0% found this document useful (0 votes)

53 views201 pages

Solutions

This document is a solutions manual for a graduate course on Quantum Mechanics, detailing various concepts and calculations related to quantum theory. It includes discussions on photon energy, momentum conservation, and the application of quantum mechanics in different scenarios, such as the two-slit experiment and the behavior of electrons at room temperature. The manual also covers mathematical principles relevant to quantum mechanics, including vector operations, Hermitian matrices, and the properties of distributions.

Uploaded by

Getachew Tesfaye

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

0% found this document useful (0 votes)

53 views201 pages

Solutions

Uploaded by

Getachew Tesfaye

We take content rights seriously. If you suspect this is your content, claim it here.

Available Formats

Download as PDF, TXT or read online on Scribd

You are on page 1/ 201

Quantum Mechanics, a graduate course

Solutions manual

HORATIU NASTASE
Chapter 0

1) Doing the integral over Bν in the formula for P , we find

2h ∞
Z π
ν3
Z
P = 2 dν hν 2π sin θdθ | cos θ|. (0.1)
c 0 e kB T − 1 0

Defining x = hν/(kB T ), we find

Z ∞ Z 1
2k 4 x3 2k 4 2k 4 π 4
P = 3 B2 T 4 dx x 4π d(cos θ)(cos θ) = 3 B2 T 4 Γ(4)ζ(4) · 1 = 3 B2 T 4 .
h c 0 e −1 0 h c h c 15
(0.2)
2) The photon has energy Eph = hν, and is split between the kinetic energy of
the photoelectric electron, and its binding energy W . By the kinetic theory of gases
2
(applied, by assumption, to the electron), mv 3
2 = 2 kB T . Thus,

mv 2 3
Eph = + W = kB T + W ⇒
2 2
hν − W hν
T = ∼ , (0.3)
3kB /2 3kB
where in the last ∼ we used the assumption of W of the same order as hν. But
νvisible ∼ 4−8×1014 Hz , h ' 6.6×10−34 m2 Kg/s , kB ' 1.4×10−23 m2 Kg/(s2 K) ,
(0.4)
so
4 − 8 6.6
T ∼ × 1000K ∼ 7 − 14 × 1000K. (0.5)
3 1.4
3) The photon has momentum pγ = h/λ and energy Eγ = hc/λ, while the
2
electron has initially negligible
p momentum and (rest) energy me c , and after has
0 0 2 2 2
momentum pe and energy (pe c) + (me c ) .
Conservation of energy is then
2
2 0
p 0 2 hc hc 2
0 2 2 2
Eγ + me c = Eγ + (pe c) + (me c ) ⇒ (pe c) = − 0 + me c − (me c2 )2 ,
λ λ
(0.6)
and conservation of momentum is
p~0e = p~γ − p~0γ ⇒ p02 2 2 0 2 0 2
e c = (pγ c) + (pγ c) − 2pγ pγ c cos θ. (0.7)
Identifying the two ways of calculating p0e , we find
1 1 h 1 h
− 0 = 0
(1 − cos θ) ⇒ λ0 − λ = 2 sin2 θ/2. (0.8)
λ λ me c λλ me c
3
4 Chapter 0

4) For a relativistic wave, ω = 2πν and k = 2π/λ = 2πν/c. If the two-slit

experiment has a distance a between the slits and is at a distance d a from
the screen, and we consider interference at a distance x from the midpoint on the
screen, we see that the difference in distance traversed by the the two waves is
l2 − l1 = ∆l ' aθ, but then θ is approximately the same as the angle made by the
interference point from the slits, i.e., θ ' x/d. In between two maxima or minima,
the difference in distance travelled is of λ, so
d cd
δx = δ∆l = . (0.9)
a νa
5) Room temperature is about Troom ∼ 300K. An electron at room temperature
would have, by kinetic gas theory,
v2 3 h h
me = kB T ⇒ λ = =√ . (0.10)
2 2 me v 3kB Troom me
Since me ' 9 × 10−31 Kg, we find
6.6 × 10−34
λ' √ m ' 0.6 × 10−8 m. (0.11)
3 × 1.4 × 10−23 × 300 × 9 × 10−31
On the other hand, visible light, with ν ∼ 4 − 8 × 1014 Hz, has
c 3
λ= ∼ × 10−6 m. (0.12)
ν 4−8
But since the accuracy of the microscope is of the order of λ, we see that a
microscope with room temperature electrons is still better than one with visible
light (has a better accuracy).
6) We could indeed apply Bohr-Sommerfeld quantization, since we have a peri-
odic motion in phase space. Defining as φ the angle around the circle traced by the
circular pendulum as it moves, we would have
Z
pθ dθ = lh. (0.13)

However, this is not very useful, since the circular pendulum is a classical object,
so the angular momentum is very large, so l is well approximated by being contin-
uous. If however we have a quantum system of the same kind, like for instance a
precession motion in a quantum system, then it would be useful.
7) Light is a relativistic system, so we would need to use Quantum Field Theory.
Since we have Quantum Mechanics, in some sense yes, we should use a form of the
Schrödinger equation, but really only the Quantum Field Theory formalism does
it.
1 Chapter 1

1) The triangle inequality for ~c = ~a + ~b is ~c2 ≤ ~a2 + ~b2 . Generalizing to vectors

|vi = |wi + |ui, we find
hv|vi ≤ hw|wi + hu|ui. (1.1)

2) Considering ~a · ~b 6= 0, ~a · ~c 6= 0, ~b · ~c 6= 0, we first normalize ~a,

~a
~a0 = √ . (1.2)
~a02
Then we orthogonalize ~b to it,
0
~˜b = ~b − (~a0 · ~b) ~a , (1.3)
~a 02

˜
so that ~a0 · ~b = 0, then normalize it,
~˜
~b0 = qb . (1.4)
~˜b2

Then we orthogonalize ~c to ~a0 ,

0
~a
~c˜ = ~c − (~a0 · ~c) 02 , (1.5)
~a
so that ~c˜ · ~a0 = 0, then orthogonalize it to ~b0 also,

˜ ~b0
~c˜ = ~c˜ − (~b0~c˜) , (1.6)
~b02
˜ ˜
so that ~c˜ · ~b0 = 0, as well as ~c˜ · ~a0 = 0, and finally normalize it,
˜
~c˜
~c0 = q . (1.7)
˜
~c˜2
3) If
X X
Â = |aiha0 | , B̂ = |bihb0 | , (1.8)
a,a0 b,b0

5
6 Chapter 1

4) Consider Hermitian matrices A = A† , B = B † . Then

[Tr(AB)]† = Tr(B † A† ) = Tr(BA) = Tr(AB) , (1.11)
so indeed, Tr(AB) is real.
5) Consider Â acting on V and B̂ acting on V 0 . Then their tensor product is
! 
X X X X
Tr(A⊗B) = (A⊗B)ij,ij = Aii Bjj = Aii  Bjj  = TrV A·TrV 0 B.
(ij) (ij) i j
(1.12)
6) If Û is unitary, Û −1 = Û † , U ∈ U (N ), we can have det U 0 = 1, in which case
we have U 0 ∈ SU (N ). But for U ∈ U (N ), we saw that | det U | = 1, so det U = eiβ .
Then, defining α = β/N , with N the rank of the group, we find that U 0 = U/eiα
has det U 0 = 1. So indeed U = U 0 · eiα , as we wanted.
0 0
Moreover, eia form a group, since eiα · eiα = ei(α+α ) , and [eiα ]−1 = e−iα .
7) For a general 2 × 2 matrix

a b
A= , (1.13)
c d
the eigenvalue equation is
det(A − λI) = 0 ⇒ (a − λ)(d − λ) − bc = 0 , (1.14)
with solutions
s 2
a+d a+d
λ1,2 = ± − (ad − bc). (1.15)
2 2
2 Chapter 2

1) We discretize the product function (f · g)(x) as (f · g)(xi ). In a discrete and

complete basis |xi i, we have
n
X
|f · gi = f · g(xi )|xi i. (2.1)
i=1

If we defined (f · g)(xi ) ≡ hxi |f · gi, then we have

n
X
|f · gi = |xi ihxi |f · gi. (2.2)
i=1

We can compare this with

n
X n
X
|f i = |xi ihxi |f i , |gi = |xi ihxi |gi. (2.3)
i=1 i=1

If we understand the product function as product of the functions at the point,

||F ||2 = hF |F i = hf |f ihg|gi. (2.4)

With integrals (on wave functions), i.e. acting on |xi ⊗ |x0 i, we have
Z b Z b0
2
||F || = 2
|f (x)| dx |g(x0 )|2 dx0 . (2.5)
a a0

3) The delta function is the distribution

δx0 : D → R , hδx0 |φi = φ(x0 ). (2.6)

If we try to define a distribution δx20 : D → R, there is no way to define it to be

finite, yet different than δx0 , so the answer is no. Indeed, then we would have

hδx20 |φi = φ(x0 ) OR φ(x0 )δx0 (x0 ) = ∞ , (2.7)

and the first case is the same as δx0 , the second is infinite.
4) The action of δ 00 (x − y) is found from the definition of the derivative of a
distribution:
∂ ∂ ∂ ∂ ∂2
h δ(x − y)|φi = −h δ(x − y)| φi = +hδ(x − y)| 2 φi. (2.8)
∂x ∂x ∂x ∂x ∂x
7
8 Chapter 2

Then we find Z
dxδ 00 (x − y)φ(x) = φ00 (y). (2.9)

5) The function of an operator is defined from the Taylor expansion, as

∞
X f (n) n
f (Â) = Â , (2.10)
n=0
n!

where f (n) is the n’th derivative at zero, and is a real number for a real function.
Then
∞
h i† X f (n) n †
f (Â) = (Â ) = f (Â† ) = f (A) , (2.11)
n=0
n!
Ân
P∞
so is Hermitian. Examples are eÂ = n=0 n! , sin Â, cos Â.
6) According to the previous exercise,

eiÂ = cos Â + i sin Â , (2.12)

and cos Â, sin Â are Hermitian, so can be diagonalized. Assume that Â is diagonal-
ized to λi eigenvalues. Then

eiÂ = (cos Â)i + i(sin Â)i = cos λi + i sin λi = eiλi . (2.13)
i

7) To find the eigenvalues of σ1 , we write the secular equation,

−λ 1
det(σ1 − λ 1l) = 0 ⇒ = λ2 − 1 = 0 , (2.14)
1 −λ
so are λ1,2 = ±1. That means that
+1
1 0 e 0
(σ1 )diag = ⇒ (eσ1 )diag = . (2.15)
0 −1 0 e−1
3 Chapter 3

1) We need
(x−x1 )2 (x−x2 )2
− −
P1 = |ψ(x1 )|2 ∼ e P2 = |ψ(x2 )|2 ∼ e
2
σ1 2
σ2
, , (3.1)
and P = P1 + P2 . That is always true, provided that the particles are independent
(so P1 is independent of P2 ).
2) In H = αp2 + βq 2 + pf (q), the first two terms are already symmetrized
with respect to ordering, only the last term remains. If f (q) = n αn q n , the fully
P

symmetrized form (or ”Weyl ordered”) of the last term is

1 n
pq + qpq n−1 + ...q n p

n+2
1
= pq n + 0 + 1 · Kq n−1 + 2 · Kq n−1 + ... + n · Kq n−1 ,

(3.2)
n+2
where in the last form we commuted the terms to put all of them in the form pq n ,
using [Q, P ] = K. That means that we have
1 n 1 + 2 + ... + n
pq n = pq + qpq n−1 + ...q n p − Kq n−1

n+2 n+2
1 n n−1 n
n−1 n(n + 1)
= pq + qpq + ...q p − Kq . (3.3)
n+2 2(n + 2)
Substituting in the classical Hamiltonian, we get the quantum Hamiltonian
X n(n + 1)
Ĥ = ĤWeyl ordered − K αn q̂ n−1 . (3.4)
n
2(n + 2)

3) If the spectrum extends a finite amount, then there is at least one accumu-
lation point, so En+1 − En → 0. If Ek+1 − Ek has a monotonic sign ∀k, then this
accumulation point is one of the boundaries of the spectrum, Emin or Emax . If
Emax = +∞ or Emin = −∞, that is not necessary, so there is no needed condition.
4) If there is a single spinless particle in a potential, the observables (commuting
operators) are energy (Ĥ) and positions xi (X̂i ), i = 1, 2, 3.
5) If Ĥ = x2 , the Schrödinger equation ~i ∂t ψ = Ĥψ , with
i
ψ(t, x) = e− ~ Et ψ̃(x) (3.5)
gives the time-independent Schrödinger equation Ĥ ψ̃(x) = E ψ̃(x), which doesn’t
have a solution.
6) If |xa,b i ≡ |2i ± |3i, then
|ai + |bi |ai − |bi
|2i = , |3i = . (3.6)
2 2
9
10 Chapter 3

The average value of the observable Â is

1 |ai + |bi ha| + hb|
Â = Tr(ρ̂Â) = A11 + ha| |Â|ai
3 2 2
|ai − |bi ha| − hb|
+hb| |Â|bi
2 2
1 1 1
= A11 + Aaa + Abb . (3.7)
3 2 2
7) Besides |ai, |bi, there should be other eigenstates, in total |ii, so we can write
X
Ĥ = λi |iihi|. (3.8)
i

Then
ρ̂Ĥ = λ1 |1ih1| + λ2 |2ih2| + λ3 |3ih3| = Ĥ ρ̂. (3.9)
Therefore [ρ̂, Ĥ] = 0, so ρ̂ doesn’t evolve in time.
4 Chapter 4

1) It is easy to see that the simplest possibility is of the type

γi = σi ⊗ σa , i = 1, 2, 3 , γ4 = 1l ⊗ σb , (4.1)

where σa and σb are chosen among the σi , and differ from each other. Then, since
σi2 = 1l, we also have γi2 = σi2 ⊗ σa2 = 1l ⊗ 1l = 1l4 . Also, σa2 = σb2 = 1l means that
also γ42 = 1l. Moreover, since {σi , σj } = 2δij (so that also {σa , σb } = 0),

{γi , γj } = {σi , σj } ⊗ σa2 = 2δij ⊗ 1l = δij 1l4

{γi , γ4 } = σi ⊗ {σa , σb } = 0. (4.2)

As an example, we can choose σa = σ1 and σb = σ2 .

i
2) The time independent Schrödinger equation is Ĥ ψ̃ = E ψ̃ (ψ = e− ~ Et ψ̃). Then
~ · B,
Ĥ = µB S ~ so
~
E = ±µB Bz , (4.3)
2
corresponding to the ψ̃ = | ↑i and | ↓i.
P3
3) For Ĥ = a0 1l + i=1 ai σi , the time-independent Schrödinger equation Ĥ ψ̃ =
E ψ̃ gives the eigenenergies as solutions of the secular equation

a0 − E + a3 a1 + ia2
det(Ĥ − E 1l) = 0 ⇒ det = 0. ⇒
a1 − ia2 a0 − E − a3
q
E = a0 ± a21 + a22 + a23 . (4.4)

4) From (4.52) in the text, the probability of oscillation is

1 − cos[(E2 − E1 )t/~] (E2 − E1 )t

p2 (t) = sin2 2θ = sin2 2θ sin2 . (4.5)
2 2~
But, since E2 ' E1 ' p ' E (note that p is the momentum of the incident
neutrino, so p is the same for E2 and E1 , so E22 − E12 = m22 − m21 ),

E22 − E12 m2 − m21 m2 − m21 ∆m2

E2 − E1 = = 2 ' 2 = , (4.6)
E2 + E1 E2 + E1 2E 2E
we have
∆m2 t
p2 (t) = sin2 2θ sin2 , (4.7)
4E~
the argument of the latter multiplied by c4 if we reinstate the c’s. Then, since
11
12 Chapter 4

for ultrarelativistic particles we have t = L/c, and using units natural for the
corresponding quantities, we have
∆m2 (eV 2 )c4 L(km)

p2 (t) = sin2 (2θ) sin2 × 1.267 . (4.8)
E(GeV )
5) From (4.60) in text, we have
∆m2 E2 + E1
E2 − E1 = ~Ω ' , = Ē → E , (4.9)
2E 2
so the unitary operator becomes
∆m2 ∆m2 2
!
− iEt cos2 θei 4E~ t + sin2 θe−i 4E~ t 2i sin θ cos θ cos ∆m
4E~
t
U (t) = e ~
2 ∆m2 ∆m2
.
2i sin θ cos θ cos ∆m
4E~
t
sin2 θei 4E~ t + cos2 θe−i 4E~ t
(4.10)
6) The state is
|φi = C [|1i ⊗ |1i + |0i ⊗ |1i + a|1i ⊗ |0i + |0i ⊗ |0i] , (4.11)
so the normalization condition is
1
1 = hφ|φi = C 2 [1 + 1 + a2 + 1] ⇒ C = √ . (4.12)
3 + a2
A state is not entangled if we can write it as |φi ⊗ |ψi, for some |φi and |ψi. We
immediately note that for a = 1, we can write the state as
|φi = C [(|1i + |0i) ⊗ (|1i + |0i)] , (4.13)
so it is not entangled. In general, we can write the state as
|φi = C [(|1i
+ |0i) ⊗ |1i + (a|1i
+ |0i) ⊗ |0i]

a 1
= C (|1i + |0i) ⊗ |1i + |1i + |0i α|1i , (4.14)
α α
so we see that we need a/α = 1 and 1/α = 1 to have a non-entangled state, therefore
indeed a = 1 is the only such case.
7) Consider the total density matrix
ρ̂0 = |φihφ| , (4.15)
and its trace over the states in the B subsystem,
ρ̂ = TrB ρ̂0 = C 2 TrB [(|1i ⊗ |1i + |0i ⊗ |1i + a|1i ⊗ |0i + |0i ⊗ |0i) ·
· (h1| ⊗ h1| + h0| ⊗ h1| + ah1| ⊗ h0|)] . (4.16)
The trace over B means B h0|ρ̂0 |0iB + B h1|ρ̂0 |1iB , giving
ρ̂ = C 2 [(a|1i + |0i)(ah1| + h0|) + (|1i + |0i)(h1| + h0|)]
= C 2 (1 + a2 )|1ih1| + (1 + a)(|1ih0| + |0ih1|) + 2|0ih0| .

(4.17)
Then the probability to be in state 1 in A, independently of the state in B, is
1 + a2
P1 = h1|ρ̂|1i = C 2 (1 + a2 ) = . (4.18)
3 + a2
13 Chapter 4

Then the probability increases monotonically,

dP1 1 1 + a2 2
2
= 2
− 2 2
= > 0. (4.19)
da 3+a (3 + a ) (3 + a2 )2
That means that the minimum probability is Pmin = 1/3, for a = 0, whereas the
maximum probability is Pmax = 1, for a = ∞.
5 Chapter 5

1) As was noted in the text,

∞
d
X dn
ea dx ψ(x) = an ψ(x) = ψ(x + a) , (5.1)
n=0
dxn

which is just the Taylor expansion of a function around a point. But then H1
becomes
Z
H1 = dxψ ∗ (x)ψ(x + a) , (5.2)

which is clearly nonlocal, as it depends on two points, x and x + a.

On the other hand, using the identity (extended from a constant to an operator)
Z ∞ Z ∞
1 1 d2
= dαe −αa
⇒ d2 = dαe−α dx2 , (5.3)
a 0 dx2 0

we can rewrite H2 as
Z ∞ Z
d2
H2 = dα dxψ ∗ (x)e−α dx2 ψ(x) , (5.4)
0

which is similar to the case of H1 , so it is clearly nonlocal as well.

2) The angular momentum L ~ = ~r × p~ is, in components, Li = ijk rj pk . Consid-
ering the generating functional depending on all the Li ,

F (~r, p~, t) = ~r · p~ + i Li (~r, p~). (5.5)

Then the variations of the phase space objects under the canonical transformation
generated by F are

δpi = j {ri , Lj }P.B. = j jkl {ri , rk pl }P.B. = j jkl rk δil = ijk j rk

δri = = j {pi , Lj }P.B. = j jkl {pi , rk pl }P.B. = j jkl pl (−δik ) = ijl j pl . (5.6)

3) We can rewrite the Lagrangian as

q̇12 q̇ 2
L= + q̇1 q̇2 + 2 − k(q1 + q2 )
2 2
(q̇1 + q̇2 )2
= − k(q1 + q2 ). (5.7)
2
Then there is only one variable really, q ≡ q1 + q2 , with Lagrangian
q̇ 2
L= − kq , (5.8)
2
14
15 Chapter 5

and momentum
∂L
p= = q̇ , (5.9)
∂ q̇
so with Hamiltonian
p2
H = pq̇ − L = + kq. (5.10)
2
The Poisson brackets of two functions, f and g, of phase space, are
∂f ∂g ∂f ∂g
{f, g}P.B. = − . (5.11)
∂q ∂p ∂p ∂q
1
Canonical quantization means (since {, }P.B. → i~ [, ]) the replacement

{q, p}P.B. = 1 → [q, p] = i~ (5.12)

and [q, q] = [p, p] = 0.

4) The total momentum is the sum of individual momenta in one dimension,
N
X
P = pi , (5.13)
i=1
√
where pi = ± 2mEi , while the total energy is
N N
X X p2i
E= Ei = . (5.14)
i=1 i=1
2m

If only the total energy E is fixed, we have a continuous degeneracy: Ei are

arbitrary, except for the constraint that their sum is E, so we have N − 1 free
parameters. On top of this, we have the 2N discrete degeneracy coming from pi =
√
± 2mEi .
The time evolution of the eigenstate is as usual when going from the time inde-
pendent to the time dependent Schrödinger equation,
i
|ψE (t)i = e− ~ Et |ψE (t = 0)i. (5.15)

5) For the Hamiltonian

p2
Ĥ = + αp + βx2 + γx3 , (5.16)
2m
the time independent Schrödinger equation Ĥψ = Eψ is
~2 ∂ 2

~ ∂ 2 3
− +α + βx + γx ψ(x) = Eψ(x). (5.17)
2m ∂x2 i ∂x
If β = γ = 0, we have
" #
√ α 2 mα2

1 ~ ∂
√ + 2m − − E ψ(x) = 0 , (5.18)
2m i ∂x 2 2
16 Chapter 5

so if we defined Ẽ = E + mα2 /2, and define ~i ψ(x) = pψ(x), we have

p √ α p
√ + 2m = ± E + mα2 /2 , (5.19)
2m 2
so the general solution is
i i
√
2m
h i
√ i
√ i
2m(E+mα2 /2) 2m(E+mα2 /2)
ψ(x, t) = e− ~ Et e− ~ 2 αx e+ ~ x + e− ~ x . (5.20)
d
6) In Cartesian coordinates, pi , i = 1, 2, 3 are integrals of motion, dt pi = 0, while
2
p
~
the energy is a derived integral of motion, since E = 2m .
In polar coordinates, in the plane made by O and the linear trajectory, since L ~ =
~r × p~, L = pθ is conserved, so the independent integrals of motion are L = pθ , pz , pr .
In spherical coordinates, pθ , pφ , pr are independent and conserved (integrals of
motion), while L ~ 2 = p~2 and E are derived ones.
θ,φ
Canonical quantization is [qi , pj ] = i~δij for all.
7) In one dimension, the free particle had general wave function
i
e− ~ Et + i x√2mE i
√
ψE (x, t) = √ αe ~ + βe− ~ x 2mE , (5.21)
2π~
so then in 3 dimensions, the wave function is
i
e− ~ Et i~p·~x
ψE,~p (~x, t) = e . (5.22)
(2π~)3/2
Then the probability to find the particle at point ~x is
1
P (~x, t) = |ψ(~x, t)|2 = . (5.23)
(2π~)3
6 Chapter 6

1 We use the momentum space wave function

and finally the Heisenberg uncertainty relation is

~2
h∆x2 ih∆p2 i = . (6.6)
4
2)
The propagator and initial wave functions are
x02
e ~ p0 x e− 2d2
r i
0 m im(x−x0 )2
U (x, x ; t) = e 2~t , ψ(x, t = 0) = p √ , (6.7)
2πi~t d π
so the wave function as a function of time is
Z
ψ(x, t) = dx0 U (x, x0 ; t)ψ(x; t = 0)
  r 2
r Z +∞  2 2
m 1 0
 1
0 imd i (p0 − mx/t)d
= p √ dx exp − 2 x 1 − − q 
2π~it d π −∞  2d
 ~t ~ 1 + imd
2
~t
)
2 2 2
(p0 − mx/t) d imx
− imd2
+
2
2~ 1 − ~t 2~t
p t 2
( )
2 imt x − m0 imx2
= q√ exp − 2~t 1 + i~t + 2~t . (6.8)
π d + i~t md
md2

But

im 1 im i~t
− =− 2 − 1+ ⇒
2~t 2d 2~t md2
p t 2 p t 2

im x − m0 imx2 1 x − m0

p0 t
− i~t
+ = − i~t
− ip 0 x − , (6.9)
2~t 1 + md 2 2~t 2d2 1 + md 2 2m
so
p0 t 2
( )
1 x−

2 m p0 t
ψ(x, t) = q√ exp − 2 i~t
exp −ip0 x− . (6.10)
i~t
π d + md
2d 1 + md2
2m

Then
( 2 " #)
2 1 x − pm0 t 1 1
ρ = |ψ(x, t)| = q exp − i~t
+ i~t
i~t
π d + md
i~t
d − md
2d2 1 + md 2 1 − md2
" 2 #
1 x − pm0 t
=√ q exp − , (6.11)
~2 t2 d 2 1 + ~ 2 t2
π d2 + m 2 d2 m 2 d4

so that the width of the Gaussian is

r
d ~2 t2
∆x = √ 1+ . (6.12)
2 m 2 d4
3) The momentum and angular momentum don’t commute,
[pi , Lj ] = [pi , jkl rk pl ] = i~ijl pl , (6.13)
19 Chapter 6

but in general we have the Heisenberg uncertainty relations

1 2
h(∆A)2 ih(∆B)2 i ≥ hψ|[Â, B̂]|ψi , (6.14)
4
so we have now (without sum over i and j)
~2 ijl l 2
h(∆pi )2 ih(∆Lj )2 i ≥ | hp i| , (6.15)
4
or, with sum over i and j,
XX
h(∆pi )2 ih(∆Lj )2 i ≥ ~2 |h~
p2 i|2 . (6.16)
i j

p2
4) Since H = 2m + αpx, we have that
[Ĥ, P̂ ] = αp̂ 6= 0 , (6.17)
so no, we cannot measure them simultaneously.
5) ∆E∆τ ≥ ~, but ∆E = 0.5eV , so
~
∆τ ≥ = τmin . (6.18)
0.5eV
6) The wave function is
 
(x−x0 )2 (x−x0 )2
− 2 − 2
2σ1 2σ2
C i p0 (x−x0 )  e e
ψ = C(ψ1 (x) + ψ2 (x)) = e~  √ + √  , (6.19)

π 1/4 σ1 σ2

R +∞
which we need to normalize, −∞
ψ ∗ (x)ψ(x) = 1, which leads to
1
2C 2 =  −1/2  . (6.20)
1 1
σ12 + σ2
2 1 + √ 2 
σ1 σ2

For simplicity we take the origin at 0 (x0 = 0), and find

Z +∞ Z +∞
hxi = dx x|ψ(x)|2 = dx xC 2 [|ψ1 (x)|2 + |ψ2 (x)|2 + ψ1 ψ2∗ + ψ2 ψ1∗ ] = 0
−∞ −∞
(6.21)
which vanishes because of the symmetry of the integral. Then
 3/2 
1 1
Z +∞
σ 2
σ 2
σ12
+ σ2
hx2 i = dx x2 |ψ(x)|2 = C 2  1 + 2 + √ 2  , (6.22)

−∞ 2 2 σ1 σ2

so that the width in x squared is

 3/2 
1
2
σ22 + σ12
2  σ1 σ12
h(∆x)2 i = hx2 i − hxi2 = C  + + √ 2 . (6.23)

2 2 σ1 σ2
20 Chapter 6

For the momentum variable,

Z +∞ Z +∞
∗ d ∗ ψ1 ψ2
hpi = −i~ dx ψ (x) ψ(x) = dx ψ (x) p0 ψ(x) + i~x + 2
−∞ dx −∞ σ12 σ2
= p0 + 0 , (6.24)

and

d2
Z
hp2 i = (−i~)2 dxψ ∗ (x) 2 ψ(x)
( dx
p0 + i~x p0 + i~x
Z +∞ 2 " #)
ψ22

∗ 2 2 ψ1 2 σ12 σ22
= dx ψ (x) p0 ψ(x) + ~ + 2 +~ x ψ1 + ψ2
−∞ σ12 σ2 σ12 σ22
Z +∞
2 2 2 ∗ ∗ ψ1 ψ2 2 ψ1 ψ2
= p0 + C ~ (ψ1 (x) + ψ2 (x)) 2 + 2 + i~x + 4 , (6.25)
−∞ σ1 σ2 σ14 σ2

and the momentum width squared is

1 1 8
h(∆p)2 i = hp2 i − hpi2 = C 2 ~2 + 2+√
σ12 σ2 πσ1 σ2
3/2 
1 + 1 
1 1 2
σ1 σ 2 i~ 1 1 
+i~ 4 + 4 √ 2 + 2 + 2 . (6.26)
σ1 σ2 σ1 σ2 2 σ1 σ2 

Then we see that indeed

~2
h(∆x)2 ih(∆p)2 i ∼ + ... , (6.27)
4

satisfying Heisenberg’s uncertainty relations.

7) Introducting time dependence, we have
Z
ψ(x, t) = dx0 U (x, x0 ; t)ψ(x0 , t = 0) = C[ψ1 (x, t) + ψ2 (x, t)]
 
2

1 (x − p 0 t/m) ip0 p0 t
=r exp − 2 exp − x−
 
√ i~t 2σ 1 + i~t ~ m
π σ1 + mσ1 1 mσ 2
1
 
2

1 − (x − p 0 t/m)  exp − ip0 x − p0 t
+r exp ,
√ i~t

2σ 2 1 + i~t ~ m
π σ2 + mσ2 2 mσ2 2

(6.28)

and finally

σ12 ~2 t2 σ2 ~2 t 2

2
h(∆x) i = C 2
1+ 2 2 + 2 1+ 2 2
2 m σ1 2 m σ2
21 Chapter 6

 3/2 
 1 + 1  
2 2 2 2

σ12 1+ ~ 2t 2 σ22 1+ ~ 2t 2

m σ1 m σ2
+ r . (6.29)


2
~ t 2 2
~ t 2 
σ1 σ2 1 + m2 σ2 1 + m2 σ2  
1 2
7 Chapter 7

1) In case I, for energies −V0 ≤ E ≤ 0, we have a discrete spectrum, since there

are 2 boundary conditions on the wave function: the usual condition at infinity to
have the decaying solution ∝ e−kx (normalizability at infinity), and now also the
condition to have normalizability at 0, which selects one of two solutions. This arises
since the effective one-dimensional (Schrödinger) potential Veff has a component
∝ 1/r2 , so goes to infinity at 0. In case II, for energies 0 ≤ E ≤ Vmax , we have a
continuous and nondegenerate spectrum, with tunnelling solutions that can escape
all the way to infinity: we have only the condition of normalizability at 0. In case
III, for energies E > Emax , we have a continuous nondegenerate spectrum as well,
due to the single condition of normalizability at 0.
2) We can say that V0 → ∞ in the previous case. But then it depends on whether
the one-dimensional effective Schrödinger potential goes to +∞ as well, or not. If
so, the same as above. If instead, Veff → −∞, it is more complicated (see later
chapters), and the negative spectrum ends at a finite value of the energy.
3) If V = −V0 δ(x), then the Schrödinger equation is

~2 d2

− − V0 δ(x) ψ(x) = Eψ(x) , (7.1)
2m dx2

and integrating over an infinitesimal domain around x = 0, we obtain

~2 0 2mV0
− [ψ (0+) − ψ 0 (0−)] − V0 ψ(0) = 0 ⇒ ψ 0 (0+) − ψ 0 (0−) = − 2 ψ(0). (7.2)
2m ~
√
For a bound state, E < 0, ~2 k 2 = (−E), so k = −E/~, and the wave functions
on both sides of the singularity must be decaying, so

ψL (x) = Aekx , x < 0 , ψR (x) = Be−kx , x > 0 , (7.3)

and continuity of ψ(x) gives ψL (0) = ψR (0), so A = B. Then the condition (7.2)
implies
V0 m
k= . (7.4)
~2
√
The normalization condition, dx|ψ(x)|2 = 1 finally gives A = k. Then the
R

spectrum is
√
−E V0 m V 2 m2
k= = 2 ⇒E=− 02 . (7.5)
~ ~ ~
22
23 Chapter 7

4) The wave function for particle in a box is

q
 2 sin nπx n even
L L
ψ(x) = q (7.6)
 2 cos nπx n odd
L L

Then
Z +L/2
2 nπx
m
hx in even = dx xm sin2
L L
−L/2
0 m odd
= 2 R +L/2

2πnx +L/2
 L −L/2 dx x2k 1−cos2 L = 2 x2k+1 L2k+1
L 2(2k+1) − 2(πn)2k+1
I m = 2k ,
−L/2
(7.7)

where we have defined

Z +πn/2 Z +πn/2
sin 2θ
I≡ dθ θ2k cos 2θ = − dθ 2kθ2k−1
−πn/2 −πn/2 2
+πn/2
2k(2k − 1)
Z
cos 2θ
= 2kθ2k−1 + dθ cos 2θθ2k−2 = ...
4 −πn/2 4
πn 2k−1 k(k − 1) πn 2k−3
=k + (k − 1)
2 2 2
k(2k − 1) (k − 1)(2k − 3) πn 2n−5
+ (k − 2) + ... + 0. (7.8)
2 2 2
Then
(
m 0 m odd
hx in even = L2k L2k
(7.9)
22k (2k+1)
− (πn)2k+1
I m = 2k

and similarly
Z +L/2
m 2 nπx
hx in odd = dx xm cos2
L L
(−L/2
0 m odd
= L2k L2k
(7.10)
22k (2k+1)
+ (πn)2k+1
I m = 2k.

In the classical limit, n → ∞ and k finite, we obtain

πn 2k−1 L2k
I∼k + subleading ⇒ I→0, (7.11)
2 (πn)2k+1
so that we obtain the classical result
(L/2)2k
hx2k i ' . (7.12)
2k + 1
If m = 2k + 1,
Z +L/2
hpm in = d(P̂ k ψ ∗ )(P̂ k+1 ψ) = 0 , (7.13)
−L/2
24 Chapter 7

since one of the brackets is a sine, the other a cosine. On the other hand, if m = 2k,
Z +L/2 Z +L/2
m k 2 2k
hp in = dx|P̂ ψ| = ~ dx|ψ (k) (x)|2
−L/2 −L/2
nπ 2k Z +L/2 1 ± cos 2πnx
2k 2 L
=~ dx
L L −L/2 2
2k
nπ~
= = (2mEn )k , (7.14)
L
which is already its classical value.
5) For a cosine potential, V (x) = V0 cos(ax), we have −V0 ≤ V (x) ≤ V0 . For the
wave function, we have (at least) n − 1 nodes for the nth eigenfunction. That means
that if n is very large (as we will see we generically have), then the wave function
is highly oscillatory.
Even though @ limx→±∞ V (x), effectively we can define U± = V0 (the maximum
value of the oscillatory potential), since there will always be a place where V0 is
reached, no matter how far away in x we go. Moreover, for energies E < −V0 , there
is no solution, as usual, and for E > V0 , there is a continuum of solutions, also as
usual.
Naively, for −V0 < E < V0 , there will be a discrete spectrum, but! each local
minimum in the periodic potential will have its own discrete spectrum, and when
putting everything together, the various discrete spectra will interact, and the (ini-
tially identical) energy levels will split, obtaining energy bands.
6) To prove (7.83), we start from (7.82). Starting with the 4th equation, we
obtain
Q k2 − κ3
= e2ik2 b . (7.15)
P k2 + κ3
Then from the 3rd equation in (7.82), we obtain

(k2 + κ3 )(k1 − k2 ) + e2ik2 (b−a) (k2 − κ3 )(k1 + k2 )

R = e2ik1 a , (7.16)
(k2 + κ3 )(k1 + k2 ) + e2ik2 (b−a) (k2 − κ3 )(k1 − k2 )
and from the 1st equation in (7.82) we obtain

ei(k1 −k2 )a (k2 + κ3 )2k1

P = . (7.17)
(k2 + κ3 )(k1 + k2 ) + e2ik2 (b−a) (k2 − κ3 )(k1 − k2 )
Finally, from the 2nd equation in (7.82), we obtain

Q
S = e−iκ3 b P eik2 b 1 + e−ik2 b
P
ei(k2 −κ3 )b+i(k1 −k2 )a 4k1 k2
= , (7.18)
(k2 + κ3 )(k1 + k2 ) + e2ik2 (b−a) (k2 − κ3 )(k1 − k2 )
which means that
|S|2 κ3
T =
k1
25 Chapter 7

16k1 k22 κ3
= 2 .
[(k2 + κ3 )(k1 + k2 ) + (k2 − κ3 )(k1 − k2 ) cos 2k2 (b − a)] + (k2 − κ3 )2 (k1 − k2 )2 sin2 2k2 (b − a)
(7.19)
To compare with the formula in the text, note that
ξKL = k2 (b − a)
4k1 k22 κ3
4ηζξ 2 =
K4
2 2
k (k1 + κ3 )
ξ 2 (η + ζ)2 = 2
K4
2
(k + k1 κ3 )2
(ξ 2 + ηζ)2 = 2 , (7.20)
K4
and the rest is algebra.
To check |R|2 + T = 1, from the value of R above, we find
|R|2
2
[(k2 + κ3 )(k1 − k2 ) + (k2 − κ3 )(k1 + k2 ) cos 2k2 (b − a)] + (k2 − κ3 )2 (k1 + k2 )2 sin2 2k2 (b − a)
= 2 ,
[(k2 + κ3 )(k1 + k2 ) + (k2 − κ3 )(k1 − k2 ) cos 2k2 (b − a)] + (k2 − κ3 )2 (k1 − k2 )2 sin2 2k2 (b − a)
(7.21)
and a bit of algebra proves that indeed |R|2 +√T = 1.
√ √
7) Defining, as usual, κ = U0 − or k = − U , and k1 = − U1 , the wave
functions for > U1 are:
√ √
ψ(x) = ei x + Re−i x , x < 0
Seik1 x , x > L
ψ(x) = (
Aeκx + Be−κx U1 ≤ ≤ U0
ψ(x) = ikx −ikx
(7.22)
Ce + De ≥ U0 .
a) U1 ≤ ≤ U0 . In this case, continuity of ψ(x) at x = 0 and x = L gives
1+R=A+B ⇒R=A+B−1,
κL −κL ik1 L −ik1 L+κL B −2κL
Ae + Be = Se ⇒ S = Ae 1+ e . (7.23)
A
Continuity of the log derivative of ψ at x = L gives
ik1
AeκL − Be−κL B 1−
κ κL −κL
⇒ = e2κL κ
ik1
, (7.24)
Ae + Be A 1+ κ

and then at x = 0 gives, substituting the value of R and of B/A,

√ 1−R A−B
i =κ ⇒
1+R √ A+B
2 i κ
A= √ √
1 + i κ − BA 1− κ
i
√
2 i κ 1 + ikκ1

= √ √ . (7.25)
1 + i κ 1 + ikκ1 − e2κL 1 − ik1 i

κ 1− κ
26 Chapter 7

Then
2
B −2κL
T = |S|2 = e2κL |A|2 1 + e
A
√ 2
2
4e2κL κ
= √ 2 √ 2
k1 2 +k1 2
1− (1 − e−2κL ) +
κ κ (1 + e2κL )
1
= " √
2
#. (7.26)
√ 2 k
1− κ2 1
√ 2
+k
√ 1 2 +k1
2
+ sinh κL √
2 /κ
+ 2 √

Note that for k1 = (meaning, U1 = 0), we get

1
T = " 2 #, (7.27)
2 1− κ2
1 + sinh κL √
2 +1
κ

and
(κ2 − )2 U02
= , (7.28)
4κ2 4(U0 − )
so we reproduce the result in the text.
b) > U0 . In this case, continuity at x = 0 and x = L gives
1+R=C +D ⇒R=C +D−1
ikL −ikL ik1 L i(k−k1 )L D −2ikL
Ce + De = Se ⇒ S = Ce 1+ e . (7.29)
C
Continuity of the log derivative of ψ at x = L gives
k1
CeikL − De−ikL D 1−
ik ikL −ikL
= ik1 → = e2ikL k
k1
, (7.30)
Ce − De C 1+ k

and then at x = 0 gives, substituting the values of R and D/C,

√ 1−R C −D
i = ik ⇒
1+R √ C +D

2k
C= √ √
1 + k − D
C 1 − k

√
2 k 1 + kk1

= √ √ . (7.31)
1 + k 1 + kk1 − e2ikL 1 − k1

k (1 − k)

Then
2
D −2ikL
T = |S|2 = |C|2 1 + e
C
√ 2
4 2k
= √ 2 2 2 √ 2
k12
1 + kk1 + 1 − kk1 1 − k − 2 cos 2kL 1 −

1+ k k2 1− k2
27 Chapter 7

1
=
k2
. (7.32)
√ 2 1− k12 (1− k2 )
k1 +
√ 2
2
+ √ 2 sin kL
2 k

Note that for k1 = (U1 = 0), we obtain

1
T = 2 , (7.33)
1− k2
1+ √
2 sin kL
k

where
(k 2 − )2 U02
2
= , (7.34)
4k 4( − U0 )
reproducing the result in the text.
8 Chapter 8

1) For small q1 , q2 , the Lagrangian becomes

q̇12 q̇ 2 α1 2 2 α2 2 2 α12 2
L' + 2 − β q − β q − β (q1 + q2 )2
2 2 2 1 1 2 2 2 2 12
q̇ 2 q̇ 2 q2 q2 2q1 q2
= 1 + 2 − 1 (α1 β12 + α12 β12
2
) − 2 (α2 β22 + α12 β12
2
)− 2
α12 β12 . (8.1)
2 2 2 2 2
Defining
a ≡ α1 β12 + α12 β12
2
, b ≡ α2 β22 + α12 β12
2 2
, c ≡ α12 β12 , (8.2)
we want to diagonalize by writing
γ δ µ ν
L' (q̇1 + αq̇2 )2 + (q̇1 − β q̇2 )2 − (q1 + αq2 )2 − (q1 − βq2 )2
2 2 2 2
γ 2 δ 2 µ 2 ν 2
≡ Q̇1 + Q̇2 − Q1 − Q2 , (8.3)
2 2 2 2
and then quantize the independent harmonic oscillators in Q1 , Q2 as usual.
To diagonalize, we note that
aq12 + bq22 + c2q1 q1 = µ(q1 + αq2 )2 + ν(q1 − βq2 )2 (8.4)
implies
b − aβ 2
ν =a−µ, µ=
α2 − β 2
b − aβ 2 c + aβ
α=β+ ⇒µ= b−aβ 2
. (8.5)
c + aβ 2β + c+aβ

At this moment, α is fixed in terms of β (and the given constants a, b, c), but β
is free. However, we still need to impose that
q12 + q22 = γ(q1 + αq2 )2 + δ(q1 − βq2 )2 , (8.6)
which fixes δ and γ, but also gives an extra constraint,
α 1 1
δ= γ,γ= α = ⇒
β 1+ β α2 1 + β
α
α
α2 + αβ = 1 + . (8.7)
β
2) For the Hamiltonian
N h i
â†n ân + αâ†n+1 ân + h.c.
X
H= , (8.8)
n=1

28
29 Chapter 8

consider the Fourier transform of the creation and annihilation operators

N
1 X 2πijn
aj = √ an e N . (8.9)
N n=1
Then the kinetic harmonic oscillator term is left invariant, since
N N N N N
1 X 2πij(m−n)
a†j aj =
X X X X
a†n am e N = a†n am δn,m = a†n an . (8.10)
j=1 n,m=1
N j=1 n,m=1 n=1

On the other hand,

N N N N
1 X X † 2πin(j−k) 2πij 2πij
a†n+1 an = a†j aj e− N ,
X X
aj ak e N e− N = (8.11)
n=1
N n=1 j=1
j,k=1

so that finally
N
2πj
â†j âj 2 + 2α cos
X
H= . (8.12)
j=1
N

3) To prove the completeness relation

N
)
1 X ∗ ∗ 3 ∗
+ [α αi+1 + β(αi + αi+1 ) + β(αi + αi+1 )] . (8.16)
N i=1 i

6) For the potential V (x) = λx4 , the Schrödinger equation is

~2 d 2

4
− 2
+ λx ψn (x) = En ψn (x) ⇒
2m dx

2m
ψn00 (x) + n − 2 λx4 ψn (x) = 0 , (8.19)
~
2m
where we have defined as usual n = ~2 E n or, defining
r
E mω 2~
˜n = , y= x , λ̃ = λ 2 3 , (8.20)
~ω ~2 m ω
we have
ψn00 (y) + [2˜
n − λ̃y 4 ]ψn (y) = 0. (8.21)

We next consider the asymptotics of the equation:

a) For y → ∞, we obtain ψn00 (y) − λ̃y 4 ψ = 0, with solution
√ 3
y
ψn (y) ' An e± λ̃ 3 , (8.22)

and b) For y → 0, we obtain ψn00 (y) + n ψn (y) = 0, with solution

√ √
ψn (y) = Ãn sin y + B̃n y , (8.23)

but since we are at y → 0, it is actually

√
ψn (y) ' Ãn y + B̃n . (8.24)

Then the ansatz that eliminates the asymptotic behaviour is

√ 3
y
ψ(y) = e− λ̃ 3 H(y) , (8.25)
31 Chapter 8

which leads to the reduced equation

p p
H 00 (y) − 2 λ̃y 2 H 0 (y) + 2(˜
n − λ̃y)H(y) = 0. (8.26)
To write a quantization condition, we Taylor expand H(y) as
∞
X
H(y) = Cn y n , (8.27)
n=0

leading to the equation

X∞ h p p i
Cn n(n − 1)y n−2 − 2 λ̃ny n+1 + 2˜
n y n − 2 λ̃y n+1 = 0 , (8.28)
n=0

and, after redefining the sums in the various terms such that we have a common
factor y n , we get
∞
X p
y n [(n + 2)(n + 1)Cn+2 − 2 λ̃Cn−1 + 2˜
n Cn ] = 0. (8.29)
n=0

We can put each coefficient to zero, leading to the recursion relations

p
2 λ̃Cn−1 + 2˜ n C n
Cn+2 = . (8.30)
(n + 2)(n + 1)
√ 3
y
In order to avoid the appearance of the divergent solution e+ λ̃ 3 at infinity,
H(y) must cut off at a finite term, thus be a polynomial, so we must have some n
for which Cn+2 = 0. However, that gives the relation
p
λ̃Cn−1 + ˜n Cn = 0 , (8.31)
which is not clear how to explore further in order to obtain the quantization con-
dition on ˜n .
7) For two harmonic oscillators of frequencies ω and 2ω, with energy E = 15~ω/2,
since in general
E = (n1 + 1/2)~2ω + (n2 + 1/2)~ω = ~ω(2n1 + n2 + 3/2) , (8.32)
we can have the possibilities:
(n1 = 0, n2 = 6); , (n1 = 1, n2 = 4) , (n1 = 2, n2 = 2) , (n1 = 3, n2 = 0). (8.33)
That means that the symmetric wave function is

(1) (2) (1) (2) (1) (2) (1) (2)
ψ(x) = C ψ0 (x)ψ6 (x) + ψ1 (x)ψ4 (x) + ψ2 (x)ψ2 (x) + ψ3 (x)ψ0 (x) .
(8.34)
9 Chapter 9

1) For a one-dimensional particle in potential V (x), from (9.13) in the text, we get
the evolution operator
i t 0 ~2 d 2
Z
Û (t, t0 ) = T exp − dt − + V (x) . (9.1)
~ t0 2m dx2
But for a Hamiltonian that is time-independent, like this one, we get
~2
h i
− ~i (t−t0 ) − 2m +V (x)
Û (t, t0 ) = e , (9.2)

and the differential equation is

~2 d 2

d
i~ Û (t, t0 ) = Ĥ Û (t, t0 ) = − + V (x) Û (t, t0 ). (9.3)
dt 2m dx2
P̂ 2
2) For a free particle, the Schrödinger picture Hamiltonian is ĤS = 2m , so the
Heisenberg picture Hamiltonian is

i P̂ 2 P̂ 2 − i (t−t0 ) P̂ 2 P̂ 2
ĤH = Û † ĤS Û = e+ ~ (t−t0 ) 2m e ~ 2m = Ĥ
S = . (9.4)
2m 2m
Then the time evolution equation for a generic operator ÂH is (from eq. (9.28)
in the text)
dÂH ∂
i~ = [ÂH , ĤH ] + i~ ÂH . (9.5)
dt ∂t
We consider ÂH = X̂ and P̂ . In the first case, we obtain

dP̂H
i~ = 0 + 0 ⇒ P̂H (t) = P̂H (t0 ) = P̂S . (9.6)
dt
In the second case, we obtain (from (9.24) in the text)

dX̂H ∂ P̂S P̂H

i~ = Û † [X̂S , Ĥ]Û + i~ X̂H = Û † Û + 0 = ⇒
dt ∂t m m
P̂H P̂H
X̂H (t) = X̂H (t0 ) + t = X̂S + t. (9.7)
m m
†
3) In the picture with Ŵ (t) = e2iωâ â(t−t0 )
, the equation for âW (t) is
d
i~ âW (t) = iˆ(∂t â)W + [i~(∂t Ŵ (t))Ŵ † (t), âW (t)]
dt
= i~[∂t Ŵ (t)âŴ † (t) + Ŵ (t)â∂t Ŵ (t)] , (9.8)
32
33 Chapter 9

but (∂t Ŵ )Ŵ † = 2iωâ† â, so

d
âW (t) = −2~ω[â† â, Ŵ aŴ −1 ] = −2~ω [â† â, â] W = 2~ωâW (t) ⇒

i~
dt
d
âW (t) = −2iωâW (t) ⇒
dt
âW (t) = âe−2iωt . (9.9)

