Professional Documents
Culture Documents
Feynman, Richard - Photon-Hadron Interactions (Westview, 1998)
Feynman, Richard - Photon-Hadron Interactions (Westview, 1998)
INTERACTIONS
PttOTON--HADRON
INTERACTIONS
RICHARD P. FEYNMAN
late, California Institute of Technology
�e�'o/.i�
.. . � ...... · ·
2 3 456 78 910
First printing, February 1998
Addison-Wesley'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 clari
ty. Over time, it was expected that these informal accounts would be replaced
by more formal counterparts-textbooks or monographs-as the cutting-edge
topics they treated gradually became integrated into the body of physics knowl
edge and reader interest dwindled. However, this has not proven to be the case
for a number 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 notes on Richard Feynman's lectures at Caltech on the topic of pho
ton-hadron interactions, in which he developed his theory of partons, were first
published some twenty-five years ago. As is the case with all of Feynman's lec
tures, the presentation in this work reflects his deep physical insight, the fresh
ness and originality of his approach to understanding physics, and his overall
pedagogical wizardry. As a result, this volume will always be of fundamental
v
vi Editor's Foreword
David Pines
Urbana, Illinois
December 1997
Special Preface
Many of us were first introduced to the concepts of the parton model from this book. At
that time the competing view was one in which there were no elementary particles. Every
particle was supposed to be a composite of every other particle. The ideas and concepts
in this book have helped pave the way for our understanding of the constituent nature of
hadrons which eventually led to the Quantum Chromodynamic (QCD) theory of quarks
and gluons. As is true of most of Feynman's books, the maximum benefit is obtained if
one has previously studied the subject in some detail. Feynman's unique perspective can
best be appreciated by readers with a solid background in the subject.
Although this book is almost 18 years old, it still is an excellent reference. It
appears on the recommended reading list of all the current QCD books. The boo k provides
a good understanding of the parton model frt>m the man who invented it. In the "pre-QCD"
or "naive" parton model the constituents within hadrons were assumed to be bounded in
the transverse direction. The probability of finding a parton within a high momentwn
hadron with a large transverse momentum was assumed to fall like a Gaussian or an
exponential. QCD tells us that this is not exactly true and gives a power law fall-off in the
transverse momentum. Because of this, many "naive" parton model expectations are
modified (in an important way) by logarithmic factors. Feynman used to laugh when his
parton model was referred to as "naive," and he would say, "At least I got it right up to
logarithms." We all miss Feynman very much and it is through books like this that his
ideas live on.
R.D. Field
March, 1989
vii
Vita
Richard P. Feynman
Born in 1918inBrooklyn,Richard P.Feynmanreceivedhis Ph.D. from Princeton in 1942.
Despite his youth, he played an important part in the Manhattan Project at Los Alamos
during World War II. Subsequently, he taught at Cornell and at the California Institute
of Technology. In 1965 he received the Nobel Prize in Physics, along with Sin-Itero
Tomanaga and Julian Schwinger, for his work in quantum electrodynamics.
Dr. Feynman won his Nobel Prize for successfully resolving problems with the
theory of quantum electrodynamics. He also created a mathematical theory which
accounts for the phenomenon of superfluidity in liquid helium. Thereafter, with Murray
Gell-Mann, he did fundamental work in the area of weak interactions such as beta decay.
In later years Feynman played a key role in the development of quark theory by putting
foiward his parton model of high energy proton collision processes.
Beyond these achievements, Dr. Feynman introduced basic new computational
techniques and notations into physics, above all, the ubiquitous Feynman diagrams which,
perhaps more than any other formalism in recent scientific history, have changed the way
in which basic physical processes are conceptualized and calculated.
Feynman was a remarkably effective educator. Of all his numerous awards, he
was especially proud of the Oersted Medal for Teaching which he won in 1972. The
Feynman Lectures on Physics, originally published in 1963, were described by a reviewer
in Scientific American as "tough, but nourishing and full of flavor. After 25 years it is the
guide for teachers and for the best of beginning students." In order to increase the
understanding of physics among the lay public, Dr. Feynman wrote The Character of
Physical Law and Q.ED.: The Strange Theory ofLight and Matter. He also authored a
number of advanced publications that have become classic references and textbooks for
researchers and students.
Richard Feynman died on February 15, 1988.
viii
Contents
Editor's Foreword v
Special Preface vii
Vita viii
Preface xiv
ix
Note 38
In Lab 115
Electromagnetic Form Factors (continued) 116
Pion Fonn Factor 118
Proton Fonn Factor for Positive q2 121
Note 122
APPENDIX A 271
APPENDIX B 276
INDEX 279
Preface
The most advanced course in graduate theoretical physics at Caltech is "Special Topics
in Theoretical Physics." Each year the professor chooses the topic with which he will deal.
This year (1971-72), having just come back from the 1971 International Symposium on
Electron and Photon Interactions at High Energies, held at Cornell University. my own
interest in the subject was aroused, and I chose to analyze the various theoretical questions
related to that conference. The lectures themselves became so extensive that the decision
was made to put them into book form, with the thought that other people might also be
interested. Thus, the report of the Cornell conference should be considered as a
companion volume to these lecture notes. The references given here are far from
complete, but a full list of references is given in the Proceedings of the Symposium,
published by the Laboratory of Nuclear Studies, Cornell University, January 1972.
The material is dealt with on an advanced level; for instance, knowledge of the
theory of hadron-hadron interactions is assumed. I have tried to analyze in detail where
we stand theoretically today. The treatment is somewhat uneven; for example, I should
have liked to study the theory of the decay of the in more detail than I was able to do.
On the other hand, there are long discussions of vector meson dominance and of deep
inelastic scattering. The possible consequences of the parton model are fully discussed.
Time did not permit me to complete the original plan which was to include the
theory of weak interaction currents which are so closely related to electromagnetic
currents.
Many thanks must go to Arturo Cisneros who edited, corrected, and extended the
lectures from my class notes. Without his effort, this book would not have been possible.
I also wish to thank Mrs. Helen Tuck for typing the lecture notes.
Richard P. Feynman
Pasadena, California
Summer 1972
xiv
Photon-Hadron
Interactions
General Theoretical Background
Lecture 1
We are assuming that we do know the photon. QED has been checked so
closely that we know that if the photon propagator were off by a factor of
the form ( 1 - q2 /A2)-l then A exceeds 4 or 5 GeV. The amplitudes are known
to about for q2 as high as (1 GeV)2• For the rest of this course we
5%
shall assume QED is exact. There is already evidence, as we shall see, that
3
4 Photo11-Hadro11 lntc'ractions
The S matrix is the transformation matrix from the "in" representation to the
"out" representation
The state I n , in> means a state which far i n the past i s asymp totically
free stab le hadrons (stab le in strong interactions only , e . g . 11° is "stab le")
described by momenta, and helicities all contained via the index n . The state
m, out has the set of indices m with the same space of indices , but represents
a state which in the future is asymptotically in si tuation m.
Thus the S matrix is really the uni t matrix but in a mixed represen t at i on,
a different labeling for incoming and outgoing states . Supposing these states
are all there are , conservation of probability requires
(i.e. , ( 1 . 3)
(In the special case that the state n represents a single stab le particle the
in and out states are the same).
General Theoretical Background
m
diagram: M
N n
The electron-photon system goes from state N to M, the hadrons from n in to
m out . We suppose the only interaction possible is by the exchange of a
photon - and this pho ton is characterized by a polarizationµ , momentum q:
This is done by removing from the measured amplitude the known (by QED theory)
factors
(1 .6)
so we now write
6 Photon-Hadron Interactions
(1. 7)
4ne- 2
To deal with this a li t tle more abs tractly , the quantity j • - can , in
u q 2
QED be described as the matrix element (between the leptons ) o f an operator ,
the vector potential operator for the leptons a (q) , all cons idered known
ll
and compatible by QED in any specific case, Thus the firs t order interaction
The 1 comes from the zero order (where we saw the S matrix is really the
f
I au (q) Ju (q) d q +)
4 4
(1 - i a+ (q) J+ (q) d q +·� (1 + i - 1 (1.8)
u u
Lecture 2
Conservation of Current .
ll II
Thus one component of a ( the one in direction q ) is always missing (unless
General Theoretical Background 7
polarizat ion e the coup ling is e / (q) - but for a free photon e is undefined -
ll ll ll
to it could be added aq\J (e � = e
ll
+ aqll) any a. This can make no e f fe ct s o
physical property Jll must satisfy . If it is not true for general q2 redefine
G �
s ince it is a local operator field theory sugges ts that i f hadrons are governed
( l ) , J ,,( 2 = 0 if 1
�
by an underlying field theory then and 2 are space
\J
like separated (symboli zed by l • 2 ). I f you wish you may assume this -
but it is very in teresting that we can prove it from our as s umption that the
s t rong system interacts with QED (subject to sys tematic errors in proof due
to taci t as sumptions . This may not be an important poin t but it is interes t ing
This is via diagrams of the type whose ampli tude depends on a computable
factor from the leptons* times a ma trix element depending on the two momen ta
and polari zat ions of the two virtual pho tons -1 / 2 Vil" (q1, q ). As defined
2
q1 ++ v,
<-+
this i s symm e trical in q , ll for Bose s tatistics among pho tons
2
does not permit us to distinguish the photons so no other function can b e
( 2 .1)
\J ll+
*
in part I of th is course "lep tons " will mean e , e
+, , only .
8 Photo11-Hadro11 l11ternctii111s
The contribution from the QED lepton side can be figured for each diagram ,
and can , as shown in QED theory , be generally wri t ten as the matrix e lement
(be tween in and out photon-lept on states) of the time ordered product of the
to restrict the integral to t > t and write this and the first two orde rs as
1 2
T • 1 - i J J (l) a / 1 ) dt -
µ 1
t
J>� v v
µ
(1, 2) a (l ) a) 2 ) dt dt
µ 1 2
(2 . 2 )
1 2
order .
- f> J.
t
1
t
2
v \JV \I
( 1 , 2) a ( 1 ) a ( 2 ) dt
V 1
dt
2
• o (2 .3)
+
ob tained by writing T T • 1 and expanding to 2 n d order . Th e 2nd integral
is over all t , t •
1 2
It can be spl i t into a part t
1
> t
2
and a part t
2
< t
1
in the lat ter relab le variab les 1 , 2 (and µ , v ) to get
Now , i f we could assume that wi thin the range of possib le QED states arb i trary
v ( 1 , 2) - J ( l ) J v ( 2 ) for t l > t
2
(2 . 5 )
µv \J
But outside the light cone for example � (2) , a(l � • 0 so the two products
are not independent , they are equal . To proceed therefore more carefully
General Theoretical Background 9
write
Now the first factor must vanish, for we could take the case that a(2), a(l)
coDDDUte first (for example one photon from an electron, another from a muon in
lowest order) hence we surely always must have
determining the real part of v (l, 2). In addition in general we must have
11v
(taking adjoint of last term)
The commutator is zero outside the light cone. Inside I believe we can make
it arbitrary (although some little further study of special cases is necessary
to verify this) hence we deduce
if (2 .9)
underlying field theory, for then our picture of coupling if t1 > t2 can
Leptons
be cut at a t between t and t1; the first coupling is J11(1) and the second
2
10 Photon-Hadron Interactions
is J {2}, so we get the product . But one may be averse to assuming that
v
s trong interactions can be described by a complete set of states (and that the
complete set can be taken as In, in> } at any arbitrary time . Nevertheless
if we continue our s tudy of the raquirements for consis tency with QED to
4 th order we can do it .
'nlis point is not important , but we do include the proof for completeness .
If ( 2 . 5 } were right then T would be {exp - i JJ11 (1} a\J (l} dt1 } T . In general
way with the position indices}. 'nle first order is made to agree with J (l}
\J
so in general we can write (U(l2 } represents the deviation of V ( 12 ) from
JJ
J �t 2 U ( l , 2 ) a (l} a (2 )
-i (1) a (1) d t 1 [';
T • e 11 [ + i
�
t
l
+ J t f t JU (l 2 3) a (l} a (2 } a ( 3 }
�
t
1 z � 3
+
Now in forming T T to check unitarity the exp -i /J a factors go out. To 2nd
and 4th order in a the U (1 2 3) term does not enter. To second order we have
- � +
u (l 2) a(2)•a(l} + J uc1 2) a(l} a(2} • o
as before .
+ �U ( 5 6)
t > t
U (7 8) a (5 } a (6 } a (7} a (S) • 0 (2 .11}
s 6
t > ts
7
General Theoretical Background 11
In the last integral we have t5 > t6 and t7 > t8 but no particular relation
of t5 and t7 there are 6 relative orders. We replace in each the variables
via t1, t2, t3, t in that order
4
1 5 5 5 7 7 7 call a(l) •A
2 6 7 7 5 5 8 a(2) • B
3 7 6 8 6 8 5 a(J) •c
4 8 8 6 8 6 6 a(4) • D.
U(l4) U(23) a(l) a(4) a(2) a(3) ADBC. A[p, �c +AB[!>, c] +ABCD
U(23) U(l4) a(2) a(3) a(l) a(4) BCAD s[C, .go + (!, �CD
• +ABCD
U(24) U(l3) a(2) a(4) a(l) a(3) BDAC. s[!>, .gc + �. �DC + AB[n, cJ +ABCD
U(34) U(l2) a(3) a(4) a(l) a(2) CDAB. c[n, A]B + [E. �DB +Ac(E, s] +
+A@, �D +ABCD (2 .12)
Now we get many obvious relations. For example the coefficient of ABCD must
be zero (take case all 4 potentials commute). We believe that the vector
potential a is an arbitrary function of space and time. We can therefore
choose it to be different from zero only in four small regions of space time
around the points 1, 2,3 and 4; call these regions o1, o2, o3, o4• For
our special interest here take the case that the variables have the following
light cone properties
[A, c] • o • [A, s]
[s, ] n • o • (n, cJ
Only [}J, c] and [A, � </- O
12 Photon-Hadron Interactions
1 is outside the light cone of 2 , 3 is outside the light cone of 4 . Omi tting
the ABCD term which we noted was independently zero and collecting what is
left in this case we get
Only the third term needs to be turned around to [!>, ajcB+ [!>, £1 [!, SJ then
all the tel"lll8 are coefficients of @, !JAD or CB(!?, A] and must all vanish.
But ultimately we are left with U (24) U(l3) [!>, A'] Cs, � or finally
U (24) U (l3) G � �
<2> , a ( 3 . (4) , a ( l � dT l dT2 dT dT4. 0
3
(2.15)
Since the commutators are sufficiently general , I think we can conclude the
integral will be zero only if the integrand is and U(l3) • 0 even if 1 is
outside the light cone of 3 .
End of Proof
We can therefore conclude
(2 . 16)
� Cx1 , ]
t 1 ) , a (i 2 , t 2 > are sufficiently general functions of 1 and 2 (when
2 is in the light cone of 1 ) . We believe this to be true . It is left to
those interes ted in more rigorous proofs to verify this , for example by
trying to cons truct a basis .
Lecture 3
(3 . 1 )
General Theoretical Background 13
furthermore
for (3. 2)
The last term comes because there could be a 6(t1 - t2) term, or by relativity
4
a 0 (1-2) term or gradients thereof--leading to just a constant or polynomial
under Fourier transform.
{ -}
These seagull terms would mean that the abstract form for T would be like
e
-i [fJµ(l)aµ(l)dT + Jsµv(l)aµ(l) av(l) dT1 + ··
T
thus adding a local term at one space time point, but second order in a(l)
as appears in QED for the interaction with a scalar particle, for example.
Of course, instead it could contain gradients, as F v(l) F (l) for example.
µ µv
{ lf f J}
There can be higher terms for higher orders. In short what we have found is
that T must be expressible in the form Tex i L(l)dT1 • where L(l) is
T
an operator depending on a(l) only, for example of form J (l) a (l) +
µ µ
+s v (1) a (1) a (1) + where J , S are operators in the hadron
µ µ v
variables. Such a form is of course, also, the inunediate result of
supposing a local field theory for hadrons.
Research Problem
What experiments could best establish existence or non-existence
of seagulls?
In QED for spin 1/2 there is no seagull, for spin 0 there is (but
with Kemmer Duffin matrices there is not - resolve this!) Since all
quantities are defined by experiment the reality of such seagulls for
hadrons is an experimentally determinable fact.
Thus we see a knowledge of matrix elements of Jµ alone should determine
all scattering amplitudes V.
some special technical difficultiea coming from the highly divergent nature
(of, for example v v) and in no other problem and so they can be avoided beet
11
by disregarding them. ('Ibey have been analyzed by Schwinger, and are called
Conservation of Current
II II II
a (1) completely, but a gradient a (1) +v x(l) (x(l) is an arbitrary function,
{(
l
J (ai1m
T
1- i J11C1> +911 xm ( )) dt - i J v11vu2> a11c1> +
)}
1
or to firat order ia x
0 (3.5)
' l 11(l) •
v < v o (3.6)
1>
II )IV
Cl 2) • etc.
I
Equation (3.5) we have already discussed; (3.6) give• aomethiag new. Uaiag (3.3)
�Lii �J
4
9 (1, 2). v
II )IV 11 V )IV
v (1) J (2 +' 11 •••gulls a c1, 2)
T
'11t11
� �J IvLii �J
but
0
II
(1) J (2 • J (1) J (2 •
T T
v v
(3.5;
II
by equation except that' dou not comute with the time ordering
operation in
lLii (1) J c2
v
�J T
• eCt
1 - t
2
> J Cl> J (2) + eCt -t > Jv (2) J (1).
II v 2 1 11
Vµ Vµ/l, 2) • f µ 3µ (1 , Jµ (2
:
} + O(t1-t ) �0(1), J/28 +
2
+ vµ seagulls µv 6 (1-2) •
6 3 Cx-1-x-
or
[J (1), J �
0 v
(2)
t •t
1 2
• -Vµ seagulls µ v
2
) (3 . 7)
Actually I have simplified the discussion of the seagulls, for gradients also
appear on the 6 4(1, 2), but the point we want to make is that the equal t ime
(by analogy to QED where the coupling is purelyJ�A µ y µ �dT to a spin 1/2 field
[J (1), J (2�
0 v
�t •t
• 0 ( 3.8)
1
2
we have no direct test of this yet (,although we do have tests of 'ii'µ Vµv(l, 2) • O ) .
Remark
�<i1>.J<i2� -o
s t a te opera t or) or
pHp - ppH • 0
Now take the vacuum expectation value. Let the energy of the inte rme d i ate
16 Photon-Hadron Interactions
but p0 •
n p!0 and if the vacuum is the lowest state we must have E
n
- E0 > 0 so
which is impossible.
seagulls in the original field operator for the lagrangian. It comes becaUBe
we do not really compare thia formal field theory directly to experiment but
remove some divergent vacuum diagr81DB at the beginning. This problem appears
entirely aaaociated with the vacuum problem and could be removed. We can indeed
have the analogy of QED; 'no seagulls' and have equation (3,8) satisfied for
every problem except the vacuum. The precise statement is that ( 3 . 8) holds
matrix.
Lecture 4
interacting system above). The J matrix may have elements between different
t0
values of the quantum numbers, but it must of course conserve charge.
fast enough) we conclude that weak interactions are involved here. Thus we
Among sets of different isospin we can describe the result by saying J has
different multiplets. Since J does not change charge it must only involve
the 3rd component, if it is isovector for example. The fact that proton and
General Theoretical Background 17
independent of isospin, nor pure isovector (where the c harges would have to be
to require I • 2, but I do not know how precisely or exten s i vely this has been
tested. Recently some evidence was claimed for the need of an I • 2 compone nt
in comparing y p -+ 11 +N and yN .,. 11 - p at energies near the A reson an c e --b ut it
appears that corrections for deuteron structure ( for the y neu tron rate is
inferred from yD data) were incorrectly analyzed.
this respect at least because the strong coupling conserves this I spin
character .)
Having available matrix elements <m j J J n> for a variety of s t ates n (and m)
µ
all of the s ame I-spin multiplet permits one by appropriate linear combination
always to isolate the pieces due to the isos calar and isovecto r part separately.
Thus we can de f i ne matrix eleme nts and therefore op erato r s for J5(q) and JVJ(q).
µ µ
But for the vector we could alao ca lcu l a t e (via Clebscb Gordan coefficients}
think some of these currents are physically important also. We think the
v+
current J (q) is the nonstrangeness changing nonparity violating part of
µ
weak interaction (an assumption known as CVC). This leads to a suggestion
by which these extensions of current are useful in a powe r ful theoretical way
(Gell-Mann).
That QED is coupled to J ' or weak interaction to JV+ is (in our point
µ µ
of view in these lectures )akind of accident irrelevant to s t rong interactions.
They just lead us to a tool to study hadrons, but hadron in terac t ions can be
hadrons, or that hadrons are such that weak perturbation fields could be coupled
(we will use the latter hypothesis), We can expect, for any two points and
components of current J:(l) and J� (2) that they co111Dute if 1, 2 are space
like separated
(4. 1)
2X
new assumption,
bow far is it possible to go to prove say that the isovector par t, or the
�: (1)' J: (2B etc. commute? In the realm of isospin what we assume here is that
the space-like commutation law is true not only for the total current
J S + JV 3 but also for the isoscalar part alone with itself, the isovector
part alone with itself, the isoscalar part with the isovector part and
directions.
That is, extend the concept of vector potential a (1) to contain another
11
index a, the type, carrying for example isospin or strangeness. Then the coupling
T • 1 - i J J= (l) a:(l) dT -
1 t JJv:�(l 2) { a:(l) a�(2 � T
dT1 dT
2 +
Thus Vab (l ,
\IV
2) • { a
\I
b
J (l)• J ( 2 )
V
l +
1r seagulls
ab
\IV
4
S Cl-2)
(4.2)
(4. 3)
the case if I spin which we know is exactly conserved. Consider the scatteri ng
of an I • 1 particle via J� +
. The charge of the final state is one higher
(4.4)
General Theoretical Background 19
x'>
(4 . 5 )
but the las t , equal time commutator is by hypothesis ( 4 . 1) zero outside the
3
light cone�it mus t be a multiple of 6 (x-x ' ) , say s ( x) 6 3 ( x-x ' ) . Equation (4 . 1)
v+ -
shows s mus t be J (x) . Hence the equal time commutation relation results ,
µ
(4. 6 )
(this assumes there are no special terms in � (x1 , x2 ) which would integrate
out , a question we shall see related to seagulls again) .
