from its mean.108). is rather general. m This last equation. px .9780521869638sol CUP/AUL November 6. px = ı h f (x) px + px f (x) .117). (3. one obtains 2 d2 2 2 2 1 σ = 2 σp − 2 x m dt m where ˆ ˆ ˆˆ V x + xV − 2 x V ˆ ˆ . (10. ˆ 2 2 σ p (t) = σ p (t0 ) = 2mη.. This proves the ﬁnal result. 2.8 If V (x) is small. If we replace V in the last equation ˆ with the ﬁrst two terms of the expansion (10. . ˆ 10.
dt 2 m2 one rigorously obtains Eq. m dt 2 η= . i.9 The Hamiltonian of the harmonic oscillator is given by Eq.e.48). if the free wave packet is taken to be minimum at time t0 . σ = m dt 2 x 2 σp d2 2 σ p = 4mω2 η − . ˆ ˆ Moreover. and the width of the band increases indeﬁnitely. ˆ m px (t0 ) m then particles initially located around the same point become uniformly distributed over a band vx · t at time t. (10.9780521869638sol CUP/AUL November 6. ˆ ˆ dt i.e. Finally. Eqs. 1 2 2 σ p + m 2 ω2 σx . x and px carry out sinusoidal oscillations of frequency ω/2π about the origin. 10.119). Then.118) may be rewritten as px ˆ d x = ˆ dt m which imply d2 x = −ω2 x ˆ ˆ dt 2 and d2 px = −ω2 px .e. (10. i. 2m d2 2 4 2 η − mω2 σx . 2008 18:21 Page47 t 47 S o l u t i o n s to se l e c te d p ro b l e m s Morevoer. ˆ ˆ dt 2 and d px = −mω2 x . (4. The second “spreading” term of the rhs of the previous equation allows a classical interpretation of the free wave packet: a bunch of pointlike particlesinitally ˆ contained within a small interval x(t0 ) about the average value x(t0 ) . Since the velocities of these particles are dispersed over an interval vx = about the group velocity of the packet vx = 1 px (t0 ) . we have Vc = mω2 . 2 2 σx (t0 ) · σ p (t0 ) = h ¯ . 4 then σx (t0 ) = 0 and ˙2 x(t) = [ x(t0 )] + 2 px (t0 ) (t − t0 ) m 2 1 2 . since 2 2σ p (t0 ) d2 2 σx = .
134). As we see in Ch. this solution extends to region II (where x1 < x x2 ) in the form 2 ψII = −ıλq exp 1 x2 dx x 1 λq 2 = −ıλq eτ exp − 1 x dx x1 1 . II. 13. 10.9780521869638sol CUP/AUL November 6.139). The point of discontinuity x1 and the turning point x2 divide the xaxis in regions I. .10 Consider the potential depicted in Fig.2). λq where τ= If we deﬁne h ¯ λq (x1 ) = √ 2m [V (x1 ) − E] and x2 dx x1 1 . respectively. k= h ¯ in region I the solution of the Schrödinger equation may be written ψI = C sin [k(x − x1 ) + δ). Instead. 2008 18:21 Page48 t 48 S o l u t i o n s to se l e c te d p ro b l e m s 2 2 which show that σx and σ p oscillate sinusoidally with frequency ω/π about 2 σx = η mω2 and 2 σ p = mη. where the phase π/4 has been added for the sake of computation. that is. we must construct the solution of the Schrödinger equation whose asymptotic form in region III represents a purely transmitted wave (in the direction of increasing x). ψIII = λ 2 cos 1 x dx x2 1 π −ı λ 4 + ı sin x dx x2 1 π − λ 4 . the WKB approximation will have the form (10. 13. or η= 2 σp m 2 = mω2 σx . this is equivalent to the condition for a state to be coherent (see Subsec.4. V (x) is a positive function decreasing monotonically from the positive value V (x1 ) to V (∞) = 0. When x < x1 . and III.4. The conditions d2 2 σ =0 dt 2 x require 2 σx = 2 σp and d2 2 σ =0 dt 2 p m 2 ω2 . when x > x1 . In that region. λq √ 2m [E − V0 ] . (10. we have V (x) = V0 = const. for x x2 . 10. The condition we impose upon its asymptotic form determines that solution (to within a constant). In order to ﬁnd the transmission coefﬁcient. According to Eqs.
x−x ψ(x . i. x) = ˙ and 1 ψ(x. we shall limit ourselves to a particle moving in one dimension. −∞ Consider the ﬁrst exponential in the previous equation: if x is appreciably different from x.e. t). L(x. one can calculate the transmission probabilities (see also Subsec.11 Even though following result can be cast in general terms. t). We therefore make the substitution x = x + η. t). i. Expanding in power series the lhs up to ﬁrst order in and the rhs up to second order in η.1). we have ψ(x.t ψ(x . In the onedimensional case with a potential V (x. this factor oscillates very rapidly. 2008 18:21 Page49 t 49 S o l u t i o n s to se l e c te d p ro b l e m s where the constants C and δ are obtained applying the continuity conditions to the wave function and its logarithmic derivative at the point x1 . Let us start from Eq. In order to derive the desired differential equation. t)ψ(x. t = t + . Only for x values that are close to x do we obtain signiﬁcant contributions. 10. ˙ 2 e ı −h V ¯ x+x 2 . Given these results.191) and the fact that for a small time interval the action is approximately equal to times the Lagrangian. making the integral over x vanish.e. 4. (3. x. This in turn requires that the barrier be at least several wavelengths thick and that the trasmission probability be extremely small (< 10−5 ). it should be noted that the present calculation is correct if V (x) varies sufﬁciently slowly in regions II and III where the WKB approximation has been made. t). t + ) = N +∞ ¯ dηe h ı mη2 2 ¯ e− h ı V (x+ η .t ) 2 ψ(x + η. Taking into account Eq. p.2. t). ψ(x . −∞ where the main contribution to the integral will come from values of η of the order √ . 390). (10.99) for the onedimensional case. t + ) = N +∞ ¯ dx e h ı m(x−x )2 2 1 2 m x − V (x. This yields 1 ψ(x. t). One then ﬁnds that k cot δ = − 1 λq (x1 ) and C sin δ = −ı λq (x1 )eτ . we obtain . we consider the case in which the time t is only an inﬁnitesimal interval after t. t ) = d xG(x . Finally. t . t + ) = 1 N +∞ dx e −∞ ı x+x hL 2 ¯ . where we have omitted the ı factor (see footnote 16. where we expect important contributions only for small values of η.9780521869638sol CUP/AUL November 6.
t) satisﬁes the differential equation ı that is precisely the Schrödinger equation for a onedimensional system.322. 2008 18:21 Page50 t 50 S o l u t i o n s to se l e c te d p ro b l e m s 1 ∂ψ = ψ(x. t) + ∂t N +∞ ¯ dηe h ı mη2 2 1− −∞ ı V (x. t). t) + η 1 ∂ 2ψ ∂ψ + η2 2 . −∞ In the limit as approaches zero.9780521869638sol CUP/AUL November 6. t) = ψ(x. Let us consider the leading terms in both sides of the previous equation.8]. 11 [Gradstein/Ryshik 1981. t) by V (x. . the normalization factor N must be chosen so that the equality holds. t) − V (x. 12 [Gradstein/Ryshik 1981.462. t) h ¯ ψ(x. ∂t 2m ∂ x 2 This equality holds to order if ψ(x. m we ﬁnally obtain the equality ψ(x. Since11 +∞ d xeax = −∞ 2 π − . We have 1 ψ(x. h ∂t 2ım ∂ x 2 ¯ h2 ∂ 2ψ ∂ψ ¯ =− + V (x.2]. t) + h ∂ 2ψ ∂ψ ı ¯ = ψ(x. 3. 3. t)ψ. t) N +∞ ¯ dηe h ı mη2 2 . In a similar way the same result can be obtained in more complicated contexts. a where a is a complex number. ∂x 2 ∂x where we have replaced V (x + η/2. t)ψ(x. since the error is of order higher than . t) − . we have N = By making use of the integrals +∞ ¯ dηe h ı mη2 2 2πı h ¯ m 1 2 . η=0 −∞ and12 +∞ ¯ dηe h ı mη2 2 η2 = −∞ ıh ¯ .
