Sie sind auf Seite 1von 159

Quantum Mechanics

Lecture Notes
Tom Kirchner
1
unter Mitarbeit von Dennis Rohrbein
Institut f ur Theoretische Physik
TU Clausthal
1
tom.kirchner@tu-clausthal.de
ii
Vorbemerkung
Ohne es selbst untersucht zu haben oder entsprechende Untersuchungen zu
kennen, wage ich die Behauptung, dass Vorworte die am gr undlichsten gelese-
nen Seiten von Skripten, Monographien und B uchern sind. Im vorliegenden Fall,
das wei ich ganz sicher, ist der einzig akzeptable Grund daf ur, dass nur hier
einige zusammenhangende (deutsche) Satze zu nden sind. Das Skript selbst
besteht weitgehend aus Formeln und (englischen) Stichpunkten. Das reektiert
meine

Uberzeugung, dass man Quantenmechanik nicht versuchen sollte zu lernen,
ohne eines oder mehrere der reichlich vorhandenen hervorragenden Lehrb ucher
zu Rate zu ziehen. Dieses Skript soll nicht mit diesen (teilweise recht umfangre-
ichen) B uchern konkurrieren, sondern lediglich eine Orientierung liefern, was die
Kernaussagen der Quantenmechanik sind, die in einer einsemestrigen Vorlesung
typischerweise behandelt werden. Ich selbst habe eine entsprechende Vorlesung
an der TU Clausthal in den Wintersemestern 03/04, 04/05 und 06/07 gehalten.
Ich hoe, das Skript wird als Leitfaden zur Nacharbeit und Pr ufungsvorberei-
tung n utzlich sein. Hinweise auf Fehler nehme ich wie stets gerne entgegen.
Tom Kirchner
Clausthal-Zellerfeld im Oktober 2007
Contents
1 Introduction 1
2 Wave Mechanics 5
2.1 Matter waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Classical wave functions . . . . . . . . . . . . . . . . . . . 5
2.1.2 Transition to Matter Waves (MWs) . . . . . . . . . . . . . 6
2.1.3 Wave Packets (WPs) . . . . . . . . . . . . . . . . . . . . . 7
2.2 Schrodinger equation . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 Schrodinger equation for a free particle . . . . . . . . . . . 13
2.2.2 SE for a particle in a potential (conservative system) . . . 15
2.2.3 SE for a particle in the classical EM eld . . . . . . . . . . 16
2.2.4 SE for N particles . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Statistical interpretation . . . . . . . . . . . . . . . . . . . . . . . 18
2.3.1 Problems with the concept of matter waves and the solution 18
2.3.2 Probabilities, norm, continuity . . . . . . . . . . . . . . . . 19
2.3.3 Probability densities in coordinate and momentum space . 20
2.3.4 Expectation values . . . . . . . . . . . . . . . . . . . . . . 21
2.4 Solution of the Schrodinger equation quantum eects . . . . . 25
2.4.1 Stationary (time-independent) SE . . . . . . . . . . . . . . 25
2.4.2 One-dimensional model systems . . . . . . . . . . . . . . . 26
3 Formalism of Quantum Mechanics 41
3.1 Mathematical framework . . . . . . . . . . . . . . . . . . . . . . . 41
3.1.1 (State-) vectors and their linear space . . . . . . . . . . . . 41
3.1.2 (linear) Operators . . . . . . . . . . . . . . . . . . . . . . . 45
3.1.3 Representation of vectors and operators . . . . . . . . . . 48
3.1.4 Non-orthogonal basis sets . . . . . . . . . . . . . . . . . . 51
3.1.5 Eigenvalue problem . . . . . . . . . . . . . . . . . . . . . . 52
3.1.6 Continuous basis sets and spectra . . . . . . . . . . . . . . 55
3.2 From wave mechanics to QM . . . . . . . . . . . . . . . . . . . . . 60
3.2.1 State . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.2.2 Eigenvalue problems etc. . . . . . . . . . . . . . . . . . . . 61
3.2.3 Commutators . . . . . . . . . . . . . . . . . . . . . . . . . 62
iii
iv
3.2.4 Stationary Schrodinger equation . . . . . . . . . . . . . . . 63
3.3 The postulates of Quantum Mechanics . . . . . . . . . . . . . . . 65
3.3.1 States . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.3.2 Observables . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.3.3 Expectation values . . . . . . . . . . . . . . . . . . . . . . 65
3.3.4 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.3.5 Commutation relations . . . . . . . . . . . . . . . . . . . . 68
3.3.6 Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.4 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . 73
3.4.1 Time-development (evolution) operator . . . . . . . . . . . 74
3.4.2 Heisenberg picture . . . . . . . . . . . . . . . . . . . . . . 75
3.4.3 Interaction picture . . . . . . . . . . . . . . . . . . . . . . 80
3.5 Symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.5.1 Representation of symmetry transformations . . . . . . . . 82
3.5.2 Symmetries and constants of motion . . . . . . . . . . . . 84
3.5.3 Translations (1-d) . . . . . . . . . . . . . . . . . . . . . . . 85
3.5.4 Other continuous symmetry transformations . . . . . . . . 87
3.5.5 Discrete symmetry transformations . . . . . . . . . . . . . 89
4 Applications of the Theory 97
4.1 The harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . 97
4.1.1 Algebraic solution of the one-dimensional harmonic oscillator 97
4.1.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.2 Approximation methods . . . . . . . . . . . . . . . . . . . . . . . 103
4.2.1 Stationary perturbation theory . . . . . . . . . . . . . . . 103
4.2.2 Variational methods . . . . . . . . . . . . . . . . . . . . . 107
4.2.3 Time-dependent perturbation theory . . . . . . . . . . . . 109
4.3 Angular momentum and spin . . . . . . . . . . . . . . . . . . . . 115
4.3.1 Angular momentum algebra . . . . . . . . . . . . . . . . . 116
4.3.2 Orbital angular momentum . . . . . . . . . . . . . . . . . 120
4.3.3 Spin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
4.3.4 Addition of angular momenta . . . . . . . . . . . . . . . . 127
4.4 The hydrogen atom (and extensions) . . . . . . . . . . . . . . . . 128
4.4.1 Separation of the two-body problem . . . . . . . . . . . . . 128
4.4.2 The QM central-eld problem . . . . . . . . . . . . . . . . 129
4.4.3 Solution of the Coulomb problem (E < 0) . . . . . . . . . 131
4.4.4 Assorted remarks . . . . . . . . . . . . . . . . . . . . . . . 135
5 Brief introduction into relativistic Quantum Mechanics 139
5.1 Klein-Gordon equation . . . . . . . . . . . . . . . . . . . . . . . . 139
5.1.1 Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.1.2 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
5.2 Dirac equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
v
5.2.1 Construction of the free DE . . . . . . . . . . . . . . . . . 140
5.2.2 Solutions of the free DE . . . . . . . . . . . . . . . . . . . 142
5.2.3 Electromagnetic potentials . . . . . . . . . . . . . . . . . . 144
5.2.4 Relativistic hydrogen atom . . . . . . . . . . . . . . . . . . 144
A Some details 147
A.1 Details on the splitting at the potential well . . . . . . . . . . . . 147
A.2 Remarks on photoabsorption . . . . . . . . . . . . . . . . . . . . . 149
vi
Chapter 1
Introduction
Classical (theoretical) physics consists of two pillars
Matter Radiation
particles (trajectories) waves (elds)
Newton Maxwell
classical mechanics classical electrodynamics
+ statistical mechanics
( classical thermodynamics)
Both pillars are relatively independent (although wave phenomena exist also for
material objects, e.g. water waves)
Experiments (1900) bring up problems:
1. Quantal behavior of radiation
cannot be explained by classical electrodynamics
Black-body radiation
Planck

s hypothesis (1900) : E
n
= nh , n = 1, 2, ...
h = 6, 626 10
34
Js
= 4, 1357 10
15
eV s
i.e. emission and absorption of energy is quantized
Plancks law agrees with experiments
1
2 Introduction
Photoelectric eect (rst experiments in 1887)
e
EM radiation
detector
metal
Observations:
- number of emitted electrons
intensity of light
- kinetic energy of electrons
frequency of light,
but independent of intensity!
Einsteins interpretation (1905):
Radiation consists of photons which carry energy
E = h
explains observation!
Compton eect (1922)
A

:H=O
=
>

observation :

()
can be understood as inelastic collisions of particles
energy of photon E = h = pc
momentum of photon p =
h
c
=
h

2. But: in other situations radiation behaves like waves do (e.g. interference


in Youngs two-slit experiment, which in 1801/1802 was very important to
establish the wave picture of radiation!)
dualism particle-wave
3. Structure and stability of atoms (for whose existence there was ample
evidence) cannot be explained by classical mechanics
Rutherfords scattering experiment (1911) lead to planetary model
of an atom:
A

radius of

nucleus

R
n
5 10
13
cm
radius of atom
R
a
10
8
cm
3
but: according to classical electrodynamics these atoms cannot be sta-
ble as accelerated charges radiate and lose (continuously) energy; life-
time of Rutherford atom 10
11
s!
Moreover: line spectra of elements cannot be explained, but only de-
scribed empirically
(e.g. for H: Balmer/Rydberg formula:
= R
_
1
m
2

1
n
2
_
, n, m N , n > m)
theoretical basis completely unclear!
Bohrs model (1913):
classical mechanics + postulates
postulate (1): existence of stable, quantized orbits i.e. he
assumes that classical electrodynamics is not in
operation on these orbits
postulate (2): emission + absorbtion of radiation transition
between two orbits h = E
n
E
m
postulates + classical mechanics (and Coulombs law) yield Balmer/Rydberg
formula!
but: problems for other atoms (He) remain and ad-hoc assumptions
are not well founded
De Broglies matter waves (1923/24):
Suggestion: one can ascribe wave length and frequency to a particle
according to Plancks/Einsteins formulae:
p = mv =
h

E = h
Combination of de Broglie relations with Bohrs 1
st
postulate yields an
interpretation because one nds that Bohrs postulate of quantized
orbits (levels) corresponds to boundary conditions of a standing wave
question: can one observe typical wave phenomena with
particles?
Interference phenomena of electrons in diraction experiments (Davis-
son + Germer, 1927)
= wave-particle dualism for radiation + matter!
4 Introduction
bad news: both pillars of classical theoretical physics
are aected
good news: problems are similar solutions may
be similar
in fact, one ends up with one theory instead of two; the quantum theory
(QT)
A simple scheme of QT
classical particles
non-
relativistic
relativistic
Schrdinger
equation
Dirac
equation
wave character; statistical
interpretation deals with dualism
e

, e
-
: antiparticles
classical electromagnetic
Iield (relativistic)
' wave'
mechanics
quantum particles interacting with
classical electromagnetic Iield
1
st
quatization
QED
e, e-, C
(classical observables operators)
2
nd
quantization
(Iields operators)
interact via elemen-
tary processes
We will be concerned mainly with the left panel (wave mechanics) in this
lecture!
Chapter 2
Wave Mechanics
2.1 Matter waves
2.1.1 Classical wave functions
...solve wave equation (WE)

1
v
2

t
2 = 0 (2.1)
remember: E and B solve WE if there are no charges + currents
( = 0 , j = 0)
plane waves (PWs) solve WE:
(r, t) = Ae
i(krt)
(2.2)
(k) = vk (dispersion relation) (2.3)
For electromagnetic (EM) waves in vaccum: v = c 3 10
8 m
s
phase of PW : (r, t) = krt (2.4)
points in space with constant phase are characterized by
kr = const. =
0
t
0
(2.5)
5
6 Wave Mechanics
t = t
0
: eq. (2.5) denes a plane perpendicular to vector k
H

plane moves in direction of k


+ plane moves in opposite direction
t > t
0
: (r, t) = kr t
!
=
0
r

(t) =

0
k


k
t with r

=
k r
k
phase velocity v
ph
=

k
(2.6)
(consistent with eq. (2.3))
2.1.2 Transition to Matter Waves (MWs)
Starting point: de Broglie relations
=
E

, k =
1

p , ( =
h
2
) (2.7)
dene a plane matter wave, that moves in direction of p by using (2.7)
in (2.2):
(r, t) = Ae
i

(prEt)
(2.8)
for a free, non-relativistic particle we have
p = mv
kl
(2.9)
E =
p
2
2m
(2.10)
use v
ph
=

k
=

k
=
E
p
=
p
2m
=
v
kl
2
? (2.11)
what does that mean?
but concept of plane matter waves is ill-dened anyway as a particle
wave needs to be localized in space!
2.1 Matter waves 7
2.1.3 Wave Packets (WPs)
dened as dense linear combinations of PWs
(r, t) =
1
(2)
3
2
_
A(k)e
i(krt)
d
3
k (2.12)
It is easy to show that (r, t) and A(k) form a Fourier transform pair, i.e.
A(k) =
1
(2)
3
2
_
(r, 0)e
ikr
d
3
r (2.13)
One can dene a Fourier transform of (r, t) at any time t by
A(k, t) A(k)e
it
(2.14)
a) One-dimensional WP at t = 0
(x, 0) =
1

2
_
A(k)e
ikx
dk (2.15)
Example 1)
k
A
0
width ,k
- ,k/2 ,k/2
A(k):
(x, 0) =
A
0

2
_ k
2

k
2
e
ikx
dk
=
_
2

A
0
sin(
k
2
x)
x
(2.16)
approximate width x
4
k
classical uncertainty (a property of Fourier transforms)
k x 4
dene momentum width p = k
x p 4 (2.17)
naive uncertainty relation (will be interpreted later on)
8 Wave Mechanics
Example 2)

)
)

(x, 0) =
A
0

2
_
k
0
+
k
2
k
0

k
2
e
ikx
dk
=
_
2

A
0
sin
k
2
x
x
e
ik
0
x
(2.18)
modulation of PW with wave number k
0
by envelope function
(x, 0) =
_
2

A
0
sin
k
2
x
x
dened via
(x, 0) = (x, 0)e
ik
0
x
Figure 2.1: Real part of (x, 0) (blue line)
and envelope function (x, 0) (red line)
Example 3) Gaussian wave packet (GWP)
(x, 0) = Ce

x
2
2x
2
. .
envelope function (x,0)
e
ik
0
x
(2.19)
width x (sometimes dened as
x

2
)
2.1 Matter waves 9
Figure 2.2: Real part of GWP (blue line) and
envelope function (red line)
A(k) =
1

2
_

(x, 0)e
ikx
dx =
(exercises)
. . . =
= Cx exp
_

x
2
(k k
0
)
2
2
_
(2.20)
again, this is a Gaussian! width k =
1
x
x p (2.21)
b) Time development of WPs
dene time-dependent envelope function:
(x, t) =
1

2
_
[A(k)[e
i(k,x,t)
dk
(k, x, t) = (k k
0
)x ((k)
0
)t +(k)

0
= (k
0
)
A(k) = [A(k)[e
i(k)
(x, t) = (x, t)e
i(k
0
x
0
t)
(x, t) characterizes time-development of WP as a whole. What is the
associated velocity?
10 Wave Mechanics
dene center of WP via stationary phase condition:
d
dk

k=k
0
= 0 x
0
=
d
dk

k
0
t
d
dk

k
0
(2.22)
Obviously x
0
moves on a rectilinear trajectory with the constant group
velocity:
v
gr
= x
0
=
d
dk

k
0
Relation between group and phase velocities:
v
gr
=
d
dk

k
0
=
d
dk
_
v
ph
k
_
k
0
=
_
v
ph
+k
dv
ph
dk
_
k
0
(2.23)
Example 1) EM-WP in vaccum
(k) = ck

d
dk
= c =

k
v
gr
= v
ph
In this case the shape of the WP does not change in time:
As is a linear function of k we can write
(k) =
0
+
d
dk

k
0
(k k
0
)
=
0
+v
gr
(k k
0
)
(x, t) =
1

2
_
[A(k)[e
i[(kk
0
)xv
gr
(kk
0
)t+]
dk
=
1

2
_
A(k)e
i(kk
0
)(xv
gr
t)
dk
= (x v
gr
t, 0)
Example 2) EM-WP in media
(k) = v
ph
(k)k
in this case, is not a linear function of k
= (x, t) ,= (x v
gr
t, 0)
i.e. the WPs shape changes (dispersion)
2.1 Matter waves 11
Example 3) Matter wave packet (MWP)
(nonrelativistic case) (confer (cf.) page 6)


k
=

k
=
E
p
=
p
2m
=
k
2m
(k) =
k
2
2m
(2.24)
v
ph
= v
ph
(k) =

k
=
k
2m
(2.25)
v
gr
=
d
dk

k
0
=
k
0
m
=
p
0
m
= v
kl
(2.26)
i.e. we can identify v
gr
with the classical velocity of the particle
Since v
gr
,= v
ph
: dispersion (in vacuum)!
How fast does the MWP disperse?
(k)
0
=

2m
(k
2
k
2
0
)
=

2m
_
(k k
0
)
2
+ 2kk
0
2k
2
0
_
=

2m
(k k
0
)
2
+
k
0
m
k
k
0
m
k
0
=

2m
(k k
0
)
2
+v
gr
(k k
0
)
(x, t) =
1

2
_
A(k)e
i[(kk
0
)x(
0
)t]
dk
=
1

2
_
A(k)e
i[(kk
0
)(xv
gr
t)]
e
i

2m
(kk
0
)
2
t
. .
this term causes dispersion
dk
dispersion can be neglected as long as

2m
(k k
0
)
2
t 1
dispersion is signicant for times

m

(x)
2
(with x
1
k
=
1
kk
0
)
12 Wave Mechanics
dene width of MWP
(t) = x

1 +
_
t
m(x)
2
_
2
t

spreading of MWP!
Example: GWP for t > 0 (cf. page 8)
(x, t) =
1

2
_

A(k)e
i(kx
k
2
2m
t)
dk
with expression (2.20) for A(k)
=
exercises
. . . = involved complex expression!
consider intensity of GWP (less involved expression)
[(x, t)[
2
=
[C[
2
_
1 +
_
t
mx
2
_
2
exp
_

_
x
k
0
m
t
_
2
x
2
_
1 +
_
t
mx
2
_
2
_
_
(2.27)
still a Gaussian, centered around
x
0
=
k
0
m
t = v
gr
t
but indeed: the width increases continuously!
Figure 2.3: Intensity of spreading GWP at three
points in time
2.2 Schrodinger equation 13
Open questions at this point:
(a) interpretation of MWPs?
(b) is there a dierential equation for MWPs?
2.2 Schrodinger equation
Some requirements for a matter wave equation:
- linear + homogeneous
i.e. if eq. is solved by
1
,
2
it is also solved by
1

1
+
2

2
, (
1
,
2
C)
(principle of superposition)
- consistent with de Broglie relations
2.2.1 Schrodinger equation for a free particle
We seek for a dierential equation that is solved by MWPs
classical wave eq. (2.1)

D = (
t
2) = 0
does not do the job!
MWP : (r, t) =
_
(k, r, t) d
3
k (2.28)
(k, r, t) =
A(k)
(2)
3
2
exp
_
i

(p r Et)
_
(2.29)


D(k, r, t) = 0 =
k
2

2
=
4m
2

2
k
2
= (k)
as depends on k the wave eq. is not solved by the MWP!
way out: note that
p
2
E
=
k
2

= 2m = const. (2.30)
ansatz:

D = (
t
) = 0
check :

D(k, r, t) =
_

p
2

2
+
iE

_
(k, r, t) = 0

(2.30)
= i
2m

= const.


D(r, t) =
_

D(k, r, t) d
3
k = 0
14 Wave Mechanics
= MWP solves this dierential equation: the free Schrodinger equation
(SE):

2
2m
(r, t) = i
t
(r, t) (2.31)
Discussion:
Solutions must be complex
(i.e. sin / cos -type functions do not solve (2.31))
is not observable!
Quantization rule
apply SE to MWP explicitly:
i
t
(r, t) =
_
E
A(k)
(2)
3
2
e
i

(prEt)
d
3
k

2
2m
(r, t) =
_
p
2
2m
A(k)
(2)
3
2
e
i

(prEt)
d
3
k
from i
t
=

2
2m
it follows that
E =
p
2
2m
(classical energy conservation)
extract recipe:
classical energy
E
p
2
2m
= 0 (2.32)
quantization
E i
t


E (2.33)
p
2

2
= p
2
(2.34)
p

i
_ = p (2.35)
(sign will be justied later on)
SE:
_

E
p
2
2m
_
= 0
2.2 Schrodinger equation 15
application of the recipe to relativistic particles
classical energy-momentum relation
E
2
= p
2
c
2
+m
2
0
c
4
(2.36)
correspondence E
2

t
2

E
2
(

E
2
p
2
c
2
m
2
0
c
4
) = 0

1
c
2

t
2
_
m
0
c

_
2
_
(r, t) = 0 (2.37)
Klein-Gordon-equation
m
0
0
classical wave equation
Looks nice, but it turns out that this equation does not work for
electrons (cf. Section 5)
2.2.2 SE for a particle in a potential (conservative system)
use recipe:
classical energy
E =
p
2
2m
+V (r) = H(r, p)
. .
Hamilton function
(2.38)
quantization
(

E

H) = 0 (2.39)
i
t
(r, t) =

H(r, t) (2.40)

H =

2
2m
+ V (r) (2.41)
note that V (r) is not quantized (i.e. we apply the trivial quantization rule
r r r)
extension to time-dependent potentials V (r, t)
E = E(t) =
p
2
2m
+V (r, t) = H(r, p, t)
16 Wave Mechanics
(

E

H(t)) = 0
i
t
(r, t) =
_


2
2m
+ V (r, t)
_
(r, t) (2.42)
(energy is not conserved, but we still have E = H classically and can
quantize this relation as before).
Note that a time-dependent potential usually corresponds to a classical
description of an external eld (e.g. the classical EM eld), i.e. the time-
dependent SE (2.42) usually describes a quantum particle in a classical
environment.
2.2.3 SE for a particle in the classical EM eld
step 1: derive classical Hamilton function
(details can be found, e.g., in [Blob] and [Kira], chap. 4.3.4)
Lorentz force
F
L
= q(E +v B) (2.43)
EM potentials are introduced via
E =
t
A (2.44)
B = A (2.45)
F
L
= q
_

t
A+
_
v (A)
_
_
(2.46)
dene generalized potential
W

= q( A v) (2.47)
show that
F
L
= W

+
d
dt

v
W

(2.48)
Lagrangian
L = T W

=
1
2
mv
2
q +qv A (2.49)
Hamilton function
H = p v L =
1
2m
(p qA)
2
+q (2.50)
with canonical momentum
p =
L
v
= mv +qA (2.51)
2.2 Schrodinger equation 17
add another external potential V
H =
1
2m
(p qA)
2
+q +V (2.52)
step 2: quantization as usual
p p =

i

(

E

H) = 0
i
t
(r, t) =
_
1
2m
_

i
qA(r, t)
_
2
+q(r, t) + V (r)
_
(r, t)
=
_


2
2m
+
q
2
2m
A
2
(r, t) + q(r, t) + V (r)
_
(r, t)

q
2mi
_

_
A(r, t)(r, t)
_
+A(r, t)(r, t)
_
in Coulomb gauge ([Jac], chap. 6.5) for source-free elds we have
A = 0 , = 0
i
t
(r, t) =
_


2
2m
+ V (r)
q
m

i
A(r, t) +
q
2
2m
A
2
(r, t)
_
(r, t) (2.53)
comment: one step towards QED would be to quantize in (2.53) the vector
potential A (but we would need some additional rules to do that).
2.2.4 SE for N particles
conservative classical system:
E = H(r
1
...r
N
, p
1
...p
N
) (2.54)
quantization:

H = H(r
1
...r
N
, p
1
... p
N
) (2.55)
with p
i
=

i

i
SE
i
t
=

H (2.56)
note that = (r
1
...r
N
, t) (2.57)
i.e. is dened on 3N-dimensional conguration space!
18 Wave Mechanics
Example: Atom with N = Z electrons
i
t
(R, r
1
...r
N
, t) =
=
_

2
_

R
2M
+
N

i=1

i
2m
_

i=1
Ze
2
4
0
[r
i
R[
+

i<j
e
2
4
0
[r
i
r
j
[
_
(R, r
1
...r
N
, t) (2.58)
4
H

A
A
K?AKI
H
this eq. cannot be solved analytically
for N > 1. It contains many aspects of
the observable properties of atoms (atomic
structure) and can be solved with good accu-
racy by numerical or approximate methods.
2.3 Statistical interpretation
2.3.1 Problems with the concept of matter waves and
the solution
a) C not observable
can we interpret = [[
2
as a density
(such that q would be a classical charge density)?
b) Spreading of wave packets (concerns also (r, t)!)
c) Self interaction of a charge density
according to classical electrodynamics (r, t) should be the source of an elec-
tric eld (Poisson eq.)! This was not taken into account in the Schrodinger
eq.!
d) What does [(r
1
...r
N
, t)[
2
mean?
Obviously, it is not a charge density in ordinary space!
e) Double-slit experiment with electrons cannot be described in terms of clas-
sical waves only (neither in terms of classical particles)
this is best explained in [Fey], Vol III
2.3 Statistical interpretation 19
Solution: Borns statistical interpretation (1926)
1
:
(r, t) [(r, t)[
2
=

