You are on page 1of 300

yj<

gg

OU ~ 160162 >m *3

OSMANIA
Call No.

S3>O-

Accession No.

*SViAuthor

U M

1_1 ^- \

Tide

C\a.Nwa.vxLvN

This book should be returned on or before the date Itust marked below.

ELEMENTARY QUANTUM FIELD THEORY

INTERNATIONAL SERIES IN PURE AND APPLIED PHYSICS


Leonard L
Schiff,

Consulting Editor

Allis

and Berlin

Thermodynamics and

Statistical

Mechanics

-Becker

Introduction to Theoretical Mechanics

Clark
Collin

Applied X-Rays Field Theory of Guided Waves

Evans

The Atomic Nucleus

Finkelnburg Atomic Physics Ginzton Microwave Measurements

Nuclear Physics Gurney Introduction to Statistical Mechanics Hall Introduction to Electron Microscopy
Green

Hardy and Perrin

The Principles of Optics Harnwell Electricity and Electromagnetism Harnwell and Livingood Experimental Atomic Physics Harnwell and Stephens Atomic Physics

Henley and Thirring Elementary Quantum Field Theory Houston Principles of Mathematical Physics Hand High-frequency Measurements Kennard Kinetic Theory of Gases Lane Superfluid Physics
Leighton

Lindsay
Livingston

Principles of Modern Physics Mechanical Radiation

and Blewett

Particle Accelerators

Middleton

Morse
Muskat
Present

Introduction to Statistical Communication Theory Vibration and Sound

An

'Morse and Feshbach

Methods of Theoretical Physics Physical Principles of Oil Production Kinetic Theory of Gases
Introduction to

Read
Schiff Seitz
Slater

Dislocations in Crystals

Richtmyer, Kennard, and Lauritsen

Modern

Physics

Quantum Mechanics The Modern Theory of Solids


Introduction to Chemical Physics

Slater
Slater

Slater Slater

Quantum Theory of Atomic Quantum Theory of Atomic Quantum Theory of Matter


and Frank
Electromagnetism

Structure, Vol.

Structure, Vol. II

Slater and Frank


Slater

Introduction to Theoretical Physics


Electricity

and Frank Mechanics Smythe Static and Dynamic


Stratton

Mesons: A Summary of Experimental Facts Townes and Schawlow Microwave Spectroscopy


Thorndike

Electromagnetic Theory

White

Introduction to Atomic Spectra

The
G.

late F.

his death in 1939. P.

K. Richtmyer was Consulting Editor of the series from its inception in 1929 to Lee A. DuBridge was Consulting Editor from 1939 to 1946; and Harnwell from 1947 to 1954.

ELEMENTARY QUANTUM FIELD THEORY


ERNEST M. HENLEY
Professor of Physics
University

of Washington

WALTER THIRRING
Director
Institute for Theoretical Physics

University

of Vienna

McGRAW-HILL BOOK COMPANY,


New York
San Francisco

INC.

1962

Toronto

London

ELEMENTARY QUANTUM FIELD THEORY


1962 by the McGraw-Hill Book Company, Inc. Printed in the United States Copyright of America. All rights reserved. This book, or parts thereof, may not be reproduced in any form without permission of the publishers. Library of Congress Catalog Card Number
61-14047.

28149

Preface

is to present that aspect of quantum field theory not obscured by mathematical difficulties and which does not require a deep understanding of special relativity. Within this scope the emphasis has been placed on particle physics rather than on other applications of quantum field theory. To make the book comprehensible to a wide range of readers, we have presupposed only a knowledge of nonrelativistic quantum mechanics. All other tools that are needed are developed in the text. Thus, in the first part both the mathematical and physical descriptions of a quantum field are introduced. The conceptual aspects of the field are stressed. However, only fields that obey Bose-Einstein statistics are examined. Observables, invariants of the field, and internal symmetries are dis-

The aim of this book


is

which

cussed.

In the second part of the book further techniques are developed by considering the interactions of a quantum field with various static sources. Those problems that are known to have exact solutions, namely, the neutral scalar theory, the pair theory, and the Lee model, are treated from both classical and quantum-mechanical points of view. In the third part both the mathematical tools and the physical insight
acquired in earlier chapters are applied to low-energy pion physics. After describing a classical approach and various other methods that have been used to analyze the problem in the past, we turn to the one model that is not based on uncontrolled mathematical approximations, namely, the static model developed by Chew and Low. In terms of this model we attempt to give the reader an understanding of pion-nucleon scattering, the static properties of nucleons, electromagnetic phenomena, "" and nuclear forces.

VI

PREFACE

In the past few years a relativistic approach, based on analytic properties of the scattering matrix, has been evolving for the treatment of Although this approach reduces to that which we interacting fields. use in the nonrelativistic limit of the pion-nucleon problem, it is a wealthier one and contains much more of the physical situation than does the static model. It will thus ultimately allow a comparison with
detailed

experimental

results.

Unfortunately,

these

developments

necessitate considerably more involved calculations than those presented here, and it is not yet clear whether a complete theory underlies them. Although relativistic treatments should ultimately remove all the short-

comings of the models discussed herein, the nonrelativistic approach will remain the basic first step to master. For a unified treatment of all the problems covered, it seemed advan-

To emphasize the correspondence between classical and quantum-mechanical viewpoints, we chose the Heisenberg representation. We have endeavored to cover the ground within our scope reasonably thoroughly, stressing the intuitive meaning of the results. We realize that rigor and simplicity are complementary aspects of a theory and have therefore tried to keep a reasonable balance between these features. We have not attempted to include a complete list of references, but we have tried to indicate where the reader can obtain further information whenever we felt that this was necessary. For additional study of the subject we refer the reader to N. N. Bogoliubov and D. V. Shirkov, "Introduction to the Theory of Quantized Fields" (Interscience Publishers, a division of John Wiley & Sons, Inc., New York, 1959), J. Hamilton, "The Theory of Elementary Particles" (Oxford University Press, New York, 1959), and S. S. Schweber, "An Introduction to Relativistic Quantum Field Theory" (Row, Peterson & Company, Evanston, 111., 1961). We should like to thank Professors H. Frauenfelder and B. A. Jacobsohn for valuable comments and Drs. Ranninger and H. Pietschmann
tageous to work in a single representation.
for critically reading the proofs.

Ernest M. Henley Walter Thirring

Contents

Vlll

CONTENTS
7.

CHAPTER

INTERNAL DEGREES OF FREEDOM


Freedom Three and More Degrees of Freedom
Fields with

58
58

7.1

Two

Internal Degrees of

7.2

63

PART
CHAPTER

II.

SOLUBLE INTERACTIONS

8.

GENERAL ORIENTATION
Field Equations

.71
71

8.1

8.2
8.3

Quantization
Scattering

and Wave Matrix

75 78 82

CHAPTER

9.

STATIC SOURCE
Interpretation of "Static" Source Energy of the Coupled System

9.1

9.2
9.3

9.4
9.5

Connection between Bare and Physical States Fluctuations of the Field


Several Sources

82 84 86
89 90
. . . .

CHAPTER

10.

PRODUCTION OF PARTICLES
General Remarks
Specific

.93
93

10.1

10.2

Examples

96
100
100
104

CHAPTER

11.

PAIR THEORY, CLASSICAL


General Remarks

11.1

11.2
11.3

Bound

States

Behavior of the
Scattering

Wave

Matrix

a
. . .

106
109

11.4

CHAPTER

12.

PAIR THEORY, QUANTUM-MECHANICAL


State

.112
112
113

12.1

Quantization and Commutation Relations in the Presence of a

Bound

12.2
12.3

Scattering

Energy Expressions in Terms of the Asymptotic Fields


Virtual Particles

.119
122

12.4

CHAPTER

13.

THE LEE MODEL: STATES WITH Q =


Introduction

.126
126

13.1

13.2
13.3

Commutation Relations and Equations of Motion


Physical Nucleons Scattering States

.127
130
131

13.4
13.5

13.6

Completeness The Phase Shift

.132
135

CHAPTER

14.

14.1

Scattering:

LEE MODEL: STATES WITH Q = -f Low Equation


.
.

139
139

14.2
14.3

iT + n Scattering Low- and High-energy Behavior of

.141
145

T(k)

PART
CHAPTER
1

HI.
15.

PION PHYSICS

INTRODUCTION
Model Commutation Relations and Equations of Motion Comparison with Other Models
Static
. .

153
153
.
.

5.

r The

15.2
15.3

.158
162

CONTENTS
CHAPTER
16.

IX
.

GENERAL FEATURES OF THE STATIC MODEL


Treatment of Stationary Motion Classical Treatment of Scattering Quantum Aspects of the Static Model
Classical

166 166 172


175

16.1

16.2
16.3

CHAPTER

17.

THE GROUND STATE


Exact Results
Perturbation Theory

179

17.1

179
183*

17.2
17.3

17.4
17.5

Tamm-Dancoff Approximation Tomonaga Intermediate-coupling Approximation


Strong-coupling Approximation
.
. .

184
. . .

.186 .192
196
198

17.6

Numerical Methods

CHAPTER

18.

PION SCATTERING
Introduction

18.1

18.2 18.3

The

Scattering Matrix Properties of the Scattering Matrices

18.4
18.5

Low- and High-energy Limits of Elastic

Scattering

18.6
18.7

Diagonalization of the T Matrix Relation of Low Equations to Experiment Approximate Solution of the Low Equation

....

198 199 201 203

.....
. . .

18.8

Summary

204 208 210 214


219

CHAPTER

19.

PROPERTIES OF THE NUCLEON

19.1

19.2
19.3

219 Expectation Value of the Field .220 Ground-state Expectation Value of Observables 222 Renormalization Constants and Other Parameters of the Static Model

19.4
19.5

Nucleon Self-energy Charge and Current Distribution of Physical Nucleon. Magnetic

224
225

Moment
CHAPTER
20.

ELECTROMAGNETIC PHENOMENA
Contributions to Charge and Current Operators The Production Amplitude General Features of the Cross Section Comparison with Experiment

20.1

....
.
.

232
232 235 242 244 247 249
249 252 255 256 259
263

20.2
20.3

20.4
20.5

.......
...

Compton

Scattering

CHAPTER

21.

NUCLEAR FORCES
Quantum-mechanical Comparison with Experiment Concluding Remarks
Static Potential,

21.1

Introduction: Classical Calculation of Nuclear Interaction Energy

21.2
21.3

.......

21.4

Appendix
List of

Symbols

Index

269

Part

One

FREE FIELDS

CHAPTER

Introduction

1.1. Relation of Quantum and Classical Field Theory. Quantum theory provides us with a set of rules which are supposed to be of unlimited generality. They can be applied to any system and will tell us how our classical concepts have to be modified and how quantum

features arise in the system under consideration. The application of these rules to fields creates quantum field theory. Elementary quantum mechanics is not a consistent theory when combined with classical field
It was pointed out by Bohr and Rosenfeld 1 that inconsistencies theory. arise unless the classical electromagnetic field is quantized. If this is not

done, then, in principle, the uncertainty relation between a position and a momentum component of a particle (e.g., an electron) can be violated. The normal Schrodinger or Klein-Gordon wave functions y can also be regarded as classical matter fields and should therefore be subject to It is the latter type of fields with which we shall be quantization. mainly concerned in this book. As in ordinary quantum mechanics, the quantization of fields is linked to the classical theory by the corre-

spondence principle. It appears that the elementary quantum excitations of fields behave like particles; this is the only description we know at present to be applicable to elementary particles as we find them in
nature. Correspondingly, quantum field theory dominates our thinking about the fundamental features of matter. In the following we shall give a brief discussion of harmonic motions and fields in classical physics. The concept of a field is very wide, embracing all physical quantities which depend on space and time, like

N. Bohr and L. Rosenfeld, Kgl. Danske Videnskab. Selskab, Mat.-fys. Medd.> and L. Rosenfeld, in W. Pauli (ed.), "Niels Bohr and the Development of Physics," p. 70, McGraw-Hill Book Company, Inc., New York, 1955.
12(8) (1933);

FREE FIELDS

temperature, electric potential, and density. The common property of these phenomena is that there is an equilibrium state, and linear equations already reflect important behavioral features, if the departure from equilibrium can be considered small. Systems which are governed

by similar types of equations


physical situations.
tions.

(elliptic,

1 hyperbolic, etc. )

show

the

same

dynamical behavior, although they

Quantum

represent completely different field theory deals with hyperbolic equa-

may

Accordingly, we may, for a first orientation, consider ihe simplest system of this type, namely, a vibrating line of atoms. Furthermore,

we

shall see that in

many

respects there

is

little

difference between

a continuous and a discrete line. Hence we shall start with the latter because it is closer to classical mechanics. 2 In the following we shall concentrate on the formal aspects of the problem, assuming that the
physical situation is familiar to the reader. 1.2. Vibrating Line of Atoms. If, as shown in Fig. 1.1, q n denotes the displacement of the nth atom of a line from its equilibrium position,

KyH
?n-2
Fig. 1.1.
i,-l
<*n

?.+!
is a,

Line of vibrating atoms. The equilibrium separation between atoms and the instantaneous displacement of the nih atom is q n
.

# denotes the time derivative of this displacement, and we have harmonic forces between nearest neighbors, then the equation of motion is

fc^nfy..*!

-i-

2O
2

(1.1)

Here we have given the atoms unit mass, and O is the constant of the force between nearest neighbors. macroscopic piece of the line is, of less rigid. If a line of atoms is displaced by dx, then the course, restoring force is merely dx W/N.

To make
after

N atoms in such a manner that qi+N = q The problem posed by the N equations (1.1) can be solved by introducing normal coordinates.
it

the system of coupled oscillators finite,

we

close the line

This

is conveniently done with the aid of the Hamiltonian formalism, which will also be used later in the quantum theoretic development. The Hamiltonian (energy) of the line is readily seen to be (p n = q n)

tf

= 2*[pl + Q ta,.i

n=l

+ i) ]

(1.2)

See, ejg.,

H. Jeffreys and B.

S. Jeffreys,

"Methods of Mathematical Physics,"


2d
ed., p. 103,

p. 499,

Cambridge University

Press,

New

York, 1946.

Wiley

See, e.g., C. Kittel, "Introduction to Solid State Physics,*' Sons, Inc., New York, 1953.