For a general operator, the evolution equation is

d
i~ ÂW (t) = −2~ω[â† â, ÂW (t)] + i~(∂t Â)W . (9.10)
dt
But from equation (9.42) in the text,
† †
ĤW = e2iωâ â(t−t0 )
~ω(â† â + 1/2)e−2iωâ â(t−t0 )
= ~ω(â† â + 1/2) = Ĥ. (9.11)

Then the Hamiltonian for evolution in the Ŵ picture is

Ĥ 0 = ĤW + i~∂t Ŵ Ŵ −1 = −~ω(â† â − 1/2). (9.12)

In turn, that means that the evolution of the state in the Ŵ picture is

i∂t |ψW i = Ĥ 0 |ψW i = −~ω(â† â − 1/2)|ψW i. (9.13)

4) For Ĥ = Ĥ0 + Ĥ1 , with

Ĥ0 = ~ω(â† â + 1/2) , Ĥ1 = λ(â + â† )3 , (9.14)

the interaction picture evolution equations for âI (t) and â†I (t) are
d
i~ âI (t) = [âI (t), Ĥ0 ] + 0 = Ŵ [â, ~ω(â† â + 1/2)]Ŵ −1
dt
= ~ωâI (t)
d †
i~ âI (t) = Ŵ [â† , ~ω(â† â + 1/2)]Ŵ −1 = −~ωâ†I (t). (9.15)
dt
The evolution of states is given by

i~∂t |ψI (t)i = Ĥ1,I |ψI (t)i = λŴ (â + â† )3 Ŵ −1 |ψI (t)i. (9.16)

But
† †
Ŵ (â† + â)3 Ŵ −1 = eiω(â â+1/2)(t−t0 ) (â + â† )3 e−iω(â â+1/2)(t−t0 )
= (â + â† )3 + [iω(â† â + 1/2), (â + â† )3 ] + ...
= (â + â† )3 − (a + a† ) + a3 − (a† )3 + a2 a† − (a† )2 a , (9.17)

so finally

i~∂t |ψI (t)i ' λ (â + â† )3 − (a + a† ) + a3 − (a† )3 + a2 a† − (a† )2 a |ψI (t)i. (9.18)

5) In the conservative (time-independent) case, the evolution operator is

i i
UI (t, t0 ) = e ~ Ĥ0 (t−t0 ) e− ~ Ĥ(t−t0 )
† † i † 3
= eiω(t−t0 )(â â+1/2) e−iω(t−t0 )(â â+1/2)− ~ λ(â+â ) (t−t0 ) . (9.19)
34 Chapter 9

6) For a Hamiltonian Ĥ(p) only, in the Heisenberg picture,

dÂH
i~ = [ÂH (t), ĤH ] + i~∂t ÂH (t) = Û † [ÂS , Ĥ(P̂ )]Û , (9.20)
dt
and is nonzero ⇔ [ÂS , P̂ ] 6= 0, so only for ÂS = f (X̂).
In the Dirac picture, we obtain
d
i~ ÂI = [ÂI (t), Ĥ0 ] , (9.21)
dt
and since we also have Ĥ0 (P̂ ), we obtain the same conclusion.
7) Because of eq. (9.44) in the text,
d
i~ ÂW = 0 ⇔ i~[∂t Ŵ · Ŵ −1 , ÂW ] = 0 ⇔ ∂t Ŵ · Ŵ −1 ∝ 1̂l , (9.22)
dt
which means that Ŵ = eiα̂ , with ∂t α̂ = 0, so ∂t Ŵ = 0. But if Ŵ is time-
independent, then the picture is equivalent to the Schrödinger picture, amounting
to just a change of basis.
10 Chapter 10

1) eiS = cos S + i sin S is highly oscillatory, so the (path) integral over it is not
well defined. To make sense, we must replace it with an honest Gaussian, e−SE , by
a ”Wick rotation” to Euclidean space. If iS = −SE , we see that we can achieve it
R t = −itE , Rsince in the Lagrangian L = T − V , the potential term becomes
with
i dt(−V ) = − dtE V , and if V ≥ 0 (as needed in well defined quantum systems),
then we have a decreasing exponential, as we wanted.
2) For a term Ĥ1 = α(P̂ X̂ 2 + X̂ P̂ X̂ + X̂ 2 P̂ ), in order to be able to use the same
formula (10.20) in the text,

hp(ti )|e−iĤ |x(ti−1 )i ' e−iĤ(p(ti ),x(ti−1 )) e−ip(ti )x(ti−1 ) , (10.1)

we need to have P̂ on the left in the Hamiltonian, so we write

Ĥ1 = 2αP̂ X̂ 2 + α(2[X̂, P̂ ]X̂ + X̂[X̂, P̂ ]) = 3α(P̂ X̂ 2 + i~X̂). (10.2)

That means that we need to replace Ĥ1 → Ĥ1 + 3αi~X̂ everywhere in the fol-
lowing.
3) For the Hamiltonian H(p, x) = 34 p4/3 + V (x), the path integral in phase space
formula, (10.21), remains valid,
( Z 0 )
Z Z t
0
U (t , t0 ) = Dx(t) Dp(t) exp i dt[p(t)ẋ(t) − H(p(t), x(t))] . (10.3)
t0

The first observation is that the equation of motion for p in the exponent is
1/3
p (t) = ẋ(t), which when substituted back in the exponent gives the classical
Lagrangian,
1
L = ẋ4 − V (x). (10.4)
4
That already tells us that the first approximation to the path integral is still the
classical action, eiScl [x,ẋ] . Next, we see that indeed
3
− p4/3 + pẋ − V
4
3 3 ẋ4
= − (p1/3 − ẋ)4 − 2ẋ(p1/3 − ẋ)3 − ẋ2 (p1/3 − ẋ)2 + −V , (10.5)
4 2 4
so that we can write
Z R h 4 iZ
0 i dt ẋ4 −V
U (t , t0 ) = Dx(t)e Dp(t)
Z
3 1/3 4 1/3 3 3 1/3 2
exp i dt (p − ẋ) − 2ẋ(p − ẋ) − (p − ẋ) (. 10.6)
4 2
35
36 Chapter 10

It would seem that this is fine, since the first exponent is eiS , but the path
integral over p(t), even shifted by ẋ, still has coefficients that depend on ẋ, so gives
an (uncalculable) function of ẋ, which means that the integrations over fluctuations
around the classical solution give a result that differs from the one in the standard
eiS path integral, and we cannot calculate them.
4) For the Hamiltonian

Ĥ = ~ω(â† â + 1/2) + λ(â + â† )3 , (10.7)

when going to the harmonic phase space path integral, we still have eq. (10.41),
∗
hα∗ |Ĥ(a† , â)|pi = H(α∗ , p)eα β
, (10.8)

so we still obtain eq. (10.46),

U (α∗ , t0 ; α, t) ( Z 0 )
i t
Z Z ∗
∗ α̇ (τ ) ∗ ∗
= Dα(t) Dα (t) exp dτ α(τ ) − H(α (τ ), α(τ )) + α (t)α(t) .
~ t i
(10.9)

5) For the partition function

1
dt0 J(t)∆(t,t0 )J(t0 )
R R
Z[J(t)] = N e− 2 dt
, (10.10)

the 2-point function is

δ2
G2 (t1 , t2 ) = Z[J] = N ∆(t1 , t2 ). (10.11)
iδJ(t1 )iδJ(t2 ) J=0

For the 3-point function, since there are only pairs of J’s, when putting J = 0,
we obtain 0, so
G3 (t1 , t2 , t3 ) = 0. (10.12)

For the 4-point function,

δ4
G4 (t1 , t2 , t3 , t4 ) = Z[J]
iδJ(t1 )...iδJ(t4 ) J=0
= ∆(t1 , t2 )∆(t3 , t4 ) + ∆(t1 , t3 )∆(t2 , t4 ) + ∆(t1 , t4 )∆(t2 , t3 ).
(10.13)
q̇ 2 2 2
6) For the harmonic oscillator Lagrangian, L = 2 − ω 2q , the partition function
is
Z R
Z[J] = Dq(t)eiS[q(t)]+i dtJ(t)q(t)
2
ω2 q2
Z Z
q̇
= Dq exp i dt + Jq −
Z Z 2 22
1 d 2
= Dq exp i dt − q(t) + ω q(t) + Jq . (10.14)
2 dt2
37 Chapter 10

Defining
d2

∆−1 (t1 , t2 ) = i + ω 2
δ(t1 − t2 ) (10.15)
dt21
dtJ(t)q(t) ≡ J · q, dt1 dt2 q(t1 )∆−1 (t1 , t2 )q(t2 ) = q · ∆ · q, we write
R R R
and
Z
1 −1
Z[J] = Dq exp − q · ∆ · q + iJ · q
Z 2
1 −1 1
= Dq exp − (q − i∆ · J) · ∆ · (q − i∆ · J) − J · ∆ · J
Z2 Z 2
1
= N exp − dt1 dt2 J(t1 )∆(t1 , t2 )J(t2 ) , (10.16)
2
which is the result we wanted, and N contains the result of the Gaussian path
integration over q(t) − i dt0 ∆(t, t0 )J(t0 ).
R
2
To find ∆(t1 , t2 ), we Fourier transform ∆−1 (t1 , t2 ) ≡ i dt
d
2 + ω 2
δ(t1 − t2 ), to
1
obtain
0
dp e−ip(t−t )
Z
∆(t1 , t2 ) = i . (10.17)
2π p2 − ω 2
7) For the partition function
Z Z
1
Z[J(t)] = N exp − dt dt0 J(t)∆(t, t0 )J(t0 )
Z Z2 Z Z
+λ dt1 dt2 dt3 dt4 J(t1 )J(t2 )J(t3 )J(t4 ) , (10.18)

we obtain the 4-point function

δ 4 Z[J]
G4 (t1 , t2 , t3 , t4 ) =
iδJ(t1 )...iδJ(t4 ) J=0
= N [∆(t1 , t2 )∆(t3 , t4 ) + ∆(t1 , t3 )∆(t2 , t4 ) + ∆(t1 , t4 )∆(t2 , t3 ) + 4!λ] .
(10.19)
11 Chapter 11

1) For the Hamiltonian H = p2 /2 + λx4 , with observable A = x3 + p3 , the evolution

equation is
dA ∂A
= + {A, H}P.B. = 3x2 p − 12λp2 x3 , (11.1)
dt ∂t
or, equivalently,
dA dx dp ∂H ∂H
= 3x2 + 3p2 = 3x2 − 3p2 = 3x2 p − 12λp2 x3 . (11.2)
dt dt dt ∂p ∂x
The quantum version is
d
hAiψ = 3hx2 piψ − 12λhp2 x3 iψ . (11.3)
dt
2) We start with the quantum version of the Hamilton-Jacobi equation for
i
ψ(~r, t) = Ae hbar S(~r,t) , which is (from eq. (11.27) in the text)
1 ~ 2 ~ 2A
~2 ∇
(∇S) + V (~r) + ∂t S − = 0. (11.4)
2m 2m A
We consider the central potential V (r), and separate variables. We first go the
time independent equation, by writing
S = s − Et , (11.5)
leading to
1 ~ 2 ~2 ∇~ 2A
(∇s) + V (r) − = E. (11.6)
2m 2m A
Using spherical coordinates, we obtain
2 " 2 #
2
1 ∂s 1 ∂s 1 ∂s
+ + 2 + V (r)
2m ∂r 2mr2 ∂θ sin θ ∂φ
~2 ∂2

2 ∂A 1
− 2
A+ + 2 ∆θ,φ A = E , (11.7)
2mA ∂r r ∂r r
where the angular Laplacian is
1 ∂2A

1 ∂ ∂A
∆θ,φ A = sin θ + . (11.8)
sin θ ∂θ ∂θ sin2 θ ∂φ2
We further separate variables as
s(r, θ, φ) = R(r) + F (θ, φ)
A(r, θ, φ) = P (r) · G(θ, φ) , (11.9)
38
39 Chapter 11

so that the wave function is

i i i
ψ(~r, t) = P (r)e ~ R(r) G(θ, φ)e ~ F (θ,φ) e− ~ Et . (11.10)
We then obtain
2 2
∂F 1 ∂F
+ = L2
∂θ sin2 θ ∂φ
1
∆θ,φ G = ˜l(˜l + 1) , (11.11)
G
where we obtain that L is a constant, and ˜l is not only a constant, but an integer
(˜l ∈ N), and its eigenfunction is the spherical harmonic Ylm (θ, φ).
Then the resulting radial equation is
2
~2 ∂ 2 P L2 − ˜l(˜l + 1)~2

∂R 2 ∂P
− + + = 2m[E − V (r)]. (11.12)
∂r P ∂r2 r ∂r r2
For completeness, note that we can further separate variables,
F (θ, φ) = lz φ + Θ(θ) , (11.13)
from which we obtain
2 r
lz2 l2
Z
dΘ
+ 2 = L ⇒ Θ = ± dθ L2 − z2 .
2
(11.14)
dθ sin θ sin θ
3) For the potential V (r) = −B/r, using the formulas from the previous exercise,
we find
2
l2

∂R
+ 2 = 2m[E − V (r)] ⇒
∂r r s
L2
Z
B
R(r) = ± dr 2m E + − 2. (11.15)
r r
and, considering lz = 0 (for motion in the plane (r, θ) only), we get
r
l2
Z
F (θ, φ) = lz φ ± dθ L2 − z2 = Lθ. (11.16)
sin θ
The Hamilton-Jacobi conditions, for the separated S,
S = s − Et = R(r) − Lθ − Et , (11.17)
are
∂S
= −t0 = const.
∂E
∂S
= θ0 ≡ 0 (11.18)
∂L
which become
Z
2m
dr q = t − t0 ⇒ r = r(t)
B L2

2m E + r − r2
40 Chapter 11

−2L/r2
Z
dr q + θ − θ0 = 0 ⇒ θ = θ(r). (11.19)
L2
2m E + Br −

r2

Since we are not interested in the trajectory r(t), but only in the deflection angle,
coming from θ = θ(r), we ignore the first equation, and concentrate on the second,
which is, explicitly,
Z r
2L
θ(r) = dr0 q . (11.20)
2
r02 2m E + rB0 − rL02

The particle comes from r = ∞, down to the inflection point, the minimum of
θ(r), so where the denominator in θ(r) vanishes, called θ1 (r0 ), then back to r = ∞.
We then obtain
s 2
L2 B B
r0 = + − . (11.21)
2m 2 2
In conclusion, we have
Z ∞
2L
θ1 = dr0 q , (11.22)
B L2

r0 r02 2m E + r0 − r 02

and, drawing perpendiculars onto the angle lines corresponding to the two limits,
we obtain that θ1 − π/2 is half the deflection angle δα, so
δα = 2θ1 − π. (11.23)
We can calculate the integral in θ1 , defining first 1/r = x, then x − mB/L2 = y,
to obtain
1
r0 − mB
L2
Ly
θ1 = 2 arcsin q
m2 B 2
2mE + L2 − mB
L2
1 mB mB
r0 − L2 L 2
= 2 arcsin q + 2 arcsin q . (11.24)
m2 B 2 m2 B 2
2mE + L2 2mE + L2
i
4) For the wave function ψ = Ae ~ S(~r,t) , the modified Hamilton-Jacobi equation
in eq. (11.27) in the text is
~ 2
(∇S) ~ 2A
~2 ∇
+ V (~r) + ∂t S − =0, (11.25)
2m 2m A
and when substituting S(~r, t) = W (~r) − Et, to obtain eq. (11.41) in the text, and
adding the continuity equation, we obtain the coupled system of equations for W (~r),
A(~r, t),
1 ~ ~ 2A
~2 ∇
(∇W )2 + V (~r) − E − =0
2m 2m A
1 ~ ~ 1 ~2
∂t A = − ∇A · ∇W + A∇ W . (11.26)
m 2
41 Chapter 11

This system can’t be uncoupled in general, so it stands for the equation for A.
~2 ∇~ 2A
However, in the geometric optics approximation, 2m A → 0, so we can solve
the first equation for W and replace it in the second, which now becomes a single
equation for A.
If moreover V (~r) = 0 (for a free particle), we obtain
√
W = 2m(~ p · ~r + const.) , (11.27)
so the equation for A is
1 ~2

p
∂t A = − ~ ⇒ A = Cei p~·~x− 2m t .
p~ · ∇A (11.28)
m
5) For a one-dimensional harmonic oscillator, V (x) = kx2 /2, the WKB approxi-
mation can be valid, yes, if the usual geometrical optics approximation is valid,
δλ λ
1, (11.29)
λ
which means: far from the turning points with E = V (r); also it gets better for
larger n (states are more classical), in which case E − En becomes also smaller.
6) For the potential V (x) = −B/x, the WKB wave function is
" s #
i x 0
Z
const. B iEt
ψ(x, t) = 1/4 exp ± ~ dx 2m E +
x
− . (11.30)
E+B x
x0 ~

The domain of validity is, as always, the domain of validity of the geometric
optics approximation,
δλ λ 1 δλ (E − V (x))
1⇒ 1. (11.31)
λ 2 E − V (x)
p
Then for E > 0, at large x, λ = ~/ 2m(E + B/x) is approximately constant,
so this is satisfied. For E < 0, we need to be far from the turning points, where
E + B/x = 0.
7) In one dimension, the equations for W and A (as written at exercise 4) become
1 ~2 A00
W 02 = E − V (x) +
2m 2m A
1 2 0 0
Ȧ = − (A W ) . (11.32)
2mA
In order to find the first order corrections to the WKB approximation, we can put
the 0th order solution (the WKB one) on the right-hand side of the first equation
above, and the resulting solution is also used in the second equation, to find the
next order of the coefficient A. Therefore, from the WKB solution,
const. A00 5 V 02 1 V 00
A= ⇒ = − , (11.33)
[E − V (x)]1/4 A 16 (E − V (x))2 4 E − V (x)
and this A00 /A is used in the first equation, obtaining
r
~2 A00
Z
W = dx E − V (x) +
2m A
42 Chapter 11

s
~2 V 02 V 00
Z
5 1
= dx E − V (x) + − . (11.34)
2m 16 (E − V (x))2 4 E − V (x)

Next, we use the 0th order (WKB) A and the 1st order W to obtain A2 W 0 to on
the right-hand side of (11.32), so
v
V 02 V 00
u 5 1
2 2 − 4 E−V (x)
u
~ 16 (E−V (x))
A2 W 0 = const. 1 +
t
⇒
2m E − V (x)
u v 0
5 V 02 1 V 00
const.
Z u
~ 2 16 (E−V (x)) 2 − 4 E−V (x) 
dt[E − V (x)]1/4  1 +
t
A=−  .
2m 2m E − V (x)

(11.35)
12 Chapter 12

1) For the Lagrangian

m ˙ 2 ) − V ((q 2 + q̃ 2 )2 ) ,
L= (|q̇|2 + |q̃| (12.1)
2
we find the symmetries:
-If q, q̃ are rotated by SO(2), as
0
q cos θ sin θ q
= , (12.2)
q̃ 0 − sin θ cos θ q̃

˙ 2 is invariant.
such that q 2 + q̃ 2 is invariant, and |q̇|2 + |q̃|

q 2πik
-If is multiplied by a Z4 element, e 4 , for k = 0, 1, 2, 3 (so both q and q̃
q̃
are multiplied by the same element), then (q 2 + q̃ 2 )2 is invariant, and so are |q̇|2
˙ 2.
and |q̃|
-Also an invariance is the Z2 that exchanges q and q̃, and the Z2 that acts only on
q as q → −q. All other symmetries are composites of these (or included in these).
For the continuous SO(2) symmetry, the representation is the fundamental, or
defining, one. For the Z4 and the Z2 ’s, the representation is the fundamental, or
defining, one.
2) In general, the Noether charge is
∂L j
Qa = (iTa )i qj . (12.3)
∂ q̇i
For the case of the SO(2) symmetry at exercise 1, qi refers to q, q ∗ , q̃, q̃ ∗ . The
complex conjugates can be dealt with by adding the complex conjugate of the result,
so we obtain

m ∗ ˙∗ q
Qa = q̇ q̃ (iTa ) + h.c. (12.4)
2 q̃

But for SO(2), for an infinitesimal angle θ = , so that we obtain the generator
(iTa ), we find

cos θ sin θ 0 1 0 1
= 1l + + O(2 ) ⇒ iTa = , (12.5)
− sin θ cos θ −1 0 −1 0

so that we obtain the Noether charge

m ∗
Q= (q̇ q̃ − q̃˙∗ q) + h.c. , (12.6)
2
43
44 Chapter 12

or, in terms of phase space variables (coordinates and momenta),

Q = pq q̃ − pq̃ q + h.c. (12.7)
Then we have
{Q, q}P.B. = −q̃ , {Q, q̃}P.B. = q , (12.8)
which means we have

δq 0 1 q q
' = − Q, , (12.9)
δ q̃ −1 0 q̃ q̃ P.B.

exactly as in the general case in eq. (12.14) in the text.

3) For the Hamiltonian
1 2 k
H= (p1 + p22 ) + (x21 + x22 ) , (12.10)
2m 2
we have the SO(2) continuous symmetry rotating the x’s and the p’s,
0 0
x1 cos θ sin θ x1 p1 cos θ sin θ p1
0 = , 0 = . (12.11)
x2 − sin θ cos θ x2 p2 − sin θ cos θ p2
Then as at exercise 2, we find that the Noether charge is
Q = p1 x2 − p2 x1 . (12.12)
Quantizing as [x1 , p1 ] = i~, [x2 , p2 ] = i~, we find that
Q̂ = p̂1 x̂2 − p̂2 x̂1 . (12.13)
Then, indeed Q̂ commutes with the Hamiltonian,
p̂1 p̂2 p̂2 p̂1
[Q̂, Ĥ] = (i~) − (i~) + kx̂2 x̂1 (−i~) + kx̂1 x̂2 (+i~) = 0. (12.14)
m m
4) For x21 + x22 = r2 , since dx21 + dx22 = dr2 + r2 dθ2 , we have that
p21 + p22 mẋ21 mẋ22 ṙ2 r2 θ̇2 p2 p2θ
= + =m +m = r + . (12.15)
2m 2 2 2 2 2m 2mr2
However, considering pθ = 0, the Hamiltonian equations of motion are
pr
ṗr = {pr , H}P.B. = −kr , ṙ = {r, H}P.B. = . (12.16)
m
Then the Ehrenfest theorem states that
hpr iψ
hṗr iψ = −khriψ , hṙiψ = . (12.17)
m
The transformed version under the continuous symmetries is:
-in the active case: |ψ 0 i = ÛSO(2) |ψi.
−1 −1
-in the passive case: r̂ → ÛSO(2) r̂ÛSO(2) , p̂ → ÛSO(2) p̂ÛSO(2) .
m 2
5) For the Lagrangian L = 2 |q̇| − V (q),
2πi
-in the case III, for V = 1/q 3 , we have Z3 symmetry, q → e 3 q.
-in the case II, for V = 1/|q|3 + 1/|q|5 = V (|q|), we have only the U (1) ('
45 Chapter 12

SO(2)) symmetry q → eiα q. There are no particular discrete symmetries, since the
symmetries of q 3 differ than the ones of q 5 .
-in the case I, for V = (1/q 3 + 1/q 5 ) + (1/q̃ 3 + 1/q̃ 5 ), for the same reason, there is
no discrete symmetry for each of the two brackets. There is only the Z2 symmetry
exchanging q and q̃. This potential also doesn’t admit any continuous symmetries
(the SO(2) symmetry was for a potential of the type V = V (q 2 + q̃ 2 ) only ).
6) a) We identify the two matrices with the generic matrix elements as
   
1 0 0 0 0. 1
A = 0 1 0 = e B = 0 1 0 = a , (12.18)
0 0 1 1 0 0
then indeed they satisfy the multiplication table of Z2 , where e corresponds to +1
and a to −1, so:
e · e = e , e · a = a , a · a = e. (12.19)
We can indeed check that A2 = A, A · B = A, B 2 = A.
b) The representation for A and B is not block-diagonal, so it is irreducible.
c) For a regular representation, D(g1 )|g2 i = |g1 g2 i. Since we have D(e) = A and
D(a) = B, we would need
D(e)|ei = |e · ei = |ei , D(e)|ai = |e · ai = |ai ,
D(a)|ei = |a · ei = |ai , D(a)|ai = |a · ai = |ei. (12.20)
But we must then identify |ei as the first basis element, and |ai as the second basis
element of the column vector. We see then that for A and B as above, it doesn’t
1 0
work. In fact, from the above conditions we see that we must have A =
0 1

0 1
and B = , for a regular representation.
1 0
d) Since we have a different dimensional space, the representation is not equiva-
lent to the roots of unity representation (±1).
7) In order to still have D(g1 )|g2 i = |g1 g2 i, we see that we need to have N × N
matrices. By comparison with the roots of unity representation, we see that if
2πi
g1 = e N , the matrix D(g1 ) takes |ei → |e2 i, |e2 i → |e3 i,... (shift by one), D(g2 )
shifts by two, etc. So:
0 0 ... 0 1 0 1 0 ... 0
   
1 0 ... 0 0 0 0 1 ... 0
   
D(e) = 1lN ×N , D(e2 ) = .. .. ... .. .. , ..., D(eN ) = 
 
.. .. .. ... .. .

.. .. ... .. .. 0 0 0 ... 1
0 0 ... 1 0 1 0 0 ... 0
(12.21)
13 Chapter 13

1) For the harmonic oscillator, H(−x) = H(x), since we have only quadratic terms,
and in quantum mechanics [π̂, Ĥ] = 0. That means that both the three dimensional
and the one dimensional harmonic oscillator are parity invariant. The parity of a
state n of the one dimensional harmonic oscillator is

π̂|ni = (−1)n |ni , (13.1)

and of the 3-dimensional one is

π̂|n1 n2 n3 i = (−1)n1 +n2 +n3 |n1 , n2 , n3 i. (13.2)

2) The probability density in the nth state is

ρn (x) = |ψn (x)|2 = |hx|ni|2 , (13.3)

so is clearly parity invariant, being quadratic (it is even positive). The Hamiltonian
is time reversal invariant, having only quadratic terms, which means ρn is real (it
has no phase), and it is time-reversal invariant and parity invariant.
3) For the one-dimensional Hamiltonian
p21 p2
H= + 2 + V (|q1 |) + V (|q2 |) + V12 (|q1 − q2 |) , (13.4)
2 2
possible continuous internal symmetries would act only on (q1 , q2 ), but there are
none, since for SO(2), we would need q12 + q22 .
4) For the algebra [A, B] = C, [B, C] = A, [C, A] = B, we need to check the
Jacobi identities to see if it is a Lie algebra. There is only one identity,

[C, [A, B]] + [A, [B, C]] + [B, [C, A]] = 0 , (13.5)

which is indeed zero, so it is a Lie algebra. It can be written as

[Xa , Xb ] = abc Xc , (13.6)

where Xa = (A, B, C), so fabc = abc . Then the adjoint representation is

c
(Ta )b = −ifab c = −iabc , (13.7)

so we obtain (since 1 = A, 2 = B, 3 = C)
     
0 0 0 0 0 −1 0 1 0
TA = −i 0 0 1 , TB = −i 0 0 0  , TC = −i −1 0 0 .
0 −1 0 1 0 0 0 0 0
(13.8)
46
47 Chapter 13

~2 + S
5) For H = α(S ~ 2) + βS
~1 · S
~2 , we have
1 2

~ −1 = −S
T ST ~ ⇒ TS
~ 2 T −1 = S
~2 , T S
~1 · S
~2 = S
~1 · S
~2 , (13.9)
so the Hamiltonian is time-reversal invariant. Under parity spin is invariant, so the
Hamiltonian is explicitly invariant.
6) SU (N ) has N 2 − 1 independent components, meaning that there are N 2 −
1 generators Ta , so the dimension of the adjoint representation is N 2 − 1. For
equivalence, we would need to same dimension for the fundamental representation,
of dimension N , so N 2 − 1 = N , which has no solutions.
7) For the Lagrangian
N N
!
X |q̇i |2 X
2
L= −V |qi | , (13.10)
i=1
2 i=1

if q ∈ R, we obtain SO(N ) invariance, since

N
T
X
(qi )0 = M (qi ) ⇒ (qi )0 (qi )0 = (qi )T M T M (qi ) = (qi )T (qi ) = (qi )2
i=1
⇔ M T M = M M T = 1l. (13.11)
If on the other hand, q ∈ C, we obtain SU (N ) invariance, since
N
†
X
(qi )0 = U (qi ) ⇒ (qi )0 (qi )0 = (qi )† U † U (qi ) = (qi )† (qi ) = |qi |2
i=1
⇔ U † U = U U † = 1l. (13.12)
14 Chapter 14

p
1) Considering that αjm = (j + m)(j − m + 1), for spin one matrices we obtain
(from eq. (14.76) in the text), for m = −1, 0, +1,

√
 
0 1 0
~ ~ 2
(J1 )mm0 = (α1,m+1 δm0 ,m+1 + α1,m δm0 ,m−1 ) = 1 0 1
2 2
0 1 0
√
 
0 1 0
~ ~ 2
(J2 )mm0 = (α1,m+1 δm0 ,m+1 − α1,m δm0 ,m−1 ) = −1 0 1
2i 2i
0 −1 0
 
−1 0 0
(J3 )mm0 = m~δmm0 = ~  0 0 0  . (14.1)
0 0 +1

In the adjoint representation, given that the Lie algebra is [Ji , Jj ] = iijk Jk , we
have
c
(Ta )b = ifab c = −abc , (14.2)

giving the generators

     
0 0 0 0 0 −1 +1 0 0
c c c
(J˜1 )b = − 0 0 1 , (J˜2 )b = −  0 0 0  , (J˜3 )b = − −1
0 0 .
0 −1 0 +1 0 0 0 0 0
(14.3)
The two representations are equivalent, meaning ∃U , such that J˜i = U Ji U −1 ,
since both representations are of dimension 3, and there are no other representations
of dimension 3. The form of the matrix U can be calculated, but it is a rather long
calculation.
2) We want to identify the form
α|
i|~ α|
i|~ α|
i|~ α|
i|~
i e + e− 2
2 ~ · ~σ e 2 − e− 2
α ~ · ~σ
α
g = e 2 αi σi = + = cos |~
α/2| + i sin |~
α/2|
2 |~
α| 2 ! |~
α|
α3 iα1 +α2
cos |~ α| sin |~
α/2| + i |~ α/2| |~
α| sin |~α/2|
= iα1 −α2 α3 (14.4)
|~
α| sin |~
α/2| cos |~α/2| − i |~α| sin |~
α/2|

with the Euler form

φ+ψ φ−ψ
!
cos θ2 ei 2 i sin θ2 ei 2
φ−ψ φ+ψ . (14.5)
i sin θ2 e− 2 cos θ/2e− 2
48
49 Chapter 14

We obtain 4 equations
θ φ+ψ
cos cos = cos |~
α/2|
2 2
θ φ+ψ α3
cos sin = sin |~
α/2|
2 2 |~
α|
θ φ−ψ α1
sin cos = sin |~
α/2|
2 2 |~
α|
θ φ−ψ α2
sin sin = sin |~
α/2| , (14.6)
2 2 |~
α|
leading to the equalities
α2 φ−ψ
= tan
α1 2
α3 φ+ψ
tan |~
α/2| = tan
|~
α| 2
2

α θ
α/2| 1 − 32 = sin2 ,
sin2 |~ (14.7)
|~
α| 2
and the first two equations can be solved for the Euler angles φ and ψ as
+α
α1 +
2 α3
α| tan |~
|~ α/2| −α
α1 +
2 α3
α| tan |~
|~ α/2|
tan φ = α2 α3 , tan ψ = α2 α3 . (14.8)
1− α| tan |~
α1 |~ α/2| 1+ α| tan |~
α1 |~ α/2|

3) Identifying the general matrix with the form with Euler angles, we obtain
φ+ψ φ−ψ
!
cos θ/2ei 2 i sin θ2 ei 2

a b
A= = φ−ψ φ+ψ , (14.9)
c d i sin θ2 e− 2 cos θ/2e− 2

which results in
θ
ad = cos2 θ
2 bc = − sin2
2
ab ac
a
d = ei(φ+ψ) , b
c = ei(φ−ψ) ⇒ e2iφ = , e2iψ = . (14.10)
cd bd
4) For the Lagrangian

L = q̇i† q̇i + det eȦ − q † Aq , (14.11)

P
with A = i=1,2,3 ai σi , we have that
Ȧ
det eȦ = eTr ln e = eTr Ȧ = 1 , (14.12)

where in the last equation we have used the fact that Tr σi = 0. Moreover, under the
SU (2) transformation q → U q, the kinetic term q̇i† q̇i is invariant as usual, whereas
the potential term is also invariant,
X X X
q † Aq = ai q † σi q → ai q † U † σi U q = ai q † σi q. (14.13)
i i i
50 Chapter 14

5) For spin 3/2, the representation is 2 · 3/2 + 1 = 4 dimensional, so via eq.

(14.76) in the text, cited above, we have
 
0 α3/2,−1/2 0 0
~ α3/2,−1/2 0 α3/2,1/2 0 
(J1 )3/2 =  
2 0 α3/2,1/2 0 α3/2,3/2 
0 0 α3/2,3/2 0
 √ 
√0 3 0 0
1 3 0 2 √0 
=   
2 0 2 √0 3
0 0 3 0
 √ 
0
√ 3 0 0
1 − 3 0 2 √0 

(J2 )3/2 = 
2  0 −2 0
√ 3
0 0 − 3 0
−3/2
 
0 0 0
 0 −1/2 0 0 
(J3 )3/2 =  0
. (14.14)
0 +1/2 0 
0 0 0 +3/2
One can show by explicit commutation that [Ji , Jj ] = iijk Jk .
6) At large j, we have J~2 = ~2 j(j + 1) ' ~2 j 2 , which is the classical value
for units of momentum
q of ~. Then also generically (if both j and m are large)
αjm ' j 2 − m2 = J~2 − J 2 . J3 = m~ also appears approximately continuous,
p
3
and
~2
(J1 )2 + (J2 )2 ' 2
(αjm δmm0 + αjm δmm0 ) ' (J~2 − J32 )δmm0 . (14.15)
2
7) For the isotropic 3 dimensional harmonic oscillator, the potential is
k 2 k
V = (x + x22 + x23 ) = r2 = V (r). (14.16)
2 1 2
Then the quantum numbers are E → n for the (discrete) energy levels, as well
as l, m for rotations.
15 Chapter 15

α · ~σ )2 = α
1) Since σ(i σj) = δij , so (~ ~ 2 , we obtain
|~
α| |~
α| |~
α| |~
α|
αi σi e 2 + e− 2 ~ · ~σ e
α 2 − e− 2 |~
α| α · ~σ |~
α|
e 2 = + = cosh + sinh , (15.1)
2 |~
α| 2 2 |~
α| 2
so that
αi σi |~
α|
Tr e 2 = 2 cosh . (15.2)
2
2) The decomposition is 1 ⊗ 1 = 0 ⊕ 1 ⊕ 2 (with dimensions 3 · 3 = 1 + 3 + 5), so
the angular momentum addition formulas in terms of Clebsch-Gordan coefficients
are
X
|11; 0mi = |11; m1 m2 ih11; m1 m2 |11; 0mi
m1 ,m2
X
|11; 1mi = |11; m1 m2 ih11; m1 m2 |11; 1mi
m1 ,m2
X
|11; 2mi = |11; m1 m2 ih11; m1 m2 |11; 2mi. (15.3)
m1 ,m2

3) The recursion relations are

p 0 0
q(j ∓ m)(j ± m + 1)h1, 1; m1 m2 |1, 1; j, m ± 1i
= (2 ∓ m01 )(1 ± m01 )h1, 1; m01 ∓ 1; m02 |1, 1; jmi
q
+ (2 ∓ m02 )(1 ± m02 )h1, 1; m01 ; m02 ∓ 1|1, 1; jmi , (15.4)

the phase convention is that all Clebsch-Gordan coefficients are real, while
h1, 1; 1, m2 |1, 1; 1, 1i is real and positive, and the normalization condition is
X
(h1, 1; m1 , m2 |1, 1; jmi)2 = 1. (15.5)
m1 ,m2

We consider only the case of lowest j, namely j = 0. Then the normalization

condition is
X
(h1, 1; m1 , m2 |1, 1; 0, 0i)2 = 1 , (15.6)
m1 ,m2 =0,±1

and the recursion condition is

q
(2 ∓ m01 )(1 ± m01 )h1, 1; m01 ∓ 1, m02 |1, 1; 0, 0i
q
+ (2 ∓ m02 )(1 ± m02 )h1, 1; m01 , m02 ∓ 1|1, 1; 0, 0i = 0. (15.7)

51
52 Chapter 15

Considering (for instance) m01 = m02 = 0 and both signs in the recursion condi-
tion, we obtain
h1, 1; 0, ∓1|1, 1; 0, 0i = −h1, 1; ∓1, 0|1, 1; 0, 0i. (15.8)
Considering m02 = 1, m01 = −1 and the lower sign, we obtain
h1, 1; −1, +1|1, 1; 0, 0i = −h1, 1; 0, 0|1, 1; 0, 1i , (15.9)
and considering m01 = −, m02 = −1 and the lower sign, we obtain
h1, 1; +1, −1|1, 1; 0, 0i = −h1, 1; 0, 0|1, 1; 0, 10i. (15.10)
Then considering m02 = 1, m01 = 0 and the lower sign, we obtain
h1, 1; 1, 1|1, 1; 0, 0i = 0 , (15.11)
considering m02 = 0, m01 = −1 and the lower sign, we obtain
h1, 1; −1, −1|1, 1; 0, 0i = 0 , (15.12)
considering m01 = −1, m02 = 1 and the lower sign, we obtain
h1, 1; 0, 1|1, 1; 0, 0i = 0 , (15.13)
and considering m01 = 1, m02 = −1 and the upper sign, we obtain
h1, 1; 0, −1|1, 1; 0, 0i = 0. (15.14)
That means that we have only 3 nonzero Clebsch-Gordan coefficients, and the
normalization condition becomes
3(h1, 1; 0, 0|1, 1; 0, 0i)2 = 1. (15.15)
Given that the phase convention is that h1, 1; +1, −1|1, 1; 0, 0i is real and positive,
we finally obtain that
1
√ = h1, 1; +1, −1|1, 1; 0, 0i = h1, 1; −1, +1|1, 1; 0, 0i = −h1, 1; 0, 0|1, 1; 0, 0i
3
(15.16)
and the rest are zero.
4) For the addition 1 ⊗ 1 ⊗ 1 = 2 (total angular momentum 2), we have (in the
3 ways of decomposing the double sum)
|1, 1,X
j12 , 1, 2, mi
= |1, 1, 1; m1 , m2 , m3 ih1, 1, m1 , m2 |j12 , m12 ihj12 , 1; m12 , m3 |2, mi
m1 ,m2 ,m12 ,m3
|1, 1,X
1, j23 , 2, mi
= |1, 1, 1; m1 , m2 , m3 ih1, 1, m2 , m3 |j23 , m23 ih1, j23 ; m1 , m23 |2, mi
m1 ,m2 ,m23 ,m3
|1, 1,X
j13 , 1, 2, mi
= |1, 1, 1; m1 , m2 , m3 ih1, 1, m1 , m3 |j13 , m13 ih1j13 ; m2 , m13 |2mi ,
m1 ,m2 ,m3 ,m13
(15.17)
53 Chapter 15

and the Racah coefficients W are defined by

p
h1, 1, j12 , 1, 2, m|1, 1, 1, j23 , 2, mi = (2j12 + 1)(2j23 + 1)W (1, 1, j13 , j12 , j23 ).
(15.18)
5) For the addition 1 ⊗ 1 = 0 ⊕ 1 ⊕ 2, we write the ket |1, 1; m1 , m2 i in terms of
Schwinger oscillators as
(a†1,+ )1+m1 (a†1,− )1−m1 (a†2,+ )1+m2 (a†2,− )1−m2
|1, 1; m1 , m2 i = p p |0i. (15.19)
(1 + m1 )!(1 − m1 )! (1 + m2 )!(1 − m2 )!
On the other hand, the ket |1, 1; j, mi is written as
(a†j,+ )j+m (a†j,− )j−m
|1, 1; j, mi = p |0i , (15.20)
(j + m)!(j − m)!

where necessarily a†j,+ is either a†1,+ or a†2,+ (identifying the up spin 1/2 creation
operators on both sides), and a†j,− is either a†1,− or a†2,− (idem for the down spin
1/2).
Therefore the Clebsch-Gordan coefficients are
h1, 1; m1 , m2 |1, 1; j, mi
† †
(a1,+ )1+m1 (a1,− )1−m1 (a2,+ )1+m2 (a2,− )1−m2 (aj,+ )j+m (aj,− )j−m
= h0| p p p |0i.
(1 + m1 )!(1 − m1 )! (1 + m2 )!(1 − m2 )! (j + m)!(j − m)!
(15.21)
6) For the Lie algebra
[Ji , Jj ] = iijk Jk , [Ki , Kj ] = iijk Jk , [Ji , Kj ] = iijk Kk , (15.22)
we can make the redefinition
Ji + Ki Ji − Ki
Qi = , Pi = , (15.23)
2 2
after which we obtain
[Qi , Qj ] = iijk Qk , [Pi , Pj ] = iijk Pj , [Qi , Pj ] = 0 , (15.24)
so two commuting copies of SU (2). That means we can define two copies of the
Schwinger oscillators, representing the algebra as
Q− = a†1,+ a1,− , Q+ = a†1,− a1,+ , 2Qz = a†1,+ a1,+ − a†1,− a1,−
P− = a†2,+ a2,− , P+ = a†2,− a2,+ , 2Pz = a†2,+ a2,+ − a†2,− a2,− . (15.25)
The representations are defined by the double (j1 , j2 ), for representations of the
two independent SU (2)’s, and are written in terms of the Schwinger oscillators as
(a†1,+ )j1 +m1 (a†1,− )j1 −m1 (a†2,+ )j2 +m2 (a†2,− )j2 −m2
p p |0i. (15.26)
(j1 + m1 )!(j1 − m1 )! (j2 + m2 )!(j2 − m2 )!
7) For a composition of 4 angular momenta of j = 1, we have
1 ⊗ 1 ⊗ 1 ⊗ 1 = (0 ⊕ 1 ⊕ 2) ⊗ (0 ⊕ 1 ⊕ 2)
54 Chapter 15

=0⊗0⊕0⊗1⊕0⊗2⊕1⊗0⊕1⊗1⊕1⊗2⊕2⊗0⊕2⊗1⊕2⊗2
= 3 · 0 ⊕ 6 · 1 ⊕ 6 · 2 ⊕ 3 · 3 ⊕ 4. (15.27)
Indeed, we can check that the dimensions work out, since (the dimension of the
spin j representation is 2j + 1)
81 = 34 = 3 + 6 · 3 + 6 · 5 + 3 · 7 + 9. (15.28)
16 Chapter 16

1) Consider the operator

1 ~ˆ 2 (2)
Ôij = X̂i X̂j − X δij T̂ij . (16.1)
3
This operator is in a symmetric traceless representation, coming from
1 ⊗ 1 = 0 ⊕ 1 ⊕ 2 (spin) or 3 ⊗ 3 = 1 ⊕ 3 ⊕ 5 (multiplicity) , (16.2)
where the decomposition is: 1= trace; 3= antisymmetric product, 5= symmetric
(2) (2) ij m
traceless product. Ôij is the latter. Define T̂m = T̂ij Cm , with Cij orthonormal
matrices translating the (ij) indices into a single index m, both of same multiplicity
(analog of the Clebsch-Gordan coefficients).
Then, we need
(2)
[Jˆk , T̂m
(2)
] = T̂ 0 h2m0 |Jk |2mi.
m (16.3)
But Jˆi = ijk X̂j P̂k (so that [Jˆi , X̂j ] = i~ijk X̂k ), meaning

ˆ (2) 2
[Ji , T̂lm ] = (−i~)ijk δkl X̂m + δkm X̂l − X̂k δlm
3
2
= −i~ ijl X̂j X̂m + ijm X̂j X̂l − ijk X̂j X̂k δlm
3
(2) ˆ 0 0 ∗
≡ T̂
l0 m0 hlm|J i |l m i
1
= X̂l0 X̂m0 − X̂p X̂p δl0 m0 (−i~)
3
2
× ijp δl0 j δmm0 δpl + ijp δl0 j δm0 l δpm − ijk δl0 j δm0 k δlm . (16.4)
3
This shows indeed the form of a tensor operator.
ˆ
2) From exercise 1, we know that P̂i P̂j − 31 P~ 2 δij is a tensor operator of spin
k = 2 (though at exercise 1 we had X̂i , but nothing changes for P̂i ). Then in the
Wigner-Eckhart theorem we have k = 2, and → α, which means that
ˆ2

0 0 1~

1ˆ
hE l || P̂ i P̂ j − 3 P δ ij ||Eli
hE 0 l0 m0 | P̂i P̂j − P~ 2 δij |Elmi = hl2; mq|l2j 0 m0 i √ .
3 2l + 1
(16.5)
3) Ylm ’s satisfy eqs. (16.41) in the text, but the second one is solved as
Ylm (θ, φ) = eimφ ylm (θ) , (16.6)
~ 2 = L2 + L2 + L2 , as
and we are left with the first one, for L 1 2 3

(L21 + L22 + L23 )(eimφ ylm (θ)) = ~2 l(l + 1)eimφ ylm (θ) , (16.7)
55
56 Chapter 16

or, by replacing L1 , L1 , L3 with their explicit forms in terms of θ and φ, we get

∂2 m2

− 2+ + im cos θ − l(l + 1) ylm (θ) = 0. (16.8)
∂θ sin2 θ
4) In cylindrical coordinates,
x1 = r sin θ , x2 = r cos θ , x3 = z , (16.9)
we have the Laplacian
1 ∂2 ∂2

~2 = 1 ∂
∇ r
∂
+ + . (16.10)
r ∂r ∂r r2 ∂θ2 ∂z 2
The Schrödinger equation for a free particle in these coordinates is
~2 ~ 2

− ∇ ψ(r, θ, z) = Eψ(r, θ, z). (16.11)
2m

∂ ∂ ∂
In terms of Lz = zij X̂i X̂j = −i~ x1 ∂x 2
− x 1 ∂x2 = −i~ ∂z , we have
" #
2 2 L̂2z
p̂r + p̂z + ψ(r, θ, z) = Eψ(r, θ, z). (16.12)
2mr2
5) The separation of variables for the above free particle in cylindrical coordinates
is
ψ(r, θ, z) = f (z)R(r)Y (θ) , (16.13)
2 2
k
and, with the ansatz f (z) = eikz , so that p̂2z f (z) = ~2m , and Y (θ) = eiqθ , so
2 2
L̂z Yq (θ) = ~ q Yq (θ), we have
q2

1 ∂ ∂
r + 2 + k 2 − 2mE R(r) = 0. (16.14)
r ∂r ∂r r
6) For two angular momenta, L ~ 1 and L
~ 2 , the Schrödinger equation for ψ(~r1 , ~r2 ) =
ψ(r1 , θ1 , φ1 ; r2 , θ2 , φ2 ), is
" #
1 L~2 ~2
L
2 1 2 2
p + 2 + p2,r + 2 ψ = Eψ. (16.15)
2m 1,r r1 r2
In the case ~r1 = ~r2 , the solution is a wave function that can be expanded in the
usual way,
lX
1 +l2 j
X
ψ = hr, θ, φ|l1 , m1 ; l2 , m2 i = hr, θ, φ|l1 , l2 ; j, mihl1 , l2 ; j, m|l1 , m1 ; l2 , m2 i.
j=|l1 −l2 | m=−j
(16.16)
7) The normalization condition is
Z ∞ Z π Z 2π
2 2
A dr|jl (kr)| dθ sin θ dφ|Ylm (θ, φ)|2 ⇒
Z ∞ 0 0
Z +1 0
1
dr|jl (kr)|2 d(cos θ)|ylm (θ)|2 = . (16.17)
0 −1 2πA2
17 Chapter 17

1) For the electron in magnetic field, we have states |l, s, ml , ms i, so ml = 0, ±1

and ms = ±1/2. B ~ couples to both L~ and S,
~ so we have 6 lines for the 6 states.
2) For a free particle,
~2 k 2
t i~
ψms (~r, t) = Ae−i 2m e k·~
r
× |ψ(t)i , (17.1)
where for a neutrino (spin 1/2) oscillating, we have, from (4.47) in the text,
E− t E+ t
|ψ(t)i = C− e−i ~ |ψ− i + C+ e−i ~ |ψ+ i , (17.2)
and
|ψ− i = cos θ|φ1 i + sin θ|φ2 i
|ψ+ i = − sin θ|φ1 i + cos θ|φ2 i. (17.3)
If we have |ψ(t = 0)i = |φ1 i, then from (4.51) in the text, we have
E− t E+ t E− t E+ t
|ψ(t)i = (cos2 θe−i ~ +sin2 θe+i ~ )|φ1 i+sin θ cos θ(e−i ~ −e+i ~ )|φ2 i. (17.4)
3) For the same free particle in spherical coordinates, we have
~2 k2
ψlm (~r, t) = e−i 2m t
Ylm (θ, φ)jl (kr)|ψ(t)i. (17.5)
4) To prove the addition formula for spherical harmonics,
+l
X
∗ 2l + 1
Ylm (~u)Ylm (~v ) = Pl (cos θ) , (17.6)
4π
m=−l

we consider a rotation of the system of coordinates,

~u0 = R · ~u , ~v 0 = R · ~v . (17.7)
Since Ylm are the unique eigenfunctions (a complete basis) of L ~ 2 and Lz , we can
0 0
expand Ylm (~u ) in Ylm0 (~u) and Ylm (~v ) in Ylm (~v ), as
X
Ylm (~u0 ) = Um0 m Ylm0 (~u)
m 0
X
Ylm (~v 0 ) = Um00 m Ylm00 (~v ) , (17.8)
m00

so that
l
X l
X X
0 ∗
Ylm (~u )Ylm (~v 0 ) = ∗
Um0 m Um 00 m Ylm0 (~
∗
u)Ylm 00 (~
v ). (17.9)
m=−l m=−l m0 ,m00