Zm p In the case of the hydrogen atom (Z = 1).9780521869638sol CUP/AUL November 6. .5 The reduced mass is given by m= me Z m p me + Z m p m e −1 Zm p me 1− .4 We have: x→+∞ lim y(x) = 0. studying the ﬁrst and the second derivatives. we have m me 0.05%. b 4a = me 1 + me from which we derive me − m m = me me me .− .− . Furthermore. we obtain a minimum at (see ﬁgure above) x1 ≡ and a ﬂex at x2 ≡ 11. 2008 18:21 Page51 t 51 S o l u t i o n s to se l e c te d p ro b l e m s x 1 x2 Chapter 11 11. b 9a b2 2a . x→0+ lim y(x) = +∞. Zm p 3a 2b2 .
i.6 Starting from Eqs. For y = η we obtain the result (11. (11. 2008 18:21 Page52 t 52 S o l u t i o n s to se l e c te d p ro b l e m s 11.23a). into Eq.e. (11.30) with z = −y.23b). that is. (11.21). (11. we have ˆ H I − Er 1 =− E0 E0 =− Er + h2 ∂ 2 h 2 l(l + 1) Z e2 ¯ ¯ − + 2m ∂r 2 2m r 2 r h4 ∂ 2 h 4 l(l + 1) h2 Z Er ¯ ¯ ¯ − + − 2 e4 ∂r 2 2 e4 2 E0 2m 2m r me2 r 2 2 2 r l(l + 1) r ∂ Z ˜ − r0 = −E − 0 2 + 0 2 2 ∂r 2 r r 1 l(l + 1) Z 1 ∂2 ˜ + − .8 Consider Eq. r l+1 e− n W (˜ ) = ξ (˜ ˜ r r). for Z = 1. and making use of substitutions (11. 11.e. r ξ (˜ ) = − − and Eq. ˜ 11. ˜ r ˜ 2 2 2(l + 1) 2(l + 1)W 2W + − nr ˜ rW ˜ nW 2W 1 2(l + 1)W 2 l(l + 1) − − 2+ − ξ rW ˜ nW r ˜ n r2 ˜ 2(l + 1) 2 W 2 l +1 W + − + 1− = W r ˜ n W r ˜ n By multiplying by W r . g(y) + (γ + y) d(y) dy 2 where g(y) = f (−y). = −E − 2 ∂r 2 2 r2 r ˜ ˜ ˜ from which. which shows that the solution of this equation has the form g(y) = F(α. γ . .26).7 Substituting Eq. d2 d f (−y) − α f (−y) = 0. f (−y) + (γ + y) d(−y) d(−y)2 It is easy to derive y d2 d g(y) + αg(y) = 0.39).24).27b). (11. (11.13c) and (11. the desired result can be derived. and (11. −y ξ. we obtain 0= − + W W (l + 1)2 1 2(l + 1) l +1 W − + + + 2+ − 2 2 2 W nr ˜ r ˜ W r ˜ n W2 2 2 r ˜ W W l +1 W (l + 1)2 1 − + + + 2+ 2 2 2 W r ˜ W r ˜ n W2 r l+1 e− n W .9780521869638sol CUP/AUL November 6. −y).18). i. we obtain ﬁnally the desired result.
10 We ﬁnd R10 (˜ ) and leave the following ones to the reader. r 11. we obtain 2 0 dr e −2 rr 0 r 2 = 1. c j = 0. (11. +∞ d xe−x x 2 = 2. 2008 18:21 Page53 t 53 S o l u t i o n s to se l e c te d p ro b l e m s 11. we have n = 1. l = 0. As a consequence (see Eq. (11. .45)). (11. we ﬁnd the normalization factor for R10 (r ) and leave the others to the reader. the possible values of l range from 0 to n − 1. This implies that (see Eq. and.12 Again. e −2 rr 0 .26)) ξ (˜ = r e−˜ . (11. for any j > 1. We have ℘10 (r ) = 4r 2 3 r0 −3 r0 2 3 N 2 = 1. we ﬁnd the radial probability density corresponding to R10 (r ) and leave the others to the reader.9780521869638sol CUP/AUL November 6. and (see Eq.9 For each value of n. we obtain r) ˜ r r R10 (˜ ) ∝ e−˜ . Finally (see Eq. 2 since m k= k=0 m(m + 1) .47)). 0 which implies 2 or N = 2r0 2 . 11. moreover.19)). for each value of l there are 2l + 1 sublevels. The total number of degenerate states is then given by n−1 (2l + 1) = l=0 2n(n − 1) + n = n2. Denoting by N the normalization factor for R10 (r ). c1 = 0. k=0 r 11. we have c0 = 1. In this case. W (η) = 1. 2 and m 1 = m. we must have +∞ N Integrating by parts.11 Again.
For most hydrogenoid atoms this ratio is small because Z is much smaller than 100.14. 1 E n = − mc2 2 c is the speed of light. 2 so that ˆ v2 2T = = c2 mc2 Zα n 2 . (6.51) and impose the condition of Prob. r0 r r0 −2 = 1 n2 . (11. the mean value of the kinetic energy becomes 1 ˆ ˆ ˆ T = H − V = E n − 2E n = m v 2 . r r0 −3 = 1 n 3l l+ 1 2 (l + 1) Notice that the latter case is divergent for l = 0. H= 2m r and Ze ˆ V = − r 2 =− Z e2 Z r0 r r0 −1 =− Z e2 m Z e2 = 2E n . We know that. The previous results may be used to obtain an interesting consequence. (11. This is due to the fact that R(0) = 0 only for l = 0.14 Impose that the condition +∞ π 2 0 2π drr 0 dθ sin θ 0 dφ ϕnlm (r .69) and Eqs. Moreover.68)–(6. but ﬁnite for l > 0. 11. .9780521869638sol CUP/AUL November 6.13 Using the explicit form of the radial wave functions (Eq. 2008 18:21 Page54 t 54 S o l u t i o n s to se l e c te d p ro b l e m s 11. 2 r r0 −1 r = n 2 5n 2 + 1 − 3l(l + 1) . θ . ϕ200 . n2 h2 ¯ where. φ)2 = 1 is satisﬁed for ϕ100 . is the ﬁnestructure constant.15 Make use of Eqs. ˆ Z e2 p2 ˆ − . 11. and α= e2 hc ¯ 1 137 Zα n 2 . and in this case the radial integral is divergent. 11. we ﬁnd r r0 2 = 1 3n 2 − l(l + 1) . = 1 n3 l + 1 2 . for a hydrogenoid atom. for a hydrogenoid atom.51)). ϕ300 .
As a consequence. for the f levels (l = 3).16 The slevels (l = 0) are obviously unaffected by the spin–orbit interaction. where κ = κl=1 (n). the correction is equal to κl=2 (n) for j = 5/2 and to −3κl=2 (n)/2 for j = 3/2. It should be noted that the proportionality constant κ grows as Z 4 . 11. if r2 < r1 .9780521869638sol CUP/AUL November 6.21 From the deﬁnitions (11. we have either j = l + 1/2 = 3/2 or j = l − 1/2 = 1/2. given that. For example. the spin–orbit correction to the 6 p level of thallium (Z = 81. the unit of mass is simply given by the electron mass m e . Finally.22 We have (1) E1 = Z6 π2 Z6 = 2 π = Z6 2π π2 dr1 dr2 e−2Zr1 e−2Zr2 dr1 e−2Zr1 dr1 e−2Zr1 dr1 e−2Zr1 dr1 e−2Zr1 r1 dr2 1 r12 2 2πr2 e−2Zr2 2 2 r1 + r2 − 2r1r2 cos θ d cos θ 2 dr2 r2 e−2Zr2 1 2 2 2 r1 + r2 − 2r1r2 cos θ 2r1r2 cos θ=−1 cos θ=1 2Z 6 = π 2Z 6 = π 1 dr2r2 e−2Zr2 [(r1 + r2 ) − r1 − r2 ] r1 r1 0 2 dr2 e−2Zr2 2r2 + ∞ r1 dr2r2 e−2Zr2 2r1 . the spin–orbit correction is particularly important in the case of heavy atoms. since the physical dimensions of energy are given by [E] = [m][l]2 [t]−2 . 2008 18:21 Page55 t 55 S o l u t i o n s to se l e c te d p ro b l e m s 11. Similarly. the energy correction is equal to κ/2. Moreover. In the former case. A = 204) gives rise to a shift of about 1500 ρA in the wavelength of the radiation ˆ emitted in the corresponding transition. we obtain me m e e4 h 4 −2 ¯ t = 2 . From this it follows that t0 = 11. we ﬁnally obtain . Finally.112). From the previous equation. m e e4 where t0 is the atomic unit of time. and the unit of electron charge is equal to the opposite e of the electron charge. 2 e4 0 me h ¯ h3 ¯ . where θ is the polar angle. since ˆ · s = 0 identically. whereas in the latter it is simply given by −κ. for the dlevels (l = 2). we immediatly infer that the unit of length is given by r0 . the second integral in the square brackets is zero. l ˆ For the plevels (l = 1). the correction is equal to 3κl=3 (n)/2 for j = 7/2 and to −2κl=3 (n) for j = 5/2.