John

&

INTRODUCTION
Hamilton's canonical equations

Pit in

~~~

""""""""

^>

LZ \Qn + 1 v j re T

j.

Cj

n in

~T~
'

Qn Tw

\ i

Q n/

Ti -n)

being equivalent to

(1.1).

We

define the

normal coordinates
q*

Q
Pn

and momenta Ps by

^N

""

(L3)

where, in accordance with the periodic boundary condition introduced r above, s takes on the values s = 2?r//A / being an integer between N/2 and A^/2. Since the qn are real, the Q s and P., themselves are complex
,

and

satisfy

With the

aid of the formula

we may

also invert (1.3) to

Inserting (1.3) into the Hamiltonian,

we find, by means of

(1.4)

and

(1.5),

(1.7)

Thus the normal coordinates serve to uncouple the oscillators, and the 1 equation of motion coming from the Hamiltonian (1.7) is simply

&=-%.
which also appears by inserting of (1.8) can be written as

"

=: 2ft sin

(1.8)

(1.3) directly into (1.1).

The

solution

QM = (WO) cos
-(0)

co a t

+ +
2

sin

cos [s(n

- n') -

e^]

sin [s(n

- n') -

o>^]
)

0) 8
1

The reader may check for himself that

Q 8 and Q can* be treated as

independent

variables.

FREE FIELDS

Equation (1.3) together with (1.8) tells us that the motion of our system is a superposition of vibrations with frequencies co s many of which are much smaller than Q (M S O2?r//A^), corresponding to the smaller rigidity of the whole line, about which we remarked earlier. This well-known fact is, for instance, the point of Debye's theory of the In quantum mechanics we shall specific heat, as opposed to Einstein's. see that the excited states of the oscillators Q s with energies (o s behave in many respects like particles. Oscillations of the type considered here are very common phenomena, appearing as sound waves in solids and liquids, as spin waves in ferromagnetics, and as surface waves in nuclei. In all these cases we find the particlelike behavior of the elementary excitations of the oscillators Q s with energies oj s In fact, the sound waves in liquid helium are the closest mechanical model of elementary

particles that
1.3.

we

have.

Continuous Vibrating Line.

In

many

cases

it

is

expedient to

look at the situation from the macroscopic, rather than from the microscopic, point of view and not to resolve the line into individual atoms. -> oo as the distance This can be done by a limiting process in which between the atoms a -*> but the length L = aN of the line remains This means, however, that the system now acquires infinitely constant. many degrees of freedom and that we need this number of variables for

its

description.

1 -> Calling x the distance from the origin of the line [q n q(x)], we obtain for the equation of motion in this limit the familiar partial

differential

equation
flx)

= Qv^(x)

(1.9)

(1.9) la appears to be the wave velocity v, so that ft must behave in this limit like Q, Thus, to obtain with infinitely many atoms a vja.

In

line

of atoms.

finite rigidity requires

an

infinite force

between neighboring

As a consequence,

the line will be capable of vibrations with

infinite frequencies.

The general solution of (1.9) subject to q(x) = q(x -f- L) can again be obtained by carrying out the limiting process on the normal coordinates
in (1.9).

With
.

2n

_ 00 < <
.

00

and
_

W = -|^C*
= qja*,
so that, for a
finite energy, q(x) will

(1.10)

t1

We

actually put q(x)

remain

finite.

INTRODUCTION

and from the

limit

of

(1.5),

CL

dx
\

d Jo o

we

find

from

(1.9), for

the equations of motion of

Qk

&=-kVQ
.

(1.11)

Thus the frequency co k is kv, which is the limit of our previous expression Hence it is just for short wavelengths that the atomic (1.8) for a) s Introduction of normal coordinates structure of the chain transpires. means solving a partial differential equation by a Fourier expansion. These results can also be deduced from the Hamiltonian (1.2), which becomes in the limit1

H = \dxq\x) +
Whereas
Eq.
(1
.

=
ij(&|

kV |6*|

2 )

d-12)

k Eq. (1.12) leads directly to to generalize the Hamiltonian formalism of ordinary mechanics to get the equations of motion for the nonenumerable coordinates q(x). This can be done by introducing functional deriva1

for the enumerable coordinates

1),

we have

tives

with the aid of Dirac's d function

(1.13)

dq(x')

dx

dx
>

The former

d nn ., and the latter is the continuum form of dq n /dq n obtained by differentiating with respect to x. Writing the Hamiltonian in terms of canonical variables /; and q,

is

H=
we

f7
Jo

'

2, , dx \p\x)
I

+
,

2 Jdq(x)]*\
\

-~

L dx J

find that Hamilton's equations

are generalized to
"

"~

~~

"

*
Jo

*
dx'

~~ V

dq(x)

dx

dx 2

SH
1

We give the whole line unit mass.

FREE FIELDS
(1 .9).

will find frequent applications with the continuous case, which is mainly almost simpler than the atomistic point of view. It allows us to elimiand to replace them by the nate the microscopic constants Q, a, macroscopic constants u, L. Some fields, however, such as the electromagnetic one and those of elementary particles, do not possess mechanical backgrounds to serve as guides in writing equations of motion. We have to appeal to the special theory of relativity to obtain the invariance properties of these The four-dimensional homogeneity and isotropy of our spacefields. time continuum are supposed to emerge from the same property of the In technical language, the infields of all the elementary particles. variance under the inhomogeneous Lorentz group is the only guiding principle which allows us to select the possible field equations for elementary particles. This daringly speculative procedure is, in fact, very successful and reveals many startling properties of elementary particles. Unfortunately, the theory of the representations of the Lorentz group is far from being elementary, so that we shall not be able to give a systematic discussion of relativistic field theory. However, we shall encounter the influence of relativity theory on our notions about particles. The requirements of Lorentz invariance gives fields remarkable Since properties which are not possessed by any mechanical system. the theory has to be invariant under arbitrary space-time displacements, the field cannot have any atomic structure but must be continuous. Furthermore, the field must fill all space and time; it has to last forever everywhere and can never be removed. Thus we arrive at a new outlook on space and matter. Space is spanned by the continuous background of the fields of elementary particles; in some respects this is Matter is just a the sequel of the ether concept of the last century. There is no local excitation of this background, something accidental. conservation of matter, and the laws governing the interactions of matter are secondary and complex. The simplicity of nature is revealed by the equations of the elementary fields, which reflect symmetry and regularity. This is quite a different picture from the mechanical one, in which matter is supposed to be fundamental and the law of This force between its constituents is the primary law of nature. explains why the present fundamental research in physics makes so much use of quantum field theory, which concentrates on exploring the properties of this background for all physical phenomena.
later, since

and agree with

These formal tools

we

shall deal

CHAPTER

The Harmonic

Oscillator

2.1. Eigenvalues of H. For the fields considered in Chap. 1 it was shown that the basic equations of motion are like those of a simple harmonic oscillator or of a set of coupled harmonic oscillators. In this chapter we shall therefore give the quantum theory of the harmonic oscillator in a form appropriate for later developments. It will appear

oscillators

in the following chapters that the quantum development for coupled and for fields is a straightforward generalization of the theory

for this system with one degree of freedom. Moreover, the typical quantum features of fields are already encountered in rudimentary form
in the

harmonic
is

oscillator,

where they are familiar from elementary

discussions.

Our problem

characterized by the Hamiltonian

H=
the

4(p

-f

o>V)
are

(2.1)

The coordinate q and the momentum p commutation relation 1


[<?,p]

now

operators which obey


(2.2)

a typical prediction of quantum theory that measurement of an observable cannot yield an arbitrary result, but only an eigenvalue of the operator associated with this observable. We must therefore seek the eigenvalues of observables such as the energy (2.1). This problem can be attacked in several ways. For instance, we can satisfy (2.2) by
It is

ential equation arising

id/dq and solve the differrepresenting p by the differential operator from the eigenvalue problem (H 0. E)y> Such an approach is not the shortest one, and for our purpose a purely

We shall

always use appropriate units with h 9

10

FREE FIELDS

algebraic

method

is

more convenient.

We

introduce the operators

which correspond to the amplitude of the


a

classical

motion:

^
(2.3)

or
(2.4)

(2*,)*

It

follows from (2.2) that the commutation relations for a and a f are

[,] = [aV] =

[V] =
(2.5),

(2.5)
1

In terms of a and a f the Hamiltonian becomes, by means of (2.4) and

H^utfa +
Since the position and
Energy

to

(2.6)

momentum

with the Hamiltonian, the operators a and # f


either.
it

operators q and p do not commute will not commute with it


In fact, from (2.5) and (2.6)
follows that
[//,<,']

and

tf>

[a,//]

=
the

a>

(2 ?)
coa

The form of
relations

commutation

(2.7) allows us to draw conclusions about the eigenvalues

of

//.

function
9

E (H This

Applying them to an eigenwith eigenvalue y E of

E)y E

= 0,

we

find

tells

us that ay E

eigenfunction of the eigenvalue

is another belonging to

a>.

Similarly,

it

Fig.2.1. Harmoniooscillatorpotential

with energy eigenvalues and groundstate

follows from the other relation (2.7) that * increases an eigenvalue by w.

wave function

yi

(o).

This shows that there are equally

THE HARMONIC OSCILLATOR


with the spacing o>. spaced sequences of eigenvalues of these sequences have to terminate somewhere, since tfa
therefore, possess no eigenvalue <0. tion requires that for a certain state y the relation 0y>
definite

11

However,
is

and

H can,

positive

This condi-

hold, in

which case we cannot get a lower eigenvalue by applying a. From (2.6) we see that y is &n eigenfunction of// and belongs to the eigenvalue ro/2. Furthermore, our condition determines ^o uniquely, so that there is" only one sequence of eigenvalues. We can summarize our findings as follows. The eigenvalue spectrum of is shown in Fig. 2.1 and is

E
n being a nonnegative integer.
2.2. Properties

nco

i<u

(2.9)

of the Eigenstates of
ay>

//.

The

state ^o correspond-

ing to

obeys

=
f

(2.10)

and the

state

y n belonging to

En is obtained by applying

n times to

vv

That
ized,

(nl)~*

is

the correct normalization factor, provided y

is

normal-

can be seen most easily by induction:


w,-,

fvtn-fv:-! J

n
this

n
==

equation finally leads to JjVJtyn Jy* Voalso be used to obtain an explicit representaSince in tion of the eigenfunctions \p n in terms of the space variable q.

Repeated use of

The present method can

this representation

p =

idjdq,

we have

and

it

follows that the normalized ground-state eigenfunction

is

-+*
The higher
excited states v* are then obtained

(2.13)

differential operator

af

by the application of the

(wq

d/dq)(2a>)~*.

the ground-state wave function is of the order o>~ J ; this expresses the quantum-mechanical zero-point fluctuation A0 of the

The width of

12

FREE FIELDS
is

position q and

shown

in Fig. 2.1.

Formally,

we can

define this

uncertainty by

dq

(2.14)

where q is the mean value of q and (2.10) and its hermitian conjugate

is

easily seen to vanish

by means of

This result
is

f
J

&Vb dq = Jf V?(
from y Q (q)

=
(2w)-

(2.15)

also apparent

=
Vo*

y Q (q).

With our methods

we

find that

VoVvo ^4

=
J

^
-

2co

a direct consequence of the uncertainty principle is not obtained by localizing the particle sharply at the origin, since this would entail a large A/? and hence kinetic energy. Since the mean values of the position and momentum are zero, the average value of the energy is | [(A/?) 2 -f <w 2 (A^) 2 ]. If this is minimized with respect to A<y, with the constraint that A/? Agr = * we
Physically, this
A<jr

is

A/?

>

J.

The lowest energy

find that the

most

That is to say, the energetically favorable compromise is close to the value for tq given by (2. 16), and the lowest energy is not zero, as in the classical case, but co/2. 1 These quantum-mechanical fluctuations are usually small but
is

minimum

eu/2 for

==

A<JT

(2co)~

'.

sometimes lead to macroscopic effects. For instance, they prevent liquid helium from solidifying under normal pressure. 2.3, Time Dependence of Motion. From these quantum features

we now turn to dynamical aspects which reflect the classical harmonic motion of the system. The time development can be described in 1 We can, for instance, consider the quantum mechanics in many ways. constant and the state vector time-dependent according to operators
V<0
(Schrodinger representation).
state vector constant
t It is

e-^'tfO)
is

(2.17)

and

possibility im to the to apply the unitary operator e

Another

to consider the

gives the exact numerical result.

only an accident that this rough argument with the uncertainty relation However, one usually gets the correct order of
ed.,

in this manner. A. M. Dirac, "The Principles of Quantum Mechanics/' 3d Oxford University Press, New York, 1947.

magnitude
1

See

P.

chap. V,

THE HARMONIC OSCILLATOR


operators 0, so that they vary with time according to
0(t)

13

= e iHi <D(Q)e-' =

ttt

(2.18)

(Heisenberg representation).
tion values,

This obviously leads to the same expectaV*(0)0(OY<0)

y>*(f)0(OMO

and

to the

same physical consequences.

In the latter case the time dependence of the operators is such that they obey the classical equations of motion, since (2.18) yields

0(0

/[//,0(0]

(2.19)

and the commutator


classical mechanics.
will

gives the

same expression

as the Poisson bracket in

Because in our problems the classical equations be of a well-known structure and tools for their solution are readily

available,

shall always stay in the Heisenberg representation. 1 to reserve the letter y for the field operators, we shall use Furthermore, Dirac's 2 bra and ket notation from now on. In it the state vector and

we

its

complex conjugate are denoted by the brackets

|>

and

(|.

To

specify

For the state vector, we may write some labels into the bracket. instance, the energy eigenstates of the harmonic oscillator can simply be
characterized by the associated quantum so that the Schrodinger equation reads

number of the

state, e.g.,

w),

In elementary

wave mechanics

this notation

corresponds to denoting a

vector by a single symbol r rather than specifying its components in a The latter appear as the scalar product particular frame, e.g., x,y, z. of the vector r with the unit vector n in the direction of the axes (in a
r n,.. Thus, x Correspondparticular frame) under consideration. the Schrodinger function y, (q) at a point in coordinate space, </', ingly, is trie component of the state \n) in the direction of an eigenvector
}

|<7')

of the operator q and

is

given by the scalar product

The components of the state \n) in another frame are linear combinations of y n (q') in the same way that the components of r are in a frame
given by the unit vectors n
x'
t :

=r

iv

= 2 =
i

(r

n<X<

'

M=

x,V,z

i~x

2
t

r,(n,
v,z

n^)
is

1 There are other useful representations, in which part of the time dependence retained by the state functions. We shall not be concerned with these here. 2 Dirac, op. cit., chaps. l-IIL

14

FREE FIELDS
instance, in

For

momentum
we
(P'

the operator/?,

space, where we use the eigenstates \p'} of obtain, in analogy with the above,
n)

6,00
which
space
is

(p'\q')(q'

n)

the usual expression for the


insert

wave function

in

momentum

if

we

Our new notation can be

illustrated for a general operator

by

analogy with the above development.

Thus, an eigenvalue equation

can be rewritten, by multiplying by


(q'\0\ n)
or, equivalently,

(q'\ 9

as

q"}(q" ^jdq" (q'\&\

n)

= *(q'

n)

dq" (q'\Q\ q")y n (q")

= ^ n(q')
\

The operator

is therefore a matrix in a particular representation. a general matrix element (m n) can be rewritten as Similarly,
\

(m

n)

=
if

These general equalities are considerably simplified is a is chosen in which diagonal matrix,

a representation

For example, we can

find

(/?'

q') as follows:

q\q'}=q'\q'}
(P'\q\q')-j(p'\q\p")(p"\q')dp"=q'(p'\q')
In

momentum

space, the diagonal representation of q

is

where

6' is

the

first

derivative of the d function.

We

therefore obtain1

oc

stands for "proportional to."

THE HARMONIC OSCILLATOR

15

Having

system. operator a

dealt with these preliminaries, we can study the motion of the According to (2.7) the equation of motion (2.19) for the
is

simply
d(i)

-icoa(t)

(2.20)

which can immediately be integrated to


a(t)

= e~

itot

a(G)

(2.21)

From this result and its hermitian conjugate we find that the position and momentum are the same functions of time as in classical mechanics
:

(2 - 22fl)

_
t

To

certain state

learn something about the time dependence of our system in a ), we shall calculate
|

which represents the probability of finding the oscillator at q' at the time t. For the states n) this will, of course, be time-independent; in our representation this follows from (2.21). To obtain something more In interesting, we have to consider a superposition of the states \n). particular, it is useful to study the state (wave packet) d) which is an
|

eigenstate of the nonhermitian operator a:

a(0)

d)

d d)
|

(2.23)

As we

shall

show,

it

classical

frequency

co

undergoes harmonic motion of period equal to the of the oscillator. In analogy with (2.13), we have
- ( "-*)t "*
(2.24)

of the same width as the ground state but e.g., a gaussian distribution displaced from the origin by the distance d. To be sure, such a state is not an eigenstate of //, and our formalism tells us immediately that the
proportion of the state
|

n) present in the packet

is

-<0
nl

a"

16

FREE FIELDS
set

Making use of the completeness of the

of states
|

n}

we

find

and

(2.25)

Fig. 2.2. Representation of

packet and

its

shown

at

motion of packet d). The motion of the center of the width A^' are represented. The distribution of the packet is also
|

0.

Thus the probability of finding the nth


Poisson law 1 with a

excited state in
|

d) follows a

mean
cod 2

value of
I

displacement

\2

^zero-point fluctuation/

We
1

shall see shortly that the

with an amplitude

d and frequency

wave packet performs harmonic motion cu. Hence the dominant state in it

p. 425,

H. Margenau and G. M. Murphy, "The Mathematics of Physics and Chemistry," D. Van Nostrand Company, Inc., Princeton, N.J., 1943.

THE HARMONIC OSCILLATOR


2

17

is the excited level for which the energy equals JoA/ , e.g., the classical energy of such a motion. In order to express d) in terms of the eigenstates of <?(/), we remember
|

from

1 (2.22) that

(2.26)

Thus we

find, for

yd [q'(t)}

<<7'(0

/>,

[V(0
which integrates to

g-^]

V-

^"'

tt<

V-

(2-27)

exp
or,

on normalizing,
Vdh'(0]

=
("")

ex P

(q'

d cos

cor ~h

2 id sin <wr)(^'

~ d cos <wr)
(2.28)

This demonstrates that our wave packet performs rigid oscillations with frequency w, as shown in Fig. 2.2. For the mean values of position and

momentum, we

get,

from
(d

(2.28),
|

q(t)

d} d)

d cos

<ot

(d

p(r)

OK/ sin

wr

Hence the
later.

state
|

d)

tion for classical

is the appropriate quantum-mechanical generalizaharmonic motion and will find frequent applications

In our subsequent studies of more complicated systems we shall follow the pattern of the above treatment for the harmonic oscillator,
since
1

we

shall

always encounter similar situations.


is

Here and subsequently, when no time


t

that

0, e.g.,

a - a

(t

shown

after

an operator, we shall imply

0)

a(0).

CHAPTER

Coupled

Oscillators

theory,

To approach quantum field 3.1. Eigenvalues of the Hamiltonian. we now treat a system of coupled oscillators quantum-mechani-

The problem we shall consider is a generalization of that dealt cally. with in Chap. 1 (Fig. 1.1), namely, oscillators on a closed line coupled not only to their neighbors but also to their equilibrium position. The Hamiltonian and the equations of motion for such a system are expressed in terms of the generalized coordinates qn and momenta pn
:

qn

= &(q n+l + q n ^ - 2q n - ^q n
)

(3.1)

Putting the second independent frequency & equal to zero would bring us back to (1.2). It will turn out later that, in the continuum limit, (1.2) corresponds to the case of a massless field, whereas (3.1) correAs in Chap. 2, we now have to state sponds to a field with "mass" Q the commutation relations, which are, according to the general rules of
.

quantum mechanics,

[i,J>

J=
=

tfi.m

[>Z,<?m]

OhPm]

=
expedient to use the

(3 2)

To work out the eigenvalues of variables defined in Chaps. 1 and 2,


normal coordinates
(1.3):

it is

new

first,

for example, to introduce the

.-?<-$
1

*-=?<"'"'*
1

(3 - 3)

See, e.g., L.
Inc.,

I.

Schiff,

Company,

New

"Quantum Mechanics,*' 2d ed., p.


18

35,

McGraw-Hill Book

York, 1955.

COUPLED OSCILLATORS

19

In terms of these variables the commutation rules deduced by means of


(1.5)

become

= [a,fi j =
[fizAJ

'<5*,m

[P,,P

j=o
= qn
,

(34)
the conditions (1.4)

Since qn and p n are hermitian operators, ql have to be generalized to

Q. a
Inserting the

Ql

P. s

= P]
we
find
1

(3.5)

new

coordinates into the Hamiltonian,


l

+ <aJfi.fi.

>

In terms of the normal coordinates the oscillators are decoupled, and


in accordance with (3.5)

we now

introduce, as in Chap. 2, the variables

(3.7)

(3.8)

Note that 0_ s
a8 a] with
,

a\

2nl/N and
a]

tions for a 8

and

and that we again have 2N independent operators N/2 < I < N/2. The commutation relafollow directly from (3.4) and (3.7):

The Hamiltonian becomes a sum of terms of the form

(2.6),

!.-.
i)

+D
(3.10)

and we may draw conclusions about


in the previous section.

its

eigenvalues and eigenvectors as

The

state

of lowest energy

0>

is

determined by the condition

a,|0>

(3.11)

20
for

FREE FIELDS
all s.

It

belongs to the eigenvalue

=2K
The
general eigenvalue
is

(3.12)

given by
(3.13)

where the n 8 are a

set of nonnegative integers. The eigenvector belonging to the eigenvalue (3.13) is a generalization of (2111):
"i," 2 ,3,
. .

N
!

.,**>

= (i! "

Mr Vir^r^r*

(4r*

o>
(3.14)

fact that the eigenvalues of the energy are integer multiples of basic frequencies lends itself to a particle interpretation. The state (3.14) behaves like one with n^ particles of energy w l9 w 2 particles of

The

Later, when we consider localized quantities, such as energy 2 , etc the energy or momentum contained in a certain volume, it will become apparent that the particle properties of the system are actually much more extensive than they now appear. Since our system just represents The elastic (sound) waves, the quanta are usually called phonons. energy of the quanta is additive, so that they behave like noninteracting Furthermore, a state is characterized only by the number of particles. in the modes with energies w s and there is no possibility of particles distinguishing the various particles in the same mode. Every mode can accommodate an arbitrary number of particles. Hence the particles obey Bose-Einstein statistics, and we have a model for particles which
-

are indistinguishable. It appears that particles are more like vibrations than like classical bodies, and any two vibrations cannot differ so long That particles lose their identity is as they have the same frequency.

one of the most revolutionary consequences of the application of

quantum theory to fields. We shall not belabor the point, since it is discussed in most books on elementary wave mechanics. The correspondence between the elementary excitations of an elastic body and an ideal Bose gas forms the basis of the theory of specific heat of solids at low temperatures. Usually it is taken as granted that to any motion with a frequency CD there belongs an energy fao. As we have seen, some mathematical development is necessary to deduce this result from first Our idealization of purely harmonic forces is, of course, principles. not always close to reality, but there are systems where the essential features of the Bose gas show up. It remains for us to study how the typical 3.2. Quantum Features. features of the oscillator, such as the zero-point motion and quantum From (3.12) energy, manifest themselves in our vibrating string. and (3.6), we see that the zero-point energy lies between Nl /2 and

COUPLED OSCILLATORS

21

+ ^o)V2; it is thus the same as that of uncoupled oscillators with basic frequencies lying between A^Q /2 and N(Q? 4- &o)V2- In a crystal lattice this zero-point energy plays an important role, but for the fields of elementary particles it has not yet been possible to relate it to observable effects. In the latter case, since the number of degrees of
freedom

N goes to infinity,

it

becomes

infinite.

Relativistic invariance

requires that it should be zero, since the state with no particles should look the same to observers in different Lorentz frames; a nonvanishing
the energy. But one day, its deeper significance will be discovered. perhaps, For the zero-point oscillation of the nth atom in the ground state of the system, we find, since (0 q n 0) 0,

energy-momentum vector spoils removed by calling the operator

this property.

It is

conventionally

(A<7 M)

^<0|^|
s,s'

>
i

i(s-s')n

N
.

2a> s

That

is to say, the square fluctuation is just the mean value of the As was to be expected, fluctuations associated with the frequencies co s the quantum fluctuations of the various modes are independent, so that

the square fluctuations are additive.

an amplitude which atomic dimensions, and they are


tions have

is

For atoms in a lattice the fluctuasomewhere between nuclear and

directly observable, for example,

the scattering of light or neutrons. weaker the forces between them the

by The lighter the atoms and the more violent are the fluctuations.

they lead to macroscopic effects for helium, which they temperature. prevent from solidifying under normal pressure even at Although this fluctuation is familiar from elementary quantum mechanics, the analogous result for fields is somewhat surprising and was only discovered in the modern development of quantum electrodynamics. We shall take this up in detail in a later chapter, where we shall study
the fluctuation effects for states in which particles are present. For the time dependence of the 3.3. Dynamical Aspects.
field

As mentioned,

operators

we

obtain, in analogy with (2.20) to (2.22),

and

q n (t)

=2777- [^"-"'WO) + =2(777-) [^ \2Ncty


s

-*-"-V(0)]

(3.17)

This form is identical with the time dependence of the classical solution of (1.8). It is the most general superposition of vibrations with eigenfrequencies a> s with the important difference that the coefficients are
,

operators.

22

FREE FIELDS

In analogy with the end of the last chapter, we can construct a standard wave packet in which the th atom is, at the time t = 0, displaced from its equilibrium position by dn -. general state dt ) of this kind is defined by

For real dn the expectation values of the positions at a time t all s. correspond to the classical motion caused by an initial displacement dn of the atoms and zero initial velocity:
for
(d
|

q n (f) \d)

= 2 1 dn N
s.n'

cos [s(n

n')

o>.f]

(3.19)

A macroscopic sound wave with a single frequency


can be represented by

co s

dn

de i8 n

'

(3.18), with the dn assuming the Calling this state c<v>, we have
|

and amplitude d complex values

and hence it corresponds to a Poisson distribution of phonons with the 2 appropriate frequency and a mean number of phonons d co s ,/2. For the time-dependent solution we find

<(0

V>

w (0

exp

(q n

'

and the average positions are


(d
|

in this case simply given

by
(3.20)
it is

qn

d)

= d cos

(s'

avO

If

we want

this classical-like

motion to be observable,

necessary

that the displacement

larger than the zero-point fluctuation can be seen from (3.15) to be of the order of amplitude. 1 Since the mean l/(ov) if all frequencies are of the order of a> s number of phonons is d2 co s ,/2 the above condition implies that it is >1. In a state with a definite number n' of phonons, the expectation value of all q n (e.g., (n' q n n')) is zero. This is usually expressed by saying that the phases of the waves associated with phonons are completely undetermined. In summary, we can say that sound waves and phonons represent the classical and quantum-mechanical aspects of vibrating systems and are generalizations of what we found for the harmonic oscillator.

d be much

The

latter

>.

CHAPTER
Fields

shall now investigate Continuously Coupled Oscillators. the quantum theoretic development works in the limit of a con-> oo, a -+ 0, but aN finite (see tinuous line. In the limit Chap. 1),
4.1.

We

how

the Hamiltonian and the field equations (3.1)

become

iS

J"

U*"

\"7-/

~
\

r\

a,

(4.D

limit the

have already remarked, in Chap. 1, that even in the continuum normal coordinates form an enumerable set. Therefore we shall first quantize the theory in terms of these variables and shall study the commutation rules for the continuous coordinates q(x) later on.
Rewriting
(3.3) in the

We

form
"

(1.10),
*

with s -* ka,
"^

k
2irl

(4 2)
*

fc

00

< <
/

00

we

find for the energy, as hi (1.12),


* * *

+Q
23

(4 3)
'

24

FREE FIELDS
(1.5),

where we have used the continuum limit of L dx e i<k -"' )x = LA

'fcfc'

This is now a sum over an infinite number of uncoupled oscillators. In terms of the labels k, the commutation rules (3.2) are

KWV] = W W
and

'

^
and
a* of (3.7) are

(4>4)

similarly those for the operators a

The introduction of

these operators allows us to write the following

expression for the energy:

k has the significance of the momentum of the parwe2 with energy a) k Hence, if v is the velocity of light and Q then the energy and momentum of a field quantum are related in the same manner as those of a relativistic particle. In the continuum form it is easy to generalize to the three-dimensional case. In the mechanical model of a displacement field which we had in mind so far, the general three-dimensional case is somewhat more complicated, since the field is then a vector and has three components. However, by only allowing displacements of a three-dimensional
ticles
.

eigenvalues of the next chapters that

In correspondence to the development of Chap. are integer multiples of the

3,
a> k .

we

find that the

We shall see in

lattice in one direction, say x, as shown in Fig. 4.1, we have the It is this discrete analogue of a scalar (hermitian) field <(*,j>,z). case which will prove to be an appropriate description for pions simpler

atomic

if

we

identify v with the velocity

pion.

of light and The three-dimensional version of (3.1)

with the mass of the

is

//(4.6)

which
1

is the Klein-Gordon equation. Anticipating future notation, we have put v

c equal to

and

We shall use r as an abbreviation for the three space coordinates (xi,x 2 ,x 3) or 2 r2 = r ). Frequently we shall write x,y, z; (r,f) for (#,j,z,f); and r for (e.g.,
|r|

(r,0)

simply as

r.

FIELDS

25

replaced
requires
<f>(x

by m.

The three-dimensional

periodicity

condition

+ L, y, z, = #*, y + L,z,t) =

<f>(x,

y,z

+ L,t) =
(4.6)

ftx, y,

2, t)

(4.7)

These conditions and the equations of motion 1 analogy to (3.17) by


x

can be

satisfied in

9(r,t).

-j

^ >
(48)

Iff

i,

*
-*)

H-y

h-\

The cube of vibrating atoms, of Fig. 4.1. Mechanical analogue of a scalar field. length L, has an atomic equilibrium separation of a. All atoms vibrate in a single direction, here chosen to be x.

The commutation

rules (3.9) are generalized to

(4.9)

We shall shortly develop a more general recipe for finding the commutation properties of the field. The Hamiltonian can be written in the familiar form (3.10), except that the sum over k is now a threedimensional one:

#=2>fak + i)
1

(4.10)

We shall

henceforth abbreviate

(o

by

o>,

w by
,

fc

CD',

etc.

26

FREE FIELDS

Thus we get the remarkable result that, by applying the rules of quantum mechanics to a field which obeys the Klein-Gordon equation, we obtain a system that behaves like an ensemble of an unlimited number of relativistic Bose particles. To be more specific, we have, by analogy with (3.14) and (3.13), a state

0,0,0,

.>

which

is

an eigenstate of the Hamiltonian


|

H with energy

,1

n l9 n n ,

.)

= 2 (akfl

=
The
eigenfunctions therefore satisfy a Schrodinger equation for an unlimited number of particles with energies given by

1 Since the application of a* to a ket with n particles yields one with n f is usually called a creation operator and a, particles, correspondingly,

a destruction operator. There are other fields for which the Hamiltonian is not the continuum analogue of coupled oscillators but the Larmor precession of electron In this case one finds that quantization leads to particles spins. obeying Fermi-Dirac statistics. The kinds of fields that correspond to particles with half-odd-integral spin are beyond the scope of this book. However, we should like to point out that quantum field theory predicts 1 the experimentally established connection between spin and statistics. to three dimensions increases the number Roughly speaking, going of degrees of freedom by a factor of 3. This change is not so drastic as -+ <x> used in this chapter. that of the limiting procedure we shall chiefly discuss the limit L -> oo, where the unFinally, physical boundary condition (4.1) is relaxed. In this limit the k vectors in (4,8) become a dense set such that the fc are now a continuous

oo to oo. shall emphasize this by denoting variable going from the destruction and creation operators in this limit by <ar(k), d*(k). Furthermore then has to be replaced by an integral. Since the

We

H This tatement and the equation that follows can also be proved directly, by means of the communication relations (4.9). 1 W. Pauli, in W. Pauli (ed.), "Niels Bohr and the Development of Physics," p. 30, McGraw-Hill Book Company, Inc., New York, 1955.

FIELDS
distance between two neighboring points in 3 density is (L/27r) , and hence in the limit

27

k space

is

2ir/L, their

In most calculations leading to numerical results, this suitably weighted sum over all degrees of freedom occurs, and hence we introduce a new notation $ for it.

This introduction of infinitely


questions, which
infinitely

many

we

shall

now

discuss.

degrees of freedom raises some First of all, since there are

many

<o

and they do not have an upper bound, the zero-point


(4.11)

energy

diverges.

This can be remedied by calling


(4.12)

the energy, since at this moment we do not know what the zero-point energy of the field is. One could ask whether the infinite sum (4.12) converges toward a limit. This raises difficult questions about nonseparable Hilbert spaces which we are not prepared to answer. The reader has to be content with the observation that for states with only a finite number of excited oscillators the application of gives a finite sum.

For the other typical quantum feature, namely, the zero-point fluctuations of <(r), the infinite number of degrees of freedom also As in (2.15) and (3.15), we find creates some difficulties.
(0
|

#r) 0)
|

c^'>r-<oi*'>^
which diverges.
obtain

(*>
in the limit

This can best be seen

-+

oo,

where we

(4.14)

The

infinite fluctuation is infinite


is
it

with an

norm when

connected with the fact that <(r) gives a state 1 applied to any state with finite energy.

Thus <(r) However,


1

not an operator in the Hilbert space we are dealing with. turns out that the average of <(r) over a finite region in
is

Equation (4.14)

a special case of this statement for the vacuum

state.

28

FREE FIELDS

space has a finite square fluctuation. z field b over a volume b by


(f>

To

see this,

we

define the average

and obtain
\ (0

S ini
I

0> /

= (27r)v )

&- 8 d*k d*r


j

flMr-r')

d*r'

e 2(0

^a

(4.15)

This teaches us that the fluctuations of the field become more and more Since violent as we decrease the volume b* over which we average. the averaging process renders wavelengths less than b ineffective, a
decrease in the volume increases the contributions to the field fluctuations.

At

first

sight this

m = 0.

magnetic potentials

seems to have drastic consequences. The electroV and A satisfy an equation of the type (4.6) with

Therefore, for the fluctuation in

V we

obtain

AP~b

which

is

enormous

if

5 That is to say, the potential e\^b created by (4rr/137) charge e the elementary charge at a distance b is much less than the quantum
.

we keep

in

mind

that in our units 1 the elementary

One fluctuations of the field averaged over a comparable region. might wonder how, in these circumstances, the electron in a hydrogen atom can possibly follow the orbit dictated by the force of the

~b~ l = me2

proton.

The answer
,

is

that

137 times the frequency of the They merely cause a small-amplitude, ground vibration of the electron, whereas the Coulomb field acts high-frequency for relatively long times in the same direction and dominates the motion. We easily2 find that the amplitude of this vibration is less than the

which for b

~ 10"

most of the fluctuations have a frequency


8

cm

is

electron in the

state.

Compton wavelength of

the electron, 10~ n cm.

Therefore this effect

displaces atomic levels less than relativistic effects, which spread the charge of the electron over a region of the size of the Compton wavelength.
theless, the present fluctuations in the
*

This will be shown for scalar field particles in Chap. 5. Neverexperiments establish the influence of the quantum 4 hydrogen atom with an accuracy of 1 part in 10
.

We remind the reader that h =


See

W.

Thirring, "Principles of

Quantum Electrodynamics," Academic

Press,

Inc.,

New

York, 1958; and T. A. Welton, Phys. Rev., 74: 1157 (1948).

FIELDS

29

In the mesodynamic application which we shall discuss, the vacuum fluctuations of the field will be particularly important, because the meson-nucleon interaction is much stronger than e. The fluctuations
constantly shake the spin and charge of the nucleon, since the pion field acts mainly as a torque on these variables rather than on the position of
the nucleon to which
it is

4.2. Derivation of Field

coupled. Equations from a Lagrangian.

We

shall

now

study the form of the commutation rules for the continuous This will give us a clue to the general quantizavariables <}>(r) and <f>(r). tion rules. and (4.9), we obtain Using (4.8)
[#r,0,

[#r,0, <Kr',O:Uf =
#r',0:U
fact

[#r,0, #r',O]<=,<
i?(*

(4.16)

')

Here we have used a

namely, that the following

< L/2.

If

For the

known from the theory of Fourier expansions, sum is effectively a d function for L/2 < x limit L -> oo we have
k

^=

d\r)

(4.17)

This expresses the completeness of the exponential functions and is the continuum analogy of (1.5). That (4. 16) is the continuum form of the canonical commutation rules was to be expected. In fact, the limit for

*->Oof(3.2), [?,,/U

Im ,is

[(*), X*')]

b(*X ?(*')]
same

a
lattice
is

where

6X

^ equals

if

x and
limit

x' are in the

space and equals

otherwise.

For the

a ->

the ratio d x ^/a


x'),

sional Dirac d function, d(x

and

(4.16)

is

just the one-dimenthe three-dimensional

generalization of this form. can now state the general rules for quantizing a field with the aid of the formal tools of the functional derivative and the d function. It is

We

convenient to start with the Lagrangian, from which we get the equations as the stationary properties of the action integral,

field

where
If

&

is

the Lagrangian density.


I.

With the boundary conditions


ed., p. 52,

See, e.g., L.

Schiff,

"Quantum Mechanics," 2d
York, 1955.

McGraw-Hill

Book Company,

Inc.,

New

30

FREE FIELDS

that the arbitrary variations 6<f> be zero at / t and / 2 functional derivative introduced in (1.13), we find

and by means of the

0+

The Euler equations of motion are therefore

vt

wW

UQ)

By means of the

functional derivatives

the Euler equations take

on
o

the classical

form

dL

dL
(4.18)

The generalization of the canonical conjugate variable

is

dL

For the Hamiltonian and the commutation relations we postulate accordingly the general formulas

[Xr,0, iKr'.Ol-i' [#r,0,

^O]

f . f

= ^V - r') = [nfr,OXr',01. r =

(4.19)

Hence the transition from discrete to continuous variables is simply done by replacing sum by integral, partial derivative by functional derivative, and Kronecker d by Dirac d.

FIELDS

31

The

field
1

equations and Hamiltonian (4.6) are derived from the

Lagrangian

KO =#r ||> - (V^) - mV ]


2

(4.20)
rr(r,r) is

With

the aid of (1.13), the canonical conjugate field

seen to be

and hence this prescription leads and commutation rules (4.4) and

to the
(4.16).

form of the Hamiltonian

(4.6)

The use of a mechanical analogy to find the field equations (4.6) may not seem very convincing when applied to, say, the pion field. To do this more systematically, the Lagrangian formalism is essential. To satisfy Lorentz invariance, the Lagrangian density, e.g., the quantity under the integral in (4.20), has to be a scalar. Our expression is, in
fact, the

most general

scalar

which

is

quadratic in the

field

and

its

derivatives.

As a further example, which we shall occasionally use to contrast with the relativistic field <, we apply the Lagrangian formalism to a field y which obeys the time-dependent Schrodinger equation. This equation is of the first order in the time, but since the field is not hermitian, f f y> the two equations for y and y> are equivalent to one equation of Y>

The application of our rules to equations of first order some care. To see this, we revert temporarily to the study of a The equation of motion with o = 1 for the single harmonic oscillator.
second order.
requires
real operator q, q
q,

can be rewritten in terms of two first-order


nonhermitian operators
iq

differential equations for the

and

J =q J ^q +
f

\q

j= J & = _/J
/

recognize that a hermitian Lagrangian which gives usual energy is

We

= J
/

and the

L= TT

The conjugate

variables

and

rr

are

^^L^^L^!
63
and hence the Hamiltonian
is
f

^t

~ =

4
2

^^ &
Jfaf

H - 7TJ2 +
1

7T

J -

L- \& =
it

+ f) +

\
total time

This Lagrangian

is

not unique, since

can be changed by adding a

derivative.
2

J and

Jt are independent variables.

32

FREE FIELDS
rules derived

However, the correct commutation


[^,.2]

from

[q,q]

are 1

[ 5, j]

= [&,&] =

and imply
[*,ir]

=i

[JBV]=J
in the

The

factor of

which appears

commutation

and ir 1 will always 2 be and * and canonically conjugate operators if a hermitian Lagrangian is used to derive first-order equations present
of motion for a nonhermitian
field.
f
y>

&

relations for the


"

A suitable hermitian Lagrangian for the Schrodinger fields y and


=
and the
field
I

is

V - V V) it

"

m Vy Vy]

(4.21)

equations derived from


iw *

are 3
2

V --w
2m

t
'

=
2m
canonically conjugate to
7T
f y and ^ are

The momentum operators


77

?r, T?

W'

2^
By means of the commutation
O(r),
77(r)]

-- y
2^

rules

(V(r),

^ f (r)]

(r

r')

(4.22)

WrX^fra^C^WXrH^O
we
therefore find

Wr.O.yV/fl/.^^r-r') = [y(r,0, vfr'.O]-r =.|V(r,0, vV,O]^r


1 have chosen a hermitian Lagrangian because it is identical with the conventional one for a harmonic oscillator. By adding a term /(*//<#)( tjg) f we can start from a nonhermitian Lagrangian which leads to the correct equations of *= /. motion but to the commutation rules [&,&] = 2, [J,7r] See, e.g., 0, [J2t f7r t]

We

Schiff, op. cit., p. 348.


2

one described by
3

This applies, of course, only to the type of system under consideration, namely, linear differential equations, which are of first order in time. This case is actually a limiting case of the Klein-Gordon field,
y(r,/)

iim

FIELDS
Finally, the

33

Hamiltonian
I

is

H=
tonian.
It is

d*r

2m

Vv>

Vw

==

cPr

2m

(Vw*

VTT*

VTT

Vw)

field equations from the Hamilonly when the canonically conjugate momenta TT, TT? appear explicitly in a hermitian Hamiltonian that the relations (1.14) The eigenvalues of the energy can be give the correct field equations. obtained by the same manipulations as before: 1

Care must be exercised in deriving the

The Schrodinger
relativistic

field

case

is

actually

somewhat simpler than

the

Klein-Gordon one;

energy and momentum of next chapter we shall study some differences between the relativistic and the nonrelativistic case which are not of a trivial kinematical nature.
1

in particular, the relation between the In the the field quanta is the classical one.

Since y

is

not hermitian, the creation part with a[

is

not needed.

CHAPTER

Observables

5.1. Energy, Momentum, and Angular Momentum. Led by our mechanical analogue we have so far investigated only two observables: the total energy and the field amplitude <f>(r t). In the continuum limit the latter was not an operator in the sense that, when applied to a state of finite norm, it leads to a state of infinite norm [see (4.14)], so that we shall need some other observables for a discussion of the physical properties of our quantized field. There are some general recipes in classical field theory for constructing quantities such as the linear momentum or the angular momentum of a field from a given Lagrangian. These and other observables, together with their commutation properties, will be studied in this chapter, and the next one will be devoted to the eigenstates of these operators. As in point mechanics, the invariance of the Lagrangian under certain transformations ensures the existence of corresponding constants of the motion. We have already encountered one example of this general principle, namely, the energy. If, and only if, the Lagrangian does not depend explicitly on time, then the energy (4.19) is constant. The reader will readily verify the formula
9

t-t
where dL/dt does not involve the implicit time dependence through ^(r,r). Similarly, if L does not depend explicitly on the coordinates r those Lagrangians studied in Chap. 4), which means that it is vu(i.e., invarianf under displacements and rotations in space, then we get six more constants of the motion, one for each parameter of the invariance group. In classical mechanics the constants (which for displacements and rotations are the total momentum and angular momentum,
34

OBSERVABLES

35

respectively) associated with invariances are simply the generators of the transformation. The invariance of the classical Hamiltonian under

such transformations ensures the vanishing of the Poisson bracket between the Hamiltonian and the generators, which implies that the latter are constant. The same holds true in quantum theory, where the Poisson bracket is replaced by the commutator. Therefore, the linear and angular momenta are generally defined to be those operators for which the commutator with any quantity gives its change under an
infinitesimal displacement
is not yet invariant under rotations, because of the cubic periodicity condition The invariance is obtained, however, (4.8) we imposed on our fields. by imposing a spherical boundary condition, e.g.,

At

this point

we have

and rotation. to remember that our problem

<(r,f)

for r

(5.2)

We

shall

have

this
It
is

condition in

momentum.

shown

discussing the total angular in the classical study of solids that the

mind when

particular form of the boundary condition is unimportant for large systems and only serves as an aid for the mathematical development. The case of physical interest is the one with L -> oo or R ~> oo. Correspondingly, the form of the boundary condition should not, and does not, enter into results of physical significance which correspond to volume and not to surface effects. The physical results will always be deduced with states wherein the field is only excited in finite regions of For these states the field operators at infinity are effectively space. With this in mind, we shall henceforth also neglect surface zero. integrals from infinitely remote surfaces. For the relativistic and nonrelativistic fields, the total momentum and

angular

momentum
P
=,

turn out to be 1

=L

(5.3)

d*r

X
X

JJ

= -i fr [Trr X

Vy;

-f-

1 These operators are restricted by conditions of hermiticity and proper behavior under Lorentz transformations. They can be obtained by analogy with classical mechanics. See, e.g., G. Wentzel, "The Quantum Theory of Fields," p. 8 and Appendix I, Interscience Publishers, Inc., New York, 1949. Here we shall merely give P and L and show that they have the correct properties associated with such

operators.

36

FREE FIELDS
aid of (4.19)

With the

we

find, e.g.,

[P, #r,f)]

and
of

similar equations for

[L,#r,0]--=irX V#r,0 That is to 7r(r,f), V etc.

say, the

commutator

operator gives the change of that quantity under an infinitesimal displacement and rotation, respectively. .Since the Hamilton ian is invariant under these operations, we have

P and L with a field

[P,//]-[L,H]-0
fact, (5.5)

(5.5)

which, because of (2.19), means that P and L are constant in time. can also be verified with the aid of the field equations.

In

We

can be converted to infinite readily see that the expressions for P and surface integrals by means of the Klein-Gordon equation. However, a calculation shows that P and L fail to commute; in fact, the simple

commutation

relations

between them are the same as in elementary

quantum mechanics,

= 1, 2, 3, or x, y, z ic iik Pk /, j, k (5.6) [P,,LJ where ijA is the totally antisymmetric tensor of third rank, e 123 being 1 There is actually a very general and e 213 being 1, for example. reason for (5.6). Since L and P generate infinitesimal rotations and displacements, the commutation relation between them must be the same as the one for the operations of rotation and displacement.
.

Similarly, the commutation rules of the out to be of the usual form


[L.,,L,]

components of

are

worked
(5.7)

/6,,,L,

Inserting the expressions (4.8) or (4.23) into (5.3), usual methods,'

we

obtain, by our
(5.8)

P= 2 Xk k

As we found

earlier, the

operators ala^ have integer eigenvalues, and


" kl >'W

this tells us that the state


ti,n.,
-

>

^
|

>

(alr (al)

n*

'

'

'
\

0>
2

is

also an eigenstate of P

and belongs to the eigenvalue


is

n^ + n k
=

2 ~h

thus supported by (5.8), which states particle interpretation 2 2 that a momentum k is associated with the energy o> (k -f)Mf The eigenvalues for the momentum are therefore the integer multiples

Our

The zero-point momentum


*

" J^ ^, k

which appears for the Klein-Gordon


.

field, is

zero by symmetry, because for every component k i9 there is one kf If Because of this, the a k are usually called the particle-destruction operators in momentum (or k) space, as opposed to the ^(r) in coordinate (or r) space, which both create and destroy particles.

OBSERVABLES

37

of the k. This resemblance to the energy eigenvalue problem has formal origin in the fact that the commutation relations (4.19) and (5.4) have the same structure. Consequently, the possible values for the angular momentum are also of the same nature. However, our standard states, which are eigenstates of P, will not be eigenstates of L, since P and L do not commute except for eigenstates with P = 0. But L and commute, so that we should be able to find simultaneous eigenstates of, say, L 3 and H. To construct such states, we should not expand in terms of plane waves (eigenfunctions of displacement), but rather in terms of spherical harmonics (eigenfunctions of rotations), since it is in the latter representation that we expect L 3 to be diagonal. To accomplish this
its

objective,

we

use the plane-wave expansion


e
ik ' T

= 7 2 (2^ Ui(r)Yr\O K l,m


l

k9 (pk

)Yr(O rt <pr)

O r , q> r are the angles between the vectors k and r and an z axis and where Ytm is a normalized spherical harmonic. arbitrary The functions Vk (r) satisfy the equation

where

Ok ,

yk and

'and are given by

U k(r) =
l

The expansion

(5.9)

reduces to the familiar expression 1


0)

r is taken to be in the direction of the z axis. However, since it a rotationally invariant expression, it must hold generally. The constants which appear in the definition of U% have been chosen such that these functions have ^-function normalization. This can be seen by the use of their asymptotic behavior
is

when

=
1

s<L>

r-x

Hm 4( r) ~ (2? sin \7T/

( kr
\

_ !?
2

This expansion is most easily obtained by comparing the asymptotic expansions of both sides after integrating with />,(cos 0) </(cos 0). See G. N. Watson, "Theory of Bessel Functions," rev. ed., p. 128, St. Martin's Press, Inc., New York, 1944, and
L.
I.

Schiff,

"Quantum Mechanics," 2d

ed., p. 77,

McGraw-Hill Book Company.

Inc.,

New

York, 1955.

38

FREE FIELDS
field

and of the

equations

f*dr

l r*Vk(r)V k ,(r)

^-L_ j*dr [vk(r)


,.

= hm
B-*oo

1 sin * - - R(k
7T
fc

-- =
-j

vk .(r)

- v{.(r)
,

/c') -

,.

,.

d(k

k)
set

fc'

Furthermore, they and the spherical harmonics Yf* form a complete of three-dimensional functions in the sense that

^=
Finally, if

3
<5

(r

- r')

we introduce
m

the expansion 1
to '

#*>*>

Jo (2a)Yi

"A ^ ^)[yWr.^lm(^"
t

rrU^
(S.Wa)

then, by comparing with the continuum limit of (4.8), a lm(k) is defined by

we

see that

With

the help of

we

thus deduce the commutation relations

=
We
9

(5.H)

can readily verify that (5.11) results in the correct commutation properties of </>(* t). In analogy to our plane-wave expansion for <(r,0, we may a ^ so derive the continuous A>variable development given above as the limit of a From discrete set which is selected by the boundary condition (5.2). the asymptotic expansion of v, which is valid only for n or / much less than kR (e.g., for fixed n and / in the limit R -> oo), we see that

U k(R) =
requires k tions we see that

= (n + Ityir/R

with n

= 0,
R

1, 2,

Under

these condi-

I-*-J
1

dk
shall

We

have here used a convention which we

keep henceforth.

It relates

discrete &-space quantities to continuous Ar-space ones. In the former, destruction operators are written as ak or aklm9 whereas in the latter they appear as a(k) or

OBSERVABLES

39

and we can rewrite the above development as

=2 k,i,m

^u

k
.

t i A & l>MXTm'] = <V<5mA*'

-i

(5.100)

with

a^ -> (^IK^a lm(k).

the discrete

and the continuum form.

In future developments we shall use bothTo obtain numerical results

from the theory, the latter is obviously more convenient. For easy reference, the relevant formulas relating discrete to continuum k space are collected in the Appendix.
In terms of the operators introduced in (5.10&),
servables of interest 1

we

get for the ob-

H-E -2
and for

4i.m**,i.>

(5.12)
n

1,m,k

L3

by making use of dYi*(pn <pdfi<pr

imYi

(Or) (p r),

we

get
(5.13)

L3
The vacuum
is

= I l,m,k
fl*.i,m

*.i,m*.i,mw

defined in terms of the


I

new

variables

by
(5.14)

0)

In the limit of

->

o>,

in

which case the boundary conditions

(5.2)

should be equivalent to the ones used in the momentum representation, (5.14) is identical with our old definition, since the new operators a are then linear combinations of the old ones. In the new representation, eigenstates of particles with energies to are obtained by applying al tltm
to the

vacuum.

Thus
|

0)

We recognize from (5.13) that these states will also be eigenstates of L3 m< t ,,, m ,. The H M e.g., the eigenbelonging to the eigenvalue

.,

i0*iiti

values of a\lm aklm , are the numbers of particles with energies eo t and that L 3 has the i9 so angular momentum lt with three-component
.

integers as eigenvalues.

To build up eigenstates of a given total momentum, we must properly combine the single-particle angular eigenstates we have constructed with the methods familiar from

2 We shall not go into this at present, but elementary wave mechanics. That there are shall carry out equivalent manipulations later. is connected with the fact several ways of constructing eigenstates of

we
1

In Z, 3 , a term

\m> which appears for the Klein-Gordon


m.
Lifshitz,

field, is

zero because of

symmetry between -f-/w and 2 L. D. Landau and E. M.

"Quantum Mechanics," chap. IV, AddisonWesley Publishing Company, Reading, Mass., 1958.

40

FREE FIELDS

that in the limit

~> oo or

-+

oo,

is

infinitely degenerate.

The

eigenstates of L 3 and //are just superpositions of eigenstates of P with the same eigenvalue of //, the coefficients being those which transform plane waves into spherical waves. 5.2. Parity. further constant of motion emerges from the invariance of under the noncontinuous orthogonal transformation of the coordinates represented by the reflection r -> r. Because this transformation cannot be generated by continuous rotations, 1 the constant associated with it, called parity, is independent of angular

momentum.
there

It is

stitution <(r,f)

-><(

deduced by the usual argument. Since the subr,f) leaves the commutation relations invariant,
effecting the substitution:

must be a unitary transformation

&+0>\
Also,

0>\0> +

(5.14a)

is

invariant under the substitution <(r,0 -> <(

r,0, so that

we

have

=H

[& + ,H]

which implies that 0*+ is a constant. However, both commutation relations are also invariant under ^(r,f) -> that one can also define a reflection
^-^(r,*)^:
1

and the

<t>(r,t) 9 so

-<K-r,0
there
is

(5.146)

an interaction can we tell which is the right reflection property of 0, that is to say, which of the two includes a term operators & is a constant. For instance, if

and

^_

is

also constant.

Only when

where
with

H and

p(r)

is
<f>

is

invariant under reflections, then only + commutes then called a scalar. On the other hand, a term

d 3 rp(r)a-V</>(r,t)

with

2>-l

commutes only with 0L, and


latter case is realized in

The operators rather than in a


==

is then called a pseudoscalar. nature by the pion field.


<

This

momentum
Since

can be diagonalized in an angular-momentum = but representation since [^ ,L]

-P.

This means that

^ has no classical analogue; there is no infinitesimal generator

for a reflection.

OBSERVABLES

41

we

find as explicit expressions

exp

[- in(l
klm

l)al lm a klm -]

(5 14c)
'

and from

these expressions

we deduce

From (5.14c) it appears that + is 1 raised to the number of particles I raised to the number of with odd angular momentum, and ^_ is with even angular momentum. Hence parity is a multiparticles plicative quantity; for several particles it is the product of the individual Scalar particles have only their orbital parity ()*, whereas parities. pseudoscalar particles also have an intrinsic negative parity. 5.3. Number of Particles and Particle Density. Another observbut has no analogy in classical mechanics able which commutes with 2 is the number of particles

&

= 2
klm.

(5-15)

Its

eigenvalues are the sums over the integers n k (or n klm ) which we interpreted as the number of particles present in a state with momentum k (or angular-momentum z component m)
:

Thus

present.

can be called the operator for the total number of particles We obviously have
[//,AT]

(5.17)

which means that no

In fact, if we particles are created or destroyed. define the operator for the number of particles of a given momentum k as ala k , so that k9 then we find that k

N =

N = ^N
k

Note that

e i([1 ae~ iG

is

defined by expanding the exponentials


/2

21

so

that, if fotft]

1,

e-i^aae ina-a
,

__ a
is

The phase
0*
|

factor in 0*
1

which

is left

0}

0).

This equation and related equations given later are always to be understood in the limit L -* o> or R -* ex?.

By means of (5.140) we can

open by (5.14a) and (5.146), 1 also compute ^ ak^^.

chosen by

a~ k

42
This

FREE FIELDS

tells us that no particles are transferred from one momentum state to another; in other words, no particles are scattered by the Hamiltonian we have been considering. Equation (5.17) will no

longer hold for systems that we shall consider later on. Like the operators considered previously, can be expressed as a

volume

integral.

If

we decompose

</>

into a positive-

and a negative-

frequency part,

*'"*

-? I
(5.18)

we

find
JV

-i
this

In the nonrelativistic limit

becomes for the Schrodinger

field the

more

familiar expression

=jd*x

y'Or.OvfcO

(5-20)

In elementary wave mechanics this is put equal to 1, which means, in our present language, that there we consider only one-particle states. We can also show that Ehrenfest's theorem 1 of wave mechanics holds in our general theory. If, in analogy with wave mechanics, we define the center of mass by the operator 2
(5.21)

NJ
we
obtain, with the help of (4.22), (5.3),

and

partial integrations,
(5.22)

R=
1

i[H,R]

= -Nm
p. 25,

See L.

I.

Schiff,

"Quantum Mechanics," 2d ed.,


in elementary

McGraw-Hill Book Com-

pany, Inc.,

New York, 1955. 2 Note that [R Pj] = id ii9 as


it

wave mechanics.

OBSERVABLES

43

Thus

momentum is equal to the total mass multiplied by the of the center of mass. The relativistic analogy of (5.22) holds velocity only for the center of energy,
the total

m 2 ^ 2]r
R=
5.4.

(5.23)

(5.24)

Local Observables. The observables considered so far were of This suggests interpreting the integral over all space. integrand as the corresponding local density and an integral over a finite volume as that part of the observable contained in this volume. 3 However, the quantities integrated over the whole volume L may fail to commute with or bilinear operators such as the momentum density P(r), and hence the states considered so far will in general not be This eigenstates of local quantities such as the momentum density.
the

form of an

<f>

become quite clear in the next chapter, in which we consider states. With respect to local quantities, there is an important difference, which we shall now consider, between the relativistic and nonrelativistic case. If we define the number of particles in a volume v as
will
JlV

(r,0

(5.25)

then, for the nonrelativistic field considered,

N(r,0

= vfaOvM
(+)
(r,f)

(5.26)

and for the

relativistic one,

AT(r,0

-i[^->(r,0*
it

- ^(r.O^fr.O]

(5.27)

In the nonrelativistic case,


(4.22) that

follows from the commutation relations

[AWO,

IWl-r =

(5.28)

whether the volumes v and v z overlap or not. This means that we can talk of a definite number (e.g., 1) of nonrelativistic particles in a volume of any size, no matter how small or how large. It is true that this number does not remain constant, since

=2m
and the surface
v,

Jv

integral to which this reduces is finite for finite volumes but this only means that the wave packet for a localized particle

44

FREE FIELDS

spreads out as time goes on.

For the

relativistic field

<f> 9

the

commuta-

tion relation (5.28) does not hold even if the volumes v and v 2 do not 1 , which spoils the overlap. This comes about because of the factor

vanishing of the commutator

|>
u
/
/

(+)

(r,0,
.

^-'(iV)]^,'

=S
k

= iA (+) (r ~
2co

r')

(5.29a)

when r ^ r The behavior of the commutator can be found

as follows
sin
2
(/c

kr

+ m 2)*

Substitution of

A:

= m sinh
!

gives
<

= -

TT r Jf r dr
47r'

"" Ul '

d6

=~

T~ f 4?rr 9r

=
We
of the
first

m -477T

Tr d)/. x f/i (jmr)

see, therefore, that the

commutator behaves

like

a Hankel function

1 (1) kind, // (/>w), which has the following properties:

r-o

4-Trr \7rrn ri