57
58 Chapter 17

Pl ∗
But U is unitary, so m=−l Um0 m Um 00 m = δm0 m00 , meaning

l
X l
X
Ylm (~u0 )Ylm
∗
(~v 0 ) = ∗
Ylm0 (~u)Ylm 0 (~
v) , (17.10)
m=−l m0 =−l

i.e., it is rotationally invariant, so can only depend on ~u · ~v = cos θ. Then we rotate

until ~v 0 ||~e3 , so only the m = 0 term contributes in the sum, so
l r
X
∗ 0 2l + 1 2l + 1
Ylm (~u)Ylm (~v ) = Yl0 (~u)Yl0 (~e3 ) = Yl0 (~u ) = Pl (cos θ) , (17.11)
4π 4π
m=−l
q q
where we have used the fact that Yl0 (~e3 ) = 2l+1
4π and Yl0 (~u0 ) = 2l+1
4π Pl (cos θ),
where now cos θ = u03 .
5) Larmor precession. The torque is
~
dL d X X
~τ = = (~ri × p~i ) = 0 + ~ri F~i . (17.12)
dt dt i i

But F~i = i q~vi × B,

~ and in fact in general we have
P

~ ×B
~τ = µ ~ , (17.13)
~ = 21 dV(~r × ~j). Thus in general, µ ~
R
and in particular, for currents, we have µ ~ ∝ L,
~ ~
~ which means that dJ ⊥ J~ and dJ ⊥ B, ~ or the same for µ
or more generally µ~ ∝ J, dt dt ~
~
instead of J, which means we get precession of µ ~
~ around B.
Quantum mechanically, that means that Lz , or Jz is constant, so in the electron
case, for given ms , e.g. of +1/2, so
|ψi = ψ(~r, t)|φ1 i. (17.14)
~ +S
6) For l = 2, since L ~ = J,
~ we have j = l − s, ..., l + s, in this case 3/2 or 5/2.
Then the states are
|3/2, mi , m = −3/2, −1/2, +1/2, +3/2 , and
|5/2, mi , m = −5/2, −3/2, −1/2, +1/2, +3/2, +5/2. (17.15)
7) To prove that
Sy
T = e−iπ ~ K0 , (17.16)
we note that K0 (iJi )K0−1 = −iJi , and the same for iJi replaced by iSi .
Sy Sy
But then e−iπ ~ Si e+iπ ~ for Si = Sy gives back Sy , for Si = Sz gives −Sz , and
for Sx gives −Sx .
18 Chapter 18

1) For the 3-body problem with 2-body potentials Vij (rij ), the Hamlitonian is
~2 ~ 2 ~2 ~ 2 ~2 ~ 2 X
Ĥ = − ∇1 − ∇2 − ∇ − Vij (rij ) , (18.1)
2m1 2m2 2m3 3 ij

where ~rij = ~ri − ~rj . We also define

~ ≡ m1~r1 + m2~r2 + m3~r3

R
m1 + m2 + m3
1 1 1
≡ + , M ≡ m1 + m2 + m3 . (18.2)
µij mi mj
We will see that the center of mass motion can be factorized as before. To do
that, we calculate

∂i3 = (∂i3 r31 )∂r31j + (∂i3 r23j )∂r23j + (∂i3 Rj )∂Rj

m3
= ∂r31j − ∂r23j + ∂R , etc. (cyclic). (18.3)
M j
Then
1 ~2 1 ~2 1 ~2
∇ + ∇ + ∇
m1 1 m2 2 m3 3
1 ~ 2 2
~ r + m3 ∇~R + 1 ∇ ~ r + m2 ∇

= ∇r31 − ∇ ~r −∇ ~R
23 23 12
m3 M 2 m 2 M
1 ~ ~ m1 ~
+ ∇r12 − ∇r31 + ∇R
m1 M
1 ~ ~ r )2 + 1 (∇~r −∇ ~ r )2 + 1 (∇ ~r −∇ ~ r )2 + 1 ∇
~2.
= (∇r31 − ∇
m3 23
m2 23 12
m1 12 31
M R
(18.4)

Therefore the Schrödinger equation Ĥψ = Eψ becomes

~2 ~ 2 ~2 ~ 2 2

− ∇R − (∇r31 − ∇ ~ r )2 − ~ (∇ ~r −∇ ~ r )2 − ~ (∇ ~r −∇ ~ r )2
23 23 12 12 31
2M 2m3 2m2 2m1
~ ~
+V12 (r12 ) + V13 (r13 ) + V23 (r23 )] ψ(~r12 , ~r23 , ~r31 , R) = Eψ(~r12 , ~r23 , ~r31 , R). (18.5)

Then we can separate the center of mass variables

~ = ψ̃(~r12 , ~r23 , ~r31 )φ(R)
ψ(~r12 , ~r23 , ~r31 , R) ~ , (18.6)

and the Schrödinger equation becomes

~2 ~ 2 ~

− ~
∇ φ(R) − ECM φ(R) ψ̃(~r12 , ~r23 , ~r31 )
2M R
59
60 Chapter 18

~2 ~ 2 2

+ − (∇r31 − ∇ ~ r )2 − ~ (∇ ~r −∇ ~ r )2 − ~ (∇ ~r −∇ ~ r )2
23 23 12 12 31
2m3 2m2 2m1 i
×ψ̃(~r12 , ~r23 , ~r31 ) + (V12 (r12 ) + V13 (r13 ) + V23 (r23 )) ψ̃(~r12 , ~r23 , ~r31 ) φ(R)~
= (E − ECM )ψ̃(~r12 , ~r23 , ~r31 )φ(R) ~ , (18.7)
which splits into
~2 ~ ~ = ECM φ(R) ~ ,
− ∇R φ(R)
2M2 2 2
~ ~ ~ r )2 − ~ (∇ ~r −∇ ~ r )2 − ~ (∇ ~r −∇ ~ r )2
− (∇r31 − ∇ 23 23 12 12 31
2m3 2m2 2m1
+V12 (r12 ) + V13 (r13 ) + V23 (r23 )] ψ̃(~r12 , ~r23 , ~r31 ) = (E − ECM )ψ̃(~r12 , ~r23 , ~r31 ).
(18.8)
2) For the two-dimensional system with 1/r potential, V (r) = −Q̃2 /r, the re-
duced Schrödinger equation is
2
~ ~2
− ∇r + V (~r) ψ(~r) = Erel ψ(~r). (18.9)
2µ
In two dimensions, the Laplacian is
2 2
~ 2r = ∂ + 1 ∂ + 1 ∂ ,
∆r = ∇ (18.10)
∂r2 r ∂r r2 ∂φ2
2
∂
where ∂φ 2 = ∆φ is the Laplacian on the circle. The wave function separates vari-

ables as
ψ(r, φ) = R(r)eimφ , (18.11)
2
∂
such that ∆φ eimφ = ∂φ 2e
imφ
= −m2 eimφ . Then the resulting radial equation is
( )
~2 d2 m2 Q̃2

1 d
− + − 2 − R(r) = ER(r). (18.12)
2µ dr2 r dr r r
Defining the reduced radial function
χ(r)
R(r) = √ , (18.13)
r
the equation for χ(r) takes the form of the one-dimensional Schrödinger equation,
" !#
d2 χ(r) 1 − 4m2 2µ Q̃2
+ + 2 E+ χ(r) = 0 , (18.14)
dr2 4r2 ~ r
with the effective potential
Q̃2 ~2 m2 − 1/4
Veff (r) = − + . (18.15)
r 2µ r2
The boundary condition is χ(r = 0) = 0 at r = 0 and normalizability at infinity,
Z ∞ Z ∞
2
|ψ| rdr < ∞ ⇒ χ2 (r)dr < ∞. (18.16)
0 0
61 Chapter 18

3) Continuing with the analysis, define as usual ρ = 2kr, where

r r
2µ|E| 2 µ
k= , λ = Q̃ , (18.17)
~2 2|E|~2
so that the reduced radial equation is
dχ2 (ρ) 1/4 − m2

1 λ
+ + + χ(ρ) = 0. (18.18)
dρ2 ρ2 4 ρ
The solution at ρ → 0 is of the type χ ∼ Aρα , from which we obtain
ρα
ρα−2 [α(α − 1) + 1/4 − m2 ] + − λρα−1 = 0. (18.19)
4
Cancelling the leading term, in order to find the leading solution, we find
α(α − 1) = m2 − 1/4 ⇒ α = ±m + 1/2 , (18.20)
but χ ∼ Aρ−m+1/2 is not normalizable at 0, so is excluded, remaining
χ ∼ Aρm+1/2 . (18.21)
The solution at ρ → ∞ is found by first observing that V (r → ∞) → 0, and so
also Veff (r → ∞) → 0, so by the general theory of one-dimensional potentials, for
a positive energy E > 0, we have
χ(ρ) ∼ Be±kr = Be±ρ/2 . (18.22)
But eρ/2 is not normalizable at infinity, so we are left with
χ(ρ) ∼ Be−ρ/2 . (18.23)
Then, we need to factorize the behaviours at 0 and infinity of χ(ρ), so we first
write
χ(ρ) = e−ρ/2 H(ρ) , (18.24)
leading to
1/4 − m2

00 0 λ
H −H + 2
+ H = 0. (18.25)
ρ ρ
Next, we factorize also the behaviour at 0,
H(ρ) = ρm+1/2 F (ρ) , (18.26)
leading to

2m + 1 F
F 00 + − 1 F 0 − (m + 1/2 − λ) = 0. (18.27)
ρ ρ
We note that if we put m = l, λ = λ̃ − 1/2, we obtain the 3 dimensional equation
for l, λ̃, which we saw had a hypergeometric function as a solution. So the solution
is
F = A1 F1 (m + 1/2 − λ, 2m + 2; ρ) + Bρ−2m−1 1 F1 (−λ − 1/2 − m, −2m; ρ). (18.28)
62 Chapter 18

Moreover, the quantization condition is done in the same way,

Q̃2 µ
r
µ
λ + 1/2 = λ̃ = nr + m + 1 ≡ n ⇒ Q̃2 2
= n − 1/2 ⇒ En = − 2 .
2|E|~ 2~ (n − 1/2)2
(18.29)
4) To prove the normalization constant (eq. (18.61)),
s
3/2 1 (n + l)!
Nn,l = (2kn ) , (18.30)
(2l + 1)! (n − l − 1)!2n
for the radial function
Rn,l (r) = Nn,l e−kn r (2kn r)l 1 F1 (−n + l + 1, 2l + 2; 2kn r) , (18.31)
we first note (eq. (18.62))
n!Γ(b + 1) b
1 F1 (−n, b + 1; z) = L (z) , (18.32)
Γ(b + n + 1) n
so
Rn,l (r) = Ñn,l e−kn r (2kn r)l L2l+1
n−l−1 (2kn r) , (18.33)
where
s
(n − l − 1)!
Ñn,l = (2kn )3/2 , (18.34)
(n + l)!2n
where we note that kn = 1/(na) = Z/(na0 ). Defining as usual 2kn r = z, we find
the condition
2 Z ∞
Ñn,l
dzz 2l+2 e−z [L2l+1 2
n−l−1 (z)] = 1 , (18.35)
(2kn )3 0
which is satisfied, since
Z ∞
(n + k)!
dze−z z k+1 [Lkn (z)]2 = (2n + k + 1) , (18.36)
0 n!
somewhat similar to the orthonormalization condition of Laguerre polynomials,
Z ∞
(m + k)!
dze−z z k Lkm (z)Lkn (z) = δmn . (18.37)
0 n!
Note that some places (like Messiah) use a different normalization of the Laguerre
polynomials, with [(m+k)!]3 instead of (m+k)! on the right-hand side of the above,
so [Γ(b + n + 1)]2 instead of Γ(b + n + 1) in the denominator of the right-hand side
of (18.32), and resulting in [(n + l)!]3 instead of (n + l)! in the denominator of the
square root of Ñn,l .
5) We have
Z ∞ 2 Z ∞
Ñn,l

1 1 2
= dr r2 Rn,l (r) = (2k n ) dz z 2l+1 e−z [L2l+1
n−l−1 ]
2
r nl 0 r (2kn )3 0
kn Z
= = , (18.38)
n a0 n2
63 Chapter 18

where we used the orthonormality of the Laguerre polynomials.

Further,
Z ∞ 2 Z ∞
Ñnl 1
hrinl = 2 2
dr r rRnl (r) = 3 2k
dz z 2l+3 e−z [L2l+1
n−l−1 ]
2
0 (2k n ) n 0
a0
= [3n2 − l(l + 1)] , (18.39)
2Z
where we have used
Z ∞
(n + l)!
dz z 2l+3 e−z [L2l+1 2 2
n−l−1 ] = 2[3n − l(l + 1)] . (18.40)
0 (n − l − 1)!
Moreover,
Z ∞ 2 Z ∞
Ñnl 1
hr2 inl = dr r2 r2 Rnl
2
(r) = dz z 2l+4 e−z [L2l+1
n−l−1 ]
2
0 (2kn ) (2kn )2
3
0
a20 2 2
= n [5n + 1 − 3l(l + 1)] , (18.41)
2Z 2
where we have used
Z ∞
(n + l)!
dz z 2l+4 e−z [L2l+1 2 2
n−l−1 ] = 4n[5n + 1 − 3l(l + 1)] . (18.42)
0 (n − l − 1)!
6) The average radial momentum is
Z ∞
~ ∂ ∂
hpr in,l = = −i~ drlr2 Rnl (r) Rnl (r). (18.43)
i ∂r n,l 0 ∂r
But, since
Rnl (r) = Ñnl e−kn r (2kn r)l L2l+1
n−l−1 (2kn r) , (18.44)
d k
and dx Ln (x) = −Lkn−1 (x), so
d
Rnl (r) = Ñnl e−kn r −kn (2kn r)l Ln−l−1 (2kn r) + 2kn l(2kn r)l−1 L2l+1

n−l−1 (2kn r)
dr
l 2l+1

−2kn (2kn r) Ln−l−2 (2kn r) , (18.45)
we have
Z ∞ Z ∞
2 2 1 2
hpr inl = −i~ (−kn ) dr r Rnl (r) + l dr r2 Rnl (r)
0 0 # r
2 Z ∞
2kn Ñnl
− 3
dz z 2l+2 L2l+1 2l+1
n−l−1 Ln−l−2
(2kn ) 0

l i~Z
= −i~ −kn + kn + 0 = (n − l) , (18.46)
n a0 n2
where in the first term we used the normalization of the wave function, in the second
h1/rinl = kn /n (from the previous exercise) and in the third the orthogonality of
the Laguerre polynomials.
7) The Hamiltonian gets shifted, Ĥ → Ĥ + µ ~ But
~ · B.
µB ~ ~ ,
µ
~= (L + 2S) (18.47)
~
64 Chapter 18

so
µB ~ ~ ~ · B).
~
Ĥ 0 = Ĥ + (L · B + 2 S (18.48)
~
Then in the Hydrogen atom analysis, we shift
E → E − µB (m + 2ms )B. (18.49)
19 Chapter 19

1) For the potential V (r) = αr cos βr, Veff (r) is dominated at r → 0 by the term
l(l + 1)/r2 , so blows up. However, the potential (so also the effective potential) has
an infinite number of negative regions for r → ∞, so an infinite number of bound
states, like the Coulomb 1/r potential.
2) For the potential V (r) = |α|
r 2 ln r/r∗ , we have effectively s > 2 in the analysis in
the text, so the potential dominates over the l(l + 1)/r2 term at r → 0, so becomes
infinitely negative. As for the analysis in the text, we have states of arbitrarily
large |E| for E < 0, and the particle ”falls into r=0”, so the quantum particle is
effectively at r = 0 (in a bound state).
3) For the Yukawa potential V (r) = − |α| r e
−mr
, at r → 0, we have a Coulomb
potential, so again Veff is dominated by the l(l + 1)/r2 term, so we have a bound
state spectrum. However, at r → ∞, we have effectively s → ∞ in the analysis
in the text, so Veff is also dominated by the l(l + 1)/r2 at infinity, being therefore
positive. The effective potential therefore starts infinitely positive at r = 0, drops
down to a negative minimum, then up to a positive maximum, and then slowly
decays to zero. By the analysis of the energy of a state
~2 |α| −mr
E∼ − e >0 (19.1)
2m(∆r)2 r
for large r, there are no bound states at large distances, so we have necessarily a
finite number of bound states (at finite distance). From the shape of the potential,
we can indeed approximate it with a harmonic oscillator around the negative min-
imum (like for the diatomic molecule), and the approximation should be valid for
the first few excited states.
4) In the equality of the free wave functions (in Cartesian coordinates and in
spherical coordinates)
∞ X
l
~
X
eik·~r = alm (~k)jl (kr)Ylm (θ, φ) , (19.2)
l=0 m=−l

by choosing ~k onto Oz, since the left-hand side is independent of φ, the right-
imφ
hand side
q must also, so we must have m = 0 (the φ dependence is in e ). Since
2l+1
Yl,0 = 4π Pl (cos θ), with cos θ ≡ u and ρ = 2kr as usual, we have
∞
ρu X
ei 2 = ekr cos θ = cl jl (ρ/2)Pl (u). (19.3)
l=0

To find cl = (2l + 1)il c0 , we differentiate both sides of the above equation with
65
66 Chapter 19

respect to ρ/2, obtaining

X
iueiρ/2u = cl jl0 (ρ/2)Pl (u)
l
X
= cl ijl (ρ/2)uPl (u). (19.4)
l

But the recurrence relations of the Legendre polynomials

(2l + 1)uPl (u) = (l + 1)Pl+1 (u) + lPl−1 (u) (19.5)
gives (after redefining the sums over l to have the common factor Pl (u), and defining
x = ρ/2)

X lcl−1 (l + 1)cl+1
iueixu = i Pl (u) jl−1 (x) + jl+1 (x) . (19.6)
2l − 1 2l + 3
l

On the other hand, using the recurrence relations of the spherical Bessel functions
l+1
jl0 (x) = jl−1 (x) − jl (x)
x
jl (x) 1
= (jl+1 (x) + jl−1 (x)) , (19.7)
x 2l + 1
we find
iux
X l l+1
iue = cl jl−1 (x) − jl+1 (x) Pl (u). (19.8)
2l + 1 2l + 1
l

Equating the two formulas for iueiux , and identifying the coefficients of Pl (u),
we find

cl cl1 cl cl+1
ljl−1 (x) −i = (l + 1)jl+1 (x) +i , (19.9)
2l + 1 2l − 1 2l + 1 2l + 3
but since this must be true at any x, we must put to zero the coefficient of jl−1 (x)
on the left and of jl+1 (x) on the right, giving the recurrence relation
cl+1 cl
=i , (19.10)
2l + 3 2l + 1
which iterated gives
cl = (2l + 1)il c0 . (19.11)

q for −V0 < E < 0 and r < r0 , the solution

5) For the spherical square well,
2m(E+V0 )
is Rl (r) = Ajl (kr), with k = ~2 , whereas for r > r0 the solution is
q
χ(r) = Be−κr , or Rl (r) = Be−κr /r, where κ = 2m|E| ~2 .
The conditions for continuity of the radial wave function and its derivative at r0
give
Be−κr
Rl (r0− ) = AJl (kr0 ) = Rl (r0+ ) =
r
0 0 0 −κr0 κ 1
Rl (r0− ) = Akjl (kr0 ) = Rl (r0+ ) = Be − − 2 . (19.12)
r0 r0
67 Chapter 19

Solving for the ratio A/B, we find

(kr0 )jl0 (kr0 ) = −(1 + κr0 )jl (kr0 ). (19.13)
Using the fact that jl0 (x) = l
jl−1 (x) 2l+1 − l+1
jl+1 (x) 2l+1 , we find the condition
ljl−1 (kr0 ) − (l + 1)jl+1 (kr0 ) + (2l + 1)(1 + κr0 )jl (kr0 ) = 0. (19.14)
But note that, as l increases (and jl ' jl+1 ' jl−1 ) the last term dominates,
so there is no more equality possible, so only for small enough l can we solve the
equation. Moreover, for each l, we find a finite number of solutions, since jl (kr)
oscillates, but not too much.
6) For the 3-dimensional isotropic harmonic oscillator in spherical coordinates,
the radial wave function is
α2 r 2
Rn,l (r) = Nn,l e− 2 (αr)l 1 F1 (−nr , l + 3/2; α2 r2 ). (19.15)
We need to prove
√ 1/2
α 2α nr !
Nn,l = . (19.16)
Γ(l + 3/2) Γ(nr + l + 3/2)
But first note that
Γ(b + 1)
1 F1 (−n, b + 1; z) = n! Lb (z) , (19.17)
Γ(b + n + 1) n
so that
α2 r 2
Rn,l (r) = Ñnr ,l (αr)l e− 2 Ll+1/2
n (α2 r2 ) , (19.18)
where we need to prove
3/2
√

nr !
Ñn,l = α 2α . (19.19)
Γ(nr + l + 3/2)
But, using the normalization condition
Z ∞ Z ∞ i2
2 1 z2
h
dr r2 Rn,l
2
(r) = Ñn,l 3
dz z l+2 e− 2 Ll+1/2
nr =1, (19.20)
0 α 0
and the integral
Z ∞ i2 3
l+2 − z2
2 h 1 Γ(nr + l + 3/2)
dz z e Ll+1/2
nr = , (19.21)
0 2 nr !
we prove the relation.
7) In two dimensions, the isotropic harmonic oscillator hamiltonian in polar
coordinates is
~2 ∂ 2 1 ∂2 m 2 ω 2 ρ2

1 ∂
Ĥ = − 2
+ + 2 2
+ . (19.22)
2m ∂ρ ρ ∂ρ ρ ∂φ 2
Then, the Schrödinger equation Ĥψ = Eψ, with separated variables for the wave
function,
ψnρ ,|m| (ρ, φ) = Φm (φ)Rnρ ,|m| (ρ) , (19.23)
68 Chapter 19

gives the φ dependence as before,

1
Lz Φm (φ) = m~Φm (φ) , Φm (φ) = √ eimφ , (19.24)
2π
and the radial equation
d2 R(ρ) 1 dR(ρ) m2 m2 ω 2 2

2m
+ + − 2 + 2 E− ρ R(ρ) = 0. (19.25)
dρ2 ρ dρ ρ ~ ~2
Defining
χ(ρ)
R(ρ) = √ , (19.26)
ρ
we find the reduced radial equation
d2 χ(ρ) m2 − 1/4 2mE m2 ω 2 2

+ − + 2 − ρ χ(ρ) = 0. (19.27)
dρ2 ρ2 ~ 2
This is the same equation as (19.72) in the text, with the the substitution l(l +
1) → m2 − 1/4, so l = − 21 ± |m|, leading to the energy quantization
E 3
= l + + 2nρ = 1 ± |m| + 2nρ ≡ 1 + n. (19.28)
~ω 2
20 Chapter 20

1) The wave function is

 
ψ (x ) ψω2 (x1 ) ψω3 (x1 )
1  ω1 1
ψA (x1 , x2 , x3 ) = √ ψω1 (x2 ) ψω2 (x2 ) ψω3 (x2 ) . (20.1)
6
ψω1 (x3 ) ψω2 (x3 ) ψω3 (x3 )
2) The potential

V̂ = V̂int (|X̂1 − X̂2 |) + V̂int (|X̂1 − X̂3 |) + V̂int (|X̂2 − X̂3 |) (20.2)

is symmetric under the permutation of {1, 2, 3}, so it is OK for indistinguishable

particles.
3) We have
|N...21i = (±1)nN |12...N i , (20.3)

where
N (N − 1)
nN = (N − 1) + (N − 2) + ... + 1 = . (20.4)
2
4) The general state of the fermionic harmonic oscillator is

|ψi = bnf |0i , (20.5)

where nf = 0 or 1. For N fermionic oscillators,

1 X
|ψn1 ...nN i = √ sgn(P )P̂ [bn1 |0i1 ⊗ ...bnN |0iN ] , (20.6)
N! P

5) When expanding the fermion field in fermionic harmonic oscillators, we must

write
Z
~ ~
ψ(~x, t) = d3 k u~k b(~k)e−iωt+ik·~x + v~k c† (~k)e+iωt−ik·~x , (20.7)

where b(~k) and c† (~k) must be different harmonic oscillators. This is so, since we
must have {ψ(~x, t), ψ † (~y , t)} = δ 3 (~x − ~y ), since ψ † is canonically conjugate to ψ
(the Lagrangian is linear in derivatives, not quadratic). Here we have

{b(~k), b† (~k 0 )} = δ 3 (~k − ~k 0 ) , {c(~k), c† (~k 0 )} = δ 3 (~k − ~k 0 ) , (20.8)

and all the other anticommutators are zero.

Then we trivially find

{ψ(~x, t = 0), ψ(~y , t = 0)} = 0. (20.9)

69
70 Chapter 20

For φ(~x, t)ψ(~y , t), we have (bose times fermi is fermi)

[φ(~x, t), ψ(~y , t)] = 0. (20.10)
6) For 3 particles of spin 1/2, we have
(1/2 ⊗ 1/2) ⊗ 1/2 = (1 ⊕ 0) ⊗ 1/2 = 3/2 ⊕ 1/2 ⊕ 1/2. (20.11)
In terms of multiplicity, we have 4 ⊕ 2 ⊕ 2, with the 4 being the totally symmetric
representation, and one of the 2 being the totally antisymmetric representation
(and the other 2 being a mixed one).
Then the spin wave functions are
χ+++
χ+−+ + χ−++ + χ++−
√
3
χ−+− + χ+−− + χ−−+
√
3
χ−−− (20.12)
for the symmetric representation (quadruplet), which therefore is multiplied by the
totally antisymmetric spatial wave function, and
χ−++ + χ++− − χ+−+
√
3
χ+−− + χ−−+ − χ−+−
√ (20.13)
3
for the totally antisymmetric representation (doublet), which therefore is multiplied
by the totally symmetric spatial wave function. The other doublet is of mixed
symmetry, therefore doesn’t appear in the wave functions.
The totally symmetric representation for the spatial wave function is
1
ψS (x1 , x2 , x3 ) = √ [φa (x1 )φb (x2 )φc (x3 ) + φa (x1 )φb (x3 )φc (x2 )
6
+φa (x2 )φb (x1 )φc (x3 ) + φa (x2 )φb (x3 )φc (x1 ) + φa (x3 )φb (x1 )φc (x2 )
+φa (x3 )φb (x2 )φc (x1 )] , (20.14)
and the totally antisymmetric representation is
1
ψA−S (x1 , x2 , x3 ) = √ [φa (x1 )φb (x2 )φc (x3 ) − φa (x1 )φb (x3 )φc (x2 )
6
−φa (x2 )φb (x1 )φc (x3 ) + φa (x2 )φb (x3 )φc (x1 ) + φa (x3 )φb (x1 )φc (x2 )
−φa (x3 )φb (x2 )φc (x1 )] . (20.15)
7) The probability is
P = |ψS/A−S |2 , (20.16)
and the exchange terms are the cross terms in the above modulus square.
21 Chapter 21

1) For the ortho-He atom with one electron in n = 1, l = 0, and the other in
n = 2, l = 0, so with spin singlet state, and symmetric spatial wave function,
1
φ = φs (~r1 , ~r2 ) = √ [ψq1 (~r1 )ψq2 (~r2 ) + ψq2 (~r1 )ψq1 (~r2 )] , (21.1)
2
in the approximation 1, the energy is just the sum of the energies,
e20 Z e2 Z
E = E (1) + E (2) = 2 + 0 2 , (21.2)
a 0 n1 a0 n2
so, for Z = 2 (He), we have
e2 5/2e2

2 2
E= 0 + 2 = . (21.3)
a0 12 2 (4π0 )a0
In the approximation 2, we need to add the electron-electron interaction, as
e2 e2
Z Z
∆E = hφs | 0 |φs i = d3 r1 d2 r2 0 |φs (~r1 , ~r2 )|2 = C + A , (21.4)
r12 r12
where
|ψn=1,l=0 (~r1 )|2 e2 |ψn=2,l=0 (~r2 )|2
Z Z
3
C= d r1 d3 r2
4π0 r12
Z Z 2
e ∗ ∗
A = d3 r1 d3 r2 ψn=1,l=0 (~r1 )ψn=1,l=0 (~r2 )ψn=2,l=0 (~r2 )ψn=2,l=0 (~r1 ).
4π0 r12
(21.5)
But
Rn0 (r) Ñn,0
ψn,l=0 (~r) = √ = √ e−kn r L1n−1 (2kn r) , (21.6)
4π 4π
and, since L10 (x) = 1 and L11 (x) = 2 − x and kn = Z/(na0 ), we find
3/2
2 2 2r
ψn=1,l=0 (~r) = √ e− a0
4π a0
3/2
2 1 − ar r
ψn−2,l=0 (~r) = √ e 0 1− . (21.7)
4π a0 a0
Then
2
27 e2
Z Z
4r1 2r2 r2
C= d3 r1 d3 r2 e− a0 e− a0 1 −
4πa60 4π0 |~r1 − ~r2 | a0
3(r1 +r2 )
7 Z 2 −
Z
2 3 3 e e a0
r1 r2
A= d r1 d r2 1− 1− . (21.8)
4πa60 4π0 |~r1 − ~r2 | a0 a0
71
72 Chapter 21

To calculate the integrals, we follow the example in the text. Indeed, there it was
shown that
e−4r1 a0
Z
Φ(~r2 ) ≡ d3 r1
r12
1 e−2k1 r2

1 −2k1 r2 1 1
= −π e + − , (21.9)
k12 k13 r2 k13 r2
for k1 = 2/a0 . Thus, we obtain (after some algebra) that
Z ∞ 2
27 e2

2
2r
− a2 r2
C= 4πr2 dr2 Φ(r2 )e 0 1−
4π(4π0 )a60 0 a0
2

2πe 13
= 1− 4 . (21.10)
a0 (4π0 ) 3
To calculate A, we need to calculate first
3r2
e− a0
Z
r2
Φ̃(r1 ) = d3 r2 1− , (21.11)
r12 a0
which (like for Φ(r1 ) in the text) satisfies (for k1 = 3r2 /a0 )
d2

r1 r1
∆r1 Φ̃(r1 ) = −4π 1 − e−2k1 r1 ⇒ 2 (r1 Φ̃(r1 )) = −4πr1 1 − e−2k1 r1 ,
a0 dr1 a0
(21.12)
so it has the particular solution of the type

r1 Φ̃p (r1 ) = (ar12 + br1 + c)e−2k1 r1 . (21.13)

From the differential equation, we fix

π 2π 2π π π
a= , b=− 2 + , c=− 3 + . (21.14)
a0 k12 k1 a0 k13 k1 2a0 k14

But the solution to the homogeneous equation is r1 Φ̃(r1 ) = C1 r1 +C2 , and adding
the physical conditions Φ̃(r1 → ∞) → 0, that implies C1 = 0, and the condition
that Φ̃(r1 ) is nonsingular at r1 → 0, we find that the coefficient of 1/r1 in Φ̃ must
be 0, so C2 = −c.
Then
2
π r1 1 1 1 −2k1 r 1 1
Φ̃(r1 ) = + 2r1 −1 + − + e + − .
r1 k12 a0 a0 k1 k1 2a0 k12 k1 2a0 k12
(21.15)
Substituing k1 = 3/(2a0 ), we obtain
4πa20
2
r1 2 4 3r1 4
Φ̃(r1 ) = − r1 − a0 e− a0 + a0 . (21.16)
9r1 a0 3 9 9
Then, after some algebra, we find
Z ∞
2 7 e2

2
3r
− a1 r1
A= 4πr1 dr1 e 0 Φ̃(r1 ) 1 −
4π(4π0 )a60 0 a0
73 Chapter 21

24 · 95 e2
=− , (21.17)
37 a0 0
so finally
e2 1 24 · 95

13
∆E = C + A = − − . (21.18)
a0 0 2 2 · 34 37
2) For the He atom with two electrons in the n = 2, l = 0 state, the ground state
is para-He, since there is no antisymmetric spatial wave function (the antisymmetric
part is zero, because the electrons are in the same state).
Thus the wave function is
ψ = ψn=2,l=0 (~r1 )ψn=2,l=0 (~r2 )χsinglet
3
1 1 r +r
− 1a 2 r1 r2
= e 0 1− 1− . (21.19)
π a0 a0 a0
The energy in the first approximation is
Ze20 e2
E = E (1) + E (2) = 2 = , (21.20)
a 0 n2 (4π0 )a0
and in the second approximation, we add
e20
∆E = hψ| |ψi
r12
r1 +r2
e−2 a0
2 2
e20
Z Z
3 3 r1 r2
= 2 6 d r1 d r2 1− 1− . (21.21)
π a0 r12 a0 a0
Thus we must calculate
2r2
e− a0
Z 2
r2
Φ̃(r1 ) = d3 r2 1− . (21.22)
r12 a0
It satisfies
2
r1
∆r1 Φ̃(r1 ) = −4π 1 − e−2kr1 , (21.23)
a0
where k = 1/a0 . It becomes
2
d2

r1
r1 Φ̃(r1 ) = −4πr1 1 − e−2kr1 , (21.24)
dr12 a0
with the particular solution of the type
r1 Φ̃p (r1 ) = (ar13 + br12 + cr1 + d)e−2kr1 . (21.25)
From the differential equation, we fix the coefficients as (substituting also k =
1/a0 )
b π c 3π d π
a = −π , = , 2 = , 3 = . (21.26)
a0 2 a0 4 a0 8
To this, we must add a general solution to the homogenous equation, r1 Φ̃h (r1 ) =
C1 r1 + C2 and, as before, we must have Φ̃(r1 → ∞) → 0, implying C1 = 0, and
74 Chapter 21

non-singularity at zero, i.e., the vanishing of the coefficient of 1/r1 , so C2 + d = 0,

i.e., C2 = −d. Finally then,
πa30
3
1 r12

r 3 r1 1 − 2r 1 1
Φ̃(r1 ) = − 13 + + + e a0
− . (21.27)
r1 a0 2 a20 4 a0 8 8
Then we obtain
e2
∆E = I, (21.28)
a 0 0
where, after some algebra, we find
Z ∞
x2

3 1 −2x 1 13
I≡ dx x(1 − x)2 e−2x −x3 + + x+ e − = − 10 . (21.29)
0 2 4 8 8 2
Then, all in all,
13 e2
∆E = − . (21.30)
210 a0 0
We can also check that the dependence on Z was linear, ∆E ∝ Z, so to restore
Z in the above, we just multiply with Z/2.
3) To find the approximation 3 to exercise 1, we must first restore the Z depen-
dence. The electron wave functions are
3/2
1 Z Zr
ψn=1,l=0 (~r) = √ e− a0
π a0
3/2
1 Z Zr r
ψn=2,l=0 (~r) = √ e− a0 1 − . (21.31)
π 2a0 a0
Then, replacing Z → Zeff , and calculating the symmetric spatial wave function,
we find
3

1 1 Zeff Z (r +r )
− eff a1 2 r1 r2
ψS (~r1 , ~r2 ) = √ 3/2 3 e 0 1− + 1−
22 πa0 a0 a0
r +r
Z 3 Z (r
eff 1 +r 2 ) 1 − 1 2

Moreover, from exercise 1,

e2 Z e2 1 24 · 95

13
∆E = hψs | |ψs i = C + A = − − , (21.36)
4π0 r12 2 a0 0 2 2 · 34 37
so that finally
2 2
Z · Zeff e20 Zeff e20 4π 1 24 · 95

5 Zeff e0 13
E=2 −2 + − − . (21.37)
16 a0 a0 2a0 2 2 · 34 37
Then, minimizing the energy over Zeff , δE/δZeff = 0, we obtain
24 · 95

2 1 13
Zeff = Z − 4π − − ≡ Z − σ. (21.38)
5 2 2 · 34 37
Substituing back in the energy, we find
e20 5
E=− (Z − σ)2 . (21.39)
a0 8
4) To find approximation 3 to exercise 2, we first write the symmetric spatial
wave function with Z put back in,
3
1 Z −Z
r1 +r2 Zr1 Zr2
ψs (~r1 , ~r2 ) = e 2 a0
1− 1− , (21.40)
π 2a0 2a0 2a0
and then replace Z by Zeff , to obtain
3

Zeff Zeff (r1 +r2 ) Zeff r1 Zeff r2
ψseff (~r1 , ~r2 ) = 3
e− 2a0 1− 1− . (21.41)
π(2a0 ) 2a0 2a0
Then, we easily check that we have
p~21 Z 2 e2
hψ2 | |ψs i = eff 0
2m 4 · 2a0
1 Z 1 Zeff
h in = 2
⇒ hψs | |ψ1 i = , (21.42)
r a0 n r1 4a0
so that finally the energy is (where the last term comes from exercise 2, with Z
replaced by Zeff )
2 2
Zeff e0 ZZeff e20 e20 4π · 13 Zeff
E=2 − 2 − . (21.43)
2a0 n2 a0 n2 a0 210 2
As at exercise 3, minimizing over Zeff , δE/δZeff = 0, we obtain
4π · 13
≡ Z + σ,
Zeff = Z + (21.44)
212
and putting it back in the energy, we obtain
(Z + σ)2 e20
E=− . (21.45)
a0
5) For the H2 molecule,
C ±A
E± (R) − 2E0 =
1 ± |S|2
76 Chapter 21

e2 e2 e2 e2
Z Z
C(R) = d3 r1d3 r2 − 0 − 0 + 0 + 0 |ψ(r1A )|2 |ψ(r2B )|2
r2A r1B r12 R
e20 e20 e20 e20
Z Z
3
A(R) = d r1 d r2 − 3
− + + ψ ∗ (r1A )ψ(r1B )ψ(r2A )ψ ∗ (r2B )
Z r2A r1B r 12 R
S = d3 rψ ∗ (rA )ψ(rB ) = hψrA |ψrB i ⇒ |S|2 < 1. (21.46)

We want to see that E+ (R) starts positive, decreases to a minimum below 0, then
goes up, asymptotically at infinity reaching 0 from below, while E− (R) decreases
until reaching 0 asymptotically from above.
The physical argument for that is that E+ corresponds to the symmetric spatial
wave function, which corresponds to the singlet spin function, ↑↓, which is stable,
so must have a minimum. On the other hand, E− corresponds to the antisymmetric
spatial wave function, thus to the triplet spin function, ↑↑, which is unstable, so
has no minimum. But at small R, C > A always, and the e20 /R term dominates, so
both start off positive. Then the only possibility is the one stated above.
6) Consider the 3 H atoms in a regular triangle. Why is it unstable? We have seen
that the stable H2 molecule is the one with spin singlet, ↑↓. But then by adding an
extra H, we will have ↑↑↓, which contains an unstable E− (R), so is unstable.
7) The Hamiltonian is
4 4 4 4
X ~2 X e20 X e20 X e20
Ĥ = − ∆i − + + . (21.47)
i=1
2m r
i,j=1 iAj
r
i,j=1 ij
R
i,j=1 Ai Aj

There are two potentially stable configurations:

-a regular tetrahedron, so RAi Aj = R is the same for all; and
-a star in a plane (regular triangle with an extra atom in the center of the
triangle), which is probably unstable to the tetrahedron.
22 Chapter 22

1) The relativistic particle Lagrangian is

r
2 ~v 2 p
L = −mc 1 − 2 = − (mc2 )2 − (mc~v )2 , (22.1)
c
~ and p0 = mc2 → mc2 − qA0 means we
so minimal coupling, p~ = m~v → m~v + q A
replace it by
q
~ 2.
L0 = − (mc2 − qA0 )2 − c2 (m~v + q A) (22.2)
d ∂L ∂L
Its equations of motion, dt v = ∂~
∂~ x , are
 
2
d  c (mvr + qAr ) 
q
dt ~ 2
(mc2 − qA0 )2 − c2 (m~v + q A)
q∂r A0 (mc2 − qA0 ) c2 q∂r As (mvs + qAs )
=q +q (. 22.3)
~ 2
(mc2 − qA0 )2 − c2 (m~v + q A) ~ 2
(mc2 − qA0 )2 − c2 (m~v + q A)

The canonically conjugate momentum is

∂L ~ r
mc2 (m~v + q A)
pr = =q , (22.4)
∂vr ~ 2
(mc2 − qA0 )2 − c2 (m~v + q A)

so the Hamiltonian is
X (mc2 − qA0 )2 q ~
H= pr v r − L = q + A · p~. (22.5)
~ 2 m
r (mc2 − qA0 )2 − c2 (m~v + q A)

If we define a Lorentz-invariant canonically conjugate momentum pµ generalizing

the p~ above, then we can write
q
H= Aµ p µ , (22.6)
m
and the Lagrangian equation of motion can be written as
d pr pµ
= q∂r Aµ , (22.7)
dt m m
which can be generalized to the 4-vector form
d
pν = q(∂ν Aµ )pµ . (22.8)
dt
77
78 Chapter 22

1 d d
Noting that √ = dτ , and and that we have defined
1−v 2 /c2 dt

muν + qAν
pν = q , (22.9)
~ 2
(mc2 − qA0 )2 − c2 (m~v + q A)
d dxµ
and that dt qAν = dt q∂µ Aν , we see that we would obtain the correct (relativistic)
Lorentz force law,
mduν
= q(∂ν Aµ − ∂µ Aν )muν , (22.10)
dτ
only if we neglect A0 and A ~ in the denominator of pµ . So in this case minimal
coupling almost works, but not quite.
We should note that the correct treatment is given by first noting that
r
2 2 dxµ dxν
Ldt = −mc dτ = −mc − ηµν dτ , (22.11)
dτ dτ
and then adding the coupling to electromagnetism as the term Aµ dxµ pulled back
R

on the particle worldline, so replacing by

dxµ
Z Z
S 0 = −mc2 dτ −ẋµ ẋν ηµν + dτ qAµ (xρ (τ ))
p
. (22.12)
dτ
2) At the quantum level, we should write a Schrödinger equation,

i~∂t ψ = Ĥψ. (22.13)

If we have the Hamiltonian from the previous exercise,

q
Ĥ = Aµ p µ , (22.14)
m
we saw that we do not expect a correct result.
Instead, the correct treatment is to write a ”square of the Schrödinger equation”,

(i~∂t )2 ψ = Ĥ 2 ψ , (22.15)

and add the relativistic relation E 2 = (mc2 )2 + (c~ p)2 , to obtain

h i
~ 2 − (mc2 )2 ψ = 0.
−~2 ∂t2 + ~2 ∇ (22.16)

Then introducing minimal coupling to Aµ , we have

" 2 2 #
iq iq
−~2 ∂0 − A0 + ~2 ∂i − Ai − (mc2 )2 ψ = 0. (22.17)
~ ~

~ · dS
~ 6= 0 through a surface S
R
3) If we have a nonzero magnetic flux, Φ = S
B
bounded by the conductor closed loop C, then
Z I
~ ~
Φ = (∇ × A) · dS =~ ~ · d~l 6= 0.
A (22.18)
S ∂S=C
79 Chapter 22

But the probability current density is

~ ∗ − q A|ψ|

~j = ~ ψ ∗ ∇ψ ~ − ψ ∇ψ ~ 2 ≡ ~j0 + ~j1 , (22.19)
2mi m
so we see that ~j1 6= 0 in the conductor loop, so I 6= 0 in the loop, so
I
I~ · d~l 6= 0. (22.20)
C

4) In order to prove that the free energy, in the London-London theory, is

Z ∞
1
~ 2 + λ2L µ20~j 2 ,
f= 2πr dr B (22.21)
µ0 0
we first write the (free) energy of the electromagnetic field,
!
Z ~2
B ~2
µ0 E
f0 = + , (22.22)
2µ0 2

and, in ~j = ~jn + ~js , we neglect the normal current ~jn , since it vanishes quickly in
the absence of external fields. For the same reason, we put E ~ = 0. Then ~j = ens~vs
(the superconducting current is composed of quasiparticles with charge e, effective
mass m∗ , density ns and velocity ~vs ), and the kinetic energy associated with it is
Z
~v 2
Z ~j 2 Z
µ0 λ2L~js2
f1 = ns m∗ s = m∗ 2s = , (22.23)
2 2e ns 2
where we have introduced, as in the text, µ0 λ2L ≡ m∗ /(ns e2 ). Then we obtain the
formula we needed for the total free energy in (22.21), from f = f0 + f1 .
Minimizing (22.21) over ~j, δf /δ~j = 0, we obtain
!
Z
δ ~
B
B~· + λ2L µ20~j = 0. (22.24)
δ~j
~
But, from ~js = −A/(µ 2 ~
0 λL ), taking ∇×, we obtain

~ = −µ0 λ2L ∇
B ~ × ~js . (22.25)
~ ~j =
Then δ B/δ ~
−µ0 λ2L ∇× δ(~x − ~y ) under the integral sign, and by partial inte-
~ × B,
gration in (22.24), we get ∇ ~ more precisely
~ ×B
+µ0 λ2L ∇ ~ + ~j = 0. (22.26)
~ of the above, then using again ∇
Taking ∇× ~ × ~j = − 1 2 B,
~ and that ∇
~ × (∇
~ ×
µ0 λL
~ =∇
B) ~ 2 B,
~ we finally get
~ 2B
λ2L ∇ ~ = B.
~ (22.27)
5) When the magnet Rgoes down through the middle of the superconducting ring,
the magnetic flux Φ = S B ~ · dS
~ increases as well. But since B~ = −µ0 λ2 ~js inside
L
the superconductor, we get that
Z Z I
~ ~ 2 ~ ~ ~
B · dS = −µ0 λL (∇ × js ) · dS = −µ0 λL 2 ~js · d~l , (22.28)
S S C=∂S
80 Chapter 22

so the superconducting current in the loop increases, as does B ~ in the loop, so

(see exercise 4) so does the total superconducting free energy f , so we obtain a
countering force Fx = −df /dx, with x being the vertical position of the magnet.
6) Consider the direction x perpendicular to the conducting plane. Then the
interacting Hamiltonian of the electromagnetic wave with the atoms is
q ˆ ~
Ĥint = − p~ · A , (22.29)
mc
and the electromagnetic wave has
~ = 2A(0)~ cos ω x − ωt ,

A (22.30)
c
where ~ is a transverse polarization, ⊥ ~x, so ~ · ~x = 0, thus ~ is in the plane. That
~ 6= 0, so there is a transition.
means that, in general, p~ · A
7) For a H atom in the n = 1, l = 0 state, the spin-orbit interaction is
e 1 dA0 ~ ~
Ĥint,LS = S·L, (22.31)
2m20 c2 r dr
~ 2 = 0 in this ground state, so L
but L ~ ·S
~ = 0, so there is no interaction.
23 Chapter 23

1) The magnetic flux is

Z Z I
~ · dS
B ~= ~ · d~l +
B ~ · d~l ,
B (23.1)
S S1 S2

and this gives the Aharonov-Bohm phase through

Z I
~ ~
B · dS = A~ · d~l. (23.2)
S C=∂S

However, since B ~ acts in opposite directions to dS ~ in the two loops of the figure
eight, if the size of the loops is identical, the two contributions to the magnetic flux
cancel, so there is no Aharonov-Bohm phase.
2) As we undo the figure eight into a circle, the total magnetic flux increases
from 0 to nonzero, and so does the Aharonov-Bohm phase, such that in the end
iq R
~ ~l
A·d
|ψ(t)i = e ~ C |ψi. (23.3)
~ i , i = 1, 2, ..., N , the generic Berry phase is a sum,
3) For positions R
Z
1X ~ (i) · dR
~i ,
γn = A n (23.4)
~ i

where each independent Berry connection is

~ (i)
A ~ ~ ~ ~ |n({R
n (Ri ) = i~hn({Rj })|∇R
~ j })i. (23.5)
i

4) For a H2 molecule in a constant electric field, moving slowly, in the Born-

Oppenheimer approximation, considering d~ = q~l being the electric dipole of the
polarized molecule (if such a thing happens), we can define the Berry connection
A ~
~ ~ = ihn(d(t))|∇ ~
~ ~|n(d(t))i , (23.6)
d d

and the nontrivial Berry phase

I
γ= A( ~ · dd~ ,
~ d) (23.7)

~ spans a nontrivial cycle in time.

which is nonzero if d(t)
5) For a constant d~ and time dependent E(t),
~ the interaction Hamiltonian is
~
Ĥint = E(t) · d~ , (23.8)
so the Berry connection is
~ n (E)
A ~ = ihn(E(t))|
~ ∇~ ~ |n(E(t))i
~ , (23.9)
E

81
82 Chapter 23

the Berry curvature is

F~n (E)
~ =∇
~ ~ ×A
E
~ n (E)
~ , (23.10)
~ is defined by
and the Berry magnetic field Fkn (E)

Fij = ijk Fkn , (23.11)

where
∂ ~ − ∂ Ai (E).
~
Fijn = ∂i Anj − ∂j Ani = Aj (E) (23.12)
∂Ei ∂Ej
∂
Since ∂E i
Ĥint = di , if we assume that there are eigenfunctions in the directions
~ called Oz, we obtain
of E,
di
Bin = . (23.13)
E2
Then, as in the case in the text, the Berry phase is
I Z Z Z ˆ
di 2 ~ di 2
γn = Ani dEi = Bin d2 Ei = − 2
d E i = −|d| 2
d Ei
C=∂S S S E E
~
= |d|∆Ω cos(E, ~ ,
~ d) (23.14)

where E ~ is in the center of the (small) S.~ Unlike the case in the text, we see the
appearence of the cosine, so the result is quasi-geometric.
B̂i
6) In order to prove that Bin = m~ B 2 in the case in the text, first consider the

definition of the Berry connection,

~
Ani = hn(B(t))| ~ ~ |n(B(t))i
∇ ~ , (23.15)
B

then the second-order perturbation theory formula (23.54) in the text adapted for
our case,
" #
X hn(B(t))|∂i Ĥ|m(B(t))i
n
Fij = i hm|∂j Ĥ|ni − (j ↔ i) . (23.16)
(Em − En )2
m6=n

Since
gµB g
∂i Ĥ = − Si = − µB σi , (23.17)
~ 2
If the spin S~ is aligned with the z direction (which we can always choose), then
|mi is an eigenstate of Sz , so hm|∂z H|ni ∝ hm|ni = 0 (since m 6= n). Therefore,
we obtain Fiz = 0, as we should. Moreover, the only nonzero component of Fij is
Fxy , so (using that Em − En = 2 g2 µB B, and the explicit forms of the σx = σ1 and
σy = σ2 Pauli matrices), for |ni = | + 1/2i fixed we have
i
Bzn = Fxy
n
= [h+1/2|σx | − 1/2ih−1/2|σy | + 1/2i − (x ↔ y)]
4B 2
1
= − 2. (23.18)
2B
83 Chapter 23

But there is a possibility to have also |ni = | − 1/2i, in which case we obtain
Bz = +1/B 2 . Al, in all, we have the formula we wanted,
B̂i
Bin = −m , (23.19)
B2
with m = ±1/2.
7) In the covariant non-Abelian case, we can define the non-Abelian Berry con-
nection associated with states |mi of the same energy (degenerate) as
(nm)
Ai (~k) = ihn, ~k(t)|∇
~ k |m, ~k(t)i ,
i
(23.20)
and in the usual way the field strength of the non-Abelian connection,
(nm) (nm) (nm) (kl) (qr)
Fij (~k) = ∂i Aj − ∂j Ai + f (nm) (kl)(qr) Ai Aj , (23.21)
in terms of the structure constants f with inidices (mn) in the adjoint of U (N ),
leading to the usual F = dA + A ∧ A.
24 Chapter 24

1) For a stationary, transverse gauge transformation, Λ = Λ(x, y), and B = Bz , we

have
~ = ∇Λ(x,
δA ~ y) , (24.1)

and the Schrödinger equation transforms to

" 2 2 #
~2 iq iq p2z i
Ĥψ = − ∂x − (Ax − ∂x Λ) + ∂y − (Ay − ∂y Λ) − 2 Xn (x, y)e ~ zpz
2m ~c ~c ~
= Eψ , (24.2)

so there is no more factorization in the (x, y) wave function.