(r, t)(r, t)
is a measure for the probability to nd a particle at time t at position r.
this is the link between the wave picture and the particle picture and
allows a consistent description of experiments with quantum particles such as
electrons or photons.
probability distribution [[
2
is - for a given initial state
0
= (t
0
) -
uniquely determined by the Schrodinger eq., but the behavior of a given particle
is random (i.e., it cannot be predicted).
particle like
events

probability
distribution
P = [[
2

wave like
behavior
2.3.2 Probabilities, norm, continuity
(r, t) : probability amplitude
[(r, t)[
2
: probability density
[(r, t)[
2
d
3
r : probability to nd particle within d
3
r
Normalization (for one particle)
N =
_

[(r, t)[
2
d
3
r = 1 (2.59)
conditions: a) quadratically integrable (square-integrable) i.e. N <
b) N(t) = N = const.
a) is fullled, e.g., for the GWP, but not for plane waves!
prove b) :
dN
dt
=
_

t
[(r, t)[
2
d
3
r = 0 (2.60)
proof :
t
[(r, t)[
2
=

t
+ (
t

)
use SE
=
1
i
_

(

H) (

H)

_
with

H =

2
2m
+ V (r) , (V = V

)

t
[(r, t)[
2
=
i
2m
_

()

_
(2.61)
1
In a footnote of an article about e

-atom scattering
20 Wave Mechanics
use Greens theorem
2

dN
dt
=
i
2m
_
S
_

d
dn

d

dn

_
df = 0
(because () =
d
dn
() = 0)(for well-behaved quadratically integrable func-
tions)
(by the way: we have proven the hermiticity of

H - see Section 3.1.2)
dierential form of norm conservation
start from eq. (2.61):
i
2m
_

()

_
=
i
2m
div(

)
div j(r, t)
. .

current density


t
+ div j = 0 (continuity equation) (2.62)
note that eq. (2.61) is also valid for non-square integrable functions, for
which N .
check it for a plane wave (r, t) = e
i(krt)
(r, t) = 1 ,
t
= 0
j(r, t) =
i
2m
(

) =
k
m
div j = 0
2.3.3 Probability densities in coordinate and momentum
space
Wave functions:
(r, t) =
1
(2)
3
2
_
(p, t)e
i

pr
d
3
p (2.63)
(p, t) =
1
(2)
3
2
_
(r, t)e

pr
d
3
r (2.64)
2
_
V
_

_
d
3
r =
_
S(V )
_

n
_
df
2.3 Statistical interpretation 21
(cf. eqs. (2.12) - (2.14)): (r, t) and (p, t) are a Fourier transform pair.
Denition of probability densities:
(r, t) = [(r, t)[
2
(2.65)
P(p, t) = [(p, t)[
2
(2.66)
Note that for a free MWP we have P(p, t) = P(p, 0), but (r, t) ,= (r, 0), i.e.
the wave packet spreads in position space, but not in momentum space. The
stationarity of P(p) reects momentum conservation of a free particle!
We can prove that
N = 1 =
_
(r, t) d
3
r =
_
P(p, t) d
3
p (2.67)
proof:

1
(r)
2
(r) d
3
r =
1
(2)
3
_
d
3
r
_
d
3
p
_
d
3
p

(pp

)r

1
(p

)
2
(p)
=
1
(2)
3
_
d
3
p
_
d
3
p

1
(p

)
2
(p)
_
e

(pp

)r
d
3
r
_
=
_
d
3
p
_
d
3
p

1
(p

)
2
(p)(p

p)
=
_

1
(p)
2
(p) d
3
p
(this is Parseval

s relation)
(eq. (2.67) follows for
1
=
2
)
Since Fourier transformation is a unique, invertible mapping, position and
momentum wave functions are completely equivalent (i.e. they carry the same
information)
2.3.4 Expectation values
probability interpretation of QM implies that notions of probability theory,
such as the expectation value of a distribution can be dened
In one dimension (classically + QM)
x) =
_
x(x) dx
_
(x) dx = 1
_
_
_
valid for a continuous
classical distribution
QM ingredient: (x) = [(x)[
2
22 Wave Mechanics
In three dimensions
x)(t) =
_
x(r, t) d
3
r (2.68)
(if
t
,= 0 x)(t) = x)(t
0
))
write
x)(t) =
_

(r, t)x(r, t) d
3
r (2.69)
r)(t) =
_

(r, t)r(r, t) d
3
r (2.70)
in general
F(r, t)) =
_

(r, t)F(r, t)(r, t) d


3
r (2.71)
momentum space
apply the same rule:
p)(t) =
_
pP(p, t) d
3
p =
_

(p, t)p(p, t) d
3
p (2.72)
G(p, t)) =
_

(p, t)G(p, t)(p, t) d


3
p (2.73)
Example:
E H) = T) +V )
=
_

(p, t)
p
2
2m
(p, t) d
3
p +
_

(r, t)V (r)(r, t) d


3
r (2.74)
can we express H) as an integral over d
3
r only (or over d
3
p only)?
consider :
x
(r, t) =
1
(2)
3
2
_
(p, t)
x
e
i

pr
d
3
p
=
1
(2)
3
2
_
(p, t)
i

p
x
e
i

pr
d
3
p
i.e. we have a Fourier transform pair

i

x
p
x

p
x
)(t) =
_

(p, t)p
x
(p, t) d
3
p
Parseval
=
_

(r, t)

x
(r, t) d
3
r
=
_

(r, t) p
x
(r, t) d
3
r (2.75)
2.3 Statistical interpretation 23
With the same argument we obtain
p)(t) =
_

(r, t) p(r, t) d
3
r (with p =

i
) (2.76)
and in general
G(p, t)) =
_

(r, t)G
_

i
, t
_
(r, t) d
3
r (2.77)
e.g. T) =

2
2m
_

(r, t)(r, t) d
3
r
Remarks:
These relations justify the denition of the sign in p (cf. page 14)
The average momentum is related to the current density (for quadrat-
ically integrable wave functions):
p)(t) = m
_
j(r, t) d
3
r
We can apply a similar argument to show that
r) =
_

(p, t)i
p
(p, t) d
3
p (2.78)
and F(r, t)) =
_

(p, t)F(i
p
, t)(p, t) d
3
p (2.79)
Note that expressions F(i
p
, t), G(

i
, t) do not work in all cases, e.g.
for the Coulomb potential [r[ appears in the denominator!
Examples:
(i) One-dimensional harmonic oscillator
classical Hamilton function:
H(x, p
x
) =
p
2
x
2m
+
m
2

2
0
x
2
E)(t) = H)(t)
=
_

(x, t)
_


2
2m

2
x
+
m
2

2
0
x
2
_
(x, t) dx
=
_

(p
x
, t)
_
p
2
x
2m

m
2

2
0

2
p
x
_
(p
x
, t) dp
x
24 Wave Mechanics
(ii) Free particle
We can write for the position expectation value:
r)(t) =
_

(p, t)i
p
(p, t) d
3
p
with (p, t) = (p, 0)e

p
2
2m
t
r)(t) =
_

(p, 0)e
i

p
2
2m
t
i
p
_
(p, 0)e

p
2
2m
t
_
d
3
p
=
_

(p, 0)e
i

p
2
2m
t
_
i
p
(p, 0) +
p
m
t (p, 0)
_
e

p
2
2m
t
d
3
p
=
_

(p, 0)i
p
(p, 0) d
3
p +
t
m
_

(p, 0)p(p, 0) d
3
p
= r)(t = 0) +
p)
m
t

d
dt
r) =
p)
m
=
k
0
m
= v
gr
This is a special case of Ehrenfests theorems (see exercises and Section
3.4.2)
We have seen that we can evaluate expectation values either in coordinate
space or in momentum space. Both representations are fully equivalent
(below we will see that we can even rewrite the SE as an equation for
(p, t))
Note the extended correspondence rule (cf eqs. (2.33) - (2.35))

r r

coordinate space
r

i

momentum space
i
p
p

p p

2.4 Solution of the Schrodinger equation 25


2.4 Solution of the Schrodinger equation quan-
tum eects
2.4.1 Stationary (time-independent) SE
Consider eqs. (2.40), (2.41); i.e., SE for a time-independent potential V (r). Note
that i
t
(r, t) =

H(r, t) is an initial-value problem (i.e. solution is unique for
a given initial state (r, t
0
) =
0
(r))
Ansatz
(r, t) = (r)f(t) (2.80)
insertion into SE eq. (2.40)

i

f(t)
f(t)
=

H(r)
(r)
= const. A (2.81)
separate eqs.:

f =
i

Af , f(t) = e

At
(2.82)

H(r) = A(r) (2.83)


What is A?
E =

H) =
_

(r, t)

H(r, t) d
3
r =
_

(r)

H(r) d
3
r
=
_

(r)A(r) d
3
r = A
_

(r)(r) d
3
r
= A
solution of SE (eq. (2.40)) has the form

E
(r, t) =
E
(r)e

Et
(2.84)
and describes a state with constant energy.
Properties:
(r, t) = [
E
(r, t)[
2
= [
E
(r)[
2
= (r)

t
= 0 = div j = 0 (and j = j(r))

d
dt

A) = 0 for all operators




A
t
= 0
the states (2.84) are called stationary states
The task then is to solve the stationary SE

H
E
(r) = E
E
(r) (2.85)
which is an eigenvalue problem (see later), and which is specied completely
only if boundary (and regularity) conditions are imposed.
26 Wave Mechanics
Weak boundary conditions (always imposed)

E
(r),
E
(r) are continuous, unique, < (bounded)
(they have implicitly been used in chapter [2.3] in order to deal with well-
dened probability and current densities)
In general, one can distinguish two types of solutions of (2.85):
a) bound states
a classical bound particle cannot escape to innity
the quantum particle must fulll
E
(r)
r
0
=
_

[
E
(r)[
2
d
3
r <
this strong boundary condition leads to quantized states and a
discrete spectrum (i.e. only a discrete subset of energies E is allowed
in eq. (2.85))
b) scattering states
classically, a scattered particle may escape to innity

E
(r)
r
nite value

[
E
(r)[
2
d
3
r
continuous spectrum (all energies E are possible)
Extract strategy for the solution of time-dependent SE for time-independent
potentials:
1. Solve stationary SE obtain E,
E
(r)
2. A reasonably well-behaved initial-state
0
(r) can be represented by a
linear combination of stationary states
3. Apply principle of superposition: the corresponding linear combina-
tion of stationary states describes solution of time-dependent SE for
all times
For bound state problems one usually deals only with the rst issue.
2.4.2 One-dimensional model systems
Stationary SE takes the form

(x) + ( U(x))(x) = 0 (2.86)


2.4 Solution of the Schrodinger equation 27
=
2m

2
E , U =
2m

2
V
Eq. (2.86) is an ordinary dierential equation of second order and of Sturm-
Liouville type.
general solution:
=

(x) =
1

1
(x) +
2

2
(x)
eigenvalue is two-fold degenerate
a) The free particle (again)
Step 1:
Solutions of eq. (2.86) for U = 0 and 0:

(x) = Ae
ikx
+Be
ikx
k =

stationary states (cf. eq. (2.84))

(x, t) = Ae
i(kxt)
+Be
i(kx+t)
=
k
2
2m
_

(x, t)[
2
dx =
_

(x)[
2
dx
continuous spectrum (weak boundary conditions only)
Step 2: Represent initial-state wavepacket (t
0
= 0)

0
(x) =
_

0
_
A(k)e
ikx
+B(k)e
ikx
_
dk
=
1

2
_

A(k)e
ikx
dk (cf. eq. (2.15))
Typically,
0
(r) is given explicitly (e.g. as a GWP), and the task is to
determine

A(k) (by inverse FT)
Step 3: Apply principle of superposition
(x, t) =
1

2
_

A(k)e
i(kxt)
dk
28 Wave Mechanics
Remarks:
(i) WPs solve time-dependent SE
(ii) On the other hand PWs are eigenfunctions of

H, but not appropriate
wave functions for particles (well: not really...)
(iii) Think of PW as (non-normalized) limit of WP with width k 0
(iv) < 0 :

(x) = Ae
x
+Be
x
(2.87)
=

boundary conditions:

(x) < x
= x B = 0
x + A = 0

(x) = 0
(free particle with < 0 does not exist (in nonrelativistic QM))
b) General remarks about piecewise constant potentials
We will consider potentials of the type U(x) = U
i
for x [a
i
, b
i
];
step
x x x
barrier well
e. g.
Figure 2.4: Dierent types of piecewise constant potentials
Scheme:
1) Solve SE (2.86) in all intervals, in which U is constant, separately
> U
i
k
i
=
_
U
i
> 0
SE :

(x) + k
2
i
(x) = 0
solution :

(x) = Ae
ik
i
x
+Be
ik
i
x
(2.88)
=

Asin(k
i
x +)
2.4 Solution of the Schrodinger equation 29
< U
i
k
i
=
_
U
i
= i
_
U
i

i
:= ik
i
=
_
U
i
> 0
SE :

(x)
2
i
(x) = 0
solution :

(x) = Ae

i
x
+Be

i
x
(2.89)
2) Matching and boundary conditions
for n intervals with constant potentials U
i
...U
n
we have (2n 2)
matching conditions
2 constants remain (xed by normalization and phase initial
condition of physical problem)
c) Potential step
N
7
1 11
U(x) = U(x)
_
_
_
0 x < 0
U x 0

(x) = (x)
I
(x) + (x)
II
(x) (2.90)
1) > U :

I
(x) = Ae
ik
I
x
+Be
ik
I
x
(2.91)

II
(x) = Ce
ik
II
x
+De
ik
II
x
(2.92)
k
I
=

(2.93)
k
II
=

U =
_
k
2
I
U (2.94)
matching conditions:

I
(0) =
II
(0) (2.95)

I
(0) =

II
(0) (2.96)
choose A = 1 , D = 0 (justication follows)
eqs. (2.95) + (2.96) yield

1 +B = C
ik
I
ik
I
B = ik
II
C

B =
k
I
k
II
k
I
+k
II
, C =
2k
I
k
I
+k
II
(2.97)
30 Wave Mechanics

(x, t) =
_

in
(x)(x) +
ref
(x)(x) +
trans
(x)(x)
_
e
it

in
(x) = e
ik
I
x
j
in
(x) =
i
2m
_

in
d
x

in

in
d
x

in
_
=
k
I
m
(2.98)
(cf. eq. (2.62))

ref
(x) =
k
I
k
II
k
I
+k
II
e
ik
I
x
j
ref
(x) =
_
k
I
k
II
k
I
+k
II
_
2
k
I
m
(2.99)

trans
(x) =
2k
I
k
I
+k
II
e
ik
II
x
j
trans
(x) =
_
2k
I
k
I
+k
II
_
2
k
II
m
(2.100)
dene reection coecient R and transmission coecient T
R =

j
ref
j
in

=
_
k
I
k
II
k
I
+k
II
_
2
(2.101)
T =

j
trans
j
in

=
4k
I
k
II
(k
I
+k
II
)
2
(2.102)
= R +T = 1

6A
4A
7
A
(if we choose A = 0, D = 1,
we obtain the same R, T)
Note that R, T have to be interpreted as probabilities for reection
and transmission. The individual event (as stated on page 17) is not
determined. Each particle is either reected or transmitted; i.e. the
wave packet splits, but not the particle!
consider WPs for further analysis
(x, t) =
1

2
_

f(k
I
)

(x)e
it
dk
I
(2.103)
2.4 Solution of the Schrodinger equation 31
solves time-dependent SE if

solves stationary SE. More explicitly:


(x, t) =
1

2
_

f(k
I
)
_
(x)
_
Ae
i(k
I
xt)
+Be
i(k
I
x+t)
_
+ (x)
_
Ce
i(k
II
xt)
+De
i(k
II
x+t)
__
dk
I
(2.104)
note: = (k
I
) =
k
2
I
2m
; k
II
= k
II
(k
I
) (eq. (2.94))
Consider the motion of centers of WP (cf. eq. (2.22), (2.23)):
A)
A
(k
I
) = k
I
x t
d
A
dk
I

k
0
I
= 0 x
A
0
=
k
0
I
m
t
moves to the right in I for t < 0 ( incoming WP)
B)
B
(k
I
) = k
I
x t
x
B
0
=
k
0
I
m
t
moves to the left in I for t > 0 ( reected WP)
C)
C
(k
I
) = k
II
(k
I
)x (k
I
)t
x
C
0
=
k
0
II
m
t
moves to the right in II for t > 0 ( transmitted WP)
D)
D
(k
I
) = k
II
(k
I
)x (k
I
)t
x
D
0
=
k
0
II
m
t
moves to the left in II for t < 0
N
J
If we choose A = 1, D = 0, we
choose the initial condition
corresponding to an incoming
WP from left to right
N
J If we choose A = 0, D = 1, the
incoming WP moves from the
right to the left
32 Wave Mechanics
due to eq. (2.104) we can speak of a splitting of the wave packet
at t 0; it is partly reected and partly transmitted.
2) < U :

I
(x) = Ae
ikx
+Be
ikx
(2.105)

II
(x) = Ce
x
+De
x
(2.106)
k k
I
=

, =

U > 0 (2.107)
(x) < x C = 0

(x) = (x)
_
Ae
ikx
+Be
ikx
_
+(x)De
x
(2.108)
choose A = 1
matching conditions (2.95) + (2.96) yield
B =
+ik
i
, D =
2ik
ik
(2.109)
current densities (cf. page 25)
j
in
=
k
m
j
ref
= j
in
j
trans
= 0
_

_
= T = 0 , R = 1 (2.110)
no current in classically forbidden region II, but nite prob-
ability density

II
(x) = [
II
(x)[
2
= [D[
2
e
2Kx
=
4k
2
k
2
+
2
e
2x
(2.111)
dene average penetration depth x
m
:

II
(x
m
) =
1
2

II
(0)
(2.111)
= x
m
=
ln 2
2
(2.112)
We can anticipate the tunnel eect: if the potential step has a
nite width, i.e. becomes a barrier, there is a chance (i.e., a nite
probability) to nd the particle that comes from one side on the
other side of the barrier (a quantum particle can penetrate a nite
barrier)
For our problem we have total reection (R = 1), but as the
particle can enter region II the reection is (slightly) delayed
2.4 Solution of the Schrodinger equation 33
WPs
(x, t) =
1

2
_
f(k)
_
(x)
_
e
i(kxt)
+B(k)e
i(kx+t)
_
+ (x)D(k)e
x
e
it
_
dk (2.113)
note that B(k)
(2.109)
= [B(k)[e
i(k)
= e
i(k)
D(k) = [D(k)[e
i(k)
=
2k

2
+k
2
e
i(k)
i.e. additional phases , !
motion of center of WPs:

in
x
0
=
k
0
m
t (as before)

ref
(k) = kx t +(k)
x
0
=
k
0
m
t +
d
dk

k
0
(2.114)
delay time:
(x
0
= 0) =
m
k
0
d
dk

k
0
=
d
dE

E
0
(2.115)
= arctan
B
1 B
= arctan
2

U
2
U 2
(2.116)
d
dE
=
2m

2
d
d
=
2m

2
_
(U )
_

1
2
> 0 , (for ]0, U[ )
(2.117)
consider limit U
=

U
II
(x) = 0 , (x > 0)
wave function vanishes in regions where U
d) Innitely deep potential well
N
7 7
A

7N
U(x) =
_
_
_
0
L
2
x
L
2
else

(x) = Ae
ikx
+Be
ikx
; k =

> 0 (2.118)
34 Wave Mechanics
boundary conditions (see above)

L
2
_
=
_
L
2
_
= 0 (2.119)

Ae
ik
L
2
+Be
ik
L
2
= 0
Ae
ik
L
2
+Be
ik
L
2
= 0

sin kL = 0 k = k
n
=
n
L
(2.120)
E
n
=

2
k
2
n
2m
=
1
2m
_

L
_
2
n
2
(2.121)
quantized energy levels!
eigenfunctions:

n
(x) = A
_
e
ik
n
x
(1)
n
e
ik
n
x
__

_
x+
L
2
_

_
x
L
2
__
(2.122)
norm alization:
N = 1 =
_

[
n
(x)[
2
dx =
_ L
2

L
2
[
n
(x)[
2
dx
= A =
1

2L

n
(x) =
_

_
x+
L
2
_

_
x
L
2
__
_
2
L
_
_
_
cos
n
L
x , n = 1, 3, 5, ...
i sin
n
L
x , n = 2, 4, ...
(note that
0
(x) sin 0x = 0)
discussion:
Quantization = standing waves: de Broglie wavelength

B
n
=
2
k
n
=
2L
n
is adapted to width of the well
ground-state energy E
1
=
1
2m
_

L
_
2
> 0
( zero point motion: bound particle cannot be at rest)
Later on we will see that this is a consequence of the uncertainty
relation
quantization of energies E
n

1
m
level spacings become very
small for heavy particles

n
(x) has n 1 nodes (general result)
2.4 Solution of the Schrodinger equation 35
parity; dened by
(x) = (x) even parity (e.g. cosine) (2.123)
(x) = (x) odd parity (e.g. sine) (2.124)
in our case U(x) = U(x) eigenfunctions
n
have well-dened
parity (see chapter (3.5.5))
e) Finite potential well
x
U
-
L
/
2
L
/
2
U(x)
I II III
U(x) =
_

_
U x <
L
2
0
L
2
x
L
2
U x >
L
2
1) < U :

I
(x) = Ae
x
+Be
x
(2.125)

II
(x) = C sin(kx +) (2.126)

III
(x) = De
x
+Ee
x
(2.127)
=

U > 0 , k =

> 0 (2.128)
boundary condition: (x) < x
B = E = 0
matching conditions: consider

I
_

L
2
_
=
II
_

L
2
_
(2.129)

II
_
L
2
_
=
III
_
L
2
_
(2.130)
and logarithmic derivative
d
dx
ln (x) =

(x)
(x)
instead of deriva-
tive

I
(
L
2
)

I
(
L
2
)
=

II
(
L
2
)

II
(
L
2
)
(2.131)

II
(
L
2
)

II
(
L
2
)
=

III
(
L
2
)

III
(
L
2
)
(2.132)
36 Wave Mechanics
From eqs. (2.131) + (2.132) we get

= tan
_

kL
2
+
_
k

= tan
_
kL
2
+
_

(2.133)
solve for :

=
kL
2
+ arctan
k

=
kL
2
arctan
k

+n

Add and subtract both equations to obtain

=
n
2
arctan
k

=
kL
2
+
n
2

(2.134)
transcendental equation determines the quantization of energy
levels
graphical analysis:
dene =
_

U
, (0 < 1)
K =

U
eq. (2.134) can be written as
n
2

KL
2
= arcsin (2.135)
solutions: intercept points of straight line and arcsin function:
8
4
-4
6
2
xi
1 0,8 0,2 0 0,4
-2
0
0,6
n eigenvalues exist if
(n1)

UL
2
< n
(at least one for n = 1)
quantization : k k
n

n
A, C, D A
n
, C
n
, D
n
2.4 Solution of the Schrodinger equation 37
eigenfunctions:
eqs. (2.129) + (2.130) together with =
n
2
(eq. (2.134))
yield:

A
n
C
n
= e

n
L
2
sin
_

k
n
L
2
+
n
2
_
D
n
C
n
= e

n
L
2
sin
_
k
n
L
2
+
n
2
_

(2.136)
Distinguish n even and n odd:
n even : A
n
= D
n
C
n
=
A
n
e

n
L
2
sin
k
n
L
2
+ for n = 2, 6, 10, ...
for n = 4, 8, 12, ...
solution :

n
(x) =
_

_
A
n
e

n
x
x <
L
2

A
n
e

n
L
2
sin
k
n
L
2
sin k
n
x
L
2
x
L
2
A
n
e

n
x
x >
L
2
n odd : A
n
= D
n
C
n
=
A
n
e

n
L
2
cos
k
n
L
2
+ for n = 1, 5, 9, ...
for n = 3, 7, 11, ...
solution :

n
(x) =
_

_
A
n
e

n
x
x <
L
2
A
n
e

n
L
2
cos
k
n
L
2
cos k
n
x
L
2
x
L
2
A
n
e

n
x
x >
L
2
38 Wave Mechanics
A
n
can be determined by
_

[
n
(x)[
2
dx = 1
= A
2
n
_

L
2

e
2
n
x
dx +A
2
n
_

L
2
e
2
n
x
dx
+ A
2
n
e

n
L
sin
2
(
k
n
L
2
)
_ L
2

L
2
sin
2
(k
n
x) dx
= A
2
n
_
e

n
L

n
+
e

n
L
sin
2
(
k
n
L
L
)
_
L
2

1
2k
n
sin(k
n
L)
_
_
A
n
= e

n
L
2
_
1

n
+
L
2

1
2k
n
sin(k
n
L)
sin
2
(
k
n
L
2
)
_

1
2
(same result is obtained for n odd)
structurally very similar to previous case (U ), but
eigenfunctions enter (slightly) the classically forbidden regions
I, III.
2) > U :