Thus, the dominant behavior for asymptotically large distances arises in ikr evaluated at the complex pole k = im. (5.29Z>) from the exponential e we find for the commutator of the local density Correspondingly, N(r t) with that at another spatial point, r', but at the same time t
9 9

with

r (+) (r

- r') = = J- 1
47r
2
|

f r = r - r dk e ikr(k 2 -f m 2 )^ r3rJ-oo G.TSL Watson, "A Treatise on the Theory of Bessel Functions," 2d ed.,

chaps.

3, 6, 7,

Cambridge University

Press,

New

York, 1958. Note that

OBSERVABLES

45

Here we used the algebraic


[>1B,CD]

identity

A\B,C]D
(+)

AC[B,D]
#r',0]

[A,C~\DB

C[^,Z)]B

and

|>
integral for F<
f)

(r,0,

J/<%

r')

The

convergent by

diverges in its present form, but it can be made bringing down a sufficient number of powers of k in the
>

denominator by differentiation with respect to r. For large distances the dominant behavior of F< h is again determined by the |r r'|

k im. Thus, for the relativistic field exponential considered, the commutator of the local density N(rj) with that at
evaluated
at

Fig. 5.1. Distribution of mesons about two nonoverlapping volumes v l and z> 2 l separated by a distance s much larger than the Compton wavelength m~ of the
particles.

another spatial point but at the same time [e.g., N(r'J)} goes to zero m~ l The if the two points are separated by a distance |r r'j same statement holds for the number of particles contained in two

only

>

Here m~ l is the nonoverlapping volumes Vi and v 2 as shown in Fig. 5. 1 Compton wavelength of the field particle, and for the TT meson, for 13 Hence it is not possible to assert that one instance, it is ~10~ cm. other definite number) is in a volume the boundary of pion (or any which is defined within the order of 10~ 13 cm or less. That would be true only if this state were an eigenstate ofNv with eigenvalue 1 for this for all neighboring volumes particular volume and with eigenvalue within 10~ 13 cm. Because of the noncommutativity of such closely neighboring V9 this is impossible. The best we can do relativistically 1 is to have v for which v is many w apart from eigenvalue for those this is connected the volume which contains the particle. Physically, with the fact that defining the boundary so sharply, Ar < m~ l requires that there be an external field partially composed of wavelengths <nr l Such a field is capable of creating new particles. Because of the identity of particles in field theory, the new particles cannot be distinguished from the old ones. Hence the state will cease to be a one-particle state.
,
.

46

FREE FIELDS

2 It appears that the fundamental principles of relativity (E = me ) and quantum theory (E = hv) give an important modification to our concepts of particles. Whereas in the nonrelativistic limit they appear as points and there is no lower limit to the size of the region into which they can be confined, in relativistic field theory the quanta of the field have roughly the size of their Compton wavelength. This is the origin of the decrease of the electromagnetic interactions when wavelengths

<

w"" 1 are involved.

m~ l , and the effect of smaller wavelengths is sphere with radius r averaged out. Hence the cross sections for scattering of photons by 1 .1f electrons decrease for photon wavelengths Similarly, this effect decreases the binding of the hydrogenic S electron, since its size does not permit it to take ftill advantage of the narrow singular part of

An

electron, for instance, acts like a charged

<w~

the

Coulomb

potential.

Summarizing, we can say that the behavior of observables in quantum The question field theory is like that of an ensemble of free particles. of the size of the particles and other features of local quantities will be
further illuminated
If

when we

discuss typical states in the next chapter.

See

W.

Inc.,

New

Thirring, "Principles of

Quantum Electrodynamics," Academic

Press,

York, 1958.

CHAPTER
States

The states we have been 6.1. Vacuum and One-particle States. mainly interested in so far have been eigenstates of the energy. The state with the lowest energy, 0), has no particles and, appropriately, is called the vacuum. Application of any of the a[ to 0) creates a state with one particle present with momentum k. The most general one-particle state is obtained by multiplying 0) with a general linear w^h different values of k. This can also be combination of operators done by means of the field variables (r,t) of Eq. (5.18) or, in the
| |
|

<f>

nonrelativistic case,

by

ip*(* 9 t).
first,

Treating the latter and more familiar case


t

we can

write
(6.1)

v (r,0|0>

We

note that our previous one-particle states a^

0> or a]am

0} are

special cases of (6.1), with

or
since these states are time-independent if they are eigenstates of the Hamiltonian. The normalization of the one-particle state (6.1),
(1
|

1)

1,

requires

= =

(0

|J/*(r,

3 ,Ov(r,0 d

r/(r',Ov,0

0)

JVr/*(r,0/(r,f)

Jd

r/*(r)/(r)

(6.2)

47

48
This
is

FREE FIELDS
the normalization condition for the

wave function

in

wave

It appears plays the role of this quantity. whenever expectation values of a quantity like the energy density //(r), the momentum density P(r), or the density of the number of particles

mechanics, and, indeed,

1 N(r) are computed.

Thus

(1

N(r)

1)

(0

f /*(r')y(r') J

dV

t T

ft
J

y> (r)y>(r)

/(r")y> (r")

dV

OX

<1

H(r)

1>

=
2m
2m

(0

f/*(rW) <*V ^Vfr)

Vy(r)

|/(r>V) <*V

<>>

and, similarly,
(1
|

P(r)

1 >

[/*(r)V/(r)

- /(r)V/*(r)]

(6.3)

shall now investigate whether the field quanta can be considered particles in the sense that they are objects localized in a certain region in As in wave mechanics, we can at a certain time have a particle space.

We

density with an arbitrary spatial distribution. To be sure, such a state is not, in general, an eigenstate of energy and momentum, but this may also be true in wave mechanics where a localized wave packet eventu-

Our nonrelativistic particles need not have a finite size at a given time /, since we can have a state for which/(r,0 is different from zero only in an arbitrarily small region [e.g., /(r,0) = <5 3 (r)]. In this case the expectation values of all densities will, according to (6.3), be zero outside this region. We see from (4.22) that such a state is even an eigenstate of densities outside this region belonging to the eigenvalue This means that there are states for which, outside a region as tiny 0. as we like, no experiment will find any trace of a particle. Nevertheless, we shall always have J = |1) (6.4) JV|1} = ^(r) 0} is easily seen to be an In particular, at t = 0, say, 1 } eigenstate of v (although not normalized) belonging to the eigenvalue 2 1 if v contains r and to To show this, we use eigenvalue if it does not.
ally diffuses.
,

r'

3
<5

(r

r')

which also proves that

has integral eigenvalues for arbitrary vol-

umes

v. *

1 //(r), P(r), and N(r) are the integrands of the corresponding integrated observ0. ables, evaluated at t 2 and it will thus be This state is not an eigenstate of H, since [y^r,/), H]

time-dependent.

When no

time dependence

is

indicated for

./V^,

we mean

STATES

49

case the

behave differently. First of all, in this not even an eigenstate of the local densities //(r), ~ )2 These quantities contain terms proportional to and P(r), or L(r). therefore lead from the vacuum to a two-particle state. Nor is the vacuum expectation value of //(r) equal to zero
relativistic field states

The

vacuum

is

<

(0

H(r)

0}

--=

i(0

<jt

4-

(V^)

+ mV* - i (0
<
|

0}
(r)<

(+)

+ V^ (+) (r) + mV+Vtf^tt 0>


(

->(r)

We may, however, redefine the densities so as to ensure the vanishing of


their
<

vacuum expectation

values.
</>

This
(

is

<+)

operators to the right of the

accomplished by writing

all

ones, e.g.,

(6 5)
'

so that the
p(r)

momentum
(

density, for example,

becomes

= -ti

)2 should be noted that this does not take care of the terms, so that vacuum is still not an eigenstate of local densities. However, the the above rearrangement does eliminate the zero-point energy for E. These

It

<

change the observables by ordinary (although infinite) nonmeasurable numbers. Furthermore, these numbers are real, so Henceforth we that the hermiticity of the observables is not destroyed. shall always assume ordered products for observables quadratic in This does not mean that the vacuum fluctuations of $ vanish. Thus 2 2 0} is still given by (4.13), and the fluctuation in the (A<) = (0
alterations only
<f>.
<
| 1

energy density
different

//(r),

with the reordering for


2

(0 // (r) 0>, and diverges even faster zero, [A/f(r)] than that of the field operator <. That is to say, in a relativistic theory

from

H (but not for H

), is

also

we are never sure that the local energy is zero. This arises again from the fact that the accurate definition of a volume requires high momenta
and energies which, in a relativistic theory, may create particles. As we go along, we shall notice that the virtual existence of particles throughout space is a most striking feature of this theory.

The one-particle states of the relativistic theory also present interesting To write our arbitrary linear combination of the creation features.

50

FREE FIELDS

operators a[ in

momentum

space,

fl

k/k

n terms of the variables $


its

we have
because

to introduce

Fk =

k (2o>)*/

and

Fourier transform F(r),

w)*

L* J

LWfry* ^

Thus, at

= 0,

we have
I

1 >

= 2 Aal
k

0>

f <* r F(r)^->(r)

0}

(6.6)

The normalization condition

does not correspond to the usual jW3r


r

|F(r)|

but rather to

<0

F*(r)F(r')</>

M
(r)<f>

-\r

0)
(+)
(r

This
is

i
J

F*(r)A

- r')F(r') d*r d*r' =

due to the factor <w, which, even with the redefinition (6.5), makes impossible to find a spatial distribution F(r) such that the expectation values of ail densities are zero outside a certain region. For instance,
it

1 k putting all/ equal to 1/L , corresponding to a spatial d function at but not for F(r), we obtain for the particle density, Eq. for/(r),

(5.27),

=S*
k

tk

VS
k'

e"*''

"}
(6.8)

This is clearly not equal to zero for r 0, so that the one-particle state so described cannot be considered localized; it is, rather, spread out over a distance of the order of l/m. On the other hand, if we choose Fk equal to 1/L 2 corresponding to F(r) = (5 3(r), then we find for AT(r), by means of the commutation relations (5.18) and (5.290),
,
1

N(r)

1>

= =

-i<0
3

+ +\Q)ti
(+)
(r)

-}
(*)4>

(T)

^\^

(+

\r)-]<t>-^

0)

J[(5
3
i<5

(r)A

(+)
(r)

(r)A

STATES

51

This

is

indeed zero for r

0,

so that the state appears to be localized at

the origin. 1

However, density becomes

in this case the expectation value of energy

4-

[V#r)]
2

+
The dominant
r

[VA

(+)
(r)]

+ + [mA(+) (r)]

mWO)
2

0>
(6.9)
-

spatial dependence of this function is given by This behavior can be understood by considering a single Any measurement of the energy creates particle located at the origin. but these have already been accounted for by the redefinition particles, This ensures that, in the absence of real (as opposed to virtual) (6.5). particles, e.g., for the vacuum state, the expectation value of the energy density is zero. However, because of the presence of the particle at the A pair of origin, something new can happen if we measure nearby. l particles may be created at a distance r < m~~ 9 one of which stays there,

>

e~ Zmr for

whereas the other annihilates the particle at the origin. The distance 1 Since over which virtual particles can spread is limited by AE AT A > 2m for creation of a pair of particles, AT < (2m)' 1 and the l Therefore the above particle cannot propagate further than (2m)~ of event will influence the energy density within distances of the type order of m~ l about the origin. By similar calculations we recognize that v has only integer eigenvalues if i?* has an extent much larger than the Compton wavelength of the field quanta. We find that [A^^-^r)] + ^(r) but contains an ~ additional term proportional to ^ averaged around r within a

Compton

wavelength:

~
Hence ^^(r) 0}
|

I <A

~V)A

(+)

Or

r')

dV]

if r is in v

f
<

-,

- r') dV

if

is

not in v

2 Jv
will not be an eigenstate of Nevertheless, if v is v than nr 3 and F(r) is a smooth distribution in a volume larger >m~ 3 about the middle of v 9 then JW 3r FOO^-^r) 0} will almost be an Since the local energy densities at different v eigenstate of this r , it if r but at the same time commute, 2 [//(r), //(r')] == points
.

much

1 Our choice of F(r) does not lead to a normalized state 1 >. that the state described by F(r) = <53(r) is not an eigenstate of
|

Note, furthermore,

Mr),

e.g.,

N(r)

1)

constant
|

1)

for r

consistent with our discussion in Chap. for this state is localized.


2
its

5.

However, the expectation value of N(r)

Note that the ordering of H(r) changes it only by an ordinary number, so that commutation properties are not changed.

52

FREE FIELDS

might seem possible to build states for which the energy is exactly This can indeed be done, but then it can be shown that Local these states do not possess a definite number of particles. quantities and the number of particles are, therefore, complementary
localized.

concepts. 6.2. Two-particle States. In order to get a feeling for some of the consequences of the Bose-Einstein statistics of the field quanta, we shall

For the sake of finally study interference and fluctuation phenomena. we shall keep to the nonrelativistic field. The results for the simplicity,
relativistic field

$ are

similar, but are complicated

by the additional

effects discussed

above.

The

interference effects already appear for the two-particle states.

Nonrelativistically, the

most general
</\

state

of

this

type

is

2)

/(r l5 r2

)y(r(r

2)

0)

(6.10)

and belongs
(6. 10)
is

to the eigenvalue 2 of TV.

The normalization condition

for

<2

2>

=
in

frt /"(r^C/fc.*)

+ /(r^)] =

(6.11)

The second term


the field quanta.

In fact,

(6.11) arises because of the Bose statistics of we could have restricted ourselves to a

symmetric/
r2
is

in (6.10), because the part of/ which is odd in ^ and does not contribute. Since y^fo) and ^(i 2) commute, ^ t (r 1 )^ t (r 2) even in r x and r 2 and gives zero on integration with an odd func-

tion.

There are some peculiar features connected with these facts which are shall do best illustrated by calculating the expectation value of N(r). this for a two-particle state for which /(r^Fg) is of the form f&Jftfa),

We

so that the particles would be independent could they be distinguished. 2

Furthermore, we assume that each /(r,) is normalized to unity, e.g., 2 3 jW r,/(r,) = 1, so that the correctly normalized over-all distribution function /is
f (r r\ /(ri ra)
'

with
^
1

(//,)
is

frtftoftf)
it is

(6.13)

If this state
if it is

an eigenfunction of the Hamiltonian, then

time-independent,

but

0. In the not, then the state function is given by (6.10) only at t following discussion, we shall consider the latter case. 2 Particles which can be distinguished will be encountered in the next chapter.

STATES

53

With

this

notation

we

find
<?rt

<2 |AT(r)| 2>

JX
x
[i

which is obtained by commuting all y's to the right and y^'s to the left and using if 0) = (0 v|t = 0. We see that if the two wave functions
\

are orthogonal, (/i,/2 ) individual ones,


as for

= 0,

the particle density


, , ,

is

just the

sum of

the

a|N(r) 2>B= /l(r) |. + /l(l) ,. independent particles and as shown in Fig. 6.la.

In particular,

ft (r)

Mr)

(b)

Part a Fig. 6.1. Interference effects in the two-particle density distribution. two noninterfering particles. Part b applies to the case of two particles
interfere.
|

shows which

The

The density |/i| 2 4- |/2 2 is that which would apply for no interference. actual density is <2 \N(r}\ 2) and shows that the particles tend to cluster in the

interfering region.

54
this
is

FREE FIELDS
the case in the classical limit of nonoverlapping

wave packets, for

particles can be identified by following their trajectories. there is an interference term, which decreases However, if (/i,/2Xj| the density wherelil/'s do not overlap, because of the denominator in (6.14), and hence increases it on the average in the overlapping region. This is demonstrated in Fig. 6.16. Because of this property, bosons have a natural tendency to stick together. Hence, we see not only that

which the

^90

^90

180

180

180

Fig. 6.2. Polar plots of the positive amplitudes /i(9) and/ (8) for hypothetical a-a 2 scattering are shown in part a. The intensities without and with interference appear in parts b and c, respectively.

single particle

can interfere with

itself,

which

is

the usual super-

position principle in
particles

wave mechanics, but

also that

two

identical

This interference, which is another expression for the symmetry requirement of the wave function, does not occur between particles of different fields (see Chap. 7) and emphasizes that identical particles are just excitations of the same field. One important consequence occurs for the scattering of two identical There the scattering intensity is not just the sum of the inparticles. tensities for the two particles but includes an exchange term of the type displayed in (6. 14). For example, if in the scattering of alpha particles 4 by Hfe the center-of-mass-system amplitude/i(0) is peaked in the forward 2 2 direction, then/ (9) must be peaked backward, but |/| 2 2 |/i may then be anomalously large around 90, as is shown in Fig. 6.2.

can interfere with each other.

+/

STATES
1

55

6.3.

Many-particle States.

Another unusual physical phenomenon

directly related to the above is the fluctuation in the number of bosons in 3 a volume v Z, , where the latter volume contains a state with a

<

definite

number n of bosons. For independent particles the distribution of the number of particles in v satisfies a Poisson law, e.g., the
is

probability of finding v particles in v

v\

where

v is the

average number of particles in the volume,


*

= fw
v

(6.15)

For a uniform distribution of particles in our normalization volume, v would be nv/L?. For a Poisson distribution the fluctuation in the number of particles in v is
2

(Ai/)

= ^-

v2

=
V

*7v"

~v =
z

(6.16)

That

is,

in the local density are proportional to the density classical particle result.

for a normal uncorrelated distribution the square fluctuations This is the itself.

obtain from a complete field theoretic treatment ? be shown that even for bosons in orthogonal states (for which the particle densities are additive) the local fluctuations of the density of To derive this, we first calculate v for the particles are not additive.
It will

Now, what do we

w-particle state

which

is

represented by

n)

=
(6.17)

with 1

(//,)
the

dt ,

(6.18)

With

same methods

as before, this gives us

*=

(n\Nv
(n

\n)
f
\

= f fr
Jv
1

v>

(rMr) n>
|

= f d*r
*>v

2
j

=l

|/,(r)|

(6.19)

We

shall

now

under consideration from

take the /'s to be orthogonal in order to separate the effect trie one discussed above.

56
If

FREE FIELDS

each /is chosen to be a plane wave,/

= e*k

'*

/", we get
(6.20)

*-5
as

was

to be expected.

For the expectation value


I

Nj

=
Jv

d*r d*r' JV(r)/V(r')

=
we
find in the usual
<n

N,

+
3

Jv

\<Pr

rfV

vVtyVOvW')

(6-21)

way
f</ r
tV*:
/;

*; n>
|

=*+
JVJ

fr' f*(T)fi(r>)[fMMi>) +/^r)/XO]

(6.22)

and therefore
(AN.)'

<n

2
/V,
|

>

= v +%

f <Pr /*(*)/&)

i^k \Jv

l[ \#r \ffr)\*\* LJ V J
i

(6.23)

(n\N.\n) The first term in (6.23) is the result obtained earlier, Eq. (6.16), for an uncorrelated distribution; the two other terms represent the fluctuation due to the interference of the various particles. In the limit v = L3 the second term vanishes by our assumption (6. 18), and the last one is just n for which as expected, since we are then dealing with an eigenstate of 2 2 (AAO = 0. For plane waves the last term is n(v/L?) = vv \L? and

where

N.=

vanishes for v

constant,

L3

->

oo.

Let us study the fluctuations of the particle density in the two extreme cases in which the volume v is very much larger and very much smaller than the dimension of the volume in which the distribution

functions/, representing the particles, are different from zero. In both cases we take v to be much smaller than L 3 since otherwise the fluctua,

quanta are represented by wave packets with wavelength much smaller than t;*, as shown in Fig. 6.3a, then the second term drops out because of the orthogonality of/ and/ ^ ,. The field then has properties of a purely noninterfering system and therefore no wavelike behavior. In the opposite extreme of long wavelength, v ^ A 3 = &,~ 3 as shown in Fig. 6.3fc, and for plane waves we obtain for the second term
tions

approach

zero. 1

If the

fc

and the

fluctuations are

(AN,)

v(y

1)

We
1

note that for a low average density of particles, v Remember that in this limit the third term on the right-hand

<

1,

the particles

side of (6.23) goes

to zero.

STATES

57

behave like classical ones. To illustrate the significance of the added interference contribution (e.g., * 2 ), we need a large density of particles, v In this extreme the particles behave like a superposition of 1.

>

waves of equal amplitude and random phases,

The

intensity for the resulting


/

wave

is

Re

r2 e-'+> 2 e**\ =
7
2

+ 2 2 cos (<, 2
<^fc )]

fa)

(6.24)

and therefore
In (6.24)

[n

+ 2 2 cos (<,

Re means

that the real part of the parenthesis

is

to be taken.

(a)
Fig.

(b)

In part a the dimension of the packet representing 6.3. Fluctuation extremes. the particle is much smaller than the volume y, and in part b it is much larger than v.

On

averaging over the phases, we find, for n

>

1,

(625)
in

agreement with

(6.23).

These large fluctuations stem from the

natural tendency of bosons to cluster. This has, indeed, been observed in dense light beams where the counting rates of bosons do not follow a Poisson law. 2 Depending on whether there are few or many bosons

within a wavelength, our system will exhibit either particle or wave


properties.
3 1 than Actually, our particles correspond to waves with amplitude (y/L )*, rather which changes n into v in (6.25).
1 ,

See E.

M.

Purcell, Nature, 178:1449 (1956).

CHAPTER

Internal Degrees of Freedom

7.1. Fields with

the fields

To

Internal Degrees of Freedom. The quanta of discussed behave like indistinguishable particles. describe systems with distinguishable particles, it is necessary to

Two

we have

introduce several

fields,

one for each of the kinds of

particles.

The

particles may differ in such aspects as mass, spin, and spin direction or even in properties such as the charge, which are not connected to space

time.

It is

the latter kind of distinction which will concern us in this

Consider, for example, two hermitian Klein-Gordon fields chapter. and <f> 2 <f>l <fl and take the Lagrangian density to be the sum $i

- r')
,

(7.1)

^(r',0]

This describes a system with two kinds of particles with different masses. Our previous considerations concerning the eigenstates of the various operators still apply, except that now each state has to be characterized two-particle state with by the number of particles of kinds 1 and 2. one particle of each type, for instance, is given by

2>

d* ri d*r*

/(r^r^-W^-W

0)

(7.2)

but/(r1 ,r2) need not

now be

symmetric, because

58

INTERNAL DEGREES OF FREEDOM

59

Hence particles of different fields do not obey Bose statistics, irrespective of whether or not they have the same space-time properties. Consequently they do not interfere with each other and do not show any
anomalous
fluctuations.
fields is

The mechanical model analogous to the introduction of two two-dimensional oscillator. The Lagrangian for this case is

two directions are equal, e.g., w l = 2 then a new constant of the motion appears, owing to the rotational nonergodic symmetry of the problem. This constant is the angular momentum around an axis perpendicular to 1 and 2. Exactly the same happens in
If the forces in the
,

the field theoretic case


substitution

when the two masses in (7.1) are equal. The and the commutation rules are then invariant under the Lagrangian

$=
fa

fa cos

+
9

<f>

2 sin q>

&

s* n

(? 4)

<h cos

The

relations (7.4) also express a rotational symmetry in a two-dimensional space, but this space has nothing to do with our space-time continuum. Nevertheless, the formal analogy suggests that fa and fa are the components of a two-dimensional vector field in an "internal

be associated new constants of the motion These constants are called arising "isospin," in analogy to angular momentum for fields with three components. We shall discuss them presently. Like those stemming from the invariance of the space-time continuum, the constants are the generators of the infinitesimal transformations, and there are as many space" with which there
will

from the rotational symmetry.

constants as there are parameters in the group which leaves L invariant. Since the invariance group (7.4) has one parameter, we have but one Because constant. To find this constant, we follow the usual pattern. (7.4) leaves the commutation rules (7.1) invariant, there must be a
unitary operator

U which

connects 1 <, and fa


1

UfaU" l UfaU~
1

= =

fa cos

9?

+
y

fa sin

<p

(? 5)

fa sin

-f fa cos

<p

unlike elementary quantum mechanics, it is not generally true a unitary operator for every transformation which leaves the commutation rules invariant. However, we shall not get into trouble with these pathoSee A. S. logical nonequivalent representations in nonseparable Hilbert spaces. Wightman and S. S. Schweber, Pkys. Rev., 98:812 (1955).

In

field theory,
is

that there

60

FREE FIELDS
infinitesimal rotations,

For

<p

dy,
l

we put
id<pQ

Uand obtain

id<pQ

U-

= l-

-Q

(7.6)

= -*&

77)
is

can, in fact, be constructed explicitly from the field operators and the generalization of the expression for the angular momentum 2 p 2q l in the mechanical model (7.3):

p^

^i]
It is

(7.8)

simple to verify by means of (7.1) that (7.6) and (7.7) are satisfied. Furthermore, like the Lagrangian, the operators //, P, and L are of a

unit quadratic form in the

components

<f> t

and are therefore invariant

under
This

(7.5),

or

[c>w]
tells

[e P]
,

[e>L]

(7 9)
.

us inversely that

is

and rotations and,

in particular,

invariant under spatial displacements is a constant in time. The latter can


relation

also be verified directly

from the
<Mi)

7 (0i^2 at
which shows that

+V

<

V^) =

(7.10)

can be reduced to a surface integral. is also found for the other constants, such as the The fact energy, which are of the form of an infinite volume integral. that their time derivatives vanish can be expressed in differential form

The same behavior

by a continuity equation*

It

follows from (7.10) that the local density Q(r,t) of

and the current

defined by

= j(r,0 (7.12) 2Vtk) 2 such a continuity equation. This suggests that they can be satisfy This interpreted as the electric charge density and current of the field. their behavior under Lorentz transis sustained by interpretation formations. The density Q(r,t) and current j transform like a fourvector, as opposed, say, to the energy density which also satisfies a
<

(W

continuity equation.
field to 2(r,r)

and

It is, therefore, possible to couple in a Lorentz-invariant manner,

an

electric

INTERNAL DEGREES OF FREEDOM

61

where V is the electrostatic and A the vector potential. Of course, whether or not the particles actually have a charge e that is different from zero can only be discovered empirically. 1 Both cases exist in For example, the KQ and K particles do not couple to the nature. electric field, whereas the TT+ and TT~ mesons do. However, when are charged, they must be coupled to the electric field via a particles current of the form (7.12), because there is no other quantity that has the right transformation property and satisfies a continuity equation. The eigenvalues of the charge operator Q can be inferred from the commutation rules (7.7), which can be written more compactly by means of a matrix

as

[60>]

=2

( T 2)j*0*

( T20)/

1=1

where matrix multiplication


right-hand side.
,

is implied in the last way of writing the Introducing those linear combinations of the fields A and 02 which diagonalize r 2

*+

^
\2)*

^"

=
\2?

(7 13)
'

we

find

and

=J

</

r(0-0+

~~

V0_

r')

In terms of these fields the transformation (7.4) becomes a simple multiplication with a phase factor (gauge transformation of the first kind 2 ):

The commutation
1

rules (7.14) are of the standard

form

(2.7),

but since

Q is not positive definite, we may conclude that Q has both positive and
we have only the number of particle density, and a continuity equation, if there is an interaction. Hence the particles associated with this field must be neutral. 2 This also shows that a single hermitian field can have no charge.
For a
single hermitian field
satisfy

this

does not

62

FREE FIELDS
see that the eigenvalue problem P, L, or Q, can always be solved by The present case is particularly simple,
//,

negative integers as eigenvalues. for a generating operator, such as

We

following the same procedure.


since
it

amounts to diagonalizing a 2 x 2 matrix rather than a

differ-

ential operator.

Since

we do not wish

the

vacuum

state to carry

any charge, we
(7.15)
^

require that

C|0>
It
9

should be noted that this requirement is satisfied without reordering the charge operator Q according to the prescription (6.5). It follows

from

(7.14) that the one-particle states

</>

0)

and

<__
|

0) are eigenstates

1 and +1, of the charge operator with eigenvalues respectively, so that <_ creates a positive particle and + a negative particle. Since Q commutes with P and //, we can also construct simultaneous eigenstates of these operators. If we define, with the notation (4.8),
<f>

ak

fl

ik

2*.

^^

<*ik

"-^fltt
a,
/?

(7 16 )

and

recognize that the transformation from a l9 a 2 to obtain

is

unitary,

we

qkg

g
3 *

*~
[ak ,ak ]

[^k ,^k ]

[a,,^]

[a k

A] =

and

(7.17) it is apparent that the eigenvalues of the operators a k a k (PkPk) are the numbers of positive n + (negative _) particles with momentum k and energy o>. It is interesting to observe that the commutation rules (7.7) or (7.14) also hold for ^(r) or <+(r) and the operator Q v of the charge in a volume v, no matter how small,

From

^+(r)

if r is inside v
if r is

outrider

n (7

'

\x\ 18)

Therefore, the charge in an arbitrarily small volume also has integral eigenvalues, in contradistinction to the number-of-particles operator

INTERNAL DEGREES OF FREEDOM

63

studied previously. The point nature of the charge quanta is also for arbitrary v and v' and is not in contrarevealed by [Q V >Q V ] diction to our earlier findings that the particles which are eigenvalues of have the size of a Compton wavelength. These particles have to be 1 pictured as forming a fluctuating cloud of pointlike charge quanta

spread over a region of size

m~ 3

Tl
.

Indeed, Q(r) does not

commute with

so that a state <_(r)


therefore,

defined such that

Qv

= 0,

and

for which,

n A M\
will not

n\ fi.*-(r)|0>

= I^-W
(

>

if r is inside v

if r is

outside,

be a one-particle
at present,

perform

to the charge quanta. 1.2. Three and More Degrees of Freedom. The rotational invariance of the two-dimensional charge space can be generalized; it then finds applications, for example, in pion physics, where we are concerned with three different particles TT~, TT, TT+ the description of which requires three fields. We shall consider the case of n different Several fields with equal masses first and shall then apply it to n = 3. fields of the same space-time properties seem to be realized in nature. + For instance, the mesons K~, K correspond to four spin-zero fields. In general, we have

eigenstates of H. to eigenstates of

In almost all the experiments we can state. we barely have enough energy to excite the lowest Hence the particles we know empirically correspond

and are complementary concepts

K K
,

(7.19)

It is

same
1

important to recognize that all terms in L must contribute with the the energy. 2 sign, in order to prevent negative contributions to

Of course, not all have the same sign of the charge. This phenomenon is sometimes called Zitterbewegung.

the solutions of the Dirac equation.

It was first found for For a discussion within the framework of

quantum field theory, see W. Thirring, "Principles of Quantum Electrodynamics," Academic Press, Inc., New York, 1958.
2

Particles with negative energy spoil the


is

dynamical

stability

of systems

if

an

interaction

turned on.

64

FREE FIELDS
general transformation which leaves (7.19) invariant
,

The most

is

UU*= U^U =
1

(7.20)

where

A is

time-independent and

satisfies

AA =1
f

and

A*

=A
dcp(r)

(7.21)

We

therefore have rotational invariance in an w-dimensional euclidean

space.

For rotations through infinitesimal angles can be written in the general form 2

abput axes

r,

(7.22)

The number of dy(r) are infinitesimal (real) parameters. (r) will be determined below. In two linearly independent matrices ^~ The restrictions imposed by (7.21), dimensions the matrix.^" is /r 2
where the
.

to order dy, are

^ = _^w
must be
real

^w* = g-y
and antisymmetric.

(7.23)

Hence

the 3T matrices

Thus the

have n(n l)/2 linearly independent matrix elements, or we can choose that many linearly independent basic n x n matrices, satisfying (7.23), in terms of which the most general matrix satisfying this equation of parameters which characterize the can be expressed. The number
general

&

is

therefore n(n

l)/2.

The operator

U for

the general infinitesimal


(r)

is

again of the form


(7.24)

where

(r)

tion relations for

must be hermitian, since U is to be unitary. The commutat which are obtained from (7.20) are

[**,] = _,
and we recognize that

fc=l

J 9-$^

[ t W,^]

-/

k=l

J **&

(7.25)

(7.8)

can be generalized to
j

(7 26)
.

Because the space-time constants are invariant under the transformation U9 we have
[|W H]
f

(r)

[f

,P]
:

(r)

[^

,L]

r = transpose, * ~ complex conjugate, use the following notation hermitian conjugate. 2 There are also constants associated with transformations not continuously connected with unity. They have interesting implications but are outside the scope ofthis book.
1

We

INTERNAL DEGREES OF FREEDOM

65

and each

(r)

is

verify that there t is equal to Q.

is

thus a constant of the motion. a conserved current for each t (T)

We
.

can also readily In two dimensions

In order to construct eigenstates of the operators t (r \ we should first observe that, in general, t (r} and t (s) do not commute. We readily find from (7.26), or directly by means of (7.19), that the commutation relations among the operators t (r) can be calculated from those among (r the matrices \

= I f J
i

iik

=
where the second
notation.

'2f^^[^ J
it
'
}

W .^r(r

'

]^

(7-27)

line of the equation has been written in matrix Since the [&~ (r \3~ (r ] are antisymmetric n x n matrices, they are linear combinations of the matrices

[jrW^r-)]

_ ^ c;:^
r r"

(r>>)

(7.28)

Inserting into (7.27), we find that the operators t satisfy the same commutation rules as the matrices F. Thus only those operators t (r) (r) which have commuting can be diagonalized simultaneously. Furthermore, we can conclude generally from (7.25) that the eigen-

values of the operators of the matrices F.


Finally,

are

times integral multiples of the eigenvalues

we

shall discuss in

more

detail the case


1

appropriate description of TT mesons. antisymmetric matrices is then


/O 0\
1

A
1\

of n = 3, which is an convenient choice of the


1

0\

J- (l)

10
\0

(2)

10001
\-l
O/

(3)

= 1-1
\

-1

O/

O/
(7.29)

They
where

satisfy the

commutation

rules

nt is

the antisymmetric tensor.

The

three

2 sponding to these matrices are called the isospin commutation relations as the angular-momentum

and
/

operators corresatisfy the same


1

representation

of the rotation operators

[,.,] = </"
(r) t
|

As
1

for

Q we demand
9

that

0}

We

use

TT

mesons and pions interchangeably.


is

In the literature the isospin

often referred to as the isotopic or isobaric spin.

66

FREE FIELDS
the

It is consistent that

vacuum
t,

is

common
also

eigenstate of all three

an eigenstate of the com(noncommuting) operators mutators with eigenvalue 0. In general, we can only have common (r) and of t z = t (m + t 2 + f< 3)2 The eigenstates of one of the t distinction of one direction in isospace is physically obtained by the electric field which is coupled to one of the three conserved currents. (r) (3) obtains the significance of the say f By this coupling, one of the t
since
it is
.

electric charge.

Its eigenstates

are created
<,-

those linear combinations of the


(<t>i

by applying to the vacuum which diagonalize f (3) These are


.*

1 the former belonging to the eigenvalue (describing The common TT*) and the latter to eigenvalue (describing 77). 2 (3) can be constructed analogously to the ones for eigenstates of t and f

*^2)/2*,

<^3,

2 angular momenta. Calling t'(t' -{- 1) the eigenvalue of f , we see that t' = and the one-meson states by the vacuum is characterized by /'==!. Two mesons can have t = 0, 1, 2. The state t' = is invariant under rotations in isospace and hence proportional to the
1

scalar product

n
|

=2

t'

- 0} =

d*rz

'^oO-i^)

0)

Wl-W^W + ^(rM'
In terms of charged particles, this
|

is

of the form

- +> +

+ -> +

|00>

particles has equal probabilities of being 1 The state with t = 1 must transform positive, neutral, or negative. like a vector under rotations in isospace and hence can be represented

and hence each of the two

by the vector product


ij
>

state (k = 3) is of the form In particular, the r3 H > h). the t = 2 states transform like a symmetric traceless tensor, Finally,
| |

n
|

2,

t'

- 2)
etc., is

The notation

^~\

a straightforward generalization of

(5.18),

Note that,
properties,

since the internal-space rotational symmetry is independent of space-time = 1, 2, 3. it follows that 9(rlt r % ) is independent of /

INTERNAL DEGREES OF FREEDOM


These

67

states can also be formed by standard angular-momentum addition formulas. 1 Note that the isospin parts of the t 0, 2 states 1 states are odd under are even and those of the / exchange of the two particles. Hence F(r ls ra) must be even in the former and odd in the latter case under exchange of i^ and r2 ,^ a result which is important for the pion cloud around the nucleon. In summary, the different particles found in nature can easily be The more specific predictions fitted into the framework of field theory. of a symmetry isomorphic to the three-dimensional euclidean group for the three pions will be important in the last part of the book. There we shall find that this invariance is not destroyed by the strongest interactions found in nature, so that the above considerations are an

important tool in pion physics. I See, e.g., E. U. Condon and G. H.


chap.
II

Shortley, "The Theory of Atomic Spectra," Cambridge University Press, New York, 1953. Parts of F which have the wrong symmetry cancel out, as was explained above.
Ill,

Part

Two

SOLUBLE INTERACTIONS

CHAPTER

General Orientation

we have been considering free fields. more interesting problem involving an additional mechanism which can create, absorb, and scatter the field In this part of the book we shall describe several such particles. interactions which are tractable and can be solved exactly. Unare rather remote from physical reality and bear only a fortunately, they faint resemblance to what one finds in nature. Nevertheless, their study is of more than academic interest, since they teach us what might happen in the more realistic cases which cannot be analyzed in detail, such as the pion-nucleon interaction, which we shall study in the last

We

8.1. Field Equations. shall now turn to

So

far

the

part of the book.

As a
p(r,t)

first

which

is

example, let us consider the case of a simple a prescribed function of space and time:
r,0

field

source

p(r,0

(8.1)

The simplest mechanical analogue to this kind of problem is an external force /(f) applied to a harmonic oscillator. The equation of motion is
then
q

<o*q=f(t)

(8.2)

As

in our previous considerations, we shall first orient ourselves about the classical solutions of such an equation and shall later consider the

As we shall see, Eq. (8.1) can be aspects of the problem. solved with the aid of the Green's function, like the well-known onedimensional case (8.2). To this end, we have to realize that if the source is the sum of several parts, then, owing to the linearity of the equations, the solution is the
quantum
71

72

SOLUBLE INTERACTIONS

solutions corresponding to the individual parts. need to find a solution for a point source only

sum of the

Hence we

and can then build up a solution of


#r,r)
It is readily seen,

(8.1)

by superposition:
t

=
fdt'

d3 r' G(r

r',

W,O
2

(8.4)

with the aid of

(8.1).

the

A solution of (8.3) can be obtained by an expansion in terms of ~ V that by eigenfunctions of the differential operator d /dt
2 2
,

(8.3), that this field actually satisfies

is,

introducing a Fourier transformation


G(r,r)

= S f %* e^<-xg(K J-ao 2rr


k
<S(r)

,k)

(8.5)

<5(r)

= S f ^e'^'-K"* k /-oo 2?r


_
k2

Substitution into (8.3) gives

k-r-J[( K ;
k

_ m *) g +

i]

(8.6)

-oo

vr

m2

+ ^-K:
and
at

(8J)

We

note that the integrand in (8.5) has two poles on the path of
at

integration,

K =

o>

(A:

+ m 2)*

K =

-w.

Without

specification of the path of integration at the singularities, there is an ambiguity in our expressions for the Green's function. On integrating along different paths, we get results which differ by the residues at the This is the well-known fact that the solution of a linear singularities. inhomogeneous equation is not unique, since a solution of the homogeneous equation can always be added to it. In fact, the contributions from the residues are of the form e*< k r -* which is just the solution of the homogeneous equation. To obtain a unique solution, it is necessary to impose boundary conditions. The ones of special importance for the problems that we shall be concerned with are characterized by We the paths of integration in the complex plane shown in Fig. 8. 1 ret and A adv Their shall denote these particular Green's functions by A integral in the complex significance can be seen by studying the For t larger (less) than zero, the factor e~ iK ^ increases (deplane.
'

creases) exponentially in the upper half plane and decreases (increases) ip the lower half. Closing the path of integration by adding an infinite

GENERAL ORIENTATION
semicircle in the upper or lower half plane, and that A adv / for / 0.^1 The

73
for

we

see that

<

rct

>

complete Green's function


*

Arot

is

given by

C Am(r,0= be*"" t
and the two solutions Arct and

tk.r

for

>

/x

(8.8)

for

<

Aadv

are related by
(8.9)

plane

plane

Fig. 8.1. Green's function contours corresponding to conditions.

two

different

boundary

With the aid of

these Green's functions,

we can

write the general

solution of (8.1) in the form


r'

A ret(r adv

r',

t')p(r',t')

d*r'

+jdt'
Here
ln
<

(8.10)

are solutions of the homogeneous equation. Their can be seen most easily if the source p(r t) differs from physical meaning zero only in a finite space-time region, bounded by the times t l and t z It follows from the above properties of the Green's function that in this out in for t > r 2 Hence ln case $ coincides with for t < t l and with
<^
9
.
<

and

out

<f>

<

r. One can see from relativistic invariance that A ret = even for t 2 t ). Both A ret and A adv can be worked out to be Hankel functions of (r 2 This is most done by relating A ret to A+; see, e.g., W. Thirring, "Principles conveniently of Quantum Electrodynamics," Academic Press, Inc., New York, 1958. However, we shall not need these expressions.
If

<

74

SOLUBLE INTERACTIONS

represents the field which was present before the source was switched out is the field which is left over after the source has been on, and turned off, as shown in Fig. 8.2. In many discussions, only the term
<

with

Aret

in (8.10)

is

used, which

means

condition that

that one imposes the boundary in oo. at t <f>

<

Although there

is

no fundamental

reason why A ret should be better than Aadv , the former is used more frequently because the initial experimental conditions at
are
t

= -co

more

easily

prepared

than

specific final conditions at/


is

+00. Another kind of problem which of interest, and which will be


is

studied later on, 1 by the equation

characterized

(8.11)

Fig. 8.2. Regions about a source

which

</>

corresponds to

<

in

and

X*0 in out
<

represents a generalized potential that acts on the field; for


F(r,r',f)

coupled oscillators it corresponds to a more complicated coupling

than one that is j ust between nearest neighbors. It is only possible to solve (8.11) explicitly for particular forms of F, but in any case the equation can be rewritten in integral form with the aid of the Green's functions:
"

A ret(r -

r',

=
If

ou out
<

(r,0

+ J dt'
t

d*r"

A ady(r then
<

r', t

<

t')V(r'
out

>

(8.12)

as defined by (8.12) have the same physical significance as in the previous example. Both kinds of problems are encountered in many branches of physics,
to zero for
oo,

V tends

-*

in

and

typical although usually in somewhat more complicated form. feature of such systems is that the energy and momentum of the field alone will no longer be a constant. This- stems from the fact that
1 This kind of equation is familiar from ordinary wave mechanics, except that one usually deals with time-independent and spatially localized potentials, ne

GENERAL ORIENTATION
dL/dt and dL/dr
explicitly or K. In fact,
will, in general,

75

on

and

and

since

not be zero, since p or V may depend L now contains an additional term with p

not even the number of field particles will be conserved,


(8. 11)

since (8.1)

and

give

The observables remain constant only in special circumstances. The angular momentum, for instance, is constant for a spherically symmetric source.
8.2. Quantization.

two

cases,

we

In order to discuss the follow the standard procedure.

quantum theory of our The Lagrangians of the


fields

systems differ from the Lagrangian of the free of terms

L by

the addition

(8.13)

and
respectively.

L'

d*r d*r' #r,OK(r,r',0#r',0

(8.14)

The

we

take

K(r,r',f)

hermiticity of L' implies K(r,r',r) Since L' does not contain K(r',r,f).

=
<

K*(r,r',f)

in either case,

the canonical commutation relations remain


[<(r,f), 0(r',r)]

3
*V5

(r

r')

(8.15)
n

To
out
<

get
,

some information about the commutation properties of fi and oo where p and K we consider the limits / -> approach zero.
<f>

Since

then coincides with


r
I I

out
<

and
t

iu
<

(8.15) implies
* i

in/

ir

_>^\

^i\R/
(p
.

oo).

ir

~^\~\ ~~~ ooi

^^^

10 ir

J&3/..

.'\
)

/0

^ ^. x

(8.16)

obey the homogeneous field equations and can therefore be expressed in terms of time-independent operators A k and Bk in the familiar form
<

The operators ^ in and

out

4
(2WL )
3 t 31

Our development

for free fields provides, therefore, a basis for the


infer that the A,

problem with interactions.

From

(8.16)

we

commutation

rules

obey the usual of destruction (creation) operators. This tells


C4
)

and B,

(J9 )

76

SOLUBLE INTERACTIONS

us, in turn, that (8.16) holds for all times


/ -*>

and not only


(8. 16) is

in the limits
first

oo.

The compatibility of
in
</>

(8.

5)

and

evident for the

case where

and

oui
</>

differ

only by an ordinary number.

In gen-

commutation rules will not be trivial and will have some important implications. At a certain time /, one can, of course, satisfy the commutation relations of the local field operators <f>(r,t) and <f>(r,t) in the customary manner:
eral, however, the equivalence of the two

(8.18)

where the operators a k (t)


equal times,

satisfy the usual

commutation

relations at

K(0, 4(0]

<5

3 ' k k
,

(8-19)

It is important to note, however, that (8.18) and (8.19) do not imply that the time dependence of the operators ak is that of the free fields. This will, in fact, not be the case, except in the limits of t -> oo.

The physical interpretation of our systems in quantum theory is developed along the same lines as in the classical field case. Since for in t -> coincides with the quanta which are created and oo, A^ and A are those particles which were present before the destroyed by source was turned on. In particular, we can define the number n operator A^ of incoming particles with momentum k as
<

<f>

N? =
in
<

AlA k

(8.20)

Because is a constant, and its obeys the free-field equation, eigenstates represent a situation wherein a definite number of particles The same conwith momentum k are initially (t ~> oo) present.
l

out u siderations apply to </> and \ which correspond to the actual At any time, and in situation after the source has been turned off.

particular

when the source is switched on, the field a k (t) and the number of particles by operators

is

represented by the

The

n latter is not a constant and will differ from k once the source is turned on. Therefore, an eigenstate of N{? has a fluctuating number of k (t) particles. Even a state with no incoming particles will not be an eigenstate of the operator

GENERAL ORIENTATION

77

and generally

by N (t) are usually called bare particles, N (t) represents both real and virtual particles. The eigenstates of N (t) which are also eigenstates of // (0

The

particles represented
k

/"V]

- I coflfrKW k

(8.21)

are called the bare states. Since k (t) and Q (t) are time-dependent, 1 The corresponding bare states at different times will be different. of kn which are eigenstates of are called physical states, eigenstates

and the corresponding particles are real particles. That the states generated by A[(i) are eigenstates of the total Hamiltonian follows simply from the fact that the time dependence of the A k (t) is that for We shall see this more explicitly in the following chapters, free fields. in terms of the operators A k where we shall express We shall find that H is simply 2 4[A k ( plus a c number. The bare states are hard to

prepare experimentally, since usually sufficient energy is not available to excite more than the lowest few states of the system. For instance, a bare vacuum state 2 0) defined by a k (t) 0) = corresponds to one for which the dress of virtual particles from the source is removed at time t. This costs a lot of energy, since 0) contains an admixture of highly
| \ |

excited physical states. In our cases the situation

is

relatively simple, since

we only have

the

source which

is

capable of emitting particles.

In a theory with non-

linear terms in the field equations, as in relativistic quantum electrodynamics, each particle acts as source for the other particles. There each physical particle is a mixture of all sorts of bare particles. In our

theories the physical states are the source plus a certain number of incoming (or outgoing) particles. These consist of the bare source plus a certain configuration of bare particles.

One may
significance.

ask to what extent the virtual particles possess physical Certainly the particles we see in cloud or bubble chambers

are always the physical particles. However, the virtual particles do We shall see that exist, inasmuch as they lead to observable effects.

they contribute to the energy and charge distribution of the system. Furthermore, as we shall explicitly see later on, the virtual particles present at the time t can be made real by suddenly switching off the source at this time. In this case <(r, / + dt) will turn out to be identical

with
1

out
<

(r, t

dt)

and with <(r,0

(<j>

stays

finite).

Since 7V out

is

states are constant

we are working in the Heisenberg representation, where all and the time dependence is put into the operators. 2 To avoid the crowding of labels, we denote bare states by ) and physical states by >.

Remember

that

78

SOLUBLE INTERACTIONS

constant, the particles created remain for all later time and represent the The virtual particles are thus those which particles detected afterward.

would be

left

over

if

the source should suddenly be turned

off.

In

1 happen to some extent, e.g., through the annihilation of a nucleon by an antinucleon or through a very fast collision between two nucleons. 2 The mesons produced in such events are just the virtual particles in the meson cloud of the nucleon, which suddenly find

practice this can

from

will in general differ Therefore, the number of particles and their energy and momentum at t -> oo will differ from those at t -> o>, and both will
field
<

themselves without their source.


in
< .

The

out

differ

from corresponding quantities

at other times.

8.3. Scattering and Wave Matrix. found that for coupled In the fields the concept of a particle requires some qualifications. table below we summarize the different sets of orthogonal states which

We

are associated with the various kinds of particles

ln
|

in, 0)

- 2 4X
^

in 0}

These are P h y sical states. They represent the dressed source plus
" incoming particles.

n?

_L_
(

(A*)** k

in, 0) '

k !)*

H
|

= = [f/,/4 k ]
in,

wk>

Ek

in,
|

w k>

coA k
out, 0)

out
|

out, 0)

- 2 #X
=
(n k \y

These are P h y sical


states
'

out, n k >

L- (frfy*
k
|

out, 0)

The ?

out, n k >

[//,5k ]

out, n k >

represent the dressed source plus " outgoing particles.

-co0k

N(t)

0)

(W

k0

These are bare

states.

0)
k)

H
1

(0

= -

They

o)

k (0

k)

represent the bare source plus n k ( real or virtual ) Particles.

It

would have
fc

to be carried out infinitely fast to produce all excited physical

states.
2

See H.

(1948); E.

W. Lewis, J. R. Oppenheimer, and S. Wouthuysen, Phys. Rev., 73:127 M. Henley and T. D. Lee, Phys, Rev., 101:1536 (1955); Z. Koba and

G. Takeda, Progr. Theoret. Phys. (Kyoto), 19:269 (1958).

GENERAL ORIENTATION

79

In certain circumstances the "in" and "out" states do not form a complete set e.g., when the source is strong enough to bind particles. In this case one has to augment the in and out states with the bound We shall discuss this in detail when the states to obtain a complete set. case arises but shall assume in the discussion below that all sets are complete. Then they are related to one another by unitary matrices. The elements of these matrices can be defined as the products of the states of one set and those of another. Equivalently, they can be
defined as the elements of a matrix which transforms the generating operator of one set into that of the other. For instance, the connection between the in and out states is established by the so-called "S matrix," or "scattering matrix," which plays a crucial role in modern field theory.

We
the

can define

it

as the unitary matrix

which transforms the A k (t) into

Bk (t):

Bk (t) = S~ l A k(t)S

(8.22)

The existence of such a matrix is inferred by the usual remark that the A k and Bk satisfy identical commutation relations. Furthermore, they have the same time dependence, so that S is time-independent. From (8.22) we infer A k S 0, out) =-0 5 0, out) = 0, in) (8.23)
|

Hence an equivalent

definition of the elements of

is

Sk ;,..,Cici ..... K.

= in kj, = (in, A^ - (out, k - (out, k;,


<
'
. . .
|
|

k' n
. .

S
| |

in, k,,
in,
|

kn)
.

A^S
,

kx

k B>

;,

Bk n
>

,
|

in,
. .

k^
.

k n>
(8.24)

k;

in, k!,

k n)
its

One of

the important features of the

matrix

is

relation to the

In systems wherein scattering cross section, which is sketched below. the total energy is conserved, the S matrix only connects states with

equal energy.

It

is

between an the form

initial state

conventional to write the matrix element of S with energy w and a final state with energy <o f in
t

S ti

d fi

- 27716(0), -

a>

)Tfi

(8.25)

From

this relation, the probability that


is

a final state/
1

develops from

the initial one

found to be
i?,<

= 47T

|<5K

co,)!},!

( 8 - 26 >

We

differs

assume that all the values of k are from that shown.

different; otherwise the normalization

80

SOLUBLE INTERACTIONS
is to observe that it time of interaction in eigenstates of the
<5

A rough manner of arguing away the unpleasant


is

connected with the

infinite
f)

energy.

The

hence we can

(0^ write
2

a>

has

its

origin in

an expression J dt e(a>
'
mf}t

-<"/>',

and

a),)]

($K

(o ) f

dt e i(ai J-oo
I

d(o><

^/)
/-oo

dt

Defining a transition probability per unit time,


dt

Wn

by

we

get the "golden rule"

^-27r^-co )|T
/

/i.|

(8.27)

The

total transition rate


i

out of the

initial state
2

can be written as
ti

W = 2irS 6(0^ where the


last equality is

a> f )

\Tfi

= -2 Im T

(8.28)

a consequence of (8.22),

i
/

SfjS/i

"<y

These formal expressions can be analyzed in more

detail

by diagonaliz-

ing the S matrix. In practice this is done by finding a sufficient number of constants of the motion whose eigenstates are also eigenstates of S.

We define a projection operator


/<

ty

(A)

onto eigenstates of S, denoted by


e.g.,
!

A, by treating the energy separately,

= g(<Wsi:-/)S $? A

S(^^c

c S^K-co^S^
(A}

?
}

(8-29)

and write

(8 30)
.

Inserting into

T = -gM S sin ^(4y xW *Jf - $ (X (8.28), we find, with [^>] H^ - 2gK) $ $1 sin ^K)
fi

^4

>,

(8.31)

^4

This formalism simpli(5(eo z.) is the phase shift at the energy w t greatly for a spherical source and only one outgoing particle per incoming one. Then S is diagonal in an angular-momentum represen-

where

fies

tation,

and

co is

the only continuous variable: 2

S
|

fc, /,

m)

-e^
2

(fc)

k,
|

/,

m)
is

A more satisfactory way

of deducing these results

to consider

For a more complete

discussion, see

M. Gell-Mann and M.

wave packets. L. Goldberger, Phys.

Rev., 91:70(1953). 2 The factor ng(w^ corresponds to the energy normalization


<*,
/,

k\

/',

w'>

rrg(a>)d(a>

- />V<W
J.

For further reference,


(1950).

see, e.g., B.

A. Lippmann and

Schwinger, Phys. Rev., 79:48

GENERAL ORIENTATION
or, in

81

terms of the generating operators,

Bum =
In the above case,

S~*A Um S

f&Aum

(8 32)

we

get

s-i A 1=0
Furthermore, if we use a normalization volume > d(o, we have dynamics k dk

and

relativistic

and
Inserting into (8.31),

(27T)

87T

we

find,

with

47T
2

1) sin

kco

d,H

(8.33)

To

divide

obtain the familiar expression for the cross section, we have to by the incident flux, that is, the number of incident particles per unit time and unit normal area (= fc/wL 8 ):
**

W* T^
k

77 2 k&
i

( 2/

!) sin2

*M

( 8 - 34>

In cases where particles are produced, there are other continuous variables, in addition to the energy, which characterize the eigenstates of S. In that case (8.34) has to be modified, but (8.27) and (8.28) are
still

valid.

Whereas the S matrix contains the information of the elementary


phase-shift analysis, the matrix connecting the in states with the bare states corresponds to what is called the "wave matrix" in elementary

wave mechanics. The latter contains information about form of the wave function in the near zone (i.e., near the
field

the detailed

can answer questions about Although such problems are largely of academic interest, they are instructive and will be studied in later
theory
it

In source). the distribution of virtual

particles in physical states.

Furthermore, we shall see that there are important relations between the 5 matrix and the wave matrix.
chapters.

CHAPTER
Static

Source

9.1. Interpretation of

"Static" Source.

As a

first

shall carry out a detailed study for a static source p(r,t) gp(r). assume this source to be centered about the origin of coordinates, to
real,

example,

we

We
be

and

/>(r)

to be normalized,

so that g represents the dimensionless strength of the source. In this case we encounter in the general solution (8.10) the expression
dt

Joo -oo
If

Aret(r,0 =

f oo

dt
J-oo

S k
/I

\TT

m*

KT

we can
00

interchange the order of integration, then


,

J.j"

^ A ret, A

(r '

^c
fe2

e*'

+m

=
2

Yf

00

\^j J

p 4im)
is

k dk eikr
J-ao (k

im)(k

47rr

To

the extent that this interchange of integration adv (r,0get exactly the same result for J dt A Aadv or A ret , equivalent to the statement that, for
(V*

permitted

we

shall

These answers are

_m)f
2

dt
ao

Aret(r,r) =

-<53 (r)

and e~mrl4nr

2 the Green's function for (V 2 ). and / integrations in for the interchange of the justification The classes of inter(9. la) rests on the meaning of a static source. 00. actions discussed in Chap. 8 demand that the source be zero at f

is

The

w K

82

STATIC SOURCE

83

This is clearly not the case for a truly static source. It turns out, however, that switching the source on and off slowly (compared with m' 1) leads to the same results as a static source, and it is in this sense
that (9.16)
is

correct.

To
p(r,t)

see this,

we

consider a source
a

= gp(T)e~*

ul

>

(9.2)

If

we make
oo

use of

(8.8),

then the time integral which appears in (8.10)


*
gp(*')
CO

becomes
/*

*~

A ret(r - r', - f>(r'>Odf = S


t

tk(r

r')

(*t

e~ aU/l sin
J-M for
sin cat
2
tt>

a)(t

t')

dt'

k D

e**

<
>
_

**

Hj a2
,

2a

'

a^

for

In the limit of a -^ 0, or in fact a

< m,

Eq. (9.3a) reduces to

-oo

A ret (r-r',r-Op(r',0^
k
CO*

,-m

Ir-r'l

(9<36)

^jTT7|
which agrees with
limit.

(9.16).

We
is

The same equation holds


is
1 during a time a"

for

Aadv

in this

which

thus note that a static source switched on and off slowly

to be interpreted as

> nr\

one and we

then obtain, from (9.36) or (9.16),

sA

ret

(r

r', t

t')p(r') dt' d*r'

(9.4)

We

thus find that

in
<

out
<

which means that the source creates no

real particles and that there is no This is connected with the scattering. 1 static form of the source and the lack of internal degrees of freedom. That the process of switching on and off creates no disturbance
(e.g.,

produces no particles) corresponds to the adiabatic theorem in elementary quantum theory, according to which a disturbance that varies slowly compared with the natural frequency of the system (which here is m) will not produce any transitions.
1

This conclusion will be clarified in subsequent chapters.

84
9.2.

SOLUBLE INTERACTIONS

Energy of the Coupled System. eigenstates of the total energy, we shall and use (8.18) to obtain in terms of
<

To
first

learn something about the express the Hamiltonian

In (9.5)

we have denoted

the Fourier transform 1 of


ik ' r

/>(r)

by

/> k ,

r)e~

d3 r

(9.6)

and

its

complex conjugate by
ln
<

/>.

We

can also express

H in

terms of

the operators

If

we make use of
-<5
3

ret (-m 2 + V2)fA (r,0 dt =

(r)

we

find that

f fr d*r<
J

[Vr Ar*OVr

n + m 2 ^ ] f^Ti rfO =
1

4n

|r

Jf

d* r Ar.Orfr)
iu
<

Because of this equality, the cross terms that occur


p(r) vanish,
2

in

H between

and

and we obtain

H = H in
with

~\-

f
in 2
)

(9.7a)

H in - i f rfr [(^ in ) 2
2J
k

-f

(V^

+ m 2(<A in 2]
)

= 2 Al\<
and
'o

-^f J ^^ d*'" P(r) 4w ^-j 2

__a 2 T

--wlr-r'l

r
<

(9-76)

p(r')

|r

That

a c number, commuting with in was to be anticipated. In and out have the free-field time dependence, the Hamiltonian, when expressed in terms of either of these operators, must
is
,

Since both

<

<

if p(r) is

normalized according to

Furthermore, p of pk is

J/>(r)*/ r

1,

then p*
p*.

is

dimensionless.

()

and

in the

continuum

limit p(k)

The Fourier transform

2 This can also be shown in momentum space. In the following, it is to be understood that the zero-point energy of the vacuum is subtracted from //.

STATIC SOURCE

85

reduce to //
// (in)

(in)

and a part which commutes with

(in)
<f>
.

We

note that
In