13). z b ). we obtain Nn − k=1 h2 2 ¯ ˆ ∇ + Vn + E e ψϕrn = ϕrn − k k 2m k k Nn k=1 h2 2 ¯ ˆ ∇ + Vn + E e ψ. 3. z 0 = z b − z a . n n n n n n n given ra = (xa . y. z). Limiting ourselves to the xb coordinate. z a ) and rn = (xb . z) and g = g(x.2 The solution of the problem can be found by making use of the mathematical identity ∇ 2 ( f g) = ∇ ∇( f g) = ∇ g∇ f + f ∇g = g∇ 2 f + f ∇ 2 g + 2∇ f ∇g. k! μn−k+1 obtaining E 1 = 32Z 6 = 5 Z. y. since the equality Nn ϕrn − k k=1 h2 2 ¯ ˆ ∇ + Vn + E e ψ = Eψϕrn k 2m k k must hold for any ϕrn . for any pairs of functions f = f (x.10) and of Eqs.27)) is equivalent to the following six changes of coordinates: n n n n n n m a xa + m b xb m a ya + m b yb m a za + m b zb . 2m k k from which the result is easily obtained.351]. xc = ma + mb ma + mb ma + mb n n n n n n x0 = xb − xa . (12. ya . . (12. 8 (1) dr1 e−2Zr1 2 − e−2Zr1 (2Z )3 r2 2 2r1 + + 1 2Z (2Z )3 (2Z )2 Chapter 12 12. 12. we compute the previous integral by making use of the expression13 u d x x n e−μx = 0 n! − e−uμ μn+1 n k=0 n! u k .4 The change of variable (Eq.21).9780521869638sol CUP/AUL November 6. k 12. 2008 18:21 Page56 t 56 (1) S o l u t i o n s to se l e c te d p ro b l e m s E1 = 2Z 6 16π π ∞ dr1r1 0 e−2Zr1 r1 r1 0 2 dr2 e−2Zr2 r2 . yc = . Now. zc = . yb . we have 13 See [Gradstein/Ryshik 1981.11) and (12. y0 = yb − ya . where the extra factor 2 comes from the contribution of all the symmetric conﬁgurations in which r2 > r1 .3 Given the simpliﬁcations (12. by making use of the deﬁnition (12.16)–(12.
provided that we have ˆ Ha ϕa (ra ) = E a ϕa (ra ). 12. 2 2 m a + m b ∂ xc ∂ x0 (m a + m b )2 ∂ xc ∂ xb ∂ x0 Taking into account only the xcoordinate terms in the Laplacians of Eq. 12. (12.5 We may write ˆ ˆ Ha + Hb ϕa (ra )ϕ B (r B ) = Eϕa (ra )ϕ B (r B ). we ﬁnally obtain the desired result. ˆ ˆ ϕb (rb ) Ha ϕa (ra ) + ϕa (ra ) Hb ϕb (rb ) = Eϕa (ra )ϕb (rb ). which represents the Morse potential for a diatomic molecule.24). i. we have − h2 ∂ 2 h2 h2 ∂ 2 ∂2 h2 ∂ 2 ¯ ¯ ¯ ¯ − =− − . 2 2 m a + m b ∂ xc ∂ x0 ∂ xa (m a + m b )2 ∂ xc ∂ x0 m2 ∂2 ∂2 ∂2 ∂2 2m b b = + 2+ . 2 2 2 2 2m a ∂ xa 2m b ∂ xb 2(m a + m b ) ∂ xc 2m ∂ x0 Adding the similar terms for the y.9780521869638sol CUP/AUL November 6. ˆ Hb ϕb (rb ) = E b ϕb (rb ).e. V(rab) _ rab rab –D .and zcoordinates. 2008 18:21 Page57 t 57 S o l u t i o n s to se l e c te d p ro b l e m s ∂ ∂ ∂ xc ∂ ∂ x0 ∂ ma ∂ = + = − . This last equation leads to E = Ea + Eb .8 See the ﬁgure below. ∂ xb ∂ xb ∂ xc ∂ xb ∂ x0 m a + m b ∂ xc ∂ x0 from which we obtain 2 ∂2 ∂2 ∂2 ma ∂2 2m a = + 2− . ∂ xa ∂ xa ∂ xc ∂ xa ∂ x0 m a + m b ∂ xc ∂ x0 ∂ ∂ xc ∂ ∂ x0 ∂ mb ∂ ∂ = + = + .
1d). from which Eq. (13. ˜ dr ˜ dr ˜ 12.9). c2 ∂t 2 which is the desired result. (13. (13.55) as vib E n = −D + n + 1 2D 1 − n+ ha ¯ 2 m 2 2 a2h2 ¯ .12 Let us ﬁrst rewrite Eq. and Eq. Comparing the ﬁrst two terms with Eq. 13. (12.55) is easily obtained. (13. consequently. ˜ = drab drab d r ˜ dr ˜ and.11 In order to solve the problem. 2m where the last term represents the anharmonic correction of the Morse potential.42). we have ω0 = a 2D .9) and making use of the expansion ˆ A= k ck l− 2 3 ak.20a) for E(r.9780521869638sol CUP/AUL November 6. 12. (12.1 eık·r e−ıωk t + a† e−ık·r eıωk t e1 ˆ ˆ k.48) to derive d dr d ˜ d = −ar . 2008 18:21 Page58 t 58 S o l u t i o n s to se l e c te d p ro b l e m s 12. and (13.2 .50). which implies vib −2m E n a2h2 ¯ 1 = − n+ 2 √ 2m D + ah ¯ 2 . c2 ∂t 2 1 ∂2 A.4 The expression (13. d d d2 = = a 2r ˜ 2 drab drab drab d2 d +r 2 .2 eık·r e−ıωk t + a† e−ık·r eıωk t e2 . ˆ ˆ k.2 Inserting Eqs. we obtain ∇ × (∇ × A) = − Using the mathematical identity ∇ × (∇ × V) = ∇ (∇ · V) − ∇ 2 V.1 + ak.9 Just make use of Eq. (12. (12. t) is calculated using Eq. we need to make use of deﬁnitions (12. which holds for any vector V.3b). m Chapter 13 13. (13.7). we arrive at ∇ 2A = 1 ∂2 A.8) into Eq.
5 Let us ﬁrst compute the squares of the electric and magnetic ﬁelds ˆ E2 = k.λ .λ ˆ ˆ B2 = k.λ ˆ e−2ık·r + ak.1 − ak. (13. and where ck = k z h ωk ¯ 2 0 l3 1 2 = hk ¯ 2cl3 1 2 . 0 13. due to the periodic boundary conditions. taking into account that eık·r = eı(k x x+k y y+kz z) . from which. since d xeık x x = l 1 ı 2π n x x e l ık x l 0 = 0.λ ˆ k.2 since k x = k y = 0. b1 = e2 and b2 = −e1 (see Eq.2 eı(k·r−ωk t) − a† e−ı(k·r−ωk t) b2 . The expression (13. We recall the mathematical expression ∇×V= ∂ Vy ∂ Vz − ∂y ∂z ı+ ∂ Vx ∂ Vz − ∂z ∂x j+ ∂ Vy ∂ Vx − ∂x ∂y k for any vector V.λ 2 Now.20b) for B(r. ˆ ˆ k.λ e2ık·r − a† ˆ2 ˆ k. t) is a bit more cumbersome to derive.λ e2ık·r − a† ˆ2 ˆ k.2 =ı k ck ak. we have ∇ × eık·r e1 = ık z eık·r j − ık y eık·r k.21)). ∇ × e−ık·r e2 = +ık z e−ık·r ı − ık x e−ık·r k.9780521869638sol CUP/AUL November 6. ˆ ˆ k.2 eı(k·r−ωk t) − a† e−ı(k·r−ωk t) e1 ˆ ˆ k.1 + ak.1 eı(k·r−ωk t) − a† e−ı(k·r−ωk t) e2 ˆ ˆ k. ∇ × eık·r e2 = −ık z eık·r ı + ık x eık·r k. ∇ × e−ık·r e1 = −ık z e−ık·r j + ık y e−ık·r k.λ a† + a† ak.1 eı(k·r−ωk t) − a† e−ı(k·r−ωk t) b1 ˆ ˆ k. we have l3 dre2ık·r = 0.λ 2 0 l3 2 e−2ık·r + ak. 2008 18:21 Page59 t 59 S o l u t i o n s to se l e c te d p ro b l e m s from which we obtain easily the desired expression for the electric ﬁeld. . ˆ ˆ k.λ ˆ k.λ . we have ˆ ∇×A=ı k ck ak.λ h ωk ¯ −ak.λ a† + a† ak. and similarly for the other components.λ 2 0 l3 hk ¯ −ak. Collecting these results together.