Equation (4. 6) and its generalization to the much wider group su 3 x su 3
are Gell-Mann ' s equal time commutator relations . They represent the first
guessed dynamical property of hadrons that is not simply a consequence of
relativistic quantum mechanics general principles .
We can also describe this from the point of view of a property of the
scattering function y&bv , Take the case one potential is + isospin , the
µ
other 1 3 • Since current is conserved you might at firs t think V µ l Vµ3+ (1 , 2) • O ,
v
but the error is that the potential v, 2 carries in a charge + . So only if the
electromagnetic potential (coupling 3) also couples to the + meson is charge
conserved . E . g . diagrams are like
L
�
a•
// I y
(4 . 7)
t'
J
> (l) . J+(2 � -
v 1T
{ 1
µL µ
v/�(l) , J:(2
T
+ 6(tl -t2 ) �o(tl , xl) , <(2� t •t (4 . 8)
l 2
so writing 6 4(1-2) on the right side of (4.7) or 6(t1-t ) 6 3(x 1-x2 ) we obtain
2
the result (4 . 6 ) since 'ii J3 O . In general seagull terms must be added , but
=
µ µ
we do not know if they exist.
The generalization to a general Lie group with generators Ga with
commutation relation ( Ga , Gb ) • • fcab
c Gc ( • Ga•b by definition) (a•b)
f ab c
is
(4 . 9 )
(4 . 10 )
These may b e obtained from noting that the generalization of a gauge transformation
a� a + VµX is , in the group a: -> a: + V x a + (x l< a ) a. Supposing T [a) is
->
µ µ µ
unchanged by such a transformation we find
T �+ Vx + x )( � - T [a]
ns(l) . Then
f
VµV:� (l, 2) • 6(t l-t 2 ) �:. J�J t 1• t2 + a (l) J� (2 T } (4.12)
Now the latter presumably does not contain a singularity as strong as 6 4 (1-2)
but if su3 is valid at small enough distances�let us say the 6 4(1-2) singularity
ofVµv8µbv is correctly given by (4. 10) , Thus we say
V •
µv8b (l 2) 6 4 (1-2) Jaxb (2) + "smooth"
11\1
•
\)
(4.1 3 )
where "smooth" is less singular than 6 4 (1-Z) . Then we can still deduce the
equal time co1111u11 tator relations under the above assumptions of smoothness of
the su3 violating term. Equating the singular terms in (4.12) and (4.12)
we find
(4.14)
(seagull terms have been ignored) .
These relations are of very great interest because they are nonlinear
requiring absolute scales, Thus (if valid) they can serve as supplying
absolute scale definitions to the currents so that the rule that weak
interaction of hadrons is + A (rather than .7A) is definable and
V V -
therefore testable. This particular teat has been made by Adler and
Weisberger using PCAC to take the pion coupling as a measure of the divergence
of the axial current. We discuss how somewhat more direct tests can be made
by neutrino scattering later on in the course (in Part II) .
Singularities on the Light Cone
The commutator �µ (1) , Jv (Z� is zero if 1 is outside the light cone of
22 Photon-Hadron Interactions
through , or near the ligh t cone , For free fie lds o f any mass in field
2
theory the commutator has a 6 (s ) type s ingulari ty across the cone . There
12
� ]
is experimental evidence from inelas tic s cat tering experimen ts o f electrons
time we shall also develop further other formal properties of the commutator
in the air , I shall illus trate one applicat ion of i t --the s imp les t , the vacuum
Its fourier trans form, into variab le q --which we will call Vµv (q) is needed ,
2
for example , to calculate vacuum po lari zation correct ions ( t o order e ) due
q \J qV
If we wri te ( 6 µ b we see b mus t go to zero as q + 0 in order to avoid
v - -2 -)
q
a new pole at q2 • O. Ac ting on conserved currents the las t term vanishes .
+ . . .
+
2 2 2 2 2 2
• s
4ne i
av + s
4ne i ,
q 2v (q 2 ) 4ne i
av + s
\l
4ne i
q 2v 4ne i
q 2v 411e i
-2- s v . - .
\l
-- -2- -2- -2-
\J
--2 - 2-
q q q q q q
• 4ne
2 1
s �--=----- s v ( 4 . 16 )
\J
q 2 ( l-411e 2 iv (q 2 ) )
General Theoretical Background 23
We note there is no mass renormalization of2 the photon, the pole is still at
q2 • 0 , but the residue is altered to 4l-4we •
"e 2 where a v(O) (which may
ia
we
turn out infinite) is lost in charge renormalization. If start expanding
near q2 • 0 to get v(q2 ) a + q� , we can write the renormalized propagator
•
to first order in e 2 as
1
The imaginary part of v(q2 ) for q2 > 0 is the "virtual photon lifetime"
and gives the rate of production of hadrons in (say) an electron-position
collision. Because the imaginary part of the amplitude represents a loss in
s
probability that a photon remains photon
*
".JVV' + • 1 • 1
..
"-"0"""' + 411e2 iv(q2 ) i.e. Prob . + 411e 2 i(v-v ) •
Lecture 5
e+ + e -+ Any Hadrons
Consider the process e+ + e - + hadrons in state m out. It is governed
by an amplitude
Hadrons
. x.
( S . 1)
24 Photon-Hadron Interactions
2
l <m I J (q) I O> l • <O I J (-q) i m > <m I J (q) I O >
out \J \J out out \J
-
lbus if we could measure the total cross section for e+ + e + hadrons in any
(q
0
• 2E , Q- • 0 , q2 • 4E ) we can directly measure Em
2
O Ut
<O I J (-q) I m
)J OUt
> <m
OUt
l J (q) I O >
\J
(qµ q - q2 o µ > e.(q0)p(q2) because we know that if q < 0 (the vacuum state
v v 0
q > 2 1.2 + m Q2 /4 or q2 > (2 m )2 and q > o. ) Thus p(q2) exists only for
0 11 11 0
zero if q2 < ( 2 m )2 •
11
Therefore
F. T .
<O I J Cl)J c2 > i o> •
\J
(5 . 3)
v
F.T.
2 2
<O I J (2 ) J (l) j o> • <O j J (q)J (-q) I O> • (q q - o q ) e(-q0)p(q )
\J v \J V \J V \JV
( 5 . 4)
Now we can work out the commutator and time ordered product. Firs t the commutator
r· 2 2
o (q2 -m2 )p(m )dm • (5 . 6 )
(2m )2
11
General Theoretical Background 25
�
jJ
1V 1 - v.12 ) !"' 2dm2 p (m2 )
V
6
I.JV
· ..Ill
i; (1,2) , (5 . 7 )
(2111,,)
which is zero outside the light cone , because C111 ( 1,2) is . It is interes ting
that the proof that the coDDDu tator vanishes outside the light cone , assuming
relativity is so simple for the vacuum expectation . (At the end of these notes
the coDDDu tator , p ropagator and their Fourier t rans forms are listed) . To get
the time ordered product we need
<O ! { Jv ( l ) J
ll
(2) }T I O>�<o� ( t 1 -t 2 ) J v
(l)J
µ
(2) + l (t 2 -t 1 ) J µ (2 ) Jv (l) i o>
we 8(t1-t2 )
f e (t ) e+i Cq0 t-Q0x> d 3xdt (2n) 3 3 (Q)
To get the F. T. need a convolution with the F . T . of
o 0!ic •
q
f8 (-t ) e+i(q0t-Q0x> d 3xdt (2n) 3o 3 (Q) •
-i
q0+it
1
8 ( -q0' )) (q 'q' q' 2 )p(q' 2 ) dq0'
IJ V
- 6
I.JV
- v
I.JV
(q) ( 5 . 8)
The integral is j us t
)dm
2 2
v
tt (q ) (q q t - 0 t tq2 ) J p(q2m 2
• t
-m +i t
-
iS
tt
(5 .9)
guess (generali zing µv from tt--however see no te below) that wri ting
...
J(2m
2 2
p (m ) dm
-i s
)2 q -m
2 2 (5 . 10 )
+i£
11
wri te
iv (q )
2
• - -
11
l �v(q2 ))
Im
J q2-m2+i£ at m2
dm2 + S (5 . 11)
This is a dispersion relat ion for v(q2) , express ing v for all q2 in terms of
(5 . 13)
The Lamb shi f t corre ction (or correction to the magnet i c moment o f the e lect ron ,
e t c . ) due to hadron vacuum polari zat ion depends on 4ne2iv ' (O) which is
l
J
a (s ) ds
w -s--
I t is expected today for reasons to be dis cussed later that a (s ) may , for
large s vary as cons tant/a so the integral would converge and be determined
experimentally .
value o f q2 (say O or "' ) we can convert to a sub tracted dispe rs ion rela t i on ,
e en
1
same trick helps i f the i nt e gral on l o ok s div
2
S does no t appear . This m rg t.
General Theoretical Background 27
would write a dispersion relation for the quan tity v (q2 ) - v (O ) , the only quan tity
of physical significance
i 2
v(q 2 ) - v (O) • �
211 J�
(2m > 2
(5 . 14)
1T
n.us we may look forward in the near future (when experimen tal results for
of the <P resonance to p(m2 ) has been observed . 'This comes from the interference
of the graphs
, '
+ µ+ · y µ
2 2
X
e+ e e
+
A
Hadrons
at a q near m . In the first order in which the effect contributes only the
,
real part of i (v (q2 ) -v(o)) is seen, this is observed as a slight oscillation
2
in a near q2 • m,.
- -
As an example we calculate the modification to cr(e+e ..,. µ+ µ ) when "Hadrons"
Substituting in to (5.12 )
c dm2
( 5 . 16 )
2
� m2 -m;) 2+r 2/4 ) (m2 -q -i E ) m2
Im (
i (v(q2 ) - v(o)) ) ( 5 .17)
Re (
i (v (q2 ) - v(o)) ) --!;;-
2
p r
J_
�
c
(5 .18)
1lie lower limi t in the integral ( 5 . 18) can be set to -"' with negli gib le
error because the resonance is very narrow . Carrying ou t the integral we find :
(5 . 19 )
(5 . 20 )
where o0 i s the cross section for the process i n the ab sence of � -pho ton
interactions .
iv(q2) .
We need not have evaluated the in tegral in (5 . 1 8) to get the real part
2
rm (iv(q2))
of What the relation (5 . 15 ) says is tha t iv ( q ) i s an analy t i c
• } p (m2 ) .
( )
function wi th n o poles i n the lower half plane and such that
Re
2
2
To ob tain iv (q ) all we need to do is guess an an alytic func t i on whose
imaginary part is 2 p (m ) and whi ch has the correc t poles in the uppe r hal f
1
(5 .2 1)
the refore
(5 . 22 )
Making the sub t raction at q2 • 0 we ge t (5 . 19 ) .
• q� QX .
1lierefore get
q
� Qx 6 ( q � )
o �
q -q +i £
s o this is OK be cause the factor q 0 � mul tip lies the s ame in tegral a s i n the
xx
- (q0• 2 -q2) q2
tt cas e . Troub le comes however in the case for then the coe ff icient is
0
2 2
q�2 - q0 2
Q for the q • does no t change dire c t ly to as required , but
x
we have ins tead an extra correct ion of which cance ls the denominator
General Tht'Ol't'firnl Backgro1111d 2')
so we ge t
( Qx� - �xxq 2 )
J I
2 2
Vxx ( q ) •
i p (m ) dm i
2 11
+
211 p (m2 ) dm2
q -Ill +i£
2 2
µvq ) J
2 2
v
i
( q ) s 2ii ( q µqv - 2 p (m ) elm .L C + iS
or 6
2 2 + 2 11 (6 -6 6 ) ( 5 . 2 3)
µv µv µt vt µv
� p (m2 ) dm2 •
q -m +i£
where C is a constant (infinite no doubt) •
We could get rid of C by a seagull type term but we are confused because
Apparently <O l [J µ ( 1) , J v ( 2 � 1 0> • 0 outs ide light cone isn ' t enough . Perhaps
J
for every other real problem. Hereafter we drop it supposing some term like
A • A d
3i
is added to the Hamiltonian to straighten it out . It is called
m2 m
2 A2 2
J
of mass minus a term w ith mass + so p (m ) need not be positive and
2 2
p (m ) dm is taken zero .
�•
J
- ip · (t i)
m - e , 1 2) m (2 )
I (x,t) •
+ 2 2
p -m +i£ (211)
4 -
� (s +
811S
H
l (ms ) (5 . 24)
H / • Jl - iN
1
�
-ims
For large s , H is asymptotically H e . For small s
+
,
I (i t) •
m - .L411 2
6 (a ) +
-n
i
4 11 s
+ i_
1611 -
m
2
i8112 En (�
2 ) + Y -
1
_
2] + (5 . 25)
y • • 5 77 2 . . . .
The c0111t
111U ator function
C (i , t) • sgn t • Re ( I (i , t )
+
) (5 . 26)
F . T . C (x , t) • sgn
2 2
q 0 & C q -m ) (5 . 27)
30 Photon-Hadron Interactions
for real s only . Outs ide the ligh t cone it is zero . For small s
- [ §Jii 2
J
m2 m4 s
C(x,t) • sgn t - 4"' + I6ii - Tiiii + ( 5 . 2 8)
··
Prob lem 5 . 1 .
Show that the total cross section to produce any hadrons from e
+ -
+ e
at energy E + E is
(4tr 2> 2
a • ; p (q2 )
q
2
Problem 5 . .
2
Problem 5 . 3 .
of muons .
Low Energy Photon Reactions
Lec ture 6
/
y + p + 1 + n
y + p + '11 0 + p
y + n + 11° + n
The first two are the mos t extensively s tudied . Today data on various
polarization and spin cases are also avai l ab le , Evidently we are measuring
2 •
things which depend on <N 11 I Jµ (q) j N > for q O. The ( total energy in the
2 2 2
C . M. ) is of ten called s . s • (p+q) • m + 2mvLAB where vLAB is the energy
of the photon in th e lab sys tem ( for us v goes from 0 to 2 s ay ) . Thes e are
+ o +
or y + p + n + A , o
Y + P + n + 'II + 'll y + p + p + n
31
32 Photon-Hadron Interactions
The mos t characteris tic f eature of the behav i o r is rapid ch ange s in angular
same resonances appear here , of course , as in 11 scat tering , The photon data
That is <EFD ! T ! ABC> must have an infinity--a singularity when the momenta
satisfy PA + Pc • Px Px
2 2
p8 - and has the value
Mx. for a real particle x .
r es idue
The of the singularity is <EF ! r l cx> • <XD I T I AB> .
2
When Px is very near
2
Mx. ( and Ex is positive) the b ehavior is ve ry much as if an ideal free particle
-1
mass Mx. p rop agated so Mx.2 +
of
2 the amplitude varies as (p
2 - iE) •
X
4 4
More specifically if s is written s fi • o fi - ( 2 11 ) i o (p f p i) Tfi near
-
that the momenta can be combined into the momentum of some s t ab le p art i cl e ,
The exact form of the singularity comes from an an aly t ic extrapolation from
slight ly uns table and of decay rate r1 it can be s hown that the modification
Low Energy Photon Reactions
1
by
Mx.2+i £
2
Px -
It has b een found that such res onant behavior of the T matrix is exhib i ted
by s trong interact ions . (Here the width r is not produced by weak external
perturbation , but is a cons equence of the s t rong interac t ion itself . ) They have
( )
the following characteris tics .
-1
T 2 2
varies in Bre i t Wigner fashion p - (ll\t -ir ll\t ) .
R
This only means something preci s e if r is small enough that the variat ions
in the res idue factors < cd l T I R> <R I T l ab> are either negligib le or well known
(as near threshold b ehavior-- there a proper variation of r with Q may have
to be included} ,
The same resonances appear in many reac tions at the same mass and width
Properties is a tab le of such resonances . What is the meaning o f the exis tence
in other fields .
a) Atomic systems , The excited levels of atoms are stable except for
coupling with ligh t . The narrowness of the ob s erved resonances in the emiss ion
spectrum of an atom , for example , is known to be due to the smal l value of the
2
coupling e •
ob served . In thi s cas e the origin of the narrowness of the resonances can
physics .
34 Photon-Hadron Interactions
None of the above effects seem to appear in particle physics . But also
resonances are generally not so narrow , the width is about a fourth of the
spacing to the next resonance in the same channe l . The meaning of the appearanc
of resonances in strong interactions is not unders tood . At firs t theoretical
physicists did not expect resonances from strong coupling field theory . Then
they realized if it were strong enou&h there were "isobaric states" . But the
exact meaning of the existence of these resonances for underlying theory is
not clear .
Lecture 7
off resonance it is
for the rate a meson couples to a proton With angular momentum 1. For a
photon a similar
term was used . x was arb i trary (at ab out , JSO GeV for a l l res onances except the
1236 where . 160 was us ed) . But changing x is like changing a (E) ,--it distorts
the form and leaves the ques tion--How b ig is the tail of a resonance?
Impoasible to answer except adl itrarily .
�
The use of r • r {Q/Q ) 112 has implications for the express ion below
0 0
threshold . These are not used by Walker (violating dispersion relations ) .
What are they?
a (at resonance)
E-Ea
in defining resonance backgrounda--or rather "effects of resonances far from
their resonant energy" . This last concept in quotes is very sub tle and hard
to define but is used glibly for all kinds of things today in theory , as we
shall see .
b ) Can T b e expressed entirely as a s um o f resonant terms ? Consider
the case a + b + c + d
a
x d
b
s
t
•
•
(p a + ) 2
�
(p c - P a ) 2
u . (p a - p d ) 2
1
36 Photon-Hadron Interactions
and have a s um
(7 . 1)
c>-<d (7 . 2)
a b
For a real experiment none of thes e are res onant s ince t < O , a l l are far llMay .
An express ion has b een found for ( 7 . 1) which when s ummed gives something
that can b e wri tten as a s um of t channel res onances (Veneziano ) . These are
( 3 ) background .
(l) T h e resonant masses and widths were taken from the w N scattering
data which had b een analyzed by ampli tudes for each channel (of angular
(3) The ''b ackground" was a s lowly varying amplitude in each channel .
made t o vary s lowly , o r a t leas t smoothly . I t was hoped that all these
background terms would b e real b ut some small imaginary was ultimat ely
needed in some channels , pos sib ly b e cause resonances are le f t out , or have
Low Energy Photon Reactions 37
inaccurate parameters. (Next time Walker does this he will relax the cond i tion
Lecture 8
· �- r
r 'Ip
A
k
0
(a diagram which does not exist for 1T --but we will say more about this later) .
(2 q-k) - e
away from the nucleon. The effect is that a nucleon has a chance to be seen
+
as a nucleon with a 1T around it spread relatively far out , as
-m r
1T
e /r
forwards and appear real ("unbound " ) having received its necessary energy and
momentum from the photon . It goes fairly sharply forward--at higher energies
at least and affects the amplitudes in all valuea of a-channel angular momentum
including high values , where for low energies and momenta no resonances are
Since there the rapid variation with t near t • 0 is essential we are near the
2 2
pole m1T • . 0 2 GeV and we can consider it trus tworthy and accurate . To be
sure , at the higher energies lower angular momenta involve it for t large
2
and so the form l / t-m etc . may b e wrong or amb iguous (we mean we want to
1T
follow the principle� the contribution of poles is well defined only for
38 Photon-Hadron Interactions
parameters near their resonance) . But here the amplitude is slowly energy
dependent and its exact value for the lowes t one or two angular momenta
(s , p waves) is los t in the background amplitudes we are adding anyW'ay .
(No background was added for very high angular momentum) .
The expression we have written for the amplitude is not gauge invariant ,
however . The ff'+ springs from a source and doesn ' t conserve charge . It must
be combined to other terms to make it completely gauge invariant . One obvious
contender is the s channel resonance corresponding to the unexcited nucleon
at 9 38 . We can put these together aa follows (let charges on initial nucleon
these give
• 4w ( 2q-k) · e 1
C u2y 5u 1 ) ew/2 g ( 8 . 1)
- 2k • q+k2
M
I
•
- � +l(-+M e µl
M
II
4 wu 2Y5 2l
k +2p 1 · k
( 1' +
4M ('l( - U > ) ul l2 g (8.2)
•
µ2 } � 2-l(+m
Mxn 4wU2 ( e2 ' + iiM <''It - l(') Y u
k 2-2p 2 · k S l /2
g ( 8 . 3)
The answer is certainly yes . They were not included by Walker but again they
are relatively slowly varying for a positive energy w. They are far off from
their resonance (which appears below threshold at negative w .)
the las t term vanishes because � l • M on the initial s tate . The second term
is like a magnetic moment term in that it has no pole as k + O; it is of lc.w
angular momentum and slow energy behavior ; it can be taken into the b ackground .
Leaving out all such moment terms , as they only contribute to undefined
background we find the pion exchange term can be written in a gauge invariant
form:
"pion exchange term" •
( 8.4 )
40 Photon-Hadron Interactions
( 8 . 5)
tt '\_J pf
y..f � pi
<f l w l i> is the amplitude for the diagram
· -{
Since s trictly this is off shell some definition by theory may be used , although
all we can be sure of is that the res idue on shell is correct . The "w 0 exchange
term" does not have 1/ (t-mw2 ) • The angular dependence is slow and shows no
forward peak .
Any form like
(2q-k) • e - a • e
( 2q-k) • k a• k
for any a would be all right , but ( 8 . 5 ) has the virtue that the singular
behavior as k + O is phys ically correct .