f o are time-independent, but both terms of the operators introduced by (8.17),

and

and

//' are not.

we

get (the zero-point

energy SJco has already been subtracted)


ff

= 5X4/ + *o
k

(9.8)

therefore, of the same form as that for the free fields, except that it is shifted down by an energy as |<? |, shown in Fig. 9.1. This energy represents a "binding" energy of the virtual particles, although the interaction energy //' is by no means a
Energy
Energy

The eigenvalue spectrum of// is,

(a)
Fig. 9.1.

The eigenvalue spectrum of H


.

is

One- and two-quantum


figures
is <'

states are represented.

shown in part a and that of in part b. The energy shift between the two

normal potential. That the physical ground-state energy <^ is less than zero shows that the problem considered belongs to the wide class of This is always interactions which decrease the ground-state energy. 1 true when perturbation theory is applicable, since then
(9.9)

Physically, this means that the ground state finds a age of the new situation to lower its energy.

way

to take advant-

The energy f is referred to as an "energy renormalization" or sometimes as a "mass renormalization." The latter term is to be understood in the light of the following considerations, which use the equivaOne lence of mass and energy in the sense of a relativistic theory.
See L. I. Schiff, "Quantum Mechanics," 2d ed., p. 153, McGraw-Hill Book Company, Inc., New York, 1955. The matrix elements are between virtual or bare states, and it can be shown that, for the present problem, (9.9) gives the same
1

answer as

(9.7c).

86

SOLUBLE INTERACTIONS
static

can consider the

source to have a mechanical mass

The Hamiltonian
states (of // interaction.
4-

then represent the source and the field without When the latter is turned on, then the lowest physical = M. The mass is state is an eigenstate of with energy -f <" the physical mass of the source and differs from the mechanical mass due to the interaction between the source and the field. It is in this sense, also, that eigenstates of // are referred to as bait states and the as physical states. eigenstates of For a point source, />(r) = <5 3 (r), we find that /ok = 1 and that the energy <f diverges linearly. There is, however, no way of observing this energy, no matter what its value, so long as the source is always surrounded by its cloud of quanta. 1 Since <* is not observable, we can
)

H then

becomes

H + H' + M
Q

A
eigen-

The bare

subtract

it

from

//,

so that

tf
\

in, 0}

= (H - <f
Bare and
is

in, 0)

=
States.

(9.10)

9.3.

Connection

between
|

physical) ground state

in,

0)

Physical again defined by

The

(real,

>l

|in,0)-0

(9.11)

eigenstates of the Hamiltonian are created by repeated of A\. They correspond to a certain number of incoming applications But in = out or A k = Bk9 so that particles with definite momenta. also correspond to the same configuration of outgoing particles. they Hence there is no scattering or creation of particles in this model. Further insight into the model can be obtained by analyzing the incoming vacuum state in, 0) in terms of eigenstates of A^O). That is to say, we are interested in the configuration of virtual particles present To that in the physical ground state of the system at the time / = 0. purpose, we express A k in terms of the a k [defined by (8.18)]. This can be done by substituting (8.17) and (8.18) in (9.4) at / = [we abbreviate
<

and the various

(f>

'

2 3

(2ro

L)

from which we

infer that

This will be

made

clear in the following discussion.

STATIC SOURCE

87

Therefore (9.11)

tells

us that

ak
|

in, 0)

g-j

in, 0)

= %k

in, 0)

(9.13)

has the same form as the definition of our standard wave harmonic oscillator (2.23). The problem of analyzing 1 in, 0) racket in terms of the eigenstates n k ) is identical with the calculation and hence, with the notation of Sec. 2.3, we obtain (2.25),

and

this

for the

(*,*,

*J in, 0>| = n exp (-n k ) -%^


2 (

(9.14)

with

nk

--

This represents the probability for finding n ki virtual particles with momentum k ls n k ^ virtual particles with momentum k 2 etc. The product form expresses the independence of the particles. The probability for finding n ki particles with momentum kj irrespective of the number of particles with other momenta is given by the sum over all other n k m
,
:
.
.

V\( V\

>

11*1

ff

1 tl

fl\ I"

(9.15)

That

we have a Poisson distribution for the number of virtual of a definite momentum. Similarly, we derive by induction particles that the probability of finding n virtual particles irrespective of their
is

to say,

momenta

is

*?(")

= I K>
I

"

J
2o>

in, 0)

*'*

-^f

(9.16*)

with

=J
i

/i

ki

S ^t J
k

(9.

16*)

which

is again a Poisson law. The same law holds for the probability of finding n particles within a certain region A in momentum space, in The number n represents the average which case n equals 2

number of field quanta which dress the source. As we shall see in the next chapter, it is this number which will be produced if the source is suddenly turned off. For a reasonably small source size (e.g., of radius
a

~ 1/^max <
1

l/) we

find

tf^gliln^^l!
47777

4v

(9.16c)

The time dependence, of course,

is

now governed by

H and not by //

88

SOLUBLE INTERACTIONS
is

Thus the average number of virtual particles that surround the source
of the order of g 2 fin.
It is instructive to

carry out the expansion of the physical ground a complete set of eigenfunctions ofN(t) at t = and to compare this with the ground state of the hydrogen atom. By means of (9.13) and (9.16), we obtain
state in
|

in, 0)

=2
n

"X
1

in
I

=
|

0)(0

in, 0)

+
k

at

0)(0

ak

in, 0)

= (0

in, 0> [

0)

+ 2 fcfll

0)

+ J i JMt*t?>l

0)

where #k is defined as in (9. 1 3).

The Fourier transform of the hydrogenatom ground-state wave function is oc(l + r?& 2)~ 2 =~ #k where rb is the Bohr radius. In our notation this state could be written as
,

|in,0>

=
!*ifll|0)
is

(9.18)

In contrast to this, the ground state of the field

a mixture of states with

various numbers (from to oo) of virtual particles. This fluctuation of the number of virtual particles is sometimes expressed by saying that
the source creates and reabsorbs virtual particles. This terminology is similar to that used for the H^ molecule, where we say that the electron The virtual particles are not always is exchanged between the protons.
/2 k, present, so that e~* virtual particle in single

which corresponds to the wave function of a

momentum

but to ne~

fi

<

The wave functions

space, is not normalized to unity for states with several particles

are simple products, which shows that the particles are uncorreFor a lated except for effects due to the Bose-Einstein statistics.

point source
#k ^ Pk/(&
2

+w

(/> k 2

)"

1) the Fourier transform of the wave function behaves approximately like e~ mr /r*. More generally,

we
is

see that the expectation value for the field <(r) in the

ground

state

just
(in,
|

#r)

in, 0)

= $ (in,
r
k

*****
|

+ *&"*"
|

in, 0>

= S fr' gp(r') J

gik'(r-r')

or

f = g \d*r'
J

g-lr-r'lm
4ir r

(9.19)

STATIC SOURCE

89

so that the cloud of virtual field quanta covers the source with a thin of extension ~l//, as shown in Fig. 9.2. (For pions, m~ l 10~ 13 This is required by the uncertainty principle, since virtual mesons cm.)
veil

longer than m~ and can, therefore, not get farther than Of course, their density is not sharply cut off after m~ l but decays nr Such behavior is similar to the leakage of a particles exponentially.

cannot
l
.

last

into the energetically forbidden zone in a decay or the leakage of light into the dense medium in total reflection.

Roughly speaking, we may say that all space outside the source is energetically forbidden for the virtual

particles

but that they can leak out because of the uncertainty relation. It is the cloud of particles surrounding the source which dresses the latter

and

is

$ o,

which thus plays the

responsible for the energy shift role of a

Fig. 9.2. Plot of <0

<f>

0> close to the

binding energy for these virtual parThe meaning of the above ticles.

source.

becomes clear if we remember

that, in the manipulations leading to (9.7), half of the contribution of the interaction energy //' [(see (9.5)] to <* was Hence we have canceled from the part stemming from 7/
.

(in,
|

HQ

in, 0)
|

-i(in,

H'
| \

in, 0>

(9.20)

and

It states

this expresses a "virial theorem" if we call //' the potential energy. that the total energy which is kinetic (7/ ) potential energy is

Indeed, by means of (9.14) just the negative of the kinetic energy. we find that the ground-state expectation value of // (9.20)
,

and

(9.21)

just the mean value of the kinetic energies (including the rest mass of With this wider concept of a potential energy, we the virtual particles). may say that virtual particles are bound with an energy which exceeds
is

their

mass m. For the square fluctuation of the


<
| 1

9.4. Fluctuations of the Field.


field,

our formula

this is
|{0
|

<t>

The reason for valid in the present case. 2 that the contribution from the source to (0 0) is canceled by 2 Here, as in the case of our 0) , which is not zero here.
(4.
1

3) is

still

standard wave packet

(in

Chap.

3),

mean number of mesons n

>

implies that the fluctuations of the field are less than its average value. have remarked before The classical field picture can then be used.

We

90
(see

SOLUBLE INTERACTIONS

Chap. 4) that

this is

not the case for the

electric field

of an elemen-

tary point charge for which

whereas the fluctuation of the field averaged over a region ~r* is 1/r; the mean number of photons in a volume A in momentum A<j5 space, bounded by k mSLK and fcmin is found from (9.16) to be

_
For reasonable kmax and *mln with e 2 /4?r = T | T /? A T J^
,

?->*_*. to*47T Jkmin


k
fcmax

(9 22)
.

47T

fcmin
el ,

(e.g.,
1.

kmin

~ l/r

b)

we

obtain,

<

ies

always create the classical system we shall see that n

However, charged macroscopic bodsituation n > 1. For the meson-nucleon 1 and that 1 is in an awkward transition

region.
It may happen that, for a particular form of the source, n -> oo, as it does for a point source. In this case the probabilities (9. 14) for finding a finite number of virtual particles are zero. This implies that states with finite numbers of real particles are orthogonal to states with finite numbers of virtual particles. Thus, a perturbation or other expansion of a real state in terms of states of virtual particles is im-

possible; in particular, this applies to the ground-state expansion (9.17). If a perturbation expansion is nevertheless attempted, then infinities are

always met. This difficulty is encountered in relativistic theories wherein the interactions must be localized. In a nonrelativistic theory, where the source may have a finite size, this problem may be circumvented.
9.5. Several Sources.

Because (9.7c)

is

quadratic in

p,

the

self-

energies of several sources are not simply additive; there will be cross In particular, for two spatially separated point sources of equal terms.
strength,
/>

= gTO +

3
<5

(r

- r )]

(9.23)

we

find

from

(9.7) that

0-^-and from
|

(9.24)

(9. 17) 0, 0)

that

e-*/ exp

---.-^d + e**)a

0, 0)

where <f is the infinite, but constant, energy for one point source, given by (~ 2/2) 2 (V^ 2)* and 0, 0> is the physical ground state of the two
|

sources.

The dependence of <^

(2)

on the source separation r

is

of the

STATIC SOURCE
1

91

form of the famous Yukawa potential. In the language of perturbation theory, Eq. (9.9), it arises from the exchange of a virtual quantum between the two sources, and its range reflects the limitations imposed on such a process by the uncertainty relation. That is, an interaction between the two sources arises only when the distance between them is
of the order of the size of the quantum cloud. One cannot, therefore, ascribe a classical path to the virtual particles, and the term "exchange" has to be taken with a grain of salt. It really implies that an overlap occurs for the virtual clouds belonging to the two sources. The limit of applicability of the classical concepts must be kept in mind in using The sign of the "potential" intuitive pictures of the "exchange." an attractive force between the two sources, since energy implies f (2) < 2<f This arises because the presence of a second source within a short distance of the first one opens channels for new processes that decrease the energy. However, this is true only if g has the same If we were to take sign for both sources.
.

Ptt
the force

[3 (r)

3
<5

(r

r )]

(9.25)

This obviously stems from the fact that, sign. on close approach, the sources neutralize each other, which decreases We shall see in the next part that for the pion-nucleon |<5f (2)|.

would change

is somewhat more complex and, depending on the and charges, the nucleon sources may have the same or the spins opposite "mesonic charge" g. Consequently the exchange of a meson will lead to an attractive "potential" in some states and to a repulsive one in others. To conclude this section, we remark that for N point sources of

system the situation

arbitrary strength,
=l
r<)

(9-26)

we

obtain

W) = I *P -2
i=i

~m\ri-r }

I gig, i>i j=i

:-, 4?r

\
-

(9-27)
1

|rt

r^|

where

^ = -^1
2
potential

k co*

Thus the

is just the sum of the potentials between pairs, which shows that the presence of other sources does not disturb the force between a given
1 We shall postpone to the last chapter a discussion of the extent to which this term can actually be interpreted as the potential energy of the source particles.

92
pair.

SOLUBLE INTERACTIONS
This

is not generally true but is connected with the fact that are not scattered in this model. quanta

Further Reading

The problem of one or more static sources can also be discussed in different but equivalent ways e.g., by just stating a unitary transformation which For this type of approach and for further diagonalizes the Hamiltonian. studies the reader may refer to the following:
G. Wentzel, "Quantum Theory of Fields," New York, 1949.
p. 47, Interscience Publishers, Inc.,

S. Tomonaga, Progr. Theoret. Phys. (Kyoto), 2:6 (1947). L. Van Hove, Physica, 18:145 (1952). T. D. Lee, Phys. Rev., 95:1329 (1954).

CHAPTER

10

Production of Particles

Remarks. Having studied the properties of virtual in the last chapter, we shall now investigate the circumstances particles in which they can be converted into real particles. To this end, we need
10.1. General

to express outgoing operators in terms of incoming ones.

According

to the general formula (8.12), the connection between the incoming

and

outgoing

field is
<

given by
out

ln
<

+ JW dt' A(r sin a>t

r', t

W.O
oo

(iO-0

where

A(r,0

- Aret (r,0 - A adv(r,0

= S(O k
The
of
I

for ~-oo

< <
t

(10.2)

relation (10.1) can be expressed in

momentum
'
*

space by making use

A(r

r', t

t')p(r',t') d*r' dt'

= S^ J k
I

S1

" "** ~~

* )

a)

X r '^') &*' di

'

where

/> k

is

/.

Pk(^o)
It

=)

Xr.

then reads

Since Bk obeys the free-field equation, only that part of the source contributes to it for which the frequency and wave number are
93

related as for a free particle. Correspondingly, in ordinary space, the wave functions of the created particles will not be confined to a region close to the source. In classical terminology, we can say that the virtual
particles are the ones contained in the near zone, particles leave the source and get into the problem of analyzing states with a certain

whereas the outgoing

wave zone of the field. The number of incoming particles


. . .

in terms

of the out

^k t

flk z

vacuum

in particular, the probability of finding in the incoming outgoing particles with momenta k ls k 2 , 3 is (9. 14), since Bk |in, 0) /V> k (w)(2wL )~*|in, 0), again given by
states

where n k now corresponds to

Hence we have a Poisson

distribution of emitted quanta in every

momentum

interval

with a

mean number of particles 1


k<A
(10.4)
20)

fc_ $

We

can also carry out an expansion exactly analogous to (9.17) and


in, 0)

obtain

S
|

out, 0)

-,

e~ n/2 exp

2 jk(G>)fllJ

out, 0)

(10.40)

with

j k(w)

=J

(10.46)

and

n=a
is
iqd generated by e
.

(I0.4c)

can also give an explicit expression for the S matrix in terms of In the case of one degree of freedom the the asymptotic field operators. 5 matrix corresponds to the unitary transformation q -+q, p ->p d,

We

which

The
i

field theoretic generalization

of

this

e~ n/ exp

*k (o>)fl k

exp

= exp
WaO-rfM).
2

Care must be exercised


e Rk+Bk

in writing
,

exponential operators.

Since

5k

and

JtJ

do

as can be seen by expanding the exponentials. not commute, ^ However, because the commutator of Uk and B[ is a c number, we have

e Bk e Sk

exp

(/k +

*k<?Tk)

'

exp

fl

A*k

PRODUCTION OF PARTICLES

95

By means of

(10.36) the matrix


field

S can

be expressed in terms of the


as

operators of the incoming

A^ and A[

2 bk
this expression is consistent

The reader can convince himself that


(10.40)

with

and that

Bk
to the

s*A k S

J3k

= sXs

If the source p(r,f) is spherically symmetric, then the only contribution S matrix arises from spherically symmetric or angular-momentum zero terms. That is, if an expansion of B k is made in terms of the

angular-momentum operators Bklm then only BkOQ contributes to the S Hence all particles are produced in S states (angular momentum zero).
,

matrix.

Before discussing this expression for typical forms of /o, we shall answer another question that might arise at this point. We know that for a time-dependent source the energy of the field is not conserved. Then what is the energy pumped into (or taken out of) the field by the source? For a vacuum of incoming particles this quantity is the -> oo. It can be expectation value of the energy in this state as / evaluated in the following manner:
<in,
|

# out |in, 0)
{

=
=
n

in,

n kl ,n k2 ,...n^,n

2
k2 ,...
I

(in,

out, n k .)(out, n k .|H


2
)|

out

|out, <,)(out,

n ki

in, 0)

= 22 "k, w
ki

|<

in

out, n k

Since

|{

in,

2 out, w ki }|

is

a Poisson distribution, use of (6.15) gives


|

(in,

H out
|

in, 0)

= 2 *>*t = S i |pk(^)| k k

(10.5)

is, the energy transferred to the field is just equal to the total energy of the quanta created by the source. In classical field theory = oo to we obtain (10.5) by integrating dH/dt = dL/dt from t t = + oo. Since p(r,t) is the only term in the Lagrangian which depends explicitly on the time, we obtain

That

Classically, to
field
in
<

an incoming vacuum there corresponds an incoming


its

(quantum-mechanically,

expectation value

is

zero).

By

96

SOLUBLE INTERACTIONS

a partial integration, substitution of (8.10), rewrite AJE" as

and use of

(8.9),

we can

AE =

f
/-*>

dt

J\d*r

<Pr' dt' p(r,t)p(r',t')

1A
dt
Iet

ret

(r

r', t

-O
A
<')]

r fdtdt'd*rd r'p( ,t)p(r',n~lA ot2

(r-i',t-t')-tf \r-T', t-

(10.6)

By going over to momentum space and using (10.2), we verify (10.5). In the last way of writing (10.6), we have emphasized that it is the
difference between the

incoming and outgoing

fields

which

is

relevant

for the energy loss. This is well known from classical electrodynamics, where a similar expression for the energy loss obtains and it turns out

that the radiative force

is

generated by

rad

out
<

in
.

Since the

expectation values of the field obey the classical equation of motion and the fluctuation terms are the same as for the free fields, it was to be

expected that we should obtain the classical result for the energy loss. Similarly, for a static source the self-energy f is found to be the energy change obtained by switching the source on or off.
10.2. Specific

Examples.

It is clear

from

(10.4) that particles are

created only if the source contains frequencies 2 >m. a point source with a periodic time dependence,
r)

For

instance, for

cos

o) Q t

(1

0.7)

and hence
This is again a conseas given by (10.4c), is zero unless o> > m. quence of the adiabatic principle according to which a quasistatic source should be almost as good as a static one. For (10.8#) h becomes 2 ~~ oc and therefore infinite. This is physically clear, since o>)| \6(co oo to t = -f oo. To such a source keeps radiating particles from t =
,

obtain a

finite result, let

the source radiate only during a finite period,


(10.76)

in

which case the number of particles is proportional to the time during which the source radiates. For a < m the four-dimensional Fourier
1

See, e.g.,

W.

Academic
2

Press, Inc.,
static

Thirring, "Principles of New York, 1958.

Quantum Electrodynamics,"

chap.

2,

Thus, the

source (which

is

switched on and orTadiabatically) has no such

frequencies.

PRODUCTION OF PARTICLES
transform of the source,
/> k (co),

97

reduces to (10.8#), and


:

we can

write

*/c

d(coQ
2(0

M)

'

(27T)

=
4xr

<S**
a
out
<

if

GO.*)

This result can also be deduced by calculating the outgoing current of


particles.
is illustrated physical significance of by calculating its For a spherical source it expectation value for the incoming vacuum. consists of spherical waves. This becomes obvious when the expansion

The

is used in eigenstates of the angular momentum, since such a source couples only to that part of the field which has angular momentum zero. 1 Furthermore, for times after the source has been switched ()Ut contains only outgoing waves, as we should expect intuitively. off,

(5.10)

<

These statements are most easily verified for a source of the form (10.76) and m = 0. Then, A(r,0 becomes ~r~ l [d(r - /) - d(r -\- t)] and we find
9

(in,

OU

out

(r0|in

-- [cos (r 0} =
477T
/,

t)a)

~ a]r ~ il

cos (r -f

t)a>

e~ alr+t]

~\

For large
persists.

positive

only the

first

term with an outgoing spherical wave

a the

Returning to our example with m ^ 0, we find that for small values of number of particles created per unit time is essentially //a, and the number created in a time interval dt is
n<x.dt^=^(col

-m

4n
This
is

)*

dt

if co

>m

(10.96)

analogous to the expression for radiation in classical electrodynamics and has a simple physical interpretation. We saw in Chap. 9 [see Eq. (9.16c)] that the number of virtual particles surrounding the source is of the order of ^ 2 /4?r. Since the wave number k is the velocity v multiplied by the energy co, the right-hand side of Eq. (10.96) can be
writtenas
(g M(vdt)(co)
z

or (number of virtual particles) x (distance particles with velocity v 1 can go in time dt) x (radius of cloud of particles)" and therefore represents that number of virtual particles which can leave with a velocity v within a time interval dt. This implies that as soon as the necessary energy is supplied, the source quanta start leaving the cloud with their In this intuitive picture of the creation final (real-particle) velocity. we must keep in mind that the stock of virtual particles in the process, field is inexhaustible, because they are automatically regenerated in the
,

This type of expansion

will

be carried out

in

subsequent chapters.

98
source.

SOLUBLE INTERACTIONS
,

The time for regeneration is of the order O>Q I and to get significantly more particles than are present in the cloud, we have to wait
longer than this time. In a sudden event

we can never get more than all the particles contained in the cloud, this number being obtained by suddenly switching off the source. Indeed, for this case,
tgp(r)

for
*

' (r
1

'<H; 10

for

< >

( 1(X10)

gives

and therefore
k

^S'S^ 2or
which
is

(10.11)

just the

number of

virtual particles (9.166) of the source

/o(r).

We explained in Chap. 9 that this result is quite general and can be used
to define the

number of virtual

The

scarcity of virtual

photons in the

particles. field

the reason

charge and

why electric processes are so slow. 1 has to make of the velocity v

&

of an elementary charge is particle with unit order of 137 violent

shakes off a photon, since this quantum is present in 1 per cent of the time. Therefore the cross section for Bremsstrahlung is 137 times less than the collision cross section. We shall see that we shall get a simple understanding of pion-nucleon phenomena by picturing the nucleon dissociated, for a certain fraction of time, into a pion and a nucleon. This fraction is quite sizable, and so is the probability of having more than one meson around. Hence, almost every time the energy is available, a meson is emitted by the nucleon, and multiple production is also a fairly frequent event. Another interesting idealization is a source which suddenly changes
collisions until
it

the cloud less than

its

velocity: J
rfr,0

f p(r)

for

(p(r

- vr)
.

for

< >

(10.12)

The

adiabatic switching-on process


p(r,f)

is

assumed

for
/

<

0; that

is,

- Hm^(r) e* cc-*0

for

<

2 Equation (10.12) is not relativistically correct, since the Lorentz contraction of the source is neglected. The change in velocity may occur because of a collision of very short duration (e.g., two nucleons colliding at very high energy). In any case, v should be <1, since otherwise even the uniformly moving source radiates (Ceren-

kov

radiation).

PRODUCTION OF PARTICLES
and hence

99

The two terms


moving

correspond to the source at rest and to the just the number of virtual particles in the difference of the fields before and after t = which accounts for the This agrees with our earlier remark that what real particles at t = oo. is radiated is the difference between the incoming and the outgoing field. For m = and for a spherically symmetric source, (10.14) becomes (w = k, and 6 is the angle between v and k)
in p k (K )

source.

Hence

it is

A
(2<rr)

2k

(10.15)
v cos 07

long as p k

which exhibits the well-known Bremsstrahlung's spectrum ccdk/k as For a point source the energy loss n kk is proportional 1. to J dk, so that in every frequency interval the same amount of energy is In this case the expression for n diverges at both ends. The radiated. upper limit is easily fixed up by taking an extended source which

averages out the very high frequencies. The integral diverges at the lower limit even for an extended source, since normalizing the source 3 = 1 implies pkss0 = 1. This divergence means that we always J d r />(r) deal with infinitely many quanta (both real and virtual) of low frequency. For the virtual particles this can be understood as follows. For a point source the self-energy (in r space) is given by J d*r, where ( is the

&
.

electric field strength,

and

this is (e /47r) J rfV/STrr

It

diverges linearly

end but converges at the upper end, in agreement with our expression (9.2), since the upper limit in k space corresponds to the lower limit in r space, and vice versa. The particle density has an extra power of k in the denominator or of r in the numerator and diverges at both limits as in (9.22). The infinity of infrared quanta corresponds to the 1/r behavior of the Coulomb field at large distance and is not removed by smearing out the source. Consequently, an infinity of
at the lower

when the asymptotic parts of the Coulomb field are The number of infrared quanta within the radius of the changed. universe 1 is <1, and the "problem" is somewhat academic. It does
quanta
is

radiated

not arise for


1

m ^ 0,

since then

</>

~ e- mr

/r

and has a

finite

range.

From

(9.22),

Taking

r, n t n

as the electromagnetic radius of the electron

(M0~13 cm), we find


47

for rmax

- lO^V

37

~ 10+

cm

However,

nmiveree

~ 10

27

cm.

CHAPTER

11

Pair Theory, Classical

In this chapter we shall turn our attention 11.1. General Remarks. to a system wherein the interaction term is quadratic in the field The general form of such a term, as discussed in Chap. 8, is variables.

L'(0

- (tPr c/V <(r,OK(r,r',0<(r',0


2 J
<5

(1 1.1)

which includes, as a special case, a local potential K(r,rV) oc 3 (r' r) However, there are cases where nonlocal potentials acting on the field. of the form (11.1) are important, as in the many-body problem of
Furthermore, some types of linear couplings may correspond to a coupling of the type (11.1). For instance, an electromagnetic 'field A interacting with nonrelativistic particles of charge e has a Lagrangian
nuclear physics. 1
effectively

L=

2m

\d*r y>*(r)[iV

*A(r)] y<*)

(11.2)

which contains a term of the form


V(r,r',0

(11.1) with
3

=
=

<5

(r-r>(r,0
2

(H-3)

p(r,f)

|y<r,0|
is

In a relativistic electromagnetic theory the situation


1

more complex;

See R.
I,

J.

vol.

chap.

1,

Eden, in P. M. Endt and M. Demeur (eds.), "Nuclear Reactions," North-Holland Publishing Co., Amsterdam, 1959.
100

PAIR THEORY, CLASSICAL


in particular, the 6 function

101

becomes smeared out over a Compton

Similarly, the relativistic pseudoscalar y 5 interaction 1 in meson theory can be shown 2 to be equivalent to a leading term of the form (1 1.1) with
particles.

wavelength of the charged

J/(r,r',f)

3
<5

(r

r>(r,0

.4)

the nucleon density and y;*( r r ) and ^o s the mass of the Many other terms appear in the transformation, one of which will be the subject of the last part of this book. Also, that part
is
>

where p

nucleon.

is equivalent to (1 1.1), In this and the following chapter we shall Chap. frequently refer to electrodynamics as an example of such a theory. However, we shall study this problem not so much because its physical

of the Lee model which can be solved explicitly

as

we

shall see in

13.

applications are of interest but because

it

concepts which we
scattering

shall encounter in

pion physics.

conveniently introduces many Thus, the model

allows us to investigate bound (excited) physical source states and

phenomena.

Following our general approach, we shall first study the solutions of the equations of motion from the point of view of the classical field and shall defer the application of quantum mechanics to the next chapter. We shall limit our discussion to the case for which V does not depend

on t.*\ Furthermore, in any explicit evaluation we 3 always take V to be separable and of the form
K(r,r')

shall, for simplicity,

-A/tfOpOr')

(11.5)

and /o(r) is spherically symmetric. The development we can also be carried out with the general form of V, but this the theory of singular integral equations. requires more tools from Since the general case does not lead to essential new features of physical The equation of motion interest, it will not be analyzed in detail.
where A
is

real

shall present

remark.
2

The reader who has never heard of this The Lagrangian is L' f y*ftvv>

is

advised to ignore the following

S.

coupling equation We could equally well introduce a dimensionless coupling constant^ ^


A
is

F. J. Dyson, Phys. Rev., 73:929 (1948); L. L. Foldy, Phys. Rev., 84:168 (1951); D. Drell and E. M. Henley, Phys. Rev., 88:1053 (1952). H Of course, we keep the liberty of switching it on and off adiabatically. 1

In the

the

constant A has the dimension of length or (energy)" Aw. When


.

arise, as

Hamiltonian is no longer positive definite, negative, the we shall see in the next chapter.

and

difficulties

may

102
(8.12)
is

SOLUBLE INTERACTIONS
written in integral form by introducing the asymptotic fields:
1

<Pr' d*r"

A ret(r -

r' f t

OK(r',r'W',O
(11.6)

out
<

(r,0

+jdt'

dB r'

<Pr"

A adv(r -

r', t

- O^(r>'W/)

The Fourier transform of


integral equation.

this

obtain Fourier transform 1 of the field

To

equation has the usual form of a linear it, we introduce the four-dimensional
<f>:

and the three-dimensional Fourier transform of the time-independent


source: 2
K(k,k')
k

-J

K(r,r>-'

k '' r/

d*r d r

The reason

that the four-dimensional Fourier transform, rather than the three-dimensional one, is introduced for the field <f> is that its time dependence is not that of a free field and is not known a priori. Writing
in terms of the

above transforms and introducing four-dimensional Fourier transforms of Aret or Aadv as given by (8.6) and (8.7), we obtain 3
,

Such an equation can be considered to be the


1

limit of an ordinary linear


normalization but shall retain

We shall henceforth use the continuum (L

<x>)

S-* (\l2-np d*k

as a symbolic notation to indicate integration over the continuous

spectrum as well as summation over the discrete spectrum, when it exists. The correspondence between the continuum and finite box normalization was made clear in Chap. 5. The normalization of \l(2iffi rather than l/(27r) 2 is made for the sake of convenience. The inverse transform is then

is again chosen so that for a separable source, as defined by The sign convention is also chosen for convenience and has A. K(0,0) the added feature that for a local potential K(r,r') oc <58(r r'), we find that F(k,k') k'. depends only on the momentum difference k 3 have introduced k as a convenient shorthand notation to indicate the paths of integration discussed in Chap. 8. The limit e -* is to be understood ; hence Me, where is any finite number larger than zero, can also be written as just /e.

The normalization

(1

.5),

We

PAIR THEORY, CLASSICAL


algebraic equation (which

103

it actually would be, had we used a finite normalization volume 1) and can be treated by similar methods. For the separable source, (11.7) contains A and Q as parameters and repreand in sents, in general, a linear relation between However, if the determinant of (1 1.7) vanishes for certain values of the parameters, or out = 0. We shall see later on then there is a solution with ln In = that the solutions of the homogenous equations (e.g., 0) correto bound states, whereas the solutions of the inhomogenous spond

<

<f>

<

<

<

latter.

equations are scattering states. For the moment we concentrate on the In general, the solution of (11.7) can be given only as an infinite series, but for V of the form (11.5) the Fredholm series terminates 2 after the second term. Indeed, we find that

2 t

*'.*)

1.8a)

p(-k)

= = p(k)

/>*(k)

for a real source


for a spherical source

(11.86)

p(|k|)

This can be integrated immediately to give

(11.9)

Now
more

that

we have obtained a

specific solution,

we can
shall
<

return to the

familiar three-dimensional Fourier decomposition.

To

rewrite

(11.9) in three-dimensional

momentum

space,

we

make

use of our

ln In terms of the knowledge of the free-field time dependence of continuum analogue of the Fourier decomposition given by (8.16), we have
.

f J(
4.

-ft

where
1

ln(+)
<^

(k,0

is

the positive-frequency part3 of

ln <^
,

proportional to

For

this case the

problem

is

treated in

G. Wentzel, Helv. Phys. Acta, 15:111


<

(1942).
2

We shall carry out


pattern.

the manipulations only for ^ in ; the ones for


5.

out

follow the

same
3

This notation was introduced in Chap.

104

SOLUBLE INTERACTIONS
,

e~
e

i(ut

and ^'"^(k,*)

is

iutt
.

The four-dimensional Fourier transform of


from zero only for

the negative-frequency term, proportional to in (r,0, as defined


<

earlier in this chapter, differs

K =

oj

and

is

f" 7- J- o 277

(2eo)*

(2co)*

Writing in terms of the positive- and negative-frequency components of in and making use of the above results, we can rewrite (11.9) in a convenient matrix notation 1 as
<

=
with
(k
|

(k

Q + k'# ln(+) (k',r) +


|

(k

Q.
| |

k')f

n( - }

(k',0

(1 1.10)

Q,

k')

- (2.)3

^,

y)

/2

>i

(1U1)

It

follows from (11.86) and (11.11) that for a real spherically symmetric

source

D*(/c

)- D T
in the

2 (/c )

We

shall

sometimes write (11.10)

shorthand notation

where the

field

equation (8.11) requires


<&
2

Q-Q<w a = FD
Here
co is

(11.13)

which can

easily

be checked.

considered a diagonal matrix

and matrix multiplication with K(k,k') is implied on the right-hand side. To find the totality of the solutions of (11.7), 11.2. Bound States. we still have to consider the homogeneous solution. The function D(k 2) defined by (11.12) is the Fredholm determinant of (11.7). As
1

Double

subscripts are integrated over in the sense of

The matrices occur-

ring here are not to be confused with the operators in the Hilbert space of quantum mechanics. 2 To differentiate between the matrix w and the function co, we place a bar over the former.

PAIR THEORY, CLASSICAL


stated earlier, if it vanishes for certain eigenvalues of in solution of (11.7) exists with 0. shall
<

105
,

k b then a

We

when

this occurs,
is
1

but for the

moment we

subsequently study simply note that such a

solution

<f>(k,K

constant

d(K Q

co b )

or This
satisfies (1 1.7)

<f>(k,t)

constant

e
g

imt

(1

1.14)

provided that

[e.g.,

D(kl)

0].

We shall shortly show that this solution represents a


<

bound

has the proper spatial behavior for such a state. In order to find when (1 1.7) has a solution with ln = 0, we have to 2 study the properties of the function D (k ). For real values of A: we
state, in that the classical field

may

use 2
(11.15)
i

x
to split

x
parts,

into

its

real

and imaginary

and D,

respectively,

47T

(11.16)

If p(k)

is

a function that goes to zero smoothly as k 3

oo,

then (11.16)

can never vanish for real values of k. To study the 2 analytic behavior of D in the complex plane, we replace k by

shows that

the complex variable 4 z boundary values of

=x

-f />.

Then

2
{

(k

and D_(k 2) are the

when
1

the real axis

is

approached from above and below, respectively.


Then

We shall assume that p(k) is spherically symmetric throughout the remainder of


P means
By
this

this
2

chapter and most of the succeeding one.

p(k)

- p(|k|)

p(k).

Cauchy's principal

part.
.

we mean

patial source, p(r\


4

that p(k) is always positive. This would not be true for a which cuts off sharply beyond a certain radial distance r = r

db/e is

then included in y.

106

SOLUBLE INTERACTIONS

(11.17) it appears that D(z) is analytic in the whole complex plane, save for a branch line along the positive real axis where the imaginary part is discontinuous. To find the zeros of Z>(z), we note

From

that

is

zero only for

0.

To

locate the zeros of D(z) 9

we

therefore need

to look only at the negative real axis.

There we have 1

\"/

\"/
2

\ft

"

V.

**** ^/

and
1.

since
:

D(<x>)
0:

if

J3gr

|p(#)|

<

oo,

we have

in these

circum-

stances
A

2.
.v

A
kik

> <
if

D has

0: Z> has

no zeros in the cut plane. one and only one zero for a negative

real value

of

otherwise none.

These conditions are shown graphically in Fig.


only for A

11.1.

It

was

to be

expected that bound states occur

<

0,

since

this corre-

sponds to an attractive interaction, and we see that our form of the


potential
state.

admits only one bound


well

It is

known from

ordi-

nary quantum mechanics that even a short-range attractive interaction will not have a bound state if the potential energy is too weak to

overcome the

kinetic zero

This is energy. A but D(0)

point also true here for


0.

<

>

Fig. 11.1. Diagram for D(x) for negative values of x and three possible

11.3. Behavior of the


trix

Wave Ma-

values of

A,

with A x

>

A2

>

Ag

and A2

The matrix 3
fields

which trans-

and A3 both

<

0.

forms the local


asymptotic

field variables into

has some interest-

ing properties which are essential for the


1

quantum

theoretic treatment.
(z),

We

E
2 3

hope that the reader

will

not confuse

x = Re

y = Im (z), and

(xj,z).

is not satisfied for a point source. In the literature, O is sometimes called the wave matrix; see C. M011er, Kgl. Danske Videnskab. Selskab, Mat.-fys. Medd., 23(1) (1945).

This condition

PAIR THEORY, CLASSICAL

107

We
but

shall see that

it is

one-sided unitary

if

bound

states 1 exist in the


(11.20) (11.21)
if

sense that2
ii

aQ
t

ii

=1 = l-<P =

where

The

the projection operator onto the bound states one-sided unitarity is a phenomenon which occurs
is

any
1

exist.

only in

an

infinite-dimensional space, since for finite matrices iiii t f 1 To prove these propositions, we write implies ii ii

always

where

(k'

I 1

Q =1+R R *' k) = -i- 2* >*W 2 - 2 i< D


fc
I

(11.22)

(fc )

fc

fc'

and

(k '|
1

R*

| '

k)

_J__OOp!W_
D.T (fc' 2)
2
fc'

2
fc

ie

so that (11.20) becomes

R{R = -(R + Rl)

(11.23)
:

This can be explicitly demonstrated for a separable potential as follows


2

d*q

A p(k')p*(
2

[k'

~~q
2 )

i)(k

(k

k'

ie)D-.(k'

)D + (k

(2^f

Ik*- q* -

-q

+/

-(k'

R+

RT +

k)

(11.24)

The same relations can be proved to hold for /?_. The verification of (11.21) proceeds the same way, and we
(k'
|

find

R + R\

k)
*)
\

P (q)\*

=
fc

Ap(fcV(fc)
2

dg
3

|p((?)|

/
2 2

A
ic
2
/c

k'

J (27r)

D+(q )D-(q

)\k'

- q2 -

- g2 +

iV

(11.25)
1

The bound
is

states

have binding energies <m.

The ground

state of the physical

excluded from these considerations. It always exists, unless the source is and we shall not consider such problems. 2 Because of the matrix multiplication introduced earlier (see footnote 1, page - k'), It is, in fact, 104), <k ftiOfc k'> is not dimension less. equal to (2ir)W(k and the 1 in (11. 20) and (11.21) is to be interpreted in this sense.
source
unstable,
|
|

108

SOLUBLE INTERACTIONS
(11.16) this last integral can be written

With

*,(*>(*)
z
fe

2
fc'

+
+

* * [_1

/.

In Jo

LD_(4

___1_1 /__J D U' 2

+ (q*)l

ic

2
/e

L
A/
2

ic 27ri

-f Jc

z D(z} (D(z) \z

k'

2
fc

-J

(11.26)

where C is a contour going below the real axis from -f oo to and above it from to oo. Since D(oo) -> 1, an integral over an infinite circle contributes nothing to (11.26) and can be added to the closed path shown in Fig. 11.2, to complete the contour. We can evaluate this

by means of Cauchy's theorem, using the analytic properties of cussed above. If D has no pole, we obtain
rk (k
>
I

D dis-

| | I

k, k)

and

if

there
T

is

one,

we

find 1

/
|

k2

~
2
/c'

ivLD-Ot'

2 )

D + (fc 2)
the
(k'

D (kl)\kl~k' "&(}
f

k\

*n

Returning to (11.22), we see that


first

case implies
f
|

n+a+

k)

(27r) 6

3 3

(k

k')

(11.270)

whereas the second one results


|

in

k)

- (27r)V(k b

k')

-U

(k')U* (k) b

(11,276)

- T-^-T*
Fig. 11.2. Integration contour for Eq.
(11.26).

P(k)

LD'(^)-I

(11.28a)
(1
1

Comparison with

.14) allows us to write the

Uk,t)
1

M Uk)e-

time dependence of

as

(11.286)

D'(kl)
2

s dD(x)ldx
A

Note that

ever, D'(kl) is

x = k * and is given by (11.19). must be negative for the second term to appear in (1 1 .28). How2 also negative, as shown by (1 1.19), so that A/D'CA; ,) is always positive.
\

PAIR THEORY, CLASSICAL


It is

109

wave function Ub of the bound state is correctly normalized. That this wave function does indeed represent a bound state follows from the
spatial

a simple matter to demonstrate, with the aid of (11.19), that the

dependence of the Fourier transform of

U (k).
b

Since

Arf

is

negative,

we

find

and at large distances from the source this wave function has the characteristic spatial dependence of a bound state,
Iimt7 6 (r')=
r

r^r L2D'/c|)J
'(/c|)J

TrA

i* e
'

Mr

where

coj,

function

= U

(m
is

Furthermore, the wave \k% |)* is the binding energy. 1 orthogonal to the scattering states in the sense that
r /PL-2-iJ (ZTT)

U 6 (k)(k *

k')

(11.29)

To conclude this chapter, we turn to the connec11.4. Scattering. From (1 1.10) and tion between the incoming and the outgoing fields.
its

analogue for outgoing

fields,
(+)in

we

obtain 2
(
|

#k,0

= -

(k
(k

+ (k Q_ (+)out Q. k')0 (k',r) + (k H+


|

Q+

- )in
(k',f)
(

k')<

(k',0

k'ty
|

- )out

k')0
out
<

(k\r)

(11.30)

have the free-field and Since this equation holds for all times (< time dependence), we can split it into positive- and negative-frequency parts to obtain, by means of (1 1.20),
+)ollt

in

^<
^(-)out

= QlQ + ^ +)ln
==Q t+Q _^-)in
(11-31)

Because [^ (+) ] r
first

one.

It is

the second form of (1 1.31) is consistent with the important to recognize that Q!..Q + is unitary whether

there are

bound

states or not.

We

find,

by means of

(1 1.21),

= =
|

1 The That (k ft k') represent scattering states will be made clear in Sec. 1 .4. left as an exercise for the reader. proof of the orthogonality (1 1.29) is 2 The bound state is now ignored, since its presence would not change the following
1
|

considerations.

110

SOLUBLE INTERACTIONS

since the term with ty drops out because of the orthogonality relation If the source is spherically symmetric, p(k) (11.29). p(|k|), as has

been assumed, then

QLQ+ can easily be diagonalized.

As

for the linear

coupling, only the spherically symmetric part of the field is then coupled. This is exhibited by expanding the field in spherical waves, in which case becomes diagonal in / and m. It is essentially unity for matrix

&

elements with
|

/fc,

/,

m)

that have

^ 0,

and for

we

find 1

',

0,

Q+

fc,
|

0, 0)

= 7- d& k <ttl r

yftfc'X*'

k)Y(k)

v (11.33)

By means of (11.15) and (11.16) convenient form:


(*',0,0|Q + |*,0.0)
r
L

it is

possible to cast (11.33) into a

more

^jx^i 4w D
2

+ (fe )J

2^ 2
Jt

Ap
(fc

^yw -

k'*)D + (k*)

,,,,
Similarly,

(11.34)

we have
(k', 0,

Q_
|

fc,

0, 0)

(fc',

0,
1

ii t

Jt,

0,

0)-D-\K
)

= (fc',0,0|Q +
and hence 2

|fc.O,0)|^
1

(11.35)

QlQ + -(Q + \
or
(fc' f

5iK- D 0^= D DJ
+
fc,
|

0,
|

n!n +
is

0, 0)

(5(/c

Since this last matrix

out fidds
1

is

particularly simple in this representation.

diagonal, the connection between the in and If we define the

See Chap. 5. Matrix multiplication

now

2 )//;

implies (l/ZTr

dk.

PAIR THEORY, CLASSICAL

111

phase

shift 6(k)

by

<*', 0,

Q!.a+

fc,

0, 0)

<5(fc

k')e*

(11.37)

then

^(-f-)out

= e2,S<*y + )in
D l(
2
/c

^<-)out == ^-artUry-Oin

and

tan

<5(/c)

--

-)

4*

_ _

(H.38)
< (ll
.

These equations

will

be of importance in the next chapter, where we

shall study the particle aspect of the problem and shall relate these expressions to scattering cross sections. Here we simply wish to

remark that
zero.

difficulties are
is

Since p(k)

encountered when the source size shrinks to then equal to 1, we find, from (1 1.39),

_A-L =

(11.40)

It will be shown in Chap. 12 how these difficulties can be circumvented for observables by a limiting procedure. As for the linear coupling, pair theory can also be solved analytically for several sources. This is a rewarding problem, since it allows one to study the scattering on many centers. Furthermore, the "potential energy" is not just the sum of potentials between pairs, as in the linear It would, however, carry us too far afield to discuss these theory. 2 problems here.
1 That this is the usual phase shift (8.32) follows from (11.38). It is in the above sense that O represents the scattering "wave function." It is normalized according to (11. 20) and (11. 21).

See, however, Wentzel, he.

cit.

CHAPTER

12

Pair Theory, Quantum-mechanical

12.1. Quantization and Commutation Relations in the Presence of a Bound State. We shall first use our previous results to check the consistency of the local field commutation relations with those for the
field. Although, in general, this equivalence is assured by the adiabatic principle, it is not so obvious for the pair theory as for the linear coupling, where the local and asymptotic field operators differ only by an ordinary number. In particular, if we immediately consider

asymptotic

more general problem, which includes a bound state, then the canonical commutation relations require that, in addition to the incoming field (1 1.10), contain a term which represents the bound state. As we saw in Chap. 11 [see (11.28c)], the spatial dependence of the bound-state wave function is
the
(/>

f
C/ 6 (r,0

ex

fr'
J
2

p(r')

co fc

- (m +

k*)*

= (m 2

2
|/c 6 |

)*

so that the corresponding particles remain concentrated around the in source in the limit t -> eventuoo, whereas those associated with An Ansatz which, as we shall see, satisfies the ally disappear at infinity. commutation relations is (always employing the matrix notation introduced in the last section)
<

#k,f)

(k

tt +

<k

0- k')0|

(k',0

C7 &

*(k)(0

(12.1)

112

PAIR THEORY, QUANTUM-MECHANICAL

113

where

in

is

the usual operator (8.17) for the incoming field and

DM J = DMl] =
With these commutation
relations

(12.2)

and the familiar ones for

in
<

we can

convince ourselves that the canonical commutation rules are satisfied.


This would not be true if the terms proportional to t/6 (k) were omitted or treated as c numbers. For simplicity, we shall carry out the proof for t 0; it will be evident that things work the same way for other

times.

From

(12.1)

and (11.90) we deduce


|

= (2ir)=

(k

n + (2o>)-*

k')X(k')

+ U (k)A
b

b (2a> b

r*

(2ir)-

(k

Q+

(I)

k')

A(k')

(12.3)

Chap.

In terms of the Fourier transform <(k,0) and <^(k,0) introduced in = ^ f (k,0), and the 11, the reality of <(r) requires that <(-k,0)

usual commutation relations become


(k',0)]

(12.4)

These relations hold


(k
|

if

a + w Q\ - n_.w
l

til
|

=
1

which is actually satisfied, as can be shown by use of (1 commutation relation


.
4-

.35).

The

last

is

id\k

- k')

(12.5)

also satisfied [see (1 1.276)], but only because we introduced the extra bound-state terms in (12.1) together with the commutation rules (12.2). Of course, if there is no bound state, then we merely put Ub = 0, and
(12.5) still holds. 12.2. Scattering.

and out

fields in

We briefly discussed the connection between the in Chap. 11. We found that this relation is most easily

expressed

when

eigenfunctions.

the fields are developed in terms of angular-momentum This connection remains the same when the theory is

114

SOLUBLE INTERACTIONS

quantized, as follows from (12.1) and the same equation in terms of


^out (to) the

Yhj s p a j r o f e q ua tions also shows that no transitions occur from continuum to (from) the bound state. We can rewrite (11.37)

as

B lm(k)

S~ l A lm(k)S

= A lm(k)

for
)

1^0
(12.6)

and deduce that 1

M
From
in

Aw(k)2d(k)A w(k) dk

1 J
(12.7)

f
lm J

and a similar expression for Nout it follows that N ln = N outt e.g., there is no production of real particles. Only that state which has an = differs from the corresponding state with incoming particle with / an outgoing particle by a phase shift 26. An incoming plane wave, which is a mixture of eigenstates of L, will be an outgoing plane wave 2 d From I). plus an outgoing spherical wave with an amplitude (e
9 9

we can deduce quantum mechanics.


this

the transition probabilities as in one-particle The scattering cross section (8.34) is the usual

expression
da(k)
dO.

sin

d(k)

k2
is

We shall now briefly comment on resonance scattering, which important feature of the pion-nucleon system. If we examine

an

A
for values of A that are small

|^)|
2

(27T)V-fc

1 enough so that D" has no

poles,

we

see

that if p(k) goes to zero sufficiently rapidly for high momenta, then <$ -*> both as k -> and as k -* oo. For finite values of the momentum, d assumes positive or negative values depending on whether A is
1

We remind the reader that

Wimw

AI

PAIR THEORY, QUANTUM-MECHANICAL

115

If A is negative and its magnitude is ingreater or less than zero. 1 creased, then, beyond a critical value, l will become negative and Df 1 will have two poles, as shown in Fig. 12.1. In such a case, the phase

2 Fig. 12.1. Plot of OjCfc ) for three negative coupling constants,

and A

A2

>

A3

shift will increase beyond 90, at the momentum/:,, defined by For momenta close to kr we can expand
2

0.

2
)
(/c

D^/c

where
= *?

In this approximation the phase shift can be expressed as


tan d(k)

r/2
(12.90)

with

A kr

P (k r )
(12.96)

For the
find,

derivative of the phase shift with respect to the


A:r ,

momentum, we

near

In the following we shall discuss the possibility that as the magnitude of the coupling strength is increased a resonance appears, followed by a bound state. This does not actually happen for 5"-wave scattering, but may occur for P waves with which we shall be concerned in the third part of this book.

116

SOLUBLE INTERACTIONS
d'

which shows that

has a sharp

maximum

near

co r

for

rapid increase of d with k near a certain value of called resonance momentum and defined by D^k*)

< o> =k
r9

f .11

A very
and
long

sometimes

= 0,

indicates that

incoming

particles in this energy range tend to stay with the source


is

that the time required for the scattered particles to be emitted

Fig. 12.2.

Shape of the cross section near the resonant energy

co r .

1 This compared with the transit time of the particles past the source. can be seen by considering an incident gaussian (spherical) wave

packet

~ exp p
with v r
out
|l/
<r.f)|

[> I
b*

v r tf] '

b\t)
t

J
2
.

=
s

kr \m, b\f)

/m*b

On

expanding d(k) near kr we


,

find for the outgoing

wave packet

expl

_tLz^LL
.

where F- 1

2d'(kr)lvr

This

is

also gaussian, but

its

peak has

shifted

1 This condition is not satisfied for 5 waves, but we shall continue the discussion below as if it were. 1 For another discussion of resonance phenomena, see J. M. Blatt and V. F. Weisskopf, "Theoretical Nuclear Physics," chap. 8, John Wiley & Sons, Inc., New
York, 1952.

PAIR THEORY, QUANTUM-MECHANICAL

117

Such a resonance is often called a "virtual" or "metastable" state with a lifetime F" 1 It dominates the scattering for particles with energies close to w r since the phase shift is then close to 90 and the cross section is near its maximum. Inserting (I2.9a) into
. .

by the time delay r-

the cross-section formula,

we

find

= 4,
,

The

is plotted for o> close to co r in Fig. 12.2. The exThe first factor (12.10) has a simple physical interpretation. pression is a geometrical one. It is the area perpendicular to the incident beam

cross section

waves of momentum k. The angular-momentum / = second term expresses the probability that a particle with energy is in the metastable state which has a width F and is centered around o>r Thus, (12.10) states that, for k t& k r only those particles which have the
available to
>
.

right energy scattered.


If

and angular momentum to form the metastable


|A|

state are

we

increase

which the phase shift d goes through 90 comes closer to k = and the time delay approaches oo. This delay is reached for A = A c defined by
/)(0) state

further, then the smaller of the values of

for

=0.

If
|

A
|

is

increased further, a

bound

state occurs.

A bound

(attractive forces) and corresponds to resonance at negative energies has as a consequence that d is negative

which occurs only for A

<

for

low energies of the incoming

particles.

This happens for A 3 of Fig.

11.1.

In the limit of high energies and for a finite source size, tan 6(k) is given by its first term in an expansion in powers of A (Born approxi1 On the other hand, for k -> there may be mation), since D( oo) = 1
.

sizable corrections to the


differs

Born approximation,

since the exact result

by the factor l/D^O). This may even become zero, as it does for a point source. This change of the cross section from its Born-approximation value has a simple physical interpretation if we consider the simplified version of electrodynamics 2 which includes only

from

it

the e*A 2 /2m find that

then (and similarly for the y 5 theory). the total inertial mass of the electron, including the inertia of the electromagnetic field, and thus that the scattering

term in

m=

(1 1.2)
is

We

D(0)m Q

and high-energy limits are <? /w and e lm^ of the higher approximations in e 2 (which respectively. For low frequencies represent virtual photons) is only an inertial one. the virtual photons move rigidly with the charge, and hence the relevant
cross sections in the low-

The

effect

That the
it

mation, as
2

That

first term in the expansion does indeed correspond to the Born approxishould, is shown, for example, by J. M. Blatt, Phys. Rev., 72:466 (1947). is a vector field does not introduce essential complications.

118

SOLUBLE INTERACTIONS
is

mass mass
for

m.

This simple result is typical gives the scattering cross section. nonrelativistic quantum electrodynamics. In the relativistic

For high frequencies they cannot follow, and the bare

theory virtual pairs of charged particles are created which change not only the effective mass of the electron but also the effective charge

(vacuum

polarization).

In our meson theoretic applications we shall find it cpnvenient to introduce a renormalized interaction constant such that the exact expression for the phase shift, extrapolated to the unphysical energy
co 0, is formally equal to the Born approximation calculated with the renormalized coupling constant. 1 If this procedure, which may seem both unnecessary and arbitrary, is carried out, then we obtain a finite phase shift d for finite momenta even for a point interaction. For a relativistic theory such a point interaction is required by Lorentz invariance; after renormalization, the results will be less sensitive to the shape of the source p(r). In the present case the above procedure

defines A f

by
A
A

WX
A fixed
renormalized coupling constant A r

[or, equivalently,

lim

<

end of the

noted at the implies a certain dependence of A on the source size. last chapter that lim d(k)/k (or A r) goes to zero for a point
A:

We

A. Thus, keeping A r fixed and finite, we obtain the following relation for A in the limit of a point source (or a sufficiently small one)

source and a fixed value of

This means that A approaches zero from negative values, irrespective of the sign of A r Although this leads to no difficulties here, with a single for two or more source, it creates a bound state with energy This implies that there is no state of lowest energy in the sources. In the Lee model, which we shall study next, the same kind of theory. phenomenon occurs and can cause severe difficulties, even for a single
.

<m

source.
1

This* procedure

is

often called "charge renormalization" because of

its

analogy

to the
S. S.

same procedure for the charge e in quantum electrodynamics. See, e.g., Schweber, H. A. Bethe, and F. de Hoffmann, "Mesons and Fields," vol. I,

p. 297,

Row, Peterson

& Company, Evanston,

111.,

1955.

PAIR THEORY, QUANTUM-MECHANICAL

119
is

In terms of the renormalized coupling constant, the phase shift


given by

this form has several virtues. For instance, the integral is weighted against high energies, where it is sensitive to the exact shape of p(r\ so that it even remains finite for a point source.

As was pointed out above,

The
co r ,

scattering may still exhibit a resonance in the neighborhood of which Eqs. (12.9)
is

if

?(k*). replaced by Ar and D^k*) by Energy Expressions in Terms of the Asymptotic Fields. We in terms of the asymptotic next turn to the problem of expressing

about a particular energy and (12.10) are applicable

12.3.

fields; as for the linear scalar coupling,

shall thus find the energy (or mass) renormalization, that is, the energy of the incoming vacuum. We shall see that the phase shift 6 is directly involved in this computation.

we

By means of the
as 1

field

equation (8.11),

we can

write the Hamiltonian

= =

J i[^(k)(k) J i[* WOO

+ Ak)" ~

^W ]V(k.k')#k')

^
(12.12*)

ifk

^(k)#k)]

where we have used <(k) = <(k,0). Inserting (12.3) and computing the second time derivative of from (11.90), we obtain, in matrix notation,
<

V b co*A b + Q.fflU 1 +
1

t/?co*4]

(12.126)

The Hamiltonian

(6.5),

The

ot (12.12) has not been reordered according to the prescription pre included. so that the usual zero-point energy Q of the noninteracting fields is i nergy 3 factor (2ir) arises because of our definition of matrix multiplication.

120

SOLUBLE INTERACTIONS
as well as (1 1.29),

By making use of (1 1.36) and its hermitian conjugate we can simplify this expression to

d*k A\k)A(k)a)

(12.12c)

3 is where shall shortly study %a) b formally equal to JJ w d k the connection of this energy to that of the incoming vacuum. Aside from this constant, is the energy of incoming and bound particles. Its spectrum in relation to that of the free field is shown in Fig. 12.3.

We

Energy

Energy

jwo

quanta

Two quanta

One quantum

One quantum
With source

Without source

Fig. 12.3. Energy spectrum with bound state is also shown.

and without a source.

For the

latter case the

The above development


since the factor
<a\

is

possible only
real.

if (o b is real, e.g., \kl\

<w

2
,

was assumed

If

o>

<

= -|(/? 2 w &92)> with an oscillator with a repulsive force, 1 In particular, the energy eigenvalues no discrete energy spectrum. then have no lower bounds, and no particle interpretation is possible. It is physically clear that a disaster starts to happen as soon as |Af -> 2 When this critical momentum is reached, the total energy of this state is zero, and an infinitesimal amount of energy can create an These particles could appear infinity of particles in this bound state.

0, then

we

are dealing and there is

as soon as the source

is suddenly switched off. Thus, if A is decreased then an instability will set in when it is negative and its adiabatically, 2 = magnitude exceeds a certain critical value determined by kl

state has a
is

Formally this resembles a "runaway solution," since the time dependence of the term proportional to ef* when w 6 becomes imaginary. This resemblance not accidental. Compare N. G. Van Kampen, Physica, 24:545 (1958).

PAIR THEORY, QUANTUM-MECHANICAL

121

Since the Hamiltonian (12.12) was not reordered, it seems from (12.12c) that in the absence of a bound state the zero-point energy of
It thus exactly equal to that for the free field. no energy renormalization for the pair theory, unlike the linear static case. In the notation of Chap. 9, we seem to
is

the Hamiltonian

appears that there

is

obtain <f 0. However, the difference between two infinite (and. therefore undetermined) terms may be different from zero even if they look the same, the result depending on how the limit is taken. To obtain a well-defined answer, we shall evaluate the energy difference between the physical vacuum states with and without a source by using a finite normalization sphere; only later shall we go to the continuum limit. We assume that no bound state is present. The boundary condition (5.9) tells us that the allowed values of A: for the / = part of the field are slightly different in the presence of the source because of the
if

phase shift d. Taking the usual standing-wave solutions, we find that without interaction the possible values of k are k n defined by
,

sin

kn R

=
is

then with the source the boundary condition


sin

= [*;* + WJ]
k' n

oi(toidkn

<kj
=
kn

(12.13)

This gives a change in the energy eigenvalues,

n
co n

r R

and hence a change


<ro

in the zero-point energy,

= ~2>'--2co = 2
fc'

2#

2-*W
fc

(12.15fl)

co

In the continuum limit this difference remains


<^

finite

and

is

=-

f
27rJo

"W - 6(k)
co

(12.156)

Thus the energy shift or renormalization ^ due to the source is different from zero because the source brings the eigenfrequencies out of tune and thereby causes a change in the zero-point energy of the field. For the static source studied in Chap. 9, the origin of f was different. Because there was no scattering in that theory, terms of the type (12.15) did not contribute. The latter effect also causes forces between

122

SOLUBLE INTERACTIONS
electric frequencies depend on effect has been verified experi-

macroscopic bodies, since the possible This amusing quantum their distance.
1

mentally. The relation (12.15) between the phase shift


energies in a finite

and the change of the

volume shows why an

2 2 V where roughly be pictured as changing o> by XjV within the volume 2 IfR) is A/K P = 0. The change of w for the low-lying states (k times the probability of finding the particle within this volume. Hence

action gives a positive (negative) phase estimate the magnitude of the phase

attractive (repulsive) interIt also suggests a way to shift.


shift.

The

interaction

can

dO)

kd(k) rw -

IV
VR*
(12.16)

2R
and
in essential
d(k)

~
(12.8).

I
j

