Академический Документы
Профессиональный Документы
Культура Документы
aOURCES, AND
ULIAN SCHWINGER
late, University of California at Los Angeles
P E R S E U S BOOKS
Reading, Marsaehusetts
Many of the designations used by manufacturers and sellers to distinguish their products are claimed as trademarks. m e r e chose designations appear in. this book and
Perseus Books was aware of a uademark claim, the designations have been printed in
initial capital letters.
Library of Congess Camlog Card Number: 98-87896
1 2 3 4 5 6 7 8 9 10-EB-0201009998
First printing, September 1998
Editor's Foreword
Perseus Books's Frontiers in Physics series has, since 1961, made it possible for
leading physicists to communicate in coherent fashion their views of recent
developments in the most exciting and active fields of physics-without having
to devote the time and energy required to prepare a formal review or monograph. Indeed, throughout its nearly forty-year existence, the series has emphasized informality in both style and content, as well as pedagogical clarity. Over
time, it was expected that these informal accounts would be replaced by more
formal counterparts-textbooks or monographsas the cutting-edge topics
h e y treated gradually became integrated into the body of physics knowledge
and reader interest dwindled. However, this has not proven to be the case for a
nllrnher of the volumes in the series: Many works have remained in print on an
on-demand basis, while others have such intrinsic value that the physics community has urged us to extend their life span.
The Advanced Book Classics series has been designed to meet this demand. It
will keep in print those volumes in Frontiers in Physics or its sister series, Lecture
Notes and Supplements in Physics, that continue to provide a unique account of
a topic of lasting interest. And through a sizable printing, these classics will be
made available at a comparatively modest cost to the reader.
These lecture notes by Julian Schwinger, one of the most distinguished theore tical physicists of this century, provide both beginning graduate students and
experienced researchers with an invaluable introduction to the author's perspective on quantum electrodynamics and high-energy particle physics. Based
on tectures delivered during the period 1966 to 1973, in which Schwinger
developed a point of view (the physical source concept) and a technique that
emphasized the unity of particle physics, electrodynamics, gravitational theory,
and many-body theory, the notes serve as both a textbook on source theory and
an informal historical record of the author's approach to many of the central
problems in physics. I am most pleased that Advanced Book Classics will make
these volumes readily accessible to a new generation of readers.
Bavid Pines
Aspen, Colorado
July 1998
Vita
$dim Schwinger
UdverSity Professw, University of CafoAa, md Prafessor of Physics at the
Angeles since 1872, was bosn in Nevv York City m
U~versityof Califoda,
fxis Ph.D. in physics fm
F e b m q 12, 1918, Profwsor S
Cdvmbia U~versityin 1939, He has itlm raived honotq doctorat= irr
from four iastitutions: hrdue U~versity(19611, H m a d U~vergity(19621,
Brmdeis University (19731, and Gustavus Adolphus &Bege (1975). fn addition
to teach8 at the U~versityof Califoda, Profesmr SGhwinger h trruCyht at
hrdue U~vmsity (B41-%-$31,and at H m m d U~versity (1945-72). Dr.
Sch~ngerwas a Rmewch Asmiate at the U~versityof C a E f e a , Berkeley,
and a Staff Member of the Mlllsachwetris Institute of Twbolow hdiation
Laboratoq. In, 1965 h o f a s ~ r
rwipient ( ~ t f iEchwd
F e p m n m d Sin; Itiro Tomon
in Physia for wark in
qumtum d e e t r d p
the C. L, M q r Matwe of Li&t Awad
vii
Special Pre
*&
Contents
l --I
1-3
1-3
1-4
Unitary Transformations
Galitmm Relatiuity
Einsreinian Relatiuity
Oitique o f Particle Theories
3-1
3-2
3-3
Mtcltiqinor Fiel&
Action
If you ~@~'fr_iiorin"em,
beat "m,
Sources,
S
PARTICLES
The concept of the p&icle has undergone drastic changes and generalisations
in the c a m e of $he hisb6cal development that l& to the! atom, to the nu~leus,
sad then 4x1 subnuclear phenomena, This htts aka been a progesrjion from
es~eotialynonrel&ivistic behaGor to an ulkra-relativistic dometin, It is interesting b appmciafe how mueh of the kinema.tical particle a t t ~ b u t e is
s implied by
the mumeb stmretuxe!of %herel8btivity goup of krm~fi3rnEfrtio~1~
among equivalent coordinate sy-stem%, In preparatioxr Ear this dimurnion we fimt review some
propt?r&iesof qusntum nneehanical unitary t r d o
1-1
UNITARY TRANSFORMATIONS
Quantum mechanics is a ~Mlbolioexprwisn of the laws of micramopic m@asurt?med. Stabs, situ~tiogsof optimum infornation, am represented by vmtom in
4; and physic~lpropdies by
a earnpiex s p w [left vectors (
A]. The freedom in physical
Knear H e r ~ t i a nomratom
.
on
de~riptiancome~pondsto the fretzciom of matlternaticaf repr6wntat;ian wswiahd
with unitary operators. These are defined through the Hermitisn adjoint
operation p by
trtu = uut = I
(1-1.1)
- - .C- - -
=( U
v-'xu.
Then d l numerical and adjoint relations among vectors and operatarn are
umhanged, We rredfy that
( 2 )= a
The adjoint relation~hip,
($1
(2lXIF) = (a'lxlb').
=
(1-1.4)
Id)'
i~
trsn~fomedinto
shows that the Hemitian operator A is mapped into the Hermitian operator a.
1
1%is produeed in one s$ep by the unitary operator U%&, in vvhich the multiplication order reflee& the sequenee of tr~nsformations,The appasih sequence
is repremnhd by U1U2,
and the two are eomgamd by defining the unitary
aprator that h needed to convert the mead equence into the first,
An infinikaimaf unitaq %ransfornnationis a transfarmation in. thr?:infiniksinral neighbarhood of the identity. It is represerrbd by
1 4- i@1121,
(1-1.15)
where
at121
-s
-Gall
(l/i)lGt, @zl
(1-1.16)
sx = (r/z">[x,
q,
An equivale~tfarm ia
u-lxu = X - &X.
where
are the parameters of another element of &hegroup. For unifary operabm the
existence of the irtvem ancl, of fhe identity is awumd. An i n f i k h i m l $mm
formation of the group with parametrtm &X, i~ comtmeM from
*ere the n finite Hemitim opclrators Qa are cailed the &c3mratomof the &roup.
One is free Lo redefine Cbt: generators by real nowindrtr Einmrr frm&~msbdiom,
with eorrmponding redefinitions of the prametem. On subjmtbg the infinitttsimid transformation operator U(6X)Lo an mbitrav unitav tradomatioa
of the group, we mu&$obtain another infirzibsinnal frawfarma$iorr. TKi irr
expresged by
~ ( k ) - ' ~ ~ v=
( hE
) @.a(&)@&,
(1-1.26)
b
where the numbem uaa(X)sre real, We shall also use a matrix m t a % i oin
~ Ghe
a-dimensional parameter ~paef3trnd write
The unitafy transformation is presenM e~lbmativefyad3
and
~(~~)U(XI)GU(X~)-'U(~~>-'
= U(X~)[G~(X~)IU(A~)"
= @m2)4(X1).
(1-1.32)
where
Ba
= -ga
T*
= ga
This gives
[G, Gbl = gbG = -G4b
and, if the irnsginary elements of the matrix gb are designated ss
The latter are a set of quadratic restrictions that must be obeyed by the numbere
g* the so-called group structure constants:
This cyclic structure also follows immediately from the cyclic form of the
+ [W&,
Gel, G,] + [[G,, G,],
(l-1-41)
= 0.
The ~tmctureeomfants speeify the compsition propedies of inhibsimal
par~mekm. Let blXPIXah the parametrsrzl of Ithe infinihsimal fmnsformation
that conoeetrs the two mquences in which Lhe tramformation8 labled by
and b2X, can be applid, Accordjng to the commuhtions relations of tho ~ o u p
gener~tom,they %regiven by
[[Gat
@blF
Gel
Gb]
p liq
q lip
+ &PI.
(1-1.47)
The onr respondence between. the unitary operators U(&) and the finite
matrices %(X), &(h) doe8 nst nece~sarilyincl& the mitary ehsraekr of the
Istfer. [Note that if the %(X) m;:unitary, or real odhagonaf, matrices, we have
%(h) == B(%), md then the Eiernnitian, or ima&ary mtkymnzetnisal r n a t ~ m
g. = ia.]
Since the structure of the g matrices csn be altered by changing the
generator basis in the parameter @p-, it is useful to have a basis-independent
ith which to judge the pomibi1iC;y of exhibiti
I;ian g matfices
%(X)rn%tfiw. If the mt of rr 42 operabrs i~
by tht!linearly
cornbimtiom XG, the g nnatficm ux1,dergo the 8ame liniear %rangtber with a similafity transfomation p r d u ~ e dby the nonsine;ular m a t k X. Bw&usethe trmt: of nnatdx produels is unchanged by the
Xathr $randomation we somider the real quadratic form
Ealilsan relativity
1-2
The invariance of the* numbers under unitary transformations on. the operator8
leads again to the form
rf= u ( x ) r ' ~ ( x ) ~ ,
(1- 1.54)
with the implication that the %(h)can be pre~entedas unitary matrices. The
corresponding Hesmitian matrices g,, whi~hare linearly independent if we exelude the uninteresting possibility that the group has an Abelian poup as a
factor, arc? an example of a finite-dimensiod realization of the G,, m the %(X)
provide a finite-dimensional unitary realization of the U(&). Conversely, if the
h
unitary [Hermitian] realization
matrix Tab is not positive-definite, no s u ~ finite
of the U(&)[&)can exist. A finite-dimensisnd realization of the group meana
that a finite number of staks can be found which are transformed among
themselves by all, operations of the goup. frz general, the aetioxl of zt unitary
operator on tz s t a b introduces new states, and the repeti-l;ion. of the operation
continues the prmess of producing additional states, This can terminae with a
finite number of states only if thztf repetition eventually cea~esto provide new
operatom, that ia, if the group paramebr sp~ect;is cio~ed. The distinction between a closed and an open group manifold is most familiar in that between
rotation8 and Lranslaf ions.
XE the matrices g, are Hermitisn, the stmetum constants gabc are antisymmetrical in a and c, M well as a and b, which implies antisymmetry in ;b
and c. This cornpieh antisymmetry can only be realized with n 2 3, For
n
3, EL, suitable normalization brings the im.a@nav structure constants to the
unique farm
gabe ==z i e ~ b c t
(1-1.55)
[G., Gal
=i
El23
= 4-1. The
C
e.acCc,
e
+ 6w X r + 6vE.
Note that the sign conventions are appropriak to the significance of 60, my, as
the displacement of the origin of the spatial coordinate frame to which a dven
point is referred. If it is the point that is moved by Sr, its new position relative
to the fixed reference system is r 6r. m e graup camposition properties of
this 10-parameter group are gpecified by compafing the wquence of tramformations
with those in the opposite order. The result of pedoming the .tramformation
sequence 1, 2, l l, 2-l, or, equivalently, for infinitesimal transformations,
I-', 2-l, 1, 2, i s
where
where K, I;, M are constants. The Jaeobi identity, iltpplicsd Co thrm mt8 of
infinitesimd treznsformatians, implies that
&v are
where we have adopted the summation convention for repeaM indices; here,
the index m = 1, 2, 3,
The eommuttllor of two generator8 ean be inkvreted, in two dbrnative
wntys, m the effect of an infinitssimal unitav transfomatian upon an operabr:
The eommutaltors involving the anwlar momentum apratior, for emmpIe, can
be w ~ t k nans
.
sUJ= (X/.i)fJ, J * 401 = 6c;~X J,
(1-2.13)
6,P = fI/z'>[P,3 Gccr] == 6 0 X P,
BUN = (I/i)[N,
J 6 4 -- 6 0 X N,
which state the rmponse of a vector to infinitesinsaf rofstions, and
&,B= (1/$7[H,
J 401 = 0,
(1-2. r4)
d the nstum
The response of the anwlar momentum to tramlation is in m ~ o r with
of angular momentum as the moment of the linear momentum, and indicabs
the exisknee of a pasition, vwhr operafor R. such that
(l/$)[R,P . 6e] = 8e
or
1% Pll = a'Skl*
We therefore write
where S is( a translationally independent contribution to fhe ~tn@ar marneaturn.
This is inbrnat a n d a r momentum, or spin. The carrect rotratianal re~ponmaf
P is w u m d by this con~truction,and that of R will follow if the components of
R are mutually commutative, and if S commubs with R weEl ss with F. To
produce the proper rotational hbavior of S, these: operators must the
obey the snljgufar momentum eonnmut~tionrelations, which; we c m abo w r i b a ~ a
S X S = HS.
(1-2. a0)
The consequence is
vvhieh gives the decomposition inta energy of motion of the whale rs;vstem and
internal enerw, The latter will generally involw indernd dynamkal variables,
which commute with R and P, combined in sueh a way that Hint is invariant
under the rotation generahd by the inbrrral angular monnenlum S.
An dementary particle is a aystem tk'ifiout inkrnd enere, or a t lea& one
for which inkrnal enerw is effectively ined under the limikd physical eircumstances under consid64ration. Let us consider n elementary gafiieles, each
described as above by variables re, p,, Sa and mms m,, a = 1, . . ,n. The
operators msoeiated with digerat pafiieles commuh, The krretie traxrsformation generators of the whole sysknt are then obtained additively as
where
and
(pat -- m.~,) = Pt
-- MR,
12
Psnl~les
Chap, 1
The oprators for the total sy~teznhave the required propeeiesi. Note that the
inbmal vafiables introduced here are not l i n e d y indqendent :
I,,
1-2
Oatilsan relativity
13
and
where
In deriving the lmt equation we have omitted such commuta;tors as [(e/e)A, ecp].
This is La be validated, not as a clmsical approximation, but through the
negligibility of the dynamical reaction back an the externat system. We also
note the commutation relation indicated by
Analogous electric dipole moments have never been observed. One value of the
gyromagnetic ratio g has a special property. In a homogeneous magnetie field
the veloeiLy and the spill vectors precess about the field axis. The two p r e ~ w i o n
mtes are equal if g =.I 2. The observed values of g are very sligfitly in excess of
2 for the electron f2(1.001tM)] and Lhe rnuon [2(1.001166)], but differ widely
for other padiele~.
1.
r"
But since the Hamiltonian is no more than quadratic in the momenla, symmetrized multiplication enables one to write
d
f (r) =
dH
= v Vj(r),
C~P
[f (P), H ] = Vf(r)
--
(l-2,52)
One easily verifies that it is the rotation generator. Them is an impadant conmquenee of that fact.: Consider the coordinate wave funetion repre~entixlgrt
p&rticular sLaLe, (rt[ ), and perform. a coordinak system rotation about the axis
provided by r, An infinitesimal rotation ifs given by
1-3
Einstginian relativity
15
and the known limitation to single or double valuedness for rotation throul,Th 21r
radians implies that eg/e is either an integer or an inkger plus i.
As a discussion of magnetic charge and its implications this is quite incomplete. However, the special opraLor system given in Eqs. (1-2.471, (1-2.481,
and (1-2.52) will soon be encountered again in a very difierrznl physical context.
The new feature associated with. the finiteness of c, the speed af light, is the
abandonment of absolute simultaneity, It is replaced, for infiniksimal trarrsformations, by
where &c0 is the displacement of the origin for the variable cl. We now designate
the space-time coordinates collectively by zp = et, r, where z0 = --X@ = et
and zk =r= a = r k . The infinitesimal coordinate transformations of the Einsteinian relralividy g o u p are
where
ady=
- &oVIL.
and
c ~ =o H
+M C ~ .
16
Chap, t
PartZelss
?Mpremnkd, as
i ~ ody
g tbs
Elnstelnian relativity
1-3
17
L C ~ ~ J . ~
forms the tensor dual to J"" with the aid of the totally antisynmretricaf ternor
srpcified by
e0128
-M
(1-3.21)
E
Thk invwrietnm prowrty fsllows from the tr~wlationaIresponse of J,x, aad the
antisymmetry of c'"',
We also n o h that
&W" = 0.
The sealiar
W 2= WPWr 2 0
is invariant under aEE Lorentg tr8nsformations. As indicated, the vecfor W',
being orthogonal to PC,eannot be time-like. The @ommudation,relations among
the components of WIrare
The behavim under wordinate disglaeements that is pregexlkd in the
equations
(l/i)[J, P 6 4 = 6r X P,
(l/i)[N, P be] = -- drPo, (1-3.26)
again indicates &heexistemer: of a position operator R, obeying
N .- PX@--- PR,
(1-3.29)
(1-3.33)
As such, S m u ~ commuh
t
with R and P while itself obying the a n v l a r mos
i also necessary to supplement P4 irt order to
mentum commutation mles. It r
generab the spin h r m of the commutator
A suitable form is
-- P'R
+ a(P")S X P.
(1-3'37)
-~P@R
X [a(S X P)]-- i[a(S X P)] X P'R
a2i(s X P) X (S X P)
= ( d a / d P ' ) ~X ( S X P ) 2P0& - a2[pZf5
- P X (S X P)], (1-3.38)
We cone1ude that
a(Po) = (PO
the alternative choice with ( P o - M)-'
The 6nsl form is
J=RXP+S,
PZO
+ M)-',
being singular a t P = 0, P' = M.
- PR
M SXP,
whieh incidentally shows that S' is B Lorenta invariant. It is worth pointing out
the converse, that operators with the slat& properties of R and S can be eonstructed from the Lorentz generaton (zO= 0):
or
w 0 =P+,
W = P@S-
(l-3,451
+
M P x ( S X P ) = M S + ~ , + P~ P - S .
W 2 = M'S'.
This discussion refers generally to M 2 > 0. We next consider the limit ss
M' -t 0 for fixed .'S The resulting relation
is a Lorents invariant. This quantity is the component of the spin along the
direction of motion, or the helicity of the particle. In view of its invariance, a
phy~icalsylshnn need exhim only an@value of helieify, or, if ~pwemfieeLion
parity has a meaning for the interactions of that system, the pseudoscalar h osn
have two values, fs. The photon, with s = 1, illustrates the latter situation,
white for the neucfinos, with g = 3, X = +s and -a refer ta emntiaUy dZ~?ren%
p&&icfes, f f only one helieity vatue is meanh@ul, or with s 2 1, even if h = trl
and --s are both resliaed, not all of the 28 1 spin magnetic quantum number
ststes exist. Accordingly, the operator S cesses to be defined (with two excep
tions) in the limit M -+0, and we must introduce m w vaGabfes for this eircumstance.
f n. order to deleb S we define the new position vector
k XP=
Then
P+S
J=~ZXP+X(P/P@),
-- PPeS/(P0)2.
N=Pz@-POR,
(1-3.52)
(1-3.53)
and to complete the verification that only X appears explicitly, we give the
contmutabr
M&*
I n the situation we have just discussed, W' = 0. There is another logical
possibrtity. With X = P SIP@assuming any accessible finite value, let S
' -+ a,
as M 1 -4 0 to produoe the limit
The h v a ~ a n t
1["2 ,
,
(l/i)(hP
+ T) X (XP + T) = P-,
(1-3.63)
R =R
S X p/(PO)'
f 1-3.M)
0,
MR
= T X P/(P')'
and R
' --t m with vsnishing M. However, M R = 0. The e
htion
mhfiom tN X N = J continue to be obeyed despite the bLrodue1ion of the T
brm, since
QX
(T X P / ( P ~ ) ~ )(T X P/(PO)~)X
= 0.
(1-3.69)
This faet, together with the unbounded nature of the h spectrum, ranginf; over
all inkgers or all integem
indicates that physically acwssible states would
exist for which (bk12is arbitrarily large.
We suggest the following verbal principle for massless particles: A zero
mass padicle is not completely XoealizabXe, but a finite degree of localizability
exists. The principle has the: following valid consequences. There is na aginless
zero mass partieb, for the eomrnutative position veetor R would be available
The same reasoning exelude s = 3 massless pa&ie1w for which space reflection
parity is meaningful. And the systems we have just discussed, with W 2> 0,
are condemned wholesale by the existenw of state8 that are unlwalized without
limit. There is a simple pattern for the kxtown or strongly caxljectured maesless
garlieles; their spins are given by 8 = Z4, cr = -1, 0,+X.
The conwpt of elementary particle in rellitivistie mechanics remains an
operational one, Lhat under the eonditiorzs of physicafi exeifa$ion available, if is
consistent to @sign a unique vazlue ta m-, spin, and ather ehttracterisfic
invariant a(ltributes of the system. Far a set of n noninkraeti~gp & ~ i ~ the
le~,
Lorentz generators of the whole system are giwn by the additive forms
++,
The operators R and S for the total system must be obtained from the conatmctim (1-3.43); one is not likely to produce them by an a pfiori definition,
Consider, for example,
1-3
Einstainian relativity
23
with the usual symmetrization understood in the last term. Thus, for a singleisolated prartiele, we have
The constraints
are compatible with the equations of motion, a t least to term8 linear in the field
strengths; this involves only the commutation relations for a free particle,
Why bid we not begin with a general theory of inkrating particles, specified
by variables r,, p,, S, a = l, . , . , n, and then proceed to follow the mation of
one particle under the infiuence of the others, as in the nonrelativistic discussion?
Quite simply, because no such &;eneraltheory exists, Apa& from the obviously
formidable algebraic task of stating the relativistic conditions on in_leraction
brms (smdl, deviations from nonrelativistic behavior pose no problem), the
atternpt founders on the failure of the assumption that there is a fixed number
of parlEticles, The relation between reiati~sticand nonrelativistic energy can
be exhibited as
1x1 fhe nonrelativistic limit where changes in H are small compared to each ma,
the conservation of P' generally demands, first, the conservation of each N,,
and then, Ghat of H. But if the kinetic and interaction energies contained in H
1-4
26
/.
is assured if
TP'(z) = T'"(x).
The three-dimensional form of these operators is
The tensor transformation response of the stress tensor Tp'(z) to an infinitesimal Lorentz transformation is given by
The brms involving two or more derivatives are such thizt l-hey do not conffibule wbem intepations are pedormeb to eonstmct one of $he Loresnde generafors, We have indicated only the minimum n u m b r of d e ~ v a t i v erequired;
~
more generd possibilitks are introduee-d by appropriitb generaliaa;tion of f , 8,
snd h, The% three functions are symmetrical wikhin each pair of indices, as
iZlu~tr~ted
by
(1-414)
f n r r f p p - f nm1plr = fna.rrq
~fhile
f and h are antisymmetrieaf under an. exchange of the pairs, as in
Anokher relation is
-aOf m n * p q ( z )
Bmnrpg(Z)
Q ~ ~ * ~ ~ ( c c ) .
(1-4.16)
(1-4.21)
V-W(Z)=O,
(1-4.23)
Then, since
-&V
- xP)Hm(z'),
H(%).
(1-4@24)
(1-4.25)
V E = j',
(1-4.27)
3%
Particles
Ch~p.3
and ecmstruet
This is the basis for an uneeftainty pfineiple stattlment about the aceuracy with
whieh values of T 1and !F2can be assigned in a even state. We firgt consider an
application where f m"nnppa
does nod enter, in comquence of the antir~ymmetryin.
the two sets of indices. Let T 1and ZT2 be partitions of Lhe total e n e w operator,
so that
(1-4.32)
@I(x)
4-@ 2 ( ~ zz=) 1.
Since derivatives of
v1
and of
v2
Now choose vl (X) to be a, unit step function, defining a semi-infinib redan which
shares a, surface \dth the complementary volume defined by etz(x). With clS an
element of area directed from the latter volume, we get
This gives a eorreet acmunt of the rate a t which the enerw in each p r t i a l
volume changes, owirtg to the enerw flux acroas the common surface. Irxeidentally, if the domain8 defined by v, and v2 had been regarded as disjoint but
approaching contact in a limit, the value abtsinied for the Il.igh&band side would
have b e n zero, while, if they had initiaEXy overlapped the eventual boundary
and then the cornman. volume had approached zero, the limiti~~g
value of the
1-4
2r)
right-hand side would have been twice the stated one. Thus, an alternative
evaluation uses the average of the two limiting definitions. Ar~stherchoice of
weigh%functions is
@
(X) .l @(X),
VZ(~) xkv(~),
(1-4.35)
which gives
When @(X) is a unit step function that defines precisely a finite volume, the
operahrs T t , T 2are the asmciated energy and its first moments* But no meaning can then be assigned to the products d,, d,v d,u, which calls seriously ints
question. the eonsiskncy of any operator field theory for which. $nanlpS1 (2) @ 0.
This gives a pPivileged position to that limited clslsl; of fundamental field variables
for which f
does vanish,
The impact of this result is only slightly weakened by the follojving property
of physical systems that have vanishing
The funetion
fmnlpq.
+-
TO'
(g),
= ~(TP'T).
aCIu(x)
(1-4.43)
P""z@
and X is evaluated at the coordinrtte origin. The state x(z)[ )
where P;E:
is produced from the vacuum state by a localized excitation. To study the
pfdSele aspects of this excitcltion tve examine its spaert-time proipltgation
characteristics through the correlation %vithan analogous excitation having a
digerent localization :
=2=
The unitary operator fhnt describes the displacement from z' to z can be
exhibited in terms of i t s eigenvatues and the associated non~egativeHermitim
projection operators,
where
( d p ) = dpo d p , dpz dp3.
The values of fl that contribute to the integral, thwe for which F(p) f 0, must
conform to the physical sp~etral.requirements,
With a given threcdilnensional momentum, (dp012 = d 3 f S pso that
The di&rentjal dcJ, is an in~rztriantm o m e n t m space m s u r e on the hypersurface --P2 = M 2 . This gives
is a real, nonnegative function. The state X / ) selects from P(p) the subspace
with the angular momentum propexrlies implied by the rotational behavior of X ,
snd f ( p ) j.I. 0 at - P 2 = M 2 a~sertsthe existence of an excitation with those
physical parameters. Merely for simplicity, we only consider a scalar field X ,
which limits f ( p ) to dependence on the scalar --p2.
There are three qualitatively difierenl possibilities that can be realietxi in
f(p) = f(MZ).
For a given spaLial momentum, the time dependence of the field correlation
function contains the isolated frequency 'p = +(p2
m2)
This excitation
i s s stable pareticle. We n o b thaL should ~ ( sobey
)
a finite-degree differentid
equation,
ul(-a"~(~=
) o,
32
Psrticlss
Chap. Z
in which
is the energy computed from mass m. Owing to destructive interkrence, the amplitude of this oscillation will drop substantially below its initial
value when
(P~/~)(I/P)*
(1-4*37)
This is an unstable particle decaying into several other particles with a proper
lifetime -l/I".
where
exp[ip.
(X
-- X')].
(1-4.59)
If the field ~ ( zis) a fundamental dynamical variable, its equal time eornmutrztion
relations have a, kinemaitical bwia, I"cs typical af a sedar field thaL
and the fatter implies the sum rufe
Imagine no\' that the field X ( % ) is uncoupled from all others, and then.
obeys a linear digerentirml equation that gives f ( M 2 ) = b ( M 2 - m:), Supipom
that when the physical couplings are restored a stttble particle still exists. Its
mass will be shifted by the interaction, m. -+ m, and f ( B f 2 ) will have multiparticle contributions in addition to the discrete mass term: f 6 ( d f a - m').
The sum wle thus requires that f < 1. If we are not inkrested in the details
of the particular excitation used to generate the psrticfe, but wish only to
describe the physical particle itself, 11-ediseard the eonfinu~usmass cantribution
to ( X ( Z ) X ( Z ~ )and
,
comespondingly adjust the scale of the correlation; function
by removing the factor f , This example has supplied the designation for the
general proceduw that transfers attention from the fundamental dynamieal
geld variables to the derived phenomenalogical ptarticle level. It is called
renormalisation,
More elaborate field earrelation functions give information about the
inkraetion of particles. Consider, for example,
84
Particles
Chap, l
from which the latter can be constructed, in principle. Two radically different
situations occur, in practice. In the first, the interaeticlns are suaciently weak
that the particles of interest appear in the excitation spectra, of the fundamentstl
variables themwlves. This is the ~zssumedsituation in quantum electradynamics
where the particles are the photon and the electron (or muon). The equations
obeyed by the field correlation functions can be solved by ~ r t u r b a t i v eor iterative methods bawd on the mallness of the eharaeteristi~coupling constant
a = 11137.Q.8. The result@may be presented in, twa different ways, a t thct
unrenormrtlized field stage, or at the renormalized particle leveicl, The fieid
version contains divergent integrals, the renormalized stabmenti3 are finih an;d
in exceptional agreement with experiment, The fairly rapid convergence of the
renormalised expressions means that experiments do not probe to very high
moment%or very short distances. The underlying hypothesis of operator field
theory concerning the conceptual possibility of descriptions at arbitrarily small
distances remains unksted by the available evidence. This hywthesis is involved in the unrenormalized, field results, but whether &henonexi~$enceof the
expressions signifies the failure of the hypothesis, or merely the inadequacy of
the perturba'cive ealculational methods that are used, is premntly unknown, It
may be that opemtor field theory is unnecessafily dognatie about the physical
silpificanee of arbitrarily small volume elements. Totafly new concepts might
enter a t very large momenta, without altering the practical succesws that have
been obtained.
The other situation is that of &rang interaction physics. Here %hehypothesis
that whole families of particles are the dynamical manifestation of a feu?fundamental field variables excludes the possibilit;y. that the excitation spectrum of
the latter contains the known. par%icles. These objects must b generakd by
combinations of the basic variables, Being iporant of the underlying dynamics
of the fundamental dynamical variables and lacking the computational methods
that could @v@the consequences of that dynamics, if it were known, one musk
fall back on speculations concerning the composib field structure of the known
partieles. And such speculations become interfwind in the more immediate
problems that are presented a t the phenomencrlo@cd level. Is it not possible
to separate particle phenomenolqy from speculations about particle stmcture?
2. If the particle is the ultimate structure, no detailed description is passible in
1-44
36
88"
1,
(1-4.66)
thereby defining the transition matrix T. Only the momenta have been made
expfieit here, and the delta function is four-dimensionsf,
The resulting farm of the unitarity condition is nonlinear, relating matrix elements of i(T - T?) to products of T' and T matrix elements. The probabilities
of transitions must have a Lorentz invariant significance. Xt is therefore asserted
that the S, and the T,matrix: elements must be invariant functions of their
arguments, This is partieuIarty simple %.hen slf parlicles are spinfess since it
requires that the '
I
'
matrix elements be a funetion only of the independent
sealars (we ignore the possibility of pseudoscafars) that can be formed from the
N = n -t- nhomrnenta, which are individualy subject to the particle mass
relations, -p: = m:. The number of such scalar combinations is 3N -- 10,
where the subtracted number counts the Lorentz trrtnsformatioh parameters,
Thus, in two-pa&icle reactions, where N = 2 2, there are only two indeCo energy and scattefing anlSfe in the
pendent scalar variables, c~rre~ponding
cenkr of mass reference frame.
The constructive principle of S-matrix theory is the gostufab of analyticity.
It is =sunned that the physical reaction amplitudes, in their dependence on the
SOURCES
The critical eornmentrs of the last section set the stage for the introduction of a
new theory of particles. I t is a phenomenologieal theory, dmignd to describe
the observd particle@,be %heystable or unstable. No speculations about the
inner stmeture of pa~ictesare introduced, bu"che road to a conceivable more
fundamental theory is left open. No abstract definition of particle is devised;
rather, the theory u8es symbolic idealization8 of the reali&ie procedures that
give physical meaning to the particle wncepL. The theztry is thereby firmly
grounded in %p&ee-time,the amna within which the experimenter manipulates
hia tools, but Lbe question of an ultimate limitation to miero~eopiespace-time
dmchption is left open, with the decision re~mvedto experiment. Correspondingly, no operator fields are ursed. The compbmentary momentum-space
description plays an important role, but the possibiliw of ultimate limitations
on this space is not exciudd, and there is no appeal to andytieity in momentum
space. The constmetive principles of the new thwry are intuitive one&causality and uniformity in space-time. What emerges is a thcsory inkrnedict-te
in position. between operator field theor;)r and 8-matrix t'fimry, which rejects the
dogmas af each, and gains thereby a caleuXationaE erne and intuitiveness that
make it a worthy coatender ts displace the mrlier formulations.
The range of the term "particle" h.m been systematicalfy extended by
expe~merrtaIdiscovery. From the stable electron. and proton, to the Iong-lived
neutron, do the rapidly decaying ?r and A, to the highly unstable p and N*,
there has k e n a progression to more energetic and sfiorbr-lived excitations* It
Is now the normal situat.ion that a particle must be creawd in order ta study it.
And, in a general sense, that h also true of the very high energy stable particles
produced in accelerators. One ean regard aff such creation acts 8s cofli~ions.
The emence of such. a collision is that it occupies a finite, and to some degree
controllable, space-time reeon wherein other particles combine to transmit to a
particular one thee properties that call it into exisknce and uniquely characterizie it. If i s psrt of the experimenter's creed that a new resonance not be admitted
Lo full status as ra particle until it h a been ob%rved wifh the same charae-t;eiristies
in a number of diflerent reactions. Thus, if a pa&icle is defined by the collisions
that er-ettb it, the details of s sgecifie reaction are not; relevant and one ean
idealize the role of the other particles in the coIlision, recognizing th& their
funetion is solely to ~uppjtythe needed baiance of properlies-they eonstitute
the source for the particle of inderest. What survives in. the idealization is a
37
38
Sources
Chap, 2
The elementary acts ta be represented as the effect of a source are the creation
of a single particle tvhere none existed previously, and the annihilation of that
single particle. Since the actual presence of other particles in realistic collisions
is abstrackly portrayed by the source, the s t a k s appearing in corresponding
and ( o + ~ I , ) ~are:
, /Q-), the vacuum
quantum mechanical amplitudes, (1,\0-)~
state before the operation af the source K; (Q+[, the vacuum s h t e subsc;quent
to the operation of the source: (sink) K; and (l,j, II,), describing a single particle
state in lvhich the momentum is specified rr-ithin a small volume element (dp).
The connection of this discrete labeling with, the continuous variable specific&Lion of momentum states is
The individual creation and annihilation acts are not analyzed; the source is
defined as a measure of the whole process, as suggested by (~veanticipate a
particular variable factor)
2-1
31)
R(x) K ( z -4E
X),
(2-1.4)
or
X(z)=K(z),
Zfi=z@-X'.
(2-1.5)
= eiPXK,(p),
R.(p) = @ - ' P X K a ( ~ ) ,
(2- 1.6)
Now,
which shows clearly that the relation between the complementaryr coordinate
and momentum source desriptioas is given by Fourier transformation:
+
+
In exhibiting N we have set z0 = 0,sinee this is the origin of time, and used an
alternative form of the symmetrised product of r with
The coordinate
SQ
Sources
Chap. 2
vvith a similar statement for K,@). The implied infinitclsinnd change of H,(%)
or
Ka(x)
if3
6K(z) = 160 r X V
6~ * (ra,
= 6%" (~,K(z)~
+ Z'V)]K(Z)
(2-1.13)
where
6z" = 6 d r x , .
(2-1. f 4)
fZ(z) = K ( z t 8 4 ,
(2-1.15)
This result,
or
R ( z ) = K ( ~ ) ,~ " = z ~ - 8 z ~ ,
(2-1.16)
when conzbined with the displacement response, assrrrb that the soume functions
of qinless pafticfes, K ( z ) , behave .dsr scaler functions under the tramformalions
of the Larente poup.
An important corollary is that the ehoiee of K(z) as a real flanetion would
have a Lolrent~-invariandmeaning. That is in, sharp earrtrast with the nonrelativistic situation, where N -+ -mr and'p + p2/2n. Then, if we consider
on1y boosts for simplicity,
and
The implied form for finite transformations is
imr f tV)]Ke(r, t )
r -- i&mv2t)~(r vt, t )
(2-1,19)
2-1
41
in which the additional terms are negtigitale urrder weak source conditions.
Furthermore, it suffices to use source-free values for the factors of ( 0 + [ 1 , ) ~ and
(~-jl,~
= )( ~
I ~ # \ o - ) ~namely
*~
expressirrg tlre ixlvarianee of the vacuum state, and
apart from phase factors \\.hick serve only to ensure dllat, in the resulting
relation,
( o + ! ~ P=
) ~-- ( l p l ~ - ) ~ * ,
(2-1.25)
both single particle states are referred to the same space-like surface. Tliis
connection bettveen ereation and arlnihifatioxr probability amplitudes earl also
be presented as
i(o+i
= [i(lpl~-)K]*.
(2-1 .xi)
Thus, with a permissible choice cf arbitrary phases, the source functioxls K,(s)
sad K,(%) are reciprocal complex conjugates. The simplest possibility, and thc
one with r~hiehr,ve begin, is
a real function, We. now unite the various detaiis and state our explicit defini-
@P) 1
K , = ( ~ U , ) " ~ K ( ~ ) ,dw, = ( 2 ~ 1 32p0
and
The experimenter's bbasic tool is a beam of particles. A very weak beam of
~pinfeljsparticles has the following causal reprrasentation. We begin .rvith tbe
vacuum slab. Then rt weak soume IK2(x), occupying a finice apace-time region,
goes into action. It most often does nothing, with the associated probability
Sources
Chap, 2
amplitude ( 0 + 1 0 - ) ~ 2
1, and occasionally produces a single particle, as
characterized by ( l , l 0 - ) ~ 1 . After the emission source has ceased to operate, the
re~ultingvacuum or single particle state persish unehawed until we e n k r the
space-tim region of an absorptian source K l ( s ) . Its effect in detecting a single
particle is described by (0+j1,)~1, and we thus return to the vacuum state. The
eompleee pmeess is represented by
where rt, refers to the weak 8ouree limitation. The functions K l ( z ) and K2(x>
are the disjoint parts of the tots1 source in this situation, which is given by
There should be nothing in the overall description to distinpisk one component
part of the source from anodher, aside from reference to the space--time region
that it occupies. This is space-time uniformity. It implies that ( o + ( o - ) ~
depends only upon K, and in the manner made explicit by the hilinear structure
in K l and K 2 . Accordingly, we write
=
[We recall that @' is the energy-momentum veetor of a particle, so that
+(p2
m2)1'2.] From these characteristics of h+(%- g'), we deduce that
2-1
The explicit constructions of A+(z - s" may appear to refer only to c~lusal
or time-like relations between the points z aad d.But in fact the_ygive meaning
ta this funetion. everywhere, The only possible dificulty would be that, when z
rand sf are in spsee-like relation, where causality hits no invariant meaning,
different values might be obtained depcfing Upon the choice of coordinate
system. This does not happen. Since dw, and eke"'-"'
are invariant structuresp
there is no harm in choosing a coordinate system for which z0 = so', and
is indepndent of the ambimous sign, for the integsal depend8 only upon
(X - X')%= (z - z ' ) ~ .As a result, there is no longer any indication in (2-1.35)
of the initial causd arrangement of sources, and that stmcturr? is applicable to
an arbitrary disposition of ~ourcm. This space-time extrapolatim must meet a
mvem k~t,
however, We are nomr able to compute the probability that, despite
the intemention of the sources, the vacuum state persists, It is
(dcc)(dz")K(x) h ( I/i)b+(z - z")K(z'),
(2-1.40)
where
is vdid e-rrerywltere,and the reference to the real part is uranecesssry since it is
implicit in the symmetry of the quadra;tic form, But probability considerations
also demand that
Chap. 2
Sourcsa
by the 8ubsdilut;ion
ilzD -
-4
lz4 - z:I,
\vhieh requires that the ordering of the real numbers zO,zO'is mapped into the
same ordering of the real numbers Q, X:, We remove a factor of i in &fining
2-1
46
Even better, since it reproduces the eorrecf limiting forms, is the inequality
One can; connect the Minkowski and Euelidean. descriptians by equating the
source strengths associated with corresponding voluple elements
and the right-hand side is a real number, which, is less than unity.
The physical vacuum amplitude can also be reeo.verc3.d from the Euelidean
version, by the earnpitex substitutions
2 4 -r
iz0,
p*
-+ -ipop
(2- X. 58)
p4
-r
the
exp
(p,)2
+ m'
-+
p2
(2-1 *W)
in which, despite various scale changes, E retains its meaning as a, prameter thaf
46
Sources
Chap, 2
(2-1 .fix)
where
2-1
47
would still be given by (2-1.37) and (2-X.38), but differcent masses would appear
in the two causal forms, Then we could no longer conclude that B+(z - 2')
had a unique edrapolation into spaee-like regions. I t is the principle of spacetime uniformity that demands equal masses for the two kinds of particle, which
are identified as particle snd antipartick. The Euclide&n postulate produet38
the same conclusion through. the absence of an invariant distinc-t;ion btween
24 - Z
: > O and 2 4 - 2; < Q, which permits only one mass pammeter to
appear.
In view of these remarh the definitions that relah sources to single-particle
production and annihilation prObabi1iw amplitudes mu^& be extended ta
where & distinwish parCicle and ranfipadicle, and
Note csrefulf y the distinction between
Accordingly, we have
Chap, 2
Sources
48
to the=
p h s e transfomstions, combined with a rigid displacement of the source, by
X@,we get
( ~ ~ * p--,-ef) i@eipX(lp+lO-)X,
~
(2-1 '77)
whieh makes explicit the mehanieal and "charge' attributes of the singleparLiele ststes.
An dterrtative presentation is obtained by replacing the complex souree
K ( z ) with two real murees, a c c d i n g fa
K ( z ) = 2-1'2[K,i~(z)
- iK,,,(z)I,
This @ves
We have now exhi*bit& twa indepadent murces, with their wsoeiated particles,
But the fact that these particles have the same mass (and spin) implies that the
dwompo~itioncan be done in an infinib variety af ways, camesponding to the
p h w transformation8 of eomplex sources, which naw appear as two-dimensional
Euelidean rotations :
and
i8
19
do not produce single-particle states of definite charge. They refer to the eomplementary property of charge symmetry-the states turn into themselves or
their negatives, respectiveIy, when positive and negative eharges are interchanged. A matrix presentation of this transformation is
K(s)
rqK(z1,
:T"":
-Qrq.