I have gone through this long discussion to show that Walker commi t ted
no theoretical error of principle in his method of fitting . His adj ustable
parameters of fit were the desired quantities <Res l J I N> and the b ackground
µ
amplitudes , adj us ted to be as small and as slowly varying as possible. In th is
way all the data is distilled to produce theoretically s ignificant numbers
<Res l J I N> . Walker gives the value of the matrix element as A:
µ
rather than <f j J • e J i> directly . Suppose a resonance of mass m1 decays into a
nucleon energy E 2 , and a photon each of momentum Q . (Q (m1 2-m2 ) / 2m1 )
2
• •
If the photon goes off in the z direction with positive helicity +l , there
are two helicities possib le for the proton - 1 /2 , +1 / 2 , in which case the
helici ty of the decaying resonance is +3/2 or +1 / 2 . A is given for each case ,
and for whether it is proton or neutron ( see Table 8 . 1) .
Quark Model of Resonances
product s x A
symme tries s s x A
x x s , x, A x
A A X S
With unitary spin there are three choices. for each quark or 27 states . There
are 10 synaetric s tatea l uuu> , l uud> etc . and only 1 antisymme tric l uds>
s s A
(aince each mus t be different) . The remaining 16 are Ba and BS . Thua the
unitary symmetry is
S • lQ decimet
X • ! octet
A • ,!-. s inglet
Only such su3 multiplets are found among the b aryons . For spin with two
poss ib ilities there is no antisy1111111 t ric s tate , The sy1111111 t ric is spin 3/2
quartet , the X (two) is spin 1/2 doublet , syllbolized by superscript 4 or 2.
Therefore collbining spin and unitar,r spin we can make the following
44 Photon-Hadron Interactions
multiplets (SU
6
multiplets ) . S can come from S• S or X · X via ch�rt above so
56 states
� s tates
2 0 s tates .
l a> _!
16 ( 2X1 - X -
2 X
3)
I a> _!
./!
(X - X ) 3
2
To get overall I S > , this orbital I X> must be combined with unitary spin and
spin I x> or 70 . Finally when the unit of angular momentum is combined with the
spin in various ways to make different total angular momentum j , all of odd
parity , we find
2!1 1 2.!3/
2 2
2111 2
2 J3/ 2
4 4 4
.Ql/ 2 J.31 2 .!5 1 2
2 l:.Q.1/ 2 3/
2 l2. 2
4
All these mult iplets (except the .§.31 2 > have been found .
with three such obj ects form s tates like I zzy> meaning the firs t obj ect is in
condition z, the second x , third y . From th is and its permutations l xzy> etc .
Quark Model of R <•so11a11c<'S
If two say xy are the same s tate y • x replace l xyz> + l yxz> by 12 1 xxz> .
If three are equal only the S survives l xxx> • l xxx> .
5
Comb ining two such s tates in a product we find
St!l.te 1'bltiplet
J.,_ Iz
( f' ! Jµe µ I 1) /F A(aev} -1/2 A�(�V)-1/2 1-.e:ICP(c-ev)-1/2
p33 (!.2?.5) 0
4�3/�· +1 +3/2 p - ./6 p -0 . 190 -O . J.78 -o .2g
�/2
....,
Di,s(l570) p 0 0 0 o .°'o t
4!s;J�.i-]i
+l/2 p 0 0 0 - 0
2 +o . 050
331( 1650 ) �l?2·1-1 +l/2 p - � ),p + � rn 40 .04.7
TA13!,;l 8 . l( con·l; . }
2
n35 ( l670 ) !.�3/-ci.r}3.• 1 -1 +s/2 p + /n +o . 09 1 +') .OSl
+ f3 K::>
pll ( 14 70 ) +J../2 p (+ Jt 7-p) A +o . 0 2 8 .)0.0:32 s::
I:'>
...,
2!:t/J.�· 0 1 ..,..
.J;-l/2 n -0 . 019 -c .oro
3 ;s:
Cl
(- � >.p) J. I:>.
+-0 . 064
�
:::-15 (1G80) 2s Jss,2+1
-5/:-_.,.,..... ,J2 +::i/2 :P � .070 ..0 . 139 �
(+ � /if) i\
:::i.:i
+!./2 :P ffi -0 . 0 11 -0 . 015 - o
(- A:> + 5 /n} �· �
� i\ ::s
I:'>
+5/2 n 0 0 0 - 0
::s
a
+l/2 n +o . 038 � .OU
(+ .fl >.p) i\
w ( 7 84) 0 !( 0 . 29 7 0 . 22 ± 0 . 0 2
38i �o
A(15 20 ) +3/2 A
+ � J?i +o . 10 9
0 . 09 7 + 0 . 0 10
2131I?E· 1-J
-
+1/2 A Ni -
+o . 012
j
- � [J3 � .m] 1
""
'I
48 Photon-Hadron Interactions
Resonance mass m1 goes to photon momentum Q and final hadrons (see I z column)
mass m . J z is the resonance spin proj ection along the direction of motion
2
of the decay photon of helicity +l .
2 2
�-2
Q • -2-
ml • >.
• n" Q ,
"
p -
-m2 Q 2
• exp ( 1 "'
F "' 1
n (mi'+111 )
2 2
1
n .. 1 . 05 (GeV) 2
Lecture 9
!!....:..1.· Two orbital excitations each of type X can make s , X, A and the total
orbital angular mowi. can b e 2 , 1 , 0 as it turns out the s tate of symmetry A
has total L • 1 , s tates S or X can have 2 or O. Hence w e c an make
2 2
A11 2 !31 2 ·
The 20 has not been seen . It has no matrix element of any operator (like
J 11 ) operating on one quark at a time to · the proton . Two quarks at leas t mus t
have their motion changed to get to the 20 from the fundamental �6 +J N
,O • 0•
We have definite need for the �6 , 2+] b ecause of a d of spin 7/2+ which
can be gotten in no other way . Two other d ' s of this set of decimets at
j • 1/2+ , 3/2+ have als o been seen at roughly the same energy . This whole
multiplet is expected as a Regge recurrence of the fundamental • �6,0�N 0
and is 2 . 1 (GeV) 2 in mass 2 ab ove . This also accounts well for the octet
at 5/2+ (m2 • 2 . 85) .
A puzzling s tate is the 1/2+ Roper resonance at 2 . 16 . I t i s beat fitted
as evidence of the js6,0�N • 2•
What is puzzling about thia state is its law
Quark Model of Resonances 49
ma s s a s compared to the
21
512 at 2 . 85 which is b e s t f i t ted in �6 , 2� N •2 •
These two s t at es are identi cal in their SU {6) properties . The ctuark model
with harmonic dynamics predicts them to be of the s ame mass s ince they b o th
corres pond to two internal exci tat ions . The l arge mass difference shows that
harmonic dynamics may be too s imp le . The Roper res onance is a b reathing mode
� 6 , 0�N 2 has
net orb i t al angular momentum.
2, 2
+
re re expected i f
•
w ere only one internal deg ree o f f re edom . The 70 ' s wou ld �
o
th ere b e expected .
Row g od is the evidence that they are there . There are three s t ates that may
require them :
+
l/ 2 N (l780) poss ibly
2
r7o,o+J
m • 3 . 16
7 / 2 N {l990 ) [ 70 , 2+:i m • J . 9 6 .
+ 2
� J
freed01D only--but it is well es t ab lis h ed . on
P ar t i c le tab les , but no t on the 1971 tab les ! ) Th e f i rs t could also be 6,o .
N 4
a hi gh er "b reathing" mode excitation of the nucleon , the s e t being nucleon
m • . 881 , Roper m 2 . 16 , N ( l 7 80 ) m •
2 2
•
2 3 . 16 .
E6 , 2 ]
The second , one the
at
+ 2
miss ing o ct t
e from the J • 3/2 . If so its 1118S s • 3 . 46 is very
s t a te so
2
dif feren t from the mas s = 2 . 85 of the J • S /2 the usual rule o f
�·
to c a re ful ly
o
A good res earch p roblem is s tudy the propert ies of t he 20
via the quark m d el and propos e the wises t way to look for it exp er imental ly .
The di f i cu l ty
f in ob serving the 20 is that if we as s ume we prob e the b ary ons
this is that the is antisymme tric in its su 6 indices while the nucleon is
20
totally symmetric . Operating on one of the quarks in the nucleon still leaves
the other two in a symmetric s tate .
There are many s tates possib le at N a Th e mos t important i s surely the
�0,1-J at �0 , 3J
3.
N �6,4J N
known .
H •
l + 2 + 2 +P 2 mw2
2m ( p l + P2 + 3 ) + -2- (9 .1)
One takes the spin and unitary spin dependences to factor out from the internal
motion .
The interaction of one quark , say 1 , with an electromagnetic field is taken
to be
where e 1 is the charge on the quark . Matrix elements of this operator (with A
the appropriate plane wave for a photon) between states n of the Hamiltonian.
(9 , 1 ) gives us
our nuni>ers for <Res l J I P > to compare to experiment .
µ
The quantities m , w , g , are parameters ; g is the gyro-moment of the quark .
Th e Dirac value g • 1 works very well an d it w as chosen to be 1. Th e value o f
w i s guessed a s 400 MeV so the level spacing will be roughly right--the value
of m is chosen (as 340 GeV) to fit the proton magnetic moment . The values of
the current elements that result are given in table 8 . 1 next to last column
Quark Model of Resonances 51
Anon-rel . • ANR . It is seen that there is remarkably good agreement with the
known values . All the s igns are right and small values are predicted to be
small , sometimes by selection rules generated by the model--sometimes by cancel-
lation of tel'111S for orbit and spin coupling . (The terms containing p in the
column for <f l J e I i> are the ones contrib uted by spin-- the formulas are from
µ µ
a relativis tic modification . They are virtually the same as the nonrelativis tic
cas e ) . Finally the orders of magnitude are generally very close� the wors t
We have us ed Bose s tatis tics . If the quarks can be free it can be proven
that spin 1/2 implies Fermi statis tics . The only way out is to assume three
kinda or "colors " of quarks for each su3 type and say that the bound s tates are
pure singlets in the new index . For example , baryons mus t be in the 3Q s tate
containing one of each color thus antisymme tric in the color , to say now that
quarks obey Fermi s tatistics require� symme t ry in the other indices . (A formal
way of dealing with this called parastatis tics can be shown to be exactly
equivalent to the three colors of Fermi quark theory . )
Lecture 10
Feynman , Kis linger and Ravndal , Phys . Rev . (1971) , tried to remedy the point
that a nonrelativistic theory was ambiguous in haw you applied it to a prob lem
of relativis tic kinematics by making a theory where the rules at leas t were
relativis tically s tated . Such a theory to be complete would have to be a
relativistic field theory which is too complicated to calculate . They made a
s imple theory of states , but at the expense of unitarity--which meant cons iderable
trouble later on--but at least relativis tically unambiguous rules could be
52 Photon-Hadron Interactions
formulated . The net result for our photon matrix elements is to give almost
exactly the asme nume r ical result as the non-relativis tic caae--so it is not
of great interest h er e . (It do es so with one less arbitrary cons tant , however . )
2
2mH p2 + m2111 x2 and that 2mH was nearly the difference of the s quares of the
•
energy m. + W (or m.aaa ) of the s t ates . Thus to get {m.aaa ) 2 to be equally spaced
2
they could use an eigen-vector operator for maa s which was a harmonic os cillato r .
The only cona tant is m.2 1112 wh i ch they call ri 2 , so the quark mass disappears on
the righ t hand side. Thus let
K • J (p a2 + pb 2
2
+ p
c
2) + ;-
n (u -
a �
)
2 + (� - u ) 2
c + {u c - u a )
2 + C
( 10 . 1)
quark a . This in r e c i p r o cal is a propaga tor l/K. Its poles (K • O) give aiass 2
as seen as follows :
( 10 . 2)
K • p2 - N
( 10 . 3)
{For the b aryon photoelectric ma trix e lements the re are no fur the r (e f f e c tive )
expression times a wave function with three dirac 4 component indices although
it does not affect those indices in no electromagnetic field . However p a2 is
interpreted as its equal c; a >2 so when an electromagnetic field is operating
the operator is
KinA • 31}a - e .' C ua
8 �( ) (
; a - e 8Aa < ua> + � - �Ai, <� >
2
)
(
+ ;c - e cA( uc > )� + �: Gua - �) 2 + (� - uc ) 2 + ( uc - ua> 2 ]
(10 . 4)
(10 . 5 )
res t s tate of the oscillator in the direction of the 4-1110ment11111 P � of the state
*
( if aµ and b µ * are creat ion operatr�s f or the two oscillators ) ,
*
P µ a µ I +> • o
( 10 . 6 )
54 Photon-Hadron Interactions
( 2) By taking + to involve Dirac spinors twice as many s tates as des ired are
included (Q and Q) . To avoid the parts of the s pinor appropriate to ant i quark
Y I +> m l +>
JI )I C
P - c10 . n
2
where m is the mas s of the s t a t e (m • P JI P JI ) . ( 3 ) The curren t is not gauge
s separa tion Nn .
2
which may not have exactly Transi tions from 8p to quar t e t
2
s tates a r e unaffected . Also t h e F ( 1688) is unaf fected for its mass is almo s t
15
exac tly 2 n above t h e proton . The uncertainty involved for a l l e lements of the
Om i t t ing such s t a tes [10 . 6) and ( 10 . 7 ) ) means a complete set i s not used
and , s ince the omit ted s tates have nega tive norms , means that mat rix elemen ts
will be generally too large b y ever increas ing factors as the mas s discrepancy
of initial and f inal s tate rises . FKR compu te all kinds of meson decay widths
(via PCAC us ing the divergence of the axial current for the pion coupling
operator) and f ind this to be the case . To compensate they divide by a factor
- -3
)
(u2 u 1 ) for the spins , and include a cutoff factor
{_ a
� ml
2
( 1 0 . 8)
\
exp
2)
n cm + m2
1
Thes e do not seriously affect our matrix elements , of curren t , for lower
states ( the theoret ical 1688 would be 40% higher without them) they have been
modificat ion of the non-relativi s t ic model is implied for these ma t rix elemen ts
of J11 •
The table also contains one meson ent ry , that for w � w + Y i here
give still lower resul t s (more like . 15 ) . The value comes from ob servation
Quark Model of Resonances 55
Bes ide these non- diagonal mat rix e lements there is , of course , the diagonal
elemen t <Proton IJ µ I Proton> . This is known for all moment um Q as the form
factor of the proton - we discuss here only the low q resul t s , the charge
(which of course comes out righ t ) and the magne tic momen t . The quark model
( a s used here) i s quite unsuc ces sful there . The rela t ivistic theo ry gives
µ • 3 . 00 Nuclear magne tons for the p ro t on and - 2 . 00 for the neutron ( this
Exper imen t is -0 . 70 ± 0 . 0 7 . (The reas on it is magne t ons of the mas s of the s tate
•
2
i s ins tead o f energy perturbat ion • µ (a • B ) we have perturbat ion o f mas s
Suppos ing that all this serves as evidence for the quark model , where do
elements can be calculated dire c t ly from the s ame model (pu t .I( � y 5'0 and a
few are known by a -decay . They come out wrong by the same factor , if theory is
numb e r of very sens i tive matrix elemen t s whic h come out as the d i f ference o f
two large terms . For details see FKR - we will discuss the photoelec tric
2
mat rix element s for q • 0 later .
b) States o f negat ive norm . These c ause a l l kinds o f amb iguit ies and
uncer t aint ies . Firs t , nobody has any idea for the factors int roduced by the
spin (except t o cons ider the wave equat ion t o be that for a two-component
spinor wave funct ion - this has no t been worked out - on the face of it , i t
s t a te s . In this regard , a sugge s t ion was made by Fuj imara , Kobayashi , Narniki ,
Progr . Theor . Phys . (Kyo t o ) .!2_ 1 9 3 ( 19 7 0 ) . FKR use the "wave fun c tion "
z ] for the gr ound s tate of the osc illators (all o thers are
2 2 2 2
exp [ t - x - y -
2
t
s imply polynomials t imes this factor . ) The explod ing e can only be c on t rolled
56 Photon-Hadron Interactions
2 2 2 2
by negative norm . FKN use instead exp [ -t - x - y - z ) which they po int
- (t
2
- x
2
- y
2 2
- z )
) ' • E� ( 10 . 9 )
each of wh ich can be in its ground s tate . The time oscillator now contributes
are not excited " or something - but at leas t i t is an alternat ive view .
Thus the wave functions in momentum variab les have the form
exp -
1 ( p
2
-
(P . p) (P • p)
p2
) ( 10 . 10 )
20
2
where P is the four-vector momentum of the overall s tate P •
�s 2 • (The
ground s tate wave funct ion is simply this factor - others mult iply by polynomials . )
i • x
between two states P 1 to P ; P - P •
1
Ma trix elements o f currents e �
2 2
q
( 10 . 11)
exp ( 10 . 12 )
Thus , for examp le , the form fac tor for the pro ton (assuming no spin factors )
2 2
where P 2 P �. ( P -P )
2
• • •
1 2 1 1
q comes out
1 1
( 10 . 1 3 )
2 4
it is seen tha t for large (negative q ) this behaves as l/q as does the real
are those whose time s tate is not exci ted , al l the matrix elements are nearly
2M M
l 2�
exac t ly the same as FKR (expect for an oc cas ional factor of M 1� / P 1• P •
2 �2� 2
M -tff
l 2
which is close to 1 ) excep t they all carry the form factor F , it is theoreti cally
much more sat i s fying than FKR be cause it is no longer chosen arb i t rarily) wh i ch
(10 . 14)
which is much l ike FKR empir ical F exce p t f o r the 4 i n the exponent ( s o i t i s
very close to FKR ' s F to the fourth powe r) . Thi s cuts o f f much too fas t and
makes the poor f i t o f the 1 6 8 8 even much more unsatis factory and a l s o des t roys other
cas es that f i t te d we ll .
the s uccess ion o f two current ope rato rs . Sca t t e r in g of photon , elect romagne t i c
ma s s di fferences or non-lep tonic weak decays require the matrix e lemen t of the
product of two curren ts . In the lat te r case the two currents are at the same
space time point and we can prove that i f the quarks are Bose the famous 6I • 1 /2
rule for these decays is explained . A natural s ugge s t ion is the s e cond order
pe r t urba tion 6K i 6K . However , in s umming over all s tates in the p ropagator l/K
a) Sum over all s t a tes including time s ta tes wi th negative norn , and include
a> Sum ove r only physical s tates (no time s ta tes exci ted} using some
y) Use the sys tem o f Fuj imara e t al. s ummin g over all s t a tes including those
There are awkward prob lems associated with each way (wi ll the commut ators
vanish outside the ligh t cone ? } and no thing precise or defiaite has yet been
proposed to compare to experimen t . We shall d i s cuss this prob lem again when we
have developed further theore t i cal too ls , e . g . dispers ion relat ions .
experimen t ? The bes t clue to s ta r t with is probably the real value of the
58 Photon-Hadron Interactions
an axial vector . An interes ting prob lem is the split o f E and A in the baryon
Next comes the correct ions that the quark model implies to itself for
phys ical cons is tency . Fo r examp le s uppose , us in g PCAC or a modified quark model
tha t has some kind o f direct coupling , that a pion , 6, and P are coup led - we
v
6
I p
This imp lies , of course , that the 6 can decay into n, P and thus the 6 has
a width . But in the original theory the 6 was an exac t energy leve l . We can get
the width by imagining the delta propagates also via vertical diagrams like
p p
the quark model gives the coupling to be used later as the beginning of some
per turb ation expans ion . This succes s fully induces an imaginary part in the
propagator irm - but at the same time corrects the real part so the expe ri-
mental mass is not the uncorrected mas s from the direct us e of the quark model .
@. (D E6, �
How much is the correction and can we use it to understan d some part of say
the I: ,A
4 2
mass dif ference , the o split in the etc1 The
involve couplings far away f rom their mas s shel l , there are many more s tates
than j us t the Pn available to the 6, etc . Finally and mos t impor tan t - maybe
we are partly counting the same thing twice - perhaps in the equation for the
Due to these uncertainties nobody has worked these things out . (For
Quark Model of Resonances 59
another examp le , h ow· much is the magnetic momen t of the proton corrected ? It
has some ampli tude to be a delta and a p i on , or even a pro ton and a pion) .
Th i s que s t i on is very serious . What is the quark model a model for? For
diagrams like a f ield theory pe r turbat ion expans ion ) to compute the f inal
hadron world?
corre c t and cons is ten t (wi th uni tarity , for example) manne r . As such i t
reaches a dead end . It shows tha t ce r t ain regulari ties exi s t among the hadron
s tates , but a precise s ta tement o f what this means or what this really
imp lies one should include in des igning a more adequate theory is s ti l l
actions .
further and b) a prob lem that every "narrow resonance approximation" has to
face .
Pseudoscalar Meson Photoproduction, High Energy
Lecture 11
We now cons ider what happens at higher energy - in parti cular from
5 to 18 GeV (as full data in the t rans i tion region 2 to 4 GeV is lacking) . Here
the analysis directly by sums of resonances is hopeless as too many are involved .
A full analysis of amplitudes does not exis t , mainly cross sec tions and a few
asymme t ry measurements with polari zed pho tons , or polarized targe ts does exis t
but our theo retical s k i l l at analyzing the ampli tude for each helici ty e t c . is
no t yet good . Therefore this will be more of an empirical survey po inting out
References
60
Pseudoscalar Meson Pl10toproc/11ctio11, Higlr £11c•rgy
nuc leon cross section at high ene rgy . The cros s sec tion fo r yp + n +n is , for
The to tal cros s section for y on p shows bumps due to the 1 2 36 , 1560 and
1 700 res onances at low energies . It is mos t ly f lat above 3 GeV , it falls
convenient to plot them as functions of t for given s (or k , the pho ton
momentum in the lab . ) . The s lope of the t dependence varies relatively s lowly
with k, and cer tain charac teris tic regions for t are no ticed . There is also
a peak in the backward direc tion , small u, where i t is best to plot u for
given s .