Ak

rC/v

agreement with

conclude our discussion of the pair the distribution of virtual particles. To theory, we shall investigate this end we need the connection between the operators ak [see (8.18)]
12.4. Virtual Particles.

To

which create the virtual particles and the A k we have been using so far. This connection is easily obtained by comparing (8.18) and (12.3). With our matrix notation and in the absence of a bound state, we find that 3 (in our standard representation l* = Q T )

- M*A + M a = MJA* + M2 A M! = Kdta+fiT + dT'O+fl^irr M = Koftj+oT* - oTta+ca^Tr)a


1

V
1

(12.17)

that the destruction operator for virtual particles is a mixture of and creation operators for real particles. This is a typical -> md(k relativistic effect; in the nonrelativistic limit (k a> k'\ ')

Note

destruction

and the number-of-quanta operator Af 2 that there are no virtual quanta surrounding the source.
ground
1

= 0,

in

= MH

This means
Relativisti-

cally,

on the other hand, we have a cloud of virtual particles even for the
state
J.

(N

in

0).

See

M.

That

this is the

Sparnaay, Phystca, 24:751 (1958). a change of e rather than o> can also be seen from dimensional

considerations.
8

In this case a and

are also related by a unitary matrix A,

A/4At

AAt = AtA =

The

for a, tf consistency of the commutation relations


1

when

and A A^
t

requires

0.

PAIR THEORY, QUANTUM-MECHANICAL

123
is

The determination of
easily done also guided

the distribution of virtual particles

most

by comparison with a one-dimensional analogue, which us for the same problem with linear coupling. The analogue for our present model is a simple harmonic oscillator, the frequency of which is changed from w 2 to a/ 2 = a> 2 -f (5a> 2 The former frequency corresponds to no source and operators a, a\ whereas the latter represents the energy in the presence of the source.
.

From
(12. 18)

(2.3)

and become

(2.4)

we

find that the quantities corresponding to

(12.19)

From

(12.17)

and

its

hermitian conjugate
a
I

we

find

in, 0}

= M*a* in, 0) M = M2M M-I


\

( 12

20 ^

In the one-dimensional case,

= 6a) 2 l(a) + a>') 2 and is a number, whereas for the pair theory To expand the eigenstates is a matrix. 2 n) of the Hamiltonian with 6co in terms of the eigenstates \n) of the |in, Hamiltonian without dco 2 we use (12.20), together with familiar tricks,
,

M M

to obtain

in,0)
(n!)

(12.21)

and by induction
2
|

in, 0)|
if

(MV
2

|(0

in, 0>|

if

is

even

is

odd
00

The quantity

|(0

in, 0)|

is

determined by the condition

ry

Then (12.21) is a matrix in k space. The same method works when 1 and (12.22) tell us that virtual particles are created only in pairs and that the probability for finding an odd number of virtual particles is therefore zero. The simplest term is
%iijt
1

\(k l9

k2

2
1

in, 0)|

i|(0

in, 0)|

[(^

2
\

Jc

2)|

(12.23)

It is

Chap.

1 1 is

for this reason that the quadratic coupling discussed in this chapter referred to as "pair theory."

and

124

SOLUBLE INTERACTIONS

Unfortunately, the explicit calculation of Af, that is, the analytic 1 As is inversion of 2 is very involved, and we shall not enter into it. to be expected, the wave function of the virtual pair decreases

exponentially with distance in ordinary space and around p within a distance m~ l


.

is

concentrated

To get more tractable expressions, we turn to the calculation of expectation values of observables in the ground state. From (12.17) we immediately find that (in, 0, which *also follows <(r,f) in, 0) generally from the invariance of the Hamiltonian under </>-></>.

More

instructive are bilinear expressions, such as the

mean number of

virtual particles:

ka\k)a(k)

in,

= (27r)3 (in, = (27r) TrM


3

>i(k) (k

Mif M*

k')

A f (k')

in, 0)

(12.24)

By means of

(12.18)

and

(11.22) this

can be rewritten as

Using our

explicit expression for

R, Eq. (11.23),
2 2
\

we

find

A' f

fkd*k
6

P (k)\

P (k')\
2

o/
(<o

4J

(27r)

\D + (k)\\k

2
'

k f2 ) 2

W
^

\ /

co

4J

(27T)

Q)o>'(a)

^v

oV)
is

Thus

the probability for finding one pair (or more)

essentially

p r UJ
a result which
is

3 ( 27T)

CD*

similar to that for the linear coupling,

if

2L
except that
pairs.

U27T)

now
A

The amplitude

the particles are not independent but are emitted in for finding the field excited is of the order of } or

r^|pw
z

A
2

rk*dk\p(k)\*
)Jo

4J(2rr)
1

of

(87T

a?

It

( 1 955).

has been carried out by A. Klein and B. H. McCormick, Phys. Rev., 98:1428 These authors also derive a closed expression for the presence of more than

one

pair.

PAIR THEORY, QUANTUM-MECHANICAL


2

125

From (12.16) this is seen to be approximately J dk <5co 2 /co and is just the sum of the amplitudes for the various normal modes to be excited. For the single-oscillator analogue we find
-| JT

W ~
CO
Z

if CO

-r

f*&

CO

However, the energy of the


sort of virial theorem.

virtual pairs

is

Formally, however, (11.1) that in pair theory


(in,
|

not simply related to <sf by a it appears from (12.12) and


L'
|

in, 0>

-(in,

in, 0}

stems only from the change in the zero-point energy. The probability for finding virtual pairs for a gaussian source dis* * tribution p(r) oc e~- r lb with a width b m~ l is of the order of (A/> 3/w 2) 2 Only a deep potential (large |A|) with sharp edges will produce pairs of virtual particles in significant numbers. For a Coulomb potential ej^irr,
<^

and

>

1 the probability for virtual pairs ~e 2 /47r per cent. The virtual 1 electron-positron pairs in the Coulomb field actually give a measurable "vacuum-polarization" effect, since they are charged and change the Coulomb field for r m~ l (^'lO* 11 cm) by about 1 per cent. Modern experimental techniques are capable of measuring such tiny effects with

m" 1 , is effective in creating pairs. Approxionly the steep part, r it there by a em~ l), we find that mating potential of our type (A

<

remarkable accuracy. 2
Further Reading
(1 942), and Progr. Theoret. Phys. (Kyoto), 5:584(1950). J. M. Blatt, Phys. Rev., 72:461 (1947). G. Morpurgo and B. F. Touschek, Nuovo cimento, 10:1681 (1955). W. Thirring, Helv. Phys. Acta, 28:344 (1955). A. Klein and B. H. McCormick, Phys. Rev., 78:1428 (1955). E. Arnous, J. phys. radium, 17:107 (1956).

G. Wentzel, Heli>. Phys. Acta, 15:111

1 In this respect these pairs behave according to the above description, even though our calculation does not apply to them. 2 See, e.g., W. E. Lamb and R. C. Retherford, Phys. Rev., 72:241 (1947).

CHAPTER

13

The Lee Model:

States with

Q=

The Lee model consists of a field linearly which has one internal degree of freedom. This degree of freedom is not a classical quantity like position and has only two eigenvalues. As an example of such a quantity, we cite the charge of the nucleon (source) which has two eigenvalues distinguishing proton (/?) and neutron (n). This degree of freedom, discussed in Chap. 7, is involved in the coupling which describes the elementary processes n <-> p + TT~, if we choose to call the quanta of the field pions. In general, if the source has more than one degree of freedom, the problem gets so complicated that it cannot be solved explicitly. We shall encounter this situation in the last part of the book. However, in the Lee model it is assumed that there is no TT+, so that the process + cannot occur. In this case -f- TT /?<-> charge conservation limits the to such an extent that the problem becomes soluble. possibilities Although artificial, the model reflects important features of the pionnucleon system. Formally, the problem has many similarities to the We shall pair theory which we have discussed in the last two chapters. lean heavily on the analogies in order to avoid repetitious formal
13.1. Introduction.

coupled to a source

manipulations.

To describe the degrees of freedom of the source, we have to introduce new dynamical variables. We define the operators r 3(r), r + (t), r_(t) = r* (t) in terms of their effects on the bare nucleons at t = 0: +
r_(0) r+(0)
|

p)

r3(0)

= = n) p) =

n)

r_(0)

p)
p)

r+(0)

r3 (0)

= = p) -)
n)

(13.1)
|

n)

These operators are independent of time in the absence of an interaction


126

THE LEE MODEL: STATES WITH

Q=

127

and

an interaction, the Hilbert space is the direct product of the usual one for the fields times a twodimensional space. If the nucleon is represented by (p,n) in the latter, then the operators 1 r are the 2 x 2 matrices
operators

for equal masses of the nucleons (// Hence, in the absence of r).

is

then independent of the

1\

T O/

.=

/O
\1

0\

r3
O/

/I

0\
(13.2)

\0

-I/

In the presence of an interaction the operators r become time-dependent, and (13.2) is a representation of r(0) in terms of bare particle states at
t

= 0.

The

between the bare particle


(r + )
-

interaction H'(r) which generates the processes states must be of the form

<->/?

+ ir*

(creation operator) -f (T_)

(destruction operator)

+ TT+ for a pion described by a Klein-Gordon then a difficulty arises because all local 2 operators such as <(r,f) field, contain both destruction and creation operators. Thus r + should be multiplied not only by a creation operator for a TT" but also by a destruction operator for a TT+ [see (7.16)], and similarly for r_. However, for a Schrodinger field, y(r) is a pure destruction operator and a suitable We base the model on the following //'(r) can easily be constructed. Hamiltonian: 3
If we wish to forbid /?<->

H=
J

2m
(13.3)

The <f o term is included to allow for a p-n mass difference; when it operates on a bare/? state and <f for a bare neutron.

it is

zero

be used to renormalize the energy of the physical n state. The com13.2. Commutation Relations and Equations of Motion. mutation rules for the operators r cannot be inferred from the canonical Since rules but can be deduced from the matrix representation (13.2).
It will

subscript we mean any of the three operators r+ , T_, r 3 D. Lee, Phys. Rev., 45:1329 (1954)] usually consists of a nonlocal interaction, because it uses a relativistic theory for the boson. Such an interaction goes beyond the scope of our investigations. However, in its significant
2

By T without a The Lee model

[T.

consequences, the model we shall use parallels the original one. 3 This differs from the Schrodinger Hamiltonian of Chap. 4 by the source terms, the <^ term, and also by the addition of the term /wvty to represent a rest energy of the field quanta. Note that g has the dimension of ZA

128

SOLUBLE INTERACTIONS

the operators r(Y) are created from r(0) by the unitary transformation (2.18), they have the same commutation relations, which are

Of course,
time,

y>(t)

and ^(t) are assumed


[>(r,0, r(0]

to

commute with

r(rt at

the

same
(13.5)

= [vV), HO] ^ and


f

and the commutation rules


[see (4.22)]:

for the operators

y>

are the usual ones

|>(rA V V,0]

OM, V(r',0]

= 3(r - ") = lV(r,0, yV.O] =


<5

3 6)
'

With these commutation rules and i& for the equations of motion

[0,#] [see

(2.19)],

we

derive

-/(M) =
(

(m
)

vV,0 +

WOOMO

(13.7a)

>(r,0

(m \

- Jl) y(r,0 + ^(r)r + (0


2m/

(13.7*)

In

momentum

space these equations become

=
2
*

T+
(277)-

(13.8ft)

with

= m + i.
2w
f J

i-

yt^j = y
/

* t

*.,
(rft)e

/j

3 **

l^77")

V(k,0

v(r,0e-**

(2?r)*

^
A: n -

and

P (k)

f ^

p(r) e

-*" dr

These

differential equations

manner analogous
Green's function

to Eqs. (8.4) to (8.10).


1

can be converted to integral ones in a By means of the retarded


eiK*

A rett rk (M

~i;J

(dK dKo

-1 -/e ~lv'

for

'

<

for*>0

THE LEE MODEL: STATES WITH

Q=

129

and

its

hermitian conjugate,

we obtain

/(k,0
tfk.0

= =

tln
y,

(k,0

ig

P
J-ao

iw{t <fc'r_(O e

~ tf)

(27T)
)

(13.90)

AM find

ig

J-oo

df r + (O <r ""-''

^
(2-77)

(13.9W

For the operators r we


-*T_(0

= * r_(0 +

gjd*r

p(r)y\T

t)Tz(t)

= *<MO + g f;f~| ,*(ky(k,OT3(0


ir + (0

(13.100)

and, by means of (13.70) and (13.76),


|T3(0

=
igffr

P (r)[^(r,0r_(0

^Or + (0]
03.11)

The

first

two equations can

also be written in integral form, but

we

shall

defer this for a while.


since
it

The last equation

expresses charge conservation,

says

(3-0
It is

Q=

ir3

d*r

v>

(r)v<r)

ir3

d*k

(k)y(k)

(13.12)

only the total charge (i.e., the charge of the source together with that of the field) which is conserved. The charge of the source sepanote that the operator rately, or of the field alone, is not conserved.

We

has half-integral eigenvalues bare particle states 2


:

<J

corresponding to the following

Q
Particle

-i
n
PTT~

-I
mr~
PTT~TT~

time the reader should be familiar with the adiabatic principle, which of justifies the physical interpretation

By

this

= y'(k,0) *-* and y in v int(k,/) = v int (k,0) <* The w which appears in the exponent here and elsewhere and the w which

(M

is

used in the denominator of the


interpreted as the
2

last

equation of page 128, for example, are to be

same symbol.

is the third is + | and has integral eigenvalues. the total isospin and will be used in the last part of the book. The - * electric charge of the nucleon is JO + r3 ), and that of the meson is because we are dealing with TT~ mesons.

The actual electric charge

component of

130

SOLUBLE INTERACTIONS
(13.7) to (13.10) are

still too complex to be However, when applied to physical particle states (eigenstates of Q and //)> they simplify further because in the present model these states are mixtures of states with n and n + 1

The equations of motion


integrated in closed form.

mesons.
13.3. Physical Nucleons.

The bare proton


P)

state

\p)

and

the

physical one are

identical,

P)

=
|

T+
T3(0

= in, = out, p] = y(r,f) p) = |P> P) = P)


|

jp>

(13.13)

and both are

eigenstates of

H,

H
|

p)

(13.14)

This stems from the fact that there is no other state belonging to the same eigenvalue of Q. If we apply the equations of motion to /?>, we can replace r 3 by 1 and obtain two coupled linear equations. In the following we shall first seek the solution of these equations in terms of the initial values of the operators. Then we shall see that this solution
|

also satisfies the correct commutation relations if we postulate that the incoming operators have the same commutation properties as the ones without interaction. To formulate this conveniently, we denote with a bar those operators which are multiplied from the right with a projection operator onto the proton state. This means that barred equations are

true only

and we

to time 2

1, applied to the proton state. In particular, f 3 (r) get for the Fourier transform of (13.9) and (13.10) with respect

when

=
=

f
J

-**" dt

y5

(In)*

-f ie

(13.15)

If

the

first,

we were to eliminate f_(# ) from the third equation and put it into we should obtain an equation almost identical with (10.7) of

the pair theory. Again, the values of for which (13.15) has solutions will reflect the eigenvalue spectrum of Hfor the states generated by the f f application of T_ and y to p). Since both T_ and y> create a state with J, we expect a discrete state corresponding to the physical

Q=
|

labels
2

Such a state corresponds to a bare vacuum state 0) in our old notation. The p) and n) serve to distinguish between the internal degrees of freedom. In the usual manner we define T_(#O) a f T_(fl e-*Ko* dt.
|

THE LEE MODEL: STATES WITH

Q =

131

m. If we neutron and a continuum of/? + IT states beginning at AT denote the energy of the physical neutron by AM, we can adjust f so that < m, in which case the neutron is stable. The spectrum for Like the bound state in pair theory, the this case appears in Fig. 13.1. neutron must correspond Energy physical = to a solution of (13.15) with Q

AM

AM < m, where
have, for

tln
y>

= O.

11

Hence

we

K =

AM,

(AM - <Qf_(AM)

AM
f_(AM)

or

AM = '

(27T)

- AM
f
(13.16)

-1/2

1/2

*
Before the source
is

Fig. 13.1. Eigenvalue spectrum of the Hamiltonian (13.3) for the states of

G =

i
is

turned on, the energy of the neutron

Since

(13.16) shows that the interaction lowers the energy of the neutron, as in the case of a static source.
it

w > m, and

was assumed that

AM < m,

13.4. Scattering States.

The

solutions of (13.15) with

K >m
(}

can

be found as in the pair theory:

w
Because of (13.16), we can eliminate the unobservable
left-hand side, the bracket of which can be written as 1

from the

(/C

03.17)

11 This is interesting, because one tends to think that a physical particle is something different from a bound state of bare particles. However, our example shows that so far there is no fundamental difference between a "compound particle" and a

physical one. 1 The factor

assumed.

is less than m, as has been so long as U plays no role in w This follows because y tln (k,r) has the free-field time dependence and

AM

AM

oc

d(K

w).

132

SOLUBLE INTERACTIONS
is

variable

oo + /O) = 1 and dD(x)/dx > 1 In fact, D( from zero only when KQ = m -f fc 2 /2w > w, (k,^ we find that (13.15) has a nonzero solution for all KQ > m and for # = AAf. The contribution from the latter point will be denoted by r ^ to satisfy the commutation relations, we shall have to multiply it by a normalization factor Z*, the meaning and value of which will be determined shortly. We can summarize our findings by

K =z=x+
Q
) is

the boundary value of an analytic function of the complex As in Chap. 1 1, we find that D(z) does not /y.

have any zeros for x


Since
in v>

< m.

different

_(X
where the
)

/e in the denominator corresponds to integrating (13.100) with a retarded Green's function. 1 The solution for fi can be found as in Chap. 1 1, by making use of the Green's function and introducing the four-dimensional Fourier transform ^(k,/^). We shall not repeat 2 this procedure here but shall merely give the results:

(k
(2TT)

0I I

tin

k')v>

(k',0

(I3.19a)

(w

t)(w

AM)

^
(13.19c)

= (k|Qt|k')
We
n

note that rl. plays the same role as the bound-state operator A b in the pair theory, but b is not normalized to 1, as we shall see presently. Similarly, the Fourier transform of (13.18) gives

MO - zVHd) + g
remembering that
explicitly that the
tin

J (2-rry

AM ^(k.o
).

(13.20)

(k,^

oc

6(w

-K

13.5. Completeness.

We

are

now

in

subspace
1

Q=

commutation relations in % if we assume that r and

a position to demonstrate (13.4) to (13.6) hold in the


in
y>

satisfy the

commutation
is

It

i>_
2

should be noted that the homogeneous equation for our (exact) solution AMr_ rather than i>_ = - r_. Matrix multiplication, defined as in Chap. 11, is implied here.

THE LEE MODEL: STATES WITH


rules of the operators without interaction

Q=
Z)~
1

J-

133

and

if

Z=

(AM), as

shown below.

In a manner similar to Eqs. (12.3) to (12.5),


f

we calculate 1

J (27r) , (w

-^ ^>ff> 1 AAf)D+(w)J
5

rf*fc
L* J (2W) 1 (w
2

- AM)D_(w)
1

^1,
'

=z
where

(w

- AM) 2

2
|

=z

ds
f

D_(w)|

2*ri

Jo

- AM AM)D(z) (z

C is a contour similar to part of Fig. 11.2 and is shown in Fig. 13.2. We note that D~\z) is anaaxis lytic save for a cut on the real -> oo it from m to oo. But for
|z|

tends to

1,

so that an integral over

an

infinite circle gives

_L f^l~~
2-rri

Fig. 13.2. Integration contour fined by Eq. (13.21).

de-

Hence, on closing the contour by a contour integral by f we get


,

circle at oo

and denoting the closed

(13.21)

-Z=
<P
|

2m
|

<b J (z

- AM)D(z)

==

Z
D(AM)

rs

p>

(13.22)

which implies that we have to put


(13.23)

D(AM)
By
similar arguments,
t/ 6 (k)C7 6 (k') *

we prove
| |

that
q)(q

+ (k Q-

Q1
| 1

k')

= (27rMk - k')
k
r

(13.24)

a formula familiar to us from pair theory.


(P
1
I

Consequently,
^ (k
3
)

O00 /OO]
l

P)

(13.25)

note that T (r) does not commute with vfr.O- Physically, because ^B creates a physical neutron, and the latter is an extended structure which interferes with the meson cloud even outside the source. This is in contradistinction to bare particles, the pointlike nature of which is expressed by the canonical commutation rules.
It is interesting to

this arises

134

SOLUBLE INTERACTIONS
Also,

is satisfied.

P*(k)r J>;'(w)

D^(AM)

1
Iiri

f
J (z

*(27r)*Lw-AM
can be checked. 1

w-AM

- AM)(z - w)J

dzD~\z)

1
'

Having established a solution of the equations of motion and the commutation relations in this sector of Hilbert space, we shall next turn
to a discussion of the physical significance of the quantities encountered The physical neutron state is created by applying T * to the there.
!

proton

state
|

in, n>

- r*
|

p)

(in,

n
|

<p

r?
in,

u M Since r ^ has a time dependence ~e * 9 it follows that 2 with eigenvalue AM, as we saw earlier. eigenstate of
1

The

n) is an factor

gives the probability of finding a bare neutron in a physical one,

in,

>-<,! [zV? L
the

+ gJ("

( (2iry (w

U **%) (w)J AM)D


+

p)

= Z'

(13.27)

and

U (k) is
b
|

wave function of the

virtual pion

when

the neutron

is

dissociated into/?

+
|

TI-:

(P

+ ^k

in > *)

= (p

+ (k
With some calculations we find

'

(13.28)

that this
it

wave function has the following

C/& (k) is normalized, as properties. the neutron dissociated : 3 finding

should be, to the probability of

(13.29)

The form of

b9 Eq. (13.196), shows that the virtual pions stay near the source within a distance (m AM)" 1 Furthermore, for these virtual the virial theorem is valid in the same form as for the static pions
.

trivial
2

The proof of the commutation relations in and will not be carried out.

the other parts of Hilbert space


|

is

not

This can also be checked directly with (13.3) and the analysis of of bare particle states, which we shall now discuss.
3

in, n} in

terms

Note

that

<Z<

1 is

required.

THE LEE MODEL: STATES WITH


source.

Q=

135

For instance, for AAf 0, // , the part of the me8on f (13.3) proportional to g, and /f J d*r \fy Vy/2/w,

int

Hamiltonian we have

(in,

n
|

in, n)

(in,

-f-

Hmeaon H int (in, n


n
\ \

in, n)

in, n)

+
2

(in,

n
\

5
| I

in, n)ef

(13.30)
X. \lft

and

(in,

n
|

meson
|

in, n)

=$ k
=

t/(k)|

H-

= Zg $
2

(in,

f
|

-^
|

in, n>

and hence
(in,

n
|

H int
|

in, n>

-2<in, n

#meson
| |

in, n>
1

The

commutation relations for the operators y>, y/ and r are satisfied shows that the states we found with Q = J form a complete set. Hence there is, for instance, only one discrete state for the system of a p and a TT~, namely,
fact that the
",

(13.30.)

This has its formal origin in the fact that (K &Af)D (K ) has only = AAf, as we saw earlier. By increasing AAf, the one zero for = m. range of /b (r) becomes larger and reaches infinity for AAf When this value is reached, the neutron becomes unstable 1 against = J decay into a proton and a pion, and the sole discrete state for Q The commutation relations are then satisfied with f ^ = 0, disappears. and we shall see that the physical neutron is then only a resonance in the pion-proton scattering; the wave function of the pion for the scattering

pions at an

k space, which means that there are from the nucleon. To end this chapter, we investigate the 13.6. The Phase Shift. in the states p + TT~. For this purpose we must relate y> tin scattering tout The preceding development could have been carried out to y; tout and f', which is identical with f equally well with y>
states
|

in,/?

TT-) is

singular in

infinite distance

*!! :

= ^<f (0) = fL (0 in, n) = out, n) s n)


ut

ut

(r)

(13.31)

Hence we have
(k
1

Q|
|

tin

k')v>

(k')

= (k Q +
|

tout
|

k')v>

(k')

(13.32]

For AAf

>

m the model provides a field theoretic description of a decay if the

various particles are suitably renamed.

136

SOLUBLE INTERACTIONS
in pair theory, for a spherical source only the

As
/

angular-momentum
e~
9

parts of

tout
v>

and y tin

differ
tout

by a phase

shift

^w

-Tin

and by analogy with Eqs.


tan d(k)

(11.31) to (11.38)

we

find

=-

-g*km

|p(fc)r

'

(27T)V
where
+ (HO

- AM)(w' -

*\*W

w)J

(13.33)

D,(w)

iD(w)

fl L

g*P
'

(w

w)(w

AM)

rfm fclpWM
L 27r (w

-A M)J

< For one bound

AM

and for k -+
1

state.

If this

positive energies,

AAf

>

we have d(k) < 0, as in pair theory with resonance at negative energies is raised to m, then lim 6(k)/k becomes positive, resem0,
*-*o
.

In this case the neutron becomes bling a weak attractive potential. unstable with a lifetime 2 F" 1 This lifetime is defined by (12.9a), and

we

find

r3

(27r)
2
fc

(w'

- AM)2
(13.346)

with

-^ + m = AAf = w r
2m

As might be anticipated from the similarity to pair theory, we can define a renormalized coupling constant, so that for w the phase

= AM
New

1 This is characteristic of bound states. See, e.g., J. "Theoretical Nuclear Physics," chap. 2, John Wiley 2 Under certain conditions

M.

Blatt

and V.

& Sons,

Inc.,

F. Weisskopf, York, 1952.

shows a behavior ^e~ rt ^


If \p(k)\
2

[V. Glaser, G. Kallen, Nuclear Phys., 2:706 (1956-1957)]. can be analytically continued into the complex plane, so that D can also be continued beyond the cut, then we find a pole on the second Riemannian sheet at a + /T, corresponding to the unstable neutron state. complex energy

AM

THE LEE MODEL: STATES WITH


shift is

Q =

137

given by the Born approximation with the renormalized coupling constant. We then find

(1335)
3
7T)

gr

= Z'-g

(W

~ AM)

and

tan d(k)

=
3

(27r)

J (w'

- AM)2(w' (

1 w)J

(13.36)

This form has the same advantages as the similar equation in pair theory. In our particular version of the theory there are no divergence difficulties even for a point source, because at high energies w is proportional
to
2 fc Of course, the nonrelativistic energy relation no longer makes In a relativistic treatment of the mesons sense at very high energies.
.

(Klein-Gordon equation), (13.35)

is

replaced by

3
(27T-)

(w

- AM) 22co

The

denominator now diverges at high momenta for a source, so that the scattering cross section would become zero for point small k. Use of the renormalized coupling constant introduces an
integral in the

extra power of w' [or G/ if (13.37) is used] in the integral in (13.36), so that even for a point source the cross section remains finite in the limit of small momenta. In this case, however, difficulties arise which may

appear even in our nonrelativistic version.

In particular, from (13.35)

we

find
2

__
? gr P f
2

gr
\

P (k)\*

}(

(w

- AM) 2

Hence, for a
relativistic

sufficiently small (spatial) source

large) value of

r,

and a finite (perhaps becomes negative and g imaginary. (In the


/O.)

-* theory and in the limit of a point source, g

This

Furthermore, Z, when in terms of gn is then negative, which means a negative expressed probability of finding a bare neutron in a physical one. To formulate
implies a nonhermitian Hamiltonian (13.3).
1

See, e.g., Lee, loc.

cit.

The

factor (2w) arises naturally


earlier.

from the use of the

Klein-Gordon equation, as we saw

138

SOLUBLE INTERACTIONS

such a theory 1 requires the introduction of an indefinite metric in Hilbert space, which goes beyond our framework. In a cutoff theory no trouble develops so long as the size of the source is limited to a value ^ Z ^ 1. which reduces Z(gr) so that Formally, we find from (13.33) that the high-energy limit of the scattering cross section is given by the Born approximation with the coupling constant g. The physical reason for this and for the lowenergy behavior will be discussed extensively in the next chapter.
1

See G. Kallen and

W.

Pauli, Kgl.

Danske Videnskab. Selskab, Mat.-fys. Medd.,

30(7) (1955).

CHAPTER

14

Lee Model

States with

Q = -\

14.1. Scattering: Low Equation. For the states not considered in the previous chapter the problem can also be reduced to integral equations for the wave functions of the virtual pion. In general,

however, such equations cannot be solved explicitly, and hence we are not going to derive them. From an experimental point of view, the wave function of the virtual pions is actually not the quantity of primary interest, since most of the information it contains is not accessible to present observational techniques. What can be measured most easily is the scattering cross section, and therefore one should concentrate on calculating the phase shift as a function of the energy. We shall now learn a very important method for deriving general properties of the phase shift, short-cutting the calculation of the complete wave function of the mesons.

= For simplicity, we restrict ourselves in this chapter to the case or neutron) by N)> and denote a physical nucleon state (proton The two physical nucleons then have the same energy (mass), E = 0; the quantity we are interested in is an element of the S matrix (8.23):

AM
|

(out,

N+

7r

,
I

in,

N+

7r > k

out

(AT

y>

(k>

tln

(k)

#>

(14.1)

Equations (13.9)

and the

similar equation for

tell

tout
y>

(k,0

71 P*00 (2*T
139

"*V-(w)
tln
v>

(14.2)

us that the state with an incoming plane wave, outgoing plane wave with the same momentum,

OO N)
|

is

an

tout

(k)

W), plus

140

SOLUBLE INTERACTIONS
is

something which
scattered
If
|

generated by
for.

T_(H>).

Obviously

this

must be the
analogy
\

wave we are looking


is

N)

the proton, then


is

we can
d(K

use (13.18)

or, rather, its

for outgoing operators

to derive

an
)

explicit expression for r_(w)

N)

Since

r^(KQ )

proportional to
tout
V>

and

since

(k,K

we

2V

out

(k)<5(K

w)

obtain
.

(277)'

77

D + (W)W

477

f
Jo

dk' d(k

tout
fc')v>

(/c')

= J3V out

(/c)

(14.3)

which shows that we have a spherically symmetric scattered wave for a Comparison of (14.2) and (14.3) with (8.24) tells us spherical source. that^ is {exp [2id(k)] 1}, which also agrees with (13.33). In general, we can deduce the S-matrix element from (14.2) and the familiar time dependence of matrix elements between eigenstates of the Hamiltonian. Thus 1
(out,

N + "V in, N + k = (N ? ut(k')|V out(k) |

77 )

- k') -

(out,

N+
w')
;

3
(5

(k

- k') -

2md(w

g
(2rr)

- k') - 277/<5(w In obtaining


(14.4),
(out,

w )r(k') we made use of the


e
|

fact that

N + TV

iHt

(out,

N + TT^

e
\

iw>i

In since the energy of both physical nucleons was adjusted to be zero. the matrix T would be a function of both k' and k, but because general,

of the spherical symmetry of the problem and energy conservation, it is here a function of a single variable k'. The relation of the T matrix to the phase shift can be found from (14.3) or directly from (8.30) and is 2

irrnk
1

477

It

must be remembered that

p(r) is

a spherically symmetric source, so that

2 See also, e.g., B. A. Lippman and J. Schwinger, Phys. Rev., 79:469 (1950). our present normalization, C == J d*k and^CE) \\nmk.

With

LEE MODEL: STATES

WITH

Q =

141

The T matrix defined by (14.4) is, of course, time-independent. It is more convenient than the S matrix for the subsequent discussion
because the singular part of the latter has been split off. Thus, the matrix will turn out to be an analytic function of the energy.

follows that the scattering amplitude T(k) is obtained N) into outgoing states with energy w. Using the fact that y(k) commutes with r_ and that y ollt N) 0, we can rewrite the last expression in the following form: 1
(14.4)
it

From

by analyzing

r__

T(k)

(14.5)

Equation
will

(14.5)

is

usually called the

14.2. TT~ -f n Scattering.

Low equation, 2 after its progenitor. Our first application of the Low equation

be to derive some general properties of the TT~ -f n scattering amplitude. The commutator in (14.5) contains a term with r + (t) and one with r_(0) in front of /?). The former generates a state with3 Q = i in fact, we see from
|

that 4

T+ (r)
this part

in, n)

Z*
|

p}
5

(14.6)

Hence

of the right-hand side becomes

gr
1

(p

gr+

n>

<ii

r_(0)

p)

Z#
J

(14.76)

emphasizes that the source must involve dynamical For a c-number source the right-hand side of (14.5) is The ic is the usual convergence factor zero, in agreement with our previous result. of Green's functions which defines the position of the pole in momentum space. 2 F. Low, Phys. Rev., 93:1392 (1955). 3 From [Q,r (t)] = r(t) we conclude that the operators r+ and T_ change the eigenvalue of Q by 1

The commutator

in (14.5)

variables in order to scatter.

4 5

Since

in,
is

The

n > = out, n>, we shall use ri) to indicate either state. not needed in the denominator of (14.70) because the pole at
|

w =

lies

in the unphysical region.

142

SOLUBLE INTERACTIONS

The other term of (14.5) generates states with Q = f, e.g., states with one outgoing pion and a neutron or two outgoing pions and a proton. At this point we have to assume that the form of the energy spectrum is not changed by the interaction and that there is no discrete state with Q = f. In that case these states have energies >m. If we denote the physical states with Q = out, />, we can write f by this part of the commutator as
|

<n

r + (0r.(0)

n)

= % (n
i

e'*V + (0>r'
<

w
| 1

out, i)<out,

i
\

r_(0)

n>

= 2 e~
i

iEit
I

out
I

T-

n>

2
I

( 14 - 8 >

where

is

the energy of the state

out,

/>

This gives us,

all together,

(27r)

(27r)

E
fc-*0

ic

first

remark concerns the sign of lim


it is

d(k)/k.

It

appears from

(14.9)

and (14.40) that


state.

>0, as for an

attractive interaction without


is

a bound
attractive

can be traced back to be the following. A single neutron emits a pion, and the proton so formed absorbs it. We found that If another ir comes along, these processes lower the energy by <? then the emission and absorption activities of the nucleons become more violent. We have seen that a source emits more eagerly if an identical boson is around, because of Einstein's well-known induced 1 This process decreases the energy, the gain of interaction emission.
.

The reason why

the TT~ interaction with a neutron

energy outweighing the pion rest energy. Thus the interaction will decrease the energy of a n + ir below the energy of an n + TT~ without That such a situation gives interaction, imitating an attractive force. an attraction, whereas for a p + TT~ the phase shift starts negative, as we saw in the previous chapter, is the crux of our present understanding of low-energy pion physics. The statements made so far are exact. However, (14.9) cannot be solved in closed form. We shall find its solution in the approximation over Q = f states, those corresponding to out,/? -f 2?r~) are that, in

2
i

neglected.
1

If these virtual production processes

of a second pion are


The

Similarly,

we can

interpret the

Yukawa potential as arising from the emission of


field

mesons by one nucleon induced by the meson


emission probability arises formally because
|<n -f
1
1

of the other.

stimulated-

2
|

/i>|

oc

(n

1)

LEE MODEL: STATES WITH


left out,
1

Q=
*

143

then the term in (14.9) that


tr~~
|

is

proportional to
T.

2 involves only
'

<out,

r_
I

n),

which can be expressed by

We find 2 8
(14.10)

is a nonlinear integral equation, it can be solved end we recognize that T(k)/\p(k)\ 2 is the boundary value of a function which is analytic in the complex w plane. Indeed, (14.10) states that the function $ of a complex variable z which obeys

Although (14.10)
exactly.

To

this

"M**)!'* "
1

27T
is

(RIO

|(W'

related to

T by

= - lim
-0
(27r)

--

(w

(14.12)

ie)

From (14.1 1) we see that l/(z) is analytic in the complex plane except At this for a pole at the origin and a cut on the real axis from to oo. is discontinuous: cut the imaginary part of

stjcwUrttw-iJ
=
f

dw'
27T
2

2
|p(fc')|

fc'm I
2
ie)|

Jm

|(w' +

\w'

je

w'

<V

1 This approximation will also be used in pion physics. In general, such an approximation will not be good, especially at high energies. Indeed, comparing the exact high-energy limit (14.17) with the one of this solution, Eq. (14.14), we see that 1 However, the solution may have a they may be completely different if Z degree of validity at low energies, where it approaches the correct limit.

<

By

similar techniques,

we

obtain for/?

+ IT

scattering the exact equation

the precise solution of which

was obtained

in the last chapter:

r
L
3

6r

rg
87T

w' 2 (w'

ww -w~
k')

1
/e)J

Because of our normalization of the one-meson states


<in,

in, k'>

= (N v tln(k) vln(k') N) = d\k |

the

sum

J of (14.9) becomes J d k in the continuum limit.

144
Since
to p:

SOLUBLE INTERACTIONS
(w
/e)

=
i)

*(u> -f /e),

the discontinuity of

is

directly related

-i lim [(w
e-*0

- $(w -

ie)]

\o(k)\ = - Ifi^l mk = 2 Im $(>v +


2
77"

is

{)

A function with this

kind of analytic behavior

provided that g r

is

not so large that the bracket has a zero for


1

w<

m.

With

To
it

this proviso, (14.11) is actually satisfied by (14.13). relate this solution to more familiar material, we shall also

obtain

by means of methods used in pair theory. The integral in (14.11) can be written as a contour integral around the cut (see Fig. 13.2) and, -> oo for furthermore, can be closed by an infinite circle, since (z) z -* oo. Hence, (14.1 1) can be written in the linear form
1

_g*r
z

1
.

27r/Jc(z')(z'-z)

where

is the path of integration shown in Fig. 13.2. Evaluating the integral with Cauchy's theorem, we see immediately that (14.11) is satisfied. Thus we arrive at the following solution for T(k) [compare

this

with

(14. 12)

and

(14.13)]:

Remembering

that

T=

-sin

id

(6e l47r

km) we
9
1

get for the phase shift

tan

W=
:

aW*
3

"
...... .
(W'

(14,5)

(27T)

- W)W"

of the same form as the phase shift in pair theory with attraction The forms of (14.14) and (14.15) point out one advantage of the Low equation because it involves only physical states, it allows the solution to be expressed directly in terms of the physically meaningful, or "measurable," coupling constant gr rather than g. We assumed that there is no resonance at negative energies, which would
This
is

[see (12.116)].

Dalitz,

For a discussion of the uniqueness of the solution, see and F. J. Dyson, Phys. Rev., 101:453 (1956); and M.

L. Castillejo, R. H. Ida, Progr. Theoret.

Phys. (Kyoto), 21:625 (1959).

LEE MODEL: STATES WITH


indicate discrete states with

Q =

145

which depends on

p(k).

resonance for w > m enhances the phase shift, whereas

This imposes a limit on g*, sufficiently large gr , within this limit, a will emerge. In any case, the denominator
f.lf

Q =

For

it

was damped for the p

ir~

Since for low energies the inelastic terms in (14.5) leading scattering. TTT to a p 277~ state act in the same direction as those for the n

system,
feature.

it is

to be expected that the exact solution will also show this Therefore we anticipate that at low or medium energies the

TT~

cross section will

become much

This can also scattering. The latter solution for p


technique
14.3.

larger than the one for/? + TT" be seen by comparing (14.14) with (14.5a). + ?r~ scattering can also be obtained by the

derive

we used to derive (14.14). Low- and High-energy Behavior of T(k). Finally, we shall from (14.9) and (14.50) the low- and high-energy behavior of T(k).
end we consider

To

this

as a function of w.

From

the factors in the denominators


t

we

see that

the term
first

2 in (14.9) is regular for w < m, since E > m. However, the term has a pole at w = with a residue g*. Therefore, this term
i

will

be dominant for

->

0,

and we have
lim t(k)

=^
W

(14.16)

w-0

TT~ system as for they? -f TT~ that for w -+ the exact result is the Born namely, 1 approximation calculated with the renormalized coupling constant. In the limit w -> oo, if the sum converges sufficiently rapidly, we may

Hence we

find the

same

result for the n

scattering [see (14.50)],

put the denominator under


factor \/w out of the

2
i

n (14.9) equal to

w.

Taking the
just the

whole expression, we see that what


This, however,
is
2

is left is

equal-time commutator.
lim t(k) w-oo

r 3 , so that
1)

we

get
(14.17)

= *- (n
W

-T,

>

= 8- (2Z W
if

The

last expression

becomes more suggestive

we

write

it

as

II In that case, (14.14) would be the solution of a Low equation in which the contribution of this discrete state is included. 2 1 It is always assumed that \p(k)\ can be continued analytically to the unphysical = or fc 2 = 2m2 energy w
.

146

SOLUBLE INTERACTIONS
is

the [Born approximation of t(k) for n + TT~] x (probability of a bare n in n)) + [Born approximation of t(k) for/? + if] x finding In this form the result is (probability of finding a bare /? in n)). analogous to the one for the p + TT~ scattering. In fact, with the Low equation we could have derived the latter without having to solve the

which

Schrodinger equation.

The physical significance of the low- and high-energy lijnits can be seen by resolving the scattering process into elementary interactions

Fig. 14.1.

Graph

for the physical neutron.

between bare particles. This corresponds to an expansion of T in powers of g. The Born approximation consists in taking the least number of processes. For the scattering, it corresponds to an abFor sorption of the original -n~ and an emission of the scattered pion. the n -f TT~ scattering, the processes must occur in reverse order from 1 that for/? + TT~ processes, as is illustrated by the graphs of Fig. 14.2. Higher orders in g give a rescattering of the pion and a dressing of the

\ \

p+ir~
Fig. 14.2.

Diagrams for the

scattering of ?r~

on protons and neutrons.


in Fig.
14.3.

neutron.

For the order g4

this

is

illustrated

Now,

2 imagine that the energy of the external pion is increased beyond that of In this case the time between emission and absorption, all virtual pions.

has to be as short as possible since the uncertainty in the nucleon's is as large as the energy of the external pion (which it w 1. absorbs and emits), and A Hence, for p + nthe contribution from the graph of Fig. 14.4a will dominate scattering,
AT",

energy,

A,

A7~ AT~

In these diagrams the lines show the paths of the various bare particles, and time In such a picture the physical neutron looks as shown in left to right. These diagrams should not be taken too literally, since the concept of a Fig. 14.1. However, they serve to illustrate classical^path does not apply to virtual particles.

flows from

the various terms in a perturbation expansion where //' effects the elementary emission (and absorption) process between bare particles. See G. C. Wick, Revs. Modern Phys., 27:339 (1955). 2 This is the one which is scattered.

LEE MODEL: STATES WITH

Q=

147

p
(a.)

n
(b)

V.*""

...---.
n

n
(C)

x
p

*>>

Fig. 14.3.

Graphs of order g* for (a) p

+ IT scattering

and (,

c)

w~

scattering.

n
(a)

;i

/"
p
n

Fig. 14.4.

Diagrams to

illustrate the

high-energy behavior of

/>

scattering.

148

SOLUBLE INTERACTIONS

over all those diagrams in which (intermediate) virtual pions are present, such as those of Fig. 14.4ft and c. The high-energy limit for p -f TT~ For the n f TT~ scattering is therefore given by the Born approximation. the scattering can also occur during the time that it is a bare system proton. Thus, in the high-energy limit we get here both scattering by a bare neutron, as in Fig. 14.5#, and scattering by a bare proton, as in Fig. 14.5ft, but in both cases without (intermediate) virtual mesons being emitted and absorbed between the interaction with the external

(a)

(6)

p
(C)

Fig. 14.5.

Graphs

to illustrate the high-energy behavior of n

+ IT

scattering.

meson. The latter contributions, illustrated in Fig. 14.5c, will again be small compared with the others, and we get the Born-approximation amplitudes from the two kinds of particles weighted by the corresponding probabilities. To discuss the low-energy limits, let us consider first the fictitious situation in which the external meson has no rest energy, so that its energy can be made much smaller than that of the virtual pions. In
this circumstance it follows from the uncertainty argument that the thie between the interactions with the external particle will last much longer than the virtual processes. For p -f IT scattering, this means that, after absorbing the pion, the nucleon practically becomes a

LEE MODEL: STATES WITH

Q =

149

But the absorption of the physical neutron, as shown in Fig. 14.6. external pion converts the proton into a bare neutron and, therefore,
is

can happen only during the fraction of time that the physical neutron a bare neutron. Since two such elementary processes are involved,

we

obtain the following expression for the cross section in this limit:

(Born approximation) x (probability of having a bare neutron in the

<:.
Fig. 14.6.

/ pnpnpnpnpnp
-., ,...,
......
....,
to illustrate the "low-energy limit" of/?
-f TT~

Diagram
.

scattering.

2 The same holds for n TT~ scattering, where only physical neutron) corrections of the form illustrated in Fig. 14.70, but not higher-order

those of the form shown in Fig. 14.76, contribute in this limit. For the above conclusions still hold in the limit w -> 0, but this is in 7^ an unphysical region, and the statements about the low-energy limits

^x.r-

x:
(a)

(6)
Fig. 14.7.

Graphs

for the discussion of the "low-energy limit" of n 4

-rr

scattering.

are true only in an extrapolated sense. Generally speaking, we can say from the way that g is introduced that it measures the strength of the elementary interaction between the bare nucleon and pion. On the

other hand, gr gives the strength of the elementary interaction between the physical nucleon and pion, 1 as appears from g r (p r+g n). Fast processes involve the bare source particles and hence g; slow ones

The

physical

and bare pions are

identical.

150

SOLUBLE INTERACTIONS
.

involve the physical source particles and gr Since at high energies the situation becomes very complex, 1 it is generally only possible to measure gr Even this quantity cannot be obtained directly, since the w is not available experimentally. However, if the lowenergy energy-scattering phase shifts can be extrapolated to this energy, or if the bracketed term in the denominators of (14.14) and (14.50) is approxi.

then

mated by an expansion in terms of gr for small, but physical, we can obtain gf directly from measured cross sections.

We shall

energies,

model have been studied from time to time. One of these, proposed by van Hove, still limits the number of degrees of freedom but allows more than one meson in the cloud of the physical nucleons. The generalization consists in allowing only one type of field quantum, 77, but two processes characterized

discuss this further in the last part of the book. Various modifications and generalizations of the Lee

by
-f
TT

V^n
n

(coupling strength
(coupling strength

ga)
gb)
is

^ V -f

TT

The theory has

several advantages over the

Lee model and

also

exactly soluble for the bound physical states. The main advantage for us is that it shows the connection between the neutral scalar theory of

Chaps. 9 and 10 and the Lee model. Thus, if the physical masses of V and n are the same and ga = g'6 then the theory reduces to the neutral scalar theory, except for a doubling of states. To discuss
the
,

the theory in detail would lead us too far astray, to the original articles. 2
1

and we

refer the reader

It is clear that-the nonrelativistic

approximation cannot be made at such energies

furthermore, nucleon-pair creation and recoil must be taken into account. 2 T. W. Ruijgrok and L. van Hove, Physica, 22:880 (1956); L. van Hove, Physica, 25:365 (1959); T. W. Ruijgrok, Physica, 24:185, 205 (1958) and 25:357 (1959).

Part Three

PION PHYSICS

CHAPTER

15

Introduction

The Static Model. There are many problems in physics which within the scope of quantum field theory and can be treated by methods similar to those we have developed or by less rigorous ones. The prototypes of discrete coupled oscillators are the atoms in a Electrons that pass through the crystal act as a disturbance crystal. that is linearly coupled to the atoms. However, the reaction of the crystal back on the electrons cannot be neglected, so that the interaction
15.1.
fall
is

a dynamic one, involving the translational degrees of freedom of the and cannot be represented as a coupling to a fixed and prescribed source. The ensuing complications make this "polaron" 1 problem an interesting one, but one that cannot be solved exactly. An example of a mechanical model with a quasi continuum is the 2 The complications quantization of the equations of motion of a liquid. of this problem arise because of the nonlinearity of the hydrodynamical
electron,

equations, so that only approximate solutions exist, e.g., at low temperatures, where quantum effects dominate the picture and the continuum 3 In nuclear physics the approach appears to give reasonable results.
surface waves of nuclei represent a two-dimensional example in which 4 Also, the many-body problem (such quantum effects are important. as represented by a Bose gas or a Fermi-Dirac gas of electrons or

nucleons), which at
1

first

sight appears to

have no

field features, is

most

T. D. Lee, F. E. Low, and D. Pines, Phys. Rev., 90:297 (1953); and T. D. Lee and D. Pines, Phys. Rev., 92:883 (1953). 2 R. Kronig and A. Thellung, Physica, 18:749 (1952). 3 L. D. Landau and E. M. Lifschitz, "Statistical Physics," chap. 6, Addison-

Wesley Publishing Company, Reading, Mass., 1958. 4 A. Bohr and B. R. Mottelson, Kgl. Danske Videnskab. Selskab, Mat.-fys. Medd.,
27:16(1953).
153

154

PION PHYSICS

1 Furthermore, in the easily tractable by methods of field theory. high-density limit an electron gas with Coulomb interactions turns out

to be equivalent to a pair theory and can therefore be solved exactly the methods we have described. 2

by

The most fascinating applications of our rules are, however, not to any material substance but to immaterial fields, the excitations of which appear to us as elementary particles. There are many examples among them, quantum electrodynamics and the theory of "weak These theories have the advantage that the coupling is interactions." sufficiently small that the physical particles and bare particles are

On the other hand, their systematic development requires the theory of the representations of the Lorentz group and is therefore outside the scope of this book. Instead, we shall be conalmost identical.
cerned with an approximate theory of the pion-nucleon interaction. This theory is one of the most exciting ones, inasmuch as it penetrates to the smallest distances yet explored by any partially successful physical theory. In the other applications we have named, the phenomena take place in regions of atomic dimensions, or of the order of the Compton wavelength of the electron. By now, we can be fairly

we have employed are valid in the latter Pion physics, however, occurs at distances of the order of regions. 10~ 13 cm, and it is interesting to note that the concepts of field theory seem to work at this small a distance. Even though much work has been done, very little is yet known about fields inside this region, nor is it known whether it is necessary to introduce the concept of a smallest These problems are among the most challenging that face distance.
confident that the principles
physicists today.

The theory of
(2) particles

facts: (1) the exact

the pion-nucleon interaction is complicated by form of the interaction is not known for certain,

two and

other than pions

nucleon.

It is fairly certain,

nucleon also contains

make up the structure of the physical for example, that the meson cloud in the mesons and hyperons, as well as antinucleons

and antihyperons. An approximate separation of part of the pion cloud from that of the other particles fortunately is possible, because the pions extend to the largest spatial dimension. As we saw earlier, the extension of the cloud is roughly measured by its Compton wavelength; in other words, it is inversely proportional to the particle mass The K meson is approximately 3| times that the field represents. heavier than the pion, and because our present concepts indicate that
"strangeness"
1

is

conserved,

its

emission requires changing the nucleon


J.

See, e.g., K. Huang and C. N. Yang, Phys. Rev., 105:767 (1957); and L. N. Cooper, and J. R. Schrieffer, Phys. Rev., 108:1175 (1957). 2 G. Wentzel, Phys. Rev., 108:1593 (1957).

Bardeen,

INTRODUCTION
into

155

a hyperon, which costs approximately another meson mass. Hence the cloud of K mesons and hyperons is not expected to extend to more than about two-ninths the distance of the pions, which implies that the volume of the nucleon is 99 per cent pion cloud. Nucleon-anti-

nucleon pairs have an even shorter range. However, their importance be that a pion can split into three pions 1 through an intermediate \pair, thus contributing to a finite size of the pion itself. 2 These considerations show that the prospects for a theory containing 3 only pions and a nucleon are fairly good. Experimental evidence indicates that the nucleon has a "repulsive core," which is not understood and which extends to about 5 x 10~ 14 cm, but that outside this

may

region pions are the principal contributors to the structure. The situation is not as clear-cut as, say, the nonrelativistic theory of the hydrogen atom. The corrections to this theory are important in a

region of the order of the Compton wavelength of the electron, or T j-j of the radius of the hydrogen atom. Correspondingly, the Baimer

formula

is a very good approximation, and fine-structure corrections are less than 0.1 per cent. Here, perhaps an accuracy of 5 per cent is

more appropriate. Even with the restriction to pions and nucleons, the picture is complicated. The reason is that in strongly interacting systems many virtual particles are present and all sorts of interactions that are allowed by invariance principles will take place. For instance, there is no reason why there should be no strong pion-pion interaction. This will
certainly also affect the pion-nucleon interaction, since the nucleon
is

surrounded by a fairly dense pion cloud. Fortunately, it turns out that in low-energy pion physics there are large effects, medium-sized The large effects have a common origin, effects, and small effects. an unstable excited state of the nucleon. The latter can be namely, obtained by the interaction term we shall use. The medium-sized effects amount to about 15 per cent corrections to the large effects and are due to pion-pion interactions, nonlinear pion-nucleon interactions, kinematical corrections, etc. Since these are missing in our model, we shall not be able to make predictions about effects which are not
1

Two

A suggestion such as this was advanced by

pions are not allowed by angular

(U.S.S.R.), 32:178 (1957), trans, in This would also contribute to a pion-pion interaction, for which there are some experimental indications at high energies. [See, e.g., L. S. Rodberg, Phys. Rev. Letters, 3:58 (1959); W. R. Frazer and J. R. Fulco, Phys. Rev. Letters, 2:365 (1959).] Quantitative results are only just being developed, and we shall therefore not consider this matter here. 3 See, e.g., H. A. Bethe and P. Morrison, "Elementary Nuclear Theory," 2d ed.,

momentum and parity conservation. I. E. Tamm, /. Expfl. Theoref. Phys. Soviet Phys. JETP (U.S.S.R.\ 5:41 (1957).

pp. 130, 132, John Wiley

& Sons, Inc., New York,

1956.

156
directly

PION PHYSICS
related

to

the

excited

state.

Techniques

for

calculating

now being developed. These methods are Finally, there generalizations of the ones we shall use for our model. are small effects (~few per cent), which may be due to strange-particle
medium-sized
effects are

interactions,

many-meson

forces, etc.,

and which are beyond the reach

of our present calculational power.

With this orientation, we turn to the question of the detailed form of the pion-nucleon interaction. The empirically known processes p <-> TT~ show (1) that there must be some p -f 7T, /><-> -f TT+, <->/?

linear coupling, since quadratic couplings, for example (which cannot be ruled out), would create meson pairs, and (2) that the coupling

involves charge degrees of freedom.


latter condition allows scattering to

We saw in the Lee model that the

occur even for a fixed source, that Of is, neglecting the translational degrees of freedom of the nucleon. course, momentum conservation requires that the latter be involved in the elementary process, but theory and experiments indicate that for many physical phenomena recoil effects are smaller than those due to the changes in the charge degrees of freedom. This can be seen as The process of emission and reabsorption of a virtual pion follows. a)- 1 = (// 2 -fwith momentum k lasts for a time of the order of A/ k 2)'-, where p is the mass of the pion. The recoil velocity of the nucleon is v = k/M, so that the fluctuations in position of the of mass nucleon, owing to the virtual processes, are of the order of

Ar~ i>Af~

2
(A;

-!-/* )-

This uncertainty is thus less than the Compton wavelength of the 2 x 10~ 14 cm. Correspondingly, the scattering cross nucleon, l/M section due to Galilean invariance and other recoil effects 1 is of the order -> oo. On the other hand, the characof (1/M) 2 and vanishes for teristic length for the emission and absorption process is the Compton 2 wavelength of the meson [actually (^ -f fc )"*], which is approximately seven times that of the nucleon. This length also determines the size of the scattering cross section due to the internal degrees of freedom. Thus, we expect to be on fairly safe ground in neglecting recoil effects

&

kinetic energies of, say, <3^. -> oo there are only two angular-momentum states accessible to the meson in the elementary process TV -* TT -f N. Since

for

meson

In the limit of

a nucleon at rest has no orbital angular momentum and spin i, the conservation of total angular momentum dictates that the pion can or 1. Since these only be emitted with orbital angular momentum
1

E.

M. Henley and M. A. Ruderman,

Phys. Rev., 90:719 (1953).

INTRODUCTION

157

two

states

other state

have opposite parity, emission can only occur into one or the if parity is conserved which occurs depends on the intrinsic
;

1 2 parity of the pion. Experimental evidence has definitely established that this parity is negative, so that the pions must be emitted into a state of orbital angular momentum / 1. This simple model is, in fact, able

all essential empirical facts. The emission of the be accompanied by a flip of the nucleonic spin, so that it is necessary to take into account the two spin states of the nucleon. This is done by simply introducing the set of Pauli matrices 3 o, which represent the spin pseudo vector of the source in the spin space. In the usual representation we have

to account for almost

pion

may

2x2

/O

1\

/O

^ll

-A
OJ

o]

'"H"

/I

'=\Q -l

Another property of the pion-nucleon interaction is that the invariance of the three-dimensional orthogonal group, which we considered in Chap. 7, is not destroyed in the presence of the source. 4 This has been amply demonstrated experimentally 5 in both pion and nuclear Since the charge physics and will be used by us as a guiding principle. of the pions must be coupled to that of the nucleons in order to space conserve the total charge (just as in the Lee model), it is essential to introduce the notion of a nucleon with two charge states represented by
the proton and the neutron. In this new space for the source, we introduce another set of Pauli matrices, T, the isospin of the nucleon, the properties of which are already familiar to us from the Lee

2x2

model.
linear

The most general coupling between mesons and nucleons that is and has the properties described above can be formulated in a

simple manner, taking into account the invariance properties of the Lagrangian. The odd intrinsic parity of the pion implies that the pion field changes sign under a reflection of the coordinates in the origin. Since L' is to contain linearly, it must be coupled to the source in such a manner as to form a true scalar; that is, L' must be even under
<

This was defined in Chap. 5. H. A. Bethe and F. de Hoffmann, "Mesons and Fields," vol. II, sec. 28c and d, Row, Peterson Company, Evanston, Hi., 1955. 3 The properties of the spin matrices are the same as in elementary quantum mechanics; see L. I. Schiff, "Quantum Mechanics," 2d ed., sec. 33, McGraw-Hill
2

&

Book Company,
4

Inc.,
is

New

York, 1955.
it

This statement

not exact;

holds only

if

electromagnetic forces, for example,


Fields," sees. 30, 31;

are neglected. 5 See, e.g., Bethe

and de Hoffmann, "Mesons and


cit.,

and

Bethe and Morrison, op.

pp. 97, 115, 128.

158

PION PHYSICS

This dictates a coupling of the form 1 a V<. reflections in the origin. If we then have a spherical source p(r) to ensure angular-momentum
conservation,

we can

write

r)a-V^(r)

(15.1)

this Lagrangian and shall see explicitly that the V operator causes emissions of pions only in states of orbital angular momentum unity. It remains to introduce "isospin conservation," which implies invariance under rotations in charge space (see Chap. 7). There are

combined rotation of a and r, In the next section we shall construct the angular-momentum eigenfunctions of the pion-nucleon system
is

As

required, this

is

invariant under a
5.2).

and

it

commutes with ^L (see Sec.

from

Since three charge states of the pion, TT+, 77, TT~. 1, 2, 3) a (a transforms like the components of a vector in charge space and
<

since the only other vector we have at our disposal obtain for the static coupling 2

is

T,

we

finally

L=IA,

2J

<ft- />(r)ra

a V^(r)
.

(15.2)

From our earlier physical discussion, we expect the source p(r) to mesons contain the effects not specifically included in (15.2), such as

therefore expected to have some shall assume that the structure with a range of perhaps i to $p. source is normalized to unity, $p(r)d 3 r 1, which should be kept in

and some

vestige of recoil.

It is

We

mind when we state that the coupling constant /measures the strength of the pion-nucleon interaction. This coupling constant is determined from experiments which we shall discuss in subsequent chapters. Of the various applications of the theory of pion-nucleon interaction,
the

most successful

is

pion-nucleon scattering.

We

shall also discuss

electromagnetic phenomena which are qualitatively accounted for by


the theory. Finally, we shall turn to the oldest historical application, namely, nuclear forces. These can be explained to a large extent by

Relations and Equations of Motion. Having decided on the interaction term, we are now in a position to write the
1

this theory. 15.2. Commutation

The
Jo,,

spin o, like the

angular-momentum operator L,
It is related

is

an

axial vector

and does
s,

not change sign under reflection.


s

to the spin of the nucleon,

by

2 The pion mass /* is introduced in this equation, so that/ becomes a dimensionless coupling constant. To distinguish between r space and isospace, we shall henceforth use English-letter subscripts for the former and Greek-letter subscripts for the

latter.

INTRODUCTION
fundamental equations of our theory.

159

We shall do this in terms of the


we introduced
in

various representations for the field variables which

Chap.

5: 1

The two expansions emphasize different aspects of the problem, and each of them will be used at some time in the future chapters. In order
to

expand

into spherical harmonics,


linear, spin

it is

useful to introduce circular, rather than

components,

(15-3)

-l

and

similarly for the isospin,

=-!

(15.4)

where 2

t^*

= (2)- (^
s
.

To distinguish between linear and are the operators diagonalizing f (3) circular components, we shall use subscripts for the former (going from
1

to 3)
1

and superscripts

for the latter (assuming the values

1,

0, 1).

In order to satisfy the canonical commutation rules the operators a are defined such that ^ oc -i(a tf). 2 The operators $ l were written as ^ in Chap. 7, where the operator f (3) is
defined.

160

PION PHYSICS

With these conventions we can write the total Hamiltonian of our model
as 1