I
(x) = e
ik
1
x
+Be
ik
1
x
, k
1
=

II
(x) = Ce
ik
2
x
+De
ik
2
x
, k
2
=

III
(x) = Ee
ik
1
x
(we have chosen A = 1 , F = 0; cf. page 25)
4 equations for 4 constants B, C, D, E follow from matching
conditions
continuous spectrum (no restriction on )
consider current densities and coecients T, R (page 26)
j
in
=
k
1
m
j
ref
= [B[
2 k
1
m
j
trans
= [E[
2 k
1
m
_

_
R = [B[
2
T = [E[
2
(R +T = [B[
2
+[E[
2
= 1)
2.4 Solution of the Schrodinger equation 39
matching conditions yield (after some calculation (see A.1))
E =
e
ik
1
L
cos k
2
L
i
2
_
k
1
k
2
+
k
2
k
1
_
sin k
2
L
T =
1
cos
2
k
2
L +
1
4
_
k
1
k
2
+
k
2
k
1
_
2
sin
2
k
2
L
(2.137)
T = 1 if cos
2
k
2
L = 1 , sin
2
k
2
L = 0 k
2
=
n
L
(cf. eq. (2.120)) De Broglie wavelength
B
=
2
k
max
2
=
2L
n
is
adapted to width of the well (cf. Sec. (2.4.2d))
one can show that a particle (represented by a narrow
WP) with wave number close to k
2
=
n
L
stays in II for a long
time (signicantly longer than a classical particle) before it
moves on. (it is quasi-bound resonance phenomenon)
1
0,2
0,8
0,4
0,6
k2
6 5 4 2 3
Figure 2.5: Transmission coecient
40 Wave Mechanics
Transition from potential well to barrier
well : k
1
=

U
k
2
=

barrier : k
1
=

U
k
2
=

2U = i

2U i
i.e. replace in eq. (2.137) k
2
by i
T =
1
cosh
2
L +
1
4
_

k
1

k
1

_
2
sinh
2
L
this is the transmission coecient for tunneling through a
rectangular barrier.
f) Summary: quantum eects
spreading of (free) WPs
splitting of WPs: T, R
WPs can enter + penetrate classically forbidden regions
delayed reection, tunnel eect
bound states in wells are quantized; zero-point motion + en-
ergy
scattering at potential wells: resonances
all items are rather general phenomena (not at all specic for
our simple models!)
Chapter 3
Formalism of Quantum
Mechanics
3.1 Mathematical framework
Wave mechanics is based on the principle of superposition. Mathematically, this
implies that wave functions are vectors.
3.1.1 (State-) vectors and their linear space
a) Denition: vector space H (over C)
denote (ket) vectors as [) (the symbol [ ) was introduced by Dirac)
A vector space is a set of vectors with two composition laws:
(i) summation of vectors:
+ : HH H
[
1
) +[
2
) = [) H
H is abelian group with respect to +, i.e.:

_
[
1
) +[
2
)
_
+[
3
) = [
1
) +
_
[
2
) +[
3
)
_
neutral element [0) H
[) +[0) = [) , ( [) [) H)
[) H , [

) H:
[) +[

) = [0)
[

) [ ) (inverse element)
abelian: [
1
) +[
2
) = [
2
) +[
1
)
41
42 Formalism of Quantum Mechanics
(ii) multiplication with C :
: C H H
[) = [) H
(
1
+
2
)[) =
1
[) +
2
[)

1
_
[
1
) +[
2
)
_
=
1
[
1
) +
1
[
2
)
(
1

2
)[) =
1
_

2
[)
_
1 [) = [)
principle of superposition: [
i
) H ,
i
C , i = 1, ..., N
[) =
N

i=1

i
[
i
) H
b) Denition: scalar (inner, dot) product in H
HH C
[
1
), [
2
)
1
[
2
)
. .

bracket

C

1
[
2
+
3
) =
1
[
2
) +
1
[
3
)

1
[
2
) =
1
[
2
)

1
[
2
)

=
2
[
1
)
[) 0 ; [) = 0 [) = [0)
([) R follows from
1
[
2
)

=
2
[
1
))
a complex vector space with a scalar product is called unitary space or
prehilbert space
consequences:

1
+
2
[
3
) =
1
[
3
) +
2
[
3
)

1
[
2
) =

1
[
2
)
The objects

[

are called bra (vectors). They are not elements of H,


but form another vector space, namely the so-called dual vector space H

of H.
obvious properties : [ =

1
+
2
[ =
1
[ +
2
[
3.1 Mathematical framework 43
Some remarks:
norm of a vector: [[[[ :=
_
[)
Schwarzs inequality: [
1
[
2
)[ [[
1
[[ [[
2
[[
orthogonality:
1
[
2
) = 0
orthonormality:
1
[
2
) = 0

1
[
1
) =
2
[
2
) = 1 , i.e.
i
[
j
) =
ij
quadratically integrable functions form unitary space.
Vector properties can be easily checked.
scalar product
1
[
2
) :=
_

1
(x)
2
(x) dx
normalization [) =
_

[(x)[
2
dx
This can be generalized to functions (r
1
...r
N
):

1
[
2
) :=
_

1
(r
1
...r
N
)
2
(r
1
...r
N
) d
3
r
1
...d
3
r
N
c) Basis states and systems
i) nite vector space dimH = N
a set of vectors
_
[
i
) , i = 1, ..., N
_
is linearly independent:

i=1

i
[
i
) = 0 only if
1
=
2
= ... =
N
= 0
= H : [) =
N

i=1

i
[
i
) ,
_
[
i
)
_
:

basis

a basis
_
[
i
)
_
is called orthonormal if
[) =
N

i=1

i
[
i
)

i
[
j
) =
ij
, i, j = 1, ..., N
ii) dimH
denition of a linearly independent set persists, but we have to clarify
what [) =

i=1

i
[
i
) shall mean, i.e., we need to establish the notion
of convergence of a sequence [
N
) =

N
i=1

i
[
i
)
N
[)
44 Formalism of Quantum Mechanics
Lets dene a distance (metric):
d(,
N
) [[
N
[[
=
_

N
[
N
)
N
0
this is a well-dened notion of convergence
Cauchy sequence
_
[
N
)
_
: > 0 N N :
[[
n

m
[[ < , n, m > N
a unitary space is called complete if all Cauchy sequences converge,
i.e., [
N
)
N
[) H
d) Hilbert space : unitary space which is complete with respect to
norm [[[[ =
_
[)
separable Hilbert space : [) H countable sequence
_
[
n
)
_
such that [) =

n=1

n
[
n
) (the basis may be orthonor-
mal (ONB)
i
[
j
) =
ij
)
The quadratically integrable functions (r
1
...r
N
) in
conguration space (r
1
...r
N
) form a separable Hilbert space
with dimH =
QM states can be characterized as Hilbert space vectors
example for an orthonormal basis in such a space (

L
2
(, )

n
(x) =
1
_
2
n
n!

x
2
2
H
n
(x), n = 0, 1, ...
(Hermite

s functions)
H
n
(x) = (1)
n
e
x
2 d
n
dx
n
e
x
2
Hermite

s polynomials
= (x) =

n=0

n
(x)
for all quadratically integrable functions (x)

m
(x)(x) dx =
m
[)
=

n=0

n
_

m
(x)
n
(x) dx
. .
=
mn
=
m
3.1 Mathematical framework 45
Making use of Diracs notation we can write
[) =

n
[
n
) (3.1)

m
[) =
m
(3.2)
scalar product with respect to ONB:

1
[
2
) =

(1)
m

(2)
n

m
[
n
)
. .
=
mn
=

1
[
m
)
m
[
2
)
in particular:
[) =

n,m

m
[
n
)
. .
=
mn
=

n
[
n
[
2
(3.3)
(Parsevals relation: characterizes a complete basis set)
Short hand notation: [
n
) [n)
3.1.2 (linear) Operators
Operators are linear maps: [)

A
[) =

A[) = [

A):

A([
1
) +[
2
)) =

A[
1
) +

A[
2
) (3.4)

A([)) =

A[) (3.5)
a) Sum of operators and multiplication with scalar
(

A +

B)[) :=

A[) +

B[) , [) H
(

A)[) := (

A[)) , C
the set of linear operators forms a vector space! (check the vector space
axioms!)
b) Product of operators

A

B[) :=

A(

B[)) =

A[

B) , [) H
(apply

B rst, and then

A)
properties:

A(

B

C) = (

A

B)

C
(

A

B) = (

A)

B =

A(

B)

1 =

1

A =

A , (

1 :

unity

operator)

A(

B +

C) =

A

B +

A

C
(

A +

B)

C =

A

C +

B

C
46 Formalism of Quantum Mechanics
but:
not every operator

A has an inverse operator

A
1
, such that

A

A
1
=

A
1

A =

1
in general

A

B ,=

B

A
commutator [

A,

B] =

A

B

B

A (3.6)
(a very important object for QM!)
some rules (easy to prove):
[

A,

A] = [

A,

1] = 0
[

A,

B] = [

B,

A]
[

A,

B] = [

A,

B]
[

A,

B +

C] = [

A,

B] + [

A,

C]
[

A,

B

C] = [

A,

B]

C +

B[

A,

C]
0 = [

A, [

B,

C]] + [

B, [

C,

A]] + [

C, [

A,

B]]
c) Operator functions
With sum and product of operators we can dene
f(

A) =
0

1 +
1

A +
2

A
2
+. . .
e.g. : e

A
:=

n=0

A
n
n!
(3.7)
note that e

A+

B
= e

A
e

B
only if [

A,

B] = 0
d) Special operators
Inverse operator: if it exists it is dened via
[) =

A[)
[) =

A
1
[)
=

A

A
1
=

A
1

A =

1 (3.8)
(

A
1
)
1
=

A
(

A)
1
=
1

A
1
(

A

B)
1
=

B
1

A
1
Adjoint operator

A

consider scalar product [) [

A)
3.1 Mathematical framework 47

is dened by property:
[

A) =

[) = [

(3.9)
properties (easy to verify) : (

A

=

A
(

A)

(

A +

B)

=

A

+

B

(

A

B)

=

B

note the conjugation relations


c-numbers


vectors [ [)
operators

A



A
(but note that [ and [) are elements of dierent spaces)
Self-adjoint (hermitian) operator :

A =

A

properties:

[) =

A[) = [

A)
write = [

A[) (3.10)
(we will use this notation from now on for all operators. In general,
it is understood that

A acts to the right.)
[

A[) R
(

A

B)

=

B

=

B

A =

A

B only if [

A,

B] = 0
[

A,

B]

= [

B,

A] = [

A,

B]
Unitary operator :

U



U
1


U

U

=

U


U =

1 (3.11)
properties:
[

) := U[) , [

) :=

U[)
=

) =

U[

U) = [


U[)
= [)


W =

U

V and

U,

V unitary
=

W
1
= (

U

V )
1
=

V
1

U
1
=

V

= (

U

V )

=

W

( unitary operators form a group)


48 Formalism of Quantum Mechanics
Projection operators (projectors)
dene elementary projector via

P
n
[) := [n)n[) , (assume n[n) = 1)
i.e.

P
n
= [n)n[ (3.12)
properties: (easy to prove)

P
n
=

P

P
2
n
=

P
n
_
_
_
these are the dening properties
of general projectors
=

P
m

P
n
=
mn

P
n
further examples of projectors:

P =
N

n=1
[n)n[ , (i.e., show

P =

P

,

P
2
=

P)


P[) =
N

n=1
[n)n[) =
N

n=1

n
[n)

n=1
[n)n[

1 ,

completeness relation

(3.13)
[) =

n=1
[n)n[) =

n=1

n
[n)
The completeness relation (decomposition of the unity operator) is
equivalent to Parsevals relation (3.3):
[) = [

1[) =

n
[n)n[) =

n
[
n
[
2
3.1.3 Representation of vectors and operators
insert unity operator at appropriate places:
a) Vectors
[) =

1[) =

n
[n)n[) (3.14)
[)
_
_
_
1[)
2[)
.
.
.
_
_
_
=
_
_
_

2
.
.
.
_
_
_
3.1 Mathematical framework 49
[ = [

1 =

n
[n)n[ (3.15)
[ ([1), [2), . . .)
= (1[)

, 2[)

, . . .)
= (

1
,

2
, . . .)
scalar product
[) =

n
[n)n[) = (

2
. . .)
_
_
_

2
.
.
.
_
_
_
=

1
+

2
+. . . =

n
(3.16)
b) Operators

A =

1

A

1 =

m,n
[m)m[

A[n)n[ (3.17)
matrix elements A
mn
:= m[

A[n) C

A
_
_
_
1[

A[1) 1[

A[2) . . .
2[

A[1) 2[

A[2) . . .
.
.
.
.
.
.
_
_
_
m[

A[) =

n
m[

A[n)n[) =

n
A
mn

n
the representation of the vector

A[) has to be calculated as a
matrix-vector product
[

A[n) =

m
[m)m[

A[n) =

m
A
mn
general matrix element
[

A[) = [

1

A

1[) =

mn

m
A
mn

n
(

2
. . .)
_
_
_
A
11
A
12
. . .
A
21
A
22
. . .
.
.
.
.
.
.
_
_
_
_
_
_

2
.
.
.
_
_
_
operator product

A

B =

1

A

1

B

1 =

mk
[m)

n
A
mn
B
nk
. .
product of matrices
k[ (3.18)
50 Formalism of Quantum Mechanics
adjoint operator
(

A

)
mn
= m[

[n) = n[

A[m)

= A

nm
= A

mn
(3.19)
adjoint operator is represented by adjoint matrix
hermitian operator

A =

A

A
mn
= m[

A[n) = n[

A[m)

= A

nm
(3.20)
c) Change of representation
consider two dierent ONBs
_
[u
i
) ; i = 1, ...
_
, u
i
[u
k
) =
ik
_
[v
j
) ; j = 1, ...
_
, v
j
[v
l
) =
jl
[) =

i
[u
i
)u
i
[)
=

ij
[v
j
)v
j
[u
i
)u
i
[) =

j
[v
j
)v
j
[)
v
j
[) =

i
v
j
[u
i
)u
i
[) =

i
U
ji
u
i
[) (3.21)
transformation is mediated by unitary matrix U!
proof:

k
U

ik
U
kj
=

k
U

ki
U
kj
=

k
v
k
[u
i
)

v
k
[u
j
)
=

k
u
i
[v
k
)v
k
[u
j
) = u
i
[u
j
) =
ij
if ONBs
_
[u
i
)
_
,
_
[v
j
)
_
have the same index set one can extract an
operator

U that has the matrix elements U
ji
= v
j
[u
i
)


U =

k
[u
k
)v
k
[ , (

U
1
=

U

) (3.22)
proof:
U
ji
= v
j
[

U[v
i
) =

k
v
j
[u
k
) v
k
[v
i
)
. .

ki
= v
j
[u
i
) = u
j
[

U[u
i
)
=

k
u
j
[u
k
)v
k
[u
i
) = v
j
[u
i
)
3.1 Mathematical framework 51
transformation of operators
u
i
[

A[u
j
) =

kl
u
i
[v
k
)v
k
[

A[v
l
)v
l
[u
j
)
=

kl
U

ik
v
k
[

A[v
l
)U
lj
(3.23)
3.1.4 Non-orthogonal basis sets
Sometimes it is useful to consider a basis
_
[
i
) , i = 1, ...
_
which is (complete,
but) not orthonormal, i.e.,
S
ij

i
[
j
) ,=
ij
S
ij
= overlap matrix is positive denite and hermitian
(S
ij
= S

ji
) inverse exists and is also hermitian!
A projector onto an Ndimensional subspace
_
[
1
) . . . [
N
)
_
is given as

P =
N

ij=1
[
i
)S
1
ij

j
[ (3.24)
proof:
i)

P =

P

P[) =
N

ij=1
[
i
)S
1
ij

j
[)
=

ij

i
[)

_
S
1
ji
_

[
j
)

=
_

ij
[
j
)S
1
ji

i
[)
_

= [

P[)

P[)
ii)

P
2
=

P


P
2
=

ijkl
[
i
)S
1
ij

j
[
k
)S
1
kl

l
[
=

ikl
[
i
)
_

j
S
1
ij
S
jk
_
. .
=
ik
S
1
kl

l
[
=

il
[
i
)S
1
il

l
[ =

P
=

1 =

ij=1
[
i
)S
1
ij

j
[ (3.25)
52 Formalism of Quantum Mechanics
representation of a ket
[) =

ij=1
[
i
)S
1
ij

j
[)

i
c
i
[
i
)
c
i
=

j
S
1
ij

j
[)
representation of an operator

A =

1

A

1 =

ijkl
[
i
)S
1
ij

j
[

A[
k
)S
1
kl

l
[

il
[
i
)

A
il

l
[

A
il
=

jk
S
1
ij
A
jk
S
1
kl
for S
ij
=
ij
= S
1
ij
the old expressions (3.14), (3.17) are reobtained.
3.1.5 Eigenvalue problem


A[u
a
) = a[u
a
) , ([u
a
) H) (3.26)
Obviously it depends on the properties of

A whether such an equation is fullled,
i.e., whether eigenvectors [u
a
i
) and eigenvalues a
i
exist.
Insert

1 in (3.26):

A[k)k[u
a
) = a

k
[k)k[u
a
) [[ l[

k
_
A
lk
u
k
a
a
lk
u
k
a
_
= 0 (3.27)
for dimH = N this homogeneous system of eqs. has a nontrivial solution if
det(A
lk

lk
a) = 0 (3.28)
the roots of eq. (3.28) are the eigenvalues a
i
, and inserting them into (3.27)
yields the eigenvectors. The eigenvectors can be normalized: u
a
i
[u
a
i
) = 1
a) Eigenvalue problem for hermitian operators (non-degenerate case)

A[u
a
i
) = a
i
[u
a
i
)
3.1 Mathematical framework 53
non-degenerate case means that each eigenvector [u
a
i
) corresponds to a
dierent eigenvalue a
i
(i.e. a
i
,= a
j
for i ,= j)
u
a
j
[

A[u
a
i
) = a
i
u
a
j
[u
a
i
)
=

Au
a
j
[u
a
i
)
= u
a
i
[

A[u
a
j
)

= a

j
u
a
j
[u
a
i
)
(a
i
a

j
)u
a
j
[u
a
i
) = 0
i) i = j : =a
i
= a

i
i (3.29)
all eigenvalues are real numbers!
ii) i ,= j : =u
a
j
[u
a
i
) = 0 (3.30)
eigenvectors are orthogonal!
further remarks:
the eigenvectors of hermitian operators form ONBs in Hilbert space
(disregarding some exceptions that will not be discussed):

i=1
[u
a
i
)u
a
i
[ =

1
u
a
j
[

A[u
a
i
) = a
i

ij
(3.31)
diagonal matrix

A
_
_
_
_
_
a
1
0
a
2
.
.
.
0
.
.
.
_
_
_
_
_
a
i

ij
=

kl
u
a
j
[l)l[

A[k)k[u
a
i
)
=

kl
U

jl
A
lk
U
ki
diagonalization is mediated by unitary transformation from
representation
_
[k)
_
to representation
_
[u
a
i
)
_
(cf. eq. (3.23))
54 Formalism of Quantum Mechanics
spectral representation of an operator

A =

A

A =

1

A

1 =

i
a
i
[u
a
i
)u
a
i
[ =

i
a
i

P
i
(3.32)
[

A[) =

i
[u
a
i
[)[
2
a
i
(3.33)
f(

A) =
0

1 +
1

A +
2

A
2
+. . .
f(

A)[u
a
i
) = f(a
i
)[u
a
i
) (3.34)
i.e. operator function f(

A) has the same eigenvectors as

A
Instead of a general proof: illustrate (3.34) for

A
2
:

A
2
(3.32)
=

ij
[u
a
j
) u
a
j
[u
a
i
)
. .
=
ij
u
a
i
[a
i
a
j
=

i
a
2
i
[u
a
i
)u
a
i
[
b) Eigenvectors of commuting operators (without degeneracies)
consider :

A[u
a
i
) = a
i
[u
a
i
)
and [

A,

B] = 0


A

B[u
a
i
) =

B

A[u
a
i
) = a
i

B[u
a
i
)
=

B[u
a
i
) = b
i
[u
a
i
)
on the other hand: assume that

A,

B have the same complete set of eigen-
vectors, i.e.

A =

i
a
i
[u
a
i
)u
a
i
[ =

i
a
i

P
i

B =

j
b
j
[u
a
j
)u
a
j
[ =

j
b
j

P
j
=

A

B =

ij
a
i

P
i
b
j

P
j
=

ij
b
j

P
j
a
i

P
i
=

B

A , (because [

P
i
,

P
j
] = 0)
summary: operators

A,

B have the same complete set of eigenvectors
[

A,

B] = 0
very important statement!
3.1 Mathematical framework 55
c) Degeneracy

A[a

i
) = a
i
[a

i
) , = 1, ..., N
i
(3.35)
(we use short-hand notation [a

i
) [u

a
i
))
eigenvalue a
i
is N
i
-fold degenerate, i.e., N
i
linearly independent eigen-
vectors
_
[a

i
) , = 1, ..., N
i
_
exist. They span an N
i
-dimensional sub-
space of Hilbert space (eigenspace)(N
i
is also possible)
they are not automatically orthogonal, but one can always orthogonal-
ize them (e.g. by Gram-Schmidt procedure)
a

i
[a

j
) =
ij

(3.36)
consider operator

B which commutes with

A (cf. page 54)


A

B[a

i
) =

B

A[a

i
) = a
i

B[a

i
)
i.e. vector

B[a

i
) is eigenvector of

A with eigenvalue a
i
. This implies
that

B[a

i
)
_
[a
1
i
), ..., [a
N
i
i
)
_
One can always nd a transformation such that

B[ a

i
) = b

i
[ a

i
)
note that eigenvalues b

i
of

B are not necessarily degenerate (like
a
i
); degeneracy may be lifted
summary: [

A,

B] = 0

A,

B have same (complete) set of
eigenvectors, but degeneracies of eigenvalues can be dierent
unity operator:

i
[a

i
)a

i
[ =

1 (3.37)
spectral representation:

A =

i
a
i
[a

i
)a

i
[ (3.38)
(note that

i
=

i=1

N
i
=1
)
3.1.6 Continuous basis sets and spectra
(One) motivation: stationary Schrodinger equation has the form of an eigen-
value equation. We have seen that in some cases the spec-
trum of eigenvalues is continuous ( scattering problems)
and not discrete!
we need to talk about continuous basis sets and spectra
56 Formalism of Quantum Mechanics
For [), [) H (H separable) we can calculate the scalar product ac-
cording to
[) = [

1[) =

k
[k)k[) =

k
On the other hand we have argued that
[) =
_

(x)(x) dx
is a legitimate scalar product for quadratically integrable wave functions
(which form a separable Hilbert space). If we compare both expressions we
nd that the second looks like the result of a limiting process of the rst
one:

k
=

k
[k)

k
k[)

k
k
k0

(k)(k) dk (3.39)
with (k) =

k[) = lim
k0
k[)

k
and

(k) = [

k)
[

k) = lim
k0
[k)
k
(3.40)
i.e. the scalar product (3.39) is the representation of the scalar product
[) in a continuous basis
_
[

k)
_
.
a) Representation of kets
[) =

k
[k)k[)
_
[

k)

k[) dk
=
_
(k)[

k) dk (3.41)
unity operator

1 =
_
[

k)

k[ dk (3.42)

[) =
_

k)

k[) dk
=

k) = (k

k) (3.43)

_
[

k)
_
are not properly normalized and are not elements of H!
Mathematically, this fact causes some trouble. Practically, one can
work with these states in a very similar fashion as with elements of H:
3.1 Mathematical framework 57
discrete representation continuous representation

k
[k)k[ =

1
_
[k)k[ dk =

1
k[k

) =
kk
k[k

) = (k k

)
(omit tilde from now on)
There are cases where a basis is partly discrete and partly continuous:

1 =

k
[k)k[ +
_
[k)k[ dk
condensed notation (sometimes used):

_
[k)k[ dk =

1
k[k

) = (k, k

) (3.44)
scalar product in continuous representation
[) =
_
dk [k)k[) =
_

(k)(k) dk
note that [), [) H , but [k) , H!
b) Representation of operators

A =

1

A

1 =
_
[k)k[

A[k

)k

[ dkdk

=
_
[k)

A(k, k

)k

[ dkdk

(3.45)
generalized matrix elements A(k, k

) = k[

A[k

)
k[

A[) =
_
k[

A[k

)k

[) dk

=
_
A(k, k

) (k

) dk

(3.46)
[

A[) =
_

(k)

A(k, k

)(k

) dkdk

(3.47)
58 Formalism of Quantum Mechanics
c) Eigenvalue equations
Discrete spectrum (as before, but this time wed like to represent the
eigenvalue problem with respect to continuous basis)

A[u
a
i
) = a
i
[u
a
i
) , ([u
a
i
) u
a
i
) H ;

