Lecture Notes On Quantum Mechanics

LECTURE NOTES
ON QUANTUM MECHANICS
Dr. Shun-Qing Shen
Department of Physics
The University of Hong Kong
September 2004
Contents
0.1 General Information . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
1 Fundamental Concepts 1
1.1 Relation between experimental interpretations and theoretical inferences . . . 2
1.2 Materials from Britannica Online . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Photoelectric eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2.2 Frank-Hertz experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.2.3 Compton eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.3 The Stern-Gerlach Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 The Stern-Gerlach experiment . . . . . . . . . . . . . . . . . . . . . . 7
1.3.2 Sequential Stern-Gerlach Experiment . . . . . . . . . . . . . . . . . . . 9
1.3.3 Analogy with Polarization of Light . . . . . . . . . . . . . . . . . . . . 11
1.4 Dirac Notation and Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.5 Base kets and Matrix Representation . . . . . . . . . . . . . . . . . . . . . . . 14
1.5.1 Eigenkets of an Observable . . . . . . . . . . . . . . . . . . . . . . . . 14
1.5.2 Eigenkets as Base kets: . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5.3 Matrix Representation: . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.6 Measurements, Observables & The Uncertainty Relation . . . . . . . . . . . . 18
ii
CONTENTS MANUSCRIPT
1.6.1 Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.6.2 Spin1/2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.6.3 Probability Postulate . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6.4 S
r
and S
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
1.6.5 The Algebra of Spin Operators . . . . . . . . . . . . . . . . . . . . . . 23
1.6.6 Observable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
1.7 Change of Basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
1.7.1 Transformation Operator . . . . . . . . . . . . . . . . . . . . . . . 27
1.7.2 Transformation Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
1.7.3 Diagonalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
1.8 Position, Momentum, and Translation . . . . . . . . . . . . . . . . . . . . . . 31
1.8.1 Continuous Spectra . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
1.8.2 Some properties of the cfunction. . . . . . . . . . . . . . . . . . . . . 32
1.8.3 Position Eigenkets and Position Measurements . . . . . . . . . . . . . 33
1.8.4 Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
1.9 The Uncertainty Relation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2 Quantum Dynamics 46
2.1 Time Evolution and the Schrdinger Equation . . . . . . . . . . . . . . . . . . 46
2.1.1 Time Evolution Operator . . . . . . . . . . . . . . . . . . . . . . . . . 46
2.1.2 The Schrodinger Equation. . . . . . . . . . . . . . . . . . . . . . . . . 48
2.1.3 Time Dependence of Expectation Value: Spin Precession. . . . . . . . 52
2.2 The Schrodinger versus the Heisenberg Picture . . . . . . . . . . . . . . . . . 54
2.2.1 Unitary operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
2.2.2 Two Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
iii
CONTENTS MANUSCRIPT
2.2.3 The Heisenberg Equation of Motion. . . . . . . . . . . . . . . . . . . . 56
2.2.4 How to construct a Hamiltonian . . . . . . . . . . . . . . . . . . . . . 57
2.3 Simple Harmonic Oscillator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
2.3.1 Eigenvalue and eigenstates . . . . . . . . . . . . . . . . . . . . . . . . 59
2.3.2 Time Development of the Oscillator . . . . . . . . . . . . . . . . . . . 66
2.3.3 The Coherent State . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
2.4 Schrodinger Wave Equation: Simple Harmonic Oscillator . . . . . . . . . . . 68
2.5 Propagators and Feynman Path Integrals . . . . . . . . . . . . . . . . . . . . 70
2.5.1 Propagators in Wave Mechanics. . . . . . . . . . . . . . . . . . . . . . 70
2.5.2 Propagator as a Transition Amplitude. . . . . . . . . . . . . . . . . . . 75
2.6 The Gauge Transformation and Phase of Wave Function . . . . . . . . . . . . 79
2.6.1 Constant Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
2.6.2 Gauge Transformation in Electromagnetism . . . . . . . . . . . . . . . 81
2.6.3 The Gauge Transformation . . . . . . . . . . . . . . . . . . . . . . . . 84
2.6.4 The Aharonov-Bohm Eect . . . . . . . . . . . . . . . . . . . . . . . . 86
2.7 Interpretation of Wave Function. . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.7.1 Whats
o
(r)? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
2.7.2 The Classical Limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
2.8 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
2.8.1 One dimensional square well potential . . . . . . . . . . . . . . . . . . 91
2.8.2 A charged particle in a uniform magnetic eld . . . . . . . . . . . . . 95
3 Theory of Angular Momentum 98
3.1 Rotation and Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . 98
3.1.1 Finite versus innitesimal rotation . . . . . . . . . . . . . . . . . . . . 99
iv
CONTENTS MANUSCRIPT
3.1.2 Orbital angular momentum . . . . . . . . . . . . . . . . . . . . . . . . 104
3.1.3 Rotation operator for spin 1/2 . . . . . . . . . . . . . . . . . . . . . . 106
3.1.4 Spin precession revisited . . . . . . . . . . . . . . . . . . . . . . . . . . 107
3.2 Rotation Group and the Euler Angles . . . . . . . . . . . . . . . . . . . . . . 108
3.2.1 The Group Concept . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
3.2.2 Orthogonal Group . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
3.2.3 Special? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
3.2.4 Unitary Unimodular Group . . . . . . . . . . . . . . . . . . . . . . . . 110
3.2.5 Euler Rotations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
3.3 Eigenvalues and Eigenkets of Angular Momentum . . . . . . . . . . . . . . . 114
3.3.1 Representation of Rotation Operator . . . . . . . . . . . . . . . . . . . 120
3.4 Schwinger Oscillator Model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
3.4.1 Spin 1/2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
3.4.2 Two-spin1/2 system . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
3.4.3 Explicit Formula for Rotation Matrices. . . . . . . . . . . . . . . . . . 128
3.5 Combination of Angular Momentum and Clebsh-Gordan Coe!cients . . . . . 130
3.5.1 Clebsch-Gordan coe!cients . . . . . . . . . . . . . . . . . . . . . . . . 133
3.6 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.6.1 Two spin-1/2 systems . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
3.6.2 Spin-orbit coupling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137
4 Symmetries in Physics 140
4.1 Symmetries and Conservation Laws . . . . . . . . . . . . . . . . . . . . . . . . 141
4.1.1 Symmetry in Classical Physics . . . . . . . . . . . . . . . . . . . . . . 141
4.1.2 Symmetry in Quantum Mechanics . . . . . . . . . . . . . . . . . . . . 142
v
CONTENTS MANUSCRIPT
4.1.3 Degeneracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143
4.1.4 Symmetry and symmetry breaking . . . . . . . . . . . . . . . . . . . . 144
4.1.5 Summary: symmetries in physics . . . . . . . . . . . . . . . . . . . . . 147
4.2 Discrete Symmetries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
4.2.1 Parity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
4.2.2 The Momentum Operator . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.2.3 The Angular Momentum . . . . . . . . . . . . . . . . . . . . . . . . . 151
4.2.4 Lattice Translation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
4.3 Permutation Symmetry and Identical Particles . . . . . . . . . . . . . . . . . 157
4.3.1 Identical particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 157
4.4 Time Reversal . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.4.1 Classical cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.4.2 Antilinear Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.4.3 Antiunitary operators . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.4.4 T for a zero spin particle . . . . . . . . . . . . . . . . . . . . . . . . . 162
4.4.5 T for a nonzero spin particle . . . . . . . . . . . . . . . . . . . . . . . 162
5 Approximation Methods for Bound States 163
5.1 The Variation Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163
5.1.1 Expectation value of the energy . . . . . . . . . . . . . . . . . . . . . . 164
5.1.2 Particle in a one-dimensional innite square well . . . . . . . . . . . . 165
5.1.3 Ground State of Helium Atom . . . . . . . . . . . . . . . . . . . . . . 166
5.2 Stationary Perturbation Theory: Nondegenerate Case . . . . . . . . . . . . . 168
5.2.1 Statement of the Problem . . . . . . . . . . . . . . . . . . . . . . . . . 168
5.2.2 The Two-State Problem . . . . . . . . . . . . . . . . . . . . . . . . . . 169
vi
CONTENTS MANUSCRIPT
5.2.3 Formal Development of Perturbation . . . . . . . . . . . . . . . . . 170
5.3 Application of the Perturbation Expansion . . . . . . . . . . . . . . . . . . . 174
5.3.1 Simple harmonic oscillator . . . . . . . . . . . . . . . . . . . . . . . . . 174
5.3.2 Atomic hydrogen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
5.4 Stationary Perturbation Theory: Degenerate Case . . . . . . . . . . . . . . . 179
5.4.1 Revisited two-state problem . . . . . . . . . . . . . . . . . . . . . . . . 180
5.4.2 The basic procedure of degenerate perturbation theory . . . . . . . . . 181
5.4.3 Example: Zeeman Eect . . . . . . . . . . . . . . . . . . . . . . . . . . 183
5.4.4 Example: First Order Stark Eect in Hydrogen . . . . . . . . . . . . . 184
5.5 The Wentzel-Kramers-Brillouin (WKB) approximation . . . . . . . . . . . . . 186
5.6 Time-dependent Problem: Interacting Picture and Two-State Problem . . . . 186
5.6.1 Time-dependent Potential and Interacting Picture . . . . . . . . . . . 187
5.6.2 Time-dependent Two-State Problem . . . . . . . . . . . . . . . . . . . 188
5.7 Time-dependent Perturbation Problem . . . . . . . . . . . . . . . . . . . . . . 191
5.7.1 Perturbation Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191
5.7.2 Time-independent perturbation . . . . . . . . . . . . . . . . . . . . . . 193
5.7.3 Harmonic perturbation . . . . . . . . . . . . . . . . . . . . . . . . . . 193
5.7.4 The Golden Rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 194
6 Collision Theory 196
6.1 Collisions in one- and three-dimensions . . . . . . . . . . . . . . . . . . . . . . 197
6.1.1 One-dimensional square potential barriers . . . . . . . . . . . . . . . . 197
6.2 Collision in three dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
6.3 Scattering by Spherically Symmetric Potentials . . . . . . . . . . . . . . . . . 204
6.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
vii
CONTENTS MANUSCRIPT
6.4.1 Scattering by a square well . . . . . . . . . . . . . . . . . . . . . . . . 209
6.4.2 Scattering by a hard-sphere potential . . . . . . . . . . . . . . . . . . 211
6.5 Approximate Collision Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 212
6.5.1 The Lippman-Schwinger Equation . . . . . . . . . . . . . . . . . . . . 212
6.5.2 The Born Approximation . . . . . . . . . . . . . . . . . . . . . . . . . 216
6.5.3 Application: from Yukawa potential to Coloumb potential . . . . . . . 217
6.5.4 Identical Particles and Scattering . . . . . . . . . . . . . . . . . . . . . 218
6.6 Landau-Zener Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
7 Selected Topics 220
7.1 Quantum Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
7.1.1 Density Operator and Ensembles . . . . . . . . . . . . . . . . . . . . . 220
7.1.2 Quantum Statistical Mechanism . . . . . . . . . . . . . . . . . . . . . 223
7.1.3 Quantum Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
7.1.4 Systems of non-interaction particles . . . . . . . . . . . . . . . . . . . 227
7.1.5 Bose-Einstein Condensation . . . . . . . . . . . . . . . . . . . . . . . . 230
7.1.6 Free fermion gas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232
7.2 Quantum Hall Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
7.2.1 Hall Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
7.2.2 Quantum Hall Eect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
7.2.3 Laughlins Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
7.2.4 Charged particle in the presence of a magnetic eld . . . . . . . . . . 240
7.2.5 Landau Level and Quantum Hall Eect . . . . . . . . . . . . . . . . . 243
7.3 Quantum Magnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244
7.3.1 Spin Exchange . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
viii
CONTENTS MANUSCRIPT
7.3.2 Two-Site Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
7.3.3 Ferromagnetic Exchange (J < 0) . . . . . . . . . . . . . . . . . . . . . 249
7.3.4 Antiferromagnetic Exchange . . . . . . . . . . . . . . . . . . . . . . . . 251
0.1 General Information
Aim/Following-up: The course provides an introduction to advanced techniques
in quantum mechanics and their application to several selected topics in condensed
matter physics.
Contents: Dirac notation and formalism, time evolution of quantum systems, angu-
lar momentum theory, creation and annihilation operators (the second quantization
representation), symmetries and conservation laws, permutation symmetry and identi-
cal particles, quantum statistics, non-degenerate and degenerate perturbation theory,
time-dependent perturbation theory, the variational method
Prerequisites: PHYS2321, PHYS2322, PHYS2323, and PHYS2325
Co-requisite: Nil
Teaching: 36 hours of lectures and tutorial classes
Duration: One semester (1st semester)
Assessment: One-three hour examination (70%) and course assessment (30%)
Textbook: J. J. Sakurai, Modern Quantum Mechanics (Addison-Wesley, 1994)
Web page: http://bohr.physics.hku.hk/~phys3332/ All lectures notes in pdf les can
be download from the site.
ix
CONTENTS MANUSCRIPT
References: L. Schi, Quantum Mechanics (McGraw-Hill, 1968, 3nd ed.); Richard
Feynman, Robert B. Leighton, and Matthew L. Sands, Feynman Lectures on Physics
Vol. III, (Addison-Wesley Publishing Co., 1965); L. D. Landau and E. M. Lifshitz,
Quantum Mechanics
Time and Venue: Year 2004 (Weeks 2 14)
11:40 -12:30: Monday/S802
11:40 -12:30: Wednesday/S802
11:40 -12:30: Friday/S802
x
CONTENTS MANUSCRIPT
Numerical values of some physical quantities
~ = 1.054 10
27
erg-sec (Plancks constant divided by 2
c = 4.80 10
10
esu (magnitude of electron charge)
: = 0.911 10
27
g (electron mass)
' = 1.672 10
24
g (proton mass)
a
0
= ~
2
,:c
2
= 5.29 10
9
cm (Bohr radius)
c
2
,a
0
= 27.2eV (twice binding energy of hydrogen)
c = 3.00 10
10
cm/sec (speed of light)
~c,c
2
= 137 (reciprocal ne structure constant)
c~,2:c = 0.927 10
20
erg/oersted (Bohr magneton)
:c
2
= 5.11 10
5
eV (electron rest energy)
'c
2
= 938MeV (proton rest energy)
1eV = 1.602 10
12
erg
1 eV/c = 12, 400
1eV = 11, 600K
xi
Chapter 1
Fundamental Concepts
At the present stage of human knowledge, quantum mechanics can be regarded as the
fundamental theory of atomic phenomena. The experimental data on which it is based
are derived from physical events that almost entirely beyond the range of direct human
perception. It is not surprising that the theory embodies physical concepts that are foreign
to common daily experience.
The most traditional way to introduce the quantum mechanics is to follow the
historical development of theory and experiment Plancks radiation law, the Einstein-
Debyes theory of specic heat, the Bohr atom, de Broglies matter wave and so forth
together with careful analysis of some experiments such as diraction experiment of light,
the Compton eect, and Franck-Hertz eect. In this way we can enjoy the experience
of physicists of last century to establish the theory. In this course we do not follow the
historical approach. Instead, we start with an example that illustrates the inadequacy of
classical concepts in a fundamental way.
1
CHAPTER 1 MANUSCRIPT
1.1 Relation between experimental interpre-
tations and theoretical inferences
Schis book has a good introduction to this theory. Please read the relevant chapter in the
book after class. Here we just list several experiments which played key roles in development
of quantum theory.
Experimental facts
Electromagnetic wave/light: Diraction (Young, 1803; Laue, 1912))
Electromagnetic quanta /light: Black body radiation (Planck, 1900); Photoelectric
eect (Einstein, 1904); Compton eect (1923); Combination Principle (Rita-Rydberg,
1908)
Discrete values for physical quantities: Specic heat (Einstein 1907, Debye 1912);
Franck-Hertz experiment (1913); Stern-Gerlach experiment (1922)
Theoretical development
Maxwells theory for electromagnetism: Electromagnetic wave (1864)
Plancks theory for black body radiation: 1 = ~.: Electromagnetic quanta (1900)
de Broglies theory: 1 = /,`: Wave-Particle Duality, (1924)
Birth of quantum mechanics: Heisenbergs theory (1926); Schrodingers theory (1926)
2
1.2 Materials from Britannica Online
1.2.1 Photoelectric eect
The phenomenon in which charge particles are released from a material when it absorbs
radiant energy. The photoelectric eect commonly is thought of as the ejection of electrons
from the surface of a metal plate when light falls on it. In the broad sense, however, the
phenomenon can take place when the radiant energy is in the region of visible or ultraviolet
light, X rays, or gamma rays; when the material is a solid, liquid, or gas; and when the
particles released are electrons or ions (charged atoms or molecules).
The photoelectric eect was discovered in 1887 by a German physicist, Hein-
rich Rudolf Hertz, who observed that ultraviolet light changes the lowest voltage at which
sparking takes place between given metallic electrodes. At the close of the 19th century,
it was established that a cathode ray (produced by an electric discharge in a rareed-gas
atmosphere) consists of discrete particles, called electrons, each bearing an elementary neg-
ative charge. In 1900 Philipp Lenard, a German physicist, studying the electrical charges
liberated from a metal surface when it was illuminated, concluded that these charges were
identical to the electrons observed in cathode rays. It was further discovered that the current
(given the name photoelectric because it was caused by light rays), made up of electrons
released from the metal, is proportional to the intensity of the light causing it for any xed
wavelength of light that is used. In 1902 it was proved that the maximum kinetic energy
of an electron in the photoelectric eect is independent of the intensity of the light ray and
depends on its frequency.
The observations that (1) the number of electrons released in the photoelectric
eect is proportional to the intensity of the light and that (2) the frequency, or wavelength, of
3
light determines the maximum kinetic energy of the electrons indicated a kind of interaction
between light and matter that could not be explained in terms of classical physics. The
search for an explanation led in 1905 to Albert Einsteins fundamental theory that light,
long thought to be wavelike, can be regarded alternatively as composed of discrete particles
(now called photons ), equivalent to energy quanta.
In explaining the photoelectric eect, Einstein assumed that a photon could
penetrate matter, where it would collide with an atom. Since all atoms have electrons, an
electron would be ejected from the atom by the energy of the photon, with great velocity.
The kinetic energy of the electron, as it moved through the atoms of the matter, would be
diminished at each encounter. Should it reach the surface of the material, the kinetic energy
of the electron would be further reduced as the electron overcame and escaped the attraction
of the surface atoms. This loss in kinetic energy is called the work function , symbolized by
omega (.). According to Einstein, each light quantum consists of an amount of energy equal
to the product of Plancks universal constant (/) and the frequency of the light (indicated
by the Greek i). Einsteins theory of the photoelectric eect postulates that the maximum
kinetic energy of the electrons ejected from a material is equal to the frequency of the
incident light times Plancks constant, less the work function. The resulting photoelectric
equation of Einstein can be expressed by 1
|
= /i . , in which 1
|
is the maximum kinetic
energy of the ejected electron, / is a constant, later shown to be numerically the same as
Plancks constant, i is the frequency of the incident light, and . is the work function.
The kinetic energy of an emitted electron can be measured by placing it in an
electric eld and measuring the potential or voltage dierence (indicated as V) required to
reduce its velocity to zero. This energy is equal to the product of the potential dierence
and an electrons charge, which is always a constant and is indicated by e; thus, 1
|
= c\ .
4
The validity of the Einstein relationship was examined by many investigators and
found to be correct but not complete. In particular, it failed to account for the fact that the
emitted electrons energy is inuenced by the temperature of the solid. The remedy to this
defect was rst formulated in 1931 by a British mathematician, Ralph Howard Fowler, who,
on the assumption that all electrons with energies greater than the work function would
escape, established a relationship between the photoelectric current and the temperature:
the current is proportional to the product of the square of the temperature and a function of
the incident photons energy. The equation is 1 = cT
2
c(r), in which I is the photoelectric
current, a and A are constants, and c(r) is an exponential series, whose numerical values
have been tabulated; the dimensionless value x equals the kinetic energy of the emitted
electrons divided by the product of the temperature and the Boltzmann constant of the
kinetic theory: r = (/i .),/T; in which r is the argument of the exponential series.
1.2.2 Frank-Hertz experiment
in physics, rst experimental verication of the existence of discrete energy states in atoms,
performed (1914) by the German-born physicists James Franck and Gustav Hertz .
Franck and Hertz directed low-energy electrons through a gas enclosed in an
electron tube. As the energy of the electrons was slowly increased, a certain critical electron
energy was reached at which the electron stream made a change from almost undisturbed
passage through the gas to nearly complete stoppage. The gas atoms were able to absorb the
energy of the electrons only when it reached a certain critical value, indicating that within
the gas atoms themselves the atomic electrons make an abrupt transition to a discrete higher
energy level. As long as the bombarding electrons have less than this discrete amount of
energy, no transition is possible and no energy is absorbed from the stream of electrons.
5
When they have this precise energy, they lose it all at once in collisions to atomic electrons,
which store the energy by being promoted to a higher energy level.
1.2.3 Compton eect
increase in wavelength of X rays and other energetic electromagnetic radiations that have
been elastically scattered by electrons; it is a principal way in which radiant energy is ab-
sorbed in matter. The eect has proved to be one of the cornerstones of quantum mechanics,
which accounts for both wave and particle properties of radiation as well as of matter.
The American physicist Arthur Holly Compton explained (1922; published 1923)
the wavelength increase by considering X rays as composed of discrete pulses, or quanta,
of electromagnetic energy, which he called photons . Photons have energy and momentum
just as material particles do; they also have wave characteristics, such as wavelength and
frequency. The energy of photons is directly proportional to their frequency and inversely
proportional to their wavelength, so lower-energy photons have lower frequencies and longer
wavelengths. In the Compton eect, individual photons collide with single electrons that
are free or quite loosely bound in the atoms of matter. Colliding photons transfer some of
their energy and momentum to the electrons, which in turn recoil. In the instant of the
collision, new photons of less energy and momentum are produced that scatter at angles the
size of which depends on the amount of energy lost to the recoiling electrons.
Because of the relation between energy and wavelength, the scattered photons
have a longer wavelength that also depends on the size of the angle through which the X
rays were diverted. The increase in wavelength or Compton shift does not depend on the
wavelength of the incident photon.
The Compton eect was discovered independently by the physical chemist Peter
6
Figure 1.1: The Stern-Gerlach Experiment
Debye in early 1923.
1.3 The Stern-Gerlach Experiment
We start with the Stern-Gerlach experiment to introduce some basic concepts of quantum
mechanics. The two-state problem can be regarded as the most quantum. A lot of important
discoveries are related to it. It is worthy studying very carefully. We shall repeat to discuss
the problem throughout this course.
1.3.1 The Stern-Gerlach experiment
Oven: silver atoms (Ag) are heated in the oven. The oven has a small hole through which
some of the silver atoms escape to form an atomic beam.
Ag: (Electron conguration: 1s
2
2s
2
2p
6
3s
2
3p
6
3d
10
4s
2
4p
6
4d
10
5s
1
). The outer
7
shell has only ONE electron (5s
1
) . The atom has an angular momentum., which is due
solely to the spin of 47
||
electron.
Collimating slit: change the diverging Ag beam to a parallel beam.
Shaped Magnet: N and S are north and south poles of a magnet. The knife-edge
of S results in a much stronger magnetic eld at the point P than Q . i.e. the magnet
generates an inhomogeneous magnetic eld.
Role of the inhomogeneous eld: to change the direction of the Ag beam. The
interaction energy is 1 = j B. The force experienced by the atoms is
:a =
0
0.
(j B) - j
:
01
:
0.
If the magnetic eld is uniform , i.e.
d1
z
d:
= 0., the Ag beam will not change its direction. In
the eld dierent magnetic moments experience dierent force, and the atoms with dierent
magnetic moments will change dierent angles after the beams pass through the shaped
magnet. Suppose the length of the shaped magnet 1. It takes a time 1, for particles to
go though the magnets. The angle is about
at
z
n
d1
z
d:
J
u
=
j
:
d1
z
d:
1
:
2
= j
:
01
:
0.
1,21
|
j
:
.
They will reach at dierent places on the screen.
Classically: all values of j
:
= jcos 0 (0 < 0 < ) would be expected to realize
between |j| and |j| . It has a continuous distribution.
Experimentally: only two values of z component of

o are observed (electron spin
j

o)
Consequence: the spin of electron has two discrete values along the magnetic
8
Figure 1.2: Beam from the Stern-Gerlach apparatus: (a) is expected from classical physics,
while (b) is actually observed experimentally.
eld
o =
;
A
A
?
A
A
=
~,2
~,2
~ = 1.0546 10
27
erg.s = 6.5822 10
16
eV.s
Plancks constant divided by 2
It should be noted that the constant cannot be determined accurately from this experiment.
1.3.2 Sequential Stern-Gerlach Experiment
SGz stands for an apparatus with inhomogeneous magnetic eld in z direction, and SGx in
x direction
Case(a): no surprising!
Case(b): o
:
+ beam is made up of 50% o
r
+ and 50% o
r
? o
:
beam?
9
Figure 1.3: Sequential Stern-Gerlach experiment
Case(c): Since o
:
is blocked at the rst step, why is o
r
+ beam made up of
both o
:
+ and o
:
beams?
Consequences of S-G experiment:
Spin space cannot be described by a 3-dimensional vector.
The magnetic moment of atom or spin is discrete or quantized.
We cannot determine both o
:
and o
r
simultaneously. More precisely, we can say that
the selection of o
r
+ beam by SG x completely destroys any previous information about
o
:
.
10
1.3.3 Analogy with Polarization of Light
It is very helpful if you compare the situation with the analogy of the polarized light through
a Polaroid lter. The polarized light wave is described by a complex function,
E = 1
0
xcos(/. .t).
Please refer to Sakurais book, P.610.
1.4 Dirac Notation and Operators
The Stern-Gerlach experiment shows that the spin space is not simply a 3-D vector space
and lead to consider a complex vector space. There exists a good analogy with polarization
of light. (Refer to Sakurais book). In this section we formulate the basic mathematics
of vector spaces as used in quantum mechanics. The theory of linear algebra has been
known to mathematician before the birth of quantum mechanics, but the Dirac notation
has many advantages. At the early time of quantum mechanics this notation was used to
unify Heisenbergs matrix mechanics and Schrodingers wave mechanics. The notation in
this course was rst introduced by P. A. M. Dirac.
Ket space: In quantum mechanics, a physical state, for example, a silver atom
with denite spin orientation, is represented by a state vector in a complex vector space,
denoted by |ci, a ket. The state ket is postulated to contain complete information about
physical state. The dimensionality of a complex vector space is specied according to the
nature of physical system under consideration.
An observable can be represented by an operator A
A(|ci) = A|ci
11
Eigenket and eigenvalue:
A(|ci) = a |ci
Eigenstate: the physical state corresponding to an eigenket, |ci.
Two kets can be added to form a new ket: |ci +|,i = |i
One of the postulates is that |ci and c |ci with the number c 6= 0 represent the
same physical state.
Bra space: a dual correspondence to a ket space. We postulate that corre-
sponding to every ket there exist a bra. The names come from the word bracket
bra-c-ket.
1. There exists a one-to one correspondence between a ket space and a bra space.
|ci ==hc|
2. The bra dual to c |ci (c is a complex number)
c |ci ==c
hc|
Inner product: In general, this product is a complex number.
h,|ci = (h,|) (|ci)
Two properties of the inner product:
1. h,|ci and hc|,i are complex conjugates of each other.
h,|ci = (hc|,i)
12
2. the postulate of positive denite metric.
hc|ci _ 0
where the equality sign holds only if |ci is a null ket. From a physicists point of view
this postulate is essential for the probabilistic interpretation of quantum mechanics.
Question: what happens if we postulate hc|ci _ 0?
Orthogonality:
hc|,i = 0.
Normalization: all vectors can be normalized!
| ci =
1
hc|ci
1
2
|ci , h c| ci = 1
The norm of |ci : (hc|ci)
1/2
Operator: an operator acts on a ket from the left side, x|ci , and the resulting
product is another ket. An operator acts on a bra from the right side, hc| x,and the resulting
product is another bra.
Equality: x = y if x|ci = y|ci for an arbitrary ket
Null operator: x|ci = 0 for arbitrary |ci
Hermitian operator: x = x
x|ci hc| x
Multiplication
Noncommutative: xy 6= yx
Associative: x(yz) = (xy)z = xyz
13
Hermitian adjoint: (xy)
= y
(Prove it!)
y|ci = hc| y
xy|ci = x(y |ci) =
hc| y
= hc| y
Outer product: |ci h,| is an operator!

The associative axiom of multiplication: the associative property holds as long
as we are dealing with legal multiplications among kets, bras, and operators.
1. (|ci h,|) |i = |ci (h,| |i) ;
2. (h,|) (x|ci) = (h,| x) (|ci) = h, |x| ci ;
3. h, |x| ci = h,| (x|ci) = {(hc| x
) |,i}
.
1.5 Base kets and Matrix Representation
1.5.1 Eigenkets of an Observable
The operators for observables must be Hermitian as physical quantities must be real.
1). The eigenvalues of a Hermitian operator A are real.
Proof. First, we recall the denitions
A|ai = a |ai ==ha| A
= ha| a
ha| A|ai = a =ha| A
|ai = a
If A = A
==a = a
.
14
2). The eigenkets of Hermitian A corresponding to dierent eigenvalues are
orthogonal.
Proof.
A|ai = a |ai ==ha
0
|A| ai = a ha
0
|ai
ha
0
| A
= ha
0
| a
0
==ha
0
|A| ai = a
0
ha
0
|ai
We have
(a a
0
) ha
0
|ai = 0 ==ha
0
|ai = 0
It is conventional to normalize all eigenkets |a
0
i of A so that {|a
0
i} form an orthogonal set:
ha|a
0
i = c
o
0
,o
00 .
1.5.2 Eigenkets as Base kets:
1. All normalized eigenkets of an observable A form a complete and orthonormal set.
{
a
(n)
}. The number of the eigenkets is equal to the dimensionality of the complex

vector space.
2. An arbitrary ket in the ket space can be expanded in terms of the eigenkets of A:
|ci =
P
n
C
n
a
(n)
E
==
D
a
(n)
|c
E
= C
n
3. The completeness relation (or closure):
P
n
a
(n)
ED
a
(n)
= 1
15
(
P
n
a
(n)
ED
a
(n)
)(
P
n
a
(n)
ED
a
(n)
)
=
P
n.n
a
(n)
ED
a
(n)
a
(n)
ED
a
(n)
=
P
n.n
a
(n)
E
c
nn
D
a
(n)
=
P
n
a
(n)
ED
a
(n)
==(
P
n
a
(n)
ED
a
(n)
)(
P
n
a
(n)
ED
a
(n)
1) = 0
This relation is very essential to develop the general formalism of quantum mechanics.
4. Projection operator: 1
n
=
a
(n)

a
(n)
1
n
|ci =
a
(n)
ED
a
(n)
ci = C
n
a
(n)
E
1
n
selects that portion of the ket |ci parallel to

a
(n)
: 1
n
1
n
= 1
n
such that 1
n
s
eigenvalues are either 1 or 0.
1
n
1
n
= (
a
(n)
ED
a
(n)
)(
a
(n)
ED
a
(n)
)
=
a
(n)
ED
a
(n)
|a
(n)
ED
a
(n)
a
(n)
ED
a
(n)
= 1
n
1.5.3 Matrix Representation:
Having specied the base kets, we show how to represent an operator, say X, by a square
matrix.
X = (
P
n
a
(n)
ED
a
(n)
)X(
P
n
a
(n)
ED
a
(n)
)
=
P
n.n
a
(n)
ED
a
(n)
a
(n)
ED
a
(n)
=
P
n.n
(
D
a
(n)
a
(n)
E
)
a
(n)
ED
a
(n)
=
P
n.n
A
nn
a
(n)
ED
a
(n)
16
Let us assume that
a
(n)
= (0, , 0, 1, ) and
a
(n)
E
=
3
E
E
E
E
E
E
E
E
E
E
E
E
E
E
C
0
.
.
.
0
1
.
.
.
4
F
F
F
F
F
F
F
F
F
F
F
F
F
F
D
Thus the outer product of

a
(n)
and

a
(n)
gives a new operator in the form of square

matrix,
a
(n)
ED
a
(n)
=
3
E
E
E
E
E
E
E
E
E
E
C
0 0
.
.
.
.
.
.
.
.
.
.
.
.
0 1
.
.
.
.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
F
F
F
F
D
.
Thus X can be expressed in a matrix form,
X =
3
E
E
E
E
E
E
C
a
(1)
a
(1)

a
(1)
a
(2)
a
(2)
a
(1)

a
(2)
a
(2)

.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
D
.
For an operator A which eigenstates are

a
(n)
A =
3
E
E
E
E
E
E
C
a
(1)
0
0 a
(2)

.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
D
Basic Matrix Operations:
1. Transpose: C = A
T
==c
I
= a
I
2. Addition: C = A+B ==c
I
= a
I
+/
I
17
3. Scalar-matrix multiplication: C = cA ==c
I
= ca
I
4. Matrix-matrix multiplication: C = AB ==c
I
=
:
P
|=1
a
I|
/
|I
1.6 Measurements, Observables &The Uncer-
tainty Relation
1.6.1 Measurements
Measurement has special meaning in quantum mechanics. It is a physical operation proce-
dure. It is quite dierent from that in classical physics. In the classic case the measurement
reects the state of an object, such as position, velocity, and energy. The process of measure-
ment does not change the state of the object. For instance we measure your body height,
which does not change your body height. In the quantum case the process of measurement
changes the state of the object. If we want to measure the spin state of electron we have to
let electron go through the Stern-Gerlach apparatus. The magnetic eld changes the spin
state, and select one of the eigenstates along the direction of the magnetic eld. This is not
a particular easy subject for a beginner. Let us rst read the words of the great master, P.
A. M. Dirac, for guidance.
A measurement always causes the system to jump into an eigenstate of the
dynamic variable that is being measured .
P.A.M. Dirac
Interpretation word by word:
1. Before the measurement of observable is made, the system is assumed to be repre-
18
sented by some linear combination
|ci = (
P
n
a
(n)
ED
a
(n)
ci =
P
n
c
n
a
(n)
E
2. When the measurement is performed, the system is thrown into one of the eigen-
states, say

a
(1)
, of the observable
|ci A measurement
a
(1)
E
3. A measurement usually changes the state. When the measurement causes |ci to
change into

a
(1)
it is said that A is measured to be a

(1)
!
1.6.2 Spin1/2 system
We are ready to derive the spin 1/2 operator we encounter in the Stern-Gerlach experiment.
The eigenvalues of S
:
are ~,2. We denote the corresponding eigenkets as |+i and |i,
respectively,
h+|+i = h|i = 1; h+|i = 0.
|+i and |i form a complete (?) and orthonormal (?) set of basis. From the completeness
relation: 1 = (|+i h+|) + (|i h|), we have
S
:
|+i =
~
2
|+i
S
:
|i =
~
2
|i
<
A
A
@
A
A
>
=
S
:
|+i h+| =
~
2
|+i h+|
S
:
|i h| =
~
2
|i h|
<
A
A
@
A
A
>
S
:
(|+i h+| +|i h|) =
~
2
(|+i h+| |i h|)
S
:
=
~
2
(|+i h+| |i h|)
Let us construct two operators:
S
+
= ~|+i h|
S
= ~|i h+|
19
From the denition, we get
;
A
A
?
A
A
=
S
+
|+i = ~ |+i h|+i = 0
S
+
|i = ~ |+i h|i = ~ |+i
;
A
A
?
A
A
=
S
|+i = ~|i h+|+i = ~|i