Euelidean. transformations decompose into two conneeted pieces, distinwished as proper and improper trsnsforrn&ions. In, contrast, the full Lorent~
group contains four connected pieces, owing to the discontinuous causal dislinction between z0 > O and z0 < O. The wider invariance introduced by the
Euclidmn postulate thus enables one to perform some discontinuous LarenL~,
transfornnations through the intermediary of continuous Euelidean transformations, The mosL important example of that is
fSO
Sources
Chap. 2
but it is the Euclidean postuiaite that suppliea the general btzsis for this inv a ~ a n c e , The reflmtion of the time coordinate inverts the causal order of
sourem and inbrchangrzs creation and annihilation. This is evident from the
momentum form of the source transformation,
K(P)
K(-P)?
and thus
K,,*K:-,
K,--K:+
or
According to its csnstmction W a time refleetion (I')and a space reflection
(F-arity), which also hm the egect of inhrchanging parCicle and %ntipa;rticlea
charge reflwtion (C)-this process is often known rts the TCP opration, but
one is Hkely to encounter any other permutation of the three constituents.
Perhaps it should be called the Shell g a m .
2-2
. ..
(d.)(dg')K,(z)d+(z
- xF)Ka(rF)
2-2
@l
which asseds fhe &bf~enceof coupling between diEerea6 singk particle exchange
re@ons. Then, since the individual sources are weak,
The slam form ztpplies to two-component real sourees, and for empXex source8
if bwomea
The ems& arrangement also ennsbles us do analyz;e the eomplete proem into an
initial multiparticle emission act, represented by the probability amplitude
(O+~O_)~.,
(2-2.1 1)
where the multiparticle label is realized by the collection of integers {B,). The
avidenf interpretation of n, is a particle o e ~ ~ p a f inumber
o~l
a~aciftt;edwith the
indicated physiesl properties. This is confirmed by the response of the multiparticle states to the source translation K(,) + K(% X), nvhich gives
((3
Io-)~
eiPX((n)Io-)~,
(Q+/ (n))K
(@+l{R)
(2-2.14)
displays the additive contributions of the particles present in the state under
consideration.
The probability amplitudes must meet the following total probability or
complekness test :
and indeed
2-2
63
No& how the vacuum amplitude has been used in two distinct ways. Through
the consideration of a causal arrangement, relative multipartick annplitudcs are
obtained :
where q == &l is the charge label that distinguishes partiele and antiparticle.
0 ~ ) --+
~ eiQveiPx( $4 Io-)~?
(O+I jnj
)K + (O+I {R)) R e - i Q p e - i P X
(2-2.21)
where
C
%,,g,
PQ
C
PQ
%p,$
(2-2.22)
The momentum labcling of individual particle states is naf the only passibility. A spherical or angular momentum specification is introduced by the
transformation
K, =
(dn)'I2 Y~,(P)KP@Z~P
where
Im
and jr, Vz, are standard symbols for spherical Bessel funetions and spherical
harmonies. The discrete angular momentum quantum numbers
cation *
G
b
h the infiGhiaaaf ~ I i m
d gXw da. Thus
md one has anly to ohan@ the labis in. (2-2.13) ia obdsin tfie souroe mpm
wnbtion of tbe ntsw mdtipartielie &W. The p n e d i s a t i o ~fr, wmplex
is &km immdab. A m d n g to the
uthd and@depnde~liwsf
which is em(a"w),
the sowee robtiorr indie~Mby
a m y e&nnaturdly b
aXm. Dimt power wriw mpamion @ves
2-2
Bg
where
~em(sfi+(zi- ~2)=
n! perm.
A+(zl
* *
A+(%
--
)
.
:
g
(2-2.37)
MS
Sourccas
Chap. 2
which generalize the initial definitions (2-1.28) while retaining the weak ssuree
limitation. I n particular, the probability for the emission of another partide,
show8 the additional stimulated emission, that is characteristic of B,E. statisf ics,
As a preliminary to picking out the pneraX transition amplitudes, we
construct the probability amplitude ( { R ) +l {n)- ) K , which has the same initial
a d final configuration and, in. that sense, is a generdi~afiontof $he vacuum
amplitude. This object is extracted by retaining only equal powers of K:, and
K z , in the expansion
which applies to each momentum cell independently. For the process of inter@@%,
then,
(iK:,iKop iKgPiKs,) -+
[l iKgpn,iKop]
c
P
({n)+i{n)-)K= exp
where
The last term is rso bvritten in order to maintain the symmetry in x and 3'. The
fiT
Note that the probability amplitude ({R)+l {nJ-)K reduces to unity for K = 0,
which means that the initial and final multiparticle staks are r e k m d to the
same time or spaelike surface, as is appropriate to a reasonabIy loealiged probe,
To find the probability amplitudes in which initial and final ~ t a h are
s no8
the same, we do not return to the general eonstrmeLion given in (2-2.40), but
directly, in the manner of the vacuum amplitude. Thus,
use (in)+l In)
coxzsider the ectusal sou=@arrangement
K(z1 =
(4 4- K 2 ( ~ 1 ,
The explicit terms indicated for a given momentum describe the mverd processes
in which, respectively, no ehange in particle number oecurs, an additiond
particle is emitted, an incident particle is absorbed. Higher p w e m , containixlg
more complicated procemes involving several particles, are relatively negligible
for suficiently small do,; the probability for emitting two particles into the
momentum range do,, for example, is - ( d ~ , ) ~ But
.
we must not let this
apparently innoeent sirnplificaLion pms without comment;. The infinikgimd
chzbracter of h, will be vitiated if there is an inordinate sensikivity to p produced
by the appearance of very large coordinate intervals (e*'X). T o put it more
physically, we recall that we are dealing with a beam of parlieleg interwting
with a probe source. What we have done is eorreet if the probe is placed well in
the interior of the beam where there is no ~ignificantposition dependence. It
trill fail if the probe is outside or near the boundaries of the beam. This is a
68
Chap. 2
Sources
where the products refer to the various particles that are (e)mitted or (a)bsorbed.
The two statements are equivalent, if it is &mitt& that ( { n 11, + I {n f 1,) -)K
and ( { n )+l{n)J K differ negligibly, owing to the infinitesimal nature of do,.
Now we face the consistency test associated with the alternative uses of the
probabiiity amplitude. From the completeness of the fins1 or initial multiparticle sfates we deduce
~ ~ I ~ W + I {=
~ I - [ )l +~ (np
I ~
+ ~ ) I K+~ ,I I~K ~ I ~ I
I(In)+lh)JKI= exp [-
1(dx)( d d ) K ( z )Re ( l / i ) A ,
ln +(X
-z3~(2')]
(2-2.55)
and
Re ( 1 /i)A l,
+(X
- X')
(2-2.56)
[gGPQ%~K~P,](2-2.57)
[I
- &)K(x!)]
- 23 has the
(2-2.58)
following
2-2
fig
(dx>(~Z?K(Z>A-(Z
-xf)~(zP)
\VC?. define,
Tben we have
whieh is the same Euelidesn form that is obtained from (o+~o-)~,as the strong
wurce genera;liga$ion of (2-1.57). The Euclidean version of the veteuum amplitude is a red number lying in the i n t e n d htween 0 and 1. The vacuum.
amplitude (o-Io+)~ i s regained through the substitution
Pot
(I.
+ i15)pa
-4
-(f
- if.)^^,
snd, in respon~e,
A+(z
- 2') -+---A-@
--- z'),
(0+
2-2
61
(2-2.77)
while
(2-2.78)
which confirm in alternative ways the stated transformations.
We shall now use the causal structure of the theory to give a complete
derivation of the unitstrity praperty. This is done within a very limited physical
context, of course, but it is clearly a general procedure. For our present purposes
we replace the causal labling K l , K z by K[-,, K,+,. (While this may seem .to
be still another use of the overburdened signs, it will turn out to be consistent
notation.) Let the time T be located between the regions defined
with the
by the two component sources. Introduce a new time coordinate for z0 > T by
reflection at T,
sncl then transform this time interval in, the manner just diacussed;
where the appearance of the last term indicates the causal arrangement. When
the transformation is carried out, the quadratic K(+, term remains unawarc! of
what happens letttw, the quadratic K(-, term responds in the known manner
[h+ -+--A-] without reference to the other terms, and the last term changes
only by a minus sign arising from
the influence of the em"' factor, which is not compensated by the refleetion that
is al80 being used. f n the latter the inkpation limit8 are reversed to maintain
82
Sources
Chap. 2
[F
= e . ~ ( i ~ ( - ) p )(W+)$]
*
({n>Io-)~'
= (0-1
(2-2.85)
we get
(0-10+)'(-) exp
[/(dz)(dz')( - ~ ) K ( - ~ ( Z ) A ( + ) ( Z - z1)iK(+)(x')]( o + I o - ) ~
=
C+)
c (0-1 { n ) ) K ( - ) ( { nIOJK(+)
)
(nl
( O - I O - ) ~ ( - ) ~ ~ ( +(2-2.86)
).
As the notation indicates, the picture has become that of a system evolving in
time from the initial vacuum state under the influence of the source K(+)(%)
and then traced back to the initial state in the presence of the source K(-)(%).
It is not the physical system that goes back in time, of course. What is reversed
is the causal order of the states that are being compared. If the two sources are
identical, we must regain the initial state; that is,
K ( - ) ( z ) = K ( + ) ( z )= K ( x )
(2-2.87)
implies
(O-l{n))K({n)10JK = 1,
( O - I O - ) ~=
~~
(2-2.88)
inI
where the last terms appear in that form to produce the necessary symmetry in
and X'. We recognize the identity (2-2.67).
2-2
63.
( {n)+l
What we mu8t verify is that all refemnce ts the source K(%) disappeam from
(2-2.862, l e ~ ~ on1
n gy
But, under the given causal circumstances, L\- --+ --id(-) and A+ -+i~'+',
which completes the verification.
is a1w useful for the direct
The probability amplitude (o-Io-)~[-I*~(+~
eornputation of various expectation vaium. L&
This is the expectation value of eiPX for the states produced from the vaeuum
by the action af the sourn K, Sin= only the relative dhplawmenl of the tvvo
muram is sipif cant, we have
&,(eiPx
-- I ) ~ K ( ~ )=
/ ' exp
[T
(e'
-- 1)
N = E n,.
(2-2. la0)
Thus, the werage total number of particles creakd and the vasuum persisLence
prab~bilityare &implyreletted, according to (2-2.17) :
The discu~~ion
of Auetuations is fwilitated by writing (the vafious indiees arc;
omitted)
2-2
86
(2-2,107)
One digerentktion with re~pectto X, at X = 1, @ves
The causal relatioxls among thege functions are the same m in the v ~ e u u m
situation although b\$)(z - s'), for example, is no longer an exclusively positive
.frequency function, I n deriving expectation values we mu& note the$ the
amplitude ((B') +l {n)-)K responds to the translation K(%)-r K(z X) with
the factsr
@x~ti(P(%')
)XI,
(2-2- 1113)
since?both initial and final s l a k s are now mlevant, Some results are
(2-2.1 14)
and
(2-2.115)
IW3
Chap, 2
Souross
( o _ I o - ) ~ ~ - ' ~=
~ ~exp
+'
(~)a-(z- z ' ) ~ , - ) ( z ~ )
-- zf)K(+)(z'f
(dz')K?+,(z)d+
( d & ) ~ ; -(z)A'+'
)
(z- z f ) K t+)(a;')
( d z ' ) ~ ~ - , ( z ) ~ ~+ 'z')K;+,(z')
(z
(a;) =.
Kf-,(z) = K(z),
(2-2.117)
Accordingly,
P
~,+I"
,M-):
=a
IK@-I~,
(2-2.121)
(e-ip
- 1)(N-)j,
(2-2.123)
2-3
67
on. using a familiar Berne1 function. expansion. The introduction of the gropag*
tion function (2-2.59), with its atkndant stmcturers, generalbw (2-2, X 16) do
the probability amplitude ({B)-1 {n)- ) K ( - ~ ~ K ~ + ~ .
2-49 SPIN 1 PARTICLES. THE FNOTQN
(o+Io-)~
= exp
- z"rJ,(x")
fdz) (d~"J@(z)A+(z
now encounter
68
Chap. 2
Sources
f l rest frame:
= 0,
= m.
(2-3.6)
The components of the symmetrical tensor that appears in (2-3.5) are then
given by
p=v=O:O,
(2-3.7)
p
= k, V = 0: 0,
~ P D (1/m2)ppp~:
p = k, V = I : bat.
The result is simply 151*,which is positive, and which contains three independent
source components, transforming among themselves under spatial rotation, as
is appropriate to unit spin.
We note that (l/m)@ is a unit time-like vector, which can be supplemented
by three orthogonal space-like vectors, 4x, obeying
4?eppxr = 8xxn.
pp4x = 0,
They give a dyadic construction of the metric tensor,
(2-3.8)
The symmetry of g@' indicates that complex conjugation of the three e$, produces
a unitary transformation on the set. With the definition
the vacuum persistence probability appears as
+ JS (4,
exp [/doput (P) *(gpv+ m - 2 ~ p ~ v ) i J ; ( ~(o+Io-)
)]
"
Jp(z) = J';(2)
which implies
(o+~o-)' = (o+~o-)'l
= (o+~o-)'l exp
[Ci l : p ~ i r 2 p x ]
PA
(o+Io-)~~.
where n,x = 0, 1, 2, ... again indicates B.E. statistics. The consistency between the two uses of the vacuum amplitude is obvious.
2-3
69
One can choose the unit spacolike vectors 6~to be real. The orthogondity
requirement
0
(2-3.15)
P epk = P 0epx
displays the role of p in providing a reference direction. If epx is perpendicular
to p, the time component e$, vanishes. Let epl be such a real unit vector,
Then
is another one, and the set is completed by
= (p"/m)(p/lpl),
We note, incidentally, that
eps
0
epa
= Ipl/me
(2-3.18)
Alternative, complex, choices are suggested by angular momentum considerations. The response of the vector JP(z) to the homogeneous infinitesimal
Lorentz transformation
?EP = xp
6&*x,
(2-3.20)
-+-
and, equivalently,
60 = ~V(P//PI).
We realize a single-particle state of helicity X :
110
Chap. 2
Sources
(6o* S)&
lfe:kt
--e& X 6 0 = i
(2-3.30)
k1=-1
(2-3.31)
ilm
where
mid the reder is warn4 not 4x1 confum the letter m, u d in subseript~to dtcrtnoh
a rnaeetic quantum. numhr, with m, appearing e1cpXieit;lg in i f s mle ws particle
m m , The above ~tmetureis mch that
2-3
Tt
(2-3.36)
On combining the various contributions, we get the required form:
dadp(p)' (g,
+ m-'p,p.)JY
Jpjm hJp0jmh,
(p) =
(2-3.371
pDimx
where h = 1, 2,s di~tinpishesthe three excitation8 with total snlgular mamea%urnquantum xlumbrs j, m,
These sources esn. be exhibited expjicidly. The vector orthor~omality
praperty enables us to evaluate
where nolv
L = X X (lJi)V,
(It is unfortunate that the cornbination of two well-established no.Latioaal
conventions produces things like ji.) Incidentally this type of aourct: van&he?3
for ji = 0. Similarly, we find that
It is seen that the sources designated as X = 1 , 2 depend only upon J(z), and
that in the form V X J(z), while far the Lhird @pe of source we have, effectively,
"112
Sources
Chap, 2
(o+Io-)~= exp
(2-3.45)
W"(%)
=: 0.
(2-allats)
we a e v e a t
Jp3
0,
(2-3.47)
Now let W obtaixz this result directly, by trsiag the photon source desc&ption
@venin (2-3.45). The consideration of a causal arrangement,
J""(3;")
impfies
(O+IO-)
= (O+IO-)
Jg
(2-3.W)
(o+[o-)".
exp
(2-3.51)
The dysdie represenlalion for g,, given in Eq. (2-3.9) is not ap;propriak here
dnw p"" is now a null vector,
(2-3.52)
pg = 0.
Then pl" 'p and p" -- pp am,rmpectively, fa timelike and spmelike vector.
They are ~uppbmentedby two orthrtgonal uniL spae-like veebm @gx,
to give the d y d i ~
coxlatrwtion
I
-
(P'
gm
C @'PX
X
@V*
P&
which k the desired partide exchrzngs form. If is also impGcsd by (2-3.54) that
the two
have aero time component and, as spatial vecton, are perpendicular
to p. This k s self-contained desefiption of the k m tramvem8 witatima that
art3 pemithd to photons.
It bm been recomized earlier that a
s concept that is invarirtnt under proper orthoehronous Lorentr transformations.
One &odd be able fo make mare evident thL wpwt of tbe photon helioity
@ t a b .A d we should like to under~bndwhy the M e i t y slates fasve appeared
psired afthough no overt reference fa e?gatiai refieetion h= been made, b t us
begin with the remark that the conservation eonditiop imposed on JP($) is
sadisfied identiaally if
J@(x)= dPMp"z),
(%a.58)
where
MP"(x)
We
-MY"(a;).
The 1aGbr property indieabs that eaeh of them objects hm only three inde
pendent components, as i l l ~ s t r ~ by
%d
far as eontinuouta change8 of coardiniFtte aysbms are conmemiond nsbtion, we have t k 3 expucit 0~)mtm~tian8
2-3
76
Why can one not modify this phobn dwriptim by odtting J',x(z>, a y ,
and themby prduee a theory with only positive hdicity phsbna~tFor tbe same
reamn that one cannot have a theory in which only p~itivelycharged p~&icltls
occur; it would violak %hepfinciple of hipwe-time uniformity, To d k u m fhb
point in more detail, consider the cantribution to %hev ~ u u r nanaplitude
ated with the emi~sionand submquent abhiavtian of a p 4 t i v e helicity photon
and in
J;s(~=
) J;l(z)t
+ 5;1(2)~
(2-3.73)
(2-8.74)
J$~(z):t- J'+s(z):.
One might try Lo resist the iderenccs that there is another eaupling involving
(g')l and J'/ ,($)g by introducing the ~paoe-timeextrapalation of (2-3.72)
with an addil;ionsl factar,
1, z0 > z",
q(zO- zO')=
0, x0 < zO',
J'+312)'
and the particles dehieribd here mwt also be of n;wo m- if a unique splleetinre
extrapoltbtion is fo be acbievd. T h w antipa&iclr?s am t;he negative helicity
photons,
J'+s(~)'= Jtl(z),
(2-3.77)
and the additional krm ean be rewfi6bn as
(dz)(dar')JI", (z)T
(2-3,78)
$vl,
76
Sources
Chap. 2
(2-3.83)
*MC~(.
1/2(~/1~1)
(2-3.M)
JW,
which makw expli~itthaG Lhe pola~ratianv ~ t o r shave been r o b b d thmutgh
the angle ~ 1 about
2
the photon direction of mofion. If fhe mation angle k p,
*JP&
(d@p)
&@
:
the tr&~sformatiorr.
hcomes
(2-3.S)
JP(%) 4 Jg(x) cos (a *Jfl(z)sin 9.
The subsfieuLion of *P for JP &so has a Elimple eEect upon the an&ar momentum labled BWTCW, J p a i d . W@fimt remark that
J(z) = V X M(%)
*M(z), (2-3.86)
+
+
which uses the eflective value --vZ --t 'p = (P@)' and the pmperty L V = 0.
The ~ubatitutiianJfi 4 * P , which is equivalent to M -+ *M,*M -., -M,
interehangea these vectorial structures and tnrmforms the two souroes w o r d ing to
J p ~ j n r l -* JlOjma,
Jpojm%
--JpOjml*
(2-3,s)
2-3
7'7
This struetum indieaks that $he scale of significant T variation ia set by Ix - X'[.
If the sources vary little, in the time intervals that are msociakd with the
distances characte~sticof the instantaneous digt~bution,one can i p o r e the r
dependence in J@(x,z0 & 47) and evaluate
&P)+(X
- X',
7)""
d~
sin polx
- X'\
PO
This ~ v e the
s foXIowirrg form to the vacuum amplitude:
(0+j0-}~= eup
(2-3.93)
where
One reeognism here the aeeumulakd phmf?;change of s state that has a time
vafiable energy, E(xa). When a stesdy-state regime is established, we are led to
asgociste with it the energy value
whieh is a s&aternentof the Coulomb and Ampbrim laws of charge and current
interactions* This shows haw the principle of space-time uniformity provides
the logical connection befvveen the properLies of photons and the chsmekristics
of quasi-stationaq charge distributions.
There is one suLtLleLy hem we should not overllook. One cannot produce a
complettlly arbitrary statie charge distribution. The local conwrvatioa condition
78
Sour-
Chap. 2
pm~4d
the sourn is wnlind fo some finite ~patialre@on. Being wro in the
i&tbl muurn s h b , the bbl ch~rgemmains Piera, We m y picture initially
p ~ i t i v and
e negative charge distribution8being separaM, maved
mmpe
anb then m o m b h d . But them is another way of viming the
abut
in4duction of a charge &t~butionintto an emp%yredon, It requirw rmomil;ly than is wusl tha$ a physied dme~ptionrefern only to the
iaig m
fi~b
re@on which h under the exp~menbr'smntrol. The iniCial
and final vacuurn ~tatesp&ain to s boundd three-dimemionaf re@an, W e
gon, oubide the walls- We thus &pp~ci&%e
that an arbitrary charge di*
Critouticrn can b produced by the t r a a ~ p dof charge acrom the boundary, into
the mdon of inter&, and that thil charge distribution can be dismanlled
uldirxrtabIy by withdradng id across the boundaq.
2-4 SPIN 2 PARTICLES, THE ORAVITON
1;
gPyT@*(jc).
When thew are removed, the residual multiplicity af five is the anticip~tLedone
for spin 2 pl.ticIerr of noneem mass, m. T o c&rvout this p r o p m we exploit
our @ x p ~ e n e*th
@ unit spin p&&ielesand wride direcffy the physically am88hbq
~$mc%urrt
for the vscuum pc?rsi~bnwprobability, It is
do,F'v(p)*~r.(p)g.k(p) Pk(p)
where
pp.(p) =
and
+ (l/m?)p,p.t
T@'(p)
pYB&.(p)
0,
= TL*(p)
- 3yrpg@@(p)Twfp)
38
(2-4.4)
(2-4.5)
0by8
B ~ ~ ( P ) T ' ' ( P ) Qv
S
(2-4.Q
2-4
78
(2-4.7)
in, which
The projmlion
and are five in numbr. The sources for specifie slates are then identified as
When hdiciv states an? uwd in the veetor & d i e mnstrueLion
+ e(lph6e;h)- B(-
= f (GheLh6
1)'6-~h#
C (-
1) "e:kle;-hl
(2-4.161
X1
C
(--I)~G-~X
X
+ @So= 0.
-26+x',;l_l
(2-4.17)
I d 8
to
with
(2-4.27)
Thig eource restdction snd the dyadic form (S3.55) are combin& to abtain the
eflective rep1&cement
3($'gP"
gl@g'@- g"gP') --t
chh?e$$:,
(24.28)
X&'
The noerttion already indicates that them is only one conceivable identification
of this v e e t a ~ a lproperty-it is enerw-momentum. Unlike photon ~ources,
which hme a unique measure through the electric charge inhrpretation, grraGton
gources are eonfranted with an independent sale ori@naling in Lhe mechanical
significance of T,.. We provide an empirical conversion factor by writing
is intrinsically positive.
Again in contrast with eleetrie charge, energy or
The establishment of a g r t ~ t o nsource dk%fibutionin m iaitial vacuum sifurttion oa;n only be realized through Lhe t r a m p & of enerm and momentum infm
the mdon of i n b r e ~ t ,through the Esoundaries th& deEdt this domah. We
E ~ ~ ~ , (=
z '-) GM
where
2-4
83
2. Let the test body be a light beam for whieh t = opZ= O. The interaction
enerw with the 8un thus exceeds it%Newlonian value (replacing mass with total
enerw) by a faetor of tw, That is afso the increase of the deBwtion angle of
light over the Nefftonian wlue, whieh is Einstein's result For a direct cdcrtlation we compare the acquired transverse momentum Mrith fhc: longitudinal
momentum of the h a m , which pmses act a distanee p from the Sun. The deflee-
3, The same inkraction reduces the speed of light by the factor l - 2(GJ(r/R),
d n ~ the
e energy of a phobn is pi (l - 2GMIR) and differentiation with respect
fo p gives the velocity. This &et has b e n oherved by mmuring time delltys
in radar echoes from the inner pl%neb, We eonsider the superior eortjunetion of
a planet, with the line of aight from the earth passing at distanee p from the
Sun. Then the abntieipated additional time delay for the echo is
whem z, and z, are the distances, from the point of closest approach to the SW,
to the earth and the planet, re~pectively. The coefficient in the differential
relation
where
8rC
8ources
Chap. 2
And, finally, there is the contribution to the energy density to@that is assoeisted
with the gravitafisnal inLeration between the planet and the Sun. This is
not locslilred on eilher mass, but is distributed in space in a way that can be
esl~ufibtedwiLh suEaientP preeisisn from the Newtonian field stren@h:
2-6
85
is the angular momentum per unit planetaq mass (it is often defignated
and
by h). Here we have
and
I n discussing unit spin and spin 2 particles, it was natural to replam the s ~ a l a r
soltree of spinlms particles by vector and knsor sources, The response of rc veetor source, for example, to a homogeneous infinite~irnalLorentz transformation
is given by (2-3.221, which we write as
L(i)S(z),
(2-5.2)
where
2Z?C
= EPpxY -
rp,
IF,l;t,,E"x =
(2-5.3)
86
Chap. 2
Sources
name1y
2' =
+ 11Pve2"),
1 2 v A~ Ae2',
(11l2)"~= IlP.I2'~,
(2-5.4)
(2-5.5)
\S-ehave
3 )=L
and
, S(a) = L(I,)s(x)
(2-5.6)
(2-5.7)
It is in this sense that the finite matrices L(1) give a matrix representation of
the homogeneous Lorentz group.
For infinitesimal transformations
l', = 6':
wve wvrite
(2-5.8)
(2-5.9)
the matrices at1 are antisymmetrical and therefore Hermitian, while the see are
symmetrical and skewv-Hermitian. It was preordained that not all the S,,
matrices could be Hermitian, for the discussion of Section 1-1 shows that the
open structure of the Lorentz group precludes any finite-dimensional realization of the group. This injunction is not applicable to the attached Euclidean
2-6
87
= -isx4,
(25.14)
(l/i)(6~paYK
(2-5.15)
~ X Y ~ ~ ( X ) I
illtustmting how the momenlrxm dependence is removed, a t the expense of introducing ske\lrrHermitiars. operator^. The follol~ingis the sn,zlogous stn;temenl
for arbitrav momentum,
@xp[=tr@(p
* s/lpt)I
e x p [ &(p
~
(2-5.18)
\{?here
sinh B = lpllrn,
cash @ =
(2-5.19)
d@(lpl)ldIpl= l/pO,
88
Soureas
Chap, 2
which describes the behavior of s under the "rotaition" "specified, by 8, Incidentally, if a particle moves ~vithmomentum p, aXong the third axis, the t r a w
formation to its rest frame is
z3 = g3 GO&8 - z0sinh 8,
Z@ = --re3 sinh B -f- xo cosh 8,
The appearance of &isk in the role of
?tk
8%
is emily understood. The commutation relations for the sp, are simplified by
introducing the linear combinations indicated by
Stl)
&(l)
sC2)
S(21
= is(2)
(2-5.25)
m=i(s"'-s'2'),
(2-5'26)
*''
and
The theorem is
2-5
89
B(P)S(P)~
(2-5.38)
and derive
d8(p) = [da (p X
$ + is) + 6.
(p0
$ + in)] ~ ( p ) (2-5.37)
On writing
S(p) =
this becomes
&!$(X)
(dz)e-'~'~(z),
(2-5.38)
3 GC#'(X~~~
- X& + &,)S(X),
(2-5-39)
where the matrices S,, are those given in (2-5.32). The analogous construction
of (0+lp)' is [cf. Eq. (2-1.25)]
( P ~ ) " ~ ( ~ + I PS(P)*B(P),
)~
(2-5.40)
which involves the Hermitian character of B(p). The infinitesimal Lorentz
transformation behavior of S(z)* is like that of S@), but with the matrices
W
90
Sources
Chap, 2
--6:.
replacing the ..s,
The use of matrix representstions with imaginary .8,
is re-quired to h consistent with rest S(%),as we have mentioned before,
The compact notation used in \vriLing (2-5.36) obscures an essential point,
We are desc~binga particle of definite spixl, but embed it in a larger system
when we employ constructions like (2-5.26). tlceordingly there must h premnt
atl the left of B(p)S(p), say, an explicit election of the states of interc;st, We
shall illustrate this, and a t Ghe mme time @ve a simple example of the eonncct b n bt\vef?nfhe pre~entmatrix approach and the earlier pro~edures,by Gfioosing
age;
== Q.
(2-5.44)
2-5
91
notation gives
(dop)-1'2~p~
= 4 t r I@
e:b(p)#Jyl (P)~(P)]
= &:Jp:~,(p),
where
b(p) "- =p
l---P@
pi'lpll*
e',;
tr [c.e?b(p)crPb(p)]
(2-5.54)
0,
sinee
pPa, == m(cosh 6 3- sinh Ba * p/lpl)
= m[b(p)-1]2,
(2-5.55)
trak = 0,
(2-5.56)
and
Next, we consider
in which the form of the second factor depends upon the HermiCian naturt? of
b(p) and the a;, The following identity expresses the role of the four matrices
2-v2g, LLBan orthonormal basis for 2 X 2 matrices:
(trapX)(tra,%f)= Ctr @X) (tr o V ) --- (tr X)(tr )'l
= 2lt.r (X Y ) - (tr X)(tr Y ) ]
= det ( X - Y ) - ded ( X
Y).
+
(2-5.58)
tr (cpbZ)
4 tr (qpb2)$ +[tr (gp@') - (tr 8)(tr Q')].
(2-5.61)
92
Sources
Chap, 2
and
- (tr bJl)(trG')]
*[tr (rN@')
= gfiV,
(2-5.63)
tvhich gives
We have nmr reproduced all the covariant p r a ~ r t i e sof the three polarization
vectors for unit spin, When the third axis in (2-5.44) is ideatifid with fhe
direction of the momen't;urn vector p, the explicik expresions obtained from
(2-5.53) are just the heficity l a b 1 4 vectom (2-3.28,29). LneidentalXy, an
using the singlet rather than the tripjet functions, we get the farm
which is the anticipated scalar combination.
AS the basis for s corresponding treatment of arbitrary integer spin, we
consider the spin combinations
where the individual makriees set upon the appropriate index of %hesource
&l1).
...pQ',"
....p(p).
(2-5.67)
2-5
93
-+-
where zZ=
(gk)'
Xn view of the btal symmetry of the tensor, this property guarantees the
vanishing of the trace for any pair of indiees. The problem thus p a s d is a
familiar one. The polynomial of degrcse 1~ given iu (2-5'72) is a solution of
Laplace's equation according to (2-5.73). With z2 set equal to unity, it is a
spherical harmonic of degree n. T o identify the coefltieients Gnmt it S U % C ~ S to
consider the single nonvanishing component S33...3
= 1. With 'g = I,
= p,
we encounter the poIy nomiaI
which mu& be proportional to Legendre" ppolynomirtf, P,(p). Hence,
S*&...p l ( p ) ~ *@ ~* z L n = S
@
P n ( p ) ~ @* ~zP.
This generalizes the construction glven for n = 2, Eq. (2-4.5), and produces
symmtfic hnsom of rank n that obey
(2-5.77)
provided sueh tensors are used in the egeetively three-dimensional context of
Eq. (2-5.70).
~ ' ' ' ~ ~ ~ ) ~ f i ~ ~ ~ Q?
- ~ ~ f i , ( ~ )
and
r*
=X
Y = flPp,(p)zY
- $?l[(% ~
*
) 1 ~ ~ ~ *
96
and, for n = 1,
Accordingly, we constmct the polarization vectors for spin n, from those hlonging to unit spin, by
(2-5,921
It may be coneluded that the source effective for emission into the specific
particle state Isbeled pX is
ab~~,,
(o+jo_)'
exp [iW(S)).
(2-5.94)
88
Sources
Chep. 2
oA1' and ck2'labels, spa& from the option of an additional minus sign, whieh
(2-5.101)
The pt3rmutation of a single pair of spin indices afiets ogpasitely the singlet and
tP.iplet combinations, comesponding to the opposite behavior, under spafisl
rdection, of the time and space components of a four-vector, We have expressed it this way, sinee if Xeaves free the choice of overall sign in the refiee%ion response, which is the alternative between a vector, and a p~eudoor axial
vector. The behavior of the tensor S,,...,s is that implied by the several vector
indices, together with the overall i factor. The concept of parity refers to the
rest frame where the surviving source components are Sx,...rs, which aet as s
unit under spatial refi.~?etion.When standard =tor behavim is eansider~d,
the parity is ( - T ) n .
This gives a sequence of integer spin particlw with deernating parities, as symbolieed by of, I-, 2+, . . The other aequence is 0-,
l+,2-, . . .
Although the only known or conjectured massless particles of integer spin
have already been discussed, we shaH nev&heless pre%nt a unifirtd treatment
..
2-5
97
The produets formed from x b n d yp are four-dimensional. Any u s of the vector pp, as in (2-5.79), vcpould give no contribution in, view of the source restriction (2-5.102). We now exploit that fcact to replace the tensor TX with another
that is equivalent to it in the context of (2-5.103). This is accomplished by the
following substitution, applied to both z p and y6",
Y = s"Pp,(l-",P)xP
and
Considered in the rest frame of the time-like veetor pp $P", the orLhogoasl
veehr pp -- p@ has only spatial components, doubling the parlicle's momentum,
and we recognize that. the subspace dected by 8,, is the two-dimensional
Euelidean plane perpendicdar to the momentum of the particle.
If only hcslieitiw X = fn are to be represented in the source eoupliing
(2-5.103), the tensor n must be irreducible with respect to forming traces in
98
Chap. 2
Sources
X)(Y
[(X
where
(2-5.111)
. y/[(x.X ) ( y
y)]'I2 = COS 4
(2-5.11 2)
(2-5.113)
P =X
m54n:
( - l ) m n (n - m - l ) !
J
m! 4m ( n - 2 m ) !
(2-5.114)
and, in particular,
dnl = -4n.
2,
(2-5.116)
A-f n
where
(+X
n f inp
.X)(112)n.nf
z
e
.
(2-5.1 17)
(2-5.1 18)
e'pilxfi= ((ti.2)
*<P
pin
- eyhl -..e2il,
(2-5.119)
2-6
Spin
$9
where
and
Ordinary matter possesses no conserved physical properties that could be
identified with the ones described by the laeal conservation law (2-5.1231, or
indeed for any n 2 3. The inability to construct their saurees strongly sfirms
the empirical absence of the particles. Bud perhaps one should not rejeet totally
the possibility of eventually encountering such properties, and the associated
padicles, under cireumstanees that are presently unattainable.
2-43 SPIM
PARTICLES.
FERMI-DIRAC STATISTICS
4 particle description,
4(qk,cl] = Eikll
i g 1 2 ' ~ 2=
i~3
l,
(2-6.3)
and we identify
sk =
$C&.
(2-6.4)
1W
Chap. 2
OOsour~ftil
T E -+ Q& is
number :
i
~
= i h~ $ipg~
~ T ~ T Q ,
(2-6.5)
They are the analogues of igk with the T and 7' matrices inbrchanged.
The two sc?h of three anticommuting mnatfices arc? mutually commutative.
These six antisymmetrieal mtri.ces, Q,p&, and the ten symmetrical matrices 1,
orpg, provide s basis for a11 4 X 4 matrices. Since the three px, are on the same
footing, we arbitrarily identify h with p2 and writet
where the
&rereal, symmetrical matrices. We note their algc?br&e pmpedies:
the last statement b i n g the ma-lization of Eq, (2-5.33). Since space reflecLion
induces n -, --n without ehanghg s, it is r e p r e m u by a matGx that eommutes with ar and anticommutes with pz, The only matricea with those eharacteristics arc: pl and ps. We choose the lathr arbibrarily and multiply this antisymmetrical matrix by i to get the real space reflection m a t h
which obeys
The spme refleetion matrix appears in another role on considering the real
matrix ms~ciated'VVith an infinitesimal Lorexltz; tr&nsfamatianfef. Eq. (2-5.9)f:
The validity of this statement for the finite tran6farmations of the groper
orthoclrronous group is assured by $he composition prverty of succeiclsive
2-6
101
transformations,
( L ~ L ~ ) ~ T .=
L L~TLTT.LIL~
~L~
= L;~.L, = 7..
(2-6.16)
The relation (2-6.15) also holds for the space-reflection transformation, since
r.Tr. = 1
(2-6.17)
a0 = 1,
is given by
LTaL = a! - 8w X
a!
- 6va0,
LTaOL= a0 - 6v a, (2-6.21)
+ 8d',)av.
This is the response of a vector to homogeneous infinitesimal Lorentz transformations. The repetition of such transformations yields the finite transformation law
L * ~ L
= rVay,
which is also valid for the improper space-reflection transformation generated
by L = 7.. Note that the symmetry of the a' and the antisymmetry of r,, as
well as their reality, is maintained by the Lorentz transformations.
We now consider the coupling between sources associated with singleparticle exchange, where the individual emission and absorption acts are represented by (2-5.36) and (2-5.42), with
The spin 3 particle has been placed in a larger framework, as evidenced by the
existence of the three matrices pk that commute with a. Two of the four components must be rejected by interposing a spin-independent projection matrix
between the two B(p) factors that are associated with the individual acts. The
possibilities afforded by the three pr are really only two in number, depending
upon whether the p matrix used commutes or anticommutes with a. In the
Chap. 2
c',
2-6
163
The conclusion is unavoidable that spin -& presents ta totally new situation.
Only one coume is open. Insbad of trying to modify the symmetfy cfismckristies of the kernel to suit the algebraic properLies of the source, we must adapt
the algebraic prope&ies of the source to the antisymmetry of the kernel. The
comparison of the two equivalent versiom af (2-6.29) with the andisymmetq
property
(24.32)
&ts(ztIX ) = -Ktlt(xp X')
will cease to be a paradox and become sn identity if
that; obey
L-Xr"L= P,?",
together with.
L-l?@rgL= l @ g i p h ? E ~ h ,
and so forth. The algebraic property r: = -1, along with 'a = 1, shows that
and
which afso gives the identification
The r matrices do not have a common symmetry, The: definition (2-6.35)
hplies that
fpT
--i@~r;I$
104
Saurcs
Chap. 2
(yk, 7 1 ) =
-4
(at, W)=
-&I.
(2-6.44)
[at,ad =
&PI, 711,
(2-6 -48)
~YOY~.
(2-6.49)
= ~+EY,'IYVI,
fZ-@*F;a)
L-'v,L
1,1'.12,13k~"gkrS
= (det orS,
(2-8.61 )
which c h a r w h ~ z e rS
s m a pseudoscalar. It is invariant for proper transform
tionss, det I = +I, and mvems 8ign for improper transformra;tions, or reBec~
from the anticannnnute~.
tions, det I = --I. The latter property f ~ l l o wdirectly.
tivity of 7 and Y &S ~peeifi~slly
noted in (2-6.55). Let us also obsenre the
pseudo or axial vector efiarz9cter of ~ T P Y ~ ,
Q
+ + -+
for which the count is l -f- 4 6 4 1 = 16.. Clsmly reXat& but &mtinccf
is the organization by symmetq properties. As suwwted by the emstruction
we consider ror, where r refers to any of the sets exhibited in (2-6.63). Then,
=
---rT~O
= -ya~;lrT~I,
(2-6.6s)
106
sources
Chap. 2
and therefore
nebf- notation,
sinh B = Ipl/m.
(2-6.75)
The projection matrix +(l TO) is constructed from any ttvo orthonormal
eigeaveedars vh,
f 'vr =
V ~ U =
~ ' 6hy,
(2-43.76)
2-6
in the dyadic f o m
+(l
107
v*u!.
where the relevant null trace of ra expresses its antisymmetry. A more general
remark follows on nsting that
t;ts
eigenvectors of a cornponenf of a,
(2-6.80)
& 1,
Cr
--v:?
@,v,* = -av:,
are satisfied by
-*(ex f
wz)~$= v&,*
* -- Z"QY&Y@*
(2-6.82)
(24.83)
v-,
Since v: is an eigenvector of Y @belonging to the eigenvslue -- 1, there are corresponding orthogonaliky properties,
= Ot
**
v62;lUf
Q,
(2-6.M)
+ Y')
in (2-6.74), we get
where
which involves the anticommutativity of Y'Y with Y' and the eigenvector
significance of v, relative to 7
'
. The same properties are used in verifying the
artlraonormzzlity of these vectars in the form
* 0Up.'
ZLpuY
No%?,according to Eg. (2-6,59),
= V:@,? =
&@.P.
IOS
Gouross
Chap, 2
This form shows the utility of defining the v. with respect to p ss s spin reference direction. Then @ = p/lpl can be replseed by the eigenvalue a, which i~
now a helicity 1Plbel. On employing the relation (s6.83) these veebrs borne,
simply,
= ior5up.
- m - ?P.
2m
This is equivalent to
(m
- Yp)(m4- ?p) == 0,
the consistency of the two definitions conveys the Hermitisn nature of 7'.
These are the precise definitions of single particle emission and absorption
sourem, nrhich have k e n built up from vari~usfachrs, Thus B(p) i s contain4
2-63
109
in uOzr,,. In the rest frame of the parkicle, U,, reduces to v,, which is an eigenThus the
veetor of TO and therefore of the space-~flectionmatrix F, = g?',
single-particle states have a definite, i f imaginary, parity. Incidentally we did
not prejudge this question by using the same matrix,
in. defining r, and the
projection factar +(l ps). I t is now clear that the latter also performs a
parity selection, and that the refieetion matri.x must be defined accordingly.
The particle sources v,, and v;fl, as linear functions of the qr(z),are also
totally anticommutative,
O;rpa, qpfua'
{%t
~ p ~ c =
r ' {(7)1~#, ~ p ' r r r )
Q.
(2-6. X W )
In particular,
(VP.)~
= 0,
(v;@)%
= 0-
(2-C5.101)
All this is quite the same as with, B. E. statistics. But; now the power series
contains just two terms: a,, = 0, 1, for, on reversing the multiplication order
of two elements, we see that
It is indeed possible to factor the coupling terms in the desired way, but strict
account must be kepG of the minus signs that are involved, This is facilitated
by the falEo6ng procedure, which we illustrate with two pa&icle sla$es, fabeled
a and b,
By always displacing sourees Lhrough an even number of factors, one avoids the
explicit appearance of minus signs, In this way we a r ~ v eaL a facbri~atiaa
where the emission sources are multiplied in some order, r e d from left to fight,
110
Sources
Chap, 2
while the absorption sources appear in the same order, but read from right t o
left, It is given general expression by the following identification. of multiparticle states :
(in)IO-Y
= (O+IO-)~
n (i,.)%g
nT
in which
symbolizes the opposite multiplication sense from
and any
standard sequence can be used for the denurnerably infinite number of particle
states. As in the B. E. discussion, the pa&icle occupation number interpretation of n,, is supported by the response to source translation, 4%)
-+ rt(z
X),
which gives
where
shows the additive contributions of the various particles that are present.