2) The gauge transformation δ A ~ = ∇Λ(x,
~ y) rotates A~ in the (x, y) plane. The
(shifted) harmonic oscillator rotates as well, like the classical motion in the (x, y)
plane, which is also a rotation.
3) For the 2+1 dimensional system with Lx , Ly , in the IQHE, each plateau has
a different n, corresponding to σn = nσ0 , but n is the number of fully occupied
Landau levels. On the other hand, the degeneracy of the Landau levels (independent
Φ
of the Landau level) is Nmax = h/e , with Φ = BLx Ly the magnetic flux.
However, the quantization of Φ, Φ = Nmax Φ0 , also leads to the quantization
of σ, albeit indirectly, via what is called Laughlin’s gauge principle: If we change
B such that ∆Φ = Φ0 in a topology with two circular boundaries, we shift all
the Landau levels, effectively moving the electrons from one circular boundary to
another, leading to the quantization of σ.
The length of the plateau, in B (as B increases), is determined by ∆Φ = Φ0 , so

h/e
∆B = . (24.3)
Lx Ly

4) In the FQHE, the fraction of occupied states in the Landau level is

ne
ν= ∈
/ Q. (24.4)
nB
That means that there is not an integer number of electrons (occupied states) in
the Landau level, meaning the electronic states are quasi-particles.
5) If we have a conductivity matrix

σxx σxy
σij = , (24.5)
−σxy σxx
84
85 Chapter 24

then the resistivity matrix is its inverse,

1 σxx −σxy
Rij = (σ −1 )ij = 2 2
. (24.6)
σxx + σxy σxy σxx
Using σ ≡ σxy + iσxx , this can be rewritten as
1 σxy − iσxx
= 2 2
. (24.7)
σ σxx + σxy
Then the symmetry exchanging ji with Ei , ji ↔ Ei , replaces σij ↔ (σ −1 )ij =
Rij , or σ ↔ 1/σ.
6) The Landé g-factor is
J(J + 1) − L(L + 1) − S(S + 1)
g =1+ , (24.8)
2J(J + 1)
where, since J~ = L
~ + S,
~ and S = 1, we have J = L − 1, L, L + 1, so the possible
values of g are
L(L − 1) − L(L + 1) − 2 L+1
g =1+ =1−
2L(L − 1) L(L − 1)
1
=1−
L(L + 1)
(L + 1)(L + 2) − L(L + 1) − 2 L
=1+ =1+ . (24.9)
2(L + 1)(L + 2) (L + 1)(L + 2)
We note that in the classical limit L → ∞, all of the g → 1, as we wanted.
7) If an atom with S, L, J moves in B(t) moving on the curve C, in general the
Berry phase is
I
γn = ~ B)
A( ~ · dB
~ , (24.10)
C

where
~ B)
A( ~ = hn(B(t))|
~ ~ ~ |n(B(t))i.
∇ ~ (24.11)
B

But since the energy difference in the magnetic field is

∆E = gµB B(t)Mj , (24.12)
we have that
Jˆi
∂Bi Ĥ = −gµB . (24.13)
~
On the other hand, we have the same formula as in the spin 1/2 case (which we
will prove afterwards),
B̂i
Bin = nJ ~ , (24.14)
B2
~ Then, the Berry phase is
where B̂ is the unit vector in the direction of B.
I Z
1 1
γn = Ani dBi = B n d2 Bi
~ C=∂S ~ S(C) i
86 Chapter 24

Z Z
B̂i 2
= −mJ d Bi = −mJ dΩ , (24.15)
S(C) B2 C

just like in the spin 1/2 case. We note that the formula is actually independent on
g!
It remains to prove (24.14). As before, we write (substituting ∂i Ĥ and Em in the
general formula)
" Ŝ
#
n i X hn(B(t))| Ŝ~i |m(B(t))ihm(B(t))| ~j |n(B(t))i
Fij = 2 − (↔ j) . (24.16)
B ∆m2J
m6=n

where we used
gµB
∂i Ĥ =Ŝi , Em − En = ∆mJ gµB B , (24.17)
~
and we note that gµB cancels between the numerator and denominator.
Then we note that Fzi = 0, where B = Bz , and that
" Ŝy
#
Ŝx
i X hn J | |m J ihm J | |nJ i
Bzn = Fxy
n
= 2 ~ ~
− (x ↔ y) , (24.18)
B (mJ − nj )2
mJ 6=nJ

and therefore Bin = B̂i Fxy

n
. Since

1) Expanding in ~/i the Schrödinger equation, we find

2 3
0 2 ~ 0 0 ~ 0 0 0 2 ~
(s0 ) + 2s0 s1 + [2s2 s0 + (s1 ) ] + [2s01 s02 + s00 s03 ]
i i i
2 3
~ 00 ~ 00 ~
−2m(E − V (x))] + s0 + s1 + s002 = 0. (25.1)
i i i
In the text the 0th order and 1st order in ~/i were solved,
p
(s00 )2 − 2m(E − V (x))
p = 0 ⇒ s00 = 2m(E p − V (x))
00 0 0 0 0
s0 + 2s0 s1 = 0 ⇒ ( 2m(E − V (x))) + 2s1 2m(E − V (x)) = 0 ⇒
1 p const.
s1 = − ln 2m(E − V (x)) + C ⇒ es1 = . (25.2)
2 [E − V (x)]1/4
Now we continue, with

2s02 s00 + (s01 )2 + s001 = 0 , (25.3)

and find
s001 + (s01 )2
s02 = −
2s00 ( )
1 1 p 00 1 p 0 2
= −p − ln 2m(E − V (x)) + ln 2m(E − V (x)) .
2m(E − V (x)) 2 4
(25.4)
2
2) For a harmonic oscillator in 1 dimension, V (x) = k x2 , we have two turning
points x1 , x2 , in the first the potential is going down, and in the second, it is going
up. So we can use the formulas in the text for the turning point identifications, at
x1
" s #
1 x1 0
Z 2
1 x
x2 1/4 exp − ~ dx 2m k
2
−E
k 2 −E x
√ " Z r #
2 1 x 0 x2 π
→ 2 1/4
sin dx 2m(E − k ) + , (25.5)
E − k x2 ~ x1 2 4

and at x2 (from the right to the left)

" s #
1 x 0
Z 2
1 x
x2 1/4 exp − ~ dx 2m k
2
−E
k −E 2
x2

87
88 Chapter 25

√ " Z r #
2 1 x2 0 x2 π
→ 2 1/4
sin dx 2m(E − k ) − . (25.6)
E − k x2 ~ x 2 4

3) Assuming a wave function centered on one of the minima, then x1 and x2 are
turning points on each side of the minimum, so at x1
1 x1 0 p
Z
1
1/4
exp − dx 2m (V0 cos ax − E)
[V0 cos ax −
√ E]
~ x
Z x
2 1 0
p π
→ 1/4
sin dx 2m(E − V0 cos ax) + , (25.7)
[E − V0 cos ax] ~ x1 4

and at x2 (from the right to the left)

1 x 0p
Z
1
1/4
exp − dx 2m (V0 cos ax − E)
[V0 cos ax −
√ E]
~ x
Z 2x2
2 1 0
p π
→ 1/4
sin dx 2m(E − V0 cos ax) − . (25.8)
[E − V0 cos ax] ~ x 4

4) For the potential V = −|α|/r3 , with effective potential

~2 |α| ~2 l(l + 1)
Veff = V + l(l + 1)r2 = − 3 + , (25.9)
2µ r 2µ r2
the extremum r0 of the potential (maximum: it grows from −∞ to a positive value,
then goes down to 0) is at

0 3|α| 2µ
Veff (r0 ) = 0 ⇒ r0 = , (25.10)
2l(l + 1) ~2
and the value at this maximum is
4 ~2 l(l + 1)
Veff (r0 ) = . (25.11)
27 2µ |α|2
Define
p
k(r) = p2m(E − Veff (r)) (region III)
k̃(r) = p2m(E − Veff (r)) (region I)
κ(r) = 2m(Veff (r) − E) region II). (25.12)
Then at r > r2 (region III), we have
1 i Rrr dr0 k(r0 ) −i r dr 0 k(r 0 )
R
ψIII (r) = p e 2 + e r2 , (25.13)
k(r)
and the turning point formula at r2 means that we have in region II (r1 < r < r2 )
Z r2
1 τ 0 0
ψII (r) = p e exp − dr κ(r ) , (25.14)
κ(r) r

where
Z r2
τ= dr0 κ(r0 ). (25.15)
r1
89 Chapter 25

Then the turning point formulas at r1 means that we have in region I (r < r1 )
A i R r1 dr0 k̃(r0 )+β Rr 0 0

+ e−i r dr k̃(r )+β .
1
ψI (r) = q e r (25.16)
k̃(r)
Note that these formulas are the same ones as in the text, in the case of a well
potential.
5) Matching ψ and ψ 0 at r = r1 now gives us A = eτ .
The transmission coefficient is
|jtrans |
T = , (25.17)
|jinc |
where

~j ≡ ~ ψ ∗ ∇ψ
~ − ψ ∇ψ
~ ∗ . (25.18)
2mi
Then

~jinc = |A|2 ~k̃(r) ~er = |A|2 ~ ~er

m q 2 m
k̃(r)

~jtrans = ~k(r) ~er = ~ ~er , (25.19)

m p 2 m
k(r)

so the transmission coefficient is

1 1 −2 rr2 dr 0 κ(r 0 )
R
T = = = e 1 . (25.20)
|A|2 e2τ
6) For V = −V0 cos(ax) and ax0 < π/2,
V 00 (x) = V0 a2 cos(ax) ' V0 a2 , (25.21)
so mω 2 ' V0 a2 , then
r
V0
ω'a . (25.22)
m
Then
Z
1
e−i(n+ 2 )ωT .
i
X 1
ψWKB (T ) = dx0 e ~ Scl √ = K̃(T ) (25.23)
det O n≥0

7) In this Minkowski space calculation, we cannot truncate, since the term

− ~i En T
e is a phase.
However, in a Wick rotation to Euclidean space (see later chapters), we can
1
truncate, since e− ~ En T decays strongly if En~T 1, and then
ωT E0 T
U (T ) ' K̃(T )e− 2 = K̃(T )e− ~ . (25.24)
26 Chapter 26

1) We have
q̇ 4 q̇ 2 (q̇ 2 + α)2 α2
L= + α − λq 4 = − − λq 4 . (26.1)
4 2 2 4
Then the conjugate momentum is
∂L
p= = q̇(q̇ 2 + α) , (26.2)
∂ q̇
and the quantization condition is
I I
1 1
pdq = n + h⇒ q̇(q̇ 2 + α)q̇dt = n + h. (26.3)
C 2 C 2
2) For the potential V = k|x|, at x → ±∞, we have
1 x 0p
Z
1
ψIII,W KB = exp − dx 2m(k|x| − E)
[2m(k|x| − E)]1/4 ~ x2
const. 4mk 3/2 2mE 1/2
→ exp − x − x . (26.4)
(2mk|x|)1/4 3~ ~
On the other hand, at x → 0, we have
Z x
1 1 0
p π
ψII,W KB (x) = cos dx 2m(E − k|x|) −
[2m(E − k|x|)]1/4
~ x1
r 4
1 2mEn k m 2
→ cos x− x +δ . (26.5)
(2mEn )1/4 ~ 2E 2En
3) For a particle in a box, −L/2 ≤ x ≤ L/2, the Bohr-Sommerfeld quantization
gives
Z +L/2 √ √

0 1
dx 2mE = 2mEL = n + π~ , (26.6)
−L/2 2
so
2
π 2 ~2

1
En = n+ , (26.7)
2 2mL2
so we only need to replace n + 1/2 with n in order to get the exact result. That
means that at large n, we get the correct result.
Using this En , the WKB wave function becomes
" Z #
1 1 L/2 0 p π
ψ(II)W KB (x) = cos dx 2mEn −
(2mEn )1/4 ~ −L/2 4

90
91 Chapter 26

s
L 1 πx nπ
= cos n+ + . (26.8)
π~(n + 1/2) 2 L 2
Then for n odd, we get ∓ sin[(n + 1/2)πx/L], and for n even, we get ∓ cos[(n +
1/2)πx/L].
That means that the wave function is OK, except for the normalization constant,
and an overall phase in the cosine.
4) For the harmonic oscillator in the Bohr-Sommerfeld quantization, in the clas-
sical n → ∞ limit, we get
mω 2 x2
ψn (x) ' const.xn e− ~ 2 , (26.9)
and then the wave function is approximately exact in the large x limit only. But
then, large x is allowed (classical oscillators reach there), and xn can be of the
mω 2 x2
same order of magnitude as e ~ 2 , so |ψn (x)|2 ∼ 1 (large probability there, since
a classical oscillator can reach there).
5) To write the equations for R2 (r) and Θ2 (θ), we need to keep the next order
in (26.29). This is
2 " 2 2
~ ∂s1 ∂s0 ∂s2 1 ∂s1
∆s1 + +2 + 2
i ∂r ∂r ∂r r ∂θ
2 #
2 ∂s0 ∂s2 1 ∂s1 2 ∂s0 ∂s2
+ 2 + 2 2 + 2 2 = 0.
r ∂θ ∂θ r sin θ ∂φ r sin θ ∂φ ∂φ
(26.10)
The separated variable ansatz is
s0 = R0 (r) + Θ0 (θ) + Φ0 (φ)
s1 = R1 (r) + Θ1 (θ) + Φ1 (φ)
s2 = R2 (r) + Θ2 (θ) + Φ2 (φ). (26.11)
But ∂s0 /∂φ = ∂Φ0 /∂φ = L3 and then d2 Φ0 /dφ2 = 0, and in turn, it means we
can choose ∂s1 /∂φ = ∂Φ1 /∂φ = 0, since we need d2 Φ1 /dφ2 = 0. Then, since
d2 R0 1 d2 Θ0

2 dR0 dΘ0
∆s0 = + + 2 + cot θ
dr2 r dr r dθ2 dθ
2 2
d R1 2 dR1 1 d Θ1 dΘ1
∆s1 = + + 2 + cot θ , (26.12)
dr2 r dr r dθ2 dθ
we find the (separated) equations
2
d2 R1 (r) 2 dR(r)

dR1 dR0 dR2
+ + + =0
dr2 r dr dr dr dr
2 2
1 d Θ1 dΘ1 1 dΘ1 2 dΘ0 dΘ2
+ cot θ + 2 + =0
r2 dθ2 dθ r dθ r dθ dθ
2L3 dΦ2
=0, (26.13)
r2 sin2 θ dφ
so again we can choose Φ2 (φ) = 0.
92 Chapter 26

6) For the central potential V = −α/r2 , there is only one positive solution to
the vanishing of the E − Veff (r),
2mE − L2
2mE + =0, (26.14)
r2
(and only for 2mE < L2 do we have even one solution), since the effective potential
starts at +∞ and decays monotonically to zero.
That means that there is only one condition imposed on the wave function (at
infinity), not two (at zero and infinity), so there is no quantization, all positive
energies are possible!
7) For the hydrogenoid atom, the radial wave function is
" r #
1 r 2mZe20 L2
Z
1
Rn (r) = exp − dr 2m|E| + − 2
[2m(E − Veff (r))]1/4 ~ r2 r r
√
" p #
2
const. m Ze 0 2m|En |
' r− 2En ~ exp − r . (26.15)
(2m|En |)1/4 ~

Taking into account that |En | = mZ 2 e20 /(2~2 n2 ), we find

const.
Rn (r) ' r−n e−kn r . (26.16)
(2m|En |)1/4
But the exact behaviour at infinity is
Rn (r) ∼ e−kn r rl 1 F1 (−n + l + 1, ...) ∼ e−kn r rn−1 . (26.17)
That means that only the leading behaviour is correct, the first subleading term
is incorrect.
When n → ∞ (would-be classical limit), we can only say that the leading term
becomes more leading, since kn = k0 /n, so e−kn r = e−k0 r/n , but otherwise there is
no improvement in the behaviour.
27 Chapter 27

1) The answer is yes, if there are both electric and magnetic sources, i.e., if the
action is Z Z
4 1
S = − d x Fµν F + d4 x(j µ Aµ + k µ Ãµ ) ,
µν
(27.1)
4
where j µ is the electric current, k µ is the magnetic current, and Ãµ is the dual
gauge field, such that the dual field strength is F̃µν = ∂µ Ãµ − ∂ν Ãµ . Then the
Maxwell duality is
Aµ ↔ Ãµ , j µ ↔ k µ , Fµν → ∗Fµν = F̃µν . (27.2)
For a set of particles minimally coupled to electromagnetism, their electric/magnetic
charges must also be dual, Qe ↔ Qm . But then, in the quantum theory, this is only
possible if the particle spectrum is also invariant.
2) For an infinitesimally thin, planar, infinite perfect conductor, we will have
that the parallel electric field at the surface of the conductor, E ~ || = 0, and the
transverse magnetic field at the surface of the conductor, B ~ ⊥ = 0.
Indeed, if the conductor is perfect (σ = ∞), inside it E ~ =E ~ || = 0 (since otherwise
~j = ∞, which is impossible), and B ~ ⊥ = 0 (since otherwise, electrons would feel a
parallel Lorentz force that would have nothing to balance it, so the same would
happen).
But then, we can use the Maxwell equations to prove the continuity conditions
at a surface, applied here to the air/conductor surface, which are ~n × ∆E ~ = 0,
~n · ∆B = 0 (so no change in E ~ || and B ~ ⊥ at the interface), as well as ~n × ∆H ~ =
~j, ~n · ∆E = σ/0 .
The first relation leads to the fact that E ~ || = 0 just outside of the surface of
the conductor, which leads to the familiar fact that an electric charge put near the
conductor will generate a mirror charge (in reality, the charges in the conductor
rearrange to create this illusion) of same value and opposite sign (Q0e = −Qe ),
situated at an equal distance from the conductor, on the other side.
To prove the second relation, one considers a rectangular contour with two lines,
one inside the conductor, one outside, both parallel to the interface, and closed by
two very small lines perpendicular to the surface (which can then be neglected).
Then the Maxwell equation ∇ ~ ×E ~ = − ∂ B, ~ integrated over the contour’s surface
∂t
gives
Z I
~ ~ ~ ∂ ~ ~
0 = (∇ × E)dS = − B · dl. (27.3)
S C=∂S ∂t
~ is parallel to the surface S (which is transverse
The left-hand side is zero since E
93
94 Chapter 27

to the surface of the conductor), due to the previous relation. Then we get that
∂ ~
∆ ∂t B = 0 on the two sides of C (inside and outside the conductor), which really
means (for a static situation) ∆B ~ ⊥ = 0. But since we already knew that B ~⊥ = 0
~
inside the conductor, it follows that B⊥ = 0 just outside it, too.
Then in the presence of the conductor the magnetic charge gets also a mirror
charge, just of the same sign (Q0m = +Qm ).
So, for an electric charge e and a magnetic charge g situated at the distance R
from the conducting plane, there are mirror charges −e and +g on the other side,
at the same distance R, leading to

~ =E
~ ⊥ = 2E
~1 = 2e ~ =B
~1 − B
~1 = B
~ || = 0.
E ~e⊥ , B (27.4)
4π0 R2
3) Consider the two particles (dyons, of identical masses m) with vectors ~r1 and
~r2 . Then the time derivative of the total angular momentum is
~
dL
= m(~r1 × ~r¨1 + ~r2 × ~r¨2 ). (27.5)
dt
They each feel Lorentz forces due to the other particle, so

m~r¨1 = q1 (E
~2 + ∇
~ ×B
~ 2) , m~r¨2 = q2 (E
~1 + ∇
~ ×B
~ 1 ). (27.6)
~ = ~r1 − ~r2 (so is the vector from particle 2 to particle 1), and choose a
Define R
reference frame where (instantaneously, at the moment t) ~v2 = 0. Then for particle
2 we can use the usual formulas,

~ 2 = q2 R̂ , B
E ~ 2 = µ0 g2 R . (27.7)
4π0 R 2 4π R2
For particle 1, however, which now has a generically nonzero velocity ~v1 , we must
use the Lorentz transformation rules,
~0 = E
E ~ || E
~⊥0 ~⊥ − β
= γ(E ~ × B)
~
||
~ =B
B 0 ~ || B
~ = γ(B
0 ~ ⊥ + β~ × E).
~ (27.8)
|| ⊥

At small enough velocities, we can ignore v 2 /c2 corrections, so put γ = 1. Then,

also denoting by E ~ 1, B
~ 1 the fields in the rest reference frame for the particle 1, the
fields in our reference frame are
~0 ' E
E ~ 1 − ~v1 × B
~1 , B
~0 ' B
~ 1 + ~v1 × B.
~ (27.9)
1 1

Then
m~r¨2 = q2 (E
~ 10 + ~v2 × B
~ 20 ) = q2 (E
~ 1 − ~v1 × B
~ 2 ). (27.10)

Finally then
~
dL
= q1 (~r1 × E~ 2 + ~r1 × (~v1 × B
~ 2 )) + q2 (~r2 × E~ 1 − ~r2 × (~v1 × B~ 1 ))
dt
q1 q2 ~ + ~r1 × (~v1 × q1 B~ 2 ) − ~r2 × (~v1 × q2 B
~ 1)
= (~r1 − ~r2 ) × R
4π0
95 Chapter 27

!
µ0 q1 g2 R̂ µ0 q2 g1 R̂
= −~v1 × ~r1 × 2
+ ~r2 × . (27.11)
4π R 4π R2

Now choosing (instantaneously, at time t) ~r2 = −~r1 (so the origin to be at the
midpoint between the two particles), so that R2 = 4r12 , we have
!
~
dL µ0 R̂
= (q1 g2 − q2 g1 )~r1 × ~v1 × 2 . (27.12)
dt 4π R

But, as in the text, we have

~ − Ri~r1 · ~r˙1
ṙ1i~r1 · R
(~r1 × (~r1 × R̂))i = . (27.13)
R3
On the other hand, we have

d Ri Ṙi R2 Ri ~ ~˙ ~ · ~r˙1
ṙ1i R2 − Ri R
= 3
− 3R ·R= 3
, (27.14)
dt R R r R

where we have used Ṙi = ṙ1i (since v2 = 0).

Then we finally find
!
~
d ~ tot − µ0 q1 g2 − q2 g1 R1
L =0, (27.15)
dt 4π 2 R

~ tot must be an
and by the same argument as in the text, the extra term added to L
integer N times ~/2, which gives

q1 g2 − q2 g1 = 4π~N , (27.16)

which is the DSZ quantization condition, except for a factor of 2 (we only get it for
even N ).
4) We have eg2 − q2 h/e = h, so g2 = h/e(1 + q2 /e). Assume particle 2 is at rest,
then

~ 2 ) = q1 q2 R̂ + ~v1 × µ0 q1 g2 R̂
~ 2 + ~v1 × B
mr̈1 = q1 (E
4π0 R2 !4π R
2

e h R̂ µ0 h q2 R̂

= + ~v1 × 1+ . (27.17)
4π e0 R2 e e R2

~
5) For B(x) = 2θ 2
e δ (~
x),
the Berry phase is
I I
γ=e ~ ˙~
Adl = ~ · dS
B ~ = 2θ. (27.18)
C=∂S S

Then eiγ = e2iθ , but rotating one particle around the other amounts to a double
interchange of the particles. That means that we get the phase eiθ for interchanging
of identical particles. This is the definition of an ”anyon”, a particle with a fractional
statistics (a phase different than ±1 at interchanging identical particles).
6) The sphere has the monopole and antimonopole on it, at opposite sides. Then
96 Chapter 27

the total magnetic charge inside the sphere is zero, even if we add (part of) the m
and m̄ charges. But then the magnetic flux through the sphere is zero, giving
I I
0= (∇~ × A)
~ · dS
~= A~ · d~l , (27.19)
S2 ∂S 2

which contains no contradictions (the integral is zero, since the sphere has no bound-
ary). So a unique A~ is well defined on the sphere.
7) Yes, we can still infer that the electron charges are quantized. The Dirac
quantization can be defined by a surface going in between the monopole and anti-
monopole, so it is still valid.
28 Chapter 28

1) The anti-Feynman propagator is

0
dp e−ip(t−t )
Z
¯ F (t, t0 ) =
∆ . (28.1)
C 2π p2 − ω 2 − i
We start with C = R, and then add a half-circle at infinity to close it, such that
the half-circle has vanishing contribution.
The contours are almost the same as for the Feynman propagator ∆F , meaning
that for t > t0 we must close the contour in the lower-half plane, as in Fig.21.
Except that now, due to ω 2 → ω 2 + i, instead of ω 2 → ω 2 − i, the contour avoids
the poles at ω = ±ω oppositely as in Fig.21: −ω is avoided from above (since the
pole is really −(ω + i/2)), and +ω is avoided from below (since the pole is really
ω + i/2).
But then, for t > t0 , we have inside the contour only the pole p = −ω + i, so
when using the Cauchy theorem, we obtain
1 +iω(t−t0 )
− e , (28.2)
2ω
and for t < t0 , we only change t − t0 → t0 − t (the sign in front would change, except
the sign of the contour changes as well, into counterclockwise). Therefore all in all,
we have
¯ F (t, t0 ) = − 1 e+iω|t−t0 | .
∆ (28.3)
2ω
Then the boundary conditions are: -at t → +∞,,
¯ F ∼ −e+iωt ,
q(t) ∼ ∆ (28.4)

and -at t → −∞,

¯ F sin −e−iωt .
q(t) ∼ ∆ (28.5)

This is the opposite of the boundary conditions for ∆F .

2) If the contour avoids the poles slightly above the real line, then for t > t0 ,
when again we must add the half-circle at infinity in the lower-half plane in order
to close it, we have both poles (p = ±ω) inside the closed contour, so by using the
Cauchy theorem we get
1 −iω(t−t0 ) 1 +iω(t−t0 )
∆(t, t0 ) = + e − e . (28.6)
2ω 2ω
For t < t0 , we must close the contour in the upper-half plane, so there will be no
97
98 Chapter 28

poles inside the closed contour, so the result is zero, ∆(t, t0 ) = 0. Therefore, all in
all, we have

0 0 1 −iω(t−t0 ) 1 +iω(t−t0 )
∆(t, t ) = θ(t − t ) + e − e . (28.7)
2ω 2ω
In this case, the boundary conditions are: at t → +∞, we get
e−iωt − eiωt
q(t) ∼ ∆ ∼ , (28.8)
2
whereas -at t → −∞, we get
q(t) ∼ ∆ ∼ 0. (28.9)

3) The equation to be solved is

d2

− 2 + ω K(τ, τ 0 ) = δ(τ, τ 0 ) ,
2
(28.10)
dτ
and the solution is for a periodic function, K(τ, τ 0 ) = ∆(τ − τ 0 ), ∆(τ − β) = ∆(τ ).
But the solution of a second order differential equation on an interval is unique.
So all we need to do is show that the solution we have satisfies the conditions.
1
(1 + n(ω))e−ωτ + n(ω)e+ωτ , τ ∈ (0, β) ,

∆free (τ ) =
2ω
1
n(ω) = βω . (28.11)
e −1
But then, using the condition ∆(τ −β) = ∆(τ ), we can extend it to τ ∈ (−β, +β),
1 h
−βω+ω(τ −τ 0 )
−ω(τ −τ 0 )
∆free (τ − τ 0 ) = e βω
e + θ(τ − τ 0
) 1 − e −βω
e
2ω(eβω − 1) i
0 0
+eβω+ω(τ −τ ) + θ(τ − τ 0 ) 1 − eβω eω(τ −τ )

(28.12)

Then we can check that for τ − τ 0 > 0,

2
d
+ ω ∆free (τ − τ 0 ) = 0 ,
2
(28.13)
dτ 2
and around τ − τ 0 = 0, we have
d2 0 θ00 (τ − τ 0 ) h βω −ω(τ −τ 0 ) βω
ω(τ −τ 0 ) i
∆ free (τ − τ ) = e − 1 e + 1 − e e
dtτ 2 2ω(eβω − 1)
0 0
θ (τ − τ )
+ω 2 ∆free (τ − τ 0 ) + −ω eβω − 1 + ω eβω − 1

βω
2ω(e − 1)
= ω 2 ∆free (τ − τ 0 ) + δ(τ − τ 0 ). (28.14)

4) For the Euclidean integral

Z Z
dE1,E dE2,E 1
I(EE ) = 2 ,
2π 2π (E1,E + ω12 )(E2,E
2 + ω22 )[(E1,E + E2,E − EE )2 + ω32 ]
(28.15)
99 Chapter 28

the Wick rotation back to Minkowski space is t = −itE , so EE = −iE, but in fact
we must rotate with only π/2 − angle, so EE → −i(E + i0 ), and
2
EE + ω 2 → −EE2
+ ω 2 − i ⇒
(E1,E + E2,E − E)2 + ω32 → −(E1 + E2 − E)2 + ω32 − i , (28.16)

so that finally the integral becomes

Z Z
dE1 dE2 1
I(E) = .
2π 2π (−E12 + ω12 − i)(−E22 + ω22 − i)[−(E1 + E2 − E)2 + ω 2 − i]
(28.17)
5) The harmonic oscillator partition function is
Z Z
1 0 0 0
ZE [J] = N exp dt dt J(t)∆E (t, t )J(t ) . (28.18)
2
The Euclidean space 4-point function for the harmonic oscillator is
δ 4 ZE [J]
G4,E (t1 , t2 , t3 , t4 ) =
δJ(t1 )δJ(t2 )δJ(t3 )δJ(t4 )
= ∆E (t1 , t2 )∆E (t3 , t4 ) + ∆E (t2 , t3 )∆E (t1 , t4 )
+∆E (t1 , t3 )∆E (t2 , t4 ). (28.19)

6) For the even function (”superfield”)

√
Φ(x, θ) = φ(x) + 2θα ψα (x) + θ2 F (x) , (28.20)

where θ2 = αβ θα θβ , in order to calculate

Z
d2 θ(a1 Φ + a2 Φ2 + a3 Φ3 ) , (28.21)

where d2 θ = − 14 dθα dθβ αβ , we need to remember that dθ1 = 0 and θ12 =
R R R

θ22 = 0 = (θ2 )2 , which means that we will only select the terms with two thetas,
specifically with θ1 θ2 , from the integrand. Moreover, note that
Z Z
1
d2 θθ2 = − dθ1 dθ2 (2θ2 θ1 ) = +1. (28.22)
2
Then
Z
d2 θa1 Φ = a1 F (x) ,
Z Z
d2 θa2 Φ2 = a2 [2φ2 F (x) + 2 d2 θ(θα ψα (x))2 ]
Z Z
= a2 [2φ(x)F (x) − dθ1 dθ2 ((θ1 ψ1 )2 + (θ2 ψ2 )2 + 2θ1 ψ1 θ2 ψ2 )]

Z = 2a2 [φ(x)F (x) − ψ1 (x)ψ2 (x)]

Z
d2 θa3 Φ3 = a3 [3F (x)φ2 (x) + 3 · 2φ(x) d2 θ(θα ψα (x))2 ]
= 3a3 φ(x)[F (x)φ(x) − ψ1 (x)ψ2 (x)]. (28.23)

7) We divide the interval from t to t0 into N + 1 steps, and use the completeness
100 Chapter 28

relations Z
dψi dψ̄i |ψi ihψ̄i |e−ψ̄i ψi = 1l , (28.24)

and insert them at each intermediate point, using

q.e.d.
29 Chapter 29

1) For the Lagrangian

1 2
(q̇ + q̇22 ) − α(q12 + q22 )2 ,
L= (29.1)
2 1
the momenta and Hamiltonian are
p1 = q̇1 , p2 = q̇2 ⇒
1
H = (p21 + p22 ) + α(q12 + q22 )2 . (29.2)
2
The primary constraint is
φ1 = q1 + βq2 = 0 , (29.3)
so we obtain the secondary constraint from
{φ1 , H}P.B. = {q1 + βq2 , H}P.B. = p1 + βp2 ≡ φ2 ≈ 0. (29.4)
Then we can check that
{φ2 , H}P.B. = {p1 + βp2 , H}P.B. = −2α(q1 + βq2 )(q12 + q22 ) = −2αφ1 (q12 + q22 ) ,
(29.5)
so gives nothing new. So the constraints are φ1 and φ2 . Moreover,
{φ1 , φ2 }P.B. = 1 + β 2 6= 0 , (29.6)
so both φ1 and φ2 are second class constraints.
2) If we have an approximately constant gravitational acceleration, like on the
surface of the Earth, we can say that U = mgq2 , where q2 is the height. Indeed
then F~ = −∇U
~ = −mg~eq .
2

For motion on a circle of radius R in a vertical plane, we have the (primary)

constraint
q12 + q22 − R2 = φ1 = 0. (29.7)
The Lagrangian is
m 2 m
L= (q̇1 + q̇22 ) − U = (q̇12 + q̇22 ) − mgq2 , (29.8)
2 2
and for momenta p1 = q̇1 , p2 = q̇2 , the Hamiltonian is
1 2
H = p1 q̇1 + p2 q̇2 − L = (p + p22 ) + mgq2 . (29.9)
2m 1
Then
HT = H + uφ1 . (29.10)
101
102 Chapter 29

The (first) secondary constraint is obtained from

2
{φ1 , HT }P.B. = (p1 q1 + p2 q2 ) = φ2 ≈ 0. (29.11)
m
Then
2 2 4
{φ2 , HT }P.B. = 2
(p1 + p22 ) − gq2 + u(q12 + q22 ) , (29.12)
m m
but it is not a new constraint (can be obtained from the previous two).
Then
4
{φ1 , φ2 }P.B. = − (q12 + q22 ) (29.13)
m
is not ≈ 0, so the constraints are second class.
3) For the Lagrangian
q̇12
L= + αq̇2 q1 − βq1 q22 , (29.14)
2
we have
p1 = q̇1 , p2 = αq1 , (29.15)

so we have the primary constraint

φ1 = p2 − αq1 ≈ 0. (29.16)

The Hamiltonian is
p21
H = p1 q̇1 + p2 q̇2 − L = + βq1 q22 . (29.17)
2
The (first) secondary constraint is

{φ1 , H}P.B. = −2βq1 q2 − αp1 = φ2 ≈ 0. (29.18)

The total Hamiltonian is

HT = H + uφ1 = H + u(p2 − αq1 ). (29.19)

The equation for the particular solution is

{φ2 , H}P.B. + U {φ2 , p2 − αq1 }P.B.

= {−2βq1 q2 − αp1 , H}P.B. + U {−2βq1 q2 − αp1 , p2 − αq1 }P.B. ≈ 0 ,(29.20)

and gives
βq2 (2p1 + αq2 )
βq2 (2p1 + αq2 ) + U (−α2 − 2βq1 ) = 0 ⇒ U = . (29.21)
α2 + 2βq1
There is no solution to the homogenous equation, since there is a single constraint
φ1 , and the equation would be

V {φ2 , φ1 }P.B. = V {−2βq1 q2 − αp1 , p2 − αq1 }P.B. = V (−2βq1 − α2 ) = 0. (29.22)

Note that there are no further secondary constraints.

103 Chapter 29

4) For Dirac quantization,

{f, g}D.B. = {f, g}P.B. − {f, χs }P.B. css0 {χs0 , g}P.B. , (29.23)

where
−1
css0 = ({χs , χs0 }P.B. ) , (29.24)

and χs are the independent second class constraints.

But the independent second class constraints are φ1 , φ2 , with

c̃12 = {φ1 , φ2 }P.B. = {p2 − αq1 , −αp1 − 2βq1 q2 }P.B. = α2 + 2βq1 ≡ A. (29.25)

0 A 0 −1/A
The matrix is , with inverse matrix , so
−A 0 1/A 0
1
c12 = − . (29.26)
A
Then
1 1
{f, g}D.B. = {f, g}P.B. + {f, φ1 }P.B. {φ2 , g}P.B. − {f, φ2 }P.B. {χ1 , g}P.B. .
A A
(29.27)
5) For
s 
2
F µν
Z
Fµν 1 µνρσ
LB.I. = −L−4 d4 x  1 + L4 − L8 Fµν Fρσ − 1 , (29.28)
2 8

the momenta are

4 0 0 0 0
δL F µ0 + L64 µ ν ρ σ Fµ0 ν 0 Fρ0 σ0 40µνρ Fρσ
Pµ = =q 2 . (29.29)
δAµ,0 F F µν
1 + L4 µν2 − L8 18 µνρσ Fµν Fρσ

Then the primary constraint is, as for the Maxwell case,

P0 = 0 , (29.30)

and the fundamental Poisson brackets are

{Aµ (~x, t), P ν (~x0 , t)}P.B. = δµν δ 3 (~x − ~x0 ). (29.31)

The Hamiltonian is
Z
H = d3 xPµ Aµ,0 − L

Z  1
= d3 x q ∂0 Ai F i0

4 Fµν F µν
8 1 µνρσ
2
1+L − L 8 Fµν Fρσ

2
L4 µ0 ν 0 ρ0 σ0
+ Fµ0 ν 0 Fρ0 σ0 40µνρ ∂0 Aµ Fνρ + L−4
2 · 64
L4 µνρσ

1 µν 2
+ Fµν F − ( Fµν Fρσ ) − 1
2 64
104 Chapter 29


Z
 1 1 0i 0i 1
= d x 3
q F F + Fij F ij + L−4
4 F µν F µν
8 1 µνρσ
2 2 2
1+L 2 − L 8 Fµν Fρσ

L4

2
− (µνρσ Fµν Fρσ ) − L−4 + P i ∂i A0
2 · 64 
−4 1 ij
L + 2 Fij F
Z 1 
−4
= d3 x P i F i0 − A0 ∂i P i + q − L ,
2 F F µν 2
1 + L4 µν2 − L8 81 µνρσ Fµν Fρσ

(29.32)

where we should think of F i0 as

s 2
Fµν F µν

1 µνρσ
F i0 = P i 1 + L4 − L8 Fµν Fρσ
2 8
L4 µ0 ν 0 ρ0 σ0
− Fµ0 ν 0 Fρ0 σ0 0ijk Fjk . (29.33)
16
But then we notice that Fµ0 is related to P µ only (even though we cannot invert
the relation perfectly): there are no extra A0 terms. The only term with A0 in the
Hamiltonian is −A0 ∂i P i , so again (like in the Maxwell case) we get the secondary
constraint
{P 0 , H}P.B. = P i ,i ≈ 0. (29.34)

And again there are no secondary constraints, since H = H(P i , Fij , A0 ), so we

get
{P i ,i , H}P.B. = 0. (29.35)

Also as before, we get a trivial algebra of constraints

{P 0 , P 0 }P.B. = 0 = {P 0 , ∂ i Pi }P.B. = {∂ i Pi , ∂ j Pj }P.B. = 0 , (29.36)

so, since all the constraints are first class, the Dirac brackets are equal to the Poisson
brackets.
Then the primary constraints are φm : φq = P 0 , and the secondary constraints
are φj : φ2 = P i ,i . Then as in the Maxwell case,
Z
H 0 = H + d3 xU P 0 (~x)
Z
HT = H + d3 xv(~x)P 0 (~x0 )
0
Z Z
= d x{...} + d3 x[−A0 P i ,i + vP 0 ] ,
3
(29.37)

and finally the extended Hamiltonian is

Z Z Z
HE = HT + d xu(~x)P ,i (~x) = d x{...} + d3 xu0 (~x)P i ,i (~x) ,
4 i 3
(29.38)

where we have defined u0 = u − A0 .

105 Chapter 29

6) For the one-dimensional fermion action

Z
S = dt(+iψ ∗ ψ̇) , (29.39)

we have the momenta

∂L ∂L
p= = −iψ ∗ , p∗ = =0, (29.40)
∂ ψ̇ ∂ ψ̇ ∗
so there are two primary constraints,
φ1 = p + iψ ∗ ≈ 0 , φ2 = p∗ ≈ 0. (29.41)
The naive Hamiltonian is
HL = q̇i pi − L = ψ̇p + ψ̇ ∗ p∗ − iψ ∗ ψ̇ = 0 , (29.42)
so
{φ1 , H}P.B. = {φ2 , H}P.B. = 0 , (29.43)
so there are no secondary constraints. The total Hamiltonian is
HT = uφ1 + vφ2 = u(p + iψ ∗ ) + vp∗ . (29.44)
The condition
{φ1 , HT }P.B. = iv ≈ 0 (29.45)
gives v = 0, and the condition
{φ2 , HT }P.B. = −iu ≈ 0 (29.46)
gives u = 0.
7) So we have HT = 0. On the other hand,
{φ1 , φ2 }P.B. = i 6= 0 , (29.47)
so both primary constraints are second class. That means that
HE = HT = 0 , (29.48)
since there are no first class constraints. Since the matrix of second class constraints
is

0 1
c̃ij = i , (29.49)
−1 0
it follows that

−1 0 1
c = c̃ =i =c, (29.50)
−1 0
so the Dirac brackets are
{f, g}D.B. = {f, g}P.B. − {f, φi }P.B. cij {φj , g}P.B.
= {f, g}P.B. − {f, φ1 }P.B. i{φ2 , g}P.B. + {f, φ2 }P.B. i{φ1 , g}P.B. .
(29.51)
106 Chapter 29

Then we obtain
{ψ, p}P.B. = 1 ⇒ {ψ, p}D.B. = 1 , {ψ ∗ , p∗ }P.B. = 1 ⇒ {ψ ∗ , p∗ }D.B. = 0.
(29.52)
It means that using Dirac brackets, we can put p∗ = 0 and p = −iψ ∗ . But still
HT = 0.
30 Chapter 30

1) The state is
1
|χa1,2 i = √ |ψ1,2 i + a|φ1,2 i
2
1
= √ √ [| ↑A ↑B i ± | ↓A ↓B i + a| ↑A ↓B i ± a| ↓A ↓B i] . (30.1)
1 + a2 2
But the most general separable state is
c (| ↑A i + α| ↓A i) ⊗ (| ↑B i + β| ↓B i)
= c [| ↑A ↑B i + αβ| ↓A ↓B i + α| ↓A ↑B i + β| ↑A ↓B i] . (30.2)
In order for this to equal the previous, we must have
{ α = ±β = a , αβ = ±1 } ⇒ ±β 2 = ±1. (30.3)
That means that β 2 = 1, so β = ±1, so a = ±1. For a = +1, we have
1 1
√ (|ψ1 i + |φ1 i) , √ (|ψ2 i + |φ2 i) , (30.4)
2 2
whereas for a = −1, we have
1 1
√ (|ψ1 i − |φ1 i) , √ (|ψ2 i − |φ2 i) . (30.5)
2 2
Equivalently,
1
ρA = TrB [| ↑A ↑B i ±1 | ↓A ↓B i + a| ↑A ↓B i ±2 | ↓A ↑B i]
2(1 + a2 )
⊗ [h↑A ↑B | ±1 h↓A ↓B | + ah↑A ↓B | ±2 h↓A ↑B |]
1
(1 + a2 )| ↑A ih↑A | + (1 + a2 )| ↓A ih↓A i

= 2
2(1 + a )
+a(±1 ±2 )| ↓A ih↑A | + a(±1 ±2 )| ↑A ih↓A |] . (30.6)
That means that we need a(±1 ±2 ) = ±(1 + a2 ), which is indeed what we had
found.
2) We have
1 1
|χa1,2 i = | ↑iA ⊗ p (| ↑i + a| ↓i)B +| ↓iA ⊗ p (±1 | ↓i ±2 a| ↑i)B ,
2(1 + a2 ) 2(1 + a2 )
(30.7)
and we can define the states after the tensor products as |ĩiB . If ±1 ±2 = 0, then
we find
1
ρA = (| ↑iA | ↑iB + | ↓iA | ↓iA ) , (30.8)
2
107
108 Chapter 30

√
so pi = 1/2, meaning that |i0 iB = 2|ĩi. Then finally
1
|10 i = √ (| ↑i + a| ↓i)B
1 + a2
1
|20 i = √ ± (| ↓i − a| ↑i)B ,
1 + a2
1
|χa1,2 i = √ (| ↑iA ⊗ |10 iB + | ↓iA ⊗ |20 iB ) . (30.9)
2
3) Consider the unitary matrix UAB defined in eq. (30.26) in the text. Then

| ↑↑i |ψ1 i | ↑↓i |φ1 i
UAB = , UAB = ⇒
| ↓↓i |ψ2 i | ↓↑i |φ2 i
a
1 | ↑↑i + a| ↑↓i |χ1,2 i
UAB √ = , (30.10)
1 + a2 | ↓↓i + a| ↓↑i |χa1,2 i
−1
so UAB disentangles them.
4) We have

ρ = C (|1ih2| + |2ih1| + a|2ih2| + |2ih3| + |3ih2|) . (30.11)

From Tr ρ̂ = 1, we obtain C = 1/a.