S
|i = ~ |i h+|i = 0
Let
|+i =
3
E
E
C
1
0
4
F
F
D
; |i =
3
E
E
C
0
1
4
F
F
D
On the base, we have
S
:
=
~
2
3
E
E
C
1 0
0 1
4
F
F
D
; S
+
= ~
3
E
E
C
0 1
0 0
4
F
F
D
; S
= ~
3
E
E
C
0 0
1 0
4
F
F
D
1.6.3 Probability Postulate
The probability for jumping into some particular

a
(n)
is given by
the probability for
a
(n)
E
: j
n
=
D
a
(n)
|c
E
2
provided that |ci is normalized (hc|ci = 1). (|ci =
P
a
(n)

a
(n)
|c
). The expectation
value of A with respect to state |ci is dened as
hAi = hc|A| ci =
P
n.n
D
c|a
(n)
ED
a
(n)
|A| a
(n)
ED
c
(n)
|a
E
=
P
n
a
(n)
D
c|a
(n)
E
2
=
P
n
a
(n)
j
n
where j
n
is the probability in the state

a
(n)
.
Selective measurement (or ltration)
20
We imagine a measurement process with a device that selects only one of the
eigenket and rejects all others. Mathematically, we can say that such a selective measurement
amounts to applying the projection operator.
1.6.4 S
r
and S
j
We are now in the position to determine the eigenkets of S
r
and S
and the operators from

the results of sequential Stern-Gerlach experiments.(see Fig.1.3)
Case (b): The eigenstates of S
r
and S
can be expressed in terms of the eigen-

states of S
:
. Similar to S
:
:
S
r
=
~
2
(|o
r
+i ho
r
+| |o
r
i ho
r
|)
From experiment == |h+|o
r
+i|
2
= |h+|o
r
i|
2
= 1,2
From experiment == |h|o
r
+i|
2
= |h|o
r
i|
2
= 1,2
==
;
A
A
?
A
A
=
|o
r
+i =
1
2
12
|+i +
1
2
12
c
Io
1
|i
|o
r
i =
1
2
12
|+i +
1
2
12
c
Io
0
1
|i
ho
r
+|o
r
i = 0
1
2
h+|+i +
1
2
h|i c
Io
1
+Io
0
1
= 0 = c
0
1
= +c
1
S
r
=
~
2
(|o
r
+i ho
r
+| |o
r
i ho
r
|)
=
~
2
(c
Io
1
|+i h| +c
Io
1
|i h+|)
Likewise,
S
=
~
2
(c
Io
2
|+i h| +c
Io
2
|i h+|)
21
How to determine c
1
and c
2
?
From experiment |ho
|o
r
+i|
2
= |ho
|o
r
i|
2
= 1,2
== c
1
c
2
= ,2
We just can determine the dierence of c
1
and c
2.
For convenience we take S
r
to be real
such that,
c
1
= 0 and c
2
= ,2.
Therefore,
|o
r
i =
1
2
1/2
(|+i |i);
|o
i =
1
2
1/2
(|+i i |i)
and
S
r
=
~
2
(|+i h| +|i h+|)
S
=
~
2
(i |+i h| +i |i h+|)
In the form of matrix.
S
r
=
~
2
3
E
E
C
0 1
1 0
4
F
F
D
; S
=
~
2
3
E
E
C
0 i
i 0
4
F
F
D
; S
:
=
~
2
3
E
E
C
1 0
0 1
4
F
F
D
The relations between S
r
, S
and S
:
S
+
= S
r
+iS
= S
r
iS
Dene
S
o
=
~
2
o
o
22
where the dimensionless o
o
are called the Pauli matrices:
r
=
3
E
E
C
0 1
1 0
4
F
F
D
;
=
3
E
E
C
0 i
i 0
4
F
F
D
;
:
=
3
E
E
C
1 0
0 1
4
F
F
D
.
The eigenvalues of these three matrixes are 1. (Please check it after class!)
1.6.5 The Algebra of Spin Operators
1. The commutation relations
[S
r
, S
] = i~S
:
;
[S
, S
r
] = i~S
:
;
[S
I
, S
] = ic
I|
~S
|
3
E
E
C
c
r:
= c
:r
= c
:r
= 1
c
:r
= c
r:
= c
r:
= 1
4
F
F
D
c
I|
is an antisymmetric tensor.
2. The anticommutation relations:
{S
I
, S
} =
1
2
~
2
c
I
3. The increasing and decreasing operators: S
= S
r
iS
[S
:
, S
] = [S
:
, S
r
] [S
:
, S
] = iS
i(iS
r
)
= S
r
+iS
= S
[S
, S
+
] = 2S
:
23
4. The square operator of

o
S
2
= S
2
r
+S
2
+S
2
:
=
3
4
~
2
3
E
E
C
1 0
0 1
4
F
F
D
=
3
4
~
2
3
4
=
1
2

1 +
1
2
1.6.6 Observable
Compatible Observables: Observables A and B are dened to be compatible when the
corresponding operators commute,
[A, B] = 0
and incompatible when
[A, B] 6= 0
Degeneracy: Suppose there are two or more linear independent eigenkets of
A having the same eigenvalues then the eigenvalues of the two eigenkets are said to be
degenerate.
If a ket |ci is an eigenket for both A and B, i.e.
A|ci = a |ci B|ci = / |ci
the ket |ci is a simultaneous eigenket of A and B. We can denote it by |ci = |a, /i ,e.g.
[S
2
, S
:
] = 0
S
2
|i =
3
4
~
2
|i ; S
:
|i =
~
2
|i
S
2
(c|+i +, |i) =
3
4
~
2
(c|+i +, |i)
S
:
(c|+i +, |i) =
~
2
(c|+i , |i)
24
Theorem 1 Suppose that A and B are compatible observable, and eigenvalues of A are
nondegenerate. Then the matrix elements ha |B| a
0
i are all diagonal. (A|ci = a |ci )
Proof. Suppose A is diagonal on the basis ket {
a
(n)
}
A
c
(n
E
= a
(n)
c
(n)
E
and all a
(n)
6= a
(n)
if a 6= : A and B are compatible, [A, B] = 0
D
a
(n)
|[A, B]| a
(n)
E
=
D
a
(n)
|[ABBA]| a
(n)
E
= (a
(n)
a
(n)
)
D
a
(n)
|B| a
(n)
E
= 0
==
D
a
(n)
|B| a
(n)
E
= 0 if : 6= :
If A and B are compatible, we can always diagonalize them on one set of basis
ket simultaneously no matter whether A has degenerate eigenvalues!
Measurement of A and B when [A, B] = 0
1. Nondegenerate
|ci |a, /i
2. Degenerate
|ci
P
I
c
(I)
o
a
0
, /
(I)
E
a
0
, /
(I)
E
Whether or not there is degeneracy, A measurements and B measurements do not
interfere. We can measure A and B accurately at the same time!
25
Incompatible Observables
1. Incompatible observables do not have a complete set of simultaneous eigenket.
Proof. Proof: Assume A and B are incompatible and {|a
0
, /
0
i} is a complete set of simul-
taneous eigenkets. Then
AB|a
0
, /
0
i = A/
0
|a
0
, /
0
i = a
0
/
0
|a
0
, /
0
i
BA|a
0
, /
0
i = Ba
0
|a
0
, /
0
i = a
0
/
0
|a
0
, /
0
i
<
A
A
@
A
A
>
(ABBA) |a
0
, /
0
i = [A, B] |a
0
/
0
i = 0 for all {a
0
/
0
}
== [A, B] = 0
This is in contradiction to the assumption!
1. Accidently, there may exist an simultaneous eigenket r a subset of simultaneous eigen-
ket for incompatible A and B.
1.7 Change of Basis
Example 1
Spin 1/2 system:
Basis I:{|o
:
: +i , |o
:
: i}
S
:
=
~
2
(|o
:
: +i ho
:
: +| |o
:
: i ho
:
: |)
S
r
=
~
2
(|o
:
: +i ho
:
: | |o
:
: i ho
:
: +|)
S
=
~
2
(i |o
:
: +i ho
:
: | +i |o
:
: i ho
:
: +|)
26
Basis II: {|o
r
: +i , |o
r
: i}
S
r
=
~
2
(|o
r
: +i ho
r
: +| |o
r
: i ho
r
: |)
S
=
~
2
(|o
r
: +i ho
r
: | |o
r
: i ho
r
: +|)
S
:
=
~
2
(i |o
r
: +i ho
r
: | +i |o
r
: i ho
r
: +|)
Connection of two basis
|o
r
: i =
1
2
1/2
|o
:
: +i
1
2
1/2
|o
:
: i
3
E
E
C
|o
r
: +i
|o
r
: i
4
F
F
D
=
3
E
E
C
1
2
12
1
2
12
1
2
12

1
2
12
4
F
F
D
3
E
E
C
|o
:
: +i
|o
:
: i
4
F
F
D
1.7.1 Transformation Operator
Suppose we have two incompatible observables A and B. The ket space can be viewed
as being spanned either by the set {|a
0
i}or by the set {|/
0
i}. i.e. A representation or B
representation. Changing the set of base kets is referred to as a change of basis or change
of representation. The two bases are connected by a transformation operator.
Theorem 2 Given two complete and orthogonal sets of base kets, there exist a unitary
operator U such that

/
(n)
= U
a
(n)
.
Unitary operator: U U
+
= U
+
U = 1
Proof. This theorem is proved by an explicit construction of u: the summation of the
out-product of two sets of eigenket forms an operator
U =
P
|
/
(|)
ED
a
(|)
27
1)
U
a
(l)
E
=
P
|
/
(|)
ED
a
(|)
|a
(l)
E
| {z }
=
/
(l)
E
2)
UU
+
=
P
|,l
/
(|)
ED
a
(|)
a
(l)
ED
/
(l)
=
P
|,l
/
(|)
ED
/
(|)
= 1
1.7.2 Transformation Matrix
In the matrix representation, the u operator can be expressed in the form of a matrix in the
base ket {
a
(l)
}
U = {
D
a
(|)
|n| a
(l)
E
} = {
D
a
(|)
|/
(l)
E
}
For an arbitrary ket in two representations
A:
|ci =
P
n
a
(n)
ED
a
(n)
|c
E
B:
|ci =
P
n
/
(n)
ED
/
(n)
|c
E
Relation of two representations:
D
/
(n)
|c
E
=
P
|
D
/
(n)
|a
(|)
ED
a
(|)
|c
E
=
P
|
D
a
(n)
a
(|)
ED
a
(|)
|c
E
28
Relation of an operator in two representations
D
/
(n)
|X| /
(n)
E
=
P
|,l
D
/
(n)
|a
(|)
ED
a
(|)
|X| a
(l)
ED
a
(l)
|/
(n)
E
=
P
|,l
D
a
(n)
a
(|)
ED
a
(|)
|X| a
(l)
ED
a
(l)
|U| a
(n)
E
=
D
a
(n)
Xa
(l)
U
a
(n)
E
== X
b
= U
X
o
U
A similarity transformation
Trace of an operator:
T
:
(X) =
P
n
D
a
(n)
c
(n)
E
=
P
n.|.l
D
a
(n)
|/
(|)
ED
/
(|)
|X| /
(l)
ED
/
(l)
|a
(n)
E
=
P
n.|.l
D
/
(l)
|a
(n)
ED
a
(n)
|/
(|)
ED
/
(|)
|X| /
(l)
E
=
P
|.l
D
/
(l)
|/
(|)
ED
/
(|)
|X| /
(l)
E
=
P
l
D
/
(l)
|X| /
(l)
E
The trace of X is independent of representation. We can also prove:
t
:
(rj) = t
:
(jr); t
:
(n
+
rn) = t
:
(r)
29
1.7.3 Diagonalization
When we know the matrix elements of B in the base {|a
0
i}, how do we obtain the eigenvalues
and eigenket of B:
1
/
(l)
E
= /
l
/
(l)
E
In the base ket {
/
(l)
}, the operator is expressed as

B
/
(l)
ED
/
(l)
= /
l
/
(l)
ED
/
(l)
B
X
l
/
(l)
ED
/
(l)
=
X
l
/
l
/
(l)
ED
/
(l)
B =
X
l
/
l
/
(l)
ED
/
(l)
In the Dirac bra-ket notation, we rewrite it as

P
n
D
a
(n)
|B| a
(n)
ED
a
(n)
|/
(l)
E
= /
l
D
a
(n)
|/
(l)
E
.
Furthermore, in the matrix form,
3
E
E
E
E
E
E
C
1
11
1
12

1
21
1
22

.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
D
3
E
E
E
E
E
E
C
C
(l)
1
C
(l)
2
.
.
.
4
F
F
F
F
F
F
D
= /
(l)
3
E
E
E
E
E
E
C
C
(l)
1
C
(l)
2
.
.
.
4
F
F
F
F
F
F
D
where
1
nn
=
D
a
(n)
|B| a
(n)
E
and
C
(l)
|
=
D
a
(|)
|/
(l)
E
As we know from linear algebra, the eigenvalues are determined by
det(B`) = 0
30
Example 2
S
r
=
~
2
3
E
E
C
0 1
1 0
4
F
F
D
det(S
r
`1) = 0 =det
3
E
E
C
`
~
2
~
2
-`
4
F
F
D
= 0
`
2
~
2
4
= 0 =` =
~
2
Eigenvectors?
1.8 Position, Momentum, and Translation
1.8.1 Continuous Spectra
When the eigenvalues of an observable is continuous, the dimensionality of the vector space
is innite. Denote the eigenvalue-eigenket relation by
|i = |i
We generalize of a vector space with nite dimension to that with innite dimension
discrete continuous
ha
0
|ai = c
o
0
o
0
|
= c(
0
)
P
|a
0
i ha
0
| = 1
R
d |i h| = 1
|ci =
P
n
a
(n)

a
(n)
ci |ci =
R
d |i h| ci
ha |A| a
0
i = a
0
c
o
0
o
||
0
=
0
c(
0
)
31
1. The Kronecker symbol c is replaced by Diracs c -function.
2. The discrete sum over the eigenvalues is replaced by an integral over a continuous
variable.
1.8.2 Some properties of the function.
1. c(r) = c(r)
2. c
0
(r) = c
0
(r)
3. rc(r) = 0
4. rc
0
(r) = c(r)
5. c(ar) =
1
o
c(r)
6. c(r
2
a
2
) =
1
2
(c(a r) +c(a r))
7. c(r a) )(r) = (c(r a))(a)
8.
R
drc(a r)c(/ r) = c(a /)
Representation of the cfunction
c(r) = lim
$4
sinqr
r
c(r) = lim
c$0
1
1/2
o
c
r
2
/c
2
Normalization in terms of the cfunction
n
|
(r) = cc
I|
r
32
+4
R
4
drn
|
(r)n
l
(r) = c
2
+4
R
4
c
I(|
)r
= c
2
2c(/
r
|
r
) = c(/
r
|
r
)
= c = (2)
1/2
1.8.3 Position Eigenkets and Position Measurements
To extend the idea of a ltering process to measurements of observables exhibiting continues
spectra, we consider the position operator in one dimension. The eigenkets |r
0
i of the
position operator x satisfying
x|r
0
i = r
0
|r
0
i
are postulate to form a complete set. The state ket for an arbitrary physical states can be
expanded in terms of {|r
0
i}
|ci =
Z
d
r
0 |r
0
i hr
0
|ci
Suppose we place a very tiny detector that clicks only when the particle is precisely at x
and nowhere else after the detector clicks, we can say the state in question is represented
by |r
0
i. In practice the best the detector can do is to locate the particle within a narrow
range (r
0
o

2
, r
0
o
+

2
) ( is very small). When the detector clicks, the state ket changes
abruptly as follows
|ci =
Z
4
4
d
r
0 |r
0
i hr
0
|ci measurement
r
o
+
2
R
r
o
2
d
r
0 |r
0
i hr
0
|ci
Assume that hr
0
|ci does not change appreciably within the narrow interval, the probability
that the detector clicks is given by
|hr
0
|ci|
2
33
Obviously
+4
R
4
|hr|ci|
2
dr = 1 = hc|ci .
In the wave mechanics, the physical state is represented by a wave function. Thus hr|ci is
the wave function for the state |ci
Wave function in position space: hr
0
|ci in |ci =
R
d
r
0 |r
0
i hr
0
|ci is the wave
function in the position space. We assume that the position eigenkets |r
0
i are complete. It
can be three dimensional, i.e. x in |r
0
i stand for three components of the position vectors.
In this word the position is an observable, and [r
I
, r
] = 0.
1.8.4 Translation
Suppose we start with a state that is well located around x, and introduce an translation
operator T(|) such that
T (|) |ri = |r +|i .
Several properties of T(|):
1. T(| = 0) = 1
2. T(|
1
)T (|
2
) = T(|
1
+|
2
)
3. T(|)T(|) = 1 ==T(|) = T
1
(|)
4. T(|)
+
T(|) = 1
From the denition we have
hr| T(|)
= hr +||
T(|)
T(|)
= hr +||r +|i = hr|ri

34
We now consider an innitesimal translation | = cr 0. We demonstrate that if T(cr) is
taken to be
T(cr) r 1 iKcr + 0((cr)
2
)
where K is an Hermitian operator K
= K. then all properties listed are satised.

Let us check them..
1. cr = 0 :
T(cr = 0) = 1
2. T(cr
1
)T(cr
2
) = T(cr
1
+cr
2
) :
(1 i1cr
1
)(1 i1cr
2
)
= 1 i1(cr
1
+cr
2
) + 0(cr
1
cr
2
)
3. T(cr)T(cr) = 1
(1 +i1cr)(1 i1cr) = 1 i1
2
(cr)
2
= 1
4. T(cr)
T(cr) = 1
(1 iKcr)
(1 iKcr) = (1 +iK
cr)(1 iKcr)
= 1 +i(K
K)cr + 0((cr)
2
)
35
Commutation relation of x and K:
xT(cr) |ri = x|r +cri = (r +cr) |r +cri
T(cr)x|ri = T(cr)r|ri = rT(cr) |ri = r|r +cri
== (xT(cr) T(cr)x) |ri = cr|r +cri
== (ixKcr +iKxcr) |ri = cr|r +cri - cr|ri
== [x, K] = i
x and K do not commute with each other!
Momentum Operator in the Position Basis.
Translation Operator for an innitesimal small displacement is dened as (T(cr) =
1 iKcr). Let T(cr) act on an arbitrary state
|ci T(cr) |ci =
Z
drT(cr) |ri hr|ci =
Z
dr|r +cri hr|ci
=
Z
dr|ri hr cr|ci r +c
r
r
=
Z
dr|ri (hr|ci
0
0r
hr|ci cr) =
Z
dr|ri (1 cr
0
0r
) hr|ci
also ==
Z
dr|ri hr| (1 iKcr) |ci
=
Z
dr|ri (hr|ci iKhr|ci cr) ==1 = i
0
0r
Consider an ideal wave function
hr|ci =
1
(2)
1/2
c
+I

1r/~
which has the momentum P
Khr|ci =
P
~
hr|ci
36
The physical signicance of K is that

K is an operator for the momentum with a propor-
tionality factor. The factor is the Plancks constant over 2 : ~ = /,2
L. de Broglies relation (1924):
2
`
=
j
~
de Broglie wave is also called matter wave, any aspect of the behavior or prop-
erties of a material object that varies in time or space in conformity with the
mathematical equations that describe waves. By analogy with the wave and par-
ticle behavior of light that had already been established experimentally, the French
physicist Louis de Broglie suggested (1924) that particles might have wave prop-
erties in addition to particle properties. Three years later the wave nature of
electrons was detected experimentally. Objects of everyday experience, however,
have a computed wavelength much smaller than that of electrons, so their wave
properties have never been detected; familiar objects show only particle behavior.
De Broglie waves play an appreciable role, therefore, only in the realm of sub-
atomic particles. The response of the wave properties of a particle to an external
force follows a basic law of quantum mechanics that, in its mathematical form,
is known as the Schrdinger equation.
The momentum operator in the position space
P = i~
0
0r
.
37
The commutation relation:
[x, p] = i~
[x, p])(r) = (xp px))(r) = i~r
0
0r
)(r) +i~
0
0r
(r)(r))
= i~r
0
0r
)(r) +i~r
0
0r
)(r) +i~)(r)
Therefore
[x, p])(r) = i~)(r)
1.9 The Uncertainty Relation
The commutation relation [x, p] = i~ shows that the position and the momentum are
incompatible. This relation leads to the famous uncertainty relation
(x)
2

(p)
2
_ ~
2
,4
Let A ad B be observables. Dene A = A hAi. The expectation value of

(A)
2
is
known as the dispersion of A. For any pair of operators, A and B, we have
(A)
2

(B)
2
_ |h[A, B]i|
2
,4.
The uncertainty relation can be understood from the postulate of measurement.
Proof.
(A)
2
(AhAi)
2
=
D
A
2
2AhAi +hAi
2
E
=
A
2
hAi
2
1) The Schwarz inequality
hc|ci h,|,i _ |hc|,i|
2
38
where the equality only holds if |ci = |,i .
(hc| +`
h,|)(|ci +`|,i) _ 0
take ` =
ho|oi
ho|oi
<
A
A
@
A
A
>
== hc|ci h,|,i _ |hc|,i|
2
To simplify the problem, we can take hc|ci = h,|,i = 1 and |hc|,i|
2
_ 1. Let
|ci = A|i
|,i = B|i
==
(A)
2

(B)
2
_ |hABi|
2
where we have used the hermiticity of A and B since they are observables
2)
AB =
1
2
(ABBA)
+
1
2
(AB+BA)
=
1
2
[A, B] +
1
2
{A, B}
=
1
2
[A, B] +
1
2
{A, B}
Anti-Hermitian of [A, B]:
[A, B]
= (ABBA)
= (AB)
(BA)
= B
= [A, B]
39
Hermitian of {A, B}
({A, B})
+
= {A, B}
3) a. The Hermitian operator has only purely real expectation value
H
= H, H|/i = /|/i
h/| H
= /
h/|
==
;
A
A
?
A
A
=
h/|H| /i = /
= /
==/ = /
b. The anti-Hermitian operator has only purely imaginary expectation value

A
= A,
A|ai = a |ai , ha| A
= a
ha|
ha |A| ai = a,
= a
== a = a
Therefore
hABi =
1
2
h[A, B]i +
1
2
(h{A, B}i)
= +i ImhABi + Re hABi
|hABi|
2
=
1
4
(h[A, B]i)
2
+
1
4
(h{A, B}i)
2
_
1
4
(h[A, B]i)
2
(A)
2

(B)
2
_
1
4
(h[A, B]i)
2
40
[x, p] = i~
(x)
2

(p)
2
_ ~
2
,4
Since the position and momentum operators are incompatible, [x, p] 6= 0, we
cannot nd a complete set of simultaneous eigenkets for x and p such that the two observables
can be measured at the same time. For instance for an eigenstate of x, x|ri = r|ri, it is
not an eigenstate of the momentum. It can be spanned in the momentum space,
p|ri = p
Z
dj |ji hj| |ri =
Z
dj hj|ri j |ji .
You may have an expectation value.
Example 3
A plane wave with a momentum j
0
:
c
0
(r) =
1
(2~)
1/2
exp[ij
0
r]
The probability to nd x is a constant in the whole space,

0
(r)
2
=
1
2t~
.
Example 4
A particle at r = r
0
:
c
r
0
(r) = c(r r
0
)
41
In the momentum space,
c
r
0
(r) = c(r r
0
)
=
Z
dj
2~
c
I(rr
0
)
Example 5
A Gaussian wave package:
c(r) =
1
1/4
r
1/2
0
exp[r
2
,2r
2
0
]
The particle is mainly conned in the regime, cr ~ r
0
. In the k space,
c(j) =
1
(2)
1/2
Z
c(r)c
Ir/~
dr
=
1
(2)
1/2
Z
1
1/4
r
1/2
0
exp[r
2
,2r
2
0
ijr,~]dr
=
1
(2)
1/2
Z
1
1/4
r
1/2
0
exp[
1
2
r
r
0
+
ijr
0
~
2
]dr
exp[
1
2
jr
0
~
2
]
=
_
2
1/4
r
1/2
0
(2)
1/2
Z
1
1/2
_
2r
0
exp[
1
2
r
r
0
+
ijr
0
~
2
]dr
exp[
1
2
jr
0
~
2
]
=
r
0
1/4
r
1/2
0
exp[j
2
r
2
0
,2~
2
]
The particle is mainly conned in the regime, cj ~ ~,r
0
. Thus crcj ~
~.
42
5 2.5 0 -2.5 -5
0.75
0.625
0.5
0.375
0.25
0.125
0
f(x) f(x)
Example 6
Momentum Determination Experiment
We assume that the particle is an atom in an excited state,
which will give o a photon that has the frequency i
0
if the atom is
at rest. Because of the Doppler eect, the motion of the atom toward
the observer with speed i means that the observed frequency is given
approximately by
i = i
0
(1 +

c
) == = c(
i
i
0
1)
Accurate measurement of the momentum : by measurement of the
frequency i requires a relatively long time t; the minimum error in the
frequency measurement can be shown to be i r 1,t. The instant
at which the photon is emitted is uncertain by t, at this instant, the
momentum of the atom decreases by ~i,c , and its velocity decreases
by ~i,:c . This makes the subsequent position of the atom uncertain
by the amount r r
~u
nc
t.Since the later the photon is emitted,
43
the longer the atom has the higher velocity and the farther it will
have traveled. This position uncertainty arises entirely because of
the niteness of t. If t were zero, and we knew the velocity and the
velocity change on emission of photon, we would know where the atom
is at each instant; it is because t is nite that we do not know when
the velocity changed and hence where the atom is later times. The
momentum uncertainty is obtained:
j
r
r : r :c(
i
i
0
1)
r
:c
i
0
i r
:c
i
0
c
In the nonrelativistic case considered here ,c << 1 == i r i
0
Therefore
r j ~ ~
Example 7
44
2 2 matrix: Suppose a 2 2 matrix X is written as
A = a
0
o
0
+a o (a
0
and a
I
are real)
=
3
E
E
C
a
0
0
0 a
0
4
F
F
D
+
3
E
E
C
0 a
0
4
F
F
D
+
3
E
E
C
0 ia
ia
0
4
F
F
D
+
3
E
E
C
a
:
0
0 a
:
4
F
F
D
=
3
E
E
C
a
0
+a
:
a
r
ia
a
r
+ia
a
0
a
:
4
F
F
D
Eigenvalues of X:
det(A `1) = 0
det
3
E
E
C
a
0
+a
:
` a
r
ia
a
r
+ia
a
0
a
:
`
4
F
F
D
= (a
0
+a
:
`)( a
0
a
:
`) (a
r
ia
)(a
r
+ia
)
= (a
0
`)
2
a
2
:
a
2
r
a
2
= 0
== `
= a
0

a
2
r
+a
2
+a
2
:
1/2
45
Chapter 2
Quantum Dynamics
This chapter is devoted exclusively to the dynamic development of state kets and/or ob-
servables. In other words, we are concerned here with the quantum mechanical analogue of
Newtons equation of motion.
2.1 Time Evolution and the Schrdinger Equa-
tion
2.1.1 Time Evolution Operator
We now discuss how a physical state evolves with time. Suppose we have a physical system
whose state ket at t
0
is represented by |ci. Let us denote the ket corresponding to the state
at some later time t by
|c, t
0
: ti
46
Our main task is to study the time evolution of a state ket:
|ci = |c, t
0
: t
0
i ==|c, t
0
: ti
As in the case of translation, we assume that the two kets are connected by the time-evolution
operator
|c, t
0
: ti = l(t, t
0
) |ci
Several properties of U:
1. The identity
l(t
0
, t
0
) = 1
2. The unitarity
l
(t, t
0
)l(t, t
0
) = 1
3. The composition:
l(t
2
, t
0
) = l(t
2
, t
1
)l(t
1
, t
0
)
Because time is assumed to be continuous, we consider an innitesimal time-
evolution operator
|c, t
0
; t
0
+cti = l(t
0
+ct, t
0
) |ci
As
lim
o|$0
l(t
0
+ct, t
0
) = 1
we expand the operator in terms of ct
l(t
0
+ct, t
0
) = l(t
0
, t
0
) +
0
0t
l(t, t
0
)|
|=|
0
ct +...
= 1 i
ct +...
47
Just like the operator 1 in the translation operator T (|),

is a Hermitian operator.
Whats the physical meaning of

?
We consider a simple example. In the position space
hr|c, t
0
, ti = hr|l(t
1
t
0
)|ci

1
(2t)
12
c
I(|r.(||
0
))
1
(2t)
12
c
I|r
t t
0
+ct == = .
In the old quantum theory, the time-frequency . is postulated to be related to energy
1 = ~. (Planck-Einstein Relation)
In the classical mechanics, the energy is related to the Hamiltonian operator

H , which is also
the generator of time-evolution. It is then natural to relate or dene

to the Hamiltonian
operator

H:
=

H,~
Thus
l(t
0
+ct, t
0
) = 1 i
H
~
ct +...
2.1.2 The Schrodinger Equation.
We are now in the position to derive the fundamental dierential equation for the time-
evolution operator. l(t, t
0
) and a physical state |ci . From the composition property of
48
l,
l(t +ct, t
0
) = l(t +ct, t)l(t, t
0
)
=
1 i
H
~
ct
!
l(t, t
0
)
l(t +ct, t
0
) l(t, t
0
)
ct
=
i
~
Hl(t, t
0
)
Take ct 0
+
, we obtain
i~
0
0t
l(t, t
0
) = Hl(t, t
0
)
Multiplying a state ket from the right sides
i~
0
0t
l(t, t
0
) |ci = Hl(t, t
0
) |ci
i~
0
0t
|c, t
0
; ti = H |c, t
0
; ti
The formal solutions to the Schrodinger equation for the time-evolution operator l(t, t
0
)
Case A: The Hamiltonian operator is independent of time,
i~
0
0t
l(t, t
0
) = Hl(t, t
0
)
i~
d
d|
l(t, t
0
)
l(t, t
0
)
= H
i~
0
0t
lnl(t, t
0
) = H
lnl(t, t
0
) lnl(t
0
, t
0
) =
i
~
Z
|
|
0
Hdt
l(t, t
0
) = exp[
i
~
H(t t
0
)]
Case B: The Hamiltonian

H is time-dependent but the

H
0
: at dierent times
commute
Z
|
|
0
dt lnl(t, t
0
) =
i
~
Z
|
|
0
dtH(t)
49
l(t, t
0
) = exp
i
~
Z
|
|
0
dtH(t)
Case C: The Hamiltonian

H is time-dependent but the

H
0
s at dierent times do
not commute, i.e.
[H(t
1
), H(t
2
)] 6= 0
l(t, t
0
) = 1
i
~
Z
|
0
dt
1
H(t
1
) +
(i)
2
~
2
Z
|
|
10
dt
1
Z
|
1
|
0
dt
2
H(t
1
)H(t
2
) +...
= 1 +
4
P
n=1
(
i
~
)
n
Z
|
|
0
dt
1
Z
|
1
|
0
dt
2
...
Z
|
r
1
|
0
dt
n
H(t
1
)H(t
2
)...H(t
n
)
Energy Eigenkets:
To be able to evaluate the eect of time-evolution operator on a general ket |ci,
we consider a special base ket used in expanding |ci . Assume that we know all energy
eigenkets of an time-independent H with eigenvalues 1
n
H
a
(n)
E
= 1
n
a
(n)
E
{
a
(n)
} forms a complete and orthogonal set of base kets. Therefore

l(t, t
0
) = exp(
iH
~
(t t
0
))
=
P
n.n
a
(n)
ED
a
(n)
c
I
1
~
(||
0
)
a
(n)
ED
a
(n)
=
P
n.
c
I
T
r
~
(||
0
)
a
(n)
ED
a
(n)
.
Once the expansion of the initial ket |ci in terms of {
a
(n)
} is known,
|ci =
P
n.
a
(n)
ED
a
(n)
ci =
P
n
C
n
a
(n)
E
.
The time-evolution operator enables us to solve any initial time problem,
|c, t
0
; ti = c
I
1
~
(||
0
)
|ci
=
P
n.
C
n
c
I
T
r
~
(||
0
)
a
(n)
E
.
50
In other words
C
n
(t = t
0
) = C
n
C
n
(t) = C
n
c
I
T
r
~
(||
0
)
.
The phase of C
n
(t) changes with time, but its modulus remains unchanged. If all eigenvalues
are not degenerate, the relative phases of C
n
do vary with time and the state ket varies with
time.
A special case: if the initial state happens to be one of energy eigenkets, say
|ci =
a
(n)
, we have
|c, t
0
; ti = c
I
T
r
~
(||
0
)
a
(n)
E
.
The state remains unchanged. Thus the basic task in quantum dynamics is reduced to nd a
complete set of energy eigenket and eigenvalues. It is very helpful if we can nd a maximal
set of commuting observables ,
B,

C,...which also commute with the Hamiltonian
[

,

1] = [

1,

C] = [

,

C] = ... = 0
[

,

H] = [

1,

H] = [

C,

H] = ... = 0
In the case, the energy eigenket can be characterized by the indices {a, /, c...} = {/} For
example,
;
A
A
A
A
A
A
?
A
A
A
A
A
A
=
|/i = a |/i ,
1|/i = / |/i , ..
H |/i = 1
|
|/i
Therefore it is of fundamental importance to nd a complete set of mutually compatible
observables that also commute with H.
51
2.1.3 Time Dependence of Expectation Value: Spin Pre-
cession.
It is instructive to study how the expectation value of an observable changes as a function of
time. To end this we treat an example here: spin precession. We start with a Hamiltonian
of a spin 1/2 system with magnetic moment j = c~,a:
t
c subjected to an external magnetic
eld

1 = 1

7
H =
c
:
t
c
o

1 =
c1
:
t
c
o
:
= .
o
:
As
o
:
=
~
2
3
E
E
C
1 0
0 1
4
F
F
D
then the corresponding eigenvalues of

H are
1
=
~
2
. (= o
:
.).
Since the Hamiltonian is time-independent, the time-evolution operator for the state ket
l(t, 0) = c
I.
S
z
|/~
The eigenkets of

o
:
are
|+i =
3
E
E
C
1
0
4
F
F
D
for o
:
= +
~
2
|i =
3
E
E
C
0
1
4
F
F
D
for o
:
=
~
2
==H |i =
1
2
~. |i
i.e., |i are also the energy eigenkets with eigenvalues ~.,2. Suppose at t = 0
|ci = C
+
|+i +C
|i
52
At a later time t,
|c, t
0
: ti = l(t, t
0
) |ci
= C
+
c
1.t
2
|+i +C
c
+
1.t
2
|i
Specically, (1) Let us suppose the initial state |ci is the spin up state |+i ,i,e. C
+
= 1 and
C
= 0, then
|c, t
0
: ti = c
I
.t
2
|+i
This is a stationery state ! (2) Suppose the initial state |ci is the spin state |o
r
+i :
|o
r
+i =
1
2
1/2
|+i +
1
2
1/2
|i
At a later time t
|c, t
0
: ti =
1
2
1/2
c
I
.t
2
|+i +
1
2
1/2
c
I
.t
2
|i
The probability for the system to be found in the state |o
r
i =
1
s
2
(|+i |i)
ho
r
+|c, t
0
: ti
=
1
2
1/2
(h+| +h|)
1
2
1/2
(c
I
.t
2
|+i +c
I
.t
2
|i)
=
1
2
(c
I
.t
2
+c
I
.t
2
) = cos
.t
2
|ho
r
+|c, t
0:
ti|
2
= cos
2
.t
2
ho
r
|c, t
0:
ti = i sin
.t
2
|ho
r
|c, t
0:
ti|
2
= sin
2
.t
2
The total probability is conserved.
|ho
r
+|c, t
0:
ti|
2
+|ho
r
|c, t
0:
ti|
2
= 1
53
The expectation value of o
r
is
ho
r
i = hc, t
0:
t |o
r
| c, t
0:
ti
=
~
2
|ho
r
+|c, t
0:
ti|
2
~
2
|ho
r
|c, t
0:
ti|
2
=
~
2
cos .t
Similarly, we have
ho
i =
~
2
sin.t
ho
:
i = 0
Question: what happens if we take

1 = 1 r?
2.2 The Schrodinger versus the Heisenberg Pic-
ture
2.2.1 Unitary operators
Unitary operators are used for many purposed in quantum mechanics. It satises that
l
l = 1. Under the unitary transformation that changes the state kets it is important to
keep in mind that the inner product of a state bra and a state ket remains unchanged,
hc|,i =
c|l
l|,
.
Another mathematical observation is that
hc| l
A (l |,i) = hc|
Al
|,i
54
that follows from the associative axiom of multiplication. This identity suggests two dierent
approaches to evaluate the expectation values of physical observables under the unitary
transformation.
2.2.2 Two Approaches
When we are concerned with a time-dependent of a physical observable

X there are two
approaches:
Approach I:
|ci |c, t
0
: ti = l(t, t
0
) |c, t
0
i
D

A
E
|1
hc, t
0
: t|

A|c, t
0
: ti
=
l
+
Al
Approach II:
|ci |ci
A l
+
Al =

A(t)
D

A
E
|11
= hc|A(t)| ci
=
l
+
Al
The expectation values of two approaches are the same!