The completeness requirement on the muttiparticfe states is stated alternatively as
(2-6.109)
where
(0-1
s a d by
(72.1)"
((C4IO->'?
(2-6.110)
with
We have been at pains to bvrite these more carefully than in the B. E. situation,
since we are now dealing with functions of antieornmuting numbers, No pre~ a u t i a n sare needed for the vscuum amplitude, which is an even function, and
we present the two completeness statements as
where we have omitted the compensating factors of z" and -i. The comparison
of the two forms suggests a rule of complex conjugation for F, D. Boureeg thaG
we shafl find is a, consistent one: complex conjugation also inverts the sense of
multiplication, as illustrated by
z 'real
)
sinee YO is imaginary. This is another aspect
Consequently, q ( z ) ~ ' ~ (is
of the matching of the statistics to the spin. Sirlee the matrices (I/i)rF are real,
the only eomplex quantity in W is &.+(X - 29,and
(2-6.118)
The relation
(2-6.1 X 9)
then gives
q(-p)ro(m -- r p ) ? ( p )
Up
C )l(p)* y~ U ~ @ ~ ; U ~ O V ( ~ )
Q
Re C s;,np..
P@
(2-6.120)
which makes essential use of the complex conjugation rule. This resull,
112
Sources
Chap, 2
e(ri/4)g~'
E
@&v@
and indeed
(2-6. f 25)
are given by
(r4= iro)
at = r0vX,
ad = pyoe
2-43
113
vacuum amplifrxde is
(2mdo,) ~ , . , ~ O v ( p ) ,
= (2mdw,) "'s(p) *you,,,.
(24.138)
This dbcus&on. of the Euclidean postulate brinp %heTCP operation t;o
mind, Thmu& the attached Euclidean p u p wa produce the tramformtion
zi" = ---g^
(2-6. 137')
qpgg =
&S
~ ( 3=
) rstq(~)p
(2-6. t 38)
where
=
e(ri/2)elle(rii2)c.4
e(w~/Zlszge(ri/l)s~, e ( r i f 2 1 @ i i e ( ~ i / ~ ) *
=
4 --iT5
( 2 4 .f 39)
de-l&ifs the ewivslent raLations through the angle r in turo peqrtndieular
planes. This matrix is antisymmetricd, imaenary, and obeys
The invariance of Che vacuum amplitude is vtsrified direekiy on uging Lhe mlstion
W turns into --W, But this sign change can be compensated by reversing the
multiplication order of all sources, which is in sceord with the representation of
causal sequence by mu1tipEieativc?-position.
The eflect, on the individual emission and absorption, sources, of the substitution
?(P)
Y~v(-P)
(2-6.143)
+
(2-6.144)
or
?pug
and
-i@vF
v;uq
-@
iflvP -@
(2-6.145)
-g,
(2-6. f 46)
-g*
where
= E,
-@
-qt
and the source transformation that constitutes part of the FCP operation produces the required reversal of multipliestion order,
2-7
PARTICLES.
NEUTRIMQS
They obey
f
CMrj=I
i
2-7
115
C
( d ~ ) Yim(~)rlp@tm,
1m
(2-7.8)
116
Sources
Chap. 2
C (dQ)'l2Ytrn(~)~*p0~m
= C (d~)~"~ljrntl*potjm?
(2-7.10)
ljm
Im
++.
COS
36 C ( d Q ) " 2 ~ l j m ~ + p ~ljm
sin 3 0 ~(p/lpl) C ( d ~ ) " ~ z ~ j m q - ~ o t ~ m
I
= C ( d ~ ) ~ ' COB
~ Z t 3 & + P ~ l jm
E
- 1,
where
wherein the Zljmqare constructed as in (2-7.5) from the eigenvectors v,,, and
the spherical harmonics refer to the angles of the unit coordinate vector. The
comparison of (2-7.13) and (2-7.15) with the left-hand member of (2-6.73)
supplies the identification
2-7
117
Unlike linear momentum states in general, angular momentum states permit a specification. of spaee-refiectim parity. The response of the parficfe
sources to
X) -+~ Y ~ ? ( Z-I)
~,
>
Z$ma =
(-l)'+"miy~l
5 -m
-pp
(2-7.23)
(&)h"
jmP(-~)*To~
5qf~=
) i(- 1)
~p*a E J-m
-Q,
(2-7,28)
and then
jmg
--B
-i( -l ) E + j + m rlpar j -m
(2-7.29)
--g-
e jl . . . j,~'~+(zl - zj,)
r"@+(z, - 25,)
(2-7.31)
expresBeS the anticommuta;tivity of F. D. sources, We aee hercl: the simlple and
necessary connection between the symmetry propedies that clnaraebrige the
two statisfcics and the elemenlary algebraic propedie~that; distinguish fhe two
kinds of soureek Appropriahty symmetfized prorfuels of individual praprtgation functions give the space-time representation of the noninteraeting multiparLi~Iesituation.
Let us diaeuss now those generalizations in which the t e r m i d vacuum
states sre replaced by multiparticle ~t%tf?s.A causal situation is considered,
containing emiwion gaurce gz, probe source q0, detection source q1 :
2-7
l 19
[/
+ i / ( d z )( d x 3 ~ o ( z ) 7 ~ @-+ bx')~a(xO](o+[O-)'O
where the index r represents any set of single-particle labels, say pcrq. The
causal analysis of this vacuum amplitude is
from which the detailed effect of the probe source can be inferred. To describe
a weak probe one must interpret the product isr({n)10-)?. If the single-pebrticle
state or mode r is initially occupied, n, = 1, the result is zero, ( v , ) 2 = 0. This
is the Exclusion Principle, forbidding the introduction of an additional particle
into an already occupied node. Otherwise,
where n,, counts the number of occupied modes that precede r in the standard
sequence, which is the number of source factors in ((n) 10-)"hat
iqt must be
moved through in order to place it in proper position. Similarly,
and we get the weak source results
( (n
+ I t ) +l(n)-)Q(-
X)n<ri~rt
{ { n - lp)+l(n}-)q Si (-I)n<riqF.
To construct the probability amplitude ((n)+[{n) -)V, one must retain only
equal powers of v f , and q 2 , in the expansion of (2-7-33),
exp
[F(isrdrlo, +
iv;,inzr)] +
In contrast with the B. E. situation, the series terminates with the indicated
product. On referring to (2-7.35, 36) we see that
where the factor n, indicates the absence of the term n, = 0. The effective substitution is, then,
= exp
[F
iYwn.iq~,]
(2-7.40)
920
8aure8s
Chap. 2
The linear relation between ? ( X ) and emission and absorption sources for any
type of mode specification can be w i t t e n 88
(dz)~ ( z ) r ~ ~ . , ( z ) .(2-7.41
Thus,
(2-7.42)
and Eq. (2-7.16) supplies another example. The related canslmction of the
propagation function, is that illustrated in (2-X17, 18) :
We now ge.t;
((n)+1 in)->L
exp
with
The form of the second term assures the antisymmetry of TOG~,,+(~- S').
Explicit causal stmctures are
[l
i q Z i q o r i s ~ l i s ~ r l(2-7.47)
,
n, in (2-7.48)
(2-7.48)
appears as
where the two sets of modes labled a and e are disjunct, sinee the individual
mode facton are linear in ql1 and q f . If B nonvanishing t e r n is to result in
2-7
l21
th& sre not oecupied finally, while the e(mitted) modes are those oeeupietd
find1y,
(in
+ l.)+l{n + L)-)"
({%l+{a)-)"
[(-l)n<'i~rl
G
nT
[(-~)'<~iq:l,
(2-7.52)
whieh i~ the generdizlation of (2-7.37).
In order to test $ha?eompledenctss of the multiparticEr? state8 in this general
context, we multiply (2-7.52) on the left by its complex conjugate and prrtst;nl
the ~ m l int a farm that reinstates (In) m an arbitrary initial 8taLe:
But, according to
(2-7.55)
is valid for df
;e
(0+/0-)7' = exp
(dz)(dz')q(z)~'$,(z)$,(z') *r"v(z')
as antieipatd.
The reduetion of unitarity to causality for spin i$ particles imitates the pattern alresdy established with spinless particles. We follow the development of
the syskm. from the initial V ~ ~ U Ustate,
H ~ under the influence of the source
qls)(~),
and then trace it back to tlne initial strtte, using the source g(,,(z),
The
stru~tureof the propagation function 6+(z - z') governs %hisevolution and
we get, as the analogue of (2-2.83),
(2-7.6 l)
and
G'+'(%
G[-)(X- X')
- zp)= C
+,(z)y~.~(~')*r~,
r
--C$r(~)*h(z')~o.
T
(2-7 -64)
The various functions are dso eanneeted by the identity
As in the spin O discussion, tfle general unitsrirty proof uses the sources q f f ) ta
generate arbitrary terminal st&s. The complotc3 removd of reference to a
subsequently ~letingG O U T C ~~ ( 2requires
)
the additional relations
which lacks only the matrices g,, = --CF.,. Since TO and ror, are antisymmetrical and r0up symmetricd, the antisymmetrical matrix q eennot multiply
them. But p must be used to reverse the antisymmetry of Y @ Y ' ~ Y ~ . This structure shauld fit into the mode function pattern detailed in (2-7.633, M) since the
latter refern only to the combination of individual emission and absorption wcta
An essential aspect ia the positiveness property
124
Chap. 2
Sources
for (2-7.69) is the following Hermitian matrix positiveness requirement associated with any particle momentum F,
which incidentally asserts the reality of ml,mz, a, and X, When a particte of
mass m > 0 is viewed in its rest frame, this condition reads
The three matrices ipT5, Y', ror5anticommute and are of unit square, from
which we infer the numerical requirements
+ + m:]'"
ma i [ ( m a ~ ) m
~:
2 0.
(2-7.73)
In addition to the conclusion that a is positive, we note that the zero value must
be attained if a projection matrix is to be produced, and accordingly
m2a2(1 - h2) = m;
+ m:.
(2-7.74)
This is represented as
a = cosh 8,
QX = sinh 8;
= 1,
+ m: = m2.
ml = m cos ( p ,
m2
= m sin cp.
(2-7.77)
,t)
= ,-c
l/ 2 ) ~ ? 5
1/~
~ )( @ ~ Q T s
X [r"(m
- r*(i/i)a,)a+(x - Z t ) j e ( 1 1 ~ ) e 4 ~ 6l el 2( 1 ~ 1 7 (2-7.78)
~.
1,
r n ~= m2 = 0.
If we now chwse
a=
there emerges
4,
2-7
+ particles,
Msiutrinos
725
The first version ixrdicrttes that we have regained (2-6.2q, with the objection
to the antisymmetry of p2 =. iyti removed by the pmsence of the additional
antisymmetrical matrix Xq. The second version is related ta the standard form
of raG+ by symmetrical matrix factors, which could be transferred to the
sources. These are sinwlar projection matrices, however, and the new sources
will be subject to the restrictive condition
128
Chap. 2
Sources
neutrino :
l
+ b.
lept.
= 0.
We now put forward the natural hypothesis that one role of Ieptonic charge is
to distinguish, and label the two leptons with a common electric charge q:
Ech. tept. =
(2-7.87)
TB*
Its consequence is the empirical equivalence between neutrino helicity and the
accompanying electric charge,
In the interest of compIeteness we shall exhibit the sources for specific
neutrino states, under the simplifying assumption of zero neutrino mass. One
uses the dyadic construction
where
~ =
and
Up1
6 r~ ~ 1 ~ ~ ~
= Up-l.
Particles of integer 4-
4 spin
127
PARTICLES OF fNTEGER 4-
SPIN
and its specifie identification is confirmed by evaluating the trace over the sixdimensional space. The resulting form of the source eouplimg in the rest frame is
rl:*(1
where
r0)(6k&
"ii rc
*@k@1)'11
~rlk
= q:B(1
7')~kt
&Q~@ZVZ
(2-8.3)
(2-8.4)
obeys
Qk-?k
= 0,
(2-8.5)
128
Sour~et
Chap. 2
+-
in which we bsve returned to the vemion given in (2-g.?), and u& $he dya&c
construction for gtr,(p). The introduction of the dyadic spinor realization for
(m - r p ) / 2 m gives the form
anwiar momentum,
(2-8. l S)
therefore: inducm
VPX
i(--f )
(812)-X
,*
~JV~(-P)
?.ph
-+
-i(- l ) t 3 1 f ) - X q p - ~ ,
(2-8.29)
(2-8.20)
where
?'(z)r0[scu
(-7 X(l/i)ax)D+(%- z')
- *% ( - ~ ~ ( 1 / i l a h ) ~ + (-z Z~)?~]~@(Z'),
a&q"(z) = 0.
1343
Sources
Chap 2
SlnGe
PP.Y(P)= a,
We also use
0.
(2-8.26)
= 2~"
(2-8.27)
@~,~eup,t,
U'=&
where
--. @ " ) . U ~ ~ P =
(a6-E- Q
' ) u ~ ~ ~
(2-8.28)
in which +(l.
tionis are
-+ XB)
= ei+l@~k,
As in the spin fj- n e u t ~ n odisewsion, one can introduce an. rtdditioml decornposition in which helieity is coupled to charge in a unique way.
Preparatory to generafi~ingthis approach to all parliclw of spin s n -f- 9,
n = f , 2, . . . , we return to the r e ~frame
t
spin projection
=I..
Vk
== Ptk
(2-8.33)
$@k@l171
S @ p n t p . ~ ~ @ ~ ~ t ,
(2-8.34)
where
a k p , t l l -=
*(8kt8,,
-4-
akrl8ta)
"=.kp6tQ
(2-8.35)
is the rest frame vemion of the n = 2 projmtion tensor that is defined generally
Pattieies of intsgsr
2-8
+ 5 spin
'l35
i j k , ...kR
@klBkl.-.k,
=z
(2-8.38)
0,
according to
0
2=
@kflk,qk,kzkg*.k,
(2-8.39)
== qkkks..sk,.
apnkl,..knp,~I...2,ct@q~ll...l,
as stated in (2-8.37). Alternativeiy, one ean verify that the trace of the projection matrix that is defined on the spaee of n three-eamponent vector8 and
two-component spinors h a the required value of 2(n +) 1 :
-+ +
+ +
+ I thme-vector indices
132
Sour~aa
Chap. 2
A particle of spin s = n $- can be describd by the symmtricd tensorspinor source $ 1 . " ' e ( 2 ) . The four-dimensional momentum space version of W i.s
-+
(2-8.45)
where
The other heliaity functions are produced nnos"cimp1y fmm this one, by rotation, ss effectively rewliad in. the algebraic construction
where $,&(E) is defined a8 in (2-5.87) but with n replaced by s, The rwults for
= 8 that are &vent irr (2-8.15) are immdiakly repmduced in this w8y. The
i;:
2-43
Particles of integer
+ f spin
133
sourees for the helicity labeled stzt;hs of thew F. I). psdieres arts identified as
n = X, 2 .
must be accompanied by the corresponding projection tensor, which is described by (2-5.104), and inded the general form is
one verifiw Ghat (2-8.22) is reproduced, while becoming aware of the egtrivalmt
form
The coupling hrnn in 2"W for a causal arrangment is obtained from (2-8.52) as
The dyadie eonstruetion (2-5. f f 7), combined with (2-5.1.21), conveds the
tensor-matrix of (2-8.55) into (the tensor indices are raised, for clarity)
134
Saurcm
Chap, 2
The f m b r 4(1 f a') locks the spin 4 hdicity to the olfiw, and we re-co@;nige
the genemEsaLion oE (s8.31) for s = n 4 :
vy.-.v,
= epkn
with the w m c b t d wmce definitions
r/z ; P ~ * . . P * ,
$p&
( 2 ~ 'h p ) %p& ~rl...r,(~)$
(2-8.59)
h = As:
=-b
a,
2-8
f 35
8=
The h m e l of this quadratic lorm ha9 a definite wsynrrrnetq under matrix tr%nsposition combined with the substitution p"" + --fl,
Aecorifingly, if the dgebraic propertim of the source are to mahh the symmetv
properlim of the kernel, we must have
n even, s == integer: [S(x), S(&))]= 0, B. E. stsdistics,
(2-9.10)
n odd, s = integer 3- 9: (S(%),&(X')) == 0, F. S). &&%&ties,
which is the general stakemen%of Lhe connection betwrmn spin and ~tstistics.
This proof will be eompjete, however, only when we h w e shmn that any a t
tempL to revem thme natural connections does violence to Lhe completeness of
the multiparticle stabs,
Let a8 eon~ider-the causal srrangement
B(x) = Sl(2) 4- 82(~),
which implim
(o+~o-)~= (O+IO-)'Z
exp
[?'(m - 7p)IaiS2(p)
d w , i ~(p~)'
a
(2-9.12)
136
Chap. 2
Sources
%P.
n (up+).,
(2-9.14)
a=l
Eq.(2-9.12) becomes
which uses the fact that even functions of the sources are commutative for
either statistics. The causal analysis
(~+lo--)~
= C @+l
l* 1
ns'(
n)l~-)~'
(2-9.21)
2-9
137
B.E.:
F* D.:
fSPx,SP3~f=0,
(&x,
&lx?>
(2-9.23)
o,
that the two statistics are distinguished. In particular, the algebraic property
F. D.:
(2-9.24)
=O
Become, respectively,
1,
(2-9.26)
and
The clem contradiction .cvi%h(2-9.36) completes the unified proof of the cannection between spin. and statistics.
2-9
*r39
combined with reversing the multiplication order of all sources, The egect of
the substitution on W comes down to the minus sign induced on eseh r@
by
the r, transformation, and thus W is multiplied by (- 1)". The reversd in the
sense of multiplication introduces a plus or minus sim, in accordance with the
st~tistics. Through the connection between spin and statistics, W, (and the
vBeuurn amplibde), is left invariant; under the compie%eTCP operation.
To study the eEect of TCP on individual emis~ion~ n absorption
d
sources)
we first notice the generalization of the spin complex conjugatioa property
(2-6.92), which depends ugan the multiplicative eornpasifion of the u,~,
Then we find that
to which a charge index can h tzdded in the h o w n way. The eorrwponding
multipsrticle transformation is
where
= n, -x.
The r@m&trices, which mirror the indefinite Minkowski metric, must be removed in the tran~form&ionto the Euelidean description. This is ~.ecomptished, for n even, by the symmetrical matrh
IrK)
Souram
Chap, 2
snd now
2-9
141
2O=z0,
%=-xr,
(2-9.56)
with
n
r8 = ( z t )
(ir3.
(2-9.57)
a==l
The uniform selection of p' = f l in the rest frame gives the definite parity
(f)P, which is real for integer spin. With n = 2, the alternatives of antisymmetrical and symmetrical spinors give the spin-parity properties 0-, I" and
0+, l+,corresponding to the sign option in (2-9.57). Otherwise, with the general use of symmetrical spinors, integer spin particles fall into the two sequences
of parity (&)(-l)'.
No rest frame is available for massless particles. In this circumstance, the
kernel of (2-9.12), referring to causal conditions, becomes
Now it is the values of the individual helieity matrices U p/lpf and the asssociated Yg matrices that specify a particle state. For a systematic classification of
almost all helicities, using symmetrical spinor sources, it suffices to identify the
value of every ir6 matrix and thereby of the individual helieity matrices. This
is sccomplished by inserting the following symmetrical real projection matrix:
Then we have
142
Sources
Chap, 2
Although this discussion applies to n both even and odd, the necessary
existence of a charge propedy in the latter situation, of h an integer 9,invites ra further classification in which the helieity is tied to the charge value,
This is produced by replacing (2-9.61) with the symmetrical real projection
matrix (the common =t sign gives two alfernatives)
For a given value of q the trace of the complete projection matrix now equals
4"(X f21Z)(X/2") = 1. Thew tare only two states, labeXed by q = & l , and the
helieity is
== ( ~ ) 4 $ @ ,
(2-9, M)
where the sign option refers to the alternatives of (2-9.65). Pn each situaion
This treatment is k3ss general than the earlier neutrino discussion for R = 1,
sine@fph%t did not require the msumption. of zero masg.
We &all close this section by examining the connection h t w w n the muftispinor description and the tensor treatment of integer spin particles, in the simplest situation of a second rank spinor Silt. It is convenient to regard the
l a t k r as matrix, and to correspondingly rewrite the structure of W as
2-9
- ~ ~ ) S ( - ~ ) ~u rp)S(p)yO
~ ( r n ] = -(pZ + m2)tr S(--p)TraS(p)70
+ m tr ( S ( - P ) ~ ~ @S(p)yOl)
[~P,
+ + tr ( [ r ~s ,( - P ) ~ ~ ~ I [ Y P~ ,( ~ 1 7 ~ 1 )
(2-9.71)
orthogonsl in the senw of the produet defined by the trace. Their normafizaGiaxrs v a v in sign \r.ith the Xfermitian or ske~v-Herrnitiannature of %liematrix,
as dictated by the space-time metric, Thus, the algebr&ieproperties of the Y,
imply that
(2-9.73)
4 tr r,r, = 4 tr YrYt;YuYS = --(firu,
~vhile
PEA g,rt@rh S r x Q v a *
(2-9.74)
The results are
E
with
K($) = 2"'(mS2(z)
+ a,P(x)),
(2-9.78)
Chap. 2
and
The K and J stmetures are the anticipated onm for spin O and spin X.
There sre additional terms, however, ~vhiehmodify the vacuum smpfitude by
the typicat. factor (S stands for SE,8 2 , S,, S,,)
This is an equivalent description. The additional phase fetctor daes not change
the vtacwm persisknee probability nor does it contribute to the coupling between sources in a causal arrangement. And it has no implication for the obsem&blewpeets of the energy asociakd wikh a quwi-static source distribution,
for they refer to the effect of relative displacement of two disjoint parts.
Physical eonsideratiom that arc? sensitive to such souree overlap terms can appear only in the fudl-ter development and specialisation, of the general souree
f ormlzlism,
Far m == 0,unit helicity parlieles should be selmted by inserting the projection matrix
Its aetion upon the second-rank spinor is given by the
matrix trans~riplion
The two terms in the symmetrical spinor of (2-9.70) commute and anticornmub, respeetivefy, with Y ~ .Only the XatLer is retained by the projection nxat~x,
which eEecti-vely sets &(p) egual to zero. As we recognize from (2-9-78), Lhe
divmgence of the veebr source JP(%)then vanishes identically and the photon
O in
deseription is regained, It would not have suffi~edto merely let m
(2-9.78), since it is dsa necessary that (I/m)dJF+ O, We have remarked that
$he antisymmetricat spinor should be supplid andogously with a rSprojection
faetor that digem from (2-9.80) in the relative sign of the two Lerms. This
selects terns in S, that commute with rS, which is uniquely the axial vector
contribution of (2-9.70). Nsw, ho~\rever,it is sufficient .t;a set m = 0 in the
eEe~tivesoume (2-9.76). It seems to be a specific property of the second-rank
spinor repremtaticm that the source of massless spin 0 pa&ides acquires the
special form of the divergence of a vector.
-+
FIELDS
&-3
SPIN O PARTICLES
Sources are intrdueed tia give an idealized description of the creation. and the
dcrtection of particles. But the puwose of this activity is to study the proprties
of the particles, snd this takes place in some region inkmmedictte htti-een the
locations of the terminal acts of creation and debction, Thus one n e d s a convenient measurn of the strength of the excithion that is produced in s w i o n
that may be far from its sources. Ttre natural way to obtain such a memure is
by investigating the eflect on a probe or test source that is introduced into the
region of inkrest. Accordingly, con8irleeng spin O particles and their real scalar
sources, a represented by
where
This combination of source and propstgation, function, measuring the eEwt of
preexisting sources on a weak k s t source, is the$eEd of the sources, 1%is defined
in an analogous tvay for any type aE particle, as indicatd by
146
146
Chap. 3
Fields
since p 2 mZ = 0 in these integrals, while the discontinuity of the time derivative across x0 = zO',
+ m 2 ~ ~-E23( =~ ~ E ( X- X'),
[-
(3-1.11)
where
(d.)
6(x - X')
C-,
(dx) b(x
- x')]g
(3-1.12)
c-, SE(x
- X')
(3-1.13)
or (xq = isa)
( I / i ) 6(2 - X')
Unlike the Minkowski situation, the two fundamental solutions of the Eudidean
differential equation are sharply distinguished by their asymptotic behavior:
-e*"R. Thus the requirement of boundedness, for X # X', uniquely selects one
solution, the one that is produced automaticaIly by the Fourier integral solution of (3-1-11],
and A+(x - z') i s recovered by the previously explored procedures. The alternative methods of imposing boundary conditions can also be applied directly
3-1
Spin 0 particles
147
where the minus sign of the second term recalls the opposite sense of time development that is involved. The two fields encountered here are
(3-1.19)
(3-1.20)
Then
> x0':
in which
(3-1.23)
we see that
+(-)(X>
where
#(+)(X)= +mt.(x)
-x')K(~)
(3-1.25)
(3-1.26)
548
Fields
Chap, J
(-a2
+ m 2 ) ~ , . t . ( z=) K(z)*
The retarded field of an arbitrary source ean be found by solving tthia equation
\vi%hthe boundary condition that the field be aero prior to the intcjrvexrtion of
the sources. 1%is remarkable that this classic boundary conditbn requires the
device of the closed Lime path for its appearance. fneidentsfly, the f o m awumed
by &Wfar the cimumstanees stated in (3-lam),namely
(3-1.33)
3-1
Spin O partictes
lrC9
where
one gets
s
the final boundary condition.
which m ~ k e expliciG
Still other kinds af fields md Green" functions appar on replacing the
vacuum state with a general multiparticle state. Rather than USE? any specific
one, we consider a pararnetriaed mixture, as in
C
(m+l @
- ) K ~Ia ( 1 = exp[iW@(K)l,
tn!
where
(3-1 -36)
and 19, is an arbitrary time-like vector with > O* OR reviewing the discussion
of ({B)+l{n) J K , pa&icuI~rlyEq. (2-2.46), we recognize that this probabgity
amplitude is linear in eaeh occupation number n,, which is merely =$ace$ by
an average value in (3-1-36), namely :
(3-1-39)
with
A6(2
- g')
= 6+(1:- X') f-
-+- G-'"~-
"'1.
(3-1 ,401
which
defined by
(3-1.42)
--a2
+ m2.
green'^
where
are relabd by
On noting that
(%,)a
(3-1 .del
2).
+ 1 = gP(np)sp
(3- 1.47)
ab-)(~- g')
= a;+'(%
- 2'
+ ia),
(3-1 +48)
0 < z0 - zO"
or
--X0
.XXTD
/2@
3-1
If one wishes to verify that the periodicity condition. does produce the
de~iredsolution of fhe Green" function diflferential equation, it is convenient ta
adopt the rest frame of the time-like vector X@,with X' = T,and satisfy the
periodicity requirement in z0by using Fourier series while retaining the Fourier
intepal. treatment of the spatial coordinates. That gives the Green" funetion
repreentation
12
T
--
@(l
i2)ip0~e-ip'lr0-."/
EI ~ ) i ; p o ~
@--(X
1/ 2 ) i p 0 T ipOlrO-so' i
IP I ~ ~ O T
(3-1 -58)
and the substitution inverse to (3-1.50) in the rest frame,
followed by removal of the reference to the rest frame, does indeed produes
Aa(z - 3'). The same results are obtained directly from the digerential equrttion for +(z) by imposing the periodicity boundary condition
2"
-(2
- 3') :
-ihk--'(X - g'),
= z0 > sop:
f +f
zo < 2" : -ia, ( X - %'C').
(3-1.62)
The designation th& we have given it exploits the following form& property
of the averaged oecupw;tion numbers,
@,)-a
==
and thus
(4-1
L'%-@ (z
- XI)
= --rhk-)(z
-- zt),
*'-"
-e (x -
57
= -A# (4-1( X
- X/).
(3-2.64)
Chap. 3
(dg) SK,-)(z)~iat-lI2)
(3-1.66)
are
(dx')~;-~(l;'(z
- zf)K[-, (z'),
since sX1.the eausall relations among the var;ious bc func-diom are hdependexrt
of p, tas is
(o+lo-)"
Spin Q) parclclas
exp[iW(K)],
( ~ z ) ( ~ L . ~ ) K * ( + ) A--, ( xx')K(zF),
163
(3-1 35)
One must not be misled by the notatiori and conclude that these fields stre in
complex conjugate relation. That is s correct rts%rtiort about the differential
equations they obey,
but t;hese equations are to be solved with the same boundary coxrditiocls-th&
of ouLgoing waves in time, by ~vhichiu meant positive frequencies in. the futufe
and negative frequencies in the past of the source,
Let us examine the structure of these fields in the two mympbtic time
regions. If the fields are evaluated at a time after the sources have ceased
operation, that causal circumstance is expressed by replacing P+(z - z') witfi
h
( - X ' Thus (z > K suggests the causal arrangement):
and
z
> K:
< K:
< K:
K *(X')
(dup)"2e-'Pzi~~+
=
P
IM
Eiefdd
Chap, 3
The two eausal evaluation8 of the fields are assseisted with particle emimian and
absorption processes, respectively. They assign the field (h,)
'j'e'~' to an individual emission act and the field (dw,)v2e-'p' to an individual absorption act.
As in. the interpretation of complex sources these field8 produce definite charge
chsngw. Depending upon the causal situation, +(z) de~cribesenrrit;td positively
charged padieles or absorbed negatively cha-ed padicles, while +*(g) rcpre~enh
emitilt;ed negatively charged patticleg or absorbed positively charged par-t;icle~.
The time cycle vacuum amplitude is
) ( d z ' ) ~ t(z)bL+'
,
(z - zf)K,+,(X')
&W(Kf-,,
The ernlicit forms are
(dz)[s~:+,(z)4,,
,(z) 4- &~,+,(z) +:+,(X)
~,(z) +,-,(X)
t
- a~,-,
+L,(X)].
(g)
(3-1
(3- 11 '84)
and
If is seen that the field struedure already given in Eq. (3-1.19) is duplicated
here, and the earlier discussion can be applied, enlarged by the substitutions
K + K", cf, -+ $*. In particular, when
which implie8 the analogous complex coxrjugate cqurtdions, we have
where the d m retarded fields are complex conjugates since A,,t;.(x - X') is a
real function. One implieation of this property is that any small deviation af
3-1
Spin O particles
186
GW(K(-,, K,+,> =
+,(g)
-- GK<-,(X))
= &W(K,-,,K,+,)*.
+,et
(3-1 239)
The replacement of' the vacuum state with a general rnuZdigarticIe s t a k can
be parametrizcd, as in (3-1-36), with tbc tveight funetion
and d+(z
- x f ) becomes
according to (2-2.59).
z and x" but there is a symmetry in \\.hich. positive and negative charges are also
interchanged,
a 4 ---a,
and
das(z
The functions &;$'(X
X@)E
ACt--aa(~'
--- z).
(3- 2.94)
+ 1 = e"'BP(gp+)a~,
+1 =
(@P--)agP
-- 2') = ea~L~',;;)(z
- Z' - $S),
A~;'(z. - z t ) = e-olhb$'(z - 2' + ip).
,at*}
(Z
-.-
X')
- X' - - X)
&e~4B(1:
= e-Qa8(z
- 2'
+X),
(3-1.rw)
Although we shall not; @ve the details, this boundary condition on the propa-gaLion funetrion her&(%
- 5'1, in conjunction with its driaerential ewation, dow
reproduce the original funetion.
The counterpart of the multipiarticle replacement of h+($ - z') with
&,@(S
- 2')
is
~ - ( x-
--t
- g)*
= h-,
me(%- X').
(3-1.1QI)
Nob that if we wished to write the last term in the aldernative way that uses
the h'*' function, it is necessary to change a into --a,according to Eq. (3-1.97).
The fields defined in the manner of (3-1.83) are now given by
3-2
Spin 4 particles
1S7
and
PARTICLES
where
G+(z
- z') = (m - ? @ ( I / ~ ) ~ ~ ) A + (Z z
').
(3-2.2)
Let us obeme innmdiately that t h e algebraic prope~iesof the Y" matrices imply :
(r@(~/i)a,
+ n)C+(z - 2')
= (-a2
- g').
(3-2.3)
This identifies G+(% --x') arj. a Green" function of the D i m mat^ diffaential
operator. Aoeording to the stmcture af &+(X --- z ' ) ~it iei the one that obeys outgoing wave time b o u n d ~eondi$ions.
The field definition fo be used here is
~
where has s different scale in the trvo expressions. Alternatively, the Green's
function is constructed from solutions of the homogeneous Dirac equation,
G+@
(X)$,@,
P@@
W )*TO,
G'-'(~ -- g') =
=
-C +,.s(z)
'+p.p(z')rO,
PSQ
(2mdo,)'izzl,,,e'ps.
(3-2.1 l)
A charge label appean since this is a general attribute of spin particles. The
inhomogeneous term of the differential equation (3-2.3) is equivalent to the
time diseontirtui$y
f he f istd cancept,
3-42
The x
<
Spin 5 psrtictsa
159
( -1
1 ( ~*) = 2-112 ( 1 ( 3 ) ( 2 )
i9~2)(2))-
(3-2.17)
9(s) =
rl(z)"y0,
Chap, 3
+(z)a,T = -a,$(z),
To confom ~ t the
h new notation, we mow ~ t c ;
and
apip.
= %;@ro.
We
m~8$
also fib
which restabs the property (3-2.7). The fields in the two causal situottions are
obtained W
with
E +P@(%)r0iv@@-,
Pip.
3-2
Spin f particles
181
and
where
P
22
P@
+Pu(s)iT;u+,
S:,-
The specifications of the particle sources follow from the earlier discussion by
identifying $(z) and $(X)?'
with the projections of the eightcomponent field
o n h the positive and negative charge space, respectively, To use the fields
#(X:)
and $(x) is surely the most familisr and the most pptrlar waty of appfyhg
the Dirac equation. Neverthdess, we regard the asymmetry of the forms
(3-2.30,32), in contrast with (3-2;. 14), as justifying, in general, the employment
of the msl sources and the multiieomponent fields thab are defined in charge and
spin space, rather than the pairs of complex sources and their sssoeiated fields.
The time, cycle vacuurn amplitude
tlf+))
are given as
Since G-(z
- 2"
-,
f n the ~peehlsi$uation
~f(-j(z)"'-- 8 [ + ) ( ~ X
) 9f~)
(3-2-45)
one evidently h=
J.c-1(2)
z=z
(3-2,46)
with
(3-2.47")
3-2
Spin 8 parOZctass
163
whiel.1 vanishes before the sauree earnrss into action, The form of W far smdX
deviations from the situation (3-2.45) is
(&~tt.,(.)
- a~,-,(z)) u"$m,.(z)
(3-2.49)
(3-2.51)
(X
- z') q(zf)
(3-2,521
to determine
C ((4+I
-)'pa&( ( R ) ) = exp[ilva8(s)l.
(3-2.53)
Far sirnplicitJI, no para-meter hsls bwn introduced to distinguish tfie w i o u s spin
are, explicitly,
and
+ (1
(-4-1 '
[r0Go@
(Z
)l?',
~ ~ p q ~ a ~ ) *$psq(zf
+ p ~ g ~ ~ )
(-1
z)lT = -7 0Go@
(Z --
X'),
(3-2.56)
\\-hieh expresses the necessary antisymmetry of Y ' G , ~ ( ~- X') under the complete transposition of space-time coodinabs and discrete indices. The relation
leb
fields
Chap. 3
t~herethe symbol q now indicates &hertntirjymmelricd charge matrix, The corresponding Green" function property is
-X)
-e-aqQa4(.~ - ZI + X).
(3-2.59)
If cr is mt equd to Bern, or, more generally, is accommodated in
by a coordinab dependent rdefinition, the boundary condition on
- 2') ~ppf-t~rs
ass a s i w ebange in respan= to a coordinate displacement by X"; this property
might be -lied sntiperiadicity. Concerning the t i m cyele generalization of
d h w rmulLs, we shall only remark that the mulidipartieXe repbcement for
O,(x
3') i s El, -@(z - z'). This statement is equivalent to the relations
ag(r
---.
-,
-@(S - 2')
G%'
-- G:~)(Z - 27,
(4-1
- 2') = -GoB
(X - X'),
(3-2. (30)
and they follow from the aver&geoccupation number property
(--J"""
(npp)--a
-.B
-@(X
- (@pq)a~
(3-2.61)
Spin 1. According to Eq. (2-3.4), unit spin particles of mass m s" 0 am deseribed. by
+ (I /m2)a,~p(~)a,(~- x')~EJ'(z')].
The consideration of
a,
(3-3.1)
(dzf)A,($
- ~')aa'(z~).
(3-3.3)
and this defived scslar field vanishes outside $he region oecupied by the murce.
The differential equation that is infemed from (3-3,3),
3-3
166
on using the relation (3-3.4). Another version of this differential field equation is
where
(3-3.7)
(3-3.S)
The differential equations that relate the vector field to its source conversely
determine the field when appropriate boundary conditions are added. The
divergence of the vector equation regains the relation (3-3.4), and thereby the
form (3-3.5). The soiution of the latter with the outgoing wave boundary condition is just our starting point of Eq. (3-3.3). In this and other examples of B. E.
systems, different boundary conditions can also be used, in straightfonvard
generalization of the spin 0 discussion.
Spin 2. Massive particles of spin 2 are described by [Eq. (2-4.20)]
+ (2/m2)asTp'(x)a+(x- xf)aX'T,x(xf)
+ ( l / m 4 ) a f i a , T ~ v ( x ) ~-+ (x')a:a'x~"~(x~)
x
- 5(T(x) - ( l / m 2 ) a , a , T v ( ~ ) )
x A + ( ~ - - x ~ ) ( ~ ( x ~ ) - ( l / m ~ ) a : a ~ ~ ~ ~ ( x(3-3.9)
'))],
in which
T ( x ) = g,,Tpv(x).
(3-3.10)
(3-3.1 l )
is obtained as
- ( l / m 2 ) a ./ ( d z r ) ~ + ( z xt)aA'~,~(x')
+ (l/m4)a,av/(dxr)a+(x - ~ ~ ) a : a : ~ ~ ~ ( d )
- *(g,,. - (1/m2)a,,av)/(d~')A+(z
- x t ) ( ~ ( . r r-) ( ~ / m ~ ) a # i ~ . ~ ( ( z ' ) .
(3-3.12)
The divergence of this tensor field is the vector
which vanishes in source-free regions. That is also true of the scalar field
where
S,W(P) = g,.
167
+ (llm2)P,Pv.
- ig
(P)
[ ~ P K ( P ) ~ V A
- &~I*(P)SK~(P)IT"'(P).
(3-3.28)
(3-3.29)
1 1
O(P) = ;;;i[;;;iPKPI- 38~k(P)]T"(P)-
(3-3.30)
and
As we have noted in Section 2-9, such contact terms contribute neither to tlte
va~uurnprsisknce probability nor b the coupling of causally disposed sources.
It is only through the additional con~idemtioninvolving the structure of field
equations that a reason for their presence and specific appearance can be addue&. The maintenance of the inharnogene~usfield equation form that we now
regard as standard nlso requires the introduction of certain auxiliary fields,
ivhieb vanish outside the regions oceupkd by sources.
The discussion of spin 3, which is intended to illustrate these remarks, \\-ill
be: facilitaM by first examining the sirnpfer situation of m-less pa&icles. We
have not done this for spins 1 and 2, since the* physically important examples
will receive individud and exknded treatments. h t it be remarked, however,
&at the appropriate field equations are & h i n d merely by settiw m = O in,
my, Eq8. (3i-3.6) and (3-3.19). Also, the fiysically neeessaq soume restrictions,
af vmisbiw divergences, are algebraic consequencm of the field equations, ars we
em recognize from the m -. O limit of Eqs. (3-3.4) and (3-3.15). But more of
thk fater. The massless spin 3 situ%tionis represenkd by [Eq. (2-5.122)I
ssinee any additional term containing p&, p,, or p, ss Erzctors will not eontribuk
in (3-3-38), oiving to the souree restriction (3-3.37). Hence, the genwal form of
the field is
in rvhich the cy~licallyrelated sets of terms are required by the total symmetry
+h,,.
The new symmetrical tensor h,+)is defermind by the
murce restriction. In order to urn the latkr, we first note thzzit
of the knsor
3-3
16s
Multiplication of Eq. (3-3.39) with p9then introduces just the combination that
is evltluaLed tts *(#h --- p&+), and we get
with
P~P'P~#FV
pkp@p'+rp. - p2pk+*
namely
(p2)'+ = p2pk+k
--
$pkp'pB+hpr.
(3-3.43)
(3-3.44)
(3-3.45)
The construction of +(p) given in (3-3.45) is derived directly from this equation
by multiplication with phpppv. It might seem that we have failed to meet the
objective of providing second-order differential field equations. Three dePivaLivers act upon Q, and the scalar field obeys a fourth-order diEerenti%lequa;t;ian.
But notice that the field equs.tion involves only this combination of fields:
where dots represent the terms that are genemted by cyclic permutation from
the gven ones. The following is an algebraic consequence of this equation,
it is consistent with the vanishing divergence of the source, but doe8 not imply it.
Xow let us consider a mwsive particle of spin 3, fir& using maltere$ the
description given by Eq. (2-5.95) :
I70
Chap. 3
Fiaids
c31
==
5;
= g h h ~ j l p , ~ g vv ~ + [ g ~ , g ~ . ~ g r ~gppyhhgg,ppfi
,~
~ ~ h ~ ~ , ~(3-3.52)
g ~ ~ h ~ l
in whielt the necessary symmetriaation in h f , p', v' h= xlsL been made explieid.
Some propertim aE this tensor are
4-
Phhfigpfrt 4-
PhP,'
7
8 . 8 ~ f
(dP>
-&sApp(-p)
qh&v(p)
( 2 ~ ) ~
(3-3.55)
37-23
Quite? apart. from the explicit appearance of numerous derivatives acting on the
source, the eoegeient of Sx does not have the value given. in. Eq, (3-3.46). Thak
is rectified, however, by considering
for then
sbp[grh'8,ptgpp,
-~g,vg~~fgp~w~]~x"'pf
c, (3-3.82)
with
we get
21-r.
4hld.Y
cont.
-P~PRP,
p2
?a2
gpvpr p8."