A =

A

_
k[

A[k

)k

[u
a
i
) dk

= a
i
k[u
a
i
)

_
A(k, k

)u
a
i
(k

) dk

= a
i
u
a
i
(k) (3.48)
eigenvalue problem becomes an integral equation
special case (but important):
A(k, k

) = f(k)(k k

) +g(k)
d
dk
(k k

)
+ h(k)
d
2
dk
2
(k k

) + . . . (3.49)
insertion into (3.48) yields with standard properties of -function
(and its derivatives)

_
f(k) + g(k)
d
dk
+h(k)
d
2
dk
2
+. . .
_
u
a
i
(k) = a
i
u
a
i
(k) (3.50)
integral eq. reduces to dierential eq.
Continuous spectrum

B[v
b
) = b[v
b
) (3.51)
v
b
[v
b
) = (b b

) (3.52)
this problem cannot be treated with mathematical rigor, as [v
b
) ,
H (one cannot even write down eq. (3.51)), but if we stretch the
rules somewhat we can describe the continuous spectrum in a formally
simple way.
as in the case of the discrete spectrum we can prove that
b R if

B =

B

v
b
[

B[v
b
) = bv
b
[v
b
) = b(b b

)
3.1 Mathematical framework 59
if
_
[v
b
)
_
is complete we have:

B =
_
[v
b
)v
b
[

B[v
b
)v
b
[ dbdb

=
_
[v
b
)b

(b b

)v
b
[ dbdb

=
_
[v
b
)bv
b
[ db (3.53)
(generalized spectral representation)
For an operator with mixed discrete and continuous spectrum we have

B =

i
b
i
[b
i
)b
i
[ +
_
b[b)b[ db

_
b[b)b[ db
(in condensed notation)
our statements about the eigenvalue problems of commutating
operators and about degeneracies translate very directly to the
continuous case.
further reading: [Gro]
d) Eigendierentials (Weyl)
One option to deal with the continuous spectrum in terms of legitimate
Hilbert space vectors is the concept of eigendierentials. They are the
formal analogies of wave packets:
[v
b,b
)
1

b
_
b+b
b
[ v
b
) db

(3.54)
with [ v
b
) = lim
b

0
[v
b
)

These kets are normalized properly:


v
b,b
[v
b,b
) =
1
b
_
b+b
b
db

_
b+b
b
db

v
b
[ v
b
)
. .
=(b

)
=
1
b
_
b+b
b
db = 1
Eigendierentials are approximate eigenvectors in the continuous spectrum.
further reading: (e.g.) [Gre] chap. 5
60 Formalism of Quantum Mechanics
Literature on the mathematical framework of QM:
a) QM textbooks [Bal], [Gri], [Jel], [Mes] and [Sha]
b) Mathematics for physicists [Arf] and [FK]
c) Mathematical textbooks [Gro] and [Heu]
3.2 From wave mechanics to QM
Wed like to express our wave mechanics of chapter 2 in terms of our new math-
ematical language. We are thereby led to the conclusion that wave mechanics
is nothing else but the representation of a more abstract theory in conguration
(sometimes called position) space. We have already seen that we can switch from
conguration space to momentum space via Fourier transformation. In our new
language this is nothing else but the special case of a unitary change of represen-
tation (cf. chapter 3.1.3). Our nal goal will be to peel o any representation
from quantum theory (cf. chapter 3.3)
But rst, lets revisit wave mechanics (in the one-dimensional world)
3.2.1 State
is characterized by quadratically integrable wave function (x)
According to chapter 3.1.6 we can write
(x) = x[) (3.55)
(x) is the coordinate space (conguration/position space) representation of
[) H
[) =
_
[x)x[) dx =
_
[(x)[
2
dx = 1
with [(x)[
2
= [x[)[
2
: probability density (cf. chapter 2.3).
Each state vector can be expanded in the continuous basis:
[) =

1[) =
_
[x)(x) dx (3.56)
Momentum space representation (according to eq. (2.64))
p[) = (p) =
1

2
_
(x)e

px
dx
=
_
e

px

2
x[) dx
!
=
_
p[x)x[) dx
3.2 From wave mechanics to QM 61
identify (unitary) transformation matrix
x[p) =
1

2
e
i

px
(3.57)
p[x) =
1

2
e

px
(3.58)
transformation between coordinate and momentum spaces is given by plane
waves (i.e. Fourier transformation)
interpretation: x[p) is the coordinate space representation of the vector [p) , H
etc. What is [p)?
3.2.2 Eigenvalue problems etc.
[p) is the eigenvector of a hermitian operator p with eigenvalue p :
p[p) = p[p) , p = p

(3.59)
similarly : x[x) = x[x) , x = x

(3.60)
normalization:
x[x

) =
_
x[p)p[x

) dp
(3.57)
=
1
2
_
e
i

p(xx

)
dp
= (x x

) (3.61)
p[p

) = . . .
(3.58)
= (p p

) (3.62)
Note that completeness of
_
[x)
_
and
_
[p)
_
is a consequence of the Fourier
theorem.
spectral representation (cf. eq. (3.53))
x =
_
[x)xx[ dx (3.63)
p =
_
[p)pp[ dp (3.64)
change of representation
consider : x

[ p[x) =
_
x

[p

)p

[ p[p)p[x) dpdp

=
1
2
_
e
i

p(p

p)e

px
dpdp

=
1
2
_
pe
i

p(x

x)
dp
=

i
d
dx
(x x

i
d
dx
x

[x) (3.65)
62 Formalism of Quantum Mechanics
(the last equality can be proven by considering, that
d
dx
x

[x) =
d
dx
_
x

[p)p[x) dp )
from eq. (3.65) it follows, that
p[x) =

i
d
dx
[x)
x

[ p[) =
_
x

[ p[x)x[) dx
(3.65)
=

i
_
(x)

(x x

) dx
=

i
d
dx

(x

)
= p
(x

)
(x

)
the coordinate space representation of vector p[) is given by

i
d
x
(x

);
a result that we know from wave mechanics (write p
(x

)


i
d
x
). Similarly
one nds: x

[ x[) = x

(x

)
3.2.3 Commutators
In coordinate space representation we have
( x p
(x)
p
(x)
x)(x) =

i
(x d
x
d
x
x)(x)
= i(x)
= [ x, p
(x)
] = i
but this result can be proven to be independent of any representation:
x

[ x p p x[) = x

[ x p[) x

[ p x[)
(3.60)
= x

p
(x

)
(x

)
_
x

[ p[x)x[ x[) dx
(3.59),(3.65)
= x

p
(x

)
(x

)

i
d
dx

(x

(x

))
= . . . = i(x

)
= ix

[)
extract:
[ x, p] = i (3.66)
Interestingly, one can change the line of argumentation and can postulate the
fundamental commutator eq. (3.66) for hermitian operators x, p.
3.2 From wave mechanics to QM 63
Without any further physical ingredients one can prove the following
x, p have continuous spectra (eq. (3.63), (3.64))
[x), [p) are normalized with respect to -functions (eq. (3.61), (3.62))
coordinate space representation of p[)
momentum space representation of x[)
plane waves form the unitary transformation (eq. (3.57), (3.58))
stationary Schrodinger equation
uncertainty relation
see, e.g. [Mes] QM I, chap. 8.1; [Fic], chap. 4.1
Lets consider the 6
th
item on the list
3.2.4 Stationary Schrodinger equation
From the time-dependent Schrodinger equation (SE) we derived eq. (2.86), the
stationary SE
_


2
2m
d
2
x
+V (x)
_
(x) = E(x) (3.67)
Lets show, that eq. (3.67) is identical to the coordinate space representation of
the abstract eigenvalue problem

H[) = E[) (3.68)


with

H =
p
2
2m
+V ( x) (we assume

V = V ( x) and

H =

H

)
proof : x[

H[) = x[
p
2
2m
+V ( x)[)
=
1
2m
_
x[ p
2
[x

)(x

) dx

+
_
x[V ( x)[x

)(x

) dx

(cf. eq. (3.49, 3.50))


1
=

2
2m
_

(x x

)(x

) dx

+
_
V (x

)(x x

)(x

) dx

1
similarly to eq. (3.65) one nds
x[ p
2
[x

) =
1
2
_
p
2
e
i

p(xx

)
dp =
2

(x x

)
and p
2
[x

) =
2
d
2
x
[x

)
64 Formalism of Quantum Mechanics
=
_


2
2m
d
2
x
+V (x)
_
(x)
rhs
= Ex[) = E(x) q.e.d.
In chapter 2.4 we learned that the eigenvalues E of eq. (3.67) (or eq. (3.68))
are the possible energies that one can measure
2
. In the next section, we will
elevate this statement to a postulate for all hermitian operators!
SE in momentum space
starting point: abstract eigenvalue eq. (3.68):
p[

H[) = Ep[)
lhs : =
_
p[
p
2
2m
+

V [p

)(p

) dp

=
1
2m
_
p

2
(p p

)(p

) dp

+
_
p[V [p

)(p

) dp

p
2
2m
(p) +
_
V (p, p

)(p

) dp

= E(p) (3.69)
SE in momentum space is not a dierential but an integral equation (in
general more dicult to solve)
if

V = V ( x):
V (p, p

) p[V [p

) =
_
p[x)V (x)x[p

) dx
=
1
2
_
V (x)e
i

(pp

)x
dx (Fourier integral)
(This expression plays an important role in scattering theory when one
considers elastic scattering, e.g., from a Coulomb potential, which leads to
Rutherfords cross section formula; cf. [LL], 137)
Question: What about the time-dependent SE?
this eq. cannot be derived form commutators etc., but has to be
postulated as an independent axiom of QM (see next section)
2
and that, in general, the spectrum of

H may be (at least partly) discrete (i.e. energy levels
are quantized).
3.3 The postulates of Quantum Mechanics 65
3.3 The postulates of Quantum Mechanics
Wed like to summarize the fundamental statements (that we have already en-
countered) as 6 postulates:
3.3.1 States
are characterized by kets [) H.
(cf. chapter 3.1.1: this is motivated by the principle of superposition)
3.3.2 Observables
Observables are characterized by linear, hermitian operators. Their (real) eigen-
values are the possible outcomes of measurements (spectra can be (partially)
discrete or continuous)
Often, one calls hermitian operators with a complete set of eigenvectors ob-
servables. We will consider only such (hermitian) operators


A =

n
a
n

P
n
(spectral representation, eq. (3.32))
3.3.3 Expectation values
(of observables) (cf. chapter 2.3.4)
dene : A) :=
[

A[)
[)
R (3.70)
if state is normalized, i.e., [) = 1 we can write, e.g.
A) = [

A[) =
_
[x)x[

A[x

)x

[) dxdx

(for a one-dimensional system)


if

A = A( x) x[

A[x

) = A(x

)(x x

)
A) =
_

(x)A(x)(x) dx (3.71)
=
_
A(x)[(x)[
2
dx
(cf. eq. (2.71) for a 3-dimensional system)
66 Formalism of Quantum Mechanics
For N particles we have

1 =
_
[r
1
...r
N
)r
1
...r
N
[ d
3
r
1
...d
3
r
N
(3.72)
and for

A = A(r
1
...r
N
) we obtain similarly to eq. (3.71)
A) =
_

(r
1
...r
N
)A(r
1
...r
N
)(r
1
...r
N
) d
3
r
1
...d
3
r
N
(3.73)
=
_
A(r
1
...r
N
)[(r
1
...r
N
)[
2
d
3
r
1
...d
3
r
N
For

A p we obtain (for one particle again to facilitate the notation)
p) =
_
[p)p[ p[p

)p

[) d
3
pd
3
p

=
_

(p)p(p) (3.74)
(this is eq. (2.72))
or:
=
_
[r)r[ p[r

)r

[) d
3
rd
3
r

cf. eq. (3.65)


=

i
_

(r)
_
_

r
(r

r)(r

) d
3
r

_
d
3
r
=
_

(r)

i
(r) d
3
r (3.75)
(this is eq. (2.76))
In the same fashion we can re-derive eq. (2.78) etc.
With respect to

As eigenbasis we have
A) =

n
a
n
[a
n
[)[
2
, (compare to eqs. (2.68), (2.72)) (3.76)
(follows from spectral representation of

A)
interpretation: a
n
are the possible outcomes of measurements (postu-
late 3.3.2), and
n
= [a
n
[)[
2
R are the corres-
ponding probabilities
Note that
n
= [a
n
)a
n
[) = [

P
n
[) = P
n
)
one calls a
n
[) C probability amplitudes
if [) [a
m
) A) = a
m
,
m
= 1
3.3 The postulates of Quantum Mechanics 67
this is a certain measurement.
In this case one can show that
(A)
2
:= (A A))
2
) = A
2
) A)
2
= 0
and vice versa (which is quite evident)
(for details see [Bloc], chap. 7.2)
3.3.4 Measurements
We cannot delve into the conceptual diculties of this issue (which is controversial
to the present day). For laboratory measurements the Kopenhagen interpreta-
tion has been and still is successful in (almost) all cases, although its foundations
are still under discussion.
Literature: [Aud] and [Omnb]
More technical accounts: [dE] and [Omna]
From postulates 3.3.1 - 3.3.3 we know the following
[a
m
)
measure

A

[)
measure

A
[a
m
)
measure

B

[b
n
)
The measurement of

A changes (disturbs) the state [). After the measurement
the system is characterized by the eigenstate [a
m
). If we measure

A again we re-
obtain a
m
because this is then a certain measurement. If we measure

B we disturb
the system again (in general) and change the state to one of

Bs eigenstates. One
calls these perturbations or changes
reduction (collapse) of the wave function.
There is no element in the theory that tells us which eigenvector is chosen,
we can only calculate (know) the corresponding probability; the collapse itself is
indeterministic (which many physicists nd unsatisfactory)
We can think of a measurement as a projection onto the corresponding eigen-
state and re-normalization of the state because after the measurement the system
is in the eigenstate with probability one!
68 Formalism of Quantum Mechanics
3.3.5 Commutation relations
As mentioned in chapter 3.2.3 we postulate fundamental commutation relations
(which cannot be derived, but which rest on plausibility arguments and corre-
spondence rules)
a) Important commutation relations
1) The fundamental commutator for one particle in one dimension is eq.
(3.66)
[ x, p] = i
2) One (spinless) particle in three dimensions
[ x
i
, p
j
] = i
ij
(3.77)
[ x
i
, x
j
] = [ p
i
, p
j
] = 0 (3.78)
3) We can then consider operator functions

F = F( x
i
, p
j
)
one can show : [

F, p
j
] = i


F
x
j
(3.79)
(using eq. (3.77), (3.78))
proof : see [Bloc] [

F, x
j
] =

i


F
p
j
(3.80)
4) Example for an operator function:
angular momentum operator dened by

l := r p (3.81)
(

l
i
=

l

i
, i = 1, 2, 3)
we can calculate various commutators on the basis of eq. (3.79) and
(3.80), e.g.
[

l
3
, x
3
] = [

l
3
, p
3
] = 0
[

l
3
, x
1
] = i x
2
[

l
3
, p
1
] = i p
2
3.3 The postulates of Quantum Mechanics 69
and
[

l
i
,

l
j
] = i
3

k=1

ijk

l
k
(3.82)
with
ijk
=
_
_
_
1 cyclic permutation (ijk)
1 anticyclic permutation (ijk)
0 two like indices (ijk)
and
[

l
2
,

l
i
] = 0
The commutators (3.82) are the basis of angular momentum algebra
and can be used as denition of angular momentum in QM (see chapter
4.3).
5) (electron) spin (chapter 4.4.4)
dene operator s = ( s
1
, s
2
, s
3
) , s
i
= s

i
i
and require [ s
i
, s
j
] = i

ijk
s
k
and [ s
i
, x
j
] = [ s
i
, p
j
] = 0 , i, j
together with eq. (3.77) and eq. (3.78) these relations form the basis
for the discussion of a one-particle system with spin.
b) General uncertainty relation
The commutator relations are intimately related to the (general) uncer-
tainty relation(s). For hermitian operators we prove
A B
1
2
[[

A,

B])[ (3.83)
two non-commuting observables cannot be measured simultaneously with
arbitrary accuracy.
proof: the square variations are dened by
(A)
2
= (A A))
2
) , (B)
2
= (B B))
2
)
in addition, we dene (hermitian) operators
a :=

A A)

b :=

B B)

1
70 Formalism of Quantum Mechanics
and consider
(A)
2
(B)
2
= [ a
2
[)[

b
2
[)
= a[ a)

b[

b)
[ a[

b)[
2
Schwarz inequality, chapter 3.1.1
= [[ a

b[)[
2
a

b =
a

b +

b a
2
. .
+ i
a

b a
2i
. .
hermitian hermitian
[ a

b[) = [
a

b +

b a
2
[)
. .
R
+ i [
a

b a
2i
[)
. .
R
[[ a

b[)[
2
=
_
[
a

b +

b a
2
[)
_
2
+
_
[
a

b a
2i
[)
_
2

_
[
a

b a
2i
[)
_
2
=
1
4
[[[ a,

b][)[
it is straightforward to show that [ a,

b] = [

A,

B] q.e.d.
Comments:
1. [) H : AB =
1
2
[[

A,

B])[ (minimal uncertainty)
In this case it follows from Schwarzs (in-)equality that
a[) = c

b[) , (c C)
additionally, we require
[ a

b +

b a[) = 0
c

b
2
[) +c[

b
2
[) = 0
1 c
_
[

b
2
[)
_
= 0
= c is imaginary!
note that
[

b
2
[) = 0

B[) = B)[)
in this case we have B = AB = 0 (certain measurement)
3.3 The postulates of Quantum Mechanics 71
2. [ x, p] = i
= px

2
(3.84)
3. [

l
x
,

l
y
] = i

l
z
= l
x
l
y


2
[l
z
)[ (3.85)
4. in general: Two Operators

A,

B are called complementary operators:
[

A,

B] ,= 0
3.83
= they cannot be measured simultaneously with certainty! What can we
say about measurements of complementary operators: (cf. page 67)
state [)

A
[a
m
)

B
[b
n
)

A
[a
k
)
measured value a
m
b
n
a
k
probability [a
m
[)[
2
[a
m
[b
n
)[
2
[a
k
[b
n
)[
2
measurement of

A is not necessarily reproduced!
5. Two operators

A,

B are called compatible operators:
[

A,

B] = 0 common set of eigenvectors (cf. chapter 3.1.5)
measurements:
(a) no degeneracy
state [)

A
[a
m
)

B
[a
m
)

A
[a
m
)
measured value a
m
b
m
a
m
probability [a
m
[)[
2
1 1

these are certain measure-
ments as state is an eigenstate
(b) degeneracy (of

A)
state [)

A

_
[a

m
), = 1, ...
_

B
[a

m
)

A
[a

m
)
i.e.
3
[

) =

c
m

[a

m
)
measured value a
m
b

m
a
m
probability

[a

m
[)[
2
[a

m
[

)[
2
1
= [c

m
[
2
72 Formalism of Quantum Mechanics
This scheme leads us to
6. Complete set of compatible observables
measure compatible operators (with degenerate spectra) until the di-
mension of their simultaneous eigenspace is 1 in this case the state of
the system is completely determined
[) = [abc...)


quantum numbers

(i.e. eigenvalues of
complete set of observables characterize state)

abc...
_
[abc...)abc...[ =

1
The number of operators which are needed to determine the state of the
system completely is related to the degrees of freedom of the system.
e.g. - one particle in one dimension: one observable ( x or p)
suces to determine the state
- one particle in three dimensions: need three compatible observables
(e.g. x
1
, x
2
, x
3
; or (see later)

H,

l
2
,

l
z
)
7. What about the energy-time uncertainty relation?
E

2
(3.86)
it does not follow from a commutator via eq. (3.83)
because there is no time operator in (textbook) QM
eq. (3.86) has to be derived and interpreted dierently
(see chapter 3.4)
3
c

m
can be determined in the following way:
[

) = C

[a

m
)a

m
[)

) = 1 = C
2

[a

m
[)[
2
c

m
Ca

m
[) =
a

m
[)
_

[a

m
[)[
2
3.4 Equations of motion 73
3.3.6 Dynamics
The last postulate which completes the formal scheme of QM is about the time
development of quantum systems.
We postulate that the dynamics is governed by the time-dependent Schrodinger
equation
id
t
[(t)) =

H[(t)) (3.87)
[(t
0
)) = [
0
)
(

H may or may not depend on time)
The time t in this equation is the usual classical time, i.e., it is what we measure
with a classical clock; it is not an observable of the quantum system!
Note that eq. (3.87) boils down to eq. (2.40) or eq. (2.42) in coordinate space.
in one dimension:
ix[d
t
(t)) = id
t
x[(t))
= i
t
(x, t)
= x[

H[(t))
p. 63
=
_


2
2m
d
2
x
+V (x)
_
(x, t)
in momentum space we have: (cf. eq. (3.69))
ip[d
t
(t)) = i
t
(p, t)
= p[

H[(t))
=
p
2
2m
(p, t) +
_
V (p, p

)(p

, t) dp

3.4 Equations of motion


We elaborate on the last topic: the time-development of quantum systems. The
time-dependent Schrodinger equation (TDSE) is an initial value problem and is
linear, i.e.
[(t
0
))
TDSE
[(t))
Lets characterize this linear map.
74 Formalism of Quantum Mechanics
3.4.1 Time-development (evolution) operator
Dene

U(t, t
0
) via
[(t)) =

U(t, t
0
)[(t
0
)) (3.88)
due to [(t
0
)) =

U(t
0
, t
0
)[(t
0
)), it follows, that

U(t
0
, t
0
) =

1 (3.89)
a) Dierential equation
TDSE eq. (3.87):
id
t
[(t))
(3.88)
= i
_
d
dt

U(t, t
0
)
_
[(t
0
))
=

H(t)

U(t, t
0
)[(t
0
))
id
t

U(t, t
0
) =

H(t)

U(t, t
0
) (3.90)

U(t
0
, t
0
) =

1
equivalent with TDSE!
b) Equivalent integral equation

U(t, t
0
) =

1
i

_
t
t
0

H(t

U(t

, t
0
) dt

(3.91)
c) Some properties (without proofs)
(i)

U(t, t

) =

U(t, t

U(t

, t

) (3.92)
(group property)
(ii)

U

(t, t
0
) =

U
1
(t, t
0
) =

U(t
0
, t) (3.93)
i.e.