9780521869638sol CUP/AUL November 6. using Eq. we obtain Eq. = 2 e ¯ −1 where we have made use of the mathematical relation ∞ nx n−1 = n=0 d dx ∞ xn = n=0 1 (1 − x)2 . 13.39).11 Take a generic matrix element of the operators eıφ and e−ıφ in the number basis. we have n eıφ m = and n e−ıφ m = n  m + 1 = δn. (13. Then.32) it is easier to work in the energy (number) representation. In fact. we have ∞ ∞ E = n=0 E n ρnn = n=0 ∞ 1 e n+ hω ¯ 2 + ∞ n=0 1 −β n+ 2 h ω ¯ Z (β) ∞ = = hω ¯ 2Z (β) e n=0 1 −β n+ 2 h ω ¯ hω ¯ Z (β) ne n=0 1 −β n+ 2 h ω ¯ hω ¯ ¯ + h ω 1 − e−β h ω ¯ 2 ¯ ne−βn h ω hω hω ¯ ¯ + β hω . n  m − 1 = δn. −β h ω 1−e ¯ 1 13. (13. ¯ ¯ −1 ∂β 2 e (b) The same result can be obtained by making use of the number distribution. (13. ¯ 2 and. (13. 13. E =− ∂ 1 hω ¯ ln Z = h ω + β h ω . (13. 2008 18:21 Page60 t 60 S o l u t i o n s to se l e c te d p ro b l e m s Substituting these results into Eq.23). we have also proven that eıφ and e−ıφ are not Hermitian. ∗ ˆ Since for a Hermitian operator O we must have Omn = Onm .2).7 (a) We have 1 ¯ ln Z = − β h ω − ln 1 − e−β h ω .6 In order to evaluate the trace in Eq. therefore.22) and using Eq. where ∞ Z (β) = n=0 1 n e−β H n = ∞ n=0 ¯ e−β h ω ˆ ∞ e n=0 n −β h ω n+ 1 ¯ 2 ¯ = e− 2 β h ω = ¯ e− 2 β h ω . since the energy is diagonal in the number basis.m+1 .m−1 0 for m = 0 for m = 0 .
Bx + E y .12 Making use of Eqs. (13.14 In an eigenstate of the number operator we obviously have n cosφ n = 0. n cos2 φ n = ˆ N = 0. We also have 1 n eıφ eıφ + eıφ e−ıφ + e−ıφ eıφ + e−ıφ e−ıφ n 4 1 2 if n = 0 . 2ı a ˆ 13. sinφ = 1 4ı 1 4ı ˆ N +1 −1 2 ˆ a + a† N + 1 ˆ ˆ −1 2 −1 2 ˆ N +1 −1 2 ˆ a − a† N + 1 ˆ ˆ −1 2 −1 2 − = = ˆ N +1 −1 2 ˆ a − a† N + 1 ˆ ˆ ˆ a a† N + 1 ˆˆ ˆ . 13. (13. which conﬁrms that the phase is completely undetermined for an eigenstate of the number operator (i. and b2 = k × e2 = −ke1 . B y + E z . n sinφ n = 0. we have cosφ.20) and (13.15 By using Eqs.17).9780521869638sol CUP/AUL November 6.e.37) and (13. . N +1 −1 2 ˆ N +1 −1 2 −1 ˆ a + a† N + 1 ˆ ˆ a ˆ 1 ˆ − N +1 2ı −1 2 −1 2 −1 2 ˆ + a† N + 1 ˆ 1 ˆ a† N + 1 ˆ 2ı 1 −ıφ ıφ = e . B = E x . ˆ ˆ ˆ ˆ and cB = hk ¯ 2c 0 l3 1 2 . = 1 4 if n = 0 2 The same values can be derived for n sin φ n . Bz . This means that both cosφ and √ sinφ for n = 0 are equal to 1/ 2. ˆ ˆ ˆ ˆ aeı(k·r−ωt) − a† e−ı(k·r−ωt) ke2 − aeı(k·r−ωt) − a† e−ı(k·r−ωt) ke1 . 2008 18:21 Page61 t 61 S o l u t i o n s to se l e c te d p ro b l e m s 13. However. it corresponds to a uniform phase distribution between 0 and 2π ).40). and we have made use of the fact that b1 = k × e1 = ke2 We also have ˆ ˆ ˆ ˆ ˆ ˆ ˆ ˆ E. one has ˆ E = cB and ˆ B = cB where cE = ı hω ¯ 2 0 l3 1 2 aeı(k·r−ωt) − a† e−ı(k·r−ωt) e1 + aeı(k·r−ωt) − a† e−ı(k·r−ωt) e2 .e .
we obtain 1 ˆ ˆ a = √ ( X 1 + ı X 2 ). 13. (4. (13.and ydirections. (13.16 The ﬁrst equation is trivial. ˆ 13.73) we may identify X 1 and X 2 as ˆ X1 = ˆ X2 = mω h ¯ 1 mωh ¯ 1 2 x.9780521869638sol CUP/AUL November 6. ˆ α a 2 − a a† − a† a + (ˆ † )2 α − ˆˆ ˆ ˆ a ˆ α a† − a α ˆ 2 . we immediately obtain ˆ ˆ α a† a α = α2 . respectively. and similarly for the other commutators. Then we have ˆ α X1 = = and ˆ α X2 = = ˆ2 α X2 α − − 1 2 ˆ α X2 α 2 ˆ2 α X1 α − 1 2 ˆ α X1 α 2 ˆ α a 2 + a a† + a† a + (ˆ † )2 α − ˆˆ ˆ ˆ a ˆ ˆ α a + a† α 2 . we have ˆ N2 α ˆ ˆˆ ˆ = α a † a a† a α ˆ ˆ = α2 α a† a + 1 α = α2 1 + α2 . ˆ 2 1 ˆ ˆ a † = √ ( X 1 − ı X 2 ).61). since. by using Eq.49a). ˆ 1 2 px . 2008 18:21 Page62 t 62 S o l u t i o n s to se l e c te d p ro b l e m s By recalling that the vectors e1 and e2 are parallel to the x. 13. and that k is in the direction of z. Bx = − a − a† ˆ ˆ 2c 0 l3 2 .17 Inverting Eqs. . the ﬁrst commutator reads h ω2 ¯ ˆ ˆ E x . ˆ 2 ˆ ˆ Comparing with Eqs.18 Let us ﬁrst write an explicit expression for the uncertainties of the quadratures in the coherent state  α . Analogously. from which we obtain αn = = α2 1 + α2 − α4 α2 = α.
to the second order in η.9780521869638sol CUP/AUL November 6. π + η2 ˆ Now.232) follows to the second order in η. For the proof of the general result. dη . we may write ˆ e−ηξ π eηξ = 1 − ηξ + η2 ˆ ˆ ˆ ˆ ˆ 1 2 13. ˆˆ α a a† α = 1 + α2 .232) as O . ξπ − πξ ˆˆ ˆ ˆ ˆˆˆ = ξ 2 π − 2ξ π ξ + π ξ 2 . let us ﬁrst compute the derivative of the lhs of Eq.19 To the ﬁrst order in η we have ˆ ξ2 2 ˆ π 1 + ηξ + η2 ˆ ˆ ξ2 2 ˆ ˆ = π − η ξ . ˆˆ from which Eq. π ˆ ˆ ξ2 ξ2 ˆˆˆ π − η2 ξ π ξ + η2 π . ˆ α a† α = α ∗ . d −ηξ ηξ ˆ ˆ ˆ ˆ ˆ ˆˆ ˆ e π e = −ξ e−ηξ π eηξ + e−ηξ π eηξ ξ . ˆ ˆ α a† a α = α2 . π . a α (ˆ † )2 α = (α ∗ )2 . ξ. ξ . ˆ ˆ ˆ dη ˆ ˆ ˆ Let us make use of the substitution O ≡ e−ηξ π eηξ . − 2 2 ˆ ˆ ˆ ˆ e−ηξ π eηξ = 1 − ηξ π 1 + ηξ ˆ ˆ = π − η ξ. ˆ dO ˆ ˆ = O . ˆ Similarly. we easily see that. the previous equation ˆ reads as ˆ dO ˆ ˆ = O. we have that ˆ ˆ ˆ ξ. that is. (13. Now we know or may easily derive that ˆ α a α = α.61). ˆ α a2 α = α2. (13. dη ˆ If we denote the rhs of Eq. we ﬁnally obtain ˆ α X1 ˆ α X2 = = 1 α 2 + 1 + α2 + α2 + (α ∗ )2 − (α + α ∗ )2 = √ 2 1 1 α 2 − 1 − α2 − α2 + (α ∗ )2 − (α ∗ − α)2 = √ . (13. Then.ξ . 2008 18:21 Page63 t 63 S o l u t i o n s to se l e c te d p ro b l e m s where we have made use of Eqs. (13. Making use of these relations. ˆ ˆ 2 2 ˆ ˆˆ = ξ. by computing again the derivative with respect to η.232) with respect to η .