An empirical rule , no t precise , but very good for comparing data at various
-2
energies is that all the cros s sections vary nearly as k in the forward
-J
di re ction , and as k in the b ackward direct ion . The reason for this is no t well
2
unders tood . That is , k da/dt is a nearly f ixed function o f t . Below t about
2
1 (GeV) the curves for different react ions are each dif ferent . They are given
on a chart from Dieb old� repo r t (reproduced here in fig . lZ-1 ) . ( This is no t
precise data , varia tions of f ( t ) with s are smoo thed over - to get exact numbers
see detailed experimen tal repo rts - but they are exce llen t representat ions of
to be mos t accurate for pho ton react ions but is seen also in many s trong in ter
- 0
action cases , fo r examp le in forward ( n p+ir n or n+p+k+I +) as we ll aa in back-
ward (n+p+pn+) (G . Fox) . It may be related to ano ther s t rong interaction rule ;
that in high energy inclus ive reactions making mesons or b aryons these particles
JP 2
are distributed in transverse momentum P.l. as e .L This is also t rue of
l.
-IP 2
photon induced inclus ive reactions . This is only true in a rough sens e , e
-2 . 2P 2
for small p J. to e J. for large PJ. •
2 2
Very supris ingly b ackward photopro duction peaks are the only ones that
seemingly do not obey the rule . They are very flat - flatter than anything else ,
62 Photon-Hadron Interactions
1•2 u• �-A
++ -
(This is not true for y,... w wh ich has a rapid fall
varyi ng as abou t e
3u
of f with u in the backward direct ion , more like e .)
-2
The k behaviour would not be expected , for the varioua traj ectories have
-2
di f f erent intercep ts and s lopes . But the k canno t be a true asymptotic rule
about ampli tudes - it mus t be a sort of accidental res ult of several terms
varying nearly that way - although quanti tatively it is always a bit of troub le
-1
to ge t i t out . This can be seen becauas if the ampli tude varied truly as s ,
or rather , in correct normalization , were constant , the pho ton coupling via
isoscalar and isovector would differ in phase by 90 ° and would not in terfere .
+ -
Therefore we would expect da /dt (yp+ w n) • da /dt (yn+w p ) and they certainly
Thus if f (s ) • 8 s
iw * m ( t)
any ampli tude f ( s ) the antiparticle amplitude is f (e •> .
-iw a a
for antiparticles it is S * e s . But for an isos calar coupling the amp litudes
for particle and an tiparticle are oppos ite , this requires 8 8 • 1c exp ( -iw a / 2 ) ,
c is real . For isovector coup ling the amp li tudes are the same because the
antipart icle has oppos ite I-spin thus S • c ' exp ( -i w a / 2 ) , c ' is real . This
v
shows that a
.
and a di f fer in phase by 90 ° .
v
In Kegge theory the exchange of a traj ectory can be in terpre ted as a sum
the direct ion of the two longitudinal momenta . Take , for example , the case
where a vector particle is exchanged and there is no spin flip at ei ther vertex
Pseudoscalar Meson Photoproduction, High Energy 63
The ampli tude goes like � (P i' e) {P i e) • P ].' P • s. The spin is therefore
P l 2
not available to carry longitudinal angular momen tum. If there is a helici ty
f lip 6A at one of the ve rtices , the spin of the exchange is again used up in the
6 I
term with the highes t power of s ; hence there is a fac tor Q A or lt'"l A .
6
If
+ 1 6 A2 1
b oth ver ti ces have a f l i p the amplitude goes as ,tt"" l A l l
6
ideally . This
in such cases the e f fect is lost by absorp tion . Absorpt ion corre c t ions for small
impact parameters reduce the t to its lowes t phys ical pos sibility ( in this
t l Al l
6 + 1 6 A 2 1 otherwise absorp tion can reduce the powe r .
Lec ture 12
6 ** p
vertex we mus t have a flip of helicity, at the p6 ver tex none is necessary
2
(because of the p - 6 mas s d i fference ) ; hence the ampli tude is ,/t'"/ ( t-m ) showing
w
2 -2
a dip at t • 0, a maxim1111 at t • m w exchange expects k cross sections
w
( s 2 a -2 , a Z 0) . Gauge invariant one pion exchange wi th ab s o rpt ion corre c t ions
We have a helicity flip at the yw vertex and one (from parity) at the
we
2) ,
w
nucleon vertex . Hence ideally expect t/ (t-m there should be a drop
To unde rstand how absorption affects the ampli tude, pe rform a Fourier trans form
pion now looks like V (e · µ b ) which is an odd function of b . The photon couples
VV
b
to this pion with amplitude that goes like (e -µ ) , the firs t gradient is
because the pion is in a P wave with res pect to the nucleon , the second gradient
VV
-
is because the photon coup les to the pion in a P wave , (e µ b ) is a function
The forward ampli tude is proportional to the integral of this f1mc tion which
is zero so we still havedt explained the observed peak . Absorp tion is here
introduced , pions which are s cattered by photons with small impact parameter
do not always get out because there is s good chance for them to interact
VV
b
strongly wi th the nucleon . The function (e -µ ) mus t be multiplied by a
function a (b ) like :
The integral of the product a (b ) VV (e -µb ) gives the forward amplitude whi ch
(U ·l)
Pseudoscalar Meson Photoproduction, High Energy 65
G . Fox has noticed the empirical rule that set ting C • 1 works very wel l in
2 2
many cases . The amp li tude becomes m / ( t-i11 ). ( Such peaks are common in
rr 11
s trong interact ions , e . g . , p n change . exchange pn�np ) .
+ -
The difference between do/dt (yp+rr n) and do / d t (yn�rr p ) shows the presence
rr n • rr + A2
+
yp � p +
yn � 11 p • rr - p + A
2
The rr and p interfere coherent ly because the p coup ling is mainly due to s pin
flip (see b elow) . The sign of coupling agrees also with what is e xpected b y
naive quark mode l . ( I t also agrees with the s ign i n a similar s t rong in terac tion
case np�n • rr+p and plf+'tll!' • rr-p . ) The expected energy dependence for various
t for the p is no t clearly seen . Perhaps for small t pions domina te . For
proportional to
-i rr a
( 12 . 2 )
( 1 -e ) a ( t ) -1
s 1n11a r (a)
s
-l
The po le i n (sinrra ) a t a•O is "killed twice" by the phase in the numerator
+ 0 0
The dip is seen more clearly in 11 p�rr n ; in y p+rr p the zero being in a double
flip amp litude may be obscured by absorp tion . In a s in gle f lip ampli tude i t
should show u p •
0..1. 0(1
A • --- ( 12 . 3 )
+
o and 0 11 are the cross sec tion for photons polarized perpendicular and
.L
1 ,I.-- . - --- -
1
(
i
'
.1
66 Photon-Hadron Interactions
a a 11 , a
s idering the exchanges which con tribute to .._ and ..L is contributed only
• e.g. p ,w ) a11
J
by natural parity exchange {P {-l ) is contributed only be
e .g . w) .
J
unnatural pari ty exchange {P • - { -l )
pseudos calar
scalar
i t i s therefore zero .
pseudoscalar
y
"'V\I V\J'\..f\
scalar
there is another change in s ign b ecause o f the pseudos calar parti cle so a
natural pari ty exchange can contribute to a.L . Similar arguments can be used
The reason for the asymme try curve can now be seen as follows . At t • 0
t (1 2 .4)
Au •
-
t-m11 2
� 0
( :)
=
The total is
t-+11111
An • � - .!_ • !. (1 2 . 5 )
t-m 2
11
2 t-m11
+ A1..P
1
2
I A..J...1 2 + I A 1 1 1 2
p p
we know that 11° couples to two photons from its decay . In fact, this effect
provides the most accurate way to measure the lifetime for 11°...-y y ; the exchanged
photon in the Primakoff effect is almost on its mass shell .
For t outside the forward peak , the reaction can come from some exchange -
we expect the w to predominate (see below) and to have a strong single flip
(flip at y11 vertex , non - flip at nucleon ver tex) . Thus it vanishes at t • O , and
does show the dip when a ( t) • 0 at t • 0.5, The energy dependence is entirely
68 Photon-Hadron Interactions
0
be equal . The ratio yn+� 0n is 1 for small t , falls to 0 . 6 near t 0 , 5 and •
yp+� p
rises again to 1 at large t .
The asymmetry for polarized photon parallel a11 and perpendicular a.L to the
production p lane has also been measured . A • (a_._- a11 ) / ( °.i_ +o11 ). The cross
section is predominantly a.L. except near the dip when a11 may be as much as 20% ,
yp+K+A yp+KH
yn 0 -2
p 1 s
w
3 3
ir o wy •
1
ir0 PY 1
� <I g
I
!:::
i <I
o
o0
o < I
00 <
. :�
{ 0 0.
I
j:::
"'"
ci
0.
>-.
� <I <I <I <1 o
000
0
0
/ < w
7{i t
C:
{\j
ci
<I
<I // >--
/
/.
/
� 0 '
0
0 0 0
0 0
0
t - Channel Exchange Phenornena
Lecture 13
71
72 Photon-Hadron Interactions
i�·
Amplitudes A(.Q.L , s) can therefore be written as
J
b -
A ( Ck , s) • e a (b , ( 13 . 1)
:�
--"-.&------ A
because for large b no interaction occurs while for small b there is lit tle
chance of getting through . In crudes t approximation we can say all comes from
a ring at radius R, leading in Fourier trans form to an amplitude like J0 (/:t R) ,
for no helicity flip . This expects a dip at t • --0 . 15 which is sometimes found.
t -Channel Exchange Phenomena 73
For a s ingle he li c i ty flip where the amplit ude mus t vary acro s s the impact
pion exchange s c a t tering , phot on produc tion of ir0 e t c . (The dip that in Regge
Tb.is model does give , with some exceptions , a rough descript ion of the
general behaviour , but it is not a comp lete theory because the energy dependence
with this have s o far , lead to complications and amb igui ties and no clear
b} The Regge po le formulas result from a sugges tion (mo tivated by an analysis
of non-relativis tic s cattering by Regge) that the amplitude function for two
body react ion A ( s , t ) be s tudied as an analytic funct ion of the energy s and
( called cut s ) is in general expected . Nearly any funct ion can be s o expanded
but the usefulne s s of the idea depends on making further assumptions . Eviden tly
as s..., the large s t a ' s would determine A, and i t was hoped that the poles would
have these , definitely above the a valuea where the cuts start , so A would
(t)
behave as S ( t ) a a or a sum of a very few terms o f this fo rm . If so, S ( t)
extended to pos i tive t would be an integer for those t equal to the mas s of a
For certain react ions (e . g . w p+w 0n) these expectations have been
p ion exchange we saw how ab s o rp t ion , a concept from viewpoin t (a) of ten
modifie� the simple pole pic ture . In the "pure" Regge view any modi fication
74 Photon-Hadron Interactions
of the simple pole formula for A(s , t) can be described, of course , by adding
cuts - but what cuts to add , and how strongly requires outs ide ideas , like
borrowings from the absorption model . The cuts star t , for large s , righ t at the
position of the poles and cannot be eas ily separated from them .
It is found empirically that a (t) is nearly a straight line . The reason
for this is unknown .
Also to describe the nearly constant total cross sections a special traj ectory
was invente d , the pomeron , with a special value one for a at t = O, without any
dynamic explanation of this part .
This theory is again incomp lete as no information is supplied fo r get ting
B ( t) , and , in practice more important , for determining the rise and behavior of
cuts . Attemp ts to avoid using the absorption model (conspiracies , " etc . ) have
failed .
Obvious ly some kind of integration of the ideas of a , b will be the mos t
profitab le .
c) The relation of the Regge form to the particles on the traj ectory can be
looked a t ano the r way (van Hove) . Suppose we have a theory which would give a
sequence of particles of ever increasing angular momentum . Then the Born terms
representing first order exchange of all these t-
channel parti cles will give
2 2
a Regge - like sum. I give an example . Le t the mass go as mn : mo 2 + nSl , where
m0 2 is the mas s 2 of a scalar , m0 2 + ".. of a vector , m0 2 + 2'1 of a tensor (spin 2 )
particle , etc . The exchange of these for a reaction A + B + C + D where for
s implicity A, B, C, D are s calar , goes as
c D
s2 + • • • ( 1 3 . 3)
-- 2
t-m3
e at
The s for vector comes from (P A + P c ) e at the upper coupling and (P B + Po) •
the lower summe d over e . Likewise s 2 for the tensor comes if we use the full
t -Channel Exchange Phenomena 75
two tensor indices to carry informat ion about P8 + P D from vertex to couple to
PA + PC to make ( ( P
2
A + P cH P8 + P D )) 2 or s in the limit . e n are squares o f
coupling cons tants . Le t m 2 • m 2 + nn , and let e equal l/n ! say , our sum i s
n o n
1
n!
set ting u • se and (t-m ) /n • a(t) the integral can be trans formed to
-nx 2
0
1 ( t ) s -a (t) - 1 -u
A • s a l0 u e du •
- 0
For large s the upper limit on the integral can be put m, to ge t the asympto tic
form
- fr s a ( t ) r (-a ( t ) ) (13 . 4)
the original series . In general any Re.gge term can be expanded by its poles
in t (coming from S {t ) and the series is then a van Hove series determining
particles and couplings . Or some guess can be made of states and residues and
a Re.gge term derived . These two forms are mathematical express ions of the same
theory , and permi t di f ferent points of view of the same th ing ; sometimes one ,
Comments
s
For negative s the series ( 13 . 3) gives an amplitude rising as e . It is
sugges ted that the coe f ficients canno t fall as quickly as l/n ! Al ternatively
one can say that the values for negative s are taken from the asymp t o tic form
Clearly in t h is sum, for negative t , we are far away from the true resonances
and are again in the position of not knowing what the real sum should be unless
we have a comp lete theory . Such analytic extension formulas are hard to use
practically - for example what does observa tion of the firs t two resonances of
- ---i + ? • ?
l
-z s
t -m t m
0 - 1
76 Photon-Hadron Interactions
empirically partially determined funct ions is a mathemat ical p rob lem I d o not
unders tand nor trus t , nor can I guess errors . To show the type of di f f i culty ,
a ( t)
suppose a s um o f t channel resonances gives Bs ; now change the coe f f icient
i • l term by 6C . +
o f the Sure ly the s um is now Bs
a (t) ( d C ); wh ich for
t -m
1
large s goes as s . The point is a modi fication o f the second term only is not
"na tural" or "smo o th " and changes the result dras tically . Whatever phys ically
causes the change in the f irs t term changes all the other terms a b i t in s uch
To s umma r ize the van Hove sum is not really us eful except theore t ically .
If you have s ome theory of all the t -channel resonances and their coup lings then
you can b e assured the theory will lead to Regge asymptotic behaviour - or i f
y o u have s ome theo ry leading t o Regge asymptotic behaviour you can , b y looking
quark mode l for meson resonances (QQ) does give an expec ted Regge behaviour in
Lec ture 14
d) s -Channel Resonances
Anothe r theoret ical idea is that all o f the behaviour at high energy also
can b e unde rs tood by j us t cont inuing our s channel res onance analys is (e . g . by
and sharper angular dis t r ibution results . Th is me ans that higher and higher
an gular momentum a-channel resonances must be con trib u t ing as s increas es . But
this is , of course , all right for we know higher angular moment \DD resonances are
does imply a great deal ab out h ow the resonances and their coup lings mus t behave
at large s .
t -Clra1111el Exe/range Phe110111e11a 77
That the entire curve comes f rom s umm ing s chann e l ( and , some wo u l d say ,
b e s o s t rong a s t o guide theory t o the one an d only exact expres s i on that can
do this - thus , much a t tention has been paid to this idea . Mos t s t atemen ts
resonances far f rom their resonan t ene rgies . Ins t ead it is noted that t h e
Such ques t i ons as the con tribut ion of u channel res onances (which a re
always o f f res onance so the imaginary p a r t is very smal l , zero for infini tely
narrow resonances ) and whe ther the pion exchange term (which is real) should
I t mus t be emphas ized that a -channel poles are exp l i ci t ly mean t , For i f
an in tegral ove r a con t inuum i s also included nea rly all f un c t ions A ( s , t ) can be
s o rep re s en te d .
from the s -channel poles ) mus t i f analytically extended , have poles a t the s ame
2
mass for the same quantum numbers as the s -channel resonances we have s u ch a
seve re res t r i c t ion that p e rhaps only a nearly unique solution exis ts , and hence
a solu tion of the s t ron g intera c t ion phys ics could res ul t . Much work has been
done in this d i re c t ion - success h as not been a t t ained . Naturally the p rinciple
might be wrong -- pe rhaps s -channe l and t-channel resonan ces mus t be separately
added - or what amoun t s to the same thing s channel cuts are necessary if the
at low ene-rgy that amp litudes are dominated b y resonances , s o that some so-rt o f
infinitely narrow res onance approximation may b e a good s t a r t ing poin t . But
is this also valid a t high energy where in most collis ions many parti cles are
p roduce d ? Can they be unders t ood as the prod ucts o f cas cade dis integration of
78 Photon-Hadron Interactions
If we suppose the photon and pion each couple to one quark - but each to a
different one , then quarks a and b have large momenta and the amplitude that the
final configurat ion a, b , c is in a s imple pro ton s tate , where they presumably
have small relative momentum is very small . Thus at high energy the most
important case is that a single quark a receives the photon and emits the pion :
a b
The net momentum c gets is only the momentum trans fer k-q and the ampli
tude a b c is a simple pro ton is large if , and only i f , this trans fer is small .
Therefore the pion goes forward in a way characterized by a function of the
momentum trans fer .
At any rate a dynamical theory of s channel resonances alone might not
find it too difficul t to ob tain apparent t-channel exchange effects .
Estimates of Coupling Constants
In a number of cases we es timated s izes of coupling constants . For
example , we said p -exchange was mainly flip from a nucleon . Such things come
from theory outs ide the impact parame ter , or Regge theory which left such
cons tants undetermined . They could come from o ther experiments (e . g . flip is
large can be go t from � nucleon scattering) but nevertheless theories are
availab le for what to expect . We used mainly su 3 , but even here F/D ratios
must be known - and these are given well by su6 , or what I have called a
quark model estimate .
The "quark model" used here is really not the same as the harmonic quark
80 Photon-Hadron Interactions
model used for the s resonances applied to t resonances . We are not dealing
with t resonances but rather with t traj ectories . The quantitative theory o f
trajectories to be expected from any (in parti�ular from the harmoni c ) quark
model has not been worked out , Rather we are us ing the "quark model" j us t
t o count - t o make up su 3 and F/D rules ,
Even then we implicitly supposed some thing that many people have used
{but to our knowledge has only las t week been explicitly formulated by
Kis l in g er ) . We assumed the p (or w) traj e ctory was coup led in the s ame p roport ion
like currents and (2) the p ,w,+ t raj e c t ories couple like the p ,w,+ :.iesons do .
Stated more precisely consider the amp l i t udes for the fo ll owi n g diagrams .
A x , ·B
,/"
//
x y y
The ratio of these amp l i t udes is < B j A> l l I < D l j l C> wherc j is the current with
p quantum numbers . We have assumed the amp li tudes have no t been correc ted for
ab sorp t ion .
Ano ther formul at ion of this "quark counting" game is to draw diagrams for
implying that baryons are QQQ and mesons are QQ in the usual way but that the
in teraction can be largely under s t oo d as a sum of terms each rep resenting the
It is clear tha t our interpre t ations of highe r energy s cat tering comb ines
t -Channel Exchange Phenomena 81
several ideas . What is evidently lacking today is some overall view that com
bines these nicely togethe r . I n the past too much at tention has been paid t o
forcing the ampli tude into one mold or another rather than combining the more
well es tablished and cent ral ideas o f each of them into some synthesis .
Vector Mesons and Vector Meson
Dominance Hypothesis
+ -
For example if the hadrons is a pair of K ' s , K K the cross section is
very small unless q2 is close to (10 2 0) 2 (ie m , 2 ) and then rises way above
background in a beauti ful Breit-Wigner curve of width 4 . 7 t . 7 MeV . The
interpre tation , of course , is that we have gone through a resonance , the ' via
82
Vector Mesons and Vector Meson Dominance Hypothesis 83
a diagram like
To analyze this have two ampli tudes , one purely hadronic <K+K- 1 4>> and
we
2
r-?"
4 2 e «jl j J j O> at the mass of the
a pho ton amplitude which we write as lf4'1fe- µ µ
4> , e is the photon po larization . The second factor may be writ ten
I'
<fj> j J j O> "' F9 elJ where e µ :!.:i •he pu.i.d& � ot .. �ion vector of the vector particle .
JJ
connection ( for any vector meson V of mass mv ' neglecting electron mass )
(a •l /137) i s
-3
2
a2
rv+ e+e - • - 3- Fv mv
2
mv (14 . 3)
4na s -
3
One can look in this energy region for other products (such as 3n , or n°y )
and find the resonance again - thus determining the hadronic amplitudes for
fj>+3n or 4>+n ° y , usually given in the form o f branching ratios .
In the case the final state is 3n there i s another resonance at 785 MeV ,
the w meson - it is studied in the same way .
Again studying Zn final s tates a large resonance near 765 of width around
125 Me V . This large width makes considerable ambiguity ( in assumptions of how
84 Photon-Hadron Interactions
r q
z
varies with ) in determining the cons tants . mass and width . Further the
curve is asymme t ric , sharper on the high energy side . A small shoulder is
nearly apparent - this effect is interpreted as interference with the w
resonance assuming there is a finite amplitude for ..,.+Z11 . (This violates isospin ,
and is an electromagnetic effect which we will discus s in the next lecture . )
Values for the various constants for the vector meson may be found in
the Particle Properties Tab le , There is now some additional data from the
Orsay Storage Rings (J , Lefrancois , 19 71 International symposium on electron
and photon interactions at high energy) . Some of the differences from Particle
Table results are new data but some are due to an altered way of reducing data ,
especially for the p. We shall have to wait for new particle tab les to thrash
these differences out .