= 2 a= -1

ld*k a**(k)a"(k)a>
/

=2

*
I

dk <al* m (k
(15.5)

I,w,a /0

The other variables occur only in // and field variables with / = 1. hence describe free particles only. The commutation rules at equal times are the usual ones

The constant <^ has been introduced, as in the Lee model, so that the physical ground-state energy can be adjusted to be zero. This simplifies the formulas which will be given in subsequent chapters and has no It should be noted that H' involves only the physical consequences.

[>,<5&]

- [<r,| - [>,<] - [<r,r] = 2ia llmn |> m n ]


,<r

at equal times

(15.6)

a/3y

is

the completely antisymmetric tensor which

was introduced

earlier.

The equations of motion


(15.5)

for the meson-nucleon system follow

from

and

(15.6) according to the general equation

-A

-f

p(r).

As usual, we assume that // is ordered and restrict The operators aa are defined similarly to ^*.

ourselves to real sources

INTRODUCTION
2 f ,2UV<*
J
(r,f)T v (f)e

161

fl

(27T)

= -*

*,, 2 f d r T J

(15.9)

Equations way:
<

(15.7)

can be rewritten as an integral equation in our standard

a (r,f)

== <*

(r0

<Pr' dt'

A ret(r

r', t

f') T

a(O a (O

or

a.(k,f)

A a (k,0

-J \

dt'

e'""-"

'

"

(15.10)

properties introduced into the Hamiltonian permit us That these several integrals of the equations of motion. quantities are constant can be verified by the use of the equations of motion or by checking that they commute with H. However, with the
to write

The invariance

down

sophistication the reader should have acquired by now, we go directly to the heart of the matter and recognize these quantities invariant. as the generators of the transformations which leave Conservation of angular momentum stems from the invariance of under simultaneous rotations of r and o. The generator for the transformation <j>(*j) -> $(r',/), where r' is related to r by a rotation about an

amount of
shall

axis

through the origin of coordinates, is the same as for the free fields. about some axis n is generated by rotation of o through an angle

= 17-^ = 0'
U

'2

e'"-*

<

15

'

>

For an infinitesimal rotation through an angle 66, U is I + /o n 60/2, and the total angular momentum, which is conserved, is [compare (5.3)]

-2 Jf
a

162

PION PHYSICS

In exactly the rotations of ^a

same way, the generator of the simultaneous isospin and ra [compare (7.26)] is found to be
T

(K)

(15.13)

Regarding the other classical constants, we have energy conservation but no momentum conservation. This comes about since no time origin is distinguished but the nucleon is fixed in the coordinate origin. Furthermore, we have parity conservation, since our Hamiltonian does not distinguish between a right-handed and left-handed coordinate
system.

The

explicit expression

is

[see (5.14c)]
(/ l,m,k
1

^_ = exp [-iV 2
l

1)4^*1 J

0>-</>(r)0>I

=is

0(-r)

^_a^l =

^_ T^I =
1

(15.14)

one constant connected with a peculiar symmetry of is not present in a more realistic extension of the If we forget about the / ^ 1 mesons and look at the reduced theory. Hamiltonian
Finally, there
this

model which

we notice that it is invariant under the exchange ra <-> a m and a\k). This exchanges / and T and leaves the commutation
invariant.

al (k) <-> relations

Consequently, there must be a constant unitary operator

leave it to the reader to find an effecting this transformation. for this operator, since it is not very useful. The explicit expression

We

general consequence of this symmetry is that, for every eigenstate of belonging to certain eigenvalues /' and T' of / and T, there is a degenerate state with eigenvalues T' and /'. Similarly, the pion-nucleon= | is equal to that for == J, / scattering phase shift for the state T

T=

f,

|-.

15.3.

Comparison with Other Models.


is,

of a linear coupling, then there are invariance properties; a simple one


the form

If we drop the requirement many forms which have the desired

perhaps, a quadratic coupling of

(15.15)

or

L'-r^rfrtf.CrWrJw,

( 15 - 16>

INTRODUCTION

163

For a point source, (15.15) reduces to the pair theory, whereas (15.16) of more complicated mathematical structure. It involves the charge degrees of freedom of the source and does not yield an exact treatment. We shall always assume that such couplings are only small corrections to the leading term (15.2). Terms of the form (15.15) and (15.16) are actually obtained when one attempts to make a nonrelativistic approximation to a relativistic pion-nucleon interaction. 1 Experiments dicis

furthermore, that pions interact with nucleons in S states, as follows from (15.15), but this interaction is considerably weaker than (15.2), although there are no convincing a priori arguments why this
tate,

should be true.

model with L' given by (15.2) only angularor P-wave mesons are coupled results in some important qualitative differences from the previous examples with 5 waves. Classically, incoming mesons with total linear momentum k and angular momentum 1 pass the nucleon at a distance k~ l For momenta k < p this distance is certainly larger than the source radius. Naively, one that mesons of this energy cannot be emitted by a nucleon, might expect since they emerge from a part of space where there is no meson source. Quantum-mechanically, the mesons in the source are not sharply
fact that in the
1

The

momentum / =

localized, but there

come from near

is a preference for those mesons which, classically, the source. It will turn out that the probability for
.

This is to be emission (or absorption) of a meson is proportional to k 2 if we look at //', which contains p(k)k and not only p(k). anticipated 2 In the direction Physically, the k dependence comes about as follows.

of emission, mesons of angular momentum 1 appear to emerge from the circumference of a circle centered about the nucleon and having a radius k~ l Quantum-mechanically, these mesons come from an area of order k~~ 2 The interaction strength is proportional to that fraction of the area which is inside the source. If we assume that the source M~ 1 9 where is the mass of the nucleon, then that fraction radius 2 This energy dependence of the effective interaction strength is (k/M) is reflected, for example, in the momentum distribution of virtual mesons in the nucleon cloud, which is proportional to k 2/ 3 rather than
. .

to 1/co 3 , as in the scalar theory. Similarly, in pion-nucleon scattering the low-energy cross section will be proportional to fc4 , since it involves

an emission and an absorption.

This

is

shown

in Fig. 15.1

and

is

in

striking contrast to y-e scattering, which starts out as a constant,

being caused by an S-wave emission and absorption. The increase of the interaction strength occurs only for k~ l
radius
1

>

source

(^M

-1
),

as

shown

in Fig. 15.2.

Hence the momentum cutoff

See F. J. Dyson, Phys. Rev., 73:929 (1948); S. D. Drell and E. M. Henley, Phys. Rev., 88:1053 (1952); and L. L. Foldy, Phys. Rev., 84:168 (1951).

164
^max

PION PHYSICS

~M

of the Fourier transform of the source determines the


strength of the source, which
is

maximum

max

^4-

p*
If the size of the source
is

\pi

reduced, then the

maximum source strength is

increased.

actually one in powers of //rmax //j.


250

Dimensionally, one would expect that an expansion in /is This is indeed the case, and it

200

100

150

200

250

300

350

Laboratory kinetic energy (mev)

(a)

80

fa
I
40

Q 20

50

100

150

200

250

300

350

Laboratory kinetic energy (mev)

(b)

Fig. 15.1. Parts

a and b show the

total scattering cross sections for

n+ and tr mesons

The respectively, as a function of the laboratory kinetic? energy. experimental curve and points are taken from H. L. Anderson, W. C. Davidon, and U. E. Kruse, Phys. Rev., 100:339 (1955). The data stem from the work of many physicists, references to which appear in the above article. The dashed curve superimposed on the low-energy data is proportional to the fourth power of
the center-of-mass

on protons,

which

will

momentum k. In part a there also appears a curve for be referred to in a subsequent chapter.

INTRODUCTION

165

Fig. 15.2. Plot of the effective strength of the pion-nucleon interaction as a function of the square of the pion momentum k. The solid line corresponds to a square cutoff and the dashed line to a Yukawa cutoff.

corresponds to an energy average of the interaction strength. Henceforth, when we talk of a strong or a weak coupling, we shall mean one for which fk mSLX /ju> > 1 or < 1, respectively. Regarding the actual form for p(r), we shall use a function which goes smoothly to zero outside the Since the detailed shape of p(r) is probably without range, l/& inax physical significance, we shall be most interested in results which do not depend on it. We shall see that at low energies the observable quantities depend on only one parameter of the source other than its strength, namely, its range. In practice, it is convenient to use a square cutoff for the source in momentum space:
.

for k for k

< >

kmax
/Cmax

where fe mttx above choice leads to trouble, one

~ l/(source radius) ~ M.
may

When
use a

the discontinuity of the

Yukawa

cutoff,

/>(/)

= =
a

or a gaussian cutoff,

p (k)

<r*

/ fc

ma

CHAPTER

16

General Features of the Static Model

16.1. Classical Treatment1 of Stationary Motion. In accordance with our usual procedure we shall first discuss the equations of motion in the limit where all quantities can be treated as commuting numbers once the equations have been obtained. 2 In this case the vectors a and T are taken to be unit vectors which define the direction of spin and isospin.

The equations system, and a


positions.

(15.7) to (15.9) then describe a nonlinear classical solution can be obtained only in the limit where all

quantities oscillate with infinitesimal amplitudes about their equilibrium Rather than solve this problem, we turn to a model with no In this model the classical equations can be solved exactly, and isospin.

we

The reason for neglecting T insight into their structure. that omission of the latter would change the dimensionality of the coupling constant and hence the cutoff dependence of the relevant
obtain

more

and not a

is

quantities.

In this section we consider solutions with in 0, e.g., the interaction of the nucleon with its own meson field. Some analogies to the corresponding problem in electromagnetic theory arise. However, the meson field is coupled to a and hence to the rotational rather than the transThus the main effect of the proper lational degrees of freedom. electromagnetic field of a charged body is to generate an addition to the inertial mass, whereas the proper meson field generates a moment of inertia of the nucleon. Analogous to the solution of the electromagnetic equations for which the field follows a uniformly moving charge, solutions will be found where a rotates uniformly and is
<f>

1 The development in this section closely follows that of W. Pauli, "Meson Theory of Nuclear Forces,** 2d ed., Interscience Publishers, Inc., New York, 1948. 2 In the classical limit, the equations of motion are obtained from Poisson brackets which have the same value as the corresponding commutators.

166

GENERAL FEATURES OF THE STATIC MODEL


behave
gular
h. field. The two theories quantum mechanics, where the momentum remains unquantized and the an-

followed by the meson


differently in

momentum must be a half-integer multiple of There the uniform rotation can occur only for certain frequencies which are determined by the moment of inertia of the meson field. If the moment of inertia is small, then the energy of this motion becomes large. When this energy is higher than the rest energy of the meson, the Fig. 16.1. Illustration nucleon will radiate a meson that is to say, this of the stationary-state excited rotational state of the nucleon becomes solution for the spin a. unstable against decay into a meson and a nucleon in the ground state. Such a motion corresponds to a damped rotation but can be sustained by taking fi n ^ 0, as we shall see subsequently. Empirically, such unstable excited levels of the nucleon have actually been discovered, and nucleon spectroscopy is an interesting new branch of physics. The variables of our model are o(/), which is a time-dependent unit vector and the classical meson field <(r,/). They are connected by the
;

equations

,-lf
I*

dt'

A ret (r dt'

r',t

(16.1)

=
f*J

"27(2^f
a(/)

and

= -2

jW P

(r'

(16.2)

The last equation shows that the meson field produces a torque acting on the spin. The direction of the torque is perpendicular to the gradient of the meson field and to the spin, so that a 2 = constant. If V< were a constant, the solution would be a gyration, like the motion of a
spinning top in a gravitational field. Although V< is not a prescribed constant in our case, we suspect nevertheless that a similar motion is Correspondingly, we seek a solution such that a rotates possible. uniformly around the 3 axis. As an Ansatz which contains the angle 6

and the frequency

as parameters (see Fig. 16.1),

we assume

(16.3)

168

PION PHYSICS
a) 8 ,

Since aa contains only the frequencies

the

meson

field

becomes 1

a^V^r)]
V being (Ko^
(')=
J</Vp(r')

(16.4)

< p or a > /*, K^ decays exponentially or reaches to infinity. In the former case there is no radiation, and we get in = 2 Like a stationary precession with 0, which we shall study first. most other calculations in classical field theory, a thorough investigation will involve quite a bit of elementary algebra.
Depending on whether
oj s
<

Inserting (16.4) in (16.2)

and using a

'i =0 and 3

d3 r f(r)

x^Xj

i(0

[a

+a

X
(o]

jd*r Vp(r)[a

Vr (r) +

ax

V^
(16.6)

X o^w,)

.)

2 =7T

r
JO

Our Ansatz
(16.6),

(16.3) implies that a(/)

(co s /cos

0)a

<*i

and hence solves


(\
f*

provided
/''Y

\ rr*e

(\

7\

Equation (16.7) determines the frequency


In the following
solutions for
s

oj s

we
p.

shall see the circumstances in

once the angle 6 is given. which (16.7) has

w <

To

this

end we have to compute the function


//,

If the cutoff

fcmax is

much

larger than

then

we can show
)2 .

that the

leading term of/(oj g) k


,..
.

is

w-independent by using the identity

= & - (^ -

^ + J^~.^

(16.8)

also has these properties.

take p(r) to be spherically symmetric and real, so that />(k) We call the coupling constant g to distinguish between this "neutral pseudoscalar" and the "symmetric pseudoscalar" case. 2 In this case the /e is irrelevant and will be dropped. 3 See footnote 1 on page 24 for an explanation of the notation.

We shall henceforth

GENERAL FEATURES OF THE STATIC MODEL

169

The

first

term in

this

decomposition gives the most divergent contribu-

tion for

& max ->

oo

2*=SF

(16.9)

whereas the next one diverges only linearly and


the source,
2
|

reflects the radius

a of

P (/c)|

f
J
r
-

The remaining term converges, and we can Collecting the contributions, we get

therefore set p(k)

1.

and

cK^

+ O^-otf-v
if

(16.11)

Our

calculation simplifies greatly

the size of the source

is

much
In

smaller than the

Compton wavelength of the

pion,

e.g., 1/0

> ^.H

mainly high-momentum components, and it can easily follow the comparatively slow motion of the spin. Expanding in successive retardation effects by using
this case the field contains

A ret(r,0
we

<XO

<5"(0

get for (16.4) the approximate

form

/-

o"(0
d/r

VK (r)

(16.12)

and hence the equation

d= ,_jl-X
677^
fl

(16.13)

This

is

the corresponding limiting form of (16.6) which implies

w =
s

or

co s

O7TM

2
fl

/i

>l\

^^ cos

(16.14)

co s

for //a sufficiently small there always exists a solution with This is no longer true in quantum theory, where the angular momentum has to be half an integer. From what we learn about the

Thus

<

//.

*[

The reason why we were not content


is

to consider only this limiting case

is

that

the complete solution

needed for an understanding of the scattering problem.

170

PION PHYSICS

motion of a rigid body in quantum theory, it is to be expected that we can obtain the quantum theoretic result if we supplement our classical
calculation by the condition that the angular momentum is quantized. shall see in the following chapter how this is borne out in quantum

We

To carry out the correspondence, we must first calculate the energy and the angular momentum of our solution by using the and J of the last chapter, but with onyssion of the formulas for
field theory.

H
2

isospin.

With
J[>

(16.4)

we

find, after

some
pa

calculation,

H =d*r

+ (V^) 2 + pi*] +

-* =4

(16.15)

477/r

La

The energy <^ was adjusted to be the energy2 of the static solution = 0). The remainder we called AZs, since it is the energy difference (to between the rotating and the static solution. We note that, for suffis

ciently large a,

>

0; that

is,

the static solution

is

the one with

lowest energy. For the angular

momentum
2

(15.12),

we

find

d*k
f

pfofa.k

/a .k

~~

LiJ

The components of J parallel to ol cancel because of (16.6), as they should. Hence the total angular momentum is a constant vector in the
z direction.
1 Note that the positive contributions come from /f and the (larger) negative contributions from H'. 2 This turns out to be independent of 0, as it should. The energy of the ground

state

is

zero.

GENERAL FEATURES OF THE STATIC MODEL

171

To

get simpler results,

we go

to the limit a~

>

/*,

which

is

satisfied in the

pion-nucleon interaction.
s

We

actually

then have

2 cos

and hence
j
For

=
of J, we find
t

(16.17)

AE expressed

in terms

corresponding to the rotational energy of a rigid body with a


inertia
spin.

moment of

/(47r^ 3a).

Its

due to the meson field following the can be easily understood and is the sum over all magnitude

The

latter is

momenta of
it

(Probability for finding a meson) can escape from the source) 2

x (energy of the meson) x

(distance

level in

The expression (16.18) can be used to estimate the energy of the next 1 quantum theory. The one after the ground state (J = ) is a
f level with an energy

2 Thus, only for sufficiently large # /47r can we meet the condition o> 6 < /* The reason for this is for this state not to decay by meson radiation. that only for sufficiently strong coupling is the meson cloud thick enough to have an appreciable moment of inertia. For too small a moment of inertia the rotation has to be very fast to acquire one unit of angular momentum. The meson cloud will then not endure the This situation is analogous to the centrifugal force and will break up. neutron-proton interaction, which is not strong enough to bind more than the triplet-spin S state (deuteron). Higher-angular-momentum

In nature we seem to states are torn apart by the centrifugal force. have this situation; that is, the coupling is not strong enough for a
stable excited state of the nucleon. However, experimentally one finds a huge resonance in the scattering of mesons for meson energies ~2/i*.
1

There J 2

J should be replaced by J(J

1)

172

PION PHYSICS
is

This

attributed to

an unstable excited

state,

whose influence on
co b

scattering will be studied in the next section. 16.2. Classical Treatment of Scattering.

For

>

p,

C(co b)

as

given by (16.11) becomes complex. Correspondingly, our Ansatz (16.3) no longer works. complex frequency implies a damping due to radiation. To sustain the motion, we introduce an incoming field in realized by an incident beam of mesons, and the precession then

<

shows up as a resonance in the scattering. For in we take a plane wave


<

in
<

(r,0

Ae'* *'*-* 9
3

A*e -^o-r-.)

(16.19)

the amplitude of which is of the order (volume of normalization)"*, so It is now that we can neglect powers higher than the first power of A.

convenient to decompose a as follows

*
Since
1

<*o

i(0

i(0

= *KK l<M + *K)e +to


is

(16.20)

we want a
at
will

solution for o which

static before the incident

wave

be of the order A, and we may drop higher powers of it. The representation of the o's corresponding to the Ansatz (16.3) is
arrives,

*o

The treatment of
#r,f)
is

the field equation with


ret

in
<

A*0 + - Jf*V *' A


/*

(r

r', t

- fXO

W)

(16.22)

completely analogous to the previous calculation, and we immediately turn to what corresponds to the positive-frequency part of (16.6):
/O)

a(w

2i
A*

A<JO

p(/c)

C(w

)a

a(a>

(16.23)

The

term on the right-hand side is the torque due to the incident The second term is the reaction of the field of the source, and t From (16.23) it appears is exactly the same expression we had before. that only the component of k perpendicular to a contributes to the motion. Putting the x axis in this direction, we find that the first term
first

field.

is

in the

direction

and the second

is

in the

xy plane perpendicular to
(a

a(coj.

Therefore

we can
<T(CO O)

satisfy (16.23) with the Ansatz

ya

- fa X

(16.24)
<

To

realize this situation,

we have

to

form a wave packet

for

in
.

GENERAL FEATURES OF THE STATIC MODEL


or, in the

173

above-mentioned frame,

Substituting into (16.23),

we

find

=
where C(w
) is

2gAa>oP(k)

~2igAC( mo)p(k)

M [;

- C (eo )]
2

(16.25)

6?7/r \ a
is similar to that of a harmonic oscillator with and under the influence of a periodic external force. Neglectdamping ing C, we have a linear motion in the y direction which is in phase with the in C represents the intorque from fluence of the proper field which generates a torque in the x direction for The result is an elliptic this motion. motion whose projection in the y diand torque
<
.

(16.26)

The motion of a

rection

is

out of phase with


If

in
<

as

is

indicated by the imaginary part of


16.2).

C
Fig. 16.2. Illustration of the solution for the spin a under the

becomes so large (Fig. C2 (w )] = 0, the motion that Re [o>J in the y direction is 90 out of phase.
()

influence

of

an

incident
.

If

/a

> co

this
0) r

happens for
2

wave of momentum k

6-rr^alg
oscillation

between the precessing k is characterized by d.

meson The phase spin o l and

and the amplitude of the


attains
its

value at this energy. This resonance frequency is 1. the one for which we found the precessing solution with cos 6 just To find the scattering cross section, we evaluate <(r,0, given by o creates only an exponentially decaying (16.22) in the limit r ~> oo. field, and at large distances the oscillating field produced by a(o> ) is the

maximum

usual outgoing radial wave

(c.c.

complex conjugate),
a(co
irfjir

ks

c.c.

e*"

/(*)

c.c.

(16.27)

where k s has the length of k and the direction of

r.

The

differential

174

PION PHYSICS
is then found in the standard manner by comparing the and the scattered flux:

cross section

incident

k,

-a

ko

.<J

(a

(16.28)

In contradistinction to the examples we considered in the second part of the book, this cross section is not isotropic. It depends on the relative orientation of the three vectors k , k s , o and contains a great deal of information. Mesons are emitted preferentially in the direction of motion of the spin. The motion due to the incident meson gives an whereas the angular distribution peaked perpendicular to a and k other term of the amplitude which originates from the field reaction favors the incident direction. All this is valuable for checking the theory experimentally, but these subtle details are not rendered correctly by our classical theory. The reason is that a spin J is remote from a classical angular momentum, its zero-point oscillations being of its own 2 = 2 = 1 Cor3). quantum-mechanically, o magnitude (classically, o we shall ignore these predictions of the theory and average respondingly, over the spin directions
,

2 2

7 cos

eo

-C

sin
2

I]

(16.29)
eo

(co )

where k
If
<o

is the length of k and k and ft is the angle between them. and the resonance energy are much less than > raax we have
s
,

(16.30)

with the resonance energy


(16.31)

and the width

r
Near the resonance the
da
__ ~~

34V
3
(60

--**
r

(16.32)

cross section assumes the familiar shape


8 /g \ k*
2

d&

15 \47T/

+ cos 2 2 ) + T /4
2
6t)

(16.33)

GENERAL FEATURES OF THE STATIC MODEL


This angular distribution
is
2

175

only qualitatively correct (the exact

quantum-mechanical result is 1 4- 3 cos 6). At co r the total cross section reaches a peak value a = 167T/&*, which is twice the quantummechanical value for a J = f resonance (see Fig. 15.1). Similarly, this kind of calculation 1 would not render accurately the charge distribution if isospin had been included. The experiments agree within their. accuracy of 5 to 10 per cent with the quantum-mechanical predictions for a J = f resonance. Thus the classical calculation is not suited for comparison with this kind of measurement. Nevertheless, the dominant fact of low-energy pion physics, namely, a resonant state with
higher angular

momentum,

Thus the basic

is correctly predicted by the classical model. features of the pion-nucleon interaction can be under-

stood in terms of the intuitive picture of a nucleon surrounded by

its

meson
16.3.

field.

classical limiting form to effects originating

the Static Model. Having discussed the of the equations of motion, we turn our attention from the quantum nature of the field and spin and isospin. Whereas spin and isospin are definitely in the region of small quantum numbers, neither of the two complementary aspects of the field is clearly realized in the pion-nucleon system, but both are needed for its understanding. Let us first study the features of the elementary processes described by //'. To obtain some insight into the isospin properties, we use circular components [see (15.4)]. The a a T to a bare nucleon state creates 3 7r(0), Tr+(<j) L ) and application of l TT mesons and in the last two cases simultaneously changes a ((/>~ ) + proton into a neutron (T~), or vice versa (T ), in such a manner as to conserve charge. We obtain 4

Quantum Aspects of

<

-(*)*

2 *A

-(* 2 *.T.
with

~ (*)* K) *=|T= J, T = 1) n) = -()* pO + ft)* mi*)


P)

= (I)*

"T + )

(16 34)
'

|n7r )
|

= -<f> \n) = -^ p) PTT+)


+
l
|

See W. W. Wada, Phys. Rev., 88:1032 (1952). In our classical calculation the nucleon has / = -J. Y ^ 1 3 = We remind the reader that 3> + j" ~=~ </>_. 4 The factor (ffl is a normalization constant. The relative phases have been chosen to agree with E. U. Condon and G. H. Shortley, "The Theory of Atomic
2
</>
<f>

<f>

Spectra," chap. 2, Cambridge University Press,

New

York, 1953.

176
Since

PION PHYSICS

2 &cTa a

is

invariant under rotations in isospin space

(e.g., it is

nucleons.

scalar in this space), the states (16.34) must have isospin J, like the In fact, the coefficients of the expressions on the right-hand

side of (16.34) are the well-known Clebsch-Gordan coefficients, formed by the addition of the pion isospin (== 1) to that of the bare nucleon The other states of nucleonic (== i), so as to add to a total isospin of \. = f, tn = i and charge of the pion-nucleon system must have T must be orthogonal to the T = J states they are 1
;

T= T=

T =
z

I,

= (i)* = - J) = (*)* T.
i)

nir
|

+
)

-f

3
|

pir-)

(f )*

The remaining two

states

of
3.

T=

f are
j3\

Inr

_ r-

ji ,T2

_ -

2)

_ (1636)

7 = i or 7" = f states, it is easy z to see that there are always equal amounts of TT, TT+, and TT~ mesons. This is an expression of the isotropy in isospin space which was built into the form of the Hamiltonian (15.5) by choosing the same-strength

We All the states of (16.34) to (16.36) are eigenstates of T 2 and Tz note that the T , r, J states contain twice as many charged f, pions as neutral ones, whereas this ratio is reversed in the T
.

T =

states.

If we average over all

From (16.34) it follows coupling constant |/| for all components a that the consequent interaction strength for emission of a charged meson is (2)- times that for a neutral one. Furthermore, the coupling constant for the emission of neutral pions has the opposite sign for protons and neutrons. This last statement is of nontrivial group theoretic origin and can be observed experimentally by studying, for example, the elastic (e.g., the final nuclear state has the same energy as the initial one) photoproduction of 77 from deuterium. The TT waves emerging from the proton and neutron will interfere destructively or constructively, depending on the relative sign of the coupling constants.
< .

As was mentioned in of H e.g., H' k,


',

the last chapter, the


is

characteristic of a P-state interaction

momentum-space dependence and will

play an essential role in the development of the following chapters. Similar to the isospin, J rf s r p(r)a V<, when applied to a nucleon, only generates states of the same total angular momentum J. There are
1 The phases have again been chosen Gordan coefficients (see, e.g., ibid.).

so as to agree with the standard Clebsch-

GENERAL FEATURES OF THE STATIC MODEL

177

again six possible states of the pion-nucleon system, corresponding to = I and / = }. These can be written in complete analogy to (16.34)

and
1

(16.35) if we designate the pion states by their /z components and the nucleon spin components by t(up) and j(down):

1,

0,

|/
I

= t.j.= = (*)* 1 + ()* to) - - (* t-D + (|) Jo) J = I, J = J-iJ, = -f)-| l-l)
1

J
I

=*t

J,

-4)

-(I)-

t-i)

+ (J)-

jo)

Although we are
classical properties

in the region of small quantum numbers, the of the angular-momentum additions are already

f , Jz |), which classically apparent. For instance, in the state / leans to the side, the fO) configuration has twice the weight of the In the J 4 Jz J) state, on the other hand, j 1) configuration.
| |

Because of conservation of angular momentum, only the J = \ states are accessible, when a nucleon emits a pion, and (16.37) then shows that in this process the nucleon flips its spin two out of three times. This is the quantum-mechanical expression of the
the situation
is

reversed.

classical spin precession

we

studied earlier.

A statement analogous to the ratio of charged to neutral pions is that

meson cloud has a spherical shape. There are twice as many 1 in the / = 1 state. as with l z = Since the former mesons /z = have a distribution proportional to | sin 2 and the latter to cos 2 0, we obtain an isotropic distribution, conforming with the general theorem 1 On the other hand, that a spin-J particle has no quadrupole moment. = I, Jz = 4 state has a meson distribution oc -|- 3 cos 2 0, as was aJ mentioned in the last section. We conclude this section with some remarks about the eigenvalue
the
I

If//' = 0, then the spectrum is simple (see, e.g., Fig. = J = |, there is the nucleon, a fourfold degenerate For T 12.3). At E = the continuum of states of one meson state with E = 0. /= f and we can have states with T = f, / = |; T = begins, T = |, J = |. In general, states of mesons coupled to the nucleon with T or J = (2n + l)/2 appear at or above an energy E = rip.

spectrum of//.

/*,

When
1

//' is switched on, all these

energy levels are shifted to lower


in the

could,

Although the statement is deduced here only for the case of one meson it can be J. generalized to any system of J

178

PION PHYSICS

1 energies, but

the structure of it is believed that for sufficiently small the energy spectrum is not changed. This means that there is still a fourfold degenerate ground state N), which we shall call the physical
|

'

After an energy gap of magnitude p, there will be the continuum of physical "nucleon -f- one meson" states. These can be obtained by applying in (k -> 0, t) to N). Since the operator ^ln has a time dependence eitft , the resulting state is actually an eigpnstate of with energy /*. States with several mesons can be generated by applying products of creation operators to the ground state.
nucleon.
<f>
|

It is to be expected that for sufficiently strong interactions the structure of the energy spectrum changes. For larger than a certain critical value, we may find discrete states of the meson-nucleon system, representing stable excited states of the physical nucj^pn. As we saw,

would correspond to a precession of the nucleon spin (or isospin) and its meson field with frequencies <//. In the classical approximation this occurs for sufficiently strong In quantum theory no one has found pa). coupling constants (/
in the classical picture these states

>

approximations) the minimum strength of the constant that is necessary to obtain bound excited states. It coupling has not even been possible to prove that these discrete states will always have E > 0, so that the ground state of the system corresponds to T=J= At this point we have to proceed semiempirically and use the fact that the nucleon we find in nature has / = T = i and that there are no stable isobars, the first excited state with / = T = f level already being in the continuum. Experience with approximate treatments tells us that this kind of behavior is also expected from the model if the empirical coupling constant and cutoff are used. Consequently we shall base our further development on the assumption that the energy spectrum is normal.
(except in
certain
.

This

is

the prediction of perturbation theory, since

where

and

En

shall see that for the

are the energies of the ground and nth states, respectively. ground state this is an exact statement.

We

CHAPTER

17

The Ground

State

17.1.

Exact Results.

For a long time

it

has been one of the main

goals of

meson theory

bare nucleon and its a dead-end road and has not yet yielded to calculations. Furthermore, even if it could be solved, the result would be of limited value. The reason is that the large effect in pion physics, namely, the resonant state of the nucleon, is not important for the ground state. For it, mediumsized effects included in the model and those excluded are as important as the resonance. As we shall see, the measurable quantities of the ground state are predicted by the model to within an accuracy of only 50 per cent. Thus, with regard to fruitfulness and complexity, the problem can be compared with a calculation of the ground state of light nuclei with inaccurate nuclear forces. One of the main advances in static meson theory has actually been a divorce from the concern with the physical nucleon. From a practical point of view, the model is most powerful in describing processes such as pion-nucleon scattering. For this, one needs only certain matrix elements between the ground state, and many quantities, such as the remain unobservable. Nevertheless, to gain some self-energy <f we shall review in this chapter the important features of what is insight, known about the ground state and what has been learned in the past by various approximation schemes. These methods form the bulk of the
,

to analyze the physical nucleon in terms of the surrounding meson cloud. This problem led into

later sections.

First

we shall study the form of important matrix elements.


N)

We have

already argued that the ground state

=
|

N,

in)

=
|

N, out)

is

fourfold degenerate.

The degenerate
179

states all

have spin and isospin

180

PION PHYSICS
\ |

Matrix elements and can be written as \p\} 9 \p\\ f>, j>. between them can be largely reduced by taking advantage of the transformation properties under various invariance groups. The fact that these states transform like bare nucleons under spatial and isospin 1 rotations can be expressed by
n

<a,;|^
iT n
-

|a',/)

<a,7|e
for arbitrary vectors n for the isospin andy

'|a',/>
n'.

= (a,;|^- n |a',/) = (a,;|^ n'|a' /)


s

and
1,

Here,

we have used

subscripts

<x

1,

the four nucleon states.

of 2 for the angular-momentum since a and T have the same Furthermore,


classification

transformation properties under rotations as J and nucleon states are eigenstates of the latter operators,
<P (P
|

and

since

our

we have
(1? 2)

T3

P>

T3

= H> =

-<

T8

JI>

By

rotations these relations can be generalized to

(*J
where
r x is

r,

',/>

r^aj

r,

a',/)
a'.

*,
Similarly,

(17.3)

number independent of ft
(a, J
|

a,

and
\

we have
(17.4)

cr,

= a', /}

ri(a, j

at

a',

/)

6^

where rj must be the same number as r x because of the invariance of the rotations in spin theory under the exchange of J and T. Simultaneous and isospin space give
'

tf

*<r.

=
|

r2 (f
.

a .ra
t

f)

(17.5)

which contains another constant

r2
. .

we

use a single subscript whenever it is convenient.

To avoid crowding of
.

1,

4 for labeling the nucleon

subscripts, states

seen that the matrix elements of J, T, o, T and their two free paramproducts between physical nucleon states involve but and r2 They are, in fact, not completely free, since they eters, T! must satisfy several inequalities. To this end, we express the physical nucleon in terms of bare states

So

far

we have
.

|>

K(a ,o,T)

f)

(17.6)

where the "dressing operator" R is a rotationally invariant combination of the o, T, and meson-creation operators, as in (16.34), for example. Since both the physical and bare nucleons have spins and isospins of J,
1 The operators cr, T are time-dependent but are here taken at t = 0, as is always understood when no time dependence is mentioned. Hence, the matrix elements of o and T between bare states |) are the standard Pauli matrices.

THE GROUND STATE


the angular

181

or

1.

momentum and the isospin of the meson cloud can only be If we label the various terms in by subscripts /, 1Z9 t, and t M9 in

that order, to indicate the quantum numbers of the corresponding part of the meson cloud, then will be a linear combination of the operators Thus, we can write the ground state ^oooo> ^i/,00* R'ooit t > ai*d R'u g it

p\ ),

for example! as
I

Pt>

Kiooo

Pt)
|

+ *noo + R'wn
where the

+ Kioii t) + Kioio Pt) Pi) + *iooo Pt) + *I m ill) t) + R'nw Pi) + *ioio Pt)
| | |

contain only meson-creation operators. So far we have assured only that \p\) is an eigenstate of Tz and Jz but in Chap. 16 we learned how to prepare eigenstates of T and /. Following the argument given there, we introduce the operators /? 0000 , ^ii,oo ^ooi e > RII Z H Z in such a way that those with different lz (or t f ) are connected by The R differ from the R' by the extraction of some Clebschrotations. Gordan coefficients, and we get, in terms of the former,
,
|

Pt)

= *oooo

+ (4)^2^0011 + (i)*P^noo + ipRmi i) - 2^ 10n


I

Pt)

*t)

~ *ooio - *iooo Pt)] Pi) - 2*K mo p|) + K 1010 nt)


I
|

(17.7)
|

Pt)]

Since the bare

vacuum

is

invariant under rotations, the matrix elements

CK
are independent of
lz

|4AU HA |0
t^ The

(17.8)

and

normalization of the physical

nucleon state requires


Coo

Coi

+ C 10 + C u =

(17.9)

Furthermore, the symmetry of the theory under exchange of space and


isospace requires

Cu
so that

fl

(17.10)

we have

the following inequalities:

<

Coo

<

< C10 = C01 < < Cu < 1


The constants
1 This
r^

(17.11)

can be readily expressed in terms of the matrices

Cu

M.

theorem. See, e.g., is a general result usually called the Wigner-Eckart E. Rose, "Elementary Theory of Angular Momentum," p. 85, John Wiley &Sons, Inc., New York, 1957.

182

PION PHYSICS
states are eigenstates

We have only to remember that the bare nucleon of the operators T and o, e.g.,
*3
|

P)

<*3

P)

=
I

P)

and that the operators -R f commute with a and T. Since the independent of lz and t Z9 we may calculate any matrix element,
t

lt

are

e.g.,

rl

_ ~

(pt |r,|pt>
,
A
I i

IT

c ^00

""T

|C f ^10
*

ic S -'!!
1

"
r2

n_ 3T3 _ (Pi "'1 pt) ~~ ^00 c IT


cr
|
i

<"' !2>

I i

(Pt

(T3

T3

c 8^10

"T

The reader may check that the evaluation of, say, (p\ a~ l p\ > leads Because of (17.9) and (17.10), only two C's to the same result for r^ are independent and, hence, can be expressed by the r's. By inserting (17.9) and (17.10) into (17.1 1) and (17.12), we derive several inequalities,
\
\

e.g.,

-J -1
there are

<. tj <, 1 <, 3r2

-J
x

<; r2
1

<: 1

+ 2t

<

2t!

- 3r

>

(1? 13)

There are no further exact statements that can be made. However, good reasons to expect C10 to be small. Since a single meson has / = t = 1, the states in question must have at least two mesons, the angular momenta of which add to 1, while the isospins compensate Therefore, the angular momentum of the mesonic (or vice versa). wave function is odd, 1 whereas the isospin is even under exchange of the

mesons (or vice versa), and their Bose-Einstein statistics requires the This radial antisymmetriradial part of the wave function to be odd. zation, together with the short-range and exponential radial decay (klTl+k * r * e~ (k * r i +k i r * ), suggests that the vacuum expectation (e.g., e~ value which defines C 01 should be small. This conjecture is strength}
)

ened by an exact calculation of such states in an extension of the Lee model which allows the exchanges of more than one meson. 2 In this

model
1

Ci <

0.01.

With the assumption that

CQl

is

negligible,

r2 are both

Cu),

completely determined by a single parameter and we obtain

Cu (since C00 =

^ and

r^l-tCu
r,
1

= 1 - tCu = HI +

(17H)
2rt )

The state with unit total angular momentum of two particles of individual = k], x k2). angular momentum 1 is odd under exchange of the two particles (e.g., k When the angular momenta add to 0, then the state is even under exchange (e.g.,

k-k^k,).
*

U. Haber-Schaim and

W.

Thirring,

Nuovo cimento, 2:100

(1955).

THE GROUND STATE

183

The importance of

rx

and

r2 is

physical nucleon states which

that they are the sole quantities of the we shall need to relate the theory to

experimental measurements. There have been many attempts, however, to obtain approximate expressions for the operators R as well as These calculations are illustrations of our for their expectation values. general development, and they allow us to calculate ^ and r a as well as other matrices in certain limiting, but for the most part unrealistic, cases. 1 In the following we shall outline the main points of various

approximations, in current usage, when applied to the ground state. 2 17.2. Perturbation Theory. In perturbation theory, the coupling constant /and, therefore, H' are considered to be a small disturbance of the free meson field. In an expansion in //' (or/), the first-order change in the ground-state wave function is due to the possible presence of just one meson, which can be emitted by the action of H'. For example,
since

\N)

= 0,

In this approximation the wave function of the virtual mesons is similar to that of the neutral static-source-model state, aside from the
last chapters.

the spin and isospin dependence, which we discussed in the It is clear that the wave function is not (but could be) The energy of the ground state being adjusted normalized to order/2
factor

k and

to zero determines

<f

to
:

'

order/ [provided

(f

H'
|

f) is zero, as it is in

the theory considered]


(Pt

H
|

pt)

2(pt
-2(pf

H'a>alflH'
"o
H'
|

pt)

"o
|

L H'

pt)

/.

(17.16)

This energy diverges cubically for a point source, and it is, therefore, very sensitive to the form or cutoff momentum of the source. We can
1

The reader who

is

interested only in practical results

may omit

the remaining
Inc.,

sections of this chapter. 2 See, e.g., R. E. Marshak,

"Meson

Physics,*'

McGraw-Hill Book Company,

New

York, 1952.

184

PION PHYSICS

also write (17.15) in the form of (17.7). to n contribute, because at 00 and

Furthermore,

J? ooo

Only the terms corresponding most one meson is present.

^1010

so that
COG

C10

=C =
01

Because the state

|/?f) is

not normalized, (17.9) and (17.13) do not

apply; from (17.12), however,

we

find

2
(fc max //*)

If/

>

1,

approximately. The perturbation approach is clearly valid normalized to order/ 2 only if Cn < 1, and the whole expansion in powers of/(e.g.,/fcmax ///) becomes impracticable unless this is the case. For example, for fcmax M, this limits/to/ < 0.1 and/ a /47r < 10~ 3 whereas we shall see later that experiments require / 2 /4^r 10" 1
.

then the inequalities (17.13) cannot even be satisfied This is due to the fact that the state function is not

Tamm-Dancoff Approximation. 1 The Tamm-Dancoff method attempts to remedy some of the shortcomings of perturbation theory. It is similar in that it limits the number of mesons in the cloud (usually
17.3.

to one) but differs in that

it

solves the Schrodinger equation in this


it

subspace.

Correspondingly,

Lee model for the

Q=

\ states.

agrees with the exact solution of the From (17.7), the general form of the
is

ground

state in the

one-meson approximation considered

pf>

= ^l + ^ L J
a

(27T)

-,
4- 3

I(k)

rj(k)

pt)

(17.20)

where

is

a normalization constant,

^r- 2
1
1.

|/(k)|

(17.21)

Tamm,

(1950).

Row,

/. Phys. (U.S.S.R.), 9:449 (1945); S. M. Dancoff, Phys. Rev., 78:382 See also H. A. Bethe and F. de Hoffmann, "Mesons and Fields," vol. II, Peterson Company, Evanston, 111., 1955.

&

THE GROUND STATE


and/(k) describes the meson momentum-space wave function.
(17.20)
is

185

When

substituted into the Schrodinger equation 1

B
I

Pt>

(HQ

+ H'- *

pt>

=
\

(17.22)

then the application of H' to the term proportional to a\k) p) creates states with two and no mesons. In the spirit of the approximation stated earlier, the two-meson amplitude is neglected, and we obtain

e/f

A - s - if\ f
I

fi

~"

jlt

We

deduce, therefore,

and

A
-

__$ =
^/(In)*, (17.25)
is

(,7.25,

In the continuum limit,

e.g.,

an

integral
is

The wave function/(k), as anticipated, equation for the number ^ to that of the Lee model. In the weak-coupling limit analogous

and/(k) and

<^

go over into the perturbation-theory

result.

On

the

other hand, for the

more

realistic limit

2/*

co

we

obtain (using the same sign as in perturbation theory)

(.7.26,

Whereas
1

2 for small/ the quantities

Cn C01
,

and

C00 correspond to those


trial state

(17.20).

This can also be considered to be a variational procedure with the This method will be developed in the next section.

186

PION PHYSICS

of perturbation theory, in the opposite limit considered above we find


/(k)

-i

^(2o>)*

-X S& f
C01
*i

and

^=
Cu

so that

Coo
*i

=l =i

= C10 = =*

(1?27)

agreement with the experimental numbers that we shall find later. Since the physical-nucleon-state function is normalized, the inIt is not clear, equalities (17.13) are satisfied in this limiting case.
in fair

however, whether the results resemble the exact solution. There is no a priori reason why two-meson and higher amplitudes should be
negligible for large fkm jp.

17 A. Tomonaga Intermediate-coupling Approximation. 1 The Tomonaga method for intermediate coupling strengths does not limit the number of mesons in the cloud. Instead, it takes advantage of the Bose-Einstein statistics of the pions, which favors a clustering and assumes that they all have the same radial wave functions. For this
reason,
it is

Eq. (5.100).

simpler to go to the angular-momentum representation, Thus, if we expand the meson field in a complete set of

orthonormal radial functions

F (k)
s

in the sense that

MmK
'

(17.28)

*o

then the approximation for the ground state consists in assuming a function (to be determined by a variational principle) that involves only one mode, say s =0. The new operators 0, obey the usual
trial

commutation

relations for every value of s:

[flSl^IvV]

fl^Al'dfnm'

(17.29)

The Hamiltonian

is

then quite generally

(17.30)

See also T. D. Lee and S* Tomonaga, Progr. Theoret. Phys. (Kyoto), 2:6 (1947). Pines, Phys. Rev., 92:883 (1953); T. D. Lee and R. Christian, Phys. Rev., 94:1760 (1954); M. H. Friedman, T. D. Lee, and R. Christian, Phys. Rev. 9 100:1494 (1955);

D.
E.

M. Henley and
2

T. D. Lee, Phys. Rev., 101:1536 (1956). This could also have been done in the last two sections.

THE GROUND STATE


with In the

187

W = f "dk o>F;(fc)F,(*)
Jo
trial state
|

/.

=f
fji

t"dk
Jo

^yyffi (127rw)
f

(17.31)

Tomonaga approximation only

the term s
for ^

is

retained 1
best

F (k) =s

taking a

t)

with a s

N =
t)

0.

The

by form of

F(fc) is

found by minimizing
<f

(N \HQ
t

H'\

t)

(17.32)

with respect to the form of F(k) and the dependence of ) on the meson-creation and -destruction operators a^ and a^. In the former procedure, we take into account the restraint
|

by introducing a Lagrangian

multiplier

A'.

We

obtain

2 V^I,

+
and thus

(N,

2 wl,
j

^Vt )

^(12^))%

-^- + W(k) =
(17 33)
'

find, for the

form of F(k\
F(k)

4^T% +
0)\0)
A)
A:

is

is a normalization constant independent of where an undetermined multiplier, related to A' by

and where A

with this form of F(k), we get the lowest eigenvalue as a This parameter is then to be determined by a further variational calculation, which will be carried out in the next section. The form of F(k) determined above is the same as that of the TammDancoff approximation (17.24), but it should be remembered that the

Minimizing

function of

A.

number of mesons
1

is

not limited.

1 and Since only / the subscripts / and s from aa

5=0 enter into


/m>

the following,

we

shall henceforth

drop

188
It

PION PHYSICS

remains to determine the ground-state eigenfynction ) and the (f and /, and, therefore, on A. The rewhich depends on duced Hamiltonian, (17.30) with s = only, corresponds to nine oscillators (a = 1, 2, 3; j = 1, 2, 3) coupled to a spin and an isospin, and its diagonalization is a problem in elementary wave mechanics. However, it cannot be carried out in closed form. This would be true even if we had the complications due only to spin (neutral pseudoscalar theory) or to isospin (symmetric scalar theory). To gain some insight into the problem, we shall consider the former case 1 in some detail. Here the reduced Hamiltonian is 2
energy
,

H=W% a]a, + gofa +


3

a])

(17.34)

Going back

to the canonically conjugate operators p

and q by

we can

rewrite

H in the form
H=
2

|(p
2

+ W V) + g'vq-iW-eo
2

(17.36)

with

-!^ j=i

and

g'

(17.36) emphasizes the formal analogy to elementary wave mechanics and corresponds to a three-dimensional oscillator coupled to a spin by means of an interaction a q. This interaction does not conserve parity, since a -> a, but q -> q under a reflection in the

The form

By carrying the analogy further, we note that the ground state should be a mixture of an angular momentum Sj and a P state. 3 These are the only states allowed by the restriction j = | for the bare
origin.

and physical nucleon, and they correspond to angular-momentum 7 = and / = 1 for the meson cloud. These states will be of the form 4 A (^) N) and a qhi(q) N). However, we shall not expand in
|

distinguish this case, we shall replace /by g, as was done in Chap. 16. 2 assume that A is real ; therefore, g = *, and F(k) as determined by (17.33) then real, too.

To

We

is

level ordering in a central potential is applicable and proves = J state. See R. G. Sachs, "Nuclear Theory," ground state actually is ay appendix 1, Addison- Wesley Publishing Company, Reading, Mass., 1953.

The theorem of

that the

THE GROUND STATE

189

terms of these states but rather in terms of those for which a is parallel or antiparallel to q, since this determines the sign of //'. Introducing the projection operators into the eigenstates of a q,
<P

l( 2 \
h

^3) q
I
|

(,7.37)

we

write the ground state as

-^l + ^_ Mg>]
q

N)

(17.38)

-I

where h
27T

satisfies

Jo

\"dq [|M<7)|

|M<7)|

MO) - MO) =
is

(17.39)

Since

^p

is

rotation-invariant, (17.38)

value J and of Jz with eigenvalue spin up or down. Evaluating

an eigenstate of J with eigendepending on whether JV) has N} = with the aid of


J-,
|

Aa B = A B +

/a

AX B

we

find

These two second-order equations cannot be solved analytically but can be readily discussed in limiting cases. Whereas for small values of g' we are led back to the perturbation we find that h__ becomes the result (e.g., h dominant), for large g' > leading term. This can be seen by realizing that (17.40) corresponds to a harmonic oscillator displaced by g'/W* from the center (Fig. 17.1). Since we are concerned only with q > 0, A_ will be a gaussian function around g'jW 2 with the same width as the ground state, whereas h + has to be kept as small as possible. Indeed, for large displacements the 2 coupling term h_/q will become negligible, and a pure h_ solution will be a good approximation. Furthermore, if the displacement is much larger than the zero-point fluctuation ~W~* only a small adjustment of A_ will be needed to meet the boundary condition at q = 0. Hence, in this limit the ground state will be approximately
9

with

f^--w

(17.42)

190

PION PHYSICS
solution of (17.40), which is predominantly A+, will have a f the two solutions will go over into

The other

Similarly, the energy of one of the be lowered by turning on //'. Indeed, a perturbation treatment of the centrifugal term shows that it will be above the ground
f-states will

much higher energy. For small g the S } and Pk level, respectively.

state

by

Fig. 17.1. Plot of the potential

V = \W*(q
\

JlL

"

- /Jl +
/

\- rr

lwj
*
/

for

h+ and

h_.

The curves have been continued into the unphysical region of negative values of q, where they are shown by dashed lines.

With increasing #' the energy approaches the ground state, as is depicted in Fig. 17.2. Of course, whether this excited state will be stable or will decay into the ground state with pion emission can be decided only by
taking into account the continuum

meson modes

(s

^ 0)

in (17.30).
is

a source with various energy levels coupled to the meson field. If the selfenergy shifts of all levels due to this coupling are the same, then the first excited state will be stable if its energy is less than /* above the
will

The problem

be similar to the Lee model, where there

ground

state.

The

excited f-state corresponds classically to the spin gyration

we

THE GROUND STATE


studied earlier.
F(k)

191

Indeed, since for strong coupling

we find

A ->

and so

&
3
/*

97T

we

find for the excitation energy estimated

above

W*

2-ir

in fairly close agreement with the classical result


9?r

from

(16.18),

4/Cmax

2 r2M~,,2 /4

H+6,

Fig. 17.2. Plot of the energy


effective

for the lowest states as a function of the 4- <? coupling strength g'. The states are labeled by their angular momenta and other appropriate quantum numbers.

pertinent features of the ground state, we first remark (17.41) implies a Poisson-like distribution of virtual mesons, as in the neutral scalar theory. However, q only creates mesons with total angular momentum and, therefore, a

To mention some
that the gaussian

form of

mixture of meson pairs.


then

Indeed,

if

192

PION PHYSICS

shows that h$(q) and h^q) always create a mixture of pairs with lz = 1 or with lz = 0, the former possibility having twice the amplitude of the
latter.

exist,

and

In the neutral pseudoscalar theory, the constant r 2 does not in the limit of very strong coupling, TI is determined by

(AT,

N = .^\N fq exp J [- w(q -j^


t)
|

i(tf|l
f

N)
.

(17.43)

For interHence, ^ in this theory and in the limit considered is mediate values of g the system of equations can be handled only by numerical methods, which we shall not discuss here. 1 Returning to the symmetric pseudoscalar theory, we can easily imagine that an analytic solution is not feasible. Because of the extra degrees of freedom brought in by the isospin, we obtain, instead of /z 3 in a (17.39), four simultaneous equations for four functions /z
,
.
. .

nine-dimensional space. However, for small coupling strengths / the perturbation result is obtained, whereas for large values of/the theory goes over into the strong-coupling approximation, which we shall
discuss next.
17.5. Strong-coupling Approximation.
2

In this limit

many mesons

surround the bare nucleon, and the zero-point fluctuations are much smaller than the mean value of the field, so that classical calculations begin to acquire meaning. In the simple treatment here, where our

main

interest

is

in the ground-state properties, all virtual

mesons are put


is

into the

same mode, and the


<7 aj ,

resulting reduced Hamiltonian (17.30)


If

diagonalized in the limit of large g.


ators/^. and
as in (17.35),

we introduce canonical

oper-

07-44)

See, however, references listed in footnote

1,

page 186.

Phys. Rev., 62:851 (1942). See also G. Wentzel, Helv. Phys. Acta, 13:269 (1940) and 14, 633 (1941); R. Serber and S. M. Dancoff, Phys. Rev., 62:85 (1942); F. Harlow and B. A. Jacobsohn, Phys. Rev., 93:333 (1954); A. Pais and R; Serber, Phys. Rev., 105:1636 (1959) and 113:955 (1959).
Pauli

W.

and

S.

M. Dancoff,

THE GROUND STATE


then the reduced Hamiltonian becomes

193

H"

= HJ + H" - <? =I
<xi

\ 2.

(Pi,

+ W%) + f'aM., - -W-# 2


/'

(17.45)

where

^foW)*

To find the eigenvalues of //, we follow the pattern of the last section and, for the moment, resort to a semiclassical treatment in which the operators q are considered to be ordinary numbers and the minimum of H(a,p = 0) is determined. We must then find the lowest eigenvalue of me 4 x 4 matrix T a o# aj To this end, we note that a rotation in both
.

1 spin and isospin space

t/o-.tr

<r...

B*

BB t

induces the transformation

UTm a,qm ,V- l


where
(in

= T a&
m

mj

(17.46)

shorthand notation)

6a
strate
it

= A^BrfK = AqB^

(17.47)

definite

T 2 symmetric matrix qq can be diagonalized by the onal matrix A TA- 1 T=


:

We claim that q can be diagonalized by this transformation and demonas follows. We observe, first of all, that the real positive
real orthog-

QQ

Aqq

(17.48)

can, therefore, extract the square root to obtain the real diagonal matrix Q with elements Q^ = 2<A^. The elements g a are the analogue of the radial variable q in the neutral theory and will, therefore, be assumed positive. The matrix B can be written by means of (17.47) as

We

B
and can

= Q~

Aq

(17.49)

T readily be shown to be orthogonal (Q -1 2 =1 BB T =

Q):
(17.50)

Q^Q Q

The eigenvalues of
matrix,

//'

which are proportional to those of the 4 x 4

(17.51)
1

The
T.

and
2

properties of the rotation matrices are determined by the hermiticity of sum over subscripts appearing twice is to be understood.

transpose matrix of q.

194

PION PHYSICS

can be found by making use of the properties of G t


0f _
!

Hence, the operators eigenvalues of 0< are

__! [0.^] O can be simultaneously


t

0.0.0^ =

= =

oyr t , namely,

diagonalized.

The

0^0* must be
are

1.

of eigenvalues of Therefore, the four possible eigenvalues of (17.51)


1,

and the product of a

set

-61

-62-63 - 62 + 63 61

Because of our assumption that


to the

+ 62 ~ 63 -fii + 62 + 63 Q > 0, the first of these


61
t

corresponds

ground

state.

To find

i9

the minimum energy of the ground state and the best form of we minimize the particular form of the potential energy of H in the
state,

ground

(17.53)

=2"ef-/'ft 2
i

with respect to

The minimum
e,

is

reached at
(17.54)

=
2

and

is

ft

=-+-ffZ

W*

3^- = 2

ft2

-*12 W
2

r/2

(17.55)

The ground

state

of

variables Qt centered about their equilibrium values/'/

H will contain a gaussian function of the radial W


2
,

e.g.,

However, to get a form like (17.41), we still need a projection operator into the eigenstate of q^Tjjt with eigenvalue Qisimple form

for this operator can be obtained only at the equilibrium position and will be useful, inasmuch as the widths of the gaussian functions are
negligible

compared with

their displacement.

For
t

Q
2

=/'/ W* we have
;

=
where the e
satisfy

A?B, -

(17.56)

THE GROUND STATE


Since at the equilibrium

195

fo.*W = ^-|i*.^
we
recognize that

y-\(
is

-v*i T* a

ij)
Q
i

( 17 - 57 >

the desired projection operator.

In fact, for

2 =/'/PF we have

*=*
Thus, in
this limit the

.^* = -3<p
state
l

ground

is

approximately
r

N)

= JT(\ -w.o,W*f'-

)exp

L -^2 2
t

i
(fi<
\

r\ - ~T i
2

rK

(17.58)

<2 Z and ty are invariant under rotations in spin and isospin the ground state (17.58) is an eigenstate of J and T with the same space, 1 t eigenvalue as JV). The term proportional to q^rjs^ is the / component of the meson cloud, and the other term corresponds to / 7 0. There are no mixed terms, and by means of (17.8) and

Since

= =

(17.12)

we obtain

CW>=i
r tl

=
r* == r2

Cl0

=
(17 59)
'

=o

* J

That r t = means that the expectation values for the ground a and T vanish. This can be verified directly, since

state

of

(N ra N)
|

ra \ (N 4
|

+ | T,T.T,
ft

JV)

Since the ground-state energy

is

zero,

we obtain
'

^o=-|^ +

(17-60)

where E' is the sum of the zero-point energy, \W, and the kineticterm *,/&, which is of the order of f W. For large values of energy /'we can neglect E'in the ground state and find, by means of (17.31) and
(17.33),

dk

W~

J_

LJo

dk
/0
f

kP
(ft> -f-

k
A)

*"

n*
(17.61)

196 with

PION PHYSICS

L.ffl-pW co(o) Jo
we minimize

-f A)

(17.62)

To

find the best value of A,

<^

with respect to

it.

Making

use of

we

find
a<r

^
<f
is

/
2

LJ8
1

The minimum value of

therefore obtained for A

and

is

(17 63)
.

Hence, in the strong-coupling

limit, the self-energy

1'2

is

one-third of

that in perturbation theory. As for the neutral pseudoscalar theory, the first excited state will be one with higher spin and isospin, whereas the states corresponding to

the other eigenvalues of o^^q^ found to be a f , }-level (/ T

lie

much
and
9

AE =
above the ground
state.

i)

-r

is

The next level higher. at the energy

is

(17.64)

If this state is unstable,

it

will

produce a

resonance in the f ,f-scattering. We have already studied this in the classical approximation, and we shall study its quantum-mechanical aspects in the next chapter. The results we have obtained here are not 1, since all mesons were assumed to be in only one exact, even iff mode and it was not shown that the other modes can be neglected. 17.6. Numerical Methods. The ground-state problem has also been attacked by more elaborate variational methods, which require considerable numerical work. 3 From these investigations we have learned the following. For small values of/ 2 and fcmax the constants tj. and r2

>

vary rather rapidly from their weak-coupling limits 1, 1 to the strongcoupling limits 0, J. Once the latter are attained, the situation remains
1

From

(17.16)

we have

See Pauli and Dancoff, op. cit. G. Eder, Nuovo cimento, 18:430 (1960); F. R. Halpern, Phys. Rev., 107:1145 (1957); F. R. Halpern, L. Sartori, K. Nishimura, and R. Spitzer, Ann. Phys., 7:154
3

(1959).

THE GROUND STATE

197

rather insensitive to/ 2 and Armax Empirically, we shall see that the best values are 1 f2 !^ The variationai calculations 0.2 and A:max 5/j. indicate that with these values we have not quite reached the strongcoupling limit, and actually tj and r 2 are experimentally determined to be | and |. Thus, it is quite conceivable that these values correspond to the exact result of the theory. However, such an agreement would not be very significant, since we are still in the region where numerical values depend sensitively on the (unrenormalized) coupling constant and the cutoff. Summarizing, we can say that it is hard to calculate and fc max with properties of the ground state for realistic values of But within sufficient accuracy to assign confident limits to the result.

the crude physical limits of validity the results of the model resemble experimental findings for the ground state.
1

This

is

not to be confused with the renormalized coupling constant

f$l4ir

r 2f/4,r~0.1.

CHAPTER

18

Pion Scattering

18.1. Introduction.

In this chapter

we

turn our attention to the

1 (/ = P-wave mesons. Hence only the L or f) phase shifts will 1 differ from zero. it is known that there are also Experimentally nonzero S- wave phases, as well as D waves at higher energies, but the dominant contribution in the energy region in which the static model makes sense arises from P waves. Part of the other phase shifts can be

In the static limit, which we are scattering of pions by nucleons. using, the Hamiltonian can absorb and emit only angular-momentum