U unitary

H hermitian
(t)[(t)) = (t
0
)[

(t, t
0
)

U(t, t
0
)[(t
0
))
= (t
0
)[(t
0
))
norm conservation (cf. chapter 2.3.2)
(iii) For stationary systems (
t

H = 0)


U(t, t
0
) = exp
_

H(t t
0
)
_
(3.94)
3.4 Equations of motion 75
(check it by applying eq. (3.90))
in this case
[(t)) = exp
_

H(t t
0
)
_
[(t
0
))
if

H[(t
0
)) = E[(t
0
))
i.e. [(t
0
)) [
E
(t
0
)) (eigenvector of

H)
[
E
(t)) = e

E(tt
0
)
[
E
(t
0
)) (3.95)
[
E
(r, t)[
2
= [r[
E
(t))[
2
= [r[
E
(t
0
))[
2
= [
E
(r, t
0
)[
2
no change in probability density (stationary state)
This scheme, together with the postulates of the the last chapter, are sum-
marized as formulation of QM in the so-called
Schrodinger picture
characteristics: states obey TDSE
observables can depend on time only explicitly
(d
t

A =
t

A); e.g. time-dependent EM eld
Since unitary transformations do not change scalar products and eigenvalue
spectra, they do not change the physical contents. Hence, we can apply a
(time-dependent) unitary transformation to the states and operators of the
Schrodinger picture in order to formulate QM in a dierent picture !
3.4.2 Heisenberg picture
dene Heisenberg state [
H
) via
[
H
) =

U

(t, t
0
)[
S
(t)) (3.96)


Schr odinger state

that
fullls TDSE eq.(3.87)
=

U(t
0
, t)[
S
(t)) = [
S
(t
0
))
= d
t
[
H
) = 0 (3.97)
transformation of operators (cf. eq. (3.23))

A
H
(t) =

U

(t, t
0
)

A
S
(t)

U(t, t
0
) (3.98)
76 Formalism of Quantum Mechanics
Consequence:
A) =
S
(t)[

A
S
(t)[
S
(t))
use inverse transform.
=
=
H
[

(t, t
0
)

U(t, t
0
)

A
H
(t)

(t, t
0
)

U(t, t
0
)[
H
)
=
H
[

A
H
(t)[
H
) (3.99)
invariant as required!
a) Equation of motion
id
t

A
H
(t) = id
t
_

(t, t
0
)

A
S
(t)

U(t, t
0
)
_
(3.90)
=

U

(t, t
0
)

H
S
(t)

A
S
(t)

U(t, t
0
)
+

U

(t, t
0
)

A
S
(t)

H
S
(t)

U(t, t
0
)
+ i

(t, t
0
)
_

t

A
S
_

U(t, t
0
)
=

U

(t, t
0
)
_
[

A
S
(t),

H
S
(t)] + i
t

A
S
_

U(t, t
0
)
(3.98)
= [

A
H
(t),

H
H
(t)] +i
t

A
H
(where we have dened
t

A
H
=

U

(t, t
0
)
t

A
S

U(t, t
0
))
Note that commutators are invariant under transformation
4
Heisenberg equation
i
d
dt

A
H
(t) = [

A
H
(t),

H
H
(t)] + i
t

A
H
(t) (3.100)
initial condition :

A
H
(t
0
) =

A
S
(t
0
)
characteristics of Heisenberg picture: states are time-independent
operators obey Heisenberg eq.
b) Eigenvalue problem
Schrodinger picture :

A
S
[a
S
i
) = a
i
[a
S
i
)
Heisenberg picture : [a
H
i
(t)) =

U

(t, t
0
)[a
S
i
)


A
H
(t)[a
H
i
(t)) =

U

(t, t
0
)

A
S
[a
S
i
)
= a
i
[a
H
i
(t))
4
[

A
S
,

B
S
] = 0 [

A
H
,

B
H
] = 0
[A
S
, B
S
] =

C
S
[A
H
, B
H
] =

C
H
3.4 Equations of motion 77
same eigenvalues a
i
as required!
c) Transition probabilities

i
= [a
S
i
[
S
(t))[
2
= [a
H
i
(t)[
H
)[
2
(invariant!)
d) Analogy with classical mechanics
Schrodinger picture is advantageous for practical calculations
Heisenberg picture is advantageous for formal development; e.g. it
shows nicely the formal correspondence between QM and classical me-
chanics
id
t
x
H
j
(3.100)
= [ x
H
j
,

H
H
] = i


H
H
p
H
j
, (j = 1, 2, 3)
We obtain a similar eq. for the momentum operator
Summary:
d
t
x
H
j
=


H
H
p
H
j
(3.101)
(correspond to Hamilton eqs.
of classical mechanics)
d
t
p
H
j
=


H
H
x
H
j
note that there is a further correspondence: if one introduces Poisson brack-
ets f, g in classical mechanics
f, g :=

i
_
f
x
i
g
p
i

f
p
i
g
x
i
_
classical eq. of motions take the form
d
dt
A
kl
(x
1
...x
N
; p
1
...p
N
; t) = A
kl
, H
kl
+
A
kl
t
, (cf. eq. (3.100))
moreover : x
i
, x
j
= p
i
, p
j
= 0
x
i
, p
j
=
ij
extract correspondence rule (rst observed by Dirac) for the transition
from classical to quantum mechanics
A
kl
, B
kl

1
i
[

A,

B]
e.g. x
i
, p
j

1
i
[ x
i
, p
j
]
(canonical quantization)
78 Formalism of Quantum Mechanics
e) Constants of motion
dened by d
t

A
H
(t) = 0
if
t

A = 0 (which is often the case)
d
t

A
H
= 0 [

A
H
,

H
H
] = 0 (3.102)
[

A
S
,

H
S
] = 0
constants of motion commute with the Hamiltonian
in this case :

A[a
i
) = a
i
[a
i
)


good quantum number

(i.e.

certain measurement

)
Examples:
(i)

H =
p
2
2m
(free particle)
[ p,

H] = 0 momentum conservation
but :
(ii)

H =
p
2
2m
+V ( x)
[ p,

H] = [ p, V ( x)]
(3.80)
=

i

V
x
,= 0
f) Expectation values and Ehrenfests theorems
Lets consider eq. of motion for expectation values (can also be derived in
Schrodinger picture)
d
dt
A) =
H
[d
t

A
H
[
H
)
(3.100)
=
1
i

H
[[

A
H
,

H
H
][
H
) +
H
[
t

A
H
[
H
)
=
1
i
[

A
H
,

H
H
]) +
t
A
H
)
Since this eq. holds also in Schrodinger picture we can write
d
dt
A) =
1
i
[

A,

H]) +
t
A) (3.103)
(same structure as eq. (3.100))
3.4 Equations of motion 79
in particular we nd
d
dt
x
i
) =
1
i
[ x
i
,

H])
(3.79)
=


H
p
i
) (3.104)
d
dt
p
i
) =
1
i
[ p
i
,

H]) =


H
x
i
) (3.105)
Ehrenfests Equations
Do they imply that the expectation values of position and momentum move
classically? Not quite!
Examples:
(i)

H =
p
2
2m
d
dt
x) =
p)
m
;
d
dt
p) = 0
x)(t) =
p)
m
t +x
0
)
classical motion!
(ii)

H =
p
2
2m
+
m
2

2
x
2
(harmonic oscillator)
d
dt
x) =
p)
m
;
d
dt
p) = m
2
x)

d
2
dt
2
x) +
2
x) = 0
(classical motion!)
(iii) but:

H =
p
2
2m
+V ( x)
in general
d
dt
p) =
V
x
) = F(x)) ,= F(x))
only if the latter inequality becomes an equality (free particle, har-
monic oscillator) do x), p) obey classical equations of motion!
g) Energy-time uncertainty relation (cf. page 72)
We have the general uncertainty relation eq. (3.83), and, in particular:
A E
1
2
[[

A,

H])[
80 Formalism of Quantum Mechanics
for any observable
If
t

A = 0 we can use eq.(3.103) and obtain
A E

2
[d
t

A)[
J ,J
,)
)
)
Dene time interval via
A

= [
d
dt

A)[
= E

2
(3.106)
Interpretation: properties of a system change noticeably during only if


2E
(stationary states: E = 0 )
application: lifetimes (e.g. of excited states) + energy widths (line widths)
3.4.3 Interaction picture
dened as lying in between Schrodinger and Heisenberg pictures
S-picture I-picture H-picture
[
S
(t))

0
(t,t
0
),

H
0
[
I
(t))

I
(t,t
0
),

V
[
H
)

A
S

A
I
(t)

A
H
(t)

(t,t
0
),

H

we consider decomposition

H =

H
0
+

V (3.107)

U(t, t
0
) =

U
0
(t, t
0
)

U
I
(t, t
0
) (3.108)
eqs. of motion : id
t

U
0
(t, t
0
) =

H
0

U
0
(t, t
0
)

U
0
(t
0
, t
0
) =

1
id
t

U(t, t
0
) =

H

U(t, t
0
)

U(t
0
, t
0
) =

1
3.4 Equations of motion 81
Use them in eq. (3.108) (which denes

U
I
):
id
t

U(t, t
0
) = i(d
t

U
0
)

U
I
+

U
0
id
t

U
I
=

H
0

U
0

U
I
+

U
0
id
t

U
I
=

H

U
0

U
I
id
t

U
I
(t, t
0
) =

U

0
(t, t
0
)(

H

H
0
. .
=

V
)

U
0
(t, t
0
)

U
I
(t, t
0
)
dene, in general, transformed operators

A
I
(t) =

U

0
(t, t
0
)

A
S
(t)

U
0
(t, t
0
) (3.109)
and states
[
I
(t)) =

U

0
(t, t
0
)[
S
(t)) (3.110)
= id
t

U
I
(t, t
0
) =

V
I
(t)

U
I
(t, t
0
) (3.111)

U
I
(t
0
, t
0
) =

1 (3.112)

U
I
can be interpreted as evolution operator in interaction picture
moreover : [
I
(t)) =

U

0
(t, t
0
)[
S
(t))
=

U
I
(t, t
0
)[
H
)
!
=

U
I
(t, t
0
)[
I
(t
0
)) (3.113)
( [
I
(t
0
)) [
S
(t
0
)) [
H
) )
Dierentiate eq. (3.113):
id
t
[
I
(t)) =
_
id
t

U
I
(t, t
0
)
_
[
I
(t
0
))
=

V
I
(t)

U
I
(t, t
0
)[
I
(t
0
))
= id
t
[
I
(t)) =

V
I
(t)[
I
(t)) (3.114)
Schrodinger-type equation in interaction picture
Similarly, we obtain an equation of motion for operators (3.109)
id
t

A
I
(t) = [

A
I
,

H
0
I
] + i
t

A
I
(3.115)
82 Formalism of Quantum Mechanics
Characteristics of interaction picture:
states and operators are time-dependent
states are driven by

V
I
(eq. (3.114))
operators are driven by

H
0
I
(eq. (3.115))
Literature: a thorough discussion of S-, I- and H-pictures can be found in:
[Fic], chap. 3.5
3.5 Symmetries
Ein Ding ist symmetrisch, wenn es eine Moglichkeit gibt, es zu verandern, und
es hinterher doch wieder so aussieht wie vorher.
H. Weyl nach R.P. Feynman, Vom Wesen physikalischer Gesetze, Kapitel 4
translation:
symmetry transformation: operation which leaves some
mathematical expressions or
laws invariant
classical physics: symmetry transformation
Noether theorem
conservation laws
(i.e. constants of
motion)
important examples: + special cases
invariance with respect to:
translations momentum conservation
rotations angular momentum conservation
translations in time energy conservation
we will nd analogous relations in QM!
3.5.1 Representation of symmetry transformations
We restrict the discussion to symmetry transformations which are mediated by
linear operators (the other important case which we omit are transformations
that are expressed by anti-linear operators).
a) Transformation of states
[

) =

S[) (3.116)
3.5 Symmetries 83
require norm conservation:

) = [


S[) = [)
=

S

=

S
1
linear symmetry transformations are represented by unitary operators
(cf. page 47)
b) Transformation of operators

A =

S

A

(3.117)
such that

A[

) = [

A[)
Consequences:
1.

A[a
n
) = a
n
[a
n
)

A[ a
n
) =

S

A


S[a
n
) = a
n

S[a
n
) = a
n
[ a
n
)
(same eigenvalues)
2.
n
= [a
n
[)[
2
= [ a
n
[

)[
2
(same transition probabilities)
i.e. if everything is transformed simultaneously, physics remains unchanged!
c) Some properties of symmetry operators

S
1. they can be written as

S = e
i

O
(3.118)

S
1
=

S



O =

O

(proof: [Bloc], chap. 9)


= [

S,

O] = 0
2.

S[
n
) = e
i

O
[
n
) = e
i
n
[
n
) , (
n
R)
i.e. common eigenvectors, and eigenvalues of

S are of modulus 1
3. Symmetry transformations (and - apart from exceptions - operators

S) form (not necessarily abelian) groups


group theory is a powerful approach to the investigation of sym-
metries (see, e.g., [Mes] QM II, App. D and chap. 15)
84 Formalism of Quantum Mechanics
3.5.2 Symmetries and constants of motion
a) Denition: operator

A is called symmetrical (invariant) with respect to

S
:

A =

S

A

=

A
[

A,

S] = 0 (3.119)
[

A[) = [

A[)
= [

S

A

[)
=

A[

)
( with [

) =

S

[) )
i.e. transformation of

A yields same expectation value as inverse transfor-
mation of state
b) Invariance of the TDSE (3.87)
transformation [

) =

S[)
consider id
t
[

) = id
t
(

S[))
= i(
t

S)[) +

Si
t
[)
= (i
t

S +

S

H)[)
invariance means that :
!
=

H[

) =

H

S[)
(i.e. transformed state fullls TDSE with the same Hamiltonian!)
i
t

S + [

S,

H] = 0 (3.120)
Note that this equation holds also in the Heisenberg picture (although we
have derived it in the Schrodinger picture). In fact, one can extract it also
from the requirement that the Heisenberg equation (3.100) be invariant
with respect to

S.
Now, lets compare eq. (3.120) with standard Heisenberg equation (3.100)
for

S.
= invariance of equation of motion d
t

S
H
= 0 (3.121)
3.5 Symmetries 85
Note that

S is no observable; but

O (eq. (3.118)) is, and it is straightforward
to show that d
t

S
H
= 0 implies d
t

O
H
= 0
If TDSE is invariant with respect to transformation

S
hermitian operators

O which are dened via

S are
constants of motion
5
in closed systems, we have
t

O =
t

S = 0


O is constant of motion [

O,

H] = 0 = [

S,

H] (3.122)
[

O,

U(t, t
0
)] = 0
(cf. eq. (3.102))
c) Recipe for discussion of stationary problems
(a) Write down

H
(b) Find symmetry transformations via [

S,

H] = 0
(c) Find generators

O (which are constants of motion)
(d) Find simultaneous eigenstates of

S,

O,

H
3.5.3 Translations (1-d)
Lets consider translations in real space as an example of a continuous symmetry
transformation.
Obviously we have:
translation
1
+ translation
2
= translation
1
+
2
= translation
2
+ translation

1
translations form a continuous abelian group!
translation operator

T(
1
)

T(
2
) =

T(
2
)

T(
1
) =

T(
1
+
2
) (3.123)
eq. (3.123) is solved by

T() = e
i

G
, (

G =

G

) (3.124)
Innitesimal translation

T() =

1 + i

G


G = i

T()

T(0)

= i
d

T
d

=0
(3.125)
5

O is called generator of symmetry transformation
86 Formalism of Quantum Mechanics
Action on states (this is an obvious postulate)

T()[x) = [x +) = [ x) (3.126)

()[x) = [x ) (3.127)


(x) = x[

) = x[

T()[) =

()x[)
= x [) = (x ) (3.128)
Transformed wave function at x = original wave function at inversely
transformed position x
Action on x

[ x[

) =

T[ x[

T) = [

T[)
=
_

[x)x[ x[x

)x

) dxdx

=
_

(x)x

(x) dx
(3.128)
=
_

(x )x(x ) dx
xx
=
_

(x)(x +)(x) dx
= [ x +

1[)
compare:

() x

T() = x +

1 (3.129)

T() x

() = x

1 (3.130)
Consider innitesimal transformation

T() x

() = (

1 + i

G) x(

1 i

G)
= x +i[

G, x] +O(
2
)
(3.130)
= x

1
= [

G, x] = i (3.131)
Dene:
p =

G , ( p has dimension of momentum)


=

T() = e

p
(3.132)
[ x, p] = i
3.5 Symmetries 87
i.e. one can derive the basic commutator [ x, p], which was postulated in
chapter 3.3.5 from consideration of symmetry transformation
p
(x)
(x) = x[ p[) =
_
x[ p[x

)(x

) dx
(3.125)
= i
_
x[
d

T
d

=0
[x

)(x

) dx

= i lim
0
_
x[

T()

T(0)

[x

)(x

) dx

= i lim
0
_
x[x

+) x[x

(x

) dx

= i
_
d
x
(x x

)(x

) dx

= id
x
(x)
standard form of p
(x)
is recovered from denition and properties of transla-
tions.
Some comments:
1. Structure of theory
1
st
step: consider symmetry transformations; i.e. dene and investigate
translations
derive [ x, p] = i
further consequences (see chapter 3.2.3)
2. if [

H,

T()] = 0 (translational invariance)
= momentum conservation (cf. eq. (3.122))
e.g.

H =
p
2
2m
[

H,

T] = [

H, p] = 0

H =
p
2
2m
+

V (x) [

H,

T] = [

V ,

T] ,= 0
3.5.4 Other continuous symmetry transformations
a) Rotations
about a xed axis abelian group
e.g.

R
z
() = e

l
z
(3.133)
88 Formalism of Quantum Mechanics
one nds that generator can be identied with corresponding compo-
nent of angular momentum operator
if [

H,

R
z
()] = 0

l
z
is conserved!
i.e. rotational invariance angular momentum conservation
general rotations:
more dicult!
But: each rotation can be composed from three (non-commuting) ro-
tations about Euler angles (cf. Classical Mechanics)
One nds correspondingly:


R() = e

l
z
e

l
y
e

l
z
(3.134)
and
[

l
i
,

l
j
] = i

ijk

l
k
, (cf. eq. (3.82))
as rotations about dierent axes do not commute.
b) Galilei transformations
H
H
5
5
H

J
consider translation of coordinate system
r

= r vt
p

= p mv
corresponding operators (requires some calculation)
r

=

(v)r

(v) = r vt

1 (3.135)
p

=

(v) p

(p) = p vm

1 (3.136)
with

(v) = exp
_
i

v(mr t p)
_
(3.137)
Galilei operator depends on time explicitly!
invariance of TDSE if (cf. eq. (3.120))
i
t

(v) + [

(v),

H] = 0
3.5 Symmetries 89
3.5.5 Discrete symmetry transformations
a) Reections parity
the simplest symmetry group (only two elements) is obtained from denition

S
0
[r) = [ r) (3.138)


S
2
0
[r) = [r)


S
0
=

S
1
0
=

S

0
(3.139)
symmetry operator is unitary and hermitian
apply it to wave function

S
0
(r) = r[

S
0
[) =

S
0
r[) = (r)
eigenvalue equations:


S
0
is unitary: modulus of eigenvalues =

1 (page 83)


S
0
is hermitian: eigenvalues real
eigenvalues 1

S
0
[

) = [

) (3.140)
+ : positive (even)

parity

: negative (odd)

parity

if [

S
0
,

H] = 0 parity conservation
(not fullled in case of weak interactions parity violation in -
decay)
b) Discrete translations Bloch-theorem
Simple model of a crystal: atomic sites are xed and form a periodic lattice;
electrons move independently in a model potential which is also periodic
task: solve (stationary) Schrodinger eq. for an electron in periodic potential
V (r +t) = V (r) (3.141)
t is an arbitrary lattice-translation vector (see [Bloc], chap. 10.5)
discrete translation (cf. eq. (3.126), (3.127))

T[r) = [r +t) (3.142)


T
(r)
(r) = r[

T[) = (r t)
if V (r) = V (r+t) [

H,

T] = 0
90 Formalism of Quantum Mechanics
search for common eigenstates (r):
eigenvalue eq. of

T (in coordinate space)

T
(r)
(r) = e
ikt
(r) (3.143)
(denes k implicitly)
together with eq. (3.142):
(r t) = e
ikt
(r)
(r) = e
ikt
(r t)
ansatz : (r) = e
ikr
u
k
(r) (3.144)
(r) = e
ikt
e
ik(rt)
u
k
(r t)
u
k
(r) = u
k
(r t) (3.145)
Bloch theorem: Eigenfunctions of periodic systems are of the form
(r) = e
ikr
u
k
(r) with u
k
(r) = u
k
(r t)
(i.e. periodic functions)
Use it in Schrodinger eq.
_


2
2m
+ V (r)
_
(r) = E(r)

2
2m
(+ik)
2
+V (r)
. .

(k)
_
u
n
k
(r) = E
n
(k)u
n
k
(r) (3.146)
the new Hamiltonian

H

depends continuously on k
E
n
(k): dispersion relations of electrons in periodic potential
(cf. dispersion relation of free particles (eq. (2.24)))
band structure
c) Permutations of identical particles
identical particles share all physical properties, e.g. mass, charge; their
permutations are connected with a symmetry property
Consider two identical particles (N = 2)
3.5 Symmetries 91
Preparation:
Some remarks on direct product states and spaces
consider states [
(1)
) H
(1)
[
(2)
) H
(2)
dene direct product:
[
(1)

(2)
) = [
(1)
)[
(2)
) = [
(2)
)[
(1)
)
[
(1)

(2)
) H = H
(1)
H
(2)
scalar product in H:

(1)

(2)
[
(1)

(2)
) =
(1)
[
(1)
)
(2)
[
(2)
) (satises axioms, chapter 3.1.1)
if
_
[
(1)
k
)
_
ONB (orthonormal basis) in H
1
and
_
[
(2)
l
)
_
ONB in H
2
=
_
[
(1)
k

(2)
l
)
_
ONB in H
i.e. [) =

kl
c
kl
[
(1)
k

(2)
l
) for all [) H
if dimensions of spaces are nite
dim H = (dim H
1
)(dim H
2
)
Operators:
if

A
(1)
[
(1)
) = [
(1)
)
then

A
(1)
[
(1)

(2)
) = [
(1)

(2)
)
(i.e.

A
(1)
does not act on [
(2)
))
= [

A
(1)
,

B
(2)
] = 0
but: there are operators that act on both [
(1)
) and [
(2)
)
Dene permutation operator

P
12

P
12
[(1, 2)) = [(2, 1)) (3.147)
(coordinate representation r
1
r
2
[) = (r
1
r
2
) )
obviously

P
2
12
=

1

P
12
=

P
1
12
=

P

12
(3.148)
92 Formalism of Quantum Mechanics
(cf. parity operator eq. (3.139))
eigenvalue problem (similar to the one of

S
0
)

P
12
[
+
(1, 2)) = [
+
(1, 2)) , (

symmetrical

) (3.149)

P
12
[

(1, 2)) = [

(1, 2)) , (

antisymmetrical

) (3.150)
states
All known Hamiltonians commute with

P
12
: [

H,

P
12
] = 0
Moreover, we postulate that all observables commute with all permu-
tation operators

P

of a many-particle system (if particles are identi-


cal)
symmetrization postulate
[

P

,

O] = 0 ,

O,

P

(3.151)
Implication: there is no way to distinguish identical particles in QM
(no observable is sensitive to permutations)
= QM states are either (totally) symmetric or (totally)
antisymmetric with respect to particle exchange, i.e.
either

P

[
+
) = [
+
) for all permutations
or

P
ij
[

) = [

) for all transpositions


This statement is very important and has no classical analogy. It must
be fullled since otherwise the projector [)[ on a non-(anti-)sym-
metric state [) would not commute with all permutation operators
(i.e. would violate eq. (3.151))!
A simple example: two non-interacting particles in common potential

H =

H
(1)
+

H
(2)
=
p
2
1
2m
+V (r
1
) +
p
2
2
2m
+V (r
2
) (3.152)
We assume that we know the single-particle solutions:

H
(1)
[
(1)
k
) =
(1)
k
[
(1)
k
)

H
(2)
[
(2)
l
) =
(2)
l
[
(2)
l
)
it follows that [) = [
(1)
k

(2)
l
) solves

H[) = E[) with E =


(1)
k
+
(2)
l
3.5 Symmetries 93
proof:

H[
(1)
k

(2)
l
) =
_

H
(1)
[
(1)
k
)
_
[
(2)
l
) +[
(1)
k
)

H
(2)
[
(2)
l
)
=
_

(1)
k
+
(2)
l
_
[
(1)
k

(2)
l
)
There is a second eigenstate with the same eigenenergy:
[

) =

P
12
[) = [
(1)
l

(2)
k
)
exchange degeneracy
[), [

) are not symmetric or antisymmetric (i.e. they dont fulll eq.


(3.149), ( 3.150))
(Anti-) symmetric solutions (for l ,= k):
[
+
) =
1

2
_
[) +[

) =
1

2
_
[
(1)
k

(2)
l
) +[
(1)
l

(2)
k
)
_
[

) =
1

2
_
[) [

)
_
=
1

2
_
[
(1)
k

(2)
l
) [
(1)
l

(2)
k
)
_
(3.153)
properly normalized for l ,= k , (
+
[
+
) =

) = 1 )
for l = k we have
[
+
) = [
(1)
k

(2)
k
) , (
+
[
+
) = 1)
[

) = 0 Pauli principle!
Antisymmetric product states
We can rewrite [

) of eq. (3.153) in the form of a determinant:


[

) =
1

[
(1)
k
) [
(2)
k
)
[
(1)
l
) [
(2)
l
)

This can be generalized (N 2):


[

) =
1

N!

[
(1)
k
1
) . . . [
(N)
k
1
)
.
.
.
[
(1)
k
N
) . . . [
(N)
k
N
)

(3.154)
Slater determinant (

) = 1, if

k
i
[
k
j
) =
ij
)
94 Formalism of Quantum Mechanics
determinant is zero if two lines or two columns are identical
Pauli principle: two particles of a system of N
(for antisym- identical particles cannot occupy the
metric states) same single-particle state [
k
i
)!
Even though we have only considered non-interacting systems so far, we
can formulate the Pauli principle for the case of interacting particles in a
similar way:
Lets consider:
Interacting two-particle systems

H =

H
(1)
+

H
(2)
+

W(1, 2) (3.155)
Schrodinger eq.