D

A
E
|1
=
D

A
E
|11
.
Approach I (The Schrodinger Picture): The state ket varies with time |ci l |ci
55
while operators do not change. Recall the Schrodinger equation for the wave function, i.e.,
the state.
Approach II (The Heisenberg Picture): The operators vary with time A
l
+
Al while the ket does not change. Recall the Heisenberg equation for operators.
We denote the operator A
(1)
in the Heisenberg picture. l(t) = l(t, t
0
). The
relation of an observable in two pictures
A
(1)
(t) = l
+
(t)A
(S)
l(t)
The state ket
|c, t
0
: ti
1
= |c, t
0
= 0i ;
|c, t
0
: ti
S
= l(t) |c, t
0
= 0i = l(t) |c, t
0
: ti
1
2.2.3 The Heisenberg Equation of Motion.
We now derive the fundamental equation of motion in the Heisenberg picture. Assume that
A
(S)
does not depend explicitly on time
d
dt
A
(1)
=
d
dt
h
l
+
(t)

A
(S)
l(t)
i
=
d
dt
l
+
(t)

A
(S)
l(t) +l
+
(t)A
(S)
d
dt
l(t)
=
1
i~
l
+
(t)

H

A
(S)
l(t) +
1
i~
l
+
(t)

A
(S)

Hl(t)
= +
1
i~
l
+
(t)

A
(S)
l(t)l
+
(t)

Hl(t)
1
i~
l
+
(t)

Hl(t)l
+
(t)A
(S)
l(t)
=
1
i~
[

A
(1)
,

H
(1)
]
i~
d
dt
A
(1)
= [A
(1)
, H
(1)
]
56
This is the Heisenberg equation of motion! In the case that

H is time-independent or the
H s at dierent times commutes we have
H
(1)
= l
+

Hl =

H
Thus
i~
d
dt
A
(1)
= [A
(1)
,

H]
2.2.4 How to construct a Hamiltonian
1. For a physical system with classical analogue we assume the Hamiltonian to be of the
same form as in classical physics; we merely replace the classical r
0
I
and j
0
I
by the
corresponding operators r
I
and j
I
in quantum mechanics. With this assumption we
can reproduce the correct classical equation in the classical limit. It is required that
the Hamiltonian in quantum mechanics should be Hermitian.
2. When the physical system in question has no classical analogues, we can only guess
the structure of the Hamiltonian. We try various forms until we get the Hamiltonian
that leads to results agreeing with empirical observation.
Two useful formulas:
[ r
I
, 1( j)] = i~
0
0j
I
1
and
[ j, G( r)] = i~
0
0r
I
G
where F and G are functions that can be expanded in powers of j
0
I
: and r
0
I
: respectively.
Example 1
57
A free particle: The Hamiltonian to describe the motion of a free
particle of mass m is taken to be of the same form as in classical
mechanics:
H =
j
2
2:
=
1
2:
(j
2
r
+j
2
+j
2
:
)
We look at the observables r
I
and j
I
(they are operators, NOT num-
bers)
d
dt
j
I
=
1
i~
[j
I
, H] = 0
d
dt
r
I
=
1
i~
[r
I
, H] =
j
I
:
=
j
I
(0)
:
So
r
I
(t) = r
I
(0) +
j
I
(0)
:
t
We know
[r
I
(0), r
(0)] = 0
but
[r
I
(t), r
I
(0)] = i
~
:
t
Example 2
A particle in a potential.
H =
j
2
2:
+\ (r)
The Heisenberg equation of motion leads to
dj
I
dt
=
1
i~
[j
I,
\ (r)] =
0
0r
I
\ (r)
dr
I
dt
=
1
i~
[r
I,
H] =
j
I
:
58
Furthermore,
d
2
r
I
dt
2
=
1
i~
[
dr
I
dt
, H] =
1
i~
[
j
I
:
, H] =
1
:
dj
I
dt
d
2
r
I
dt
2
= \\ (r)
Newtonian Equation?
The expectation value of this equation is known as the Ehrenfest
theorem. It has the classical mechanics analogue. The reason is that
the Planck constant disappears in the equation.
2.3 Simple Harmonic Oscillator.
2.3.1 Eigenvalue and eigenstates
We consider a simple harmonic oscillator in a one-dimensional case. The basic Hamil-
tonian is
H =
j
2
2:
+
:
2
.
2
r
2
where x and p are Hermitian since they are physically dynamic variables. The quantum
condition is
[r, j] = rj jr = i~
As the rst step to solve the problem, we evaluate the commutator basket of x and p with
H
[r, H] =
1
2:
[r, j
2
] =
i~
:
j
[j, H] =
:.
2
2
[j, r
2
] = i~:.
2
r
59
Adding the two equations, we have
[r +ij, H] = ~:.
2
r +
i~
:
j
= ~:.
2
(r +i
1
1
:
2
.
2
j)
where is a constant. Here we take
=
1
:.
We have
[r i
1
:.
j, H] = ~.(r i
1
:.
j)
As
[r +i
1
:.
j, r i
1
:.
j] =
2~
:.
we take
a =
:.
2~
1/2
(r +i
1
:.
j); a
+
=
:.
2~
1/2
(r i
1
:.
j)
such that
[a, a
+
] = 1.
Furthermore,
a
+
a =
:.
2~
(r i
1
:.
j)(r +i
1
:.
j) =
H
:.

1
2
Thus
H = ~.(

+
1
2
)
where
= a
+
a,
Obviously,
[H,

] = 0 i.c.
60
N and H can be diagonalized simultaneously. We denote the eigenkets of N by its eigenvalues
n. So
|:i = :|:i
and
H |:i = (: +
1
2
)~. |:i
Whats n?
We rst note that
[a, ] = aa
+
a a
+
aa = [a, a
+
]a = a
[a
+
, ] = a
+
As a result,
a
+
|:i =
n
[

, a
+
] +a
+

o
|:i
= (: + 1)a
+
|:i
Likewise,
a |:i = (: 1)a |:i

Thus a
+
|:i is also an eigenket of

with eigenvalue : + 1 (increased by one). a |:i is an
eigenket of

with eigenvalue :1 (decreased by one). So we call a the decreasing operator
and a
+
the increasing operator. From the point of view of energy, the increase (decrease)
of n by one amounts to the creation (annihilation) of one quantum unit of energy. Hence
the terms creation and annihilation operator for a
+
and a are deemed appropriate. Since
61
a |:i is an eigenket with : 1 . We write
a |:i = C
n
|: 1i
(h:| a
+
)(a |:i) = h: 1| (C
n
C
n
) |: 1i
a
+
a
= : = |C
n
|
2
_ 0
We take C
n
real. then,
a |:i = :
1/2
|: 1i
Likewise,
a
+
|:i = (: + 1)
1/2
|: + 1i
Suppose we keep on applying a to both sides of eq.(1),
a
2
|:i = [:(: 1)]
1/2
|: 2i
a
3
|:i = [:(: 1)(: 2)]
1/2
|: 3i
The sequence of the eigenket must terminate at : = 0 as : _ 0. If n is a non-integer, the
sequence shall not terminate and leads to eigenkets with negative eigenvalues. This is in
contradiction with : _ 0 . To conclude n is a nonnegative integer. The lowest energy state,
i.e. the ground state of the oscillator is
1
n=0
=
1
2
~.
62
The eigenstate eigenvalue
|0i
1
2
~.
|1i = a
+
|0i
3
2
~.
|2i =
1
2
12
(a
+
)
2
|0i
3
2
~.
|3i =
1
(3!)
12
(a
+
)
3
|0i
7
2
~.
.
.
.
.
.
.
|:i =
1
(n!)
12
(a
+
)
n
|0i (: +
1
2
)~.
Position and Momentum
r =
~
2:.
1/2
(a +a
+
)
j = i
:~.
2
1/2
(a +a
+
)
From the orthonormality conditions
h:
0
|a| :i = :
1/2
c
n
0
n1
:
0
a
+
= [: + 1]
1/2
c
n
0
n+1
In the matrix form,
a =
3
E
E
E
E
E
E
E
E
E
E
E
E
E
E
C
0 1 0 0
0 0 2
1/2
0
0 0 0 3
1/2

0 0 0 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
F
F
F
F
F
F
F
F
D
63
a
+
=
3
E
E
E
E
E
E
E
E
E
E
E
E
E
E
C
0 0 0 0
1 0 0 0
0 2
1/2
0 0
0 0 3
1/2
0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
F
F
F
F
F
F
F
F
F
F
F
F
F
F
D
The matrix elements of x and p are
h:
0
|r| :i =
~
2:.
1/2
(:
1/2
c
n
0
n1
+ (: 1)
1/2
c
n
0
n+1
)
h:
0
|j| :i = i
:.~
2
1/2
(:
1/2
c
n
0
n1
+ (: 1)
1/2
c
n
0
n+1
)
In the Position Representation: the wave function
We rst derive hr|0i
a |0i = 0
== hr|a| 0i = 0
==
Z
dr
0
hr|a| r
0
i hr
0
|0i = 0
Recall
a =
:.
2~
1/2
(r +i
j
:.
)
64
Figure 2.1: Energy eigenfunctions for the rst six states of the harmonic oscillator.
hr| r| r
0
i = rc(r r
0
)
hr|j| r
0
i = i~c(r r
0
)
0
0r
== (r +
~
:.
d
dr
) hr|0i = 0
Take ~,:. = r
2
0
. The solution of the equation is
hr|0i =
1
1
4
r
1/2
0
exp(
1
2
r
2
r
2
0
)
In general
hr|:i =
*
r
(a
+
)
n
(:!)
1/2
0
+
=
*
r
(a
+
)
n
(:!)
1/2
r
+
hr|0i
=
r
r+1
2
0
1
4
(2
n
:!)
1/2
(r r
0
d
dr
)
n
c
1
2
2
0
65
Figure 2.2: The position probability density for the state n=10 of a harmonic oscillator
(solid curve) and for a classical oscillator of the same total energy (dashed line).
2.3.2 Time Development of the Oscillator
So far we have not discussed the time evolution of oscillator state ket or observables like a
and p. Everything we did is supposed to hold at some instant of time, say t = 0. Now we
come to see the time evolution of x and p. In the Heisenberg picture,
dj
dt
=
1
i~
[j, H] = :.
2
r;
dr
dt
=
1
i~
[r, H] =
j
:
.
d
2
j
dt
2
=
1
i~
[
dj
dt
, H] =
1
i~
[:.
2
r, H] = .
2
j;
d
2
r
dt
2
=
1
i~
[
dr
dt
, H] =
1
i~
[
j
:
, H] = .
2
r;
66
Equivalently,
da
dt
= i.a ==a(t) = a(0)c
I.|
da
+
dt
= i.a
+
==a
+
(t) = a
+
(0)c
I.|
r(t) +
i
:.
j(t) = (r(0) +i
j
(0)
:.
)c
I.|
r(t)
i
:.
j(t) = (r(0) i
j
(0)
:.
)c
I.|
r(t) = r(0) cos .t +
1
:.
j(0) sin.t
j(t) = :.r(0) sin.t +j(0) cos .t
For any energy eigenket, we have
;
A
A
?
A
A
=
h:|r(t)| :i = 0
h:|j(t)| :i = 0
For a superposition of energy eigenstates such as
|ci = C
0
|0i +C
1
|1i
we have
hc|r(t)| ci =
~
2:.
(C
0
C
1
c
I.|
+C
0
C
1
c
I.|
)
1/2
= 2
1/2
A
0
|C
0
C
1
| cos(.t +c)
2.3.3 The Coherent State
How can we construct a superposition of energy eigenstates that most closely imitates the
classical oscillator? In wave-function language, we want a wave packet that bounces back
67
and forth without spreading in shape. It turns out that a coherent state does the desired
job.
A coherent state:
a |ì = `|ì
Properties of |ì
1. The coherent state |ci =
P
4
n=0
)(:) |:i
|)(:)|
2
is of the Poisson type about some mean value n
|)(:)|
2
= (
:
n
:!
exp( :))
2. It can be obtained by translating the oscillator ground state by some nite distance.
3. It satises the minimum uncertainty relation at all time
D
(r(t))
2
ED
(1(t))
2
E
= ~
2
,4
4. ` = ( :)
1/2
2.4 Schrodinger Wave Equation: Simple Har-
monic Oscillator
The Schrodinger equation (see p66 in Schis book)
H =
j
2
2:
+
:
2
.
2
r
2
= 1
68
To solve the problem we rst introduce the dimensionless variable = r,r
0
,
~
2
2:r
2
0
d
2
d
2
+
:
2
.
2
r
2
0
= 1
d
2
d
2
+
:
2
.
2
~
2
r
4
0
=
2:r
2
0
~
2
1
d
2
d
2
+
2
=
21
~.
= `
We take
r
2
0
=
~
:.
.
As the potential approach innite when x + we require the wavefunction approach to
zero. From the equation we have

n
c
2
/2
.
We assume
H()c
2
/2
.
Then
H
00
2H
0
+ (` 1)H = 0
The solutions are the Hermite polynomials,
H
00
n
2H
0
n
+ 2:H
n
= 0
such that
` = 2: + 1; 1
n
= (: +
1
2
)~. : = 0, 1, 2, ....
Alternatively, we may write the Hamiltonian in the form
d
2
d
2
+
2
= `
1
2
+
d
d

d
d

d
d
+
d
d
= `
69
a =
1
_
2
+
d
d
; a
=
1
_
2

d
d
[a, a
] = 1
2a
a + 1
= `
The ground state
a
0
= 0.
2.5 Propagators and Feynman Path Integrals
2.5.1 Propagators in Wave Mechanics.
Time-evolution problem with a time-independent Hamiltonian can be solved once we expand
the initial ket in terms of the eigenkets of an observable that commute with H.
|c, t
0
, ti
= exp(
i
~
H(t t
0
)) |c, t
0
i
=
P
{o
0
}
exp(
i
~
H(t t
0
)) |a
0
i ha
0
|c, t i
=
P
{o
0
}
exp(
i
~
1
o
0 (t t
0
)) |a
0
i ha
0
|c, t i
70
Multiplying both sides by hr| on the left, we have
hr|c, t
0
: ti
=
P
o
0
exp(
i
~
1
o
0 (t t
0
)) hr|a
0
i ha
0
|c, t i
=
Z
dr
0
P
o
0
exp(
i
~
1
o
0 (t t
0
)) hr|a
0
i ha
0
|r
0
i hr
0
|c, t
0
i
=
Z
dr
0
1(r, t, r
0
, t
0
0
) hr
0
|c, t
0
i
where
1(r, t, r
0
, t
0
0
) =
P
o
0
hr|a
0
i ha
0
|r
0
i c
1
~
J
d
0 (||
0
)
Denote by c
o
(r, t) the wave-function
c
o
(r, t) = hr|c, t
0
: t i
we have
c
o
(r, t) =
Z
dr
0
1(r, t, r
0
, t
0
0
)c
o
(r
0
, t)
We call the kernel of the integral operator K as the propagator in the wave mechanics. In
any given problem the propagator depends only on the potential and is independent of the
initial and nal wave-function. It can be constructed once the energy eigenvalues and eigen
functions are given. Clearly the time evolution of the wave function is completely predicted
if K is known and c
o
(r
0
, t) is given initially. The only peculiar feature, if any, is that when
a measurement intervenes, the wave function changes abruptly in an uncontrollable way
into one of the eigen functions of observable being measured.
Two Properties of K:
1. For t t
0
, K satises the Schrodingers time-dependent wave equation in the variables
71
x an t with x and t
0
xed.
i~
0
0t
1(r, t, r
0
, t
0
0
) = H1(r, t, r
0
, t)
2.
lim
|$|
0
1(r, t, r
0
, t
0
0
) = c(r r
0
)
Proof.
lim
|$|
0
1(r, t, r
0
, t
0
0
) = lim
|$|
0
P
o
0
hr|a
0
i ha
0
|r
0
i c
I
T
d
0
~
(||
0
)
= lim
|$|
0
P
o
0
hr|a
0
i ha
0
|r
0
i = hr|r
0
i = c(r r
0
)
The propagator K is simply the Greens function for the time-dependent wave equation
satisfying
(
~
2
2:
\
2
+\
(r)
i~
0
0t
)1(r, t, r
0
, t
0
0
) = i~c(r r
0
)c(t t
0
) (2.1)
With the boundary condition
1(r, t, r
0
, t
0
0
) = 0 for t < t
0
The c-function c(t t
0
) is needed on the right hand side of the Eq.(2.1) because K varies
discontinuously at t = t
0
.
Example 3
A free particle in 1D
The Hamiltonian is expressed as
H =
1
2:
1
2
72
The momentum

P commutes with H. Hence |1i is a simultaneous
eigenket of the operators

P and

H.
1 |1i = 1 |1i
H |1i =
1
2
2:
|1i
Thus
1(r, t, r
0
, t
0
0
)
=
P
hr|ji hj|r
0
i exp
i
j
2
2:~
(t t
0
)
=
Z
dj
2~
exp
i
jr
~
i
jr
0
~

j
2
2:~
(t t
0
)
=
+4
Z
4
dj
0
2:~
exp
i
t t
0
2:~
(1
:(r r
0
)
t t
0
)
2
+i
:(r r
0
)
2
2~(t t
0
)
= lim
o$0
+
+4
Z
4
dj
2:~
exp
c|j| i
t t
0
2:~
(1
:(r r
0
)
t t
0
)
2
+i
:(r r
0
)
2
2~(t t
0
)
:
2i~(t t
0
)
1/2
exp(i
:(r r
0
)
2
2~(t t
0
)
)
Certain space and time integral derivable from K are of considerable.
Without loss of generality, we set t = 0 in the following
73
(1) r = r
0
G(t) =
Z
dr1(r, t, r, 0)
=
Z
dr
P
o
hr|ai ha|ri c
I
T
d
t
~
=
Z
dr
P
o
ha|ri hr|ai c
I
T
d
t
~
=
P
o
ha|ai c
I
T
d
t
~
= T
:
(c
I1|/~
)
(2) Let us consider the Laplace-Fourier transform of G(t).
G(1) = i
4
Z
0
dtG(t)c
IJ|/~:|/~
(- 0
+
)
= i
4
Z
0
dt
X
o
c
I(JJ
d
+I:)|/~
=
X
o
1
1 1
o
+i-
(- 0
+
)
The complete energy spectrum is exhibited as simply poles of

G(E)
in the complex plane.
74
2.5.2 Propagator as a Transition Amplitude.
To gain further insight into the physical meaning of the propagator, we rewrite it in an
alternative form, which is related the concept of transition amplitude.
1(r, t, r
0
, 0) =
P
o
hr|ai ha|r
0
i c
IJ
d
(||
0
)/~
=
X
o
D
r
1T
d
t
~
a
ED
r
c
1T
d
t
0
~
r
0
E
=
X
D
r
11t
~
a
ED
a
c
11t
0
~
r
0
E
= hr, t|r
0
, t
0
i
Where |r, ti and |r
0
, t
0
i are to be understood as an eigenket and of the position operators
in the Heisenberg picture. Roughly speaking, hr, t|r
0
, t
0
i is the amplitude for the particle to
go from a space-time point (x,t
0
) to another one (x,t)
Path Integral as the Sum Over Paths
Without loss of generality we restrict ourselves to one-dimensional problem. The
entire time interval between t
I
= t
1
and t
}
= t
is divided into N-1 equal parts.

t
t
1
= t =
t
t
1
1
(, = 1, ...)
Exploiting the composition property we obtain
hr
, t
|r
1
, t
1
i
=
Z
dr
1
Z
dr
2
...
Z
dr
2
hr
, t
|r
1
, t
1
i hr
1
, t
1
|r
2
, t
2
i
... hr
2
, t
2
|r
1
, t
1
i
75
Figure 2.3: Paths in x-t plane.
Before proceeding further, it is protable to review here how paths appear in classical
mechanics. Suppose we have a particle subjected to a force eld, say gravitional eld,.
\
(r)
= :qr
(r
1
, t
1
) = (/, 0)
(r
, t
) = (0,
2/
q
1/2
)
The classical Lagrangian is written as
1
c
(r, r
0
) =
1
2
:
dr
dt
2
\ (r)
According to Hamiltonians principle, the unique path is determined by minimizing the
action
c
|
.
Z
|
1
dt1
c
(r, r) = 0
Feynmans Formalism
The basic dierence between classical mechanics and quantum mechanics should
be apparent: In classical mechanics, a denite path in x-t plane is associated with the
76
particles motion; in contrast, in quantum mechanics all possible paths must play roles
including those which do not bear any resemblance to the classical path.. Yet the dierence
should disappear when ~ 0 . To formulate Feynmans approach, let us go back to
hr
n
, t
n
|r
n1
, t
n1
i with t = t
n
t
n1
(t 0). We write
hr
n
, t
n
|r
n1
, t
n1
i =
1
n(t)
exp(i
o(:, : 1)
~
)
where
o(:, : 1) =
|
r
Z
|
r1
dt1
c
(r, r)
and w(t) is determined by
hr
n
, t
n
|r
n1
, t
n1
i |
|
r
=|
r1
= c(r
n
r
n1
)
Because at any given time the position kets in the Heisenberg picture form a complete set,
it is legitimate to insert the identity operator written as
Z
dr|r, ti hr, t| = 1
at any time we desire. For example,
r
}
, t
}
|r
I
, t
I
=
Z
dr
r
}
, t
}
|r, t

r, t|r
I
, t
I
=
Z
drdr
0
r
}
, t
}
|r
0
, t
0
hr
0
, t
0
|r, ti
r, t|r
I
, t
I
= ...
and so on. If we somehow guess the form of hr
2
, t
2
|r
1
, t
1
i for a nite time interval by
compounding the appropriate transition amplitude for innitesimal time interval. This
property leads to formulate a space-time approach to quantum mechanics. Our task is to
77
evaluate the t 0 limit of S(n,n-1). As t 0,
o(:, : 1) =
|
r
Z
|
r1
dt(
1
2
: r
2
\ (r))
= t(
1
2
:(
r
n
r
n1
t
)
2
\ (
r
n
r
n1
2
))
For a free-particle case, \ = 0,
hr
n
, t
n
|r
n1
, t
n1
i |
|
r
=|
r1
= c(r
n
r
n1
)
we obtain
1
n(t)
=
:
2i~t
1/2
where we have used
4
Z
4
:
2i~t
1/2
c
I
r
21~ t
2
d = 1
and
lim
|$0
:
2i~t
1/2
c
I
r
21~ t
2
= c()
To summarize, as t 0, we obtain
hr
n
, t
n
|r
n1
, t
n1
i =
:
2i~t
1/2
c
I
V(rr1)
~
The nal expression for the transition amplitude with t
t
1
is
hr
, t
|r
1
, t
1
i
= lim
$+4
(
:
2i~t
)
.1
2
Z
dr
1
dr
2
...dr
2
Y
n=2
c
I
V(rr1)
~
=
r
.
Z
r
1
D(r(t
1
))c
I v
t
.
t
1
J
c
(r,r)J|/~
This is Feynmans path integral!
78
2.6 The Gauge Transformation and Phase of
Wave Function
2.6.1 Constant Potential
In classical mechanics, it is well known that the zero point of potential energy is of no
physical signicance. The force that appears in Newtons second law depends only on the
gradient of the potential; an additive constant is clearly irrelevant. e.g.
:
d
dt
= ) :
:q c1
5c
5c
t
when c c +c
0
:
d
dt
= ) :
d
dt
= )
Let us look at the time evolution of a Schrodinger picture state ket subject to some potential
\
(r)
== \
(r)
+\
0
H =
1
2
2:
+\ (r) ==H =
1
2
2:
+\ (r) +\
0
|c, t
0
: ti ==
^
|c, t
0
: ti =?
To be precise we consider a time-independent H
|c, t
0
: ti = c
I
1
~
(||
0
)
|c, t
0
i
Thus
^
|c, t
0
: ti = c
I
\
0
(tt
0
)
~
|c, t
0
: ti
For a stationary state, if the time-dependence computed with V(x) is
c
IJ(||
0
)/~
79
Figure 2.4: Quantum-mechanical interference to detect a potential dierence.
then the corresponding time dependence computed with \ (r) +\
0
is
c
I(J+\
0
)(||
0
)/~
Even though the choice of the absolute scale of the potential is arbitrary, potential
dierence are of non-trivial physical signicance and in fact, can be detected in a very striking
way. A beam of charged particles is split into two parts, each of which enters a metallic
cage. A particle in the beam can be visualized as a wave packet whose dimension is much
smaller than the dimension of the cage. Inside the cage the potential is spatially uniform.
Hence the particle in the cage experience no force. If we desire, a nite potential dierence
between the two cages can be maintained by turning on a switch. The nal state which the
two beams reach at interference region is
|)i =
1
2
1/2
|c, t
0
: ti
1
+
1
2
1/2
|c, t
0
: ti
11
Thus
h)|)i = 1 + cos(\
1
\
11
)t,~
This is an observable eect! This is a purely quantum mechanical eect and has no classical
analogue.
80
2.6.2 Gauge Transformation in Electromagnetism
We rst recall the Maxwells equations
;
A
A
A
A
A
A
A
A
A
A
?
A
A
A
A
A
A
A
A
A
A
=
\ 1 = 4j;
\H =
4t
c
J +
1
c
d1
d|
;
\ 1 = 0;
\1 =
1
c
d1
d|
1 = 1 + 41
H = 1 4'
We introduce a scalar potential c and a vector potential

;
A
A
?
A
A
=
1 = 5

1 = 5c
1
c
d
d|
Then the Maxwells equations are reduced to

;
A
A
?
A
A
=
5
2
c
1
c
2
d
2
d|
2
c = 4j
5
2

1
c
2
d
2
d|
2
c =
4t
c
J
with
5
+
1
c
0c
0t
= 0
Under the gauge transformation.
;
A
A
?
A
A
=

+5
c c
1
c
d
d|
with
5
2

1
c
2
0
2
0t
2
= 0
81
the Maxwells equations keep unchanged! In quantum mechanics, the Hamiltonian for a
particle of electric charge e ( c < 0 for electron in Sakurais book) , subjected to the
electromagnetic eld is taken from classical physics to be
H =
1
2:
(j
c
c
)
2
+cc
=
1
2:
(j
2
c
c
j
c
c
j +
c
2
c
2
2
) +cc
6=
1
2:
(j
2
2c
c
j+
c
2
c
2
) +cc
because p and A do not commute, [j, ] 6= 0. It is straightforward to examine that the
Hamiltonian in this form is Hermitian, H = H
+
. To study the dynamics of a charged
particle subjected to c and A, let us rst proceed in the Heisenberg picture:
d
dt
r
I
=
1
i~
[r
I
, H] =
1
:
(j
I
c
c
I
)
where i = r, j, .. Dene
= :
dr
dt
= j
c
c
We call that j is canonical momentum and is kinetic (or mechanical) momentum. The
commutation relation of
I
and
is
[
I
,
] =
i~c
c
(
0
0r
I
0
0r
I
)
= i
~c
c
c
I|
(5A)
|
Hence the Hamiltonian can be rewritten as
H =

2
2:
+cc
82
We now study the Schrodingers equation with c and A in the position space.
D
r
(j
c
c
(r))
2
c, t
0
: t
E
=
Z
dr
1
Z
dr
2
D
r
(j
c
c
(r))
r
1
ED
r
1
(j
c
c
(r)) |r
2
i hr
2
c, t
0
: t
E
= (i~ 5
c
c
(r))
2
hr|c, t
0
: ti
Thus the Schrodingers equation is derived as
[
1
2:
(i~ 5
c
c
(r)
)
2
+cc] hr|c, t
0
: ti = i~
0
0t
hr|c, t
0
: ti
Denote
,(r) = hr|c, t
0
: ti
and
j = ,
(r),(r)
The time derivative of the density is
0
0t
j = (
0
0t
,
), +,
0
0t
, = 5j
0
0t
j +5 j = 0
where
j =
~
:
1
n
(,
5,)
c
:c
|,|
2
For
, =
1/2
c
Is/~
, =
j
:
(5:
c
c
)
83
2.6.3 The Gauge Transformation
In quantum mechanics, we consider the gauge transformation
;
A
A
?
A
A
=
c c
+5
e.g. Consider a uniform magnetic eld in z direction
1 = 1 . = (
0
0r
0
0j
r
) .
To derive the eld, we can choose
(1).
r
=
1
2
1j,
=
1
2
1r,
(2).
r
= 1j,
= 0.
These two vector potentials are connected by a gauge transformation
+5(
1rj
2
)
Let us denote by |ci the state ket in the presence of A, and | ci the state ket in the presence
of . The Schrodingers equations for these two state kets are
[
1
2:
(j
c
c
)
2
+cc] |c, t
0
: ti = i~
0
0t
|c, t
0
: ti
[
1
2:
(j
c
c
c
c
5)
2
+cc]
^
|c, t
0
: ti = i~
0
0t
^
|c, t
0
: ti
Whats the relation of |c, t
0
: ti and
^
|c, t
0
: ti ?
We come to construct an operator G such that
| ci = G |ci
with GG
=1.
G
[
1
2:
(j
c
c
c
c
5)
2
+cc]G |c, t
0
: ti = i~
0
0t
|c, t
0
: ti
84
We require
(1). hc|r|ci = h c|r| ci :
G
rG =r
(2).

c|j
t
c
|c
=
D
c|j
t
c
| c
E
:
G
j
c
c
c
c
5
G =
j
c
c
[j, G] = +
c
c
5
G
(i~\G) = +
c
c
5
i~\lnG = +
c
c
5
G = exp
h
i
c
~c
i
Therefore, if we take
| ci = c
I
c
~ c
|ci
then
c
I
c
~ c
(
1
2:
(j
c
c
c
c
5)
2
+cc)c
I
c
~ c
=
1
2:
(j
c
c
)
2
+cc
The two wave-function are related via
,(r, t) = c
I
c
~ c
,(r, t)
85
Figure 2.5:
In the form , = j
1/2
exp[io,~],we have
+\
j j
o o +
c
c
2.6.4 The Aharonov-Bohm Eect