B • Branching ratio
New Particle Tab les
4> B { cf>+n°y) • (Z . l :!: . 7) x 10 -Z B ( tj>+i(IC-) • 46 . 4 :!: Z . 8%
B (tj>+11°y ) • ( . Z5 :!: . 0 9 ) x 10 -z B ( 4>+ 1S_ - 35 . 4 :!: 4%
�)
• :!: Z . Z) x 10 :!:
B { tj>+311) ( 14 . 7
-Z B ( cj>+311) - 18 . Z 5%
r tj> total = 4 . 7 :!: . 7 MeV 4 .0 :!: .3 Me V
• :!:
Z
:!:
z
p mp • 780 6 mp 76 5 10
r • 15 3 :!: r • :!: zo
p
total 13 MeV total 1Z5
p
r +e e •
+ -
r +e e :!:
+ - . 5 keV 7.5
p
• 6.1 :!: . 9 keV
p
but Orsay makin g "corrections for finite width" reduces the data in some other
way :
g p 2 /411 2 . 56 • :!: . 22
2
g w /4 11 - 19 . 2 :t 2
Note . Now that we have data for the rates t+ny and .,..11 y we can compare the
predictions of the quark model with harmonic dynamics of Feynman , Kislinger
and Ravndal .
Quark Model Orsay
r � ny = 1 . 79 x 10 -4 GeV 1 .0 :t .3 + 10 -4 GeV .
r w11y • 1 . 92 x 10 - 3 GeV .6 :t .2 x 10 -3 GeV .
Lecture 15
We will make a di g res s ion from our main discussion of vector mesons to
_!. (uu + dd) isospin zero but has a small admixture of I p 0> • ...!. (uu - dd)
rz rz
isospin one in i t . This mixing is due to electrodynamics .
We consider the w - p system as a two -s t a te sys tem where Cp0 is the ampli
tude to be in a p0 state and Cw0 is the amplitude to be in an w0 s tate .
( 15 . 1)
H •
( m
w
- lli
2 6 )(
- 784-i6
-2-
irp
6 m - 6
p
l w > • l w0 >
+
6
19+56i l p o > ( 15 . 2)
86 Photon-Hadron Interactions
I
o p +2n
I • 2 .97 x 10 -3 o 2 MeV -2
Prob . w+2n
Prob . w+all • 19+5 6 i w+all ( 15 . 3)
0 12 n.
we ob tain the following electromagnetic self energies
p 0 : • (4a + a) - 21 (4b + b) • 5a - 5b /2
w0 : • ( 4a + a) - } ( 4b + b) • 5a - 5b /2
p+ : • (4a + a) + 2b • 5a + 2b ( 15 . 4)
• 5a + 2b
*
K o • (a + a) - b • 2a - b
w
0
Sa - 5b /2 3a - 3b /2
w0 3a - 3b/2 Sa - Sb /2
Vector Mesons and Vector Meson Dominance Hypothesis 87
S ince K *+ - K *0 • 3a + 3b
+ 0
p - p - 9b / 2
*+ - K *0 ) - (p + - po ' )
we have 3a - 3b/ 2 • (K (15 . 5)
in terms o f measurab le quantities .
o' o
No te that in eq . (15 . 5) we have writ ten p · instead of p , this is only
to indicate that p o ' does not contain the contribution from the annihilation
2
term which we calculate next . To first order in e the change in mass is given
by
2
llm 2 2- <p j J j o> <o j J j p >
-
4TTe v v
p
- (15 .6)
q
or
4TTe 2 F 2
llm 2 . - z- (15 . 7)
p m p
p
= 1 . 53
p
llm MeV
Let x be the amplitude for _!. (uu - dd) .,. , the amplitude for _!. (uu + dd)
./2
y .,. y
./2
is x/3 ( see lecture 16) . The mass matrix due to the annihilation term is there fore
proportional to
Po w
0
2 2 ( 15 . 8)
Po x x /3
;
2
w
0 x /3 x2 9
w0
1 .53 . 51
MeV (15 .9)
.51 . 17
88 Photon-Hadron Interactions
Adding the contributions from the electromagnetic self mass and from the
annihilation term to the non-diagonal matrix element we find
01
5 = .5 MeV + (K*+ - K*o ) - (p+ - p ) ( 15 . 10 )
and s ince
( 15 . 11)
5 • -1 . 0 2 MeV + (K*+ - K* o ) - ( p+ - p o )
0
from data in the particle tables , (K*+ - K*0 ) • - 8 ±3 MeV and ( p + - p ) =
-
2 4 ± 2 . 1 MeV .
. But we do not trust these results , especially the p mass
difference , we only say that they indicate that 5 is likely to be negative .
'
. r;;:i
The yp coupling of the form y+ Tre- F e lJY e ' is no t gauge invariant and
lJ
2
would give a finite mas s for the photon via
y p y p y
�
The coupling is A YB P • However we can couple the fields Fµ v of the photon and
lJ µ
p , for examp le . It leads to
( 15 . 12 )
The latter term has no pole singularity and is "los t " in the background of
other than pole term and effectively we can use a residue j us t proportional
to ey • e p ( i . e . 2 C p mp 2 Fp ) •
=
Lecture 16
Vector mesons (continued)
We now return to our discussion of vector mesons . We can see what su 3
2
gives for the ratios of the couplings 8v given at the end of lecture 14 . We
Vector Mesons and Vector Meson Dominance Hypothesis 89
use the simple quark model ( for cowiting only) and assuming the + is purely
st range quarks , J couples to QQ pair proportional to its charge
• • SS + - 1/ 3 • - 1/3
Noting that these are values for the coppling F , and that g is the reciprocal
we f ind
: g (IJ
-2 -2 9 : 1 2
: g' :
V arious people have tried to correct this for su 3 breaking , but nobody
really knows how, There are two questions . The first is , to what extent is
the ' pure ss? I see no way to determine that . The low value of the ' + 2w
b ranching ratio is interpreted in the quark model by saying that the ' being
made of purely s trange quarks finds it hard to go into non-strange obj ects (w) ,
More important is the question o f su 3 break ing because the masses are not
equal , For one example of ambiguities, should we compare F ' s directly with
v
the su 3 predicted ratios , or should it be Fv /m , or F /m or what , which
2
v v v
should bear the s imple ratios? su 3 cannot tell us and nobody really knows how
to calculate i t , although various guesses have been made . (I prefer F,, / ll\,
for a theoretical reason , but one which is really not profowid or necessary} .
These ques tions probably do not strongly affect the p ,w ratio , as the masses are
c lose together . As you see the ratio 9 : 1 is not very far of f , (and , in spite
of the wicertainties , the ' is also about right ) .
p (the only one of the three ' ' w, p which couples to the pion) as
1
""'" fp 11 11 (p l11+p211 > q2 2+ir m
-m
F
p
p pp (16 .2)
�
m2
{f p 1111 / gp ) 2 � {p
-mpq +Lr P mP l11+p2 11 >
when q2 ::::. mp 2 • One would ordinarily expect other terms {such as a direct
coupling of photon to pions or various other "intermediate states") to be added
to this , lbis , of course, would not change the behaviour near the pole (which,
unfortunately for the p is not so definite as the width rp is rather large, which
makes many difficulties in practice in fitting it, nevertheless in this theo -
retical discussion we shall neglect it) . Another way to say it is to suppose
the numerator has a factor t{q2 ) which varies with q2 such that t{mp 2 ) · � or
again to say the "constants" fP 1111 or gp vary with q2 only their value at q2 •
mp 2 being defined. All these ways say the same thing so we shall not argue
about them.
Nevertheless a bold hypothes�.s has been suggested (vector meson dominance)
that this expression with fP 11 11 /gp constant is all there is to it - there are n�
other terms. I see no good physical reason for this. It is made in analogy to
another mysterious hopothesis of the same kind, PCAC , that works . lbere we do
an analogous thing with a pion pole - but only extend it from q2 m112 .02 • •
at the pole, to q2 0, Here we suggest eq. (16 . 2 ) is valid not only near q 2
• •
We know for long wave lengths the pion looks like a simple point charge
so that as q2 0 (use crossing to convert <11+11- I J l o> to <11+I J 1 11+> we have
+
II II
2
< 11+ I J11 1 11+> • (p1 + p 2 11 > for q + O ,
11
(16 .3)
hence for (16 .2) to be true we must have {neglect problems generated by rp ,
or suppose rp depends on q 2 and vanishes at the (2m ) 2 threshold and below) .
11
• 1
Experimentally fp21111 / 4 11 • 2. 43
Vector Mesons and Vector Meson Dom inance Hypothesis 91
= 2 . 56
well wi hi the 10%
t n \lll ce r t a i n t i es t q u t ti s .
in d e t e nnining h ese an i e
2
It would also imply that the form factor F l q 1 for the pion, written Tr
example in interactions with the nucleon the part due to isovector would be
<N I J I N> isovector
2
..£....
l m
( 16 . 4)
µ gp • -
is defined only at the p pole, and there has an electric and magnetic part so
we can write as <N j c1y µ + c2 f (y �-(y µ ) IN> Thus , expressing the lsovector
•
part of the current coupling in terms of the usually defined fonn we f a c t ors
1 )
( 6 .5
are testing (1) for large negative q 2 - far from the p pole and it fails .
Obviously VMD canno t be an exact principle for all q2 But maybe it is nearly
valid from q 2 mp 2 to O . We must discuss extensions of the idea and
• f u r th e r
evidence.
A further conclusion from the nucleon case above is that c 1 /g • 1. There
p
is no way to get he p t up ling he nucle n , c1 or c2 , without ambiguity
co to t o at
present (or ever? is it a matter of definition <N j p l N> cannot have all the
particles physically mass h l l ! ) . Some kno ledg ,
on the s e nN scattering w e from
for example, exists for the ccupling of the p-like trajectory - or for what we
call p-exchange - but the extrapolation from p-trajectory to p-pole exchange
is ambiguous . The latter (p-pole exchange) has an amplitude varying s but as
the terms associated with p trajectory exchange do not , and they are not supposed
to in Regge t or ts l f .
he y i e
coupling, Since the mass of the w is nearly equal to that of the p , the
isoscalar and isovector form factors should have the same dependence as q2
they do even though this dependence is not the (l-q2m 2) -l expected .
p
Lecture 17
(l7 .2)
That the left side of (17 . 2 ) gives the total of all the ways that J µ can
couple is the general statement of the hypothesis of vector dominance. We have
already seen how it predicted the p n n coupling constant. Can we do a similar
thing for the $KK coupling, say $K+K- ? We know that the current operator at
q 2 � 0 might be written (p1µ+ Pz µ > with coefficient 1 because the charge is 1 ,
so we have the sum o f lg (K+K- / V) muat equal 1 . Thus i f we write
v
(l 7. 3)
we would have
(17 . 4 )
The problem now is that although we do know the three gv (from experiment) we
have too many unknowns in ( 1 7 . 4 ) to relate it to experiment . (The rate that
$ +K+K- implies fv + - / 4 w = 1 . 4 7 ) , We need some rule to partition the bits of
K K
the d quarks , or the s quarks (as though each had unit charge) , Then , for
.!.
likewise an w-type Jw • 12 (Ju + Jd ) and a ' type , which since ' • ss is jus t J' •
(17 . 6 )
el
Now obviously w e suppose the three p ieces into which J i s split i n (1 7 . 6)
are the corresponding vector meson resonant pieces of (17 . S) . This may be writ ten
in a very s imple way , if the gv really have the ratios of the quark model (Lecture
( -m;2
14) we are saying, of course
<Y j Jv j X>
µ µ
• cans t , <YV I X> ( 17 . 7 )
q 2 -m +ir m
v v
Ju . i + .!.2 Jy + JB
z
Jd • -J + .!.2 Jy + J B (17 , 8)
JB - J
y
+ JB
(because if you substitute for example the quantum numbers of the u quark , Iz •
+ 1/2 , Y • + 1/3, B • 1/3 for the currents you get one for J , zero for J
u d
and
s
J , etc , )
Hence
94 Photon-Hadron Interactions
(17 .9)
lb.is then defines precisely what we mean by the currents in (14.5) and
(14 .6) .
For example, let us apply this to the K+K- de ay of the + . We need the
c
(17 .10)
Thus combining this with (14.9) and (14 . 7) we see we mus t have
defining how the total of (1) i n {17.4) is partitioned. A t any rate we predict
(f + 2K+K- /4n} /(g+ 2 / 4n) 1 / 9
•
2
which agrees perfectly with the f
+ KK/4n • 1.47 from + + K K and s, 2 / 4w 1 3.3
+ -
a
from + -+ e+e-.
How s low is + + J n compared to w + 3n7 ' goes 18% t o 3n, b u t r is only 4 MeV
so the partial width is only . 7 MeV while i t is 10 MeV for the w . Phase s pa ce
is lar ger for the + but more detailed analys is would require knowledge of how
and P1 , p 2 , p 3 the four vectors of the three pious the amplitude must be cons tant
e e p p . The " cons tan t" for the ; has to be about 0 ,l of i ts value
µ vap µ P lv 20 3P
for the w .
Lecture 18
rather how can we de fine this in terms of quantum numbers or rules without
Vector Mesons and Vector Meson Dominance Hypothesis 95
referring to the quark model? We don't know how. We try to say that it couples
weakly to states which have no strange particles in them, so not to 3n but yes
to KK. Yet such an idea "a state that has no strange particles in it" is not
readily definable . Due to strong interactions a state like KK has no overall
quantum numbers to distinguish it from 3n ; in fact via virtual interactions
the KK should couple strongly to 3n (for example through a virtual w , KK ++ w +-+ 3n . )
There fore the ques tion remains , what keeps the � from coup ling s t rongly with 3n 7
From point
the of view of a to that s state
quark model we might t ry s ay the s
which mix the $, state. These diagrams cannot be calculated, (the above
w
12
Of course it is always possible that the world is simply complicated,
that the ss
comb ina tion but the
is not i s o lated by principle in the beginning ;
very virtual interactions select some linear combination of ss and !..._ (uu+dd) to
12
be the $ (and the orthogonal combination for the w) and it just comes out that
the combination is ss .
Then perhaps it is really possible to understand why $ + ny is so small,
but is the smallness of $ + 3n really obvious in this model of the state?
But still more striking is the fact that this "accident" is repeated again!
96 Photon-Hadron Interactions
For the mesons of spin 2+ where the f (1260) goes nicely into 2n or 4n and into
KK. The f ( l514) goes predominate ly into Kif and even into KK*+ KK* and no t into
nn in spite of the even larger phase s pace , (The quark model rates (Feynman ,
Ki slinge r , Ravndal) us ing pure !._ (u'U+dd) for the f ( 1260) and ss for f (1514) are
ri
all high , b ut the relative proportions are right . )
The experimental and quark model (FKR) rates for 2 + mesons are :
Experimental FKR
Kif "' 8 13
2 n +2 n - '0
f ( 15 14) + Kif 52 10 3
Kk• + I<K• 7 15
nn < 10 0
nnn 13 ± 7
nn < 30
NOTE : There are in addition two hadronic phenomena associated with the �
which are unexpected un les s the � is , like as, weakly coupled to hadronic s t a tes
: �· :
I t is very weak , as though the � were not coupled to the N , N* (a nucleon
traj ectory) j unc tion . Again p + p + ' + nn is very s trongly suppressed
( relative to p + nn for example) as though the ' ' s s and s cannot be made from
the non -s trange quarks in the o ther parti cles .
Comple te vector dominance implies that the photon cannot interact with a
in general) interacts with the hadrons . Thus we would expec t to find some kind
There are several ques tions involved in us ing the formula . Firs t , there
2
are theore tical questions of mean,ing . S ince y has ; • O, the vector meson
Vector Mesons and Vector Meson Dominance Hypothesis 97
2
on the right s ide should be a vir tual vector meson of q • O , for which s t rictly
no de finition can be given . It might be argued that s ince mass on the initial
s tate incoming particle might have the leas t effect at h igh s , (where E ';:: p
anyway for finite mas s ) this relation may be mos t nearly correc t for real vector
mes ons on the right -hand s ide at very high s . Another related ambiguity is
that the photon has only two polarizations , helicity ± 1, so the equation has
theorists have come to the conclusion it is in the a -channel frame that the
large s als o . Thus we shal l , to avoid much comp licated dis cus sion , limit our-
The next ques tion is how we are to ob tain the vector meson cross sections
or ampli tudes V + A + B, after all V meson beams are not available . Some t imes
theo ret i cal arguments are available , but i f they are too complicated they are
not useful to tes t VDM . The most useful s imple cases are :
p + Nucleus + p + Nucleus .
reversed reac tion n + N + p + N , which should have the same amplitude . This
the appropriate mass region of the two outgoing pions . Some correc tions mus t
be made for pairs of pions , say in mutual s waves , which are not due to
virtual p decay . However the data at low t has been nicely analyzed in detail
theoretically and we can describe the results , and compare to VDM expectations .
0 0
and y N + n N is related direct ly by VDM to n -p + v n where v is a vector
we drop i t , and the w f interference . VDM then predic ts for example that (by
98 Photon-Hadron Interactions
+
(yp+w n) +
da -
2 dt
1 [ da (yn+ 11 p) ]
dt
These [P 11 J are density matrix elements to project out the helicity � 1 reactions .
-2 -2
p "'
Data in the "' reaction are lacking , but g / g is about 9 / 1 so the second term is
(w
da - o
pro b ab ly never large . In addition , at 8 GeV p + p n) is itself about
dt
10 t imes
da
dt
(w -
p + "' n) in the forward dire c tion , so the "' contribution migh t
P -lil interference term ( for an amplitude of order .1 interfering with 1 can make
a 20% effect , but its s quare is only 1% ) . Tha t is the reason that the comparison
is made to the sum of yp w+ n and yn + 11 p cros s sec tions - for in that sum
-
+
the p -w interference term ( f rom isospin cons iderations) cancels out ( as does the
p -+ interference) ,
The cros s sections for all the helicity combinations of the p have been
measured ( and reported in the form of dens ity matrices) by measuring the angular
+ -
distributions of the pair of pions lf lf resulting from the dis integration of
the polarized p2 • We also have data with polarized photons ( perpendicular and
parallel to the plane of produc tion and so can make two comparisons . The vector
41fe 2
da
(yN wN) • -z ( p l l + p l-1 ) dt c w -p + p n)
.J. da o
dt + ( 18 . 3 )
g
p
da1 2
( yN ( w p + P m)
"'"'d't
1 4we [ p - p da - o
l-1 J dt
+
11N) · -- ( 18 . 4)
g 2
ll
p
and the p ' s are density mat rix elemen ts in the helicity frame , The results of
�r- - ttf t
4 1/2 [crh + CTy11] + 112 ["";, · err,!
.. <:!• 300
�
t !V!
Y,2 (P,II-P,H J (g_di<r) + ,(P,, •P,.,)(�)
I'
,u
� p � p I'
lj
200 E
!:'!
D
:I.
-t D
:I.
200
lll;; lll;;
<:!.. <:!.. I
---'r�ww-------------------------
100 -
l1i 1/IfI + 11
.. .. .. ..
:I! :I!
'
100 • 0
!!! !!!
I
•
0
I + .. . . . . . . . .
----1..:._"_ _J _ l__ _ _J___J_ __J____L
0 o. 4 o.e o. 4 o. e
k GeV / c k GeV/c
The results sre seen to agree in exactly the forward direction and generally
for the unnatural parity exchange a11 , but clearly disagree away from t • 0 for
".i · Thus VDM as a general theorem cannot be exactly correct (barring the poss i
Thus photons do not couple exactly as off mass sbe.11 vector mesons . But
j us t how do they couple? To make progress we should s t udy the nature of the
deviations f rom a s t rict VDM and find their characteris tics . For example had
develop - we could say why these various cross sections agree for a" and for
the forward direction , and why not for � ? Perhaps some clues can b e ob tained
The values of t we are s tudying here are so smal l that single pion
were exactly true without absorp tion corrections we would expec t a re lation
diagrams correspond:
11 \_J N 11 LJN
Y /\ N p nN
di ffe'rence of p , y and selection
2
The only difference ( as ide from the mass
of the correct helici ty ampli tudes to compare) would be that the y is coupled
to the p
lµ
+ p
2µ
of the pions via a facto � and the p via f
P1111• Hence
to replace (18. 1) ,
and corresponding relations for the cross sections (i .e . (18. 3) ,
(18.4) 2
with g p replaced by f 2 ) . Therefore insofar as these relations agree
p 1111
for small t we have no t made
any new tes t of the vector dominance relations other
than that f
P 11W
2
� g
p
2
, a test that we have al ready seen works very well . If
that equality of coupling cons tants was an
acciden t then the low t agreement
"'" 'uld
only mean that s ingle pion exchange dominates both reactions at small t -
someth ing we expect to be true on other grounds .
On the other hand even at low t , .r.:f • m11 for example, t he two theories
deviate - how could that come about ? The one pion exchange term mus t , of course ,
be corrected for absorption . The differences between a yN an d a p N at low t most
likely can be attributed to differences in the degree to which absorption
u.odifies the one pion exchange expectation .
In the first place (see Leith ' s report) a fair fit to the p data is given
by a one pion exchange model co rre c ted fo r absorption (due to P . K . Williams ,
Phys . Rev . Dl 1812 (1970)) for the amplitudes that concern this gives us
(18.6)
and
(18. 7)
A.J. � 0
disagreement with experiment) . We really do not exactly know why the sub-
tracted term for absorption should always be the full 1/2 . Possibly in
this case extra contributions (isoscalar or isovector?) are contributing to
the photon case to increase the 1/2 appropriate to the p case. The point
here is an effect different for p and y and hence different from expectations
of VDM (unless it could possibly arise from the w term) . However the correct
cons tant term to use is determined experimen tally by the cross section for
P. K . Williams had suggested such a factor for absorption effects but expected
a much slower fall off (like e) (t -mn > ) . To what is such a rapid fall off due?
2
also of the possibility that the source of the pion is indefinite by the size
of the nucleon or the will explain it. The point of explaining it, however,
p,
is that the drop off is much slower for the photon case . Therefore whatever
the cause it works differently in p and y pseudoscalar production. In its study
lies the possibility of understanding physically where the ideas of VDM go
wrong. The problem cannot be difficult - the effect shows up for small t and
1 02 Photon-Hadron Interactions
hence for large impact parameter and therefore in a realm where physical
Lecture 19
Diffractive P roduction of p , w, t
here . I will not go into as much detail in the results as we are accus tomed
the cross section approaches a cons tant at high energy and has the typical
dependence of such scat tering . But how does the photon di ffract into a p ? One
answer is provided by VDM , from the expected ampli tude relation (18 . 1) we expect
2
do o 411e
( N + P N) (19 . 1)
dt y " -- 2
g
p
(because of I spin change it is expected that w0N + p 0N or t0N + p 0N will fall
rapidly with energy so only the p term remains ) . We do not know directly the
scat tering .