The von Neumann entropy is S = − Tr ρ ln ρ, so in order to calculate it, we must
diagonalize ρ. As a matrix, we have
 
0 1 0
1
ρ= 1 a 1 . (30.12)
a
0 1 0

Then det(ρ − λ 1l) = 0 gives

p
1± 1 + 4/a2
λ3 = 0 , or λ1,2 = , (30.13)
2
so we obtain the diagonal (we can check that Tr ρ = λ1 + λ2 = 1)

ρ̂ = λ1 |10 ih10 | + λ2 |20 ih20 | , (30.14)

so
S = − Tr ρ̂ ln ρ̂ = −λ1 ln λ1 − λ2 ln λ2 . (30.15)

5) We have

ρ̂tot = |χa1,2 ihχa1,2 | ⇒ ρ̂A = TrB ρ̂tot

1 a(±1 ±2 )
= | ↑iA h↑ |A + | ↓iA h↓ |A + (| ↓iA h↑ |A + | ↑iA h↓ |A ) (30.16)
.
2 1 + a2
The von Neumann entropy of ρA is SA = − Tr ρA ln ρA . In the case ±1 ±2 = 0,
we find
X
SA = − pi ln pi = ln 2. (30.17)
i
109 Chapter 30

6) For the state

|ψi = C (| ↑↑↑i + | ↓↑↓i + | ↑↓↑i) , (30.18)
we have, like for any pure state,
ρABC = |ψihψ| ⇒ SABC = Tr ρABC ln ρABC = 0. (30.19)
√
Moreover, Tr ρ = 1 gives C = 1/ 3.
Then
ρAB = C h↑ |ρABC | ↑iC + C h↓ |ρABC | ↓iC
1
= [(| ↑↑i + | ↑↓i) (h↑↑ | + h↑↓ |) + | ↓↑ih↓↑ |] , (30.20)
3
so that
2 2 1 1 2
SAB = − Tr ρAB ln ρAB = − ln − ln = − ln 2 + ln 3. (30.21)
3 3 3 3 3
Moreover,
ρB = A h↑ |ρAB | ↑iA + A h↓ |ρAB | ↓iA
1
= [(| ↑i + | ↓i) (h↑ | + h↓ |) + | ↑ih↑ |] ,
3
ρBC = A h↑ |ρAB | ↑iA + A h↓ |ρABC | ↓iA = ρAB . (30.22)
Then
2
SBC = SAB = − ln 2 + ln 3 , (30.23)
3

2 1
and for SB , we need to diagonalize ρB = 13 , more precisely find the eigen-
1 1
values. We obtain
r
1 1 1
det(ρ̂ − λ 1l) = 0 ⇒ 9λ2 − 9λ + 1 = 0 ⇒ λ1,2 = ± − . (30.24)
2 4 9
Then
SB = −λ1 ln λ1 − λ2 ln λ2 , (30.25)
since we also have SABC = 0, we can check numerically the resulting strong sub-
additivity relation SABC + SB ≤ SAB + SBC , i.e.,

2 1 1 1
2 ln + ln 3 ≥ 0 + λ1 ln + λ2 ln , (30.26)
3 2 λ1 λ2
and it satisfied.
7) The initial density matrix is
1
ρAB = |φiAB AB φ| = (|+z i|−z i − |−z i|+z i) (h+z |h−z | − h−z |h+z |) . (30.27)
2
After the measurement of A on z, there is 1/2 probability for the state to be
|+z iA |−z iB , and 1/2 probability for the state to be |−z iA |+z iB . But since
1
|±z i = √ (|+x i ± |−x i) , (30.28)
2
110 Chapter 30

that means 1/2-1/2 probability for one of two states,

|+x iA + |−x iA |+x iB − |−x iB |+x iA − |−x iA |+x iB + |−x iB
√ √ , √ √ . (30.29)
2 2 2 2
Then after the measurement of B on x, we have probabilities:
|+x iA + |−x iA |+x iA + |−x iA
1/4 : √ = |+z iA , 1/4 : − √ = −|+z iA ,
2 2
|+x iA − |−x iA |+x iA − |−x iA
1/4 : √ = |−z iA , 1/4 : √ = |−z iA , (30.30)
2 2
so after the final measurement of A on z, we have probabilities 1/2-1/2 for |+z i or
|−z i, as expected.
31 Chapter 31

1) The question is: can we have, even at large angles,

|CQM (~a, ~b) − CQM (~a, ~c)| > 1 + CQM (~b, ~c) , (31.1)
given that CQM (~a, ~b) = − cos θ(~a, ~b)? In other words, can we have
| cos θ(~a, ~b) − cos θ(~a, ~c)| > 1 − cos θ(~b, ~c)? (31.2)
The answer is YES. We can consider the same planar set-up, and take θ(~a, ~b) '
π/4, θ(~b, ~c) ' π/2 and θ(~b, ~c) ' 4π/4. Then, indeed,
√ √
2 2 √
| + | = 2 > 1 − 0 = 1. (31.3)
2 2
2) For example, trivially changing + to −:
N5 + N6
P (~a−, ~b−) = P5 + P6 = P
k Nk
N5 + N7
P (~a−, ~c−) = P5 + P7 = P
k Nk
N 2 + N6
P (~c−, ~b−) = P2 + P6 = P , (31.4)
k Nk
and, since
N5 + N6 ≤ (N2 + N6 ) + (N5 + N7 ) , (31.5)
we obtain
P (~a−, ~b−) ≤ P (~c−, ~b−) + P (~a−, ~c−). (31.6)
In quantum mechanics, the condition to be contradicted is as for the + case,
!
1 θ(~
a , ~b) 1 θ(~
a , ~
c ) θ(~
c , ~b)
sin2 ≤ sin2 + sin2 . (31.7)
2 2 2 2 2
Or, a bit more nontrivially, we have
N7 + N8
P (~a−, ~b+) = P7 + P8 = P
k Nk
N6 + N8
P (~a−, ~c+) = P6 + P8 = P
k Nk
N 2 + N6 )
P (~b+, ~c+) = P2 + P6 = P , (31.8)
k Nk
and, since
N6 + N8 ≤ (N2 + N6 ) + (N8 + N7 ) , (31.9)
111
112 Chapter 31

we obtain
P (~a−, ~c+) ≤ P (~b+, ~c+) + P (~a−, ~b+) , (31.10)
which in quantum mechanics is
" #
1 ~ c) a, ~b)
2 θ(~
a, ~c) 1 2 θ(b, ~ 2 θ(~
cos ≤ sin + cos . (31.11)
2 2 2 2 2

At small angles, this becomes

θ2 (~b, ~c) + θ2 (~a, ~c) ≥ θ2 (~a, ~b) , (31.12)
which is the same as before.
There are, of course, many other examples.
3) For a more generic (not just the maximal one) violation of
|C(a, b) + C(a0 , b) + C(a, b0 ) − C(a0 , b0 )| ≤ 2 , (31.13)
with CQM (~a, ~b) = − cos θ(~a, ~b), we consider small angles, obtaining that we need to
violate
1h 2 ~ i
2− θ (~a, b) + θ2 (~a0 , ~b) + θ2 (~a, ~b0 ) − θ2 (~a0 , ~b0 ) ≤ 2, (31.14)
2
so we need to have
θ2 (~a0 , ~b0 ) > θ2 (~a, ~b) + θ2 (~a, ~b0 ) + θ2 (~a, ~b0 ). (31.15)
But with the 4 vectors ~a, ~b, ~a, ~b0 being coplanar and arranged in the previous
order, we have
θ(~a0 , ~b0 ) = θ(~a0 , ~b) + θ(~b, ~a) + θ(~a, ~b0 ) , (31.16)
which satisfies the needed condition.
4) From
M̃ = (A + A0 )B 0 + (A − A0 )B = AB 0 + A0 B 0 + AB − A0 B (31.17)
instead of M , as before, we obtain
Z Z
| dλρ(λ)M̃ (λ)| ≤ h|M̃ |i = 2 ρ(λ) = 2 , (31.18)
λ λ
or
|hABi + hAB 0 i + hA0 B 0 i − hA0 Bi| ≤ 2 ⇒
|Chid (a, b) + Chid (a, b0 ) + Chid (a0 , b0 ) − Chid (a0 , b)| ≤ 2 , (31.19)
which is really identical (up to a redefinition b ↔ b0 ) with the previous inequality
(Bell-CHSH).
5) Consider that A, A0 , A00 = ±1. Then
A + A0 = 0 , A − A0 = ±2 OR A + A0 = ±2 , A − A0 = 0
A + A00 = 0 , A − A00 = ±2 OR A + A00 = ±2 , A − A00 = 0
A0 + A00 = 0 , A0 − A00 = ±2 OR A0 + A00 = ±2 , A0 − A00 = 0 , (31.20)
113 Chapter 31

so for instance
|[(A + A0 ) + (A + A00 ) + (A0 − A00 )]B + [(A − A0 ) + (A − A00 ) − (A0 − A00 )]B 0 | ≤ 6 ⇒
|2hABi + 2hA0 Bi + 2hAB 0 i − 2hA0 B 0 i − 2hA00 Bi| ≤ 6 ⇒
|C(a, b) + C(a0 , b) + C(a, b0 ) − C(a0 , b0 ) − C(a00 , b)| ≤ 3. (31.21)
To violate this in quantum mechanics, consider that a, b, a0 , b0 are at small angles
from each other and a00 is at π from ~b. Then the violation is as before.
6) We can consider as an ”experiment” the transition from quantum to classical
at a later point. That ”measures” the wave function of the Universe. We obtain
probabilities and compare to the ”experiment” (our classical Universe).
7) We have very small deviations, at each moment in time. Then the probabil-
ity that the (genetically the) same person is born, let alone that it has the same
characteristics (personality, etc.), is ludicrously small.
32 Chapter 32

1) Start with ρ̂ = |ψihψ|, with Tr ρ̂ = 1. Then

1
ρ̂A = TrB ρ̂ = [|1ih1| + 2|2ih2| + |3ih3|]
4
1
ρ̂B = TrA ρ̂ = [|1ih1| + |4ih4| + |2ih2| + |3ih3| + |2ih3| + |3ih2|] , (32.1)
4
where we can check that Tr ρ̂A = Tr ρ̂B = 1.
2) The Liouville-von Neumann equation is
∂ ρ̂
i~ = [Ĥ, ρ̂] , (32.2)
∂t
and, using Ĥ|ii = Ei |ii, we find
∂ ρ̂
i~ = (E2 − E3 )|2ih3| + (E3 − E4 )|3ih4| + (E4 − E5 )|4ih5|. (32.3)
∂t
Then, for small times, we have
it
ρ̂ ' ρ̂0 − [(E2 − E3 )|2ih3| + (E3 − E4 )|3ih4| + (E4 − E5 )|4ih5|] . (32.4)
~
3) For a large number N os spins 1/2, we have Ω = 2N , so

S = kB ln Ω = N kB ln 2. (32.5)

We need to find a quantum mechanical ρ̂ such that

S = −kB Tr[ρ̂ ln ρ̂] = N kB ln 2. (32.6)

Consider then the normalized density matrix

1 X
ρ̂ = N |a1 i ⊗ |a2 i ⊗ ... ⊗ |aN iha1 | ⊗ ... ⊗ haN |. (32.7)
2 a ,...,a =1,2
1 N

Then indeed Tr ρ̂ = 1 and

X
S = −kB Tr ρ̂ ln ρ̂ = −kB pa1 ...aN ln pa1 ...aN
a1 ,...,aN
1 1
= −kB (2N ) N
ln N = N kB ln 2. (32.8)
2 2
4) For N harmonic oscillators,

En = Ei = ~ωi (ni + 1/2) , (32.9)

114
115 Chapter 32

and if they are connected to a reservoir of temperature, in the canonical ensemble,

we can use the Maxwell-Boltzmann distribution,
X X
Z= e−βEn = gn e−βEn
{ni } n
N
! N
Y X Y
−βEi
= e = Zi ,
i=1 ni i=1
∞ ~ωi
X
−β~ωi (ni +1/2) e−β 2
Zi = e = . (32.10)
ni =0
1 − e−β~ωi

Then the free energy is (using β = 1/(kB T ))

N N
" ~ωi
#
X 1X e−β 2
F (T, V, N ) = −kB T ln Z = − kB T ln Zi = − ln . (32.11)
i=1
β i=1 1 − e−β~ωi

The heat capacity is

∂S ∂S
CV = T = −β . (32.12)
∂T V ∂β V

But, from F = U − T S, we have

N N
∂F β ∂F X X β~ωi
ln 1 − e−β~ω + kB

S=− =+ = −kB , (32.13)
∂T V T ∂β V i=1 i=1
eβ~ωi − 1

so the heat capacity becomes

N
X (β~ωi )2 eβ~ωi
CV = kB . (32.14)
i=1
(eβ~ωi − 1)2

5) For a grand-canonical ensemble, with a reservoir of both T and µ, we obtain

X P N
Y
Z= e−β i (Eni −µi Ni ) = Zi
{ni } i=1
∞ ~ωi
X
−β(Ei −µni ) e−β 2
Zi = e = . (32.15)
ni =0
1 − e−β(~ωi −µ)

Then the thermodynamic potential Ω is

N
" ~ωi
#
1X e−β 2
Ω(T, V, µ) = −kB T ln Z = − ln , (32.16)
β i=1 1 − e−β(~ωi −µ)

the entropy becomes

N N
∂Ω β ∂Ω X h i X β(~ωi − µ)
S=− = = −kB ln 1 − e−β(~ωi −µ) + kB ,
∂T V,µ T ∂β V,µ i=1 i=1
eβ~ωi − 1
(32.17)
116 Chapter 32

so the heat capacity is

N
∂S X [β(~ωi − µ)]2 eβ(~ωi −µ)
CV = −β = kB 2 . (32.18)
∂β V,µ i=1 eβ(~ωi −µ) − 1
6) For relativistic fermions we use the FD distributions, so
Z
1
hni ≡ n = d3 pfF D (~
p)
(2π)3 Z
1
hi ≡ ρ = d3 pE(~p)fF D (~
p). (32.19)
(2π)3
Using d3 p = 4πp2 dp and p2 = E 2 − m2 , we obtain
Z ∞ √
1 E 2 − m2 E
n(β, m) = 2
dE β(E−µ)
2π m √e +1
Z ∞
1 E 2 − m2 E 2
ρ(β, m) = dE β(E−µ) . (32.20)
2π 2 m e +1
At T → ∞, or β → 0 (for ultrarelativistic particles), T m, µ, so we can put
m = 0, so
Z ∞ Z ∞
1 E2 1 x2 (kB T )3 3
n(β) = dE = dx = ζ(3)
2π 2 Z0 eβE + 1 2π 2 β 3 Z0 ex + 1 2π 2 2
∞ ∞
1 E3 1 x3 (kB T )4 7 π 4
ρ(β) = 2
dE βE = 2 4
dx x = . (32.21)
2π 0 e +1 2π β 0 e +1 2π 2 8 15
7) The total entropy at temperature T for this system is

e−β Ĥtot 1 −βE1

ρ̂tot = = e |1iA ⊗ |1iB h1A | ⊗ h1B | + e−βE2 |2iA ⊗ |2iB h2A | ⊗ h2B |
Z Z
+e−βE3 |3iA ⊗ |4iB h3A | ⊗ h4B | + e−βE4 |4iA ⊗ |3iB h4A | ⊗ h3B | .

(32.22)
Then the density matrix of subsystem A, tracing over subsystem B, is
1 −βE1
ρ̂A = TrB ρ̂tot = e |1iAA h1| + e−βE2 |2iAA h2|
Z
+e−βE3 |3iAA h3| + e−βE4 |4iAA h4| ,

(32.23)
and by imposing the normalization Tr ρ̂A = 1, we get
4
X
Z= e−βEi , (32.24)
i=1

and so the enanglement entropy at temperature T is

X
SA = −kB TrA (ρ̂A ln ρ̂A ) = −kB pi ln pi
i∈A
4 4
1 X Ei e−βEi X e−βEi
= + kB ln Z. (32.25)
T i=1 Z i=1
Z
33 Chapter 33

1) We need to prove that the number of distinct strings of length n is

n! X
∼Q ∼ 2nH(x) , H(x) ' − 4p(x) log2 p(x). (33.1)
x (np(x))! x=1

To see that, note that at large n, there are ' np(x) letters ax , out of n. Then
n!
the combinatorics of groups of npx ax ’s gives Q (np x )!
, since (np(x))! is the number
x
of permutations of these objects. So the combinatorics is the number of ways of
picking these letters = total number of permutations, divided by the permutations
of each object. √
√ Then, the Stirling approximation gives N ! ' 2πN N −1/2 e−N , and we neglect
2πe−N . Then we note that 2nH(x) = x p(x)−np(x) , and by substituting the
Q

Stirling approximation in the combinatorics, we reproduce this result.

2) The mutual information is 0, I(x, y) = 0, ⇔ H(X) + H(Y ) − H(X, Y ) = 0, or
h− log2 p(x)i + h− log2 p(y)i = h− log2 p(x, y)i , (33.2)
which means that there is no interaction between x and y, so
p(x, y) = p(x) · p(y) , (33.3)
so p(x|y) = p(x,y)
p(y) is independent on y.
3) To prove the concavity of the von Neumann entropy: if p1 , ..., pn ≥ 0 and
P
i pi = 1, then
!
X X
S pi ρi ≥ pi S(ρi ) , (33.4)
i i

where S(ρi ) = − Tr(ρi log ρi ), consider first only ρ1 and ρ2 . Then, for t ∈ [0, 1],
ρ(t) = tρ1 + (1 − t)ρ2 , S(ρ(t)) = − Tr(ρ(t) log ρ(t)) (33.5)
implies

dS ρ̇(t)
= − Tr[ρ̇(t) log ρ(t)] − Tr ρ(t) . (33.6)
dt ρ(t)
But Tr ρ(t) = 1, so Tr[ρ̇(t)] = 0, so we are left with the first term, in which we
note that ρ̇(t) = ρ1 − ρ2 is independent on t. In order to take a second derivative
of t, that will then act only on ρ(t), we must re-express it as
Z ∞ Z ∞
1 1 d 1 1
log ρ(t) = ds − ⇒ log ρ(t) = ds ρ̇(t) .
0 s + 1 s + ρ(t) dt 0 s + ρ(t) s + ρ(t)
(33.7)
117
118 Chapter 33

Then finally we obtain

∞
d2 S
Z
1 1
=− ds Tr ρ̇(t) ρ̇(t) . (33.8)
dt2 0 s + ρ(t) s + ρ(t)
1
, we have Tr[B 2 ] ≥ 0, we obtain that
R
But since by defining B = ρ̇(t) s+ρ(t)

d2 S
≤0, (33.9)
dt2
so the von Neumann entropy is concave.
4) For an example of S(A + B) < S(A) + S(B), consider the case of A + B in
the pure state |ψi that is entangled, e.g.
1
|ψi = √ (|1i|1i + |1i|2i + |2i|1i) , (33.10)
3
so that ρAB = |ψihψ|, so S(AB) = 0, but SA , SB > 0. Indeed,
1
ρB = TrA ρ = [2|1ih1| + |2ih2| + |1ih2| + |2ih1|] = ρA = TrB ρ. (33.11)
3
Then
2 2 1 1
SA = SB = ln + ln > 0. (33.12)
3 3 3 3
For an example of S(A + B) = S(A) + S(B), consider noninteracting states, e.g.,
product states. E.g.,

2 1 2 1
ρ = ρA · ρB = |1ih1| + |2ih2| A ⊗ |1ih1| + |2ih2| . (33.13)
3 3 3 3 B

In this case, we have SAB = SA + SB .

and the right-hand side is

1) Example 1: consider the universal set of gates (NOT, AND, OR, INPUT). Then
to go from (1, 0, 1, 0, 1, 0) to (0, 0, 1, 1, 1, 1) we can consider the two realizations of
the computation:
-(N OT1 , N OT4 , N OT6 ); or
-(1 AN D2 , IN P U T2 , IN P U T3 , 3 OR4 , IN P U T5 , 5OR6 ).
Example 2: Consider the universal set of gates (NOT, AND, OR, XOR). Then
consider the realization of the computation:
-N OT1 ⊗ N OT4 ⊗ 5OR6 .
Therefore, the complexity of the computation is less or equal to 3.
2) For the 5-qubit space, going from

|0i + |1i |0i + |1i |0i + |1i |0i + |1i |0i + |1i
|ψR i = √ , √ , √ , √ , √ (34.1)
2 2 2 2 2
to
|0i − |1i |0i + |1i |0i − |1i |0i + |1i |0i − |1i
|ψT i = √ , √ , √ , √ , √ (34.2)
2 2 2 2 2
can be done without need of an error. Note that |ψT i is not entangled.
We use the complete set of Hadamard gate T , Phase gate P and Toffoli gate T .
Note that
2
2 1 1 1 1 2 1 0
H = √ = 1l , X = 1l , P = ⇒ P 4 = 1l. (34.3)
2 1 −1 0 −1
Then
|0i + |1i |0i − |1i
P2 √ = √ , (34.4)
2 2
so one example of a product of gates giving the computation is
P 2 (1) ⊗ P 2 (3) ⊗ P 2 (5) , (34.5)
so the complexity of the computation is less or equal to 6.
3) A basis OI for a computation acting on a 4-qubit space in the Nielsen approach
is
OI = Oi1 i2 i3 i4 = σi1 ⊗ σi2 ⊗ σi3 ⊗ σi4 , (34.6)
such that the Hamiltonian is
X 3
X
H(s) = Y I (s)OI = Y i1 i2 i3 i4 Oi1 i2 i3 i4 , (34.7)
I i1 ,i2 ,i3 ,i4 =1

120
121 Chapter 34

and the Schrödinger equation is

dU (s)
i = Y i1 i2 i3 i4 (s)σi1 ⊗ σi2 ⊗ σi3 ⊗ σi4 U (s). (34.8)
ds
4) For the function F (Y I ) = I |Y I | to be a good cost function (Finsler func-
P

tion), we need to satisfy 4 conditions:

a) F ≥ 0 and F = 0 ⇔ Y I = 0 is true.
b) F is continuous (F ∈ C 0 ), but not smooth (F ∈ C 1 ).
c) F is homogenous: F (λY I ) = λF (Y I ).
d) F satisfies the triangle inequality:
X X
F (Y I + Y 0I ) = |Y I + Y 0I | ≤ (|Y I | + |Y 0I |) = F (Y I ) + F (Y 0I ). (34.9)
I I

So it would be a good function if it were smooth, but it isn’t, so the Riemannian

geometry defined by it is singular at zero, meaning it is not good.
5) Consider the Hamiltonian H = −J i ~σi · ~σi+1 , and W = σk1 , for k fixed.
P

Then, to show that W (t) spreads away from k, we must calculate it. Since
∞
X (it)n
W (t) = eiHt W e−iHt = [H, ..., [H, W ]...] , (34.10)
n=0
n!

using [σia , σjb ] = δij i abc σjc , we obtain

P
c
X
[H, W ] = −Ji a1c σkc (σk+1
a a
+ σk−1 )
a,c
= +iJ σk2 σk+1
3
− σk3 σk+1
2
+ σk2 σk−1
3
− σk3 σk−1
2

, (34.11)

so [H, W ] contains σk−1 , σk , σk+1 . It is then easy to see that [H, [H, W ]] contains
σk−2 , σk−1 , σk , σk+1 , σk+2 , [H, [H, [H, W ]]] contains σk−3 , ..., σk+3 , etc. So indeed,
as t increases, and higher commutator terms become relevant, we have σi further
away from σk becoming relevant, so W (t) spreads.
6) For quantum chaos,

hW † (t)V † W (t)V i = F (t) = 1 − eλL t , (34.12)

where λL is the Lyapunov exponent.

Here W (t) is a small perturbation that starts depending on time (see the previous
exercise). The correspondence with classical mechanics is that classical trajectories
are mapped to quantum wave functions, or more precisely to quantum states.
So, classical chaos is the fact that two initially nearby classical trajectories diverge
with time, and should be mapped to the fact that the corresponding quantum states
(with a small perturbation added at an initial time) divergence from each other with
time.
The two states are

|ψ̃i ≡ W (t)V |ψi ⇒ hψ̃| = hψ|V † W † (t)

|ψ 0 i ≡ V W (t)|ψi ⇒ hψ 0 | = hψ|W † (t)V † , (34.13)
122 Chapter 34

and then F (t) = hψ 0 |ψ̃i is the overlap of the two states, different from hψ̃|ψ̃i = 1 by
a small amount, since initially (at time t → 0) [V, W (t)] → 0 (is very small, leading
to |ψ 0 i ' |ψ̃i), yet it grows with time.
7) From (34.39) in the text, λL ≤ 2πkB T /~, so we infer that λL → 0 as T → 0.
But this doesn’t imply that t∗ → ∞, even though λL ∼ 1/t∗ . The reason is that
(34.37), F (t) ' 1 − eeλL t , is only valid at T 6= 0, yet t∗ is defined from C(t) =
2(1 − ReF (t)) ∼ 1 even at T = 0.
35 Chapter 35

1) There is a large quantum state (”cat”), |ψlarge iA and an atom in a quantum

|deadiA ⊗ |decayiB + |aliveiA ⊗ |no decayiB

√ . (35.1)
2
To simplify it, by eliminating observers, notice that: there is also an interaction
with the environment (C) (since A interacts with C), which means that these states
become unaccessible, due to decoherence. Since |ψlarge i is needed for quantum com-
puting, the quick decoherence is bad. So we need quantum error correcting codes.
2) At the end of decoherence, from the point of view of S, there is no pure
quantum state. We have
X
ρS = |hψ|ii|2 |iihi| , (35.2)
i

so the probability of the state |φi is

X
hφ|ρS |φi = |hψ|iihi|φi|2 , (35.3)
i

with no interference terms.

For instance, if |ii = |deadi and |ii = |alivei, then

hi|ρs |ii = |hψ|ii|2 , hj|ρs |ji = |hψ|ji|2 , (35.4)

(hi| + hj|)ρs |(|ii + |ji) = hi|ρs |ii + hj|ρs |ji , (35.5)

which is the classical sum of probabilities.

3) If hi |j i =
6 δij ,
At step 1), |ψi⊗|i, with |..|2 = hψ|ψih|i = hψ|ψi, evolves into i |i ihi|ψi, with
P

|...|2 = i,j hi |j hψ|jihi|ψi =

P
6 hψ|ψi in general, so we have a non-unitary evolution.
At step 2), |ψi ⊗ |i evolves into
X
|ii ⊗ |i ihi|ψi , (35.6)
i

123
124 Chapter 35

X
= hi |i ihψ|iihi|ψi =
6 hψ|ψi , (35.7)
i

so in general we again have non-unitary evolution. Then, after decoherence, we have

which is not diagonal, not with hψ|iihi|ψi, and is still quantum.

4) An example of integrable quantum mechanical model is the Heisenberg spin
1/2 chain. It has ”magnon” excitations that scatter via S-matrices reducible to
2-point S-matrices.
Thermalization means correlations between excitations, but for this (and other)
integrable system(s), due to the reduction to 2-point scattering, the excitations
move around without changing shape and without correlating.
Another example is the sine-Gordon model, with soliton excitations that maintain
their shape through interactions.
h =P∞ (infinite dimensional
5) For D i Hilbert space), we still have
D=∞
ρ̂d = exp − m=1 λm P̂m and λm = − ln(|Cm |2 ) and, as in the text, still ρ̂d =
PD=∞ 2
m=1 (|Cm | |mihm|). So the system is still a GGE: for integrable systems, we have
P̂m = integrals of motion, and since there are an infinite number of them, there are
an infinite number of integrals of motion, so the system is integrable.
6) In order to diagonalize
L h i2
â†i âi + λ a†i + ai − a†i+1 + ai+1
X
Ĥ = , (35.9)
i=1

we first expand the creation and annihilation operators in a (discrete) Fourier series,
L−1
b 2πinj
√n e L ,
X
aj = (35.10)
n=0
L

which preserves the commutation relations: If [ak , a†j ] = δkj , the same is obtained
from [bn , b†m ] = δmn :
L−1 2πi(nk−mj) L−1 2πin(k−j)
e L e L
[ak , a†j ] =
X X
[bn , b†m ] = = δkj . (35.11)
n,m=0
L m=0
L

Moreover, the kinetic term in the Hamiltonian is invariant,

L−1 L−1 2πij(m−n) L−1
e L
a†j aj
X X X X
= b†n bm = b†n bn . (35.12)
j n,m=0 j=0
L n=0

But the other term is still not in a nice form, since

L−1
1 h 2πinj 2πin 2πinj
2πin i
(aj +a†j )−(aj+1 −a†j+1 ) =
X
√ e L e L − 1 bn + e − L e− L − 1 b†n .
n=0
L
(35.13)
125 Chapter 35

We must still redefine

cn,1 + cn,2 cn,1 − cn,2
bn = √ , bL−n = √ , (35.14)
2 2
which still preserves the commutation relations, since now again we find

[cn,1 , c†m,1 ] = δnm = [cn,2 , c†m,2 ]. (35.15)

The transformation also leaves the kinetic term invariant, since

b†n bn + b†L−n bL−n = c†n,1 cn,1 + c†n,2 cn,2 . (35.16)

We now substitute the latest transformation into

L/2
1 h 2πinj 2πin 2πinj
2πin
(aj + a†j ) − (aj+1 − a†j+1 ) =
X
√ e L e L − 1 bn + e − L e− L − 1 b†n
n=0
L
2πin 2πin i
− 2πinj 2πinj
+e L e− L − 1 bL−n + e L e L − 1 b†L−n ,
(35.17)

and (after some algebra) finally obtain

L−1
X βn †
Ĥ = cn,1 cn,1 + cn,1 c†n,1 + c†n,2 cn,2 + cn,2 c†n,2
n=0
2
2 2
† †
+βn cn,1 + cn,1 − cn,2 − cn,2

2πn
βn = 1 − 2λ 1 − cos
L
2πn
αn = 2λ 1 − cos . (35.18)
L

But, for a general Hamiltonian of the form

aa† + a† a (a ± a† )2 α aa† + a† a

α
a2 + (a† )2 ,

H=β ±α =β 1+ ±
2 2 β 2 2β
(35.19)
(so in our case α = αn , β = βn and ± refers to cn,1 , cn,2 ) we can make a Bogoliubov
transformation,
b = α̃a + β̃a† ⇒ b† = α̃∗ a† + b̃∗ a , (35.20)

which preserves the commutation relations, so [b, b† ] = 1 ⇔ |α̃|2 − |β̃|2 = 1.

We want to make this transformation in order to get rid of the b2 and (b† )2 terms
in the Hamiltonian, and remain just with b† b, i.e., to diagonalize it. The condition
for these terms to vanish is

α α
1+ α̃∗ b̃∗ = ± ((α̃∗ )2 + (β̃ ∗ )2 ). (35.21)
β β
126 Chapter 35

√ √
For α, β real (like in our case), we define α̃ − β̃ ≡ 1/ ω, α̃ + β̃ ≡ ω, leading to
s s √ √ √ √
1 − α/β 1 + 3α/β ω + 1/ ω ω − 1/ ω
ω1 = , ω2 = ⇒ α̃ = , β̃ = .
1 + 3α/β 1 − α/β 2 2
(35.22)
Then finally we obtain the diagonal Hamiltonian
bb† + b† b (1 + α/β)2 − 2α2 /β 2 bb† + b† b
H= βp ≡ ωn . (35.23)
2 (1 − α/β)(1 + 3α/β) 2
We note that the frequency is the same for both ± signs. We then apply to our
case (for α = αn , β = βn , and ± referring to cn,1 , cn,2 , so b becomes now c̃n,1 , c̃n,2 ),
obtaining the diagonal Hamiltonian
c̃†n,1 c̃n,1 + c̃n,1 c̃†n,1 c̃†n,2 c̃n,2 + c̃n,2 c̃†n,2
L−1
!
X
H= ωn + . (35.24)
n=0
2 2

7) Since thermal particle creation arises from a vacuum state, we can deduce that
energy is not (globally) conserved, so the space is either curved, or topologically
nontrivial. It cannot happen in Minkowski space, which conserves energy.
36 Chapter 36

1) Consider the perturbation Ĥ1 = λx to the harmonic oscillator potential, H0 =

2
k x2 . Then the exact solution is given by
2
λ2

k λ
Ĥ = x+ − , (36.1)
2 k 2k
which is also a harmonic oscillator of constant k, so the energies En are only shifted,
λ2
En0 = En − , (36.2)
2k
λ λ
whereas the wave functions are shifted by x = k = mω 2 , so
λ (0)
hx|ni = hx + |n i. (36.3)
k
In first order perturbation theory,
Z +∞
En(1) = hn (0)
|H1 |n (0)
i=λ dx|ψn (x)|2 x = 0 , (36.4)
−∞

so the energy is unchanged, basically because the Hermite polynomials obey Hn (−x) =
(−1)n Hn (x), so |ψn (−x)|2 = |ψn (x)|2 , meaning the integrand is odd, and the inte-
gral vanishes. This is in agreement with the exact result.
On the other hand, the wave function changes: we have
X H1,mn
|n(1) i = (0) (0)
|m(0) i. (36.5)
E
m6=n n − Em

The wave function is

r 1/4
− mωx
2 mω mω
ψn (x) = An e 2~ Hn x , An = , (36.6)
~ π~(2n)(n!)2
and the Hermite polynomials Hn (y) obey
n n
y2 d y2 2 d 2
Hn (y) = e 2 y − e− 2 = (−1)n ey e−y ⇒
dy dy
1 d
ψn+1 = p y− ψn . (36.7)
2(n + 1) dy
Moreover,
Z +∞
∗
H1,mn = λ dx xψn (x)ψm (x)
−∞

127
128 Chapter 36

Z +∞ r r
mωx2 mω mω
= λAn Am dx xe− ~ Hn x Hm x , (36.8)
−∞ ~ ~

and by the same argument, since Hn (−x) = (−1)n Hn (x), if n + m is even, the
integrand is odd, so the integral vanishes. That leaves the case of n + m = 2k + 1.
(0)
Since we also have En = ~ω(n + 1/2), we have
X H1,n+2k+1,n
|n(1) i = |n(0) + 2k + 1i. (36.9)
−~ω(2k + 1)
k

The first term in the sum is

Z +∞
H1,n+1,n = λ dx xψn (x)ψn+1 (x)
−∞
Z +∞ n " n+1 #
λ~ 2 −y 2 y2 d −y 2 y2 d −y 2
=− dy yAn e e e e e ,
mω −∞ dy dy
(36.10)

so
Z +∞ n " n+1 #
H1,n+1,n λ 2 2 d 2 2 d 2
=− A2 dy ye−y ey e−y ey e−y .
~ω mω 2 n −∞ dy dy
(36.11)
Then
r
(1) λ mω
hx|n i= (nrs.)hx|n + 1(0) i
mω ~
d
y − dy
r
λ mω
= (nrs.) p ψn(0) (x). (36.12)
mω ~ 2(n + 1)

We see that the approximation is not so good.

2) For the Hydrogenoid atom, perturbed by λĤ1 = λe−µr , the eigenenergy per-
turbation is
Z ∞ Z
(1) −µr 2 2 −µr 2
En = hnlm|λe |nlmi = dr r Rnl (r)λe dΩYlm (θ, φ)
Z ∞ 0

=λ dr r2 e−µr Rnl
2
(r)
0 Z
∞ 2
2
dr r2 e−2κn r−µr (2κn r)2l L2l+1

= λÑnl n−l−1 (2κn r) . (36.13)
0

But, since
Z ∞
dxe−x(s+1) xµ+β Lµk (x)Lµk (x)
0 
A2

1+µ+β
Γ(1 + µ + β)Γ(1 + µ + k) dk  1 F1 2 , 1 + µ+β
2 , 1 + µ; B 2
=  (36.14)
,
(k!)2 Γ(1 + µ) dhk (1 − h)1+µ B 1+µ+β
h=0
129 Chapter 36

1+h
where A2 = 4h/(1 − h)2 , B = s + 1−h , we finally obtain

A2

λ(l + 1) dn−l−1 1 F 1 l + 3/2, l + 2, 2l + 2; B 2
En(1) =   , (36.15)
n dhn−l−1 (1 − h)2l+2 B 2l+3
h=0
µ
where s = 2κn and

A2 4h 1 + (1 − s)h
2
= 2
, B= . (36.16)
B [1 + s + (1 − s)h] 1−h
For the ground state wave function, we have
(1)
X H1,n0 0
|ψ0 i = (0)
|n0 i ,
(0) 0
(36.17)
0 E
n 6=0 0 − E n 0

(0)
with En0 = E0 /n02 . Due to spherical symmetry, dΩY00 Ylm = δl0 δ0m , so
R
Z
H1,n0 0 = r2 drR10 (r)Rn0 l (r)λe−µr δl0 δm0 , (36.18)

meaning that we have

∞ Z ∞
(1) λ X n02
|ψ0 i = r2 drR10 (r)Rn0 0 (r)e−µr |n0 00i. (36.19)
E0 0 n02 − 1 0
n =2
1 −2r/a0
But since R10 (r) = 3/2 e , and 2/a0 = κ1 , the integral becomes
a0

∞
(κ1 /2)3/2
Z
2 −z 1+ µ+κ+1
z dze 2κ 0
n L1n−1 (z) , (36.20)
(κn0 )3/2 0

and since
Z ∞
β −st α Γ(β + 1)Γ(α + n + 1) −β−1 1
dt t e Ln (t) = s 1 F1 −n, β + 1, α + 1; ,
0 n!Γ(α + 1) 2
(36.21)
we obtain
∞ 3
r
(n0 − 1)! n02

(1) λ 3/2
X
0 2κn0
|ψ0 i = (κ1 /2) 2n ×
E0 0 =2
n0 !2n0 n02 − 1 2κn0 + µ + κ1
n
0 2κn0
×1 F1 −n + 1, 3, 2; |n0 00i. (36.22)
2κn0 + µ + κ1
3) To write the second order perturbation theory for E0 , we have

(2)
X |λ(e−µr )0m |2
E0 = (0) (0)
, (36.23)
m6=0 E0 − Em
(0) (0)
where m = (nlm0 ) and Em = E0 /n2 , so

(2)
X |(λe−µr )0;n,l,m0 |2
E0 = (0)
. (36.24)
n,l,m E0 (1 − 1/n2 )
130 Chapter 36

Moreover,
Z ∞
(λe−µr )0;n,l,m = dr r2 λe−µr R10
∗
(r)Rnl (r)δ0l δ0m
0
3/2 Z ∞
1
= r2 dre−κ1 r−µr−κn r L1n−1 (2κn r)
a0 0
3 r
3/2 2κn (n − 1)!
= λ(κ1 /2) 2n ×
2κ n + µ + κ 1 n!2n
2κn
×1 F1 −n + 1; 3; 2; . (36.25)
2κn + µ + κ1
4) For the first excited state, n = 2, we have a degenerate state, with (nlm)
states (200) = |1i, (210) = |2i, (21 + 1) = |3i, (21 − 1) = |4i. Then
Z ∞
H1,αβ = λ(e−µr )αβ = λÑ2l 2
r2 drR2l
2
(r)e−µr δll0 δmm0 = λhe−µr i2l,2l δll0 δmm0 ,
0
(36.26)
so is diagonal in |αi = |1i, |2i, |3i, |4i. We have
(
(1) λhe−µr i20,20 , |αi = |1i
E2 = (36.27)
λhe−µr i21,21 , |αi = rest.

and we have done he−µr i20,20 at exercise 2.

5) We have
1
Ĝ(z) = , (36.28)
z − Ĥ0 − λĤ1
for a = z − Ĥ0 and b = λĤ1 , so we get, by expanding in λ,
1 X 1 n
Ĝ(z) = = λn (z − Ĥ0 )−1 Ĥ1
(z − Ĥ0 )(1 − λ(z − Ĥ0 )−1 Ĥ1 ) n≥0 z − Ĥ0
X
= λn Ĝ0 (Ĝ0 Ĥ1 )n . (36.29)
n≥0

q.e.d.
6) To prove
r
l2 − m2
hlm| cos θ|l − 1 mi = hl − 1 m| cos θ|lmi = , (36.30)
4l2 − 1
we note that it is equal to
Z π Z 2π
∗
= dθ sin θ dφYlm (θ, φ)Yl−1,m (θ, φ) cos θ
Z0 π 0

= dθ sin θPlm (cos θ)Pl−1,m (cos θ)

0 r
Z 1
l 2 − m2
= dw wPlm (w)Pl−1,m (w) = . (36.31)
−1 4l2 − 1
q.e.d.
131 Chapter 36

7) For the Stark effect for n = 3 in the Hydrogenoid atom, we first define the
degenerate basis,
|3; 1i = |3, 0, 0i , |3; 2i = |3, 1, 0i , |3; 3i = |3, 1, 1i
|3; 4i = |3, 1, −1i , |3; 5i = |3, 2, 0i , |3; 6i = |3, 2, 1i
|3; 7i = |3, 2, −1i , |3; 8i = |3, 2, 2i , |3; 9i = |3, 2, −2i. (36.32)
Then we still have
hnl1 m1 |Ĥ1 |nl2 m2 i = dEhl1 m1 | cos θ|l2 m2 ihrinl1 ,nl2 , (36.33)
and still
r
l2 − m2
hlm| cos θ|l − 1, mi = hl − 1, m| cos θ|lmi = , (36.34)
4l2 − 1
and the rest are zero. For hri3l,3l−1 , we define A = hri31,30 and B = hri32,31 , and
in our case we have
r
1
h1| cos θ|2i = h00| cos θ|20i =
3
2
h2| cos θ|5i = h10| cos θ|20i = √
15
1
h3| cos θ|6i = h11| cos θ|21i = √ = h1 − 1| cos θ|2 − 1i = h4| cos θ|7i.(36.35)
15
Then
A
 
0 √
3
0 0 0 0 0 0 0
 √A 0 0 0 2B
√ 0 0 0 0
 3 15 
√B
 
0 0 0 0 0 0 0 0
 15 
0 0 0 0 0 0 √B 0 0
 15 
Ĥ1 = dE ×  0 2B
0 0 0 0 0 0 0 . (36.36)
 
√
αβ  15 
0 0 √B 0 0 0 0 0 0
 15 
0 0 0 √B 0 0 0 0 0
 
 15 
0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0
The eigenvalues and eigenstates are found through diagonalization. There are 2
zero eigenvalues, and for the other 7 × 7 matrix, we write the secular equation,
det(H1 − λ 1l) = 0 , (36.37)
which can be solved with Mathematica.
37 Chapter 37

1) For the Hydrogen atom in the ground state, perturbed with λH1,mi = λe−mA+λ̃t ,
for t < 0, the probability to go to the m’th eigenstate is
Pf i (t) = |b(1) 2
n (t)| , (37.1)
where
Z τ
i~b(1)
m (τ ) = dteiωmi t H1,mi (t). (37.2)
0
2
For the Hydrogen atom, ωmi = Em~−E0 = −|E0 | 1/m~ −1 , so
Z τ
−mA+t(λ̃+iωmi ) e−mA (λ̃+iωmi )τ
i~b(1)
m (τ ) = dte = e − 1 ⇒
0 λ̃ + iωmi
e−2mA
2

Pf i (t) = |b(1) 2
m (t)| = 2 2 2 )
e 2λ̃τ
(cos ωτ − 1) 2
+ e 2λ̃τ
sin ωτ (. 37.3)
~ (λ̃ + ωmi
2) For the Hydrogen atom in the ground state, perturbed suddenly (at t = 0)
with Ĥ1 = Kr, to find the transition probability to the n = 2, l = 0 state, we first
find
3
H1,mi = Khri10,20 , Em − Ei = E0 (1/22 − 1) = |E0 | , (37.4)
4
and then substitute in the general formula
4|H1,mi |2 ωmi τ
Pmi (τ ) = sin2 , (37.5)
(Em − Ei )2 2
which at large τ gives
2π
Pmi (τ ) → τ |H1,mi |2 δ(Ef − Ei ) = 0 , (37.6)
~
since Ef 6= Ei .
3) To prove the formula
1 sin2 ax
lim = δ(x) , (37.7)
a→∞ π ax2
2 2
we note that for ax → 0, sin ax sin ax
(ax)2 → 1, so then ax2 → a → ∞. However, for x 6= 0
2 2
(finite) and a → ∞, both sin ax and x are finite, so the limit gives 0. Then, indeed
the limit is proportional to δ(x), it only remains to fix the proportionality constant.
But
Z + Z +∞
sin2 ax sin2 y
dx = dy =π, (37.8)
− ax2 −∞ y2
132
133 Chapter 37

so the proportionality constant is 1/π.

4) For a radial square well, and the electron in the ground state, so within the
square well, and a Hmi = H=constant, we have Fermi’s golden rule,
dPmi (t) 2π 2π
= |H1,mi |2 δ(Ef − Ei ) = |H|2 δ(Ef − Ei ) , (37.9)
dt ~ ~
and since the extra H takes the square well ground state above zero, it can transition
to the free space beyond (so Ef = Ei is possible).
5) The differential cross section is
dσ 2πm0
= p − p~0 |2 ,
|V (~ (37.10)
dΩ ~
where Z
1 i V0
V (~
p) = d3~re− ~ p~·~r V (~r) = , (37.11)
(2π~)3/2 (2π~)3/2
since V (~r) = V0 δ 3 (~r). Then
dσ 2πm0 V02
= = constant. (37.12)
dΩ ~ (2π~)3
6) Writing the Coulomb potential as the limit of a Yukawa potential,
zZ|e0 |2 −λr
V (r) = lim e , (37.13)
λ→0 r
we can make the Fourier transform of the Yukawa potential, writing d3~r = r2 drdφ(−d(cos θ))
and p~ · ~r = pr cos θ, we have
−λr Z ∞ Z +1
e−λr
Z
1 3 − ~i p re
~·~ 2π i

3/2
d ~
r e = 3/2
r 2
dr d cos θe− ~ pr cos θ
(2π~) r (2π~) Z 0 −1 r
∞
1 dr h −r(λ+ ip ip
~ ) − e−r (λ− ~ )
i
= √ e
~ 2π~ 0 − ip ~
2 1
= √ p , (37.14)
~ 2π~ |λ + i ~ |2
so that
2zZ 1
p) = lim √
V (~ ~2
p
. (37.15)
2
λ→0 ~ 2π~ λ +
~2

7) For the adiabatic introduction of the perturbation, with

Ĥ1
Ĥ1 (t) = , t < 0. (37.16)
1 + γ 2 t2
h i
1 1
Writing 1+γ 2 t2 = (1+iγt)(1−iγt) = 21 1−iγt
1 1
+ 1+iγt , we have
τ τ
eiωmi t eiωmi t eoωmi t
Z Z
H1,mi
i~b(1)
m (τ ) = dt 2 2
H1,mi = dt + . (37.17)
−∞ 1+γ t 2 −∞ 1 − iγt 1 + iγt
For τ → ∞, we can add for free a semicircle in the upper-half plane, to close the
134 Chapter 37

contour, since there eiωmi t → 0. For the counterclockwise loop, we get a minus, so
in all, by the residue theorem, we get for the integral
H1,mi iωmi γi
− e . (37.18)
2
Otherwise, for τ < 0, we leave it as it is, and
H1,mi τ
Z
1 1
b(1)
m (τ ) = −i dte iωmi t
+ , (37.19)
2~ −∞ 1 − iγt 1 + iγt
and the transition rate is
τ 2
|H1,mi |2 d
Z
dPf i d (1) 1 1
= |bm (τ )|2 = dteiωmi t + . (37.20)
dτ dτ 4~2 τ −∞ 1 − iγt 1 + iγt
38 Chapter 38

1) The decay width is

Z
2π X X
Γ= df2 ... |H1,mi |2 ρ(E, f2 , ..., fn ) , (38.1)
~ n
fn

and the energy shift is

X Z Z X |H1,mi |2
∆Ei = − J dEn df2 ... ρ(En , f2 , ..., fn ). (38.2)
n
En − Ei
fn

So if we have Γ finite at nonzero, |H1,mi |2 6= 0, so in principle it is possible to

have ∆Ei = 0, if we have En − Ei with an equal number of + and − terms (since
Ei is in the middle of En ).
If we have ∆E finite and nonzero, we cannot have Γ = 0, since it is positive
definite, and |H1,mi |2 (for some m, since ∆Ei is finite and nonzero) is 6= 0.
2
2) If ρ(En ) = AeαEn , if ∆Ei < ∞, since it is ∼ n dEn |H 1ni | αEn
P R
En −Ei Ee , it follows
2 −αEn
that |H1,ni | decreases exponentially, faster than e .
P R ρ(En )
R If Hni is independent on En = Ei , then from ∆Ei < ∞, n dEn En −Ei < ∞,
3)
so d(log En )ρ(log En ) < ∞, so ρ(En ) decreases fater than 1/ log En .
4) To argue for ∆E · ∆t ∼ ~, based on the Breit-Wigner distribution,
|H1,mi |2
|b(1)+(3)
m |2 = , (38.3)
~Γ 2

(Em − Ei − ∆Ei )2 + 2

we note that the lifetime is hti = Γ1 = ∆t, and the uncertainty in energy is the
width of the Lorentzian distribution, which is ~Γ, so
1
h∆Ei = ~Γ = ~ ⇒ h∆Eih∆ti = ~. (38.4)
h∆ti
5) The sudden approximation is
(
0, t<0
H1 (t) = , (38.5)
H1 , t≥0
and then
i t 0
Z
ÛI (t, t0 ) = T exp − dt H1,I (t0 ) , (38.6)
~ t0
where
(
0, t<0
H1,I (t) = i
− ~i Ĥ0 (t−t0 )
, (38.7)
e ~ Ĥ0 (t−t0 ) Ĥ1 e , t>0
135
136 Chapter 38

and it is natural to put t0 = 0, so nothing needs to be changed, except H1,I is

independent of t.
Then from (38.41) in the text,
X (−i/~)s X Z t Z t n i
bn (t) = δni + dt1 ... dts T e ~ En (t1 −t0 ) H1,S,nm1 (t1 )
s! t0 t 0
s≥1
{m k}i o
− ~i Em1 (t1 −t0 ) i
e ... e+ ~ Ems−1 (ts −t0 ) H1,Sms−1 ms (ts )e− ~ Ei (ts −t0 ) (38.8)
.

Again, it is natural to put t0 = 0, and H1,S is independent of t.

