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.
Consulting Editor
Allis
and Berlin
Thermodynamics and
Statistical
Mechanics
Becker
Clark
Collin
Evans
Nuclear Physics Gurney Introduction to Statistical Mechanics Hall Introduction to Electron Microscopy
Green
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 Highfrequency Measurements Kennard Kinetic Theory of Gases Lane Superfluid Physics
Leighton
Lindsay
Livingston
and Blewett
Particle Accelerators
Middleton
Morse
Muskat
Present
An
Methods of Theoretical Physics Physical Principles of Oil Production Kinetic Theory of Gases
Introduction to
Read
Schiff Seitz
Slater
Dislocations in Crystals
Modern
Physics
Slater
Slater
Slater Slater
Structure, Vol.
Structure, Vol. II
Electromagnetic Theory
White
The
G.
late F.
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.
of Washington
WALTER THIRRING
Director
Institute for Theoretical Physics
University
of Vienna
INC.
1962
Toronto
London
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 BoseEinstein statistics are examined. Observables, invariants of the field, and internal symmetries are dis
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 quantummechanical points of view. In the third part both the mathematical tools and the physical insight
acquired in earlier chapters are applied to lowenergy 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 pionnucleon 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 pionnucleon 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 quantummechanical 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.
Contents
Vlll
CONTENTS
7.
CHAPTER
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
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
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.
11.1
11.2
11.3
Bound
States
Behavior of the
Scattering
Wave
Matrix
a
. . .
106
109
11.4
CHAPTER
12.
.112
112
113
12.1
Bound
12.2
12.3
Scattering
.119
122
12.4
CHAPTER
13.
.126
126
13.1
13.2
13.3
.127
130
131
13.4
13.5
13.6
.132
135
CHAPTER
14.
14.1
Scattering:
139
139
14.2
14.3
.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
.
16.1
16.2
16.3
CHAPTER
17.
179
17.1
179
183*
17.2
17.3
17.4
17.5
184
. . .
.186 .192
196
198
17.6
Numerical Methods
CHAPTER
18.
PION SCATTERING
Introduction
18.1
18.2 18.3
The
18.4
18.5
Scattering
18.6
18.7
Diagonalization of the T Matrix Relation of Low Equations to Experiment Approximate Solution of the Low Equation
....
.....
. . .
18.8
Summary
CHAPTER
19.
19.1
19.2
19.3
219 Expectation Value of the Field .220 Groundstate Expectation Value of Observables 222 Renormalization Constants and Other Parameters of the Static Model
19.4
19.5
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
Quantummechanical Comparison with Experiment Concluding Remarks
Static Potential,
21.1
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 KleinGordon 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, McGrawHill 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
(elliptic,
1 hyperbolic, etc. )
show
the
same
Quantum
may
Accordingly, we may, for a first orientation, consider ihe simplest system of this type, namely, a vibrating line of atoms. Furthermore,
we
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
?n2
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
we
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,
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
we may
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)
The
solution
QM = (WO) cos
(0)
co a t
+ +
2
sin
cos [s(n
 n') 
e^]
sin [s(n
 n') 
o>^]
)
0) 8
1
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 wellknown 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.
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.
an
infinite force
between neighboring
As a consequence,
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
remain
finite.
INTRODUCTION
limit
of
(1.5),
CL
dx
\
d Jo o
we
find
from
(1.9), for
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 )
d12)
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
1),
we have
tives
(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
are generalized to
"
"~
~~
"
*
Jo
*
dx'
~~ V
dq(x)
dx
dx 2
SH
1
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 fourdimensional 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 spacetime 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
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,
discussions.
Our problem
H=
the
4(p
f
o>V)
are
(2.1)
now
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
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
10
FREE FIELDS
algebraic
method
is
more convenient.
We
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
H^utfa +
Since the position and
Energy
to
(2.6)
momentum
and
tf>
[a,//]
=
the
a>
(2 ?)
coa
The form of
relations
commutation
of
//.
function
9
E (H This
E)y E
= 0,
we
find
tells
us that ay E
is another belonging to
a>.
Similarly,
it
Fig.2.1. Harmoniooscillatorpotential
wave function
yi
(o).
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
is
normal
fvtnfv:! 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
this representation
p =
idjdq,
we have
and
it
is
+*
The higher
excited states v* are then obtained
(2.13)
differential operator
af
(wq
d/dq)(2a>)~*.
the groundstate wave function is of the order o>~ J ; this expresses the quantummechanical zeropoint 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
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).
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.
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 quantummechanical 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 timedependent according to operators
V<0
(Schrodinger representation).
state vector constant
t It is
e^'tfO)
is
(2.17)
and
Another
to consider the
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,
13
= e iHi <D(Q)e' =
ttt
(2.18)
(Heisenberg representation).
tion values,
y>*(f)0(OMO
and
to the
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)
gives the
same expression
Because in our problems the classical equations be of a wellknown 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
>
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')
is
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. lIIL
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)
wave function
in
momentum
if
we
by
(q'\ 9
as
n)
= *(q'
n)
= ^ 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
find
(/?'
q') as follows:
q\q'}=q'\q'}
(P'\q\q')j(p'\q\p")(p"\q')dp"=q'(p'\q')
In
momentum
is
where
6' is
the
first
We
therefore obtain1
oc
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)
= 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 timeindependent; 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

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

is
<0
nl
a"
16
FREE FIELDS
set
of states

n}
we
find
and
(2.25)
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.
excited state in

d) follows a
mean
cod 2
value of
I
displacement
\2
^zeropoint fluctuation/
We
1
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.
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
(227)
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)
is the appropriate quantummechanical 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
that
0, e.g.,
a  a
(t
shown
after
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 quantummechani
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)
it is
new
first,
.?<$
1
*=?<"'"'*
1
(3  3)
See, e.g., L.
Inc.,
I.
Schiff,
Company,
New
35,
McGrawHill Book
York, 1955.
COUPLED OSCILLATORS
19
become
= [a,fi j =
[fizAJ
'<5*,m
[P,,P
j=o
= qn
,
(34)
the conditions (1.4)
Q. a
Inserting the
Ql
P. s
= P]
we
find
1
(3.5)
new
+ <aJfi.fi.
>
we now
(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):
(2.6),
!..
i)
+D
(3.10)
its
The
state
of lowest energy
0>
is
a,0>
(3.11)
20
for
FREE FIELDS
all s.
It
=2K
The
general eigenvalue
is
(3.12)
given by
(3.13)
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 BoseEinstein 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.
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 zeropoint motion and quantum From (3.12) energy, manifest themselves in our vibrating string. and (3.6), we see that the zeropoint 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 zeropoint 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 zeropoint oscillation of the nth atom in the ground state of the system, we find, since (0 q n 0) 0,
this property.
It is
conventionally
(A<7 M)
^<0^
s,s'
>
i
i(ss')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
is
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
and
q n (t)
*"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)
co s
dn
de i8 n
'
(3.18), with the dn assuming the Calling this state c<v>, we have

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 timedependent solution we find
<(0
V>
w (0
exp
(q n
'
by
(3.20)
it is
qn
d)
= d cos
(s'
avO
If
we want
this classicallike
motion to be observable,
necessary
larger than the zeropoint 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 quantummechanical 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
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
+Q
23
(4 3)
'
24
FREE FIELDS
(1.5),
'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)
The introduction of
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 threedimensional case. In the mechanical model of a displacement field which we had in mind so far, the general threedimensional case is somewhat more complicated, since the field is then a vector and has three components. However, by only allowing displacements of a threedimensional
ticles
.
3,
a> k .
we
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
pion.
is
//(4.6)
which
1
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 threedimensional
periodicity
condition
+ L, y, z, = #*, y + L,z,t) =
<f>(x,
y,z
+ L,t) =
(4.6)
ftx, y,
2, t)
(4.7)
can be
satisfied in
9(r,t).
j
^ >
(48)
Iff
i,
*
*)
Hy
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
(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 KleinGordon 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
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 FermiDirac statistics. The kinds of fields that correspond to particles with halfoddintegral 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, McGrawHill 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.
many
we
shall
now
discuss.
many
<o
energy
diverges.
the energy, since at this moment we do not know what the zeropoint 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 zeropoint 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
+
oo,
where we
(4.14)
The
with an
norm when
connected with the fact that <(r) gives a state 1 applied to any state with finite energy.
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)
state.
28
FREE FIELDS
To
see this,
we
and obtain
\ (0
S ini
I
0> /
= (27r)v )
flMrr')
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
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
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 smallamplitude, ground vibration of the electron, whereas the Coulomb field acts highfrequency 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"
cm
is
electron in the
state.
Compton wavelength of
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
.
W.
Thirring, "Principles of
Press,
Inc.,
New
FIELDS
29
In the mesodynamic application which we shall discuss, the vacuum fluctuations of the field will be particularly important, because the mesonnucleon 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
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)
')
< 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
otherwise.
For the
a >
and
(4.16)
is
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
See, e.g., L.
Schiff,
"Quantum Mechanics," 2d
York, 1955.
McGrawHill
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
0+
vt
wW
UQ)
By means of the
functional derivatives
on
o
the classical
form
dL
dL
(4.18)
is
dL
For the Hamiltonian and the commutation relations we postulate accordingly the general formulas
^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
Lagrangian
(4.20)
rr(r,r) is
With
seen to be
and hence this prescription leads and commutation rules (4.4) and
to the
(4.16).
(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 timedependent 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,
and
J =q J ^q +
f
\q
j= J & = _/J
/
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
derivative.
2
J and
32
FREE FIELDS
rules derived
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 firstorder equations present
of motion for a nonhermitian
field.
f
y>
&
is
V  V V) it
"
m Vy Vy]
(4.21)
are 3
2
V w
2m
t
'
=
2m
canonically conjugate to
7T
f y and ^ are
?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
We
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 KleinGordon 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
The Schrodinger
relativistic
field
case
is
actually
the
KleinGordon 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
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
tt
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
<(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
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,//][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 KleinGordon equation. However, a calculation shows that P and L fail to commute; in fact, the simple
commutation
relations
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.
.
components of
are
worked
(5.7)
/6,,,L,
we
obtain, by our
(5.8)
P= 2 Xk k
As we found
earlier, the
>
^