19 and of the fact that ˆ ˆ ˆ O. 13.9780521869638sol CUP/AUL November 6. ˆ ˆ ξ (η) is the solution of this differential equation for which ξ (0) = 1. by making use of the results of Prob. and compute its derivative with respect to η: d ˆ ˆ ˆ ˆ ˆ ˆ ˆ ξ (η) = Oeη O eη O + eη O O eη O dη ˆ ˆ ˆ ˆ ˆ = O + eη O O e−η O ξ (η). t is an eigenstate of the annihilation operator. the last two equations show that O and O satisfy the same differential equaˆ ˆ tion. 13. for any boundary condition O ≡ O and the lhs and the rhs of Eq. t = ae− h H t  α = e− h H t e+ h H t ae− h H t  α .232) are equal. ˆ ˆ ¯ ¯ To evaluate e+ h H t ae− h H t we use the formula (13.23 Let us consider an initial coherent state  α . which implies d ˆ ˆ ˆ ˆ ˆ ξ (η) = O + O + η O. which. As a consequence. We want now to verify that  α. O n! n . We have ¯ ¯ ˆ ¯ a  α. for η = 1. O. (13. Its time evolution is given by ¯  α. However. 13. t = e− h H t  α . O = 0. Since the comˆ . proves the desired result. O dη ˆ ξ (η). ˆ ˆ ˆ = O . 14 See [Gardiner 1991. O we obtain ˆ ˆ ˆ ˆ ˆ ˆ eη O O e−η O = O − η O . ˆ ˆ ˆ mutator O we ﬁnally obtain η ˆ ξ (η) = e ˆ ˆ O+ O e η2 ˆ ˆ 2 [ O. 138–39] and [Messiah 1958. . O commutes with O + O .20 Let us ﬁrst write14 ˆ ˆ ˆ ξ (η) = eη O eη O . O ] . by integrating the previous equation. 442]. ˆ ¯ ı ı ˆ ı ˆ ı ˆ ı ˆ ˆ ı ˆ ˆ ˆ ˆ e− O O e O = n (−1)n ˆ ˆ O.234). O. 2008 18:21 Page64 t 64 S o l u t i o n s to se l e c te d p ro b l e m s ˆ ˆ Now. ı ˆ where 1 ˆ H = h ω a† a + ˆ ˆ ¯ 2 is the singlemode freeﬁeld harmonicoscillator Hamiltonian. O .
9780521869638sol
CUP/AUL
November 6, 2008
18:21
Page65
t
65
S o l u t i o n s to se l e c te d p ro b l e m s
The evaluation of the ﬁrst two terms of the inﬁnite sum in the formula above with ˆ ˆ ˆ O = −ıω a† a + 1 t and O = a allows us to write ˆ ˆ 2
¯ e+ h H t ae− h H t = ˆ ¯ ı
ˆ
ı
ˆ
n
(−ıωt)n a = ae−ıωt , ˆ ˆ n!
where we have also made use of the result of Prob. 4.11. Using this result, we obtain
¯ ¯ a  α, t = e− h H t e−ıωt a  α = αe−ıωt e− h H t  α ˆ ˆ ı
ˆ
ı
ˆ
= αe−ıωt  α, t , which shows that the initial coherent state  α under time evolution remains a coherent state, i.e. the state  α, t =  α(t) with eigenvalue α(t) = αe−ıωt . 13.24 Using the completeness relation (13.71) we have, for any state  ψ F of the radiation ﬁeld,  ψF = 1 π d 2α  α α  ψF .
In particular, for a generic coherent state  β we have β = 1 d 2α α  β  α π 1 1 1 2 ∗ 2 d 2 αeα β− 2 α  α , = e− 2 β π
where we have made use of expression (13.69). 13.27 Given the most general expression of the density matrix ρ = j w j ψ j ψ j with ˆ w j = 1 (see Eq. (5.20)) and { ψ j } representing an orthonormal basis, we may j write Q(α, α ∗ ) = 1 π w j  α  ψ j 2 ≤
j
1 π
wj =
j
1 , π
since  α  ψ j 2 ≤ 1. 13.28 Any density matrix may be written as ρ = j w j ψ j ψ j , { ψ j } representing ˆ an orthonormal basis. On the other hand we have j w j = 1. Then, we may write ˆ  n ρ m =
j
wj n  ψj ψj  m wj n  ψj
j
≤
ψj  m .
Making use of the fact that, for any real x and y, 2x y ≤ x 2 + y 2 , we have wj n  ψj
j
ψj  m ≤ ⎡ 1⎣ = 2
j
1 wj 2
n  ψj
2
+ ψj  m
2
⎤ wj n  ψj
j 2
+
j
wj ψj  m
2⎦
.
9780521869638sol
CUP/AUL
November 6, 2008
18:21
Page66
t
66
S o l u t i o n s to se l e c te d p ro b l e m s
Since n  ψj we ﬁnally have ⎡ 1⎣ wj n  ψj 2
j 2
≤ 1 and
ψj  m ⎤
2
≤ 1, ⎛ ⎞ w j ⎠ = 1.
j
2
+
j
wj ψj  m
2⎦
≤
1⎝ 2 2
13.30 Using the identity 1 2π we have
+∞ −∞ +∞ −∞ +∞ −∞
d xeı x y = δ(y),
+∞ −∞
dpx W (x, px ) =
1 πh
ˆ dx x + x ρ x − x
dpx e2ı
px x h
ˆ = x ρ x = ℘(x).
Chapter 14
14.1 Let us write  R = Then, we have ρ SR = c0 2  0 0  ˆ
n, j ∗ + c0 c1 0 j
a j r j ,  R0 =
∗
j
a 0 r j , and  R1 = j
1 an a 1 j n, j ∗
j
a1 r j . j rj
0 an a 0 j
rn
∗
r j + c1 2  1 1  rn
∗ r j + c0 c1 1
rn
∗
1
n, j
0 an a 1 j
0
n, j
1 an a 0 j
rn
rj .
Now, by tracing the reservoir out, we obtain the reduced density matrix ˆ ρS = ˜
k
ˆ rk ρ SR rk
0 an n ∗ + c0 c1 0 2
= c0 2  0 0  1
+ c1 2  1 1 
∗ n ∗ + c0 c1 1
1 an
2
0 1 an an n 2
0
n
1 0 an an
∗
.
The sums ˆ P0 =  R0
n
0 an
2
and
0 an a 0 j
n
1 an ∗
ˆ ˆ are the traces of P0 and P1 , respectively, where ˆ r j , P1 =  R1 R1  =
n, j ∗ 1 an a 1 j ∗
R0  =
n, j
rn
rn
rj , and
0 1 and, by the normalization condition, both are equal to 1. The terms n an an ∗ 1 0 are equal to R1  R0 and R0  R1 , respectively. In fact, n an an
R1  R0 =
n, j
0 an a 1 j 1 an a 0 j n, j
∗
r j  rn =
n
0 1 an an 1 0 an an n
∗
, .
R0  R1 =
∗
r j  rn =
∗
Collecting these results and writing them in matrix form, we obtain Eq. (14.5).
9780521869638sol
CUP/AUL
November 6, 2008
18:21
Page67
t
67
S o l u t i o n s to se l e c te d p ro b l e m s
14.2 It is easy to see that ˆ ˆ O O =
l,m, j,k ∗ ∗ Ol,m O j,k {l, m j, k} = Ol,m O j,k δl, j δk,m ∗ Ol,m Ol,m = l,m l,m † ˆ ˆ Oml Ol,m = Tr O † O ,
= and
ˆ { j, k O} = { j, k
l,m
Ol,m l, m}
=
l,m
Ol,m δl, j δk,m = O j,k .
14.5 Let us write the density matrix ρ in terms of the expansion (13.92), that is, ˆ ρ=π ˆ
n,m
ˆ Qn,m a†
n
am . ˆ
Then, we may write ρ a† = π ˆˆ
n,m
ˆ Qn,m a†
n
a m a† ˆ ˆ
n
= a† π ˆ
n,m
ˆ Qn,m a†
am + π ˆ
n,m
ˆ mQn,m a†
n
a m−1 ˆ
dρ ˆ , = a† ρ + ˆ ˆ da ˆ where we have taken into account the result a m a† = a† a m + m a m−1 ˆ ˆ ˆ ˆ ˆ of Prob. 4.12. The ﬁnal equality on the rhs of Eq. (14.53) may be derived by taking into account the expansion (13.91) that yields α 14.6 We have that ˆ ˆˆ ˆ ˆ ˆˆ α a ρ a† α = α a† ρ a + a† since ρ a† = a† ρ + ˆˆ ˆ ˆ dρ ˆ da ˆ and a ρ = ρ a + ˆˆ ˆˆ dρ ˆ d a† ˆ . dρ ˆ d a† ˆ +ρ+ ˆ d 2ρ ˆ dρ ˆ α , a+ ˆ da ˆ d ad a† ˆ ˆ ∂Q dρ ˆ α =π . da ˆ ∂α
By making use of the previous results we obtain the desired solution.