6) From the above, we can obtain the first order time-independent perturbation
theory:

Z t
i X i i
bn (t) = δni − dt1 e ~ En (t1 −t) H1,Snm1 e− ~ Em1 (t1 −t0 )
~ m1t0
Z t1
iX i
= δni − H1S,nm1 e ~ (En −Em1 )(t−t0 )
~ m t0
1
X H1S,nm i
= δni − 1
e ~ (En −Em1 )(t−t0 ) − 1 . (38.9)
m
En − Em1
1

Then
i
X X
|ψi = cn (t)|ψn i = bn (t)e− ~ En t |ψn i , (38.10)
n n

and we obtain
i
X H1S,nm i i

|ψi = e− ~ Ei t |ψi i − 1
e− ~ Em1 t − e− ~ En t |ψn i , (38.11)
m
En − Em1
1

which is just the time-independent result, once we add to it the standard time
dependence.
7) To continue the same to the second order formulae,
Z t Z t
(−i~)2 X
∆b(2)
n (t) = dt1 dt2 H1S,nm1 H1S,m1 m2 ×
2! m ,m t0 t0
1 2
i i
×e ~ (En −Em1 )(t1 −t0 ) e ~ (E
m1 −Em2 )(t2 −t0 )
i
i
X e ~ (En −Em1 )(t−t0 ) − 1 e ~ (Em1 −Em2 )(t− t0 ) − 1
= H1S,nm1 H1S,m1 m2 ,
m ,m
(En − Em1 )(Em1 − Em2 )
1 2
(38.12)
− ~i En t
and, as above, this gets multiplied by e |ψn i. The result matches the time-
independent one,
 
X 1 X H1,mp H1,pn H 1,nn H1,mn
|n(2) i =  −  |mi. (38.13)
m
E n − Em E n − Ep E n − E m
p6=n
39 Chapter 39

1) The relativistic energy of a particle can be expanded as

p 1 (pc)2 1 (pc)4
E= (m0 c2 )2 + (pc)2 ' m0 c2 + 2
− + ... (39.1)
2 m0 c 8 (m0 c2 )3
To couple to electromagnetism, we add the electromagnetic potential, and write
~ so
p~kin = p~can − q A,
1 ~ 4
p − q A)
H = m0 c2 + V (~r) + (~ ~ 2 − 1 (~
p − q A) + ... (39.2)
2m0 8 m30 c2
~ and A
To the usual A ~ 2 terms,

~2 ~ 2 ~ ~ ~ q ~2
− ∇ + iq A·∇+ A , (39.3)
2m0 m0 2m0
we now add
4 q ~ · p~)~ 2q 2 ~ 2 2 ~ 2 ) + ...
3 (A p2 − (A p~ + (~p · A)
8 m0 c 2 8m30 c2
i q~3 ~ ~ ~ 2 q 2 ~2 ~ 2 ~ 2 ~ · ∇)
~ 2 ) + ...
= ( A · ∇)( ∇ ) − (A ∇ + (A (39.4)
2 m30 c2 4m30 c2
2) To find the expressions for the corrections (due to the above terms) to the
linear and quadratic Zeeman effect, to
q ~ ~ ·B
~ ,
− A · p~ → −µB L (39.5)
m0
giving the linear Zeeman effect, we now add
1 q ~ · p~)~ p~2 ~ · B].
~
+ 3 2
(A p2 → − 2 2 × [−µB L (39.6)
2 m0 c 2m0 c
Also, to
2
q2 ~ 2

m eB
+ A →+ hr2 sin2 θin , (39.7)
2m0 2 2m0
giving the quadratic Zeeman effect, we must add
q2 ~ 2 p~2 + (~ ~ 2 ].
− [A p · A) (39.8)
4m30 c2
The correct treatment will be done later.
3) To show that the splitting of lines according to the Zeeman or the Paschen-
Back rules gives the same number of lines, we remember that the |LSJmJ i basis
137
138 Chapter 39

is equivalent to the |LmL SmS i basis via the Clebsch-Gordan coefficients, so there
must be an equal number of states,
L+S
X
(2J + 1) = (2L + 1)(2S + 1) , (39.9)
J=|L−S|

so the number of lines, for transitions between states, is also the same.
4) For the transition element H1,f i for the Hydrogenoid atom hit by an electro-
magnetic wave, in the dipole approximation, between states with n = 1, l = 0 and
n = 2, l = 1, we first note that for electric dipole radiation, we must have

j 0 = j − 1, j, j + 1 , m0 = m ± 1 , (39.10)

which is indeed consistent with l = 0 transitioning to l = 1. Then,

Z
H1,f i = −A0 ωij~ · d3~rρf i~r , (39.11)

where

E2 − E1 |E1 | 1
ωf i = = 1− , (39.12)
~ ~ 4
and
∗
ρf i = hn = 2, l = 1|ρ̂|n = 1, l = 0i = qψn=2,l=1 (~r)ψn=1,l=0 (~r). (39.13)

Also note that ρ~r = ~j is the electric current. Then we need to calculate (for
~ · ~r = z cos θ, i.e., only z component)
Z
∗
q d3 r~ · ~rψn=2,l=1 (~r)ψn=1,l=0 (~r)
Z ∞ Z
∗ 1
=q r2 dr dΩr cos θR21 (r)Y1m (θ, φ)R10 (r) √
0 4π
= qhlm| cos θ|l − 1m − 1ihri21,10 , (39.14)
q
2 2
and hlm| cos θ|l − 1mi = l4l−m 2 −1 .

5) For the Hydrogenoid atom with n = 3, l = 2, m = 0, the possible transitions

by monochromatic wave, are:
-in first order (electric dipole): j 0 = j − 1, j, j + 1, m0 = m ± 1 means we can go
to l = 1, 2, 3, with m = ±1.
-in second order: a) magnetic dipole is also a vector, so the same rules apply as
above. b) for electric quadrupole Qij , j 0 = j − 2, j − 1, j, j + 1, j + 2, so we can go
to l = 0, 1, 2, 3, 4 and m = ±1, ±2.
6) The total absorption cross section for exercise 4 is
Z X
σabs = 4πα~ωif |hψf |~(~k) · ~r|ψi i|2 df2 ... ρ(Ef , f2 , ..., fr )dEf δ(Ef − Ei − ~ω).
fr
(39.15)
139 Chapter 39

But there is no degeneracy, so

Z
2π X
σabs = |H1,f i |2 df2 ... dEf δ(Ef − Ei − ~ω). (39.16)
~
fr

7) The differential cross section is

dσ 4παωf i ~c
= |hψf |~ · p~|ψi i|2 , (39.17)
dΩ (mωf i )2
and for the Hydrogen atom in the ground state, E1 ' −13.6eV , so a monochromatic
wave with E = 14eV lifts it into the continuum, so the final state is a free wave,
p·~
ei~ r
h~r|ψf i = (2π~) 3/2 , and |ψi i is the Hydrogen in the ground state.

Then the matrix element is

ei~p·~r ~ ~
Z
1
M ≡ hψf |~ · p~|ψi i = d3 r ~ · ∇R10 (r) √ . (39.18)
(2π~)3/2 i 4π
But the free wave is rewritten in spherical coordinates as
1 X
3/2
(2l + 1)il jl (kr)Pl (cos θ) , (39.19)
(2π~) l

~ 10 (r) = R0 (r)∇r,
and ∇R ~ and if ~ · p~ has only a z component, then we obtain z =
10 q r
l2 −m2
cos θ. Also using hlm| cos θ|l − 1mi = 4l2 −1 and rest zero, only for l = 1, m = 0
we get a nonzero result, so we have 12 δl1 . Then finally,
Z ∞
3i 0
M= r2 drj1 (kr)R10 (r). (39.20)
2(2π~)3/2 0
40 Chapter 40

1) To estimate h2|e−a∂x |1i for the harmonic oscillator, using WKB and Bohr-
Sommerfeld quantization, we first remember that the wave function from the B-
S+WKB approximation is
mω 2
x2
ψn ' const. × xn e− 2~ (40.1)

at large x, so
Z +∞
mω 2 mω 2
x2 (x−a)2
h2|e−a∂x |1i = h2|1(x−a)i ' dx x2 e− 2~ (x−a)e− 2~ +intermediate x.
−∞
(40.2)
If a is large, then the above approximation becomes better.
2) For V (x) = − x2 A+a2 and |1i, |2i the first two eigenstates in this potential, then

f12 ≡ h1|x|2i ∼ e−ω21 Imτ , (40.3)

where
r
x0
2(E − V (x))
Z
dx ~ω1,2
τ≡ , v(x) = , E1,2 ≡ E ± . (40.4)
v(x) m 2
Here x0 is the pole in the upper-half plane in V (x). But there is a unique such
1
pole at x0 = ia, since V (x) = − (x+ia)(x−ia) . Near this pole,
r
A 1
v(x) ' √ , (40.5)
iam x − ia
so
r x0 =ia √
I
iam
τ (x) ∼ dx x − ia. (40.6)
A
But with the parametrization x −√ ia = reiθ , dx = −ireiθ dθ (in the negative θ
direction), the integral is 3 r , and i = 1+i
4 3/2 √ , so
2
r
ma 4 3/2
Imτ ∼ r . (40.7)
2A 3
3) Instantons are not directly associated with the motion of a real particle (clas-
sical, or quantum). The path integral is a sum over quantum paths, and is given by
the instanton approximation, so there is only an implicit relation.
4) Consider x1,2 the minima of U (x), and the series of instanton paths x1 − x2
140
141 Chapter 40

(the one-instanton), x1 − x2 − x1 − x2 (2-instanton), x1 − x2 − x1 − x2 − x1 − x2

(3-instanton), etc. If the instantons are non-interacting, then
2
P1 ∼ e− ~ ImS1 ; , P2 = P12 , P3 = (P1 )3 , etc. , (40.8)

so
∞
X 1
P1n = − 1. (40.9)
n=1
1 − P1

5) For V = α(x2 − a2 )2 , to calculate

2
P1 ∼ e− ~ Im(S[x1 ,x0 ]+S[x0 ,x2 ]) , (40.10)

we consider the inverse (Euclidean) potential VE = −α(x2 −a2 ) and classical motion
in it, from x1 = −a to x2 = +a through x0 = 0.
The Euclidean action is
Z [x2 ] " 2 #
1 dx 2 2 2
SE = dtE + α(x − a ) , (40.11)
[x1 ] 2 dtE

with equation of motion

d2 x
= +αx(x2 − a2 ) , (40.12)
dt2E
and by multiplication with dx/dt and integration, we obtain the ”virial theorem”,
1 dx 2

2 dt = V (x), which when integrated gives
Z Z
dx 1
Z
dx √ 1 x+a
dt = p = √ 2 − a2
⇒ t 2α = ln , (40.13)
2V (x) 2α x 2a x−a

which inverts to
√
1 + e2 2αat
x(t) = a √ . (40.14)
1 − e2 2αat

Putting it into the on-shell Euclidean action, we obtain

Z x2
SE = dt[2α(x2 − a2 )2 ]
x1
√ √ !2
+∞
e4 2αat
e2 2αat
Z
4
= 4a dt √ 4 1+ , (40.15)
−∞ 2
1 − e2 2αat

and finally the probability is e−SE .

6) For the potential V (x) = A cos2 ax, the Euclidean (inverted) potential is
π
VE = −A cos2 ax, with maxima at x1,2 = ± 2a , and local minimum at x0 = 0. Then
the solution is found from (see previous exercise)
1 1 tan ax 2 +1
Z Z
dx dx
t= p = √ =√ ln ax , (40.16)
2V (x) 2A cos ax 2A a tan 2 − 1
142 Chapter 40

so √
ax 1 + e 2Aat
tan = √ , (40.17)
2 1 − e 2Aat
such that at x → −∞, tan ax π ax
2 = 1, so x = + 2a , whereas for x → +∞, tan 2 = −1,
π
so x = − 2a .
Then the on-shell Euclidean action is (see the previous exercise)
Z +∞ Z +∞ Z +∞ " √ #2
2 2e 2Aat
SE = dt 2V = 2A dt cos ax = 2A dt √ , (40.18)
−∞ −∞ −∞ 1 + e2 2Aat

and the transition probability is e−SE .

7) The formal fluctuation correction to the instanton at the previous exercise is
one over the square root of the determinant of the kinetic operator. But
d2 d2
− 2 + V 00 (xcl (tE )) = − 2 − 4Aa2 cos(2axcl (tE ))
dtE dtE
√
 !2 
2 2Aat
d 2e
= − 2 − 4Aa2 2 √ − 1 , (40.19)
dtE 1 + e2 2Aat

so the formal fluctuation correction is

  √ !2 −1/2
 d2 2e 2Aat 
det − 2 − 4Aa2 2 √ − 1 . (40.20)
 dtE 1 + e2 2Aat 
41 Chapter 41

1 1
1) For the 2-state system with basis |1i, |2i and Hamiltonian H = , the
1 1
first form of the variational method starts with the wave function
X
|ψi = aα |αi = a1 |1i + a2 |2i , (41.1)
α

and the energy of the state

a∗α aβ Hαβ
P
α,β
hEiψ = , (41.2)
a∗α aα
P
α

and is greater or equal to the true ground state energy E0 , so by minimizing,

∂hEiψ /∂aα = 0, we find the approximation. In our case,
(a∗1 + a∗2 )(a1 + a2 ) a1 a∗2 + a2 a∗1
hEiψ = ∗ ∗ =1+ . (41.3)
a1 a1 + a2 a2 a1 a∗1 + a2 a∗2
We choose a1 , a2 ∈ R (since also Hαβ ∈ R), so then
∂Eψ 2a2,1 2a1 a2 2a1,2
= 2 − = 0 ⇒ a21 = a22 , (41.4)
∂a1,2 a1 + a22 (a21 + a22 )
and plugging back into the energy we find Eψ = 2.
To compare with the exact solution, of (Ĥ −E 1l)ψ = 0, for which det(Ĥ −E 1l) =
(1 − E)2 − 1 = 0, we also find E = 0, 2, so the variational method finds the first
excited state, instead of the ground state energy, as we wanted.
2) For the Ritz variational method, for the harmonic oscillator, with
r
2 2 2 mω
ψ1 (x) = e−y /2 , ψ2 (x) = e−y , ψ3 (x) = e2y , y = x , (41.5)
~
we write
|ψi = c1 ψ1 + c2 ψ2 + c3 ψ3 , (41.6)

and we can choose ci ∈ R, since ψi ∈ R. Then, in the general formula

∗ ∗
P P
i,k ck ci hψk |H|ψi i i,k ck ci Hki
hEiψ = P ∗ ≡ P ∗ , (41.7)
i,k ck ci hψk |ψi i i,k ck ci ∆ki

we minimize over coefficients,

∂Eψ
= 0 ⇒ det(Hki − E∆ki ) = 0. (41.8)
∂c∗k
143
144 Chapter 41

To do that, we need to calculate

the Hki and ∆ki coefficients. Using the fact that
~2 d 2 mω 2 x2 d2
Ĥ = − 2m dx2 + 2 = 2 − dy2 + y 2 , we have
~ω

Z +∞
r
Z +∞ r
~ −y 2 π~
∆11 = dxψ12 (x)
= dye =2
mω mω
Z−∞
+∞
r −∞ Z ∞ r r
~ − 3y2
2
2 π~
∆12 = dxψ1 (x)ψ2 (x) = 2 dye =2
mω 3 mω
Z−∞
+∞
r Z0 ∞ r
~ 2 2 π~
∆12 = dxψ2 (x)ψ2 (x) = 2 dye−2y = √
mω 2 mω
Z−∞
+∞
r Z0 ∞ r r
~ − 5y2
2
2 π~
∆13 = dxψ1 (x)ψ3 (x) = 2 dye =2
mω 5 mω
Z−∞
+∞
r Z0 ∞ r
~ 2 2 π~
∆23 = dxψ2 (x)ψ3 (x) = 2 dye−3y = √
−∞ mω 0 3 mω
Z +∞ r Z ∞ r
~ −4y 2 π~
∆33 = dxψ3 (x)ψ3 (x) = 2 dye =
−∞ mω 0 mω r
Z +∞ r
~ ~ω +∞ d2 ~3 ω √
Z
y2 2
− 2 2 − y2
H11 = dxψ1 (x)Ĥψ1 (x) = dye − 2 +y e = π
−∞ mω 2 −∞ dy r m
Z +∞ r r
~ ~ω +∞ d2 ~3 ω 3π
Z
y2 2
H12 = dxψ1 (x)Ĥψ2 (x) = dye− 2 − 2 + y 2 e−y =
−∞ mω 2 −∞ dy r m r 2
Z +∞ r
~ ~ω +∞ 2
~3 ω 8π
Z
2 d y2
H21 = dxψ2 (x)Ĥψ1 (x) = dye−y − 2 + y 2 e− 2 =
−∞ mω 2 −∞ dy r mr 3
Z +∞ r
~ ~ω +∞ 2
~3 ω π 13
Z
2 d 2
H22 = dxψ2 (x)Ĥψ2 (x) = dye−y − 2 + y 2 e−y =
−∞ mω 2 −∞ dy rm r2 8
Z +∞ r
~ ~ω +∞ 2
~3 ω 5π
Z
y2 d 2
H13 = dxψ1 (x)Ĥψ3 (x) = dye− 2 − 2 + y 2 e−2y =
−∞ mω 2 −∞ dy r m r 2
Z +∞ r
~ ~ω +∞ 2
~3 ω π 1
Z
2 d y2
H31 = dxψ3 (x)Ĥψ1 (x) = dye−2y − 2 + y 2 e− 2 =
−∞ mω 2 −∞ dy r m r 10 5
Z +∞ r
~ ~ω +∞ 2
~3 ω π 11
Z
2 d 2
H23 = dxψ2 (x)Ĥψ3 (x) = dye−y − 2 + y 2 e−2y =
−∞ mω 2 −∞ dy r m r3 4
Z +∞ r
~ ~ω +∞ 2
~3 ω π 7
Z
2 d 2
H32 = dxψ3 (x)Ĥψ2 (x) = dye−2y − 2 + y 2 e−y =
−∞ mω 2 −∞ dy rm 34
Z +∞ r Z +∞ 2
~ ω √ 49
3

~ ~ω 2 d 2
H33 = dxψ3 (x)Ĥψ3 (x) = dye−2y − 2 + y 2 e−2y = π .
−∞ mω 2 −∞ dy m 128
(41.9)
ˆ
q we calculate det(Ĥ − E ∆). In the matrix, we take
For the ground state energy,
3
out the common factor of ~ mωπ , obtaining
 q q q q 
2E 3 2E 2 5 2E 2
1− ~ωq 2 − ~ω 3 2 − ~ω 5
 q
det  2 2E 2 13 2E √1 11
√ − 2E √1
− −  = 0. (41.10)
 √

 3 ~ω 3
q 8 2 ~ω 2 4 3 ~ω 3 
2 3/2 1 2E 2 7 2E √1 49 2E 1
5 4 − ~ω 5
√
4 3
− ~ω 3 128 − ~ω 2
145 Chapter 41

2E
Defining a ≡ ~ω ,
the equation becomes, explicitly,
r r !r
13 a 49 a 5 2 2 7 a
(1 − a) √ −√ − + −a (a − 1) √ −√
8 3 2 128 2 2 5 3 4 3 3
r r ! r r r ! r
3 2 11 a 2 1 5 2 13 a 2 1
+ −a √ −√ −a − −a √ −√ −a
2 3 4 3 3 5 10 2 5 8 2 2 3 10
r r !r
3 2 2 49 a 7 a 11 a
− −a (1 − a) − − (1 − a) √ −√ √ −√ = 0. (41.11)
2 3 3 128 2 4 3 3 4 3 3

We note that, among the 6 terms, 4 are canceled by putting a = 1, which is the
exact ground state of the harmonic oscillator (E = ~ω
2 ), and we are left with (for
a = 1)
r " r r ! r r ! #
2 9 5 2 5 3 2 7
− √ − − √ , (41.12)
5 10 2 5 8 2 2 3 4 3

which means that the exact ground state is very close to the lowest solution of the
approximate equation.
2
3) For the practical variational method, with trial wave function ψa (x) = e−ay ,
in order to find the ground state energy, we write

~ω ∞ 2
d2 2
dye−ay − dy 2
e−ay
R +∞ R
2 + y
∗
−∞
dxψ (x)Hψ(x) 2 0
hEiψ = R +∞ = R∞
−∞
dxψ ∗ (x)ψ(x) 0
dye−2ay2
" √ #
~ω 2
= (1 − 4a2 ) + 2a . (41.13)
2 4a

Varying it with respect to the parameter a, we find

√ √
∂hEiψ √ 2 2
0= = 2 − 2 − 2 ⇒ a2 = √ , (41.14)
∂a 4a 4(2 − 2)
and plugging back into the energy, we find the approximate ground state energy,
√
q
~ω 1/4
hEi = 2 2− 2, (41.15)
2
which is close to the true energy, ~ω 2 .
4) For the third state of the harmonic oscillator (second excited), |n = 2i, with
2
trial wave function ψa (x) = y 2 eay , we have
Z +∞ √
2 1 3 π
dxψa (x) =
(2a)5/2 4
Z−∞+∞
~ω +∞ d2
Z
2 2
Hψψ = dxψa (x)Ĥψa (x) = dye−ay y 2 − 2 + y 2 y 2 e−ay
−∞ 2 −∞ dy
√ √ √
~ω 2 π 10a 3 π 1 − 4a2 15 π
= 2 − + + , (41.16)
2 (2a)3/2 4 (2a)5/2 8 (2a)7/2 16
146 Chapter 41

so that

~ω 7a 5
hEiψ = + 2 . (41.17)
2 3 8a
Minimizing over a, we obtain
1/3
∂hEiψ 30
= 0 ⇒ 2a = , (41.18)
∂a 7
and substituting back into the energy, we find
1/3
7 30
hEiψ = ~ω , (41.19)
8 7
which is not too close to the correct value 52 ~ω.
2
5) For the potential V = kx2 + α|x|3 , and the trial wave function ψ(x; a, b) =
2
|y|a e−by , we have the Hamiltonian
r !
~ω d2 2 3 2α ~
H= − 2 + y + |y| , (41.20)
2 dy k mω
q
and defining A ≡ 2α k
~
mω , we find
R +∞
a −by 2 d2 3

~ω −∞
dy|y| e − dy 2 + y 2
+ A|y| 3
|y|a e−by
hEiψ = R +∞
2 dy|y|2a e−2by2
−∞
1 − 4b2

~ω a(a − 1)2b
= − + 2b(2a + 1) + (a + 3/2)(a + 1/2)
2 a − 1/2 2b
A
+ (a + 5/2)(a + 3/2)(a + 1/2) . (41.21)
(2b)3/2
The equations for the minimum of the energy (to find the correct approximation)
are
∂hEiψ ∂hEiψ
= = 0. (41.22)
∂a ∂b
6) For the details for the minimization of the trial wave function R(r) = Ae−r/a
for the Hydrogen atom, we write
~2 ~ 2 Ze20 ~2
2
~2 l(l + 1) Ze20

d 2 d
Ĥ = − ∇ − =− 2
+ + − , (41.23)
2m r 2m dr r dr 2mr2 r
so that
Ze20
h 2 2 i
R∞ −r/a ~2 l(l+1)
0
r 2
dre − ~
2m dr 2
d
+ 2 d
r ∂r + 2mr 2 − r e−r/a
hEiψ = R∞
0
r2 dre−r/a
Z ∞ 2
2~2 2Ze20 1 2~2 l(l + 1) 1

1 ~
= t2 dte−t − 2
+ − +
2 0 2ma ma2 a t ma2 t
2 2
~ 1 Ze0
= + + 2l(l + 1) − . (41.24)
ma2 2 a
147 Chapter 41

Minimizing over a, we find

Ze20 2~2

∂hEiψ 1
=0⇒ = + + l(l + 1) , (41.25)
∂a a ma2 2
and then substituting back into the energy, we find
~2

1
hEiψ = − + l(l + 1) . (41.26)
ma2 2
But, of course, we need spherical symmetry, so l = 0, so finally
~2
hEi0 = − , (41.27)
2ma2
and then
~2 a0
a= 2 = . (41.28)
mZe0 Z
The constant A comes from normalization, as before,
Z ∞ 3/2
2 2 −2r/a 2 1
A r dre =1⇒A= √ . (41.29)
0 a 2
7) For the minimization of the same Hydrogen atom with trial wave function
R(r) = r2 A
+b2 , we first put l = 0 as in the previous exercise, and then we find

Ze20
R∞ h 2 2 i
r2 d 2 d 1
0
dr 2
r +b 2 − ~
2m dr 2 + r dr − r r 2 +b2
hEiψ = R∞ 1
0
r2 dr (r2 +b 2 )2
2 Ze2
− ~ 2 (2I1 − 10I2 + 8I3 ) − 2b0
= 2mb π , (41.30)
2 − I1
R∞ dx
where we have defined In = 0 (x2 +1) n+1 . Then, when minimizing, we get

∂hEiψ 2~2
=0⇒b=− (2I1 − 10I2 + 8I3 ) , (41.31)
∂b mZe20
and when substituting back into the energy, we find the minimum (approximation
for the ground state energy)
Ze20
hEi0 = − . (41.32)
4b π2 − I1
42 Chapter 42

1) For the 4s and 4p atomic orbital wave functions, we find (from

q the general for-
mula Rnl (r) = Ñnl e ρ Ln−l−1 (ρ), with ρ = 2κn r and Ñnl = (n−l−1)!
−ρ/2 l 2l+1
2n(n+l)! (2κn )
3/2
)
r
3!
ψ4s (r, θ, φ) = (2κ4 )3/2 e−ρ/2 L13 (ρ)
8 · 4!  
r cos θ
2!
ψ4p (r, θ, φ) = (2κ4 )3/2 e−ρ/2 L32 (ρ) × sin θ cos φ , (42.1)
8 · 5!
sin θ sin φ

where the Laguerre polynomials in the above are (from

dk z dm+k
Lkm (z) = (−)k dz k e dz m+k e−z z m+k )

L32 (z) = 60(z 2 − 10z + 20) , L13 (z) = −4(z 3 − 12z 2 + 36z − 24). (42.2)

2) Using the shell models and Hund rules for C 6 , O8 , Hg 12 , we find

C : 1s2 , 2s2 , 2p2 : px , py : ↑, ↑

O : 1s2 , 2s2 , 2p4 : 2px , py , pz : ↑↓, ↑, ↑
Hg : 1s2 , 2s2 , 2p6 , 3s2 , (42.3)

where we have shown explicitly what happens only for the non completely filled
orbitals.
3) For the element O (Oxygen), with 1s2 , 2s2 , 2p4 (see above), considering that
s has l = 0 and p has l = 1, and in the 2p4 , two electrons are parallel, and two are
antiparallel, so S ~ + 1/2
~ = 1/2 ~ + ~0 = ~1 (the filled s orbitals don’t contribute) and
~ = ~1 + ~1 + ~1 + ~1 (the filled orbitals don’t contribute), but two of the ~1’s cancel, so
L
~
L = 2, and then J = 1, 2, 3 for the LS coupling.
On the other hand, for the j − j coupling, again the filled s orbitals don’t con-
tribute, and we have ~j1 = ~1 + 1/2 ~ = 3/2,~ ~j2 = ~1 − 1/2
~ = 1/2,~ ~j3 = 1/2
~ + ~0 = 1/2,
~
~ + ~0, and then ~j1 + ~j2 = ~1 or ~2, and ~j3 + ~j4 = ~1, and when finally summing
~j4 = 1/2
the two to get J, ~ again we have J = 1, 2, 3 (note that J~ = ~0 is excluded, since there
is total spin).
4) The method of atomic orbitals for O, in the basis of only filled orbitals gives
3
X
|ψa i = Ck |φk i , (42.4)
k=1

where
|φ1 i = |1s2 i , |φ2 i = |2s2 i , |φ3 i = |2px i. (42.5)
148
149 Chapter 42

Then the minimum energy is found, as usual, from the secular equation
3
X
Ck (Hik − Eδik ) = 0 ⇒ det(Hik − Eδik ) = 0. (42.6)
k=1

5) For O2 , where each Oxygen has the orbital structure (1s2 , 2s2 , 2p4 ), the
method of molecular orbitals starts with molecular orbitals (the notation means
the electron is in the first orbital with respect to the first nucleus, and the second
orbital with respect to the second)

φ1 = φ1s2 (1),1s2 (2)

φ2 = φ1s2 (1),2s2 (2)
φ3 = φ2s2 (1),2s2 (2)
φ4 = φ1s2 (1),2p2x (2)
φ5 = φ2s2 (1),2p2x (2)
φ6 = φ2p2x (1),2p2x (2)
φ7 = φ1s2 (1),2p2y (2)
φ8 = φ1s2 (1),2p2z (2)
φ9 = φ2s2 (1),2p2y (2)
φ10 = φ2s2 (1),2p2z (2)
φ11 = φ2p2y (1),2p2y (2)
φ12 = φ2p2z (1),2p2z (2) , (42.7)

with each having two equivalent electrons (product wave functions).

Then, the molecular orbitals are (with 16 electrons in pairs of 2)

φ113366 11 12 = (φ1 )2 (φ3 )2 (φ6 )2 φ11 φ12 , φ122366 11 12 = φ1 (φ2 )2 φ3 (φ6 )2 φ11 φ12 , etc.
(42.8)
6) The LCAO method for O2 applied to each atomic orbital in O is:
The molecular orbital is

~ 1 ) + Ci2 χ2 (~ri , R
φai i (~ri ) = Cir χr (~ri , R ~ 2) , (42.9)

for instance

χ1 = 1s2 , χ2 = 1s2 ⇒ φ1i

χ1 = 1s2 , χ2 = 2s2 ⇒ φ2i
χ1 = 2s2 , χ2 = 2s2 ⇒ φ3i
χ1 = 1s2 , χ2 = 2p2x ⇒ φ4i
χ1 = 2s2 , χ2 = 2p2x ⇒ φ5i
χ1 = 2p2x , χ2 = 2p2x ⇒ φ6i
χ1 = 1s2 , χ2 = 2py,z ⇒ φ7i
χ1 = 2s2 , χ2 = 2py,z ⇒ φ8i
χ1 = 2p2x , χ2 = 2py,z ⇒ φ9i
χ1 = 2py,z , χ2 = 2py,z ⇒ φ10i . (42.10)
150 Chapter 42

7) For Bohr quantization for the common electron in benzene, we write

I
p dq = nh , (42.11)

applied to motion on a ring of radius a, so

~
mv · 2πa = nh ⇒ v = n . (42.12)
ma
Substituting in the kinetic energy, we find the energy levels,
mvn2 ~2 2
En = = n . (42.13)
2 2ma2
43 Chapter 43

1) For a nucleus spinning around an axis, it will deform to be an ellipsoid, but in the
liquid droplet model we would still have the volume proportional to A, V = V0 A (V0
being the volume of a single nucleon). This will replace the particular R = r0 A1/3 ,
valid only for a sphere.
2) The rotating nucleus become an ellipsoid. To find the excentricity, we start
from the observation that the difference in pressure between the exterior and the
interior of the curved surface of the ellipsoid is

1 1
∆p(out − in) = σ + , (43.1)
R1 R2
where σ is the surface tension and R1,2 are the two local radii of curvature of the
two-dimensional surface. With respect to the axis of rotation, we define a North
Pole A, where the distance to the center O is smallest, and an equator, with a point
B on it, where the distance to the center O is largest.
Then at the North Pole, the two local radii of curvature are equal, R1 = R2 =
R + ∆R, whereas at the Equator, the radius in the Equator direction is still R1 =
R + ∆R, but the radius in the meridian direction (towards the North Pole) is
R2 = R − ∆R. That means that there is an extra pressure at the Equator, and that
is due to the effect of the rotation, pushing matter outwards. But the pressure at
the center O is the same, whether we integrate it from A to O, or from B to O.
Then, when comparing the two ways to integrate, we should get the same result
at the center. That means that the integral of the centrifugal pressure from B to
O should equal the extra pressure ∆p. We have for the centrifugal pressure
d2 F
d2 F = d2 mω 2 r = ρ · 2πrdrdy · ω 2 r ⇒ dp = = ρω 2 rdr. (43.2)
2πrdy
Integrating, we have
R2
∆Pextra = ρω 2 . (43.3)
2
But then,
R2

2σ 1 1
∆p(A − O) = = ∆p(B − O) = σ + − ρω 2
R + ∆R R + ∆R R − ∆R 2
2
2σR 2R
= 2 − ρω . (43.4)
R − ∆R2 2
That gives
∆R R2 ∆R ρR3 ω 2 3ω 2
2σ 2
' ρω 2 ⇒ = = M. (43.5)
R 2 R 4σ 16πσ
151
152 Chapter 43

3) Can we have a shell model with a V (r, z)? The Hamiltonian is

~2 ∂ 2 ∂2

1 ∂
Ĥ = − + + + V (ρ, z) , (43.6)
2µ ∂ρ2 ρ ∂ρ ∂z 2
and we want a separated solution,
ψmz ,m,nρ = Znz (z)Φm (φ)Rnρ ,|m| (ρ) , (43.7)
where
1
Φm (ρ) = √ eimφ , Lz Φm (φ) = m~Φm (φ). (43.8)
2π
Indeed, if V (ρ, z) = V (ρ) + V (z), then
2 2
~ ∂
Hz Znz = − + V (z) Znz = Ez Znz . (43.9)
2µ ∂z 2
So yes, we can have a shell model, at least for V (ρ, z) = V (ρ) + V (z), in which
case we have separation of variables, so we get quantum numbers, corresponding to
the shells.
Then, in the particular case of the harmonic oscillator with different constants
in the polar ρ and z directions,
ρ2 z2
V (ρ) = µω12 , V (z) = µω22 , (43.10)
2 2
we obtain
α2
2z
2

Znz = N (α2 )e− 2 Hnz (α2 z)

Ez = ~ω2 (nz + 1/2) , (43.11)
and then the radial equation becomes
d2 R 1 dR m2 µω12 2

2µ
+ + − 2 + 2 [E − ~ω2 (nz + 1/2)] − 2 ρ R = 0. (43.12)
dρ2 ρ dρ ρ ~ ~
√
Writing, as usual, R = χ/ ρ, in order to get rid of the term with a single
derivative, we obtain
d2 χ m2 − 1/4 2µ µω12 2

+ − + [E − ~ω (n
2 z + 1/2)] − ρ R=0, (43.13)
dρ2 ρ2 ~2 ~2
so, with l = −1/2 + |m|, we obtain the quantization condition
E ω2 3
− (nz + 1/2) = l + + 2nρ = 1 ± |m| + 2nρ , (43.14)
~ω1 ω1 2
so the energy becomes

ω2
E = ~ω1 (nz + 1/2) + 1 ± |m| + 2nρ . (43.15)
ω1
If ω2 /ω1 = k ∈ N, then
E = ~ω1 [knz + 2nρ + 1 + k ± |m|]. (43.16)
153 Chapter 43

Then the degeneracy is

n
X l+1 (n + 1)(n + 2)
dn = = , (43.17)
k k
l=0

and the magic numbers for the shells are

n
X (n + 1)(n + 2)(n + 3)
dn = . (43.18)
m=0
3k

4) For a spherical square well with V = 0 for r < R, the solution for r < R is
r
χnl (r) 2m(E + V0 )
= Nn,l jl (kr) , k = , (43.19)
r ~2
and the solution for r ≥ R is
r
χnl (r) 2mE
= Ñn,l jl (k̃r) , k̃ = . (43.20)
r ~2
Then the matching equations at r = R are

Nn,l kl (kR) = Ñn,l kl (k̃R)

Nn,l kjl0 (kR) = Ñn,l k̃jl0 (k̃R) , (43.21)

so the quantization condition is

kjl0 (kR) k̃j 0 (k̃R)

= l . (43.22)
jl (kR) jl (k̃R)
5) For
r2

VN (r) = −V0 1 − 2 er/R2 , (43.23)
R
with R2 R1 , but not by a lot, for r → 0 (r R2 ),

r2

r
VN (r) → −V0 1 − 2 1− , (43.24)
R1 R2
whereas for r → ∞ (r R2 ), we have VN (r) → 0, so we have a better approxima-
tion to the real case. That means that at large n, we are closer to the reality, with
respect to the harmonic oscillator approximation considered in the text. Instead of
being linear, En = ~ω(n + 3/2) ∼ n, they will accumulate, En − En+1 → 0.
6) We have
1 dVN (r)
Hint,spin−orbit = −const. (2~s · ~l) , (43.25)
r dr
and
mω12

V0 1 dVN (r) 2V0 r
= , = 2 e−r/R2 1− , (43.26)
R12 2 r dr R1 2R2
154 Chapter 43

while, as before,
(
l, j = l + 1/2
2~s · ~l = . (43.27)
−l − 1 , j = l − 1/2
Then,

r l
Hint,spin−orbit = −(const.0 ) × e−r/R2 1− × , (43.28)
2R2 −l − 1
so this is still < 0 for j = l + 1/2 and > 0 for j = l − 1/2, but now only for
r < R2 , that is, for the first numbers n. Otherwise (for larger numbers n), the two
are switched, and much smaller.
7) For the pairing energy for a 2-particle delta function potential, for 2 nucleons
in the ground state 1s1/2 , l = 0, s = 1/2, so j = 1/2, and from equations (49.47)
and (43.48) in the text, we obtain
Z
2g dr
∆E = h1/2, 1/2, 0, 0|V̂ |1/2, 1/2, 0, 0i = |φ10 (r)|2 |φ10 (r)|2 . (43.29)
8π r2
But
√ 1/2 √
α 2α 1 1/2 2 2 4α 2α − α2 r2
φ10 (r) = 1 √ 31 √ L0 (α r ) = √ e 2 , (43.30)
2 π 22 π π 3
p
where α = mω/~, so
gα7 √ √
∆E = 5
256 2(− π). (43.31)
9π
44 Chapter 44

1) To show that (44.11) in the text, i.e.,

2V0 2V0 2V0 ω0 − ω
N (t) = N (0) − P (0) sin Ωt + N (0) − (cos Ωt − 1) , (44.1)
~Ω ~Ω ~Ω Ω

solves the Bloch equations, specifically the first one, we calculate

dN (t) 2V0 2V0 2V0 ω0 − ω
=− P (0) cos Ωt − sin Ωt N (0) − Q(0) , (44.2)
dt ~ ~ ~Ω Ω

and we match against the right-hand side of the first Bloch equation, which is

1 ω0 − ω
− V0 2i sin ωt cos ωt Q(0) + P (0) sin Ωt
i~ Ω
ω0 − ω 2V0 ω0 − ω
+ N (0) − Q(0) (cos Ωt − 1)
Ω ~Ω Ω
2V0 ω0 − ω
+ sin ωt P (0) cos Ωt + N (0) − Q(0) sin Ωt
~Ω Ω
+2iV0 cos ωt {cos ωtP 0 (t) − sin ωtQ0 (t)}]
2V0 0
=− P (t)
~
2V0 2V0 ω0 − ω
=− P (0) cos Ωt + N (0) − Q(0) sin Ωt , (44.3)
~ ~Ω Ω

where on the second line we noticed a Q0 (t) and on the third a P 0 (t), which then
add up to the simple form on the fifth line.
2) Continuing with the second and third Bloch equations, we obtain first for the
left-hand side of the second Bloch equation

dQ(t)
= sin ωt(Ṗ 0 − ωQ0 ) + cos ωt(Q̇0 + ωP 0 )
dt
2V0 ω0 − ω
= sin ωt −Ω sin ΩtP (0) + Ω cos Ωt N (0) − Q(0)
~Ω Ω
ω(ω0 − ω)
−ωQ(0) − P (0) sin Ωt
Ω
ω(ω0 − ω) 2V0 ω0 − ω
+ N (0) − Q(0) (cos Ωt − 1)
Ω ~Ω Ω
2V0 ω0 − ω
+ cos ωt (ω0 − ω)P (0) cos Ωt + (ω0 − ω) sin Ωt N (0) − Q(0)
~Ω
Ω
2V0 ω0 − ω
+ωP (0) cos Ωt + ω sin Ωt N (0) − Q(0) , (44.4)
~Ω Ω
155
156 Chapter 44

to be compared with the right-hand side of the second Bloch equation,

0 0 2V0 2V0
ω0 (P (t) cos ωt − Q (t) sin ωt) + sin ωt N (0) − P (0) sin Ωt
~ ~Ω
2V0 2V0 ω0 − ω
+ N (0) − Q(0)
~Ω ~Ω Ω
2V0 ω0 − ω
= cos ωtω0 P (0) cos Ωt + N (0) − Q(0) sin Ωt
~Ω Ω
ω0 (ω0 − ω)
+ sin ωt −ω0 Q(0) − P (0) sin Ωt
Ω
ω0 (ω0 − ω) 2V0 ω0 − ω
+ N (0) − Q(0) (cos Ωt − 1)
Ω ~Ω Ω
2V0 2V0 2V0 2V0 ω0 − ω
+ N (0) − P (0) sin Ωt + N (0) − Q(0) ,
~ ~Ω ~Ω ~Ω Ω
(44.5)
and we can check that the terms on the two sides match.
Next, for the left-hand side of the third Bloch equation,
dP (t)
= − sin ωt(Q̇0 + ωP 0 ) + cos ωt(Ṗ 0 − ωQ0 )
dt
2V0 ω0 − ω
= − sin ωt cos Ωt(ω0 − ω)P (0) + sin Ωt(ω0 − ω) N (0) − Q(0)
~Ω Ω
2V0 ω0 − ω
+ωP (0) cos Ωt + ω sin Ωt N (0) − Q(0)
~Ω Ω
2V0 ω0 − ω
+ cos ωt −Ω sin ΩtP (0) + Ω cos Ωt N (0) − Q(0) − ωQ(0)
~Ω
Ω
ω(ω0 − ω) ω(ω0 − ω) 2V0 ω0 − ω
− P (0) sin Ωt + N (0) − Q(0) (cos Ωt − 1) ,
Ω Ω ~Ω Ω
(44.6)
to be compared with the right-hand side of the third Bloch equation, giving
2V0
−ω0 [Q0 (t) cos ωt + P 0 (t) sin ωt] + cos Ωt {N (0)
~
2V0 2V0 2V0 ω0 − ω
− P (0) sin Ωt + N (0) − (cos Ωt − 1)
~Ω ~Ω ~Ω Ω
2V0 ω0 − ω
= ω0 sin ωt −P (0) cos Ωt − sin Ωt N (0) − Q(0)
( ~Ω Ω
2
2V0 2V0 P (0)
+ cos ωt N (0) − sin Ωt
~ ~ Ω
2
2V0 1 2V0 ω0 − ω
+ N (0) − (cos Ωt − 1)
~ Ω ~Ω Ω
ω0 (ω0 − ω)
−ω0 Q(0) − P (0) sin Ωt
Ω
ω0 (ω0 − ω) 2V0 ω0 − ω
+ N (0) − Q(0) (cos Ωt − 1) , (44.7)
Ω ~Ω Ω
and we can check that the terms on the two sides match.
157 Chapter 44

3) For the radiation

V̂ (t) = V̂0 eiωt + V̂0 e−iωt e−γt , (44.8)

the equivalent of Fermi’s golden rule is derived as follows. From eq. (44.24) in the
text,
1 t 0
Z
(iω−γ)t0 † (−iω−γ)t0 0
c(1)
n (t) = dt V 0,ni e + V 0,ni e eiωni t
i~ 0
1 1 − e[i(ω+ωni )−γ]t 1 − e[i(ωni −ω)−γ]t †

= V0,ni + V0,ni . (44.9)
~ ω + ωni + iγ −ω + ωni + iγ
That means that ~ωni = En − Ei is now replaced with
~ωni ± ~ω + iγ = En − Ei ± ~ω + iγ , (44.10)
leading to a modified Fermi’s golden rule of
!Z
2π |V0,ni |2
Z
dPni X
= † 2 dE ... ρ(En , ...)δ(En − Ei ± ~ω + iγ). (44.11)
dt ~ |V0,ni |

4) We want to find a limit for which the first order perturbation theory in the
harmonic potential is consistent with the exact, two-level calculation. It is easy to
see that this is V0 → 0 (but keep small perturbations, so that the approximation is
exact), and ω → ωni = E2 −E~
1
.
Then, for first order perturbation theory,
1 1 − ei(ω+ωni )t 1 − ei(ωni −ω)t †

(1)
cn (t) = V0,ni + V0,ni
~ ω + ωni −ω + ωni
(ωni −ω)t (ωni −ω)t (ωni −ω)t
ei 2 e−i 2 − ei 2
†
→ −V0,ni , (44.12)
~ω − (En − Ei )
where we have neglected the first term in the limit, and so the transition probability
becomes
† 2
2iV0,ni sin(ω − ωni )t/2
Pn(1) (t) = |c(1) 2
n (t)| →
~ω − (En − Ei )
(2V0,ni )2 sin2 ω − En −E t
i

~ 2
= . (44.13)
|~ω − (En − Ei )|2
This gives the same result as from the exact two-level system in the same limit
V0 → 0, ω → ωni , since then Ω → ω − ω0 , and
2
(2V0 )2

2V0 Ωt E2 − E1 t
P2 (t) = sin2 → sin2
ω − . (44.14)
~Ω 2 (~ω − (E2 − E1 ))2 ~ 2
5) For a possible quantum term αA ~ 2 , we want to analyze in terms of photons,
like in the text for the linear term in A. ~ We have
h
~ ~0 0 ~ ~0 0
~j (~k) · ~l (~k 0 ) aj (~k)al (~k 0 )ei(k+k )~r−i(ω+ω )t + a†j (~k)a†l (~k 0 )e−i(k+k )~r+i(ω+ω )t
X
αA ~2 ∝
~
k,~
k0 ,j,l
158 Chapter 44

i
~ ~0 0 ~ ~0 0
+aj (~k)a†l (~k 0 )ei(k−k )~r−i(ω−ω )t + a†j (~k)al (~k 0 )e−i(k−k )~r+i(ω−ω )t . (44.15)
√ √
Now, since as before we have hN − 1|a|N i = N and hN + 1|a† |N i = N + 1,
we have two terms,
p
hN − 2|aj al |N i = N (N − 1) , (44.16)
corresponding to an absorption of two photons (at the same time), and
hN + 2|a†j a†l |N i = (N + 1)(N + 2)
p
(44.17)
corresponding to an emission of two photons (at the same time), but we also have
the terms
hN |a†j al |N i = N , and hN |aj a†l |N i = N + δjl δ(~k − ~k 0 ) , (44.18)
where nothing happens.
6) If there are no photons, Nj (~k) = 0 in the radiation field, we interpret the rate
of spontaneous emission as describing emission of the photon from the vacuum. The
limitations of this situations are that this is a highly quantum process, so it has no
classical limit, and therefore must be treated fully within Quantum Field Theory.
7) In deriving the Planck formula, we used the Boltzmann distribution for atoms,
which seems counterintuitive, since we describe a quantum process, but we consid-
ered many atoms at equilibrium (the distribution is over atomic states), in which
case we have a classical distribution for them.
45 Chapter 45

2
1) For the potential V (x) = V0 e−αx , for E ≤ V0 , we want to calculate the first
two terms in the perturbative expansion for ψ1 (x) (∼ e−kx at x → −∞) and ψ2 (x)
(∼ eikx at x → +∞). We have the self-consistency equation
1 x
Z
ψ1 (x) = eikx + dx0 sin[k(x − x0 )]U (x0 )ψ1 (x0 ) , (45.1)
k −∞
(0)
where U (x) = 2mV (x)/~2 , solved perturbatively as ψ1 = eikx and
Z x
(1) 2mV0 1 0

ik(x−x0 ) −ik(x−x0 )
02 0
ψ1 (x) = 2
dx e − e e−αx eikx
~ 2ik −∞ Z
x Z x
2mV0 1 ikx 0 −αx02 −ikx 0 2ikx0 −αx02
= e dx e − e dx e . (45.2)
~2 2ik −∞ −∞

Similarly,
Z +∞
1 0 0

ψ2 (x) = eikx − dx0 eik(x−x ) − e−ik(x−x ) U (x0 )ψ2 (x0 ) , (45.3)
2ik x
(0)
so ψ2 (x) = eikx and
1 2mV0 +∞ ik(x−x0 )
Z
(1) 0 02 0
ψ2 (x) = − 2
e − e−ik(x−x ) e−αx eikx
2ik ~ x
1 2mV0 ikx +∞ 0 −αx02
Z Z +∞
−ikx −αx02 ikx0
=− e dx e −e e e .(45.4)
2ik ~2 x x

2) We have
∂x2 G2 (x, x0 ) = −k sin[k(x − x0 )]θ(x0 − x) + cos[k(x − x0 )]θ0 (x0 − x)
= −k 2 G2 (x − x0 ) + δ(x − x0 ) , (45.5)
so G2 (x, x0 ) is also a Green’s function for the Schrödinger operator, and moreover
∂x2 (G1 − G2 ) = −k 2 (G1 − G2 ) , (45.6)
so G1 − G2 is a solution to the homogenous Schrödinger operator.
3) For the Lippman-Schwinger equation in the momentum representation, we
multiply by hk 0 | the abstract equation
1
|ψi = |ki + V̂ |ψi , (45.7)
E + i − Ĥ0
obtaining
1
ψ(k 0 ) = δ(k − k 0 ) + hk 0 | |k 00 ihk 00 |V̂ |k̃ihk̃|ψi
E + i − Ĥ0
159
160 Chapter 45

Z
0 1
= δ(k − k ) + dk̃V (k 0 , k̃)ψ(k̃). (45.8)
E + i − Ĥ0 (k 0 )
The perturbative solution gives

1 1 1 1
ψ(k 0 ) = δ(k − k 0 ) + hk 0 | + + V̂ + ... V̂ |ki
E + i − ĤZ
0 E + i − Ĥ0 E + i − Ĥ0 E + i
0 1 0
= δ(k − k ) + 0
dk̃V (k , k̃)δ(k̃ − k)
Z Z E + i − H0 (k )
˜ 0 1 ˜ ˜
+ dk̃ dk̃V (k , k̃) V (k̃, k̃)δ(k̃ − k) + ... (45.9)
E + i − H0 (k̃)
4) For V = V0 , |x| ≤ L/2, V = 0, |x| > L/2, and E ≤ V0 , we have:
~2 d2
-For |x| > L/2, the Schrödinger equation is − 2m dx2 ψ(x) = Eψ(x), so the solu-
tion in this region is
~2 k 2
ψ(x) = Aeikx + Be−ikx , = E. (45.10)
2m
2 2
~ d
-For |x| ≤ L/2, the Schrödinger equation is − 2m dx2 ψ(x) = (E − V0 )ψ(x), so the
solution in this region is
~2 κ2
ψ(x) = Ceκx + De−κx , = |V0 − E|. (45.11)
2m
For ψ1 (x): For x < −L/2 (incoming), we have ψ1 (x) = Aeikx , so from the
matching conditions at x = −L/2, we get

Ae−ikL/2 = Ce−κL/2 + De+κL/2

ikAe−ikL/2 = κ(Ce−κL/2 − De+κL/2 ) , (45.12)

from which we find

D κ − ik
= e−κL
C κ + ik
L
(κ−ik) κ + ik
C = Ae 2 . (45.13)
2κ
Matching at x = +L/2 with ψ(x) = Eeikx + F e−ikx , we find

CeκL/2 + De−κL/2 = EeikL/2 + F e−ikL/2

κ(CeκL/2 − De−κL/2 ) = ik(EeikL/2 − F e−ikL/2 ) , (45.14)

from which we find

−2κL
ik
) + iκ(1 − e−2κL ) − κ(1 − e−2κL ) − ik(1 + e−2κL )

F −ikL κ κ(1 + e
e = ik −2κL ) + iκ(1 − e−2κL )] + κ(1 − e−2κL ) + ik(1 + e−2κL )
E κ [κ(1 + e
Ce κL/2
1 + D/Ce−κL
E = ikL/2
e 1 + F/Ee−kL
−2κL κ−ik
κ + ik 1 + e κ+ik
= AeL(κ−ik) . (45.15)
2κ 1 + F/Ee−ikL
161 Chapter 45

Then, at x → +∞,

F
ψ1 (x) = E eikx + e−ikx . (45.16)
E
For ψ2 (x): ψ2 (x) = Eeikx at x → +∞.
Matching at x = +L/2, we have

CeκL/2 + De−κL/2 = EeikL/2

κ(CeκL/2 − De−κL/2 ) = ikE ikL/2 , (45.17)

from which we find

D κ − ik
= eκL
C κ + ik
κ + ik
C = Ee(ik−κ)L/2 . (45.18)
2κ
Then, at x = −L/2, we have

Ae−ikL/2 + Be+ikL/2 = Ce−κL/2 + De+κL/2

ik(Ae−ikL/2 − Be+ikL/2 ) = κ(Ce−κL/2 − De+κL/2 ) , (45.19)

from which we find

ik
2κL

B κ κ(1 + e ) + ik(1 − e2κL ) − κ(1 − e2κL ) − ik(1 + e2κL )
= ik
+ ik(1 + e2κL )
A κ [κ(1 + e 2κL ) + ik(1 − e2κL )] + κ(1 − e2κL )

(ik−κ)L κ + ik
1 + e2κL κ−ik
κ+ik
A = Ee . (45.20)
2κ 1+ B A e ikL

At x → −∞, we have

B
ψ2 (x) = A eikx + e−ikx . (45.21)
A
Then the transfer matrix is
α(k) β ∗ (k)

K= , (45.22)
β(k) α∗ (k)
where

ψ1 (x → +∞) = α(k)eikx + β(k)e−ikx ⇒

−2κL κ−ik
E κ + ik 1 + e κ+ik
α(k) = = e(κ−ik)L
A 2κ 1 + F/EeikL
F EF
β(k) = = . (45.23)
A AE
ˆ
ˆ with Ŝ = ei∆ , we write δ̂ = α0 + αi σ
5) To calculate the S-matrix, and ∆,
(decompose the 2x2 matrix like this), and write
!
β∗
α
1
∗ α ∗ 1 β∗ − β β∗ + β
Ŝ = = 1l + σ 1 + iσ 2
− αβ∗ α1∗ α∗ 2α∗ 2α∗
162 Chapter 45

ˆ ~ · ~σ
α
= ei∆ = eiα0 cos |~
α| 1l + i sin |~
α| , (45.24)
|~
α|
and by identifying the two forms, we get
1 iα0 α2 iα0 β∗ + β
= e cos |~
α | , α 3 = 0 , e sin |~
α | =
α∗ |~
α| 2α∗
α1 iα0 β∗ − β
e sin |~ α| = −i , (45.25)
|~
α| 2α∗
so further, α3 = 0, and defining β = β1 + iβ2 ,

|α(k)| arccos |α1∗ |

eiα0 = , |α2 | =
α∗ (k)
q
β2
1 + β22
1
α1 β2 1
=− , |~
α| = arccos ∗ . (45.26)
α2 β1 |α |
6) The 2-magnon ansatz is
X
|ψ(p1 , p2 )i = ei(p1 x1 +p2 x2 ) + S(p2 , p1 )ei(p2 x1 +p1 x2 ) |x1 x2 i , (45.27)
1≤x1 <x2 ≤L
PL
to substitute in the Schrödinger equation for H = 2J j=1 (Pj,j+1 − 1), and we
have
X
Ĥ|p1 p2 i = 2J eip1 x1 +p2 x2 ) + S(p2 , p1 )eip1 x2 +p2 x1 )
1≤x1 <x2 ≤L
× (|x1 − 1, x2 i + |x1 + 1, x2 i + |x1 , x2 − 1i + |x1 , x2 + 1i) .(45.28)

We redefine the summation in each term (x1 − 1 → x1 , x1 + 1 → x1 , x2 − 1 →

x2 , x2 + 1 → x2 ) in order to have the same |x1 x2 i element, and so we obtain

Ĥ|p1 p2 i = 2J2(cos p1 + cos p2 − 2)|p1 p2 i = (E1 + E2 )|p1 p2 i , (45.29)

except in order for that to be true, we note that (Pj,j+1 − 1)|x1 , x1 + 1i, for j = x1
gives 0 instead of the needed

|x1 + 1, x2 i + |x1 , x2 − 1i − 2|x1 , x2 i , (45.30)

so we need that

ei(p1 x1 +p2 (x1 +1)) + S(p2 , p1 )ei(p1 (x1 +1)+p2 x1 )
×(|x1 + 1, x 1 + 1i + |x1 , x1 i − 2|x1 , x1 + 1i)
= eix1 (p1 +p2 ) eip2 + S(p2 , p1 )eip1 = 0 , (45.31)

which is satisfied by
1
2 cot p22 − 1
2 cot p21 + i
S(p2 , p1 ) = . (45.32)
1
2 cot p22 − 1
2 cot p21 − i

7) To solve the Bethe Ansatz Equations for n = 0, we write p1 + p2 = 0, so

163 Chapter 45

p2 = −p1 , and then from the BAE we have

cot p21 + i cos p21 + i sin p21
eip1 L = S(p1 , −p1 ) = = = eip1 , (45.33)
cot p21 − i cos p21 − i sin p21
which leads to the solution
2πn
p1 = , p2 = −p1 . (45.34)
L−1
We then substitute this into the 2-magnon wave function to obtain
L
X 2l + 1 πn
|ψ(n)i = |ψ(p1 (n), −p1 (n))i = Cn cos πn |x2 + l, x2 i , Cn = 2e−i L−1 .
L−1
l=1
(45.35)
46 Chapter 46

1) For V = V0 > 0 for r ≤ R and V = 0 for r ≥ R, and E > V0 , l > 0, and a square
integrable solution, we have the Schrödinger equation
∂ 2 ψ 2 ∂ψ l(l + 1)
+ − ψ + k2 ψ = 0 , (46.1)
∂r2 r ∂r r2
where weq
have implicitly assumed ψq= REl (r)Ylm (θ, φ), and REl = χEl /r, ρ = 2kr,
and k = 2mE
~2 for r > R and k̃ = 2m(E−V ~2
0)
for r < R. Then we have

∂ 2 χ(ρ)

l(l + 1) 1
+ − + χ(ρ) = 0 , (46.2)
dρ2 ρ2 4
which has the solution
χ(ρ)
REl (ρ) = = jl (ρ/2) = jl (kr) , (46.3)
r
with the correct, square integrable behaviour at r =R0, since jl (kr) ∼ rl for r → 0,
and square integrability of ψ means 0 |χ| drdΩ = 0 |ψ|2 r2 drdΩ < ∞, so ψ goes
2
R

slower than 1/r at r → 0, satisfied by the above solution.