>
(alr (al)
n*
'
'
'
\
0>
2
is
also an eigenstate of P
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
" J^ ^, k
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 particledestruction 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
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
U k(r) =
l
The expansion
(5.9)
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>
rx
( 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,
Inc.,
New
York, 1955.
38
FREE FIELDS
field
and of the
equations
f*dr
l r*Vk(r)V k ,(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 threedimensional 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)
we
see that
With
the help of
we
=
We
9
(5.H)
can readily verify that (5.11) results in the correct commutation properties of </>(* t). In analogy to our planewave 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
keep henceforth.
It relates
discrete &space quantities to continuous Arspace ones. In the former, destruction operators are written as ak or aklm9 whereas in the latter they appear as a(k) or
OBSERVABLES
39
=2 k,i,m
^u
k
.
i
(5.100)
with
the discrete
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
HE 2
and for
4i.m**,i.>
(5.12)
n
1,m,k
L3
imYi
(Or) (p r),
we
get
(5.13)
L3
The vacuum
is
= I l,m,k
fl*.i,m
*.i,m*.i,mw
new
variables
by
(5.14)
0)
In the limit of
>
o>,
in
(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 threecomponent
.
integers as eigenvalues.
To build up eigenstates of a given total momentum, we must properly combine the singleparticle 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
field, is
zero because of
"Quantum Mechanics," chap. IV, AddisonWesley Publishing Company, Reading, Mass., 1958.
40
FREE FIELDS
~> 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:
&+0>\
Also,
0>\0> +
(5.14a)
is
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
<Kr,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)aV</>(r,t)
with
2>l
This
momentum
Since
P.
for a reflection.
OBSERVABLES
41
we
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.
(515)
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 angularmomentum z component m)
:
Thus
present.
can be called the operator for the total number of particles We obviously have
[//,AT]
(5.17)
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
21
so
that, if fotft]
1,
ei^aae inaa
,
__ 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?.
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
field the
more
familiar expression
=jd*x
y'Or.OvfcO
(520)
In elementary wave mechanics this is put equal to 1, which means, in our present language, that there we consider only oneparticle 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,
pany, Inc.,
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)
N(r,0
= vfaOvM
(+)
(r,f)
(5.26)
relativistic one,
AT(r,0
i[^>(r,0*
it
 ^(r.O^fr.O]
(5.27)
[AWO,
IWlr =
(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
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
>
u
/
/
(+)
(r,0,
.
^'(iV)]^,'
=S
k
= iA (+) (r ~
2co
r')
(5.29a)
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
commutator behaves
like
a Hankel function
ro
4Trr \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 )^ r3rJoo G.TSL Watson, "A Treatise on the Theory of Bessel Functions," 2d ed.,
chaps.
3, 6, 7,
Cambridge University
Press,
New
OBSERVABLES
45
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 oneparticle 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
<
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
<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
See
W.
Inc.,
New
Thirring, "Principles of
Press,
York, 1958.
CHAPTER
States
The states we have been 6.1. Vacuum and Oneparticle 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 oneparticle 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,
we can
write
(6.1)
v (r,00>
We
0> or a]am
0} are
or
since these states are timeindependent if they are eigenstates of the Hamiltonian. The normalization of the oneparticle 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
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
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) JV1} = ^(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')
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]
timedependent.
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 twoparticle 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)<
(+)
>(r)
vacuum expectation
values.
</>
This
(
is
<+)
accomplished by writing
all
ones, e.g.,
(6 5)
'
so that the
p(r)
momentum
(
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 zeropoint 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),
(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
), 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 oneparticle 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
we have
because
to introduce
Fk =
k (2o>)*/
and
w)*
L* J
LWfry* ^
Thus, at
= 0,
we have
I
1 >
= 2 Aal
k
0>
f <* r F(r)^>(r)
0}
(6.6)
F(r)
but rather to
<0
F*(r)F(r')</>
M
(r)<f>
\r
0)
(+)
(r
This
is
i
J
F*(r)A
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 oneparticle 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
0,
the origin. 1
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
5.
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. Twoparticle States. In order to get a feeling for some of the consequences of the BoseEinstein 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
by the additional
effects discussed
above.
The
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
for
<2
2>
=
in
frt /"(r^C/fc.*)
+ /(r^)] =
(6.11)
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 twoparticle state for which /(r^Fg) is of the form f&Jftfa),
We
Furthermore, we assume that each /(r,) is normalized to unity, e.g., 2 3 jW r,/(r,) = 1, so that the correctly normalized overall distribution function /is
f (r r\ /(ri ra)
'
with
^
1
(//,)
is
frtftoftf)
it is
(6.13)
If this state
if it is
timeindependent,
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
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 vt = 0. We see that if the two wave functions
\
= 0,
is
just the
sum of
the
aN(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 twoparticle 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
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 aa 2 scattering are shown in part a. The intensities without and with interference appear in parts b and c, respectively.
single particle
itself,
which
is
position principle in
particles
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 centerofmasssystem 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.
+/
STATES
1
55
6.3.
Manyparticle States.
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
v\
where
v is the
= 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
wparticle state
which
is
represented by
n)
=
(6.17)
with 1
(//,)
the
dt ,
(6.18)
With
same methods
*=
(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
take the /'s to be orthogonal in order to separate the effect trie one discussed above.
56
If
FREE FIELDS
= e*k
'*
/", we get
(6.20)
*5
as
was
to be expected.
Nj
=
Jv
=
we
find in the usual
<n
N,
+
3
Jv
\<Pr
rfV
vVtyVOvW')
(621)
way
f</ r
tV*:
/;
*; n>

=*+
JVJ
(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 righthand
<
1,
the particles
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.
>
The
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
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
>
1,
(625)
in
agreement with
(6.23).
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
See E.
M.
CHAPTER
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,
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
Consider, for example, two hermitian KleinGordon 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 twoparticle 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)
now be
symmetric, because
58
59
Hence particles of different fields do not obey Bose statistics, irrespective of whether or not they have the same spacetime 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 twodimensional 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
,
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 twodimensional space, but this space has nothing to do with our spacetime continuum. Nevertheless, the formal analogy suggests that fa and fa are the components of a twodimensional 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 spacetime continuum, the constants are the generators of the infinitesimal transformations, and there are as many space" with which there
will
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
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
components
<f> t
under
This
(7.5),
or
[c>w]
tells
[e P]
,
[e>L]
(7 9)
.
us inversely that
is
in particular,
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
by a continuity equation*
It
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
an
electric
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
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(00+
~~
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
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,
//,
We
differ
ential operator.
Since
we do not wish
the
vacuum
state to carry
any charge, we
(7.15)
^
require that
C0>
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
</>
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
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 numberofparticles operator
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
m~ 3
Tl
.
commute with
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 oneparticle
at present,
perform
to the charge quanta. 1.2. Three and More Degrees of Freedom. The rotational invariance of the twodimensional 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 spacetime properties seem to be realized in nature. + For instance, the mesons K~, K correspond to four spinzero 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
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.
quantum field theory, see W. Thirring, "Principles of Quantum Electrodynamics," Academic Press, Inc., New York, 1958.
2
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
timeindependent and
satisfies
AA =1
f
and
A*
=A
dcp(r)
(7.21)
We
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
.
^ = _^w
must be
real
^w* = gy
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
is
where
(r)
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 spacetime 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
65
and each
(r)
is
is
We
.
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
'
]^
(727)
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
are
we
shall discuss in
more
A
1\
0\
J (l)
10
\0
(2)
10001
\l
O/
(3)
= 11
\
1
O/
O/
(7.29)
They
where
satisfy the
commutation
rules
nt is
The
three
2 sponding to these matrices are called the isospin commutation relations as the angularmomentum
and
/
representation
[,.,] = </"
(r) t

As
1
for
Q we demand
9
that
0}
We
use
TT
66
FREE FIELDS
the
It is consistent that
vacuum
t,
is
common
also
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
<,
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 onemeson 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
'^oOi^)
0)
WlW^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
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 internalspace rotational symmetry is independent of spacetime = 1, 2, 3. it follows that 9(rlt r % ) is independent of /
67
states can also be formed by standard angularmomentum 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 threedimensional 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
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 pionnucleon interaction, which we shall study in the last
We
So
far
the
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 wellknown 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
(8.1)
by superposition:
t
=
fdt'
d3 r' G(r
r',
W,O
2
(8.4)
(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
,
is,
,k)
(8.5)
<5(r)
krJ[( 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 wellknown 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
>
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
two
different
boundary
we can
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 spacetime 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
<
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
characterized
(8.11)
which
</>
corresponds to
<
in
and
X*0 in out
<
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 timeindependent 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
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
L by
the addition
(8.13)
and
respectively.
L'
(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,
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
out
<
and
t
iu
<
(8.15) implies
* i
in/
ir
_>^\
^i\R/
(p
.
oo).
ir
^^^
10 ir
J&3/..
.'\
)
/0
^ ^. x
(8.16)
obey the homogeneous field equations and can therefore be expressed in terms of timeindependent operators A k and Bk in the familiar form
<
out
4
(2WL )
3 t 31
Our development
From
(8.16)
we
commutation
rules
and B,
(J9 )
76
SOLUBLE INTERACTIONS
in the limits
first
oo.
The compatibility of
in
</>
(8.
5)
and
case where
and
oui
</>
differ
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)
commutation
relations at
K(0, 4(0]
<5
3 ' k k
,
(819)
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 freefield 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 timedependent, 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
 \ 
is
we only have
the
source which
is
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)
(<j>
stays
finite).
Since 7V out
is
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
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
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
<
The
out
differ
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
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
out, n k >
L (frfy*
k

out, 0)
The ?
out, n k >
[//,5k ]
out, n k >
co0k
N(t)
0)
(W
k0
states.
0)
k)
H
1
(0
= 
They
o)
k (0
k)
It
would have
fc
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
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 socalled "S matrix," or "scattering matrix," which plays a crucial role in modern field theory.
We
the
can define
it
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 timeindependent. From (8.22) we infer A k S 0, out) =0 5 0, out) = 0, in) (8.23)