2;i;;li ~
(3-3.66)
Chap. 3
where
The addition of these terms to the righGhand side of Eq, (3-3.62) removes the
third der-jivativea of 2, and replacm them by fimt derivatives of @. And %her@
is an additional contribution in (3-3,M), which can be d d e d Lo the lefbhaxld
side as
l
-2% 2 P 2z+-ZOmflp2(m2
- P ~ ) P ~ S ~ . (3-3.69)
=f
eont ,
N o b that the malar field effectively vanishes as m -+ Q, and (3-3.4fi) is recover&, When supplemertted by outgoing w a w boundary eondi$ions, %het;
unique solution of the set (3-3.70) is the field 4&,, ~ v e in
n Eq. (3-3.56).
Spin 4. According to Eq. (2-8.10), this system is described by
61'(-
p)r0$@(~)
IS, therefore,
1
= 2 ; i b - Yp)p,
+ *YP(~'YYf p.)lrlV
(8-3.72)
3-3
173
and
- p,) (m - Y p ) ,
(m - ~p)(m?;- ~n,),
-C-
4- ?P)
==
(3-3.78)
(rp 4-
m)& =
4-
&r,
qv
1
+- 3
pp,Ip. - S(mr. +
PP)~?',
(3-3.79)
from which we immediately obtain
Like the field equation, for spin +,this is (the momenlum space e q ~ v a l e nof)
l a
firr~t-orderdigerentiaf,eqtr%%ion
~viththe source appearing m the inhomogeneous
hrm. The solution of the equation under outgoing wave boundaq conditions i~
the field given by (3-3.7'3). As is indicsted by the m -.sl O limit of (3-3-76}, the
necessafy source resl.riction for massless particles,
With the examples of spins 2,3, and # b&m us we can recognize the possibiliky of simple dgt3braic redefinilions of the seurces Lhat preseme the general
stmcturtt, of the field equations, but introduce or modify contaet brms in Lhe
expression for W , Thus, let
On introducing this redefinition into the expression for W ( T ) ,say Eq. (3-3.241,
the additional g,. terms supply only contact contributions. The explicit statemen&is
where W(2")has the same funetiond form as W(T). The implied field transformation inferred from
Since field and souree are frtznsformed linearly, and Ioeally in spacetime, the
daerenlisl field equation mainfains its general farm, buL with changed caeEeierrts. This is illu~krakdfor a = 4, where
(3-3.92)
urhich maintains the general structure of the fidd equations but & a n p the
eontael germs, cannot be aged, to Iremove the latbr, As we ohmwed, second~rderdiffewntial equations lacking in source derivstive~are not obtained if
canttbet term8 are o ~ t ; k d ,
3-3
176
ean be written as
+L
or
= $g
+ av,rpr(..
where
The transformed version of the field equation (3-3.80) is
(?P $
-- a(pPyY$:-/-
in whieh
a' = 3u(a - 1)
+ 1,
(3-3.100)
+ 1).
(3-3.101)
4(3a2- 2a
Eatice that there are just two situations in which
a" =
The fimt gives the original field equation, and the second produces
a=*,
(3-3.105)
whieh gives
%.
Spin Again, we turn first to the massless situstion in order to get the simplest
indication of the field equation structure. The appropriate specialisation of the
176
Fields
where
- f~"(-~)~'~~(-rp)r~4;~9(~)
-- f s ( - - ~ ) ~ ~ ( - ~ p ) s ( p )(3-3.107)
l,
?(P) a Q ~ W V ~ ~ ( P )
(3-3. XQS)
and
12,1.l"""(~)
==
0.
(%3. 109)
&g"'(
--P)$&P(P)
(3-3 * 110)
is not unique, 8.lnce $he source is restricted by (3-3, tW), ftAs general form is
is eonskbnt with the souree restriction, but dws not imply it.
By f o l l a ~ n gthe ins6ruetions given in SecCion 2-8, we a ~ v ade the f o l l o ~ n g
scrum dwripdion for a,masgive. padiele of q i n #:
Le%us also @ t a bhere the dditiorrd contact brms that are req&ed Lo bfing
the field equations into first-order fsrm without source derivatives. They are
derived from
Tht! multispinor description provides a unifid approach do a11 spin values, wirth
the first-order digerential equation af spin set.ting the pattern af field equations.
I n order to reafiae t h i ~sfftndard, ho~sever,contael Germs must be introduced in
d l situations save that of spin +. Consider, for example, tho descriptian of unit
spin thad L provided by $he symmetric spinor of the second r&nk2as contained in.
The appropriate contact term has already been inlrodueed, and the soume
differ8 in normaXiz;%ianfrom that used in (2-9.8), as indieat4 generally by
q(p) = (2m) 2'"-"S(p).
(3-4.2)
(rzp
and addition gives
Crr
~ >
( ~ 2 ~ 1
2e
(3-4.9)
The simple algebraic property just recorded provides izxl e1ementtaw snd
generally useful eantrol over our procedures. The eigenvttlue relEtLZLOnship~,
TIP = f r,p, are invariant statements of the restframe possibilities, r! = f 720,
where the plus si= se1ecl-s the appropriak aubspsee for the dwcfiption. of the
particle, Correr~pondinl;Sty,setting ~~p ==: ? % p in (3-4.6) reduces the l%tterto
the form of the spin Dirac equation, while the choice r2p .- -rip effectively
removes the caordinizh de~vativmand. supplies a field %h&vankhes outside
%hesource, The ~trueturegiven in (3-4.1) b c o m e ~mare obvious, for, with
TzP = TIP,
which puts into evid~neethe contact term and rrormli~ationthat are needed
do attain (3-4.8).
The effectiveness of such considerations becomes clearer on turning t a the
nnulti~pharof rank 3, where the choice h p = Yzp = Yap leads to the wad=tion of
Multispinor fields
179
(m - TIP)(^
p2
- 72p)(m- 7 3 ~+
) 3m
+ m2 - ie
+(P> =
+ 3m - f C 7.p]
n(p)
(3-4.13)
that is defined bv
There are only two general alternatives open to the ?,p. Either they are all
equal, or one of them has a sign opposite to the other two. These situations are
characterized by
it can be verified directly. Note that the field statement of Eq. (3-4.16)can also
be presented as the unrestricted equation
ll&0
Chap. 3
FOsids
where the latter illuslrahs the set of three equation8 th& are relatd by eyelic
permu%
a%ion.
It is aIso inbre~tingto eliminate the three auxiliary fidda and pm~entthefield equation in the form
or, slbm&tively,
The Iakter may be earnpared with the second-oder di8erent;hl eqamiLio.on for
unit spin :
[ P ~ + , ~ - - - { P ~ + ~ ~ c Y ~ P ) ~ I (ImJ -. =+C.,,,
But now there is ambiguity in giving (Yl p ) 2 s more general inkrpretation; shall
it be --p2, or Q
~ . p ~ @ pIn? fsct, we shall use s ~pecificlinear combination
of the two, 80 ~hogenthat p W e mof moment&in 3/ are held h the minimume
182
Fields
Chap. 3
in whieh $he last sunnmafiion is extended over $istinet pairs, ar < 6,"a
p',
tit: art6 @ p', with ~ 1 repetilious
0
counting. This is equivalenf to s faurt;hh
order diR6rendial equaf ion,
Turning ts fifth-rank multispinors, we first note that
while
+g
which obey
(TIP
5m)~[a@l[a"~~
-k Yl~$ladSlfla'~"
0,
(3-4.53)
C'
a'<B6
+(Y~'P
Y@'~)$[aflj[a~@~l
= 0,
(3-455)
which imply
The next in the sequence, ~vhicfnis as far s s it shall be developed here, are
where the k r m s left unwritten are those referring to three index pairs, This set
has the property that
186
Chap. 3
FIatds
That is, beginning with the quadr&ic source expression for W, ~vherer is the
approprirzk repregentation of the metric, fields X are defined through the consideration of an in6n;itmimal test souree, The noalacal space-lime relation
between field and source that is conveyed by G is then converted into a local
differential one, which is symbolized by the operator F. Alternative expressions
for W are
and
(Srx - gxrFx),
To this point, the field X has been a derived quantity, a eorrvenient shorthand for M, It now acquires independent status, in the fallowing ~ense,Forego
the knowletfge of any connection bedtvwn X and S, and subject them to independent vetriation in (3-5.7). This gives
There is a more symmetrical form that contains only first derjvatives of the
field. No addit;ional surface integral term is assigned to the partial inkgration.
This can best be appreciated with the aid of the assoeiakd Euelidertn description
where fields decrease exponentially at large distances from the source, sinee that
is the characteristic of AE(x - X", m 7C O. Even for massless pa&ieles the
(z behavior is sufficient to suppress infinitely remote surface integral
eorttributions. Accordingly, rre ~vrite
@
(@(z))
l2
(3-5.12)
+ ?n2)&(z)= K ( r ) - a,Kp(z),
(3-5.1 4)
wyhich exhibits the same kind of efieetive sealrtr source that Etas already been
encountered in. Eq. (2-9-76)- Tbe corresponding action expression, is
f f the first equation of (3-5.13) is regarded as a definition, (6, loses its independent
position and we recover the action %pression (3-5. XI), ~viththe effective scaler
source indicated in (3-5-14), and the additional contact term - ~ ( d z ) + l < ' ~ , .
188
Fields
Chap. 3
a,$,
- a,4, - G,,
nr,,
-nr,,,
a,G"'
+ m2+w= J ~ . (3-5.26)
(3-5.28)
(3-5.30)
(3-5.35)
ahcpPh
- apG, + ma(+pp
+ +g,.+)
= F.,
(3-5.361
Kip, == *(Kx,,
K& = KhFp,
+ Kgkv -- KgFx
K ~ s ,-4, Rpgk,),
(3-5,s)
"
(3-5.42)
ahH,,h
= Tpv-- +g,.T,
(3-5.43)
afl+rvf
a&hB
H,v&= L:.,,
(&5*4)
in which
trP(r/i)a,4- mIJ/-(s)= v @ )
(3-5.49)
Action
3-45
t91
Spin
4.
+ $1
The second and third terns effect an explicit symmetrization between the application of the derivatives to the ~ g h and
t to the left,
This is _automar;Cicfor the first and fast k r m s of (3-5.55).
c = -+($@yro[r"(l/i)ak
- 2#ppr0rp(l/i)a"k.
+ C],
(3-5.57)
+ m]+r. - 2vyr0((l/i)8,rk#hv
+ 2pvror,[m - ru(l/i)d.]th+h.
(3-5.58)
....
Spins Q,*, 1, g, 2, #,
Under this heading are collected the multispinor
descriptions, Thus, the following ig applicable to sipins 0 or 1 :
It shaufd not b fargotkn thst these even-rank spinors are commutative quttntities (B. E. statistics), matching the symmetry of ?!?g and the antisymmetry
of Y : Y ~ ( Y ~f 79. The second-order differential equation (3-4.26) can also be
adopted as the bmis of an &ion princble. Let us write it as
where the fachr 112m is supplied to make the two action expressions direetly
compar@ble;otherwi*, it esn. be abs~crrkdinto a common scale factor for field
~tndsource. The value of the W that is implied by (3-5-62) can be pre~entedm
which a s s e d that the two actions diger in eontent only by a contact term. The
Iatler removes the contact term that was added in Eq. (S4.1). Another such
remark is based on. the eornmutativity of the s y m m e t ~ c dmathx i ~ ~ ~ 2 "
with ?!?g, as well as with the matrices (ror@),.
On decomposing $ and into
the aetion
component8 with the aid of the projection matrices +(l
expression (3-5.62) completely separates into two independent pshs. Thus,
it is possible to use a reduced form crf the action. principle which contains only
one of these field components and its assaeiaLed source* The latter aetion should
be multiplied by two in order to retain the same scale for sourees and fields,
That the above procedure only changes eontatzt terms is vel.ifid by considering
Action
3-5
193
+,incidentally, by
(dz)[r(z)7@16(z) C (#(2))1,
diEerv from the original action only by a contact term. Next, supposc? that one
of the two projection matrices
YO
consistent projectioxi of field and source onto a subspace has bwn brought
about. The signifieanee of this new aetion is given by
a,
only the contact term has been altered, and the same physi~alsystem is h i n g
described. But ~ ~ h e f hthis
e r second-order formulation,af spin has any pra&icaf.
merit will not be diseu~edhere. One remark is in order, however, The r6dependent contact term in (3-5.70) i s imaginary (roir,is antisymmetrical and
real) and should be subtracted from the second-order action to: prBerve the!
detailed physical equivalence of the ttvo descriptions, Sinee thia subtractive
term is given contact form through the use of the source v, and not J, it emphas i ~ ethat
s the ~eeond-orderfomulatian could not be adopLed m the fundamental
spin description.
7M
Chap. 3
Fields
Spins -4- or
m represented by
where
r=
:
7 = 7:y;yg
and
in which
c = [m - i$
C r:(l/i)a,ln*
(3-5.75)
where the last term can dsa be wfitkn [cf. Eq. (3-4.19)l
The symmetricd field and ssource spinors of unit spin me dven general form by
(da;)fJ""br,
4- +M""%,
+ cl,
+
(3-5.82)
This .is the firsborder form (3-5.27,28), with the derivatives symmetr.rzed in
application; to
vtt~-t;orand knsor fields;, Similarly, the 8ntisymrnetricttl
spinam af zero spin are presented rias
(3-5.86)
-- a,Kr),
(3-5.87)
and
2n-'l2$ = ir9r0@, 2rn-"'t
= ir&~'(K
where the f structurefs are the appropriate projections of (3-5.61). With thee
rduced fields and sources understood, the aetion (3-5.62) becamm (a factor
of 2 is supplied)
The trace evaluations give directly the action expressions (3-5.18, 19) and
(3-5.11, 12) with the effective vector and scalar sources, explicit in (3-5.86,87),
th&t have been s t ~ t e dpreviously. One can also mske the opposite choices in
these projections, and we record those action form which are, for spin 1 and 0,
Actian
19c7
and
The effective sources that appear here are M,. - (1/2mZ)(arJ. -- aJ,) and
-- (11m2)d,K, respectively.
Wr; ahall close this section with a few varied comment8, First we recall that,
although we have not illustrated it here, the possibility exists of redefining
sources and fields by linear transformations which change the detailed ~ t m c t u m
of the field equatians and, therefore, of the Lapange. functions. Then it is noted
that all discurnion has been concerlled with the vacuum smplitude (O+~O_)~.
The shift of atbntion to other transformation functions k convittyed by a cfiange
of boundary conditions in the action principk. Let us be mplicit &bout the time
cycle transformation function (o-/o-)'-~'+. Here the action separates into
two snalogous terms, with opposite signs, that are associated wilh the two smsw
of time development:
K,
798
Chap, 3
Fields
which, as shall, be discussed in greahr detail laCer, deseribs the photon, and
rtlferring to a massless gatPticb of zero spin (helieitd. Notice that the Lagr&nge
funetion of the spinless particle comes entirely from the mms term of the original
Lagrange function, and would have been overlooked had one merely get m = O
in. (3-5.19).
The other example is spin 2, where we express the tensor field and its
source by
= h,.
--
(2-lt2/m)(a,A.
and
a,TF9 = m 2 - ' l Z ~ & ,
@p&],
(3-5.96)
(3-5.97)
the latkr being Eq. (2-4.21). These structures are such that
When (3-5.96) is inserbd into the actiorz (3-5.30,31), and the limit nz 0
performed, three independent parts are obtained, Two of them restate the unit
and zero hellicity actions, Eqs. (3-5.94) and (3-5.95). The third one is
As we shall also discuss later in more detail, it is the (or a) gravifon action
expression. This time the photon Lagrange function emerges completely from
the mass term of the spin 2 particle, as indicated by
but the scalar field action has contributions from b t h parts af the original
Lagrange function. Xt is an interesting unification of the actions repre~enting
massless particles of various helieities to connect them with one action expression
for a massive particle, Also impXied are the relationships between different spins
necessary to amive ad a common description for a given helieify.
Ghsrga
199
The vacuum amplitude (o+/o-)' is unaltered by a rigid translation or rotation of the sources; for charged particles it remiins unchanged by B univ m d phase transformation of the wurces, Physical information is obtained
through the relative tmnsformation of diflferend pa&s of the source distfibutisn.
Relakive tramfation gives information about energy-momentum, relative rotation about angufar momentum, and relative phase displacement .tc?acbesabout
charge. When the source distribution is causally arranged, with the dbjoint
pieces treated a%units, one acquires knowledge of int-egral physical qu%ntitiesb t a l enerm, total charge. At the next stage one considers transformatiam that
vary arbitrarily in space-tinrle, thereby supplying more locdized data about the
various physical quantities. Fields are the instrument far eonve~ringthese data.
To illustrate these remarks, let US consider spinless charged gttrtieles of
mass m, as desel.ibed by the action
where K ( z ) and +(z) are now two-component objects in an appropriate Euelidean charge space, Invariance under the constant phase transformation af
the source,
K ( z ) = eiY9K(z),
(3-6.2)
follows from the existence of the compensating field transformation
since all the Euclidean products in (3-6.1) are unchanged by s common rotation.
Next, led p become an arbitrary funetion of position. For simpfielty we consider
an infinitesimal phase transformatisn and write this generali~ationof (3-6.2) as
Fields
Chep. 3
where
j@(.) = a"+(.)ig#(.).
The eonsiskney of the two evaluations implies, with the aid of an irrtepation
by pards, that
(&G, 10)
gPjp(z)= (z)iqK(z).
This is verified directly, on using the field equations. When the righbfrand side
iis zero, which is true in source-frw regions, we recogni~ethe local stakment of a
con~rv&tion
law. If the charge matrix Q is diagonrtliged and ~ompfexsources
intrdtreed, the action expression. and %heproprties of jfi became
and
d p j f i= i(@*K- K*+),
(3-6.12)
where
th& #* is naC the complex conjug%teof 9. S t a b m e n t ~anaEoto all thew apply to the vacuum time eyele action with agproprbte afge
braic sims in, 6t.F" Lo indicah fhe sense of time flow.
Using this more general, framework, we now re-examine the causal situation
with K = K Z ITz, where the phme of K 2 i~ changed by a eo&ant and that
of XI i s held fixed. For the infixliksimd dransformstion b i n g consider&, we
know that
~ t hthe
, remind=
PUB
This wei&ted average of the charge values rewmbXes an. expectation value.
I n d e d it is one, if we consihr the time cyele function, with the phase of K,+,
displaced, a f k r which $he two sourea are identifid with K,
3-6
-a,
GP(%),
Charge
2@9
which identifies
dgIr;iIr=
(Q):*
In the latter situation jP(z)is computed from the real or mutually conjugatpl
mtarded fields, and is a real function.
f t is evident that j"(z) provides a space-time account of t h distribution
~
and
flow of charge-it is the charge Aux veclor or current vector, We sh&llevaluate
it for a single-pahicle state. On referring to Eqs. (3-1.79-81), it is seen that the
fields in the region between emission source K Z and absorption Bource .Kl, W*
eiakd with a positively chsrged particle of momentum p, are
To compute the tots1 charge one can integrate the eharge density jyx) over all
202
Fields
Chap. 3
/(dx)lLap d% ~ X P ( ~X)P
(do,) '12e'pzi~zp-,
4(x) = (dwp)112e-ipzi~*
IP-,
(3-6.26)
and
jW(x)= -2pp dwp(iK~p-)(iK2p-),
(3-6.27)
with an analogous verification that the total charge of the particle is -1. The
retarded fields of the time cycle description that are associated with a given
momentum, and positive or negative charge, are [cf. Eq. (3-1.84)J
where the surface integral refers to any space-like surface that is subsequent to
the source region. On any surface that precedes the source, the retarded fields
and the current vanish. The explicit form of the right-hand side is
3-6
(I/i)[bmt.(z
--.
X')
- i?rrCt.(zL z)]
";=-
(I/i)[b,,t.(z
Charge
203
--.
(3-6.33)
and
This supplement to the charge density, V n(x), ~vheren k = m'*, adds a twodimcnsianal surface ixrtegral to the charge associated with a three-dimensional
volume:
(3-6.36)
Thc calculation of tatnl cllarge is not affeckd, therefore, nor is the value of the
flux vector assigned to a uniform ~ituationsince this is also fixed by total charge
corisidcrstiorrs. Wlxy can one not ignore the ambiguity and just accept the
currexrt expression that is naturally associated with %beLagzlrtge funetion?
One reason is that alternative Lsgrange functions can produce: diaerent currents.
This is illustrated by the unit spin, situation.
Tlre seco~ld-orderL ~ r a n g efunetiorr (3-5.19) and the first-order Lagrange
and
In the abrsense of the source M,,, these current expressions are equivzttent. But,
when we use the Lsgrange function ($5.231, there results
Let us apply this to the region betweert tbe %M-o causally sep&raM souree8
.l
JS,
!,
where the fidd is [Eq. (3-3.3)]
=1
with the latter form applicable in soureefrtze: regions, whik (3-5.41) gives
(%$.m,51)
action
jp(,) = &~.(z)y@r~@(z),
(%6*47)
a,jP(z) = +(z)r0i9?(z),
($8.48)
bgekher wi&
which is also s consequence of the field equations. The field in the interval
23-43
and
= C [@p.p(~)itllp.p
P@@
Chargs
266
is [Eq. (3-2.14)]
+pop(~)*itl:prgl,
where, it is recalled,
= (2m d~~)"~u,,,e'~',
UpuqY UP@Q =
1,
u;u:7(rp
= -2pp
+ m,7')
g,,,
+2m~~~,r~r~u,.,.
(3-6.84)
since the mLisymmetrieaX m a t ~ xQ; removes the mateh between the antieommutativity of the sources and the antisymmetry of the kernel ?'G+(% - X').
According1y,
which exhibits the charges p = & l assigned to the sin&@-particlestabs. Although this interpretation is quiLs etear, it may be helm1 Lo give EL mar@formal
discussion, b ~ ~ upm
e d tfre analogue of the rdsttian (%ti.lS), or
Chap. 3
+C
jr = +#~(j ~ g ) q +
+L.Y+(Y:
- T;Y&~,B~
a<@
3+[231~~:&~231f
(3-6.61)
with matrix notation regtored for the third spinor index, are satisfied since
frr^(l/i)a,
-l-
= 0.
(3-6.65)
The current that is defived from the first Lerm of (3-6.61) by inserting (3-6.63) is
3-6
Charge
207
m,, =
-- -2 -l
9~@"'.~'k
3 2m 2
The o;,term is the only such structure that does not involve coordinate dcrivn-.
tives. Ana;Xogous but more elaborate comparisons can be medc between alterneLive descriptions for particles with spins 8, 5, , . . .
The technique of vsriabfe phase transformation. hag been used t o give a
more detailed space-time dewription for the average charge distribution, It
also supplies such information about charge fiuctualions, We shall illustrate this
for spinless particles, confining the discussion to the simplest measure of ffuctuations. Consider, then, the time cycle vacuum amplitude with the sources
K(-I(Z~== K(%)?
(%6,70)
which is
u, gives
This equation, with cp = 0, has been discussed Now let us differentiate once,
before setting u, = 0,with the consequence that
where
j?+,(z) == P4t-r-I
(X)~~~P(+I[X),
(3-6.76)
one of the 4(+, fields is taken a t (P == 0, and beeornes + r e t , ( ~ ) . For the other we
In k r w of the~efield8 we have
which urn the charge prop&y
= 1.
In the circumstances to whieh (3-6.79) refers, which are made explieit by the
apprance af the funetion eret,(g)#
the advan& field vani~hw. And the
contfibution ts 4 of $+fat. caned#, XeaGxlg the re81 fom
Since the lstter is a solution of the homogeneous field egustion, one finds that
A oalculstion of the total eharge fiuetustion can be performed by integrating
over the muree :
(h)
(dzf)~(z)h'+'(z-- z f ) ~ ( z 8=
) C K , , I ~ . (3-6.87)
P@
The latter is the expected total number of particles emitted by the source,
and thia fluctuation formula
((Q - (~))')f
= (N+ f ~
-)f
(3-6.88)
3-7
Mechanical propertisrrr
205)
The wLion for spinless particles is invariant under zc rigid source translation,
(3-7.1)
The following are the proposed eneralizations of tberse expressions when 6XY
becomes an arbitrary function of position, 6 z v ( ( z ,
only the first term on the right would appear for tz rigid translation. An equive
bnQgresentation is
62 == d,(6zp&) - tPya,&X,,
(3-7.7)
where
tPw(z)
= aP4(z)av+(s) gpve
(+(z)) = $"(X).
(3-7 -8)
mrhere the fins1 form involves the use af the field equation. The chsne;e induced
in the action by the source variation of (3-7.4) is computed alternatively ss
270
FSelbs
Chap. 3
and
The compsrisoa of the two versiorls implies t;hat
whicl~can be verified directly. This is the local stahment of a vwtorial conservation law when the right-hand side is zero, whieb is true in soume-frm
~giona,
In the causal circumstance indicated by K = K Z K Z ,B rigid infinitesimd
displacement of K 2 induces
( ~ + l o - )=~ C (O+I
{E)
where
j /o-)~'
+C(o+iIa))Ri[lt i b X * P , j ( { n J / ~ - ) ~ ~(3-7.13)
,
P,(
which can be writbn
(4) = C
Pp@,,,
P(I
(3-7.14)
Similarly, in the time cycle situation the displacement of K(+, and its subsequent
identification with Kt-t = K gives
where ESY(z),
kvhi~his computed from the real retarded fields, is also red.
The distribution and Aow of energy a~ldmomentum is described by t"(z).
It is the enerm-momentum flux vector, or stress tensor. Let us evaflxak it for
the state of a single particle, chosen to be neutral, for simplioity. In the region
betgveen the causally separated sources, the field that is associated with at
particle of mometrtum p is
3-7
Mechanical preperties
211
The first parL of Lhe: field, associated with KZibelf, does not corrlribute:
since the gradient of the symmetrical function &+(X -- z F )is tan antisymmetrical
function. Accordingly, with a slight rearrangement we get
(dz)(dz')iK l (s)(l/ i ) a ' ~ ' (~X ' -- zt)iK2(z')
in1 )KliKtpi~zp(inl
@+I i%
+ 1,)
i l)({% l,]
Io-)~',
f 3-7.28)
252
Fields
Chap. 3
The analogue of (3-1.27) for the time cycle description is the expetation value
is
e = --+Pad
+ ace#. -
(~7.35)
with
pp
vaV+
+
---- f14V+
= tPp,
(3-7.36)
(3-7.37)
%nd
(3-7.38)
We aIm have
a,pv -.. -KaY# J, ( a , ~ @ )3+ ~K ~ ( ~ F ay+p),
+ ~
(3-7.39)
which e m be given oLher forms with the aid of %hefield equstionfs. The only
fieldaependent term on the rlghbhartd side can be w ~ t h ss
n -av@, nultiplied
by the effective source K - 8,KP. In the absence of the veetoriaf soK,,
the two vensiom af the ~ t r e mbnsor coincide,
The Lapange functions (3-5.19,28) for unit spin ps&ielm contain vector
field@,and their derivativw in the cud combination. Using %heh t b r ss the,
3-7
Mechanicat properties
213
(3-7.42)
which gives
6& = 6zv4,d: - GFrGvhd, 62, - m2+p$va, 8%.
= iaV(6xVdG)
- tpva, SzV
(3-7.43)
and
(3-7.44)
Here
1''
= G@"'~
==:
+ m'+@< + grv&
IF@,
(3-7.45)
+ 46:
l = Gp'GPv-4= -m2$'+),.
(3-7.46)
where the t a t term does not eontributc:to the volume inkepakions %hatevaluate
the total energy and momentum emitted by the source,
The alternative Lagrange function (3-5,23) digem from (3-5.19) by the
divergence of a vector [Eqs. (3-5.%, 2511, and Chat remains true of the variations
6C. Such additiond terms da not contribute to &W, But let us note that the
relsfion
(dx)dh[f (%)aFd z v ]
(3-7.59
3-7
Mechani~sfgropeftias
218
Tadeed, the distinction between field and source variations disappears completely were we to adopt
How does this freedom to choose the form of the displacementinduced v%riatiorts
r%Eectthe identification of the stress tensor, defined generally by
which vanish for rigid transfatioxls and rotations. This tensor is a, memure of
the dilation produeed by the displacement, and includes the scalar measure
c3, 62". The same dilation tensor appears in (3-7.M), in consequence of the
symmetry of the stress tensor. Consider, for definiteness, the effect of the additional field variation
on the action expression (3-5.18, 19). Sinee
This is also what is requird in order to maintain consistency with the dimet
25(3
Fisjtds
Chap, 3
which gives
tPp= ap$av
-- +a2(+2)I.
(8-7 '74)
which, in source-free regions, is the result obtained in (s7.9) through the use
of the field equations.
The? question of uniqueness intrudes agrtin in this exrample. The rearrange
ment that eonneets (3-7.72) and (3-7.73) mighf have been haadled digereatly :
This wmld eh&&&
3-7
Mgchaniesl proplttties
,237
This is true for angubr mamntum also. The d d e d term i s p r e ~ e n t din the
form of (3-7.52) by choosing
fh
' = &[B"3p
+hydfl-- 28'ya"+2.
(3-7.81)
Observe that this expression is symmetrical in p and v, but does not have the
antisymmetv that is stakd in (3-7.51)- The annulment of the last krm in
(3-7.50) comes about, instend, through the differential identity (it is (3-7.79)
again)
akarfk" 0.
(3-7,52)
The rejection of stress knsor terms that invalve digerential iderttities is thus
an essentid aspect of the computation rules. But we art? noiv going to sese that
stress tensors are quite analogous to charge flux vectors, Arty current vector
'j can be replaced by jw a,mp",with arbitrary &ntis~;mmt?kicaf
ntF"". Qmmight
ag;ree ts rejeet such additional terms in studying a given Lstgritnp function, but
the existence of different Lagrange funetions for the same system, leading to
current expressions that differ in just this wayl shows that the arbitrariness is
intrinsic.
The arbitrariness in symmetrical stress tensors i s expressed by the possibilify of repfacing tp"(x) \\-ith
where
and the symmetry restriction
nZ~v""IX
assures that
mm"",x' .+ rtzx",l^"
a , a , a , ~ a ~=~ ~o.~
and it will be verified later that the total angular momentum is equalXy unaflected by the additional term. In the simple example provided by (%Ten),
As in the discussion of currents, the exisknee of two sets af firsborder field
equations for spin 2 particles provides s valuable proving vound for unique
RWS questions,
2l8
Fidds
Chap. 3
The tensor variation structure (3-7.40) is compatible with both antisymmetry and symmetry of the tensor. Accordingly, the symmetrical tensor field
of spin 2 particles can be assigned the displacement variation
Two significant derivative combinations of this tensor are given in (3-5.37) and
(3-5.44). For the first, antisymmetrical combination we deduce
and
-aApx
(3-7.92)
(3-7.93)
where
- ~ " ~ a 62,,
~ a ,
(3-7.94)
~ P V=' p '
+ +'G,vv.
(3-7.95)
(3-7.96)
= 1'"'
(3-7.98)
3-7
NIs~hsnlcalpropertim
219
When we confine our aLtention tr, sauree-free regions w considerable simplification. occurs since all vector and scalar field combinations rranigh in these
circumsdances:
and further reduetion c m be acconnplished with the aid of the field equations
ahck" = m2+pp,
Thus,
axc;pAp .= 0.
+ f G""C~'~
= m2$pK+vK
ah(Gh'*+'.)
(3-7.102)
(3-7.103)
gives
from which we obtain
to = --m2@'+,..
In discussing the alternative action expression (3-5.40), we shall p c e e d
directly to source-free eondi.tions by omitting all scalar and vector fields in the
tagrange function, This i s jmtified, despite the varistion that is going to Be
perform&, since all such fields occur in pairs, one faetor of whieh continues to
vanish after the infinitesimal coordinak displacement has been etppfieid. Then,
with
(3-7.106)
&'X
a p + v ~ 4- ~ V + & X ---- d ~ + p . ,
we have
(3-7.107)
e = ~H~'%H,.,- +mzyp6&.
and
"LI;:
-- H X " + ~ ~ ) ~ ~ ~ l p
(3-7.108)
220
Chap, 3
Flsida
+,
The spin Lerm only contains 3, 6%" - d, 6q,. There i s no symmetrical hnsor
%hater-tn be deviged from matrices, other tbwn g&" multiplid by the unit matrix.
A specific mui%ipleof d, &X' appe~m,analogous to the seatar source reBponss!
of (3--7,4). The corr~spondingfield va;ri~tion,~vhicbis &sign4 to leave intact
$(h)?ro+fis
(3-7.1 14)
&\d.(%) == &'(~)ar$(~) iQo"@+(z)&8~,(2),
But the symxrretv of the second de~viativepicks out the matrix comGnalion
and, since the
Farthermore,
[rk,+opv]
=i(ghF
-~Bh~7r),
~
and we get
66: == a,(lEa;'C)
- tPVd,
tvith
(3-7,118)
(3-E .f 19)
(3-7.120)
t = -m4+r0$ - fVro+,
(3-7.121)
3-7
lVIschanicr@
proper?les
Xrct,
where
tg = +qr"+(rp(l/i)a'
+ rp(l/i)ap]q
(3-7.1 36)
is the stress tensor of the simple sginor or Dirae equation for source-free conditions. The additional terms can be exhibikd in the form of Eq. (S7.831, with
3-7
NIechanfoet properties
223
6 d t ( z b-dow(z),
62' = dwA'(z)z*,
(3-7.138)
the aceomp~ayintgdilation
The latter v&&%, of eourmJ when Gu~,(s)ia a constant, dea~ribixlga rigid
rohtion, The implied actian variation is
where
jhhr
= Z~th'
--
%'tkpe
This re8uIt is not independe~tof the divergence ewation for the ntress bngor,
since
ghjx"~
ghaxtkv
-- xpaxthfi + pp - p,
(3-7.145)
Wer ~ h dnow
l confirm that the evaluation of total ~ 1 ~ @ momentum
ar
is anafleeM by the mbitrariness of the @%re88
&mar. AL~e~mpanying
the redefinition
The verifieatioxr that the additional ~ c a x l d - d e r i v i ekrm gives no eontfibu&ionto l&&Pfi'is identical to that of Eq. (3-7.87)--only the oyclic symmetry
prapr,rLy (3-7.85) is invokect. The lafter also impbw this slakanent of antisynnme%qin X and a:
[ m h ~ , a ~ m b . a r ~+
m a ~ . h v l = o8
(3-1.152)
from which fdlows the vanishing of the s u d ~ eintevd for the Xwd term
of (3-.7.151).
Ps&icle s t a w %ha%
are labled by three4imensioztrtl angular momexttum
quisb~tumnurnbr~,ar~therthan by Xinew momentum, have been exhibited for
spirms 0, *, 1. T b i r use i~
suEciently similar to the charge and enerw-momentum
&aewiom that we Eskdl not enbr into detail8 here,
In ontrwk do $he fidd translations and rotations of phygieal inkre&,
$da$iorrr~ would =em da be a d y a device thaL assists in the identification of
merw-momentum fluers. There is, however, a 8ubset of these tran~formations
that plays a more physical role in (the spc;ie,ialcireumtance of
18 is the ~ o u oZ
p hatropic dilations that is characteris& by
These conditions are vemy. re~krietive.We fimd nob the seafar relation
3-7
and
et
Mschanlcai propcrrtiba
226
But even more is obtained by applying the operator a' to (3-7.153), namely
Thus &p(x) is Einnihd La a finear funetion of the ~oordinahg,
The corresponding form of 6xp, apart from infinite~imddran~Xa?;tiong
%ndrotations, is the quadratic=function
Tagether with tran~lationsand rotations, them fransfarmations form s
group of 15 parsmeter&. It has the structure of the rotation group in 4 2
dimensions, in the senw that the homogeneous Larentg group is the rotation
coup in 3 I, dimensions, Perhaps the quiekest way to recogni~ethis is through
the introduction of homogeneous coordinahs,
-+
fixed,
The quadratic form of (3-7.161) also admits refieetiow, including 116 -+ --g&,
which hezs the following effect upon the xp coordintaks:
226
Fietdzii
Chap, 3
3-8
with
C@ = tpyz,,
C"
Magnetic charge
= 1fik(2~%2'
- ~"Yz').
227
(3-7.173)
C ~ v
==
-22hjhpu,
2zhek.
and the existence of conserved total quantities is indicated, for Lhe time cycle
description, by
(3-7.177)
The physical eontetlt of these conservation statements will be discussed in the
corltext of the most familiar massless particle.
3-43 THE ELECTRQMAGI\1ETIC FIELD.
MAGNETIC CHARGE
Although frequent. reference has been made t a the m --+0 limit for unit spin
particles, it is important to give ark independent discussion of the field associated
with the massfess, unit helicity particle-the photon, The slarting point is
Eq, (2-3,45), written as
(3-8.2)
strict account must be tnkezt of the source restriction, which demands that
one shouId not identify the eoefieiexlts of 6Jp(ts). The correct conclusion, is that
they differ by any expression that leads to a vanishing integral in consequence
of the restrickion (3-5.3). The general form of such an expression is
Chap. 3
and therefore
(3-8.6)
The aspect of A,(x) that is governed by the arbiCrary scalar function h(s)
is p k k d out by forming the divergewe of (3-8.6). This gives
a.kV(z) = a2x(z),
(3-8.7)
and the application of the differential operator --aZ to (3-8.6) then provides
A,(%) 3- dpX(z)t
(3-8.9)
since %heddition of ra- gradiend term to a vector does not alkr the: curl of the
vector. Thst the divergeneetess nature of J P is built into the field equations is
8ho emphasizled, for
8,Jr = d,iZ,PV E 0,
(3-8.12)
owing Co the sntisymmetry of F@". The curl construction of F,, is given, another
form in the differentid equa;tions
- +l.
(3-8.15)
(3-8.16)
3-43
Magnetic-charge
229
The Lagrange function is explicitly gauge invariant, and so is the action beclzuse
of the &iflerentialconsemation property of J@. This time we have bemn with a,
&FP.=
6 ~ ~ a h F , . Fk.a,
+ F,,9,
6zk,
(3-8.23)
involve8 only gauge invariant quantities, And the vector field prescription
which w&sbased initially on the tramformation prapedies of the psdient of s
malar function, is maintained under a gauge traasform8tion. The direet evlzlluatisn of the imducd &@&ion
variation is
23Q
Chap. 3
Fields
with
and
The comparison of the two evaluations j v e s
tok
(E X HIr.
(3-8.30)
and similarly
J , F ~ ~ ( ~ Z "-~ g"z')
= fd,tgk)(2z%'
E
a,eE"V,
(3-8.3 f )
-- g"z2)
(3-8.32)
The nstum of the eamfiponding integrd mnsewrttian b w s mogt closely rewmbles t h ~ oft
Thus, witb the weighting factor provided by the enerw density, the werage
value of x2 varies quadratically in time, with unit coefficient of (~'1%.The inkrpretation in Germs of the motioa of the particles &at carry the energy ig
clear: photons move a t the speed of light. The coefficient of z0 and the constant
term supply information abouC the initial. correlation between position and
velocity and tbe initial average value of X'. This view of i(dx)eO is consistent
with its ~ignificaneein terms of the momentum distribution:
3-8
Magnetic charge
23t
The field strengths F,, and the vector potential A, are placed on the same
footing in the following action principle:
where
= -%Fpv(apAp- aVA,)
+ $p%,,
(3-8.38)
avFNv
= P,
dpAp
or
a V ~ p=
v
where
*JP =
p ,
-a,,
- a,A,
= F,,
+ M,,,
a, * p v = *JP,
*MPv,
8,
*JP E
0,
(3-8.39)
(3-8.40)
(3-8.41)
+ *J, *M".
(3-8.43)
On setting Mpv = O and identifying the field strength tensor with the curl of
the vector potential, the previous results are recovered.
If the photon source function Jp(z) has the interpretation of an electric
current, according to the first set of the Maxwetl equations (3-8.40), is *JP($),
as realized in (3-8.41), a magnetic current? The answer is negative. It is consistent with this, but hardly decisive, that the total vdue of the apparent
magnetic charge is zero,
provided MN"as the kind of spatial localizability that attaches to the source
concept. The essential remark is that, through a redefinition of the field strength,
the magnetic current is transformed into an equivalent electric current. Indeed,
the equations (3-8.40, 41) are also given by
which contains the effective electric current already exhibited in (3-8.43). But
this short-lived possibility does raise a fundamental question concerning the
existence of real magnetic charge, distributed and flowing in a manner that,
explicitly or in context, differs from (3-8.41).
To study this question, we go back to the beginning, to the source. Is it
possible to distinguish two fundamentally different kinds of photon sources?
But the two kinds must also be closeIy related, for the structure of the Maxwell
232
Fialds
Chap. 3
JP
*JP
+ *JC^sin p,
---JPsin p + *JP
p,
-+ J P cos p
COS
+ *FPr@inp,
*F@' -+ -Pu
sin p + *F@*cos p.
F*'
FP'
COS p
(3-8,471
3-43
Tha ~Iectromsgneticfield,
Magnetic charge
233
can be extended to include the third unit vector prardlel to p, thereby introducing
the unit dyadie:
Of the two k r m s on the righbhand side, the second vanishes when gP is a photon
mamentum ohying (3-8.582, and the first is indepnden-d of the gpcific choiee
af vecbr &(p) that obeys the restriction (3-8.57). This is in, fact the proof of
covarianee for cmsaf eireurnstanees, But same explanation is caitled for, A elarss
of functions Ghat obey (3-8.57) is given by
where n, is an arbitrary constant vector, ff n, points along the time axis, fa2
example, we h w e the situation of Eq, (3-8.56). That eharae.t*rtrizatioaof f,(p)
is not eavariant; after a Lorentz, transformation is perfomed, n, will have nonvanishing ~patialcompsnent~,although it is still a, timelike vector. 1%ias here
that the arbitrariness of n, enters, for we can replace the time-like vector by ons
with only a kennporab component. It is &rough sueh coupling of the ehoice of n,
234
Chap. 3
Fields
et
- X@')
&(X
- X)).
(~8.65)
The soluGion is not unique. Two alternative solutions that correspond to retarded and advane& bounday conditions are
where
represents the Heaviside step function (the capital of the Greek letter is H,
as the eapitrtl of i5 i s D, in the Chalcidian alphabet). Another choice aligns the
vector n" tvith the third spatid axis, for example, Then
and dternativt: soIutions are
fa(2
- 2')
d ( ~ ' - z")
6(g1
- 2:)
-?(%g
--
za)l
(3-8.69)
Tha ~Is~tromagnstic
field.
3-8
Magnetic charge
235
stances, If one of the f" functions in (3-8.66) were adopted, an, additional causal
elernenL, which i~ arbitrary and physically irrelevan%,would be injected into
the description, ]in contrast, the kind of funetion illustrated in. (3-8.69) is
bmporatly inert;, and its arbitrary aspecls are confined to spatial directions.