Then the solution is
(
jl (k̃r)Ylm (θ, φ) , r<R
ψ(r, θ, φ) = . (46.4)
[Ajl (kr) + Byl (kr)]Ylm (θ, φ) , r > R
Note that the asymptotics of the Bessel functions are
r
2 (l + 1/2)π π
Jl+1/2 (z) ∼ cos z − −
r πz 2 4

2 (l + 1/2)π π
Yl+1/2 (z) ∼ sin z − − . (46.5)
πz 2 4
and of the spherical Bessel functions
r
π
jl (z) = Jl+1/2 (z)
2z r
π
yl (z) = (−1)l+1 Yl+1/2 (z) , (46.6)
2z
leading to

1 (l + 1)π
jl (kr) ∼ cos kr −
kr 2
1 (l + 1)π
yl (kr) ∼ sin kr − , (46.7)
kr 2
164
165 Chapter 46

and
1 ±i(kr− π (l+1))
jl (kr) ± iyl (kr)(−1)l+1 ∼ e 2 . (46.8)
kr
The matching conditions for the wave function at r = R are

jl (k̃R) = αjl (kR) + βyl (kR)

k̃jl0 (k̃R) = k (αjl0 (kR) + βyl0 (kR)) , (46.9)

out of which we find α, β, and then, using the asymptotics at infinity of the spherical
Bessel functions, we have

αjl (kr) + βyl (kr) = A(jl (kr) + i(−1)l yl (kr)) + B(jl (kr) − i(−1)l yl (kr)) ⇒
α − βi(−1)l+1 α + βi(−1)l+1
A= , B= . (46.10)
2 2
2) The relation (46.15),

~ 2π eikr 2π e−ikr
eik·~r ' δ(~nk − ~nr ) − δ(~nk + ~nr ) , (46.11)
k r k r
is understood, as shows its proof, as a distribution in ~nr , i.e., by d2~nr f (~nr )× it,
R

and its says that the bulk of the integral comes from the ~k parallel or antiparallel
to the ~r region.
3) If f~k (~nr ) in
~ eikr
u~+ (~r) ' eik·~r + f~k (~nr ) (46.12)
k r
would be independent of ~nr , it would not contradict unitarity, since then we would
still have
ik ik
Ŝ = 1l + f~k ⇒ Ŝ † = Ŝ −1 = 1l − f~ , (46.13)
2π 2π k
and Ŝ would still be unitary. In other words, the scattered wave is not larger than
the incoming wave. Note that the optical theorem is
Z
k
Imf~k = |f~k |2 d2~nr , (46.14)
2π
and it is still OK.
R
4) For the case at exercise 1), since we have by definition A = S · B, and more
precisely
ik
Ŝ = 1l + f, (46.15)
2π
it follows that, if f is small (so we can ignore f 2 terms), we have

f ' A − B = −βi(−1)l+1 , (46.16)

and
Z Z
σtot = d2~nr |f |2 = d2 Ω|f |2 . (46.17)
166 Chapter 46

5) Considering
e−ikr
ikr
2π e 2π
ψ= δ(~nk + ~nr ) + a + δ(~nk − ~nr ) , (46.18)
k r k r
with a, b ∈ R, we can interpret it as a scattering solution. Then
ik ik
f = A − B = a − b , Ŝ = 1l +f = 1l + (a − b) , (46.19)
2π 2π
R
and A = S · B works, except for the b(a − b) term, which can be neglected. Then
Z Z
σtot = d ~nr |f | = d2 Ω|f |2 = 4π|a − b|2 .
2 2
(46.20)

The case with a, b ∈ C doesn’t work, since then we have no optical theorem.
6) To calculate the generalized Green’s function G0 (z; ~r, ~r0 ) for complex energy
z ∈ C, we write
1 1
Ĝ0 (z) = ⇒ Ĝ0 (z; ~r, ~r0 ) = h~r| |~r0 i. (46.21)
z − Ĥ0 z − Ĥ0
k2 2 ~2 k 2
We then define, analogously to E = ~2m , z = 2mz , then follow everything up to
eq. (16.61) in the text, with k 2 ± i replaced by kz2 ∈ C, so
Z +∞ 0 0
0 1 eiq|~r−~r | − e−iq|~r−~r |
G0 (z; ~r, ~r ) = − 2 qdq . (46.22)
8π i|~r − ~r0 | −∞ q 2 − kz2
This has poles at q = ±kz ∈ C. If Im kz > 0, then this is like G+
0 , if Im kz < 0,
this is like G−
0 , so finally
1 0
G0 (z; ~r, ~r0 ) = − 0
e±ikz |~r−~r | , (46.23)
4π|~r − ~r |
with the ± corresponding to Im kz > or < 0, so that always the exponential is
decreasing.
7) For the equivalent of eq. (46.69) in the text, i.e.,
e±ikr 2m
Z
~˙ ~0 0
ψ± (~r) = eik~r − d3~r0 e∓ik ·~r V (~r0 )ψ± (~r0 ) (46.24)
r 4π~2
for G0 (z; ~r, ~r0 ), we substitute it in
Z
2m
ψz (~r) = φz (~r) + 2 d3~r0 G0 (z; ~r, ~r0 )V (~r0 )ψz (~r0 ) , (46.25)
~
~
and use φz (~r) = eikz ~n·~r ≡ eikz ·~r , so
Z ±ikz |~ r0 |
r −~
~
kz ·~ 2m 3 0e
ψz (~r) = e r
− d ~r V (~r0 )ψz (~r0 ). (46.26)
4π~2 4π|~r − ~r0 |
0 0
Using e±ikz |~r−~r | ' e±ikz r e∓ikz ~nr ·~r , we finally find
2m e±ikz r
Z
i~
kz ·~ ~0 0
ψz (~r) = e r
− 2
d3~r0 e∓ikz ·~r V (~r0 )ψz (~r0 ) , (46.27)
4π~ r
where ~kz0 = kz ~nr .
47 Chapter 47

1) For V (r) = −V0 δ 3 (~r), we want to calculate the first 2 terms in the Born series.
We note that the potential is spherically symmetric so, with q = |~k 0 − ~k|, we have
Z
2m ~ 0 ~ 2m 0 ~0 ~
(1)
f =− 2
V (k − k) = − 2
d3~r0 ei~r ·(k −k) [−V0 ]δ 3 (~r)
4π~ 4π~
2m
= V0 , (47.1)
4π~2
so in first order we have

(1) eikr (1) 2m eikr

ψ± (~r) = f± = V 0 , (47.2)
r 4π~2 r
while in second order we have
Z
(2) 2m ± (1)
ψ± (~r) = d3~r0 G (~r, ~r0 )ψ± (~r0 )
4π~2 0 !
0 0
e±ik|~r−~r | e±ikr
Z
3 0 2m 2m
= d ~r 2 − V0 (−V0 )δ 3 (~r0 )
~ 4π|~r − ~r0 | 4π~2 r0
2 ±ikr

2m 2e 1
= V 0 , (47.3)
4π~2 r r0 r0 →0

so it is divergent.
2) For the potential V (r) = A/r2 , we want to calculate dσ/dΩ in the Born
approximation. We have
2m
f (1) = − V (~k 0 − ~k) , (47.4)
4π~2

where ~k 0 = k~nr and ~q = ~k 0 − ~k. Because of spherical symmetry,

4π ∞ 0 0 A
Z Z
0
V (q) = d3~r0 ei~r ·~q V (r) = r dr 02 sin qr0
q 0 r
4πA ∞ dz eiz − e−iz 4πA +∞ dz eiz
Z Z
= =
q 0 z 2i q −∞ z 2i
4πA π
= , (47.5)
q 2
where the integral was done by extending it in the complex plane, adding for free
a semicircle at infinity in the upper-half plane, thus forming a closed contour, and
using the residue theorem to equate it to 2πi times the residue of the integrand in
the upper-half plane (or more precisely, to πi times the residue on the real line).
167
168 Chapter 47

Then we have
2 2 2
dσ (1) 2π 2 A

2m 2m
= |f (1) | = [V (q)]2 = . (47.6)
dΩ 4π~2 4π~2 q
3) To understand how is it possible for the Born approximation to quantum
mechanics for scattering in Coulomb potential to give the same result as classical
scattering, we note that what this means is that the classical result is exact, and
there are no quantum corrections to it. This means the Coulomb case is a very
special case, and it is not expected to be true in general.
4) For the potential V (r) = A/(r2 + a2 ), we want to calculate the first two terms
in the Born series. By spherical symmetry, we have
2m
f (1) = − V (q) , (47.7)
4π~2
and
4π ∞ 0 0 A
Z Z
0 A
V (q) = d3~r0 ei~r ·~q 2 = r dr 02 sin qr0
r + a2 q 0 r + a2
4πA +∞ zdzeiz
Z
=
2iq −∞ (z + iqa)(z − iqa)
2π 2 A iqa
= e , (47.8)
q
where again we have extended the integral over R to an integral over a closed
contour, by adding for free the semicircle at infinity in the upper-half plane, and
then evaluated it using the residue theorem as 2πi times the residue in the upper-
half plane, specifically at z = +iqa. We note that the result is consistent with the
result at exercise 2, since the a → 0 limit gives V = A/r2 , and then the V (q) ’s
match.
Then
2m 2π 2 A iqa
f (1) = − e , (47.9)
4π~2 q
and the first term in the Born series is
(1) e±ikr (1)
ψ± = f , (47.10)
r
whereas the second term in the Born series is
!
Z ±ik|~ k0 |
r −~
(2) 2m e (1)
ψ± = d3~r0 − ψ± (~r0 )V (~r0 )
4π~2 4π|~r − ~r0 |
2 2 iqa Z 0 0
e±ik|~r−~r | e±ikr

2m A e 3 0
= d ~
r , (47.11)
4π~2 q |~r − ~r0 |(r02 + a2 ) r0
0 0 0
and, using e±ik|~r−]~r | ' e±ikr e∓i~r ·~r as in the text, we obtain
2 2 iqa ±ikr Z ~0 0 0
e∓ik ·~r ±ikr

(2) 2m A a e 3 0
ψ± ' d ~r 0 0
4π~2 q r r (r + ia)(r0 − ia)
169 Chapter 47

2 2 iqa ±ikr 0
2π ∞ 0 1 − e±2ikr
Z
2m A e e
= dr
4π~2 q r k 2i(r0 + ia)(r0 − ia)
2 2 iqa ±ikr 0 Z ∞
2m A e e 1 1 dz dz
= 2π −
4π~2 q r 2i 2ika 0 z − ika z + ika
1 ∞ dze±2iz
Z
−
2i 0 (z + ika)(z − ika)
2 2 iqa ±ikr
2m A e e πi 1 −2ka
= 2π + e , (47.12)
4π~2 q r 2ka 2
where in doing the second integral over dz we again added a semicircle at infinity
and used the residue theorem.
−µr
5) For V (r) = V0 e r , the Lippman-Schwinger equation for the T-matrix is
T̂ = V̂ + V̂ Ĝ0 T̂ , (47.13)
and in coordinate space, acting on momentum space, it is
h~r|T̂ |~ki = h~r|V̂ |~ki + h~r|V̂ Ĝ ~
Z 0 T̂ |ki −µr
e−µr i~k·~r Vo e
= V0 e + d3~r0 G0 (~r, ~r0 )h~r0 |T̂ |~ki , (47.14)
r r
and Z
h~r|T̂ |~ki = d3~k 0 ei~k · ~rT (~k, ~k 0 ). (47.15)

6) For V (~r) = −V0 δ 2 (~r), as we saw V (~k) = −V0 . The Lippman-Schwinger equa-
tion for T̂ is
( 1l − V̂ Ĝ0 )T̂ = V̂ ⇒ T̂ = ( 1l − V̂ Ĝ0 )−1 V̂ , (47.16)
so
1 1 1
h~k|T̂ |~k 0 i = h~k| V̂ |~k 0 i = V (~k 0 )h~k| |~k 0 i = −V0.
1 − V̂ Ĝ0 1 − V̂ Ĝ0 δ(k − k ) + V0 G0 (~k, ~k 0 )
~ ~ 0
(47.17)
7) There is no domain of validity of the Born approximation in the case of the
Coulomb potential, since the Born approximation is always valid in the case of the
Coulomb potential.
48 Chapter 48

1) For V (~r) = −V0 δ 3 (~r), in the Born approximation we have

2m 2mV0
f (1) = − V (q) = + . (48.1)
4π~2 4π~2
On the other hand, we expand
X
f (1) = (2l + 1)al (k)Pl (cos θ) , (48.2)
l≥0
so
2mV0
a0 = f (1) =
, al = 0 , l > 0. (48.3)
4π~2
2) For V = −V0 , r ≤ R, V = 0, r > R, in the Born approximation
2m
f (1) (~k, ~k 0 ) = − V (~k − ~k 0 ) , (48.4)
4π~2
and
Z ∞ Z R
0 0 0 0
V (q) = 4π r dr V (r ) sin qr = −4πV0 r0 dr0 sin qr0
0 0
4πV0
=+ (qR cos qR + sin qR). (48.5)
q2
Then
2mV0
f (1) (~k, ~k 0 ) = − (qR cos qR + sin qR)
~2 q 2
X e2iδl (k) − 1
= (2l + 1)Pl (cos θ) . (48.6)
2ik
l≥0
q
1−cos θ
Since q = 2k sin θ/2, so qR = 2kR 2 , and we have

2mV0 R2 X

1 1
f (1) (~k, ~k 0 ) = − (−1)m
(qR)2m
+
~2 (qR) (2m)! (2m + 1)!
m≥0
r 2m
2mV0 R2 (1 − cos θ)m

2 X kR
=− 2 (−1)m √ (2m + 2).
~ 2kR 1 − cos θ 2 (2m + 1)!
m≥0
(48.7)
It is complicated in practice to match this against the general expansion, so we
will not continue.
3) We have
Sl (k) = e2iδl (k) (48.8)
170
171 Chapter 48

and
2
dσ 2mV0
= |f (1) (~k, ~k 0 )|2 = (qR cos qR + sin qR)2 . (48.9)
dΩ ~2 q 2
4) If δl (k) ∈ R, we want to deduce something about Tl (k):
e2iδl (k) = Sl (k) = 1 − 2πiTl (k) ⇒ Sl† (k) = e−2iδl (k) = (Sl (k))−1 , (48.10)
so unitarity means
Sl Sl† = 1l ⇒ (1 + 2πiTl† )(1 + 2πiTl ) = 1 ⇒ 2πi(Tl† (k) − Tl (k)) + (2π)2 |Tl (k)|2 = 0.
(48.11)
5) The scattering length is
2mV0
a = − lim a0 (k) = −f (1) = − . (48.12)
k→0 4π~2
6) Is the relation
1
−Fl− (k)fl+ (k; r) + Fl+ (k)fl− (k; r)

φl = (48.13)
2ik
well defined for k ∈ C? Yes, it is, since for k ∈ C, φl , fl+ , fl− are still linearly
dependent, and we can still define φl , fl± in the same way, through
lim e±ikr fl± (k; r) = 1 , (48.14)
r→∞

but in this case, since k ∈ C, fl± (k; r) is not a pure phase at infinity, since |eikr | =
6 1.
F + (k)
7) If limr→0 lk2 is a constant for l even, to find the k → 0, r → ∞ behaviour
with k of φl (k; r) for even l we write
1
−Fl− (k)fl+ (k; r) + Fl+ (k)fl− (k; r) ,

φl (k; r) = (48.15)
2ik
and at r → ∞ fl± (k; r) ∼ e∓ikr , while for k → 0, Fl+ (k) ∼ k 2 × constant, and
Fl± (−k) = Fl∓ (k), so
Fl− (−k) Fl+ (k)
lim = lim = constant ≡ A , (48.16)
k→0 k2 k→0 k2
and for l even, Fl± (−k) = Fl (k), so finally then
eikr − e−ikr
φl (k; r) ∼ Ak = Ak sin kr. (48.17)
2i
49 Chapter 49

1) We use the Lippman-Schwinger equation for ψ, in the form of eq. 49.16 in the
text,
Z
2m ~0 0
fk (θ) = 2
d3~r0 e∓ik ·~r V (~r0 )ψ± (~k; ~r0 ) , (49.1)
4π~
to write an expression for σl . Indeed,
dσ
= |fk (θ)|2 , (49.2)
dΩ
P
and σtot = l≥0 σl , where
π
σl = (2l + 1)|e2iδl (k) − 1|2 = π(2l + 1)|al (k)|2
k2
Z 1 2
= π(2l + 1) d(cos θ)Pl (cos θ)fk (θ)
−1
Z 1 Z 2
2m ~0 0
= π(2l + 1) d(cos θ)Pl (cos θ) d3~r0 e∓ik ·~r V (~r0 )ψ± (~k; ~r0 ) .(49.3)
−1 4π~2
2) For the details of going from (49.19) to (49.20) in the text, from
2mik ∞ 0 02
Z
(1)
Rl (k; r) = jl (kr) − dr r jl (kr< )hl (kr >)V (r0 )Rl (k; r0 ) , (49.4)
~2 0

where r< , r> refers to the smaller and larger between r, r0 , we note that Rl (k 0 ; r)
(the left-hand side) is from the expansion of ψ in (2l + 1)Pl (cos θ) (see eq. (48.24)
in the text), and jl (kr) on the right-hand side is from the expansion in (2l +
~
1)Pl (cos θ) of eik·~r , and the second term on the right-hand side is from the expansion
ikr
of f (1) (~k, ~k 0 ) e r in l≥0 al (2l + 1)Pl (cos θ) (note that at r → ∞, thus for r0 < r,
P
(1) ikr (2) −ikr (1,2)
we have hl (kr) ∼ eikr i−l , hl (kr) ∼ e ikr il , and hl = jl ± inl ).
The integral on the right-hand side equals
Z r Z ∞
(1) (1)
dr0 r02 jl (kr0 )hl (kr)V (r0 )Rl (k; r0 ) + dr0 r02 jl (kr)hl (kr0 )V (r0 )Rl (k; r0 ) ,
0 r
(49.5)
and for r → ∞, the second integral can be neglected, finally arriving at
2m ∞ 0 02
Z
al = − 2 dr r jl (kr0 )V (r0 )Rl (k; r0 ). (49.6)
~ 0
q.e.d.
3) For V (r) = −λδ(r − a), we want to prove that Rl (k; r) is given by (49.31) and
the Jost functions by (49.32) in the text.
172
173 Chapter 49

From the Lippman-Schwinger equation for the radial Green’s functions in eq.
49.28,
Z
(0) (0)
g̃l (r; r0 ) = g̃l (r; r0 ) + dr00 r002 g̃l (r; r00 )V (r00 )g̃l (r00 ; r0 ) , (49.7)

we obtain
(0) (0)
g̃l (r; r0 ) = g̃l (r; r0 ) − λa2 g̃l (r; a)g̃l (a; r0 ) , (49.8)
so
(a;r 0 )
0 (0) (0) g̃l
g̃l (a; r ) = g̃l (a; r0 )−λa2 g̃l (a; a)g̃l (a; r0 ) ⇒ g̃l (a; r ) = 0
(0)
, (49.9)
1 + λa2 g̃l (a; a)
which can be put back in the Lippman-Schwinger equation to obtain (49.30) in the
text, namely
(0) (0)
(0) g̃l (r; a)g̃l (a; r0 )
g̃l (r; r0 ) = g̃l (r; r0 ) − λa2 (0)
. (49.10)
1 + λa2 g̃l (a; a)
Then, from (49.27) in the text,
Z
+(0)
Rl (k; r) = jl (kr) + dr0 r02 g̃l (E; r; r0 )V (r0 )Rl (k; r0 ) , (49.11)

where the left-hand side corresponds to g̃l and the first term on the right-hand side
(0)
corresponds to g̃l , then (49.30) in the text implies
(0)
g̃l (E; r; a)jl (kr)
Rl (k; r) = jl (kr) − λa2 +(0)
, (49.12)
1 + λa2 g̃l (E; a; a)
which is (49.31) in the text.
For the Jost solutions, φ∓
l (k; r) ∼ e
±ikr
at r → ∞, but

(1) eikr −l (2) e−ikr l

hl (kr) ∼ i , hl (kr) ∼ i , (49.13)
ikr ikr
so
(1,2)
fl∓ (k; r) ∼ ±ikrhl (kr). (49.14)
q.e.d.
4) Expanding the optical theorem for f (θ = 0) in l,
X X
f (θ) = (2l + 1)Pl (cos θ)al ⇒ f (θ = 0) = (2l + 1)al Pl (1) , (49.15)
l l

so we obtain
m2 k
X Z
(2l + 1)(Imal )Pl (1) = [(2π) 3
] dΩ|h~k|T̂ |~k 0 i|2 . (49.16)
2~2
l

Expanding the T-matrix in l,

X
h~k|T̂ |~k 0 i = (2l + 1)Pl (cos θ)Tl , (49.17)
l
174 Chapter 49

where θ is the angle between ~k and ~k 0 .

For the Born series,
XZ 2m
f (θ) = d3~r0 2 G0 (~r; ~r0 )V (~r0 )ψ (n) (~r0 )
n
~
X 2m Z Z
= 2
d ~r G0 (~r; ~r )V (~r ) d3~r00 G0 (~r0 ; ~r00 )V (~r00 )... ,
3 0 0 0
(49.18)
n
~

and for the T-matrix

X
h~k|T̂ |~k 0 i = h~k|V̂ Ĝ0 V̂ ...Ĝ0 V̂ |~k 0 i. (49.19)
n

5) For a potential with finite range and V = V0 > 0 for r < r0 and V = 0 for
r > r0 , with E > V0 , we write the solution
(
Cl jl (kr) + Dl nl (kr) , r > r0
REl (r) = , (49.20)
C̃l jl (k̃r) + D̃l nl (k̃r) , r < r0
2 2
k̃
where ~2m = E −V0 . Imposing regularity at r = 0, since jl (x) ∼ xl and yl (x) ∼ x−l ,
we must put to zero its coefficient, so D̃l = 0. Then at r = r0 we have the matching
conditions
j 0 (kr0 ) cos δl − n0l (kr0 ) sin δl j 0 (k̃r0 )
q0l = kr0 l = k̃r0 l . (49.21)
jl (kr0 ) cos δl − nl (kr0 ) sin δl jl (k̃r0 )
6) For the hard sphere, we have
jl (kr0 )
tan δl = , (49.22)
nl (kr0 )
so
π 2 4π 1
σl = (2l + 1) e2iδl (k) − 1 = 2 (2l + 1)
k 2 k 1 + tan21δl (k)
4π 1
= 2 (2l + 1) n 2 (kr ) . (49.23)
k 1 + 2l 0 jl (kr0 )

7) We consider the high energy limit, k → ∞, for σl of the hard sphere, and
obtain
jl (kr0 ) (l + 1)π
tan δl (k) = → cot kr0 − , (49.24)
nl (kr0 ) 2
so it oscillates! That means that, as a function of l, the contribution to
4π 1
σl = (2l + 1) (49.25)
k2 1 + tan21δl (k)
varies.
50 Chapter 50

1) For the step potential, with E < V0 , the l = 0 solution regular at r = 0 is

χ(r) = Ñ sinh κr , (50.1)

while in the general case (for general l) we have the equation

d2 χ 2m ~2 l(l + 1)

+ E − V 0 − χ = 0. (50.2)
dr2 ~2 2mr2
For r > r0 , the solution is
χ
= R = Cl jl (kr) + Dl nl (kr) , (50.3)
r
so we have
jl0 (kr0 ) cos δl − n0l (kr0 ) sin δl
q0l = kr0 , (50.4)
jl (kr0 ) cos δl − nl (kr0 ) sin δl
while for r < r0 , we have the modified spherical Bessel function solution
r
π
il (x) = Jl+1/2 (ix) , (50.5)
2x
which is nonsingular at x = 0, so on this side we have

i0l (k̃r0 )
q0l = k̃r0 , (50.6)
il (kr0 )
to be matched against the value at r > r0 .
2) We want to find the first correction to (50.30),
~2 κ2 ~2 1
Ibd.state = = , κ'− , (50.7)
2m 2ma2 a
between the binding energy of the bound state close to zero and the scattering
length. But
~2 κ2 ~2 ~2

1 2r0
κ= ⇒ Ibd.state = = 2 ' 1+ . (50.8)
r0 − a 2m 2ma2 1 − r0 2ma2 a
a

3) To find the equivalent of (50.30) in the text, if we have still r0 a, but

E = |V0 |/2 and very small (kr0 1), we write

~2 k̃ 2 V0 3V0
=E− = , (50.9)
2m 2 2
175
176 Chapter 50

and kr0 1 means that the outside solution can again be rewritten as Be−kr , and
= C(r − a) (at low energy), so
~2 κ2 ~2
T = = . (50.10)
2m 2m(r0 − a)2
4) Extending from l = 0 to general l the analysis of the bound state for Sl (k),
δl (k), al (k), we have for a single bound state
k + iκ
Sl (k) ' − = e2iδl (k) , (50.11)
k − iκ
and
Sl (k) − 1 2 1
al (k) = '− = ⇒
2ik −κ − ik k cot δl (k) − ik
k cot δl (k) ' −κ. (50.12)

5) In the complex k ∈ C plane, we still have δl (k) ∈ R, since if δl (k) ∈ C, it

means |Sl (k)| =
6 1, so that would be an inelastic scattering.
6) For a single bound state,

k + iκ ilπ F + (k)
Sl (k) = − e = e2iδl (k) = l− eilπ , (50.13)
k − iκ Fl (k)

and from the theorem stating that the zeroes of Fl+ (k) in the lower-half plane (Im
k < 0) are on iR (the imaginary axis) ↔ bound states, and are simple zeroes, so

Fl+ (k) = A(k + iκ) ⇒ Fl− (k) = Fl+ (−k) = −A(k − iκ). (50.14)

7) We want to prove

2 −4kn2
Nnl = , (50.15)
dFl+ (−iκ)
Fl− (−iκ) dκ
κ=κn

where, for physical bound states, the physical reduced radial function is

χl (kn = −iκn ; r) = Nnl φl (kn = −iκn ; r). (50.16)

We have
1
−Fl− fl+ (k; r) + Fl+ (k)fl− (k; r) ,

φl = (50.17)
2ik
where fl± (k; r) ∼ e∓ikr for r → ∞, so

Fl− (k) e−ikr Fl+ (k) e+ikr

χl
Rl = ∼ Nnl − + (50.18)
r 2ik r 2ik r
at r → ∞. Then, for k = −iκn ,
Fl (−iκn ) e−κn r Fl+ (−iκn ) e+κn r

R̃l ∼ Nnl − + , (50.19)
2κn r 2κn r
177 Chapter 50

which needs to be normalized,

Z ∞
drr2 R̃l2 (k; r) = 1 , (50.20)
0

so for consistency we must have Fl+ (−iκn ) = 0 (so that there is no diverging eκn r
component). Then, for k 6= −iκn (κ 6= κn ),
d +
Fl− (iκ) = Fl+ (−iκ) = F (−iκ) , (50.21)
dκ l κ=κn

and the normalization condition becomes

− 2 Z ∞
2 [Fl (−iκn )]
2
dr fl+ (−iκn ; r) = 1 ,

Nnl 2
(50.22)
4κn 0
so
2 4κ2n
Nnl = ∞ 2 . (50.23)
[Fl− (−iκn )]2 0 dr fl+ (−iκn ; r)
R
51 Chapter 51

1) To write down the approximate form for Sl (k) with two resonances and 2 bound
states, we note that for bound states, near them we have
k + iκi ilπ
Sl (k) ' − e . (51.1)
k − iκi
On the other hand, for resonances of Sl (k), we saw that if kpole = k1 − ik2 , then
kzero = k1 + ik2 , which in turn means kzero = −k1 + ik2 , and kpole = −k1 − ik2 .
In general then, for the resonance,
k − k1 − ik2
Sl (k) ' . (51.2)
k − k1 + ik2
If the resonances and bound states are isolated, they don’t interfere, and then we
expect to have
(1) (1) (2) (2)
k + iκ1 k + iκ2 k − k1 − ik2 k − k1 − ik2
Sl (k) ' −eilπ , (51.3)
k − iκ1 k − iκ2 k − k (1) + ik (1) k − k (2) + ik (2)
1 2 1 2
(2) (1) (1) (2)
but, as we saw, the resonances come in pairs, so k2 = k2 , k1 = −k1 .
Another possibility is to have a sum of terms instead of the product, which will
also be an approximation (the correct result will be, of course, neither formula in
general)
2) If only σ1 is near the resonance, and in general
4π k22 4π (Γ/2)2
σl (k) = (2l + 1) 2 = (2l + 1) , (51.4)
k2 (k − k1 )2 + k2 k2 (E − E1 )2 + (Γ/2)2
then only for σ1 we have k ' k1 . Since, by assumption, near the resonance k2 k1 ,
and σl ≤ σ1 , then (if the other σl ’s are far from the resonance) σtot ' σ1 .
3) If Sl (k) has a single pole (”resonance”), but with k1 ∼ k2 , from (51.15) near
the pole
4π k22
σl (k) ' 2 (2l1 ) (51.5)
k (k − k1 )2 + k22
is still valid, and also
~2 2 ~2
Eres = (k1 − k22 ) , Γ/2 = 2k1 k2 . (51.6)
2m 2m
~2
Moreover, for k ' k1 , again E − E1 ' 2m 2k1 (k − k1 ), but in general
~2 2 ~2 2
Eres = (k1 − k22 ) = E1 − k . (51.7)
2m 2m 2
178
179 Chapter 51

We also still have

k2 Γ/2
tan δl = − =− , (51.8)
k − k1 E − E1
and
E − E2 + iΓ/2
Sl (k) = , (51.9)
E − E1 + iΓ/2
so we still have
4π (Γ/2)2
σl (k) ' (2l + 1) , (51.10)
k2 (E − E1 )2 + (Γ/2)2
with maximum at E = E1 , just that now

~2 2
Eres = E1 − k 6= E1 . (51.11)
2m 2
4) If we have two resonances close to each other for σl , and at resonance E ' E1
(on the real line), the radial wave function is

eikr e−ikr

Rl (k; r) = C Sl (k) +
" r r #
(1) (2)
k − k1 − ik2 k − k1 − ik2 eikr e−ikr
'C (1) (2) r
+ , (51.12)
k − k1 + ik k − k2 + ik r
2 2

so
eikr e−ikr
χl (k; r) ' (1) (2)
+ (1) (2)
.
(k − k1 + ik2 )(k − k1 + ik2 ) (k − k1 − ik2 )(k − k1 − ik2 )
(51.13)
Consider k = k1 + δk, such that we have a wave packet

eiδk(r−v1 t)
Z
E1
ψ(r, t) = ei(k1 r− ~ t) dδk (1) (2)
(δk + ik2 )(δk + ik2 )
eiδk(r−v1 t)
Z
E1
−e−i(k1 r− ~ t) dδk (1) (2)
" (δk − ik2 )(δk − ik2 ) #
(1) (2)
E1 ek2 (r−v1 t) ek2 (r−v1 t)
= −2πiei(k1 r− ~ t) (2) (1)
+ (1) (2)
θ(v1 t − r)
i(k 2 − k 2 ) i(k2 − k 2 )
(1) (2)
" #
−i(k1 r−
E1
t) ek2 (r+v1 t)
ek2 (r+v1 t)
+2πie ~
(1) (2)
+ (2) (1)
θ(−v1 t − r).
i(k2 − k2 ) i(k2 − k2 )
(51.14)

5) In the case at the previous exercise, in the ”in” region, for t < −r/v1 , there
are 2 waves,
(1) (2)
k2 (r+v1 t)
E1
t) e − ek2 (r+v1 t)
ψ(r, t) = 2πie−i(k1 r− ~
(1) (2)
, (51.15)
i(k2 − k2 )
180 Chapter 51

in the intermediate region −r/v1 < t < r/v1 , again we have ψ = 0, and in the
”out” region, for t > r/v1 , again we have two waves,
(1) (2)
k2 (r−v1 t)
E1
t) e − ek 2 (r−v1 t)
ψ(r, t) = −2πiei(k1 r− ~
(2) (1)
. (51.16)
i(k2 − k1 )
6) For the Levinson theorem, nlb only means poles of Sl (k) on the imaginary line,
or also a bit off it? nlb is the number of energy levels = number of bound states =
number poles of Sl (k) on the imaginary line.
If the poles are not on the imaginary line, but a bit off it, k = iκ + δk,
eikr 1 e−ikr e−kr+iδkr 1 eκr−iδkr
ψ∼ − ∼ − . (51.17)
r Sl (k) r r Sl (iκ + δk) r
If the bound state is a pole at iκ + δk, then
e−κr+iδkr
ψ∼ , (51.18)
r
so we have an extra phase, otherwise it seems OK, which would seem to indicate
the pole could be slightly off as well. However, in practice, then REl (r) would not
be real anymore, so the factor sin(kr − lπ/2 + δl ) will get changed. So, in fact, we
cannot have the pole a bit off.
7) If Sl (k) has a single resonance very close to R for all l, by considering l ∈ C,
S(l; E) has a pole = zero of B(l; E) = Fl− (k), so the Regge trajectory collapses to
a point on the real line (all l’s, which should be = αi (E), now are at a single point).
52 Chapter 52

1) For the potential V (x) = V0 /(x2 +a2 ), scattering with an energy 0 < E < V0 /a2 ,
the turning points are
r
V0
x1,2 = ± − a2 . (52.1)
E
Considering a wave coming in from the left, for x < x1 we have the transition
rule at x1
√ " Z s #
1 x 0

2 V0 π
i1/4 sin ~ dx 2m E − 2 −
h x + a2 4
2m E − x2V+a
0
2
x1
" s #
1 x 0
Z
1 V0
→h i1/4 exp − ~ dx 2m −E . (52.2)

V0 x1 x2 + a2
2m x2 +a2 − E

Then, at x2 , we have
" s #
Z x2
1 1 0 V0
i1/4 A exp + ~ dx 2m −E
h x + a2
2
2m x2V+a
0
2 − E x

√ " Z s #
1 x 0

A 2 V0 π
→h i1/4 sin ~ dx 2m E − 2 + , (52.3)
x + a2 4
2m E − x2V+a0
2
x2

where
" s #
Z x2
1 0 V0
A = exp − dx 2m −E . (52.4)
~ x1 x2 + a2

2) For the 3d problem with V (r) = V0 /(r2 + a2 ) and 0 < V0 < V0 /a2 , the
transition point is
r
V0
r0 = − a2 , (52.5)
E
and the wave function of angular momentum l is
√
kr
ψ = Ylm (θ, φ) W , (52.6)
r
√
where krW = χ(r), and in the WKB approximation, for r < r0 ,
Z x0
1
WWKB (x ≡ ln kr) = p exp − dx0 κ(x0 ) , (52.7)
κ(x) x

181
182 Chapter 52

where

2 2mr V0
k (x) = 2 E− − (l + 1/2)2 , k 2 (x) = −κ2 (x) , (52.8)
~ r2 + a2
and for r > r0 ,
Z x
2 π
WWKB (x = − ln kr) = p sin dx0 k(x0 ) + . (52.9)
κ(x) x0 4

3) In the case at the previous exercise, we have

Z ∞ "r #
WKB π lπ 0 2
2m V0 (l + 1/2)2
δl = + − kr0 + dr k − 2 02 − − k , (52.10)
4 2 r0 ~ r + a2 r02

and
4π 2
X
σl = (2l + 1) sin δ l , σ tot = σl . (52.11)
k2
l

If r0 a (so if V0 /E a2 ), we have
"r # "r #
Z ∞ Z ∞ ∞
0 A 0 A 2 /k 2 A2 1 A2
dr k 2 − 02 − k = k dr 1− 02
−1 ' 2 0 =− ,
r0 r r0 r 2k r r0 2k 2 r0
(52.12)
where
2mV0
A≡ − (l + 1/2)2 . (52.13)
~2
4) For V (r) = V0 /(r2 + a2 ), in the eikonal approximation,
Z +∞
mV0 +∞
Z
m p dz
∆(b) = − 2 dzV ( b2 + z 2 ) = − 2
2~ k −∞ 2~ k −∞ z + a2 + b2
2
mV0 π
=− √ . (52.14)
2~2 k a2 + b2
If V = 0 for r ≥ r0 , for b < r0 we have
Z +√r02 −b2
m V0
∆(b) = − 2 dz 2
2~ k −√r02 −b2 z + a2 + b2
p
mV0 1 r2 − b2
=− 2 √ arctan √ 0 . (52.15)
~ k a2 + b2 a2 + b2
If r0 is large, then in the eikonal approximation,
Z r0 h i
f (+) (θ) = ik db bJ0 (kbθ) e2i∆(b) − 1 , (52.16)
0
√
r02 −b2 π
and, for r0 a, b, we obtain arctan √
a2 +b2
' 2, so

mV0 π
∆(b) ' − √ , (52.17)
2~2 k a2 + b2
183 Chapter 52

so
Z r0 →∞
(+) πiV0
f (θ) = ik db bJ0 (kbθ) exp − √ −1
0Z ~2 k a2 + b2
∞ 2

i πimV0 θ/~
= 2 dz zJ0 (z) exp − √ −1 . (52.18)
kθ 0 a2 k 2 θ2 + z 2
−µr
5) For the Yukawa potential V (r) = V0 er , in the eikonal approximation, we
can use the formula for a finite range r0 ∼ 1/µ for the cross section,
Z r0
σtot = 8π db b sin2 ∆(b) , (52.19)
0

where
√
mV0 +∞ e−µ b +z
Z 2 2

∆(b) = − 2 dz √
2~ k −∞ b2 √
+ z2
mV0 +1/µ e−µ b +z
Z 2 2
mV0 2
'− 2 dz √ ∼− 2 , (52.20)
2~ k −1/µ 2
b +z 2 2~ k µb
so that
2 Z ~2 k
Z 1/µ
1 mV0
2 8π mV0 mV0 1
σtot ∼ 8π db b sin = 2 dx x sin2 , (52.21)
0 µb ~2 k µ ~2 k 0 x
where the integral is almost a number (if ~2 k/mV0 → ∞).
6) For a Coulomb potential V (r) = A/r, but with V = 0 for r ≥ r0 , in the
eikonal approximation
Z r0
eik
σtot ' 8π db b sin2 ∆(b) , (52.22)
0

and
√ √
Z + r02 −b2 + r02 −b2
mA dz mA 1 z
∆(b) = − 2 √ √ =− 2 arcsinh √
2~ k − r02 −b2
2
b +z 2 2~ k b b − r02 −b2
r
mA r02
=− arcsinh −1, (52.23)
~2 kb b2
and for r0 /b 1, the arcsinh becomes ' ln 2rb0 , so that
Z r0 Z r0
mA 2r0
eik
σtot = db b sin2 ∆(b) ' db b sin2 ln , (52.24)
0 0 ~2 kb b
and this is also divergent as r0 → ∞, as is the classical Rutherford formula.
7) We want to check if there are any resonance for Coulomb scattering, and the
analytical properties of Sl (k). We have
Sl (k) − 1
al (k) =
X 2i
f (θ) = (2l + 1)al (k)Pl (cos θ)
l
184 Chapter 52

Γ(iα) αe−iα ln(kr(1−cos θ))

= , (52.25)
Γ(−iα) k(1 − cos θ)
where
me20
α= . (52.26)
k~2
Note that e−iα ln(kr(1−cos θ)) = [kr(1 − cos θ)]−iα .
For cos θ = 0, we obtain l = 0, so then we can identify
S0 − 1 Γ(iα) −iα
= k , (52.27)
2i Γ(−iα)
Γ(iα)
where we rewrote e−iα ln k = k −iα . The ratio Γ(−iα) has poles at iα = −n, and
zeroes at iα = n. The poles correspond to
me20 i me20
= iαn = n ⇒ k n = i , (52.28)
~2 kn ~2 n
which are the correct values for the bound states.
But there are no resonances (poles or zeroes near the real line). We also note
that we have the correct analytical properties for
Γ(iα) −iα
k (52.29)
Γ(−iα)
as al (k).
53 Chapter 53

1) Mapping the black disk eikonal into the general δl (k), via δl (k) ↔ ∆(b)|b=l/k ,
we remember that

Sl (b < r0 ) = 0 (Im∆(b) = ∞) , Sl (b > r0 ) = 1 , (∆(b) = 0) , (53.1)

inel
so that σtot = πr02 . But for the mapping in the eikonal approximation, ∆(b > r0 ) =
0 becomes δl (k) = 0 for l > kr0 , and Im ∆(b < r0 ) = +∞ becomes Im δl (k) = +∞
for l < kr0 . Then
(
π
π 2 (2l + 1) , l < kr0
σlinel (k) = 2 (2l + 1)|e2iδl (k) − 1|2 = k . (53.2)
k 0, l > kr0

Then we have
kr0
inel π X π 2 2
σtot (k) = (2l + 1) = kr 0 (kr0 + 2) = πr0 1 + . (53.3)
k2 k2 kr0
l=0

2) We have
2
dσ ikr0 k 2 r02 2
= |f (θ, r0 )|2 = − J1 (−qr0 ) = J (−qr0 ). (53.4)
dΩ q q2 1

We note that as θ → 0, we have (since ~q = ~k − ~k 0 and q = 2k sin θ/2 ' kθ),

dσ k2 r04 2
dΩ → 4 . Integrating over dΩ = 2π sin θdθ = 4πd sin θ/2, we get
Z π Z 1
inel d sin2 θ/2 2 dy 2
σtot = πr02 2 J1 (−2kr0 sin θ/2) = 2πr0
2
J1 (−2kr0 y). (53.5)
0 sin θ/2 0 y

3) For a projectile A colliding with a Hydrogenoid atom in the Born approxima-

tion, exciting the electron to the first excited state,
2
dσ (1) k0 k 0 2mA ~ 0
= |f (1) |2 = hk , n|V̂ |~k, n0 i , (53.6)
dΩ k k 4π~2
2 me40
where (En = − Zn2 2~2 )
r
~2 k 02 ~2 k 2 Z 2 me40 3Z 2 me mA e40

1 0
= − 1− ⇒ k = k2 − , (53.7)
2mA 2mA 2~2 4 4~4
and
Ze20 e20
Z Z
r0
h~k 0 , n|V̂ |~k, n0 i = 3 0 −i~
d ~r e q ·~
d 3 ∗
~r1 ψn=2,l=0 (~r1 ) − 0 + 0 ψn=1,l=0 (~r1 )
r |~r − ~r1 |
185
186 Chapter 53

Ze20 4π
Z
1 3 ∗ −i~
q ·~
r1
= −δ 2,1 + d ~
r ψ
1 n=2,l=0 1 (~
r )e ψ (~
r
n=1,l=0 1 )
q2 Z
2 Z ∞ Z π −iqr1 cos θ
e 4π e ∗
= 0 r12 dr1 2π sin θdθ R2,0 (r)R1,0 (r)
q Z0 0 4π
2 ∞
e 4π ∗
= 02 r1 dr1 sin(qr1 )R2,0 (r1 )R1,0 (r1 ). (53.8)
q 0