Hence an equivalent
is
Sk ;,..,Cici ..... 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
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
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
a final state/
1
develops from
found to be
i?,<
= 47T
<5K
co,)!},!
( 8  26 >
We
differs
80
SOLUBLE INTERACTIONS
is to observe that it time of interaction in eigenstates of the
<5
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 Joo
I
d(o><
^/)
/oo
dt
Wn
by
we
^27r^co )T
/
/i.
(8.27)
The
out of the
initial state
2
can be written as
ti
a> f )
\Tfi
= 2 Im T
(8.28)
a consequence of (8.22),
i
/
SfjS/i
"<y
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.
ty
(A)
= g(<Wsi:/)S $? A
S(^^c
c S^Kco^S^
(A}
?
}
(829)
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 angularmomentum represen
where
fies
tation,
and
co is
S

fc, /,
m)
e^
2
(fc)
k,

/,
m)
is
to consider
discussion, see
M. GellMann and M.
k\
/',
w'>
rrg(a>)d(a>
 />V<W
J.
see, e.g., B.
A. Lippmann and
GENERAL ORIENTATION
or, in
81
Bum =
In the above case,
S~*A Um S
f&Aum
(8 32)
we
get
si 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.
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
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
Joo
S k
/I
\TT
m*
KT
we can
00
J.j"
^ A ret, A
(r '
^c
fe2
e*'
+m
=
2
Yf
00
\^j J
p 4im)
is
k dk eikr
Jao (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
_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),
becomes
/*
*~
tk(r
r')
(*t
e~ aU/l sin
JM for
sin cat
2
tt>
a)(t
t')
dt'
k D
e**
<
>
_
**
Hj a2
,
2a
'
a^
for
< m,
oo
A ret (rr',rOp(r',0^
k
CO*
,m
Irr'l
(9<36)
^jTT7
which agrees with
limit.
(9.16).
We
is
for
Aadv
in this
which
to be interpreted as
> nr\
one and we
sA
ret
(r
r', t
(9.4)
We
in
<
out
<
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
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
In (9.5)
we have denoted
/>(r)
by
/> k ,
r)e~
d3 r
(9.6)
and
its
complex conjugate by
ln
<
/>.
We
H in
terms of
the operators
If
we make use of
<5
3
(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
<
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
__a 2 T
wlrr'l
r
<
(976)
p(r')
r
That
a c number, commuting with in was to be anticipated. In and out have the freefield 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)
2 This can also be shown in momentum space. In the following, it is to be understood that the zeropoint energy of the vacuum is subtracted from //.
STATIC SOURCE
85
reduce to //
// (in)
(in)
(in)
<f>
.
We
note that
In
and
and
we
= 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
(a)
Fig. 9.1.
is
shown in part a and that of in part b. The energy shift between the two
normal potential. That the physical groundstate energy <^ is less than zero shows that the problem considered belongs to the wide class of This is always interactions which decrease the groundstate 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, McGrawHill 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
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

in,
0)
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
<
(f>
'
2 3
(2ro
L)
from which we
infer that
This will be
made
STATIC SOURCE
87
Therefore (9.11)
tells
us that
ak

in, 0)
gj
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
(*,*,
(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)
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)
The Fourier transform of the hydrogenatom groundstate 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>
=
!*ifll0)
is
(9.18)
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
momentum
but to ne~
fi
<
are simple products, which shows that the particles are uncorreFor a lated except for effects due to the BoseEinstein 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'(rr')
or
f = g \d*r'
J
glrr'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,
<f>
binding energy for these virtual parThe meaning of the above ticles.
source.
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 groundstate 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
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
(in
Chap.
3),
>
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
electric field
of an elemen
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^
,
(9 22)
.
47T
fcmin
el ,
(e.g.,
1.
kmin
~ l/r
b)
we
obtain,
<
ies
However, charged macroscopic bodsituation n > 1. For the mesonnucleon 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 groundstate 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.
(2)
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 pionnucleon <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
arbitrary strength,
=l
r<)
(926)
we
obtain
W) = I *P 2
i=i
~m\rir }
:, 4?r
\

(927)
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
According
and
outgoing
field is
<
given by
out
ln
<
r', t
W.O
oo
(iO0
where
A(r,0
= S(O k
The
of
I
for ~oo
< <
t
(10.2)
momentum
'
*
A(r
r', t
= S^ J k
I
S1
" "** ~~
* )
a)
X r '^') &*' di
'
where
/> k
is
/.
Pk(^o)
It
=)
Xr.
then reads
Since Bk obeys the freefield 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
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
momentum
interval
with a
fc_ $
We
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
of
this
e~ n/ exp
*k (o>)fl k
exp
= exp
WaOrfM).
2
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
S can
A^ and A[
2 bk
this expression is consistent
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 angularmomentum zero terms. That is, if an expansion of B k is made in terms of the
angularmomentum 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 timedependent 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
out, <,)(out,
n ki
in, 0)
= 22 "k, w
ki
<
in
out, n k
Since
{
in,
2 out, w ki }
is
(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
<
(quantummechanically,
expectation value
is
zero).
By
96
SOLUBLE INTERACTIONS
and use of
(8.9),
we can
AE =
f
/*>
dt
J\d*r
1A
dt
Iet
ret
(r
r', t
O
A
<')]
(ri',tt')tf \rT', 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
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
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 selfenergy f is found to be the energy change obtained by switching the source on or off.
10.2. Specific
Examples.
It is clear
from
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
in
which case the number of particles is proportional to the time during which the source radiates. For a < m the fourdimensional Fourier
1
See, e.g.,
W.
Academic
2
Press, Inc.,
static
Quantum Electrodynamics,"
chap.
2,
Thus, the
source (which
is
frequencies.
PRODUCTION OF PARTICLES
transform of the source,
/> k (co),
97
we can
write
*/c
d(coQ
2(0
M)
'
(27T)
=
4xr
<S**
a
out
<
if
GO.*)
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
(r0in
 [cos (r 0} =
477T
/,
t)a)
~ a]r ~ il
cos (r f
t)a>
e~ alr+t]
~\
For large
persists.
positive
only the
first
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 righthand 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 (realparticle) 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
,
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
/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
&
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 pionnucleon 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
its
velocity: J
rfr,0
f p(r)
for
(p(r
 vr)
.
for
< >
(10.12)
The
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
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 wellknown 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 selfenergy (in r space) is given by J d*r, where ( is the
&
.
and
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
m ^ 0,
since then
</>
~ e mr
/r
and has a
finite
range.
From
(9.22),
Taking
r, n t n
for rmax
 lO^V
37
~ 10+
cm
However,
nmiveree
~ 10
27
cm.
CHAPTER
11
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
(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 manybody 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)
(11.1) with
3
=
=
<5
(rr>(r,0
2
(H3)
p(r,f)
y<r,0
is
more complex;
See R.
I,
J.
vol.
chap.
1,
Eden, in P. M. Endt and M. Demeur (eds.), "Nuclear Reactions," NorthHolland Publishing Co., Amsterdam, 1959.
100
101
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.
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
as
we
shall see in
13.
it
concepts which we
scattering
shall encounter in
pion physics.
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')
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
S.
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
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/)
this
To
equation has the usual form of a linear it, we introduce the fourdimensional
<f>:
J
K(r,r>'
k '' r/
d*r d r
The reason
that the fourdimensional Fourier transform, rather than the threedimensional 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 fourdimensional Fourier transforms of Aret or Aadv as given by (8.6) and (8.7), we obtain 3
,
<x>)
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
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)
(11.86)
p(k)
(11.9)
Now
more
that
we have obtained a
specific solution,
we can
shall
<
return to the
To
rewrite
(11.9) in threedimensional
momentum
space,
we
make
use of our
ln In terms of the knowledge of the freefield 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
ln <^
,
proportional to
For
problem
is
treated in
(1942).
2
out
follow the
same
3
104
SOLUBLE INTERACTIONS
,
e~
e
i(ut
and ^'"^(k,*)
is
iutt
.
K =
oj
and
is
f" 7 J o 277
(2eo)*
(2co)*
Writing in terms of the positive and negativefrequency 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
(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
shorthand notation
where the
field
QQ<w a = FD
Here
co is
(11.13)
which can
easily
be checked.
and matrix multiplication with K(k,k') is implied on the righthand 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
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.
105
,
k b then a
We
when
this occurs,
is
1
moment we
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].
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
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
=x
f />.
Then
2
{
(k
when
1
is
this
2
p(k)
 p(k)
p(k).
Cauchy's principal
part.
.
we mean
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
0.
To
we
therefore need
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.
11.1.
It
was
to be
<
0,
since
this corre
It is
known from
ordi
nary quantum mechanics that even a shortrange attractive interaction will not have a bound state if the potential energy is too weak to
overcome the
kinetic zero
<
>
Fig. 11.1. Diagram for D(x) for negative values of x and three possible
Wave Ma
values of
A,
with A x
>
A2
>
Ag
and A2
The matrix 3
fields
which trans
and A3 both
<
0.
quantum
theoretic treatment.
(z),
We
E
2 3
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
107
We
but
it is
onesided unitary
if
bound
sense that2
ii
aQ
t
ii
=1 = l<P =
where
The
the projection operator onto the bound states onesided unitarity is a phenomenon which occurs
is
any
1
exist.
only in
an
infinitedimensional space, since for finite matrices iiii t f 1 To prove these propositions, we write implies ii ii
always
where
(k'
I 1
(11.22)
(fc )
fc
fc'
and
(k '
1
R*
 '
k)
_J__OOp!W_
D.T (fc' 2)
2
fc'
2
fc
ie
(11.23)
:
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
The ground
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
+ (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'
ivLDOt'
2 )
D + (fc 2)
the
(k'
D (kl)\kl~k' "&(}
f
k\
*n
case implies
f

n+a+
k)
(27r) 6
3 3
(k
k')
(11.270)
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
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.
\
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
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 /PL2iJ (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
fields,
(+)in
we
obtain 2
(

#k,0
= 
(k
(k
Q+
 )in
(k',f)
(
k')<
(k',0
k'ty

 )out
k')0
out
<
(k\r)
(11.30)
have the freefield and Since this equation holds for all times (< time dependence), we can split it into positive and negativefrequency parts to obtain, by means of (1 1.20),
+)ollt
in
^<
^()out
= QlQ + ^ +)ln
==Q t+Q _^)in
(1131)
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
As
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)
it is
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',0,0Q +
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
out fidds
1
is
now
2 )//;
implies (l/ZTr
dk.
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 == ^artUryOin
and
tan
<5(/c)

)
4*
_ _
(H.38)
< (ll
.
These equations
will
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),
_AL =
(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).
cit.
CHAPTER
12
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 boundstate 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
113
where
in
is
DM J = DMl] =
With these commutation
relations
(12.2)
in
<
we can
times.
From
(12.1)
= (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)
(12.4)
if
a + w Q\  n_.w
l
til