8 We have that 2 2 2 s2 = s x + s y + s z 2 2 2 2 2 2 = ρeg + ρge + 2ρeg ρge − ρeg − ρge + 2ρeg ρge + ρee + ρgg − 2ρee ρgg 2 2 = ρee + ρgg + 4ρeg ρge − 4ρee ρgg + 2ρee ρgg = 1 + 4 ρeg ρge − ρee ρgg . from which it follows that ρge ρeg < ρee ρgg . i. therefore. and. we have for ρgg : ˙ ˙ ˆ ˆˆ ˆˆ ˆ ˆ ˆ ˆ ˆ ρgg = g ρ g = γ 2 g σ− ρ σ+ g − g σ+ σ− ρ g − g ρ σ+ σ− g ˙ ˆ = 2γ e ρ e = 2γρee . where we have made use of the fact that. 14. if ρ is a pure state. ˆ σ+  g =  e . we may calculate the time derivatives of the elements of ρ: ˆ ˙ ˆ ˆ ˆ ˆˆ ˆ ˆ ρee = e ρ e = −γ e σ+ σ− ρ e + e ρ σ+ σ− e ˙ = −γ ˆˆ ˆ ˆ g σ− ρ e + e ρ σ+ g ˆ ˆ = −γ e ρ e + e ρ e = −2γρee . we must ˆ 2 ˆ have that Tr(ρ ) = 1. ˆ σ+  e = 0. ˆ σ−  g = 0. from which it follows that ρge ρeg = ρee ρgg . 2008 18:21 Page68 t 68 S o l u t i o n s to se l e c te d p ro b l e m s 14. Now it is easy to show that 1 + 4 ρeg ρge − ρee ρgg is equal to one in the case of pure states and strictly smaller than 1 in the case of mixtures. we also ˆ ˆ must have that 2 2 ρgg + ρge ρeg + ρeg ρge + ρee < 1. ˆ Then. since ρee + ρgg = 1 (they are the diagonal 2 2 elements of the density matrix). .e. In the case in which ρ is a mixture. also ρee + ρgg + 2ρee ρgg = 1. which immediately gives the desired result. This means that the sum of the diagonal elements of ρ 2 must ˆ be equal to 1.9 We recall that σ−  e =  g . from which it follows that 1 + 4 ρeg ρge − ρee ρgg < 1. Analogously. In fact. 2 2 ρgg + ρge ρeg + ρeg ρge + ρee = 1.9780521869638sol CUP/AUL November 6. we have that Tr(ρ 2 ) < 1.
η) the lefthand side and R(ξ . η) satisfy the same ﬁrstorder differential equation with the same boundary condition and therefore they must be equal. we have used the following procedure: ρgg (t) = 1 − ρee (t) = 1 − e−2γ t ρee (0) = 1 − e−2γ t 1 − ρgg (0) . s y (t) = ıe−γ t ρeg (0) − ρge (0) = e−γ t s y (0).10 This equation plays an important role and it can be proved in many ways. Both the functions L(ξ .9780521869638sol CUP/AUL November 6. Let us call L(ξ . The solutions of these differential equations are then given by ρee (t) = e−2γ t ρee (0). • Both terms satisfy the same differential equation ∂ L(ξ . ˙ The solution of these differential equations are ﬁnally sx (t) = e−γ t sx (0). η) and R(ξ . A ﬁrst possibility is to verify that: • Both terms are equal to the identity at ξ = 0. η) the righthand side. η) ˆ ˆ = ( O + η O )R(ξ . From these results. ˙ ˙ ˙ ˙ ˙ s y = ı ρeg − ρge = −ıγ ρeg − ρge = −γ s y . . 2008 18:21 Page69 t 69 S o l u t i o n s to se l e c te d p ro b l e m s For the time derivative of the matrix element ρeg . for deriving the second solution. ∂ξ The proof of the ﬁrst equation is trivial (it is essentially the deﬁnition of the exponential). η). ρeg (t) = e−γ t ρeg (0). ∂ξ ∂ R(ξ . which are in agreement with Eqs. ρgg (t) = 1 + e−2γ t ρgg (0) − 1 . 14. η). we have ˙ ˆ ˆˆ ˆˆ ˆ ˆ ˆ ˆ ˆ ρeg = e ρ g = γ 2 e σ− ρ σ+ g − e σ+ σ− ρ g − e ρ σ+ σ− g ˙ ˆ ˆ ˆ = −γ g σ− ρ g = −γ e ρ g = −γρeg . (14. η) ˆ ˆ = ( O + η O )L(ξ . where. ˙ ˙ ˙ sz = ρee − ρgg = −2γρee − 2γρee == −4γρee = −2γ (1 + sz ).83). it follows that the time derivatives of the Bloch vector’s components are sx = ρeg + ρge = −γ ρeg + ρge = −γ sx . sz (t) = ρee (t) − ρgg (t) = e−2γ t ρee (0) − 1 − e−2γ t ρgg (0) + e−2γ t = e−2γ t sz (0) + e−2γ t − 1. while the proof of the second equation can be obtained by inspecting the different terms.
The term η2 may come from both the jth and the lth terms of the product ˆ (where all the other factors give a contribution equal to 1/δ O). We can now perform the limit δ → 0 and obtain ξ dξ1 0 ξ1 ˆ ˆ ˆ ˆ ˆ dξ2 eξ1 O O e(ξ2 −ξ1 ) O O e(ξ −ξ2 ) O . we can write L(ξ . δ→0 δ→0 ˆ ˆ ξ ξ ˆ ˆ ξ Without loss of generality we can evaluate the limit δ → 0 by restricting to the sequence of δ where ζ = ξ/δ is an integer. j=1.4) R(ξ1 . for small ξ . for any real δ = 0.ζ We can now easily compute the expansion in powers of η of the previous formula: the rhs is the product of terms linear in η. We can thus substitute the sums with integrals. (14. so we have to compute the Taylor expansion of the lhs. • The function R(ξ .91) is to check that the lhs and the rhs do coincide term by term in the Taylor expansion around η = 0. However the most instructive (and constructing) way to verify Eq.ζ a=1. Using this deﬁnition of ζ we have L(ξ .l=1. j<l ˆ ˆ (1 + δ O)⎠ O ⎝ j−1 ˆ (1 + δ O)⎠⎠ c=l+1. 8.ζ .9780521869638sol CUP/AUL November 6. . The two properties identify in a unique way the function and therefore the two functions must coincide. In the limit δ → 0 each individual term in the sum over j and l is irrelevant and we can assume that both j and l are of order δ −1 . 2008 18:21 Page70 t 70 S o l u t i o n s to se l e c te d p ro b l e m s A second proof can be obtained by performing the following steps: • One veriﬁes that. ˆ ˆ R(ξ . η) satisﬁes the semigroup property (see Sec. neglecting terms going to zero with δ we obtain ξ ξ dξ1 0 ξ1 ξ ˆ 1 ˆ ˆ dξ2 (1 + δ O) δ O (1 + δ O) ξ ξ2 −ξ1 δ ˆ ˆ O (1 + δ O) ξ −ξ2 δ . η) = 1 + ( O + η O )ξ + O(ξ 2 ). The function L(ξ . η) trivially satisﬁes the same equations. Let us compute the coefﬁcient of the order η2 . η) = (e( O+η O )δ ) δ . To this end it is convenient to notice that. η) = R(ξ1 + ξ2 .ζ . η)R(ξ2 . In particular. as can be proved by combining the different terms present in the lhs. η) = lim δ→0 ˆ ˆ (1 + δ( O + η O )). η) = lim (e( O+η O )δ ) δ = lim (1 + δ( O + η O )) δ . we can write ˆ ˆ L(ξ .l−1 ˆ (1 + δ O) ˆ ˆ O (1 + δ O) l− j−2 ˆ ˆ O (1 + δ O)ζ −l−1 . j<l ⎝⎝ ˆ ˆ (1 + δ O)⎠ O ⎝ =δ 2 j. j−1 b= j+1. η). We thus ﬁnd that the 2 is just given by coefﬁcient of η ⎛⎛ ⎞ ⎛ ⎞ ⎛ ⎞⎞ δ2 j.l=1. The Taylor expansion of the rhs is trivial.