Then the differential cross section is

2
r
dσ (1) 3 Z 2 me mA e40 2mA e20 ∞
Z
∗
= 1− r1 dr1 sin(qr1 )R 2,0 (r1 )R 1,0 (r1 ) . (53.9)
dΩ 4 k 2 ~4 ~2 q 2 0
4) For N = 4 fundamental particles, a, b, c, d, the channels are:
I : a + b + c + d , II : (ab) + c + d , III : (cd) + a + b ,
IV : (ac) + b + d , V : (bd) + a + c , V I : (ad) + b + c ,
V II : (bc) + a + d , V III : (ab) + (cd) , IX : (ac) + (bd) ,
X : (ad) + (bc) , XI : (abc) + d , XII : (abd) + c ,
XIII : (acd) + b , XIV : (bcd) + a. (53.10)
Consider the ”in” channel to be V III : (ab) + (cd), and the ”out” channel to be
XIV : (bcd) + a, then
Ein = ab + cd − Wab − Wcd
Eout = a + bcd − Wbcd , (53.11)
where Wab , Wcd , Wbcd are binding energies. Consider that the (bcd) is least bound,
so
Wbcd < Wacd < Wabd < Wabc , (53.12)
then
ab + cd − Wab − Wcd = 0a + 0bcd − Wbcd , (53.13)
and other cases are similar. Then the condition for the ”out” channel to be open is
ab + cd = Wab + Wbc − Wbcd + 0a + 0bcd > Wab + Wbc − Wbcd . (53.14)
5) For the target a fundamental particle, and the projectile composite, the initial
state is
|Eαi = |EA0 , α0 , ~nA i ⊗ |B i , (53.15)
and the final state is
|EA , α, ~n0A i ⊗ |B i. (53.16)
The Lippmann-Schwinger equation is
|E, α0 ; ~nA +i = |EA , α0 .~nA i ⊗ |B i + Ĝ+
0 (E)V̂ |E, α0 ; ~
nA +i , (53.17)
and we have, in the coordinate representation,
~ · |Eα+i = v + (~r, R)
h~r| ⊗ hR| ~
187 Chapter 53

~
X eik0 r
~ +
= eik·~r un0 (~r)gB (R) fnn0 (~k 0 , ~k)un (~r)gB (R)
~ (53.18)
,
n
r

where n are open channels. The differential cross-section is

dσnel0 dσninel k0
= |fn0 ,n0 |2 , = |fn,n0 |2 . (53.19)
dΩ dΩ k
6) For the ”in” case with two parallel projectiles, a + b (total momentum k), and
the ”out” case with a single general projectile, c (momentum k 0 ), we write
dσ inel k 0 mA
= |fα0 Γ0 ;αΓ |2 . (53.20)
dΩ k mA0
The channel Γ has a + b and target B, and the channel Γ0 has c and target B 0 ,
so the Lippmann-Schwinger equation is
|Eα; Γ±i = |Eα; Γi + Ĝ0 (E)V̂ Γ |Eα; Γ±i , (53.21)
where
0
fα0 ,Γ0 ;α,Γ ∝ TαΓ0 ,α
,Γ
(E) = hEα0 Γ0 |V̂Γ |EαΓ+i = hEα0 Γ0 − |V̂Γ0 |EαΓi. (53.22)
7) For electron scattering off a Helium atom, we have N = 2, and its 2 electrons
have (n1 , l1 , ml1 , m2 ), (n2 , l2 , ml2 , m2 ).
In the general formula
d exch
Tαα0 (E) = Tαα 0 (E) ± N Tαα0 (E) (53.23)
we have explicitly
d d

Tαα 0 (E) = tp
~0 ,n01 ,l10 ,m0 0 0 0 p,n ,l ,m ,n ,l ,m × δm01 m1 δm02 m2 + δm01 m2 δm02 m1 δm03 m3
l1 ,n2 ,l2 ,ml2 ;~ 1 1 l1 2 2 l2
exch exch
Tαα 0 (E) = tp
~0 ,n01 ,l10 ,m0 0 0 0 p,n ,l ,m ,n ,l ,m δm01 m1 δm02 m2 δm03 m3
l1 ,n2 ,l2 ,ml2 ;~ 1 1 l1 2 2 l2

+δm01 m2 δm0 2m3 δm03 m1 + δm01 m3 δm02 m2 δm03 m1 + δm01 m3 δm02 m1 δm03 m2 . (53.24)
Then we have
dσm01 m02 m03 ;m1 ,m2 ,m3
= (...)|td (δδ + ...) ± texch (δδδ + ...) |2 , (53.25)
dΩ
explicitly
dσ+++,+++ dσ−−−,−−−
= = |2td − 4texch |2
dΩ dΩ
dσ++−,++− dσ−−+,−−+
= = |2td |2
dΩ dΩ
dσ++−,+−+ dσ++−,−++ dσ−−+,−+− dσ−−+,+−−
= = = = |2texch |2
dΩ dΩ dΩ dΩ
dσ+−+,+−+ dσ−+−.−+− dσ−++,−++ dσ+−−,+−−
= = =
dΩ dΩ dΩ dΩ
dσ+−+,−++ dσ−+−,+−− dσ−++,+−+ dσ+−−,−++
= = = = = |td − texch |2
dΩ dΩ dΩ dΩ
dσ+−+,++− dσ−+−,−−+
= = |2texch |2 . (53.26)
dΩ dΩ
54 Chapter 54

1) Yes, the wave function (or something like it, the quantum field, in the relativistic
case) satisfies the Klein-Gordon equation. The (relativistic) Dirac equation is
∂
i~ ψ = i~cγ 0 γ i ∂i ψ + imc2 γ 0 ψ. (54.1)
∂t
Squaring it, we get the KG equation:
~ 2 ψ − (mc2 )2 ψ = 0.
[~2 c2 ∂µ2 − (mc2 )2 ]ψ = −~2 ∂t2 ψ + ~2 c2 ∇ (54.2)

But then: taking the square root to go back, we get ±mc2 : the rest energy con-
sidered with arbitrary sign (but negative energy states are ”occupied”, giving the
”Dirac sea”, according to Dirac’s interpretation).
2) The Dirac equation in 1+1 dimensions is formally the same in terms of the
arbitrary gamma matrices,
~ ∂
γ0 ψ + ~γ i ∂i ψ + mcψ = 0 , (54.3)
c ∂t
or
(γ µ ∂µ + m)ψ = 0 (54.4)

for ~ = c = 1, where
{γ µ , γ ν } = 2η µν . (54.5)

But in 1+1 dimensions we know such γ µ : 2 of the Pauli matrices give

{σ i , σ j } = 2δ ij , (54.6)

(all 3 of them satisfy σi σj = δij + iijk σk ), so it is the above condition, just in

Euclidean space (0+2 dimensions). To go to Minkowski space (1+1 dimensions),
we write, for instance

σ 1 = γ 1 , σ 2 = iγ 0 ⇒ {γ µ , γ ν } = 2η µν . (54.7)

3) We have

γ5 = iγ 0 γ 1 γ 2 γ 3 ⇒
γ52 = −γ 0 γ 1 γ 2 γ 3 γ 0 γ 1 γ 2 γ 3 = +γ 0 γ 1 γ 2 γ 0 γ 1 γ 2
= γ 0 γ 1 γ 0 γ 1 = −(γ 0 )2 = +1
{γ5 , γ3 } = i(γ 0 γ 1 γ 2 (γ 3 )2 + γ 3 γ 0 γ 1 γ 2 γ 3 )
= i(γ 0 γ 1 γ 2 − γ 0 γ 1 γ 2 ) = 0 , (54.8)
188
189 Chapter 54

where we have used the definition of the Clifford algebra to anticommute matrices
and square them, and similarly
{γ5 , γ2 } = {γ5 , γ1 } = {γ5 , γ0 } = 0 , (54.9)
so that finally we have the relations missing to form the 5-dimensional Clifford
algebra,
[{γ5 , γM } = 2η5M , M = (5, µ) , {γµ , γν } = 2ηµν ] ⇒ {γM , γN } = 2ηM N . (54.10)
4) To couple the KG equation to electromagnetism, we start with the Dirac
equation coupled to electromagnetism (the relativistic form),

~
γµ ∂µ − qAµ + mc ψ = 0 , (54.11)
i
and since the square of the Dirac equation gives the KG equation, we square the
Dirac equation (multiplied by γ 0 ) coupled to electromagnetism, to give

0 µ ~ 0 0 ν ~ 0
γ γ ∂µ − qAµ + mcγ γ γ ∂ν − qAν + mcγ ψ = 0 ⇒
" i i #
2
~ ~
∂µ ∂ µ + q 2 Aµ Aµ + (mc)2 − q 2∂ µ Aµ ψ = 0. (54.12)
i i

5) In the shell model for the nucleus, with the nucleon potential being Vn (r) =
−µr
−V0 e r , the relativistic corrections to the Hamiltonian for the nucleon are found
just like the corrections to the Hamiltonian for the Hydrogenoid atom, namely the
relativistic corrections to the Schrödinger equation are
p2 )2
(~ 1 ~ ~ 1 dVn ~2
Hn0 = − 3 2
+ L·S + ∆Vn , (54.13)
8m c 2m r dr 8m2 c2
−µr
applied for Vn = −V0 e r , so that
e−µr µ2

1 dVn V0 −µr 1 −µr 3
= 2e µ+ , ∆Vn = −V0 ∆ = −V0 e 4πδ (r) + .
r dr r r r r
(54.14)
6) We check the missing steps in the relativistic corrections to the Hydrogenoid
atom. We have
p2 )2
(~ 1
hH1 inljmj = h− p2 )2 inljmj ,
inljmj = − 3 2 h(~ (54.15)
8m3 c2 8m c
and
2 2
2 ~ ∂ 2 ∂ l(l + 1)
p~ = + − ,
i ∂r2 r ∂r r2
4 4 4
4m c (αZ) 1 3
p2 )2 inljmj =
h(~ 3
− , (54.16)
n l + 1/2 4n
and
1 Z2
h i nl = , (54.17)
r2 n3 a20 (l + 1/2)
190 Chapter 54

which indeed gives

(αZ)4

1 1 3
hH1 inljmj = − mc2 − . (54.18)
2 n3 l + 1/2 4n
We have
~2 ~ 2 ~ 2 2
(L ~ = J − L − S = ~ [j(j + 1) − l(l + 1) − s(s + 1)] ,
~ · S) (54.19)
2 2
and 0 for l = 0, so that
~2 Ze20
H2 = [j(j + 1) − l(l + 1) − s(s + 1)] , (54.20)
4m2 c2 r3
and 0 for l = 0, so
Ze20 ~2 1
hH2 inljmj = 2 2
[j(j + 1) − l(l + 1) − s(s + 1)]h 3 inl (54.21)
4m c r
and 0 for l = 0. To calculate h r13 i, we use the recursion relation
s+1 s s
hr i − (2s + 1)a0 hrs−1 i + [(2l + 1)2 − s2 ]a20 hrs−1 i = 0 , (54.22)
n2 4
applied for s = −1, giving
1 1 1
a0 h i − [(2l + 1)2 − 1]a20 h 3 i = 0 , (54.23)
r2 4 r
solved as
1 Z 1 Z 2Z 2 Z3
h i = h i = = ,
r3 a0 l(l + 1) r2 a0 l(l + 1) (2l + 1)n3 a20 a30 n3 l(l + 1)(l + 1/2)
(54.24)
which means
Ze20 ~2 Z 3 (me20 )3

l
hH2 inljmj =
4m c −(l + 1) ~ n l(l + 1)(l + 1/2)
2 2 6 3

(Zα)4 mc2

+1
= 3 , (54.25)
4n (l + 1)(l + 1/2) −(1 + 1/l)
which matches what we wanted to prove, at large l (+1/(−1 − 1/l) vs ±1).
7) For the relativistic corrections to the Hydrogenoid atom in magnetic field, we
want the corrections to
(~ ~ 2
p − q A) ~q
− ~ ,
~σ · B (54.26)
2m 2m
and in a constant magnetic field, A ~=B ~ × ~r/2, so the interaction Hamiltonian has
the usual linear term in B
q ~ ~ q ~ ~ q ~ ~ ~ ,
Hint = − B·L− B·S =− B · (L + 2S) (54.27)
2m 2m 2m
so the calculation follows the previous one (in the text), with the interaction given
by the Landé g-factor,
qB
∆E = −g mj . (54.28)
2m
191 Chapter 54

~ of the corrections
Then the first relativistic correction comes from coupling to A
in the text, which gives an extra contribution only in H1 , where we now have
* +
~ 2 ]2
p − q A)
[(~
− , (54.29)
8m3 c2

~= ~ r
B×~
out of which, from A 2 , and after the rearrangement, we get
~ · L)
(q B ~ 2 q 2 B 2 ~2 m2l
− 3 2
→− , (54.30)
8m c 8m3 c2
so the contribution to the energy is quadratic in B.
55 Chapter 55

1) Translational invariance and rotational symmetry constrain the 3-body potential

from QFT to be
V = V (|~rij |) = V (|~r1 − ~r2 |, |~r2 − ~r3 |, |~r3 − ~r1 |). (55.1)
2) For the LCAO approximation for an atom with 3 electrons, the expansion
restricted to 4 basis elements is, for instance
ψ(~r1 , ~r2 , ~r3 ) = [C1 (φ100 (~r1 )φ100 (~r2 )φ200 (~r3 ) + perms.)
+C2 (φ100 (~r1 )φ200 (~r2 )φ200 (~r3 ) + perms.)
+C3 (φ100 (~r1 )φ100 (~r2 )φ210 (~r3 ) + perms.)
+C4 (φ100 (~r1 )φ200 (~r2 )φ210 (~r3 ) + perms.)] , (55.2)
where
φa (~r) = φnlm (~r). (55.3)
3) The coefficients from the previous exercise can be rewritten as
C1 (t) = CE1 E1 E2 (t) , C2 (t) = CE1 E2 E2 (t) ,
C3 (t) = CE1 E1 E3 (t) , C4 (t) = CE1 E2 E3 (t) , (55.4)
their symmetry properties are (since the electrons are fermions, we get minus signs)
Ca1 a2 a3 (t) = −Ca2 a1 a3 (t) = −Ca3 a2 a1 (t) = −Ca1 a3 a2 (t) , (55.5)
and in the occupation number picture they are rewritten as
C1 (t) = CE1 E1 E2 (t) = C210... , C2 (t) = CE1 E2 E2 (t) = C120... ,
C3 (t) = CE1 E1 E3 (t) = C2010... , C4 (t) = CE1 E2 E3 (t) = C1110... . (55.6)
4) for 4 bosons with one-body potential V (r) that is approximately harmonic
oscillator, we can expand ψ in terms of harmonic oscillator states with n ≤ 3 as
ψ = C50... (ψE0 (~r1 )...ψE0 (~r5 ) + perms.)
+C410... (ψ
X E0 (~ r1 )...ψE0 (~r4 )ψE1 (~r5 ) + perms.) + ...
= Cn0 n1 n2 n3 0... (t) (ψ1 (~r5 )...ψ5 (~r5 ) + perms.) .(55.7)
n0 ,n1 ,n2 ,n3 ,n0 +...+n3 =5

5) For the case at the previous exercise, we write T (1) in the occupation number
basis as (N = 5 for 5 bosons)
5 r
X
0 n0 !...n3 ! 0
T(1) = hi|T̂(1) |i i Cn0 ...ni −1...ni0 +1...n3 (t) , (55.8)
0
5!
i,i =1

192
193 Chapter 55

where T(1) is diagonal in i, i0 basis (harmonic oscillator Hamiltonian), so

hi|T̂(1) |i0 i = Ei δii0 , (55.9)

therefore
5 r
X n0 !...n3 ! 0
T(1) = Ei Cn0 n1 n2 n3 (t). (55.10)
i=1
5!

6) For the same case, for the Coulomb 2-body potential

A
Vij = , (55.11)
rij
the explicit first 6 nontrivial terms in the 2-particle operator in the occupation
number basis are
r
X 1 A 0 0 n0 !...n3 ! 0
V(2) = ni (nj − δij )hij| |i j i Cn0 ...ni −1...ni0 +1...nj −1...nj0 +1...n3 0... (t)
2 rij 5!
i,j,i0 ,j 0
r
1 1 (n0 − 1)!(n1 − 1)!(n2 + 1)!(n3 + 1)! 0
= n0 n1 h01| |23i Cn0 −1n1 −1n2 +1n3 +1 (t)
2 r01 r 5!
1 1 (n0 + 1)!(n1 + 1)!(n2 − 1)!(n3 − 1)! 0
+ n2 n3 h23| |01i Cn0 +1n1 +1n2 −1n3 −1 (t)
2 r23 r 5!
1 1 (n0 + 1)!(n1 − 1)!(n2 − 1)!(n3 + 1)! 0
+ n1 n2 h12| |03i Cn0 +1n1 −1n2 −1n3 +1 (t)
2 r12 r 5!
1 1 (n0 − 1)!(n1 + 1)!(n2 − 1)!(n3 + 1)! 0
+ n0 n2 h02| |13i Cn0 −1n1 +1n2 −1n3 +1 (t)
2 r02 r 5!
1 1 (n0 − 1)!(n1 + 1)!(n2 + 1)!(n3 − 1)! 0
+ n0 n3 h03| |12i Cn0 −1n1 +1n2 +1n3 −1 (t)
2 r03 r 5!
1 1 (n0 + 1)!(n1 − 1)!(n2 + 1)!(n3 − 1)! 0
+ n1 n3 h13| |02i Cn0 +1n1 −1n2 +1n3 −1 (t).
2 r13 5!
(55.12)
7) For 3 fermions with one-body potential approximated by a harmonic oscillator,
the first 4 terms in T(1) , acting on the Slater determinant, we have

T(1) = hφEi |T̂i |φEi0 iCE1 ...EN (t) , (55.13)

where hφEi |T̂i |φEi0 i ∝ δEi Ei0 is diagonal, and leaves the Slater determinant invariant.
56 Chapter 56

1) To find the eigenvalues of eαb in single-particle states, namely

is also diagonalized, so
eαb |βi = eαβ |βi. (56.4)
2) The Hamiltonian

ti b†i bi + V b†i b†j bi+1 bj−1

X X
Ĥ = (56.5)
i i<j

is interpreted physically as follows: the first term is the kinetic energy Ĥ0 , which
has no interactions. Then b†i bi+1 means hopping from site i + 1 to site i, and b†j bj−1
hopping from j − 1 to j, so the interaction is between 2 hoppings.
3) For a system with 7 one-particle states and 3 bosons in it, the Fock space,

(b†1 )n1 (b†7 )n7

√ ... √ |0i , (56.6)
n1 ! n7 !
the numbers are unrestricted. However, since we want n1 + ... + n7 = 3 (we have
3 bosons only), that means there are 73 states (each boson can be in any of the 7
states), so 73 multi-particle states (of course, with identifications for when different
bosons are in the same state). P
4) To find the eigenvalue of the fermionic e( i ai )α operator in the Fock space,
we note first that
√
ai |...ni ...i = (−1)Si ni |...ni − 1...i , (56.7)
and on the other hand, for a fermionic state,
†
|...α...i = eai α |..., 0, ...i = (1 − αa†i )|.., 0, ...i , (56.8)
so we have
ai |...α...i = +α|...0...i = α|...α...i , (56.9)
194
195 Chapter 56

where we commuted the operator ai over (1 − αa† ), used a|αi = α|αi and α2 = 0
and then recomposed the state.
Finally then, we have
P X X
e( i ai )α |β1 ...βn i = (1 − α ai )|β1 ...βn i = (1 − α βi )|β1 ...βn i
P i i
= e( i βi )α
|β1 ...βn i. (56.10)
5) For the details of the fermionic Schrödinger equation on the occupation num-
ber states, we note that (56.14) in the text is the same for fermions, except that
|ni − 2i ≡ |0i, so no last term. But also (56.18) in the text follows, since there are
an even number of b’s and b† ’s. So (56.19) in the text is the same, hence (56.20) in
the text gives
X † 1 X † †
Ĥ = ai hi|T̂ |i0 iai + ai aj hij|V̂(2) |i0 j 0 iai0 aj 0 . (56.11)
0
2 0 0
i,i iji j

6) For the potential V (~ri − ~rj ) = V0 δ(~ri − ~rj ), the field operator is
Z Z Z
1
ψ̂σ† (~r)ψ̂σ† 0 (~r0 )V̂ (~r, ~r0 )ψ̂σ0 (~r0 )ψ̂σ (~r)
X X
†
Ĥ = d ~r3
ψ̂σ (~r)T̂ (~r)ψ̂σ (~r) + d ~r d3~r0
3

σ
2 0
Z Z σ,σ
V0
ψ̂σ (~r)ψ̂σ† 0 (~r)ψ̂σ0 (~r)ψ̂σ (~r)
X X
3 † 3 †
= d ~r ψ̂σ (~r)T̂ (~r)ψ̂σ (~r) + d ~r
σ
2 0
σ,σ
X † V0 † †
= ti ci ci + ai aj aj ai . (56.12)
i
2

7) The Coulomb occupation number Hamiltonian is

X ~2 k 2 e20 X X †
Ĥ = a~† a~kσ + a~ a† 0 ak0 ,σ0 a~k,σ , (56.13)
2m kσ V q ,σ ~
k+~ k −~q ,σ 0 ~
0
~
k,σ ~
k~ ~ σ,σ
k0 q

and the Feynman diagram is for an incoming fermion with momentum ~k 0 and σ 0
emitting a photon with momentum ~q that is absorbed by another fermion with
momentum ~k and σ, so the final state has fermions with momentum ~k 0 − ~q and σ 0
and one with momentum ~k + ~q and σ.
57 Chapter 57

1) For bosons instead of fermions, the set-up is as follows:

|δψi = δb†i1 b†i2 ...b†iN |0i + b†i1 δb†i2 ...b†iN |0i + ...b†i1 ...b†iN −1 δb†iN |0i , (57.3)
and we have
δb†i = −i Kji b†j .
X
(57.4)
j

If δb†i contains other b†j (so that Kji 6= 0 for i 6= j), then hδψ|ψi = 0, so we obtain

hδψ|Ĥ|ψi = 0 , (57.5)
but
hδψ|Ĥ(2) |ψi = h0|...bi ...bj ...|Ĥ(2) |...b†i0 ...b†j 0 ...|0i = hij|Ĥ(2) |i0 j 0 i. (57.6)
That means that terms with one index on the left equal to one index on the right
must vanish, and the equality to zero will give the Hartree-Fock equation. We have
the relevant terms in Ĥ(2)
1X 1X
hij|V̂(2) |ij 0 ib†i b†j bi bj 0 + hij|V̂(2) |i0 iib†i b†j bi0 bi
2 0 2 0
ijj iji
1X 0 † † 1X
+ hij|V̂(2) |jj ibi bj bj bj 0 + hij|V̂(2) |i0 jib†i b†j bi0 bj
2 0 2 0
ijj
X iji
= hik|V̂(2) |i ki + hik|V̂(2) |ki i b†i b†k bk bi0
0 0

i,k,i0

196
197 Chapter 57

!
b†i b†k bk hik|V̂(2) |i0 ki b†k bk hik|V̂(2) |ki0 i
X X X
= + bi0
i,i0 k k
" #
b†i
X X X
0 0
= Nk hik|V̂(2) |i ki + Nk hik|V̂(2) |ki i bi0
i,i0 k k
b†i hi|V̂H−F |i0 ibi0 .
X
≡ (57.7)
i,i0

Then the terms with one only one b†i changed must vanish, giving the equation

hi|T̂(1) |ji + hi|V̂H−F |ji = 0 , i 6= j ⇒

(T̂(1) + V̂H−F )|ii = Ei |ii , (57.8)

which is the bosonic equivalent of the Hartree-Fock equation.

3) Given the Hartree-Fock equation, we can rewrite the Hartree-Fock Hamilto-
nian as
X † X
ĤH−F = ai ai0 hi|T̂(1) |i0 i + hi|V̂H−F |i0 i = Ei a†i ai , (57.9)
i,i0 i

since the quantity in brackets is diagonal, so we get Ei hi|i0 i = Ei δii0 .

4) To calculate the Hartree potential for Helium in the ground state, we write
Z Z
(1)
X
VH = VH−F (~r) = d3~r0 |φk (~r0 )|2 V(2) (~r, ~r0 ) = d3~r0 ρ(~r0 )V(2) (~r, ~r0 ) , (57.10)
k,occ.

where
X e20
ρ(~r) = |φk (~r)|2 , V(2) (~r, ~r0 ) = , (57.11)
|~r − ~r0 |
k,occ.

and for He in the ground state there are 2 electrons in the 100 state, so
X Z e20 |φ100 (r0 )|2
Z
VH,He ground state = d3~r0 |φ100 (|~r0 |)|2 0
= 2e 2
0 d3~r0
|~r − ~r | |~r − ~r0 |
k=1,2
4r 0
e− a0
3 Z 1 Z ∞
4e20 2 02 0
= d cos θ r dr q
r a0 0 2 0
−1 0 1 + rr − 2 rr cos θ
3 Z ∞
r0 r0

2 2 0 0 − a0
4r 0
= 4e0 r dr e 1+ − 1− , (57.12)
a0 0 r r
where we have used that for the Helium atom (Hydrogenoid with Z = 2) we have
3/2 R +1 dx
φ100 (r) = a20 2e−2r/a0 √14π and did the integral over cos θ, using −1 √a−bx =
2
√ √
b( a + b − a − b).
At large r (r a0 ), we can take r > r0 in the integral (since the integral is cut
0
off by e−4r /a0 ), and obtain
3 Z ∞
4e20 2 4r 0 e2
VH ' r02 dr0 e− a0 = 0 , (57.13)
r a0 0 r
198 Chapter 57

which is the Coulomb potential of a single positive charge, like for the He nucleus
screened completely by the other electron. At intermediate r the result can be
written in terms of incomplete gamma functions, or otherwise be left as the integral
above.
5) For Lithium, Z = 3 and the electrons are in the configuration 1s2 , 2s1 . Then
then 2-particle exchange term in the Hartree-Fock approximation is
(2)
X φ∗ (~r0 )φk (~r) e2 1
VH−F (~r, ~r0 ) = e20 k
0
= 0 [2R10 (r0 )R10 (r) + R20 (r0 )R20 (r)]
|~r − ~r | 4π |~r − ~r0 |
k,occ.
3
e20 3r0

1 3 − a3 (r+r 0 ) 1 − 2a3 (r+r0 ) 3r
= 8e 0 + e 0 2 − 2 − ,
4π |~r − ~r0 | a0 8 a0 a0
(57.14)
where we have used
3/2 3/2
Z − Zr Z Zr Zr
R10 (r) = 2e a0
, R20 (r) = 2− e− 2a0 . (57.15)
a0 2a0 a0
6) The Hartree-Fock equation for the 2-point Green’s function is
(2) (2) (2)
G(2) ' G0 + G0 · VH−F · G(2) , (57.16)
so explicitly, in the (~r, σ) representation, is
(0)
GσσZ0 (~r, t; ~r0Z, t0 ) ' ZGσσ0 (~r, t; ~r0 , t0 )+
(0) (2)
X
d3~r00 dt00 d3~r000 Gσσ00 (~r, t; ~r00 , t00 )VH−F (~r00 , ~r000 )Gσ00 σ0 (~r000 , t00 ; ~r0 , t0 ).
σ 00
(57.17)
7) The Dyson, or version of the Bethe-Salpeter equation, is
(2) (2) (4) (2) (2)
G(4) = G0,1 · G0,2 + KB−S · G0,1 G0,2 · G(4) , (57.18)
and its explicit form in the the (~r, σ) representation, is
G(4) r1Z, t1 , ~r2 , Zt2 ; ~r3 , t3Z, ~r4 , t4Z) = GZ(0)
σ1 σ2 ,σ3 σ4 (~ r1 ,Zt1 ; r~3 , t3Z)G(0)
σ1 σ3 (~ σ2 σZ
4
(~r2 , t2 ; ~r4 , t4 )
X
+ d3~r10 d3~r20 dt01 dt02 d3~r100 d3~r200 dt001 dt002 ×
σ10 ,σ20 ,σ100 ,σ200
(4)
×KB−S σ1 σ2 ,σ0 σ0 (~r1 , t1 , ~r2 , t2 ; ~r10 , t01 ; ~r20 , t02 )Gσ10 σ100 (~r10 , t01 ; ~r100 , t001 )×
1 2
×Gσ20 σ”2 (~r20 , t02 ; ~r200 , t002 )Gσ100 σ200 ,σ3 σ4 (~r100 , t001 , ~r200 , t002 ; ~r3 , t3 , ~r4 , t4 ). (57.19)
58 Chapter 58

1) For an example of the fact that the wave function changes for a non-Abelian
Berry phase, yet it is consistent, consider the following: If the states are fully de-
generate, there is no observable way to discern between them, so changing from one
(ab)
to the other by Peiγ is OK.
2) For the steps leading to (58.12), starting with the Hamiltonian
2
X p~i − q A(~~ ri ) X X
H= + V(~ri − ~rj ) + qA0 (~ri ) , (58.1)
i
2m i<j i

with the (canonical) transformation

θ
Y
U= e−i π α(~ri −~rj ) , (58.2)
i<j

~ i,
we note that, since p~i = ~i ∇

U −1 p~i − q A(~ ~ ri ) + U −1 ~ ∇
~ ri ) U = p~i − q A(~ ~ iU
i
~ ri ) − ~θ
X
= p~i − q A(~ ~ i α(~ri − ~rj ) ,
∇ (58.3)
π i<j

and in 2 spatial dimensions, since arg(~x) = arctan y/x, we have ((x1 , x2 ) ≡ (x, y))
xj
∂i arg(~x) = −ij ⇒
|~x|2
(~ri − ~rj )b ẑ × (~ri − ~rj )
∇ai α(~ri − ~rj ) = azb = , (58.4)
|~ri − ~rj | |~ri − ~rj |2
where ẑ is the unit vector perpendicular to the 2 dimensional plane.
Then we finally have
X ẑ × (~ri − ~rj )
U −1 p~i − q A(~ ~ ri ) − ~θ
~ ri ) U = p~i − q A(~ . (58.5)
π i<j |~ri − ~rj |2

q.e.d. (58.12) in the text.

3) From
2π
Fµν = µνρ J ρ , (58.6)
k
taking the (01) component gives
2π 2π
E 1 = −F01 = − 012 J 2 ⇒ E 1 = − J 2 . (58.7)
k k
199
200 Chapter 58

~ so what we obtain is
But, in general, we have the matrix relation ~j = σ · E,
(considering also what happens when we exchange the 1 and 2 indices)
k ij
σ ij = , (58.8)
2π
in other words, the integer quantum Hall effect (IQHE).
4) The Chern-Simons action is, in the Lorentz invariant form,
Z
k
SCS = d3 x ijk Ai ∂j Ak . (58.9)
B=∂M 4π
But by the Stokes theorem for general form integration (noting that the integrand
is a 3-form), this is equal to
Z
k
SCS = d4 x µνρσ (∂µ Aν )(∂ρ Aσ )
M Z 4π
k
= d4 xµνρσ Fµν Fρσ , (58.10)
16π M
where we have used that µνρσ ∂µ ∂ρ Aσ = 0, when we acted with the extra (anti-
symmetric) derivative on the CS integrand.
But note that the action is gauge invariant, since it is written only in terms of the
field strength Fµν , and it is topological, since there was no metric needed in order
√
to write it (not even the usual invariant integration measure d4 x −g), we used
R

the Levi-Civitta symbol µνρσ instead. That means that the integral is invariant
under small perturbations of the metric, so it is proportional to an integer, i.e., it
is topological.
5) We didn’t consider the equation of motion for Aµ when considering the CS
action or anyons because Aµ is an external source, so appears only linearly in the
full action. Then, when integrating out aµ , we obtain the CS action in (58.43) in
the text,
Z
0 1 1
Seff = d3 xijk Ai ∂j Ak , (58.11)
r 4π
which is quadratic, but contains no propagating degrees of freedom. In fact, it is
a response action, meaning that it defines the response of the system with respect
to the external source. That means that we don’t consider its equation of motion,
since Aµ lacks the dynamical kinetic term.
6) For FQHE we considered Fµν ∝ δ 2 (~r) unphysical because of flux conservation,
but then fµν ∝ δ 2 (~r) is OK, since the statistical gauge field is only defined in 2+1
dimensions. There is no physical 4th direction for it, like there is for Fµν .
7) The Moore-Read wave function is
Y
ψMoore−Read ∝ (zi − zj )m , (58.12)
i<j

and it implies that there are m vortices at each position zi .

One vortex has a wave function with an ansatz f (r)eiα , where α is the polar
angle in the plane, and then, for consistency (so that the arbitrary phase eiα is
201 Chapter 58

irrelevant there), we must have f (r = 0) = 0. Having m vortices at the position

r = 0 then corresponds to f˜(r)eimα , so that there are m units of winding of the
solution around the physical circle. In order to have the vortex at position ~r0 , we
shift r → |~r −~r0 |, and eiα refers to the polar angle around r0 , so ~r −~r0 = |~r −~r0 |eiα .
Then, indeed, in complex coordinates we have z = reiα and z − zi = |z − zi |eiα ,
and at z = zi we have |z − zi | = 0, so the Moore-Read solution has vortices there.
But (zi − zj )m = |zi − zj |m eimα has m vortices at zj from the point of view of zi ,
and m vortices at zi from the point of view of zj .

Notes On Quantum Mechanics PDF
100% (1)
Notes On Quantum Mechanics PDF
317 pages
Calculus For Engineers, 4th Edition
100% (14)
Calculus For Engineers, 4th Edition
1,216 pages
Quantum PDF
No ratings yet
Quantum PDF
318 pages
(Tang C.) Solutions Manual. Fundamentals of Org
No ratings yet
(Tang C.) Solutions Manual. Fundamentals of Org
65 pages
Main 2122
No ratings yet
Main 2122
207 pages
Quantum Mechanics Course Guide
100% (3)
Quantum Mechanics Course Guide
297 pages
The Physics of Quantum Mechanics (Instructor Solution
No ratings yet
The Physics of Quantum Mechanics (Instructor Solution
101 pages
Solutions To Selected Problems: K K K K N J J J JK
100% (2)
Solutions To Selected Problems: K K K K N J J J JK
79 pages
Aqm-David Gross
No ratings yet
Aqm-David Gross
128 pages
Physics For Scientists & Engineers
100% (20)
Physics For Scientists & Engineers
1,298 pages
Problem Sets All
100% (1)
Problem Sets All
14 pages
Skills in Mathematics Trigonometry For JEE Main and Advanced 2022
100% (6)
Skills in Mathematics Trigonometry For JEE Main and Advanced 2022
399 pages
QM2
100% (1)
QM2
317 pages
Mathematical Methods and The Born-Oppenheimer Approximation
No ratings yet
Mathematical Methods and The Born-Oppenheimer Approximation
22 pages
Wave and Oscillation PDF PDF
100% (10)
Wave and Oscillation PDF PDF
417 pages
Solutions Manual of Quantum Mechanics by Auletta & Fortunato - 1st Edition PDF
No ratings yet
Solutions Manual of Quantum Mechanics by Auletta & Fortunato - 1st Edition PDF
5 pages
Approximation Methods: Physics 130B, UCSD Fall 2009
No ratings yet
Approximation Methods: Physics 130B, UCSD Fall 2009
90 pages
An Introduction To Linear Algebra For Science and Engineering - 3rd Ed - Norman
100% (17)
An Introduction To Linear Algebra For Science and Engineering - 3rd Ed - Norman
592 pages
G125057 Parts Catalog
100% (2)
G125057 Parts Catalog
135 pages
Indefinite Definite Integration XII
No ratings yet
Indefinite Definite Integration XII
30 pages
Final Report Smart Streetlighting
No ratings yet
Final Report Smart Streetlighting
22 pages
Quantum Mechanics for Physicists
No ratings yet
Quantum Mechanics for Physicists
58 pages
Skills in Mathematics Differential Calculus For JEE Main and Advanced 2022
91% (11)
Skills in Mathematics Differential Calculus For JEE Main and Advanced 2022
561 pages
SM6 Im
No ratings yet
SM6 Im
6 pages
2012 May Solutions
No ratings yet
2012 May Solutions
6 pages
Part Rec Engine Saa6d170e-5 SN 611472 PC 1250 PT DH
No ratings yet
Part Rec Engine Saa6d170e-5 SN 611472 PC 1250 PT DH
28 pages
Introduction To Mathematical Proofs - A Transition To Advanced Mathematics (PDFDrive)
100% (12)
Introduction To Mathematical Proofs - A Transition To Advanced Mathematics (PDFDrive)
406 pages
Climate Related Risk Management
No ratings yet
Climate Related Risk Management
40 pages
Adv QM3
No ratings yet
Adv QM3
8 pages
Skills in Mathematics Integral Calculus For JEE Main and Advanced 2022
88% (8)
Skills in Mathematics Integral Calculus For JEE Main and Advanced 2022
319 pages
From Classical To Quantum Fields PDF
100% (11)
From Classical To Quantum Fields PDF
951 pages
Geometry
100% (10)
Geometry
370 pages
Multi-Variable Calculus A First Step PDF
100% (10)
Multi-Variable Calculus A First Step PDF
337 pages
Integral Calculus
100% (10)
Integral Calculus
493 pages
koopmanVonNeumann02 241031 223740
No ratings yet
koopmanVonNeumann02 241031 223740
11 pages
Week2 25
No ratings yet
Week2 25
8 pages
Tevian Dray - The Geometry of Special Relativity-A K Peters - CRC Press (2012)
100% (2)
Tevian Dray - The Geometry of Special Relativity-A K Peters - CRC Press (2012)
148 pages
B1.1+Boeing+737-600 700 800 900+ (CFM+56) +72+Engine+General
100% (1)
B1.1+Boeing+737-600 700 800 900+ (CFM+56) +72+Engine+General
73 pages
2020 - Book - SolvingProblemsInMathematicalA Part 1 PDF
100% (7)
2020 - Book - SolvingProblemsInMathematicalA Part 1 PDF
375 pages
Essential Calculus
100% (10)
Essential Calculus
156 pages
Differential Calculus
94% (17)
Differential Calculus
625 pages
Drag Reduction by Application of Aerodynamic Devices in A Race Car
No ratings yet
Drag Reduction by Application of Aerodynamic Devices in A Race Car
20 pages
Introduction To Geometry
92% (24)
Introduction To Geometry
580 pages
50 Challenging Calculus Problems (Fully Solved) - Chris McMullen
100% (16)
50 Challenging Calculus Problems (Fully Solved) - Chris McMullen
236 pages
Homework Assignment 8: Solutions
No ratings yet
Homework Assignment 8: Solutions
7 pages
Advanced Quantum Mechanics: Problem Sheet 1
No ratings yet
Advanced Quantum Mechanics: Problem Sheet 1
9 pages
Ghosh: CHM 112M: Lecture 3
No ratings yet
Ghosh: CHM 112M: Lecture 3
6 pages
A
No ratings yet
A
73 pages
(Industrial and Applied Mathematics) Martin Brokate, Pammy Manchanda, Abul Hasan Siddiqi - Calculus For Scientists and Engineers (2019, Springer)
94% (16)
(Industrial and Applied Mathematics) Martin Brokate, Pammy Manchanda, Abul Hasan Siddiqi - Calculus For Scientists and Engineers (2019, Springer)
655 pages
Chapt 5
100% (1)
Chapt 5
15 pages
Quantum Mechanics Homework
No ratings yet
Quantum Mechanics Homework
8 pages
QM16 SHOQuestions
No ratings yet
QM16 SHOQuestions
10 pages
Optical Theorem
No ratings yet
Optical Theorem
5 pages
Introduction To Linear Algebra For Science and Engineering 1st Ed
90% (58)
Introduction To Linear Algebra For Science and Engineering 1st Ed
550 pages
Transformers CRO
No ratings yet
Transformers CRO
11 pages
Quantum Coherent States Guide
No ratings yet
Quantum Coherent States Guide
9 pages
PHAS0022 PS 2020 2 Solutions
No ratings yet
PHAS0022 PS 2020 2 Solutions
8 pages
WELLCARE
No ratings yet
WELLCARE
9 pages
Vardhaman Mahaveer Open University, Kota: Differential Geometry and Tensors-1
100% (1)
Vardhaman Mahaveer Open University, Kota: Differential Geometry and Tensors-1
358 pages
Installation - OM Manual IDT (Transformadores de 8.25)
No ratings yet
Installation - OM Manual IDT (Transformadores de 8.25)
106 pages
Midterm Solution 2018
No ratings yet
Midterm Solution 2018
11 pages
Diff Erential Geometry of Curves and Surfaces: Kristopher Tapp
100% (12)
Diff Erential Geometry of Curves and Surfaces: Kristopher Tapp
370 pages
Quantum Simple Harmonic Oscillator
No ratings yet
Quantum Simple Harmonic Oscillator
10 pages
Eg 1
No ratings yet
Eg 1
5 pages
Mathematical Physics
89% (19)
Mathematical Physics
32 pages
Commutators and Quantum Compatibility
No ratings yet
Commutators and Quantum Compatibility
25 pages
QM1 Problem Set 1 Solutions - Mike Saelim
No ratings yet
QM1 Problem Set 1 Solutions - Mike Saelim
5 pages
591 Homework v1 - 3
No ratings yet
591 Homework v1 - 3
45 pages
A Course in Modern Mathematical Physics
95% (43)
A Course in Modern Mathematical Physics
618 pages
Essential Calculus Skills Practice Workbook With Full Solutions
95% (85)
Essential Calculus Skills Practice Workbook With Full Solutions
528 pages
Quantum Mechanics Basics & Equations
No ratings yet
Quantum Mechanics Basics & Equations
6 pages
Quantum Mechanics Observables Solutions
No ratings yet
Quantum Mechanics Observables Solutions
11 pages
Wire Size Selection For House Wiring
No ratings yet
Wire Size Selection For House Wiring
24 pages
Pneumatic Metal Cutting Machine
No ratings yet
Pneumatic Metal Cutting Machine
14 pages
Calculus Better Explained PDF
93% (14)
Calculus Better Explained PDF
76 pages
Civil Engineering Lab Guide
No ratings yet
Civil Engineering Lab Guide
85 pages
Quantum Mechanics I Solutions 6.: Hint
No ratings yet
Quantum Mechanics I Solutions 6.: Hint
12 pages
Finance and Accounting - ALP
No ratings yet
Finance and Accounting - ALP
8 pages
Automatic Controls For Industrial Refrigeration Systems
No ratings yet
Automatic Controls For Industrial Refrigeration Systems
1 page
Elecrical Lesson 1 - Earthing Systems 2
No ratings yet
Elecrical Lesson 1 - Earthing Systems 2
7 pages
Log DWG @1!7!2024
No ratings yet
Log DWG @1!7!2024
2 pages
OU Open University SM358 2012 Exam Solutions
No ratings yet
OU Open University SM358 2012 Exam Solutions
20 pages
A Course in Mechanics by Dr. J. Tinsley Oden Part II - Homework 3 - Solutions
No ratings yet
A Course in Mechanics by Dr. J. Tinsley Oden Part II - Homework 3 - Solutions
7 pages
E0.02 ELECTRICAL SCHEDULES Rev.0
No ratings yet
E0.02 ELECTRICAL SCHEDULES Rev.0
1 page
F 11 Place 2
No ratings yet
F 11 Place 2
8 pages
Top View: Pemerintah Kabupaten Nagekeo Dinas Perhubungan
No ratings yet
Top View: Pemerintah Kabupaten Nagekeo Dinas Perhubungan
1 page
Project Title: Automatic Dam Gate Control System With Caution Alarm
No ratings yet
Project Title: Automatic Dam Gate Control System With Caution Alarm
2 pages
3000 Solved Problems in Physics
95% (74)
3000 Solved Problems in Physics
782 pages
Resume Fahmi Shalas
No ratings yet
Resume Fahmi Shalas
5 pages
235 Divani Kebirden Sechmeler Qazallar Rubailer I II III IV (Shefiq Can) (Otuken Yayinlari 2000)
No ratings yet
235 Divani Kebirden Sechmeler Qazallar Rubailer I II III IV (Shefiq Can) (Otuken Yayinlari 2000)
541 pages
e Bill
No ratings yet
e Bill
1 page
1001 Algebra Problems
96% (71)
1001 Algebra Problems
292 pages
GW HT 100 136k User Manual en
No ratings yet
GW HT 100 136k User Manual en
19 pages
Data Sheet 160-3-1500-15 Chasquipampa
No ratings yet
Data Sheet 160-3-1500-15 Chasquipampa
1 page
PTP 2
No ratings yet
PTP 2
7 pages
KDS-Product Sheet Venturi-V4
No ratings yet
KDS-Product Sheet Venturi-V4
2 pages
Valve Purging System Guide
No ratings yet
Valve Purging System Guide
5 pages
Academic Profile: Iván García-Kerdan
No ratings yet
Academic Profile: Iván García-Kerdan
5 pages
SKM Egg Products Exp India Ltd-532143-March-2004
No ratings yet
SKM Egg Products Exp India Ltd-532143-March-2004
44 pages
Trigonometry
No ratings yet
Trigonometry
442 pages

Solutions

Uploaded by

Solutions

Uploaded by

Quantum Mechanics, a graduate course

1) Doing the integral over Bν in the formula for P , we find

Defining x = hν/(kB T ), we find

4) For a relativistic wave, ω = 2πν and k = 2π/λ = 2πν/c. If the two-slit

1) The triangle inequality for ~c = ~a + ~b is ~c2 ≤ ~a2 + ~b2 . Generalizing to vectors

2) Considering ~a · ~b 6= 0, ~a · ~c 6= 0, ~b · ~c 6= 0, we first normalize ~a,

Then we orthogonalize ~c to ~a0 ,

4) Consider Hermitian matrices A = A† , B = B † . Then

1) We discretize the product function (f · g)(x) as (f · g)(xi ). In a discrete and

If we defined (f · g)(xi ) ≡ hxi |f · gi, then we have

We can compare this with

If we understand the product function as product of the functions at the point,

||F ||2 = hF |F i = hf |f ihg|gi. (2.4)

3) The delta function is the distribution

δx0 : D → R , hδx0 |φi = φ(x0 ). (2.6)

If we try to define a distribution δx20 : D → R, there is no way to define it to be

hδx20 |φi = φ(x0 ) OR φ(x0 )δx0 (x0 ) = ∞ , (2.7)

5) The function of an operator is defined from the Taylor expansion, as

eiÂ = cos Â + i sin Â , (2.12)

7) To find the eigenvalues of σ1 , we write the secular equation,

symmetrized form (or ”Weyl ordered”) of the last term is

The average value of the observable Â is

1) It is easy to see that the simplest possibility is of the type

{γi , γj } = {σi , σj } ⊗ σa2 = 2δij ⊗ 1l = δij 1l4

As an example, we can choose σa = σ1 and σb = σ2 .

4) From (4.52) in the text, the probability of oscillation is

1 − cos[(E2 − E1 )t/~] (E2 − E1 )t

E22 − E12 m2 − m21 m2 − m21 ∆m2

Then the probability increases monotonically,

1) As was noted in the text,

which is clearly nonlocal, as it depends on two points, x and x + a.

which is similar to the case of H1 , so it is clearly nonlocal as well.

F (~r, p~, t) = ~r · p~ + i Li (~r, p~). (5.5)

δpi = j {ri , Lj }P.B. = j jkl {ri , rk pl }P.B. = j jkl rk δil = ijk j rk

3) We can rewrite the Lagrangian as

{q, p}P.B. = 1 → [q, p] = i~ (5.12)

and [q, q] = [p, p] = 0.

If only the total energy E is fixed, we have a continuous degeneracy: Ei are

5) For the Hamiltonian

so if we defined Ẽ = E + mα2 /2, and define ~i ψ(x) = pψ(x), we have

1 We use the momentum space wave function

and finally the Heisenberg uncertainty relation is

so that the width of the Gaussian is

but in general we have the Heisenberg uncertainty relations

For simplicity we take the origin at 0 (x0 = 0), and find

so that the width in x squared is

For the momentum variable,

and the momentum width squared is

Then we see that indeed

satisfying Heisenberg’s uncertainty relations.

1) In case I, for energies −V0 ≤ E ≤ 0, we have a discrete spectrum, since there

and integrating over an infinitesimal domain around x = 0, we obtain

ψL (x) = Aekx , x < 0 , ψR (x) = Be−kx , x > 0 , (7.3)

4) The wave function for particle in a box is

where we have defined

In the classical limit, n → ∞ and k finite, we obtain

(k2 + κ3 )(k1 − k2 ) + e2ik2 (b−a) (k2 − κ3 )(k1 + k2 )

ei(k1 −k2 )a (k2 + κ3 )2k1

and then at x = 0 gives, substituting the value of R and of B/A,

Note that for k1 =  (meaning, U1 = 0), we get

and then at x = 0 gives, substituting the values of R and D/C,

Note that for k1 =  (U1 = 0), we obtain

1) For small q1 , q2 , the Lagrangian becomes

consider the Fourier transform of the creation and annihilation operators

On the other hand,

3) To prove the completeness relation

6) For the potential V (x) = λx4 , the Schrödinger equation is

We next consider the asymptotics of the equation:

and b) For y → 0, we obtain ψn00 (y) + n ψn (y) = 0, with solution

but since we are at y → 0, it is actually

Then the ansatz that eliminates the asymptotic behaviour is

which leads to the reduced equation

leading to the equation

We can put each coefficient to zero, leading to the recursion relations

and the differential equation is

dX̂H ∂ P̂S P̂H

F (~r, p~, t) = ~r · p~ + i Li (~r, p~). (5.5)

δpi = j {ri , Lj }P.B. = j jkl {ri , rk pl }P.B. = j jkl rk δil = ijk j rk

Note that for k1 = (meaning, U1 = 0), we get

Note that for k1 = (U1 = 0), we obtain

and b) For y → 0, we obtain ψn00 (y) + n ψn (y) = 0, with solution

hp(ti )|e−iĤ |x(ti−1 )i ' e−iĤ(p(ti ),x(ti−1 )) e−ip(ti )x(ti−1 ) , (10.1)

[Xa , Xb ] = abc Xc , (13.6)

where Xa = (A, B, C), so fabc = abc . Then the adjoint representation is