=
1
.35).
The
last
is
id\k
 k')
(12.5)
also satisfied [see (1 1.276)], but only because we introduced the extra boundstate 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 angularmomentum This connection remains the same when the theory is
114
SOLUBLE INTERACTIONS
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)
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
the transition probabilities as in oneparticle 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 pionnucleon system. If we examine
an
A
for values of A that are small
^)
2
(27T)Vfc
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
Wimw
AI
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
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
= *?
r/2
(12.90)
with
A kr
P (k r )
(12.96)
For the
find,
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'
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
Fig. 12.2.
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
wave packet
expl
_tLz^LL
.
where F 1
2d'(kr)lvr
This
is
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.
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
. .
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 angularmomentum / = 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
>
.
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
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
<
for
particles.
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
.
Born approximation,
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 Bornapproximation value has a simple physical interpretation if we consider the simplified version of electrodynamics 2 which includes only
from
it
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 highenergy 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
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)
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
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
111.,
1955.
119
is
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.
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
12.3.
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
= =
+ 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 zeropoint 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
For the
is
possible only
real.
<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
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).
121
Since the Hamiltonian (12.12) was not reordered, it seems from (12.12c) that in the absence of a bound state the zeropoint 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
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 welldefined 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 standingwave solutions, we find that without interaction the possible values of k are k n defined by
,
sin
kn R
=
is
= [*;* + WJ]
k' n
oi(toidkn
<kj
=
kn
(12.13)
n
co n
r R
= ~2>'2co = 2
fc'
2#
2*W
fc
(12.15fl)
co
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 zeropoint 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
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 lowlying 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
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 )
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 numberofquanta 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
A/4At
AAt = AtA =
The
when
and A A^
t
requires
0.
123
is
The determination of
easily done also guided
most
by comparison with a onedimensional 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
(12.19)
From
(12.17)
and
its
hermitian conjugate
a
I
we
find
in, 0}
( 12
20 ^
= 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
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
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
mean number of
virtual particles:
ka\k)a(k)
in,
>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
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
essentially
p r UJ
a result which
is
3 ( 27T)
CD*
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.
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 singleoscillator analogue we find
 JT
W ~
CO
Z
if CO
r
f*&
CO
virtual pairs
is
in, 0>
(in,
in, 0}
stems only from the change in the zeropoint 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 electronpositron pairs in the Coulomb field actually give a measurable "vacuumpolarization" 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).
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
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)
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
1\
T O/
.=
/O
\1
0\
r3
O/
/I
0\
(13.2)
\0
I/
In the presence of an interaction the operators r become timedependent, and (13.2) is a representation of r(0) in terms of bare particle states at
t
= 0.
The
interaction H'(r) which generates the processes states must be of the form
<>/?
+ ir*
(destruction operator)
+ TT+ for a pion described by a KleinGordon 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 pn 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
[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)
to
commute with
r(rt at
the
same
(13.5)
y>
>(rA V V,0]
OM, V(r',0]
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 \
(13.7*)
In
momentum
=
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
A rett rk (M
~i;J
(dK dKo
1 /e ~lv'
for
'
<
for*>0
Q=
129
and
its
hermitian conjugate,
we obtain
/(k,0
tfk.0
= =
tln
y,
(k,0
ig
P
Jao
iw{t <fc'r_(O e
~ tf)
(27T)
)
(13.90)
AM find
ig
Joo
df r + (O <r ""''
^
(277)
(13.9W
= * r_(0 +
gjd*r
p(r)y\T
t)Tz(t)
(13.100)
=
igffr
P (r)[^(r,0r_(0
^Or + (0]
03.11)
The
first
we
shall
says
(30
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
<J
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
last
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.
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
mesons.
13.3. Physical Nucleons.
state
\p)
and
the
identical,
P)
=

T+
T3(0
jp>
(13.13)
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.

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
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
AM < m,
The
K >m
(}
can
w
Because of (13.16), we can eliminate the unobservable
lefthand 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
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 freefield 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.
< 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 fourdimensional 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)
= (kQtk')
We
n
note that rl. plays the same role as the boundstate 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 (2rry
AM ^(k.o
).
(13.20)
(k,^
oc
6(w
K
13.5. Completeness.
We
are
now
in
subspace
1
Q=
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.
Q=
Z)~
1
J
133
and
if
Z=
(AM), as
shown below.
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,
an
_L f^l~~
2rri
de
circle at oo
(13.21)
Z=
<P

2m

<b J (z
 AM)D(z)
==
Z
D(AM)
rs
p>
(13.22)
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)
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)*LwAM
can be checked. 1
wAM
 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
in,
><,! [zV? L
the
+ gJ("
( (2iry (w
p)
= Z'
(13.27)
and
U (k) is
b

virtual pion
when
the neutron
is
dissociated into/?
+

TI:
(P
+ ^k
in > *)
= (p
+ (k
With some calculations we find
'
(13.28)
that this
it
(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.
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.
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
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 pionproton 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
*!! :
ut
(r)
(13.31)
Hence we have
(k
1
Q

tin
k')v>
(k')
= (k Q +

tout

k')v>
(k')
(13.32]
For AAf
>
136
SOLUBLE INTERACTIONS
in pair theory, for a spherical source only the
As
/
angularmomentum
e~
9
parts of
tout
v>
and y tin
differ
tout
by a phase
shift
^w
Tin
(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
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.,
[V. Glaser, G. Kallen, Nuclear Phys., 2:706 (19561957)]. 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
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.
.
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
we
find
2
__
? gr P f
2
gr
\
P (k)\*
}(
(w
 AM) 2
Hence, for a
relativistic
large) value of
r,
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
cit.
The
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 highenergy 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
W.
Pauli, Kgl.
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, shortcutting 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
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
\
N)
we can
d(K
use (13.18)
to derive
an
)
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 Smatrix element from (14.2) and the familiar time dependence of matrix elements between eigenstates of the Hamiltonian. Thus 1
(out,
77 )
 k') 
(out,
N+
w')
;
3
(5
(k
 k') 
2md(w
g
(2rr)
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
p(r) is
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
WITH
Q =
141
The T matrix defined by (14.4) is, of course, timeindependent. 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
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
gr
1
(p
gr+
n>
<ii
r_(0)
p)
Z#
J
(14.76)
emphasizes that the source must involve dynamical For a cnumber source the righthand 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)
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
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
out,
/>
all together,
(27r)
(27r)
E
fc*0
ic
first
d(k)/k.
It
appears from
(14.9)
>0, as for an
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 wellknown induced 1 This process decreases the energy, the gain of interaction emission.
.
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 lowenergy 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
Similarly,
we can
interpret the
of the other.
stimulated
2

/i>
oc
(n
1)
Q=
*
143
is
proportional to
T.
2 involves only
'
<out,
r_
I
n),
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
stjcwUrttwiJ
=
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 highenergy 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
was obtained
r
L
3
6r
rg
87T
w' 2 (w'
ww w~
k')
1
/e)J
in, k'>
the
sum
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)]
is
{)
provided that g r
is
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
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.
Q =
145
which depends on
p(k).
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
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~
become much
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 Highenergy Behavior of T(k). Finally, we shall from (14.9) and (14.50) the low and highenergy behavior of T(k).
end we consider
To
this
as a function of w.
From
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)
w0
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
2
i
n (14.9) equal to
w.
Taking the
just the
is left is
equaltime commutator.
lim t(k) woo
r 3 , so that
1)
we
get
(14.17)
= * (n
W
T,
>
= 8 (2Z W
if
The
last expression
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 highenergy lijnits can be seen by resolving the scattering process into elementary interactions
Fig. 14.1.
Graph
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.
scattering of ?r~
neutron.
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.
Q=
147
p
(a.)
n
(b)
V.*""
....
n
n
(C)
x
p
*>>
Fig. 14.3.
+ IT scattering
and (,
c)
w~
scattering.
n
(a)
;i
/"
p
n
Fig. 14.4.
Diagrams to
illustrate the
highenergy 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 highenergy 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 highenergy 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
+ IT
scattering.
meson. The latter contributions, illustrated in Fig. 14.5c, will again be small compared with the others, and we get the Bornapproximation amplitudes from the two kinds of particles weighted by the corresponding probabilities. To discuss the lowenergy 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
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:
<:.
Fig. 14.6.
/ pnpnpnpnpnp
., ,...,
......
....,
to illustrate the "lowenergy 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 higherorder
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 lowenergy limits
^x.r
x:
(a)
(6)
Fig. 14.7.
Graphs
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
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 energyscattering 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
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
furthermore, nucleonpair 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 twodimensional example in which 4 Also, the manybody problem (such quantum effects are important. as represented by a Bose gas or a FermiDirac 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. highdensity 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 pionnucleon 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
the pionnucleon interaction is complicated by form of the interaction is not known for certain,
two and
nucleon.
It is fairly certain,
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
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 twoninths the distance of the pions, which implies that the volume of the nucleon is 99 per cent pion cloud. Nucleonanti
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 clearcut 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 jj of the radius of the hydrogen atom. Correspondingly, the Baimer
formula
is a very good approximation, and finestructure 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 pionpion interaction. This will
certainly also affect the pionnucleon interaction, since the nucleon
is
surrounded by a fairly dense pion cloud. Fortunately, it turns out that in lowenergy pion physics there are large effects, mediumsized 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 mediumsized effects amount to about 15 per cent corrections to the large effects and are due to pionpion interactions, nonlinear pionnucleon 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
(U.S.S.R.), 32:178 (1957), trans, in This would also contribute to a pionpion 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).
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 strangeparticle
mediumsized
effects are
interactions,
manymeson
forces, etc.,
With this orientation, we turn to the question of the detailed form of the pionnucleon 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
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 angularmomentum 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.
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
pion
may
2x2
/O
1\
/O
^ll
A
OJ
o]
'"H"
/I
'=\Q l
Another property of the pionnucleon interaction is that the invariance of the threedimensional 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, McGrawHill
2
&
Book Company,
4
Inc.,
is
New
York, 1955.
it
This statement
not exact;
holds only
if
and
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 angularmomentum
conservation,
we can
write
r)aV^(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 angularmomentum eigenfunctions of the pionnucleon system
is
As
required, this
is
invariant under a
5.2).
and
it
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 pionnucleon interaction. This coupling constant is determined from experiments which we shall discuss in subsequent chapters. Of the various applications of the theory of pionnucleon interaction,
the
most successful
is
pionnucleon scattering.
We
Relations and Equations of Motion. Having decided on the interaction term, we are now in a position to write the
1
The
Jo,,
angularmomentum operator L,
It is related
is
an
axial vector
and does
s,
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 Englishletter subscripts for the former and Greekletter subscripts for the
latter.
INTRODUCTION
fundamental equations of our theory.
159
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
it is
components,
(153)
l
and
=!
(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
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 groundstate 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&]
at equal times
(15.6)
a/3y
is
was introduced
earlier.
from
and
A
f
p(r).
As usual, we assume that // is ordered and restrict The operators aa are defined similarly to ^*.
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)
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
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 righthanded and lefthanded 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.
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 pionnucleon=  is equal to that for == J, / scattering phase shift for the state T
T=
f,
.
15.3.
(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 pionnucleon 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 Pwave 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. Quantummechanically, the mesons in the source are not sharply
fact that in the
1
The
momentum / =
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 Quantummechanically, 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 pionnucleon scattering the lowenergy cross section will be proportional to fc4 , since it involves
This
is
shown
in Fig. 15.1
and
is
in
being caused by an Swave 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.
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
maximum
max
^4
p*
If the size of the source
is
\pi
increased.
Dimensionally, one would expect that an expansion in /is This is indeed the case, and it
200
100
150
200
250
300
350
(a)
80
fa
I
40
Q 20
50
100
150
200
250
300
350
(b)
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 lowenergy data is proportional to the fourth power of
the centerofmass
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 pionnucleon 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
~ l/(source radius) ~ M.
may
When
use a
Yukawa
cutoff,
/>(/)
= =
a
or a gaussian cutoff,
p (k)
<r*
/ fc
ma
CHAPTER
16
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.
(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
momentum must be a halfinteger 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 stationarystate 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 timedependent 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
we assume
(16.3)
168
PION PHYSICS
a) 8 ,
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
<
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),
(co s /cos
0)a
<*i
provided
/''Y
\ rr*e
(\
7\
oj s
we
p.
w <
To
this
If the cutoff
fcmax is
much
larger than
then
we can show
)2 .
that the
is
= &  (^ 
^ + J^~.^
(16.8)
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
169
The
first
term in
this
tion for
oo
2*=SF
(16.9)
a of
P (/c)
f
J
r