Since causa1i"ty is a fundamental widing principle, we reject the use of funetions
such m those in. (3-8.66). Without being commitkd to the specific examples of
(3-8.69, '?Q), we do insist that f"(x - X') have a, spsce-like direction and be
loealiaed in its time-like coordinate excursions.
The desired space-time exLrapoltzLrion is glxven by
Xn verifying that this properly represents the initial carnal situation, we encounter the Fourier transforms
and
whieh are involved in reproducing the last two terms of (3-8.51)' The latkr
are interchanged by the substitution:
JP
(P)
*J,(p),
*JP(p)-+ -$#(p),
(3-8.77)
To test W for this symmetry property it is convenient to introduce four-dimensional momentum xlotatiart :
The effect, on the last term, of the substitution (3-8.771, combined with
+ --p, and p ++ X, is
p,
==
;F;(yz).
which are, inThere are alm two independent kinds d gauge arbitrarin~m~
corporsM in the fie!d exprewiona
+ 4, *h($),
where @@""as been used to form dud tensors. The following identity should
be aobd [it i ~ the
3
eonGenf of Eq. (3*-8.13)J
3-43
Magnetic ehsrga
where the Iwt form involves the inhrehange of the indiees h: and
gauge inv&nisrttfields
V,
237
The fwa
then obey
dvF""(x)
JP($),
3, *Fp"(%)
*J'(x),
(3-8.89)
But only if *Fp"(;e)is the dual of F""(%) can we, procbim these to be the general
f o m of Maxwellk eequa;ttions, with electric and magnetic eurrent~,
A direct proeedurr; for &is purpose is to evaluate the curls of the two vechr
patentials A,($), *R,(s) and compare the results in (3-8,88), Here is another
identity that is valid for any antisymetrieal hnsor Q@,,
(3-8.90)
1xr consequence,
with a similar expression. involving *A,(s), and the use of the differentbl
equations
gives
The necessary dual relationship is exbibikd here. Notice that the gauge invariant field stren&ha are also independent of the arbitrary veetor $,' It is
e~identthat thme kasors obey Maxwell's equations, The converBe is also %me;
the solution of the MaxwelX equations with outgoing wave boundav eonditiom
is just (8-8.93). To verify this the identity (3-8.90) is applied, hthe form
238
Chap. 3
Fields
which. produtees
-a2Fpv= a , ~ , --
(3-8.96)
The dwired ~olutioaiis that stakd, with its dual, in Eq. (8-8.93).
Apast from the characteristic freedom of ga;uge tr~nsforrnalions,the fm
vector potentiebb ean be exhibikd in kms of the field slren@hs. Fir&, let ua
sbmrve that
(3-8.97)
The conmquenf v~nishingof a,(%) is exploited to derive from Eq. (3-8.m) tba&
which also uses the diWFerential equation obeyed by f"(z - zf). Thus,
3-9
Xn 8rriving at the last expressions the following property of the dual is used:
which ~ltlsoimplies that
The latdtjlr passea the action propedy. But one must etppreciak the context,
describing the independent field variables. In (3-9.Q, for emmpte, the fieMs
A,, F,, are subject to independen%variation, while the a;ymbol *A, stands far
the functionat of the field s6renGt.l. Gensor stakd in (3-8.103),
(3-9.8)
Chap, 3
Fields
mf of M%xwellt~
(3-s.xo>
The ~pgymme%ry
in-vot~edlin e q l o y i q either A, or *A, as i~dependentfields is
overcome with yet a fhird aetion expresgi~n:
= 6,
~P(P)K*(P)
=0
and
K"d
= 0,
which also uses the positiveness of -f@(p)f,(p) that expreases the choice of np
and both sets of Manuell's equations have been derived in s symmetrical way
from the i4~;ftian
expres~ion(3-9.12).
By this time the bypodhetical alert reder of li~tiessddeatian, hencefadh
aefanycnicdly known as Harold, can no longer reatrain h
exchawe ensue^.
3-9
243
H.: You showed in the previous section that the apparent magnetic ~fZ&rg@
given in (3-8.41) could be transformed aws~y. It was intirnahd that a different
kind of mapetic current would be forthcoming. Yet the action pdnciple of
($9.6) and the Geld equations (3-9.9) etre identieral in form to (3-8.57) and
(3-8.391, with
and indeed
How then can you claim that true magnetic charge is rrow k i n g disc
S.: Mistake me no$, goad Sagredo, er, Haratd. The fundion, (3-9.a) does
differ-in context-from the source function of (3-8.41), far it lateks that depw
of facalizability which is characteristic of sources. Consider, for example, the
choice of $,(s - z') with only the spatial companenf
whieh need not be zero, That is in contrast with *he null value of (s8.441,
which ~vasbmed on the sptstial loedizttbility of "dd,,fz),
Had we umd the add
fC^
function of (3-8."i"), the explicit form of *Kfo3(z)would be digerenl, bud xlat
the value of the surbce i h g r a 1 that produces the totd magnetic charge. Ths,
it is through the special properties of the class off functions that we make the
transition from mere semblance to the redity of magnetic charge. At the setme
time this poses a fundamental problem since the detailed description would seem
to depend upon the arbitrary choice of the f function, for whieh there is no
physical basis. Surmounting that formidable difficulty is the task to which we
now rtddress ourselves,
Let us introduce into (3-8-81), which is the diEerential statement of the dependelice of W on the source functions, those expressions for bJp(z)and 6 *JP(%)
242
Chap, 3
Fields
up(%)
= ~,(~z~(z)J@(x))
- Jv(~)a,azP(z)
=
-a,[
~x~(z)J~ ( szV(z)
z ) JP(Z)]
(g9,
26)
and
6 "J@(z)= -a,f&rc""(s)*JP(.)
- 6zV(.) *JP(%)].
(3-9.27)
The conservation requirement8 (3-8.82) are identically satisfied, This insertion gives
forms. If electric and magnetk currents are proportional with a wziversd constant, the f p term vft~fi~hes,
as it should since tbis ia a rotated vemion of pure
electric charge; when electric currents are causally separabd from magnetic:
eurrents the fB term vanishes, weording to the restriction s n the etas8 of fC"
functions that confines it to space-like vectors connecting points in space-like
relistion.
It is the situation of electric and magnetic charge coexisting with different
space-time distributions that poses the problem of nonphysical P dependence.
To make this very explicit, supposef""(z - z') is chosen a-s in (3--8.701, a spatid
veetar of fixed direction with its sawort, its nonvani~hingdomain, confined to a
fine of that; direction. Those points in the two source distributions that can be
conmcted by this line canlfibub to 6W, When the direction of the line is varied
continuously, bW and W itself also vary, continuously, thereby denying to W
any physical meaning. Is Lhis the death h e l i of mslgnetic ehargt?? No. Them
is a subt;le possibility concealed bere, I t depends upon the precise fact that not
W but expfiWjis the physically significant quantity. If, in altering the direction
of f p continuously, W were indeed to change, but change discontinuously-by
multipIes of 2~-the exponential would remain unaltered and the mathennatied
arbitrariness of p should be wri%houtphysical consequence. This is impossible,
of course, when, as is assumed above, the sources are eantimuously distribrtkd
objects, Instead, they must h w e a panuJar atmcture, gving values of the p
inbgral that differ by finite amounts according a&rs the f@ line does or does not
penetrate the kernels of that structure. And, since the magnitude of the integral
it3 ~ l s measured
o
by the product of electric and magnetic charge, this combination
cannot be arbitrav but must be re~trictedto certain discrete vdues. Theae are?
remarksfile conclusions-charge is eomplebly locdized and q u a n t i ~ din magnitude, The sweeping nature of such inferences should b-tt, emphssi~ed. We are
encountering =strictions on the structure of photon sources that are required
for the consi~~tency
of a theory of electric and magnetic charges. Sourees are
introduced as idealiaations of realistic physical mechanisms, idealizsllions thaL
dispense with individual characteristics but respct all general laws. Xn uncovering fundamental restrictions on sources, we are revealing general laws af nature,
Sueh was the argument when the divergenceless nature of the vector pftoton
soufee, demanded by the null photon mass, was interpreted as the rtssertion of s
general eonsewation law, that of electric charge,
A realization of electric and magnetic currents in hrms of the motion of
point charges is given by
(instead of "8, syrnbola such as g are aho used hut we wish to emphasize? the
symmetry between electric and msgncsLic quantilies). The eau~almotion of the
points is conveyed by the restrictions
since the point % p ( & ) is infinitely remote from zp a t the teminalis of the integration, The evident identification of the e, and *e, as charges attached Lcl the
individual moving points is consistent with the evaluation of the total charges,
ars in
do, dsP(s)6(1: -- ~ ( 8 ) )=
e,,
(3-9.33)
where the inkgation sweeps the whole four-dimensional domain with dcr, dzp(s)
acting abs volume element.
We cttnnot sirnply insert theae expressions into W(J V ) ,however, The
latter was devised for continuously distributed sources and should not be applied
to s collection of point charges without reexamination of the physieab sipificance
244
Chap. 3
Fields
+ * J ~ ( x*A.(%
)
*U
- 3FpV(z>
(apAv(~* X ) - avAp(z * X ) ) + ~ F p v ( ~ ) F pfv X)],
(~
W ( X )= / ( ~ x ) v ( x ) A , ( x f X )
(3-9.34)
where the appearance of f X signifies the procedure of equal averaging for expressions containing + X p and -Xp. This action continues to be stationary for
field variations about the solutions of the Mamell equations:
* + *Jp(x)*A,,(z * X)].
W(X) = & / ( ~ X ) [ J ~ ( X ) A . X( X
)
(3-9.36)
Wo(X)= #(c:
+ *e.')/dr ds'
D+ (%.(S)
- zo(d) * X).
(3-939)
The mathematical existence problem which the X device is designed to overcome is concentrated in Wo(X). I n the neighborhood of 8 - 8'
0, D+ would
be singular without the addition of the space-like K to its argument. This
cos px
h= nsturesl ulpper frequency limits if the motion of the particle: ia without di*
(S9.42). To
csntixluity, and the limit X' -+ 0 can, be introduced directly i ~ t Q
discuss
w,(X) = Re W@(&)
Doe8 wa(X) have a pfiysical sipificance"2t does nod. This quantity is woGi~kdwith a single poinl; eh8rgf: or partide. 8inse the pa&icles that cornprim
a source have prescribed motions they are being idealized as very m m i v ~
parlicles, which are uninflueneed by the effe~tsthey praduee. The desc~ption
of their indi~duafmechanical propertie8 lsgically precedes the discussion of
inler~ctions, The nature of this description can be infemed from the rwultp,
concerning stress h n ~ o rand
s their vdues in single-particle states: P"= 2 h , p p p ' .
As we have? expldnd, &is is a simgXificatian valid in. the intt3rior of a barn
whesc? fhe variation of momentum rand the associated firtih spatial exkensian
can km neglwbd. To reinshh these, we identify pp with the p a d i e d of a @me
function
~zndintroduce s variable weight funetion,
tp"(2) = p(z)aYrp(~)d"p(s);
the m-
rest~ction,
dpcparp $ m2 = O
aFa. N o b that
and the foeal rneehanicd consemation laws are satisfied by the consemation of
padicle A l u ,
ar(~d)"v)Q.
(3-9.48)
This interpretation &o supplies the value sf the intepal :
(3-9.49)
Within this picture of prescribed motion it is cornistent to take
6(2
-- S(&)) = 0,
(3-9.51)
and
(3-9.52)
I n transferring these results to the connection bekvveerz action and stress Lensor,
one must not forget the meaning of 6s,(s). It arose m a generalization, of the
~ g i displtaeements
d
given ta soureefri, which were inbnded to simulak $he &hplacement of a referexlee aurface and are therefore in the opposite mnse. Thus,
when tranglating h t o the motion of point particles a minus sign must be s&ed:
(S9.55)
3-9
Chrrgs quantixation,
Mass normatization
247
and supgies the action expression for a single particle, labeled a, performing a
preserihd motion,
(3-9.58)
The phenomenological orientation of source theory has the folfotving eorollary, Physical parameters identified under restrickd physical circumstances do
not change their meaning when a wider elsss of phenomena is eonsidered. The
mass parameter m, is determined from the response of the particle to weak,
slowly varying, prescribed forces as in beam deflection experiments. When
eleclrornagxtetic interactions among several particles are considered, this parameter i s not assigned a different value. It has already been fixed, normalized, by
experiment. Thug the single-particle term (3-9,44) mast not be added to
(3-9.58), thereby changing the value of m,. There is no question here of assign?
ing some fraction of the total mass to an electromagnetic origin, What is a t
issue is the consistency bet\veen the various levels of dyrlamicat description
through kvkich one passes in the course of the evolution of the theory. The
prescribed forces of the most elementary level become assigned to the motion
of particles a t the next stage, but in neither one is there any reference to individual particle structure and the phenoxnenoEagicaf parameter m, must be common
to both. The eonclusiorl is that the real zt*,(X) krms, which contribute neither
to the vacuum persistence probability nor to the couplings among sources, must
h struck out. Here, then, is the action to be assaciakd rr-ith a point charge
realization of photon sources:
W = Lim [W(X) -h-90
w,(X)].
Q
x f,(z.(s)
- I&(s~)
f X)
dZbr'S"
---
ds"
-C
4W,(X)
a
The antisymmetrical product of two vector displacernexlts defines a two-dimensional element of area,
4xf dz; - 6~:: dx: = dcz",
(3-9.61)
and the antisymmetrical product of three displacementa produces a thmedimensional volume element, or the equivalent directed sudace eXement for
the coordinates zz - z,;
d *C:' dzb, = d ~ t b .
(3-9.62)
Chap, 3
4 d@:'[e,F,,(z.
-C
@a
tab
- *e,es)
f X)
S 'e, *F,,(z.
de:afN(x.
- ~b
f X)]
f X)
- C 6wa(X)
a
&"+d
where n is an integer. In order to ensure thabonphysieal etennmb do not intervene during the limiting process ?P--+ 0, we demand that this hold for aImosG
d l X@. The scale of p iis fixed by the diEerenLiaI equation (3-8-62), or the equhalent integral statement
(3-9 "65)
referring Lo 8ny surface that encloses the origin. The diserekne8s required by
(3-9,64) implies thak the suppod offp on any such sadace i~ ca~lfin~d.
ko a f i ~ k
number of points. And, in virtue of the qnnnnetry properLy (3-8,72),
fhat number must be an even integer, 2v. We may visualize $his number of
filaments drawn out from the o ~ $ nin a way $had assims to e a ~ hfilament its
image in the origin, Let the contribution to the sudam integral (3-9.65) th& i s
~~aciaLt?d
with an individual paint a,a! .- 1, , , . , 2 ~ be
, designs64 r, so that
The basic 8ilua;dian far (3-9.M) is that @ ( X ) , for example, incfudes a single point
a,while @(-X)
contains no support; point of p, Then
and the addition of sueh, sxpresaiong repmsents any other possibility. Xn. par-
G h s r ~ quentization.
s
3-9
Mass normalixat4on
249
or, making explicit that the paints of support oeeur in pairs with equa! values
of r, and n,,
4 dfl~'[eaFL".(~.
f X) -k *e, *F,,(x.
f X))
&W.(&)
a
This might seem to pose anotiier problem, hosvever, Although WE: retain the
symbol $W, it is no longer the change of a quantity W and the question of
uniqueness arises. Consider rz corltinuous deformation of the trajectories that
finally retur~isthem to the initial eonfigun-ttion, thereby defining a surface
exlelotjing s three-dimensiond voIume. As the covariant generafizatiorr of the
three-dimensional reletioxl
(3-9.75)
and similarly
4 do" *F,,
--
(3-9.76)
250
Chap, 3
Fields
of (3-9.77) record the amounts of electric and magnetic charge within the various
volumes, Here the basic situation occurs when particle b lies within the volume
@@(X), for example, but is outside of a,(-X).
The associated contribution to
d W is *(e, *eb - *e,eb), a multiple of 27r according to (3-9.69). This aammation
of the single-vafuedness of expfz'Wfwas inevitable; it was only of some interest
to see how the charge quantization condition brought it about.
The charge quantization demanded by magnetic charge provides a most
satisfying explanation for one of the more striking empirical regularities in
nature. Uespik the widest variation in ather proper-t,ies possessed by partieles,
the magnitude of the unit of pure eleetric charge is universaf. It is measured by
the fine structure constant
a == e2/4a 1=: 1/137.036.
(3-9.75)
and
This is very large indeed, being the equivalent of the electrie charge 2(f 37)e.
However, one might think, if only for a moment, that this great asyntnniely
could be apparent since there is the freedom to redefine; dl eleetrie and magnetic
charges by the rotation of (3-8.47) :
2 (ez + *ez)(ei
Now consider the following invariant slatemend. For all known p~;t"ticles,
(ei f * e , 2 ) / 4 ~is small compared to unity. Comparisoll of the inequality (3-9.83)
with the charge qusntization condition (3-9.71) then shows that the integers
n a b must all be zero. The corresponding points with coordinates e,, *e, are
confined to a single line, which thus acquires an sbsoluk significance. It is
conventional to identify that line with the axis of pure e1eet;rie charge. The
complete l-eduction of the line to equally spaced points demands the existence
of a. second class of particles for which (e.2 *e:)/4r is large compsred to unity.
Among such particles there is no necessity for an. absolute charge line although,
3-9
261
if the integers of the charge quantixation condition assume only moderate values,
the charge points will clusbr near a, line, which is the conventional axis of pure
magnetic eharge..
It is remarkable that we have been led to the existence of two types of
charged particles that are characterieed internally by relatively weak and reIadively strong forces, for this corresponds to the empirical distinction between
leptons snd hadrons, respectively. Certainly hadrons-mesons and baryonsare? not magnetically charged particles, nor do their interactions possess a
strength as great as (3-9.80). Rather, we view them as magrretica1l.y neutral.
eornposites of particles that carry both eleet~icand magnetic charges, with the
observed strong interactions of hadrons emergir~gas residuals of the considerably
stronger magnetic forees, lvhich thus far have successfully prevented the experimental recognitioxr of free magnetic charge, I t is essential far this explanation
that a magnetically neutral composite appear as an ordinary electrical particle,
If we have a group of particles wit11 charges e,, "c, such that
E n.~
(3-9.85)
From our various tzssunrpdions, which are grounded in the symmetry. hlwws
electric and magnetic charge, we have inferred $hat the charge units on 8 d u d
char@ padicle are the same fracLion, 1/N, of the uniks of pure e l e c t ~ cand
magnetic c h a ~ e . Among the possibilities, 2, 3, . . . , which value hw nature
elected for the integer N ?
But fir& we musk digess $0 discuss the relation b t s v a n the tati is ties of
eompwitc3:petrticlm and their constituents. One approach uses the spin-statiskics
wnneetiam. A compasitc?b m e d of an odd number of particles with inbger 4
spin (F.D. 8%&ti~tiesf
h= a regultant spin anguIar momentum thatt is also
inbger "f- &. This eamposite particle o b y s F.D. sta%istics,I f there are an even
n u m b r of eonstituent particle8 with inbger 3 spin, the compsite p
inbgral spin and is a B.E. particle, It is as though a F.D, (B.E.) particle @&fie@
a, nninus (plus) sip and these ;9iws are mdtiplied to give the statistics of s
composite stm~ture. This is more than s mnemonic, far the $us and d n u s
signs identify the dgebmie propedies of the individual m u m 8 &at are m m pfict-d to produce the eEeetive sourge of the composite sysbm. Now, as we have
mentioned, there are two varieties of hdrong; mesons, which are B.E. parkiclm,
and baryons, w h i ~ h&reXi".f). particles, If bLh dyps of hadrons are to km con~ t m c t e dm mametically neutral eornposites of dual ehargd pa&ieles, the lattctr
eannot all be BB. particles, The simplest msumptioa is that they are all F.D.
partides; ~n wen number of such constituentss p d w m a B.E. particle, an d d
number builds a F.D. particle.
Cm the dud charged particles exhibit only one 8trength of magnetic charg~?
N o b %hatboth sign8 of the magaetic charge, linked to sign ehange~in e l m t ~ e
charge, will occur. This is the antiparticle concept, with both charge^ involved
in order to maintain the structure of the two @etaof Maxwell's equaliom, which.
ha-ve the field stren@h bnsor in common. If the only values of magnetic ebarge
are fl/N) "e and -(l/N) "e, they must be combined to produce a neutral
composite, and sueh p a h of F.D, d u d charged padicles are B.E. particles;
bavons cannot be manufactured in this way. Eence there must be at lemt two
digerent eharge magnitudes, According to the magnetic analowe of the elmtrie
lattice cowtruelion (%9.90), the mametie charges on dual charged padicles
with the same electrie charge must differ by a multiple of *e, the unit of pure
magnetic charge. It would seem to be a rertsonable charachrieation of dual
charged pa~iclesto describe them as carving charges that artit smaller in magnitude than the uniLs of pure charge. If that is granted, just two values of magnetic eharge are admitted, With a conventiond sign choice, they are --(X/N) *e
and [CN - l )f N ] *e. The possibje values of electric cha~grjwe analogoug:
-(l/N)e and [ ( N ---- 1)/Nje. Either electric charge can. be assigned to either
efioiee of magnetic charge, giving four dual charge combinations, although them
may be duplieatioxls of these assignmenb.
In ezddiLion to neutralizing a magnetic charge by its negative, which builds
a maon, we can now balance .the mapetic charge f ( N - IL)/RT] *.a against N - l
units of the magnetic eharge -( l / N ) *e. This is a composite of N F,D, particle@,
and N = 2,3, . . must be odd if the resuit is fo be a F.D. baryon. The simples-1;
possibility, which we adopt, i s N -. 3, Thus, bsryons are v i w d ~ZSe o x n p ~ ~ i h
of three entiLies that bear the magnetic charges, in. *e units, af 3, ---*,
-9, We
learn, incidexllally, from *e = 3 *ea, that
-*
refer to duplicates of the same particle, or to BiRerend particles with s common value of
magnetic charge. To this we can only offer the observation that, withoul
reference to antiparticles, the magnetic charge average over all distinct dual
charged particles will not be zero in the first possibility, but does vanish in the
second one wbercr eharge -g has twiee the multiplicity of charge 8. We accept
the situation of greater synrmetq, and extend it to electric eharge as well. Thus,
whether we speak of electric charge in units of e or msgneLic & a r e in unik
of "e, there are three options with values g, -9, -4, I t is natural to regard
these nine possibilities as differ& slates of a fundzbrnerztal dual eharged particle.
To emphasize its basic dyadic eharwter in regard to charge, this pa&icIe is
called the dyon,
Although the hypothetical picture of magnetic charge ills the bmis of hadroaie
behavior is still quite incomplete, we haye alredy far outrun. our ability to test
it, particularly rsince et quantitat;ive phenommologiical analysis of the properties
ol hadrons is not yet before us3. We must turn away from these heady s p e d a -
P54
Fields
Chap, 3
tiom and bgin the study of ordinaq elwtgeal pa&ides in dynsmied eontexte.
Hawever, Harold finds ffimwIf compelle-d h eomment.
The comemed nature of the photan eleed~esource JP(%) sets the patbrn for
any realkation of ~ u e hsource8 by an deetfie current metar wociated with
~pecifictype of particle. The electric currenf-a that we h&vealredy comidered
for v a ~ o u sspin choices fail ta meet this standard since they are conserved only
oubido soume rM0n.s. Let us rope& that discussion for spinlem pahicfes, using
%heslightly diRerent procedure thiat is b
on (;fie ae%ionexpremion
with
3-1 0
Electromagnatie interactions-
s ~ u r models
~e
265
where
d&fF(rz:- X') = &(s - X')
defines a, no$ unfamiliar class of functions. When the support of f@(z- 2') is
restricted to space-like intervals, the subtracted term in (3-10-8) vani~hesad
any time for which the sources are esusaly inoperative, To keep uniformity of
treatment between jCI(z) and JP(z),we shall relate the canwrved vector, now
designated J:a,,, (X), to an arbitrary vectorial. function J p ( z )by
The vector potential Ap(z) must multiply the total current, in the action
expression. That can be rearranged to give
in which
A',(z) = A,(z) - a,
(dzk>fp(z
- xj) A ,(x",
(3-10.12)
which ha8 no apparent pbyslcaf sigrrifiesnee here. Note that the construction. of
A:(%) from A,(s) is a gauge transformation, such that the new vector potential
i a charaekriaed by
(3-10.14)
(dzt)f"(z - z f )A:(.') = 0.
This is a unique characterization, for, if the general gauge transformation
26B
Fields
Chap. 3
(a, - ieqA,(z))+(z)
--t
(5-10.~)
The field, equations deduaed from the sLsLionay wtion principle by varying
@ and rf, are, re~petiv?rfy,
where the gauge covariant wmbination stays intact since the sign rever~alOf
the derivative on partial integatian is matehed by the antisymmetry of the
charge matrix g. In performing the variation of A: we must not violate the
gauge restriction on the veetar pokntial,
where ?(S) is arbitrary as far ss the action principle is concerned, But tha
divergence of this equation gives
a,dP(x)
+-a,ji"(z)
(3-10.27)
= ?(X),
+ @-ie@Acz)
e-ieqA(zj
c35P f d ?
(3- 10B)
where
and A,(rt;) is the veefor potential in an a r b i t r a ~gauge. 'This transformation
does two things. I t replaces A: in b: by
~ : ( z )4- apA(z) = A,(%),
(3- 10.31)
which is the inverse of the gauge transformation (3-10.12), and the transfer81
of the uncornpensakd phase factor to the saurecs replaces them by
,n(x)=eie~b'z)~(2),
K , A ( ~ ) = ~ ~ ~ ~ (~ z( +) ) K (3-10.32)
With the introduction of the arbitrary veetor ptential A,(z), we return ta the
uut; of Jta,,(z), The additional label will be omilted, however, for one can
understand from the cantext %.hether JP($) is sn arbitrary vector, since the
vector potential is limit4 to a particular gauge, sr is a conserved vector, since
the vector potential admits gauge transformations. The new sction expression is
The gauge invariance of the Lagrange function is now matched by that of all
+-
A(.)
and
KA
eiegh(l)
4X
),
induces
+ A(z)
(z),
"4- &(X)
K: (z) -t eiegh'"' K A, )
(3-10.34)
(3-10.35)
While the charged partide field equations that are implied by the action
(3-10.33) continue to be given by (3-10.23) with the sources K", K:, the eleetromagnetic field equation. presents ab different aspect. In contrast with the action
of Eq, (3-10,XS), &A, is arbiLrae md the charged padiele sources are furrctionds of the vector pakntial. The implieation of the latter fiaet is indictzkd by
( d z ) + ( z ) ~ ~ (=
z ) (dz)4 (z)iepxA( X ) &A(2)
(3-10.36)
3-1 0
EIectromagnetic interactiona-
source models
259
dse-i'pn - nE"/pn,
pn 3C5 0.
(3-10.44)
[/
where A.,(x) combines the field associated with J $ and the initial photons with
that having analogous reference to the final photons. Xn view of the causzal.
arrangement of sources, wherever A,(x) is of interest in (3-10.45), it is a solution
of the source-free Maxwell equations or, in momentum space,
which shows the equivdence, for the purpose of evaluating (3-10.471, of the
time-like jp function with the space-Iike
/ (dz')4(x')lCA
(X') =
(dx')+(X') exp
[- iep/
( d z ) f p(X
- X') A, (X)] K ( z r ) ,
where x' serves as the reference point at which charge eq disappears in the source
and emerges on the particle of interest.
The members of the class of fC"functions given in (3-10.49) differ only in
the choice of the tirne-like unit vector nC",which represents the motion of the
280
Fields
Chap. 3
charge in the source model. When fhe coordinate system identifies with the
time axis, f (p) has only spatial components that are independent of p@,
and
where
There is one choice off@that avoids the reference to an external unit veetor by
devising the latter from relevant physical. paramekm. It requires an exknsion.
of the structure sf f @
fa include akebraic funetions of derivatives that act upon
the source funetion K(2). We indicate this repfacement in (3-10.50) and
describe its meaning by writing
where the Imt form is the analowe of (%10.44), one that is equiv81ent for the
ea;tleulation of phofan processes*
The discussion of spinless particlm is pa&ieufarly. simple. A rsysbnr without(
intrinsic angulm momentum ewn anEy exhibit scalar properties in ita rmt frame.
In the electromagnetic conkxd this permits manopole momeat-ch
forbids multiple moments, More generftfly, a particle of spin s, in its mg&frame,
@an possess multipole momenk ta the rnrtximunl order 28, That is, a spin 3
pa&iele can have arbitrary dipale moments; a, particle of u ~ spin
t can have
arbitrary dipole and quadrupale msmenk; and so forlh. A sufi~ientlygeneral
eument expre~ionfor spin $ is
This way of writing the eoeBeient in the term thwt hras the form d , d ' antieipab~
the _identification of g as the wrclmsgnetie ratio, the? magnetic moment in tbe
unit &e/2m relative do the spin angular momentum fEq. (1-2.4)). Th8t be-
comes clearer on. wing the i;dlen.tity (3-6.67), applicable in source-free regions, to
remite (3-1 0.57) ara
No such progerty has yet been detected, however. Since .the second te-m of
the cument is identictitlly divergenceless, we still have [a, factor of e is i m h d
campared to Eq. (3-6-48)]
(3-X 0.62)
a,jr(z) = ~ . ( z ) ~ ~(g).
ieq~
The currend (f.E-10.57)is ineorporai;ted in the fot10~ngslcti~nexprwion,
an8logous do (3- 10,Is),
g(+,A,) = -&Fp'F,,
- ++ro[rB(--$a, - epA,)
4-
m]$
(3-10.63)
2#2
Chap. 3
fields
Thua, the identity of the two eoncepls is impo~edby imisfing that the aetion b
invariant under the unifid gaugephae tr&nsfomationwith
The replawment of desvatives on chargebeafing field8 by gauge e~vmiant
derivatives &coontpli~hes.t;hia for the whole poup of gauge drawfomna%ions,
Pvhich ia Ablian in struetm. And the possibility of adding independently gaup
hvas~bntbrms, M in (3-10,63), conveys the a r b i t r q aapeab of %heIcinennatieaI
t dt?fini$ian. It is generally believed &a;t them is something padieularly
and a;a%uralabout the ctjtectromametie coupling produed by wing ody
the gauge covariant aubgtitution, and there is gmfh in thb. But i-d magt not be
forgotbn that ttlkrnative de~criptionse G ~ for
t the same spin vdue, and by
haowing a common procedure we rsr~veat diEeren%electronrametic prope&iw.
Thus, the third-rank apinor dmariplion of @pin+,b m d on the L a g ~ a g efun*
tion ($5.73) with gaum cova~antderi
,&vm the eument of Eq. (3-6.6X),
&p& from the f m b r of e, and the
orrdi% g value, m eora~nedin
(W.68,69), is 8. If the very striking nem+quality, +g S 1, that is abwpvd
for the electron and $he muon has atay single moral, it ia the apmiall relevance
of &hesimple Dh&cspinor equer,Cionfor the description of them p&ielw.
= M,,,
DvGpY m2+@- aFpiep.+, = JP,
(3-10.73)
Lf we are ixrtereskd in the intrinsic electramagnetie properties of the parkiele, and not; those induced by the electromagnetic field, it, S U E t~
G ~implify
~~
(3-10.n) with the aid of the uncoupled particle field equations:
ths lrtst of which, i8 an innpodant but not independent statement, This giva
fm regions is eonvqed by
&)f A,(ar+'ieq+v)
The identity
- ft.
---.
a -t- b)iF,p(sbPz"eq+')
+ (blm2)ah~,.(aE^dieg4v~1.
(ap+%ieq$') = (dp#'iep$" )+ ap(+'ieq+')
(3-10.77)
(3-10.78)
haws also that the field ~Lrengthderivative in (3-10.77) should ba symmetriad in the indices X and v.
Chap, 3
Fields
where the d y a d i ~"ErEis symmetrlrzed, and we have dao picked out the term^ that
deseribe the propagating particle in a crausd arrangement. WiLh Lke coupling
of the scalar potential A' to the charge fe serving as s reminder of the nomaligation, the linear coupling of the spin vector to the magnetic field identiifies
the g value :
g== 1 . - a + b ,
(3-10.81)
while the quadratic spin term @;iveg the quadrupole momexrt Q, in the unit,
(&e)/m2,as
& = 2b.
(S10.82)
The idividual results obtained for g values when oaly the gauge covariant
derivative is used (s = fr, g = 2, 8; s
1, g = I), are given unifQrmXgby the
geneml nnulti~pinarLagrenge function (s5.78). The current %ha%
the latker
implies in source and field-free @paceis
=I.
*C@,,
Extended soureas.
Soft photons
265
++
EXTENDED SOURCES,
SOFT PHOTONS
Complementary to the pt-inciple of space-time uniformity is a principle of uniformity for phenomena that differ only in the values of energy-momentum that
are engaged. The source concept was inkoduced as an ideali~ationof collisi~ns
in. which precisely the right balantte of enerw-momentum or, invrtriantly expressed, msss is transferred to create a specific particle, But the sarne laws of
physics are operative when less mass, or more mass, is transferred. Long ago,
in Section 2-43, we used an extrapolation to quasi-statie source distribulions,
which are incapable of emitting particles, in order ta connect the properties of
photons with the Coulomb-Amp&ri%nlaws of charge and current interactions.
Perhaps in our recent preoccupation with the very familiar equations of Mmwell, we may have forgotkn the initial logical ba&s far that contact. And now,
through our concern tt-ilh the electric currents that are assoeiztted with the
rtpwation of charged particle sources, we are moving in the opposib direction.
The physical situation is quik simple. The creation of a ebarged pvticXe generally involves the transfer af that charge from other particles having different
states of motion. Accelerhed charges rsdiate. Hence, unless precise eontrol is
exercised over the energy-momentum balance, the charged particle has t% nanzem
probability of being accompanied by photons. If we were to take too narrow ail
view of the source concept and decline to extend it to this mrxltipztrticle emission
act;, we would divorce the dynamical, significance of ehacrge from its kinematiesl
aspects.
26Q
Fields
Chap, 3
for one csn identify $be probe source 6J" with JT. Since the field A,(4) is to be
evslunkd for $,(E) = Q, it is given by
apart from an irrelevant gauge term. The process in which m are interested
involves the @%us&coupling of three sources: J"; KK1,
snd &. Tbe emission
source K2 is u s 4 to inject into the system the mornmtunn P p that, is redized
m two particles,
P" = F-+-k@,
(3- l l.7)
where
Thus
This sowce is aperating in the extended sense, and we shall urn the designation
'extended source' to distinguish its mode of action from that of KIPwhich detects
the partide by absorbing mass m. A souree uLili~edin that way, performing
only its initial mission, is a 'simple source.' Now, the current of Eq. (3-11.6) is a
quadratic functional af the particle source and therefore ~ontainsa porkion
fiz(atthat is bilinear in K 1 and K 2 . ~ h igives
i
a factor on the righehand side
of (3-1 1.5) thaL a'Iredy has; the required three sources, All ather te
different processes than the one af intertlst, whkk is displayed m
The omission of %nother f erm involving K l ( f "ieq&z (t") expresses the caustll
) related to iits source by
tzrrangement, The field + 2 ( ~ i~
or, in momentum space,
The fact that P2 m 2 # O [Eq. (3-11.9)) means that the field 4z(z) has no
propagation efiaraete~sdics,and is localised in the neighborhood of the source
Kz(z), Thus the field cba(z) will have no overlrtp with a sufieiently remote
detee-tdion. source Kl(x), which is the assumed causal situation. The term
'virtual particle' is used to extrapolate ordinary particle concepts to such
sihations where the energy-momentum balance is not suitable to the creEtlion
of a 'real' particle. With our new terminology we can characterize the content
268
Chap. 3
Fields
of (3-11.11, 12) by saying that the extended source may emit a virtual particle
which quickly is transformed or decays into a real particle and a (real) photon,
or it may emit both final particles in one act, although the photon originates
a t a different point than the particle.
The precise meaning of these phrases is conveyed, on comparing (3-11.1 1, 12)
with (3-11.3), by
where the first derivative refers to the X' coordinates. An equivalent momentum
version, which also introduces (3-11.14), is
left.
k * ~ z ~ ( W n ( P ) = 0,
which is valid for p2
(3-11.17)
in which we have also introduced the form (3-10.44) for j,(k). The interpretation is clear. From the viewpoint of the soft photon, the charge eq has made an
instantaneous transition from uniform motion with velocity n, to uniform
motion with velocity p,/m. This is expressed by the photon emission source
Notice how the two contributions, one associated with the particle source, the
other with the particle, are fitted together in an equivalent photon source. This
is an illustration of the self-consistency that is demanded of the source concept.
The source is introduced as an idealization of realistic dynamical processes.
3-? 1
Extsndd raurcssr.
Soft photons
2@
The dynsmieal theory that ia erected on this foundation must, under appropriate
m~t~cfiong,
validate iLs ~tartingp i n t , Thus we learn, not s u ~ ~ s i w l %ha$
y,
the aimpie photon murce dewnption becomes wficable to w realistie syr~tenn
when there i8 ne@igible re~.(tionassociated d t h the rsmimian or abmrption
procem*
We should a h recognize the phpical significance of the cavarianf f , funetion &ven in (3-10.56), which we now \$?ribas
where the Ifitkr version refers to soft photons. The eEwtive phof;on source
vanishes; the ehotrge hw not changed vejoeity and doe8 not r d i a h , This i~
the most natural Csf 80ur~emdels, in which the ernittd particle dekrminm
ing
the velocity of the charge in, the murce and thereby supp
the a~comptlny
radiation. That mppre~ionis not limiM to mft photons, however, I f we imrf
the unapproximated version of ifp(k, P) in (3-11.16), it becomes (kg = 0):
The prabability amplitude for the emission of the two particltts labX1ed kX, pq
requires, beyond (3-11.23), the additional factors (dwk)'I2 and (dw,)li2, together
with the explicit slection of charge h e ttnd the photon pofari%a&ionX- The?
latter is produced by scalar multiplication with the polarisstion vector et:, and
There is anothm point that ean be illustrakd by the eBective wurce
(3-11 1.16). Equiv~lentta a pmicle source mde1 charaeterizr4 by fH(k) is the
and this becomes c$& = O in the appropriate coordinate frame. The significant
obwmwtion i8 that, an mdtiplying (3-1 l . 16) by one of t h w plarizr&tionvmbrs,
270
Chap, 3
Fields
in which 6K(z) -4 K l ( x ) and +(z) is related to the aowee Kz(s)by the field
eqtltttiarrs
(a,
- ieA,(z))@(zf
= +&(z).
(3-1I.W)
Since both p;a&iele sources already appear in (&X1.29), the clws of proeeams wile
wish -t;o aelect are exhibited by
where the notation emphasizes the dependence of .titre parti~fefield +z(z) upon
the veetar pokntial tf?(l)that represents the emitted photons in mlation ts
their deteetion souree JVfE
Let us fir& recovw the known ~inglephoton, result; in this new way. For
this we need the part of @i1(x) that is linear in the vector potential. The field
equation (3- l 11.3 I) retains just that amaunt of informa"cion when; it is simplifie$ to
).
Tfie first term on the right represents thtl rdiationless enzission. of the particle,
and the geeond one reproduces (3-11.1 l). The nth hrna of the power series
expansion of @ $ ( X ) in A'(t) describes %photon emission processes. If we agree
ta consider only soft photons, all such processes can be combined in. t l ~C O R R P ~ L G ~
farnub which, as we would now expect, is equivale-nt to w photon sotlrcjr?
dmeriptian.
Extended saurces,
3-f1
Soft phatone
273
(dzt)A$(z,X') exp
(dglfF(z"
--)A,([) Kz(z"
),
(3-1 X ,352
[-- (a - i
e p (2))
~
+ m2]A:(z, X')
= 6(2
-- z').
(3-11.36)
(3-1 1.37)
= exp
in which the integration path is a straight fine canneeding x and x@,,as paramet;~zedby
This vector pokntial has two ather ~ignificantpropertiers, fn regions; far from
the eleetrornagnetie sour-ce J v( t),
and, general1y,
(Z
- z')PA;(z)
= 0.
272
FIelzls
Chap. 3
(-aa
+ m )A+ (z,
2
A"
X')
= 6(2 - zt)
+ 2epAL
(l/i)apb+(z
(3)
-- z') +
(3- 11-45)
But &+(z - z'), being an invariant funetion, depends only upon (z- z
'
)
,
and its gradient is a multiple of the veetor (z -- .
'
)
z
We learn that h$(%, z
')
has no term linear in AL.
is
More can be mid; if the field strengths %retreahd as harnogenmus,
appmpriab to goft photons, of negligible momenta. Then
- zf),
The anll;ujtar momentum structure of the linear field streneh, trsrm assures it;s
commutstivity with a2; it also eommutea with the qusdnttic combination of
coordinate8:
[tF@r(g,a,- %,a,), ~ z ~ F :=~ Pz ~F] ~ ~ S ~
=
(3-1 1.50)
since
= F,.F""~,
(3-1 13 1 )
is an. antisymmetrical function of p and v. AI1 this, and the rotationd invariance
of 6(2), shows that the differential equation. (3-11.48) can be sinnplifid Lo
We shall not stop now to solve the above equation, I t suffices to know that
~ T (--z X') is an even function of field strengths, for this means that the field
depndence of bhe latter funetion can be neglected relative to its partner in
(3-1 1-37), since, earnpared La veetor potentids, field strengths contain an
additions1 photon momentum factor.
Introducing the8e soft photon simplifications, wrearrive ttt
The straight lim inkgral that occurs here bgins a t %bandmoves, in 8 dimtion
ned by the vector (z - g':")@%ward
,
an effeeCiv~!lyinfinikly dista;nd point,
3-1 f
273
~incethe photon emieion processes are localized near the extended s o m e K%,
And, if the eoupling betwwn the padicle sources iis to be appreciable, the g*
me$~caldisplacement (;e - x')" muat coincide cIowly in direction with that of
&hemomentum vee-t;or of the exchanged parkiele. Accordi~gly,
+ fmv2,
p = m*.,
/v
(3-1Z'$7)
6x1writing
where dQ is the solid an&e within, which the phohn moves, we get
This photon eaerw intepal doe8 not exist matbematica;lly, divergng both at
the upper and lower limits. But clearly there are physieal r e s t ~ e t i o nrtf~ bath
en&, When one reaches energies sL which the photon eeaws t;o be soft, the
evaluation ( S l 1 . W ) no longer applies, and a lower limit is
'by the minimum
delectable photon enerw of the experimentd amangement. Onee upon a time,
the m~themsticaldivergence at zero energy was taken literally, and this soft
photon phenomenon boame known as the 'infrared catastrophe.' As s. eatastrophe, it nzhs rathw low on the scde. Consider %he&Berence that is impfied
in the vdue of ( N ) , depending upon whether the softmt photon considered has
s wavelength of visible light, --10-' cm, or has a wavelength comparable to the
nominal radius of the universe,
cm. Since v2 < 1, that difference is
+e
where
[?(l/.;)la
+ dtbz(z>= a%(%)
The comparison with the exchange of one particle and one photon under noninteraction conditions,
supplies. the effective two-padiele source that represents the emission. of the
3-1 1
276
)v
eff.