A crude use of the quark model supposing the quarks to scat ter independently
of their spin direction and to be similarly dis t ributed inside the p 0 and 11°
suggests a (p 0N) • 0 (11°N) . This latter is not known by direct experiment but
1 N � 11 N)
- - +
0 (11 N + N) ] ,
+
isospin gives it as It does turn out that the
Z ( 0 ( 11 + 11
total cross sec tion for yN + p 0N varies with energy in j us t the way 2 [ 0 ( 11 N)+ 0 ( 11 N) ]
1 + -
behaves , In fact the total cros s sect ion is given correctly by this rule and
in which the y becomes two pions not at the p resonance and one pion s cat ters like
y
�
Vector Mesons and Vector Meson Dominance Hypothesis
11
11
103
11
N N
ins tead of
11
y
11
A strict adherence to VDM would , as far as l can see expect the analogue
but the point is in es timating j us t the effect of the s imple p elastic scattering
so we can use our 11 ° scattering analogy .
1bis idea that yN + pN tells us about pN + pN permits us to summarize the
behavior in another way . 1be polari z ation of the p coming out can be measured .
l t is very nearly purely transverse and polarized as the incident y ray . 1bis
is excellent evidence that the process of elas tic s cattering (also called
pomeron exchange) does not change the helicity in the center of mass sys tem .
w production . Here the diffraction term is smaller so other processes are also
effective , in particular one pion exchange at the lower energies . 1be natural
parity exchange has besides the pomeron exchange also poss ib le � exchange so
the purely diffractive part has not been separated out yet for a clear tes t .
+ production . 1be experiments show some inconsis tency but data is availab le .
One way to test the relations of VDM i& to compare the forward differential
cross section for yN + VN with the forward differential cross section for
VN + VN . 1bis latter can probably b e es timated fairly well from the optical
theorcom using quark model estimates for the total cross sections for VN .
2
the g
$
seems too b i g . It could mean the quark es t imate of a 4N i s too b ig -
10 mb would f i t b e t te r .
Lec ture 20
Di f f raction production of ve ctor mes ons can also be s een , and more
Pho toprocesses and ve c t o r dominance " , Cornell Confe ren ce , 19 71 . ) There are
difficulties for the w since the final 3n s tate is hard t o measure , and the p
is so wide that theore ti cal que s t ions of in terp re t ing data are involved . The
cleare s t case expe r imen tally is $ product i on , however much more data exis t s
for the p , In these exper iments on p p roduc tion obse rvations are made o f 2n
effect with w + -
+ Z n j us t as in e e production .
The 4> data is now goo d . The dependence o n A , the ma s s number o f the
a
4>P
nucleus , gives us some i dea of , and the ab s o lute cross s e c tion pe rmits a
a
2
g / 4 n is its Orsay value 1 3 . 3 then we can only conclude that may be
4>
perhaps from - . 3 to - . 5 and a4>N in the range 8 t o 14 mb (at about 7 GeV) ,
pos s ib ly a b i t lower than the quark model rules would sugges t , but lit t le
can be said .
The p data is more extens ive . Here the large width causes some confus ion
/4n • 2 . 6 and a p N • 27 ap N
2
the "good value s " (a t 8 . 8 GeV) g mb . ( But is not
p
well de termined and res ults depend on i t . ) There fore the phenomenon o f ve ctor
meson produc t ion exi s t s and behaves closely like a diffraction proces s , and
coul d wel l be in quan t i t ative agreement with the VDM . Does this tes t the
Vector Mesons and Vector Meson Dominance Hypothesis 1 05
mo de l ? I think not because formula ( 19 . 1) applied to dif frac tion s cat tering
can be derived ( at very high energy ) from ano ther hypothesis - namely that
the elas tic scattering of the p is much larger than diffractive dissociation
pe rhaps only 30% of the e las t i c s c a t t e ring f rom the nucleon s o there is
no reason not to assume it for the p , Fur thermore for the p we shall only
spin 1- ( like the photon ) is small compared to the elas tic . There is also
any isospin one , 1- s tate , produced in a diffrac t ion dissociative way does
For production on nuclei our assump tion ( of the dominance of elas tic
diffraction dissociation ) is ever more true as A rises - for the elas tic comes
from diffraction from the shadow of the entire nucleus - whereas the
We can see how this works most clearly in the case of scattering from
a large nucleus where our assumption is most nearly valid . Ins tead of
The waves of the incoming p come in from the left and fall upon the
nucleus where they are ab sorbed . A short dis tance (which can be many
wave lengths if s + � beyond the targe t say at the dot ted line the wave
function is nearly the original p wave funct ion with a hole as a function o f
106 Photon-Hadron Interactions
b , the impact paramete r . Beyond that the waves slowly diffract into the
shadow in a way determined by Huygens principle . If 1 - a(b) is the ampli-
t ude to find a p at b (so a is near 1 for b smaller than the nuclear radius ,
and near 0 beyond) the amplitude for an outgoing wave of transverse momentum
Q is ( l-a (b ) ) e iQ • b d2b • f (Q) . The amplit ude at P is
f
ikr 2
1/1 (P) • f (l-a (b) ) __ d b (20 . 1)
e approximately
r
Of course the 1 gives only the forward beam and doesn ' t interes t us - i t
is j us t a s i f the p came out only across the face of the nucleus . To be sure ,
if the p were not a poin t particle but say made of parts ( indicated by the
braided line) we could no t precisely define every thing in terms of an amplitude
to give a p , but would need a function of all the parts as variab les . But
clearly for b outside the nucleus the parts are in the same relative amp litude
as in a p - only near the edge is there some distortion of the re lative behavior
of the parts - and so a proj ection possible into some other than the p . But
the amplitude a (b ) is clearly that for a p except near the edge ; so that elas tic
scattering is much larger ( for nuclei , at least ) than diffraction dissociation .
N ow we have no ted from experiment by e+e - that whenever we have a p
/
pho ton) and therefore there is a source of the electromagne tic field - a current
given by j µ Y 4'fle F •amp. to find p polarized in direction µ , (F • - M /g ) .
2
=
P P p
Thus in our prob lem at each point P there is a current source of electro-
.p
magnetic field of size l/l(P) - thus the total amplitude for finding a photon
going out with wave vector kout is f l/l (P ) e out d P . If we subs titute our
-ik 3
-ik r/ r and e -ikout. p j us t
express ion for ljJ (P) in this the convolution o f e
-ik ·b 2 m 2 and k 2 • 0) so we even tually ge t the
1
gives e out 2 • (But k
•
k -k t
2 p out
outgoing amplitude f8� photon is
�F p f (1-a (b ) ) e
-i Q out .p 2
d b (20 . 2)
---
mp 2
where Qout is the transverse part of kout " The integral is the same as for the
p elastic scat tering so the amplitudes are proportion. al - the cross sections
2 2
mus t bear the ratio 4 'fle /gp •
One can of course carry the mathematics out in more detail separating
Vector Mesons and Vector Meson Dominance Hypothesis 107
the z and transve rse momentum care fully to see j ua t how the as s ump t ion o f
wave in to the shadow is omi t ted . Then the wave going forward behind the
ikP z
z
shadow is j ua t e ( 1-a (b ) ) where the in cident p is of energy E and
p
k ::
z
j�p
• E "'111
p
2
/ 2E . The amp li tude to make a photon of ene rgy E , transverse
iQ · b
2
( 1-a (b ) ) e d b
2E
411e F
J'X
f (Q) -
p
f (Q) (20 . 3)
2 2
m _Q
p
A l/ 2E factor has been included to t ake account of the relat ivis tic l / lfi
2 2
normalizations for the incoming and outgoing particles , The m - Q is wrong
p
2
p
here , the dif frac tion into the shadow makes it m (as though the
p
k were E for p ' s moving in th is dire c t ion ) .
z
It is evident tha t the polarizat ion of the 'I is the same as that of the p
for this proces s . Th i s feature has been checked expe rimen tally very care fully
and is ve ry st riking , the p polarizat ion p roduced by polarized pho tons is nearly
Of ten it has been s ugges ted that VDM predicts its formula (19 . 1 ) via a
�
diagram like
1'1ucleus
m
2 2 + -
q • where i t is measured in e e . It is a tes t of VDM to find tha t this
p
constant is unchanged in this range . But we· have seen , by cons iderat ion o f
the reverse reac tion , neglecting dif fraction dissociation that the appropriate
mp
2
cons tant should be that of and no extrapolat ion is involve d .
J. Mandula has con firmed my argumen t by an argument from dispers ion theory
the edge , the parts of the p may not be in precisely the same relative mo tion
to produce a p,
but they might s till produce a photon , Clearly this is small ,
2
It means an in terfering term o f order FP * /mP * times the amplitude to make p *
where (FP * i s the coupling of p * to photons ) summed on p * . For nuclei at leas t
it must be small coming only near the edge . The point is not so much that F *
P
is small (which it may be , as so little p* seems to be produced by photons on
nuclei (which measures F p *2 a p *N ) but that the number of p * produced by p via
diffraction dissociation is small .
Other Tes ta o f VDM
If VDM were correc t we expect the pho tons to have amplitudes l/ "v to be
various vector mesons so that the total cross section of y ' s on p is l / g,, 2
times the total cross section of each of the vector mesons . We expect
2.
2
° t 0 t ( Vp)
0 tot (yp) • 411e (20 . 4)
p , 111 , 41
-- 2
"v
(neglecting possib le interference e ffects of w and 41) , This is dominated by
the p (because of the 9:1:2 ratio of s.y2
is small) . We es timate and a ( 41P )
a ( p p ) • o (wp) by the quark model as equal o (11 p) (a s cheme which we che cked
°
t t t
in our a (yp -+ Vp) con s i de rat ions above ) . We find experimen tally that o tot (yp)
is too large by 40% , as though o ther processes were available to the photon to
interact with the proton besides going through virtual vector mesons .
IL
get the inequality
2
1
1
do/ dt (yp �
2
2
do
-+ yp) 4 11 e - (yp -+ Vp) (20 .6)
Bv
p ,w, dt
The quantities on the right hand B ide are available directly by experiment .
The dependence of do /dt (yp
0
-+ Vp) on s ( from 2 . 7 to 5 . 2 GeV) and on t ( from
ia
2
O to . 4 Ge V ) are well represented by the right hand s ide , but it always only
about 1/ 2 of the experimental cross section ! Thus VDM fails here .
This ia in accord with the 40% error in the total cro s s s e c t i on text , for
the optical theoreom relates do/dt (yp -+ yp) in the forward direction to the
s quare o f o tot (yp) (assuming something about phases ) .
Vec tor Mesons and Vector Meson Dominance Hypothesis 109
Lecture 2 1
Shadowing in Nuclei
How should the total cross section for y+ Nucleus vary with the mass
nuni>er of the nucleus A ? We know for collisions with hadrons nuclear matter
is nearly opaque (because a hadron-nucleon is comparable to the spacing of the
nucleons) and therefore the cross section for A nucleons is not the sum of
a contribution from each (hence a tot A a nucleon) because nucleons in the
�
front shadow those in back so ultimately for large A it goes as the area of
the nucleus or A2 /3 .
On the other hand for photon-nucleus collisions� first glance would
suggest that the nucleus now being transparent (ay nucleon is mu ch smaller than
nuclear spacing in a nucleus) each nucleon would see the full beam and hence
the cross section would go as A (times the simple nucleon photon cross section) ,
On the other hand the VDM shows that this latter conclusion cannot be
where the nucleon is located - but VDM reminds us that this ia not so. A y
can become virtual hadrons (e.g. 2 w , but most importantly a p ) far away in
front of the target, and the virtual hadrons propagate a long way to the nucleon
to interact, A real photon is a pure ideal photon plus a virtual hadron with
amplitude l/6E according to perturbation theory where 6 E is the energy difference
of the state of given momentum as photon and as hadron. Thus if it is a p
the energy ia � while that of the photon is v so 6E · VR - v "' mp 2 / 2 v ,
very small for large v . This is also the distance ahead of the nucleon where
the photon-hadron conversion occurs . If this is large compared to the mean free
path of p in the nucleus then shadowing occurs (A2 1 3 ) , if it is small, no
shadowing (A)•
The physical idea can be seen best first for a simple model consisting of
JJO Photon-Hadron Interactions
two thin s l abs a, b of nuclear matter one in front of the o ther by d . We shall
by pho tons via VDM. I f a i s not p resent the amp li t ude to find a p at b
i (k k ) (x-d)
( 21 . 1 )
Y P
-
_ r;:::r
4Tre F e
'V P
( the phase has been taken relative to the pho ton phase at x • d i . e . the
above express ion contains additional factor exp (-ik d) ) where the k , kp
y y
an
2
are the k-ve c tor for y or p at the s ame frequency . Hence - k • m / 2v for
y p p
k
large v (•k ) .
y
When a is present (2 1 . 1) is valid only for d > x > O. For x < 0 we have
(2 1 . 2 )
;
The total ampli tude (in tegrate x) is there fore proportional to
P Fe k -k
y p
(1 _
f
2
a
e
- i C k -k ) d
Y P (2 1 . 3)
problem with a continuum of layers would cancel out for (ky -kp )d > 1.
Thus the criterion for scattering involves shadowing should be complete
v,
for v + "'
One extend this easily to a thick layer in one dimension. We seek the
�an
amplitude to find a at x (for that will later be used by squaring and summing x
p
(the phases have been taken relative to the photon phase at x, i.e. the above
expressions contain the additional factot exp ( -iky x) ) .
Integrating over Y the amplitude at x is
1 -i(ky -kp ' )x { -i(k -k ' )x )
ky -kp e + ky -k1p ' \1-e Y P (21 .5)
Here kp ' is kp in the medium. At very high v it has an imaginary part (representing
absorption) which is finite and fixed, the real part goes , of course with v , but
may differ from it by a finite amount. Th.us as v + ky -kp k a fixed
"'
•
' °""
number (whose imaginary part gives the absorption cross section and whose
p
b een made . They show a v dependence of a to t (A) /Ao (nucleon) for vario us nuclei
1
-- c . 83
a t o t ( A)
- Cu
-
Ao t (N) , __
--
. 65
.s .53
to
. . ___ Pb
. 25
5 10 15 2 0 v (GeV)
The data does not show much v dependence - (but it is not far off ,
expe rimen t al errors are large enough , and the expe rimen t is very h a rd at
lower energies ) but no t outs ide expec t a tions - howeve r the asymp totic values
of a t ot (A) /Ao tot (N) are c lo s e r to . 87 for C, . 75 for Cu and . 60 (wi th large
error) for Ph .
These a re all unders tandable i f the total cro s s sec tion is considered
70% VDM and 30% something els e ; (as we have s e a n is t rue) and i f the something
Au
2
( in for Q • 0 . 5) in s t rong disagreement with VDM predi c t ions . If we corre c t
VDM by al lowing a certain percentage of "other pro ces ses " which do n o t show
(VDM cont ribution) mus t fall as ph o t ons b e come more vi r tual . Thus th e results
by the nearby po les at the mass o f the vector mesons . Sub s t an tial parts of the
interaction can be unde rstood as being due to intermediate virtual vec tor
mes on coupling - but , o f course , not all the coup ling can b e so des cribed
( for example only 70% of the total yp cross section at high ene rgy is so
des cribed ) •
• 4Gev
2 2
not a large background and not much else res onant has b een found to q
and by the fit of cons tants f ::> g . (This will be dis cus sed further in
p nn p
lec ture 2 4 , in reference to the dispers ion theory o f the pion form facto r . )
such as the ab sorp tion of the particle leaving the nucleus . The VDM theory
Diebold , High Energy Physics Con ference at Boulder , April 1969 ) . There is at
present no explanation known for this discrepancy - its s tudy would make a
nice prob lem . I doub t tha t the fact that the y couples only partially to VDM
is sufficien t to account for the l arge dis c repancy . Some i tem in the theory o f
Lec ture 2 2
Nucleon : The matrix e lement for the intera c t ion of an e le c t ron , s ay , and a
proton involves the diagonal e lement of the curren t operator <p j J I p > .
lJ
Us ing
relativis t i c invariance and gauge invariance this can be writ ten in the form
where u u
are 4-spinors des c rib ing the in and out proton of momen ta p , p
1 2 1 2
2
momentum t rans fer q • p - p • F and F are functions of q only . Other
2 1 1 2
linear comb inat ions o f F and F have also b een de fine d , mos t used are ( the
1 2
so-called e lectric and magne tic form factors )
G • F + 2MF
M 1 2
clearly as q
2
� O F
1
i s the charge : F
1 pro ton
(O) • 1, F
1 neut ron
(O ) • 0
and F
2
are the anomalous magne tic momen t a . Hence G ( O )
E
• 1 for p , 0 for n
G (O) • \J P is the total magne t i c moment .
M
These have been meas ured by s ca ttering elect rons from pro tons e las t i cally .
1 14
Electromagnetic Form Factors 115
(22 . 1)
ep 3
�''r(" Pl
The cross section aris ing from this is :
[ GE t � +
J
2 2
a (9 ) • ()
CINS 9 ------- +
1 + t
2 2
2t GM tan ( 0 / 2 ) (22 . 2)
( Rosenb luth)
2
t -
e 4 cos 0/2 (22 . 3)
2Eo
+
2
M
0/2 ( 1 sin 6 / 2) 2 2
-l
4E s in 0 /2
_o _____
• ;;_..,
1 + T s in 0/2
2E 2
( i
o f completely empirical approximations , namely that
2 -2
1 + Ti
The latter function is called the dipole function GD .
Most accurately known is �p /�p ' I t vaguely follows the function GD .
JJ6 Photon-Hadron Interactions
0.9 2
n / 4M
= µ
F become identical, so \Ulless GEp and G vanish here (which is very \Ullikely)
2 Mp
we must have GEp GMp at q2 • 4� . Thus GEp /(� /µ ) 2. 79 at q2 • + 4M 2
• •
P p
and •l at q2 O . •
The simple ratios among GMp ' GMn, GEp ' GE n which are nearly right would be
expected by pure vector meson dominance (assuming the mass of p and w equal , and
that � couples only weakly to the proton) . However, all the form factors would
be multiples not of GD but of the meson propagator � 2 /(� 2 -q2) (1 + Q2 /.5 8) -1 • •
This supposes that the photon couples to a point charge nucleon via the p . Thus
it has the wrong large Q2 behavior Q-2• Its small Q2 behavior is 1 - 1.7 Q2
instead of 1 2.6 Q2 • It is, of course, not expected that VDM continue to
-
But there is little doub t that the large size of the charge square radius
(i. e . the coefficient -2.6 of Q2 ) is contributed to a large extent by virtual
p ' s. It is not really clear why these ratios do maintain their values at
Q2 • 0 out to Q2 • 2 or 3 GeV2 as they do to 10 or 20%.
Lecture 23
Electromagnetic Form Factors (continued)
We can gain some insight into the meaning of form factors by looking at
Electromag11etic Form Factors 117
the non-relativistic case . For a s p in les s sys tem the form fa c t o r is j us t the
f (Q ) • J p(R) e
iQ •R d 3if •I p (r ) 4 11r2 dr { si �Qr
r} ( 3 . 1)
2
p
where r 2 is the mean square charge r adius
J p
p
rp 2 (r ) 411r 2 dr
2
• r
An e xp onen t i a l e-ar for the charge density gives the d p o le dis t ribution
i
f (Q) • ( 1 + Q2/ a2 ) -2 • The Q-4 dependence says the slope at o rigin must be
fini te not z e ro . It is , howeve r , no t a va l i d conclusion to use non-relativis tic
views to s tudy large Q2 • Nevertheless the con clusion is e s s en t i a l ly corre c t .
Q + m
2
Relativis tically the fact that as the form factors go t o ze ro
me ans that there is z ero p rob ab i l i ty to find the proton alone as s point charge .
This co uld result in field theory ei the r from ther e being no ideal field entity
(parton) with quan tum numbers o f the proton , or e ls e , if there were such the real
pro ton is always dressed , it h as zero qmpli tude of being an i de a l p ro ton .
a spinless Yukawa p otent ia l generated by the weak exchange of some s c alar meson ,
The two propaga tors each give l/Q2 for l a r ge Q2 hence l/Q4 for the form
1 1 . _a_
2
( ap +Q )
2
(bQ) b 3Q4
l
the second
1 1 a
at large Q2 •
asymptotic behavior depends not on the s ingularity when j us t two come toge ther ,
As we sh a ll see , s uch pic tures are too s imple and inadequate . We are on ly
making ve ry naive models to compare to the non-relat ivis ti c ideas . They wil l
not agree with the ine las t i c scattering.
Lecture 24
2
exchange is mos t important - and tha t this virtual 11 has as ne a r ly zero q
virtual to the nucleon is known in this region {gN 1111y5 , g2 15) so the only
11 •
unknown is the photon-pion coupling between a real and one that is nearly real,
11
is just about exactly {for no known very good reason) the isovector part of
the nucleon form factor F 1P -F 1n which also fits all lower Q2 • It will also fit
2 l 2
fairly well to the pure VDM or p meson pole {l + g:_ • 56 ) , which is . 32 at Q
- -1.2, •
2
but not at all well to the form factor GD of q which is too low { .15 at Q -1 .2) . •
We have already discussed this form factor for positive q 2 where it has been
measured in the reaction e+e - + 11+11-. We have an experimental curve which is
dominated by the p pole expression {with small understood corrections for the
w-p interference, etc. ) . It is interesting that an extrapolation so far into
negative Q2 still works reasonably well. The rate e+e -+ 11+11 - measures , of course,
only the absolute square J F 11 (q2 > J 2 not the real and imaginary part separately.
However the shape of the curve (p resonance) is so simple that our physical
understanding may make a separation into real and imaginary parts not hopeless
(e.g. we wrote previously, for the part near the resonance (-mp 2 / (q2 -'mp 2 + ir p p )) m
and had some guesses as to how r varied with q2 • If this can indeed be done we
would have information now known or soon available on F11(q2 ) over a large range
of q 2 both positive and negative. We can then check the hypothesis that every
one believes - that F 11 (q2 ) must be an analytic function of its variable q 2 with
a branch cut along the positive real q 2 - i.e. that F11 satisfies a dispersion
relation.