The remaining term converges, and we can Collecting the contributions, we get
1.
and
cK^
+ O^otfv
if
(16.11)
Our
is
much
In
pion,
e.g., 1/0
> ^.H
mainly highmomentum 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
form
/
o"(0
d/r
VK (r)
(16.12)
d= ,_jlX
677^
fl
(16.13)
This
is
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
<
//.
*[
is
that
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,
is
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.
171
To
we go
to the limit a~
>
/*,
which
is
satisfied in the
pionnucleon interaction.
s
We
actually
then have
2 cos
and hence
j
For
=
of J, we find
t
(16.17)
AE expressed
in terms
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
(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. neutronproton interaction, which is not strong enough to bind more than the tripletspin S state (deuteron). Higherangularmomentum
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
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
<
in
<
(r,0
Ae'* *'** 9
3
A*e ^or.)
(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.
*
Since
1
<*o
i(0
i(0
(16.20)
we want a
at
will
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
in
<
(r
r', t
 fXO
W)
(16.22)
completely analogous to the previous calculation, and we immediately turn to what corresponds to the positivefrequency part of (16.6):
/O)
a(w
2i
A*
A<JO
p(/c)
C(w
)a
a(a>
(16.23)
The
term on the righthand 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
is
in the
xy plane perpendicular to
(a
a(coj.
Therefore
we can
<T(CO O)
ya
 fa X
(16.24)
<
To
we have
to
for
in
.
173
abovementioned frame,
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
in
<
as
is
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
6rr^alg
oscillation
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
(c.c.
complex conjugate),
a(co
irfjir
ks
c.c.
e*"
/(*)
c.c.
(16.27)
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 zeropoint oscillations being of its own 2 = 2 = 1 Cor3). quantummechanically, 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)
r
Near the resonance the
da
__ ~~
34V
3
(60
**
r
(16.32)
d&
15 \47T/
+ cos 2 2 ) + T /4
2
6t)
(16.33)
175
quantummechanical 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 quantummechanical 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 lowenergy pion physics, namely, a resonant state with
higher angular
momentum,
is correctly predicted by the classical model. features of the pionnucleon interaction can be under
its
meson
16.3.
field.
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 pionnucleon 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
= (I)*
"T + )
(16 34)
'
n7r )

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>
New
York, 1953.
176
Since
PION PHYSICS
2 &cTa a
is
(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 righthand
side of (16.34) are the wellknown ClebschGordan 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 pionnucleon 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 )*
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 samestrength
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.
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.
< .
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.).
177
again six possible states of the pionnucleon 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)   (* tD + () Jo) J = I, J = JiJ, = f) ll)
1
J
I
=*t
J,
4)
(I)
ti)
+ (J)
jo)
Although we are
classical properties
in the region of small quantum numbers, the of the angularmomentum 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 quantummechanical expression of the
the situation
is
reversed.
we
studied earlier.
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 spinJ 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
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 mesonnucleon 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
where
and
En
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.
it
goals of
meson theory
bare nucleon and its a deadend 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 pionnucleon 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 selfenergy <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 have
=

N,
in)
=

N, out)
is
fourfold degenerate.
The degenerate
179
states all
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,7e
for arbitrary vectors n for the isospin andy
'a',/>
n'.
and
1,
Here,
we have used
subscripts
<x
1,
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
(*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)
r2
. .
we
To avoid crowding of
.
1,
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 mesoncreation 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 timedependent 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.
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 mesoncreation 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
Pt)
*t)
(17.7)

Pt)]
vacuum
is
CK
are independent of
lz
4AU HA 0
t^ The
(17.8)
and
Coi
+ C 10 + C u =
(17.9)
Cu
so that
fl
(17.10)
we have
<
Coo
<
(17.11)
Cu
M.
theorem. See, e.g., is a general result usually called the WignerEckart 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
<"' !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
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 BoseEinstein statistics requires the This radial antisymmetriradial part of the wave function to be odd. zation, together with the shortrange 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.
CQl
is
negligible,
r2 are both
Cu),
Cu (since C00 =
^ and
r^ltCu
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.,
kk^k,).
*
U. HaberSchaim and
W.
Thirring,
(1955).
183
The importance of
rx
and
r2 is
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 firstorder change in the groundstate 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 staticsourcemodel 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
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
is
may omit
the remaining
Inc.,
"Meson
Physics,*'
New
York, 1952.
184
PION PHYSICS
Furthermore,
J? ooo
^1010
so that
COG
C10
=C =
01
/?f) is
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
TammDancoff Approximation. 1 The TammDancoff 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.
it
subspace.
Correspondingly,
Q=
\ states.
agrees with the exact solution of the From (17.7), the general form of the
is
ground
state in the
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.
&
185
When
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 twomeson 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,
e.g.,
an
integral
is
The wave function/(k), as anticipated, equation for the number ^ to that of the Lee model. In the weakcoupling limit analogous
and/(k) and
<^
result.
On
the
more
realistic limit
2/*
co
we
(.7.26,
Whereas
1
Cn C01
,
and
(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
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 physicalnucleonstate 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 twomeson and higher amplitudes should be
negligible for large fkm jp.
17 A. Tomonaga Intermediatecoupling 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 BoseEinstein 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 angularmomentum representation, Thus, if we expand the meson field in a complete set of
F (k)
s
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
[flSl^IvV]
fl^Al'dfnm'
(17.29)
The Hamiltonian
is
(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.
187
W = f "dk o>F;(fc)F,(*)
Jo
trial state

/.
=f
fji
t"dk
Jo
^yyffi (127rw)
f
(17.31)
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 mesoncreation 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)
'
form of F(k\
F(k)
4^T% +
0)\0)
A)
A:
is
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.
the following,
we
shall henceforth
drop
188
It
PION PHYSICS
remains to determine the groundstate 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
,
a])
(17.34)
Going back
and q by
we can
rewrite
H in the form
H=
2
(p
2
+ W V) + g'vqiWeo
2
(17.36)
with
!^ j=i
and
g'
(17.36) emphasizes the formal analogy to elementary wave mechanics and corresponds to a threedimensional 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 angularmomentum 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
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
^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
rotationinvariant, (17.38)
Aa B = A B +
/a
AX B
we
find
These two secondorder 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 zeropoint 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
fstates will
state
by
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
ground
state.
The
we
191
we find
A >
and so
&
3
/*
97T
we
above
W*
2ir
from
(16.18),
4/Cmax
2 r2M~,,2 /4
H+6,
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 Poissonlike 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
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,
i(tfl
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
,
.
. .
ninedimensional space. However, for small coupling strengths / the perturbation result is obtained, whereas for large values of/the theory goes over into the strongcoupling approximation, which we shall
discuss next.
17.5. Strongcoupling Approximation.
2
In this limit
many mesons
surround the bare nucleon, and the zeropoint 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
into the
we introduce canonical
oper
0744)
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,
193
H"
= HJ + H"  <? =I
<xi
\ 2.
(Pi,
(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
.
t/o.tr
<r...
B*
BB t
= T a&
m
mj
(17.46)
shorthand notation)
6a
strate
it
= A^BrfK = AqB^
(17.47)
definite
We claim that q can be diagonalized by this transformation and demonas follows. We observe, first of all, that the real positive
real orthog
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)
Q):
(17.50)
Q^Q Q
The eigenvalues of
matrix,
//'
(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
0.0.0^ =
= =
oyr t , namely,
diagonalized.
The
0^0* must be
are
1.
set
61
6263  62 + 63 61
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
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)
195
fo.*W = ^i*.^
we
recognize that
y\(
is
v*i T* a
ij)
Q
i
( 17  57 >
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)
is
zero,
we obtain
'
^o=^ +
(1760)
where E' is the sum of the zeropoint 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.fflpW co(o) Jo
we minimize
f A)
(17.62)
To
<^
with respect to
it.
Making
use of
we
find
a<r
^
<f
is
/
2
LJ8
1
and
is
(17 63)
.
1'2
is
onethird 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
lie
much
and
9
AE =
above the ground
state.
i)
r
is
is
(17.64)
it
will
produce a
resonance in the f ,fscattering. We have already studied this in the classical approximation, and we shall study its quantummechanical 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 groundstate 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 weakcoupling 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).
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
f$l4ir
r 2f/4,r~0.1.
CHAPTER
18
Pion Scattering
18.1. Introduction.
In this chapter
we
1 (/ = Pwave 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 angularmomentum
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 strongcoupling 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.,
3B,
2
Row, Peterson
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 nucleonantinucleon8
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 Pwave 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
angularmomentum
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 angularmomentum 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>
Xtf>
(18.2)
generalization of this description for a state with n real mesons, which will be needed shortly, is
in,
The
^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 angularmomentum operators:
lew  Ko = x
Similarly, in terms of the outgoing
meson operator
B,
we have
= 0) = *k(0) i
4(0)
f
J oo
(18.4)
The operator V is
1
hermitian,
VK (f) =
we
shall henceforth set the
To
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

<*'
BK>*K
~

'(out,
k'f
0
(18.6)
we made
dependence of
VK (f)
and
S and T matrices
in,
differ
=
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
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
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
coi
T*
n
*n,K'?< n,K
eof
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)
ie which appears [e.g., after substitution of denominator of the second term in (18.13) plays no role.
and
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 mesonnucleon coupling with both
202
PION PHYSICS
(i.e.,
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 negativeenergy region. The theorems (18.15) are rather concerned with the analytic properties o the / matrix
mesons
connection
is
HI
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
R given
VK

by
I
RrxtK = S (' VK
>
\VK
\$
'

>
VK
.