ascribed to recoil (kinematical) corrections, but this is not the complete Of the P waves, only the J | phase shift is important, |, T story. since it has a resonance in a region where the static limit may still be sensible. This resonance is not unexpected on the basis of the classical and strong-coupling models we have discussed earlier. We shall see that it can be predicted quite naturally on the basis of the Low equa2 This method has tions, which we shall use to describe the scattering. the advantage that the detailed form of the mesonic wave function is not needed. Although a complete expression for the S matrix is not

obtained, the main features of pion scattering can be deduced. More 3 important is the fact that the method to be described is the only one which is not based on uncontrolled mathematical approximations.
1

See, e.g.,

H. A. Bethe and F. de Hoffmann, "Mesons and Fields,"

3B,
2

Row, Peterson

& Company, Evanstan,

vol. II, chap.

111.,

1955.

F. E. Low, Phys. Rev., 93:1392 (1955). Relativistic dispersion relations can be shown to reduce to the equations to be discussed, in the limit of infinite nucleon mass and neglect of nucleon-antinucleon8

pair creation.

198

PION SCATTERING
1

199

18.2. The Scattering Matrix. The derivation of the Low equation analogous to that developed in Chap. 14 for the Lee model. Additional complications are introduced here by the presence of the 36 + TT), at a independent states of a nucleon and a P-wave pion, in, fixed energy. These states can be written in our standard way as
is
|

in,

N+

7Tk >

= Al(k)

N)

EE XjJ

N)

(18.1)

For

this

development the expansion of the field in

angular-momentum

variables

avoid crowding of subscripts, we shall often make use of the notation introduced in (18.1); that is, we shall let the subscript denote the nine possible angular-momentum and isospin indices of the meson, as well as the (semi-) continuous variable k. Introducing a single subscript f to distinguish the four possible nucleon states, we can write (18.1) as
jOL

(k)

is

most appropriate.

To

|in,lrf>

Xt|f>

(18.2)

generalization of this description for a state with n real mesons, which will be needed shortly, is
|in,

The

W ,f>^|in,k 1 ,k 2 ,...,kw ,|>

^t ...^Jf
2

The scattering matrix relates the out and in introduced above, this 36 x 36 matrix is
SkT,k|

states.

In the notation

(out,

k'f
|

in,

k|>

'

<f

B K .A}c

f>

(18.3)

To

obtain the relation between the in and out meson operators, we use (15.10) rewritten in terms of angular-momentum operators:

lew - -Ko = x
Similarly, in terms of the outgoing

meson operator

B,

we have

ko-Bkoso that finally (at


t

= 0) = *k(0) i

-4(0)

f
J -oo

dte' imt VK (t)

(18.4)

where the source term has been abbreviated by

The operator V is
1

hermitian,

VK (f) =
we
shall henceforth set the

To

increase the legibility of the formulas to follow,

meson mass ^ equal to unity. This means that all energies are measured meson masses and all lengths in Compton wavelengths of the pion.

in units of

200

PION PHYSICS
"
f
|

Substitution of (18.4) into (18.3) gives


SfcT.kf

<*'

BK>*K

~
|

'(out,

k'f

-0

- d w <W - 2irid(co - a> ')<out, K'f = <W ^^ - 27rtf(co - co')TKTfK


To
obtain the final form of (18.6),

(18.6)

we made

use of the. explicit time

dependence of

VK (f)

in the Heisenberg representation [see (2.18)]

and

integrated over time.

For inelastic processes, the


conservation d function:
Sn.Ks

S and T matrices
in,

differ

only by the energy

=
X

<

out

>

ki> k2
.
.

'
-

*;. f
.

>

>

=
|

-27ri(5(ix
t=i
>

<)
(18.7)

(out, Ki.K^,

,K' n

f MO)
|

= ~2rrid(E n - co)rBiK|
on

By means of the field


shell (to

equations, (18.6) can be rewritten

the energy

w') as
(out,

f
/o

this equation that some simplifications become note that the momentum dependence of the T matrix apparent. can be factored out, leaving a matrix of spin and isospin operators between physical nucleons. This factorization is a special property of
It is in

the last

form of

We

the static model, which we shall call on in subsequent sections. To carry the discussion further, we rewrite the T matrix for elastic
scattering by introducing a complete set of intermediate states as we did in the Lee model. If we then use
<*'
I |

out, n) 9

P*'<0

out, n)

we obtain 1

- e-^T.Vr

(18.9)

Note
1

that in the second term of this equation the crossed but the nucleon ones are not.

meson

indices are

The jsum over n includes a integration in the continuum.

sum over
That
is,

isospin

and spin

states as well as

an energy

J.T,T Z

PION SCATTERING
18.3. Properties of the Scattering Matrices.

201

The form of the nondraw several important conclusions. a. The condition S*S = 1, which always holds, can be Unitarity. demonstrated when one physical meson is present, as follows explicitly
linear equation (18.10) allows us to
:

(in,

K'' S S
| |

in,

Kf)

=S

(in,

f K'f S
\

n)(n

S
\
\

in,

- 2i S T* Kr TniKS
n

5(E B

)]

= dK K 6 n
.

(18.11)

The vanishing of

the bracket in (18.11) can be shown by substituting into the difference TK ^ >K^. The second term of (18.10) (18.10) fKT cancels out, and the first term gives a sum

T^
T
l

n K'? A n,K$
t

-co-i

T*
n

*n,K'?< n,K

-eo-f

J"
\

In a similar manner we can demonstrate that 55 f = 1. The nonlinear character of the first term of (18.10) is thus connected with the unitarity of S. Although the second term does not contribute directly to the unitarity condition, it is vital to ensure the crossing symmetry discussed
next.
b. Crossing Symmetry. The crossing symmetry of the T matrix is most conveniently expressed by considering separately the dependence of T on the variable a) in the denominator, in contradistinction to the 2 l dependence on k p(k)co~ which is explicitly known. To this end, we
9

define

as
(18.13)

which depends on the complex variable z and has the property that
(18.14)

In this physical region the


(18.14)] in the
It is

ie which appears [e.g., after substitution of denominator of the second term in (18.13) plays no role.

needed, however, for the hermiticity condition

and

for the crossing

symmetry of the

matrix, which

is

formulated

mathematically as
This theorem of analytic continuation is directly verifiable by subIt does not depend explicitly on the form of VK stitution into (18.13). and can be shown to be valid for any meson-nucleon coupling with both

202

PION PHYSICS
(i.e.,

1 absorption and emission

not for the Lee model but for the

neutral scalar theory, etc.). The symmetry is related to the fact that the theory is invariant under interchange of ingoing and outgoing

by (18.156). It has no real inbetween a physical scattering amplitude and one in an unphysical negative-energy region. The theorems (18.15) are rather concerned with the analytic properties o the / matrix

mesons

in the sense expressed explicitly

tuitive basis, since the

connection

is

for real or complex energies.

HI

Fig. 18.1. Singularities of the matrix

in the

complex z plane.

The analytical properties and are determined by the energy spectrum of the intermediate states n. The sum over these states from the ground state and a conconsists of one contribution at En = tinuum from 1 to oo. Thus t has the spectral form
c.

of the

Poles, Branch Points, and Branch Cuts. t matrix are explicitly given by (18.13)

where the

first

term

arises

weighting functions.

from the ground state and F and G are These can be found from (18.13) and (18.15),

Hence

has a pole at the origin with residue


.
| |

R given
VK
|

by
I

Rrx-tK = S (' VK

>

\VK

\$-

'
|

>

VK

.
|

(18.18)

where the summation over

{ is to

be taken over the four physical

ground

states.

The matrix

also has cuts

from

to oo

and

to

oo,

as indicated in Fig. 18.1. The contribution of the negative-axis cut arises from the crossing symmetry and was thus absent in the Lee model.
1

M. Gell-Mann and M.

L. Goldberger, Phys. Rev., 96:1433 (1954).

PION SCATTERING
18.4.

203

As in the Elastic Scattering. to define consistently a renormalized coupling possible constant/,, such that the physically not accessible limit of t^ K t K as In z -> is given by the Born approximation, except that/r replaces/.
Low- and High-energy Limits of
it is

Lee model,

term R/z dominates

the limit given above, the singular Reall others.


the definition of

Xv

v^

membering by (18.5), we

VK given

Xx v

^^
xx

^"'
><^x

see that only matrix elements of ar between physical nucIcon states are involved. These have

^''
^L
*
(

\^

already been
section,

studied
find

in

the

last

aj

and we

- (ar)K (ar) K
where

'
. \

{)

(18.19)
Fig. 18.2.

(J)

is

implied by the matrix


(a-r)
is

Graphs for the lowest-order

multiplication of (or)*, and Aside from the factor r|, this

Frturbation calculation of-the matrix

the

perturbation-theory result. The matrix element for the latter can be computed from the two (Feynman) diagrams of Fig. 18.2, and we find

R?K',IK
0)

_ H'(K')H'(K) ft)

H'(K)H'(K')

which corresponds to (18.19) with

\\

1.

Hence
(18 20)
'

/'IH*>=/!tf'|H0
Exactly as in the Lee model, the interpretation of

that, in the idealized limit, the time between emission and absorption of the external meson takes much longer than that of all virtual pions. The intermediate nucleon is therefore a real one for all essential purposes, and we obtain the Born approximation, except for a reduction This factor has the same probability interpretation as in the factor r 2 Lee model. The constant/, represents the interaction strength of the components of the physical nucleon weighted with the probability with which they occur.
is
.

204

PION PHYSICS
it is

In the high-energy limit,

possible to neglect

En relative to z in the
VK
,
I

denominator of

(18.13).

Thus

= - S ('
Z n

VK>

out, n>(out, n

$)

'

<f

VK

out, aXout, n

Fx

f
(18.21)

,Fx ]|f)

This is of the form of t in the Born approximation, except that the matrix element is taken with the physical nucleon states rather than with the bare states. That is to say, in the limit considered, the scattering amplitude is given by Born-approximation scattering from the various bare nucleon states multiplied by the amplitudes with which they occur in the physical nucleon. These can be easily found by observing that, aside from a known momentum dependence, the commutator in (18.21)
is

proportional to
|>a cr,.,Ta ,(r r ]

= 2i(dM
We

jj k

ak

-f <^'faa y r y )
'

(18.22)

where

(-1)

according to whether a, ft, y is an even ( 1) or an odd a/?y therefore find, with the aid of (17.3), permutation of 1, 2, 3.
is
1

that in the limit of infinite energy the

matrix

is

tj.

times the Born

approximation:
'
I

[(<")*-,

(")*]

*)

rj

lim zfuorn
Z-*-oo

(18.23)

Whereas the zero-energy theorem is perhaps the most important instrument for linking theory and experiment, as we shall see, the highenergy limit is of purely academic interest, since it is outside the realm of validity of the model. Clearly in this limit, the neglect of recoil and pair creation (and the use of a finite source size) cannot make sense. 18.5. Diagonalization of the T Matrix. To use the formalism developed at the end of Chap. 8, we shall now diagonalize the 36 x 36 matrix t. Because the interaction conserves angular momentum and isospin out = / = J in r out = T = T in), we expect that this can be done (e.g., / by transforming from the one-meson states (K) to a representation
,
|

are diagonal. 1 Indeed, since these variables the energy determine the one-meson states completely, we have plus
in

which

T2 T J 2 Jz
,

z,

(out,

r, r;

/',

in, r,

z,

/, /,>

(in,

r,r;,

jv; \s\m,

T,

TV

J,J Z )

We hope that the reader will not confuse the matrix T and isospin T.

PION SCATTERING

205

To

relate the
8.

phase

shift 6 JT to the

T matrix, we

refer to the
is

end of
simply

Chap.

Since here

5 =
r*

f
I

Jo

dk, the factor irg(E) in (8.30)

Furthermore, we can project the T matrix into the corresponding k/co. jr eigenstates of / and T by means of projection operators ^p ,1f

(18.25)

J,T

TTO)

with

(18.26)

and

(18.27)

In the one-meson subspace, T and / assume only the values J and f Since 6 JT is d TJ we have only three different phase shifts \j, d^ d^ 9
.

and d^.

If

S assumes

the form

we introduce the labels 1, shown in Fig. 18.3.


8
12

2, 3 for these

phase
32

shifts,

then

The
24.

projection operators into


28
36

,14

16

20

Fig. 18.3.

Form of the S matrix. The off-diagonal elements are diagonal ones are as given in the figure.
If

all zero,

and the

The

Compare (14.40), which differs from (18.25) elastic-scattering cross section is


a

by anormalization
|sin 6JT

fact or

of 4nk*.

t x

3 2 87r co

Ar4

|71

=J

WT

^r|

206

PION PHYSICS

these subspaces are constructed in the standard

and J

way with T
)l

T/2

+a/2:1f

- T )(J 2

J)

+ (T
.

-a - JX
a

-a-0]
tX2

(18.28)

The normalization
subspace

factors chosen are such that for the

one-meson
(18.29)

<P<*><P<">

= d uv y
t
1

(v}

where
/

w, v

1, 2,

The matrix elements of


correspond to the
(15.13)
/

and

in the

=
/
1

one-meson subspace simply

representation and are [compare (15.12),


in)

and

(5.13)]
(in,

in,
|

/a',

e =
>

w^ ^
r
|

ln)

(in,^,a7|/i

|in,7V,f)==/^,,,^ r

^
\

'

so that

we

have,

e.g.,
(ln)
(in)

(in, f, ay

the matrix Returning to elements of the projection operators (18.28) in an angular-momentum

M.(f *&. a,*, our shorthand labeling, we can write for


/
|

in, a'/, *')

f ')

(18.31)

representation

We
shall

2(ar) K ,(ar) K

f)
\

(18.32)

**K-*K

+ WrWK
'

shortly encounter expressions ^P^^ A" where the meson subscripts are exchanged but the nucleon subscripts f and f remain in their usual order. Since this operation just changes the sign of / and t, these operators can be expressed as linear combinations of the old ones:

We

find

by direct substitution that


/

A
,

is

-a
7

= U -8
\
16
t

(18.33)

U Wheffeas
value at time
1

T
/

is

constant in time, both


<x>

and T are time-dependent.

It is their

that

is

needed here.

The

labels /

and a

refer to the angular

momentum and

isospin of the

meson,

respectively.

PION SCATTERING

207

The properties of A, namely, A = 1, Det^ == \A\ = show that A has the eigenvalues +1, +1, 1. With this development we can immediately obtain
2

1,

Tr

A = +1,

behavior of the
(18. 19) gives

r matrix.

the zero-energy Thus, substitution of (18.29) and (18.32) into

-P'*'*

<i)

-A

A (i) (v)

V'K'K)

()

The quantity X -1:

(A

(1)
,

A (2) , A <3) )

is

an eigenvector of A with eigenvalue

X/1- -X X - (-8,-2,

To

obtain quantities with simple analytic properties, we introduce


(18.35)

for which

we

obtain,

from

(18.16)

and

(18.17),
1

2
v

=2
v

1
I

^>A<>
0)

+1
7T

Jl

f^r*" L ('

"*"^') CO lc

or

>

ft

(z )

.
Z

+
7T/1

ft)

+2x
u

CO

+ ZJ

(18.36)

Hence the functions A (v) (z) can be continued into the complex plane, where they have the properties
h \z)
(v

(t;)

fc

*(z*)

=2A
u

(u)(v)

(u

\-z)

(18.37)

They

are related to the phase shifts by


lim h \z)
(v

- ^>> sin dv(a>) -^13

(18.38)

as can be seen by comparing (18.35) with (18.25). related to the total cross section by

The function h

is

(18 39)
-

208
if

PION PHYSICS
define the total cross section for a
127T -- Sln
.

we

P-wave channel by

This definition
target

is

such that the cross section (8.34) for an unpolarized

and

definite isospin is

Im
co

It

should be noted that, for

>

2, particles

can be produced and that

the phase shifts describing the one-particle channel become complex. Nevertheless, (18.39) can still be used, since it depends only on the general equation (8.28). With these expressions we can cast the Low

equation into the final form


h
(D
(o> )

=_ S+1
co
TT

dco'
2

I"

a^co')
co
i

Ji k' p (k')L(o'

- A ^av (a>')l v + co J
(

0)'

f *(> = _ 3fi + i Jl -^[_*1^ + 2 4afe2] LcO ^ + J


CO
77
>

(18 .40)

/0^(/C

CO

CO

CO

18.6. Relation

of
all

Low

Equations

to

Experiment.

In

the

Low

equations (18.40), directly observable

quantities except the coupling constant/? are at least in principle. The first terms on the right-

hand

side are the Born-approximation ones with the renormalized coupling constants and can be found only by extrapolations to the = 0. These terms are negative in the 1- and unphysical point co 2-states and positive in the 3-state, corresponding to a repulsion in the former and an attraction in the latter. This feature can be understood Since Fig. in terms of the lowest-order Feynman graphs, Fig. 18.2. 18.2fe has an intermediate nucleon with T = J, / |, it only contributes

(8) is due to scattering in the 1 -state. Therefore, cr solely to Fig. 18.2#, and the reason for the attraction is the same as for mr~ scattering in the Lee model. In terms of perturbation theory, it arises because the

intermediate state has higher energy than the initial one. On the other hand, the scattering in the 1 -state arises mainly because of Fig. 18.26, and in the 2-state the opposite sign of the p- and w-coupling constants to the 77 changes the sign of the contribution from Fig. 18.20. In the total scattering amplitudes the integrals in (18.40) give an

PION SCATTERING

209

increase over the Born-approximation term in the 3-state and a decrease in the other states, because <r3 is by far the largest of the cross sections.

same as in the scattering by a short-range potential, where Born-approximation amplitude overestimates the effect of a repulsive potential and underestimates that of an attractive one.* (l) = (3) whereas empiri2/z According to the Born approximation, h (1) (3) so that at physical energies the correction terms < |/* |, cally |/z must be substantial. We shall calculate this effect, as a first approximation, by expanding the real parts of the integral in (18.40) in powers of a). Keeping only the lowest term in the expansion, we obtain an
This
the
is

the

2 "effective-range" approximation to the phase shifts.

Thus,

Re h

=
(O

(1

o>r u )

(I8.41a)

ru U

= _1
7rA
(M)

f"

*'
[, L
(ft/)

2 r

+ A ^av(a>'y] J
l

Ji a>'k'p\k')

and

i Re4i = M CO h
of k*p\k) cot
(6 M /A

cotd u

^^
1277(0

~
/
2

**

A <M>

?'

(18.416)

a function of o> should approach a energies. intercept of such a "Chew-Low" = is 47r//r2 and should be the same for all plot extrapolated to co phase shifts. Experimentally, the 1- and 2-phase are unfortunately too small to be measured to sufficient accuracy. In Fig. 18.4 we plot the middle part of (18.416), with p(k) = 1 and the experimental 3-phase, 3 The experimentally determined points are seen to as a function of <w. lie on a reasonably straight line which intercepts the ordinate at
3o>) as

A plot

(M)

straight line at

low

The

LL
4?r

= 0.087

0.0 1

(18.42)

1 With this analogy in mind, an estimate of the departure from the zero-energy form (renormalized Born approximation) has been given by V. F. Weisskopf, Phys.

wave function

Rev., 116:1615 (1959), under the assumption that the logarithmic derivative of the Accord<1 0> at the source radius depends weakly on the energy. <f>
| \

ing to this procedure the 3-phase shift goes through 90 at about the right energy, whereas the other phase shifts remain small. 2 See, e.g., J. M. Blatt and V. F. Weisskopf, "Theoretical Nuclear Physics/' p. 62, 3 John Wiley Sons, Inc., New York, 1952. Since the Im //<"> oc A: it can be

&

neglected at low energies. 3 See S. W. Barnes, B. Rose, G. Giacomelli,

J. King, K. Miyake, and K. Kinsey, Phys. Rev., 117:226 (1960), from which Fig. 18.4 has been taken. The experimental See also points arise from many sources, which are listed in the above reference. S. J. Lindenbaum, Ann. Rev. Nuclear Sci., 7:317 (1957).

210

PION PHYSICS

According to (18.40) the slope of the straight line is the meson mass times the effective range r3 and can be expressed in terms of integrals over the total cross section. This actually holds within the expected accuracy, as will be shown in the next section. Furthermore, the phase
shift passes through 90 o> r at w That this also 2.1. l/r3 agrees with experiment can be seen by a comparison with Fig. 15.1, in which the total cross section for TT+ p scattering is .seen to pass
*

through a resonance at a laboratory kinetic energy of corresponding to a center of mass co & 2.1.
3

~190 Mev,

7b cot63

0.5

1.0

1.5

2.0

Chew-Low plot of (18.416). The intercept of the straight line with the The curve is taken from S. W. ordinate is 3/[4(/*/47r)] and with the abscissa is r 3 Barnes, B. Rose, G. Giacomelli, J. King, K. Miyake, and K. Kinsey, Phys. Rev., 117:226 (1960). The sources for the experimental points that appear in the figure are listed in the above reference. The numbers given are the experimental energies.
Fig. 18.4.
.

18.7. Approximate Solution of the Low Equation. Having determined ft from experiment, we should like to find (w) without putting in further experimental data. This can be done, but only with
<5
tt

occasional appeal to known empirical facts. First of all, we know that within the region of validity of our model, the scattering is mainly elastic. This is rigorously so for eo 2, but we shall assume it to hold

<

for all energies.

Then

the phase shifts d u are real,

and we can

write

ff "

fc

i*"

|apW
1

-&T

(18.43)

PION SCATTERING
This approximation
is

211

tantamount to including only zero- and onethe intermediate states. 1

meson states in As was done

the

sum over

in the corresponding useful to introduce the inverse of h:

problem of the Lee model,

it is

Zu **

zh \z)

r2;() t\ (u

' V (18.44)

finite at

Since h (u) (z) has no zeros for complex z (there Im h 0) and zh(z) is z 0, g is analytic save for cuts along the real axis from 1 to oo

>

and

to
1

oo.

The

gu (Q)

and hence,

factors in (18.44) have been chosen so that in analogy with (18.16), g can be written 3

(MMSa)

The

5 M and 05 M correspond to the imaginary of g on the positive and negative real axes. We readily obtain parts from (18.450), if we make use of (18.37), e.g., gu (z*) = g*(z),
real weighting functions

gu (co
.

-}-

ie)

- g u (co (
>

ie)

-Kt

o>

')

'*)

= 2i Im gu (a>) = -2/c = 2/Img u ( o>) = 2/G><5

tt

(18.450)
tt

(G>)

Within the one-meson approximation, we furthermore


(18.38),

find,

from

Im ^( W) and thus obtain


the other hand,
1

= - 3f

k o (k\ A f ZRW *4?r


co

3 2

(M)

(18.46)

8f M (w)

we need

without knowing <5 (co). To determine 05 M on the phase shifts in an unphysical region. This
tt

This procedure should not be confused with the Tamm-Dancoff approximation,

where the expansion of the physical nucleon is in terms of the number of mesons around a bare nucleon, not a physical one. Here, the expansion of a physical nucleon and a meson is made in terms of physical states. In this way we obtain the
correct low-energy behavior. 2 There may be poles on the real axis between
1

and

Their significance

is

discussed by Castillejo, Dalitz, and Dyson. (See the first reference We shall not examine these singularities. 3 In general, the power of z in front of the integral need not be 1.

on page

144.)
if

However,

we see

that lim

gu =

constant.

Hence

the choice z 1 , which also leads to the correct

perturbation-theory result.

212

PION PHYSICS
(18.37),

can be obtained from the crossing symmetry


for

which becomes,

gw
}

g*(z)

(18.47)

2 = 1 and B for the matrix A (u)(v \ the eigenvalues of B uv are 1. This equation gives only the Img"^ z); to find Img( z), we also need the Reg~ 1 ( z), which is not obtainable directly. Thus, even in the one-meson approximation, it is not possible to find an exact solution. However, an approximate solution has been suggested by

As

Chew and Low, 1 who

replace

B by

#',

B&B'=
which
differs

-1

\=B--

-1

-1

(18.48)

from

B
B'
2

by a small numerical displacement and also

satisfies

=l
we

TrB'

\B'\

(18.49)

With

this

approximation, the problem

is

soluble for

g and #3 but not


,

yet for

2.

From

(18.47)

find

g2

-v ^ X-z)

= g -v \z)
^
l

g2 \z)

-if

\j_

ga \z)

-V

( 18

'

and

therefore (18.456) predicts

There are no

real

mathematical arguments to justify (18.48).

How-

ever, using empirical phase shifts, we can show that what has been omitted is at least not large compared with what was taken into account.

Hence,* within our rough model, the procedure

may

be a reasonable

1 G. F. Chew, Theory of Pion Scattering and Photoproduction, in Physik," Springer- Verlag, Berlin (to be published).

"Handbuch der

PION SCATTERING
illustration,

213
seri-

ously.

We find

but the following numbers should not be taken too

(18.520)

fr

oj

k'*p\k')t
co'
2

4
-co
co
/

8
co'

\
ft>/

4?r 3ir Ji

\co'

and therefore an expansion

in

powers of

gives (at

low energies)

(18.526)

477 7T

/l

C0

/3

1 (l) is familiar to us from scattering amplitudes for h For pair theory with repulsion or /?TT~ scattering in the Lee model. h (3) it is like that of pair theory with attraction or mr- scattering in the

The form of the

Lee model.

It predicts

a resonance in the f,f-state for sufficiently

strong coupling, as observed. To get an idea of the resonance energy, we calculate the effective ranges rx and r3 predicted by (18.52). The 4(r x -f f 3 ), which follows from range r 2 can then be found from r 2

the crossing

symmetry

(18.37) applied to (18.410).

In this

way we

obtain

(18.53)

A somewhat different approximation from that considered


replacing
(18.40).

above, of

by B\

We

consists in keeping only cr 3 in the Low equations readily find that in this case the value of r l is approxi-

In seen to arise in the range r 2 either approximation the magnitude and sign of r x and /*3 turn out to be approximately the same. If we assume that the effective-range approximation is still approximately valid at the resonance energy co r then we
rl
;

mately the instead of

same

as that given

above but that


is

r^

rx

and

r3

5^/4
.

the

main uncertainty

find

co r

l/r3

For a square

cutoff, e.g.,

for k
for k

< >

/c

max

fenax

Except for obvious changes due to our dealing with P-wave mesons.

214

PION PHYSICS
find

we

YM
.

(18.54)

ATTTT

The experimental value of a> r ^ 2 requires w max ^ 5, if we make use of (18.42). It is interesting to note that a> r is exactly at the energy of the
f, f -level in

the strong-coupling limit, which

is

given by (V7.64):

47T7T

Since in this approximation (17.59) tells us that/ r co r The width T of the level is [see (12.90)]
.

= //3,
-

we

"

find

^ 2(co - co) tan d = (co r r

o>)

~*

^-

(1

l.l

(18.56)

3 4rr

which

is

cutoff-independent

if

kr

is

not close to & max and which


is

is

of the

same form as that of the


which was discussed
in

classical calculation (16.32).

The shape of

the cross section close to the resonance energy

of the usual type,

Chap.

15.

To conclude this chapter, we shall point out the 18.8. Summary. most important features of pion-nucleon scattering. At low energies, where sin 6 oc /*&*/, the cross section is expected to be proportional to

*cc/^
and
this is

(18.57)

borne out experimentally, as shown in Fig. 15.1. In this the "effective coupling strength" (fjk) is weak. region The form of the cross section can be understood as follows. Consider a

box of volume

L3

which contains a meson of momentum k

in

addition to a nucleon at rest. Since the meson is not localized, its 1 probability of getting within a sphere or radius co" centered at the 3 3 nucleon is ~a>- L~ Now, it turns out that the nucleon can emit and
.

absorb mesons when they are within a distance co- 1 which is the size of the cloud rather than of the source, and since it cannot distinguish the incoming meson from one of its own, it may absorb the former instead of the latter. The nucleon, however, is only meson-active for a fraction of time A, in the sense that there is only a certain probability that the nucleon emits and absorbs mesons.
,

PION SCATTERING

215

If we represent the history of the nucleon graphically, as in Fig. 18.5, 1 then the lengths of the meson lines are t ^ these lines are distributed at random, and A is the ratio of their lengths to that of the nucleon line. In the scalar theories, we found that if there is pre-

dominantly one meson during the active period, then A & mean number of mesons ^ g 2 /^- For P-wave mesons, the interaction

probability is strongly energy-dependent, and as long as equal to the effective coupling strength to a physical nucleon
Pion v

A <

1,

it

is

Nucleon

^
Graph

\+^a)-i-*\
to represent a physical nucleon.

Fig. 18.5.

and can therefore

Now the absorption of the incoming meson violates energy conservation last only for a time t ^ AE = M~ Within this
I
1

1 time, the nucleon must recreate a meson with energy w, but not The event thus constitutes a scatternecessarily in the same direction.

Because the absorption and emission ing of the pion by the nucleon. occur only within the quantum-mechanical fluctuation allowed by the
interval

uncertainty relation, we cannot tell when they happen within the The amplitudes t, and the events may occur in opposite order. for the two types of processes (emission-absorption, absorption-

emission) interfere, since the intermediate state is not controlled by measurements, but the interference terms do not change the order of

magnitude of the cross section (except in the 3-state at higher energies, near the resonance). For our qualitative argument, the various events can be considered to be independent of one another, and the probabilities

true

quantum-mechanically). = probability) process is (77


^process

for the separate stages multiplied (although this is not quite Then the probability for the total

== ^ (meson to get within meson cloud of nucleon) x v\ (that this occurs while the nucleon is meson-active) x ?? [for nucleon to emit (or l absorb) a meson within time co~ ]

^process

The
1

cross section

is

equal to this over-all probability divided by the

the

This classical description is somewhat acausal, inasmuch as the nucleon absorbs meson only when it knows that it will emit it in time.

216
flux

PION PHYSICS

of incoming mesons and multiplied by the rate at which mesons The latter is equal to the velocity of the meson, fc/w, 1 Since the flux of ingoing divided by the diameter of the cloud, a*leave the cloud.
.

mesons

(e.g.,
,

the probability that they traverse unit area per unit time)
is

*
is CD

L3

u * the cross section

&

ITATT
7<*>

by the geometrical area of the exceeded, our considerations become then exceed unity. The meaningless, since the various probabilities exact formal development shows that this limit will be attained only in the 3-state, and at higher energies. To the extent that the 3-resonance

Of course,
2

the cross section

is

limited

cloud,

1/co

When

this limit

is

the direction of the incident meson momentum, then Jz = s z = J, and we found there that the angular distribution of the mesons in 2 This should be compared the J = f, J.z = 1 state is oc 1 +3 cos #. which is also peaked forward and with the classical result (16.33), backward but which is much flatter. The predictions of charge of the incoming and independence are found by determining what part For scattering on protons, scattered states (e.g., n-p) have / = T = f we thus obtain, from (16.35),
.

dominates the scattering, it is possible to make other simple statements concerning the cross section. The significance of the energy dependence near the resonance in terms of probabilities was discussed in Chap. The angular distribution of the mesons can be inferred from our con1 2. If we take the axis of quantization along siderations following (16.37).

<Vp-*r+n

*-*-*

**--**-*

= 9:2:1

(18.58)

The experimental
w+p
a

curves of Figs. 15.1 and 18,6

predictions are fairly well satisfied.

show that the above The maximum of the curve for

=f

that the resonance is indeed in scattering in Fig. 15.1 demonstrates state, since the cross section reaches its maximum allowable
.

value of 87T/& 2

The simple
phases
<5

static

and

u=

model we have expounded also predicts that the 6 2 will be equal, negative, and small:

tan

<5

w-

-k P

*t

- 0.05

but not accurately Experimentally, these phases are indeed small, determined.

PION SCATTERING

217

Of course the scattering contains many finer features. For example, from (18.28) and (18.32) it follows that the T matrix can be written in the form 1 a + bo k X k', where k and k' are the momenta of the incident and scattered mesons, respectively. This predicts that the
recoil

proton

may

be polarized along a direction perpendicular to the

scattering plane, since the a and b terms will add or subtract, depending on the direction of the spin relative to the scattering plane. In addition, as we stated earlier, the *S-wave phase shifts are not zero and become
35

30

5
20

--""
30 60 90
120

150

180

Center of mass angle (deg)

Fig. 18.6. Differential cross sections for elastic

and charge-exchange

scattering of

The experimental Anderson, W. C. Davidon, M. Glicksman, and U. The curves are proportional to 3 cos 2 # 4(1955).
189-Mev pions by hydrogen.
9:2:1, as predicted by (18.58).

points are taken from H. L. E. Kruse, Phys. Rev., 100:279 1 and are plotted in the ratio of

progressively more important as the energy approaches zero below the resonance energy. Above the resonance, recoil effects may be expected to complicate the scattering further, and the model that we have considered begins to make less and less sense. Nevertheless, as we have
1 This is also the most general form allowed by parity conservation, since a and b can be functions of k 2 and k k'.

218

PION PHYSICS

seen, the simple static model is able to account for many details of the pion-nucleon scattering over a reasonably wide range of energy.

Further Reading
F. E. Low, Phys. Rev., 97:1392 (1955). G. F. Chew and F. E. Low, Phys. Rev., 101:1570 (1956). G. C. Wick, Revs. Modern Phys., 27:339 (1955). G. F. Chew, Theory of Pion Scattering and Photoproduction, in "Handbuch der Physik," Springer- Verlag, Berlin (to be published). H. Lehmann, K. Symanzik, and W. Zimmermann, Nuovo cimento, 1:205
(1955).

Cini and S. Fubini, Nuovo cimento, 3:764 (1956). S. Fubini, Nuovo cimento, 3:1425 (1956).

M.

CHAPTER

19

Properties of the

Nucleon

19.1. Expectation Value of the Field. The development of the last chapter allows us to gain further insight into the properties of the nucleon. The expectation values of several observables, such as the number of mesons in the cloud, their charge distribution, and their contribution to the magnetic moment of the nucleon, can be expressed in terms of the renormalized coupling constant and the scattering cross

sections.

In this section we shall return to a plane-wave expansion, since most observables are more conveniently expressed by it. We consider first the quantity a (k,0 at t 0, which is given by (15.10) or

The

interaction

Ka(k,f)
state

is

For the ground

we have

the counterpart of that defined by (18.5). >4 a (k) ) 0, and by making use of

the Heisenberg time dependence of affyrjtf),

we

find

<F

".00

f>

=~

In performing the integration in (19.2), we made use of the fact that the adiabatic switching on of the field implies that
f
dt
J -co

f
*<**= Urn
et>0
-*0
\

dte (i(0+ a)*

J-oo

219

220

PION PHYSICS
field

The expectation value of the meson

becomes

87T

(19.3)
,,-lr-r'l

Hence, we get the result of perturbation theory, except that/is replaced by/r Although is not directly measurable, this result is instructive, since it is the quantity (19.3) which is connected with the large-distance of the nuclear forces (see Chap. 21) and (e.g., one-meson-exchange) part also with zero-energy rrN scattering. In order 19.2. Ground-state Expectation Value of Observables. to study observables quadratic in the <'s, we use a complete set of to derive eigenstates out, n) of
. <
|

H
o

f J-

dt dt'

f*P (k)p(k')
/

~ (?
}

rg ,q

k|

ou, nXout, out, nou, n rga


)

) (E n + a>X n + w/ We have again made use of the Heisenberg-operator time dependence in

obtaining (19.4). When the intermediate state out, n) is the ground state, the matrix For elements in (19.4) can be expressed immediately in terms of r
|

other intermediate states the expression on the right-hand side can be This can be done with (18.11) and related to the total cross section.
(18.39),

-2 Im T. K = S

T*T
'
|

= 27T $' tf
'

VK

.
|

out, nXout,

n\VK

\S)

%f

- <o)
"

O7T CO

^2

where the sum S'


order to use (19.5),

is

we have

In ver a ll states excluding the ground state. to transform the projection operators into

our plane-wave representation.

This
'
I

is,

in a

more
I

explicit notation,

3W,i k =
where

wOVr-'m'^w
y

k)

(m k)
|

=
(l)f

PROPERTIES OF THE NUCLEON

221

This gives us

k'

-o

k' a k)

Equation (19.5) can be written in the form

S
n

{' r^a
|

n'
|

out, w)(out, n

rjs

n
|

f)

<5(co w

where n and

n'

are unit vectors in the directions of k


7 (0)

ing (19.56) with

+
2

OJ ) (0) q

--,

+
C0

and

k'.

Multiply-

and integrating over

q)

00

<*>_, q

we

get

(19.6)

As a

first

application of (19.6),

mesons around the nucleon.

With

we calculate the mean number of virtual (19.6) we find

f_^_
3

/>

(/c)/c

r3/ r
Leo
2
2

J(27r)

2co

" J_ f rf^ 2 37rJi ^p (^)


1

c^K)
^co q

4a2 (co q )
(w a

+
2 2

4cr3 (o> tf )1

=
No

3
77

r^^/c/cVwr/
Jo
OJ

f
2

00

gl (a> a )

L47TW

367T Jl

^p

(^)

J + co) + 4<r K) + 4(T (^ )1 J (a), + O))


3
q
2

(19.7)

summation over f > is implied in (19.7), and (N) is independent of The first term is the renormalized Born approximation, 1 f >. which differs from our familiar expression ~f2 (d*k/w 3) of Part two by the factor fc 2 which arises because we are dealing with P-wave mesons. Since our method always refers to the physical particle, / appears r rather than/ The integral J da> q represents contributions of higher
|

the state

order in/ 2
1

We shall abbreviate

this

by

RBA

henceforth.

222

PION PHYSICS

In exactly the same

way

as above,

we obtain

3 r
7T

r fi
L47Tft>

if*
2

Jo

36?T Jl

rr,(rn \ -4- dfr^ffn ^ -1- dre~(fi)

n
(19.8)

(a)

+ a)

Q)'

^-^J.

and, with appropriate changes,


f ')

=2f
at

=S1
6 f*
77

Ut
(t)

'

" )(OUt "


'

-\-

En
f
2

dfc feV^fc)!"
co

/;
|

(to,
(o) -{

Jo

L47rce)

367r Ji

M Q)
(19.9)

(19.8) and (19,9), we see that the Born-approximation term of <// > is half that of -<//') but that the double integral in Q is less '. than half that of Hence, we have the "virial theorem"

Comparing

<#o>

-K#'>

and
//'

W<

-<Ho>

(19.10)

which means that the interaction


lighter. 19.3. Renormalization Constants

makes the nucleon dynamically

and Other Parameters of the Static

Model.

To

learn
r l5 r2 ,

more about

<f

we must

parameters

2
,

w max which
o> w
:

fi rst digress to discuss the are relevant to the static-model

description of the nucleon.

Some of the relations we need are obtained

by integrating

(19.56) over

The expectation value can be worked out with (17.5), (17.3), and For example,

(18.32).

KK

(19.12)

PROPERTIES OF THE NUCLEON

223

To
.

cast the integral in (19.11) into a comparable form,


/

we

use (19.56):
IK'K)
(19.13)

~~
'

IK'K "H T

ij- 2^ 2

^ 3)

I)

Collecting these results and equating the coefficients of

1,

and a

t,

we

get

+ +

4o-2

er,

+ 4(r3 - 2(T3 +
<r

1-rJ

(19.14a)
(19.146)
(19.14c)

=
"

U
[r2

i\

-2<72

~rl

In principle, these relations can be used to calculate the unrenormalized coupling constant and the renormalization constants from observable In practice, the difficulty arises that in (19.14) there must quantities.
the

be significant contributions to the integrals from high energies, where model would not be expected to describe physical reality. For instance, taking the combinations (a + 2b 3c) of (19.14), we find
dco n
1

f
(19.15)

4J~k
since
<r

we know that (1

2tj
cr

3r 2 )
3

>

from

(17.13).

Experimentally,
J
cr

2 is

only a few per cent of

up

to 300

Mev, so that

da) n /[k n p (k n )]

must

This difficulty does get large contributions from higher energies. not occur in the Low equations (of Chap. 18) for pion-nucleon scattering or in the expression for (N), where the high-energy contributions to the integral are suppressed by one or more extra powers of co in the denominator. Nevertheless, if we insert the low-energy-scattering data into the
relations

we obtained, or shall obtain, 1 and leave the high-energy behavior open, we can get a fairly good idea of what the various parammust be
if

eters

We

shall not

the theory is to agree with experiments at low energies. 2 go into this analysis but shall only quote that for

= 0.08 =
0.22

and

47T

f2

r2

0.37

(19.16)

47T

= 0.58
the meson's charge contribution to the nucleon,

The data below make use of


shall discuss shortly.

which we
2

See

S.

Fubini and

W.

Thirring, Phys. Rev., 105:1382 (1957).

224

PION PHYSICS

the various relations

and and

r2
p.

and inequalities can be satisfied. Of course, rx are not independent and, in principle, could be calculated from/ The discussion given in Chap. 17 shows that it is not incon-

r x and r 2 have the given numerical value, if p and/ are chosen as in (19.16). For the physical r significance of r x and r 2 in terms of probabilities, we have to refer back

ceivable that the renormalization constants

to

Chap.

17.

From

e& 0.4

we

find, for instance, that the probability

of finding a bare proton (neutron) in the physical proton is 70 per cent From r2 ^ 0.5 we deduce that the probability of finding (30 per cent). a bare proton with spin up, or a bare neutron with spin down, in a = J is 75 per cent. physical proton with Jz 19.4. Nucleon Self-energy. With the cutoff (19.16) we can also compute the expectation values (19.7) to (19.9). In these expectations the denominators in the integral over do) n are sufficiently large that the main contribution to this integral arises from lower energies where the f ,f-resonance dominates the scattering. We, therefore, neglect a and <7 2 and for cr 3 we insert two-thirds of the measured n+p total cross section. The factor of f arises as follows. Whereas the ir +p system is always in a T = | state, it can be in either a / = f state or a / = J state. The With the above approxiprobability for the former is f [see (16.37)].
,

mation the remaining k integration


(N)

is

elementary, and
I

we

find
(19.17 a)

&

The major contribution to (19.17) arises from the RBA terms, whereas the correction terms amount to about 20 per cent. Because the over d Bk has a A: 4 term in the numerator, there are large contriintegral butions from energies close to <w max * n particular to (19.17). The numbers should, therefore, not be taken too seriously, because the static model is of doubtful validity at energies ^<o max Nevertheless,
,
.

the orders of magnitude, at least of (AO, should be correct. That this is the case can be inferred, for example, from the experimental data of

proton-antiproton annihilation, where we find that the average number of pions produced is ~5* We are tempted to picture this event as The nucleon cores annihilate, giving rise to the minimum follows.
tion) or three.

number of pions, which is two (assuming energy-momentum conservaThe other two or three mesons are then supplied by the meson clouds of the two heavy particles. This number compares favorably with the mean number of mesons evaluated above. Returning to our expression for ^ [see (19.8) and (19.9)], we can derive another instructive inequality. First, we note that all the
1

E. Segre,

Am.

Rev. Nuclear Set., 8:127 (1958) (see especially pp. 148-149).

PROPERTIES OF THE NUCLEON


correction terms (under the integral J dco n ) contribute with the Furthermore, for all physical energies co n , we have sign.

225

same

^ -T-?
co(co

+ J
ft)

-~
^

/
(ft)

.*

co n

V2

^ "I 2
ft>

( 19 '

18 >

which, together with (19.140),

tells

us that

3 f
4?7 TT Jo

dfcrtW ^
ft)

^r

3 f

4?r 77 Jo

rffcpy**
ft)

Thus, the exact 1 lies between the Born approximations taken with the renormalized and unrenormalized coupling constants, respectively.
In the strong-coupling (or classical) limit,
(17.63),

~/ ~
which
is

2
1

(*dk
JQ

-P \k)k*

where/
9/?l ~
[

= 9/;f, we had, from

~~

47T

TT

or

477 77 Jo

the geometric mean between the two limits. Charge and Current Distribution of Physical Nucleon. Magnetic Moment. Finally, we shall evaluate the charge and current distribution associated with the meson cloud and the resulting magnetic moment. These quantities can be measured in detail (e.g., by means of high-energy electron-nucleon scattering) and are, therefore, of special interest. To this end we need the expectation values of (f fl a '(k )a a (k) f ), the amplitude for finding pairs in the nucleon, which we derive e.g.,
19.5.
'

with the aid of the equation

(19.21)

Remembering
find,

are constant, since

that expectation values taken between stationary states { 0, and that a and V commute, we [//,$] f )

by means of (19.1),

0)

0)'

K'(fc')

out.

)<out,
ft)

n
|

ft)

-f

ft)'

L
|

ft)^ _|_

<'

Kg(k)

out, K)(out, n

wijj)]

226

PION PHYSICS
similarly,

and

by hermitian conjugation,

O)

0)'

~i +

$
tf'
I

<g/

ut> n)(out>
a) n

n
i

-f
\

ca'

Ka (k)

out, nXout, n
<*>n

F ,(k') \
tt

f>1
-I

+M

More specifically, using the results of Sec. 19.2 and the notation and techniques developed there, we have the result

X
2 r

Ta .(q.k')T,(q.k)
co'

4f3 Ji
fc

jo..
2

B p (fc n )

"L WB

r^- t -.^ + ^t.^+


to'
,

in

With these expressions we are able to calculate local quantities quadratic For the current (7.12), j, = -efaVfa 2 ^i)> we find
<

<f>.

^ =- S
k,k'

(4(00) )

al(-k')a 2(k)

ai (

k,k'

COCO

lL + 1
ft>a>'

97r Ji

k n p\k n )

(co n

*)]
(19.24)

where we have used


rxa

kr2o k

+Ta
2

k'^a k

kXk'

(19.25)

PROPERTIES OF THE NUCLEON

227

We can

We note
J

further reduce (19.24) by carrying out the angular integrations. that


'

d*k d*k' e < <k+k ) "7(k

- k')(<J

k')
*

VX

IW fk' k k J

~ (k k (k + k')
'

'

This can be most easily seen by considering the y-component and


rewriting the integral as

I (V X
i

a),

J[fk

fk'

(k,

kftk,

fyfM) e
=
x and

k+k > r

If

we make a change of

variables to

(k

k')

(k

-f k')

= K,

the integral becomes

where F1 and
because (V

F2 are arbitrary functions. The F2 term will not contribute x a) operating on it gives zero. For F we find
l

We

can therefore rewrite (19.24) as

= eV X

(S

TS

Oj

^A *^-

} fefe

k)
8

(k

^ <k+k

'

"

[A +![ Law'

09.26)

9irJi

Since the integral in (19.26) depends only on |r|, the current of the meson field is perpendicular to a and r corresponding to a magnetization
It is positive for the proton and equal but density a/(r) (see Fig. 19.1). With (19.16) we find that about 70 per cent for the neutron. opposite term oc< of the current comes from the ')<' V< f) <f>

RBA

This is which gives a contribution at large distances oca x Ve~ 2r /r2 in qualitative agreement with the results from electron-scattering
.

228

PION PHYSICS

1 experiments, which show that the current which creates the magnetic moment of the proton and neutron is spread out over a region with 13 J. mean-square radius ~0.7 x 10~ cm

Similar behavior characterizes the charge distribution, which can be calculated along the same lines, 2
tf

=
2
-

"
(2?r)3

(19. 27 a)

2or3
f

.wco

9vr

+
r 2a

co)(a) n

(19.27/7)
co

-J
(19.28)

since

kr2a

k'

2/r3 k

k'

The charge density ( Q^(r) ) is also equal but opposite in sign for a 2r 2 proton and neutron, the dominant RBA term behaving like e~ /r but opposite-in-sign charge density agrees asymptotically. The equal with a naive picture, in which the virtual processes p -> n 4- TT+ and n -+ p + 7T~ create a positive cloud around the proton and a negative cloud around the neutron. Hence, the neutron, though neutral as a
| \

whole, has some electric structure. It is, in this respect, like a hydrogen atom,
positive inside

To
add

and negative outside. get the total charge, we have to the charge of the bare nucleon

to the charge of the cloud. It seems reasonable to assume that the former has a distribution ocp(r) and, hence,
will

be
e(f\

P(r)\)

Fig. 19.1. Relationship of meson current to the spin direction.


1

With

rx

0.7/o(r)

for

given by (19.16), we obtain the physical proton and

R. HoTstadter, F. Bumiller, and M. R. Yearian, Revs. Modern Phys., 30:482

(1958).
2

The terms proportional


etc.,

to ^(k')

f8 (k), ^(k>2 (k),

etc.,

which

arise

from

^2,

cancel in the expression (19.27#) for

PROPERTIES OF THE NUCLEON


0.3/>(r)

229

Hence, physical anticipated charge distribution of the nucleons, including the charge of the bare nucleon, should have a radial dependence like that shown in Fig. 19.2. This is
not

for

the

neutron.

the

borne

out

by electron-scattering experiments.

Whereas the

mean-square radius

of the charge distribution of the proton


that of the neutron
is

is

0.75 x 10~ 13 cm, as expected,

much

less,

if

not zero.

The most

plausible
itself is

answer to

this puzzle at present is that the

charge of the pion

-Bare nucleon

QP (D
-Bare nucleon

lArr,

Fig. 19.2. Static charge distributions of the physical nucleons.


z spread out over a distance of about \. Averaging the neutron charge distribution over this distance will considerably reduce the mean-square radius, whereas that of the proton will not be affected materially. The integrated charge and the total magnetic moment of the pion cloud can be directly obtained from our results. For the former we find

e(S

r3

>->/ 7T/0

147TCO

._Lf2 367T Jl

da> n

2a3

k n p\K)
(19.29)

Since the integrated charge, including the nucleon contribution, Q, QN, with

is

Q =

'l

+rs
)

and the Q of the proton or neutron state must be 1 or 0, Eq. (19.29) is It also gives an upper limit for o> max since the another relation for r x of the meson cloud must not exceed e. Even if we neglect all charge
.

See Hofstadter, Bumiller, and Yearian, op. cit. This is thought to arise through a pion-pion interaction. and J. R. Fulco, Phys. Rev. Letters, 2:365 (1959).
2

See, e.g.,

W. R.

Frazer

230

PION PHYSICS

cross sections except er3 , then for too large a cutoff wmax , the integral J d*k becomes too big. These observations were crucial in determining term is just the values of co max and r x quoted in Sec. 19.3. The

RBA

2e/3 of the one for (N) [see (19.7)], as one would expect, since (16.34) tells us that two-thirds of the time the proton is dissociated according to TT. ->n -f- n + and for one-third of the time into

For the magnetic moment we

find,

by means of a partiaMntegration,

k nP *(k n )

expect the magnetic moment, when measured in nuclear magnetons, to be of the order of the (mean number of mesons) x (mass ratio of the nucleon to that of the meson), since the magnetic moment of a spin-zero particle with / = 1 is the mass ratio multiplied by the magneton of a spin- J nucleon. Experimentally,

One would

so that the

do justice
to

an

moment of the meson cloud is in the right direction. But to to the experimental physicists who measured these quantities 1 more accuracy of five figures, we have to do better than that.

argument is as follows. For either nucleon we find from (16.34) that both the probability of the virtual mesons being charged and the = 1 are f Furthermore, what counts probability of their having an lz is the relativistic mass the meson total energy o>), so that (e.g.,
refined
.

m==
9 7o
is

expected.

If N/co

is

interpreted as

N(k) dkfa> we obtain from the

RBA

of (19.7)

23
which is one-half of the RBA of (19.30).

The reason is that our argument

However, implies that in j we keep only the terms proportional to cFa. the pairterms aa and cPcfi contribute to the an equal amount, so

RBA

For references, see, e.g., J. W. M. DuMond, Ann.phys., 7:365 (1959); and J. M. Blatt and V. F. Weisskopf, "Theoretical Nuclear Physics," p. 31, John Wiley & Sons, Inc., New York, 1952.

PROPERTIES OF THE NUCLEON


that this

231
If the

more

sophisticated argument
f
Jl
00

is

wrong by a

factor 2.

contribution from
(19.16)
is

da> n

is

included, then the exact result with


nuclear magnetons

^=

(r3a)

1.3

(19.31)

This rather small value is mainly due to the fact that most of the mesons have high energy, so that the effective mass ratio is ~M/co m&K
rather than M. To discuss the significance of (19.31), isovector and scalar parts of by

we introduce

the

where experimentally
<m v

2.35

W,,
4,,

= 0.45

The meson cloud contributes 1.3 to $R so that an extra contribution of 1 nuclear magneton for $R W and $Rg must be accounted for by other phenomena. These may be the magnetic moments of the bare protons, antinucleons, heavy mesons, and hyperons, etc. They seem to give
shall see in the next chapter quite substantial corrections to Jt. that the static model is unable to predict electromagnetic phenomena at short distances (inside the core), which should not surprise us. All we

We

can say is that the first figure of the magnetic moments can be understood in terms of a simple model, but it will take a while until we can account for the other four measured figures.
Further Reading

M.

S. Fubini,

Cini and S. Fubini, Nuovo cimento, 3:764 (1956). Nuovo cimento, 3:1425 (1956).

H. Miyazawa, Phys. Rev., 101:1564 (1956). Fubini and W. Thirring, Phys. Rev., 105:1382 (1959). G. Salzman, Phys. Rev., 105:1076 (1959).
S.

^ The See strong-coupling result is the correct RBA x Dancoff, Phys. Rev., 62:85 (1942). Note that Wl s = in the strong-coupling limit, since <o> = core contributions assumed.
.

W.

Pauli

and

S.

M.

0, irrespective

of the

CHAPTER

20

Electromagnetic

Phenomena

Charge and Current Operators. Electromagnetic important tools for studying the pion-nucleon 1 system. The interactions of photons with matter are well understood and can, therefore, be used to give us further insight into the structure of the nucleon and its interaction with pions. Part of this study has already been carried out in the last chapter; here, we wish to extend it to meson production and to photon scattering (Compton effect). To this end, we first must define the total charge density and current The charge density includes contributions from in the static model. 2 mesons,
20.1. Contributions
effects are

to

C,(r)

etfrh

- <^

2)

(20.1)

and from the bare nucleon


(20.2)

The

spatial distribution of
is Jr

particles

Q N is assumed to arise from the virtual which make up the source. 3 Similarly, the meson current

operator

= K^V^ - ^V^)

(20.3)

and that of the bare nucleons arises from their normal (Dirac) moment and the contribution of other virtual particles. This part should be
in the static model which will appear shortly. plane-wave description of the meson field is more useful in this chapter, because Q n(r) and j w(r) will be needed in other than P states. 3 Strictly speaking, the bare neutron is then also expected to have a bare-nucleon charge density with an average value of 0, but this is neglected.
1

There are reservations

232

ELECTROMAGNETIC PHENOMENA
expressible in terms of only the nucleon operators T that it is divergence-free and hence that
lv(r)

233

and a; we assume
(20.4)

= VxMr)
and we

where

related to part of the magnetic moment, \LN determine its form. The only feature of j^ which shortly
(T) is

shall

we

shall

need

before

on

define the operator more carefully is that it does not meson-creation and -destruction operators, so that
l>,(k) J*]

we

depend

OldOJ*]

The two

contributions to the charge and current considered above cannot represent the total current, since the continuity equation
<2(r,0

+ V.j(r,0-0
source.

(20.5)
field

cannot be

satisfied

inside the

Thus, making use of


//',

equations (15.7) and (15.8) with the interaction

we

find
(20.6)

&(r,0

+V

j,

eRrJfih

- T (0 AWO V P (r)
X

and

rfV [^(O^r',0

-T

<k(r',0]V/>(r')

(20.7)

In the limit of a point source this gives

with

+V = */aVMn(OA(r,0 ~ T2(0&(r,0] j,
-

+V

(20-8)

Thus, (20.5)
current" 1
j7 .

is

satisfied if the total current includes the "interaction

When
charge
the

a n+

is

physical reason emitted by a proton,

The

why we need
it

requires a current to Since the density of carry the charge from the nucleon to the cloud. P-wave mesons is zero at the origin, they cannot immediately take over

Q N (r) suddenly disappears and meson cloud. The continuity equation thus

j z is the following. into a neutron, and the changes is transferred from the source to

In a scalar theory, where H' is not the charge from the nucleon. mesons interact), (e.g., 5-wave proportional to a V< but merely to the continuity equation is satisfied without any j z For an extended source the continuity equation is still not satisfied 3 We can inside the source, even if (5 (r) in (20.8) is replaced by />(r).
<f>
.

introduce additional 2 currents to


1

make

the continuity equation valid

replacing

This current can be deduced formally by introducing complex fields and by '. This generates an addition A j/ to H'. feA^ in V^ by V< 2 See R. H. Capps and R. G. Sachs, Phys. Rev., 96:540 (1954); and R. H. Capps and W. G. Holladay, Phys. Rev., 99:931 (1955).

234

PION PHYSICS

inside the source, but these are not uniquely defined and appear to be unimportant. At this stage, we have to appeal to a relativistic local

theory, which in the static limit gives the three kinds of currents we considered. With these, (20.5) is satisfied for an extended source when
1 integrated over any region which contains the source

+ V.J]=0

(20.9)

static

Summarizing this discussion, we can say that the predictions of the model for electric phenomena may not be as strong as those for

purely pionic problems. Having established the form of the current we needed, we shall The expectation values investigate the properties of the various terms. of } and QJ(f) between physical nucleons, which involve the P-

wave parts, were investigated in the last chapter. Since, however, the operators are bilinear in <, they contain not only other angular momenta but also cross terms between S and P waves or between
and

waves, for example. These types of terms are particularly important for electric-dipole transitions, which require a change of Since we shall be parity and of angular momentum by one unit. concerned with photons of wavelengths X < 10~ 13 cm, or &photon *> 200 Mev, a multipole expansion may, however, not be suitable for this term. The current j 7 involves principally S waves, 2 because it is zero
outside the source, and inside
it,

^photon

''source

~T

**

*^ r

^photon

1U

Cm

only for higher photon energies that the other angular momenta begin to play a role for jz There is little we know about j iV since the naive expectation that it is just the current of a Dirac particle for the bare proton and zero for the
It is
.

bare neutron gives a considerable departure from the observed magnetic Fortunately, in the following, the expectation value of the total current between physical nucleon states will be important, and this can be related to the observed magnetic moments. Thus, let us consider a Fourier component k of the matrix elements of j total s= j

moments.

between nucleon states. This must be a polar vector that can depend The continuity equation which tells us that only on a, k, and r 3
.

This equation is not very strong and is also satisfied without j/. Because of this, it is also expected that j/ will not contribute appreciably to the magnetic moments.
2

ELECTROMAGNETIC PHENOMENA
therefore restricts the expectation value to
1

235

= tf
The

static limit (e.g., k -> 0) of the bracket term in (20.10a) corresponds to the total magnetic moments of the proton and neutron. This follows from the fact that

limi f[rX
k->o2 J

j(r)]

fj

k->0

We

thus note that


(20.106)

Since the

RBA term of j w was shown in Chap.

19 to be cce~ 2r /r z outside

the source and since the other contributions arise from the source and should be even more concentrated around the origin, we expect the form
factors

and Fs to be approximately equal to their static value for and we shall use these values. Since we know the expectation value of j w we could use (20.100) to determine j v more specifically. However, our applications will always involve the expectation value of the total current, so that we need not do this. The first process that we shall 20.2. The Production Amplitude.

Fv

k <

2,

the photoproduction of mesons, that is, reactions of the type This process involves not only the pion-nucleon -f TT. y system but also the radiation field. For its proper treatment we have to analyze the Hamiltonian

turn to

is

+ N -> N

H = Hf + x + H + H
Y

int

(20.11)

We shall actually consider j(-k), since this is more useful for later discussions.

2 These form factors are the same as those in electron-proton scattering, and the bears this out. See R. Hofstadter, F. Bumiller, and analysis of these experiments M. R. Yearian, Revs. Modern Phys., 30:482 (1958).

236

PION PHYSICS

where

is

the Hamiltonian of the free photons

and //int

is

the inter-

action of the photons with the charged particles. shall not go into the details of the quantization of the radiation

We
1

but shall merely mention the pertinent facts. The free Hamiltondescribes the free photons which are massless particles with spin Y 1 but with spin componentsy z only parallel or antiparallel to its momentum. In an angular-momentum expansion, we find that* there are no For / one-particle states with zero angular momentum. they may have either parity. Photons with parity ( I) are called electric /-pole l+1 are called photons, and those with parity ( l) magnetic /-pole The interaction Hamiltonian is photons.
field

ian

Hmt

=
-JrfVA(r).j(r)

(20.12)

where the vector potential A(r) is built up, in the usual manner of emission and absorption operators for photons. Our goal is to find the transition probability from an initial state

+ to a final state -f TT. We shall calculate it with the aid of the "golden rule" (8.27). This is simplified by the facts that //int is weak and that in a perturbation 2 development in powers of e 2 we can replace Tfi by the corresponding matrix element of mi Accordingly, we shall work out
y

(out,

N+

77
I

Hint

y)

=-

J (ZTT) (Zk)

~i
TT
|

{out,

N
|

-f

77
|

j(r)

JV>

= e(out, N +
Here we have assumed an
zation vector
c,

j(-k)

N)(2k)~

(20.13)

initial photon with momentum k and polarinormalized to one particle per unit volume, so that 3

(0

I '

A(r) V

I '

-(2fc)*(2ir)

We shall first derive an integral equation for the matrix element of the
1 The quantization is similar to the treatment developed in this book, except that the spins of the photons are 1 and all photons are transverse. See, e.g., L. I. SchifT, "Quantum Mechanics," 2d ed., pp. 196-210, McGraw-Hill Book Company, Inc., New York, 1955.

The pion interaction is treated "exactly." As in the case of the atomic photoeffect, the vector field A can also be considered a classical field. This effectively gives the same result as (20.13), in which we have
3

already taken the matrix element with respect to the photon variables.

ELECTROMAGNETIC PHENOMENA
Fourier transform
j(

237

k) of the total current.

By means of
1

(19.1)

we

can obtain a relation similar to the


(out,

Low

equation:
1)

N+

7r

k a

J(-k)

N)

(('

[*.(k'),

K-k)]

'
|

[a a (k')

+.(k') e*'dt, j(]


|

f>

+S
n
|

it

'

tf

j(-k)

out. n)<out, n t,
)

F.(k')

fil

'

c'
|

K.(fc')

out. iXout. "

j(-k)

fi

[tf L

E
|

(O

(r
|

j(-k)

out, liXout. n
w

Ka (k-)

f)1

+w

(2Q14)

expression for

The one-nucleon expectation value can readily be evaluated with With the help of j, which we discussed earlier.

the

we

find for the

commutators

f (27T) J
,

d 3 r [a a (k'),

(20.15)
1

We remind

the reader that

5'

implies a

sum

over continuum states only.

238

PION PHYSICS
the

The nucleon expectation value of can be found by means of (19.3),


<r
|

sum of

the above commutators

wk'), K

and that of the current


<f
If

j is

[see (20.10)

with F(fc 2)
3