H[) = E[) cannot be separated, but sym-
metrization postulate (3.151) is still valid.
[

) =

kl
c
()
kl
[
(1)
k

(2)
l
) (3.156)
i.e. the physical solutions are either symmetric or antisymmetric and
can be expanded in the basis
_
[
(1)
k

(2)
l
)
_
(note that [

) is no longer a single Slater determinant)


E.g. in coordinate representation we can write:
r
1
...r
N
[
()
) =

()
(r
1
...r
i
...r
j
...r
N
) =
()
(r
1
...r
j
...r
i
...r
N
)

(r
1
...r
i
...r
i
...r
N
) =

(r
1
...r
i
...r
i
...r
N
) = 0
Pauli principle!
Symmetry considerations show that only (anti-) symmetric many par-
ticle states exist in QM. Do they both exist in nature?
Yes, they do. Moreover, both kinds of states are related to the spin
(see chapter 3) of the particles:
Spin-statistics theorem (Pauli, 1940)
Symmetrical states Spin = 0, 1, 2, ...
bosons (photons, mesons)
Antisymmetrical states Spin =
1
2
,
3
2
, ...
fermions (e

, p, n, quarks)
3.5 Symmetries 95
Shell structure of atoms follows from antisymmetric nature of elec-
trons, which cannot occupy the same state.
Bosons do not avoid each other, they can occupy the same state, e.g.
the ground state.
Such states have been realized recently for bosonic atoms (and, even
more recently, molecules)
Bose-Einstein condensation
96 Formalism of Quantum Mechanics
Chapter 4
Applications of the Theory
4.1 The harmonic oscillator
Important model system, because
describes systems in the neighborhood of a stable equilibrium (see [Kira],
chap. 3.2)
mathematically exactly solvable in classical and quantum mechanics
4.1.1 Algebraic solution of the one-dimensional harmonic
oscillator
Hamiltonian:

H =
p
2
2m
+
c
2
x
2
=
1
2m
( p
2
+m
2

2
0
x
2
) (4.1)
aim: solve

H[
n
) = E
n
[
n
)
Rewrite

H:

H =
1
2m
_
( p +im
0
x)( p im
0
x) + im
0
( p x x p
. .
=

i
)
_
=
0
_
( p +im
0
x)( p im
0
x)
2m
0
+
1
2
_
(4.2)
97
98 Applications of the Theory
To proceed we need
a) Creation and annihilation operators
Denition:
a

:=
1

2m
0
( p +im
0
x) (4.3)

creation operator

a :=
1

2m
0
( p im
0
x) (4.4)

annihilation operator

( a

= a
(4.2)


H =
0
( a

a +
1
2
)
(4.5)
alternative decomposition of

H:

H =
1
2m
_
( p im
0
x)( p +im
0
x) + im
0
[ x, p]
_
=
0
_
( p im
0
x)( p +im
0
x)
2m
0

1
2
_
=
0
_
a a

1
2
_
(4.6)
Combine eqs. (4.5) and (4.6)

0
= a a

1
2

0
= a

a +
1
2
subtract
= [ a, a

] =

1 (4.7)
add
= a a

+ a

a =
2

H
Denition: Anti-commutator
[

A,

B]
+
=

A

B +

B

A

[ a, a

]
+
=
2

H

0
(4.8)
4.1 The harmonic Oscillator 99
Some useful relations:
[ a,

H] =
0
[ a, a

a]
=
0
_
[ a, a

] a + a

[ a, a]
_
=
0
a (4.9)
[ a

,

H] =
0
[ a

, a a

]
=
0
_
[ a

, a] a

+ a[ a

, a

]
_
=
0
a

(4.10)
b) Energy spectra


H a[
n
) =
_
a

H [ a,

H]
_
[
n
)
= a(

H
0
)[
n
)
= (E
n

0
) a[
n
) (4.11)


H a

[
n
) =
_
a


H [ a

,

H]
_
[
n
)
= a

(

H +
0
)[
n
)
= (E
n
+
0
) a

[
n
) (4.12)
Conclusion:
If [
n
) is eigenvector of

H to the eigenvalue E
n
, then
[
+
n
) := a

[
n
) is eigenvector to the eigenvalue E
n
+
0
, and
[

n
) := a[
n
) is eigenvector to the eigenvalue E
n

0
.
If [) is a normalized eigenvector of

H
_
[) = 1
_
to the eigenvalue E,
then:
E = [

H[) =
0
[ a

a +
1
2
[)
=

0
2
+
0
[ a

a[)
=

0
2
+
0
a[ a)
E

0
2
=
0

) 0
E

0
2
i.e. the spectrum is bounded from below.
100 Applications of the Theory
Is E =

0
2
possible? Yes!
proof by reductio ad absurdum
assume : E
0
= min
_
E
n
_
>

0
2


0
2
< E
0
=
0
[

H[
0
) =

0
2
+
0
a
0
[ a
0
)
0 < a
0
[ a
0
) =

0
[

0
)
= [

0
) eigenvector of

H to E
0


0
2
contradiction!
Conclusion: ground state [
0
) is eigenvector to the eigenvalue E
0
=

0
2
E
0
=
0
[

H[
0
)
=

0
2
+
0
a
0
[ a
0
)
=

0
2
= a[
0
) = 0
From (4.12) it follows that the rst excited state has the eigenenergy
E
1
= E
0
+
0
=
3
0
2
, the second E
2
= E
1
+
0
=
5
0
2
etc.
eigenvalue spectrum of the harmonic oscillator
E
n
=
0
_
n +
1
2
_
, n = 0, 1, 2, ...
(4.13)
c) Eigenstates
the ground state is dened via
a[
0
) = 0 ,
0
[
0
) = 1 (4.14)
rst excited state is generated by [

0
) := a

[
0
)

0
[

0
) =
0
[ a a

[
0
)
=
0
[ a

a[
0
) +
0
[[ a, a

][
0
)
=
0
[
0
) = 1
[
1
) [

0
) is normalized eigenstate to E
1
4.1 The harmonic Oscillator 101
second excited state [

1
) := a

[
1
)

1
[

1
) =
1
[ a a

[
1
)
=
1
[ a

a[
1
) +
1
[[ a, a

][
1
)
=
0
[
0
) +
1
[
1
) = 2
normalized eigenstate to E
2
=
5
0
2
[
2
) =
1

2
a

[
1
) =
1

2
_
a

_
2
[
0
)
In general: If [
n
) is the normalized eigenstate to E
n
=
[
n+1
) =
1

n+1
a

[
n
) is the normalized eigenstate to E
n+1
proof (of normalization):

n+1
[
n+1
) =
1
n + 1

n
[ a a

[
n
)
=
1
n + 1
_
1

n
[

H[
n
) +
1
2
_
=
1
n + 1
_
n +
1
2
+
1
2
_
= 1
Similarly, one can show that [
n1
) =
1

n
a[
n
) is the normalized eigenstate
to E
n1
Summary:
[
n
) =
1

n!
_
a

_
n
[
0
) is the normalized eigenstate to E
n
(4.15)
4.1.2 Discussion
a) Coordinate space representation of the eigenvectors
Option 1) starting point: consider coordinate space representation of
a[
0
) = 0
=
1

2m
0
( p im
0
x)[
0
)
x[ p[
0
) im
0
x[ x[
0
) = 0
chap. 3.2.2
i

0
(x) im
0
x
0
(x) = 0
102 Applications of the Theory
this dierential equation has the (normalized) solution

0
(x) =
_
m
0

_1
4
e

m
0
2
x
2
(4.16)
excited states: x[
n
) =
1

n!
x[(a

)
n
[
0
)

n
(x) =
1

n!
1
[2m
0
]
n
2
_

i
d
dx
+im
0
x
_
n

0
(x)
=
one can show:
. . . =
_
m
0

_1
4
_
1
2
n
n!
e

m
0
2
x
2
H
n
_
_
m
0

x
_
(Hermites functions, see chapter 3.1.1)
Option 2) solve Schrodinger equation in position space
Literature: [Fli], chap. 12; [Gre], chap. 7; [DL], chap. 5.3
b) Properties of the eigensolutions
(i) Energy spectrum is discrete and equidistant
(ii) Zero-point energy E
0
=

0
2
> 0 uncertainty-relation
(cf. chap. 2.4.2d)
one can show that for the nth eigenstate:
x
n
p
n
=
E
n

0
=
_
n +
1
2
_
in particular x
0
p
0
=

2
(minimal uncertainty)
(iii) number of nodes + parity: same as for innitely deep potential well
(cf. chap. 2.4.2d)
c) Some special features
Literature: [Mes] I, chap. 12.2; [Scha], 13
d) Meaning of the creation and annihilation operator
the creation operator enhances the excitation
a

[
n
) =

n + 1[
n+1
) E
n
+
0
E
n+1
the annihilation operator reduces the excitation
a[
n
) =

n[
n1
) E
n

0
E
n1
4.2 Approximation rules 103
conventional interpretation of the oscillator:
one particle with mass m in ground or excited state with energy E
n
eld-quantum interpretation:
system consisting of n eld-quanta with the constant energy =
0
in mode
0
E
0
= no particle (but zero-point oscillation)
E
1
= one particle
.
.
.
E
n
= n particles
a, a

create and annihilate these particles (eld-quanta)


example: photons = eld-quanta of the EM-eld
e) Multi-dimensional and coupled oscillators
Literature: [Bloc], chap. 5.5.3; [Gre] I, chap. 14
4.2 Approximation methods
Only a few QM problems can be solved exactly. In most cases one has to resort
to approximation techniques (or numerical methods or a combination of both).
Here, we sketch only the most fundamental approximation methods; perturba-
tion theory and variational methods.
4.2.1 Stationary perturbation theory
Idea: Start with a simpler problem and use its known solution to construct
an (approximate) solution of the problem at hand
a) Formulation
task: solve stationary Schrodinger Equation (i.e.
t

H = 0)

H[
n
) = E
n
[
n
) (4.17)
decompose

H =

H
0
+

V (4.18)
assume that

H
0
problem can be solved

H
0
[
0
n
) = E
(0)
n
[
0
n
) (4.19)
Wed like to seek solutions cf. eq. (4.17) in terms of a Taylor (like) expan-
sion based on the eigenvalues and eigenstates of the undisturbed problem
104 Applications of the Theory
eq. (4.19). Therefore, we require that the perturbation

V be small. Lets
introduce a parameter :

V

W with 1 (4.20)
(4.17)

H
0
+

W
_
[
n
()) = E
n
()[
n
()) (4.21)
Taylor expansion about = 0:
E
n
() = E
(0)
n
+
dE
n
()
d

=0
+
1
2
d
2
E
n
()
d
2

2
+. . . (4.22)
[
n
()) = [
0
n
) +
d
d
[
n
())[
=0
+. . . (4.23)
We need to nd expressions for the derivatives in eqs. (4.22), (4.23):
consider derivative cf. eq. (4.21):
d
d
_

H
0
+

W E
n
()
_
[
n
()) = 0

H
0
+

W E
n
()
_
[

n
()) +
_

W E

n
()
_
[
n
()) = 0
(E

n
=
dE
n
d
etc.)

m
()[

H() E
n
()[

n
()) +
m
()[

W E

n
()[
n
()) = 0
i) m = n
= E

n
() =
n
()[

W[
n
()) (4.24)
ii) m ,= n
=
m
()[

n
()) =

m
()[

W[
n
())
E
n
() E
m
()
(4.25)
In order to use eq. (4.25) for an expansion of [

n
) in terms of the or-
thonormal basis
_
[
n
)
_
we have to consider the coecient
n
()[

n
())
in addition. If we assume that
n
()[

n
()) =

n
()[
n
()) (i.e. we
choose real states which is not a restriction) we can show that

n
()[

n
()) = 0
proof :
d
d

n
()[
n
())
. .
=1
=

n
()[
n
()) +
n
()[

n
())
= 2
n
()[

n
()) = 0
4.2 Approximation rules 105
[

n
()) =

m
[
m
())
m
()[

n
())
=

m=n

m
()[

W[
n
())
E
n
() E
m
()
[
m
()) (4.26)
We consider also the 2
nd
derivative term in eq. (4.22):
d
2
d
2
E
n
() =
d
d
E

n
()
(4.24)
=
d
d

n
()[

W[
n
())
=

n
()[

W[
n
()) +
n
()[

W[

n
())
(4.26)
= 2

m=n
[
n
()[

W[
m
())[
2
E
n
() E
m
()
(4.27)
Higher order terms can be calculated by dierentiating expressions (4.26),
(4.27) successively. We stop here and insert (4.24)-(4.27) in the Taylor
expansions (4.22), (4.23):
E
n
() = E
(0)
n
+
n
(0)[

W[
n
(0)) (4.28)
+
2

m=n
[
n
(0)[

W[
m
(0))[
2
E
n
(0) E
m
(0)
+ . . .
= E
(0)
n
+
0
n
[

V [
0
n
) +

m=n
[
0
n
[

V [
0
m
)[
2
E
(0)
n
E
(0)
m
+ . . .
[
n
()) = [
0
n
) +

m=n

0
m
[

V [
0
n
)
E
(0)
n
E
(0)
m
[
0
m
) + . . . (4.29)
eqs. (4.28), (4.29) are the standard expressions for the lowest-order correc-
tions.
b) Remarks:
1. Derivation and result are valid only if E
(0)
n
,= E
(0)
m
(i.e. no degeneracies)
For degenerate perturbation theory QM literature, e.g. [Jel], chap.
7
2. Convergence of perturbation series?
Cannot be answered in general. In some cases, perturbation expan-
sions do converge, in some they do not, and in some other cases the
106 Applications of the Theory
perturbation series turns out to be a so-called semi-convergent (asymp-
totic) series.
Consistency criterion for convergence (check eq. (4.29))

0
m
[

V [
0
n
)
E
(0)
n
E
(0)
m

1 , (for n ,= m)
further reading: e.g. [Jel], chap. 7.2
3. In practice, exact calculations beyond 1
st
order are often not feasible
due to (innite) sums over all basis states (cf. eq. (4.29)).
4. Variants of stationary perturbation theory (as well as more formal
derivations) exist.
Literature: [Jel], chap. 9; [Mes], QM II, chap. 16; [DL], chap. 11
c) Example: Perturbed harmonic oscillator (1-d)

H
0
=
p
2
2m
+
m
2

2
0
x
2
(4.30)
can be solved exactly (chap. 4.1)
1. linear perturbation

V = ax , (a R) (4.31)
We need - cf. eq. (4.29) - matrix elements

0
n
[

V [
0
m
) = a
0
n
[ x[
0
m
)
can be calculated e.g. by using (well-known and easy-to-prove)
recursion relations for Hermites polynomials and functions

x
0
n
(x) =
_
n + 1
2

0
n+1
(x) +
_
n
2

0
n1
(x)

0
n
[ x[
0
m
) =
_

m
0
_
_
n + 1
2

n+1,m
+
_
n
2

n1,m
_
i.e. almost all matrix elements vanish!
to 2
nd
order:
E
n
E
(0)
n
+E
(1)
n
+E
(2)
n
=
0
(n +
1
2
) + 0 +
_

a
2
2m
2
0
_
(4.32)
4.2 Approximation rules 107
It turns out, that all higher-order terms vanish and eq. (4.32) is the
exakt solution! This can be seen from a dierent viewpoint:

H =

H
0
+

V =
p
2
2m
+
m
2

2
0
x
2
+ax
=
p
2
2m
+
m
2

2
0
(x +b)
2

m
2

2
0
b
2
(with a = m
2
0
b)
the oscillator potential is not really disturbed, but only displaced!
N
>
=

all eigenenergies are shifted by


m
2

2
0
b
2
=
a
2
2m
2
0
!
2. Anharmonic oscillator

V = c x
3
(4.33)
i.e., calculate matrix elements
0
n
[x
3
[
0
m
). In this case, the 1
st
order
energy correction vanishes as well, but higher orders do contribute.
Further discussion of undisturbed and disturbed harmonic oscillator:
[CT] I, chap. 5 and II, chap. 11.4
4.2.2 Variational methods
a) Rayleigh-Ritz principle (of minimal energy)
Consider energy functional (functional=linear map vector scalar)
E[] =
[

H[)
[)
(4.34)
108 Applications of the Theory
The expectation value of

H with respect to any trial state [) is larger
than (or equal to) the true groundstate energy, i.e., the lowest eigenvalue
of

H
.
proof :

H[
n
) = E
n
[
n
)
with
n
[
k
) =
nk
and

n
[
n
)
n
[ =

1
We can expand each trial function:
[) =

j
c
j
[
j
)
consider E E[] E
0
=
[

H E
0
[)
[)
=

jk
c

j
c
k

j
[

H E
0
[
k
)

j
[c
j
[
2
=

j
[c
j
[
2
(E
j
E
0
)

j
[c
j
[
2
_
_
_
= 0 if [) = [
0
)
> 0 otherwise
energy functional E[] is bounded by ground-state energy.
One exploits this principle in the following way:
Choose a set of trial states [)
Find the minimum of E[]
According to the Rayleigh-Ritz principle the state that minimizes the
energy is the best approximation within the set!
b) Simple application
consider [) = [())
E[] =
()[

H[())
()[())
= E()
search for minimum; i.e., calculate
dE()
d
= 0
4.2 Approximation rules 109
example: 1-dim harmonic oscillator

H =

2
2m
d
2
x
+
m
2

2
0
x
2
Ansatz : (x, ) = Ae

2
2
x
2
E() =
[A[
2
_

2
2
x
2
_


2
2m
d
2
x
+
m
2

2
0
x
2
_
e

2
2
x
2
dx
[A[
2
_

2
x
2
dx
= . . . =

2

2
4m
+
m
2
0
4
2

dE
d
=

2

2m

m
2
0
2
3
= 0
2
0
=
m
0

(x,
0
) = Aexp
_

m
0
2
x
2
_
(cf. eq. (4.16))
E(
0
) =
1
2
2
0
_

4
0
2m
+
m
2
0
2
_
=

0
2
in fact, this is the true ground-state wave function (and energy).
One can rene the method by introducing trial wave functions that depend on
more than one parameter or by considering other ansatze, such as the expansion
of trial wave functions with respect to known nite basis systems, etc.
In fact, the variational method may be considered as the most important
approximation technique for the determination of ground states.
4.2.3 Time-dependent perturbation theory
a) Formulation
task: solve TDSE (i.e.
t

H ,= 0)
i
t
[(t)) =

H(t)[(t)) (4.35)
decompose Hamiltonian (cf. eq. (4.18))

H(t) =

H
0
+

V (t)


H
0
+

W(t) (4.36)
110 Applications of the Theory
assume that

W(t t
0
) = 0
t t
0
:

H
0
[
j
) =
j
[
j
)
assume [(t
0
)) = [
0
) , (initial state) (4.37)
Ansatz : [(t)) =

j
c
j
(t)e

j
t
[
j
)
=

j
c
j
(t)[
j
(t)) (4.38)
Insertion cf. eq. (4.38) into eq. (4.35):

j
_
i c
j
+
j
_
e

j
t
[
j
) =

j
c
j
e

j
t

H[
j
) [
k
(t)[
i c
k
=

j
e
i

(
k

j
)t
c
j

k
[

W(t)[
j
) (4.39)
coupled-channel eqs. (still exact if basis is complete)
If

W(t > T) = 0 c
k
(t > T) = const. and
p
k
= [c
k
[
2

t>T
= [
k
[)[
2

t>T
= const. (4.40)
transition probabilities
0

k
Ansatz for solution of eq. (4.39):
c
k
(t) = c
(0)
k
(t) + c
(1)
k
(t) +
2
c
(2)
k
(t) + . . . (4.41)
Insertion into eq. (4.39) yields:
i
_
c
(0)
k
+ c
(1)
k
+
2
c
(2)
k
+. . .
_
=

j
_
c
(0)
j
+c
(1)
j
+
2
c
(2)
j
+. . .
_
e
i

(
k

j
)t

k
[

W(t)[
j
)

0
: i c
(0)
k
= 0 c
(0)
k
(t) = const.

1
: i c
(1)
k
=

j
c
(0)
j
e
i

(
k

j
)t

k
[

W(t)[
j
)

2
: i c
(2)
k
=

j
c
(1)
j
e
i

(
k

j
)t

k
[

W(t)[
j
)
4.2 Approximation rules 111
These equation can be solved successively:

0
: c
(0)
k
(t) = const. =
k0
(cf. eq. (4.37)) (4.42)

1
: i c
(1)
k
=

j0
e
i

(
k

j
)t

k
[

W(t)[
j
)
= e
i

(
k

0
)t

k
[

W(t)[
0
)
c
(1)
k
(t) c
(1)
k
(t
0
)
. .
=0
=
i

_
t
t
0
e
i

(
k

0
)t

k
[

W(t

)[
0
) dt

(4.43)
(as = 0 at t = t
0
)
Accordingly:

2
: c
(2)
k
(t) =
1

j
_
t
t
0
dt

_
t

t
0
dt

e
i

(
k

j
)t

e
i

(
j

0
)t


k
[

W(t

)[
j
)
j
[

W(t

)[
0
) (4.44)
(. . .)
Remarks:
(a) Exakt calculations beyond 1
st
order are in general impossible due
to innite sums (cf. page 105)
(b) Interpretation
t
0
t
1
st
order [
0
)

W
[
k
) direct transition
2
nd
order [
0
)

W
[
j
)

W
[
k
)
transition via virtual intermediate states (2 steps)
further reading (and more appropriate visualization in terms of generic
diagrams): [Mes], QM II, chap. 17
b) Discussion of 1
st
order Fermis Golden Rule (FGR)
To 1
st
order time-dependent perturbation theory we have (cf. eq. (4.42),
(4.43)):
c
k
(t)
k0

_
t
t
0
e
i
k0
t

V
k0
(t

) dt

(4.45)
with
k0
=
k

0
=

k

transition frequency

V
k0
=
k
[

V [
0
) =
k
[

W[
0
)

transition matrix element

112 Applications of the Theory


Two useful (and related) examples:
(i) Sudden perturbation (somewhat academic)
J
8

V (t) =
_
_
_
0 t t
0
0

V t > t
0
k ,= 0
c
k
(t) =
i

_
t
0
e
i
k0
t

V
k0
(t

) dt

=

k
[

V [
0
)
i
_
t
0
e
i
k0
t

dt

k
[

V [
0
)

k0
_
e
i
k0
t
1
_
transition probability
p
0k
(t) = [c
k
(t)[
2
=
4[V
k0
[
2

2
f(t,
k0
) (4.46)
f(t,
k0
) =
sin
2
k0
t
2

2
k0

k0
0

t
2
4
(4.47)
Figure 4.1: y = f(t,
k0
)
4.2 Approximation rules 113
signicant transitions occur only around
k0
= 0 within width
=
2
t
One can consider limit t :
f(t,
k0
)
t

t
2
(
k

0
) (4.48)
p
0k
t

2t

[V
k0
[
2
(
k

0
)
(ii) Periodic perturbation

V (t) =
_
_
_
0 t t
0
= 0

Be
it
+

B

e
it
t > t
0
(4.49)
(note that

V =

V

)
c
k
(t) =
1
i
_
t
0
e
i
k0
t

V
k0
(t

) dt

=
1

k
[

B[
0
)

k0
+
_
e
i(
k0
+)t
1
_
+

k
[

[
0
)

k0

_
e
i(
k0
)t
1
_
_
if t
2

(i.e. ):
p
0k
(t) =
4[B
k0
[
2

2
_
f(t,
k0
+) +f(t,
k0
)
_
(4.50)
t

2t

[B
k0
[
2
_
(
k

0
+) + (
k

0
)
_
with B
k0
=
k
[

B[
0
) and eq. (4.47)
resonances at
k0
= (signicant transitions occur only around
these resonance frequencies)

k0
=
k
=
0

DM
A

stimulated emission (of energy)


(only possible if
0
is not the ground state)
114 Applications of the Theory

k0
= +
k
=
0
+
A

DM
A

(stimulated) absorbtion (of energy)


If one considers a quantum particle in a (weak) classical electromag-
netic (EM) eld (eq. (2.52)) one can rewrite the Hamiltonian such
that it takes the form

H(t) =

H
0
+

V (t) with eq. (4.49) for

V . One
can interpret the stimulated emission of the energy packet as the
emission of a photon with energy and the absorption of as the
absorption of a photon from the radiation eld. The latter process
corresponds to the photoelectric eect (cf. page 2) if the nal state is
a (true) continuum state (i.e., if an electron is ionized).
problems:
(i) p
0k
for t and
k0

(ii) p
0k
(t) oscillates (is proportional to t
2
for
k0
= )
(i) and (ii) seem unphysical!
solution: consider transitions into continuum of nal states
(it turns out that this is always justied for realistic
systems due to nite line-width of excited states)
A

,A
A

,A
P
0k,k
=
_

k
+

p
0k
(

k
)(

k
) d

k
(4.51)
with (
k
): density of states
_
in interval [
k
;
k
+ ]
_
One nds with eqs. (4.48) - (4.50)
P
0k,k
=
2t

2
_
[B
k0
[
2
(

k
)
_
(
k

0
+) + (
k

0
)
_
d(

k
)
=
2

_
[B
k0
[
2
(
k
)t

0
=
+[B
k0
[
2
(
k
)t

k
=
_
dene transition rate w
0k,k
=
d
dt
P
0k,k
4.3 Angular momentum and spin 115
Fermis Golden Rule (FGR)
w
0
emission
k,k
= w
0
absorption
k,k
=
2

[B
k0
[
2
(
k
)

k
=
0

(4.52)
Note that w
0k,k
is constant (i.e. does not depend on t).
For actual calculations three ingredients are needed (see App. A.2):
1) Explicit form of

B
starting point: Hamiltonian for a particle in the EM eld (eq.
(2.73))
2) Initial and nal states to calculate B
k0
3) Density of states (
k
)
Literature: [Scha], chap. 11; [BS], chap. IV
Note that the notion of photons for the interpretation of stimulated emission
and absorbtion has no signicance as long as we are dealing with classical
EM elds. From a theoretical point of view photons enter the game only
if the EM eld is quantized ( quantum electrodynamics (QED)). This
quantization does not change the nal expressions for stimulated emission
and absorbtion, but it shows that there is another process which cannot be
described in our semiclassical framework: spontaneous emission, i.e., the
emission of a photon (and transition to a lower-lying state) without any
external EM eld.
A relatively simple account on the quantization of the EM eld can be
found, e.g., in: [Fri], chap. 2.4 or [?], chap. 1-2
higher formulations: [Scha], chap. 14; [Mes], chap. 21 (and of course,
QED textbooks)
4.3 Angular momentum and spin
We have introduced orbital angular momentum of a particle by (eq. (3.81))

l = r p
consequences:

l
x
,

l
y
,

l
z
hermitian
[

l
i
,

l
j
] = i

k

ijk

l
k
Turn the argument upside-down and dene a general angular momentum
operator

J as a vector operator:

J = (

J
x
,

J
y
,

J
z
)
116 Applications of the Theory
with hermitian components which satisfy
[

J
i
,

J
j
] = i

ijk

J
k
(4.53)
All properties of angular momentum in QM follow from this denition.
4.3.1 Angular momentum algebra
Denition:

J
2
=

J
2
x
+

J
2
y
+

J
2
z
(4.54)
[

J
2
,

J
i
] = 0 , (i = 1, 2, 3) (4.55)
proof:
[

J
2
,

J
z
] = [

J
2
x
+

J
2
y
,

J
z
]
= [

J
x
,

J
z
]

J
x
+

J
x
[

J
x
,

J
z
] + [

J
y
,

J
z
]

J
y
+

J
y
[

J
y
,

J
z
]
=

i
_

J
y

J
x
+

J
x

J
y


J
x

J
y


J
y

J
x
_
= 0 etc.