η).91) is a convergent series (and that the lhs exists). ρ + H .13 First. ˆ Time derivation leads to ˙ ˙ ρ= ψ ˆ where ˙ ψ = Then. the reader may verify that the exponential e−ı 2 t j is equal to 1 for even values of j ( j = 2k) and to −ı for odd values of j ( j = 2k + 1). we have 1 ˜ 1 ˜ ˙ ˆˆ ˆ ˆ ρ= ˆ Hρ − ρH† ıh ıh ¯ ¯ 1 1 ˆ ˆ ˆ ˆ H0 + H ρ − ρ H0 + H ˆ ˆ = ıh ıh ¯ ¯ 1 ˆ 1 ˆ H0 . 2008 18:21 Page71 t 71 S o l u t i o n s to se l e c te d p ro b l e m s which is essential the result stated in Eq. we have 2 π ψ + ψ ˙ ψ .2). and the second and third terms of the rhs. so as to obtain . If not (the perturbative expansion is often not convergent). 14. h ¯ e − α 2 2 e−ı 2 t j e− 2 j = α √ 2 j! j=0 2 ∞ π α2 ∞ k=0 j 2α 2k α 2k+1  2k + 1 (−2ı) √ √  2k + (2k + 1)! 2k! k=0 ∞ ∞ =e 2 − α 2 ∞ (1 + ı) k=0 ∞ α 2k  2k  2k + (1 − ı) √ √ (2k)! 2k! k=0 ∞ α 2k + (1 − ı) k=0 α 2k+1 α 2k+1  2k + 1 − (1 + ı)  2k + 1 √ √ (2k + 1)! (2k + 1)! k=0 . (14.ρ .1–10. 10.11 Let us start from the identity ρ = ψ ψ  . the rhs should be interpreted as an asymptotic expansion (see also Secs. It should be clear to the reader how to generalize the result to higher (and lower) values of k. (14. Now we group the ﬁrst and last terms. ıh ¯ ˙ ψ = ı ˜ ˆ ψ  H †. We should note that all the proofs presented here have a rather formal character: we are implicitly assuming that the rhs of Eq. apart from a redeﬁnition of the integration variables. but we shall not show this further derivation.9780521869638sol CUP/AUL November 6. 10.8) for the function L(ξ . The same result could also be obtained starting from path integral representation (see Sec. where we have made use of the identities 2 = 1 + ı + 1 − ı and −2ı = 1 − ı − (1 + ı). Then. 1 ˜ ˆ H ψ . ˆ ˆ = + ıh ıh ¯ ¯ 14.91) for k = 2.
To prove this.15 Any distance between two elements a. The property d(a. Finally. 2008 18:21 Page72 t 72 − α 2 2 S o l u t i o n s to se l e c te d p ro b l e m s e e−ı 2 t j 1 j = √ α √ j! 2 j=0 2 ∞ π j (1 − ı) − α2 √ e 2 2 (−α)k k √ k! ∞ k=0 (1 + ı) α2 αk √  k + √ e− 2 k! 2 ∞ k=0 π π 1 = √ e−ı 4  α + eı 4  −α 2 where we have made use of the mathematical identity π (1 ± ı) = e±ı 4 √ 2 and of the fact that (see Eq. When applying the deﬁnition of the Fubini–Study distance to the distance between the coherent states in Box 14. since it is an instance of the triangular inequality (see also Subsec. which is satisﬁed. consider that   ψ1 −  ψ2 2 = 2 − 2 ψ1  ψ2 . since  ψ1  ψ1 2 = 1. Similarly. it sufﬁces to assume. Also the property that d(a. 1. (13. since  ψ1  ψ2 2 =  ψ2  ψ1 2 . we obtain . b) = d(b. c) ≥ d(a.9780521869638sol CUP/AUL November 6.58)) . from which it follows that it sufﬁces to multiply both the lhs and the rhs of the above inequality so as to obtain   ψ1 −  ψ2  +   ψ2 −  ψ3  ≥   ψ1 −  ψ3 . 2.70)).3. Indeed we have 1 −  ψ1  ψ2 2 + 1 −  ψ2  ψ3 2 ≥ 1 −  ψ1  ψ3 2 .2) a + b ≥ a + b. so that dfs = 1 − e−αe ıφ −αe−ıφ 2  = 1 − e−4α 2 sin2 φ . (13. b) + d(b. a) = 0 is easily satisﬁed by the Fubini–Study distance.1. c) is also satisﬁed. e− α2 2 ∞ k=0 (±α)k  k =  ±α . the Fubini–Study distance is always positive when the two states are different except the trivial case in which the difference is only given by the global phase. √ k! 14. the triangular property d(a. a). that the involved state vectors are normalized and real. we must make use of the square modulus (Eq. b has to satisfy four wellknown properties. without loss of generality. since  ψ1  ψ2 2 < 1. Expanding in power series to the ﬁrstorder in sin φ for φ dfs 2α sin φ = d. Then.
9) ∞ ℘e (t. perform the square modulus of Eq. Making the inverse Fourier transform of Eq.71) are orthogonal. αi ) = 2π 2 +∞ π dζ −∞ 0 dθ ζ e−ı √ 2ζ (αi sin θ+αr cos θ) χ℘ (ζ .1 If you do not succeed. which then turns to 1 W (αr . θ ) = d X ℘(X .72). 9. (15. θ ). and inserting this into the previous equation. αi ) = 2 π +∞ dξr dξi e2ı(ξr αi −ξi αr ) χ (ξr . we ﬁnally obtain 1 W (αr . since all the basis states in Eq. We start from Eq. χ℘ (ζ . (15.64).7 It is easy to prove that (see also Prob.61). 2008 18:21 Page73 t 73 S o l u t i o n s to se l e c te d p ro b l e m s Chapter 15 15. from which the solution follows.9 First of all. θ )eıζ X .9780521869638sol CUP/AUL November 6.21) ˆ ˆ ˆ −1 ˆ ˆ x p Dx p ψ n = Dx p ˆ Dx p Dx p ψ n ˆ = Dx p ˆ ψ n ˆ = Dx p (−1)n ψ n n = (−1)n ψx p .11 We sketch the main steps of the derivation. 2 we also have dξr dξi = 1 ζ dζ dθ . 15. αi ) = 2π 2 +∞ +∞ π 0 dX −∞ −∞ dζ dθ ζ ℘(X . −∞ Changing the variables according to 1 ξr = − √ ζ sin θ 2 and 1 ξi = √ ζ cos θ . 2 . using Eq. θ )e √ ıζ X − 2(αr cos θ+αi sin θ) . please refer to the quoted specialized literature. 13.59). First. and. (15. φ) = n=0 ψe. (15. so as to obtain (see Sec. (15. we note that the exponential factor e−ξ α +ξ α may be rewritten as e2ı(ξr αi −ξi αr ) .n (t) . 15. we may rewrite the W2 function as 1 W (αr . that is. ξi ). ∗ ∗ 15.
. ( z ∈ C. 16. (6. 2 The expectation value on the singlet state (16.27). 1 + cos 2θ . and by the fact that each Pς j cannot be negative. n=0 ı z − z ∗ = −2 (z).29) vanishes. c2 .1 From Eqs.8 The problem is solved when considering that (see Eqs. we ˆ immediately have that for any j = k all Pς j ˆ 16. with arbitrary c1 .4 Let us take  ς =  ςk as an element of a complete set of orthogonal vectors { ς j }.g.  ψ1 as the linear combination  ψ1 = c1  ς1 + c2  ς2 .154)) ˆ σ 1 · a = ax and ˆ σ 2 · b = bx 0 1 1 0 + by 2 0 1 1 0 + ay 1 0 −ı ı 0 0 −ı ı 0 + az 1 1 0 0 −1 1 0 0 −1 1 + bz 2 . then the rhs of Eq.12) of the product of these two scalar products gives nine terms. by Eq. x2 ) = 2π +∞ ¯ dpe h (x1 −x2 +x0 ) p ı −∞ +∞ h h = √ δ(x1 − x2 + x0 ) = √ 2π 2π = 1 2π +∞ +∞ d xδ(x1 − x)δ(x − x2 + x0 ) −∞ ı d xδ(x1 − x) −∞ +∞ −∞ ¯ dpe− h (x−x2 +x0 ) p 1 =√ 2π d xψx (x2 )ϕx (x1 ). 2 sin θ cos θ = sin 2θ . Writing. (16. which in turn ϕ = 0. ˆ Then. ˆ = Pς2 ϕ ˆ implies that Pψ1 ˆ = Pψ2 ϕ = 0. of which the ﬁrst three have the form . (16. make use of the following facts: ∞ cn+1 2 = 1 − c0 2 .3)–(16.5) we have 1 (x1 . we obtain the desired result. cos2 θ = 2 Chapter 16 16. 2008 18:21 Page74 t 74 S o l u t i o n s to se l e c te d p ro b l e m s Then. (16.6 Since Pς1 ϕ ϕ ϕ vanish. e.9780521869638sol CUP/AUL November 6. −∞ 16.