= [rpeq
(3-1 1.67)
-rP
m
eff.
and, on writing
rk = r P
- l ] eqn2( P )
+ m - ( ~ +pm),
we get
P
eff.
(3-1 1.68)
=-(YP+~)
-YP
eqs2( P )
(3-1 1.70)
But this is to be used in the context of Eq. (3-11.65) where the field t,bl(x)
represents particles far from their detection source, and the Dirac differential
operator in (3-11.71) produces the required null result. Alternatively, we can
use the momentum form (3-11.70) and recall that
(X)
where
E irl:,.,(2m
dup)" 2 ~ - ' p z U *P ~ ~ P
(3-1 1.72)
P"!?
U~.,YO(Y~
+ m ) = 0.
(3-1 1.73)
A: ( E ) =
(dwk)"
&X,
*
(3- l 1.74)
kX
with
since both factors are useful in producing a simplification of (3-11.67).
A relevant algebraic property is
r"(m - r P ) = ?'(m - r p )
= 2p~' flvik,
- YpYk
+ [ ( r p+ m)rP+ P],
(3-11.76)
where both terms in the square bracket can be omitted for our purposes. Simi-
Fields
-o@'ik;,rk = (rprk
since"k
+ k@)rk
[kfi~kj,
s = ~
(3-1 1.77)
QO,and
1%is evideat %h&,in the Emit of sof* photsns, there is an ef-Teedivephobn source
which is identicsl with the one encounkred for zero spin. This i s ta, be expc?cLr?d,
for every spin value. The suecwsive multipole moment efXecb involve ixlerettsing
powers of the photon momentum, and all become negligible compsred to the
charge aeeeleration rsdiabtion for suficientfy soft, photons, But the particular
choice of f@(k;) fhst removes the acceleration radiation no longer suppresses
photon emission completely, since the spin-dependent effects of magnetie dipole
moment8 remain in (3-1 1.79), and no ~pmializationof g can, annul bofh term^.
We have illustrated the e x k n d d source concept in the eonbxt of emission.
1%can all be repeated w i m the exkaded; source acb h absorb ab padicle and a,
photon. But the= inverse procet3ses are also d a t e d by the TCP operation,
concerning which nothing bm been said reeenfly. The eEw&of the Eutllide&n
basd coordinste transformstion
on sources and fields is given by
5,
q(rr?)=rsv(z),
$(Zc)==Y&#(z)*
3-1 2
fntsrsctian skefston,
277
But the compIc?& statement of the W P opration ineludes the revergal of sll
factam. The anticommutativity of the sources and fields msociaM with the
spin +,F", D, particle provides the addition& minus siw needed fo produce the
anticipratd invariance of the action under the TCP transfammation.
The TCP operation inverts the causal order, and inbrchanges emission
and abgarption processes. On applying the transformation. to (3- l l .M),
one
quickly ve~fiesthat the whole ~truetureis mainlaind, and it is therefore only
neceBsary to change the eeusal labels. The same rem~rkapplies to the momerrturn version (3-1 l .67), of course-, except that we folIo~vthe practice of r a v e ~ i n g
the sips of all momenta when absorption proeews are being deseribd, which
the transformation automatieaIIy supplies. What has been shown in the spin 4
framework is of general validity,
3-?2 tMTERACTLtOll5 SKELETON,
SCA-ERINQ
CROSS SECTIONS
In view of the nonlinearity of thia system, the construction of the fields in Cerm
of the sources wil be given by doubly infinite poxyermfie5, That is atss the
nature of the action tvhrtn the fields are eliminated snd W is expremed as a f u n e
tionsl of the sources, The successive terms af this series, W,,, with n particle
and v phohn saurces, represent increasingly colnplieated physical praces~es
whieh are thus mbo~t~ledged
to occur, but will not be given &heirfinal dweription
rztt t h i ~first level of dynamical evolution. That is the meaning of an inbraetion
~keteton. At later s&agesof the dynamiml development, proceses already
present in skeletal form are provided with more complete descriptions, and
mme additional processes are recognized. I n thiss ~mtion,we propose to carry.
%hedigeu~sionof the simplest terms in the interaction skeleton to the p i n t of
displaying their observational implications.
There are t~\-oasymmetrical ways t;o eliminate the fields. In the fimt, one
introduces the formal solution of the parti~lefield equation:
( d s f ) @ $ ( z ,z f ) q A(X'),
[r(--G-
(z) )
(3- 521.2)
where
j&ne,(z)
jP(z) -
(dz"lf""(Z
- x")aj"(~~)
(3-12.5)
+ jP(z')],
(S12.7)
(@@P
-- i k ~ s ( k ) ) @ ' ~ D +(g*.
( k ) - fk(k)ik.)
- ikJ,(k) -- f,(k)ik. -- Skdh(klfk(k))D+(k)
(3-12.81
31-12
tntersction skeleton.
279
gauge condition
fp(k)~:fk),.
= 0.
(3- 12.9)
or in the equivalent form that uses the nonconserved currents and ~ $ ( z- z)',.
The nonlinear field equation for J/ that is derivd from this tzetion is that of
(3-12,1), ~ t A,h replwed by (3-12.4) or (3-12.7).
Wbieh of them asymmetric hrms it is mos&convenient to consider depends
upan the process of intermt. Suppose, for example, that no photons arts in
evidence. Then one ean e t J P = 0 in (3-12.10) and marnine the nonlinear
prope&ies of the pa&iele fiftld, If the causal situation is such that interaetion~
aeew far from the padicle emission, and dekction saurees, which is part of the
amangemat of s s e ~ t b r i n geqefiment, the p Wmi in jtoa8.-causally tied
to the sourcecan be i ~ o r e d .The inkrsetiot-t.tern of (3-12.10) contains few
particle field8 and therctby ett least four sowee factors, When we consider
prmesws that involve only four sources, as in particleparticle scattering, fhe
stationary aspecCs of the =Lion principle permit us to identi.fy $ with the field
which are a t Xeast cubic in the source, what is thereby lacking in W fim no let35
tf-trtn six powers of the source since firstorder effects of the field change are:
annulled Lhrough the stationaw action prope&y, Thus we have ideIldifieib
where
jr(,) = IC(~)~~?@~.~PJ.(Z)
and JI(1;) is the field given in (3-12.11). Analogous r e u l b hold for any other
spin value. With spinless p&rticlr~?s,
far example,
Chag 3
Fidhs
and
(dz')A+(%
--
X')
K(%').
(3-12.16)
The refe~eneeto the vector pokntid in the p&icle sowee has been dropped,
d t h the undersLanding that (3-12.17) vvill be applied ta pmes3ses in which
~ ( 2is
) umd as a ~imple
pafiicle source, all partielephohn interaatiom occunlng
far from any of the sowees, To eixhlbit the individud W%,,we must expand in
power series the A@dependence of G$(%, S') and extract the term containing v
vmtor pobntials. For this p u r p e it is useful ta r e h k the Grwn% function
equatiorr of (S12.2)
(--ria+ m)@$(z,z8)= 6(2 - g') e P ~ ~ ( z ) ~ $ ( z , z '(3-12.19)
),
which is e o n v e ~ dCX7 an i n % v ~ equfttion
I
by the formal mlution
G:(.,
2')
C+(% - z')
$-
(dt)C+(z
- 8 e q r(6)~ ~ $ (S').
t,
(3-12.20)
3-9 2
281
+ of (3-12.11) :
(dz)(d~')#(~}r
' e g r ~(s)G+ (z
(3-az.24)
- z')eq~A(2')
X " ) ~ Q Y A (Z")#(Z"}.
-- A + ( ~ ~ +( ~A pA) -- e Z A 2 ) ] - ' ~ +
+ ~ + ( e q ( p A+ A p ) -- e2A2)&+
+ h + ( e p ( ~+~Ap)
--
e 2 ~ Z ) ~ + ( e P ( pAAp )
--
+- .
e2A2)4+
(3-1 2.28)
3.
which means th& the careful orctering of factors can be ignord if the vector
potential has a vanishing four-dimensional divergence, as is the sjitualion Eor
(Slt2.18). Bokntirtls having this property am said to be in. the Lorcsmtz gauge.
since the vaAous power8 of the emimion and detection sozlrces Bewe to direetlig
identify initid and finat stake^,
The vaeuum prob8blliLy ampfitude is detemined by the action
(3-12.35)
3-1 2
where
tF""(Sl,Sz)=Wf(Sr+Sz)-Wf(Sx)-Mcl"(S2).
(3-1Z.37)
Invariance8 of the action imply selection mles for the tramidion probabili$ies,
Ri@d translations or constant p h a e transfomations of all souxces, far example,
which do not change Wy(Sx,S2), must leave the righthand side of (3/-12.38)
unaltered. The emission and absorption s o m e prducts are multiplied by
reeipro~alphase canstrtnts, re1a;ted to momentum and charge in these examples.
The individual transitiolrz. probabifitim must vanish if the phase constants do
not cancel, expressing the neeessaw eonservatisn of momentum or eharge in the
interaction process. The fwtor that imposes momentum conservation,
will emerge fmm a space-time integration over the inksactition ~ g i o n .We make
this csxplicit by writing
thereby defining the elements of the transition matrix. T. Thc inbgral is no&a
four-dimensional delta function since the integration domain is not infinite, T o
sppreciate this we must recall that the precise specification of individual momexlta used here is an idealization that holds well ~vithina partiele barn, but
faits near the bounda~es. Where the initial and final beams overlap to &vct
esusail definition to the inkeraction region, (3-12.40) is applieabfe, and limiting
the integration to that finib volume is sn ~pproxirnateway of recopizing %he
realities of the situation, It is probability that is physically significant, and we
are actually concerned with
(dz) exp [i
(nh - tl,)p.z]
(dz)(dzf) exp [i
(nb
- n.)p.(z
- X')]
The 5 integration ean now be identified as a delta function, and the X intepal
memures the Wtal inbraetion volume V, within the uaeedainfiea attached to
the bounday liayers. The proporttionality of the transition probabili-ty to the
volume of the four-dimensional inhraction region ixldicabs that the impadant
quantity is the cwffieient of propontionality, the transition prab~bilityper unit
faw-dimensional volume, or, per unit time in a unit three-dimensional volume.
This ratio is
which suppliw the physical interpretation of the transition matrix.
Led us b & n the svcific discussions of skeletal inhracdiona with the scatkring af spinless pa;rtictes, as describd by (3-12.13, 15, f 6). The field #(x) is
requird in the inderaction region, which is eausaHy intermdiah bcl.t~sreenthe
e ~ s s i o nssoumt? K2(z) and the dekction source K l ( x ) . The b t a l field is the
@uperpositionof pads related to fLfie8e wurees,
*(X)
= cBl(x)
where
+ +zCzZt
(dz')~'-'(g
(3-1 2 . 4 )
- zf)KE(z'),
(3-12.45)
and the particular forms of A+($ -- z') disclose the caussl situation, The prscessj
we %reeoneerned with invdves the action of two emis~ion,sources and two
absowtion sourca. Thus, when (3-1 2.13) is considered, with the c
we m u ~ retain
t
only tham eontribufiom having the required overall characteristic,, &s conveyed by the causaX indices. Those term8 are
(D-
= 2-"'(1,
i).
(3-12.49)
3-1 2
2815
they obey
pP
p;
= 'p;eqr,
pp,.
= ~'@p.*
where
+Pg (z)=
( d o p )' l 2(pqei~'
is the field associated with the specific particle Iabeled pqr, which enters the inbr&&ion re$on after its creation by the source K%,,. Sinnilarly, +P,(z)* is the
h f d of the particle labeled pq which, after leaving the interaction region, is
annihilated by the detection source K:,.
The charge structure of the various partid currents that compo~e(3-12.46)
is of importance. In j$,(z), far example, the charge frtctar assoeiakd bvitlh two
incident particles of charges qhand g" 'is
and the necessary equality of p' and p" implies thet no eharge accumulates in
the interaction region. These am different ways of satisfying charge eanstirvation in the scattering process. The seeond term of Eq. (3-12.47) does not contribute to the sewttedng of particles with like ch~rgeand m examine that
process first.
where
L explicitly symmetried in p%, p: and d s a has the mquird pl, p: csymmetq &meovemjl momentum conservation implie8 th&t
W&&
3-1 2
287
measure: of their relative ffux is suggested by the requirement that it mwt vanissfr
when the vectors are: proportional, and the beams run with the same velocity.
This definition, is
2 2 11%
F
~ ( g a s b ) -t
(3-1 2.64)
which does produce a real positive quantity since the Aux vectors are time-Xike.
If we write this out in terms of particle density so and particle velocity v = #/so,
the flux definition hcomes
which introduces the masses of the particles, Other f o m ~can be used, particul~rlyone involving the total mass M, the invarisnG measure of %het o t d
moxnenturn,
= -(pn-tpb) = m: -i- Zpapa,
(3-12.69)
namely
F = dw. d&b2[MZ-- (m, rnb)']li'[M' -- (m. -- rnb)'Izi2. (3-12.70)
where the m o d vemian refers to the rest frsme of P, in urbich P' == M, The
magaitude of the relstive momentum
Pa
-B&
(3- 2 2.72)
is @ven by
1
- [M' --2M
(m.
(3-12.73)
h carqixlig out the enerm inbsa;tioxl thst selectis this value one must d k
where dft is the element of wlid angle for the relative momenturn. The immdiale
resuit is
1
1
3Sa2 M a P f 2 - (ma
X-.--""-.
mb)'~'''[~~
da, (3-12.75)
which rdtrees the &fierentid aspeet to the angles that 8pesif.y %hedirectian of
her
ee
m We nob that the same squareroot kinematics1 mf a c b r ~aceur in the final 8tah ixlbpation (3-12.75) and in, %h(?:
incident flux
(S12.70). Thew relatively camplicaM factors will e~ncelfor a purely elstic
seatbring proew where initial and find perticks are the same.
The tramition xnat~xelement (3-12.62) provides a eimple application of
$he cross section. definitiorr, giving dirwtly
In the latter form, B is the defiection anlgle, and the full equivalence of the anglers
3-1 2
289
The
W - B apremes the indistinguishability of the B. E. padicles,
rduction is wrformed in the rest frame by noting that each of the four partiefe
energiea equids +M and this gives, for example,
8 and
Of psrfcieular interest are the vePy high and very low energy limits:
Note that at sn~a;llseatkrictg angles the latter reduees $0 the Rutherford differential cross metion for the scwtkring of dislin~ishablepwfticles,
where mch current contributes two eqllia;l terms Lo a given process, correspondiag
to the symmetries expressed by p1 C-' p:, q -+ -q and p a t ) p;, q -+ --p. The
implied %ransitionmatrix element is
290
Fields
Chap. 3
Notice the simple connection between the matrix elements (3-12.62) and
(3-12.83); they are interchanged by either of the substitutions
M2
- 2m2
1
M"
- 4m2 sin2 (Bf2) I -
4m2
11/12
cos 8
COS
3-1 2
Interaction skeleton.
where
+ A$(z),
291
(3-12.90)
/
A$(z) = i (dz')f'~'+'(z - X')J2.(zf),
/
(3-12.91)
and then retaining those terms that have one photon and one particle emission
source along with one photon and one particle detection source. They are
w12
/(dz) (dzl)rl( z ) [ e q 2 ~( x~ ) A + ( ~
- zf)eq2~~2(z')
- /(dx)4l(x)2e2~i(x)A2(z)02(z),
(3-12.92)
in which we have adopted the simplification that. expresses the use of the Lorents
gauge for the vector potential. Let us recall that
Of the two terms that do not refer to polarization vectors, one vanishes because
the source is divergenceless and the other, a gradient in coordinate space, can
be removed by a gauge transformation. Thus, it is in a special class of gauges
that we write
where
A$x(x) = (dud
and they are Lorentz gauges, since
112
ikz
ek~e
w h i ~ haltf~oU S ~ Sthe fact that the differential operators pp rand p" a& dirmtly
upon momentum eigenfunctions* The transitian matrix element is obfaimd aa
I r , ~ , l p , , ) = (dur, d ~ pdut,
, do,,)"22e24:~,~,.e~,~,,
(3-12.98)
?E=
P I -4-
kl
pz 4-
h2,
(3-12.101)
and therfsfare
-plkl
while
-plkz = - p 2 k z =
(3-112.102)
(3-12.103)
Invariant expremions for the particle energies in the center of mass system, the
rest frame of P,are
We should also nok tbft crossing symmetries exhibited by the transition matrix
element (3-X2,98,99). Sinee the sign sf q is irrelevant, there is invariance,
under the substilution
spcseificalfy of h,,
for which the equdity of klplwith k2pz, and of k,pz with Iz2pf,is decisive.
Concerning tbe photons, the use of linear polarizations with Feal potari~atioa
wetars implies the transfomation
3-1 2
;193
end the transition matrix element sfiwld be invariant under the interchangtt.
This induces the exchange of p and v in Vp,, which indeed 8how6 the requir&
invariance.
The tensor br,, also has the fallowing imporlant proprlies:
They bring about the neeemary con~rvationof the egwtive soure@ for the
emission of the final photon and the absorption of the incident one:
which sugplie~the unit for a, diRerential seatkring cross swtion, Then, ailnee
2eZ = ~ K O Iwe
, get directly
- m2
M %+ m2 -I- cos B
M2 - m2
(3-12.117)
l + ~ z + r ncas
~B
vvhich u w the center of mms wwthring angle evaluation for plkz. The dif"Terentiwl and totd crass wctions for the extreme mergetic limits are
R+# with
One can wrify directly that the same resuft for the surnm&tioaand average over
polarisations is obtained in this way. Applied to the final pkotons, (3-12.121)
ghes
3-1 2
298
and, then
If we sef
12
O = p, 4- kl = P 2
kg,
which gives
The cross section. vanishes if one polarization vector lie8 in. the scatkring plancl
white the other is perpendicular to the plane, When bath veetars are perpendicular to the scattering plane,
--
M2
m2
+ + cos 6
m2
- m"
.l+
~ ~
cos B
$ m 2
Fields;
Chap. 3
The geometrical facton thst appear here, cos' +B and sin' 48, are familiar ss
probabilities, for unit anwXiar momentum with magnetic quantum number +l
in tl given direetion, that a measurement m d e in a direction at the relative
angie 8 will yidd magnetic quantum numbem +l and -1, re~pectively, 'Shere
ia also a dynamical wei&ting fpbctor that is unity at low energies, M rrr m, and
suppresses helieity changes a t very high energies. The tots1 differential cross
seetion, vvhiGh is independenl of the initial heiicity, is the gum of the partial
erass geetions in (3-12.131) :
==
Mint,
(3-12.1133)
the two partial cross sections are equai, leading to zero average angulsr mamentum in the dimetion of the scatter& photon. This
ie, of eourBe, the
same as the one given in (3-12.130) at which the scattered photon is linearly
polarised.
Other processes involving two particles and two photons are contained in
WzZ. When we SeXect terms with tvva partielet e ~ m i o soureas
n
and two photon
deketion s m e s we are considering pa&icle-a~%iparticle
andhilati~ni n h t;uro
3-7 2
lntsraction sksteton.
2337
photo-; two photon emimion sourees bgether with tmvo pafiicle deteictisa
~ ~ u r e indicah
es
the inverm process, the crestion of a padieleantigartick! pair
through collision of two phobns, The h t k r , for example, is deacl.ibed by
_,
snd iJxgk,iJi;~;
to get
where
+ ( * M ~- m2)"'
cos 81,
(3-12. f 37)
where 8 is the ande htween particle and photon, relative mamenta. The location of the: tbrmhold for the reaetian rat M = 2m i8 app8rent in the square
root that gives the mapitude of the cenhr of mass particle momen$um, a11
parfiele and photon ener@esb i n g equal to $M. Using red plariealion vectors,
the matfix elemeat (3-12,135,136) is obtained from (%12.98,99) by the
craasing tr&mformrti.t;iorr
p2 -+ --p$,
klXl -+ -kbA&.
(3-12.138)
Since the final pa&icles naw differ from the initial ones, the ratio of the
kinemadicrtf square root facdam appear8 mpfieilly in the differential cross secs
assipments, first using tke linear
%ion. We shall give it for v a ~ o u polarizst;lom
po1ar;igatiaxzs that are parallel or wrpendieular to the plane of the remtion:
the cross section vanishes when the two polarization vectors are st right angles.
Provided the threshold energy is exceeded by at least the factor 2'j2, M 2 > gmZ,
there is an angle at which the diEerential crass section for parallel polahgations vanishes,
(3-12.141)
The dominant reaction thus shifts horn equal helicifies near the thre~bsldtZ)
opposite helieilies at vew high ener~es,The crossing poixlL aceurg at
3-1 2
and again the predominant helicity relationship of the photons change8 in going
from low to high enerdes, I n the annihila.t;ionaf slow particles, M c=i gm, there
i s no relative angula~momentum for the photons to carry away and equal
bdicities for the oppositely moving photons dominates. A t very high energies
the photons sustain the mrttximum. angular mmentum along their common line
of motion. Nevertheless both: differential cross sections are isotropic, and the
tatal eross sections are
The variation of the cross section with inverse relative s p e d v, when. v f<1,
means only that the rate of the annihilation process per unit volume is pmportional to the product of the beam densities. The eomputzlLion of the tatd eroes
section for reactions in which the final state contains two identical particles,
such as the B. E, phatons in this mnihilation process, needs one note of caution.
I n summing individually over the find tstaLes of both particles, which is here tlke
summation over heficities and the integration over all dirrsctions of motion,
every physically distilnet slate of the Lwo padicles is counted twice. That trap
has been avoided in stating the cross swtions of 63-12.1463).
3-33 SPIN
-& PROCESSES
Let W b @ n with tbe scatkhnf~of rspin particle8 that have like charges, As
withod reference to specific
in the @pia0 discussion, the relevant part of
charge v%lue~,
it3
The q fabl, which is common .t;a df spixrors, It% ben, omi_l;l&. Note t2b0 that
the. id%idorder of the tatally antieom.muta%ivemure@f a e b r ~has h e n r e
amwd wiLbout the inkwention of minus signs:
Even when one is not interested in specific spin values in the inifiai and finaf
states, perhaps the simplest general procedure is to evaluate the higerentiaf
cross sections for the various helicie assignments, That is already suggeskd by
the photon pola;riz;ation eonsidergfions of the preceding metion, urbere the oudcome of poIarization summations and averages required some algebraic reduction
to attain the result that was produced directly By considering the various helicities (or Xinear polarizations). This simplification lvas particularly m a r k d in
high energy photon scattering where the helicity strongly preferred not to
change, The same tendency appears in the present situation, which we may as
well call electron-electron scattering since that is the outstanding realization.
The general construction of the spinor up,warj giwn in (2-6.90) as
where the v, sse v @eigenvectors with eigenvalue 3-1. When they are chosen to
be eigenve~torsof U pllp/ as well, identibing cr with the helicity, we get
where the latter is the high energy limit in which helicity becomes linked to the
f n the cenbr of mass frame where X
aX pa&icle energies equal
eigenvalue of
+M, consider the following high energy evaluations:
= iY5a,
(3-13.1 11)
We see that (~~,,,r~r'u,,,) vanishes if ol = --@a. The helicity does not change
in these products a t high energy because rar' commutes ~vithvs.
Accepting the restriction ol= a2,a: = g;, we find that the product a g
pearing in the first term of (3-13.7) is
Chap, 3
which is antisymmetrical in the indices crz, G; and in g,,g;. But a word of eau$ion about notation is cdled for here. Although we have M-rittenvoz, 8&y, one
must not forget the implicit reference to the direction of the momentum p2
slfong which the spin is projected to give the component a2,m a t we have just
referred to as antisymmetry in g 2 , a; is, properly speaking, antisymmetry in
~
2 and
~ p 2h i . The combinittion (3-13.14) does not vanish when the helicities
crz and cr; are equal,.
We consider fimt the situation of equal, initial helicities. Then precisely
the combination that is evaluated by the identity (3-1 3.14) appears in (3-13.13)
and, owing to the antisymmetry just mentioned; the two terms of ($13.7) are
combined into one, with the factor
The spinm v,, and u,;are in the same relation, but ~vithrespect to the direction
af p, == --pi, wrhieh is rotated by the sngb 8 about the y-axis, for example.
That is expressed by
which, as a combination of matrix elements, is also the determinant of the unimodular rotation matrix. Since $he various factors of 2m and df cancel in the
matrix element, leaving 2eZ,the result i s immediate :
3-7 3
Spin
3 procssses
303
For the situation of opposite initial helicities, the combination Lhat appears
in (3-3-18), apart from a minus sign, is
The two contributions of (3-13.7) are now wsocia-ted with different final states,
which do not interfere in dihrential cross sections, The helicity labels are,
.
respectively, cl = crz, c; = o; = --gz and ox = a; = --c%,a: = ~ 2 AlternaLiveIy expressed, unit angular momentum along the initial direction of motion
can lead Lo either of the magnetic quantum numbers, +l or -1, along Lhe
common direction taken by the scattered particles, Leaving aside the multiplier
of 2, the factors contributed by (8-13.20) in the two situations are
(y:e(l
a1
= G;,
i2)isrvU+) ( v : e ( 1 / 2 ) i @ @ ~
v+) = cosZ*8,
= Cr2:
(v-e
(1/2)illa
%+)
(3-13.21)
* (1/2)iBo.
(U-e
@v+) = sin2 48,
Then the spinors g,, essentially reduce to the rot= $1 eigenveetorsv.. Helicity
t
degcription, the referral of $1 spins fo a common.
ceases to be the m o ~ u&ul
direction in space taking its place. fnded, with that chaiee,
and the individual scattering praeosmss t ~ k pl&ce
e
withouL ehsnge of mrwnetic
quantum numhr. In this low m e r limit,
~
spin and orbitd motion are dynsmially i n d e ~ n d e n t in
, contrmt ta very high enerw conditions where they
are tightly linkd. When the initial, and therefore the final, mapetie quantum
numbers are equal, both te
af (SL3.7) cantfibuk, with rever~edsign; for
ogpo~iteinitiaX mapetie quantum numbers one or the other of the two t e r w is
effective depending upon the msignment of oppalsite mgnetic quarrtum numbem
in the find state. The spin-averagd differential cross section, is, themfore,
a@,
the differential cross section (3-13.19) applies to either spin state af zero magnetic quanhm number.
Not mueh more effort is required to obtain scattering Gross sections for
arbitrary energy, using the helicity classification of states. Now helicity changes
in scattering do occur, as exhibikd in. the general evaluation of
0 -+ 0, yes:
1P
0 --+& l , yes:
X
+ 1,
both-:
1 "4 - 1 , both:
For the last two processes, the constant --4 is the contribution asso~ialkedwidh
helicity changes, Adding these terms and supplying the appropriate factor gives
which interpolates between the high and low energy famh (3-13-24) and
(3-13.26) respectively. While resembling the zero spin result in (3-12.77), it
differs in detail, except tat high energies.
Chap, 3
Fletds
where the two equal contributions Ghat refer La a spcifie pair of appsidely
charged parLicles h w e already bwn ~0lle~Cf3d.The transition matrix eLemenf
that is defined by %hecoefficient of
4-
(U:;.;
-97
Q r *
@Pl@lP)(%p..g'I7
(p1
+~
0
~ f i ~ p ; . ;-P)
$ 1 ~
The: crossing re1a;tion btween this matrix element; and the one for like charge
scaLkring again follow fmm the unification of e ~ m i o and
n abmrption prw
in %hefield $(X), as convey4 by the formal substitutions
When ~lppliedts (3- .13.7), the h~ubstitutions
produee (3-13.34), with the additional &nu@ sign coming from the reamange
men& rrecaBaq to realize (3-13.33), the stand&rd mul%iplicationorder of the
sources [i' is omitted here] :
since the two soufees asaociatd with charge ---g must revem Iheir relative
pasition.
3-1 3
Spin
processes
3W
Ia order ta treat both terms of (3-13.34) in the same way, we use the relation
of (2-6,134),
up* -'I = i o ~ ~ u ~ . ~ ~
(3-13.38)
which gives
where the now mahhing charge labels have been omitted. The effect of the
additional r5 factor is ilfustrated by
The first statement depends upon the opposite motion of the two particles in
Lhe cenkr of mass frtzme, There is an irreconcilable conflict btween the numerical fector, demanding the equality of the helicities --@l and o:, and the
spinor product, which requims equal magnetic quantum numbers and therefore
opposite values of the helicities --g2 and er;. The situation is snalogous to that
for a spin 1 padiele, where the time component of the vector field rranishea in
the rest frame.
As we recogmizr?.from (3-13,.i10), high energy collisions with the same initial
helicities (a2=. g;) are dominaLed by the first term of (3-13.34), which diRers
from its andague in like particle seateering only in sign; therefore [(3-13-29)]
The high energy evaluation of (3-13.39), in the psincipal circumstsnee a2=. ---c~.;,
= I-bj, is
6 1
according to (3-13.B) and the nuU property noted in (3-13.40). The firs* b r m
of (3-23.34) contributes only to the process ~viiitbcpl = crz, and, reedling that
(PI-4- p : ) g = -AfZ,
(3-13.43)
Chap. 3
R-eget
The average of (3-13.41) and (3-13.44), the diEerential cross mction for unp o l a ~ m dpadicles, is
This, Loo, is identical with the high energy spin O differential erom section. As
in (3- 12.88), the simple f &etor cos4 +B relates high energy electron-positron
scatkring to high energy electron-electron scattering,
Owing to the disparity of the denorninatom in. (3-13.34) a t low en.er@e~,
only the first brm is significant for d4 2%and
= cos'
4s
Spin 4 procaraes
3-1 3
%H
The relation between. the two high enerm differential erass sectiom h- bcomc?
quite transparent.
When the substitutions (3-13.48) are introduced in (3-13.30), the general
dectromi-electron. differential cross section for unpolarisd barns, the result i s
the eorre8pondin.g electron-positron differential eross section (the kinematics1
factor 1/M2does not take part in these transformations, of course):
Although written in a slightly different way Chat exhibits the dominant, low
energy and high energy behaviar, the fimt, square bracket, term is idenbiezlbl with
the spin 0 diEerentia1 eros9 section for scatfclering of opposite charges. The latkr
was only stated implicitly in (3-12.87), one half of which is the entry in the
square bracket of (3-13.51). The two additional terms in this equation %re
relatively negligible at both low and high energies, but they can be of quantitative significance at intermediate energies,
In order ta illustralcr the scatter;ing of dikrent kinds tlE charged partictes,
we shall also consider the interatetion Between at gpin O partisle and a spin g
particle, The appropriate electric current vector is the sum of tbow associated
with the two types of particles 8xf.d the interaction term in
is
where all primed yuankities refer to the spin O par$icle. This will also extend to
the masses, m and mf9auf the spin i$ and spin O particle, respectively, A, simplification. can be introduced with the aid of the h t a l momentum,
==
PI
4-~ "$32l3-P$,
(3-13.54)
for
?(pH
df-5) = 2 W -
snd
(3- 13,55)
-- my0)up2.2,
(3- 13-56)
in ~vf-rlchthe last form refers La the center of mass frame. The introduction of
310
Fields
lvhere the kinematics of the situation, state that the eleetron enerw and momentum magnitude, whieb remain unalkred by the collision in the eenbr of
m m frame, are
At high energies, where the: individual p%&iclemass- are negXi@bXe, the electron
he:licit;y is maintained in scatlefing,
and
dc
n2 cos2 4,
--------.
din
sin4 48
>> m, m':
When, a t low energies, the electron. spin is referred to a fix& direetion no change
in magnetic quantum number aecurs on sealeering, and
The general result, s u m m d over final spins and averaged over the initial
spins (the latter process is unneeessaq here) is
It is interesting to consider the two limiting siGuatisns in which the mass of one
particle becomes v e q l&rge,not only compared to the other m s g , but Ls the
total eaergy of the second particle, ff m' iis that l a r e m%,it is m m eonvenient to introduce
df
- m'
-4
0,
(3- 13.63)
where the electron mew in the center of rnam fiame i e indistinguiehable from
the enerw in the coordinate system where the h e ~ v ypadicle remaiiols zr;t re&.
The analogous limit in whieh it is the spin .5; particle Chat has become v e v heavy
has the same form as (3-X3.M), kvithout the trigonometric faetor in the numerator :
spin 0:
&a
fbo now superfZuous prime on m has been. omitted. The two diflerential cross
sections have been identified with the moving particle, the very massive one
acting only a%a stationary charge. The possibility of applying t.t, gource dt3;scrip
tion to this circumstance will be developed in the next section.
But first let us examine some processes involving photons and spin -& pa&ides, as eonLained in the expregsion (3-12.a) for Wa2. Electron-positron annihilation into ttwa photons is described by
(dz)(dz')~.,
(z)roeqr~
I ( X ) G+(z - zr)eq"/A1(z')J.z (X'),
-,iq
(3-13.66)
gives
The B. E, symmetry in
kih; is evident, and the F. D. antisymmetry in
p ~ @ z pp;@;
,
--p can be verified (recall that rr = --~~r,r@)
with the aid of the
kixlennatical relations
JVe shall find it more coxrvenient to tvrite the dyrramical factor of (s13.67) as
In the center of mass flrarne the energies of all electrons and photons equal
*M,and
--2p2kl = *Jf[dl
-2p2k; = $fii[Jf
(3-13.70)
(e2/m)@'v%,.[c- e*,
i.
(?Or
= irs@)
\%-herethe unneeded causal Xabels have been dropped. Only terms with an even
number of Ys fachrs survive here, and tl-e have used the fact that e X ef must
be directed along k. The equality of the helieities staks that the two magnetic
quantum numbers are opposite, in the antisymmetrical way impIied by Lhe
faetor G'. Accordingly, only the singlet state of zero htef spin can, undergo twophoton annihilation, for slo\vly moving particles, The corresponding aero sngutar
momentum state of the two photons, a linear combination of the two equai
helicity states, is identified by the perpendicular polarization vectors of the
%\Wphotons, When we recall that
the differential cross section per unit solid angle for a given pair of perpendicularly $arized photuns, with the particles in the singlet state, is oMained as
(uz/m (l/v). To compute the total annihilation cross section for unpolarized
particles we must supply the follo~vingadditional factors: 2, for the number of
polarizations available to one photon, the other polarizatisn b i n $ fixed by the
requirement of perpendicularity; 4, the Aatistical -tveigbf of the singlet sts;te;
2w, the total solid angle aecessibte to either photon \vi%houtduplication of the
-.;
-,
f L is convenient to use the photon msmentum k as the spill reference directionthe z-axis. Then the arthogonal particle spirkors describing magnetic quantum
numbers of &4 in the p direction can be ~vrittenas
v, = v+ cos fiB
+ v-
sill $4,
U?.. = --D:
sill 18
+ u*
cos i 8 .
(3-13.79)
tvl~iclrhave the effect of rnisirkg and Iotvering particle magnetic quantum num-
314
Chap. 3
Fields
bers by unity;
*(g.
+ i.,)a-
= v+,
v:*(@,
- i@@,
= V;,
-+
=,
v:+(@,
+ iq,) = v*,
(3-1 3.81)
all other combinations being zero. Thus wet can ewily work out the values af
(3-1 3.78) for any choice of photon helicities. With X = --Xt = +l, for example,
tvs get
t,r;Izile null results are obtained far equal helicities, h = X L & l . As in the
high energy annihilation photons carry only the mmimum
spin 0 discu~~ion,
angular momentum, & 2, along their line of mation. Again there i s an elementary
interpretation for the geometrical. factors of (3-f3.82,&3) which appear ixr
transition probabilities ns sin2 B cos4 48 and sin2 B sin4 +B. These are the spin 2
probabilities that connect rnagneti~quantum number +l in a given directioxl
wilh magnetic quantum numbers +2 and - 2 in another direetion at, the relative
angle 8. The transition amplitude factor i/sin2 B also appears for spin 0, in
conjunction with the geometrical factor sin2 B, \vhich produces an isotropic
differential cross seetion. No\\-, hoi~ever,the spin averaged differential cross
section is
Spin 4 processes
3-1 3
at @
==
Slti
making explicit the origin of Lht: angle reestriction. (3-13."1"1) If ls-auld m m thaL
one had only to replwe (3-13.M) 1~ifE.1
and this is correct. But there is more here than mels the eye.
When the improvemends of (gL3.86)are: introduced in the denominators of
(3-13.78) and thereby in (3-13.82,83), the resal$ is
and its addition to (3-13.90) effeekivefy produces (3-13.88). Node th8L the
differential cross section for forkvard and baekrvard emission comes entirely
from this last process, The value of that cross section per unit solid angle,
$(a2/m2),differs only by a factor of 2 from the result of the low energy calculation, tvhen. the kinernatical factor 2/v is replaced by its high energy value of
unity [Eq, (3-1 3,73)].
The general evaluation of (3-13.69) involves little that has not, been encountered af;high energies, apart from the frevenl appearance of the parameter
The helicity eonstmction of the spinors in (3-13.69), dogether with the factor
3m/ez, gives it as
Spin 4 processes
317
The only transition not considered in the high energy discussion is tbe one with
zero initial magnetic quantum and fins1 magnetic quantum number of &2. 11;
has the anticipated geometrical factor, sin' 8. The immediately obtained form
of the differential cross section for unpolarized particles is
Explicit here is the contribution of the only pfoeesses that oecur at sin 8 = 0,
thoslc with zero initial and find magnetie quantum numbers:
I a2
(sin B = 0 ) = - - (I f
4 rn2~
~~h
They are also the snIy ones that survive at low energ;)r. It is the funetion 1
that produces the variation by a factor of 2 in proceeding from Iou. energy
( K = 0) to high energy (K = 1). Another presentation of the differential cross
section is
The laat term can be neglected Ett high energi:es, a d we recover (3-13.88). The
total annihilation cross section is evaluated as
which reduces to the limiting forms (3-13.74) and (3-13.89) in the appropriate
circumtanw.
Apart frclm changing the kinemaka1 faetor K'"-" inta K, the same digerential
cross section (3-13.101) rtpplies to the inverse process of electrcm-positron
creation in the collision of two photons, Factor8 associated with the summaGiarr
M8
Fields
Chap. 3
over final potariaations md the averaging af initial onw do not change ~inee
both particles, electron and photon, have two possible po1aril;stions. But there
is a dtifference in the evrtfuation af the tokal er0863 metioa, far electron and
positron are distinct particles and the full solid angle of 43r applies. This gives
the tab1 pair ere&ion cross scletion
If we apply them to the first version of (3-13.84), we get (the kinematics1 factor
l / M Z is not involved)
Spin prmesm
3-3 3
+-B
u:
whereas the formal repfacement of p with --p ia (3-13.110) pro due^ %heneg*
five af this rwdL. We did not eneounhr this phenomenon in retbting elmtronelectron sctatb~ngto electron-positron scattering since two spin. # ~ubstitutiom
are used there.
The high enerm limit of the efttctron-photon diEerenti8l crow sc?ction. far
unp1sriz;ed particle8 is, therefore,
Chap, 3
Fields
factor 1 / ainee
~
naw inifial and final particles are identical. Thia give8 directly
The Imf tQ?rmdoes not cantribute EtL hi& enerdcts, where (3-13, P 14) is reg&ined,
nor a%low energies where fhe Thornson cross ~ ) e e t i emerges,
~n
3-14
SOURCES AS SCAnERERS
The photon wurees that appear with inereming powers in Lhe interactioxls
'I;tTzl, ]FV221.. . can also be umd in the exbnded gense fo give an idwEz;ed
description of charged particles. As we have already sugge~%ed,
this Bimplifid
treatment is appropride for a particle that is sufieiently heavy fo be uniaftueneed by its ecattering partner. Conaider then s point charge of stre~@hZe
"Ghat ia stationed a t the orilyin,
When dealing with an immobile matterer, et11 reference to momentum camervaLion ie last, but enerw tzon~mettiongurviviE?~.The defirrition af the transition
matfix i ~ai eome~pondindysimplified version of the general definition (5-12.40),
mta-ining only the time in@& factor, and (3-12.43) similarly degenerak~to
a statement of the tramition pmbabilily p r unit time that digplays an the
ri&&hsnd side a single fwtor of 2z and the one delta function thati establishe~
enerm consenrslion. In the premnt situation, then, the tr&mition m a t ~ xifs
322
Fields
Chap 3
QP
which does indwd agree with (3-13.65), apart from the us@of
af the stationary scatterer.
The similar consideration for spin begins with
m the charge
wbich is expressed by
1,,u,,)
= 2m(dw,, d ~ , , ) " ~
where %hefactor slin f i e second entry reproduces the algebraic s i p s that are
r?xhibi;t;ed in (3-13.17). For either ehoice of a2 the summation over (zl dvet3;
the differential eraas section
Only the field $ ( X ) refers to propagating particles and thert3fore Wgz dso
describes an electron scatkring process, as do all the &her Wz,, Thus, the
expansion in powers of the static potential A@is no longer a classification into
sue~essivelymore complicated prooesses, but represents successive latpprlroxirn*
tions ta the complete treatment of the rnation of the padiele in the Coulomb
field of the point source. The inkraction slreletan here acquires more substance,
and thereby indicates one aspect of the dynamical scheme that is generally
lacking a t the first dynamical level, namely, the? possibility for unlimited repetition of a particular interaction mechanism.
Sinee all po\vers of A' contribute to the scattering proeess one should like
t o avoid that power series expansion and work direetly with the appropriate
, or G$(x, 2'). Unfortunately, the ability to solve
Green's function, ~ $ ( z X')
the Green" function equations in a reasonably closed form is restrickd to the
nonrelativistic limit, in the physically interesting situation of a point source and
the Coulomb potential. The latter has a simple connection with the diflerential
equation (3-11.36) for A$, whieh here assumes the three-dimensional form
when one introduces the time Fouricr transform in this time translationally
invariant situation:
(p0)2-m2d2mE,
pa--+m,
(3-1 4.18)
and the term quadratic in the scalar potential is neglected. It is this omission
that produces the essential difference between Lhe two regimes, whieh other~rise
are connected by the reciprocal correlation of (3-14.18). T h u ~if, we begin with
the nonrelativistic farm of the differential cross section and i~ztrodueethe
inverse of thc substitutions (:3-14.18), we get
= (dw,, dw,,)'/2(~a)z
l~herc
1
( d x ) exgfi(p2 - pl) . x] ----
[XI"
==;
417
Fietds
Chap 3
The fmfar Zip implies that the comec%iondiminishes the cross smtion for pa&i~lm
of like eharge and in~remwit for the serttbfing of agpossite charge#. f n dealing
only once with the egmt of the quadratic inkr~etianbrm and ignoring %he
phase f a e m tbti reprment the consquenee of r e p a b d linear interwtisns, we
have obtained only the first term of a rnultipli~ativepower eEieb4 in Zaq. Gs
thi~3first tern displetys, there must a l ~ be
o at leaat one faetor of %hepaft;icfe @p&,
&nmthe corrmtion i8 8 dfttivi~lrtiephenomenon.