(18.18)
{ is to
ground
states.
The matrix
from
to oo
and
to
oo,
as indicated in Fig. 18.1. The contribution of the negativeaxis cut arises from the crossing symmetry and was thus absent in the Lee model.
1
M. GellMann and M.
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 Highenergy Limits of
it is
Lee model,
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
the
perturbationtheory 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')
\\
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
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 Bornapproximation 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
matrix
is
tj.
approximation:
'
I
[(<")*,
(")*]
*)
rj
lim zfuorn
Z*oo
(18.23)
Whereas the zeroenergy 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 onemeson states (K) to a representation
,

are diagonal. 1 Indeed, since these variables the energy determine the onemeson 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
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 onemeson 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
2, 3 for these
phase
32
shifts,
then
The
24.
,14
16
20
Fig. 18.3.
Form of the S matrix. The offdiagonal elements are diagonal ones are as given in the figure.
If
all zero,
and the
The
by anormalization
sin 6JT
fact or
of 4nk*.
t x
3 2 87r co
Ar4
71
=J
WT
^r
206
PION PHYSICS
and J
way with T
)l
T/2
+a/2:1f
 T )(J 2
J)
+ (T
.
a  JX
a
a0]
tX2
(18.28)
The normalization
subspace
onemeson
(18.29)
<P<*><P<">
= d uv y
t
1
(v}
where
/
w, v
1, 2,
and
in the
=
/
1
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
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
A
,
is
a
7
= U 8
\
16
t
(18.33)
U Wheffeas
value at time
1
T
/
is
It is their
that
is
needed here.
The
labels /
and a
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.
P'*'*
<i)
A
A (i) (v)
V'K'K)
()
(A
(1)
,
A (2) , A <3) )
is
X/1 X X  (8,2,
To
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
(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
Pwave channel by
This definition
target
is
and
definite isospin is
Im
co
It
>
2, particles
the phase shifts describing the oneparticle 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
=_ 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)'
(18 .40)
/0^(/C
CO
CO
CO
18.6. Relation
of
all
Low
Equations
to
Experiment.
In
the
Low
quantities except the coupling constant/? are at least in principle. The first terms on the right
hand
side are the Bornapproximation 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 2states and positive in the 3state, corresponding to a repulsion in the former and an attraction in the latter. This feature can be understood Since Fig. in terms of the lowestorder 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 2state the opposite sign of the p and wcoupling 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 Bornapproximation term in the 3state 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 shortrange potential, where Bornapproximation 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
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 "ChewLow" = is 47r//r2 and should be the same for all plot extrapolated to co phase shifts. Experimentally, the 1 and 2phase 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 3phase, 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 zeroenergy 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 3phase 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
&
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
ChewLow 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
<
Then
and we can
write
ff "
fc
i*"
apW
1
&T
(18.43)
PION SCATTERING
This approximation
is
211
the
sum over
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>
')
'*)
tt
(18.450)
tt
(G>)
find,
from
=  3f
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
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 lowenergy 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
perturbationtheory result.
212
PION PHYSICS
(18.37),
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 onemeson approximation, it is not possible to find an exact solution. However, an approximate solution has been suggested by
As
replace
B by
#',
B&B'=
which
differs
1
\=B
1
1
(18.48)
from
B
B'
2
satisfies
=l
we
TrB'
\B'\
(18.49)
With
this
is
soluble for
yet for
2.
From
(18.47)
find
g2
v ^ Xz)
= g v \z)
^
l
g2 \z)
if
\j_
ga \z)
V
( 18
'
and
There are no
real
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.
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
(18.520)
fr
oj
k'*p\k')t
co'
2
4
co
co
/
8
co'
\
ft>/
4?r 3ir Ji
\co'
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
Lee model.
It predicts
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
In this
way we
obtain
(18.53)
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 effectiverange approximation is still approximately valid at the resonance energy co r then we
rl
;
same
as that given
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 Pwave 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
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
o>)
~*
^
(1
l.l
(18.56)
3 4rr
which
is
cutoffindependent
if
kr
is
is
of the
The shape of
Chap.
15.
To conclude this chapter, we shall point out the 18.8. Summary. most important features of pionnucleon 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
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 mesonactive 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 Pwave mesons, the interaction
probability is strongly energydependent, 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.
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 quantummechanical 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 (emissionabsorption, 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 3state 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
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 mesonactive) x ?? [for nucleon to emit (or l absorb) a meson within time co~ ]
^process
The
1
cross section
is
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
&
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 3state, and at higher energies. To the extent that the 3resonance
Of course,
2
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., np) 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
=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
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 *Swave phase shifts are not zero and become
35
30
5
20
""
30 60 90
120
150
180
and chargeexchange
scattering of
The experimental Anderson, W. C. Davidon, M. Glicksman, and U. The curves are proportional to 3 cos 2 # 4(1955).
189Mev 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 pionnucleon 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 planewave 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
we have
the counterpart of that defined by (18.5). >4 a (k) ) 0, and by making use of
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
\
Joo
219
220
PION PHYSICS
field
becomes
87T
(19.3)
,,lrr'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 largedistance of the nuclear forces (see Chap. 21) and (e.g., onemesonexchange) part also with zeroenergy rrN scattering. In order 19.2. Groundstate 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
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 righthand 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
is
we have
In ver a ll states excluding the ground state. to transform the projection operators into
This
'
I
is,
in a
more
I
explicit notation,
3W,i k =
where
wOVr'm'^w
y
k)
(m k)

=
(l)f
221
This gives us
k'
o
k' a k)
S
n
{' r^a

n'

out, w)(out, n
rjs
n

f)
<5(co w
where n and
n'
+
2
OJ ) (0) q
,
+
C0
and
k'.
Multiply
q)
00
<*>_, q
we
get
(19.6)
As a
first
application of (19.6),
With
f_^_
3
/>
(/c)/c
r3/ r
Leo
2
2
J(27r)
2co
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
(^)
(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 Pwave 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
way
as above,
we obtain
3 r
7T
r fi
L47Tft>
if*
2
Jo
36?T Jl
n
(19.8)
(a)
+ a)
Q)'
^^J.
=2f
at
=S1
6 f*
77
Ut
(t)
'
\
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 Bornapproximation 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)
Model.
To
learn
r l5 r2 ,
more about
<f
we must
parameters
2
,
w max which
o> w
:
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)
223
To
.
we
use (19.56):
IK'K)
(19.13)
~~
'
IK'K "H T
ij 2^ 2
^ 3)
I)
1,
and a
t,
we
get
+ +
4o2
er,
+ 4(r3  2(T3 +
<r
1rJ
(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
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 pionnucleon scattering or in the expression for (N), where the highenergy contributions to the integral are suppressed by one or more extra powers of co in the denominator. Nevertheless, if we insert the lowenergyscattering data into the
relations
we obtained, or shall obtain, 1 and leave the highenergy 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,
which we
2
See
S.
Fubini and
W.
224
PION PHYSICS
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
to
Chap.
17.
From
e& 0.4
we
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 Selfenergy. 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 ,fresonance dominates the scattering. We, therefore, neglect a and <7 2 and for cr 3 we insert twothirds 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)].
,
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
protonantiproton 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 energymomentum 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.
225
same
^ T?
co(co
+ J
ft)
~
^
/
(ft)
.*
co n
V2
^ "I 2
ft>
( 19 '
18 >
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 strongcoupling (or classical) limit,
(17.63),
~/ ~
which
is
2
1
(*dk
JQ
P \k)k*
where/
9/?l ~
[
~~
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 highenergy electronnucleon 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.
'
(19.21)
Remembering
find,
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
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)
kr2o k
+Ta
2
k'^a k
kXk'
(19.25)
227
We can
We note
J
 k')(<J
k')
*
VX
IW fk' k k J
~ (k k (k + k')
'
'
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,
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
= 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 electronscattering
.
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. meansquare 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 oppositeinsign 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)\)
With
rx
0.7/o(r)
for
(1958).
2
to ^(k')
etc.,
which
arise
from
^2,
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 electronscattering experiments.
Whereas the
meansquare radius
is
much
less,
if
not zero.
The most
plausible
itself is
answer to
Bare nucleon
QP (D
Bare nucleon
lArr,
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 pionpion 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 twothirds of the time the proton is dissociated according to TT. >n f n + and for onethird of the time into
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 spinzero 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
RBA
of (19.7)
23
which is onehalf of the RBA of (19.30).
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.
231
If the
more
sophisticated argument
f
Jl
00
is
wrong by a
factor 2.
contribution from
(19.16)
is
da> n
is
^=
(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 strongcoupling result is the correct RBA x Dancoff, Phys. Rev., 62:85 (1942). Note that Wl s = in the strongcoupling 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 pionnucleon 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)
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. planewave 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 barenucleon charge density with an average value of 0, but this is neglected.
1
232
ELECTROMAGNETIC PHENOMENA
expressible in terms of only the nucleon operators T that it is divergencefree 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 mesoncreation 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,00
source.
(20.5)
field
cannot be
satisfied
inside the
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)
with
+V = */aVMn(OA(r,0 ~ T2(0&(r,0] j,

+V
(208)
Thus, (20.5)
current" 1
j7 .
is
When
charge
the
a n+
is
The
why we need
it
requires a current to Since the density of carry the charge from the nucleon to the cloud. Pwave 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., 5wave 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>
.
make
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 electricdipole 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
Since the
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 pionnucleon 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 electronproton 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
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 angularmomentum expansion, we find that* there are no For / oneparticle 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. 196210, McGrawHill 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
By means of
1
(19.1)
we
Low
equation:
1)
N+
7r
k a
J(k)
N)
(('
[*.(k'),
Kk)]
'

[a a (k')
f>
+S
n

it
'
tf
j(k)
out. n)<out, n t,
)
F.(k')
fil
'
c'