(24 .2)
(µ •m11)
(211) 2 where at least j F 11 (q2 > J 2 is
measured. A check on a guess for the imaginary part would be to compute the real
part from (24.2) for q2 > 0 and check J F , J 2 (alternatively a much easier equivalent
method is to fit I F J 2 by choosing physically appropriate analytic expressions
11
for the complex function F11 (q2 ) ) . Then, having established a reasonable
Im F11 {q2 ) one calculates F11 (q2 ) for negative q2 (where it is real) to compare to
experiment. We already see that it will work well - in the positive q2 region
the main feature is the p resonance and up to rather large q2 no other resonance,
1 20 Photon-Hadron Interactions
becoming available from pion-virtual pion scattering analyses . This may help
in analyzing the dispersion relation.
How should we expect Fn (q2 ) to go for large q 2 ? It must fall off because
the probability of producing just two n's where so many other states are
available must be very low. The isospin current generated by the pair of pions
must be coupled to other hadrons and we require this coupling not to radiate .
As we shall understand better later, this leads one to expect F to fall as some
n
power of q2 for large q2 , but we do not know the power. However, these con-
siderations make it not so suprising (granting the experimental behavior of
Fn (q 2 ) for positive q2) that for negative q2 even out to -1.2 the expression
is close to the pole formula (which arises from (24.2) if most of the imaginary
p
part is near mp 2 ) .
Sum rules result from the following considerations. If the unsubtracted
"'
-� I m Fn (q' 2 ) dq' 2 1
" J
• (24 . 3)
(2µ) 2 q' 2
a "sum rule".
If we put q2 � "' in (24. 2 ) , we would expect Fn (q2 ) to go as l/q2 times
- -1 1 Im F(q' 2 ) dq' 2 (if F (q2 ) went more slowly than l/q2 we would have to
n n
say Jim F(q' 2 )dq' 2 diverges , as indeed we would expect for Im F(q' 2 ) would
probably also fall slower than l/q' 2 ) . If we expect Fn (q2 ) to fall faster
than l/q2 (we know no sure argument for this) we would have another relation
In practice (24.4) is more useful than (24.2) to compare F11 (q2 ) for
negative q 2 to its experimental value since uncertainties from the contributions
for large q1 2 where data is unavailable are much smaller,
We can now see using (24.4) why from what we already know experimentally
for F 11 (q 2 ) for positive q2 the value for negative � 2 should be close to the
2
p pole value even rather far out. In (24.4) we know llDF(q ) is dominated by
the pole - there is not much I F11 (q2 ) 1 2 for larger q2 (at least no resonance to
p
4Gev 2 > and (24. 4) does not contribute integrand for still larger q 2 , for q 2
even as negative as -1. 2 . Thus F (0) - F (q 2 ) is given by the analytic extension
11 11
of the pole formula nearly. But F11 (0) is (the accident of the VDM constraints
p
f /g 1, see lecture 16) given by the p pole formula nearly, as we have seen
p ll ll p
•
there q 2 4. We would expect again that F1 Cq2 ) and F2 (q2 ) are analytic
�
gives J� ...i 2
Ai. l dA giving a continuum of singularities the lowest A€owever
Ao Q + A
being Q2 -A 2 the slowest exponential tail. This idea is valid relativistically .
•
0
•
(It has been shown that the position of the singularity is the same as the position
of the singularities in Feynman diagramsof a process as if real particles were
virtual) .
For the proton the virtual state N + n+ exists with a extending out as
n
(a gradient of) e -µ r/r. Its square, the charge density involves polynomials in
l/r times e -2 µr , the longest exponential is therefore 2 µ, and the threshold at
2 µ.
A more interesting case is one say like the E, of mass M1 which virtually
can emit a K (mass m) to be a nucleon mass M2 , with M2 M1 • We obtain the
<
singularity by studying the perturbation diagram but can see its position
Electromagnetic Form Factors 123
potential can we tolerate before we get a divergence beca\18e the fie l d rises
Thua we have a field of momentum Q ( i t is pure imaginary • ia} and are j ll8 t
able t o make the K meson from the t. Th e virtual state i s a particle of mass M2
momentum O , and one of mass m momentum + Q 2 where the initial s tate is one of
denominator
2 2 2
2 2 2 2 2
Q •-4m + (M1 - M2 - m ) /M2
-a •
if Hz < M1 - m (o therwise the square roots are not pos itive . ) The threshold
2 2 2
2 2 2
is at q +a •
That i t is below the energy 4m needed to make a pair , which
was the first guess , we call this an " anomalous " threshold .
2 2 2
For M2 > M - m the equation cannot be s olve d , the s ingularity now
1
2 2
really comes (normal threshold} when the pho ton can firs t make a pair q 4m ••
124 Photon-Hadron Interactions
which is very poor - it cuts off too fast at high -q 2 (there is no reason to
expect the model valid for large Q2 ) . For small q2 it goes as about 1 - .7 q2
instead of 1 + 2.6 q2 In the model the photon coupling is direct (processes
like VDM are not considered if they were we would add a factor (l-q2 /mp 2 ) -l or
1 -1 . 7 q 2 giving better understanding ) . The ratio GE µ /GM comes out 1 + q2 /2'1'1
2
all odds the worst disaster of their model -(it comes from the ad hoc way they
dealt with the spin of the quark unaffected by the photon) that one would think in
honesty they would include it in their paper. They didn't because, surprisingly,
they didn't think to calculate it when they wrote their paper!
Electron - Proton Scattering. Deep Inelastic Region
Lecture 25
2
It is controlled by the amplitude <'U2 Y µ u1 ) -
411e (x l
µ
2- J I P > in which only
q
the factor (x IJ IP > is unknown , and so we can measure this matrix element (or
II
rather its square) . The mos t novel feature is the ability to vary the q 2 o f the
virtual photon away from q 2 • 0 ( real photon experiments , which we already
2
discussed ) to q negative , and in fact to the far negative region . In addition ,
very energetic (high mass ) s tates x can be excited in the high energy machines .
A great deal o f information is becoming available , and there are many theoretical
2
questions especially for the new large q and high mass s tates x .
Th e simples t experiments to do are those which d o not require studying the
x s tates at all (sum over possible states x ) but j us t s tudy the ele ctron beam -
125
126 Photon-Hadron Interactions
its def lection and energy determines the momen t11111 transfer and its energy loaa
determines (in the lab system} the energy of x above the proton mass . these
experiments were done first and are moat complete - we discuss first therefore their
results and theoretical interpretation (which will occupy us for many lectures } .
Then we shall return and discuss what we can expect the final s tates x to look
like and how they do behave experimentally as far as ia known .
2 2
(25 . 2 )
L <p j Jv (-q) j x><x j J µ (q) j p> 2116 (Mx - (p+q) )
"\'
K µv •
x
(spin of p averaged , is also µv symme t ric part of expression for any spin) .
(Note , the mas a 2 of the final states which we are generating is expressed in terms
2
of our variables q , v as
2
Mx2 • (p+q) 2 • ; + 2 Mv + q 2 q is negative .
By varying v for given q2 we can vary Mx2 and look , for example , for final state
resonances , )
Electron-Proton Scattering. Deep Inelastic Region 127
D •n. n •n
q qv
....).. -aII - 4
2 2 2 µ
M K
\I V
w 1 (q , v ) � ( 6 ,.:;- -
11 II
- • 4 w2 (q , v) (p (p . , )
-q µ v 2. )
...,. � · 2-
q
2 q q
(25 . 3)
Several properties o f W1 , W2 :
• 0
2
The apparent , but unwanted poles at q come from the way we have
wri tten our expression and of course cannot be real . To remove the highest pole
2 2 4 2
we mus t have ( p • q ) w2 • -M q w1 as q 2 + 0 o r w2 • -q /v w 1 order q as q + 0 , +
2 2 2
w2 Cq , v ) + -q /v w 1 CO , v) .
2 2 2
The direct total absorp tion cross section for an imaginary pho ton of mass q 2
to hit a p ro ton would be
2
a • 4 11e "\"
� l <x l e J j p> j 2 2 116 (Mx 2 - (p+q) 2 )
2k2M µ µ
• 411e 2
e JI e V K \I V (25 . 4 )
2k2M
e t • q /v -q -
z /:f ' ez •
(25 . 6 )
wl
•
f1
+ v2 2) (�)
a +a (25 . 7 )
-q
w2
t s
\
It is convenient to de fine
128 Photon-Hadron Interactions
Combining the electron spur to the expression for Kµv we ob tain the total
cross section in the lab .
-1L - _,..._
2
__
.. a_..,.... 2
8cos 8/2) w2 + 2 (sin2 8/2 ) w� (25 . 8)
dndE ' 2
4E sin4 8/2
(25 .9 )
£ • � + 2 tan2 0/2 (1 + � J
-l
-q
and often data is given for a t + co s without resolving them , £ is often small .
2
o (v + q /2M)
Lecture 26
2
to o (q + 2Mv) - and depending on <\t• GE as we have previously studied . This
2 2
contribution falls as q2 rises as 1 / (1 - q / . 71) •
Next , for fixed q2 , as M 2 varies , we see a "firs t " resonance at Mx 12 36 , •
x
Electron-Proton Scattering. Deep Inelastic Region 129
a "second" one near Mx • 1520 and finally a " third" around 1700 and some indication
of a fourth at 1900 . Theo retically we should see more resonances , e . g . 1535 but
undoubtedly the "second" peak is an unresolved mixture of these two while the o e n
near 1700 sometimes called the 1688 resonance is probab ly that with four others
expected near that energy , The resonance at 140 7 (the Roper resonance) has not been
seen in these experiments .
These resonances can be fit ted as Bre i t -Wigner p eaks on a background which
2
gradually dominates into a smooth curve as Mx increases .
How do the resonance strengths vary with q 2 ? At very low q 2 the beh avior
depends theoretically on the angul ar momentum of the state and star ts as an
appropriate power of Q (Q 2 -q 2 ) . For higher q2 , however , the s t rength of the
•
resonances all fall more or less as does the elastic peak . In fact if one plots
the ratio (da /dn) with -q
2 one finds curves which ri s e rap i d ly
res / (do /dQ) elas tic
from threshold ( pho toelectric) values to saturation (in the vicinity of one or
2
just below) for q 2 > ab out 1 GeV •
t rue , This feature , that as -q2 + m , v + � such that -q 2 /2Mv • x is kept fixed ,
vW2 Cq 2 , v ) approaches a func tion of x, F (x) is known as Bj orken scaling . We shall
have a gre a t deal t o s ay late r of i ts theore tical s i gn i fi cance ,
20 / w-2
I Resonance Region
q2 II S caling Region
III W1 , w2 S e p ara t e d
10 �3 4
---
w c
2 2 In the
In the re gion 4 , s caling is observed above Mx • (2 . 6 ) •
2
region 12 w > 4 for Mx > 2 -q > l; vw is a constant within errors and s cales
>
2
in any variable .
130 Photon-Hadron Interactions
l
f0 vW2p . 18 l for the proton
dx • t
l
It is consistent with a straight line (l - x) from x • 0 to . 75 . lbere are
indications that it curves away above the line for larger x and could reach a
value as high as . 4 ( ? ) for x + l, but higher values are probably ruled out.
In the region where w1 and w2 can be separated, values of R can be found -
they vary from 0 to O . S with large errors so no definite trend can be seen.
If R is assumed constant it is . 18 . 10 , the data is also compatible with
t
called the parton model briefly; then we shall discuss general properties of w2 ,
w1 and see that they are connected to the commutator of two currents . Finally
we come back to the parton model in more detail, discussing suggestions that
the partons be identified with quarks , etc . - and then discuss the relation of
this to a more abstract representation called Gell-Mann's light cone algebra.
lbere then follows miscellaneous discussion, ending with attempts to understand
Electron-Proton Scattering. Deep Inelastic Region 131
Parton Model
abs t ract cons iderations are therefore more s a t i s fying , b ut fi rs t we dis cus s
the more e lement ary viewpoin t , I n dis cus s ing these mat ters it is bes t to
keep in mind all the princi p les of relativity , quantum mechanics , uni tarity ,
satis fies all these principles s imul taneously . There is no known s imp le model
wh ich does this except field theo ry (and that may not do it - all examp les
In field theory the wave fun ction for a s tate , such as a pro ton , could be
given by giving the amplitudes to find various kinds of b are field parti cles
moving wi th various momen ta . These b are field parti cles we call "par tons " .
Some times some phenomena can be unders tood dire c t ly in terms of this wave
function but us ually ma trix e lemen ts to wave functions of o ther s t ates mus t be
analyzed . The wave funct ion , however , is not eas ily trans formed from one
re lativi s t i c sys tem to ano ther ( the Hamiltoni an mus t be known ) be cause it rep re
sents a slice at a given t ime . Ce rtain properties are there fore more eas i ly
1 32
Parton Model 1 33
seen from the wave funct ion in one sys tem than in anothe r . The deep ine las t i c
s c a t te ring behavio r i s bes t un de r s t ood from the wave fun c t ion for the p ro ton
In what variables will the wave function have a limiting form? From a study
o f the character o f typical field hamiltonian in a few examples , and from a study
of very high energy inclusive hadron collisions we conclude that i f the momentum
in abs o lute uni t s k then the ampli tudes depend only on �. k as P rises - (except
for � so small that �P i s a few GeV , � of order 1 GeV/P are called ''wee" -
the dis t ribution of"wee" partons is probably bes t described in absolute p z
k variables . As we shal l see the "wee" region is hard to analyze - b u t the
main features o f the deep ine las t i c scattering does no t involve them) .
The fac t that k� i s fini te is !!.2!_, at leas t in any obvious way , a direct
consequence o f field theory - ( in fact perturbation theory does not give this
result , and therefore must no t be reliable here) it is simply guessed at from the
ubiqui tous result o f high energy colliss ions that the transverse momenta
available to the products averages about p� (.4
-2 • GeV)
2 •
But granting this , the s caling is expected �p from field theo ry (although
today we coul d base thi s also directly on experiment , it is the s caling l aw
for the longitudinal momentum of the products in very high energy collisions ) ,
Some s ugge s t i on o f how i t works is this . The ampli tude that a s tate of energy E
is seen to be made up o f two parts of energy En • Er + E2 is dominated ( in
per turb a t ion th e o ry) by a f acto r
(26 .1)
z p1 + p .
But the component of momentum of the parts is that of the whole P •
2
Set p l • f; l P2 • P, � 2 p' � 1 + � 2 l, E - --.//Pl 2+M2 - P �2 p
� + = M
I f 1 i s th e mass of
Lecture 27
Parton Model ( continued)
We envisage the proton of momentum P as being made of partons of momenta
E; i P all sharing ill various proportions ti the momentum o f the proton , all E; lie
between 0 and 1 (else some other would have to be negative since I ti • l) .
We shall therefore think of the incoming proton as a box o f partons sharing
the 1110mentum and practically free , Another way to look at this is to take a
dynamic view of the parts in the res t sys tem and assume finite energy of inter -
action a1110n g parts so as time goes on they change their momenta , are created or
annihilate d , etc . , in finite times . But moving at large 1110mentum P these times
are dilated by the relativistic trans formation so as P rises things change more
and mo re s lowly , until ultimately they appear not to be interacting at all .
When the proton is hit by the photon the interac tion operator Jµ couples to one
parton or another and knocks it to a new state of momentum p 2 • p 1 + q (p • nearly)
l
E;P
2
If the parton had mass m we would expect a rate proportional to
(2 7 . 1)
-
Pi
Parton Model 135
the four-momentum of the proton (strictly it is valid for the z component only,
but we have seen that implies it for the t component E 1 • f;E; and the transverse
components being only finite are relatively small) .
Therefore if
f(x)dx • number of partons with momentum between x and x+dx each weighed
by the charge squared (in units of electron charge) we have
K)JV • !.M ( 4 P )J Pv W2 - 4 6 )JV M2w1 )
· J f(C) (4 C2 PI' Pv _
(27 . 4 )
c
(the first l / C comes from the normalizing factors 2E/2E 1 ) . Inside the 6 function
we have (CP+q) 2 -DI2 - fl • 2E; (P • q) + q2 + f; 2 !r2 - m2 - but as -q 2 .,. .. and P ·q
fl ;
2Mv
for large v
4 P P W - 4 6
2
)JV 2 )JV M W1 =
Thus
1 36 Photon-Hadron Interactions
• f ( x)
2
2MW ( q v )
1 1 ( 2 7 .5)
If the fraction of the partons which a spin 1 / 2 (weighed by their charge squared)
(l+ \J
at momentum x is y (x) we ge t w unchange d , 2MW • y (x) f (x) . The ratio
v 2 1
w -w
- -"----
__-q
cr 2 1
R m
s a
(2 7 . 6 )
In the region s t udied R is o f o rder .18 t 10 so y is less than .2, not many
� xf (x)
squared is 1 in our units and thus we could s ay only 18% of the moment um is
momentum carried by the quarks could be higher. We shall discuss such a model in
more de tail , b ut it turns out even then it is neces sary to assume something e l s e
Lecture 28
making further assumptions about the wee region we can unders tand other aspects
o f w1 and w2 • It mus t be realized we are now elaborating on our original parton
ideas developing them further to understand more features of the f (x) curve . The
data for vW2 seems to approach a cons tant ( . 32 ) as x + 0 which means that f (x)
would go as . 32/x showing s mean nuui>er of partons ris ing as x falls into the
wee region (such that the number of wee partons is finite and independent of P ,
and the mean total number of partons in a state of momentum P rises logarithmically
wi th P ) . lbis is not entirely unexpected , it is the same as the dis tribution of
p roducts in hadronic collis ions . The way this appears to happen can be gathered
by studying the field theory equations at high energy , and also the perturbation
theory of bremsstrahlung . In the latter case neutral particles are generated
J
in a dk/E • "' dPz / pz 2+� -tta
2
dx/x distribution . These neutrals can generate
+
pairs so the small x region contains large and nearly equal numbers of particles
and antiparticles . The field equation approach sugges ts the same thing; and
further that the low region is generated in higher order perturbation from the
higher momenta by a series of cascades x + x' + x" going down in x . In either
case we conclude the character of the pai J:B will be as a whole neut ral and
therefore the same for proton and neutron . lbese expectations have been made
by other means also , for example , the dx/x leads to a constant cross section
for virtual photons (of fixed but large , negative mass squared) , That these
cross sections should be a cons tant is expected from cons iderations that these
pho tons have an amplitude to be virtual hadrons ( like the p) and hadrons give
cona tant cross sections (the magic word "Pomeron" is used to "exp lain" this ) ,
the same for proton and neutron . Thus we expec t vW2 n to equal vW2 as x as
P
+ 0
indeed they turn out to do ; both f (x) for p and n app �oach the same . 3 2/x for
small x. We also know experimentally that at q 2 • 0 a • a to 3% at 16 GeV .
YP yn
lbe fact that f (x) seems to go as . 32/x as x + 0 implies that the total
cross section for virtual photons of energy v on protons is s cons tant as energy
goes to m for fixed and large negative q 2 , j us t as the real photon cross section
does . We can expect , as long as we are on the high side of the wee region , that
x • b/P with b large enough that , as our meas urements tell us , f (x) ""'" . 3 2 / x .
Hence the transverse virtual photon cross section is
1
w f (x ) • !!...!._ f (x)
2 2 2 2 2
0t !!!_ �
•
k 1
•
/22· 2M v 2M (28 . 1)
,._j v · -q
1 38 Photon-Hadron Interactions
in the s caling limi t . And for x on the high s ide of the wee region (-q 2 large
enough , )
we have just seen is experimentally found, above . But we can cer tainly guess
that , for each -q 2 the cross section is constant but dependent on -q 2 for large v ,
a
2 2 2 2 2
t 4 w e C (-q ) . We do not at presen t know C (-q ) excep t fo r q
• 0 and large •
2
-q • We need a good theory for this function .
Although the number of partons is infinite , the momentum contained in
them is fini te , of course , because the total momentum of all partons neutral
and charged is 1 (units of P ) . We have for the momentum carried by all charged
�
partons weighed by e 2 xf (x) dx • . 18 t . 0 1 .
0
Formula for R
We may get an idea of how R behaves as we approach the scaling limi t by
calculating a s directly in this region . We work in the system in which q is
pure space-like
p (E , 00) p - l..:.i. • (E , O , O , O )
µ II
P,
q 2 qll
q q
q (0 , -2Px , 0 , 0 ) 6 ll - - ll V .. diag . ( 1 , o,
-
q2
-1 , -1)
V
2
Use Equation (2 7 . 4) with all q put back into I M l
11
-
11
M (4P11 P V w2 - 46 llV ; W1 ) •
Parton Model 139
whatever that means , but at least i t is the square of some fini te energy . o s
and R are given by
4 (k�+m2 )
( 4E 2w2 -4 m2w1 ) • 2Mf (x) 2
-q at x
R •
2
-q
R • (28 . 3)
at x
The expected falling trend with l/ (-q2 ) is not evident in the data , but
the errors are large . If we estimate k2 � . 25 (GeV) 2 independent of x as in
2
hadronic collis ions , and m of the same order (else we could not explain why
kJ. is so small) , we find R � (2:!:4ll) I (-q 2 ) ; for -q 2 • 7 (near center of data)
2
Lecture 29
'Ole region near x • 1
We can ge t some qualitative ideas of the region near x • 1 by firs t s t udying the
extreme case x • 1 , the elastic form factors , at large -q 2 We take the
coordinate sys tem wi th q • (0 , -Q , 0 , 0 ) a pure space-like vector . P the
µ
momentum o f the proton • (E , P , O, 0) mus t have P • Q/2 , after the collis ion
the proton has moment um P � • P µ" 'lµ • (E, -P , 0 , 0) 1 with z component in the
opposite direction . For large Q, hence large P, we can des cribe the ini tial
s tate o f the proton as an ampli tude for various con figurations (in x, kl ) . Le t
us suppose for example , the configurat ion contains two non-wee partons and two
wee , and draw a pic ture :
x2
!