T b comegpanding discussion of spin. scatteriag proc&s mmewhat mom
indhctly since both the deaird relettivisfie correction; and the repetition of the
effectiva nonrelativistic inbraetion am combined in TTzzl
Eq. (3-14.15). Taken
as i t ~ & @ n%be
d ~Xathr
,
implies the following %r%mition
matfix modification,
--
*(PI%f P?),
(3-14.n)
witbin the context of the spin m a t ~ xelement u a d irr (3-14.24). Aceardingly,
the mtuaf earneelion contained in (3-t4.N) is (but see below)
+
PO
where
with
Before diseuming this integd, let us obsente that, when hefieity staks are used,
which meam that the earreetion is confixlied to tramitions in, whieh the helicity
does not change. NOW
the last b r m of (3-14.311, with %heeomfsnt, fachr
+(pI -- p2)%,can be identified ae altering the phase associated with helicitypresehng trawitions. Put another w%y, this term is ima&n%qand, to %he
accuracy with which we are wol;king, it does not inkrfere Bvith the prheipal
ne:glwbd, &long with the
eont~butisiantto the scattering matrix taad can
imrtginav parts of the other term, The remaining real hrms of X, the only
significant ones for the cross section modification. we
where the return t;o eoardixlah space fiss b e n advsntagww, Also utilizt=d is
the thre-dime?r;tsian.almomenturn i n b ~ a l
(dp)
ed((p@j2-m71j9zt
*
(3-14.35)
444
The mdifiesltim in the transition amplitude for scattering without helicity
change isJ conveyed by
( 2 ~ p2
) +
~ mn -- (p')a
cos 98
-4-
ah2
*8
1
sxn2 46
- is
1/2
cos
infegrates the vector n over the unit sphere. The unit vectsrs nt,na g p c i f y the;
directions of pl, p ~ .The inkgra-f e m b ~ r . ~ d t e n
where f (al nZ) and f (Bl - nz) are perpendicular vecton of length cos f B
and sin .$8, reapeetively, Basing a choice of spfie~calcoordinaLes on .them @vea
cos 98 (ess *B - p)
sin%)Bcos2 p
ge
dp 1 -- p cos #B
cos +e - p
I
-- 21log .
sin 88
(3-14.43)
MU.,, (3-14.45)
= 2m(d@~l
d@P*)fi2
where
M = = f+2'go"-nl
Xng
and
The computation of the btttl differential cross srtction for an arbitrzlry idtid
spin invofve~
ng X
B%
= in Clv,
and
The obmrvationatt s i p that neither p, nor pb is zero thus eomm from a dependenee of the final inbn~ityupon the mlative orientrttion of the two aatfefing
plsnes. In pa&iculsr, if thf? two phnm are the same geometrically but deflectiaw
in opgosik seases am cornpar&, .= &v,, the ratio of the inbwity for &flee%ion8in the 8atme seme to -tlaat for the opposilti! s e w is
r &tan one when the individad 8eatkfing a & ~are iiientjeal,
p, == pb. The preference for sucemive deflection in the same m m will be
b i d with any choice of individual seatte~ngaaglm if, W in the pment
di~ussion,the pohfisation paramettt3.r hadJ a definife! sign at aU angfes:
This
it3
3-3 4
Ssurees as scatterers
329
where A$(%)indicates the vector potential associRted with the chwge Ze. W~iing
the form of the latter that is stated in (3-14.2), the transition matrix element
appears as
where
and npis the unit time-like vector that has the single component no = 1 in the
rest frame of the charge 2%. Enerw conservation takes the form
iL is used in verifying that fb eEectivc: pkobn emission source is conserved,
for this is the algebraic pmperty
First let us observe the simplifieations that appear far soft photons, where
the photon momentum k is negligible compared to p1 - p2 and
= P.:
This @ves
whieh is the trntnsitioo matrix element for scattering in the Coulomb Eidd,
multiptied by the probabiliLy amplitude for photon emission by the source that
represents the instantaneous trangition of the eharge ep from velocity p z l m to
velocity pl/m. This conforms vvith expetation. We should remark, h o r n @ ,
fsr future reference, that the connplek negbet of the pboton meehaaied proper-
536
Fields
Chap. 3
ties a t sufficiently low frequencies is justified for finite particle deflection angles,
but require8 m r e careful eonrsideration when the deflecdian angle is very s m d .
Glomly rebted fo the soft photon situation, but &skinet fram it, is the low
enerw or nonrelativistic Iimit. Here, the photon momentum is negligible but
any frastion of the initial kimtie enera can be r a d i a w as the photon energy
and the fufther integration over alt pa.pticlc?se&tering angles gves a cross ection
fsr the photon energy distribution:
It is also int;eratiag to consider the digerential emss per unit solid angle dQ
l;fitzL
&Einto the maximum energy aet by the initial kinetic energy T = pg/2m.
The i n b g a l can be evalusltd in general, moat simpXy by f t z c b ~ n gthe denominetor into 1 - z, 1 $ z, 1 -- xe", 1 - X@-'@, but we shall only present
(3- f 4-69)
where
although wc? have not troubled to iatroduee the invariant equivalent of the
initial pa&icle momentum. The four-clirnensionaI delta function slaks that
382
Fields
Chap, 3
and, in the rest frame of ?P,which is the coordinate sysCern of physical inbrest,
--D.
The requirement nk2 = 0,which asserts tbe static nature of the field in the
pbysicd, or &attached, wordinate syslern, becomes in the particle rest frame
and therefore
kg = k$
+ kgl[(l - v2)/u2],
where
kg = kK
+ k:,.
and k i is sufficiently small, it would seem that the virtual photons dould be
approxirnaM by red ones.
There is one app~retltdifficulty, hawever, Playing the role of pol8riaation
rrecbr for the incident photon, is the vecbr np,which is indeed such that nlcz = 0.
But we should etxgecb that the pola~zl~fion
vmtor of a red photon is, or can be
cho~en,without time component or component along the propagation direction,
which i s here the negative z-axis. This suggesGs performing whrat should be a
gauge tran~formation:
(3-14.83)
n' -+ nC - (?/kg)%,
w h i ~ his comtrucbd bhave vanishing time component in the pa&iele rest frame.
The z or loxl@tudinal component of the new vec-tor is then
and, provided
r k ~ l k>
;> I/T,
Now,
k z ~ g=
kip,
- *ki,
kzpl = klp2
+ +ki
(3-14.87)
and therefore
whieh indicates that the substitution of real photons for the viPtu~t1pElotOns will
be justified if suitable upper limits are placed on kg 2;: k:. A suggestion of the
magnitude of this upper limit is obtained by comparing, in $he gauge ef, &,pz = 0,
the pa&icle r e ~ frame
t
values
and
namely
k~
< m.
(3-14.91)
We shag eonfine the discurnion to the diEerential crag@section that $ive8
the mew spectrum of the d t L t n l photons in the Z eoordinete system. Since
we are ROW f i d y establi~hdin two &Rerent eoodinate systems moving rehtive to eaeh other at pmctieally the spectd of light, a few notational distinctions
are needed. The Z h m e photon e n e r a will be denoted by
K = --nkl= r(k!
+ vkl,) r: rk!(l
-- eos 8),
(3-14.92)
where B is tbe photon scattering mgfe in the paPticfe frame, The kinematics of
tbe phaton s c a t k r i q process in that refewnce frame, as derivd from
0 = (p,
+ kz -- k112 + 'm
= -2m(k$
-- k:)
f 2kyk$(1
- cos 8),
(3-14.93)
334
Chap. 3
Fialds
is expressed by
k? =
k:
(3-14.94)
+ (k$/m)(l - cos B)
E2 = rm,
E, = E2 - K
(3-14.95)
are
-K=
E2
(k!/m)(l - cos B)
1 (k:/m)(l - cos B)
% =k! g P
E1
(3-14.96)
and
The latter shows that the incident photon energies k: that can produce a scattered photon of energy K (two different coordinate systems are used here) will
be restricted by
E2 dK k:
- - = - sin B dB.
El El
m
Considered in the rest frame of the incident particle, the differential cross
section for photon-particle scattering is
where the factor in the denominator arises from division by the photon flux
multiplied by the particle density, 2k: dok22mdo,,. The polarization summation
and average is
X I As
= f (l
+ cos2 B),
(3-14.101)
which also follows from (3-12.124), with n = p2/m. The final momentum integration can be performed with the aid of the kinematical relation
and
or, with
(dlez) = r dk$ dk2,dki,
and &hen
The suggestion implicit in (3-14.91), h, S m, is that no significant interactions oceur for larger values of the Cr~nsvemmomenhm, We propose to
e x a ~ n Chip,
e que8ti:on.
A quick indication of the quan$itaLively eorrect resullt is obtained i f one
merely aempk %hatthe eEee.t;irre replacement for &, ia indepndent of' K .
Then it sufiees Lo conrzpgre (3-14.112) with the Merenth1 crass swtion Bppr*
p ~ a l eto soft photons. This diwwsion tsktls place entirely in the &attached
physical coordinate system. The paliafization summtioo, in $he digerential
e-mm swtion derived frsm (3-24.61) i s
This expres~ioncon$ains the only reference ta the direction of Lhe emitttld pholon,
md we shall integrate over all solid angles. Removing the factor l / K Z ,that
iategal is
where R is now the uniL propagation vwtor of $he photon, and high enerw, @oft
photon simplificatiom have not yet been irtfrducd. We first obgeme t h ~ t
m denominabrs, it is ub3ef~ltO
To inhgrak the km containing G
(EE-J E2 --- E )
Sources as seattrrrarsl
3-1 4
337
Lvhich still needs to be integrated over the deflection angle 8, But now we must
recall the M-arningthat the soft photon sirnplifications need to be qualified for
very small angles. I n contrztst to the singularity of (s14.118) at fl = 0, the
minimum value attained by (pl k l - p2)2, ~vhichoccurs for scattering and
emission in the fortvard directiorr, is
by
rn2~
2E sin H@>> -t
2_E;"lE2
C3-14,121)
[shere (3-1 4. X 18) can be used without correction* Ernployirlg the variable
y
"=:
(Elm)sin $B,
(3- 14.122)
338
Chap. 3
fieldra
we begin the integration at a conservative upper limit to the real photon disc u ~ ~ i okmaiX
n , << m, and thus
Ymin
r==
(kmax12m) K< 1.
This gives
(3-14.124)
dv(l
+ v2) log f. -X
---
u2
and
16 z2a3dK
-
(3-14. f 26)
3 m2 K
Whexl the real photon contribution (3-14.1 12) is considered under soft photon conditions, E l --. El2, the addigion of the two part^ precisely cancels
fog (mp,,,)
H, and the inference is that, $enemlly,
d@
--m
virtual
+-
H. : f have not forgotten that you decided to omit all reference to historical
nlatters, but your use of il, parametric device for combining denomintztors
prompts me to ask about a smafl historical point. The technique of introducing
parameters to unite denominator products into a single denominator is invariably ascribed h Feynman in the literature, 1%it not true, however, that
the usual intent of that device, to replace space-time integrations by invariant
parametric integrafs, \vas earlier exploited by you in a related exponential
version, and that the elementary identity combining two denominators,
(3-14.1 16) in fwt, appears quite explicitly in ra paper of yours, published in the
same issue that contains Feynman's contribution?
X. :Yes.
In the physical coordinak system, at high energy, radiation processes occur
predominantly near the forward or longitudinal direetion. We express this
through a dceompasition into longitudinptl and transverse components, sts illus-
The virtual photon contribution to the differential cmss section that is produced
by (3-14.72,73) is
~.---"---..%.=--
virtual
r2 E2 K
(S14.f 32)
U B ~ ebpprop~~k
R~
variable tra~glalions,we hwe
Flelds
Chap, 3
which doe8 inded differ from (3-14.124) only by the faehr E1/g2that i~ nededt.
Co combine properly with the general real photon contribution (3-14.112) and
prduce (3-14.127).
To give sn analogous discussion for spin 1 psrtioles requires, first, the explicit
f o m of the electron-photon diRerentital cross wction in the r a t frame of the
incident efeetron, That is waitstble Lo us through transformafion of (3-13.1 l?),
%hee n k r of nnw e r a s section, but, there is some i n k r s t in a direct derivation.
The transition matrix element is
whieh urns purely spatial polarization vectors and a simplified aodation. The
m a t ~ xfaetor in square brackerts reduces to
where the latter exploits the fact that uz i s an eigenvector of ?%with eigenvalue
-4-1, and introduces the aotafion nl, z far the unit propagation vwfmirs of the
photons. C o n ~ i d e e ~real
g polarization vecGors for simplicitcy, the tr%nsition
nnad~xelement becomes
(d%, d o k , do,,dwt,)1122e2u:[-el
e2
+?,(g
e p nl
e2
+a*e2aeal
X el))uz. (S14.139)
term and the spinor8 are omitbd we ge6 the comespondirtg spin O
If the
expremion. The summation of Lhe tr%nsi.tionprobability over final spins can
be pe$armed with the aid of (3-13.11Q), giving the pino or faehr
The matrix prduet is reducd by omitting a11 krms that eontain a rli fa~tOr,
since uz is a
eigenvector, or s a faetor, the latter expressing the averaging
over aft initial spins, A quite sho& edcufation then gives the foffo~4ngfar
X [l
- cos B(el
e2)'
+ el
e2n1ezag et
g
- el X
ez alet X
ozJ,
(3-14.142)
where
II~
(S14.143)
= cos 8,
C@Z
X h3 *
ez) m21
= (et X e2)%
cos @
E@l
-- el X e2 * nlel X et nz
--el . e2n1* e2aS elf
(&l4.144)
and we g@$
replacing (et e2)' in the spin O cross section. When summed and averaged over
photon polarizstions the differential cmss section that appears in plme of
(Sli4.103)3s
For the purpom o f evaluating the pbobn ennission emss section, this is
wdm
mK -
log 2BiEt
whieh is to h aurnmed over the photon polakaation and the find electran spin,
and averaged over the initid spin. As in the dimurnion of pfxoton-elechn
s~afhringtat small angles, transitions with electron heliciw ehsngw %res i p
fiea ant, The cafeulatian can. h prformed dvtitnbgeau~lyby methods tha$
have dresldy k n illustraled, using photon helicify states and the photon
e ~ s s i o ndirection for reference, and expreming the efeetmn helieity states wilh
fhe aid of suitable rota;tion mal;rices. We shall only give the re~u1GsBere, which
are classified with respect to helieity change:
For soft photons, helicity ehsnges are fehtively negligible and the spin O &metare is reproduced. The sXighLIy different integral assacigM wi.t;h KeXieity
ebanges is
1
clu
virtual, yea
--
v2)
which add La
&
virtual
This virtual phohn part and the real fiaton contribution of (3-14.148) cambine:
ta give the final high ernerw f a m of the diRerentia1 cross =%ion for phobn
e ~ m i a r by
r an elecfran defleetd in a Coulomb field,
The proeegs that converts a pfrofon into a pair af oppositely charged para croming
tiele~,in the neighbarhood of a tafstianary charge, is m l ~ by
transfomations to the reiaction jurzlt eonsiderd. T h m %ramforrnatiorrsare
In order to recan~truetthe absoXuk squard transition matrix element, we
take the known differential, era88 section for phohn emis8ian, d ~ ~ refer~ng
~ ~ j ~ ~ .
CQ definitr: spins and pola~zations,and f a m
where the second high eneru version also emphmizes that we arc?inlerf;sM only
in the energy specification of the particles. Under fhe croming transformation,
dwk,. The differential
the kinematieal factor do,, dok,h,, beoomes dw,, bp;
er088 =etion refeming to an inciaent photon beant woufd require division by
2K dwr,, K = kg, and thus
ap8l.t from the spurious minus sign tkat accompanies this formsl substilution,
with spin 8 particles. When cross sectians involving summatioas over final
helieitiea and averages over initial ones are used, appropriate correc%ionsmust
be made far the different weight faclom. With spin 3 partieia this if9 '1;18f required, in thew reactions involving one initid pa&iclc?:and fwo final particles,
ainee b t h elwtron and photon hsve t ~ r ohelicity staks. For spin O padiolm,
howwer, the photon emission crass section, summed aver the phobn p~1arizaLion, contains an additionztl faetor of 2 relative ta the photon sbmq%ianer088
section, where photon polarisations are averaged. The implied pair produetion
cross sections are
8 z~ ~ QE ~ E ~
spin 6: cFo = 3 m2 K 2 K
(3-14.159)
K =E
+ Eft
(3-14.
XW)
In u8ing exbndd photon. sources 4x1 reprmnt heavy eh&-& padielw, tbe
the pdnf t h ~ new
t
kinds of p&&iele~am eneounkrd. They are idealixd
vemione; of cornpiLe s y ~ t e m ,S i n ~ eh y b o g e ~ ea b m ~
are the mast familiar
examplr?,they will be bmed H-pafiielee.
We first conerider a sbtic mume d i ~ t ~ b u t iJ"(x),
a n and the
p h ~ t i a fAl@(x)in Bame convenient p g e . The time tramlwtiona1 iinva~ance
of Green" funetiom, A+ for emmple, is conveyed by
where
The B. E. slynnme;try
af
&a
Notice that the joint sign reversal of p" and p' interohanges the forms of the
two differential op.etra;tors. Hence the eigexlfumtions can be so chomrr that
where a'* iindieaw a rf3X~hdSt?t af quantum rrumbrg, If G' iaeludes a magnetic
quantum number, for example, am refern to %henega$ive of that quantum
numbw. If i~ pomible to chwse the eiigenfuncfiona in a way tha&identifiw at*
with a', While not usuauy convenient for individual problem that Gbaiae
simpXifiw gener8f disewion~. To avoid canfugion, we &all a h undem%and
that p" is s positive quantity unless there is a specific indication otheraise.
* (p0' $ p'" --
2 e q f '~( X ) )
+,eoppopf ( X )
= 0,
(3-1 5.6)
~~rhich
is incorporated in the statement af arthonormdity :
(dx)
+pop,r,p(x)
(3-15.7)
f n the &Beme of the staLie source, this properQ is obeyed by the Imotvn eigen-
+ , o * ~ , t , ~ f ( ~ )
supposed to be isolated,
Sufficiently near a particular energy eigenvatue
however slightly, fmm all: the others, the Green's function is dominakd by the
eomwponding eigenfuncdions, and (3-15.9) implies that,
:
&+(X, X',
P@) --
*+p@#qfcar ( X ' )
C +p@tgr@p(x)
pot +
q"O"
(3-15.11)
A representation of the Green's function that ia valid near any part of the
physical enerw speclrum, or its negative, is given by
346
Chap, 3
Fistda
Wheu the eigexrfuactioxrs (3-15.8) are inserted in (3-15,12), the 'Emojvn form of
the free particle propagation function is recovered. If \re nolv allow
to
assume both positive and negative values, the Green" funetion can be premnted
more compactly as
~vl~ere
( f ) signals the extended meaning of
and
We urc intcjtrested, in this section, only in tltst podion of the energy sptlcLrum
which is inaceemible to a free particle:
< m, Such s$atf?scan exist, localized
in the neighborhood of the source, if &hereis force of attraction btween the
particle and source, of sufficient strength and range, Tn the familittr situation
of the long-rang& Coulomb interitctiotr between oppo~ihlysigned charges, no
minimum strength is required, %ndan unIimidd number of sueh bound s t ~ k s
exists. These are the H-particles, Wllat are the emissiorr arrd sbsoqtion sources
for H-particles?
The insertiorr of the Green's funetion (3-15.18) into the souree coupling t e r n
are sources associated rvith the particular H-particle label4 by pat, p', with of
appearing as an %dditionalindex snalogous Lo pin. Thsf the= sourcw mfer
only to time conveys the immobjlity of the very massiw W-particles, The
repeated operatjon of these sources will inject any number of parLIeles info
bound statea. 8ince no a~countis being 8ven of the inter&cfctionsamong the
particles, we ~haffbe concerned only with the propertie8 of rr single particle,
bound to the murect anid forming an H-padiele. Nti3verthelesa, id i s desirable to
verify that probabili6-y requiremenk are satisfied in the dynsmieally simpfifid
many-pa&icle situation.
The usual consideration of tc causal arrangement of emission and absorptioxl
source8 leads h
and the eausaf arrangement restricts the energy summation in (3-15.23) to the
physical, positive values. The mulLiparLicle Aabs produced by the caueat
analysis of the vacuum amplitude have the usual canstmction in krms of gauree
producls, and probability normalilration implies that
X,
-pO)lT = -yOG+(x,
(3-15.28)
X',
and o h y s
OX", more
compactly,
~ ~ o e =( - ~~ - ~~ " l )(
- ~ ~ )
1
p@'(1 - ie)
(3-15.37)
in which
~~o.(zO
= n(xO- zO')v(po')i~b$?
(zO- zO')
- *(X'' - X ~ ) ~ ( - ~ ~ ' (X') ~ A 2").
~ ~ O ~(3-15.39)
1( d ~ ) $ ~ o( x~)~* ~ Oa *t ( x x ~ )
s p ~ , q , a (xO)
t
=
-qta'(x0)*,
and
,i q & o , q t a ~ i v ~ p ~(0
~+1s0~)'
a"t ]
(3-15.42)
qa
The completeness of the multiparticle states, which have the usual source
product representation, implies that (p0' > 0)
gives
Now, let the static source that represents a heavy charged particle be
supplemented by a simple photon source. The terms in FV that contain one such
3M)
Fields
Chap. 3
The static source defines a coordinate syskm in which nc"can be chown to have
only s Lime component. Then fc"(k) h= only spatial components, \vhich are
proportional to the vector k, and the gduge condition reads:
tile scalar potential A0(z) in the rdiation gauge i s necessarily given by the
instantaneous Coulomb potential of the charge distributiort,
But the coxlverse is not true. If it is required that AQ(z)shell be the instantaneous Coulomb potential, presumably the intent of a Coulomb gauge, the
inference is that the time derivative of "C" A(z) must vanish, No restriction is
thereby placed on any static compnent of the veetor potential, A(z).
I t is $he radiation gauge la, ~vhicfithe sta;tie potential AP(x) refer^, The
vector potential A(x) can be used to represent the field of nuclear magnelic
dipale moments, leading to the hyperfine structure af H-prtrticles. Xn the
follotving, however, attention will be confined to the static chmge density and
its scalar potential. This avoids notatiorlial conflicts with the pokntials that
are associated with the aimple photori sources. The latkr are only needed far
from %heemission or detection sources, There, they reduce to the vechr poten-
A(t.1 =
C
IAkX(z)iJzth
kX
with
ALh(x)=
( d ~ kliZet*eiL'
)
4-
i~:th~kk(z)'l,
(3-1 5.55)
V.Akh(z)=O.
(3-15.56)
We now distinguish A+(%, X?), which contains the static scalar potential
A0(x),from A$($, 2'). The latter also describes the effect of the vector potential
A(x) that represents photons. The differential equation for the Green's function
&$(X, X') can be presented as (p = --$V)
(--
2')
(3-15.57)
The use of the Green's function A+(%,z') converts this into an integral equation,
..
(3-25.59)
Thus,
and so forth.
The particle field +(X) is related to H-particle sources by
(3-15.66)
After integration over the sharply selected photon energy, this can be expressed
as the Einstein B-eoeEeient, which relates the transition probability per
unit time to the photon energy density per unit angular frequency range:
k0(2k0 dwr/dka). Averaging over the incident photon polarization and direction
of motion @yes the nonrefati-visticexpression
is a reminder that, apart from the kinmadittal factors involved in the definitions,
the transition probabilities for single photon emission and absorption, are inkrchanged by the photon crossing transformation, k p 4 -kp, The emission and
tllbsorption rates are equal when the definitions refer to single photons of definite
polarization. And, as we learned long ago in the simpler coxrtexl of a probe
source, if n photans of the appropriate frequency are present initially, the
absorption rate is multiplied by n and the emission rate by n 1. The latter
represents the combination of stimulated and spontaneous ernimion processes.
The analogous discussion for spin $- particles begins with the Greenpg
function difierential equation
The succe~aivephoton interaction krm8 arc: exhibibd, with the aid of the
padicle field
(3-1 5.77)
as; [compare Eq, (%12,26)f
s') =
'
[$p@larG*(~)e-ipa
iqtp~'a8ap
4-
*e
itl:p0~pfa8#p0tqfar(~)
1.
(3-1 5.86)
pQfqglo6
Jtpofa<~) m
approximab eigenveebm of
= i Y a ~ is
, relatively negligible. It
and one mcopiaes the orbital contribution to the magnetic moment, wbiGh
ad& Ga the @pinmagnetic mantent in the mannrsr rc:presenM by g = 2. If we
neglecl thia mametic dipale radiatim, and the rdated eXeet~equadwpofe
rdiation, which is tke other b r m on the right-hand side aE (3-15.83), them
rmains only the rd4ttion of the electric dipXe moment, eq'x. This i s radiation
associded vvith aceebmbd charges, and is indepndent of spin. Inded, witEr
the sirnpiifieations irnpIjed by retaining only the firf;t two te
and replacing spin 4 eigenfunctions by nonrelativistic Brave functions, in &ccordarrce with the normalization (3-15.33), we have
This coincides with the correspondixlg spin 0 limit, (3-15'68). A similar consideration, relai.Led by the photon crossing transformation, applies to the tabssrptian process,
We shdl discuss
orlily in the context of photon scatbring. Using the
spin 0 structure (3-15,6t), we insert the causal field decomposition
i112~12&+(~,
xl,
- kg)p;e-"l
"'j~,o~~,~~(x'f
(3-15.88)
po'--k$=p'"-k~.
X',
p'),
with 'p
(3-15.89)
Far simplicity we consider only the namrelativistic limit, where the photon
momenta are neglected, as are the terms in &+(X, X', )'p having denominators
p'f $ p' N 2m, in contrast with the denominrttors''p -'p = E' - E. This
gives (using a slightly simplified notation)
3W
Fields
Ghsp. 3
from .inrhich less spwific erass sections are obtained by summing over aband
8veraGng over a''? by summing over XI and avr?r&ng over X2, and by i n b u ~ t i n g
over all solid angles,
At ghobn enerrgies %hatare Xsrp in comparison with H-pa~iclt,binding
e n e r ~ e only
~ , the 1mt tern of (3-15.91) survive^ and, with a' = a", one recognizehs the Thornmn cross section, which describe8 the scaftr;et;riagof (on a, rekL i ~ t i eseale) low enerw photons by a free particle of charge rf=eand mass m,
Another limiting conneckion with Thom~onscattering should appear a t very
low frequencie~,smsll in eompaeson with the mergy inkrvals beI;wwadigerenf
H-grarticlea. m e n a, photon of essentially zero frequency is sca,&&redelmtically
by a, part4eular B-padicle, the dynarnid connectians \;vith other W-pasticleg
are not in evidmee, and the seatlering shoutd be descr;ibetd: by the Thornson
fornuLa appropriak to the H-par$icXc?icharge and masts. Sinee the latter hafj
been ideafi~edas infinite, the elsstic scattering erass ection should vani~has
k@ -+ 0. This implies a set of relations, known as sum rules, which can be
va~ouslygreesented. The immediate form implied by (3-15.91) is
Here is another:
(3- f 5.92)
(3-15.95)
The 1st version shows the mathematical origin of the Bum rules; they are
matrix elements of the cammutation relation
The elementary ori@n of the sum rules does not detract fiorn their gignificanee its conditions of consistency for the phenomenological particle des~rip%ionof eomposik systems. That is emphasized by removing the idealization
of infinite mass to obtain the necessary result involving the charge (Z" - I)@
and mass M of the H-particle, viewed as a, composite of the two particles with
charge and mass assignment8 $ivm by -e, m (electron) and Ze, &g - m (nucleus). Tfre gcattering amplitude fhat appears in (S15.91) describes the pme e s ~ in
e ~which the electron absorbs the incident photon and emits the seatbred
photcllm. To this will now be added the reprwentatian of the processes in which
the nucleus alone perEorms these acts, and of those in which both par-tieles are
involved. Altfiough we have not developed the relevant general famalism, the
necessary modifications here are quih clear. The ma;trix product terms of
(3-15.91) describe two successive interactions with the electric current, to
which both particles now make conlributians:
where
is the rdative momentum in the center of m s s frame. In addition, there is a
eoxltributian in which the scattering takes place in, one wt that is associahd
with an individual particle; it is extended by
I n carrying out the reduction of Ghe matrix product, the relation betuveen
relative momentum and relative velocity is now given by the reduced mass,
m(M -- m ) / J f . Removing the factor of e2, we find that what replaces the
amplitude of (3-15.91) for s, realistic H-particle as ka --t O is, apart from the
gola~zittionvector product,
Here, -ed is the internal eleetric dipole moment of the system in which df
position ve~torsrefer do the cenhr of mass veetor
If is the amplitudes for individual scaLLhring by the two particles that are
add&, and not their cross sections, since this simplified treatment neglects the
photon, monnenbrn and thereby amumes that the photon wavelenglh is large
eompard t;o the particle separation. That restriction, is easiXy removed by
in~ertingthe relative phase factore and, with increasing frequency, the coherence
between the two scattering amplitudes disappeam,
It is also possible to derive (3-15.101) directly, by using s digerent gauge
whieh is specifiedty adapted to the long wavelength regime.. If the electric field
of the photons is homogeneous over the inkrior of the H-pahielc! and procems
involving the magnetic field are negligible, s suitable choice of potentiwfs is
A'(X,Z@)= -x*E(R,zO),
A(x,zO)=O,
(3-15.2M)
where
E(R, zO)=
( d ~ "2ik'[ekh
i)
exp(& R
kX
- i~:t&e:&exp(-ik
-ik'z')i~~~~
+ ikOzOf].
(%I 5. 107)
The reference to the rest frame, the state of zero momentum, redurn8 %histO
"G+(x,
X',''p
+k : ) ~
te"2*''
Again, we only consider the nonrelativistic limit. But this time the k r m in
p@ = 2m cannot be neglected.
the GreenP8funetion with denominabtors
We shrall need the explicit sbtemenl of completeness for the eigenfunetiam. It
em be: inferred by eamparilrg Lhe high enerp;y limit of the Gseen's funetion
@+(X, X', P O ) ,
Lim (-rap'~+(x,X',
pO--toc
= 6(x
- X'),
(3-15.113)
Chap. 3
fietds
D+(x,
X',
p') =
+,o~,~,~
(x)$pot,r.pfn')*ro
,ka"
3-1 6
361
vides our first encounter with unsthle particles. The distinction between.
stable and unstable particlm is a matter of time sede. Within suitably restrickd
time intervab, the mechanism producing particle instability is ineffective and
the stable partiele description is applicable, provided, of course, that enough
time is still available for the accurate dekrmination of the characteristic particle
properties. Ot;herwige, no single-particle description is mestningfut. The Hpa&icles supply examples of sLPLble and unstable parlicles, The particle of
minimum enerw is st&le. Those of greater energy are capable of emitting one
or more photons, thereby transforming themselves eventually into the tllbsolukly
stable variety. The initial description of H-paPficles wunned their stability,
and i s applicab-le over a rerstricM time scale, The descrip%ionis false for very
long time inhrvals because it mserts that weak H-particle sources emit and
absorb single H-particles that propagate unaltered between these acts, But,
given enough time, an unstable H-particle will transform itself into another
H-particle and a photon. These two particles are also c%pableof recombining
to form a single H-particle, Thus, a description of the coupFing b&ween weak,
causally arranged H-particle sources that does not refer to the real exisknee af
two or more particles propagating between them is physically incompbte. I t is
the inclusion of such multipart;icXe excbantgeis between sources and the consideration of some of the physical consctquenees that wit1 occupy us in this
section.
The first task is the identification of effective sources for the emission and
the absorption of an H-particle and a photon, This is analogous to the discussion
of Section 3-1 1. The description of s noninteracting photon and H-padicle ia
given by (using the apin 0 example)
Comparison with the vacuum amplitude term describing single photon emission,
as contained in (3-15.W), gives
2==
eff.
8 ( ~
(3-16.2)
and the same form applies, with appropriate causal labels, Co the a b s o ~ t i o nof $I
photon and an E-particle. Since this eRective photon souree is meant Lo bg/
multiplied by a vector potential in the radiation gauge, its appearance is simplified in comparison with the structure of (3-1 1.15). On replacing J";Zt ) K z( X )
and J2r(Et)K2(~"
iin (S16.1) by these egective combinations, we obtain a
desc~ptionof the causal coupling between H-particle ssourcw that is merfiabd
which extracts the tramverse parts of the mulfiplying currents. Thus, the
coupling krnn in the vacuum amplitude is
PO
3-1 6
363
where, as indicated, only positive values of''P appear. The resulting form of the
vacuum amplitude coupling term is
with
where
= &-porraFr,
-pofar(~o'
- z').
(3-16.11)
The emission and absorption sources of the vacuum amplitude term (3-16.8)
occur in the combination
- zO'),
We recognize in the central factor the prop~ga4ionfunction
evaluated under the causal restriction z0 > zO'. The space-time extrapolation
Fields
Chap. 3
r-pae,anre-pa~a~(-PO)
rpa~a~,pa~~a~~(PO)p
(3-16.13)
and the symmetry propedy is then satisfied by
The effective limitations on P' must also be removed if this propagation function
is to be meaningful for arbitrary sources. The P' integral can be defined to
simulate the initial consideration of extended sources, by excludi~ljlgneighborIf this is done symmetrically about these
hoods of the values p'' and p"'.
values and then the limit of arbitrarily small excluded in&xlr8!% considered,
with a speeial provision for''p = p'",
the result is to use the principal value of the singular P' integral. I n contrast to
other reeips $hat m~igncompbx values fo sinwlar intepala, this proeduro
h- the mtisfactory feature of preserving $he essential wociation of eonnplex
numhrs with the propagation function d p e ( ~ ' - z").
For S more expli~ittmt of these extrapolations, we examine how &hesimple
propagation function is modified, by choosing p'ra' = p""af', and considering
x0 > zol,g?@"
3:
The physjcally inkresting regime be&= afker a time Iapm of many periods,
p"(z' -- zO')>> 1. Then the integral is dominated by the immediate neighborand one can introduce a simplificstion by
hood of the singularity a t PO =
replacing r (Pa)with
3-1 6
366
The result is
which intrduces an amplitude that diminishes in time, without alhring the
time varying phase. This is in acmrd with the phenoxnenolo8cal viewpoint of
source theov. The H-particle energies that have been t?~~eur&tely.
identifid
over the finite time intervals r,a...(zo - xO') << 1 do not ehange their values
when the time scale is enfarged. We are not eoncernd here with e x a m i ~ n g
how the theoretical understanding of the energy spectrum changes EH we move
to another level of dynamical description. For our pre~en1purposes the numhrs
''p are given, whether by theory or by experiment is immaterial.
The unit value of the absolute square of exp[-ipO'(zO - X@')] represents
the cedainty with wlnieh stable particle will be found in the same e n e r o state
afkr any lapse of ttime, The square of the amplitude fmtor in (sI6,fZ.O)
scribtls the changing probability &at an undable EX-parLicle (r,ol,f > 0) shall
still exid after the time interval x' - x" -. f,
There is an initial deercse, at a r a k dven by r,o,ai.But this result become8
unsatisfacbv at larger time values. The persistence probability of the Hpadicle, according tx, (3-16.21), rewhes Eero at s finite time; it then inereme8
and eventually becomes larger than unit;y. The probability formulas is evidently
linnihd in. physieal applicability to smdl values of r,ap,lt.
Wlka~Lis still missing in the physieal %count is this: We b g a n with an
ex6ended H-part;icfe source emitting a virtual H-particle that quicHy trane
formed into ta real E-particle and ab photon. This situation endured until both
p8rtiafes reached the neighborhood of %heexknded detection. souretr;where they
recombined to f o m a virtual E-particle th& is absorbed, But, @ven enough
time, the recombination do form a vir-lual H-pargide can occur far from detaction
sources with this excitation rapidly deesmgosing back into real, part;ieles, The
cycle can h repeated mfl~nytimes before the virtual H-pafiicle is findly absorbed
by the debction. source, Qthewiw expressed, the fields appearing in the coupling
term (S16.5) originate, nod only directly in the ssurcw, but also indirectly
through other, efleetive sources which sre assoeiakd with the virtual H-particle8
that form far from the sources Ghrough the propagation of resl particles,
The qualitative description. in the last sentence is @ven a quanlitaiive
meaning by the following integral equation for the field &,...(zO) :
where %hem t r i x funadion TX describes the mechanism whereby, for the tmf
dime, zt virtuaf H-pztdiele goes through the cycle of transforming into a real
%M
Chap, 3
Fields
R-particle and phulon, then back into a (no6 necessttrily the same) vi&uf
H-particle that is detect& by the pro& source used to define the field. The
exciting fidd that appear8 in fhe inhgral expression surxznrg~zesthe @fleetof
%heinitial source excitation and of the u n l i ~ t e drepetiLions of Lhe~erevemible
conversions and is, therefore, considering all pDfa'together, the very field that
is being construeled. This point of view is similar to a xnuXLipfe setzt%!ring
analysis in terms of the last eoXXision. Xf this integral equation were t a be solvd
By iterafion, we would indeed be conside~ngmcceg~ivelymore; elabora;ts repetifions of the same bmis grocesa, The compa~sonwith fhe h a m descriptian aE
ane such action then identifies the matrix n. Tkis coqarison is ffl.cilitaM
by wfiting.
X &p:az
zO').
(8-16.24)
Al%bou& these are ra%hergeneral equ&ions we shall prsduee only an appsoxinrrak solution that is aipplieabfe
ordinary circumstances, as indicactrtd
by the specialization,
(3- 16.27)
rp~tofiapo..~t(~') 6.~.~rpa~(Po).
3-1 6
367
368
Firstds
Chap. 3
3-1 6
M9
is given by [we use the nonrelativistic approximation (3-16.67) but retain the
relativistic origin of energy]
The time t2 is R fiducial poillt \vithin the source K 2II(z"), and correspondingly
tlxe definition of H-particle emission source appeam as
The use! of a reference point that is interior to the source rather than arbitrarilJ;
&oen is always posgible and can be useful in identifying the xnwhanical propep
ties of states. It hcorrseg mnndataw in demribing unstable pa&icles.
376
Ctrrrp, 3
Fields
tz
ga~rticularunstable
The initial and fixlai times are now explicit in the specification of stateg, although
only t ==: t - t is significant?,as we have emphasized by using E l as the reference
time for the photon fielid. The time integration is evaluated as
tltld the traxtsition probabilitjr, summed aver photon polarization8 and emission
directions, but still differential in the photon energy, becomes
711
(3-16,50)
3-1 6
where
k f 11 = E11 - El
and, of eoursc?,
~ K=
I
The total probability is produced by carrying out the k' irltegration. That is
approximated under the assumption of \~\.eakinstability, r 1 1 << k f 11, by replacing k v l f 11 w~ithunity and evaluating the integrals as
-=
5vhicE-i is the required value. The spectral distributiox~of ttre emitkd photon is
also exhibited on evaluating (3-16.50) at a time rrlt >> 1, such that the radiativ~
transition has certainly occurred, The result,
is the familiar Lorentzisn shape th%tidentifies the decay constant 711,the reciprocd of the mean lifetime, tvith. the tvidth of the spectral line at half-maimurn.
This is the shape of a spectral line emitted in a transition to the stable
W-particle. But what if the final H-particle is also unstable? Now eonsider n
third H-psrtiele 111, ~vhiehcan only decay into X I , with the subsequent transmutation of the latter into the stable variety 1. In this situation two photons
are emitted and we must use W z zto describe the process. There are tutoanalogous terms in the relevant probability amplitude which are related by the
B. E. symmetry of the phobns. But, apart from the special circumstance
kf"lr 2rl h:& 1x1, only one of them terms is appreciable depending upon tvhich of
the photons has its frequency near kp 11 while the other frequency is close to
kfI 111. Thus, it suffices to regard the photons as distinguishable through fheir
frequerlcies and use only one of these terms. The probability amplitude for the
rvhole process is
tf"l1ere the time propagation functions detail the sueeessivc enusul nets of tllc
drama, In writing this expression, we have proeeedd ss thaugh the H-padicles
of types II and III were unique, although additional indices a11, a111are necessaw. T h a e detdls can be inse*d
and do not d e e t the resulb, under the
physical circumstances indicated in (3-16.27). The z0 time integretioll is thc
one already performed, with zO' supplying the latver limit instead of tn. The
Lime inkgrd faetor of (S16.26) is, therefore,
The implied transition prsbability Chat refcrs only Lo the spectrail distribution
of the photons is
El of
The successive emissions are not independent. I t is the energy k'
the p h o h snd pa&icie i n b which I1 dec%ys,rather than the energy E ~ Ithat
,
determines the spectral distribution of kO'. When one integrates over kO', the
result is just (3-16,55), h i e h means that W-partide 11 is certainly produced
at wme time by decay from Iff, after whi& the previous disemion applies.
The sns\\*erto the question concerning the spe~traldistribution of the ptrobn
radiakd in a Cransitian bett-veen unstable H-partidea is obtained by. intiegrating
over k'. It is instructive to write this integral as (E = ka f E I )
3-1 6
373
wiLh a width given by the sum af the individual H-parLicle widths. This conclusion is parfieufarly transparent if one recognise8 that the double enerw iatepal
of (3-16.W) is equivslenl to s single time irtkgral:
It ~-outdbe hard aot to suspect the existence of another approach that is capable
of producing this formula directly. We shdl find it, not surprisindy, in the
time cycle description.
But, first, 1eL us give an. analogous discussion of photon scattering, in order
to verify that the unphysicat infinite ttross section a t exact resonance has been.
removed by the explicit recognition of H-parLicle instability, Elmtic wscattering
by the stable E-pa&icle L will, be comidered, Then it ~u&ee1;3tro introduce
modified E-particlis propagation functions in (3-15.88), which will be used only
in the aonrelativistie limit and in the gauge of (3-15.lM). The sigrmifiesnf
change is the ineroduction in (3-15.101) (\v@ ignore the l / n l term) of the
substitution
while E -- (E1 - k') remains unaltered. To understand this it is necessary to
be somewhat more general than (3-16.37), where r,ot (P') is considered only for
PO =
We return to (3-16.29) and proceed as in (3-16.351, but with
rPo.