K.(fc')
j(k)
fi
[tf L
E

(O
(r

j(k)
out, liXout. n
w
Ka (k)
f)1
+w
(2Q14)
expression for
The onenucleon expectation value can readily be evaluated with With the help of j, which we discussed earlier.
the
we
commutators
f (27T) J
,
d 3 r [a a (k'),
(20.15)
1
We remind
5'
implies a
sum
238
PION PHYSICS
the
sum of
wk'), K
j is
[see (20.10)
with F(fc 2)
3
F(0)]
Kk)
*>
'
*'(
*

^~ 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
Kk)
out, n)<out, n
Kg(kQ
1)1
j
(E w
co)(2a>f
J/
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 righthand 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
j ff
(10~
13
cm)
is
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'),
of
the
photon
with
the
magnetic
moment of the
nucleon.
in
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 nN scattering, we the amplitude in expect that the rescattering gives an enhancement of 3state and a suppression in the other states. the Furthermore, it turns out that these corrections mainly affect the magneticmoment 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,
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
we have
and
,
(f J(~~k)
f)
(S
} }' ) N the first term gives the same N ]' (20.15) but zero contribution to the magneticmoment 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 magneticmoment 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 Swave scattering, which is not contained in our model,
significantly
would
section*.
change this contribution and therefore the lowenergy 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 KrollRuderman 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 planewave 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
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 ,fstate, 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
<2<u *"
1
<5
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)
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 mesoncloudcreation
process (Fig. 20. la) the cross section
is
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 angularmomentum 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 lowenergy process would be the absorption of an electricdipole 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.
300Mev 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 antiparallelspin 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. chargedmeson 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)
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
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
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 highenergy 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 lowerenergy 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
mesons needs some refinements, 1 mainly due to recoil The additional effects, but is quite well described by the theory. contribution of the electricdipole terms and of higherorder multipoles
excitation of
TT
See M.
J.
106:1337(1957).
246
TT, as
PION PHYSICS
would have been obtained according to the argument given at The lowenergy data fit (20.21) with fe* 0.07, in good agreement with the coupling constant found
scattering.
from mesonnucleon
20
16
$,
30
60
90
120
150
180
40
80
120
160
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
production
175
legend for Fig. 20.4). The sources for the higherenergy data are also cited in the legend for Fig. 20.4. The dashed curve is a straight
schmidtClermont, 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 Sstate production. The higherenergy curve is taken from "Proceedings of the 7th
line,
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 Sstate 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 Sstate 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,2state 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 centerofmass 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
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
CHAPTER
21
Nuclear Forces
Classical Calculation of Nuclear Interaction In this last chapter we shall use the static model to calculate the longrange, 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 clearcut test of the static model. static potential should be able to account for all the lowenergy properties (e.g., deuteron binding
21.1. Introduction:
Energy.
energy quadrupole moment, and nucleonnucleonscattering phase On the basis of range arguments shifts) of the twonucleon 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 shorterrange 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 Pwave 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
was obtained
Chap. 16
a
is
250
PION PHYSICS
straightforward to generalize this to a symmetrically coupled field to obtain, for static a and T,
(21.2a)
with
y (r)
'
'
(21.26)
r
r
I
H' becomes
(21.3)
H'
fl
a
f[p(r)
/
rfr
 r )]a V&r.iPr
By
we
two sources)
= /*
a=i
I;
TaaTfta
L(r>(r'
3
r )(aa
V )(a V
.
r)
r
r
l
p(r)
(21.4)
(5
(r),
this gives
477=i
aa
if
3 L
/3
\r%
3
r
where
ST
is
force):
3qa
roq 6
'
rQ
__
ro
finite sources, the longdistance 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 perturbationtheory calculation of the twobody 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
'
'
raq .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
sign,
T& potential (21.7) originates from T depending on whether the isospins of the
tt
Its
expectation value
is 1
in the
of the
Fig. 21.1.
Diagram
exchange
.
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 /A5 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
TO
15
2.0
2.5
3.0
/^
'
Spin
triplet, isosinglet
states
and
spin singlet, isotriplet states
(6)
Fig. 21.2. Plot of the longestrange 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
IN
VARIOUS STATES
is an isosinglet, tripletspin 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
attractive but
Quantum
mechanical.
'
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
BornOppenheimer (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 nucleonnucleon system. difficult to justify a BornOppenheimer 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, McGrawHill Book Company,
1955.
L.
I.
Inc.,
NUCLEAR FORCES
it is
253
all consist
ently,
and
it
was
also in this
way
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
^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 twomesonexchange diagrams of Fig. 21.5 have , and their contribution can be related to mesonnucleon
/ \r \ \
Fig. 21.5. Typical
x
?r/
/
N
v \ \
forces.
We shall only mention how they can be computed, since they are strongly influenced by mesonmeson interactions and non2 adiabatic corrections. Our results will be derived in the spirit of a HeitlerLondon approximation. Thus, for calculating the energy of the twonucleon system, we shall, in first approximation, use a state wherein both nucleons have
scattering.
(1 7.6),
as 3
= KJ?
fa ,f b ,r
(21.8)
field theory.
defined to be an energyindependent 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)
h.c.
in the limit r *
oo.
We
ft
shall
now
fl
energy
<fa ,f ,r
/ff ,ft,r
= nro)
more
refined trial state
states.
1
To
we have
to use a
{
which
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)
It
is
of
(twomeson
Furthermore, we have
(H
+ 2*
)