Consider a particle of charge e going above or below a very long impenetrable cylinder, as
shown in Fig.
Inside the cylinder is a magnetic eld parallel to the cylinder axis, taken to be
normal to the plane of Fig. So the particle paths above and below enclose a magnetic ux.
Assume the magnetic eld
1 =
;
A
A
?
A
A
=
1 . if j < j
o
0 if j j
o
Hence the particle does not experience the Lorentz force outside cylinder.
=
;
A
A
?
A
A
=
(0,
1
2
1j, 0), if j < j
o
(0,
1
2
1j
2
o
, 0), if j j
o
86
Our object is to study how the probability of nding the particle in the interference region
1 depends on the magnetic ux. For pedagogical reason we prefer to use the Feynman
path-integral method to attack this problem.
In the presence of the magnetic eld, the Lagrangian can be obtained
1
(0)
c
=
1
2
:(
dr
dt
)
2
1
2
:(
dr
dt
)
2
+
c
c
dr
dt
The corresponding change in the action is given by

o
(0)
(:, : 1) o
(0)
(:, : 1) +
c
c
|
r
R
|
r1
dt
dr
dt
= o
(0)
(:, : 1) +
c
c
r
r
R
r
r1
d
where d
| is the dierential line element along the path segment. So when we consider the
entire contribution from x
1
to x
n
.We have
Q
n
exp(io
(0)
(:, : 1),~)
==
Q
n
exp(io
(0)
(:, : 1),~) exp(
ic
~c
r
r
R
r
r1
|)
The entire transition amplitude is
R
obout
1
[r
(t)
]
c
I
:
(0)
(.1)
~
c
I
c
~ c
r
R
1
.J
l
d|orc
+
R
btlou
1
[r
(t)
]
c
I
:
(0)
(.1)
~
c
I
c
~ c
r
R
1
.J
l
|cIor
The probability for nding the particle in the interference region B depends on the modulus
square of the entire transition amplitude and hence on the phase dierence between the
contribution from the paths going above and below. i.e.
j 1 + cos(,
1
,
2
)
87
where
,
1
,
2
=
c
~c
(
r
r
R
r
1
d
|
obout
r
r
R
r
1
|
btlou
)
=
c
~c
I
d
|
=
c
~c
ZZ
\ da
=
c
~c
c
1
where c
1
stands for the magnetic ux inside the impenetrable cylinder. This means that
as we change the magnetic eld strength, there is a sinusoidal component in the probability
for observing the particle in region B with a period given by a fundamental unit of magnetic
ux, namely
2~c
|c|
= 4.135 10
7
Gauss-cm
2
2.7 Interpretation of Wave Function.
2.7.1 Whats [
c
({)?
Since
|c, t
0
: ti =
Z
d
r
|ri hr|c, t
0
: ti =
Z
d
r
|ri
o
(r, t)
o
(r, t) is regarded as an expansion coe!cient of |c, t
0
: ti in terms of the position eigenkets
{|ri}. The quantity
j(r, t) = |
o
(r, t)|
2
is dened as the probability density. So j(r, t)dr is the probability that the particle appears
in a narrow range round r at time t. This is the so-called the probabilistic interpretation of
88
wave function.
The Continuity Equation.
0
0t
j(r, t)
= (
0
0t
(r, t))(r, t) +
(r, t)(
0
0t
(r, t))
=
1
i~
(H
)+
1
i~
H
Assume
H =
~
2
2:
\
2
+\
0
0j
j(r, t) = i
1
2:
~(
(\
2
) (\
2
)) +
1
i~
(\ \
As
\
2
=  (\
)(\)
\
2
= $\
) ($(\
)
0
0t
j = i
~
2:
$
\\
) +
1
~
1
n
(\ ) j
0
0t
j +\ j =
1
~
1
n
(\ ) j
j =
i~
2:
(
\\
) =
~
:
1
n
(
$
The probability ux.
When \ is purely real, we have
0
0t
j +\ j = 0
This is the conservation law for probability. When V is complex, which is often used for
nuclear reaction for particles absorbed by nuclei, it is accounted for the disappearance of
89
particles. Except for the probability density, the wave function contains more physics than
expected. Let us write it as
(r, t) = j(r, t)
1/2
exp[io(r, t),~]
with j 0 and o real.
\ = j
1/2
\j
1/2
+
i
~
j\o
== j =
j
:
\o
The spatial variation of the phase of the wave function characterizes the probability ux;
the stronger the phase variation, the more intense the ux. The direction of j at some point
r is seen to be normal to the surface of a constant that goes through that point. Since j is
a ux, j can be rewritten as
j = j(\o,:) = jv
This physical meaning is still unclear. It is not a velocity traditionally!
2.7.2 The Classical Limit
Substituting = j
1/2
c
IS/~
into a Schrodinger equation leads to
~
2
2:
(\
2
j
1/2
+
2i
~
\j
1/2
\o)
1
~
2
j
1/2
|\o|
2
+ (
i
~
)j
1/2
\
2
o +j
1/2
\
= i~(
0j
1/2
0t
+
i
~
j
1/2
0o
0t
)
Taking ~ 0 , we obtain
1
2:
|\o(r, t)|
2
+\ (r) +
0
0t
o(r, t) = 0
This is the Hamilton-Jacobi equation in classical mechanics. In this sense, the Schrodinger
equation goes back to the classical mechanics in the limit ~ 0. In the classical mechanics
90
Figure 2.6: One-dimensional square well potential with (a) perfectly rigid walls (b) nite
potential.
o(r, t) stands for Hamiltonian principal function. In a stationary state with time dependent
c
IJ
t
/~
, the Hamiltons principle function S is separable
o(r, t) = \(r) 1t
2.8 Examples
2.8.1 One dimensional square well potential
Perfect rigid wall
\ (r) =
;
A
A
?
A
A
=
0, |r| _ a
+, |r| a
Since the potential is innite at |r| a the wave function must vanish at the points r = a.
The wave function for |r| _ a is simply
~
2
2:
d
2
n
dr
2
= 1n.
91
Introduce a dimensionless quantity r = a,
d
2
d
2
n =
2:a
2
~
2
1n = c
2
n
where c = (2:a
2
1,~
2
)
1/2
. The general solution is
n() = sinc +1cos c.
Application of the boundary conditions at r = a, i.e., = 1, gives
sinc +1cos c = 0
sinc +1cos c = 0
There are two possible classes of solution
(1). = 0 and cos c = 0
c = (: +
1
2
) (2.2)
n = 1cos c = 1cos
(2: + 1)r
2a
(2.3)
1 =
~
2
2:
c
2
a
2
=
(2: + 1)
2
2
~
2
8:a
2
(2.4)
(2) 1 = 0 and sinc = 0
c = : (2.5)
n = sinc = sin
:r
a
(2.6)
1 =
~
2
2:
c
2
a
2
=
:
2
2
~
2
2:a
2
(2.7)
Finite potential step
\ (r) =
;
A
A
?
A
A
=
0, |r| _ a
\
0
, |r| a
92
The wave equation at |r| < a
~
2
2:
d
2
n
dr
2
= 1n
The wave equation at |r| a
~
2
2:
d
2
n
dr
2
+\
0
n = 1n
~
2
2:
d
2
n
dr
2
= (1 \
0
)n
The general solution for 1 < \
0
n(r) =
;
A
A
?
A
A
=
Cc
or
, r a
1c
+or
, r < a
where , = [2:(\
0
1)]
1/2
,~.
For 1 \
0
n(r) =
;
A
A
?
A
A
=
C
1
sin/r +1
1
cos /r, r a
C
2
sin/r +1
2
cos /r, r < a
The wave equation at |r| < a is the same as that in the rigid wall potential, and the general
solution is.
n() = sin/r +1cos /r.
where / = (2:1,~
2
)
1/2
. We now impose on the solutions (1) and (2) the requirements that
u and du/dx be continuous at x=a.
sin/a +1cos /a = Cc
oo
/sin/a /1cos /a = ,Cc
oo
sin/a +1cos /a = 1c
oo
sin/a +1cos /a = ,1c
oo
93
from which we obtain
2sin/a = (C 1)c
oo
2/cos /a = ,(C 1)c
oo
21cos /a = (C +1)c
oo
2/1sin/a = ,(C +1)c
oo
Two classes of solutions
(1) A=0 and C=D
/ tan/a = ,
(2) B=0 and C=-D
/ cot /a = ,
Parity
The Schrodinger equation
~
2
2:
d
2
n(r)
dr
2
+\ (r)n(r) = 1n(r)
If the potential is symmetric about r = 0, \ (r) = \ (r), we have
~
2
2:
d
2
n(r)
dr
2
+\ (r)n(r) = 1n(r)
The u(x) and u(-x) are solutions of the same wave equation with the same eigenvalues
E. Addition or subtraction these two equations gives n(r) n(r), the solution of the
Schrodinger equation. To simplify the notation, n(r) = n(r). Such wave function are
94
said to have even or odd parity. Note that
cos r = cos(r)
sinr = sin(r)
2.8.2 A charged particle in a uniform magnetic eld
Consider a charged particle in a uniform magnetic eld. Assume the particle is conned in
a two-dimensional plane, and the magnetic eld is perpendicular to the plane, say along the
z direction. The Hamiltonian is given by
H =
1
2:
(j
c
c
)
2
with
1 = 1 . = (
0
0r
0
0j
r
) .
To derive the eld, we can choose
(1).
r
= 1j,
= 0.
(2).
r
=
1
2
1j,
=
1
2
1r.
Case (1):
r
= 1j,
= 0.
H =
1
2:
h
(j
r
c
c
r
)
2
+ (j
c
c
)
2
i
=
1
2:
(j
r
c1
c
j)
2
+ (j
)
2
Notice that [j
r
, H] = 0, i.e. p
r
is a good quantum number. The general form of the
wavefunction is
= c(j)c
I
0
r
95
The Schrodinger equation H = 1 is reduced to
1
2:
(j
0
c1
c
j)
2
+ (j
)
2
c(j) = 1c(j)
Alternatively
"
j
2
2:
+
c
2
1
2
2:c
2
(j j
0
)
2
#
c(j) = 1c(j)
with j
0
=
c
0
t1
. Thus the problem is reduced to the simple harmonic oscillator.
(2).
r
=
1
2
1j,
=
1
2
1r.
H =
1
2:
h
(j
r
c
c
r
)
2
+ (j
c
c
)
2
i
=
1
2:
(j
r
c1
2c
j)
2
+ (j
+
c1
2c
r)
2
In this case both p

r
and p
are not good quantum numbers. We cannot solve this problem

like in the case (1). Take
r
= j
r
c1
2c
j;
= j
+
c1
2c
r
[
r
,
] = 2i~
c1
2c
= i
~c1
c
H =
1
2:
2
r
+
2
=
1
4:
[(
r
+i
)(
r
i
) + (
r
i
)(
r
+i
)]
[
r
i
,
r
+i
] =
2~c1
c
96
Take
a =
r
c
2~c1
(
r
i
)
a
=
r
c
2~c1
(
r
+i
)
Thus [a, a
] = 1.
H =
~c1
2:c
a +aa
=
~c1
:c
a +
1
2
=~.
a +
1
2
97
Chapter 3
Theory of Angular Momentum
This chapter is concerned with a systematic treatment of angular momentum and related
topics.
3.1 Rotation and Angular Momentum
We consider the rotation in the space of a physical system in a state represented by a ket
|ci or the wave function ,
o
(r). We describe a rotation by a linear operator 1, which is so
dened that any vector r is rotated into the new vector 1 r. The rotation changes the ket
|ci into the new ket |c
0
i or change the wave function ,
o
(r) into the wave function ,
o
0 (r
0
),
which means
,
o
0 (1r) = ,
o
(r)
D
+
(1)D(1)
= hr|ci
h1r|c
0
i = hr|ci
98
Figure 3.1: Example to illustrate the noncommunativity of nite rotation.
3.1.1 Finite versus innitesimal rotation
From elementary physics, we know that
(1) Rotations about the same axis commute.
1
:
(
6
)1
:
(
3
) = 1
:
(
3
)1
:
(
6
) = 1
:
(
3
+

6
)
(2). Rotation about dierent axes do not commute.
1
r
(
2
)1
:
(
2
) 6= 1
:
(
2
)1
r
(
2
)
Consider a vector V with three components \
r
, \
and \
:
. When we rotate
the vector , we obtain a new vector, V
1
with \
1r
, \
1
,and \
1:
. These two vectors are
99
connected by a 3 3 matrix
3
E
E
E
E
E
E
C
\
1r
\
1
\
1:
4
F
F
F
F
F
F
D
=
3
E
E
E
E
E
E
C
1
rr
1
r
1
r:
1
r
1
1
:
1
:r
1
:
1
::
4
F
F
F
F
F
F
D
3
E
E
E
E
E
E
C
\
r
\
\
:
4
F
F
F
F
F
F
D
The requirement that \
1
be real when \ are real means that the elements of R are real.
The length of the vector \ and \
1
do not change, which means
(\
1r
, \
1
, \
1:
)
3
E
E
E
E
E
E
C
\
1r
\
1
\
1:
4
F
F
F
F
F
F
D
= \
2
r
+\
2
+\
2
:
\
T
1
\
1
= \
T
1
T
1\ = \
T
\
1
T
1 = 11
T
= 1
where T stands for a transpose of a matrix: 1
T
I
= 1
I
. To be denite we consider a
rotation about .
axis by an angle c
\
1:
= \
:
\
1r
= \
r
cos c \
sinc(= \
0
cos(c +,))
\
1
= \
r
sinc +\
cos c(= \
0
sin(c +,))
where \
r
= \
0
cos , and \
= \
0
sin,. In the matrix form, we obtain
1
:
(c) =
3
E
E
E
E
E
E
C
cos c sinc 0
sinc cos c 0
0 0 1
4
F
F
F
F
F
F
D
100
Similarly
1
r
(c) =
3
E
E
E
E
E
E
C
1 0 0
0 cos c sinc
0 sinc cos c
4
F
F
F
F
F
F
D
1
(c) =
3
E
E
E
E
E
E
C
cos c 0 sinc
0 1 0
sinc 0 cos c
4
F
F
F
F
F
F
D
We are particularly interested in an innitesimal form, from which a great deal can be
learned about the structure of R. Take sinc r c, cos c r 1

2
2
, we have
1
r
(c) =
3
E
E
E
E
E
E
C
1 0 0
0 1 -
2
,2 -
0 - 1 -
2
,2
4
F
F
F
F
F
F
D
= 1 +O(-) +O(-
2
) +....
1
(c) =
3
E
E
E
E
E
E
C
1 -
2
,2 0 -
0 1 0
- 0 1 -
2
,2
4
F
F
F
F
F
F
D
1
:
(c) =
3
E
E
E
E
E
E
C
1 -
2
,2 - 0
- 1 -
2
,2 0
0 0 1
4
F
F
F
F
F
F
D
Elementary matrix manipulation lead to
1
r
(-)1
(-) 1
(-)1
r
(-) = 1
:
(-
2
) 1.
If we just remain the quantities of rst order in - , any two rotations about dierent axes
commute because 1
:
(-
2
) r 1
101
In quantum mechanics
Because rotations aect physical systems, the state ket corresponding to a rotated
system is expected to look dierent from the state ket corresponding to the original system.
Given a rotation operator

1, characterized by a 3 3 matrix R, we associated with an
operator D(1) in the appropriate ket space such that
|ci
1
= D(1) |ci
where |ci
1
and |ci stand for the kets of the rotated and unrotated systems, respectively.
To construct the rotation operator, it is again fruitful to examine rst its properties under
an innitesimal rotation
D( :dc) = 1 i
J :
~
dc
we dene J
|
the k- component of the angular momentum. A nite rotation can be obtained
by compounding successively innitesimal rotation about the same axis
D
:
(c) = lim
$+4
[D
:
(
c
)]
= lim
$+4
(1 i
J
:
~

c
lim
$+4
(1 i
J
:
~

c
)
~/(I
z
)
I
J
z
~

= exp(i
J
:
~
c)
Denition: lim
r$+4
(1 +
1
r
)
r
= c
In order to obtain the angular momentum commutator relations we need more
concepts. For every 1, there exists a rotation operator D(1) in the appropriate ket space
1 ==D(1)
102
We postulate that D(1) has the same group properties as R.
1. Identity:
1 1 = 1 ==D(1) D(0) = D(1)
2. Closure:
1
1
1
2
= 1
3
==1(1
1
)1(1
2
) = 1(1
3
)
3. Inverse:
11
1
= 1 ==D(1)D(1
1
) = D(0) = 1
4.Associativity:
1
1
(1
2
1
3
) = (1
1
1
2
)1
3
D(1
1
)D(1
2
1
3
) = D(1
1
1
2
) D(1
3
)
As
1
r
(-)1
(-) 1
(-)1
:
(-) = 1
:
(-
2
) 1
we examine
D
r
(-)D
(-) D
(-)D
r
(-)
= (1 i
J
r
-
~

J
2
r
-
2
2~
2
)(1 i
J
-
~

J
2
-
2
2~
2
)
(1 i
J
-
~

J
2
-
2
2~
2
)(1 i
J
r
-
~

J
2
r
-
2
2~
2
)
= [J
r
, J
]-
2
,~
2
= 1 D
:
(-
2
)
D
:
(-
2
) = 1 i
J
:
~
-
2
== [J
r
, J
] = i~J
:
103
Similarly, we obtain
[J
I
, J
] = i~c
I|
J
|
or
J

J = i~
J
Comparing with a conventional vector v
v v = 0.
3.1.2 Orbital angular momentum
If the vector

c is of innitesimal length and only quantities of rst order in

c are retained,
the relation \
1
= 1\ may be written
\
1
= \ +
c

\
where
1 =
3
E
E
E
E
E
E
C
1 c
:
c
c
:
1 c
r
c
c
r
1
4
F
F
F
F
F
F
D
To derive the formula we have used the innitesimal rotation, for example,
1
:
(c) =
3
E
E
E
E
E
E
C
cos c
:
sinc
:
0
sinc
:
cos c
:
0
0 0 1
4
F
F
F
F
F
F
D
= 1 +
3
E
E
E
E
E
E
C
0 c
:
0
c
:
0 0
0 0 0
4
F
F
F
F
F
F
D
104
and the innitesimal rotation about any axis are commutable.
Now we wish to nd a transformation |ci
1
that changes the ket |ci into the ket
|ci
1
|ci
1
= D(c) |ci
== hr| D
+
(c) |ci
1
= hr |ci
== h1r |ci
1
= hr |ci
== hr |ci
1
=
1
1
r |c
Thus we have
D(
c),
o
(r) = ,
o
(1
1
r)
r ,
o
(r c r)
r ,
o
(r) (c r) 5,
o
(r)
== D(
c) = 1
i
~
1 c +
where
1 = r

1
1
r
= j1
:
.1
(= i~(j
0
0.
.
0
0j
))
1
= .1
r
r1
:
(= i~(.
0
0r
r
0
0.
))
1
:
= r1
j1
r
(= i~(r
0
0j
j
0
0r
))
These operators satisfy
[1
I
, 1
] = i~c
I|
1
|
105
3.1.3 Rotation operator for spin 1/2
We have already checked that
[o
I
, o
] = i~c
I|
o
|
Consider a rotation by a nite angle c about the z-axis. If the ket of a spin 1/2 system
before rotation is given by |ci , the ket after rotation is given by
|ci
1
= D
:
(c) |ci
with
D
:
(c) = c
I
:
z
~

To see the physical meaning of D
:
(c), we compute
D
+
:
(c)o
r
D
:
(c)
= c
I
:
z
~

o
r
c
I
:
z
~

= c
I
:
z
~

~
2
(|+i h| +|i h+|) c
I
:
z
~

=
~
2
c
I
2
|+i h| c
I
2
+c
I
2
|i h+| c
I
=
~
2
[(|+i h| +|i h+|) cos c +i (|+i h| |i h+|) sinc]
= o
r
cos c o
sinc
D
:
(c)
3
E
E
E
E
E
E
C
o
r
o
o
:
4
F
F
F
F
F
F
D
D
:
(c) =
3
E
E
E
E
E
E
C
o
r
cos c o
sinc
o
r
sinc +o
cos c
o
:
4
F
F
F
F
F
F
D
=
3
E
E
E
E
E
E
C
cos c sinc 0
sinc cos c 0
0 0 1
4
F
F
F
F
F
F
D
3
E
E
E
E
E
E
C
o
r
o
o
:
4
F
F
F
F
F
F
D
106
Compare with
\
1
= 1\.
3.1.4 Spin precession revisited
The Hamiltonian is
H =
c
:c
S B
= .S
:
where . = |c| 1,:c. The time evolution operator is given by
l(t) = exp[iHt,~] = exp[io
:
.t,2] .
For an arbitrary state |ci, at a later time t
|c, ti = l(t) |ci
= l(t)(|+i h+| +|i h|) |ci
= c
I.|/2
|+i h+|ci +c
+I.|/2
|i h|ci
The operators change as
hS
r
i
|
= hS
r
i
|=0
cos .t hS
i
|=0
sin.t
hS
i
|
= hS
i
|=0
cos .t +hS
r
i
|=0
sin.t
hS
:
i
|
= hS
:
i
|=0
107
Conclusion: the period for the state ket is twice as long as the period for spin precession:
t
precession
= 2,.;
t
state ket
= 4,..
3.2 Rotation Group and the Euler Angles
3.2.1 The Group Concept
The branch of mathematics that is appropriate for a full treatment of symmetry is the theory
of groups. Here we give a few basic denitions: A set of objects a,b,c... form a group if a
process can be dened that enables us to combine any two of the objects, such as a and b,
to form an object ab, and if the following conditions are satised:
1. All results of combination, such as a,b, are members of the group.
2. The group contains an identity or unit member e that has the properties ac = ca = a,
where a is any member of the group.
3. Each member a has an inverse a
1
also in the group, such that aa
1
= a
1
a = c.
4. Group combination is associative, so that
a(/c) = (a/)c
The members of the group are called elements. Though we frequently refer to the
combination as multiplication, this does not mean ordinary multiplication. For example,
the set of integers, positive, negative, and zero, form a group if the law of combination is
ordinary addition.
108
A group is abelian if multiplication is commutative, so that a/ = /a for all pair
of elements. Otherwise the group is non-abelian.
Two groups are said to be isomorphic to each other if there is a unique one-to-
one correspondence between elements of the two groups such that products of corresponding
elements correspond to each other.
Examples:
(1). The elements are 1 and -1 and the law of combination is multiplication.
(2). {c, a, a
2
...a
= c}: a is the root of r
= 1.
3.2.2 Orthogonal Group
We discussed rotations of a vector. The rotated vector and unrotated vector are connected
by a 3 3 real and orthogonal matrix
\
1
= 1\
All rotation matrices form a group.
1. Combination of two 1
1
and 1
2
is a new matrix 1
1
1
2
2. Identity: 1 = 1
3. The inverse matrix: 1
T
= 1
1
and 1
T
1 = 11
T
= 1
4. Associativity:(1
1
1
2
)1
3
= 1
1
(1
2
1
3
)
This group is named SO(3), where S stands for special, O stands for orthogonal,
3 for three dimension. If the vector is n-dimensional, the R form a SO(n) group.
109
3.2.3 Special?
Consider
1
:
(c) =
3
E
E
E
E
E
E
C
cos c sinc 0
sinc cos c 0
0 0 1
4
F
F
F
F
F
F
D
Its determinant is
det(1
:
(c)) = 1
In fact,
det(1) = 1 ==o
-1 is not one of elements in SO(n). We postulate that D(1) has the same group properties
of R. The determinant of D(1) is
det[D(1)] = det
exp[i
J :
~
c]
!
= exp
"
det(i
J :
~
c)
#
= 1
since
det[J
|
] = 0.
3.2.4 Unitary Unimodular Group
Another rotation we discussed is for the spin 1/2 system, where the rotation matrices are
2 2. We can write a unitary unimodular matrix as
n(a, /) =
3
E
E
C
a /
/
4
F
F
D
where a and b are complex satisfying |a|
2
+|/|
2
= 1. All these matrices form a group.
110
(1) Closure:
n(a
1
, /
1
)n(a
2
, /
2
) = n(a
1
a
2
/
1
/
2
, a
1
/
2
+a
2
/
1
)
where
|a
1
a
2
/
1
/
2
|
2
+|a
1
/
2
+a
2
/
1
|
2
= 1
(2) Identity:
n(1, 0) =
3
E
E
C
1 0
0 1
4
F
F
D
(3) Inverse:
n
+
(a, /)n(a, /) = 1
(4) Associative: multiplications of matrices are associative.
If we set
Re(a) = cos
c
2
Re(/) = :
sin
c
2
Im(a) = :
:
sin
c
2
Im(/) = :
r
sin
c
2
the n(a, /) can be rewritten as
n(a, /) ==c
I
or
2

As det[n(a, /)] = 1, this group is called SU(2) group. SU(2) and SO(3) have a two-to-one
correspondence: In SU(2). a rotation by 2 produces -1. a rotation by 4 produces 1. In
SO(3) , a rotation 2 produces 1. More generally, U(a,b) and U(-a,-b) correspond to one
matrix in R (in SO(3)).
111
Figure 3.2: Euler rotations.
3.2.5 Euler Rotations
Three steps:
(a) : R
:
(c) y j
0
(b) : R
0 (,) z .
0
(c) : R
:
0 () y j
00
In terms of 3 3 orthogonal matrices the product of the three operations can be written as
1(c, ,, ) = 1
:
0 ()1
0 (,)1
:
(c)
There appear both rotations about body axes and about the space-xed axes. This is rather
inconvenient! It is desirable to express the body-axis rotations in terms of space-xed axis
112
rotation. Fortunately, we have
1
0 (,) = 1
:
(c)1
(,)1
1
:
(c)
1
:
0 () = 1
0 (,)1
:
()1
1
0
(,)
Therefore,
1(c, ,, ) = 1
:
(c)1
(,)1
:
()
As an example we calculate D(c, ,, ) for a spin 1/2 system.
1(c, ,, ) = 1
:
(c)1
(,)1
:
()
= c
I
o
3
2
o
c
I
o
2
2
o
c
I
o
3
2
~
Recall that
exp(i
o :
2
c) = cos
c
2
io :sin
c
2
To prove the identity we use the (o :)
2
= 1.
1(c, ,, ) =
3
E
E
C
cos
o
2
i sin
o
2
0
0 cos
o
2
+i sin
o
2
4
F
F
D
3
E
E
C
cos
o
2
sin
o
2
sin
o
2
cos
o
2
4
F
F
D
3
E
E
C
cos
~
2
i sin
~
2
0
0 cos
~
2
+i sin
~
2
4
F
F
D
=
3
E
E
C
c
I
c+
2
cos
o
2
c
I
c
2
sin
o
2
c
I
c
2
sin
o
2
c
I
c+
2
cos
o
2
4
F
F
D
113
3.3 Eigenvalues and Eigenkets of Angular Mo-
mentum
The commutation relation between three components of

J are already derived:
[J
r
, J
] = i~J
:
[J
, J
:
] = i~J
r
[J
:
, J
r
] = i~J
These relation are often rewritten in a more compact form

J J = i~J
In this section we work out the eigenvalues and eigenkets of angular momentum. To this
end, we introduce a new set of operators:
(1). J
2
= J
r
J
r
+J
+J
:
J
:
(2). J
= J
r
iJ
[J
+
=(J
)
+
]
J
2
commutes with all three J
|
(/ = r, j, .)
[J
2
, J
|
] = 0
As J
r
, J
and J
:
do not commute, we cannot diagonalize J
r
, J
and J
:
simultaneously.
However we can choose one J
|
to be diagonalized with J
2
. By convention, we choose J
:
.
We denote the eigenvalues of J
2
and J
:
by a and b, respectively
J
2
|a, /i = a |a, /i
J
:
|a, /i = / |a, /i
114
To determine the values of a and b, it is convenient to work with the ladder operator, J
.
[J
:
, J
] = i~J
r
[J
:
, J
r
] = i~J
[J
:
, J
] = i~J
r
[J
:
, J
r
+iJ
] = ~(J
r
+iJ
)
The commutation relations become
[J
:
, J
] = ~J
[J
+
, J
] = 2~J
:
[J
, J
2
] = 0
First of all , we have
a
0
, /
0
J
2
a, /
= ac
oo
0 c
bb
0 ;
ha
0
, /
0
|J
:
| a, /i = /c
oo
0 c
bb
0
Then
J
:
J
|a, /i = {[J
:
, J
] +J
J
:
} |a, /i
= (~J
+/J
) |a, /i
= (/ ~)J
|a, /i
From this relation, we conclude
J
|a, /i = C
(a, /) |a, / ~i .
115
Note that the bra of J
|a, /i are
ha, /| J
= ha, / ~| C
(a, /)
ha, /| J
|a, /i = |C
(a, /)|
2
J
2
can be expressed in terms of J
and J
:
J
2
=
1
2
(J
+
J
+J
J
+
) +J
2
:
= J
J
+
+J
2
:
+J
:
~ = J
+
J
+J
2
:
J
:
~
a = |C
+
(a, /)|
2
+/(/ + ~) = |C
(a, /)|
2
+/(/ ~)
|C
+
(a, /)|
2
= a /(/ + ~) _ 0
|C
(a, /)|
2
= a /(/ ~) _ 0
If |a, /i is one of eigenkets, then
J
|a, /i = C
(a, /) |a, / ~i
If C
+
(a, /) or C
(a, /) is not equal to zero, |a, / ~i is also one of eigenkets suppose that we
keep on applying J
to both side of the equation above. We can obtain numerical eigenkets

with smaller and smaller or larger and larger / :~ until the sequence terminates at some
/
max
and /
min
such that
;
A
A
?
A
A
=
C
+
(a, /
max
) = 0
C
(a, /
min
) = 0
So
|C
+
(a, /
max
)|
2
= a /
max
(/
max
+ ~) = 0
|C
(a, /
max
)|
2
= a /
min
(/
min
~) = 0
116
a = /
max
(/
max
+ ~) = /
min
(/
min
~)
== /
max
= /
min
Clearly, we must be able to reach |a, /
max
i by applying J
+
successively to |a, /
min
i a nite
number of times. i.e.
|a, /
max
i (J
+
)
n
|a, /
min
i
|a, /
min
+:~i
We obtain
/
max
= /
min
+:~
As a result,
/
max
=
:
2
~
and
a =
:
2
(
:
2
+ 1)~
2
It is conventional to dene
/
max
= /
min
=
:
2
~ = ,~
and
|a, /i ==|,, :i
such that
J
2
|,, :i = ,(, + 1)~
2
|,, :i
J
:
|,, :i = :~ |,, :i
with : = ,, , + 1, ...,.
117
Take C
(a, /) real, we have

J
+
|,, :i = [,(, + 1) :(:+ 1)]
1/2
~|,.:+ 1i ;
J
|,, :i = [,(, + 1) :(:1)]

1/2
~|,.:1i .
The eigenkets {|,, :i} form a basis for angular momentum operator
h,: |,
0
, :
0
i = c
0 c
nn
0 .
In the form of matrix, elements of J
h,
0
:
0
|J
| ,:i = ~ [,(, + 1) :(:1)]

1/2
c
0 c
n
0
n1
.
As a result,
h,
0
:
0
|J
r
| ,:i =
,
0
:
0
J
+
+J
,:
=
~
2
[,(, + 1) :(:+ 1)]
1/2
c
0 c
n
0
n+1
+
~
2
[,(, + 1) :(:1)]
1/2
c
0 c
n
0
n1
h,
0
:
0
|J
| ,:i
=
~
2i
[,(, + 1) :(:+ 1)]
1/2
c
0 c
n
0
n+1
~
2i
[,(, + 1) :(:1)]
1/2
c
0 c
n
0
n1
h,:
0
|J
:
| ,:i = :~c
nn
0 ;
,:
0
J
2
,:
= ,(, + 1)~
2
c
nn
0
118
Our choice of a representation in which J
2
and J
:
are diagonal has led to discrete sequences
of values for the corresponding labels j and m. The innite matrices thus obtained are most
conveniently handled by breaking them up into an innite set of nite matrices, each of
which is characterized by a particular value of , and has 2, + 1 rows and columns.
(1) , = 1,2: : = 1,2
J
r
=
~
2
3
E
E
C
0 1
1 0
4
F
F
D
J
=
~
2
3
E
E
C
0 i
i 0
4
F
F
D
J
:
=
~
2
3
E
E
C
1 0
0 1
4
F
F
D
J
2
=
3
4
~
2
3
E
E
C
1 0
0 1
4
F
F
D
(2) , = 1, : = 1, 0
J
r
=
~
2
12
3
E
E
E
E
E
E
C
0 1 0
1 0 1
0 1 0
4
F
F
F
F
F
F
D
J
=
~
2
12
3
E
E
E
E
E
E
C
0 i 0
i 0 i
0 i 0
4
F
F
F
F
F
F
D
J
:
= ~
3
E
E
E
E
E
E
C
1 0 0
0 0 0
0 0 1
4
F
F
F
F
F
F
D
J
2
= 2~
2
3
E
E
E
E
E
E
C
1 0 0
0 1 0
0 0 1
4
F
F
F
F
F
F
D
(3) , =
3
2
, : =
3
2
,
1
2
J
r
=
~
2
3
E
E
E
E
E
E
E
E
E
E
C
0 3
1/2
0 0
3
1/2
0 2 0
0 2 0 3
1/2
0 0 3
1/2
0
4
F
F
F
F
F
F
F
F
F
F
D
119
J
=
~
2
3
E
E
E
E
E
E
E
E
E
E
C
0 i3
1/2
0 0
i3
1/2
0 2i 0
0 2i 0 i3
1/2
0 0 i3
1/2
0
4
F
F
F
F
F
F
F
F
F
F
D
J
:
=
~
2
3
E
E
E
E
E
E
E
E
E
E
C
3 0 0 0
0 1 0 0
0 0 1 0
0 0 0 3
4
F
F
F
F
F
F
F
F
F
F
D
J
2
=
15
4
~
2
3
E
E
E
E
E
E
E
E
E
E
C
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
4
F
F
F
F
F
F
F
F
F
F
D
3.3.1 Representation of Rotation Operator
Have obtained the matrix elements of J
:
and J
, we are now in position to study the matrix

elements of rotation operator D(
R)D
()
nn
0
(1) =
,:
0
c
I n/~
,:
Since [J
2
, J
|
] = 0, we have
[D(1), J
2
] = 0
So D(1) |,, :i is still an eigenket of J
2
with the same eigenvalue ,(, + 1)~
2
J
2
D(1) |,:i = ,(, + 1)~
2
D(1) |,:i
We just consider the matrix elements of D(1) with the same ,. For every , there are 2, +1
values of m. D(1) should be a (2, + 1) (2, + 1) square matrix. For a rotated ket
1(1) |,:i =
X
n
0
|,:
0
i h,:
0
|1(1)| ,:
0
i =
X
n
0
1
()
nn
0
(1) |,:
0
i
120
As is well known, the Euler angles may be used to characterize the most general rotation.
We have
1
()
nn
0
(c, ,, ) =
D
,:
0
c
I
J
z
~
o
c
I
J
~
o
c
I
J
z
~
~
,:
E
= c
I(n
0
on~)
D
,:
0
c
I
J
~
o
,:
E
For , = 1,2
J
=
~
2
3
E
E
C
0 i
i 0
4
F
F
D
nD
,:
0
c
I
J
~
o
,:
Eo
=
3
E
E
C
cos
o
2
sin
o
2
sin
o
2
cos
o
2
4
F
F
D
To derive the matrix we can make use of the identity: (J
,})
2
= 1,4.
For , = 1,
J
=
~
2
1/2
3
E
E
E
E
E
E
C
0 i 0
i 0 i
0 i 0
4
F
F
F
F
F
F
D
{
D
,:
0
c
I
J
~
o
,:
E
} =
3
E
E
E
E
E
E
C
cos
2
o
2