(P') retained, and get
showing the general form of the imaginary term. This distinction is unnecessary
near resonance, p' -- p'', or EX k'
E, but it is needed far from resonance
conditions. Otherwise we should have, incorrectly, added an im~tginarykrxn
to E -- (E1 -- k'), where%?
rE(E1 - k')
0,
(3- 16.615)
sinee no photon emission can occur if the Lots1 energy is less than E x .
H-particle If becomes strongly excited -when
is typicd of any rwonant scatbring process. The bmic resananf crass seetion
is 4.1rg2, h m 4rr/(kf x1)2, which is multiplied by the number of resonant states,
grx, and divided by the multiplicity of the initid particles. Tfist is jusf the
fmtor of 2, referdng to the two phobn pofsri~stions,sinee H-particle I has bmn
amurn& to be unique.
The promise to exhibik another and, more direct derivation of (3-16.62) will
be fulfilled, even to the point of generalizing this formulst so that it, refers to any
pair of urntable H-padicles, which are capable of dwaying in other sequences
khan XI1 + 11 X. Here is the statement of the mare general problem. The
arbidra~
unstable H-particle f I1 is creELCed near time zero. I t can decay to a particulrzr unstable H-particle I f as we11 W in other ways, and these secondary unstable particles continue the eabsedc?until the stable particle I is reach&. Wh&
is the differential probability for finding B photon of frequency ko 11 k f r 111,
wiLhout reference to the ather photons of different frequency &hat are also
emitbd? For s spwified polarization, that probability is expressed by
which msumes 8 time in,terval long exlough to have the probability attain. its
final value. Let us supply two additional hetom, Gtkp
the pmbability amplitude
for detecting the photon kX, and -i&
it8
,complex conjugate. This produces
376
The transition to the time cycle is made after time t2. Time t1 is now emounhred
on the return path, which is certainly 'later' than tz, and q(tl -- i2) is replac4
by unity. Also, time t is reached 'efLer9irne tr and q(t - t l ) mu& be replaced
by q(tl - t). Since both t and tI refer to the return time path, there is no sign
change in the integral. The proper treatment of the r Lerrns is fixed by %he
physical necessity of maintaining the damping, the weakening of the ~oupling
~ C inereasing
h
dime interval, All this gives the subsfitution:
Chap. 3
Fields
snd of --gkk
from E(t2) and
With dk0/2?r removed, the factor in front of the double time integral, summed
over polarizatiarrs, is the A-eoeseient for the dmay XXI -+XI, Ia the h k r m k
of a more uniform no%%tion
we: shailt now denote it by 711 ux. T o aimplify W
..time irrbgrals we introduce new v t t ~ a b l e ~ :
which mnges from -m t~ QD, and t<, the snadier of the two %inn@#
whiah vahw
from O to m, Then, %hetransformations
d have ~ e p r d ~ ~ c x f
When XII can only radiab to If, 7x1 1x1 = "Frrr, a ~ we
(3-16.62). Mart3 generally, the probability of emitting any hquency in the
neighborhood of k f 111
~ is (TI1 I I ~ / Y I I ~ <
) 1, ~ecarding
tO
3-1 5
377
and this expresses the competition between the specified transition and all1
othem that 311 can undergo, The sum af these fractions over a11 decay mode8
of If1 is equal to unity.
The time cycle extension of W z z also gives a direct derivation of the remnance scattering emss sction (3-llieCi9), or, rather, ifs generaliszttion in which
I1 beeomes an W-particle that can decay in ways other than down to the gtable
particle 1, and rr becomes the corresponding total. cross swltion. A photon is
incident on X, and eventuizlly one again finds I, aecompzznic;d by one or more
photons. The total probability for these phenomena, \vith s given interaction
time, is
When the initial particXes are introduced by appropriate sources, two of each
kind, this beeomes a Lime cycle vacuum amplitude, deseribd by iW2z. The
result is obtained from (3-16.77) by replacing XI1 with the stable I, and using
the field of an incoming photon insted of (&X6,78) :
The integrand depends only upon the time variabb 1 =. l z -- t l , and the integration over t< is identified with the duration of the inter~etion. The total erom
section is found by dividing the photon flux 2k0 dwr into the transition probability per unit; time, On recognizing thaf
e2(k!11)'
/(IIajx * e 1)i2 =
T ~ I ~ YXI,
I
(3-16*88)
kf 11, as
,P
be obtained from the total cross section by multipfyirrg the lathr tYith an additional rx II/rrrfachr. That is indeed the result dducrsd from (2-16.67) whm
the quantity mntaind in (3-16.68) is $;iven. its pneral interpretation W the
partial width r 11 :
The above discussion is ixreompfek since no mention has been made of the
ddi.tionsl t e r n in IFzz that is dewnded by the cmsaing symmetry of tbe photons. It is produced by reversing the sign of k'. This tRrm is certainly nonresonant. But, more important is the appeesance of the initial energy as EI - k a ;
the value that should be a~signedto the damping constant of H-pa&icle I1 is
not rI1 but %em, as in (3-16.65). Then the resuXting dime inlegrd gives
6(kf 1- k 4
0.
All the developmends of this aeetiotoxl have used the exampie of spinless
particle8 that are bound to form El[-particlw. A similar treatment for spin
p a ~ i c l e swould run in exact paraIXet, with occasisn~iinserLians ar deletion8 of
c(p') factors, for example, to represent the changed statistics. The nonrelativistic results are identical.
The natural instability of H-particles has direckd attention to the necessity
of considering multigartiele exchanges, in addition to single-particlepropaetion.
It is a complementary aspect of $he principle of spaee-time uiformity that
couplings identified through the examination, of red proeesms eontinue to be
meaningful when. appfid Lo virGual processes, This says thad multiptzrtiele
exchanges are significant, although the energy CO produce sevemX red par-tietes
m6y not be available. Thw, &henext stage of d y x l a ~ c a levolution. is %be
~ystematicgeneraliz;ation of all single-partick exchanges b e t w ~ nsources to
those involving two particllcls, including their unlimited repetition. Before embarking on. this masive progfam, however, wre shall give a relatively b ~ e disf
cug~ionof the ~avitationalversion of ~ u e hconeepk as primitive interactions
and gauge invariance.
3-17
3-$7
379
a, 6 ~ ~ ~ (=x0.)
The corresponding arbitrariness in the identifieatiion of h,,(z) is exhibited in
and therefore
The introduction of the source restriction, through the divergence of this equation isslate?a the a~pectof the field h,,(z) that is governed by the arbitrary
(z) vector,
(3- 17.7)
a,(hpv(x) - ~ ~ " h ( z=
) )a2ty(z).
Returning to (3-1 7,4), we deduce
-+
equation.
it is (3-3,17), with m = 0. The structure of the left-band side of t h i ~
is such thst its divergence is identically zero. The vanishing divergence of the
source tensor now appeam a8 an algebraic consequence of the field ~ U ~ Z ~ ~ Q X ~ F
380
Chap, 3
Flalds
Since the arbitrariness of the vector E,(z) is still maintained in these field equations, they are unaffected by a redefinition of the field h,,(%) having the form
The gauge i n v ~ i a n c eof the Iefbhsnd side of (3-17.14) is not realized through
the invariance of r,,~,but rather
and
Note, however, that thege gwgr?, transformation rmpsnses do not invdve first
derivatives of Lfre Ex fz).
Another ~ y a k mof first-order differential equations fit is (3-3.20, 211, with
m = Oj is praducd by the definitions
@~XE(%)
= - w V X ~ (=
~ ) aph~W(z)- dlh~@(je)
(3- 17.17)
and
oy(z) = wPhh(~)
= aph(z) - dhhyh(s),
(3-17.18)
name1y,
ah@,.k(z) - a.w,(z) = K(T~.(S)- + ~ , , T ( Z ) ) .
(3-17.19)
@ph~(~)@~xP(z)ak(ap~r(~)
- ~P&(x))
and
@p
(4
+ @p
(z)4- a p a k t k ( ~
- )d2& ( X ) .
(3-17.2131)
(3-17.21)
It is observed that Lbe divergence af the vector field m,($) is gauge invariant, A
comparison of the form of the divergenee inferred from (3-17.18) with (3-17.9a)
shows that
d,wg(x) = +KT(%),
(3-17.22)
which is dso the eantraction of (3-17.19), sinee
3-1 7
389
(3-17.24)
such that
rx(z) - XT'(X)
= ~wx(z).
(3-17.25)
Another connwtion between the two third-rank tensors, which implies this one, is
(3- 17.21))
i(rP(z)i- 'r(z)),
(3-17.30)
i t s contraction
dyhpv(z)
with
Apart from a, divergence krm, this Lapange funetion is the analogue of (3-5.411,
with m -. 0. For simplicity, we cfo not include a source for the third-rank
tensor field, in contrast with (3-5.40). The stationary requirement for va~ations
of hp', or h""" ---- +@@"h,
recavers (3-17.14)) and variaLions of F,,x, &er rearrangement~indieakd by the structure of (3-3.45), reproduce (3-17.12). The
Laf~;r&nge
funetion is not gaum invariant,
382
Chap. 3
Fields
- Arrh)e
( ~ 1 7 . ~ )
When writfen out as a qudrtttic funetion of the fint defivativm of h,,(z), this
fi-ange funetion differ@from that of Eq. (3-5.99) only in the 1-6 germ:
which itlustrat@ the freedom to add divergence terms, This athmative has
d r e d y been n o M in Eqs. (3-5.31,32).
The similar development that i s bmed on the fimborder diEerentiaX equsGions (3-17.17, 19) stafts with
which msures the inva~anceof the tbction. Whm wh,, Io~wi k independenf
sfatus and is defined by (3-17.f7), the Lavange function clan be chomn M
which is the quadratic function of the fir& defivakives of h,,(tt) that is pmdued
by averadng the two alternatives of (3-17.35) :
3-1 7
383
The stress tensor Fy(s) has been given a kinematical definition, m~hichis
not unique, through the response to infinihsimal coordinate deformations,
The two concqts are identified by requiring invariance of the action under the
unified gauge-coordinate transformation,
The use of the total stmss tensor !Py P",as the fwtor of h,, in the mtim,
is the introduction of a primitive interaction. Same modification of tB"is needed
since it is not conserved inside pa&iek, aaurces, and a gravitational model of
particle mmes must; be introduced. But let us dekr the biscugsion of that question and proceed with the development, whieh is modeled so ctosdy on the
eleclromagnetic one, in order to reach Lbe point of divergence between the two
very different physical sy~tems. Consider %heexample of spinless pa&icles,
using the simplesl stress Lensor form, Eq. (3-7.81,
As in the electromagnetic anetlogue, the eoupling term &,P"will be combined
with the particle Lagrange function, -to form
384
Chap. 3
Fields
+ *(a, ~ x , ( x )+ a, ~ x , c , ( x ) ) ,
= hpV(x)
including
E(T)
+ a, 6 # ( ~ ) .
=h(~)
(3-17.52)
(3-17.53)
= /(d5) ( l
which is satisfied if
(dz)( l
+ ~ ( z ) ) + ( z ) ~ ,(3-17.54)
+ h(x)) = ( d z ) ( l + E(@).
(3-17.55)
(3-17.56)
This can only mean that h(x) is restricted to be a very small quantity, permitting ha, 6 9 to be neglected as a second-order object.
The situation is similar for the quadratic derivative term of the action.
We first notice that
isolates the factor 1
Then,
- 2hpv(x)
(3-17.60)
The gravitrrtionslfZeld
3-1 7
388
(s17.61)
~ " ( 2=
) s""~)a.z@ahz@.
(3.- r 7.62)
or
det gpv(z)
(3-17.66)
is
(3-X 7-67)
and therefore
(-@(z)) " 2 ( d ~ =
) ( - g ( ~ ) )l l Z ( d ~ ) .
(3- 17.68)
in kms of
which i s (3-1x32) with dl reference to third-rank tensors s%a%d
386
Fields
The missing constant term can be added in (3-17.72) since it changes the
Lagrange function by a divergence. Then the strong field generalization is
clearly indica;ted:
(3-17.75)
Z K (g(%),
~
~ ( 4 ) (--II(x)) li2gpv(2)Rpu(2),
with
Rhv 8hrtv- aPr;h ~ ; ~ r ~k~r:h,
$~
(3-1 7.76)
E
This will indeed contribute an invariant action if yMvRp.is a scalar with respect
to arbitrary coordinate transformations, The required covsriant knsor behavior
of R,,(z) must emerge from the transformation law of the three-index symbol
)
The latter should resemble a third-rank tensor but cannot be entirely
of this nature, according to the ~veilk.field transformation of (3-17.45) which
~ontainssecond derivatives with respwt to coordinates. A suitable generalizatioxr is stated by
r;,(i-);1.~" rig(~);i,~Pa,~g
a,a,zk.
(3-17.77)
This transformation Xatv is such that a coordinate covariant derivative of firstrank contrsvariant vectors can be defined:
v,vP(~)
= (a. + r,(z))~v"(z
= aa.VP(z)+ r:&(z)vk(z). (3-17.78)
The matfix notation facilitates the consideration, af
[v,, v,lw"
where
=$X
V&,
(3-17.79)
(3- 17 .so)
R,,"~ =. a,r:~- aVr:&
3- T:,T";~ - r:,r,Ph
is indeed a fauxrth-rank tensor, which is antisymmet~ealin p and, v. We can now
recognize the tensor character of
+ r",
= V& 8 ~ ; ~V@
vk(-S(~))
H2
ak( - @ ( X ) )
- (-g(%))
112riv(~).(3-17.86)
v,[(-o) 1j2s"l - a i ~ , [ ( - ~ )
lizy'Y]
=
0,
(3- 17.88)
which imp1ies
v~[(-~)'~=
~ ~0." ]
(3-57.89)
From the latter pmperty one d e ~ v e s ,successively, the vanishing of the covariant derivalives for g ( ~ ) g'"(~),
,
and g , , ( ~ ) . The last statement,
leads to the explicit construction
mfhieh is the strong field generalization of (3-17-12). The weak field vemion af
(3-f7*90) appears in (3-17.29). As one can verify directly, the vani~hingcovariant derivative of g ( ~ )implies, according to (3-17,86),%h&
which generalizes (3-1 7.13). This form ensures that Bp,,
as defined in (3-17.76),
is rz symmetrical tensor.
The variation of gp"(z)in the pul;.eXy gravitational contribution to the action
induee~
where
4, R,,
R = FvR,v,
$gPvEp
and we have used the deterninandal property
X
(3- 1%94)
The variation of ggvfz) in the matter part, of the action, defines a hnsor
t,,(z) that generaIizes the stress tensor,
3-1 7
389
it lis Eimtein" pgra;vitationat field equation The sfress tensor divergence eondition (3-17.103) appesm again, now as an identity demande-d by the ~tmeture
of the p a ~ t a t i s n a field
l
equation.
The replacement of spin O particles as the model of mavitating matbr by
okher inkger spin, parYtielm is relatively straighithmard, A rather special but
interesting example is provided by photons. The Lagranp function
is innmediahly getneraliaed to realize invrariance under arbitrary caordinab
tramsformatiom, while maintaining electromagneticgtauge inva~ance,by writing
(3-17.1 IQ)
112 v
P fi, SX = 0
(3-1 7,115)
or
l(%)= gb,(z)t""(x)= 0,
(3-17.116)
Chap. 3
Fields
-+
(3-17.120)
(3--17.~2~
We conclude from these resulk $hat posrsible dditiontzl brms in %heLagiznge funetion of matter have fhc? form
"'
where mNF9is a tensor, refening to the matter field and the gravitational fieid,
that h@ the symrnetries previousl-y noted. An illustration for s p h OIgenerfziized
from (3-7.88), is
For definibeness, the coacient is ehosen so that the new ~trmstensor, in the
absenee of the gravihtiomtal field,
.,8
a,$aR
- f g,v(a"
(3-17.127)
1%would be indere~tingto ve-fify that this system, with m = 0, is conformally iwari%nk,in the sense of (3-17.114) supplemented by an itppropria6e
response for +(X). One sws $hat
X(z) is eonstanl, To complede the test If is 8uBFieienl .t;o cornider
an, infinitesimd variation of X(%) from uniiGy, 6X(z), Then,
is suitable, if
where
is to be computed from
&l,v
= &kg,,*
'l2]
= 2gP bX
and
5/""vr;,
.=;
-gwap
&X,
(3- 17,135)
thus assuring the invariance of the action, for m = 0, under the ~ o u pof
conformal transformations.
Harold interjwts a question,
where R = g@"R,,retains its meaning in terms of the g,, and their derivatives.
It would seem thsf we have acquired rt new nnasslws particle of spin 0, repre
~ n h by
d the scalar field ~ ( 2 ) .But something is amiss, In a weak field situsLion, with
g(%)
1 -4- ~ ( z ) ,
(3-17. 140)
the dominant f/;, Ldtrms in this Lagraxlgr: funetion are
The p derivative term has the wrong sign. And the source of the p field, proportional h R ~ i ? , vanishes according to (3-17.107). All this indicaks &at
the p field does not descrirbe a physicd excitation. It can be transformed away,
by introducing the conformal transformation \%rith(X(%)) I t 2 equal to to(z), which
reduces the latkr to unity.
Nevertheless, the eonfsrmal invariant version is valuable in pointing out a
new direction, As we have b e n learning in high energy particle physics, nature
does not always seleet what we, in our ignorance, lvould judge to be the most
symmetrical and harmonious possibility. Perhaps the formal inva~anceunder
conformal transformations is broken in suck ai. way that tz, ma8sless, zero spin
particle does exist, IDespik the principle of noriloeality for mmsle-ss padicles,
one cannot object to such a psrlticle on experimental grounds if it interacts with
matter sufficiently more weakly than a ~ a v i t o n ,In order to realize this suggestion, we must add an additional contribution tro the Lagrange function that
eaeictively reverses the sign of the ~r derivative t e r n and msigns it an arbitrary
coeacient. That is not enough, however, for the tr field would still have no
.
can be done
source; it is necessary to destroy the combination R $- ~ t This
arbitrtzriily, but $he possibilities are illustrated by two elementaq stfkrnative
procedures: remove the g2 faetor that multiplies m 2 ; remove the g2 factor that
multiplies R. The firs%procedure gives a version of %hie?"
B-D theory. The
second one has the same practical csnsequenees, and seems somewhat simpler.
Xt is described below.
The modified Lagange function is
+-
+-
where t,, is the total stress tensor, adding to the matter contribution that af
the er field,
394
Fields
Chap. 3
The equatianv (3-11.143-145) are independent of the made1 used for matter,
provided the matter part of the L%grarrgefunction has been made conformally
invafisxrt by the local introduction of the a field, implying
which is reivritden in the sfstGed form by eliminating 1,. It is also obtained direetly
by applying the sction principle to the conformal response of the noninvariank
f i a ~ a n g efunction,
giving
-(-s)-"2a,[(-g)
CUK
l~ZB@uau@l]
= --@R -1.
14-a!
(3-17. 1.50)
The quickest
to draw the practical consequences of the modified theory
is by returning to the souree procedures of' Section 2-4, now supglemenkb by a
k r m referring do spin O particles. The respective spin (helicity) 2 and spin O
sources are appropriately normalized ss [ ~ / ( 1 a ) ]'j2Tp'and [ja~/(1 C X ) ] ' ~ ~ T
where the latter is bmed on the field 4(02 - 1). This produces
1
wz=--
EL:
2 14-ar
fdz)(drt.')[PP(z)D,(z- z")T,,(z"
- *T(x)l>+(z - zt)T(zt)
-+ +ffT(s)l>+(a: - zt)T(zf)f,
(3-17.151)
and the inkractian enerm with a fixed body of mass M, replacing (2-4.36,37), is
Under the eireumstanees trr << to@,the Newtonisn potential energy is retained,
along with the gravitational red shift. For light, with t k k = to', the deflection
and the slowing of the spwd of light are reduced by the factor 1/(1 $ a). In
diseus~ingperihefion precession, the klnetie e n e w correction factor L (22"/m)
isr hanged it~to1 [(l --- a)/(b a)](2T/w), which gives
3-3 7
+ a).
395
where the funetions r,,, s, s,,, IP are all of order unity which is to be judged here
on a logarithmic scale. Then, if we exhibit; only the scale f a t o r s on opposite
sides of the two field equations (3-17.1.1-3, 1441, they read
==
X,
398
Chap, 3
Fields
To be consistent with the purely time-dependent tensor field, the scalar field is
also of that character, a(t), Xf it is assumed Lhgt the matder stress tensor has
only the energy component, too = Pm, the field equations imply
where the dot cjtesignates time derivative. We shall be csnknt ta pick out a
particular solution:
= (t/T)"3,
~ ( i=
) (a/3)'j2 log (l/$@),
p, = 0,
(3-17.163)
- --
3-"t
397
of mszgnitude less than this value of p, bvhich is not inconsistent with the observ*
tional data. The only sensifive dependence on a oecurs in to, the time a t which
Lhe laws of physics bmonne quafitatively similar to those now prevailing, in that
c($)> 0, t >z to, To the e-xtent that there is evidence for the winbnanee of
these laws over a significant fraction of the age of the univerm, a is correspondingly bounded from above. The nominal value a = 0.06 gives to ~ o - ~ T .
The ease with which integer spin L s ~ a n g efunetions acquire general codoe8 not extend to psrtieles
ordinette invltriance by suitably introducing
of integer
spin, T o appreciate the difference let us folloui. the earlier wwk
fidd procedure, now using the apin 4 Lagrange function
++
t,. = ~ + ~ O b [ ~ , ( l l i ) ar.(l/i)a,lJ.
v
to form
g($,h ) = g($)f t,,hp'
(3- 17.1438)
(3-17.169)
(3-17. 17a)
(3- 17.172)
Chap. 3
and
(3-17,178)
g,.(z) = e,"(s)g.be!(z) = e;(z)@~.fr)
Regarding the fist form of the l a ~ equation
%
aa a, matfix prduet, vc-e infer the
dekrminantal relation
-B($) = (det
(3-17.179)
or
(-g(z)) ' l 2 = det @,.(X) = &(g).
(347.18C1;)
The response of
+(S)
3-1 7
399
where
L~TOL
= l"flb.
= ?02
(3-17.191)
(3-17.192)
in which
dwab
==t
-doba
lCa dlcb,
(3-17-193)
and therefore
L-'d,L
+iF,a,lebgab.
(3-17.194)
A fundamental mixed t e n ~ a is
r defined by the cornrnutrtLor
which has the character of an antisymmetricat Lensor of the seeond rank for
general coordinate transformations, and of an antisymmetl.iea1 tensor of the
second rank with respect to focal Lorenta transformations. A scalar in both
senses is construct4 by
e'"(z)ePb(z)Rpv.b(x)
= R(z),
(&17.200)
and provides the basis for a gravitational Lagrange function:
I n the weak field limit, where the linguistic distinction of %heindices is removed and
@"(X)c?i g@@- h p a ( % ) ,
(3-17.262)
Flslda
Chap, 3
ii
where
@aeb
"-
A. = e-'a.(eeL)
@cab
"k gbcha
- W.*
- @&BAG
-(nut
Q:,
-+
(3-17.m)
(3- 17.m7)
EH
3-1 7
401
This is recognized aa the statement that the covariant derivative of gf"" vanish=,
, : ' l with the quantities of (3-1 7.91), known usually a s Christoffel
and identifies
symbols. After this, it is abundantly clear %hat the two objects defined in.
(3- 17.124) and (3- 17..f99) are connected by
where the correctness of the algebraic sign can be verified in the weak field
limit* Thus, the two gravitational Lagange functions, $(g, l") and c(@,O)
are idemticd.
Returning do the spin $ Lagrange function (3-X7.181), we insert the eoordinate d e ~ v a t i v egeneralization stated in (3-17. 195) and obtain
where, i d should be noted, $he total antieommutativity of the field extracts the
antisymmetrical part of the matrices u0r*obC.This removes the terms with
a = b or a = c, trhieh are proportional to the symmetrical matrices roro.
Then, since
+ m]$,
(3-17.219)
where
*&
= aC*bcdWabe.
(3- 17.220)
The notation g($,e) might have been elaboraled as C(#, e, w). When this
structure i8 added to the gravitational Lagrange function (3-17,201), in whieh
eg and
are used as independent variables, the W,,& dependence of g(+,e, W )
produces an additional term in (3- 17.203), ~vhich removes the symmetry
property noted in (3-l"7.212). This is a natural form of the theory, But we shdl
prefer, far simplicity only, to regard a a b e as defined by (3-17.21 1) and therefore
not subject to redefirzition through the appearance of @,be in the matter Lagrange
function. I n order to identify the stress tensor t,, directly, we consider the
special variation
6eg = -$-ligPve,,,
(3-17.221)
die21 is consistent with the construction of gP"ronrr the e,: and gives
Chap, 3
and thus,
where the first two teirns can be united through the reintroduction of the spinar
eovariant derivative (3-17.f 95).
The sedar defived from this k n m r is
3-1 7
403
for then
z==
~ ' C x g),
t
(3-17.235)
shows the requird dynsannkal equivafeme of the field8 y,,(a), #(s) a d 51,,(2),
+(X),
Under weak field conditions we write
where the inva~ancepropedy (3-1 7.235), staged for infixritmimal f retnsforma%ions(2""= f l - &Y),requires that
a,
(dz"
where f"(z
- x')
)(dx")fIr
(3-17.239)
)is
aufc1(z - X') =
&(X
- a').
(3-17.XO)
Tbe matter strms kngor that is rtow dc?rived, slakd in. the absence of %he
pavitation field for simplicity, is the consewed objeek
(3-17.241)
Asmrning that the graviton detection sources do not overlap fhe K support re@on,one can use Lhe sourwfree, weak gr~viLtls1ionalfield equations (3-17.13,14)
La derive
40Ei
Alternativety, one might have b w n with the la& form, where the additional
k r m mrvw tO remove the responw of + ( X ) to infinitesimal coordinate transformation~.Similar discussions can be given for any other type of matter source
and fidd, wifh appmpriate &Lention to their transfornnation properties.
The weak field form of the gravitorr source term in the action is
(d.)
TRdnI"pCo> 4 (dx)TY(z)~lr.P(z)
P
--)
(3-1 7.246)
where
d J I T ~ ( x=
) 0,
Fields
Chap, 3
H. How Can it be the end of the bmkl You have haray hewn. There are
any numbw of addifionail topics X should like to 8% developed from the viewthe revimers,
point of source theory. And think of the field day you wiIl
who usually prefer to list all the mbjeets not included in a volume r ~ t h e rthan
discuss what it dws contain.
S. Quite tme. But we have now reslelned the point of tr&nsifionLo the next
dynamical fevd. And, since this volume is already- of ft re-mnable ~ize,and
many of the ideas of source thmfy are in it, if" hardly fully developed and applied,
it m m s better to put it before the public as the first valume of a s e ~ e s .Hopefully, the next volume will be prepsped in time to meet the pawing demand
for more Souree Thwry.
The first volunle was described as a resewch document, and a textboak. Unfortunately, the beginning student was given no guidelines to tell him into whicfi
category a particular section fell. hecordingly, here are some suggestions for a
first encounter with source the or!^, and relativistic quantum mechanics,
a) In Chapter I , omit Section 1-4.
b) In Section 2-1, the derivation of the Lorentz transformation behavior of the
Appendix
The nature of the system is not changed thereby since, on extending the action
vanishes, apart fram a pssible source
grinGiple to $, we learn that QI, term. But, on adding (A-1) and (A-21, the squares of the first derivativs
cancel, producing the Lagrange function (3-15.161, from which the fimt-order
field equations fallow, This procedure is the analope of one for ordinary
mechanics that begins wit h the quadratic Lagangian
Index
AeLiont 186, See aka Lerange function
additivity of, for noninteracting
particles, 256
for arbitrary spin, 191
coardixl~teinvariamt gravihn sourw
term in, 405
discontinuous ehange of, and eharge
quantigation, 242
electrom~gnetie,displacement charge
of, 248
with field strengths as variables, 340
modificsticln for point charge@,
%3-244
409
410
index
C, 49
Cauchy principal value, 46
Causal analysis of vacuum amplitude,
51-52, 109, 119
Causal control of H-particle transitions,
362
Causality, 313, 37
and unitarity, 61, 122
Causal structure of propagation functions,
42, 5& 559, l20
Center of mass, 35
G.G.8, syatem, iv
Chalcidian slababet, 234
Charge, 47-48
accelerated, and rdistian, 265
conservation of, 255
in interaction, 285
distribution of, 77
dynamical and kinernatical arspeet;gs, 2 s
eigenveetors of, 284
electric: %xisof, 250
conservation of, 72
twa-dimensional lattice far, 251-252
universality of unit of, 250
electric land magnetic, coexistence of,
242
exmctation V ~ U Z Irelation
,
ta flux.
vector, 200
and fermions, f 13
leptonic, 125
Iocalisation of, 243
magnetic, 231
unit of, 250
of multiparticle states, 53
purely magnetic, unit of, 252
Charge fluctuatim ftux vector, 207
marge Buetuation~and p h m
transformations, 207
Charge flux vector: local conservation
of, outside sources, 200
for spin O particles, 200
for spin particles, 204
for spin 1 particles, 203
for spin 2 particles, 204
for third-rgnk spinors, 206
Cbsrge matrix, 48
Charge quantization, 239
condition for, 249
Charge reflection mratrix, 49
Charge symmetry, 49
Christoffel symbols, 401
Clifford-Dirac algebra-, 105
Coherence in scattering, 358
Collisions, 37
caussll control of, 36
Commutator, 2
Complernentarity, 38
of source descriptions, 39
Complettlness, 41, 53, 58, 59, 121, 158
of H-partieb states, 348
for particles of arbitrary spin, 137
of spin particle states, 110
of spinors, 319
Complex conjugation :of Fermi-Dirac
sources, t 10
of sources, 131
Complex fields, 153
charge interpretation of, 1%
of time cycle description, 154
Components of a vector, 2
Composite systems, 344
consistency of phenomenologieal
description, 357
Compton scattering. See Scattering, of
photoas by charged particles
Conformal group, 225
and conmrvittion laws, 2226
for electromagnetic field, 230
represented by %otatians,2225
and stress scalar, 226
Conformal invariance of action for
massless spin 0 particles, 39%
Conformal transformations, 388-3W
Gonmrved current, construction of, 255
Constructive principles: of S-matrix
theory, 35
of source theory, 31
Contact terms, 144
and field digerential equations, If38
introduced by source redefinitions, 173
in multispinor description, f 77
for spin g particles, 176-177
for spin 3 particles, 171
Coordinrcte displacement, local Lorentl;
transformation induced by, 399
Coordinak invariance, general, and
particles of integer 4 spin, 397
Coordinate invariance, infinitesimal, of
spin O Lagrange function, 384
Coordinate systems: in @ace-time, 7
transformations between, 332
Correction factors for light phenomena,
and perihelion precession, 395
Coulomb gauge. See Rdiation gauge
Coulomb inleraction, 77
Coulomb pobntial, 320-321
Coulomb scattering : connection with
nonrelativistic limit, 323
and dynamical levels, 322-323
of spin O particles, 321
of spin particles, 322
Coulomb correction to, 324
phaw shift between helicity
transitions in, 327
spin dependence of, 328
Govariant derivative: of contravariant
vector, 386
412
Index
by realistie EX-partichs, 3%
Rutherford, 289
for scatbring, of circularly polarizled
photons by spin O charged
particles, 286
of linearly polarized photons by
spin 0 charged particles, 295
of spin 0 rtnd unpolarised spin 3
pmticles, 310
of win. Q particle by massive spin
9 psrticle, 31 t
of anpolarized spin 4 particle by
massive spin O particle, 31 X
far &pinO particle pair creation, by
cireularly polari~edphotons, 298
by Iineetriy polarized photons, 297
for spin O particle pair annihiltiltian
into circularly polari~edphotons,
299
for spin 0 particles, of like charge, 288
relation btween like and unlike
charges, 8 0
for f wo-photon annihilation of
unpalarized electron and positron,
317
unit of, 296
far unpolarizred etectron-positron
scattering at arbitrary energies,
309
for unpolariaed photon %tatbring by
f pin O efiarged psrticfes, B 4
Dif3Ferential equations, firstorder :for
electromagnetic field, 228
for srbitrary-rank multi~pinors,f 85
Index
413
82
Energy-monnentum : con~rvationof, 8I
expectation value, relation to flux
vector, 220
of multiparticle states, 53
prwxistence Bvikhin sources, 403
Energy-momentum eon~rvationand
kinemstical integrals, 288
Eneru-momentum flux vectar, 209
lac& conmrvation of, out~idesources,
210
Enmgy operator, 8
q u a t i o n of motion, proper Lime, 23
Euclidesn Green's function, 146
Euclidean postulate, 44
and arbitrary spin, 189
and *in 3 particles, 111
Euclidean propagation function, 44
inequalities for, 45
Euclidean space, attached Ito Minkowgki
space, 43
Exclusion prineipb, 109, 119
E w c t a t i o n values, fr13
Exbnded sources, 265
Exbriar algebra, 106
f@: cfsss of, 233
Field, electromagnetic, 26
Field equrttiona. See alr~oBiSerenthl
equeztions
of broken eonformal invariance theary,
393
Dirw, with electromagnetic
inbraetion, 262
nonlinear, 279
gravitational, 388-381)
MaxweU, with chsrged particle current,
257
in grltvitational field, 388
nonlinear, and physical proeessa,
B7-278
for spin 0 padiebs with prinaitivcs
ellectromagnetie interaclion, 256
for spin I. particles with interactions,
263
Fields, 145
msoeiabd wit h individual emission,
absorption act, 154
auxiliary, 171, 177, 180, 182, 183
causal evaluation of, 153
complex, X53
far spin. 3 particles, 15%160
of Lime cycle; description, 154
correlation funetions of, 30,32
electronnagnetic, 227
elimination of, 278
equd time commutation relations of, 32
I[",,x, of gravitons, 380
grsvitationd, 378
graviton, as wesk gravitational, 385
helieity decomposition of in massleas
limit, 197
of helicity 8 prtrtieles, differentid
equations for, 176
of helicity 3 psrticles, dif"ferclntia1
equations far, 169
H-particle, 351-352, 354
integral equation for, 365
Maxwel.1, conformal Lr%nsformationof,
390
multispinor, 1'7'7
WBhr, of ~ ~ B V ~ L O R S380
,
operator, 24
r e d and imaginary, 2CZS
retarded, 147
of Dirac equation, 162
spin 0, for causal arrangement, 284
of spin 0 particles, 145
spin 9, causal expressions far, 159
response to local Lorentz
transf ornnations, 398-399
of time eycle description, f 61
of spin particles, 157
differential equations for, 160
of spin 1 particles, 164
differential equations for, 164
of spin 8 psrticles, E72
difffsrential equations for, 173
of q i n 2 particles, 165
digerential equations for, l66
of spin g particles, 175
diEerentia1 equations for, 177
of ;spin 3 particles, 167
differential equations for, X72
tensor, of gravitons, 378-379
of time cyele description, 147, 152
of unstable H-particles, 369
Field strength Censor: d u d to, 228
electromagnetiic, 228
Fine stxrrcture constant, 250
Fluctuations, 64
Friedmann models, 396
g, for arbitrltry multispixlor description,
265
dependence an description, 262
for spin 3 particle, 280
for spin l particles, 264
Galiban relativity group, composition
properties of, 8
generator commutation rehtions, 9
Gawe conditions, 238, 258
on MftxweXl Green" ffunction, 279
Gauge covariant derivative, 262
Oauw invariance, B8
grwitalional, 380
Gsuge transformations, 228
Abelian group of, 262
grtrvitstional, 380
Cwa kinds of, 2363
Generators, 3
f l i p , 400
F;,, 385
l'$, construction of, 387
X",: transformation law of, 386
rak,
400
399
Gravitational red shift, 82
Gravitational slowing of light, 83
Graviton, 80
wtian for, 198
etdditional, ernitkd by graviton
source, 405
emission from nnatbr source, 404
helicity states of, 81
a parable, 81
vacuum probability amplitude for, 80
Grsviton source concept, 403
Grftviton source problem, 403
Green" function, 146. See also
Pragw8ticzn funet ion
advanced, 148-149
of Dirae equa;tion, 162
antiperiodic, 165
associated Euelidean, f 46
boundary condition for, 146
for charged spin Q particle, 271
for charged spin g particle, 218
Dirac, causal expressions for, 158
of Dirae equation, 157
Birac, momentum integrals for, 158
of Laplaeek equation, 260
Maxwell, olzeying gauge condition,
278-279
modified periodic, 155
periodic, 149
Fourier series construction, af, 151
retarded, 148
of Dirac equation, 162
%&b,
of neutfinos, 125
spin I states of, 69
spin 2 staks of, 79
Hermitisn adjoin%,1
Hermitisn g matrix reafigation, crihrian,
for, 6
Hermitisn operator@,1
infinibsirnetl, 2
Homogeneous electromagnetic field, and
spin O Crwn" function, 272
H-partides, 344
fields of, 351-352, 354
I' matrix, 3fi3.
ixthgrlat equation for fields of, 365
modifid propag~tionfunction, a(i3
persistence probedbility, 3Ci8
photon scattering by, 355
Il matrix, 365366
propagation function, time limitstion
on, 365
rdiative transitions btween, 350
skeletlllt irtterwtions with photons,
351,354
sources of, M?,349
unstable, fields of, 369
photon emission by, 368
time kbavior of propagation.
function, 368
s s unstable particles, 360-361
virtud, 362
as effective wurees, 365
Hubble e w s n i o n parameter, 396
Ido, Zchir6, 81
Infinitesimal rotiztions, response to, 9
1n6niLesimaI Lrsnslations, response to, 10
Infra-red catastrophe, 273
Interaction skeleton, 277
Interaction volume, 283
f nvariance transformations, 199, 209
Invariant flux, 287
InvarianL momentum apace measure, 31
f rreducible processes, 283
Isotropic dilations, 224
Jacobi identity, 3
Jordan, P,, 392
Klein-Nisfiina formula.
DiBerential
cross mction, for Compton
scatbring by electron a t rest
Lsgrange function, See also Action
arbitrariness of, 188
far srbitrary-rank spinar, 195
with broken confomal inva~ance,393
conformally invariant, of nnwive
pin O partides and gravilatiana1
field, 392
of electromagnetic field, 229
gauge invariancto of, 256
gravitational, 385, 399
of grsviton field, 381
of gravitons, response to gauge
t ransformations, 381-382
of interacting Dirac and Msxwell
fields, 261
for massless spin O particles, eonformal
transformation khabviar of, 391
of matter, arbitrariness in grsvitational
field, 391
modified, af spin O particles in
gravitational field, 391
of photons in gravitfttionai field, 389
for second-rank spinor, 191
for spin 0, 187
of spin O particles, in gravitational
fieid, 385
with primitive electromagnetic
interaction, 2565
with primitive gravitational
interaction, 383
for spin *, 191
conformaXly invariant form, 402
in gravitatianaI field, 401
in weak gravitational field, 397
for spin l , 188
for spin 1 particles with primitive
interaction, 263
for spin g, 191.
for spin 2, 189
for spin g, X91
for spin 3, 190
for third-rank spinor, 194
of rank 2, 177
relation to knsors, 195
of rsnk 3, 178
of rsnk 4, 180
of rank 5, l82
symmetry and spin, 185
Muon, 34
Neutrinos, 125
helieity of, 20
Neutron star, 403
Newtonian interaction, 82
Nonunitary transformations, 87
Operator field theory,iii, 24
consistency of, 28
dynamics in, 33
particles in, 30
speeulstion in, 34
and strong interactions, 34
Orbit, equation of, 85
Orthonormal spin-angle funelions, 115
Ortfxonormal vector functions, 70
P, 50
Pair creation by a photon in a Coulomb
field, 343
Pair creation by two photons for spin O
particles, 286-297
Parameter : Hubble, 396
of scalar-tensor theory, observations
concerning, 395
Parametric device for combining
denominators, 336
Parity, 50
for arbitrary spin, 141
intrinsic, 1 17
orbital, 117
spaee-reflection, 20, 95
for spin particle angule~rmomentum
states, 117
for spin % particles, 109
Particle fiux vector, 208
Particle occupation number, 52
average, 149, 155
of Fermi-Dirac system, 163
in Fermi-Dirsc statistics, 101)
Particles, 1
with arbitrary integer spin, 85
sources for, 92
causal source description of s beam of,
41-42
charged, two types of, 252:
composite, 36
statistics of, 252
creation of, 37
critique of theories of, 24
detection of, 38
w i t h dual charges, 252
elementary, 11, 22
executing prescribed motions, 245
of integer -/- spin, X27
and general coordinate invariance,
397
inberetcling, 12, 23
in a macroscopic environment, f 2
massless, 19
of arbitrary integer spin, 96-97
of helieity g, 129
af helicity 3, 99
of integer $ helicity, 133
in operator field theory, 30
phexlomenolagical theory of, 37
reactions of, 32
real, 267
source of, 37
space-time description of exchange of,
54-55, 118
of spin 0, 38
elzarged, fields associated with,
285
of spin *, 99
of spin 1, 67
of spin 2, 78
stable, 31
unstable, 12, 32, 360-361
virtual, 2437
Pauli matrices, 99
Perihelion precession, 83
in. scalar-tensor theory, 395
Periodic Green" function, 1149
causal forms of, 150
Fourier series construction of, 151
modified, 155
+-
Index
++
deffnition of source, 41
fields of, 145
primitive electromagnetic interaction
for, 256
propagation funetian of, 42-43
scattering, 284
of like chargesj 285-.286
of opposite cbsrges, 289
of photons by, 290-291
strsng sources for, 50
Spin matrices: as, 127
a",U
, = 0, , . ,3, 101
a,,@ = 1 ,..., 4,112
eigenvectors of: u, 107
eigenvectors of: v, 106
T 5 , 104
Y", X03
F,, 103
rat, 133
a,,, 104
@&*E, 112
387
diftFerential identity for, 388
B,,k, 390
@p,,
R,,, 388
a,&,
399
R & ~ 386
*A~
~trese,t,,, 388
symmetrical:
t,,, 3E)8
t;, 398
tracelessr, 83
Tensor-spinor, 132
Tetrad vector field. See also Gn-tvitatio~tt~X
field, et:
conformal behavior of, 402
Thornson erorss wetian, %4, 320, 356
Tints cycle description, 62
fieids of, 147, 161
for parsmetrised multiparticle stabs,
1511, 1516
+-