9780521869638sol CUP/AUL November 6. Indeed. we have 0 ax b y σ1x σ2y ˆ ˆ 0 = 0 ax b y √ 2 1 0 0 1 ⊗ 1 0 1 1 ⊗ 1 −ı 0 2 − 2 1 0 1 ⊗ 1 0 ı 2 2 ax b y = 2 × ⊗ 0 1 −ı 0 − 2 − 0 1 1 0 ⊗ 1 ⊗ 1 0 0 ı 2 ax b y [−(1 · ı) − (1 · (−ı))] = 0. 2008 18:21 Page75 t 75 S o l u t i o n s to se l e c te d p ro b l e m s 0 ax bx σ1x σ2x ˆ ˆ 0 = × = = 0 ax bx √ 2 1 0 0 1 1 0 ⊗ 1 1 0 1 1 0 − 2 2 0 1 0 1 0 1 1 0 ⊗ 1 1 0 1 0 2 ax bx √ 2 0 −a x bx  0 ⊗ 1 − 2 ⊗ 1 0 1 2 0 = −ax bx . 16. dλρ(λ) + − dλρ(λ) = 1. 16.b = = Since + d x  f (x) .9 The solution is obtained when one considers that. for any integrable function f (x). .65). we have d x f (x) ≤ and Ab (λ)Ac (λ) − 1 = 1 − Ab (λ)Ac (λ) 16. = 2 and similarly for all other cross terms. we immediately obtain the desired result. dλρ(λ)Aa (λ)Bb (λ) = 1 − δ + dλρ(λ) − − dλρ(λ). so that we may ﬁnally conclude that 0 ˆ ˆ σ 1 ·a σ 2 ·b 0 = −(ax bx + a y b y + az bz ) = −a · b.11 Making use of Eq. (16. Similar calculations show that we also have 0 a y b y σ1y σ2y ˆ ˆ 0 = −a y b y and 0 az bz σ1z σ2z ˆ ˆ 0 = −az bz .10 We have that a . we have that dλρ(λ)Bb (λ)Bb (λ) = + dλρ(λ)Aa (λ)Bb (λ) − − dλρ(λ)Aa (λ)Bb (λ). The six cross terms are instead all zero.
b  ≤ which amounts to  a. dλρ(λ)Aa (λ)Bb (λ) − dλρ(λ)Aa (λ)Bb (λ) − 2 dλρ(λ)Aa (λ)Bb (λ).9780521869638sol CUP/AUL November 6. b = −1.55) as a. dλρ(λ) [Aa (λ)Bb (λ) (1 ± Aa (λ)Bb (λ))] dλρ(λ) [Aa (λ)Bb (λ) (1 ± Aa (λ)Bb (λ))). b − a. b  ≤ 2 + a . 16. . 2008 18:21 Page76 t 76 S o l u t i o n s to se l e c te d p ro b l e m s which together with the fact that + dλρ(λ)Bb (λ)Bb (λ) = dλρ(λ)Aa (λ)Bb (λ) − − dλρ(λ)Aa (λ)Bb (λ). these results. which implies both dλρ(λ) [Aa (λ)Bb (λ) (1 ± Aa (λ)Bb (λ))] ≤ dλρ(λ) [Aa (λ)Bb (λ) (1 ± Aa (λ)Bb (λ))] ≤ we also have  a.69) in its formulation:  a. b . together with Eqs. from which the CHSH inequality follows. b − a.66) and (16. (16. Aa (λ)Bb (λ).62). let us ﬁrst rewrite Eq. (16.13 By making use of Eq.12 Let us do the substitution a = b in inequality (16. give the desired solution. b . b − a. the result is easily obtained. allows us to derive dλρ(λ)Bb (λ)Bb (λ) = Now. 16.53). we also have that dλρ(λ)Aa (λ)Bb (λ) − 2 ≥ − dλρ(λ)Aa (λ)Bb (λ) − 2 − dλρ(λ)Aa (λ)Bb (λ). (16. b = − Since we have that [Aa (λ)]2 = 1 ≥ Aa (λ)Bb (λ). b − a. dλρ(λ) (1 ± Aa (λ)Bb (λ)) dλρ(λ) (1 ± Aa (λ)Bb (λ)). Since b . b + a . b ± a . b  ≤ 2 ± a . Since Aa (λ)Bb (λ) = ±1. dλρ(λ) (1 ± Aa (λ)Bb (λ)) + dλρ(λ) (1 ± Aa (λ)Bb (λ)).
and φ3 . (16. (16.2 In the case of pure states.172) we obtain −1 ≤ π12 (λ. yk = πk (λ. for instance. which represents the probability that the three particles are detected at D1. D2. a. b }. (16. This proves the result because the logarithmic function of 1 is 0. 2008 18:21 Page77 t 77 S o l u t i o n s to se l e c te d p ro b l e m s 16. (17. a. (16. which may be rewritten as Eq.9780521869638sol CUP/AUL November 6.73). φ2 . φ3 ) = 1/4. b) − π12 (λ.8 Applying one of the unitary operators (17. b ) + ℘12 (a . we recover the information contained in particle 1. which is independent of the phases φ1 .75). yields with the probability distribution − 1 ≤ ℘12 (a. b ) + π12 (λ. a ) − π2 (λ.144b) and obtain ℘D2 D3 (φ1 . (16. (16. which may be rewritten in the form c Indeed. a . l). In order to test if the twoparticle interference is present. and D3. we have to consider the probability that two detectors (say D2 and D3) click independently of what happens in the third arm of the inteferometer. Chapter 17 ˆ 17. and X = Y = 1. namely wk = 1. b ) − π1 (λ. with j = {a. (17.4) we have wk δ jk . In this case.74). where in Eqs. we have to sum Eqs. and using Eqs. b) ≤ 0. we have ρ = Pk =  bk bk .71). b) + ℘12 (a . b ) − ℘1 (a ) − ℘2 (b) ≤ 0. 3 3 3 .171) we have taken xk = πk (λ. 16. a }. which means that in ˆ Eq.143)) is evidenced by the sine term in Eq. 17.144a) and (16. a .76). −1 0 −1 0 0 1 0 −1 0 −1 3 0 1 3 1 0 3 1 0 3 −c −c −c c c c −c c ⎫ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎬ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎪ ⎭ 1 0 +c 1 0 1 = 1 c c . (16. b) + π12 (λ. φ2 .144a). with l = {b. with the consequence that the diagonal matrix has only an element different from zero.19 The threeparticle interference in the GHSZ state (Eq. j).73) to the corresponding state of particle 3 as represented in Eq. 1 3 3 = c c . and (16.14 From Eqs. b) − ℘12 (a. Integrating this inequality over ρλ . (16.
we evaluate each of the four components of the state above for each of the four possible Boolean functions f j ( j = 1. . even if the time required for a division is very small (say. ( 1 −  0 ). . . U4 – corresponding. ˆ 17. yaxes – in order to convert the state of 3 into the to 180 state of 1. 4). i. f1 : f2 : f3 : f4 :  00 −  01 +  10 −  11 =  0 Ê +  1  01 −  00 +  11 −  10 =  0 Ê +  1  00 −  01 +  11 −  10 =  0 Ê −  1  01 −  00 +  10 −  11 =  0 Ê −  1 ( 0 −  1 ). it is subexponential). the best known classical 1/3 algorithm takes a time of the order of 10l (i. so that Bob need do nothing further to reproduce the state of 1. 17.e. What Bob must do.10 The most direct algorithm that can be used to factorize an integer number of l digits consists of checking whether the integer is divisible by each number from 2 to √ √ M = 10 l . ( 0 −  1 ). As a matter of fact. if l us equal to 100 decimal digits. this algorithm requires a number of steps given by √ M = 10 l .109) on which we want to evaluate the Boolean function may be rewritten as 1 1 ( 0 +  1 ) ( 0 −  1 ) = ( 00 −  01 +  10 −  11 ). For example. N which shows that the growth is exponential.12 The distance between ρ 1 and ρ 2 is given by ˆ d(ρ 1 . ρ 2 ) = ˆ ˆ 1 Trρ 1 − ρ 2  ˆ ˆ 2 1 ˆ = Tr (r − s) · σ  4 1 = r − s. x. in the example above we have not chosen the most efﬁcient algorithm. 2 ˆ since (r − s) · σ has eigenvalues ±r − s. ◦ rotations around the z. 10−20 s). the state of particle 3 is the same as of particle 1 except for an irrelevant phase factor. 2 2 Now. ( 1 −  0 ).e. . In the three other cases. so that we obtain ˆ Tr (r − s) · σ  = 2r − s. obviously depends on the (classical) communication of Alice’s result.e. the total time required to factorize the original number is τ 1010 × 10−20 = 1030 s. Of course.11 The state (17. Therefore. This leads to the desired solution. 2008 18:21 Page78 t 78 S o l u t i o n s to se l e c te d p ro b l e m s In the case of the ﬁrst measurement outcome (the singlet one). U3 . i. Bob ˆ ˆ ˆ must apply one of the unitary operators U2 . 17. 1013 times the age of the universe. . respectively.9780521869638sol CUP/AUL November 6.
ˆ ˆ . ˆ ˆ ˆ ˆ ˆ ˆ d(ρ 1 . U ρ 2 U † ). we have the important result that the distance between two single qubit states is equal to one half the Euclidean distance between them on the Bloch sphere.e. 2008 18:21 Page79 t 79 S o l u t i o n s to se l e c te d p ro b l e m s In other words. unitary transformations preserve it. i.9780521869638sol CUP/AUL November 6. ρ 2 ) = d(U ρ 1 U † . Since rotation on the Bloch sphere leaves this distance unaffected.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue listening from where you left off, or restart the preview.