1
2
12
sin, sin
2 o
2
1
2
12
sin, cos ,
1
2
12
sin,
sin
2 o
2
1
2
12
sin, cos
2
o
2
4
F
F
F
F
F
F
D
To derive the matrix we can make use of the identity: (J
,})
3
= J
,}.
3.4 Schwinger Oscillator Model.
There exists a very interesting connection between the algebra of angular momentum and
the algebra of two independent simple harmonic oscillators. Let us consider two types of
121
oscillator: plus and minus type. The creation and annihilation operator are denoted by a
and a
+
, respectively,
[a
+
, a
+
+
] = 1, [a
, a
+
] = 1
These two oscillators are uncoupled,
[a
+
, a
+
] = [a
, a
+
+
] = 0
Dene the number operators
+
= a
+
+
a
+
= a
+
We have
[
+
, a
+
] = a
+
[
, a
] = a
[
+
, a
+
+
] = a
+
+
[
, a
+
] = a
+
Since [
+
,
] = 0 , we dene the simultaneous eigenket of

+
and
by |:
+
, :
i such
that
|:
+
, :
i =
(a
+
+
)
n
+
(:
+
!)
1/2
!
(a
+
)
n
(:
!)
1/2
!
|0, 0i
122
with
a
|0, 0i = 0
a
+
+
|:
+
, :
i = (:
+
+ 1)
1/2
|:
+
+ 1, :
i
a
|:
+
, :
i = (:
)
1/2
|:
+
, :
1i
We dene
J
+
= ~a
+
+
a
; J
= ~a
+
a
+
J
:
=
~
2
(a
+
+
a
+
a
+
) =
1
2~
[J
+
, J
] =
~
2
(
+
_
)
We can prove that these operators satisfy the angular momentum relations
[J
:
, J] = ~J
Dene the total number operator N by

=
+
+
,
we can prove
J
2
= J
2
:
+
1
2
(J
+
J
+J
J
+
)
=

2
(
2
+ 1)~
2
.
123
Therefore, we have
J
2
|:
+
, :
i = ~
2
:
+
+:
2
(
:
+
+:
2
+ 1) |:
+
, :
i
J
+
|:
+
, :
i = ~a
+
+
a
|:
+
, :
i
= ~[(:
+
+ 1):]
1/2
|:
+
+ 1, :
1i
J
|:
+
, :
i = ~[:
+
(:
+ 1)]
1/2
|:
+
1, :
+ 1i
J
:
|:
+
, :
i =
~
2
(:
+
:
) |:
+
, :
i
These expression can be reduced to the familiar forms for angular momentum provided that
:
+
== , +:
:
== , :
:
+
, :
_
==|,, :i =
(a
+
+
)
+n
[(, +:)!]
1/2
(a
+
)
n
[(, :)!]
1/2
|0, 0i
So we can check
J
2
|,, :i = ,(, + 1)~
2
|,, :i
J
+
|,, :i = [,(, + 1) :(:1)]
1/2
~|,, :+ 1i
J
|,, :i = [,(, + 1) :(:1)]

1/2
~|,, :1i
J
:
|,, :i = :~ |,, :i
124
3.4.1 Spin 1/2 system
We dene the spin 1/2 operators o
and o
:
in terms of the creation and annihilation
operators a
+
c
and a
c
(o =, ). For o = 1,2,
:
%
+:
&
= 2: = 1.
We have two eigenkets for o
:
|+i = a
+
%
|0i
|i = a
+
&
|0i
The operators are expressed as
o
+
= ~a
+
%
a
&
o
= ~a
+
&
a
%
o
:
=
~
2
(a
+
%
a
%
a
+
&
a
&
)
Since [a
%
, a
+
%
] = [a
&
, a
+
&
] = 1, we usually call a and a
+
the boson operators. In the case of
o = 1,2, a
2
c
= (a
+
c
)
2
= 0. So they are called hard-core bosons since no two bosons can be
occupied at the same site. (Please check whether the commutation relations still hold if
n
a
%
, a
+
%
o
= a
%
a
+
%
+a
+
%
a
%
= 1
n
a
&
, a
+
&
o
= 1.
n
a
&
, a
+
%
o
= 0.
i.e., a are fermionic operators.)
125
3.4.2 Two-spin1/2 system
We now consider two-spin-1/2 system: o
1
and o
2
. For each o
I
(i = 1, 2), we have
|i+i = a
+
I%
|0i
|ii = a
+
I&
|0i
There are four possible congurations for combination of two spins
|1 + 2+i = |1+i |2+i = a
+
1%
a
+
2%
|0i
|1 + 2i = |1+i |2i = a
+
1%
a
+
2&
|0i
|1 2+i = |1i |2+i = a
+
1&
a
+
2%
|0i
|1 2i = |1i |2i = a
+
1&
a
+
2&
|0i
We are now in a position to construct eigenstates of the total spin
o = o
1
+o
2
and its .
component
o
:
= o
:
1
+o
:
2
.
We rst examine the two kets: |1+, 2+i and |1, 2i .
o
:
|1+, 2+i = (o
1:
+o
2:
) |1+, 2+i
= ~|1+, 2+i
o
2
|1+, 2+i = 2~
2
|1+, 2+i
We denote
|: = 1, : = 1i = |1+, 2+i = a
+
1%
a
+
2%
|0i
126
Similarly,
|: = 1, : = 1i = |1, 2i = a
+
1&
a
+
2&
|0i
Recall that
J
|,, :i
= [,(, + 1) :(:1)]
1/2
~ |,, :1i
|: = 1, : = 0i
=
1
2
1/2
~
o
|: = 1, : = 1i
=
1
2
1/2
(a
+
1%
a
+
2&
+a
+
1&
a
+
2%
) |0i
We have obtained three eigenkets for S and S
:
. There should exist another eigenket since
there are four possible congurations for two spins. The fourth eigenket must be a combi-
nation of |1+, 2i and |1, 2+i
|ci = a |1+, 2i +/ |1 2+i
It is easy to check
o
:
|ci = 0
We make use of property of orthogonality of eigenkets
h: = 1, : = 0 |ci = 0
From this equation, we obtain
a = / ==a = / =
1
2
1/2
Furthermore,
o
2
|ci = 0 ==|ci = |: = 0, : = 0i
127
1
2
1/2
(a
+
1%
a
+
2&
a
+
1&
a
+
2%
) |0i
3.4.3 Explicit Formula for Rotation Matrices.
Schwingers scheme can be used to derive in a very simple way, a closed formula for rotation
matrices. We apply the rotation operator D(1) to |,:i. In the Euler angle rotation the
only non-trivial rotation in the second one about y- axis . So
D(1) = D(c, ,, )
o=~=0
= c
I
o/~
Applying D(1) on |,, :i, we have
D(1) |,:i
= D(1)
(a
+
+
)
+n
[(, +:)!]
1/2
(a
+
)
n
[(, :)!]
1/2
|0i
=
[1(1)a
+
+
1
1
(1)
]
+n
[(, +:)!]
1/2
[1(1)a
+
1
1
(1)
]
n
[(, :)!]
1/2
D(1) |0i
J
=
1
2i
(J
+
J
) =
1
2i
(a
+
+
a
a
+
a
+
)
We have
J
|0i = 0
and
D(1) |0i = |0i
We come to calculate
)
(,) = D(1)a
+
D
1
(1) = c
I
J
~
o
a
+
c
+I
J
~
o
128
d
d,
)
+
(,) = i
1
~
c
I
J
~
o
[J
, a
+
+
]c
I
J
~
o
=
1
2
c
I
J
~
o
a
+
c
I
J
~
o
=
1
2
)
(,)
Similarly
d
d,
)
(,) =
1
2
)
+
(,)
d
2
d,
2
)
(,) + (
1
2
)
2
)
(,) = 0
The solutions are
)
(,) = cos
,
2
C
1
+ sin
,
2
C
2
The boundary conditions are
)
(,)|
o=0
= a
+
=

C
1
= a
+
d
d,
)
(,)
o=0
=
1
2
)
(,)|
o=0
==

C
2
= a
+
We have
)
(,) = cos
,
2
a
+
sin
,
2
a
+
D(1)
3
E
E
C
a
+
+
a
+
4
F
F
D
D
1
(1) =
3
E
E
C
cos
o
2
sin
o
2
sin
o
2
cos
o
2
4
F
F
D
3
E
E
C
a
+
+
a
+
4
F
F
D
Recalling the binomial theorem,
(r +j)
n
=
X
n=0
!
:!( :)!
r
n
j
n
D(1) |,:i =
[cos
o
2
a
+
+
+ sin
o
2
a
+
]
+n
[(, +:)!]
1/2

[cos
o
2
a
+
sin
o
2
a
+
+
]
n
[(, :)!]
1/2
|0i
=
X
n
0
d
()
n
0
n
(,) |,:
0
i
129
d
()
n
0
n
=
X
|
(1)
|n+n
0 [(, +:)!(, :)!(, +:
0
)!(, :
0
)!]
1/2
(, +:/)!/!(, / :
0
)!(/ :+:
0
)
(cos
,
2
)
22|+nn
0
(sin
,
2
)
2|n+n
0
Notice that d
()
n
0
n
(,) = h,:
0
| D(1) |,:i .
3.5 Combination of Angular Momentum and
Clebsh-Gordan Coe!cients
One of the important problems in quantum theory is the combination of the angular mo-
menta associated with two parts of a system (such as the spin and orbital angular momenta
of one electron) to form the angular momentum of the whole system. The addition of two
vectors
1
and
2
forms a triangle. The third vector =
1
+
2
. The maximal value of v is
1
+
2
and the minimal value of is |
1
2
| . We start with two commuting angular mo-
mentum operators J
1
and J
2
: all components of J
1
commute with all components of J
2
.The
orthonormal eigenkets of J
2
1
and J
1:
are |,
1
:
1
i and J
2
has no eect on them. Similarly,
|,
2
:
2
i are the eigenkets of J
2
2
and J
2:
, and J
1
has no eect on them. Denote
|,
1
,
2
:
1
:
2
i = |,
1
:
1
i |,
2
:
2
i
130
we have
J
2
1
|,
1
,
2
:
1
:
2
i = ,
1
(,
1
+ 1)~
2
|,
1
,
2
:
1
:
2
i
J
1:
|,
1
,
2
:
1
:
2
i = :
1
~|,
1
,
2
:
1
:
2
i
J
2
2
|,
1
,
2
:
1
:
2
i = ,
2
(,
2
+ 1)~
2
|,
1
,
2
:
1
:
2
i
J
2
2:
|,
1
,
2
:
1
:
2
i = :
2
~|,
1
,
2
:
1
:
2
i .
The innitesimal rotation operator that aects both subspaces is written as
(1
1
i
J
1
:cc
~
) (1
2
i
J
2
:cc
~
) = 1
1
1
2
i
J
1
1
2
+J
2
1
1
~
:cc
So we dene the total angular momentum
J = J
1
1
2
+1
1
J
2
= J
1
+J
2
It is noted that
J J = i~J
Since J
2
, J
2
1
, J
2
2
and J
:
are compatible, we denote their simultaneous eigenkets by
|,
1
,
2
,:i = |,:i
Recall that there are (2,
1
+ 1) eigenkets |,
1
:
1
i and (2,
2
+ 1) eigenkets |,
2
:
2
i . Therefore
there are (2,
1
+ 1) (2,
2
+ 1) kets
|,
1
,
2
:
1
:
2
i = |,
1
:
1
i |,
2
:
2
i (= |:
1
, :
2
i)
These kets form a complete and orthogonal set of basis:
1
X
n
1
=
1
2
X
n
2
=
2
|:
1
, :
2
i h:
1
, :
2
| = 1
131
The connection between |,:i and |:
1
, :
2
i is
|,:i =
X
n
1
n
2
|:
1
, :
2
i h:
1
, :
2
|,:i
{h:
1
, :
2
|,:i} are the Clebsh-Gordan coe!cients. Note that
(J
:
J
1:
J
2:
) |,:i = 0
(::
1
:
2
) h:
1
, :
2
|,:i = 0
It is apparent that h:
1
, :
2
|,:i is zero unless : = :
1
+:
2.
The largest value of m is ,
1
+,
2
and this occurs only when :
1
= ,
1
and :
2
= ,
2
. Therefore the largest , is ,
1
+ ,
2
, and
that there is only such a state
|,
1
+,
2
,
1
+,
2
i = |,
1
, ,
2
i
|, = ,
1
+,
2
: = ,
1
+,
2
i = |:
1
= ,
1
, :
2
= ,
2
i
The next largest value of m is ,
1
+ ,
2
1 and this occurs twice: when (1) :
1
= ,
1
and
:
2
= ,
2
1. and (2):
1
= ,
1
1 and :
2
= ,
2
. One of the two linear independent combination
of these two states (: = ,
1
+,
2
1) must be associated with the new state with , = ,
1
+,
2
and : = ,
1
+,
2
1. Since for each , value there must be values of m ranging from ,
1
+,
2
to ,
1
,
2
. The other combination is associated with , = ,
1
+,
2
1. By an extension of
this argument we can see that each j value, ranging from ,
1
+,
2
to |,
1
,
2
| by integer steps,
appears just once. Each , value is associated with 2, + 1 combinations of the original kets
1
+
2
X
=|
1
2
|
(2, + 1) = (2,
1
+ 1)(2,
2
+ 1)
which is equal to the number of |:
1
, :
2
i as expected.
132
3.5.1 Clebsch-Gordan coe!cients
The Clebsch-Gordan coe!cients form a unitary matrix. Furthermore the matrix elements
are taken to be real by convention. An immediate consequence is h,:|:
1
, :
2
i = h:
1
, :
2
|,:i .
The orthogonality conditions of the matrix are
(1). h:
1
, :
2
|:
0
1
, :
0
2
i = c
n
1
n
0
1
c
n
2
n
0
2
X
.n
h:
1
, :
2
|,:i h,:|:
0
1
, :
0
2
i = c
n
1
n
0
1
c
n
2
n
0
2
(2). h,:|,
0
:
0
i = c
0 c
nn
0
1
P
n
1
=
1
2
P
n
2
=
2
h,:|:
1
, :
2
i h:
1
, :
2
|,
0
:
0
i = c
0 c
nn
0
(3). h,:|:
1
, :
2
i 6= 0 only if : = :
1
+:
2
Recursion Relations
We apply J
to the left side of Eq.

|,:i =
2
X
n
0
1
=
1
2
X
n
0
2
=
2
|:
0
1
, :
0
2
i h:
0
1
, :
0
2
|,:i
J
+
|,:i
=
2
X
n
0
1
=
1
2
X
n
0
2
=
2
(J
1+
+J
2+
) |:
0
1
, :
0
2
i h:
0
1
, :
0
2
|,:i
[,(, + 1) :(:+ 1)]
1/2
|,:+ 1i
=
P
[,
1
(,
1
+ 1) :
0
1
(:
0
1
+ 1)]
1/2
|:
0
1
+ 1, :
0
2
i h:
0
1
, :
0
2
|,:i
+
P
[,
2
(,
2
+ 1) :
2
(:
2
+ 1)]
1/2
|:
0
1
, :
0
2
+ 1i h:
0
1
, :
0
2
|,:i
133
Multiplying h:
1
, :
2
| from the left side
[,(, + 1) :(:+ 1)]
1/2
h:
1
, :
2
|, :+ 1i
= [,
1
(,
1
+ 1) :
1
(:
1
+ 1)]
1/2
h:
1
1, :
2
|,:i
+[,
2
(,
2
+ 1) :
2
(:
2
+ 1)]
1/2
h:
1
, :
2
1|,:i
Similarly,
[,(, + 1) :(:1)]
1/2
h:
1
, :
2
|,:1i
= [,
1
(,
1
+ 1) :
1
(:
1
1)]
1/2
h:
1
+ 1, :
2
|,:i
+[,
2
(,
2
+ 1) :
2
(:
2
+ 1)]
1/2
h:
1
, :
2
+ 1|,:i
Construction Procedure
Combination of the two recursion relations and the orthogonality conditions can
leads to all Clebsch-Gordan coe!cients . We start with the sector with the largest value of
m :
(1) : = ,
1
+,
2
, , = ,
1
+,
2
h,
1
, ,
2
|,
1
+,
2
,
1
+,
2
i = 1
(2) : = ,
1
+,
2
1
(a) :
1
= ,
1
, :
2
= ,
2
1
(b) :
1
= ,
1
1, :
2
= ,
2
, = ,
1
+,
2
, = ,
1
+,
2
1
134
| ,
1
+,
2
,
1
+,
2
i = |,
1
, ,
2
i
[(,
1
+,
2
)(,
1
+,
2
+ 1) ((,
1
+,
2
)(,
1
+,
2
1)]
1/2
|,
1
+,
2
,
1
+,
2
1i
= [,
1
(,
1
+ 1) ,
1
(,
1
1)]
1/2
|,
1
1 , ,
2
i
+[,
2
(,
2
+ 1) ,
2
(,
2
1)]
1/2
|,
1
, ,
2
1i
h,
1
1, ,
2
|,
1
+,
2
,
1
+,
2
1i =
,
1
,
1
+,
2
1/2
h,
1
, ,
2
1|,
1
+,
2
,
1
+,
2
1i =
,
2
,
1
+,
2
1/2
h:
1
, :
2
|,
1
+,
2
1, ,
1
+,
2
1 i
These two coe!cients are determined by the orthogonality conditions
h,, , 1|, 1, , 1i = 0
h,
1
+,
2
, ,
1
+,
2
1|,
1
,
2
1i h,
1
,
2
1|,
1
+,
2
1, ,
1
+,
2
1i
+h,
1
+,
2
, ,
1
+,
2
1|,
1
1 ,
2
i h,
1
1 ,
2
|,
1
+,
2
1, ,
1
+,
2
1i = 0
,
2
,
1
+,
2
1/2
h,
1
, ,
2
1|,
1
+,
2
1, ,
1
+,
2
1i
+
,
1
,
1
+,
2
1/2
h,
1
1,,
2
|,
1
+,
2
1, ,
1
+,
2
1i = 0
r =
,
1
,
1
+,
2
1/2
j =
,
2
,
1
+,
2
1/2
135
;
A
A
?
A
A
=
h,
1
+,
2
1, ,
1
+,
2
1|,
1
+,
2
1, ,
1
+,
2
1i = 1
r
2
+j
2
= 1
;
A
A
?
A
A
=
h,
1
,
2
1|,
1
+,
2
1, ,
1
+,
2
1i = +
1
+
2
1/2
h,
1
1 ,
2
|,
1
+,
2
1, ,
1
+,
2
1i =
1
+
2
1/2
The convention here is that the rst matrix element, which has the form h,
1
, , ,
1
|,, ,i , is
real and positive.
(3) : = ,
1
+,
2
2
(a) :
1
= ,
1
:
2
= ,
2
2
(b) :
1
= ,
1
1 :
2
= ,
2
1
(c) :
1
= ,
1
2 :
2
= ,
2
, = ,
1
+,
2
, = ,
1
+,
2
1
, = ,
1
+,
2
2
h:
1
, :
2
|,
1
+,
2
, : i and h:
1
, :
2
|,
1
+,
2
1, : i are determined by the recursion relations
and the coe!cients in (2). The coe!cients h:
1
, :
2
|,
1
+,
2
2, : i are determined by the
orthogonality conditions.
In this construction procedure, the only di!cult part is the use of orthogonally,
which becomes progressively more complicated as the rank of the submatrix increases. How-
ever it needs be employed only once for each submatrix, and it is easier to work out an ex-
ample with particular numerical values for ,
1
and ,
2
than the general case just considered.
136
(a) J
|,
1
+,
2
, ,
1
+,
2
1i
h,
1
,
2
2|,
1
+,
2
, ,
1
+,
2
2i =
,
2
(2,
2
1)
(,
1
+,
2
)(2,
1
+ 2,
2
1)
1
2
h,
1
1,
2
1|,
1
+,
2
, ,
1
+,
2
2i =
4,
1
,
2
(,
1
+,
2
)(2,
1
+ 2,
2
1)
1
2
h,
1
2,
2
|,
1
+,
2
, ,
1
+,
2
2i =
,
1
(2,
1
1)
(,
1
+,
2
)(2,
1
+ 2,
2
1)
1
2
(b) J
|,
1
+,
2
1, ,
1
+,
2
1i
h,
1
,
2
2|,
1
+,
2
1, ,
1
+,
2
2i =
,
1
(2,
2
1)
(,
1
+,
2
)(,
1
+,
2
1)
1
2
h,
1
1,
2
1|,
1
+,
2
1, ,
1
+,
2
2i =
,
1
,
2
[(,
1
+,
2
)(,
1
+,
2
1)]
1
2
h,
1
2,
2
|,
1
+,
2
1, ,
1
+,
2
2i =
,
2
(2,
1
1)
(,
1
+,
2
)(,
1
+,
2
1)
1
2
3.6 Examples
3.6.1 Two spin-1/2 systems
3.6.2 Spin-orbit coupling
PROBLEM
1. Find the eigenvalues and eigenvectors of o
=
3
E
E
C
0 i
i 0
4
F
F
D
. Suppose an electron is in
the spin state
3
E
E
C
c
,
4
F
F
D
. If o
is measured, what is the probability of the result ~,2?

137
2. Let n be a unit vector in a direction specied by the polar angles (0, c). Show that
the component of the angular momentum in the direction n is
L
n
= sin0 cos cL
r
+ sin0 sincL
+ cos 0L
:
=
1
2
sin0
c
I
L
+
+c
I
L
+ cos L
:
.
If the system is in simultaneously eigenstates of L
2
and L
:
belonging to the eigenvalues
|(| + 1)~
2
and :~, i.e., ||, :i ,
(a) what are the possible results of a measurement of L
n
?
(b) what are the expectation values of L
n
and L
2
n
?
3. Obtain an explicit expression for
l
1
(!) = exp(i!n S,~)
in the form of a 2 2 matrix when S is the spin operator with o = 1,2. Let the unit
vector n have the polar angles 0 and c. Show explicitly that your matrix for l
1
(!) is
unitary and that it is equal to -1 when c = 2.
4. Prove a sequence of Euler rotations represented by
1(1,2)(c, ,, )
= exp
i
c
2
o
3
exp
i
,
2
o
2
exp
2
o
3
=
3
E
E
C
c
I(o+~)/2
cos
o
2
c
I(o~)/2
sin
o
2
c
I(o~)/2
sin
o
2
c
I(o+~)/2
cos
o
2
4
F
F
D
.
Because of the group properties of rotations, we expect that this sequence of operations
is equivalent to a single rotation about some axis by an angle c. Find c.
138
5. Calculate the matrix of Clebsch-Gordan coe!cients in the case of ,
1
= 1,2 and ,
2
=
1,2. According to the matrix, write explicitly the eigenstates of the total angular
momentum and its z component in terms of (,
1
, :
1
) and (,
2
, :
2
).
6. Use the raising or lowering operator to nd the following Clebsch-Gordan coe!cients
3
2
,
1
2
= a |1, 1i
1
2
,
1
2
+/ |1, 0i
1
2
,
1
2
.
139
Chapter 4
Symmetries in Physics
Symmetry is a fundamental attribute of the natural world that enables an investigator to
study particular aspects of physical systems by themselves. For example, the assumption
that space is homogeneous, or possesses translational symmetry, leads to the conclusion that
the linear momentum of a closed isolated system does not change as the system moves. This
makes it possible to study separately the motion of the center of mass and the internal motion
of the system. A systematic treatment of the symmetry properties and the conservation law
is useful in solving more complicated problems. It provides a deeper insight into the structure
of physics.
In this chapter we consider the geometrical symmetries that may be associated
with the displacements of a physical system in space and time, with its rotation and inversion
in space, and with the reversal of the sense of progression of time.
140
4.1 Symmetries and Conservation Laws
4.1.1 Symmetry in Classical Physics
An alternative way to describe a classical system is to introduce Lagrangian L:
d
dt
0L
0
I
0L
0
I
= 0
which is called the Lagrange equation. For a conservative system,
L = T V
where
V = V(
1
,
2
, ...
).
For example, the Lagrangian for a harmonic oscillator is
L =
1
2
: r
2
1
2
/r
2
.
From the Lagrange equation, we obtain
: r = /r.
This is the Newtonian equation of motion! When the potential is related to velocity or
generalized velocity, the Lagrangian is constructed in a dierent way. For instance, the
Lagrangian for a charged particle subjected to an electromagnetic eld
L = T c +

c
.
If the Lagrangian is unchanged under displacement in the general sense,
I

I
+c
I
141
i.e.,
0L
0
I
= 0,
from the Lagrange equation, it follows that
d1
I
dt
= 0
where that canonical momentum is dened as
1
I
=
0L
0
I
.
In other words, the canonical momentum is conserved, i.e. the law of conservation.
4.1.2 Symmetry in Quantum Mechanics
In quantum mechanics, we have learned to associate a unitary operator, say U, with an
operation like translation and rotation. It has become customary to call U a symmetry
operator regardless of whether the physical system itself possesses the symmetry to U.
When we say that a system possesses that symmetry to U, which implies that
l
Hl = H
For a continuous symmetry, we take an innitesmal displacement - and expand the operator,
l = 1
i-
~
G
where G is the Hermitian generator of the symmetry operator. Thus,
[G, H] = 0.
By virtue of the Heisenberg equation of motion, we have
dG
dt
= 0.
142
Hence G is a constant of the motion. In the language of quantum mechanics, we say that G
and H can be diagonalized simultaneously. Suppose |qi an eigenstate of G. Then, the state
at a later time
|q, t
0
, ti = l(t, t
0
) |qi
is also an eigenstate of G.
Proof.
G|q, t
0
, ti = Gl(t, t
0
) |qi
= l(t, t
0
)G|qi
= ql(t, t
0
) |qi .
4.1.3 Degeneracy
Suppose that
[H, l] = 0
for the symmetry operator l and |:i is an energy eigenket with eigenvalue 1
n
. Then, l |:i
is also an eigenket of H with the same energy eigenvalue.
Proof.
Hl |:i = lH |:i
= 1
n
l |:i .
Suppose |:i and l |:i represent two dierent kets. Then these two states with the same
energy are said to be degenerated.
143
Remark 3 The proof is very trivial, but the concept play a far more important role in
quantum mechanics.
4.1.4 Symmetry and symmetry breaking
If |:i is non-degenerated, l |:i and |:i must represent the same physical state although
they can dier a trivial phase factor, i.e.,
l |:i = c
Io
|:i .
In the case, we say that the state |:i also possesses the symmetry. If
l |:i 6= c
Io
|:i ,
we say that the symmetry is spontaneously broken in the state although the Hamiltonian
possesses the symmetry. The spontaneous symmetry breaking occurs when the state is
degenerated. The spontaneous broken symmetry is one of the most important concepts in
modern physics, from elementary particle physics to condensed matter physics.
Example 1
for a hydrogen atom problem,
H =
1
2:
1
2
+\ (r).
The Hamiltonian is rotational invariant
[1(1), H] = 0.
One of eigenstates is denoted by |:, ,, :i with 1
n
. Then
1(1) |:, ,, :i =
X
n
0
1
()
nn
0
|:, ,, :
0
i
144
is also an eigenket of H with the same energy 1
n
. For each pair of :
and , there are 2, + 1 values of m:
: = ,, , + 1, ...., ,
Therefore |:, ,, :i should be at least (2, + 1)-fold degenerated.
Example 2
Two-spin Problem
The Hamiltonian for two spins S
1
and S
2
is written as
H = JS
1
S
2
.
Denote the eigenstate for H by |o
|o|
, 'i
H |o
|o|
, 'i =
J
2
h
(S
1
+S
2
)
2
S
2
1
S
2
2
i
|o
|o|
, 'i
=
J
2
[o
|o|
(o
|o|
+ 1) 2:(: + 1)] |o
|o|
, 'i
The total spin can be
o
|o|
= 2:, 2: 1, ..., 0.
As the M can be taken to be
' = o
|o|
, o
|o|
+ 1, ..., o
|o|
.
and the eigenvalue of H is independent of M, so the state |o
|o|
, 'i is
(2o
|o|
+ 1)-fold degenerated.
The ground state energy:
145
(1) Antiferromagnetic J 0, o
|o|
= 0
1
= :(: + 1)J
(2) Ferromagnetic J < 0, o
|o|
= 2:
1
= :
2
J
SU(2) symmetry: When the system is rotated the rotation
operator
l = c
I
(S
1
+S
2
)r
~

As
[H, S] = 0,
we have
l
Hl = H.
For the ground state
(1) J 0, o
|o|
= 0
l |o = 0, ' = 0i = |o = 0, ' = 0i
The state also possesses the SU(2) symmetry.
(2)J < 0, o
|o|
= 2:
l |o = 2:, ' = 0i 6= |o = 2:, ' = 0i
The ground state is (4: + 1)-fold degenerate. The symmetry is spon-
taneously broken.
146
4.1.5 Summary: symmetries in physics
Space Translation r r +cr Momentum
Time evolution t t +ct Energy
Rotation r r
0
Angular Momentum
Space Inversion r r Parity
Time reversal t t Charge conjugate
Permutation (12) (21) Quantum Statistics
Gauge e
I0
Charge
Space translation
l
:
(j)rl
:
(j) = r +j
l
:
(j) = exp[i
P
~
j]
l
:
(j)Hl
:
(j) = H
[1, H] = 0
Time evolution
l
|
(t)tl
|
(t) = t +t
l
|
(t) = exp[i
H
~
t]
l
|
(t)Hl
|
(t) = H
|c(t)i = exp[iHt,~] |c(0)i
147
Rotation
r
0
= 1r
|ci
1
= D(1) |ci
D(1) = exp
ic
J n
~
(1)HD(1) = H
[J, H] = 0
Gauge Invariance
H = i~
0
0t
This symmetry is related to the conservation of charge, i.e., the number of the particles.
4.2 Discrete Symmetries
In this section we consider two symmetry operators that can be considered to be discrete
4.2.1 Parity
The parity operation, as applied to transformation on the coordinate system, changes a right-
handed (RH) system into a left-handed (LH) system. Let us denote the parity operator ,
which satises
;
A
A
A
A
A
A
A
A
A
A
?
A
A
A
A
A
A
A
A
A
A
=
|ci |ci
r = r
{r, } = 0
= 1
148
Figure 4.1:
= 1
=
1
=
The physical meaning is that under this operation
\ = (r, j, .) =\
0
= (r, j, .)
\
0
= 1\
where
1 =
3
E
E
E
E
E
E
C
1 0 0
0 1 0
0 0 1
4
F
F
F
F
F
F
D
How does an eigenket of the position operator transform under parity?Assume
r|ri = r|ri
r|ri = r|ri
r|ri = r|ri
r(|ri) = r(|ri)
149
Thus we conclude that
|ri = c
Io
|ri .
Applying the parity operation twice
2
r = ( r) = r
2
and
r
2
|ri =
2
r|ri = r
2
|ri
Thus
2
= 1.
Therefore its eigenvalue can be only +1 or -1.
4.2.2 The Momentum Operator
Space translation followed by parity is equivalent to parity followed by translation in the
opposite direction:
T (dr) |ri = |r +dri
= `|r dri
= T (dr)`|ri
= T (dr)|ri
Since
T (dr) = 1 i
1
~
dr +...,
it follows that
1 = 1
150
or
{, 1} = 0.
4.2.3 The Angular Momentum
Dene
L = r 1,
then
L = L.
For 3 3 matrices, we have
1
(o:I|)
1
(:o|o|Ion)
= 1
(:)
1
()
where
1
()
=
3
E
E
E
E
E
E
C
1 0 0
0 1 0
0 0 1
4
F
F
F
F
F
F
D
which implies that parity and rotation operation commute. Therefore
1(1) = 1(1).
1(1) = 1 iJ :-,} +
J = J
o = o
1 = 1
151
Let us now look at the parity property of wave function whose state ket is |ci :
(r) = hr|ci .
Suppose |ci the eigenket of parity
|ci = `|ci . (` = 1)
Then
hr| |ci = `hr |ci
hr |ci = `hr |ci
(r) = `(r)
The wave function is also parity.
Example 3
Solution for a square-well potential
~
2
2:
d
2
l(r)
dr
2
= 1l(r)
with |r| _ a. In the replacement r r,
~
2
2:
d
2
l(r)
dr
2
= 1l(r)
Thus if U(x) is one solution, U(-x) must be also one solution. The
solutions are
l(r) = c
I|r/}
where /
2
= 2:1,}
2
. We can construct
152
Odd Parity: l(r) l(r) = l
oJJ
(r)
Even Parity: l(r) +l(r) = l
tutn
(r)
sin/r,~ and cos /r,~
Example 4
Simple harmonic oscillator: The basic Hamiltonian is
H =
j
2
2:
+
1
2
:.
2
r
2
.
We have
H = H
and
x = x
p = p
<
A
A
@
A
A
>
=
;
A
A
?
A
A
=
= a
a = a
The eigenkets are
|:i =
1
:!
n
|0i
The ground state wave function is
hr|0i =
1
1/2
r
0
exp
1
2
r
2
r
2
0
So
hr|0i = hr|0i
How about hr|:i?
hr||:i = (1)
n
hr|:i .
hr||:i = (1)
n
hr|:i
153
At r = 0, we have hr = 0|:i = 0 for odd n.
Example 5
Double well potential
Let us now look at the parities of energy eigenstates.
Theorem 4 Suppose [H, ] = 0 and |:i is an eigenstate of H with eigenvalue E
n
H |:i = 1
n
|:i
then (a) if |:i is nondegenerate, |:i is also a parity eigenket; (b) if |:i is degenerate, |:i
may not be a parity eigenket.
Proof. We can construct two states
1
2
1/2
(1 ) |:i .
4.2.4 Lattice Translation
154
FIGURE (a) Periodic potential in one-dimension with periodicity a. (b) The periodic
potential when the barrier height between two adjacent lattice sites becomes innite.
(from Sakurai, 1994)
Consider a periodic potential in one dimension:
\ (r a) = \ (r).
We introduce a translational operator t(a) such that
t
(a)rt(a) = r +a;
t
(a)\ (r)t(a) = \ (r +a).