O
N

one component can be measured si-


multaneously with the absolute square.
Usually, one picks the z-component.
Eigenvalue problems of

J
2
,

J
z
:

J
2
[
jm
) =
j
[
jm
) (4.56)

J
z
[
jm
) =
m
[
jm
) (4.57)
We abbreviate the common eigenstates by their quantum numbers:
[
jm
) [jm) , ( jm[j

) =
jj

mm
)
What can we say about the eigenvalues
j
,
m
?
[

J
2
i
[)

J
i
=

J

i
=

J
i
[

J
i
) 0 , ( [) )
4.3 Angular momentum and spin 117
in particular:
jm[

J
2
[jm) =
j
0 (4.58)
jm[

J
2
x
+

J
2
y
[jm) = jm[

J
2


J
2
z
[jm)
=
j

2
m
0

_

j

m

_

j
(4.59)
(i.e. spectrum of

J
z
is bounded by eigenvalues of

J
2
!)
dene ladder operators (cf. creation and annihilation operators in chap.
4.1)

J
+
=

J
x
+i

J
y
(4.60)

=

J
x
i

J
y
(4.61)

+
=

J

,

J

=

J
+
(4.62)
Some useful relations (easy to prove):
[

J
z
,

J

] =

(4.63)
[

J
z
,

J
n

] = n

J
n

(4.64)


J
+
=

J
2


J
2
z

J
z
(4.65)
[

J
+
,

J

] = 2

J
z

J
+

J

=

J
2


J
2
z
+

J
z
(4.66)
[

J
2
,

J

] = 0 (4.67)
lets play with them:

J
z

J
+
[jm)
(4.63)
= (

J
+

J
z
+

J
+
)[jm)
= (
m
+)

J
+
[jm) (4.68)


J
+
[jm) = [jm

)
with

J
z
[jm

) = (
m
+)[jm

)
_
_
_

J
+
[jm) is another
eigenvector of

J
z
with
eigenvalue raised by
Norm of the eigenvector

J
+
[jm):
[[

J
+
[jm)[[
2
= jm[


J
+
[jm)
(4.65)
= jm[

J
2


J
2
z

J
z
[jm)
=
j

2
m

m
= [[
2
118 Applications of the Theory
There is one case where [[

J
+
[jm)[[
2
= 0 ; namely when

2
m

m
= 0

j
=
m
(
m
+)
On the other hand we know that the spectrum of

J
z
is bounded, i.e., there
is a smallest and a largest eigenvalue (
min
,
max
). In particular, the eigen-
value eq.

J
z
[jm
max
) =
max
[jm
max
)
exists. Lets apply

J
+
to [jm
max
):


J
+
[jm
max
) = ?
it cannot raise eigenvalue by as
max
is already the largest eigenvalue.
The only way out is

J
+
[jm
max
) = 0
which implies (see above)

j
=
max
(
max
+) (4.69)
We can play the same game with

J


J
z
[jm)
(4.63)
= (
m
)

J

[jm)
[[

[jm)[[
2
= jm[

J
+

J

[jm)
(4.66)
=
j

2
m
+
m
There is a smallest eigenvalue
min

[jm
min
) = 0 [[

[jm
min
)[[
2
= 0

j
=
min
(
min
) (4.70)
From eq. (4.69) and (4.70) it follows that

max
=
min
(4.71)
4.3 Angular momentum and spin 119
We can move through the spectrum of

J
z
by applying ladder operators,
e.g., we can start at
max
and move towards the lower boundary
min
by:

J
z

J
n

[jm
max
)
(4.64)
= (

J
n


J
z
n

J
n

)[jm
max
)
= (
max
n)

J
n

[jm
max
)
= n
0
N
0
:
max
n
0
=
min
=
max
(4.72)
(i.e. we reach
min
after n
0
steps; note that we cannot miss it due to eq.
(4.70))
spectrum of

J
z
is quantized!
(4.72)

max
=
n
0
2
j =
min
(4.73)
with j = 0,
1
2
, 1,
3
2
, ...
(4.70)
j
=
2
j(j + 1) (4.74)
and
m
= m , (m = j, ..., j)
Summary: angular momentum spectrum is discrete (quantized):

J
2
[jm) =
2
j(j + 1)[jm)

J
z
[jm) = m[jm) (4.75)
j = 0,
1
2
, 1,
3
2
, ...
m = j, j + 1, ..., j 1, j
. .
(2j + 1) values spectrum of J
2
is
(2j + 1)fold degenerate
ladder operators

J
+
[jm) =
_
j(j + 1) m(m + 1)[jm + 1) (4.76)

[jm) =
_
j(j + 1) m(m1)[jm1) (4.77)
120 Applications of the Theory
4.3.2 Orbital angular momentum
(cf. chap. 3.3.5)

l = r p
is an angular momentum operator
One can show (see below) that eigenvalues are integer numbers:

l
2
[lm) =
2
l(l + 1)[lm)

l
z
[lm) = m[lm)
l = 0, 1, 2, 3, ...
spectroscopic notation s p d f, ...
eigenvalue problem in coordinate space

l
z,(r)

lm
(r) r[

l
z
[lm) = i(y
x
x
y
)
lm
(x, y, z)
in spherical coordinates
1

l
z,(r)

lm
(r) =

i

lm
(r, , )
!
= m
lm
(r, , ) (4.78)

lm
(r, , ) =
l
(r, )e
im
(4.79)
require that
lm
(r, , + 2) =
lm
(r, , ) = m integer!
correspondingly one nds after some calculation

l
2
(r)

lm
(r, , ) =
_


2
sin

(sin

)

2
sin
2

2
_

lm
!
= l(l + 1)
lm
(4.80)
this partial dierential equation is solved by the so-called
Spherical harmonics Y
lm
(, )
Y
lm
(, ) =

2l + 1
4
(l m)!
(l +m)!
P
m
l
(cos )e
im
(4.81)
P
m
l
(x) =
()
m
2
l
l!
(1 x
2
)
m
2
d
l+m
dx
l+m
(x
2
1)
l
(4.82)
1
x = r sin cos
y = r sin sin
z = r cos
4.3 Angular momentum and spin 121
associated Legendre polynomials
i.e. :

l
2
Y
lm
(, ) =
2
l(l + 1)Y
lm
(, ) , l = 0, 1, 2, ...
(cf. eq. (4.79), (4.80))

l
z
Y
lm
(, ) = mY
lm
(, ) , m = l, ..., l
Y
lm
form a complete set of orthonormal functions (quadratically inte-
grable) on unit sphere;
i.e. f(, ) =

lm
c
lm
Y
lm
(, ) , f(, )
and
_
Y

m
()Y
lm
() d =
ll

mm
(with = (, ))
Y
00
=
1

4
Y
10
=
_
3
4
cos
Y
11
=
_
3
8
sin e
i
Y
11
=
_
3
8
sin e
i
(...)
More details can be found, e.g., in: [Kirb], chap. 2.2.3; [Jac], chap. 3.5, 3.6; [Mes]
QM I, Appendix B.4; [Lin]
(note that some authors use slightly dierent denitions of Y
lm
or P
m
l
)
4.3.3 Spin
What about half-integer angular momenta? Are they only acceptable mathematical
solutions of the angular momentum eigenvalue problem or do they exist in na-
ture? Lets consider the mathematics for the simplest case rst and answer the
question about its physical relevance afterwards!
a) Angular momentum algebra for j s =
1
2
eigenvalue equations (cf. eqs. (4.75)):
s
2
[sm
s
) s
2
[
1
2
,
1
2
) =
3
2
4
[
1
2
,
1
2
) (4.83)
s
z
[sm
s
) = s
z
[
1
2
,
1
2
) =

2
[
1
2
,
1
2
) (4.84)
122 Applications of the Theory
two orthonormal spin functions exist:
[
1
2
, +
1
2
) [ )

spin up

[
1
2
,
1
2
) [ )

spin down

_
[ ), [ )
_
is orthonormal basis in this two-dimensional space
A general spin function [) can be written as
[) = [ ) +[ ) (4.85)
Consider ladder operators (cf. eq. (4.76), (4.77))
s
+
[s, m
s
) =
_
s(s + 1) m
s
(m
s
+ 1) [s, m
s
+ 1)
s

[s, m
s
) =
_
s(s + 1) m
s
(m
s
1) [s, m
s
1)
explicitly:
s
+
[ ) = s

[ ) = 0 (4.86)
s
+
[ ) = [ ) (4.87)

spin ips

[ ) = [ ) (4.88)
Matrix representation (2 2)
Vectors:
[ )
_
1
0
_
[ )
_
0
1
_
[)
(4.85)

_
Operators:
s
z
s
z
=
_
[ s
z
[ ) [ s
z
[ )
[ s
z
[ ) [ s
z
[ )
_
(4.83),(4.84)
=

2
_
1 0
0 1
_
(4.89)
4.3 Angular momentum and spin 123
correspondingly one nds from eqs. (4.83), (4.84) and (4.86) - (4.88):
(and s
x
=
1
2
(s
+
+s

), s
y
=
1
2i
(s
+
s

))
s
2
=
3
4

2
_
1 0
0 1
_
(4.90)
s
x
=

2
_
0 1
1 0
_
(4.91)
s
y
=

2
_
0 i
i 0
_
(4.92)
Denition: Paulis spin matrices

x
=
_
0 1
1 0
_
,
y
=
_
0 i
i 0
_
,
z
=
_
1 0
0 1
_
One can show that every 2 2 matrix can be represented as a linear
combination of
x
,
y
,
z
and 1 =
_
1 0
0 1
_
.
b) Magnetic moments and spin hypothesis
Classical orbital angular momentum (of a charged particle) gives rise to a
magnetic moment .
l (4.93)
Naive argument: consider charge q with circular motion:
= iF
..

current area

=
qv
2r
r
2
e

=
qrv
2
e

=
q
2m
l
Better argument: [Kirb], chap. 3.3.3
energy in external magnetic eld
W = B (4.94)
force F = W =
_
_

x
B
x

y
B
y

z
B
z
_
_
(4.95)
deection in inhomogeneous eld!
One can measure magnetic moment via deection of (charged) particle
in inhomogeneous B-eld
124 Applications of the Theory
Stern-Gerlach experiment (1922)
source
Ag atom
beam
z
B-Iield
inhonmogenous
screen
observation:
two distinct spots
(where Ag atoms arrive)
Interpretation:
quantize magnetic moment eq. (4.93) according to
=

l , ( =
q
2m
) (4.96)
measure z-component of i.e. eigenvalues of
z
(proportional to eigenvalues of

l
z
)

z
= m , m = l, ..., l (4.97)
the quantization of

l
z
explains why only distinct maxima are found
instead of a continuous distribution. The number of spots must cor-
respond to the number of possible eigenvalues of
z
.
= Stern-Gerlach experiment can be understood if one assumes the
existence of an intrinsic angular momentum s (of an electron) with
quantum numbers
s =
1
2
, m
s
=
1
2
,
1
2
, (spin doublet)
(spin hypothesis by Goudsmit + Uhlenbeek, 1925)
This intrinsic angular momentum - the spin - is a new degree of free-
dom of a quantum particle without classical analogy. It is, however,
connected with a magnetic moment (otherwise it would not be observ-
able by Stern-Gerlach apparatus)

s
= s (4.98)
One nds that constant is dierent from =
e
2m
for the orbital
angular motion (by a factor of 2; i.e. =
e
m
for an electron). This
4.3 Angular momentum and spin 125
factor - as well as the Spin
1
2
structure itself - can be derived with-
out any ad hoc assumption or hand-waving arguments from the Dirac
equation - the relativistic equation of motion for an electron (cf. chap-
ter 5), which is why some authors consider spin a relativistic eect.
Note that Stern-Gerlach experiment was also performed with hydrogen
atoms (i.e., one-electron systems) with the same result.
Further remarks:
Spin hypothesis explains also the splitting of the spectral lines (of
an atom) in an external (homogeneous) magnetic eld ( Zeeman
eect)
Not only electrons carry spin. In fact, quantum particles with inte-
ger and half-integer spin quantum numbers exist ( spin-statistics
theorem, chap. 3.5.5c).
c) Wave functions with spin
How to describe a quantum particle that carries spin and that has position
(momentum) degrees of freedom?
Spin is independent of motion of the particle, i.e.,
[f( x
i
, p
i
), g( s)] = 0
(for all operator functions)
state is a direct product state
[) = [)
..
orbital motion
[)
..
spin
(4.99)
(r) = r[) = r[)
_
[ ) +[ )
_
= (r)
_

_

_

(r)

(r)
_
(4.100)
two-component spinor wave function
Note that [) = [)[) = 1 1 = 1
126 Applications of the Theory
Consider a special decomposition:
[) =
_
[r)r[) d
3
r
=
_
_

(r),

(r)
_
_

(r)

(r)
_
d
3
r
=
_
_
[

(r)[
2
+[

(r)[
2
_
d
3
r
_
(r) d
3
r

spin densities total density
Expectation values of spin operators s:
s) = [ s[)
=
_
_

(r),

(r)
_
_
s
11
s
12
s
21
s
22
__

(r)

(r)
_
d
3
r

_
S(r)) d
3
r
e.g. s
z
=

2
_
1 0
0 1
_
S
z
(r)) =

2
_
[

(r)[
2
[

(r)[
2
_
(4.101)
net spin at position r (spin-excess)
d) Wave equation for particles with spin: Pauli equation
How to incorporate spin-dependent terms in Schrodinger equation?
starting point: Hamiltonian for particle in EM eld (eq. (2.52))

H =
1
2m
_

i
qA
_
2
+q
assume that particle (electron with q= -e) has spin
1
2
magnetic
moment
s
(eq. (4.98) energy W
s
(eq. (4.94))
add this energy
W
s
=
s
B =
B
B (4.102)
Pauli matrices
with
B
=
e
2m

Bohr magneton

(4.103)
to Hamiltonian which then acts on spinor wave functions:
4.3 Angular momentum and spin 127
i
t
=

H (4.104)

H =
1
2m
_

i
+eA
_
2
e +
B
B (4.105)
=
_

(r)

(r)
_
Pauli equation
Remarks:
Pauli equation describes Zeeman eect
Spin-orbit coupling, i.e., the interaction of the spin of a bound electron
with the magnetic eld that originates from the relative motion of the
nucleus can also be included in Pauli equation
Pauli equation can be derived from Dirac equation (chapter 5) for
v c
4.3.4 Addition of angular momenta
Consider sum of angular momentum operators

J =

J
1
+

J
2
(4.106)


J is also angular momentum operator, i.e., has hermitian components which
fulll eqs. (4.56), (4.57).
Examples:

j =

l +s total angular momentum of a particle

L =

N
i=1

l
i
total orbital angular momentum of Nparticle system

S =

N
i=1
s
i
total spin of Nparticle system
all such angular momentum operators share the characteristic spectrum (eq.
(4.75))
Question: How are quantum numbers and eigenvectors of

J
2
,

J
z
connected with
those of

J
2
1
,

J
1z
,

J
2
2
,

J
2z
?
angular momentum coupling
Literature: [Mes], QM II, chap. 13.5; [Lin]; [CT], chap. 10
128 Applications of the Theory
4.4 The hydrogen atom (and extensions)
Two-body problem for nucleus (n) and electron (e): (cf. chap. 2.2.4)

H =
p
2
n
2m
n
+
p
2
e
2m
e

Ze
2
4
0
[r
e
r
n
[
(4.107)
m
n
= 1836m
e
; m
e
9.1 10
31
kg
4.4.1 Separation of the two-body problem
Classical coordinate transformation (cf. [Kira], chap. 5.1.1)
(r
e
, p
e
, r
n
, p
n
) (R, P, r, p)
denition : M = m
e
+m
n
m
n
=
m
e
m
n
m
e
+m
n
m
e
R =
m
n
r
n
+m
e
r
e
M
r
n
P = p
e
+p
n
= M

R p
n
_
_
_
center of mass
motion
r = r
e
r
n
p = r =
m
n
p
e
m
e
p
n
M
_
_
_
relative
motion
QM transformation analogously
(r
e
, p
e
, r
n
, p
n
) (

R,

P, r, p) (4.108)
insertion in eq. (4.107) yields

H =

P
2
2M
+
p
2
2
+V ( r) ,
_
V (r) =
Ze
2
4
0
r
_
(4.109)
=

H
CM
+

H
rel
eq. (4.109) is the Hamiltonian of a non-interacting two-(quasi-)particle
system can be separated into one-particle problems (cf. chap. 3.5.5c):
Ansatz : [) = [
CM

rel
) (4.110)
(i.e. (r, R) =
CM
(R)
rel
(r) in coordinate space)
4.4 The hydrogen atom 129
Schrodinger eqs.

H
CM
[
CM
) = E
CM
[
CM
) (4.111)

H
rel
[
rel
) = E
rel
[
rel
) (4.112)
and

H[) = (E
CM
+E
rel
)[) (4.113)
SE (4.111) can be solved without diculty:

CM
(R) = Ae
iKR
K =
1

P
E
CM
=

2
K
2
2M
_

_
free particle
motion
eq. (4.112) can also be solved analytically, but before we sketch the solution
we consider some general properties/features of the quantum central-eld
(V (r) = V (r)) problem.
4.4.2 The QM central-eld problem
Consider

H
rel
=
p
2
2
+V (r) (4.114)
One can show that

H
rel
is invariant with respect to any rotation

R (cf. chap.
3.5.4a), i.e.
[

H
rel
,

R] = 0
= [

H
rel
,

l] = 0 (4.115)
conservation of angular momentum!
In particular, the operators

H
rel
,

l
2
,

l
z
form a complete set of compatible operators
(cf. page 72), i.e.,
[

H
rel
,

l
2
] = [

H
rel
,

l
z
] = [

l
2
,

l
z
] = 0 (4.116)
they have a common set of eigenstates. The eigenstates of

l
2
,

l
z
are the spher-
ical harmonics Y
lm
(4.81);
ansatz
rel
(r) = R
l
(r)Y
lm
(, ) (4.117)
Insertion into (4.112) for Hamiltonian (4.114) yields radial SE:
_
p
2
r
2
+

2
l(l + 1)
2r
2
+V (r) E
_
R
l
(r) = 0 (4.118)
130 Applications of the Theory
with p
2
r
=

2
r
2

r
(r
2

r
)
and operator identity p
2
= p
2
r
+

l
2
r
2
(can be proven in spherical coordinates in coordinate space)
useful denition : y
l
(r) = rR
l
(r) (4.119)
(4.118)
y

l
(r) +
_
U(r)
l(l + 1)
r
2
_
y
l
(r) = 0 (4.120)
_
E =

2
2
, V (r) =

2
2
U(r)
_
The radial eq. (4.120) is very similar to the 1D-SE (2.86). There are, however,
two important dierences:
(i) Eective potential in eq. (4.120)
U
eff
l
(r) = U(r) +
l(l + 1)
r
2


angular momentum barrier

(cf. classical central-eld problem)


(ii) boundary conditions
[
rel
(r)[
2
= [R
l
(r)[
2
[Y
lm
(, )[
2
<
in particular for r = 0
regularity condition
y
l
(0) = 0 (4.121)
r
(a) E < 0 (bound spectrum)
_
[
rel
(r)[
2
d
3
r =
_

0
r
2
R
2
l
(r) dr
_
[Y
lm
(, )[
2
d
=
_

0
y
2
l
(r) dr <
(quadratically integrable solutions required)
y
l
(r)
r
0 (boundary condition)
(b) E > 0 (continuous spectrum)
oscillatory solutions y
l
(r) for r
4.4 The hydrogen atom 131
4.4.3 Solution of the Coulomb problem (E < 0)
_
note: for E > 0 the solution leads to Rutherfords scattering formula
(which is identical in classical mechanics and QM)
_
denition :
2
= > 0
a =
4
0

2
e
2
0.53 10
10
m
for m
e
, a a
0
is the

Bohr radius

radial eq. (4.120):


_
d
2
dr
2

l(l + 1)
r
2
+
2Z
ar

2
_
y
l
(r) = 0 (4.122)
transformation: x = 2r

d
2
dx
2
=
1
4
2
d
2
dr
2

_
d
2
dx
2

l(l + 1)
x
2
+

x

1
4
_
y
l
(x) = 0 (4.123)
asymptotic solutions:
1. x

_
d
2
dx
2

1
4
_
y
l
(x) = 0
y
l
(x) = Ae

x
2
+Be
x
2
because of y
l
(x ) = 0 B = 0
2. x 0

_
d
2
dx
2

l(l + 1)
x
2
_
y
l
(x) = 0
y
l
(x) =
A
x
l
+Bx
l+1
because of y
l
(0) = 0 A = 0
This consideration motivates the following ansatz:
y
l
(x) = x
l+1
e

x
2
v
l
(x) (4.124)
132 Applications of the Theory
Insertion into eq. (4.123) yields new dierential eq. for v
l
(x):

_
x
d
2
dx
2
+ (2l + 2 x)
d
dx
(l + 1 )
_
v
l
(x) = 0 (4.125)
The quadratically integrable solution of (4.125) (Kummers or Laplaces dif-
ferential eq.) are known; they are the associated Laguerre polynomials:
L
k
p
(x) =
p

j=0
()
j
_
(p +k)!

2
(p j)!(k +j)!j!
x
j
more specically:
v
l
(x) = L
2l+1
nl1
(x) ,
_
n
r
= n l 1 0
n 1 l
_
with n
n
=
Z

n
a
, n = 1, 2, ... (4.126)
The detailed solution shows that the integrability of the solutions requires =
Z
a
to be positive, integer numbers quantization of (i.e. quantization of the
energy)
2
y
nl
(r) = A
nl
r
l+1
e

n
r
L
2l+1
nl1
(2
n
r)
and properly normalized wave functions take the form

rel
(r)
nlm
(r) =
(n l 1)!
_
(n +l)!