F(0)]

K-k)

*>

'

*'(

*
|

^-~- TO,

^-p 8R,)

1)

(20.17)

we put the meson energy CD' = k (as is dictated by energy conservamake use of (19. 1), and remember that V A = implies k c = 0, find for the photoproduction amplitude we
tion),

2k'q.(k'-k)1
,

(R

fc)2

3^
, p(k')

cTtf'
n

K,(k')
(

out, n)(out, n

j(-k)

|)

L
I

o)

i)(2a/y
|

<y

K-k)

out, n)<out, n

Kg(kQ

1)1
j

(E w

co)(2a>f

J/

The one-nucleon terms


if

are again equal to the renormalized

Born

approximation the experimental values for the magnetic moments are The significance of these various contributions is the following. used.

The

first term on the right-hand side of (20.18) creates only charged mesons and corresponds to an absorption of the photon by the meson, Of the two contributions, one arises as in the ordinary photoeffect. from the current j (proportional to k'), corresponding to an absorption by the meson in the cloud, and the other from the current j 7 (proportional to a), corresponding to an absorption of the meson in the process of creation. These two contributions are shown diagrammatically in The process of Fig. 20.16 mainly creates mesons in an S Fig. 20.1. state, whereas that of Fig. 20. la contains all angular momenta, owing 2 to the retardation denominator l/[(k' k) + 1]. This arises because
ff

the spatial extension of

j ff

(10~

13

cm)

is

not small compared with the

ELECTROMAGNETIC PHENOMENA

239

wavelength of the photon needed to create a real meson, so that a multipole expansion for this factor is not rapidly convergent. The terms with the magnetic moments correspond to an absorption of the meson by the magnetic moment of the nucleon accompanied by the emission of a meson, as illustrated in Fig. 20.2. They are the only terms which create neutral pions as well. One would expect these terms to be of the order \\M compared with those of Fig. 20.1, but since Wl n = 4.7, they may become comparable to the others.

(a)

P(n)
(b)
Fig. 20.1.

P(n)

Diagrams

to represent the

Fig. 20.2.

Diagrams corresponding to

photoproduction of charged mesons caused by the terms in the matrix element which arise from (a) [fl a(k'),

the photoproduction of neutral and charged mesons due to the interaction

of

the

photon

with

the

magnetic

moment of the

nucleon.

The contributions contained

in

$' correspond to a rescattering of the


M

meson once

1 soluble, task.

is a complicated, but Fortunately, the dominant rescattering correction can be obtained without solving the integral equation (20.18) for the proit

is

created.

Their evaluation

From our experience with n-N scattering, we the amplitude in expect that the rescattering gives an enhancement of 3-state and a suppression in the other states. the Furthermore, it turns out that these corrections mainly affect the magnetic-moment term
duction amplitude.
(Fig. 20.2)

reason

is

meson
1

and not so much the photoelectric effects (Fig. 20.1). The that in the process of Fig. 20. la the photon is absorbed by a of the cloud, which is at some distance from the source and,

R. Omnes, Nuovo cimento, 8:316 (1958).

240

PION PHYSICS

therefore, has a reasonable chance of escaping without further interaction. Diagram 20.16 leads to mesons in S states which escape without further interaction. 1 On the other hand, the terms represented

by diagram 20.2 are considerably affected by corrections of higher order in/. These can be calculated by introducing a current j' which N9 depends only on the operators c and T (and not on the meson operators)
:

j^(-k) ~
With

J.

+L

From

the definition of this current


|
|

we have
and
,

(f J(~~k)

f)

(S

} }' ) N the first term gives the same N ]' (20.15) but zero contribution to the magnetic-moment terms (20.17). That is to say, the amplitude for this part obeys the

Hence, on writing

&( =(}
|

'

k)

I)

[0 a (k'), j^(

k)]

commutator

integral equation
f

=
1

efr
3

p(\k

- kl)

(27r)

2a/

'
|

VJP)
k)

out, nXout, n

j(-k)

- j^(-l
\

(2a>')\E n
,

-coo>)

ie)

j(

ijy(

k)

out, w)(out, n
n

(2oi')*(

(20.19)

This is identical with (20.18), except that the magnetic-moment terms are missing. argued that the rescattering corrections for the terms in (20.19) are insignificant, and to a fair approximation we keep here only the renormalized Born approximation. Turning to the matrix

We

elements of j^,

we

first

observe that

One might think that the S-wave scattering, which is not contained in our model,
significantly

would

section*.

change this contribution and therefore the low-energy cross This does not occur, as was shown by S. D. Drell, M. H. Friedman, and F. Zachariasen, Phys. Rev., 104:236 (1956). The limiting form of the cross section obtained at low energies also holds in relativistic treatments and is known as the Kroll-Ruderman theorem. See N. M. Kroll and M. A. Ruderman, Phys. Rev.,

93:233(1954).

ELECTROMAGNETIC PHENOMENA

241

can be directly expressed in terms of scattering amplitudes, since it involves only matrix elements of a and T between a nucleon and a "nucleon pion" state. To this end, we have to expand in projection

operators of eigenstates of angular momentum and isospin. Of these, only the matrix element for the transition into the f, {-state becomes comparable with (20.19), whereas the other terms remain small. With the aid of (18.6), we can express the matrix element

This is most directly done by using a plane-wave representation. There we have to take the projection operators (19.50) and change the density of state factors in (18.25) from
in terms of scattering amplitudes.

X * (out,
.

k'a

I '

r3a

X
,

If)
'

s* n

p(k
a

c)
/a

'

k'

(k

c)]

From

this relation

we immediately
(2(0)

obtain 1

,k'a|.j v (-k)|>

Lj

_ ~ Jr TO, -M. 6*81114, *


x
This
is

3 '"

_
X
)]

[2k'

/a

k'

(k

(20.19a)

term of (20.18) multiplied by the ratio of the actual scattering amplitude to the RBA scattering amplitude. Regardjust the

RBA

ing j s we find that its matrix elements are insignificant compared with the leading terms that we have evaluated so far, namely, (20.19) and This is partially due to the fact that j s cannot lead into the (20.190). $TC W < J(^ n ). Thus, j s will be f ,f-state, and because p Since (20.190) is then the only term 2 which in the following. dropped
,

<

produces
ing:

TT'S,

we can

relate the

77

cross section directly to pion scatter-

<2<u *"
1
<5

Remember that the kinematics are now such

that

co

==

o>'

(k'

2
-4-

1)*

and

to be taken at that energy. 3 2 Actually there is also a small production amplitude of 7r's in in our model. scattering, which is not contained
is

5"

states

due to the

242

PION PHYSICS

Altogether, we get the following result as a fair approximation to the rather complicated production amplitude:

[2k'

10

k'

(k

eXKlSiAo)"

(20.20)

20.3. General Features of the Cross Section.

"golden rule," the cross section


(20.20).

With the aid of the can be worked out from the amplitude

The result is fairly complicated, but the physical significance and the important features of the various contributions can be understood by simple probability arguments. For the meson-cloud-creation
process (Fig. 20. la) the cross section
is

expected to be of the order of

photon gets into the meson cloud) x (probaa meson is present) x (probability for absorption of the photon by a meson in a P state) x (probability that the meson leaves 1 the cloud per unit time) x (incident photon flux)a
(probability that the
bility that

(20.210)

This cross section shows a strong increase with meson momentum and The contribution is vanishingly small at the production threshold. from j/ (Fig. 20. \b) is similar but lacks the factors k' 2 and k 2 since the meson is produced in an S state, so that 1
,

f 2 P 2 k'

The energy dependence of these two terms is illustrated in Fig. 20.3. These estimates again reflect the features of the exact expressions only In general, the at low energies, where the probabilities are small.
threshold
conservation.
is dictated by parity and angular-momentum The various possibilities are summarized in Table 20.1. table we learn that the linear increase of <rz with momentum

behavior

From
1

this

More accurately, one obtains at oc Ar'/o>, which tta energy dependence plotted Close to the threshold for meson production o> ** 1 , and the difference in Fig. 20.3. between the two expressions is small.

ELECTROMAGNETIC PHENOMENA
Table 20.1

243

Momentum
Type of
pion

+p =

4- n

dependence of cross
section

Scalar pion

Electric dipole

-1)

1,

J,

A:'

Magnetic dipole

(& = +1)

(/

= 0, / = J = 2, / - f

f)
k'

Pseudoscalar

Electric dipole

= - 1) =
+1)

= 0, J = =
l,

pion

(kj
Magnetic dipole (0>
/

|,

1.0

0.8

I
i

0.6

0.4

^
0.2

0.8

1.0

1.2

1.4

1.6

1.8

Fig. 20.3. Qualitative energy dependence of the photoproduction cross section from the meson cloud, a (see Fig. 20. la), and from the interaction current, a/, (see Fig.
20.16).

a manifestation of the pseudoscalar nature of the pion. For scalar dominant low-energy process would be the absorption of an electric-dipole photon by the meson in an S state which is then 3 emitted in a P state. This has a threshold behavior not ack' but ocA:' This difference is also reflected in the Tike the atomic photoeffect. angular distribution, which is isotropic for pions emitted in S waves and
is

particles, the

244
2

PION PHYSICS

oc sin for P waves, since the angular the photon momentum. Near the resonant energy (o>' 2, or

momentum is

in the direction

of

system),

the

important.

300-Mev y rays in the laboratory contribution from the isovector current j F becomes Whenever one state dominates the production process, the

angular and energy dependence of the cross section is quite simple. Thus, the angular distribution at the f, |-resonance is obtained by averaging over the two possible spin states of the initial nucleon. The nucleon spin can be quantized along the photon spin direction, which can be taken to be parallel to the photon momentum. Further averaging over the photon spin is unnecessary, and we obtain [see
(16.37)]:

The

factor of (J) 1 for the antiparallel-spin case arises because the initial

state has a probability of J- for being in the / f state and of f for being in the J state. The total angular distribution for mesons produced ^

= in the J =

f state

is

thus

1/"V

/-klll'rtlll

^^

'

contains twice as many neutral pions Since the T f, z J- state as charged ones [see (16.35)], we obtain 2 for the ratio of neutral- to Near the resonant J f state. charged-meson production in the T the energy dependence of the cross section from (20.19a) is energy,

T =

given by

=
20.4.

IT,
with F,

,\~y

/8^ L

-ro 2

JF

,'

(20.23)

Comparison with Experiment.

The experimental

total cross

section for the photoproduction of charged and neutral mesons is shown in Fig. 20.4. Both the resonance and the threshold behavior of

model.

the cross section behave qualitatively as predicted by the simple static For neutral mesons the experimental cross section is much

ELECTROMAGNETIC PHENOMENA
less

245

than the charged one at low energies, but it catches up close to the resonance energy. Over this whole range of energies, the TT Q cross section is well described by (20.19&) and (20.23). The solid curve of
Fig. 20.4
3.0
is

just the contribution

from

jv

as given

by

(20.190).

The

xsr(K.M.)
2.5

r+(W.T.P.V.,T.K.W.)

}r*(B.B.C.S.T.)

2.0

D
5
i
1.5

L0

0.5

150

200

250
f

300

350

400

450

Energy of photon, laboratory system (Mev)

Fig. 20.4. Total cross section for

w and TT photoproduction on protons, as a function of photon laboratory energy. The TT experimental data were obtained by L. S. Koester and'F. E. Mills, Phys. Rev., 105:1900(1957); and by W. S. McDonald, V. Z. Peterson, and D. R. Corson, Phys. Rev., 107:577 (1957). The solid curve for

the 7r photoproduction corresponds to the contribution (20.190) due to the current jv alone (see Koester and Mills). The high-energy TT+ data are an average of those found by R. L. Walker, J. G. Teasdale, V. Z. Peterson, and J. I. Vette, Phys. Rev., 99:210 (1955); and by A. V. Tollestrup, J. C. Keck, and R. M. Worlock, Phys. Rev., 99:220 (1955). The lower-energy data were obtained by M. Beneventano, G. Bernardini, D. Carlson- Lee, G. Stoppini, and L. Tau, Nuovo cimento, 4:323 The dashed curve is an arbitrary one to fit the experimental data and to (1956).
illustrate the

low-energy behavior (see Fig. 20.3).


TT +

mesons needs some refinements, 1 mainly due to recoil The additional effects, but is quite well described by the theory. contribution of the electric-dipole terms and of higher-order multipoles
excitation of

causes the total


1

TT

+ cross section to be larger than one-half that of the

See M.

J.

Moravcsik, Phys. Rev., 104:1451 (1956); G. F. Chew, Phys. Rev.,

106:1337(1957).

246
TT, as

PION PHYSICS

the end of the last section.


/r/4?r

would have been obtained according to the argument given at The low-energy data fit (20.21) with fe* 0.07, in good agreement with the coupling constant found
scattering.

from meson-nucleon

20

16

$,

30

60

90

120

150

180

40

80

120

160

Center of mass angle (deg)

Center of mass angle (deg)

production laboratory energy of 300 Mev. The experimental points stem from W. S. McDonald, V. Z. Peterson, and D. R. Corson, Phys. Rev., 107:577 (1957); D. C. Oakley and R. L. Walker, Phys. Rev., 97:1283 (1955); and Y. Gold-

the

Fig. 20.5. Differential cross section for of TT mesons at a

Fig. 20.6. Differential cross section for of TT+ mesons at the

production

175

Mev (open circles) and at 260 Mev.


are due to Beneet al. (see

The former data


ventano

legend for Fig. 20.4). The sources for the higher-energy data are also cited in the legend for Fig. 20.4. The dashed curve is a straight

schmidt-Clermont, L.

S.

Osborne, and

M. Scott, Phys. Rev., 97:188 (1955). Dashed curve represents 2 + 3 sin2 #, as predicted from a pure J = T = f
contribution to the production process.

corresponding to pure S-state production. The higher-energy curve is taken from "Proceedings of the 7th
line,

Annual Rochester Conference,"


58, Interscience Publishers, Inc.,

p.

II-

New

York, 1957.

distribution of TT mesons in the resonance region also with our model and is given fairly well by (20.22). The comagrees parison is made at 300 Mev in Fig. 20.5. For TT+ mesons, the model are produced, with a predicts that, near threshold, only S-state mesons This is seen distribution. consequent spherically symmetric angular

The angular

to be so within experimental error at 175 Mev (25 Mev above threshold) At higher energies, there are interfering contributions in Fig. 20.6. from the electric and magnetic absorptions. Since the latter has a

ELECTROMAGNETIC PHENOMENA
id *
9

247

the interference term changes rapidly near the resonance. phase e This is reflected in a shift of the maximum of the angular distribution from backward to forward directions in this region. A comparison between the angular distribution and the predictions of the static model are shown below the resonance energy, at 260 Mev, in Fig. 20.6. Thus, the theory successfully predicts subtle effects, such as the relative phase of the electric and magnetic amplitudes. Considering the uncertainty of the static model for the description of electromagnetic phenomena, as well as the neglect of S-state nucleonmeson interactions, the agreement above is remarkable. 1 In particular, it must be stressed that the photoproduction of charged and neutral pions provides two new independent measurements of/, and that both agree with the value deduced from pion scattering. Another electromagnetic phenomenon 20.5. Compton Scattering. for which the f,2-state is important is the scattering of photons by nucleons. This is an effect of second order in e, and we shall mention only the pertinent points. The normal (Thomson) scattering involves the translational degrees of freedom in the same way as the Compton The incident photon shakes the proton, which effect on the electron. then emits the scattered photon. Because of the larger mass of the 2 2 10~ 31 proton, this gives an exceedingly small cross section ~e /M cm 2 However, because of the excited state of the nucleon, there is the possibility of a resonance scattering in which the incident photon T = I state, which subsequently decays by emission of excites the J

the scattered photon. 2 In the classical picture this means that the photon acts on the magnetic moment of the nucleon, thereby creating a

forced gyration of the nucleon spin. This will give a resonance scat3 tering of the usual form [compare (20.23)]

(E y

- Ef + (F/2)
;

where A is the wavelength of the incident photon, or !/.,; Ev = reso300 Mev in the laboratory system and the other symbols nance energy are as defined in (20.23). The cross section is much less than the

2 geometrical limit TrA , since the excited state will preferentially decay by the emission of a pion (F A F), but it is much larger than the Thomson

<

model with experiment, the energies which appear always assumed to refer to the center-of-mass system. 2 For an analysis based on the static model, see, e.g., W. J. Karzas, W. K. R. Watson, and F. Zachariasen, Phys. Rev., 110:253 (1958). 3 See J. M. Blatt and V. F. Weisskopf, "Theoretical Nuclear Physics," p. 394,
In the comparison of the static
in the formulas are

John Wiley

& Sons, Inc., New York,

1952.

248

PION PHYSICS
2
2

Modern experimental techniques make it possible scattering (e /A/ ). to verify these predictions of the theory in spite of the small cross sections involved.
Further Reading

G. G.

F. F.

Chew and F. E. Low, Phys. Rev., 101:1579 (1956). Chew, Theory of Pion Scattering and Photoproductior*,

in

"Handbuch

der Physik," Springer- Verlag, Berlin (to be published).

CHAPTER

21

Nuclear Forces

Classical Calculation of Nuclear Interaction In this last chapter we shall use the static model to calculate the long-range, or external, region of the forces between two nucleons due to the "exchange" of mesons. Although historically the Yukawa potential was the beginning of meson theory, this problem is not the most clear-cut test of the static model. static potential should be able to account for all the low-energy properties (e.g., deuteron binding
21.1. Introduction:

Energy.

energy quadrupole moment, and nucleon-nucleon-scattering phase On the basis of range arguments shifts) of the two-nucleon system. alone, the external region of the potential should be due to the exchange of a single pion and should therefore be given fairly accurately by the The exchange of two mesons should be partially restatic model. sponsible for the shorter-range behavior but will be much more strongly influenced by recoil effects (neglected by our model) and by the nucleon source structure (e.g., km& x). Furthermore, other than P-wave mesons 2 term. As we will contribute to it, e.g., mesons in 5 states from a discussed in an earlier chapter, the exchange of K mesons will give rise to a still shorter range force, and these effects are presumably hidden in
<

the source.

We can calculate the classical interaction energy between two sources a distance r apart in a manner analogous to that of Chap. 9. Because the interaction energy is proportional to p 2 rather than to />, we obtain a cross term in the energy of two sources
:

P(r,r

p(r)

p(r

in

(21.1)
[see (16.4)] for
It

The

classical field solution

was obtained

Chap. 16

a
is

neutral pseudoscalar field in the presence of a single source. 249

250

PION PHYSICS

straightforward to generalize this to a symmetrically coupled field to obtain, for static a and T,
(21.2a)

with

y (r)

e \d*r' p(r') J 477

'

'

(21.26)
|r

r
I

In the presence of two sources, the interaction Hamiltonian

H' becomes
(21.3)

H'

-fl
a

f[p(r)
/

rfr

- r )]a- V&r.iPr

By

substituting (21.2) into (21.3),

we

generated by one source at the position of the other one


to the

find that the interaction energy is (a and b refer

two sources)

= /*

a=i

I;

TaaTfta

L(r>(r'
3

r )(aa

V )(a V
.

r)

|r

r
l

For point sources,

p(r)

(21.4)
(5

(r),

this gives

477=i

aa

if
3 L

/3
\r%

3
r

where

ST

is

the interaction energy between

two dipoles (tensor


.

force):

3qa

roq 6

'

rQ

__

ro
finite sources, the long-distance behavior of the interaction energy the same, but the ^-function potential is spread out over the source and gives for the last term of (21.5)
is

With

-/

a=l

2r

aa r 6fl

aa a 6 P(r>(r' f
J

- r dV
)

(21 .6)

In a perturbation-theory calculation of the two-body force, the lowestorder contribution arises from the diagram shown in Fig. 21.1 due to the exchange of a single meson between bare nucleons. The resulting
potential
is

'

'

--r|aq .ka

&

.k

*^n.

(21.7)

NUCLEAR FORCES
and corresponds exactly to that found in the
classical limit.

251

On

the

basis of our discussion in earlier chapters, we should expect the dominant contribution of the static model to be given by the same potential shall see that this is actually the case. with/ replaced by/r
.

We

The charge dependence of the


which has the opposite
isotriplet state

sign,

T& potential (21.7) originates from T depending on whether the isospins of the
tt

nucleons are parallel or antiparaliel.


singlet state. on the relative orientations

Its

expectation value

is 1

in the

3 in the isoand That the force depends

of the

isospins was to be expected, since the possibilities of meson exchange

are different in the two states.


instance,

For two protons can exchange

Fig. 21.1.

Diagram

exchange
.

for the one-mesoncontribution to nuclear

whereas in the isosinglet forces. nucleons can exchange TT O 'S and charged mesons. Correspondingly, the force is stronger in the The opposite sign is connected with our earlier remark isosinglet state. the sign of the coupling constant to the proton and concerning neutron. The spin dependence, which comes predominantly from a factor (oa r cjj, r )/rJ5, reflects the overlap of the meson clouds, each of which is proportional to (a r n ). For typical relative orientations of classical spins and of the distance, the factor a a r o6 r /A-5 has the For the various quantum states, the values illustrated in Fig. 21.3. behavior of the force is illustrated in Fig. 21.2 and Table 21.1. These features of the force are borne out by the structure of the deuteron,
only
TT'S,

state,

77-

ictoh

'Spin triplet, isotriplet states

Spin singlet, isosinglet states

TO

1-5

2.0

2.5

3.0

/^
'

Spin

triplet, isosinglet

states

and
spin singlet, isotriplet states

(6)
Fig. 21.2. Plot of the longest-range contribution to the nuclear potential with o.08. (b) Plot of the (a) The tensor contribution to the potential energy, f*/^

central force.

252

PION PHYSICS
Table 21.1

CHARACTER OF ONE-MESON-EXCHANGE FORCE

IN

VARIOUS STATES

is an isosinglet, triplet-spin state and is elongated in its spin Such a state corresponds to direction (positive quadrupole moment). the deepest value that the second-

which

-I <*

I&

order potential can attain. For other states this potential can also be attractive, but less strongly so.

This

is

fact that the total n-n force

in accord with the empirical is also


is

attractive but

not strong enough

to give a bound state. 21.2. Static Potential,


(a)
(6)
(c)

Quantum-

mechanical.

Fig. 21.3. Illustration of the value of a rob ro/ro f r various nucleST

'

onic space and spin configurations. = 0. (c) S'T (a)S' = (b) S' T T

proceed with the actual derivation of the potential, several remarks are in order. The derivation and use of

Before

we

-1.

limit are

the potential in the static-source one in the spirit of the

Born-Oppenheimer (molecular) ap1 That is, the internucleon potential is calculated for proximation. nucleons a fixed distance r apart, where r is considered a parameter, and this potential is then used in a Schrodinger equation to calculate the It is clear that it is more properties of the nucleon-nucleon system. difficult to justify a Born-Oppenheimer treatment here than in the molecular problem, where the expansion is in terms of (mJM)* J, rather than (/*/A/) 1 ~f, where is the nucleon mass. Nevertheless,

Schiff,

New York,

M. Born and J. R. Oppenheimer, Ann. Physik, 84:457 (1927). See also "Quantum Mechanics," 2d ed., p. 299, McGraw-Hill Book Company,
1955.

L.

I.

Inc.,

NUCLEAR FORCES
it is

253

only in the static limit that a potential can be defined at


1

all consist-

ently,

and

it

was

also in this

way

that the force

was defined

for the

It is clear that the potential derived neutral scalar theory in Chap. 10. in the above manner neglects corrections of order p/M as well as those

due to nuclear momenta p of order p/M.


i

^r ~"
!''

Fig. 21.4. Typical contributions to nuclear forces that are responsible for

changing

/into./;-

difficult to calculate.

corrections to (21.7) of higher order in/ 2 are substantial and Fortunately, for large distances, diagrams of the form shown in Fig. 21.4 are dominant, and their effect is simply to

The

range

change /into/.. The two-meson-exchange diagrams of Fig. 21.5 have |, and their contribution can be related to meson-nucleon

/ \r \ \
Fig. 21.5. Typical

x
?r/
/
N

v \ \

two-meson-exchange diagrams for nuclear

forces.

We shall only mention how they can be computed, since they are strongly influenced by meson-meson interactions and non2 adiabatic corrections. Our results will be derived in the spirit of a Heitler-London approximation. Thus, for calculating the energy of the two-nucleon system, we shall, in first approximation, use a state wherein both nucleons have
scattering.

an undistorted meson cloud.


|fa ,f b ,r
1

This state can be written, with


>

(1 7.6),

as 3

= KJ?

|fa ,f b ,r

(21.8)

field theory.

defined to be an energy-independent quantity that follows from definitions within this context, see K. Nishijima, Suppl. the above "consistent" Progr. Theoret. Phys. (Kyoto), 3:138 (1956). Whether For a discussion of this point, see potential is reliable is quite a different question.
potential
is

For various

J.

M. Charap and
2

See

S.

S. Fubini, Nuovo cimento, 14:540 (1959) and 15:73 (1960). Machida and T. Toyoda, Suppl. Progr. Theoret. Phys. (Kyoto), 3:106

(1956).
3 For the following it must be remembered that the "dressing operator" R contains only meson -creation and a and T operators. Since the operators a and T = 0. belonging to different nucleons commute, we have [/? a ,/? b ]

254

PION PHYSICS
is

and

an eigenstate of the

total

Hamiltonian 1

*//'

+
'

'-*
(21.9)

//a

H' b

=J

</ /c

a a (k)[/Taa(k)

+ *ik '"/Ka6 (k)] +

h.c.

in the limit r -*

oo.

We
ft

shall

now
fl

simply calculate the ground-state


>

energy
<fa ,f ,r

|/f|f ,ft,r

= nro)
more
refined trial state
states.
1

To

refine this procedure,

we have

to use a
-{-

which

also contains admixtures of "nucleon

incoming pion"

They

describe the distortion of the pion cloud and correspond to diagrams of the type shown in Fig. 21.5.

Remembering
(// 4-

H'

2*0)*.

ft, ft, r )

(Hi

- <QK

ft, ft, r )

(21.10)

we obtain

HR a R

fa q
,

ft

FQ)
,

{[[ff,Kj

KJ +

*.//;*>

*>ffJR.

*.*,<//

2<Q}

ft,

f& , r

(21.11)

The double commutator is zero shorter range ( < |) and contributes


2 exchange) of the potential.

to the order considered.

It

is

of

to the next higher order

(two-meson

Furthermore, we have

(H

+ 2*

)
|

^, (
fc

TO)

[~aI(k)/Faa(k)

/e- k

Fab (k)al(k)]

fa ^, r
,

(21 .12)

on the right-hand Because a*Va commutes with Rb and rfVb with /? a the two terms of (2 2) can be canceled against corresponding terms in (2 1 1 ),
since all other terms contain destruction operators
1 . 1

side.

1 .

and we get 3
<.,
ft, r
|//| ft, ft,

>

-2
a

f <^[<f
/

^(k)

ft) (ft, r

a a (k)

ft, r

>

O*"'*']
1

(21-13)

The exact form of Ka(k) is given by (19.1), and the subscripts a, b imply that T a or Q6> T are to be taken. The abbreviation h.c. means hermitian conjugate. 2 See R. E. Cutcosky, Phys. Rev., 112:1027 (1958) and 116:1272 (1959); lu, I. V. Novozhitov, /. Exptl. Theoret. Phys. (U.S.S.R.), 32:1262 (1957) and 33:901 (1957) [trans, in Soviet Phys. JETP, 5:1030 (1958) and 6:692 (1958)].
o
e>

The

factorization
|

Ra R b
|

a , ft)

ft

>

16

>

is

possible since

|fl ,

b)

is

the

direct product of

a)

and

ft).

NUCLEAR FORCES

255

The expectation value of a a(k)


f
2

is

given by (19.2), and

we

finally obtain

f
(21.14)

-.TT3
which, as predicted,
2

(2?r) J
is

exactly the potential (21.4) or (21.7) with the

coupling constant/ replaced by the renormalized one/Jf. 21.3. Comparison with Experiment. Having found the exact value for the strength of the potential, we are in a position to make a quantitative comparison with experiment. To this end we have to take into account the translational degrees of freedom of the nucleons and add their kinetic energy to the potential (21.7). The two-nucleon wave function & a $6 (r ) is defined by writing the two-nucleon state
2)

-2
b

^cfeO-o)

ftp

*,

FO>

(21.15)

an eigenstate of the total energy, then % obeys the Schrodinger equation


If

2>

is

[-*.*.'*'

V*

+ *W'*'('a)] fc.'*'fra) = Efafa)

(21.16)

To

space.

solve this equation, we have to diagonalize i-^ in spin and isospin This is carried out in most books 1 on nuclear physics and leads

from the deuteron ground state to the well-known equations which couple S and D states. Since our potential (21.11) describes only the
it has to be adjusted for r < 1, in a manner. However, the quadrupole mopartially phenomenological ment and the binding energy of the deuteron are sensitive principally to the behavior of the wave function % at distances r Q > 1 and should be fitted by the potential (21.14) with the coupling constant found from 2 pion-nucleon scattering. A detailed discussion of (21.16) shows that

long-range part of the force,

it is

unless the coupling constant is within the range 0.065 <, f* /4?r < 0.09, very difficult to fit the above static properties of the deuteron. The low-energy scattering parameters, that is, the scattering length and
effective ranges in

S and P states,

are also consistent with the solution of

(21.16) for r

>

1.

Recently ways have been found to extract directly from the nucleonscattering data that part which comes from the exchange of one pion.

The method

consists in extrapolating the scattering cross section to unphysical energies and angles where the intermediate pion becomes
1

See, e.g., J. M. Blatt and V. F. Weisskopf, "Theoretical Nuclear Physics," Sons, Inc., New York, 1952. chap. 2, John Wiley 2 See J. Iwadare, S. Otsuki, R. Tamagaki, and W. Watari, SuppL Progr. Theoret. Phys. (Kyoto), 3:32 (1956).

&

256
real.

PION PHYSICS

In this unphysical situation it can propagate over large distances, The residue at this pole is scattering amplitude has a pole. directly related to the renormalized coupling constant, and the experimental data 1 give /|/47r 0.07 0.01, in agreement with the value deduced by other processes.

and the

energies the situation becomes more complicated, since velocity-dependent forces such as a spin-orbit force enter on the scene. Such forces are obviously outside the scope of the static model. Correspondingly, application of the latter to processes like pion production in nucleon-nucleon collisions is dubious, since at energies above the threshold for this process the picture cannot be described by a simple
static

At higher

model.
the static
structure.

Remarks. In summary, we can say that model is a theory with a reasonably transparent mathematical
21.4. Concluding
It is

remarkably successful in tying together data concerning various low-energy mesonic phenomena. The fact that the renormalized coupling constants determined from pion scattering and photoproduction and from nuclear forces agree within 1 5 per cent shows that quantum field theory is capable of penetrating into the subnuclear world. But one should not be blinded by this success. The model is an obvious simplification of the true state of affairs. A more accurate theory must also treat nucleons as quantized fields and has to face the Unfortunately, not only does this mean complications of relativity. additional computational difficulties, but the whole mathematical structure of such theories is unknown. 2
Disregarding these questions and guided by perturbation theory, we can postulate a simple analytic behavior for the scattering amplitude. 3 Using the resulting so-called Mandelstam representation, we can 4 deduce equations for the scattering amplitude which are generalizations of the Low equation we discussed. They contain not only all kinematical corrections but also antinucleon-nucleon pairs, pion-pion There ensue further branch lines and new channels at interactions, etc. higher energies. Since the coupling of the three internal pion states to the nucleon already makes an exact solution of the Low equation
impossible in the static model,
1

it

is

clear that these equations are

See

M.

J.

Moravcsik

in

G. R. Screaton
J.

Scottish Universities'

Summer

(ed.), "Dispersion Relations," p. 117, School, 1960, Oliver and Boyd, Edinburgh, 1961;

and

P. Cziffra,

M. H. MacGregor, M.

Moravcsik, and H. P. Stapp, Phys. Rev.,

114:880 (1959).
2 Except for one unrealistic case. See W. Thirring, Ann. phvs., 9:91 (1958), and Nuovo cimento, 9:1007 (1958); V. Glaser, Nuovo cimento, 9:1005 (1958). 3 See S. Mandelstam, Phys. Rev., 115:1741 (1959) and 115:1752 (1959). 4 See G. F. Chew in G. R. Screaton (ed.), "Dispersion Relations," p. 167, Scottish Universities' Summer School, 1960, Oliver and Boyd, Edinburgh, 1961.

NUCLEAR FORCES

257

However, neglecting meson-meson interexceedingly complicated. -+ oo, we obtain the actions and going to the limit of equations of the In this way the large effects which static model, when this limit exists. are correctly predicted by the static theory follow from a more funda-

mental approach. It is even hoped that in this way medium-sized effects can be unambiguously calculated. Certainly much further work is required before all the complications of relativity and quantum theory, taken together, are worked out.
Further Reading (on potentials)

M.

Taketani,
(1952).

S.

Machida, and

S.

Onuma, Progr.

Theoret. Phys. (Kyoto), 7:45

E.

K. A. Brueckner and K. M. Watson, Phys. Rev., 92:1023 (1953). M. Henley and M. A. Ruderman, Phys. Rev., 92:1036 (1953). S. Gartenhaus, Phys. Rev., 100:900 (1955). P. S. Signell and R. E. Marshak, Phys. Rev., 106:832 (1957) and 109:1229
(1958).

Appendix

The formulas compiled below express various quantities in the different representations used in the text. They are all equally important and exhibit different features of the theory. Coordinate space has the most intuitive appeal and was first used with cubic or spherical
boundary conditions. These are not realized in actual experiments and are considered only a mathematical aid. Correspondingly, we quickly passed to an infinite volume and specified certain initial conditions.
in plane waves diagonalizes energy, momentum, and In problems with a spherical source, the momentum of the field is no longer a constant, but the angular momentum is. In this case a spherical-wave expansion reduces the problem further. Several conventions are used for defining the field variables in these various We choose those for which the commutator of the representations. field operators is just equal to a Kronecker d for discrete variables and a Dirac d function for continuous ones. The advantage of this convention is that in the expressions for energy, momentum, etc., there is a sum or integral without further numerical factors. This is achieved at the expense of having several factors in the expression of the local field in terms of these variables.

The expansion
charge.

259

o
-h

JI
i

7
IS
63 Q

%
p
e

S
tr
r*

& S
u
f
t

1
"5

s
C O

^
e/3

^
55

u
G a
\

260

H
s

.2

1
3C

261

List of

Symbols

There are a few places in the text where the same symbol is used to denote two different physical quantities. We hope that in these cases the choice will be clear to the reader from the context in which the

symbol
first

is

used.
in

The numbers
appears.

parentheses denote the section in which the symbol

General Notation

Greek Symbols

263

264

LIST OF

SYMBOLS
Greek Symbols

Photon
Small region in momentum space Advanced Green's function Retarded Green's function Phase shift
Infinitesimally small positive quantity

(20.2)
(10.1)
(8.1)

Aret
),

(8.1) (8.2)

**)

(11.1) (20.2)

Polarization vector of electromagnetic


field

Ifi

Totally antisymmetric tensor Probability Transition probability

(5.1)

(6.3)
(8.2) (5.1)

Angle in expansion of field Constant in classical meson theory

(16.1) (16.2)

A
A

Scattering angle

Transformation matrix

in

charge space

(7.2) (5.4)

Wavelength Dimensional coupling constant Undetermined multiplier Renormalized dimensional coupling constant
Eigenvalues of the Born-approximation scattering matrix

01.1)
(17.4)
(12.2)

(18.5)

Mass of pion Part of nucleon magnetic


Frequency

(15.1)

moment
finite

(20.1)
(5.4)

Number
f
77

of particles in a

volume
for nucleon

(6.3)

Combined spin-isospin indices Momentum-density operator

(17.1)
(4.2)
(7.2) (8.1)

k.

/<k)

Pion Source distribution in space Source distribution in momentum space Cross section

(9.2)
(8.2)

o
T f

2x2 isospin matrices


Isospin operator when it operates proton in the Lee model

Spin operator

(15.1) (13.1)

on the

(13.3)

Klein-Gordon field (operator) i<ut ) Positive-frequency part of <^> (oc e~ of </> (oc e itot) Negative-frequency part Charged field operators oo Field at t
Field at
t

(4.1) (5.3) (5.3) (7.1)

= =

(8.1)
(8.1) (5.1)

-f oo

Azimuthal angle

(9.3)

Two-nucleon
Schrodinger

relative

wave function

(21.3)
(4.2)

field

operator

LIST OF

SYMBOLS

265

266

LIST

OF SYMBOLS
Other Symbols

Momentum-space wave function of field quantum


Meson-nucleon dimensionless coupling
constant in symmetrical pseudoscalar theory Reduced meson-nucleon coupling constant Renormalized meson-nucleon coupling constant Weighting function of g n Green's function Weighting function of t matrix Four-dimensional Fourier transform of the Green's function Coupling constant Renormalized coupling constant

(6.1)

(15.1)

(17.4)
(18.4)

(18.7)
(8.1)

(18.3)
(8.1)

g
gr

(9.1)

(13.6)

gu
(5

H
H(T) H*>

Function proportional to l//r (M) Weighting function of g v Hamiltonian Hankel function of the first kind
Local-energy-density operator Projected amplitude of / matrix
tf

(18.7)
(18.7)
(1.3)

(5.4)
(6.1)

(18.5)
<9-

*o

2)

2 3 |p(/c)| /[(27r) 7(A:)]

and analytic con-

(14.2)

tinuation of this function

Intensity

(6.3)

J
Ji
i

Total angular momentum Spherical Bessel function Current operator Abbreviation for j, a, k indices Conjugate variable to time

(16.1)
(5.1)
(7.1)

(18.2)
(8.1)
(1.3)

n\

Wave number, momentum Resonant momentum Source cutoff momentum


Length of line Lagrangian

(12.2) (15.3)
(1.3)

(4.2)
(5.1)

Orbital angular momentum operator certain function

(17.5)
(5.1) (4.2)

M
AM

Angular momentum Lagrangian density


Physical

mass of source Mechanical mass of source


in the

(9.2)
(9.2)

Matrices

(12.4)

Energy of physical neutron

Lee

(13.3)

model

LIST

OF SYMBOLS

267

268

LIST OF SYMBOLS

Other Symbols

Index

Adiabatic principle, 83, 96, 98/i., 219 Analytic properties of scattering matrices,
143, 144,201,202,207,211 Anderson, H. L., 164/1. Angular distribution, in photomeson production, 232/i., 236, 238, 240, 242 in pion scattering, 174, 175, 216, 217 Angular momentum, 34-40, 260 commutation, 36, 60, 64
definition, 35 eigenstates, 39, 176, 177

Born

approximation

(see

Perturbation

theory)

Born-Oppenheimer approximation, 25 Bosegas, 153, 154


Bose-Einstein statistics, 20, 26, 52, 88, 1 82, 1 86

Bound
52

states (see States)

Boundary conditions, for Green's function,


periodic, 5, 7, 25
spherical, 35, 38

Bra, 13

expansion of field in, 38, 39, 159, 259 in nuclear forces, 249, 251, 252 in photomeson production, 232/?., 236,
238, 240, 242 pion theory, 156, 158, 161, 170, 171, 182, 198, 199, 204 projection operators, 206 Antinucleons, 154, 155 annihilation, 224 Antisymmetric tensor, 36
in

Branch cuts of T matrix, 202 Bremsstrahlung, 98, 99 Brueckner, K. A., 257. Bumiller, F., 228, 229, 235
Canonical variables, 7 Capps, R. H., 233.
Castillejo,

Carlson-Lee, D., 245//. R. H., 144w., 21 In. Center-of-mass operator, 42, 43

Arnous,
Bardeen,

E., 125/f.

Charap, J. M., 253w. Charge, 60, 129

commutation
J.,

rules, 60, 61

\54n.

particles, 77, 78, 129 Barnes, S. W., 209/*., 21 On.

Bare

conservation, in Lee model, 129 in pion physics, 157

Beneventano, M., 245/i., 246/t. Bernardini, G., 245 Bethe, H. A., 118/i., 155/1., 157/f.,

degrees of freedom, 58-67, 1 26 (See also Degrees of freedom)


density, 60
184/i.,

of nucleon, 219, 221, 225, 226, 228,


229, 232-234
eigenstates, in Lee model, 129, 142 of vacuum, 62, 66 Charge independence in pion physics, 176,

198.
Blatt, J.

M.,

116/1., 117ii., 125/z., 136w.,209//.,

230/1., 247/i.

Bohr, A., 153/f. Bohr, N., 3/i., 26n. Born, M., 252/1.

216, 251 (See also Isospin)

269

270
Charged

INDEX
particles, 61,

62

Creation operators,

9,

26, 33, 38, 39, 75

creation operators, 62 point nature, 62, 63

Charged pions, 65 Chew, G. F., 212, 218w.,

Cross section (see Scattering) Crossing symmetry, 201, 202, 207, 212 Current, 60, 232-235, 237-241
245/*.,

248., 256.

commutation

relations for, 233, 237, 238

Chew-Low

plot, 209,

210

distribution of nucleon, 225-227

Christian, R., 186/i.


Cini, M., 218/i., 231//.

Cutcosky, R. E., 254 Cziffra, P., 256


1

Circular components, of isospin, of spin, 159

59
Dalitz, R. H., 144/1., 211/f. Dancoff, S. M., 184/i., 192., 196n., 231/t.

Classical behavior of particles, 56, 89 Classical treatment, 34

for nucleon potential, 249, 250 for pair theory, 100-1 1 1 for pion physics, 166-172, 192 angular momentum, 170, 171, 177 bound state, 171

Davidon, W. C, 164, 217 Debye, P., 6 Degrees of freedom, 9, 19, 27 internal, in Lee model, 126 in many dimensions, 63-65
in
in three
in

energy, 170, 171

equation of motion, 167-169 excited state, 167, 170-174, 214 neutral theory, 166 scattering, 172-175 source, 168, 169
strong coupling limit, 171

pion physics, 156 dimensions, 65-67 two dimensions, 58-63

Clebsch-Gordan

coefficients, 176, 181

Commutation
9, 30,

relations, canonical variables,

32

creation

and destruction operators,

10,

24, 25, 38, 75, 259

current, 233, 237, 238

de Hoffmann, F., 118., 157/1., 184., 198/1. Demeur, M., lOO/i. Destruction operators, 9, 26, 33, 38, 39, 75 Determinant, Fredholm, 104-106 Deuteron, 255 Dirac, P. A. M., 7, 12., 13, 63. Dirac notation, 13, 14 Dirac's 6 function, 7, 29 Displaced harmonic oscillator, 15, 22 Displacement generator, 36 Distinguishable particles, 58
Drell, S. D., 101/i., 163/1., 240. Dressing operator, 180, 181, 253, 254

distinguishable fields, 58, 62, 63


field

operators, 29, 32, 160 in interaction, 75, 86, 113

DuMond,
Dyson, F.
Eden, R.
215

J.

W.

M., 230w.
163.,
211/1.

isospin, 65, 160

J., 101/1., 144/f.,

particle density, 43,


spin, 160

44

J., lOO/i.

Compton scattering, 232, 247


Condon, E.
U., 67w., \15n.

Effective coupling constant, 164, 165, 214, Effective range, 209, 213

Constants of motion, 34, 59


for distinguishable particles, 64 for pion physics, 161, 162 Continuity equation, 60, 233, 234

Ehrenfest theorem, 42
Eigenstates, of angular momentum, 176, 177 of charge, 129, 142 of Hamiltonian, for coupled oscillators,
10, 19,

Continuous vibrating Cooper, L. N., 154/f.

line,

6-8

20

Coordinates, normal, 4 Core, repulsive, 155 Corson, D. R., 245 Coulomb field, 99

for free fields, 26, 39 for harmonic oscillator, 10-12

(See also Isospin ; States; specific theories,


e.g.,

Lee model)

Coupled harmonic Coupling constant,


215

oscillators, 4, 5,

23-29

effective, 164, 165, 214,

Eigenvalues of Hamiltonian, for coupled oscillators, 18-21 for free fields, 26, 39
for for

renormalized, in pion physics, 203, 209, 210, 223, 246, 255, 256 (See also Renormalization) Creation of particles, 93-99, 208, 232, 235-

harmonic oscillator, Lee model, 127

10, 11

for neutral static sources, 84, for pair theory, 120-122 for pion physics, 177, 178
;

8*5,

90, 91

247

(SeealsoEnergy Energy renormalization)

INDEX
Einstein, A., 6,

271

142

Free

field,

commutation

relations for field

Electric-dipole transitions, 234, 243, 245,

operators, 25, 29, 75

246
Electrodynamics, quantum, 77, 98,
235, 236
1

17, 154,

Electromagnetic field, 8, 28, 46, 60, 61, 90, 96-98, 100, 101, 125, 166, 232-248 Electron gas, 153, 154 Elementary particles, 154
Endt, P. M., lOOrt. Energy, of excited state in pion theory, 170,
171, 191, 196,

equations of motion, 29-32 expansion, in plane waves, 25, 259 in spherical waves, 38, 39, 259 Hamiltonian, 24, 30

Hamiltonian density, 49
Lagrangian, 31
linear

momentum, 35
density, 49

linear-momentum

213,214
1

of ground

state,

60

zero-point energy, 49 zero-point fluctuations, 27-29 Frequency, 7


negative

(See also Energy renormalization) of quanta in Lee model, 128 of resonance in pion theory, 173, 174, 196
transferred to field, 95, 96 (See also Eigenvalues; Hamiltonian)

and

field, 42,

positive 103, 104

components
44

of

commutation
Friedman, M. H.,

rules, 42,
186/i.,

240.

Energy renormalization,
131
in

in

Lee model, 127,


84-86,

Fubini, S., 2l8n., 223/1., 231n., 253n. Fulco, J. R., 155/1., 229/1. Functional derivative, 7, 29, 30

neutral
91

static-source theory,

in pair theory, 121 in pion theory, 160, 224, 225

Galilean invariance, 156

Gartenhaus,

S.,

257 n.
61

intermediate coupling, 187-189 numerical calculations, 196, 197

Gauge transformation,

perturbation theory, 183, 225 strong-coupling theory, 195, 196, 225 Tamm-Dancoff treatment, 185 Equations of motion, classical pion theory,

167-169 coupled oscillators, 4, 5, 23, 24 Euler, 30 Lee model, 128, 130-132


neutral static source, 71, 73
pair theory, 74, 102 pion theory, 160, 161

Gaussian cutoff for source, 165 Gaussian distribution, 15 Gell-Mann, M., 80/7., 202. Giacomelli, G.,209, 210 Glaser, V., 136., 256. Glicksman, M., 217 Goldberger, M. L., 80., 202. Goldschmidt-Clermont, Y., 246 Green's function, 71, 73
advanced, 72, 73 retarded, 72, 74

Ground

state (see Eigenstates;

Vacuum)

Excited state, 196,214 (See States)

Haber-Schaim, U., 182.


26
Halpern, F. R., \96n. Hamiltonian, continuous vibrating coupled oscillators, 4, 18
free field, 30,
line,

Fermi-Dirac
253

statistics,

Feynman diagrams,

203, 208, 215, 250, 251,

Fields (see specific fields, e.g., Free field) Fluctuation, in number of field quanta, 55-

260 harmonic oscillator, 9, 10 Lee model, 127, 128, 132, 133


neutral static scalar source, 84

57,88
in position of nucleon, 56 zero-point, 12, 20, 21, 27, 28,

89

Foldy, L. L.,

101/t., 163//.

pair theory, 119, 120 pion physics, 160, 176, 224, 250, 260 in intermediate coupling, 186, 188
in strong coupling, 192-194 radiation field, 235, 236 Schrodinger field, 33

Form

factor, 235

Fourier series, 7, 29 Fourier transform, 72, 96, 97, 102

W. R., 155/1., 229. Fredholm determinant, 104-106 Fredholm series, 103


Frazer,

Free

field,

23-31

several sources, 90, 250 Hamiltonian density, 49, 51 Hamilton's equations, 4, 7 Hankel function, 44, 73

angular

momentum,

35

Harlow,

F.,

192.

272

INDEX
oscillator,

Harmonic

9-17

Karzas,

displaced, 15, 16
eigenstates, 10-12 with external force, 71

Keck,

J.

W. J., 247 C, 245

Ket, 13

ground

state,

1 1

King, J., 209//., 210 Kinsey, K., 209n., 210


Klein, A., 124/z., 125n.

zero-point fluctuations, 12 Hamiltonian, 9-1 1 number of particles, 125


operators, commutation relations, 10 14 spatial representation, 11, 13,

Klein-Gordon equation, 24, 25 Klein-Gordon wave function, 3 Koba, Z., 78. Koester, L. S. 245
Kroll, N.,

time dependence, 12-17 two-dimensional, 59

240.

Kroll-Ruderman theorem, 240w.


Kronig, R., 153/j. Kruse, U. E., 164,217
Lagrangian, 29 canonical variables from, 29, 30
for distinguishable particles, 58, 63
field

wave packet, 15
Heisenberg representation, 13, 200 equations of motion, 13 Heitler-London approximation, 253 Henley, E. ML, 78/7., lOlw., 163/*., 186.,
257n. Hilbert spaces, 27 Hofstadter, R., 228w., 229/1., 235/7. Holladay, W. G., 233w. Hydrodynamics, 153

equations from, 29, 30

for free fields, 31,32

Hamiltonian from, 30
for neutral static-source theory, 75 for pair theory, 75, 100, 101 for pion-nucleon interaction, 158

Hydrogen atom, 88
Hyperons, 154, 155
Ida, M., 144//. Indefinite metric, 138

Lagrangian density, 29 Lagrangian multiplier, 187

Lamb, W.

E., \25n.

Induced emission, 142 Infrared divergence, 99


Interactions (see Source; Pair theory) e.g.,
specific theories,

Landau, L. D., 39., 153w.


Lee, T. D.,
78//.,

92n., \27n., 153/1.,

186.

Lee model,

101, 118, 126-150, 156, 182, 185,

190, 199, 200, 202, 203, 208, 211, 213

Interference between field quanta, 52-54,

charge conservation, 129

56 Intermediate-coupling approximation, 186192

commutation

relations for operators, 127,

128, 132, 133


definition, 126, 127

energy renormalization, 187-189


excited state, 190, 191

degrees of freedom, 126


eigenstates of charge, 129, 142

Hamiltonian, 186, 188


neutral pseudoscalar theory, 188-192 pion wavefunction, 186, 187, 189, 191

energy of quanta, 128 equations of motion, 128 solution, 130-132

renormalization constants, 192 strong coupling limit, 190, 191 Invariants (see Constants of motion) Isobars, 178

Low

equation, 139, 141, 144-146 modifications, 150 neutron, 130, 131, 133-136 phase shift, for TT~W, 140, 144
for 777;, 135-140 sign, 136, 142

(See also States, bound)


Isospin, 59, 65, 127, 157, 159 conservation, 158, 162, 164
eigenstates, 175, 176

proton, 129-131
relativistic version, 127,

projection operators onto, 206


in pion physics, 157, 158 Isotopic spin (see Isospin)

renormalization,

137 of coupling constant, 136, 137, 144, 149 of wave function, 132-134, 137, 138

Jacobsohn, B. A., 192.


Jeffreys, B.*S., 4n.

scattering, Born approximation, 137, 138 145, 146, 148-150

Jeffreys, H.,

4.

high-energy behavior, 145, 146, 148150

Kallen, G., 136., 138.

low-energy behavior, 149

136,

145,

148,

INDEX
Lee model, scattering matrix, 135-145
states, bare, 129,

273

Moravcsik,

M.

J., 245/7., 256/1.

130

Morpurgo, G.,

125/z.

continuum, 131
physical, 129-131, 134-136 scattering, 131, 132, 142, 143

Morrison, P., 155/7., 157/7. Mottelson, B. R., 153/7.


Negative frequency components of field, 42 Negatively charged particles (see Charged
particles)

T matrix,
virial

140-145 theorem, 134, 135


218/7.

Lehmann, H.,

Lewis, H. W., 78/i. Lifschitz, E. M., 39/7.,

153/7.

Lindenbaum,

S. J.,

209

Line, continuous vibrating, 6-8

of vibrating atoms, 4-6 Linear momentum (see Momentum) Lippmann, B. A., 80/7., 140/1. Localization of field quanta, 48, 50, 51 Lorentz group, 8, 154 Lorentz invariance, 8

Neutral pseudoscalar theory, 166-175, 188192 Neutral static-scalar theory, 71, 82-92, 150, 215 adiabatic theorem, 83 asymptotic fields, 75 energy, 84, 85, 90, 96 energy renormalization, 85, 86
field

equation, 71

fluctuations, 89, 90

Low,

F. E., 141/1., 153/7., 198/1., 212/7., 218.,


in

Green's function, 72, 73

248/1.

ground
Lee model, 139,
141, 144-

state, energy, 84, 85,

96

Low equation,
146
in in

momenta of particles, 87 number of particles, 86-89


spatial extension, 89

photomeson production, 237


pion physics, 198, 199, 208, 210-214, 256

McCormick, B. H., 124*., 125. McDonald, W. S., 245, 246 MacGregor, M. H., 256/7.
Machida,
S., 253/1.

wave function, 88 Hamiltonian, 86-89 eigenstates, 86 interpretation, 82, 83 Lagrangian, 75 scattering, 83, 86 several sources, 90-92
potential between, 91, 92
virial

Magnetic dipole transitions, 243, 246 Magnetic moment of nucleon, 219, 225, 230,
231

theorem, 89
(see

Neutron

Nucleon)

Many-body problem, 153


Margenau, H., 16/i. Marshak, R. E., 257/t.

Mass

renormalization

(see

Energy

re-

Nishijima, K., 196/7. Nishimura, K., 253/7. Nonrelativistic field (see Schrodinger field) Normal coordinates, 4, 5, 18

normalization)

Matrix multiplication, 104, Mesons, 154, 155


(See also Pions)
Mills, F. E., 245/1.

107/i., 110/7.

Novozhilov, lu. I. V., 254/7. Nuclear forces (see Potential) Nuclear surface waves, 1 53 Nucleon, 139 bare, 179, 180, 224, 228

Miyake, K., 209., 210. Miyazawa, H., 231. M011er, C, 106.

Moment
171

of inertia in pion theory, 166, 167,


5,

Momentum,

260

charge distribution, 219, 221, 225, 226, 228, 229 current distribution, 225-227 Feynman diagram, 215 intermediate-coupling theory, 186-192 magnetic moment, 219, 225, 230, 231

angular (see Angular

momentum)
64

commutation
definition, 35

rules, 36, 60,

mesons in, 180, 181, 220 number, 219, 224 normalization of wavefunction, 181
numerical method, 196, 197 perturbation theory, 183, 184 physical, 179, 180, 181, 224-231 quantization, 256 strong-coupling theory, 192-196 Tamm-Dancoff theory, 184-186

density, 43 eigenstates, 36 eigenvalues, 36, 37

of

field

quanta, 24, 34-40


distribution, 87

zero-point, 43, 49

Momentum

274

INDEX
particles,

Number, of

41-43, 75

charged, 62
density, 43, 44, 52, 54 fluctuation, 55

Pair theory, scattering, 103, 109-111, 113119

incoming, 76
for interacting fields, 76

high-energy behavior, 117 low-energy behavior, 117, 118 phase shift, 111, 114-117, 119, 122 resonance, 115-117, 119
several sources,
virtual particles,
1 1 1

outgoing, 76 in small volume, 76

122-125

of pions in nucleon, 219, 224 Numerical methods in pion theory, 196, 197
Oakley, D. C, 246 Omnes, R., 239/t.
One-particle states, 47-52

wave matrix,
Pais, A., 192/1.

104, 106-1J1

Parity, 40, 41, 155, 157, 158, 162, 236 conservation in pion physics, 162
Particles,

20

bare, 77, 78, 129


density, 99

Onuma,

S., 257/i.

Oppenheimer, J. R., 78., 252/z. Osborne, L. S., 246 Oscillators, coupled harmonic, 4, 5, 23-29 creation and destruction operators, 19, 21, 22, 24 displaced, 22 equation of motion, 5, 23, 24
generalized coordinates, ground state, 19
1

production, 93-99, 208, 232, 235-247


real, 77, 78,

99
122-125,

virtual, 77, 78, 86-89, 94, 97, 99,

155
(See also
Pauli,

Number, of particles; specific


3//.,

theories)

W.,

26/i.,

138.,

157/i.,

166/1.,

192ii., 196/1., 231/1.

Hamiltonian,

18,

23-25

Periodic boundary conditions, 5 Perturbation theory, 178/i.


for

eigenvalues, 18, 26
eigenvectors, 19, 26 harmonic (see Harmonic oscillator)

ground

state,

of neutral static-source

theory, 90

Lagrangian, 29

normal coordinates,

18,

particle interpretation,

23 20

wave packet, 22 zero-point motion, 20, 27, 28


Pair theory, 74, 100-125, 213 bound states, 103-106, 108, 109, 117 Hamiltonian with, 119, 120 operators, 113

of nucleon, 183, 184, 189, 196, 221, 222, 225 for nucleon-nucleon potential, 250, 251 for photomeson production, 236, 238 for pion scattering, 203, 204, 208, 209 Peterson, V. Z., 245, 246 Phase shift, 80, 81 in Lee model, 135-138, 139, 140, 144 in pair theory, 111, 114-119, 122 sign, 122
pion theory, 162, 205, 207, 208, 210, 214-216 relation to T matrix, 205
in

quantization with, 112 time dependence, 108 wave function, 109 classical treatment, 100-111 commutation relations for field operators, 113 energy of particles, 120 energy renormalization, 121 equations of motion, 102
field operators, 103, 104, 109, 112, 113,

Phonon, 20, 22 Photomeson production, 232, 235-247


amplitude, 238 angular distribution, 243, 244, 246, 247 angular momentum, 232., 236, 238, 240,

242
cross section, 241-247 dipole contribution to, electric, 234 ,243, 245, 246 magnetic, 243, 246
effects in, recoil,

122

Fredholm determinant, 104-106


Hamiltonian, 119, 120 Lagrangian, 75, 100, 101 in pion physics, 163 quanturrTmechanical treatment, 112-125 renormalization of coupling constant, 118 5 matrix, 114

245

rescattering, 239-241,

244

resonance, 244, 246

energy dependence, 242-244 Hamiltonian, 235, 236 Low-type equation for, 237 perturbation theory, 238

INDEX
Photon, 236
Pi

275
4

Quantization rules, 29-32, 75-78, 112, 256

mesons

(see Pions)

Quantum field, relation to classical field,

3,

Pines, D., 153.,

186.
Radiation field, 235, 236 Radius of nucleon, 229 Recoil effects in pion physics, 156, 198, 217, 245, 253 Relativistic field (see Free field)
Relativity, 8

Pion-nucleon interaction, 153-165 angular momentum states, 156-158, 163 constants of motion, 161, 162 effective strength, 163-165 equations of motion, 161, 162 Hamiltonian, 160 Lagrangian, 158 nonlinear, 155, 162 relativistic, 163 static theory, 1 56

Renormalization, of coupling constant, in Lee model, 136, 137, 144, 149

symmetry, 162, 181


Pion-pion forces, 155 Pions, 59, 63, 65-67, 91, 98 charge states, 158

203 of wave function in Lee model, 132-134,


137, 138

in pair theory, 1 in pion physics,

Renormalization constants in pion theory, 180-183, 203, 209, 210, 222-224


in intermediate coupling, 191

exchange between nucleons, 249, 252-255


parity, 157

production, 208 (See also Photomeson production) Plane wave, expansion, 37 Point sources (see Sources) Poisson bracket, 13, 35 Poisson law, 15, 16, 22, 55, 57, 87, 94, 95 Polarization in pion scattering, 217 Polaron, 153 Poles of T matrix, 202, 203
Positive frequency components, 42 Positively charged particles (see Charged
particles)

in perturbation theory, in 191

numerical calculations, 196, 197 184


strong coupling,
1

in

Tamm-Dancoff approximation, 186


55

Repulsive core,
in

Resonance scattering, in pair theory, 115-1 19


pion physics, 172-174, 196, 198, 213, 214, 216 Retherford, R. C., 125. Rose, B., 209/1., 210
Rose, M. E., 181. Rosenfeld, L., 3. Rotation, generators, 36 Rotational symmetry, 59, 60, 64

Potential,

Feynman diagrams,

250, 251, 253

isospin dependence, 251 between neutral static-scalar sources, 91,

Ruderman, M. A.,

240//., 257/1.

249 between nucleons, 249-257


angular

Sachs, R. G., 188., Salzman, G., 23 In.


Sartori, L.,

233.

momentum,

249, 251, 252

classical calculation, 249,

250

Scalar

field,

196. 24, 40
field;

one-pion exchange, 249, 250, 252, 255 perturbation theory, 250, 251 for point sources, 250
tensor, 250-252

(See also Free


theory)

Neutral static-scalar

Scattering, in Lee model, 131, 132, 135-150 in neutral static-scalar theory, 83, 86
in pair theory 109-111 of pions by nucleons, 158, 198-218

two-pion exchange, 249, 253, 257 Yukawa, 249 Production of pions, 208
(See also Photomeson production) Products, ordered, 49
Projection operator, onto

analytic properties of matrices, 201,

202,207,211
angular distribution, 174, 175, 216, 217 angular momentum, 198, 199 charge independence, 216

bound

states, 107,

220,221,223
onto eigenstates of S matrix, 80, 81, 205, 206
in intermediate-coupling

Chew-Low

plot, 209, classical treatment, 172-175

210

pion theory, 189

cross section, 156, 205w., 207, 208,

in strong-coupling pion theory, 195

210, 214-217, 223, 224


effective range,

Proton (see Nucleon) Pseudoscalar field, 40


Purcell, E.

209,213

M., 57.

203, 208 high-energy limit, 204

Feynman diagrams,

276

INDEX
Spherical harmonics, 37-39, 159, 160, 220 expansions in terms of, 37-39, 159, 160 (See also Angular momentum)

Scattering, of pions by nucleons, low-energy limit, 203, 207, 214-^216

Low equation, 198, 199, 208 perturbation theory, 203, 204, 208, 209 phase shifts, 205, 207-212, 216
in

Spin operator, 157,


Spitzer, R.,

158/1.,

159

196.

photomeson production, 239241, 244

Stapp, H. P., 256/1.


States, bare, 77, 78, 85-89, 129, 130

polarization, 217 recoil effects, 198, 217

bound, 79 in Lee model, 130, 131, 134, 135


in pair theory, 101, 109-106, 108, 109,

renormalized coupling constant, 203,


209, 210

resonance, 198,213,214,216 scattering matrix, 198-201, 205 source term, 199,200,213 7 matrix, 200-208, 212 Scattering matrix, 78-81 for neutral static source, 93-95 in pair theory, 114
in

112, 113, 117 in pion theory, 171, 178 excited, 167, 170, 171, 190, 191, 196,

214

(See also Energy)


ingoing, 79, 94 many-particle, 55-57, 199 metastable, 117
one-particle, 47-52 outgoing, 79, 94

pion physics, 198-200, 205


I., 18/i., 29/i., 37/i.,42/i.,85/i.,

Schiff, L.

157.,

physical, 77, 78, 86-89, 129-131, 134, 135,

236/f., 252/1.

160
scattering, 131, 132, 142, 143

Schrieffer, J. R., 154/1.

Schr6dinger field, 31-33, 127 angular momentum, 35 center-of-mass, 42 commutation rules, 32, 128
creation operator, 33 destruction operator, 33

two-particle, 52-54
virtual, 117

(See also Eigenstates; Stoppini, G., 245

Vacuum)

Strong-coupling theory, 171, 178, 189-191,

231.
energy renormalization, 195, 196, 225 excited state, 196, 214 Hamiltonian, 192-194 pion wave function, 194, 195 renormalization constants, 195 Sum symbol, 27

equation of motion, 32 Hamiltonian, 33 Lagrangian, 32 linear momentum, 35 number of particles, 42, 43 one-particle state, 41, 48
two-particle state, 52-54 Schrbdinger representation, 12 Schrodinger wave function, 3, 13 Schweber, S. S., 59/i., 118/1.

Symanzik, K.,

218/i.

T matrix,

Schwinger,
Scott,

J., 80/i., 140/i.

analytic properties, 143-144, 201203, 207, 212 high-energy behavior, 145, 148-150

M., 246

Segre, E., 224/z. Self-energy (see Energy renormalization) Serber, R., 192*. Shortley, G. H., 67/i., 175/i. Signell, P. S., 257/1.

in Lee model, 140-145 low-energy behavior, 145, 146, 148-150, 203, 207 in

pion theory, 200-208, 212, 220 branch cuts, 202


crossing symmetry, 201, 202, 207, 212
diagonalization, 204-208 one-pion subspace, 206 relation to phase shift, 205

Sources for pions, 158, 199, 223

160/1., 163, 168, 169,

cutoff, 164, 165,213,219 point, 72, 90


in

zero-energy behavior, 203, 207


1 1 1
,

Lee model, 137


1 1 8,

in pair theory, 74, 100, 101 ,

119

Takeda, G., 78. Taketani, M., 251n.

pion physics, 250 several, 90-92, 111,249 time-dependent, 87, 95-98


in

Tamm,

I.

E., 155/z., 184/z.

Tamm-Dancoff approximation, 184-187,


211/z.

(See also specific theories, e.g., Neutral static-scalar theory)

Sparnaay,

M.

J.,

\22n.

energy renormalization, 185 pion wave function, 185 renormalization constants, 186

INDEX
Tau,
L.,

277

245
J.

Wada, W. W.,
G., 245

175/i.

Teasdale,

Tensor

force,

250-252
153/t.

Thellung, A.,

Thirring, W., 28.,

46.,

63/i.,

73/?.,

96n.,

125., 182., 223., 231.

Thomson scattering, 247 Time dependence, 12-17,


of of of of of

21,
1

22
3

bound

state,

108

Heisenberg operators,

operators in interacting fields, 76 SchrSdinger state vector, 12

Walker, R. L., 245, 246 Watson, G. N., 37/i., 44/z. Watson, K. M., 257w. Watson, W. K. R., 247 Wave function, of bound state, 108, 109 of quanta in ground state, 88, 124 (See also Nucleon) of two nucleons, 255 Wave matrix, 78, 81, 104, 106-111, 132 Wave packet, 15-17, 22 Wave property of field quanta, 56, 57

source, 87, 95-98 Tollestrup, A. V., 245

Wave

velocity, 7

Waves, outgoing, 97
Weisskopf, V.
247/1.

Tomonaga,

S., 92n.,

186.

R,

116/f.,

136.,

209/1., 230/t.,

intermediate coupling (see Intermediate-coupling approximation) Touschek, B. F., 125/z.

Tomonaga
Toyoda,

Welton, T. A., 28 Wentzel, G., 35n., 92n.,


154/1.

103/i., lll/i., 125/1.,

T., 253/1.

Transition probability, 79 Transition rate, 80

Wick, G. C.,

146/1., 218/1.

Uncertainty principle, 12 Unitarity of wave matrix, 107, 108

Width of resonance, 115-117, 174, 214 Wightman, A. S., 59. Wigner-Eckart theorem, 181.
Worlock, R. M., 245 Wouthuysen, S., 78.

Vacuum,

39, 47,

49

bare, 77
polarization, 125

Van Hove, L., 92n., 150/1. Van Kampen, N. G., 120*.


Variational treatment for state function,

Yang, C. N., 154//. Yearian, M. R., 228n., 229., 235/f. Yukawa cutoff, 165

Yukawa

potential, 91,

249

186-196 Velte, J. I., 245 Vibrating atoms,

4-6 Vibrating line, continuous, 6-8 Virial theorem, for Lee model, 134, 135 for neutral static-scalar theory, 89
line of,

for pair theory, 125 for pion theory, 222 Virtual particles (see Particles)

Zachariasen, F., 240., 247 n. Zero-energy behavior of T matrix, 145, 149, 150, 203, 207 Zero-point energy, 27, 49 Zero-point fluctuations, 12, 20, 21, 27, 28, 89 Zero-point momentum, 43, 49

Zimmerman, W.,
Zitterbewegung,

218/t.

63/i.