Thus the Hamiltonian is invariant under the translation
t
(a)Ht(a) = H.
As t is unitary, we have
Ht = tH.
t and H commute with each other. So t can be diagonalized with H simultaneously. t is
unitary, but not hermitian. We expect the eigenvalue to be a complex number of modulus
155
1. Denote |:i a state ket in which the particle is localized at the n
||
site. When t is applied
to it
t(a) |:i = |: + 1i .
To construct an eigenket for t, we consider a linear combination
|0i =
+4
X
n=4
c
In0
|:i .
This state is an eigenstate for t :
t(a) |0i =
+4
X
n=4
c
In0
t(a) |:i =
+4
X
n=4
c
In0
|: + 1i
= c
I0
+4
X
n=4
c
I(n+1)0
|: + 1i
t(a) |0i = c
I0
|0i
A more realistic example:
h:| H |:i = 1
0
h:| H |: 1i = t
and all other elements are equal to zero. That is
H |:i = 1
0
|:i t |: + 1i t |: 1i
H =
X
n
{1
0
|:i h:| t |: + 1i h:| t |: 1i h:|}
H |0i =
X
n
{1
0
|:i h:| t |: + 1i h:| t |: 1i h:|} |0i
= (1
0
2t cos 0) |0i
156
4.3 Permutation Symmetry and Identical Par-
ticles
4.3.1 Identical particles
Identical particles cannot be distinguished by means of any inherent properties, since other-
wise they could not be identical in all respect. In classical mechanics the existence of sharply
denable trajectories for individual particles makes it possible in principles to distinguish
between particles that are identical except for their path since each particle can be followed
during the course of an experiment. In quantum mechanics, the nite size and the spread-
ing of the wave packets that can describe individual particles often make it impossible to
distinguish between identical particles because of their positions, especially if they interact
with each others to an appreciable extent.
Distinguishable and indistinguishable identical particles
Two identical particles are indistinguishable if the wave functions for the two particles
overlap, and distinguishable if the wave functions do not overlap.
Permutation Symmetry
Let us consider the Hamiltonian of a system of two identical particles.
H =
1
2:
1
2
1
+
1
2:
1
2
2
+\
oI:
(|r
1
r
2
|) +\
tr|
(r
1
) +\
tr|
(r
2
)
Symmetry: to exchange the positions of two particles: 1 = 2, the Hamiltonian
is invariant
H ==H.
157
Dene the permutation operator
1
12
|r
1
i
1
|r
2
i
2
= |r
2
i
1
|r
1
i
2
1
12
H(r
1
, r
2
)1
12
= H(r
2
, r
1
).
Clearly
1
12
= 1
21
1
2
12
= 1
Consider a state ket for two particles:
|/i
1
|/
0
i
2
1
12
|/i
1
|/
0
i
2
= |/
0
i
1
|/i
2
We can construct two states
1
2
[(|/i
1
|/
0
i
2
|/
0
i
1
|/i
2
)]
which are eigenstates of P
12
with eigenvalues 1, respectively.
Our consideration can be extended to a system made up of many identical par-
ticles
1
I
|r
1
, r
2
, r
I
, r
, , r
i
= |r
1
, r
2
, r, , r
I
, , r
i
Clearly
1
2
12
= 1
158
just as before, and the allowed eigenvalues of P
I
are 1. It is important to note that, in
general
[1
I
, 1
ln
] 6= 0.
Connection between spin and statistics
Half-odd-integer spin particles are fermions: electrons, protons,
Integer spin particles are boson: mesons,
4
He nucleus,
Although the allowed values of P
I
are 1, system containing N identical particles
are either totally symmetrical or antisymmetric under the interchange of any pair.
Boson:
1
I
| /o:o::i = | /o:o::i
Fermions
1
I
| )cr:io::i = | )cr:io::i
It is empirical fact that a mixed symmetry does not occur.
The exclusion principle
An immediate consequence of the electron being a fermion is that the electrons must satisfy
the Pauli exclusion principle, which states that no two electrons can occupy the same state.
Let us consider a system containing N electrons without interaction
H(1, 2, , , ) = H
0
(1) +H
0
(2) + +H
0
().
Assume l
o
(:) are the wave functions of H
0
(:) with energy -
o
H
0
(:)l
o
(:) = -
o
l
o
(:).
159
The product of N one-particle eigen functions are the energy eigen wave function of total
system..
l(1, 2, , ) = l
o
(1)l
o
(:) l
~
()
Hl = (-
o
+-
o
+ +-
~
)l.
As electrons are fermions, we have to express the wave function in an antisymmetric form.
One way is to express as a determinant of the Us
l
.
(1, 2, , )
= det
3
E
E
E
E
E
E
E
E
E
E
C
l
o
(1) l
o
(2) l
o
()
l
o
(1) l
o
(2) l
o
()
.
.
.
.
.
.
.
.
.
.
.
.
l
~
(1) l
~
(2) l
~
()
4
F
F
F
F
F
F
F
F
F
F
D
.
This function is antisymmetric and satises the Schrodinger equation with energy
-
o
+-
o
+ +-
~
.
If two or more of the Us are the same,
l
.
(1, 2, , ) = 0
which is the so-called Pauli exclusion principle.
To simplify the notation we introduce the creation and annihilation operator c
160
and c such that
H
0
c
o
|0i = -
o
c
o
|0i
c
o
|0i = 0
c
o
c
o
+c
o
c
o
= 1
c
o
c
o
+c
o
c
o
= 0
The Shcrodinger equation can be reduced to
H |l
.
i = (-
o
+-
o
+ +-
~
) |l
.
i
with
|l
.
i = c
o
c
o
c
~
|0i .
Example 6
Two-particle problem
A system contains two particles, which is described by the
Hamiltonian
H = H
1
+H
2
where
H
1
=
1
2:
1
1
2
1
+
1
2
/
1
r
2
1
H
2
=
1
2:
2
1
2
2
+
1
2
/
2
r
2
2
(1) :
1
6= :
2
, /
1
6= /
2
(2) :
1
= :
2
, /
1
= /
2
161
4.4 Time Reversal
In this section we study a discrete symmetry, called time reversal. The terminology was rst
introduced by E. Wigner in 1932.
4.4.1 Classical cases
Let us rst look at the classical case: a motion of particle sujected to a certain force. Its
trajectory is given by the Newtonian equation of motion,
:
d
2
r
dt
2
= \\ (r).
If r(t) is the soltion of the equation, then r(t) is also the solution of the equation. In
another word, when we make a transformation t t, the Newtonian equation of motion
keeps unchanged. Of course we should notice the change of the boundary condition or initial
conditions for the problem.
Maxwell equations:
4.4.2 Antilinear Operators
4.4.3 Antiunitary operators
4.4.4 T for a zero spin particle
4.4.5 T for a nonzero spin particle
162
Chapter 5
Approximation Methods for
Bound States
As in the case of classical mechanics, there are relatively few physically interesting prob-
lems in the quantum mechanics which can be solved exactly. Approximation methods are
therefore very important in nearly all the applications of the theory. In this chapter we shall
develop several approximation methods.
Approximation methods can be conveniently divided into two groups, according
to whether the Hamiltonian of the system is time-independent or time-dependent.
5.1 The Variation Method
The variation method can be used for the approximate determination of the lowest or ground-
state energy level of a system when there is no closely related problem that is capable of
163
exact solution, so that the perturbation method is inapplicable.
5.1.1 Expectation value of the energy
We attempt to estimate the ground state energy by considering a trial ket,

, or a trial
wave function. To this end we rst prove a theorem of a great of importance. We dene the
expectation value

H such that
H =
0|
0
.
Then we can prove the following
Theorem 5 The expectation value of a trial state ket is always not less than the ground
state energy
H _ 1
0
.
Proof. We can always imagine that any trial state can be expanded as
=
X
|
|/i
/|
where |/i is an exact eigenket of H. The expectation value
H =
0|
=
P
|,l
0|/
h/|H||i
||
0|
=
P
|
0|/

/|
1
|
0|
_
P
|
0|/

/|
1
0
0|
0
= 1
0
.
In the last step we use the property 1
|
_ 1
0
.
164
The variation method does not tell us what kind of trial state ket are to estimate
the ground state energy. Quite often we must appeal to physical intuition. In practice, we
characterize the trial state ket by one or more parameters `
n
can compute

H as a function
of `
n
. Then we minimize the

H by setting the derivatives with respect to the parameters
zero
0

H
0`
n
= 0.
The result of the variation method is an upper limit for the ground state energy of the
system, which is likely to be close if the form of the trial state ket or the trial wave function
resembles that of the eigenfunction. Thus it is important to make use of any available
information or physical intuition in choosing the trial function.
5.1.2 Particle in a one-dimensional innite square well
We attempt to estimate the ground state energy of the innite-well (one-dimensional box)
problem dened by
\ =
;
A
A
?
A
A
=
0, for |r| < a
+, for r a.
The exact solution are, of course, well known:
hr|0i =
1
a
1/2
cos
r
2a
,
1
0
=
~
2
2:

2
4a
2
.
But suppose we did not know these. Evidently the wave function must vanish at r = a.
Further the ground state should has even parity. The simplest analytic wave function that
satises both requirements is
hr|0i = a
2
r
2
.
165
There is no variational parameter. We can compute

H as follows
H =
R
drh0|ri

2
2n
hr|0i
R
drh0|ri hr|0i
=
10
2
1
0
- 1.01321
0
.
A much better result can be obtained if we use a more general trial function
hr|0i = a
~
r
~
.
where is regarded a variational wave parameter. Straightforward algebra gives
H =
( + 1) (2 + 1)
2 1
~
2
4:a
2
,
which has a minimum at
=
1 + 6
1/2
2
- 1.72,
This gives
H =
5 + 26
1/2
2
1
0
. - 1.002981
0
The state with an odd parity?
hr|0i = r(a
~
r
~
) .
5.1.3 Ground State of Helium Atom
We use the variation method with a simple trial function to obtain an upper limit for the
energy of the ground state of the helium atom. The helium atom consists of a nucleus of
charge +2c surrounded by two electrons, we nd that its Hamiltonian is
H =
~
2
2:
\
2
1
+\
2
2
2c
2
1
r
1
+
1
r
2
+
c
2
r
12
= H
1
+H
2
+\
12
166
where r
1
and r
2
are two position vectors of two electrons with respect to the nucleus as origin,
and r
12
= |r
1
r
2
| . If we neglect the interaction energy
t
2
:
12
between the two electrons, the
ground state wave function of H would be product of two normalized hydrogenic wave
functions
(r
1
, r
2
) =
7
3
a
3
0
c
(2/o
0
)(:
1
+:
2
)
(5.1)
with a
0
= ~
2
,jc
2
and 7 = 2.
Why 7 = 2?
We shall use Eq.(5.1) as the trial wave function and permit Z to be the variation
parameter. The kinetic energy of each electron is
1
|1
=
Z
dr
1
dr
2
(r
1
, r
2
)
~
2
2:
\
2
1
(r
1
, r
2
) =
c
2
7
2
2a
0
.
The Coulomb energy of each electron is
1
c1
=
Z
dr
1
dr
2
(r
1
, r
2
)
2c
2
r
1
(r
1
, r
2
) =
2c
2
7
a
0
.
The interaction energy of two electrons is
1
In|
=
Z
dr
1
dr
2
(r
1
, r
2
)
c
2
r
12
(r
1
, r
2
) =
5c
2
7
8a
0
.
Here we make use of the formula
1
r
12
=
1
r
1
4
X
l=0
r
2
r
1
l
1
l
(cos 0), r
1
r
2
1
r
12
=
1
r
2
4
X
l=0
r
1
r
2
l
1
l
(cos 0), r
2
r
1
where 1
l
is the spherical harmonics and 0 is the angle of there two vectors.
167
Totally, the expectation value of the Hamiltonian is
H =
c
2
7
2
a
0
4c
2
7
a
0
+
5c
2
7
8a
0
=
c
2
a
0
7
2
27
8
7
=
c
2
a
0
"
7
27
16
27
16
2
#
.
The expectation is minimized at
7 = 27,16 - 1.69.
and
H =
27
16
2
c
2
a
0
= 2.85
c
2
a
0
.
The experimental value for a helium atom is 2.904c
2
,a
0
, so our estimation is about 1.9 high.
5.2 Stationary Perturbation Theory: Nonde-
generate Case
5.2.1 Statement of the Problem
In this section we shall discuss the Rayleigh-Schrodinger perturbation theory, which analyses
the modications of discrete energy levels and the corresponding eigenfunctions of a system
when a perturbation is applied.
We consider a time independent Hamiltonian H such that it can be split into two
parts, namely
H = H
0
+\.
168
The perturbation theory is based on the assumption that when V=0, both the exact eigen-
values 1
(0)
n
and the exact energy eigenkets

:
(0)
of H
0
are know
H
0
:
(0)
E
= 1
(0)
n
:
(0)
E
.
We are required to nd approximate eigenkets and eigenvalues for the full Hamiltonian
problem
(H
0
+\ ) |:i = 1
n
|:i .
In general, \ is not the full potential operator. We expect that the \ is not very large
and should not change the eigenvalues and eigenkets of H
0
very much.
It is customary to introduce a parameter ` such that we solve
(H
0
+`\ ) |:i = 1
n
|:i
instead of the original one. We take 0 _ ` _ 1. The case of ` = 0 corresponds to the
unperturbed problem and the case of ` = 1 corresponds to the full strength problem we
want to solve.
5.2.2 The Two-State Problem
Before we embark on a systematic presentation of the basic method, let us see how the
expansion in ` might indeed be valid in the exactly soluble two-state problem we have
encountered many times already. Suppose we have a Hamiltonian that can be written as
H = H
0
+`\
= 1
(0)
1
1
(0)
ED
1
(0)
+1
(0)
2
2
(0)
ED
2
(0)
+`\
12
1
(0)
ED
2
(0)
+`\
21
2
(0)
ED
1
(0)
169
where 1
(0)
n
and

:
(0)
are the exact eigenvalues and eigenstates for H

0
.The index 0 stands
for the unperturbed problem. We can also express the Hamiltonian by a square matrix
H =
3
E
E
C
1
(0)
1
`\
12
`\
21
1
(0)
2
4
F
F
D
.
What is the eigenvalues of H?
1
1
=
1
(0)
1
+1
(0)
2
2
+
5
7
1
(0)
1
1
(0)
2
2
!
2
+`
2
\
12
\
21
6
8
1/2
;
1
2
=
1
(0)
1
+1
(0)
2
2

5
7
1
(0)
1
1
(0)
2
2
!
2
+`
2
\
12
\
21
6
8
1/2
.
So this problem can be solved exactly. Let us suppose `\ is small compared with the
relevant energy scale, the dierence of the energy eigenvalues of the unperturbed problem:
`|\
12
|
1
(0)
1
1
(0)
2
Using the formula

(1 +r)
1/2
= 1 +
1
2
r
1
8
r
2
+ , (5.2)
we obtain the expansion of the energy eigenvalues in terms of `
1
1
= 1
(0)
1
+
`
2
\
12
\
21
1
(0)
1
1
(0)
2
+ ;
1
2
= 1
(0)
2

`
2
\
12
\
21
1
(0)
1
1
(0)
2
+ .
As the expansion in Eq.(5.2) holds for r < 1, the expansion is available in the case of small
\ .
5.2.3 Formal Development of Perturbation
We are dealing with the perturbation of a bound state. The assumption that V is small
compared with the energy scale suggests that we expand the perturbed eigenvalues and
170
eigenkets as power series in \ . This is most conveniently accomplished in the terms of the
parameter ` such that the zero, rst, etc., powers of ` correspond to the zero, rst, etc.,
orders of the perturbation calculation.
The perturbed wave function and energy level are written
|:i =
:
(0)
E
+`
:
(1)
E
+`
2
:
(2)
E
+ ;
1
n
= 1
(0)
n
+`1
(1)
n
+`
2
1
(2)
n
+
and are substituted into the wave equation to give
(H
0
+`\ )
:
(0)
E
+`
:
(1)
E
+`
2
:
(2)
E
+
1
(0)
n
+`1
(1)
n
+`
2
1
(2)
n
+
:
(0)
E
+`
:
(1)
E
+`
2
:
(2)
E
+
.
Since the Equation is supposed to be valid for a continuous range of `, we can equate
the coe!cients of equal powers of ` on the both sides to obtain a series of equations that
represent successively higher orders of the perturbation.
H
0
1
(0)
n
:
(0)
E
= 0 (5.3)
H
0
1
(0)
n
:
(1)
E
= (1
(1)
n
\ )
:
(0)
E
(5.4)
H
0
1
(0)
n
:
(2)
E
= (1
(1)
n
\ )
:
(1)
E
+1
(2)
n
:
(0)
E
(5.5)
H
0
1
(0)
n
:
(3)
E
= (1
(1)
n
\ )
:
(2)
E
+1
(2)
n
:
(1)
E
+1
(3)
n
:
(0)
E
(5.6)
.
.
.
Zero-order perturbation
The Eq. (5.3) means that

:
(0)
is one of the unperturbed eigenkets, as expected.

We can obtain it directly by taking ` = 0. The eigenvalue is 1
(0)
n
. Since we are dealing
171
with the perturbation of a bound state, the state ket

:
(0)
is discrete. It is assumed to
be nondegenerate as well, although others of unperturbed eigenkets may be degenerate or
continuously distributed in energy.
First-order perturbation
Multiplying an eigen bra of the unperturbed Hamiltonian from the left hand on
both sides of the Eq.(5.4), we obtain
D
:
(0)
H
0
1
(0)
n
:
(1)
E
=
D
:
(0)
(1
(1)
n
\ )
:
(0)
E
(5.7)
D
:
(0)
H
0
1
(0)
n
:
(1)
E
=
D
:
(0)
(1
(1)
n
\ )
:
(0)
E
(5.8)
where : 6= :. The Eq. (5.7) gives
1
(1)
n
=
D
:
(0)
:
(0)
E
= \
nn
.
It is convenient to calculate

:
(1)
by expanding it in terms of the unperturbed eigenkets
:
(1)
E
=
X
n
c
(1)
n
:
(0)
E
where the summation runs over all possible eigenkets of H
0
. To simplify the notation, we
dene
\
nn
=
D
:
(0)
:
(0)
E
From the Eq.(5.5),
H
0
1
(0)
n
X
n
c
(1)
n
:
(0)
E
= (1
(1)
n
\ )
:
(0)
E
X
n
c
(1)
n
1
(0)
n
1
(0)
n
:
(0)
E
= (1
(1)
n
\ )
:
(0)
E
multiplying
:
(0)
from the left hand side, we obtain

c
(1)
n
=
\
nn
1
(0)
n
1
(0)
n
.
172
Second-order perturbation
The second-order correction to the energy is
1
(2)
n
=
D
:
(0)
:
(1)
E
=
X
n6=n
\
nn
\
nn
1
(0)
n
1
(0)
n
.
Similarly, we expand

:
(2)
in terms of the unperturbed eigenkets
:
(2)
E
=
X
n
c
(2)
n
:
(0)
E
.
To calculate c
(2)
n
, we apply
:
(0)
from the left hand on both side of the Equation and obtain
D
:
(0)
H
0
1
(0)
n
:
(2)
E
=
D
:
(0)
(1
(1)
n
\ )
:
(1)
E
+
D
:
(0)
1
(2)
n
:
(0)
E
.
That is
c
(2)
n
=
1
1
(0)
n
1
(0)
n
nD
:
(0)
(1
(1)
n
\ )
:
(1)
E
+
D
:
(0)
1
(2)
n
:
(0)
Eo
.
Summary
In short, the explicit expression for the energy expansion is
1
n
= 1
(0)
n
+`\
nn
+`
2
X
n6=n
\
nn
\
nn
1
(0)
n
1
(0)
n
+ .
The expansion of the perturbed eigenket is as follows
|:i =
:
(0)
E
+`
X
n6=n
\
nn
1
(0)
n
1
(0)
n
:
(0)
E
+`
2
X
n6=n
X
l6=n
\
nl
1
(0)
n
1
(0)
n
\
ln
1
(0)
l
1
(0)
n
:
(0)
E
`
2
X
l6=n
\
nn
1
(0)
l
1
(0)
n
\
nn
1
(0)
l
1
(0)
n
:
(0)
E
+ .
173
Remark 6 The state ket |:i is not normalized. A normalized state ket should be
1
h:|:i
|:i
Example 1
Two-state problem
5.3 Application of the Perturbation Expan-
sion
To illustrate the perturbation method we developed in the last lecture, let us look at several
examples. All examples can be solved analytically. We can compare the perturbation results
with the exact results.
5.3.1 Simple harmonic oscillator
The rst two examples concern a simple harmonic oscillator whose unperturbed Hamiltonian
is the usual one
H
0
=
j
2
2:
+
1
2
:.
2
r
2
.
As we already know the Hamiltonian can be solved in this way
H
0
= ~.
a +
1
2
174
where
a
:.
2~
1/2
r i
j
:.
a =
:.
2~
1/2
r +i
j
:.
with the eigenvalues

1
n
= ~.
: +
1
2
.
Example 2
\
0
=
:
2
:.
2
r
2
This problem can be solved exact in this way:
H = H
0
+\
0
=
j
2
2:
+
1
2
:(1 +-).
2
r
2
.
The eigenvalues are
1
n
= ~(1 +-)
1/2
.
: +
1
2
.
In the perturbation approach, we just concern the ground state energy.
The rst order perturbation is
1
1
= h0| \
0
|0i = - h0|
1
2
:.
2
r
2
|0i =
-
4
~.
The second order perturbation is
1
2
=
X
|6=0
|h/| \
0
|0i|
2
1
0
1
|
=
|h2| \
0
|0i|
2
1
0
1
2
=
-
2
16
~.
Thus the perturbation correction to the ground state energy is
41 = ~.
-
4

-
2
16
+ 0(-
3
)
.
175
Compare it with the exact result
Example 3
\
0
= `r
This problem can also be solved exactly:
H = H
0
+\
0
=
j
2
2:
+
1
2
:.
2
r
2
+`r
=
j
2
2:
+
1
2
:.
2
(r r
0
)
2
1
2
:.
2
r
2
0
where
r
0
=
`
:.
2
.
Making a replacement, j = r r
0
and 0
r
= 0
, we have
H =
~
2
2:
d
2
dj
2
+
1
2
:.
2
j
2
1
2
:.
2
r
2
0
.
The eigenvalues are
1
n
= ~.
: +
1
2
`
2
2:.
2
.
The rst order perturbation is
1
1
= h0| \
0
|0i = 0
The second order perturbation is
1
2
=
X
|6=0
|h/| \
0
|0i|
2
1
0
1
|
=
|h1| \
0
|0i|
2
1
0
1
1
=
`
2
2:.
2
.
176
5.3.2 Atomic hydrogen
The Schrodinger equation for a hydrogen atom is
(
~
2
2j
\
2
7c
2
r
) = 1
In the spherical coordinates,
\
2
=
1
r
2
0
0r
r
2
0
0r
+
1
r
2
sin0
0
00
sin0
0
00
+
1
r
2
sin
2
0
0
2
0c
2
.
The wave function can be written as
= 1(r)Y
ln
(0, c)
where Y
ln
(0, c) is a spherical harmonic. (See Schi, P.76 or Sakurai, P. 454). It is the
eigenfunction for orbital angular momentum operators
1
2
Y
ln
(0, c) = |(| + 1)~
2
Y
ln
(0, c)
1
:
Y
ln
(0, c) = :~Y
ln
(0, c)
The Schrodinger equation is reduced to
~
2
2j
1
r
2
0
0r
r
2
0
0r
|(| + 1)
r
2
7c
2
r
1 = 11.
The solution of this equation is
1
n
=
j7
2
c
4
2~
2
:
2
=
7
2
c
2
2a
0
:
2
where : = :
0
+| + 1 and a
0
= ~
2
,jc
2
. The rst three radial functions are
1
10
=
7
a
0
3/2
2c
2:/o
0
1
20
=
7
2a
0
3/2
2
7r
a
0
c
2:/2o
0
1
21
=
7
6a
0
3/2
7r
a
0
c
2:/2o
0
177
Degeneracy
: = 1: the ground state is nondegenerate
: = 2: the rst excited state is four-fold degenerate. (1) | = 0 and : = 0; (2)
| = 1 and : = 1
Example 4
Gravitational energy shift
Let consider an ordinary hydrogen atom (proton + elec-
tron) whose unperturbed Hamiltonian is
H
0
=
~
2
2j
\
2
7c
2
r
where j = :
t
'
,(:
t
+ '
) is the reduced mass. Now the proton

and electron interact not only through the electrostatic potential, but
also by means of the gravational interaction. The perturbation due to
the gravitational force is
\ = G
:
t
'
r
where G is the gravitational constant. The ground state wave function
of atomic hydrogen is
= 1
10
(r)Y
00
(0, c)
where
Y
00
(0, c) =
1
4
1/2
To rst order in the perturbation theory, the energy shift of the ground
178
state of atomic hydrogen due to this perturbation is
1
(1)
1
= h0| \
0
|0i
=
Z
dr |Y
00
(0, c)1
10
(r)|
2
\ (r)
=
Z
t
0
d0 sin0
Z
2t
0
dc
Z
4
0
drr
2
|Y
00
(0, c)1
10
(r)|
2
\ (r)
=
Z
4
0
drr
2
4
a
3
0
c
:/o
0
G
:
t
'
= G
:
t
'
a
0
The ground state energy is
1
(0)
1
=
c
2
2a
0
.
The relative energy shift is
1
(1)
1
1
(0)
1
=
2G:
t
'
c
2
- 8.7 10
40
.
Needless to say, it is not necessary to calculate higher order correction
in the present case.
5.4 Stationary Perturbation Theory: Degen-
erate Case
Until now we have assumed that the perturbed state diers slightly from an unperturbed
bound state. The perturbation method for a nondegenerate case we developed fails when
the unperturbed state are degenerate.
179
5.4.1 Revisited two-state problem
Starting from the two-state Hamiltonian
H =
3
E
E
C
1
(0)
1
`\
12
`\
21
1
(0)
2
4
F
F
D
,
we consider the special case of 1
(0)
1
= 1
(0)
2
, that is, the two states are energy degenerate.
In fact, this problem can be solved exactly; the energy eigenvalues are,
1
1
= 1
(0)
1
+`(\
12
\
21
)
1
2
; (5.9)
1
2
= 1
(0)
1
`(\
12
\
21
)
1
2
. (5.10)
However we cannot get this result by using the nondegenerate perturbation theory. A blind
application of the nondegenerate perturbation formula obviously runs into di!culty because
1
1
= 1
(0)
1
+
`
2
\
12
\
21
1
(0)
1
1
(0)
2
+ ;
1
2
= 1
(0)
2

`
2
\
12
\
21
1
(0)
1
1
(0)
2
+ .
become singular if \
12
is nonvanishing and 1
(0)
1
= 1
(0)
2
. So we must modify the method of
the nondegenerate case to accommodate such a situation. One way to avoid the catastrophe
is to choose our base kets in such a way that `\ has no o-diagonal matrix element. In the
two-state case, we choose a new base ket
:
(0)
E
= a
n
1
(0)
E
+/
n
2
(0)
E
the transformed perturbation becomes
`
\ = `
3
E
E
C
(\
12
\
21
)
1/2
0
0 (\
12
\
21
)
1/2
4
F
F
D
.
180
The solution to the coe!cients a
n
and /
n
are
/
1
a
1
=
(\
12
\
21
)
1/2
\
21
;
/
2
a
2
=
(\
12
\
21
)
1/2
\
12
and the state should be normalized
|a
n
|
2
+|/
n
|
2
= 1.
The transformation matrix is
l =
3
E
E
C
a
1
/
1
a
2
/
2
4
F
F
D
.
In this way the degeneracy of the perturbed state is removed. We can consider the perturba-
tion correction according to the nondegenerate perturbation theory. Note that we choose the
perturbation potential is diagonal when the unperturbed state is degenerate. The following
problem is left as an exercise:
\ =
3
E
E
C
`\
11
`\
12
`\
21
`\
22
4
F
F
D
as a perturbation to a two-fold degenerate state.
5.4.2 The basic procedure of degenerate perturbation
theory
We see from the two-state example that the degeneracy can be removed by choosing a new
set of base kets. Here we present a more detailed approach. Suppose now that there are
two states, |:i and |:i, that have the same unperturbed energy. Then we cannot apply the
nondegenerate perturbation formula when 1
n
= 1
n
unless it happens that \
nn
= 0. We
181
rst consider the case in which \
nn
6= 0. the initial state is not specied by its unperturbed
energy; the state may be |:i or |:i or any linear combination of them. Let us suppose
that the perturbation \ removes the degeneracy in some order, so that for nite ` there are
two states that have dierent energy. We start from the perturbation equations in the last
lecture
H
0
1
(0)
n
:
(0)
E
= 0 (5.11)
H
0
1
(0)
n
:
(1)
E
= (1
(1)
n
\ )
:
(0)
E
. (5.12)
Out of innite number of combinations of the two states we choose the particular pair which
depends on \. From the Eq.(5.11), we obtain
|ci = c
n
:
(0)
E
+c
n
:
(0)
E
;
1
(0)
o
= 1
(0)
n
= 1
(0)
n
.
Substitute the combined state into the Eq.(5.12) and take the inner product of this equation
successively with