3
2
l+
1
2

l+2
n

a
r
l
e

n
r
L
2l+1
nl1
(2
n
r)Y
lm
(, ) (4.127)
R
nl
(r)Y
lm
(, ) ,
n 0
l n 1
l m l
quantization condition (4.126) leads to:
E
n
=
RZ
2
n
2
, n = 1, 2, ... (4.128)
R =

2
2a
2
13.6 V

Rydberg

constant (4.129)
2
One can nd the quadratically integrable solutions of (4.125) explicitly by using the ansatz
v
l
(x) =

i
b
l
i
x
i
and by taking the boundary (and regularity) conditions into account.
4.4 The hydrogen atom 133
The lowest-lying hydrogen eigenfunctions (orbitals)
n l m n
r
= n l 1
nlm
(r) E
n
1 0 0 0 1s
1

_
Z
a
_3
2
e

Zr
a
RZ
2
2 0 0 1 2s
1
4

2
_
Z
a
_3
2
_
2
Zr
a
_
e

Zr
2a
RZ
2
4
2 1 0 0 2p
0
1
4

2
_
Z
a
_3
2
_
Zr
a
_
e

Zr
2a
cos
RZ
2
4
2 1 1 0 2p
1
1
8

_
Z
a
_3
2
_
Zr
a
_
e

Zr
2a
sin e
i RZ
2
4
E s p d
n 1
n 2
n 3
0
1 s
2s
3s
2p
-1
2p
0
2p
1
3p
-1
3p
0
3p
1
3d
-2
3d
-1
3d
0
3d
1
3d
2
Figure 4.2: Energy spectrum of the Schrodinger-Coulomb problem. Note that
Coulomb potential supports innitely many bound states (E
n
n
0).
Degeneracy of energies (depend only on n)
given n l = 0, 1, ..., n 1
given l m = l, ..., l

n1

l=0
(2l + 1) = n
2
each energy level E
n
is n
2
-fold degenerate. Note that all central-eld prob-
lems share (2l + 1)-fold degeneracy originating from rotational invariance. The
fact that energies do not depend on n
r
, l separately, but only on n = n
r
+ l + 1
134 Applications of the Theory
is specic to the Coulomb problem (one denotes n as the principal quantum
number, and n
r
as the radial quantum number which determines the number of
nodes in the radial wave functions).
Often one considers a radial probability density dened by

nl
(r) = r
2
R
2
nl
(r)
_
[Y
lm
(, )[
2
d
= y
2
nl
(r) (4.130)

nl
(r) dr: probability to nd electron at radial distance [r, r +dr]
Figure 4.3: Radial hydrogen 1s, 2s, 2p wave functions (blue) and probability
densities (red)
Figure 4.4: Radial hydrogen 3s, 3p, 3d wave functions (blue) and probability
densities (red)
4.4 The hydrogen atom 135
4.4.4 Assorted remarks
a) More (mathematical) details about the Coulomb problem can be found in
any QM textbook!
b) Hydrogen-like ions
We have solved not only the (Schrodinger) hydrogen problem (Z = 1), but
also the bound-state problems of all one-electron atomic ions (e.g. He
+
,
Li
2+
,...) for Z = 2, 3, ...
c) Exotic systems
... are also solved
(a) positronium (e
+
e

)
(b) myonium (
+
e

)
(c) myonic atom (p

)
In these cases one has to take care of the dierent masses compared to the
hydrogen problem (p
+
e

); (m

207 m
e
)
d) Corrections
The spectrum determined by eq. (4.128) is the exact solution of the Schrodinger-
Coulomb problem, but not exactly what one sees experimentally. The rea-
son is that the Schrodinger equation is not the ultimate answer, e.g., it has
to be modied to meet the requirements of the theory of special relativity.
Therefore, corrections show up (cf. chapter 5), which lead to a (partial)
lifting of the degeneracy.
e) Many-electron atoms
... cannot be solved analytically!
One approach to approximate solutions: stationary perturbation theory.
Lets consider the He atom (2 electrons) with xed nucleus (i.e. m
n
).

H =

2
2m


2
2m

Ze
2
4
0
r
1

Ze
2
4
0
r
2
+
e
2
4
0
[r
1
r
r
[
=

H
0
+

V
12
(4.131)
0
th
order approximation:

V
12
= 0


H
0
is the non-interacting Hamiltonian.
Note that we have to take care about the Pauli principle because we
deal with a many-particle fermionic system.
136 Applications of the Theory
The antisymmetric ground-state wave function reads:

0
=
He
+
1s
(r
1
)
He
+
1s
(r
2
)
1

2
_
[ ) [ )
_
. .

spin singlet

(4.132)
I.e., we have lled the K-shell (n = 1) with 2 electrons with opposite
spin such that

0
be antisymmetric.
Since the Hamiltonian (4.131) does not include spin-dependent inter-
actions only the spatial part of
0
is needed to determine the ground-
state energy (from eq. (4.128)):
E
(0)
0
= 8R = 108.8 eV
E
expt
0
= = 79 eV
This is obviously a poor approximation
1
st
order approximation
We can improve the calculation by consideration of the rst-order cor-
rection according to eq. (4.28):
E
(1)
0
=
0
[

V
12
[
0
)
=
e
2
4
0
_
d
3
r
1
_
d
3
r
2
_
1

2
_
Z
a
0
6__
e

2Zr
1
a
0
e

2Zr
2
a
0
[r
1
r
2
[
= . . . =
5
8
Ze
2
a
0
for Z = 2 we obtain E
(1)
= 34 eV, and
E
(1)
0
= E
(0)
0
+E
(1)
= 74.8 eV
which is a considerable improvement!
Better starting point for 1
st
order perturbation theory

H =
_

H
0
+v
eff
(r
1
) + v
eff
(r
2
)
_
+
_

V
12
v
eff
(r
1
) v
eff
(r
2
)
_
=

H

0
+

V

12
idea: subtract a mean-eld potential v
eff
from the electron-electron
repulsion

V
12
, solve modied

H

0
-problem, and calculate

0
[

12
[

0
) as
the 1
st
order correction. As some parts of the electron-electron inter-
action are now included in

H

0
,

V

12
should be a smaller perturbation
than

V
12
, and the 1
st
order result should be more accurate.
4.4 The hydrogen atom 137
The systematic consideration of this idea leads to the so-called inde-
pendent particle model (IPM), and, in particular, to the Hartree-Fock
method (one has to nd reasonable mean-eld potentials v
eff
, and in
Hartree-Fock theory on uses the variational principle to nd the best
one).
Within the IPM the main characteristics of the (shell-structure of)
atoms (and the periodic table) can be understood.
Another road that can be taken to nd better approximations for E
0
for a many-electron atom is to use the variational principle to construct
approximate two-electron wave functions (beyond IPM models). This
has been very successful for small atoms, but is computationally costly
for true many-electron atoms/systems.
(the most accurate solutions to date - e.g. for He - are obtained along
these lines)
f) Bound atoms molecules
The simplest idea of a molecule is one with xed nuclei. Then, we have to
solve the SE for the electrons in a multiple-center Coulomb potential. The
more complicated geometry makes this more challenging, but such calcula-
tions can be done, e.g., on the level of the IPM (Hartree Fock), and many
aspects of the structure of molecules (e.g., the existence of stable bonds)
can be understood in this fashion. More quantitative/complete calculations
require renements, i.e., electronic structure calculations beyond the IPM,
and an account of the nuclear motion.
A (qualitive) discussion of molecules and chemical bonds can be found in
[Bloc], chap. 14.
138 Applications of the Theory
Chapter 5
Brief introduction into
relativistic Quantum Mechanics
Literature: [BD]; [BS]; [Schb], II; [Jel], II and [Scha], chap. 13. The latter two
sources provide condensed accounts on relativistic Quantum Mechanics.
5.1 Klein-Gordon equation
5.1.1 Formulation
Relativistic energy-momentum relation (classical)
E
2
= p
2
c
2
+m
2
0
c
4
(5.1)


Quantization

: E i
t
(correspondence rules) p

i

E
2

2

t
(free) wave equation (Klein-Gordon equation (KGE))

t
2(r, t) =
2
c
2
(r, t) +m
2
0
c
4
(r, t) (5.2)
rst derived by Schrodinger in winter 1925/26
5.1.2 Discussion
1. KGE is invariant under Lorentz transformations
2. Time development is determined via initial conditions (t
0
),

t
(t
0
) since
KGE is (partial) dierential eq. of second order in t (and in r).
139
140 Brief introduction into relativistic Quantum Mechanics
3. Continuity equation?
one can derive
t
+divj = 0 (cf. eq. (2.62)
with j =
i
2m
(

) (as usual; cf. eq. (2.62))


but : =
i
2mc
2
(

t

t

) (5.3)
problem: (rt) 0 (i.e. not positive denite)
probabilistic interpretation is not possible (or at least not obvious)
4. Ansatz:
(r, t) = Ae
i(krt)
insertion in eq. (5.2) yields together with de Broglie relations
E = =
_
c
2
p
2
+m
2
0
c
4
0 (5.4)
Meaning of E < 0 solution?
5. Add Coulomb potential to free KGE and solve it (in spherical coordinates)
yields wrong ne structure of hydrogen spectrum (i.e. contradicts
experimental ndings)
6. In 1934 KGE was recognized as correct wave equation for spin-0 particles
(mesons).
5.2 Dirac equation
In 1928, Dirac found a new wave equation which is suitable for electrons
(spin
1
2
-particles): the Dirac equation (DE)
5.2.1 Construction of the free DE
Ansatz : i
t
=

H
D
(5.5)
i.e. stick to the form of the TDSE; a partial dierential eq. of 1
st
order in t such
that (t
0
) is the only initial condition (see postulate 3.3.6)
Requirements:
1. DE must be compatible with energy-momentum relation (5.1)
2. DE must be Lorentz-covariant
3. Continuity equation with probabilistic interpretation
5.2 Dirac equation 141
4. Stick to quantization rules!
Dirac recognized that these requirements cannot be satised by a single scalar
equation, but by a matrix equation for a spinor wave function with N components.
Ansatz :

H
D
= c p +m
0
c
2
(5.6)
=
c
i
3

j=1

x
i
+m
0
c
2
with N N matrices
x
,
y
,
z
, and spinor wave function
=
_
_
_

1
(r, t)
.
.
.

N
(r, t)
_
_
_
as solution of (5.5)
requirement (1) is met if each component
i
solves KGE (5.2)
iterate eq. (5.5):
i
t
(i
t
) =

H
D
(

H
D
)

2

t
2 =
_
c
i

x
j
+m
0
c
2
__
c
i

x
k
+m
0
c
2
_

=
_

2
c
2

jk

x
j

x
k
+

i
m
0
c
3

j
_

j
+
j
_

x
j
+
2
m
2
0
c
4
_

=
2
c
2

jk

k
+
k

j
2

2
x
j
x
k
+
2
m
2
0
c
4
+
mc
3
i

j
(
j
+
j
)
x
j

comparison with KGE yields conditions for


i
, :

k
+
k

j
= 2
jk
(5.7)

j
+
j
= 0 (5.8)

2
j
=
2
= 1 (5.9)
Further conditions and consequences:

j
, hermitian (because

H
D
shall be hermitian)
= real eigenvalues
5.9
= eigenvalues are 1
From (5.7)-(5.9) it follows that
j
, are traceless, i.e.
tr
j
= tr = 0
1
1
The trace of a matrix A is dened as the sum over the diagonal elements. The trace does
not change when A is diagonalized. Hence tr A =

eigenvalues.
142 Brief introduction into relativistic Quantum Mechanics
Together with eigenvalues 1 this implies that dimension N is even
N = 2 is too small as there are only three (but not 4) anti-commuting
(eqs. (5.7) and (5.8)) matrices for N = 2 (the Pauli matrices)
try N = 4
derive explicit representations from these conditions

i
=
_
0
i

i
0
_
, =
_
_
_
_
1 0
1
1
0 1
_
_
_
_
(5.10)
with Pauli matrices
i
= free DE takes the form
i
t
_
_
_
_

4
_
_
_
_
= (c p +m
0
c
2
)
_
_
_
_

4
_
_
_
_
(5.11)
and one can derive a meaningful continuity equation:

t
+divj = 0
with =

=
4

i=1

(r, t)(r, t)
and j = c

( i.e. j
k
= c(

1
,

2
,

3
,

4
)
_
0
k

k
0
_
_
_
_
_

4
_
_
_
_
)
5.2.2 Solutions of the free DE
Ansatz :
j
(r, t) = u
j
e
i(krt)
, j = 1, ..., 4
after some calculation one nds:
there are 4 linear independent solutions.
Two correspond to E = +
_
p
2
c
2
+m
2
0
c
4
and two to E =
_
p
2
c
2
+m
2
0
c
4
5.2 Dirac equation 143
they have the form (E > 0):
u
(1)
=
_
_
_
_
1
0

2
_
_
_
_
, u
(2)
=
_
_
_
_
0
1

2
_
_
_
_
(and similarly for E < 0)
with
1
,
2
,

1
,

2
vc
0 : small components
u
(1)
is interpreted as spin up
u
(2)
is interpreted as spin down solution
0
m
0
c
2
-m
0
c
2
E
e e
hole
51 1 keV Ior e

Dirac sea
Figure 5.1: Energy spectrum
Diracs interpretation: (hole theory)
In the vacuum all negative energy states (in the Dirac sea) are occupied.
Hence, if electrons are present at E > m
0
c
2
they cannot fall down into the
Dirac sea because of the Pauli principle.
On the other hand, one can imagine that it is possible to excite one electron
from the Dirac sea to E > m
0
c
2
. Such an excitation corresponds to a hole in
the Dirac sea, which can be interpreted as the presence of a positively charged
particle - an anti-particle (i.e. a positron). This process - electron-positron pair
creation - has indeed been observed, and also the reversed process - destruction of
electron-positron pairs and -ray emission (the latter to balance the total energy).
In fact, the rst experimental detection of positrons in 1932 was considered a
strong proof of Diracs theory.
144 Brief introduction into relativistic Quantum Mechanics
5.2.3 Electromagnetic potentials
minimal coupling prescription (cf. chap. 2.2.3)
p

i
qA = p qA
E i
t
q
5.11
i
t
=
_
c ( p qA) + q +m
0
c
2
_
(5.12)
indeed, one can show that eq. (5.12) is Lorentz-covariant
5.2.4 Relativistic hydrogen atom
Consider eq. (5.12) with A = 0 and
q =
Ze
2
4
0
r
Ansatz : (r, t) = (r)e

Et
(5.13)
(cf. eq. (2.84))
yields

_
c p +m
0
c
2

Ze
2
4
0
r
_
(r) = E(r) (5.14)
can be solved analytically!
Result for bound spectrum ( ne structure)
quantized energy levels:
E
nj
= m
0
c
2
_
1 +
(Z)
2
(n
j
)
2
_

1
2
(5.15)

j
= j +
1
2

_
_
j +
1
2
_
2
(Z)
2
, j =
1
2
,
3
2
, ... (5.16)
(j can be identied as quantum number of total angular momentum)
=

mca
0
=
e
2
4
0
c

1
137
(5.17)
ne-structure constant
Expansion cf. eq. (5.15) in powers of Z 1:
E
nj
= m
0
c
2
_
1
(Z)
2
2n
2

(Z)
4
2n
3
_
1
j +
1
2

3
4n
_
...
_
(5.18)
5.2 Dirac equation 145
1
st
term: rest energy
2
nd
term: non-relativistic binding energy (4.128)
3
rd
term: lowest order relativistic corrections ne structure splitting
of energy levels
E m
0
c
2
n 1
n 2
n 3
1 s
2s2p
3s3p3d
2p
3/2
3d
5/2
3p
3/2
3d
3/2
3s
1 /2
3p
1 /2
2s
1 /2
2p
1 /2
1 . 8 x 1 0
-4
eV
1 s
1 /2
Schrdinger Dirac
Figure 5.2: Energy spectrum of the Coulomb problem
Further corrections of energy spectrum:
hyperne structure (coupling of magnetic moments of electrons + nuclei)
10
6
eV
QED eects (Lamb shift)
10
6
eV
Instead of solving eqs. (5.15)-(5.17) exactly and subsequently expanding the
solutions one can consider a weakly relativistic limit of the stationary DE (5.15)
and solve it in 1
st
order perturbation theory. This procedure yields the result
(5.18) once again, but together with some interpretations concerning the nature
of the relativistic corrections.
One obtains

H
D
vc


H
Schrodinger
+

H
1
+

H
2
+

H
3
146 Brief introduction into relativistic Quantum Mechanics
with

H
1
=
p
4
8m
3
c
2
mass-velocity term corresponds to relativistic correction of non-relativistic
kinetic energy

H
2
=
Ze
2
8
0
m
2
c
2

1
r
3
(s

l)
spin-orbit-coupling a natural consequence of relativistic electron dynamics!
(cf. chap. 4.3.3)

H
3
=

2
8m
2
c
2
V
Coulomb
(r) =
Ze
2

2
8m
2
c
2

0
(r)
Darwin term Zitterbewegung
(its appearance is connected with existence of negative energy solutions)
Appendix A
Some details
A.1 Details on the splitting at the potential well
(x) =
_

_
e
ik
1
x
+Be
ik
1
x
I
Ce
ik
2
x
+De
ik
2
x
II
Ee
ik
1
x
III

(x) =
_

_
ik
1
e
ik
1
x
ik
1
Be
ik
1
x
I
ik
2
Ce
ik
2
x
ik
2
De
ik
2
x
II
ik
1
Ee
ik
1
x
III
matching conditions at x =
L
2
:
:

ik
1
L
2
+Be
ik
1
L
2
= Ce

ik
2
L
2
+De
ik
2
L
2
k
1
e

ik
1
L
2
k
1
Be
ik
1
L
2
= k
2
Ce

ik
2
L
2
k
2
De
ik
2
L
2


1
k
1
+
2e

ik
1
L
2
= C(1 +
k
2
k
1
)e

ik
2
L
2
+D(1
k
2
k
1
)e
ik
2
L
2
[
1
2
e
ik
1
L
2

1
2
_
C(1 +
k
2
k
1
)e
i
2
(k
1
k
2
)L
+D(1
k
2
k
1
)e
i
2
(k
1
+k
2
)L
_
= 1 (*)
147
148 Appendix
matching conditions at x = +
L
2
:
:

Ce
ik
2
L
2
+De

ik
2
L
2
= Ee
ik
1
L
2
k
2
Ce
ik
2
L
2
k
2
De

ik
2
L
2
= k
1
Ee
ik
1
L
2


1
k
2
+
2Ce
ik
2
L
2
= E(1 +
k
1
k
2
)e
ik
1
L
2
[
1
2
e

ik
2
L
2

C =
1
2
E(1 +
k
1
k
2
)e
i
2
(k
1
k
2
)L
(**)

=
D =
1
2
E(1
k
1
k
2
)e
i
2
(k
1
+k
2
)L
(***)
Insert (**) and (***) into (*):
1 =
1
2
_
E
2
(1 +
k
1
k
2
)(1 +
k
2
k
1
)e
i(k
1
k
2
)L
+
E
2
(1
k
1
k
2
)(1
k
2
k
1
)e
i(k
1
+k
2
)L
_
=
E
4
e
ik
1
L
_
(2 +
k
1
k
2
+
k
2
k
1
)e
ik
2
L
+ (2
k
1
k
2

k
2
k
1
)e
ik
2
L
_
= Ee
ik
1
L
_
cos(k
2
L)
i
2
(
k
1
k
2
+
k
2
k
1
) sin(k
2
L)
_
E =
e
ik
1
L
cos(k
2
L)
i
2
_
k
1
k
2
+
k
2
k
1
_
sin(k
2
L)
T = [E[
2
=
1
cos
2
(k
2
L) +
1
4
_
k
1
k
2
+
k
2
k
1
_
2
sin
2
(k
2
L)
A.2 Remarks on photoabsorption 149
A.2 Remarks on photoabsorption
Starting point for the discussion of the interaction of a hydrogen atom (chap.
4.4) and the EM-eld

H =
1
2m
( p +eA)
2
e
Ze
2
4
0
r
=
p
2
2m

Ze
2
4
0
r
. .

H
0
+
e
2m
_
p A+A p
_
+
e
2
2m
A
2
e
. .

V (t)
monochromatic, sourcefree electromagnetic eld can be characterized by vector
potential in Coulomb gauge
A(r, t) = n
_
A
0
e
i(krt)
+A

0
e
i(krt)
_
, (k =

c
)
A = 0 = (n k)
= 0
if A
2
A


H =

H
0
+

V (t)

V (t) =
e
2mi
( A+A )
=
e
mi
A(r, t) =
e
m
A p (Coulomb gauge!)
=
e
m
A
0
e
i(krt)
n p +
e
m
A

0
e
i(krt)
n p
cf. (4.49)
=
_
e
m
A

0
e
ikr
n p
_
e
it
+
_
e
m
A
0
e
ikr
n p
_
e
it
=

B e
it
+

B

e
it
p
abs
ok
[B

k0
[
2
=
e
2
m
2
[A
0
[
2
[
k
[e
ikr
n p[
0
)[
2
transition matrix element
T
k0
=
k
[e
ikr
n p[
0
)
dipole approximation
e
ikr
= e

c
z
1 , justied for k a
1
0
dipole-velocity form of the transition matrix element
T
Dip
k0
=
k
[n p[
0
)
150 Appendix
use commutation relation p =
im

[

H
0
, r]
T
Dip
k0
=
im


k
[

H
0
n r n r

H
0
[
0
)
=
im

(
k

0
)
k
[n r[
0
) dipole length form
= im
k
[n r[
0
)

H
0
is a central-eld problem for the hydrogen atom

j
(r) = R
j
(r)Y
l
j
m
j
()
choose n = e
z
n r = z = r cos =
_
4
3
rY
10
T
Dip
k0
=
_
4
3
im
_

0
r
3
R
k
(r)R
0
(r)
_
Y

l
k
m
k
()Y
10
()Y
l
0
m
0
() d
The angular integral over 3 spherical harmonics is a well-known special case of
the so-called Wigner-Eckart theorem. It is nonzero only if
m
k
= m
0
[l
k
l
0
[ = 1
(dipole) selection rules
Literature: [Mes] QM II, Appendix C; [Lin]
For
0
=
1s
(r)
possible transitions 1s 2p
0
, 3p
0
, . . . (E < 0)
and continuum states with l = 1
!F

F
Figure A.1: dipole characteristic cos
2
dependence
Bibliography
[Arf] Arfken, G.: Mathematical Methods for Physicists. Academic Press.
[Aud] Audretsch, J.: Verschrankte Welt. Wiley-VCH.
[Bal] Ballentine, L. E.: Quantum Mechanics. World Scientic.
[BD] Bjrken, J.D. and Drell, S.D.: Relativistische Quantenmechanik.
B.I.-Hochschultaschenbuch.
[Bloa] Blochincev, D. I.: Grundlagen der Quantenmechnik.
[Blob] Bl ochl, P.: Theoretische Physik II: Elektodynamik. Internetadresse:
http://www2.pt.tu-clausthal.de/atp/education.html.
[Bloc] Bl ochl, P.: Theoretische Physik III: Quantum Mechanics. Interne-
tadresse: http://www2.pt.tu-clausthal.de/atp/education.html.
[Boh] Bohm, D.: Quantum Theory. Prentice Hall.
[BS] Bethe, H. A. and Salpeter, E. E.: Quantum Mechanics of one- and
two-electron atoms. Plenum.
[CT] Cohen-Tannoudji, C.: Quantenmechanik I und II. de Gruyter.
[Daw] Dawydow, A. S.: Quuantenmechanik. Wiley-VCH.
[dE] dEspagnat, B.: Conceptual foundations of Quantum Mechanics.
Perseus Books.
[Dir] Dirac, P. A. M.: The Principles of Qunatum Mechanics. Oxford
University Press.
[DL] Dreizler, R. M. and L udde, C. S.: Theoretische Physik 3. Springer.
[Fey] Feynman, R. P.: Vorlesungen, Band 3. Oldenbourg.
[Fic] Fick, E.: Einf uhrung in die Grundlagen der QT. Aula.
[FK] Fischer, H. and Kaul, H.: Mathematik f ur Physiker II. Teubner.
151
152 Appendix
[Fli] Fliebach, T.: Quantenmechanik. Spektrum.
[Fl u] Fl ugge, S.: Rechenmethoden der Quantentheorie. Springer.
[Fri] Friedrich, H.: Theoretical atomic physics. Springer.
[Gas] Gasiorwicz, S.: Quantenphysik. Oldenbourg.
[Gre] Greiner, W.: Theoretische Physik IV. H. Deutsch.
[Gri] Griffiths, D. J.: Introduction to Quantum Mechanics. Prentice Hall.
[Gro] Gromann, S.: Funktionalanalysis I, II. Aula.
[Hei] Heisenberg, W.: Physikalische Prinzipien der Quantentheorie. B.I.-
Hochschultaschenbuch.
[Heu] Heuser, H: Funktionalanalysis. Teubner.
[Jac] Jackson, J. D.: Klassische Elektrodynamik. de Gruyter.
[Jel] Jelitto, R.: Theoretische Physik IV, V. Aula.
[Kie] Kiefer, C.: Quantentheorie. Fischer.
[Kira] Kirchner, Tom: Theoretische Physik I: Klassische Mechanik. Inter-
netadresse: http://www2.pt.tu-clausthal.de/qd/skripten.html.
[Kirb] Kirchner, Tom: Theoretische Physik II: Elektrodynamik. Interne-
tadresse: http://www2.pt.tu-clausthal.de/ed/skripten.html.
[Lin] Lindner, A.: Drehinpulse in der Quantenmechanik. Teubner.
[LL] Landau, L. D. and Lifschitz, E. N.: Quantenmechanik. Harri
Deutsch.
[Mer] Merzbacher, E.: Qunatum Mechanics. Wiley and Sons.
[Mes] Messiah, A.: Quantenmechanik I, II. de Gruyter.
[Nol] Nolting, W.: Grundkurs Theoretische Physik V. Springer.
[Omna] Omnes, R.: The Interpretation of Quantum Mechanics. Princeton Uni-
versity Press.
[Omnb] Omnes, R.: Understanding Quantum Mechanics. Princeton University
Press.
[Rol] Rollnik, H.: Quantentheorie. Vieweg.
A.2 Remarks on photoabsorption 153
[Scha] Schiff, L. L.: Quantum Mechanics. McGraw-Hill.
[Schb] Schwabl, F.: Quantenmechanik. Springer.
[Sha] Shankar, R.: Principles of Qunatum Mechanics. Kluwer Academic.
[WH] Wachter, A. and Hoeber, H.: Repetitorium Theoretische Physik.
Springer.
[Zeia] Zeilinger, A.: Einsteins Schleier. C. H. Beck.
[Zeib] Zeilinger, A.: Einsteins Spuk. Bertelsmann.