^, (
fc
TO)
[~aI(k)/Faa(k)
/e k
Fab (k)al(k)]
fa ^, r
,
(21 .12)
on the righthand 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
(2113)
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
is
we
finally obtain
f
(21.14)
.TT3
which, as predicted,
2
(2?r) J
is
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 twonucleon wave function & a $6 (r ) is defined by writing the twonucleon state
2)
2
b
^cfeOo)
ftp
*,
FO>
(21.15)
2>
is
[*.*.'*'
V*
(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 wellknown 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 pionnucleon scattering. A detailed discussion of (21.16) shows that
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 lowenergy scattering parameters, that is, the scattering length and
effective ranges in
S and P states,
(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 velocitydependent forces such as a spinorbit 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 nucleonnucleon 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 lowenergy 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 socalled 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 antinucleonnucleon pairs, pionpion 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
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.
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 mesonmeson 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 mediumsized 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.
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 sphericalwave 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.
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)
Ifi
(5.1)
(6.3)
(8.2) (5.1)
(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 Bornapproximation scattering matrix
01.1)
(17.4)
(12.2)
(18.5)
(15.1)
moment
finite
(20.1)
(5.4)
Number
f
77
of particles in a
volume
for nucleon
(6.3)
(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
Spin operator
(15.1) (13.1)
on the
(13.3)
KleinGordon field (operator) i<ut ) Positivefrequency part of <^> (oc e~ of </> (oc e itot) Negativefrequency part Charged field operators oo Field at t
Field at
t
= =
(8.1)
(8.1) (5.1)
f oo
Azimuthal angle
(9.3)
Twonucleon
Schrodinger
relative
wave function
(21.3)
(4.2)
field
operator
LIST OF
SYMBOLS
265
266
LIST
OF SYMBOLS
Other Symbols
(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
Localenergydensity operator Projected amplitude of / matrix
tf
(18.7)
(18.7)
(1.3)
(5.4)
(6.1)
(18.5)
<9
*o
2)
(14.2)
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\
(12.2) (15.3)
(1.3)
(4.2)
(5.1)
(17.5)
(5.1) (4.2)
M
AM
(9.2)
(9.2)
Matrices
(12.4)
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, 3440, 260 commutation, 36, 60, 64
definition, 35 eigenstates, 39, 176, 177
Born
approximation
(see
Perturbation
theory)
Bound
52
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,
Arnous,
Bardeen,
E., 125/f.
commutation
J.,
rules, 60, 61
\54n.
Bare
Beneventano, M., 245/i., 246/t. Bernardini, G., 245 Bethe, H. A., 118/i., 155/1., 157/f.,
198.
Blatt, J.
M.,
230/1., 247/i.
Bohr, A., 153/f. Bohr, N., 3/i., 26n. Born, M., 252/1.
269
270
Charged
INDEX
particles, 61,
62
Creation operators,
9,
Cross section (see Scattering) Crossing symmetry, 201, 202, 207, 212 Current, 60, 232235, 237241
245/*.,
248., 256.
commutation
ChewLow
plot, 209,
210
59
Dalitz, R. H., 144/1., 211/f. Dancoff, S. M., 184/i., 192., 196n., 231/t.
for nucleon potential, 249, 250 for pair theory, 1001 1 1 for pion physics, 166172, 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, 6365
in
in three
in
equation of motion, 167169 excited state, 167, 170174, 214 neutral theory, 166 scattering, 172175 source, 168, 169
strong coupling limit, 171
ClebschGordan
Commutation
9, 30,
32
creation
10,
de Hoffmann, F., 118., 157/1., 184., 198/1. Demeur, M., lOO/i. Destruction operators, 9, 26, 33, 38, 39, 75 Determinant, Fredholm, 104106 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
DuMond,
Dyson, F.
Eden, R.
215
J.
W.
M., 230w.
163.,
211/1.
44
J., lOO/i.
Effective coupling constant, 164, 165, 214, Effective range, 209, 213
Ehrenfest theorem, 42
Eigenstates, of angular momentum, 176, 177 of charge, 129, 142 of Hamiltonian, for coupled oscillators,
10, 19,
line,
68
20
Coordinates, normal, 4 Core, repulsive, 155 Corson, D. R., 245 Coulomb field, 99
Lee model)
oscillators, 4, 5,
2329
Eigenvalues of Hamiltonian, for coupled oscillators, 1821 for free fields, 26, 39
for for
renormalized, in pion physics, 203, 209, 210, 223, 246, 255, 256 (See also Renormalization) Creation of particles, 9399, 208, 232, 235
10, 11
for neutral static sources, 84, for pair theory, 120122 for pion physics, 177, 178
;
8*5,
90, 91
247
INDEX
Einstein, A., 6,
271
142
Free
field,
commutation
246
Electrodynamics, quantum, 77, 98,
235, 236
1
17, 154,
Electromagnetic field, 8, 28, 46, 60, 61, 90, 9698, 100, 101, 125, 166, 232248 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, 2932 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
linearmomentum
213,214
1
of ground
state,
60
(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,
components
44
of
commutation
Friedman, M. H.,
rules, 42,
186/i.,
240.
Energy renormalization,
131
in
in
Fubini, S., 2l8n., 223/1., 231n., 253n. Fulco, J. R., 155/1., 229/1. Functional derivative, 7, 29, 30
neutral
91
staticsource theory,
Gartenhaus,
S.,
257 n.
61
Gauge transformation,
perturbation theory, 183, 225 strongcoupling theory, 195, 196, 225 TammDancoff treatment, 185 Equations of motion, classical pion theory,
Gaussian cutoff for source, 165 Gaussian distribution, 15 GellMann, M., 80/7., 202. Giacomelli, G.,209, 210 Glaser, V., 136., 256. Glicksman, M., 217 Goldberger, M. L., 80., 202. GoldschmidtClermont, Y., 246 Green's function, 71, 73
advanced, 72, 73 retarded, 72, 74
Ground
Vacuum)
FermiDirac
253
statistics,
Feynman diagrams,
Fields (see specific fields, e.g., Free field) Fluctuation, in number of field quanta, 55
57,88
in position of nucleon, 56 zeropoint, 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, 192194 radiation field, 235, 236 Schrodinger field, 33
Form
factor, 235
Free
field,
2331
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
917
Karzas,
displaced, 15, 16
eigenstates, 1012 with external force, 71
Keck,
J.
Ket, 13
ground
state,
1 1
KleinGordon equation, 24, 25 KleinGordon wave function, 3 Koba, Z., 78. Koester, L. S. 245
Kroll, N.,
240.
wave packet, 15
Heisenberg representation, 13, 200 equations of motion, 13 HeitlerLondon 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
Hamiltonian from, 30
for neutral staticsource theory, 75 for pair theory, 75, 100, 101 for pionnucleon interaction, 158
Hydrogen atom, 88
Hyperons, 154, 155
Ida, M., 144//. Indefinite metric, 138
Lamb, W.
E., \25n.
186.
Lee model,
commutation
renormalization constants, 192 strong coupling limit, 190, 191 Invariants (see Constants of motion) Isobars, 178
Low
equation, 139, 141, 144146 modifications, 150 neutron, 130, 131, 133136 phase shift, for TT~W, 140, 144
for 777;, 135140 sign, 136, 142
proton, 129131
relativistic version, 127,
renormalization,
137 of coupling constant, 136, 137, 144, 149 of wave function, 132134, 137, 138
Jeffreys, H.,
4.
136,
145,
148,
INDEX
Lee model, scattering matrix, 135145
states, bare, 129,
273
Moravcsik,
M.
130
Morpurgo, G.,
125/z.
continuum, 131
physical, 129131, 134136 scattering, 131, 132, 142, 143
T matrix,
virial
Lehmann, H.,
153/7.
Lindenbaum,
S. J.,
209
of vibrating atoms, 46 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, 166175, 188192 Neutral staticscalar theory, 71, 8292, 150, 215 adiabatic theorem, 83 asymptotic fields, 75 energy, 84, 85, 90, 96 energy renormalization, 85, 86
field
equation, 71
fluctuations, 89, 90
Low,
248/1.
ground
Lee model, 139,
141, 144
96
Low equation,
146
in in
McCormick, B. H., 124*., 125. McDonald, W. S., 245, 246 MacGregor, M. H., 256/7.
Machida,
S., 253/1.
wave function, 88 Hamiltonian, 8689 eigenstates, 86 interpretation, 82, 83 Lagrangian, 75 scattering, 83, 86 several sources, 9092
potential between, 91, 92
virial
Magnetic dipole transitions, 243, 246 Magnetic moment of nucleon, 219, 225, 230,
231
theorem, 89
(see
Neutron
Nucleon)
Mass
renormalization
(see
Energy
re
Nishijima, K., 196/7. Nishimura, K., 253/7. Nonrelativistic field (see Schrodinger field) Normal coordinates, 4, 5, 18
normalization)
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
Moment
171
Momentum,
260
charge distribution, 219, 221, 225, 226, 228, 229 current distribution, 225227 Feynman diagram, 215 intermediatecoupling theory, 186192 magnetic moment, 219, 225, 230, 231
momentum)
64
commutation
definition, 35
mesons in, 180, 181, 220 number, 219, 224 normalization of wavefunction, 181
numerical method, 196, 197 perturbation theory, 183, 184 physical, 179, 180, 181, 224231 quantization, 256 strongcoupling theory, 192196 TammDancoff theory, 184186
of
field
zeropoint, 43, 49
Momentum
274
INDEX
particles,
Number, of
4143, 75
charged, 62
density, 43, 44, 52, 54 fluctuation, 55
incoming, 76
for interacting fields, 76
highenergy behavior, 117 lowenergy behavior, 117, 118 phase shift, 111, 114117, 119, 122 resonance, 115117, 119
several sources,
virtual particles,
1 1 1
122125
of pions in nucleon, 219, 224 Numerical methods in pion theory, 196, 197
Oakley, D. C, 246 Omnes, R., 239/t.
Oneparticle states, 4752
wave matrix,
Pais, A., 192/1.
104, 1061J1
Parity, 40, 41, 155, 157, 158, 162, 236 conservation in pion physics, 162
Particles,
20
Onuma,
S., 257/i.
Oppenheimer, J. R., 78., 252/z. Osborne, L. S., 246 Oscillators, coupled harmonic, 4, 5, 2329 creation and destruction operators, 19, 21, 22, 24 displaced, 22 equation of motion, 5, 23, 24
generalized coordinates, ground state, 19
1
99
122125,
155
(See also
Pauli,
theories)
W.,
26/i.,
138.,
157/i.,
166/1.,
Hamiltonian,
18,
2325
eigenvalues, 18, 26
eigenvectors, 19, 26 harmonic (see Harmonic oscillator)
ground
state,
of neutral staticsource
theory, 90
Lagrangian, 29
normal coordinates,
18,
particle interpretation,
23 20
of nucleon, 183, 184, 189, 196, 221, 222, 225 for nucleonnucleon 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, 135138, 139, 140, 144 in pair theory, 111, 114119, 122 sign, 122
pion theory, 162, 205, 207, 208, 210, 214216 relation to T matrix, 205
in
quantization with, 112 time dependence, 108 wave function, 109 classical treatment, 100111 commutation relations for field operators, 113 energy of particles, 120 energy renormalization, 121 equations of motion, 102
field operators, 103, 104, 109, 112, 113,
242
cross section, 241247 dipole contribution to, electric, 234 ,243, 245, 246 magnetic, 243, 246
effects in, recoil,
122
245
rescattering, 239241,
244
energy dependence, 242244 Hamiltonian, 235, 236 Lowtype equation for, 237 perturbation theory, 238
INDEX
Photon, 236
Pi
275
4
mesons
(see Pions)
3,
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
Pionnucleon interaction, 153165 angular momentum states, 156158, 163 constants of motion, 161, 162 effective strength, 163165 equations of motion, 161, 162 Hamiltonian, 160 Lagrangian, 158 nonlinear, 155, 162 relativistic, 163 static theory, 1 56
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
Repulsive core,
in
Potential,
Feynman diagrams,
Ruderman, M. A.,
240//., 257/1.
233.
momentum,
250
Scalar
field,
196. 24, 40
field;
onepion exchange, 249, 250, 252, 255 perturbation theory, 250, 251 for point sources, 250
tensor, 250252
Neutral staticscalar
Scattering, in Lee model, 131, 132, 135150 in neutral staticscalar theory, 83, 86
in pair theory 109111 of pions by nucleons, 158, 198218
twopion exchange, 249, 253, 257 Yukawa, 249 Production of pions, 208
(See also Photomeson production) Products, ordered, 49
Projection operator, onto
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 intermediatecoupling
ChewLow
210
209,213
M., 57.
Feynman diagrams,
276
INDEX
Spherical harmonics, 3739, 159, 160, 220 expansions in terms of, 3739, 159, 160 (See also Angular momentum)
Low equation, 198, 199, 208 perturbation theory, 203, 204, 208, 209 phase shifts, 205, 207212, 216
in
158/1.,
159
196.
resonance, 198,213,214,216 scattering matrix, 198201, 205 source term, 199,200,213 7 matrix, 200208, 212 Scattering matrix, 7881 for neutral static source, 9395 in pair theory, 114
in
112, 113, 117 in pion theory, 171, 178 excited, 167, 170, 171, 190, 191, 196,
214
Schiff, L.
157.,
236/f., 252/1.
160
scattering, 131, 132, 142, 143
Schr6dinger field, 3133, 127 angular momentum, 35 centerofmass, 42 commutation rules, 32, 128
creation operator, 33 destruction operator, 33
twoparticle, 5254
virtual, 117
Vacuum)
231.
energy renormalization, 195, 196, 225 excited state, 196, 214 Hamiltonian, 192194 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 oneparticle state, 41, 48
twoparticle state, 5254 Schrbdinger representation, 12 Schrodinger wave function, 3, 13 Schweber, S. S., 59/i., 118/1.
Symanzik, K.,
218/i.
T matrix,
Schwinger,
Scott,
analytic properties, 143144, 201203, 207, 212 highenergy behavior, 145, 148150
M., 246
Segre, E., 224/z. Selfenergy (see Energy renormalization) Serber, R., 192*. Shortley, G. H., 67/i., 175/i. Signell, P. S., 257/1.
in Lee model, 140145 lowenergy behavior, 145, 146, 148150, 203, 207 in
119
Tamm,
I.
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,
250252
153/t.
Thellung, A.,
46.,
63/i.,
73/?.,
96n.,
21,
1
22
3
bound
state,
108
Heisenberg operators,
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, 106111, 132 Wave packet, 1517, 22 Wave property of field quanta, 56, 57
Wave
velocity, 7
Waves, outgoing, 97
Weisskopf, V.
247/1.
Tomonaga,
S., 92n.,
186.
R,
116/f.,
136.,
209/1., 230/t.,
Tomonaga
Toyoda,
T., 253/1.
Wick, G. C.,
146/1., 218/1.
Width of resonance, 115117, 174, 214 Wightman, A. S., 59. WignerEckart theorem, 181.
Worlock, R. M., 245 Wouthuysen, S., 78.
Vacuum,
39, 47,
49
bare, 77
polarization, 125
Yang, C. N., 154//. Yearian, M. R., 228n., 229., 235/f. Yukawa cutoff, 165
Yukawa
potential, 91,
249
46 Vibrating line, continuous, 68 Virial theorem, for Lee model, 134, 135 for neutral staticscalar theory, 89
line of,
for pair theory, 125 for pion theory, 222 Virtual particles (see Particles)
Zachariasen, F., 240., 247 n. Zeroenergy behavior of T matrix, 145, 149, 150, 203, 207 Zeropoint energy, 27, 49 Zeropoint fluctuations, 12, 20, 21, 27, 28, 89 Zeropoint momentum, 43, 49
Zimmerman, W.,
Zitterbewegung,
218/t.
63/i.