:
(0)
and

:
(0)
, we obtain
\
nn
1
(1)
o
c
n
+\
nn
c
n
= 0
\
nn
1
(1)
o
c
n
+\
nn
c
n
= 0.
The two solutions of this equation are
1
(1)
o
=
1
2
(\
nn
+ \
nn
)
1
2
h
(\
nn
\
nn
)
2
+ 4 |\
nn
|
2
i
1/2
These two solutions are equal if and only if
\
nn
= \
nn
\
nn
= 0.
182
In this case we say the degeneracy is not removed in the rst order perturbation and we
have to consider the second order perturbation. On the other hand if either or both of two
equations are not satised, the two valued of 1
(1)
o
are distinct, and each can be used to
calculate c
n
and c
n
. In this case, we can use the nondegenerate perturbation theory to
determine the higher order perturbation. The coe!cients c
(1)
l
are determined by
c
(1)
l
(1
(0)
l
1
(0)
n
) = \
ln
c
n
\
ln
c
l
for | 6= :, : if we assume that c
(1)
n
= c
(1)
n
= 0.
How to solve triplet degenerate problem?
The basic procedure of the degenerate perturbation theory:
(1) Identify degenerate unperturbed eigenkets and construct the perturbation
matrix \ , a q q matrix if the degeneracy is g-fold.
(2) Diagonalization the perturbation matrix by solving, as usual, the appropriate
secular equation.
(3) Identify the roots of secular equation with the rst-order energy shifts; the
base kets that diagonalize the \ matrix are the correct zero-order kets to which the perturbed
kets approach in the limit ` 0;
(4) For higher orders use the formulas of the corresponding nondegenerate per-
turbation theory except in the summations, where we exclude all contributions from the
unperturbed kets in the degenerate subspace.
5.4.3 Example: Zeeman Eect
The change in the energy levels of an atom caused by a uniform external magnetic eld is
called the Zeeman eect. We now consider the change of rst order in the eld strength for
183
a hydrogen atom
A constant magnetic eld can be represented by the vector potential
A =
1
2
Br
since B = 5A. The energy related to the vector potential is
4H =
ic~
jc
A 5+
c
2
2jc
A
2
=
c
2jc
B L +
c
2
8jc
2
1
2
r
2
sin
2
0
where L = r p, 0 is the angle between r and H, and e is a positive quantity. Since we
wish to work only to the rst order in B we can put
\ =
c
2jc
B L.
The energy eigenfunctions of the unperturbed hydrogen atom are usually chosen to be
eigenstates of 1
:
with the eigenvalues :~. We choose the magnetic eld is in the z-direction,
then the rst order perturbation is
1
(0)
1
= h:| \ |:i =
c
2jc
1:~.
So the degeneracy of the 2| + 1 states is removed in rst order.
(1). s-wave: | = 0, and : = 0;
(2). p-wave: | = 1, and : = 1, 0, and 1.
5.4.4 Example: First Order Stark Eect in Hydrogen
The change in the energy levels of an atom cause by a uniform external electric eld of
strength E is called the Stark eect. The perturbation V is now the extra energy of the
184
nucleus and electron in the external eld and is ready shown to be
\ = c1r cos 0
where the polar axis and E are in the direction of positive . and c is again a positive quantity.
In a hydrogen atom the Coulomb potential is rotationally invariant. The wave function for
any spherically symmetric ponetial energy, when expressed in spherical harmonics, have
even or odd parity according as the azimuthal quantum number l is even or odd. Since
the perturbation is odd parity with respect to inversion, the ground state has even parity
and has no rst-order Stark eect. The rst excited state (: = 2) of hydrogen is four fold
degenerate; the quantum numbers | and : have the value (0, 0), (1, 1), (1, 0), (1, 1). Since
the perturbation \ commutes with the z-component of angular momentum,
[1
:
, \ ] = 0,
we obtain
(:
I
:
|
) h|, :
I
| \ ||
0
, :
|
i = 0.
The nonvanishing matrix elements of \ are
\
1
= h1, 0| \ |0, 0i = h0, 0| \ |1, 0i
= c1
Z
drn
210
(r)r cos 0n
200
( r)
=
c1
16a
4
0
Z
+1
1
cos
2
0d cos 0
Z
4
0
dr(2
r
a
0
)c
:/o0
= 3c1a
0
.
185
The rst order energy correction is determined by the secular equation
1
1
\
1
0 0
\
1
1
1
0 0
0 0 1
1
0
0 0 0 1
1
= 0.
The four roots of the secular equation are
1
1
=
;
A
A
A
A
A
A
A
A
A
A
?
A
A
A
A
A
A
A
A
A
A
=
|\
1
|
0
0
+|\
1
|
,
so that half of the four fold degeneracy is removed in the rst order.
5.5 The Wentzel-Kramers-Brillouin (WKB) ap-
proximation
5.6 Time-dependent Problem: Interacting Pic-
ture and Two-State Problem
When the Hamiltonian depends on the time, there are no stationary solution of the Schrdinger
equation. Thus our identication of a bound state with a discrete energy level and stationary
eigenket must be modied.
186
5.6.1 Time-dependent Potential and Interacting Picture
We consider a Hamiltonian H such that is can be split into two parts,
H = H
0
+\ (t),
where H
0
does not contain time explicitly. The problem of H
0
or \ (t) = 0 is assumed to be
solved completely, i.e.,
H
0
|:i = 1
n
|:i .
Assume \ (t) = 0 at t = 0. the state ket is given by
|ci =
X
n
c
n
(0) |:i .
In the system of H
0
, the state ket at a later time will evolve into
|ci =
X
n
c
n
(0)c
IJ
r
|/~
|:i .
Our question is how the state ket of H (not H
0
) changes as time goes on.
Now we must work with the time-dependent Schrodinger equation
i~
0
0t
|c(t)i = (H
0
+\ (t)) |c(t)i . (5.13)
Our procedure consists in expressing
|c(t)i =
X
n
c
n
(t)c
IJ
r
|/~
|:i . (5.14)
Substituting Eq. (5.14) into Eq. (5.13), we obtain
i~
0
0t
c
|
=
X
n
h/| \ |:i c
n
c
IJ
!r
|/~
(5.15)
where the Bohr frequency is dened as
.
|n
=
1
|
1
n
~
.
187
This equation is exactly equivalent to the original problem.
Interaction picture
Two kinds of changes have been made in going from Eq. (5.13). First, we have
changed the representation from specied in terms of the coordinates to being specied in
terms of the unperturbed energy eigenvalues. Second, we have changed from the Schrodinger
to the interaction picture.
h/| \
1
|:i = h/| c
I1
0:
|/~
\
S
c
I1
0:
|/~
|:i
= h/| \
S
|:i c
IJ
!r
|/~
.
5.6.2 Time-dependent Two-State Problem
Exact soluble problem of time-dependent potential are rather rare. However, a two state
problem with sinusoidal oscillating potential can be solved exactly.
The Problem is dened by
H
0
= 1
1
|1i h1| +1
2
|2i h2| (1
2
1
1
)
\ (t) = c
I.|
|1i h2| +c
I.|
|2i h1| .
In the interaction picture, the Eq. (5.15) can be written as
i~
0
0t
3
E
E
C
c
1
(t)
c
2
(t)
4
F
F
D
=
3
E
E
C
0 c
I4.|
c
I4.|
0
4
F
F
D
3
E
E
C
c
1
(t)
c
2
(t)
4
F
F
D
The equation is reduced to
i~
0
0t
c
1
(t) = c
I4.|
c
2
(t) (5.16)
i~
0
0t
c
2
(t) = c
I4.|
c
1
(t) (5.17)
188
where 4. = . +1
1
1
2
.
How to solve this equation?
From Eq. (5.16) we have
c
2
(t) = (c
I4.|
)
1
i~
0
0t
c
1
(t)
Substituting c
2
into Eq.(5.17),
i~
0
0t
(c
I4.|
)
1
i~
0
0t
c
1
(t)
= c
I4.|
c
1
(t)
0
2
0t
2
c
1
+i.
0
0t
c
1
=

2
~
2
c
1
Assume c
1
= c exp[i`t], we have
`
2
.` =
2
,~
2
;
` =
1
2
.
p
.
2
+
2
,~
2
So the general solutions are

c
1
= c
+
c
1
2
(4.+.
T
)|
+c
c
1
2
(4..
T
)|
c
2
=
~
2
(4. +.
1
)c
+
c
1
2
(4..
T
)|
~
2
(4. .
1
)c
1
2
(4.+.
T
)|
c are two coe!cients and determined by the initial conditions and
.
1
=
4
2
~
2
+ (. +.
12
)
2
1/2
.
If initially, at t=0, only the lower is populated so that
c
1
(0) = 1, and c
2
(0) = 0
then the probability for being found in each of the two states is given by
|c
2
(t)|
2
=

2
2
+ (. +.
12
)
2
~
2
,4
sin
2
1
2
.
1
t
,
|c
1
(t)|
2
= 1 |c
2
(t)|
2
189
Figure 5.1:
This is Rabis formula, after I.I. Rabi.
Spin Magnetic Resonance
The two-state problem has many physical applications: Spin magnetic resonance,
maser etc. Consider a spin 1/2 system subjected to a t-independent uniform magnetic eld
in the z-direction and, in addition, a t-dependent magnetic eld rotating in the xy plane.
1 = 1
0
. +1
1
( rcos .t + j sin.t)
with 1
0
and 1
1
constant.
We treat the eect of the uniform t-independent eld as H
0
and the eect of the
rotating eld as \ (t).
H0 =
c~1
0
2:
t
c
(|+i h+| |i h|)
\ (t) =
c~1
1
2:
t
c
[cos .t (|+i h| +|i h+|) + sin.t (i |+i h| +i |i h+|)]
Four Nobel Prize winners who took advantage of resonance in two-level systems
Rabi (1944): on molecular beams and nuclear magnetic resonance;
190
Figure 5.2:
Bloch and Purcell (1952): on B eld in atomic nuclei and nuclear magnetic moments;
Townes, Basov, and Prochorov (1964): on masers, lasers, and quantum optics;
Kastler (1966): on optical pumping
5.7 Time-dependent Perturbation Problem
5.7.1 Perturbation Theory
We now return to Eq. (5.3), replace \ by `\ , and express the c
n
as the power series in `
c
n
= c
(0)
n
+`c
(1)
n
+`
2
c
(2)
n
+ .
The substitution yields the set of equations
0
0t
c
(0)
n
= 0
i~
0
0t
c
(s+1)
n
=
X
|
h:| \ |/i c
(s)
|
c
+IJ
r!
|/~
.
191
The zero-order coe!cient are constant in time. We shall assume that all except one of the
c
(0)
n
, say : = :, are zero, so that the system is initially in a denite unperturbed energy
state. Thus the solutions to the zero order coe!cients are
c
(0)
n
= h:|:i = c
nn
.
Integration of the rst-order equation gives
c
(1)
n
= (i~)
1
Z
|
4
h:| \ (t) |:i c
IJ
rr
|/~
dt (5.18)
The integral constant has been chosen in such a way that c
(1)
n
vanish at t = t
0
. To rst order
in the perturbation, the transition probability corresponding to the transition : :, (that
is, the probability that the system, initially in the state :, be found at time t in the state
: 6= :) is therefore
1
nn
=
c
(1)
n
2
, : 6= :.
It is also worth noting that for t t
0
the coe!cient c
n
of the state : is given to rst order
in the perturbation by
c
n
(t) - c
(0)
n
(t) +c
(1)
n
(t)
- 1 + (i~)
1
Z
|
4
h:| \ (t) |:i dt
- exp
i
~
Z
|
4
h:| \ (t) |:i dt
So the principal eect of the perturbation on the initial state is to change the phase.
192
5.7.2 Time-independent perturbation
The result take particular simple form if the perturbation \ is independent of time, except
for being switched on suddenly at a given time. We then obtain
c
(1)
n
=
i
~
h:| \ |:i t
c
(1)
n
=
h:| \ |:i
~
1 c
IJ
rr
|/~
1
nn
Hence
c
n
c
1T
r
t~
|:i - c
(T
r
+\
rr
)t~
|:i
This is in agreement with the result using the stationary perturbation theory. the rst order
transition probability from : to : is given by
1
(1)
nn
= 2
h:| \ |:i
~
sin(1
nn
t,2)
1
nn
2
It is worth noting that this is valid for nondegenerate case.
5.7.3 Harmonic perturbation
Eq. (5.18) takes a particularly simple form if the perturbation \ depends on harmonically
on the time except for being turned on at one time and o at a latter time. We shall call
these two times 0 and t
0
, respectively, and assume that we can write
h/| \ (t) |:i = h/| \ |:i sin.t
where h/| \ |:i is time independent. The rst order amplitude at any time t at or after t
0
c
(1)
n
=
h:| \ |:i
i~
c
I(.
rr
+.)|
0
1
.
nn
+.

c
I(.
rr
.)|
0
1
.
nn
.
The structure of Eq. (??) suggests that the amplitude is appreciable only when the denom-
inator of one or the other of the two terms is particularly zero. The rst term is important
193
Figure 5.3:
when . - .
nn
and the second term is important when . - .
nn
. For the present, we
specialize to a situation in which the initial state : is a discrete bound state and the nal
state : is one of a continuous set of dissociated states. The rst order probability of nding
the system in the state |:i after the perturbation is removed is
c
(1)
n
(t t
0
)
2
=
4 |h:| \ |:i|
2
~
2

sin
2
[(.
nn
.) t
0
,2]
(.
nn
.)
2
. (5.19)
5.7.4 The Golden Rule
The factor sin
2
[(.
nn
.) t
0
,2] , (.
nn
.)
2
is plotted in Figure. the height of the main
peak increases in proportional to t
2
0
. Thus if there is a group of states |:i that have energy
nearly equal to 1:+~., and for which |h:| \ |:i| is roughly independent of :, the probability
of nding a system in one or another of these states is proportional to t
0
.
The transition probability per unit time is given by integrating Eq. (5.19) over
194
: and dividing by t
0
n =
1
t
0
Z

c
(1)
n
(t t
0
)
2
j(:)d1
n
where j(:)d1
n
is the number of nal states with energies between 1
n
and 1
n
+ d1
n
. the
concept of an energy density j(:) of nal state is sensible, since we are considering the case
in which the transition is to one or another of a continuous set of dissociated states. We
now take advantage of the fact that the breadth of the main peak in Sincr becomes small as
t
0
becomes large, and we regard |h:| \ |:i| and j(:) as quantities su!ciently independent
of En so that they can be taken outside in the integral. The probability is approximately
written as
n =
2
~
j(:) |h:| \ |:i|
2
where we have made use of the result
Z
+4
4
sin
2
r
r
2
dr = .
This formula, rst obtained by P. A. Dirac, was later dubbed by E. Fermi The Golden
Rule of perturbation theory.
195
Chapter 6
Collision Theory
Problems for which the energy eigenvalues are continuously distributed usually arise in
connection with the collision of a particle with a force eld. In a collision problem the
energy is specied in advance, the behavior of the wave function at great distance is found
in terms of it. This asymptotic behaviors can then be related to the amount of scattering
of the particle by the force. There are so few systems of physical interest for which exact
solutions can be found that approximation methods play an important part in applications
of the theory. Various methods that are useful in scattering problems are considered in this
chapter.
196
Figure 6.1:
6.1 Collisions in one- and three-dimensions
6.1.1 One-dimensional square potential barriers
We consider rst the one-dimensional collision of a particle with the square potential barrier.
In this problem we are interested in a particle that approaches from the region of negative
x and is reected or transmitted by the carrier. In the corresponding classical problem, the
particle is always reected if its energy is less than that of the top of the barrier, and it is
always transmitted if the energy is great. We shall see that, in the quantum problem, both
reection and transmission occur with nite probability for most energies of the particle.
Asymptotic behaviors
We are interested in representing a particle that approaches from the left with
energy 1 0 and may be turned back by the potential barrier or penetrate through it.
Thus the asymptotic behavior in the region where \ (r) = 0 is as follows: for r < 0 we want
the wave function to represent a particle moving to the left (reected particle) as well as
to the right (incident particle): for r a, we want the wave function to represent only a
197
particle moving to the right (transmitted particle).
The wave function in the region where \ (r) = 0 is
~
2
2:
0
2
0r
2
n = 1n.
Our asymptotic solution are
n(r) = c
I|r
+1c
I|r
, r < 0
n(r) = Cc
I|r
, r a
where
/ = j,~ =
2:1
~
2
1/2
Normalization
Recall the denition of the possibility current density
J =
~
2i:
[
\(\
) ] .
The possibility current densities at r < 0 and r a are
J(r) =
~/
:
||
2
|1|
2
, r < 0
J(r) =
~/
:
|C|
2
, r a.
The two currents are independent of the position and should be equal due to the conversation
of the practice.
Scattering coe!cients
The character of the solution inside the potential barrier depends on whether 1
is greater or less than the potential barrier \
0
.
198
The case 1 \
0
The solution in the potential barrier is
n(r) = 1c
Ior
+Gc
Ior
, 0 < r < a
where
c =
2:(1 \
0
)
~
2
1/2
.
The continuity of n and 0n,0r at r = 0 and r = a required by the boundary condition
provides four relations between the ve coe!cients. At r = 0,
+1 = 1 +G
/(1) = c(1 G)
At r = a,
1c
Ioo
+Gc
Ioo
= Cc
I|o
c
1c
Ioo
+Gc
Ioo
= /Cc
I|o
We can eliminate 1 and G and solve for the ratios 1, and C,.
1
/
2
c
2

1 c
2Ioo
(/ +c)
2
(/ c)
2
c
2Ioo
C
=
4/c
1 c
I(o|)o
(/ +c)
2
(/ c)
2
c
2Ioo
The reection coe!cient is
2
=
1 +
41(1 \
0
)
\
2
0
sin
2
ca
1
.
The transmission coe!cient is
2
=
1 +
\
2
0
sin
2
ca
41(1 \
0
)
1
.
199
Figure 6.2: The transmission coe!cient of a square barrier.
The case 1 < \
0
The reection and transmission coe!cients for 0 < 1 < \
0
are obtained by replace c by i,
in the wave function in the potential barrier.
, =
2:(\
0
1)
~
2
1/2
.
The reection coe!cient is
2
=
1 +
41(\
0
1)
\
2
0
sinh
2
ca
1
.
The transmission coe!cient is
2
=
1 +
\
2
0
sinh
2
ca
41(\
0
1)
1
.
6.2 Collision in three dimensions
Scattering cross section
200
The angular distribution of particles scattered by a xed center of force or by
other particles is conveniently described in terms of a scattering cross section. Suppose that
we bombard a group of : particles or scattering centers with a parallel ux of particles per
unit time and count the number of incident particles that emerge per unit time in a small
solid angle 4.
0
centered about a direction that has polar angles 0
0
and c
0
with respect to
the bombarding direction as polar axis. This number will be proportional to , : and 4.
0
provided that the ux is small enough so that there is no interference between bombarded
particles and no appreciable diminution of the bombarded particles by their recoil out of
the target region, and provided also that the bombarded particles are far enough apart so
that each collision process involves only one of them.
The dierent scattering cross section o
0
(0
0,
c
0
): the number of incident particles
that emerge per unit time in 4.
0
can be written
:o
0
(0
0,
c
0
)4.
0
.
The total scattering cross section:
o
0
=
Z
o
0
(0
0,
c
0
)d.
0
.
For a collision of a particle with a xed scattering center, the denition of the dierential
scattering cross section is equally valid in the laboratory and center-of-mass coordinate sys-
tems, since a scattering center that is xed has an innite eective mass and so that the
center of mass of the system does not move. For a collision between two particles of nite
mass, however, the dierential cross section applies in general only to the laboratory coor-
dinate system and to the observation of the scattered incident particle. It does not describe
in the observation of the recoil bombarded particle in the laboratory system, although it is
of course possible to obtain a dierential cross section for the recoil particle from o
0
(0
0,
c
0
).
201
Figure 6.3:
202
Relation between angles in the laboratory and center-of-mass systems
the relation between the dierential cross section and angles in the laboratory
system and in the center-of-mass system can be found by translating the laboratory system
in the direction of the incident particle with su!cient speed to bring the center of mass to
rest. Assume that a particle of mass :
1
and initial speed strikes a particle of mass :
2
that is initially at rest. The center of mass moves with the speed
0
=
:
1
:
1
+:
2
.
Thus in the center-of-mass system the speeds of two particles are
00
=
0
and
0
. If the
collision is elastic, they evidently recede from the center of mass after the collision with the
same speed. Thus 0
0,
c
0
and 0
,
c are related by
00
cos 0 +
0
=
1
cos 0
0
00
sin0 =
1
sin0
0
c = c
0
From these Eqs., we obtain
tan0
0
=
sin0
+ cos 0
=

0
00
=
:
1
:
2
Relation between cross sections
The relation between the cross section in the laboratory and center-of-mass co-
ordinate systems can be obtained from the denitions, which imply that the same number
of particles are scattered into dierential solid angle d.
0
about 0
0
, c
0
as are scattered into
203
d. about 0, c.
o
0
(0
0
, c
0
) sin0
0
d0
0
dc
0
= o(0, c) sin0d0dc.
With the help of the relation 0
0
, c
0
and 0, c, we obtain
o
0
(0
0
, c
0
) =
1 +
2
+ 2 cos 0
1/2
|1 + cos 0|
o(0, c).
It should be noted that the total cross section is the same for both laboratory and center-
of-mass systems and also for both the outgoing particles, since the total number of collision
that takes place is independent of the mode of description of the process.
Dependence on
For < 1, 0
0
increase from 0 to as 0 increase from 0 to . For = 1, 0
0
= 0,2
and varies from 0 to ,2 as 0 varies from 0 to ; in this case
o
0
(0
0
, c
0
) = 4 cos 0
0
o(20
0
, c
0
)
and no particle appear in the backward hemisphere in the laboratory system. For 1,
0
0
rst increase from 0 to a maximum value cos
1
(1,), which is less than ,2, as 0
increases from 0 to cos
1
(-1/); 0
0
then decreases to 0 as 0 increases further to . In this
case o
0
(0
0
, c
0
) is usually innite at the maximum value of 0
0
, although this singularity gives
a nite contribution to the total cross section; no particles appear beyond the maximum 0
0
in the laboratory system.
6.3 Scattering by Spherically Symmetric Po-
tentials
In this section, we assume that the potential \ is a function only of r, and we nd the
connection between the solution separated in spherical polar coordinate and the asymptotic
204
form.
Asymptotic behaviors
The dierential scattering cross section in the center-of-mass coordinate system
can be found from
~
2
2j
\
2
+\
n = 1n
where j = :
1
:
2
,(:
1
+:
2
). The scattering is determined by the asymptotic form of n in
the region where \ = 0
n = [c
I|:
+
1
r
)(0, c)c
I|:
] (6.1)
where / = j,~ and is the speed of the incident particle. The probability current density
J is given by
J(r) =
~
2ji
(\)
The gradient operator can be expressed in polar coordinates as

\ =
0
0r
r+
1
r
0
00
+
1
r sin0
0
0c
!.
Since the second and third terms are small when r is large, the current for large r is in the
radial direction and
J
:
=
~/
j
|)(0, c)|
2
,r
2
The number of scattered particles entering the solid angle (or detector) per unit time is
d. = J
:
r
2
d.
=
~/
j
|)(0, c)|
2
d..
From the denition of the cross section, it follows
o (0, c) = |)(0, c)|
2
.
205
Dierential cross section
The asymptotic behavior of the wave function determines the dierential scatter-
ing cross section but cannot itself be found without solving the wave equation throughout
all space. In the spherical polar coordinate,
~
2
2j
1
r
2
0
0r
r
2
0
0r
+
1
r
2
sin0
0
00
sin0
0
00
+
1
r
2
sin
2
0
0
2
0
2
c
n = (1 \ )n.
The radial and angular parts of the solution can be separated by taking
n(r, 0, c) = 1(r)Y
ln
(0, c)
where
Y
ln
(0, c) =
ln
1
n
l
(cos 0)c
In
is the spherical harmonic. The problem now possesses symmetry about the polar axis, so
that n, ), and o are independent of the angle c. That is, we just consider the case of : = 0.
In the case the P
n=0
l
is reduced to the Legendre polynomials. Thus, the general solution
can then be written as a sum of products of radial function and Legendre polynomials,
n =
4
X
l=0
(2| + 1)i
l
1
l
(r)1
l
(cos 0)
=
4
X
l=0
(2| + 1)i
l
r
1
l
(r)1
l
(cos 0)
where 1
l
is the Legendre polynomial of order |, and
l
satises
d
2
l
d
2
r
+
/
2
l(r)
|(| + 1)
r
2
l
= 0
where
/ =
2j1
~
2
1/2
l(r) =
2j\ (r)
~
2
0, as r +
206
In the limit of r+, the solution
l
is one of the form
l
c
I|:
l
sin(/r +c
0
l
)
In some specialized problem, U(r) can be neglected for r greater than some distance a, and
a may be small enough so that the l term in not negligible. The general form for 1
l
(r) has
the form
1
l
(r) = c
Io
I
[cos c
l
,
l
(/r) sinc
l
:
l
(/r)]
c
Io
I
/r
sin(/r
|
2
+c
l
), r . (6.2)
Here ,
l
is the special Bessel function;
,
l
(j) =

2j
1/2
J
l+1/2
(j)
1
j
cos(j
| + 1
2
), j
j
l
(2| + 1)!!
, j 0
:
l
is the special Neumann function;
:
l
(j) = (1)
l+1

2j
1/2
J
l1/2
(j)
1
j
sin(j
| + 1
2
), j

(2| 1)!!
j
l+1
, j 0.
To identify the form of ), we require an expansion of c
I|: cos 0
in Legendre polynomials;
c
I|: cos 0
=
4
X
l=0
(2| + 1)i
I
,
l
(/r)1
l
(cos 0).
207
Comparison of Eq.(6.1) with the general form 1
l
(r) gives
c
I|:
+r
1
)(0, c)c
I|:
=
4
X
l=0
(2| + 1)i
I
,
l
(/r)1
l
(cos 0) +r
1
)(0, c)c
I|:
=
4
X
l=0
(2| + 1)i
I
l
(/r)
1
sin(/r
|
2
+c
l
)1
l
(cos 0).
l
are the coe!cient for dierent |, and should take the form
l
= c
Io
I
.
Thus it gives
)(0) =
1
2i/
4
X
l=0
(2| + 1)(c
2Io
I
1)1
l
(cos 0).
Thus the dierential cross section is
o(0) = |)(0)|
2
=
1
/
2
4
X
l=0
(2| + 1)(c
2Io
I
1)1
l
(cos 0)
2
Total elastic cross section
The total elastic cross section is given by
o = 2
Z
t
0
o(0) sin0d0
=
4
/
2
4
X
l=0
(2| + 1) sin
2
c
l
.
Here we make use of the orthogonality property of the Legendre polynomials
Z
+1
1
d cos 01
l
(cos 0)1
l
0 (cos 0) =
2
2| + 1
c
ll
0 .
208
The total cross section can also be related to )(0). it follows from the generating function
for the Legendre polynomials that 1(|) = 1 for all |, so that
)(0) =
1
i2/
4
X
l=0
(2| + 1)
c
2Io
I
1
Comparison with the total cross section shows

o =
2
i/
[)(0) )
(0)] =
4
/
Im)(0).
This relation is known as the optical theorem.
The physical interpretation of the optical theorem is as follows: In order for
scattering to take place, particles must be removed in an amount proportional to o from
the incident beam, so that its intensity is smaller behind the scattering region (0 - 0) than
in front of it. This can occur only by the interference between two terms in the asymptotic
expression of the wave function.
Phase shifts
The angle c| is called the phase shift of the lth partial wave, since according
to Eq.(6.2) it is the dierence in phase between the asymptotic forms of the actual radial
function 1
l
(/r) and the radial function ,
l
(/r) in the absence of the scattering potential. the
phase shifts completely determine the scattering, and the scattering cross section vanishes
when each of c
l
is 0
o
or 180
o
.
6.4 Applications
6.4.1 Scattering by a square well
In general, for a given potential, phase shifts are calculated from a numerical solution of the
radial equations. In a few cases an analytical solution is possible, in particular for scattering
209
by an attractive square well, for which the reduced potential l(r) is
l(r) =
;
A
A
?
A
A
=
l
0
(< 0), r < a;
0, r a.
Inside the well (r < a) the radial equation is
d
2
dr
2
+
2
r
d
dr

|(| + 1)
r
2
+1
2
1
l
(1r) = 0
where we set 1
2
= /
2
+l
0
. The regular solution of this equation is
1
1
l
(1r) =
l
,
l
(1r), r < a.
In the exterior region, the radial equation is
d
2
dr
2
+
2
r
d
dr

|(| + 1)
r
2
+/
2
1
l
(/r) = 0
The general solution is
1
J
l
(/r) = 1
l
[,
l
(/r) tanc
l
:
l
(/r)] , r a. (6.3)
The solutions for r < a and r a can be joined smoothly at r = a by requiring that
1
1
l
(1a) = 1
J
l
(/a);
d1
1
l
(1r)
dr
:=o
=
d1
J
l
(/r)
dr
:=o
Eliminating the normalization constants gives
tanc
l
(/) =
/,
0
l
(/a),
l
(1a) 1,
l
(/a),
0
l
(1a)
/:
0
l
(/a),
l
(1a) 1:
l
(/a),
0
l
(1a)
.
At low energies, the scattering is dominated by the | = 0 wave. Using the formula
,
0
(j) =
sinj
j
:
0
(j) =
cos j
j
210
we have
tanc
0
(/) =
/ tan1a 1 tan/a
1 +/ tan/a tan1a
.
In the low-energy limit, /a 1,
c
0
(/) = /a
1
tan1a
1a
Then the | = 0 partial cross section is

o
l=0
=
4
/
2
sin
2
c
0
- 4a
2
1
tan1a
1a
2
6.4.2 Scattering by a hard-sphere potential
Another simple, but interesting example is the hard-sphere potential
l(r) =
;
A
A
?
A
A
=
+, r < a;
0, r a.
Since the scattered particle cannot penetrate into the region r < a, the wave function in
the exterior region must vanish at r = a. Since the scattered particle cannot penetrate into
the region r < a, the wave function in the exterior region, given by Eq.(6.3), must vanish at
r = a, from which
tanc
l
(/) =
,
l
(/a)
:
l
(/a)
.
In the low-energy limit, /a 1,
tanc
l
(/) =
(/a)
l
(2| + 1)!!
,
(2| 1)!!
(/a)
l+1
=
(/a)
2l+1
(2| + 1)!!(2| 1)!!
211
Hence |tanc
l
(/)| quickly decreases as | increases. As a result, the low-energy scattering is
always dominated by the s-wave. Since ,
0
(j) = sinj,j and :
0
(j) = cos j,j, we have
c
0
= /a.
So the cross section is
c
0
= 4a
2
.
At high energies (/a 0), we nd
c
l
=
|
2
/a
so that
c
|o|
-
4
/
2
l
max
X
l=0
(2| + 1) sin
2
(
2
2
/a)
- 2a
2
.
It is interesting to compare this result with that obtained from classical mechanics.
6.5 Approximate Collision Theory
See Sakurais book: section 7.1, 7.2, and 7.3
6.5.1 The Lippman-Schwinger Equation
Let us begin with the time-independent formulation of scattering processes. The Hamilto-
nian is written as
H = H
0
+\ =
1
2
2:
+\.
212
H
0
describes the particle is far away from the scattering center. We have two Schrodinger
equations in the absence and presence of V:
H
0
|ci = 1|ci
(H
0
+\ ) |i = 1|i
When \ 0, |i |ci . We assume the two equations have the same energy eigenvalue.
Combining the two equations
(1 H
0
) |i = \ |i + (H
0
1) |ci
If (1 H
0
)
1
exist, we have
|i = (1 H
0
)
1
\ |i +|ci
Unfortunately, the operator (1 H
0
)
1
is singular. To avoid the di!culty, we introduce an
innitesmal complex i- (- 0
+
) such that 1 1 +i-. In this way the equation becomes
(+)
E
= |ci +
1
1 H
0
+i-
\
(+)
E
(6.4)
In the position space, Eq.(6.4) is reduced to
hx
(+)
E
= hx |ci +
Z
dx
0
hx|
1
1 H
0
+i-
|x
0
i hx
0
| \
(+)
E
= hx |ci +
2:
~
2
Z
dx
0
G(x, x
0
) hx
0
| \
(+)
E
213
We evaluate the kernel of integral G(x, x
0
).
G(x, x
0
) =
~
2
2:
hx|
1
1 H
0
+i-
|x
0
i
=
~
2
2:
Z
dphx| pi hp|
1
1 H
0
+i-
|pi hp |x
0
i
=
~
2
2:
Z
dphx| pi hp|
1
1 j
2
,2:+i-
|pi hp |x
0
i
=
~
2
2:
Z
dphx| pi
1
1 j
2
,2:+i-
hp |x
0
i
=
~
2
2:
Z
dp
(2~)
3
exp[ip (x x
0
),~]
1 j
2
,2:+i-
where we have used
hx| pi = (2~)
3/2
exp[ip x,~]
hp| xi = (2~)
3/2
exp[ip x,~].
Take 1 = ~
2
/
2
,2: and j = ~,
G(x, x
0
) =
Z
dq
(2)
3
exp[iq (x x
0
)]
/
2
2
+i-
=
Z
d
2
dcsin0d0
(2)
3
exp[i |x x
0
| cos 0]
/
2
2
+i-
=
i
4
2
|x x
0
|
Z
+4
0
d
exp[i |x x
0
|] exp[i |x x
0
|]
/
2
2
+i-
=
1
4
exp[i/ |r r
0
|]
|r r
0
|
In the last step we used the residue theorem:
I
d.
. .
0
= 2i
I
c
)(.)d. = 2i
X
a
1:
1
= 2i(sum of enclosed residues).
214
Figure 6.4:
The equation becomes
hx
(+)
E
= hx |ci
2:
~
2
Z
dr
0
c
+I|
|
xx
0
|
|x x
0
|
hx
0
| \
(+)
E
.
Assume that V is a local potential,
hx
0
| \ |x
00
i = \ (r
0
)c(x
0
x
00
)
hx
(+)
E
= hx |ci
2:
~
2
Z
dr
0
c
+I||xx
0
|
|x x
0
|
\ (r
0
) hx
0
(+)
E
.
As the observation point is far away from the scatterer,
|x| = r |x
0
| = r
0
|x x
0
| = (r
2
2rr
0
cos c +r
02
)
1/2
= r(1
r
0
r
cos c +....)
From this approximation, we have
c
+I|
|
xx
0
|
|x x
0
|
-
c
I|:
r
c
Ikr
0
.
215
So,
hx
(+)
E
hx |ci
2:
~
2
c
I|:
r
Z
dx0c
Ikr
0
\ (r
0
) hx
0
(+)
E
.
Comparing with the asymptotic form of the wave function,
hx
()
E
1
(2)
3/2
c
Ikr
+
c
I|:
r
)(k
0
, k)
We have
)(/
0
, /) =
2:
~
2
Z
dx
0
exp[i/r
0
]
(2)
3/2
\ (r
0
) hx
0
|
(+)
E
6.5.2 The Born Approximation
The formula contains the unknown wave function. We have to introduce the approximation.
(+)
E
= |ci +
1
1 H
0
+i-
\
(+)
E
= |ci +
1
1 H
0
+i-
\
|ci +
1
1 H
0
+i-
\
(+)
E
= |ci +
1
1 H
0
+i-
\ |ci +.....
Assume the eect of the scattering is not very strong, we replace hx
0
|
(+)
by hx
0
| ci in
the integral,
hx
0
|
(+)
E
hx
0
| ci = exp[ik x
0
].
This is the rst-order Born approximation. The amplitude is
)(/
0
, /) =
1
4
2:
~
2
Z
dx
0
c
I
(
kk
0
)
x
0
\ (x
0
)
which is the Fourier transform of the potential except for a constant.
For a spherical symmetric potential, assume

k k
0
= = 2/ sin0,2. (/ = /0 as
216
the energy is conserved.)
) =
1
4
2:
~
2
Z
2t
0
drr
2
Z
2t
0
dc
Z
t
0
sin0d0c
Ij: cos 0
\ (r)
=
2:
~
2
Z
+4
0
drr\ (r) sinr
6.5.3 Application: from Yukawa potential to Coloumb
potential
As an illustration of the Born approximation, we consider the Yukawa potential
l(r) = l
0
c
:
jr
)(k
0
, k) =
Z
drl(r)
r sinr
= l
0
1
2
+j
2
where
= |k
0
k|
Z
drr\ (r) sinr =
l
0
j
Z
+4
0
drc
:
sinr
=
l
0
j
Z
+4
0
drc
:
c
Ij:
c
Ij:
2i
=
l
0
i2j
1
j i

1
j +i
=
l
0
2
+j
2
.
So the dierential cross section is
o(0) =
2:l
0
j~
2
1
[2/
2
(1 cos 0) +j
2
]
.
217
When c = 0, the Yukawa potential is reduced to the Coloumb potential, l
0
,j 77
0
c
2
.
We obtain
o(0) =
2:77
0
c
2
2
~
4
1
16/
4
sin
4 0
2
=
77
0
c
2
1
2
1
16/
4
sin
4 0
2
This is the Rutherford scattering cross section. From Rutherford formula we conclude that
the charge-charge interaction in atomic scale obey Coulomb law, i.e., 1,r.
6.5.4 Identical Particles and Scattering
Collisions between identical particles are particularly interesting as a direct illustration of
the fundamental dierences between classical and quantum mechanics.
Classical case
The dierential cross-section is
o(0, o) = |)(0, c)|
2
+|)( 0, c +)|
2
Scattering of two identical spinless bosons
o(0, o) = |)(0, c) +)( 0, c +)|
2
Scattering of two identical spin-1/2 fermions
o(0, o) =
1
4
|)(0, c) +)( 0, c +)|
2
+
3
4
|)(0, c) )( 0, c +)|
2
218
6.6 Landau-Zener Problem
This two level problem can be introduced in this course.
219
Chapter 7
Selected Topics
7.1 Quantum Statistics
Quantum statistical mechanics is the branch of physics dealing with systems in mixed states;
it is the quantum analogue of classical statistical mechanics. It should be noted that statis-
tics enters at two levels in quantum statistical mechanics: rst, because of the statistical
interpretation of the wave function and second, because of our incomplete knowledge of
dynamical state of the system.
7.1.1 Density Operator and Ensembles
Question: How to describe quantum mechanically an ensemble of physical systems for which,
say .
o
=60 are characterized by |ci and the remaining .
o
=40 are characterized by some
other ket |,i .
220
Example: a general state for S=1/2 system
|ci = c
+
|+i +c
|i
which characterizes a state ket whose spin is pointing some denite direction. We should
not confuse the probability weight of the states |+i and |i , .
, and |c
|
2
.
Ensemble average and density operator.
Pure ensemble: |ci
Mixed ensembles:
.
1
c
(1)
.
2
c
(2)
.
3
c
(3)
.
.
.
.
.
.
with the normalization condition
X
I
.
I
= 1.
The ensemble average
hi =
X
I
.
I
hc
I
|| c
I
i
=
X
I
.
I
X
n
hc
I
| :i h:| | c
I
i
=
X
I,n
.
I
|h:|c
I
i|
2
a
n
Alternatively,
hi =
X
I,n
.
I
h:|c
I
i hc
I
|:i h:|| :i
=
X
n
h:|
X
I
{.
I
|c
I
i hc
I
| } |:i
= Tr(j)
221
where
j =
X
I
.
I
|c
I
i hc
I
| .
j is the density matrix operator and is independent of representation.
Two properties:
(a). j is Hermitian;
(b). j satises the normalization condition.
Tr(j) = 1.
A pure ensemble: .
n
= 1 for some |c
n
i
j = |c
n
i hc
n
|
which is just a projection operator.
Several examples of density operator:
(1) A completely polarized beam with S
:
+
j = |+i h+| =
3
E
E
C
1
0
4
F
F
D
1 0
=
3
E
E
C
1 0
0 0
4
F
F
D
(2) A completely polarized beam with S
r
+
j = |o
r
, +i ho
r
, +|
=
1
2
1/2
(|+i +|i)
1
2
1/2
(h+| +h|)
=
3
E
E
C
1
2
1
2
1
2
1
2
4
F
F
D
222
(3) A partially polarized beam with 75 S
:
+ and 25 S
r
+
j = 0.75 |+i h+| + 0.25 |o
r
, +i ho
r
, +|
=
1
8
3
E
E
C
7 1
1 1
4
F
F
D
7.1.2 Quantum Statistical Mechanism
Order and Disorder
The density matrix of a pure ensemble
j =
3
E
E
E
E
E
E
E
E
E
E
E
E
E
E
C
0
.
.
.
1
.
.
.
0
4
F
F
F
F
F
F
F
F
F
F
F
F
F
F
D
The density matrix of a completely random ensemble
j =
3
E
E
E
E
E
E
E
E
E
E
C
1
.
.
.
1
4
F
F
F
F
F
F
F
F
F
F
D
We dene a quantity
o = tr (j lnj)
=
X
n
j
nn
lnj
nn
For a pure ensemble
o = 0
223
For a completely random ensemble
o = ln
This quantity is related to the entropy in thermodynamics,
o = /tr (j lnj)
where k is the Boltzmann constant.
Basic assumption: Nature tends to maximize o subject to the constraint that
the ensemble average of the Hamiltonian has a certain value.
To minimize o:
co = 0
i.e.,
cj
nn
(lnj
nn
+ 1) = 0
with the conditions
(a). hHi = n = tr(jH)
(b). trj = 1
we obtain
cj
nn
[(lnj
nn
+ 1) +,1
n
+] = 0
j
nn
= c
oJ
!
~
=
1
P
n
c
oJ
r
c
oJ
!
The partition function:
7 =
X
n
c
oJ
r
= tr(c
o1
)
224
Thus the density matrix operator is
j =
1
7
c
o1
The parameter , is related to the temperature T as follows
, = 1,/T
Example: Spin-1/2 particle in a magnetic eld
Consider a spin-1/2 particle having a magnetic momentum
j = qj
1
o,~
subjected to a constant magnetic eld along the z-axis.
H = qj
1
o
:
,2 = .o
:
The density operator
j =
1
7
c
o1
=
3
E
E
C
1
2
c
o.
0
0
1
2
c
+o.
4
F
F
D
where
7 = c
o.
+c
o.
The average value of the z-component of spin at the temperature T is
ho
:
i =
~
2
c
o.
c
o.
c
o.
+c
o.
=
~
2
tanh(,.)
225
7.1.3 Quantum Statistics
Consider a system with two identical particles. Each particle has three non-degenerate
states: |1i , |2i , |3i with E
1
, E
2
, E
3
. For classical identical particles: there 9 possible
congurations
|1, 1i |1, 2i |1, 3i
|2, 1i |2, 2i |2, 3i
|3, 1i |3, 2i |3, 3i
The wave function is
=
J
1
(1)
J
(2)
For bosons: there are 6 possible congurations
|1, 1i , |2, 2i , |3, 3i
(|1, 2i +|2, 1i) ,2
1/2
(|1, 3i +|3, 1i) ,2
1/2
(|3, 2i +|2, 3i) ,2
1/2
The wave function
=
J
1
(1)
J
(2) +
J
(1)
J
1
(2)
For fermions: there are 3 possible congurations
(|1, 2i |2, 1i) ,2
1/2
(|1, 3i |3, 1i) ,2
1/2
(|3, 2i |2, 3i) ,2
1/2
226
The wave function
=
J
1
(1)
J
(2)
J
(1)
J
1
(2)
7.1.4 Systems of non-interaction particles
We shall now discuss the properties of systems of large numbers of non-interacting objects
which are equivalent and possess the same energy levels.
Maxwell-Boltzmann Statistics
(Classical identical particles)
If E
(j=1, 2, ) denotes an energy level of one of the particles, the total energy
of the system E can be written
1 =
X
where the sum runs over all the eigenenergies of a single particle, and where n
is the number
of particles with the energy E
. The total number of particles is

=
X
I
:
.
In a Maxwell-Boltzmann system, any number of particles can be in each level, so that nj can
either be zero or any positive integer. If all the energies Ej are dierent, each permutation of
the particles results in a dierent wave function, thus each energy level of the total system
is N!-fold degenerate. However if nj1, the interchange between these particles do not alter
the wave function. Thus the number of distinct states corresponding to a given value of the
total energy of the system is
q
11
J
=
!
Q
I
:
I
!
227
The partition function can be expressed as
7 =
X
{n
}
q
11
J
exp(,
X
)
=
X
{n
}
!
Q
I
:
I
!
exp(,
X
)
=
5
7
X
exp(,:
)
6
8
2
The average distribution is
h:
I
i =
1
7
X
{n
}
q
11
J
exp(,
X
):
=
0 ln7
0,1
I
=

7
c
oJ
1
= c
oJ
1
o
Bose-Einstein Statistics (for bosons)
The wave function of many bosons is symmetric and the energy level of the system is non-
degenerate
q
1J
J
= 1, for any :
Here we use the grand canonical ensemble ( +)

h:
I
i =
1
7
X
{n
}
q
1J
J
exp(,
X
c
X
):
228
where
7 =
X
{n
}
exp(,
X
c
X
)
=
Y
3
C
X
{n
}
exp(,:
c:
)
4
D
=
Y
(1 exp(,1
c))
1
In the case,
h:
i =
0 ln7
0,1
I
=
1
exp(,1
+c) 1
Fermi-Dirac Statistics (for fermions)
The wave function of many bosons is antisymmetric, and the energy level of the system is
non-degenerate
q
J1
J
= 1, for :
= 1 or 0
= 0, otherwise.
Here we use the grand canonical ensemble ( +)
h:
I
i =
1
7
X
{n
}
q
J1
J
exp(,
X
c
X
):
229
where
7 =
X
{n
}
q
J1
J
exp(,
X
c
X
)
=
Y
3
C
1
X
n
=0
exp(,:
c:
)
4
D
=
Y
(1 + exp(,1
c))
In the case,
h:
i =
0 ln7
0,1
I
=
1
exp(,1
+c) + 1
7.1.5 Bose-Einstein Condensation
One of the main features of boson system is that bosons tend to occupy the lowest energy
state at low temperatures. At high temperatures, both distribution laws for bosons and for
fermions become equal to that for the Maxwell-Boltzmann case approximately. The density
of particles on a certain state always tends to be zero. For a boson system, as
h:
i =
1
exp(,1
+c) 1
+
if both E and c are equal to zero, the density of the particle at E=0 can be nonzero in the
high density limit at low temperatures.
Consider a non-interacting boson gas in a three-dimensional system. The spec-
trum of energy is
1 =
~
2
2:
/
2
230
The density of particle is
j = j
0
+
1
(2)
3
Z
d/
3
:(1)
where j0 is the density of particles in the lowest energy state.
1
(2)
3
Z
d/
3
:(1) =
Z
d11(1):(1)
where
1(1) =
1
4
2
2:
~
2
3/2
1
1/2
This is the density of state at E.
j j
0
=
1
4
2
2:
~
2
3/2
Z
+4
0
d1
1
1/2
c
oJ+o
1
=
1
4
2
2:/T
~
2
3/2
)
1/2
(c)
where
)
1/2
(c) =
Z
+4
0
dr
r
1/2
c
r+o
1
.
Since c _ 0, the largest possible values of )
1/2
(c) occurs when c = 0 and numerically
)
1/2
(c = 0) = 2.315.
It follows;; that at T < T
0
where
T
0
=
~
2
2:/
4
2
j
2.315
2/3
the function (j j
0
),j becomes less than unit, and the system increasingly condensed into
the lowest energy state.
Bose-Einstein condensation was predicted by Albert Einstein and Satyendra Nath
Bose at 1924 to 25. It was rst observed in Helium-4 liquid. Recent years it was observed in
some systems with cooling atoms. It is one of the most active branches in condensed matter
physics. If you are interested in the recent development in this eld, please visit Web site:
231
http://www.aip.org/pt/webwatch/ww9703.html.
7.1.6 Free fermion gas
The free particle Schrodinger equation in three dimensions is
~
2
2:
0
2
0r
2
+
0
2
0j
2
+
0
2
0r
:2
|
(r) = c
|
|
(r)
If the particle are conned to a cube of edge L and we require the wave function to be
periodic in x, y, z direction with L, i.e.,
|
(r +1, j, .) =
|
(r +1, j, .)
|
(r, j +1, .) =
|
(r, j +1, .)
|
(r, j, . +1) =
|
(r, j, . +1)
the wave functions are
|
(r) =
1
(2)
3/2
c
I|:
provided that the components of the wave vector k satises
/
r,,:
= 0,
2
1
,
4
1
,
6
1
,
Any component of k is of the form 2n,1, where n is a positive integer. The energy spectrum
with n is
c
|
=
~
2
2:
/
2
r
+/
2
+/
2
:
As two indistinguishable fermions cannot have all their quantum numbers identical, each
energy level (or state) can be occupied by at most one particle. Thus we start lling the
energy level from that with the lowest energy (i.e., k=0 here) until all N particles are
accommodated.
232
The occupied energy levels may be represented as points inside a sphere in the
k-space. The energy at the surface of the sphere is the Fermi energy; the wave vectors at
the Fermi surface have a magnitude k
J
such that
c
J
=
~
2
2:
/
2
J
.
The total number of energy levels inside the sphere is
=
1
(2,1)
3
4
3
/
3
J
Then
/
J
=
6
2
,\
1/3
.
In short, a boson gas condensates into the lowest energy state, and a fermion gas
forms a Fermi sphere in the k-space at absolute zero temperature.
7.2 Quantum Hall Eect
The quantum Hall eect is a remarkable phenomenon discovered experimentally in 1980s
in which the Hall conductivity of a two-dimensional system of electrons is found to have
plateaus as a function of variables which determine the number of electrons participating
in the eect. The integer quantum Hall eect was observed by von Klitzing et al in 1980.
Simple theory suggests that the Hall conductivity at he plateaus should be an integral
multiple of c
2
,~ and the experiments agree with that prediction to within an accuracy of
nearly 0.1 ppm. An application of great importance is to metrology, the quantum Hall eect
promises a method of providing very precise resistance standards that are insensitive to the
particular sample and the details of its fabrication. With a more powerful magnetic eld
and lower temperature, Daniel C. Tsui and Horst L. Stormer discovered that the plateaus
233
of the conductivity is a fractional multiples of the basic unit c
2
,~. Within a year of their
discovery, Rovert B. Laughlin has succeeded in explaining their result. Through theoretical
analysis he showed that the electrons in a powerful magnetic eld can condense to form a
kind of quantum uid related to the quantum uids that occur in superconductivity and in
liquid helium.
What makes these uids particularly important for researchers is that events in
a drop of quantum uid can aord more profound insights into the general inner structure
and dynamics of matter. The Royal Swedish Academy of Sciences awarded The 1998 Nobel
Prize in Physics jointly to Tsui, Stomer and Laughlin for their discovery of a new form of
quantum uid with fractionally charged excitations.
7.2.1 Hall Eect
The Hall eect occurs when the charge carriers moving through a material experience a
deection because of an applied magnetic eld. This deection results in a measurable
potential dierence across the side of the material which is transverse to the magnetic eld
and current direction.
234
In Figure, a voltage V drives a current I in the positive X direction. Normal
Ohmic resistance is V/I . A magnetic eld tin the positive z direction shifts positive charge
carriers in the negative y direction. This generates a Hall potential (\
1
,1) in the y direction.
Quantitatively, Hall eect indicates that a transverse electric eld will be pro-
duced. Charges will accumulate on the transverse edges until a strong enough electric eld
is developed to force remaining charges to continue undeected. The condition is
1 =
\
J
c
1.
The drift velocity is given by
\
J
= J,:
t
=
i,
:c
where n is the density of charge carrier. The electric eld is
1 = \
1
,d
where l is the thickness of material and d the width. Hence we obtain an expression for the
density of charge carriers in a substance
: =
i
c

1
J
=
i1
\
1
|cc
Therefore this eect can be used to determine the density of charge carriers in conductors
and semi-conductors, and has become a standard tool in physics laboratories over all the
world. The Hall conductivity
o = i,1
=
:cc
1

:
1
235
7.2.2 Quantum Hall Eect
In 1980 the German physicist Klaus von Klitzing discovered in a similar experiment that
the Hall resistance does not vary in linear fashion, but step-wise with the strength of the
magnetic eld. The steps occur at resistance values that do not depend on the properties
of the material but are given by a combination of fundamental physical constants divided
by an integer. We say that the resistance is quantized. At quantized Hall resistance values,
normal Ohmic resistance disappears and the material becomes in a sense superconducting.
For his discovery of what is termed the integer quantum Hall eect von Klitzing was awarded
the Nobel Prize in 1985.
236
In their rened experimental studies of the quantum Hall eect, using among
other things lower temperatures and more powerful magnetic eld, Stomer, Tsui and their
co-workers found to their great surprise a new step in the Hall resistance which was three
times higher than von Klitzings highest. They subsequently found more and more new
steps, both above and between the integers. All the new step heights can be expressed with
the same constant as earlier but now divided by dierent fractions. For this reason the new
discovery is named the fractional quantum Hall eect..
237
7.2.3 Laughlins Theory
A year after the discovery of the fractional quantum Hall eect, Laughlin oered a theoretical
explanation. According to his theory the low temperature and the powerful magnetic eld
compel the electron gas to condense to form a new type of quantum uid. Since electrons
are most reluctant to condense (They are what is termed fermions) they rst, in a sense,
238
combine with the ux quanta of magnetic eld. Particularly for the rst step ( = 1,3)
discovered by Stomer and Tsui, each of electrons captures three ux quanta thus forming a
kind of composite particle with no objection to condensing. (They become what is termed
bosons). Quantum uids have earlier occurred at very low temperatures in liquid helium and
in superconductors. They have certain properties in common, e.g. superuidity, but they
also show important dierences in behaviors. Apart from its superuidity which explains
the disappearance of Ohmic resistance at the Hall resistance steps the new quantum uid
proposed by Laughlin has many unusual properties. One of the most remarkable is that if
one electron is added the uid will be aected (excited) and a number of fractionally charged
quasiparticles created. These quasiparticles are not particles in the normal sense but a
result of the common dance of electrons in the quantum uid. Laughlin was the rst to
demonstrate that the quasiparticles have precisely the correct fractional charge to explain
the fractional quantum Hall eect. Subsequent measurements have demonstrated more and
more fractional charged steps in the Hall eect, and Laughlins quantum uid has proved
capable of explaining all the steps experimentally. The new quantum uid strongly resists
compression; it is said to be incompressible.
Further reading
1. B. Davis, Splitting the electron, New Scientist, 31 Jan, 1998, p36
2. G. P. Collins, Fractionally charged quasiparticles signal their presence with
noise, Physics Today, Nov,1997, p17,
3. S. Kivelson, D. H. Lee and S. C. Zhang, Electrons in atband, Scientic
American, March,1996, p64.
239
7.2.4 Charged particle in the presence of a magnetic
eld
We consider a charged particle moving on a plane in the presence of a perpendicular uniform
magnetic eld B. The mass of the particle is M and charge e. The Hamiltonian is
H =
1
2:
(1 +
c
c
)
2
=
1
2:
[(i~
0
0r
+
c
c
r
)
2
+ (i~
0
0j
+
c
c
)
2
]
The vector potential A, is such that its curl is equal to B:
1
:
= (\A)
:
We will work in the isotropic gauge
A =
1
2
Br
i.e.
r
=
1
2
1j
=
1
2
1r
In the gauge
H =
1
2:
(i~
0
0r

c
2c
1j)
2
+
1
2:
(i~
0
0j
+
c
2c
1r)
2
Denote
r
= i~
0
0r

c
2c
1j
= i~
0
0j
+
c
2c
1r
240
The commutator is constant,
[
r.
] = i
~c1
c
.
The Hamiltonian is expressed as
H =
1
2:
{(
i
r
)(
+
r
) +i[
r
,
]}
=
1
2:
[(
i
r
)(
+
r
) +
~c1
c
= ~
c1
'c
[a
+
a +
1
2
]
where
a
+
=
c
2~c1
1/2
(
i
r
)
=
c
2~c1
1/2
(i~
0
0r
+
c1
2c
ij i~
0
0j
+
c1
2c
r)
= 2
1/2
(
0
0 .
+
1
4
.)
a = 2
1/2
(+
0
0.
+
1
4
.)
where . = (r +ij),|
0
and . = (r ij),|
0
. The length unit |
0
= (~c,c1)
1/2
.
Hence the Hamiltonian is reduced to
H = ~.
c
(a
+
a +
1
2
)
where
.
c
=
c1
'c
.
a and a
+
satisfy [a, a
+
] = 1.
Comparing to the simple harmonic oscillator the motion of a charged particle
in a uniform magnetic eld is equivalent to it. A set of the wave function has the form
)
n
(., .) = .
n
c
: :/4
241
The energy eigenvalue is
1
n
= ~.
c
(: +
1
2
)
The wave functions are also eigenstates of angular momentum operator
1
:
= i~(r
0
0j
j
0
0r
)
= ~(.
0
0.
.
0
0 .
).
with the eigenvalue
1
:
)
n
(., .) = :~)
n
(., .).
The energy level 1
n
is called :
||
Landau level
The Landau levels have a huge degeneracy which is determined by
=
c
0
=
1
2
~c
t
=

2|
2
0
To make this degeneracy more apparent, we assume the system has the shape of a square
with dimension L. We introduce another pair of operators
/
+
= 2
1/2
(
0
0.

1
4
.)
/ = 2
1/2
(
0
0.

1
4
.)
([/, /
+
] = 1)
The two operators commute with H,
[/, H] = [/
+
, H] = 0.
Also , the operator / annihilates the wave function )
n
(., .), just like the operator a.
/)
n
(., .) = 0
242
Dene
T = exp[i2|
0
/,1]
T = exp[2|
0
/
,1]
Thus, )
n
is an engenstate of

T
T)
n
= )
n
A complete set of eigenstates of the :t/ Landau level {)
n,n
}can be constructed (: =
1, ...
)
)
n,n
(., .) = T
n
)
n
(., .)
= exp[i2:|
0
/,1])
n
(., .)
From the denitions of T and

T
TT = T

T exp[i2|
0
/,1, 2|
0
/
,1]
= exp[i2,
]T

T
Therefore the states )
n,n
(., .) have
H)
n,n
(z z) = ~.(: +
1
2
))
n,n
(., .)
T)
n,n
(z z) = exp[i2:,
])
n,n
(., .)
7.2.5 Landau Level and Quantum Hall Eect
The degeneracy of the Landau energy levels is
fold. The ratio of the number of electrons

to
=
t
,
243
is called the lling number of Landau level. When = 1,it means that the rst Landau
level is fully lled and when = 2, the second landau level is also lled. There is a simple
relation between the lling number and the quantized Hall conductivity. Suppose the total
charge of the sample is
Q =
c
The current J is then equal to
J =
ci
i is determined by the magnetic eld B and electric eld E
i
c
=
1
1
So the current density j = J,d1
, =

c
d1
i =
i
c
1d1
1c
=
i

~c
t
1c =
c
2
~
1.
The Hall conductance is
o
n
=
c
2
~
.
The integer quantum Hall eect occurs at = i:tcqcr, (1,2,...) and the fractional
Hall eect occurs at = )ractio:a|: 1/3, 1/5, 2/3, ...
7.3 Quantum Magnetism
Magnetism is inseparable from quantum mechanics, for a strictly classical system in thermal
equilibrium can display no magnetic momentum, even in a magnetic eld. The magnetic
momentum of a free atom has three principal sources: the spin with which electrons are
244
endowed, their orbital angular momentum about the nucleus; and the change in the orbital
moment induced by an applied magnetic eld.
Ordered arrangements of electron spins
7.3.1 Spin Exchange
Ferromagnetism is obtained in solids when the magnetic moments of many electrons align.
Antiferromagnetism and spin density waves describe oscillatory ordering of magnetic mo-
ments. The classical dipolar interaction between the electron moments (which is of order
10
5
c\ ) is for too weak to explain the observed magnetic transition temperature (which are
245
of order 10
2
10
3 0
/ in transition metal and rare earth compounds)
The coupling mechanism that gives rise to magnetism derives from the following
fundamental properties of electrons:
The electrons spin
The electrons kinetic energy
Pauli exclusion principle
Coulomb repulsion
Before we introduce the physical origin of the magnetic coupling between elec-
trons in solids, we simply review some standard denitions and basic relations of second
quantization.
For an orthonormal single-particle basis, {|c
I
i}
c
I
|c
= c
I
.
The creation operator of state i is a
I
and its Hermitian conjugate is annihilation operator
a
I
. Both are dened with respect to the vacuum state |0i
I
such that
|c
I
i = a
I
|0i
I
a
I
|0i
I
= 0.
The number operator is dened as
:
I
= a
I
a
I
For bosons,
h
a
I
, a
i
= a
I
a
a
I
= c
I
{a
I
, a
} = 0
246
For electrons with spin : = 1,2, we have to introduce a pair of operators, c
I,c
where o =,
such that
{c
Ic
, c
c
} = c
I
c
cc
{c
Ic
, c
c
} = 0
The spin operator can be expressed as
o
I
= o
Ir
+io
I
= ~c
I%
c
I&
o
I
= o
Ir
io
I
= ~c
I&
c
I%
o
:
I
=
~
2
(c
I%
c
I%
c
I&
c
I&
)
The commutation relations:
[o
I
, o
I
] = 2~o
:
I
[o
:
I
, o
I
] = ~o
I
7.3.2 Two-Site Problem
The Hersenberg spin exchange interaction is written as
H = JS
1
S
2
To obtain the eigenstates of H, we check the following relations,
(1) [S
2
1
, H] = 0,but [S
1
, H] 6= 0
(2) [S
2
2
, H] = 0,but [S
2
, H] 6= 0
(3) [S
1
+S
2
, H] = 0
247
Therefore S
2
1
, S
2
2
,and S
2
|o|
= (S
1
+ S
2
)
2
and its z-component S
:
|o|
are good
quantum numbers, but o
:
1
, o
:
2
are not. Hence we can denote the simultaneous eigenkets of
S
2
|o|
, S
:
|o|
, S
2
1
and S
2
2
by
|o
|o|
, o
:
|o|
, o
1
, o
2
i
such that
S
2
1
|o
|o|
, o
:
|o|
, o
1
, o
2
i
= o
1
(o
1
+ 1) |o
|o|
, o
:
|o|
, o
1
, o
2
i
S
2
2
|o
|o|
, o
:
|o|
, o
1
, o
2
i
= o
2
(o
2
+ 1) |o
|o|
, o
:
|o|
, o
1
, o
2
i
S
:
|o|1
|o
|o|
, o
:
|o|
, o
1
, o
2
i
= o
|o|
(o
|o|
+ 1) |o
|o|
, o
:
|o|
, o
1
, o
2
i
S
:
|o|1
|o
|o|
, o
:
|o|
, o
1
, o
2
i
= o
:
|o|
|o
|o|
, o
:
|o|
, o
1
, o
2
i
Fortunately, the state kets are also the eigenkets of H:
H |o
|o|
, o
:
|o|
, o
1
, o
2
i = 1|o
|o|
, o
:
|o|
, o
1
, o
2
i
where
1 =
J
2
[o
|o|
(o
|o|
+ 1) o
1
(o
1
+ 1) o
2
(o
2
+ 1)]
and
o
|o|
= |o
1
o
2
| , |o
1
o
2
| + 1, ...o
1
+o
2
248
Since the energy eigenvalues are independent of o
:
|o|
can be o
|o|
, ...o
|o|
the energy eigen-
states are (2o
|o|
+1)fold degenerated. From the point of view of symmetry, the degeneracy
of the eigenstates originates from the invariance of H under the ol(2) symmetry rotation,
lHl
= H
where
l = exp[iS
|o|
nc,~]
The ground state:
The lowest energy state is determined by the sign of J: when J 0, o
|o|
should
be taken to be minimum, otherwise o
|o|
should be taken to be maximal.
The case of J 0 :
o
|o|
= o
1
o
2
, (o
1
o
2
)
The two spins are antiparallel, which is called antiferromagnetic. The ground state energy
1
= J(o
1
+ 1)o
2
The case of J < 0
o
|o|
= o
1
+o
2
The two spins are parallel, which is ferromagnetic
1
= |J| o
1
o
2
7.3.3 Ferromagnetic Exchange (M ? 0)
Ferromagnetic exchange coupling originates from the direct Coulomb interaction and the
Pauli exclusion principle. In the second quantized form, the two-body Coulomb interaction
249
is given by
\ =
1
2
Z
drdj (r, j)
s
(r)
s
0
(j)
s
0 (j)
s
(r).
The eld operator
s
(r) =
X
I
c
I
(r)c
Is
.
The interaction can be expressed as
\ =
1
2
Z
drdj (r, j)
c
I
(r)c
(j)c
|
(j)c
l
(r)c
Is
c
s
0
c
|s
0 c
ls
=
X
I
l
II
:
Is
:
Is
+
X
I,I
0
l
II
0 (:
I%
+:
I&
)(:
I
0
%
+:
I
0
&
)
+
X
I,I
0
J
J
II
0 c
Is
c
I
0
s
0
c
Is
0 c
I
0
s
+
where
l
II
0 =
1
2
Z
drdj |c
I
(r)|
2
|c
I
0 (j)|
2
(r, j)
J
J
II
0 =
1
2
Z
drdj (r, j)c
I
0
(r)c
I
(j)c
I
0 (j)c
I
(r)
The exchange interaction J
J
acts as a Hersenberg interaction:
J
J
I
0
X
c
+
Is
c
+
I
0
s
0
c
Is
0 c
I
0
s
= 2J
J
I
0
X
(o
I
o
I
0 +
1
4
:
I
:
I
0 )
The positivity of J
J
I
0
can be proved as follows:
(1) Complete screening:
= c(r j)
250
In the case,
J
J
II
0 =
1
2
Z
dr|c
I
(r)|
2
|c
I
0 (r)|
2
0
(2) Long-ranged Coulomb interaction
=
c
2
|r j|
Assume c
I
is the plane wave
J
J
II
0
Z
drexp[i/ r]
c
2
|r|
= 4
c
2
/
2
0
This is ferromagnetic!
7.3.4 Antiferromagnetic Exchange
Two-site (or atom) problem with two electrons
Lets consider two orthogonal orbitals localized on two atoms labelled by i = 1, 2. Tunnelling
between the two atoms (or states) is described by a hopping Hamiltonian
H
|
= t
X
s
(C
+
1s
C
2s
+C
+
2s
C
1s
)
For simplicity, we consider an on-site interaction
H
u
= l
X
:
I%
:
I&
To explore the physical origin of anti-ferromagnetic coupling, we consider a special case:
n t. In the case, we choose H
u
to be the zero-order(or unperturbed) Hamiltonian and
H
+
to be the perturbation.
For H
u:
there are six possible congurations which are eigenstates of H
u
251
(1) E=0, |1 , 2 i , |1 , 2 i , |1 , 2 i , |1 , 2 i ;
(2) E=U, |1 , 1 i , |2 , 2 i
Denote |ci the unperturbed state with energy 0 and |:i denote the two state
with E=U (n=1,2). In terms of the c-operators,
|ci = c
+
1s
c
+
2s
|0i
and
|1i = C
1%
C
1&
|0i
|2i = C
2%
C
2&
|0i
In the rst-order perturbation theory
hc|H
|
| c
0
i = 0
In the second-order perturbation theory
D
c
H
(2)
c
E
=
P
n=1,2
hc|H
|
| :i h:|H
|
| ci
hc|H
u
| ci h:|H
u
| :i
=
1
l
X
H
|
:

:
H
|
|:i h:| is a projection operator

|1i h1| = :
1%
:
1&
(1 :
2%
)(1 :
2&
)( )
|2i h2| = (1 :
1%
)(1 :
1&
):
2%
:
2&
( )
252
H
|
1

1
H
|
c
0
= hc| (C
2%
C
.
1%
+C
2&
C
.
1&
+ ~.c)
:
1%
:
1&
(1 :
2%
)(1 :
2&
)
(C
2%
C
.
1%
+C
2&
C
.
1&
+ ~.c) |c
0
i
= hc| (C
2%
C
.
1%
+C
2&
C
.
1&
):
1%
:
1&
(1 :
2%
)(1 :
2&
)
(C
1%
C
.
2%
+C
1&
C
.
2&
) |c
0
i
= hc| (C
1&
C
.
1%
C
2%
C
.
1&
C
1%
C
.
1&
C
2&
C
.
2%
) +:
1%
:
2&
+:
1&
:
2%
) |c
0
i
= hc| 2o
1
o
2
+
1
2
(:
1%
+:
1&
)(:
2%
+:
2&
) |c
0
i
Therefore the eective Hamiltonian H
(2)
can be written as an isotropic antiferromagnetic
Hersenberg spin exchange form.
H
(2)
= +
4t
2
l
(
o
1

o
2
1
4
)
As 4t
2
,l 0, the exchange coupling is antiferromagnetic! The ground state of this two-site
problem is spin singlet, i.e. o = 0. Our discussion on the two-site problem can be easily
generalized to a many-site system. The Hersenberg model is dened on a lattice
H =
X
J
I
o
I
o
where i and , are the lattice sites and usually are of the nearest neighbour pair. o
I
can be
taken any value of half-integer, o = 1,2, 1....
One of the most important application of spin superexchange with current inter-
253
ests is the high temperature superconductivity. the so-called t J model is extensively
discussed over the last decade.
H = t
X
hIi,c
c
Ic
c
c
+
4t
2
l
X
I,
(
o
I

o
1
4
:
I
:
)
which is limited within the Hilbert space excluding double occupancy of electrons on the
same site.
254

Lecture Notes On Quantum Mechanics

Cargado por

Información del documento

Título original

Derechos de autor

Formatos disponibles

Compartir este documento

Compartir o incrustar documentos

Opciones para compartir

¿Le pareció útil este documento?

¿Este contenido es inapropiado?

Copyright:

Formatos disponibles

Lecture Notes On Quantum Mechanics

Cargado por

Copyright:

Formatos disponibles

LECTURE NOTES

xy|ci = x(y |ci) =

Outer product: |ci h,| is an operator!

ha| A|ai = a =ha| A

}. The number of the eigenkets is equal to the dimensionality of the complex

gives a new operator in the form of square

it is said that A is measured to be a

|+i = ~|i h+|+i = ~|i

and the operators from

can be expressed in terms of the eigen-

}, the operator is expressed as

In the Dirac bra-ket notation, we rewrite it as

= hr +||r +|i = hr|ri

= K. then all properties listed are satised.

b. The anti-Hermitian operator has only purely imaginary expectation value

Case C: The Hamiltonian

} forms a complete and orthogonal set of base kets. Therefore

The expectation values of two approaches are the same!

H s at dierent times commutes we have

a |:i = (: 1)a |:i

is divided into N-1 equal parts.

Then the Maxwells equations are reduced to

2.6.4 The Aharonov-Bohm Eect

The corresponding change in the action is given by

In this case both p

are not good quantum numbers. We cannot solve this problem

= c}: a is the root of r

These relation are often rewritten in a more compact form

to both side of the equation above. We can obtain numerical eigenkets

(a, /) real, we have

|,, :i = [,(, + 1) :(:1)]

| ,:i = ~ [,(, + 1) :(:1)]

, we are now in position to study the matrix

] = 0 , we dene the simultaneous eigenket of

Dene the total number operator N by

|,, :i = [,(, + 1) :(:1)]

to the left side of Eq.

is measured, what is the probability of the result ~,2?

(a)\ (r)t(a) = \ (r +a).

where |/i is an exact eigenket of H. The expectation value

are the exact eigenvalues and eigenstates for H

Using the formula

is one of the unperturbed eigenkets, as expected.

by expanding it in terms of the unperturbed eigenkets

from the left hand side, we obtain

in terms of the unperturbed eigenkets

with the eigenvalues

) is the reduced mass. Now the proton

So the general solutions are

The gradient operator can be expressed in polar coordinates as

Comparison with the total cross section shows

Then the | = 0 partial cross section is

. The total number of particles is

Here we use the grand canonical ensemble ( +)

fold. The ratio of the number of electrons

|:i h:| is a projection operator

También podría gustarte