2
�
Be fore
Xl
After x2
! Figure 29 . l
Now we ask that the final state ( the "after" picture) b e only a proton -
that is, a factor < Proton at -P I "after s t ate ">, the (comple x conj ugate of)
amplitude tha t the after s tate appe�rs in the pro ton wave function . But a le ft
moving proton can never ( for large P) look like the after pic ture because it
contains a backward moving parton (x2 ) , (and any estimate of i ts mass square is
2
order P 2 , requi ring order P 2 b inding energy contributions to ge t the mass down
2
to Mp ) .
'nlus insofar as the proton looks like "before " it contributes very little
where one par ton has nearly all the momentum , x • 1 and all the res t are· wee ,
say below l/P .
Before
li---1---· �
1
After Figure 29 . 2
Parton Model 141
Now the s ingle fas t one is reversed by the incoming pho ton , the energy
is not changed much , and the final configuration is pos sible fo r a pro ton
because wee partons ( those wi th absolute p finite say below order 1 GeV) can
z
go backwards as well as fo rwards . The refore the s cattering ampli tude ( form
factor) is proportional to the square o f the ampli tude that a proton looks
like "be fore " , hence proportional to the probabili ty that the pro ton has all
partons but one in the wee region .
We now see quali tatively why this should be some inverse power of Q in
the following manner . If the b reml5trshlung analysis of the "sea" (dx/x)
region has any qualitative reality we can judge the probability that no parti cles
have their x above x of order l/Q to be e -cn whe re ii is the mean number expec ted
0
above this ir,, and c is some cons tant to allow for correlations (unlike first order
bre-trahlung which is purely Poisson , perhaps each "photon" could make two of
some o ther object , etc . , so that the mean number of particles would be higher but
proportional to some vaguely defined "mean number o f s t atis tically independent
event s ") .
Now we have seen , for small x at leas t , n goes as xa/x where a is
J
l
another cons tan t - hence ii, the mean number above x0 is a x �
• a ln x -
1
o
a ln x where x1 is some unknown upper limit where the dx/x ideas b reak down .
0
Thus for small enough x , e -en varies as exp ( ca ( ln x - ln x 1 ) ) as a cons tant
0 0
times a power of x • That is , the probabili ty no "sea" particles are outside
0
the wee region falls as a powe r of Q .
There may b e some special partons to make up total quantum numbers which
are not parts of the sea which mus t also be in the wee region . But s caling
studies of the field equations as well as o ther theore tical arguments (related
to Regge-type behavior) sugges t very s trongly that they would show behavior o f
the form xa dx/ x where a ! O, and thus thei r probability to be found in the
region below x would also fal l as a power of x0 •
0
We conclude that the probability the proton looks like our parton near
I
x 1 with all the others be low some low limit x falls as a power of x , s ay
•
0 0
as xY .
0
In particular then for x wee of order l/Q , we find the p robability
0
varying as Q4 •
Thus the form factor GM of a proton (and GE als o , see below) should fal l
as a power of Q , Q-Y . Experimentally y is probably a lit tle larger than 4 .
142 Photon-Hadron Interactions
The fact that the form factor falls toward zero as Q + m means that a
proton (as P + �) has a zero amplitude to appear as a single parton . This
may be because no sin g le parton has the precise quantum numbers of the p roton ,
but even i f i t did , a more cogent reason is that the probability of finding a
parton without o ther fields generated by interaction from it (represented by the
sea dx/x) is zero , (j us t as is expected from the theory of b re1111Str ahlung) .
Our analysis of the ratio R • 0 8 /o t des cribed in the previous lecture can
be as well applied to the elastic piece x • 1 . We therefore find
2
as 4 (k� + m :!: ti)
0t I elastic
•
-q I x - 1
+ m2 :!:
2
where k.L. ti is the value for the s ingle parton with nearly all the momentum .
But this ratio exp ressed in terms of G ' G is simply
E M
Now we can see how f (x) should vary for x near 1 , say x • 1 -y with y small .
Let us find the total probab ility that x exceeds 1 i . e . 1 f (x) dx . We mus t -y {�
again find one parton with nearly all the momentum, the total momentum of all
the o thers cannot exceed y , That i s not the same as the probability that none
can ex ceed x0 • y which we worked out as y -Y , but a little cons ideration of the
problem soon shows that it does vary wi th the same powe r , but with a di fferent
cons tant in fron t . (This i s mo s t easily seen b y a kind o f dimensional analysi s ;
o r alternative ly one can work out the case o f a dx/ x Poisson dis tribution in
detail to check i t . ) Hence
J1-y f (x) dx .. yy
1
Parton Model 143
for small y, or
f(x) ( 1 - x) y -l
�
for near l . That is , we expect the power law of the elastic scattering to be
x
question about the other way and ask what the fact that the resonances fall
off as the elastic peak with -q2 tells us about the wave function of these
resonances .
If we excite say the 6+ our picture again cannot look like figure 29 .1
for the "after" picture could not be a 6+ moving rapidly to the left because
it contains some partons going in the wrong direction. We must instead have
as in figure 29 . 2, but this time instead of asking whether the last picture
looks like a proton1 we ask: with what amplitude does it look like a 6+? We
again expect a to have a sea region, etc. , and to have a difficulty measured
6+
by a factor Q-y'/ 2 (in amplitude, for a proton Q -y / 2 )to appear with a single
non-wee parton. We conclude from experiment the y ' probably equals and
y -
can probably invent a posteriori argument to justify this . (For example , the
sea may be very similar in the two cases because they are generated by the same
parton - that is , the chance this parton is bare of non-wee particles is what
the factor measures - and it is the same parton in either case.) The wee partons
left over from the original right moving protons have only a certain amplitude
to be in the correct proportions (of numbers non-Q dependent momenta, etc.) to be
appropriate to a left-moving proton - and a different amplitude to be appropriate
to a 6+ . It is the ratio of these non-Q dependent amplitudes (squared) which
determines the ultimate ratio of the probability of producing 6+ to proton at
large Q2 • (It is always possible in principle of course that for some particular
144 Photon-Hadron Interactions
resonance this amplitude is zero and such a resonance would fall fas ter with Q
than the elastic peak . )
These relations o f power laws for elas tic peak , resonances , and asymptotic
f (x) are so interes ting that it is important to no tice that they can be expressed
in a general principle ( I am st rongly indebted to J . D . Bj orken for discussions
on this idea) quite apart from any parton theory interpretation . S tated in
this way they are seen to be rather profound and represent a fundamental new
property of high energy collisions ( for corresponding relations are expected
relating exclus ive and inclusive hadron reactions at high energy) .
2 2 2
Suppose we plot vW2 for a given large q against Mx - M • 2Mv (l-x)
2 2 2
'"" -q (1 -x) (x is near 1 and 2Mv ""' -q ) . We note the scale of Mx - ; and
2
1-x are j us t proportional . We find a set of resonances at app ropriate Mx
Mx 2 -M·�������--t-�
2 2 2 2
and the tail of the "background"curve ( 1-x) y-l dx or (-q ) -y (Mx -M ) y -l d (Mx ) .
2 • 2
Now when we increase q al l the resonances fall (but stay in the same place Mx ) ,
2
as (q ) -y as we have seen , but the si ze of the "background" curve (at a given Mx )
2
2
also falls as (q ) Y . That is the "background" (by background I mean the high x
tail of the scaling region function f (x) ) bears a fixed ratio to the resonances .
p
It is not possible by going to higher q 2 (or in• hadronic collisions to higher s )
to separate even more clearly the resonances from the background , nor do the
resonances fall into the background and get lost . Put another way , the back-
ground can be consis tently viewed as due to overlapping resonances or tails of
resonances itself and all resonances ultimately fall off the same way (as a
2
power of q as q2 increases ) . Bohr would like this , he would say there are two
complementary ways o f viewing the background , as continuum single particle
states leading to scaling expectations , or as large numbers of resonances - and
nature conspires to make i� impossible to decide in the intermediate region
2
(large Mx - ; ) which is corre c t .
Parton Model 145
Lecture 30
Argumen t that y ' m y
We will the assumption that for any state (e . g . p or A)
make as momen tum
distribution of Q .
in dependen t mo re explicitly for our case ,
Rathe r , ass ume we
amon g wee momen tum ( e . g . below 2 GeV ) and nothing in between falls somehow with
<one p ar on at x ""' 1 ,
t es we
k'sJ Q>• Pp (Q) fp , i (k ' s )
at momenta p ro ton at
(the index i indi cates parton type) an d s imi larly for a resonance . Hence the
a
sc a ttering of photon, wh i ch turns arolmd fast parton from Q Q ' (e . g .
the to
l., ,.,
Before p Q
. �
After A
•
Q'
t . . ->J
We explicitly suppose the las t factor is not (For certain resonances , zero .
wi th Q, but study the ones which are not spe a in this way , ) The answer
we ci l
deduce that both Pp (Q) and FA {Q) be of the form 1/ (2Q)y/Z (a con s t ant
mus t
a (p , A ) • \
L .
ff' i (k} f Ai (-k) d3k
p,
i 2
depends on the s t ates p , A, but not on q •
1 46 Photon-Hadron Interactions
If two states have di fferent powers of Q in their F (Q) • Q-r /2 then they
.J"
3
mus t be orthogonal in the sense 7r, Ck) fb (-k) d k 0 - usual ly because of •
2
and imagine we meas ure the missing mass of the resonance by � • • P X2
2 2 2 2
( PA + PB • P c > """ s ( 1-x) + MA +Ha -M Thus plot resonance data as M"" •
2
C •
2 2 2 2
+MC -MA -� � s (l -x) . As s varies resonances s tay at fixed M"" , of course , but
2
�
tleir size goes down very likely as a power of s, say as s ar cil- > dtl- ( a is negative) ;
as expected for exclusive reactions , where a is the leas t negative possib ility for
an exchange of something that acco1D1ts for the quantum number changes from A to C .
But since we are no t really looking at the products � this can be thought o f
a a an inclusive reaction where , i t has been argued , ( R . P . Feynman , Phya . Rev .
Lett . !! 14 15 (1969 )) the probability should be a flDlction of x only . Hence
2
this s caling region near x • 1 will go into the resonances for large M only if
for x near 1. (There are situations , for example , where C and A are both pro tons
that it is technically hard to get to high enough x to avoid contamination of p ' s
disintegrated from other hadron resonances C', etc . , but these matters canno t
concern us in detail here ; the result baa not been readily checked , there fore . )
Lecture 3 1
We a ummar ize what we or surmise about vW and w1 in the high v region .
2
know
2 but 2
Region Large v, large -q - q / 2Mv • x finite . 2MW ia a function of
I)
1
2 2
x only , called f (x) . w2 • (-q /v ) W1 •
Parton Model 147
2
Region II) Large v , fixed -q as v + • ,
2 • 2
2
Mv g ( -q Z
function of q )
-q
2 2 2
Region I I I )Large v , large -q 2 but finite M -M • 2 Mv - ( -q ) ,
x
2 2 -y 2 z
2MW • (2Mv) -y+l function of M only • (-q ) h (Mx -M ) ZMv
1 x
2
(The las t factor 2Mv is the normalization from dMx to dx} .
In order that the three regions fit together we have ,
(I ; II) for small x , f (x) goes as a/x (a • . 3 2)
2
for large -q , g ( -q 2 ) goes as a .
( I I ; III) for x near f (x) goes as A(l-x) y-l
1, y 4 or 5 ""
2 � 2- 2 2 �
for large Mx - , h ( Mx M ) goes as A ( Mx - ) y
-l
2 2
but is not equal to it . ) In region III we expect o s /o t to fall with increasing
2 2
v as l/v . In the special region of small q : w • -q w /v 2 + order q /v ;
4
2 1
2
as� q •
2 -y
General remark about the power law (q ) , namely it may have a logarithmically
-
falling coefficient (q2 ) y / (atnq 2+b ). In the discussion of the previous lecture
we saw powers where y could depend on the quantum numbers of the wee group , like
total isospin , angular momentum etc , These are discrete so the y i ' s are discrete
and there is a lowes t one . However , there is transverse momentum which is not
discre te ; ycould depend on the t rans ve rse momentum of the fas t p a rt on say as
• 2
Y Yo + Y 1k� so the result is
� -
(q 2 , ( y o+y 1k� ) dk; .
Y
( qz , - o
( 31 . 1)
Y1 ln q
2
Partons as Quarks
Naturally we should like to go further and find out mo re about what quantum
148 Photon-Hadron Interactions
numb e rs the p ar tons carry . As it turns out from present experiments only a few
things can b e learned , but future exper imen ts can add informa t i on wh i ch greatly
res t ri c ts the pos s ib il i ties . We dis cuss this b y choo s in g an example , tha t
ch arged par tons carry t h e quan t um numbers o f quarks . ( In s uch a mode l Ge l l -Mann ' s
equal time curren t commutat ion relat ions are automati cally s a t i s fied . ) We are
not making the three q uark mo de l , the number o f quarks may be , indeed mus t be ,
infini te .
One is immedia tely s t ruck with the que s t ion as to whe the r our explana tion
of s caling is pos s ib le at all i f the partons do not have in te ger char ges . The
idea was that the ou tgoing par ton (now quark) could es cape wi thout further la rge
hadrons .
-------.. a
p
But it is a q uark ( kn owing no low energy hadrons are of charge 2 / 3 say) it canno t
turn int o hadrons un les s it p i cks up the ext ra 1 / 3 charge by dragging along
ano the r parton quark . I f th is e x t ra par ton is from the ongoing ones o f momentum
o f order P the momen tum changes ( d i re c tion B to A) would b e o f o rder P and the
as sump t ion that b inding forces are s ma l l , apparen t ly required in our derivation ,
would f ai l . On the o ther h and , i t i s pos sible perhaps that the extra quarks
are found among the wee par tons , whi ch a re natural to any had ron wave fwic tion and
s ince hadrons going in direc tion A o r B share the s ame wee region they could
exch ange quarks there , making up the neces sary integers . Thi s is not clearly
satis factory but i t is an exciting adventure to t ry the idea that ch arged par tons
are s imply quarks , see consequences , devise experimen ts t o tes t them; and if they
are found to succeed , to re turn with more zest to the ques t ion of how Nature
b. Hadrons do no t
c. S caling works .
Parton Model 149
the very leas t we are giving an example of how experiment could lead to fur the r
identifi cation of the character of partons .
To des cribe the parton dis tribution in a pro ton we would have s ix functions .
1 • 3
2
! - 1
[ u (x) - u ( x) ] dx - 3
1
j Cd (x) - -
1
d (x) ] dx - 3 1 1
[ s (x) - s(x) ] dx .
0 0
The z isospin is +1/ 2 so
1
j
f • } [ u (x) - -u (x) ] dx - 21 [ d (x) - d{x) ] dx .J
0 0
The s t rangeness is zero so
[ 1
[ s (x) - s(x) ] dx - o .
These equations have the solution
]cu<x>
0
- u(x) ] dx . 2
t [ d (x)
0
- d(x) ] dx - 1
J [ s (x)
1
- s(x) ] dx • 0 ( 31 . 2 )
0
That is , the ne t number of each kind o f quark is j us t the number of that kind in
The observed function f p (x) from 2 MW1
the s imple non-re lativis tic 3-quark model .
(or vW /x) is the number of each kind weighed by the square of the charge :
2
hence
- 1 1
f p (x) = 9
4
(u (x) + u (x) ) + 9 (d (x)+d-(x) ) + 9 (s (x)+s-(x) ) ( 31 . 3)
Measurements on the ep s cat tering o f course give us only this sum and we cannot
ext ract the separate terms from this alone . On the o ther hand we have data on
neutron s cat tering too . The neutron is given by the same formula excep t u (x)
would now be the number of up-quarks in the �· However , by isospin
reflection the neutron funct ions are got from those of the proton by rep lacing
u by d, d by u. Thus , if u(x) continues to be as defined , the number of
150 Photon-Hadron Interactions
"ups " in the proton is also the number of downs in the neutron , we have
• 9
1 (u (x) + -u(x) 4 (d (x)+d- -
fn ( x) ) +9 1
(x) ) + 9 (s (x)+s (x) ) (31 . 4)
P
We have data on both these functions . The ratio fn / f behaves like
1
fn
r
p
-�
It has been s ugges ted that possibly fn / fp 1-x falling to zero as x O . We +
see immediately that thi s is impossible if charged partons are quarks ; s ince
u {x ) , 'il{x) e tc . , are all positive fn / f cannot fall below . 25 . The original data
P
only went up to about x -. . 8 and had fallen to perhaps . 3 so . 25 -was a possible
limi t as x + 1 but we have s ince learned that there was a mistake in the computer
program and i t is closer to .4 for x near 1 .
However , j us t for an exercise , what could we deduce i f fn / f was shown to be
P
j us t 1/4 as x l? That means d (x) falls to zero fas ter than u (x) as x + 1 . Thi s
+
2
would result i f w e s upposed that the y (in (q ) -y or ( 1-x) y-l ) f o r wee partons o f
total isospin zero is smaller than the y f o r isospin one (only one non-wee
parton) . Then the leading term (lowest y) would require tha t the fast parton
have the same isospin as the pro ton , up-quark only (since the wees are isospin
zero) . Howeve r , in that case the probability of exciting the 6 resonance (of
isospin 3/2 ) would fall fas ter than the elas tic peak for one can only make a
6 from wees with isospin one (as the leading quark can only con tribute 1/2) .
Thus we would expec t the 6 resonance to fall off with q 2 fas ter than others
( data is insufficient to indicate whether this is true or false in fact ) .
We can argue (if partons are quarks ) that if fn / f p falls below 1 i t "almost
surely" falls to 1/ 4 as follows : We first assume that almos t sure ly ( ? ) the
values for y for different quantum numbers among the wee are not equal and that
one therefore dominates . First among the s impler ones o f zero strangeness quarks
we have y0 for I • 0 and y 1 for I • 1 . Hence depending on whi ch i s smaller we
have even tually (as x + 1) a pure I • 0 or pure I • 1 among the wees . For I • 0
the non-wee quark mus t be (1/2 , 1/2) or u, this gives immediately fn / fp •
Parton Model 151
2 2 •
( charge of d) / (charge of u) • 1/ 4 . If I 1 the non-wee quark c an b e u or d
with weight given by the appropriate Clebsch-Gordan coefficient p • lfT3 (+l) d -
lf13 (O ) u , n • lf13 (O) d - lfl3 (-l ) u there fore
fn ( 2 / 3} (4 / 9 } + ( 1 / 3} ( 1 / 9 ) 3 if I a 1.
fp - (2/ 3) ( 1/ 9) + ( 1/3) (4/9 ) 2
Hence on the assump tion o f unequal y ' s e i ther 1/4 or 3/ 2 is the limi t ( there fore
The reader can show that the ratio o f G for proton and neutron comes out
M
2 2
-2 for isospin z ero wees , for large -q ( i t s tarts at µ p /µ n -1 . 4 at q O} . = •
The case that the total angular momentum among the wees is ze ro should be tried
2
firs t . Show also that GM for the proton is posi tive for large -q , so no sign
2
change is necessary as q rises , which is sa tis factory because experimentally
GM for the pro ton never seems to fall to zero . Isospin one for the wees gives
GMp / GM + 0 and GM is positive at large -q
n n 2
•
Lecture 32
Momentum Carried by the Quarks
Another interes ting s tudy is the total momentum carried by each kind of
J x fp (x) dx J x fn ( x) dx
be the total momentum carried by up (plus anti-up} quarks , etc . Then we know
• . 18 , • . 12 so
1 52 Photon-Hadron Interactions
. 18
. 12 (32 . 1)
U + D + S 1
Solving these equations we have U • . 2 1 , D • . OJ , S • . 76 . This result is clearly
unreasonable - that 3/4 of the momentum of the non-s trange obj e c t proton is to
be found in the s trange quarks . This indicates most s t rongly that all partons
cannot be quarks - there mus t be some neutral ones carrying momentum , say N . This
momentum can be thought of as in the field , i f any , by which the quarks in teract ;
(possib ly for example a neutral pseudovector) which if represented in fie ld theory
by inte rmediate field partons are usually called gluons .
In the region near x �
0 the functions fp and fn diverge as . 32/ x , corres
ponding to infinite numbers of quarks and antiquarks . (The net number of quarks
is finite , see eq . 31 . 2 and discussion . ) Further fp an d fn must have the same
behavior so if u(x) • a/x for small x then -;;(x) • u(x) • d (x) d{x) . We do
•
not think we could p rove s (x) and its equal s(x) mus t equal that value also
because S U ( J) is not perfect , but it may not be far away ( if indeed it is not equal) .
An interes ting quantity is the measure of how u and d approach each othe r
�
J
as x O ; although each is inf ini te , the difference is finite and integrab le .
Experiment gives about . 17 for the integral
1
[ f (x) - fn (x) ) dx but the erro r
0 P
•
Some people (ss for exmaple Pas chos , Weisskopf and Kuti) have tried to guess
the functions u, d, e tc . , by making s impli fying assump t ions whi ch are chosen for
their s implicity rather than their phys ical ne cessity . Such predictions may or may
not s ucceed , i f they fai l all we say is nature is not so s imple . Unlike the ideas
of the parton model we have used so far , the additional assump tions made in these
modelP have in general no physical backing . In case any o f you are interes ted
we give an exmaple o f s uch a theory here : I t is that the wees or dx/x region
Parton Model 153
is made en tirely of uncorrelated pairs , and tha t on top of thes e there are three
"valence quarks " of the kind of the three quark model and the wave function o f
these is a simple product of independen t pieces . (We do not expect such
simplicity . ) That is , i f a (x) is the typical sea dis t ribution and v(x ) that of
a valence quark we would write
u • a + 2v u • a
d • a + v d • a
s a s • a
fo r the p ro ton . j 1
v (x) dx • l ( to get 31 , 2 ) . Near x + O i t is a (x) that goes
to in fini ty (as . 9 4/x) but x v ( x) + 0 ss x + O. This predicts for the total
momen ta carried by quarks S • 2D -u so the data ( 32 , l) gives U • . 36 , D • .18
•
J x fn dx to exceed t J x fp dx .
and S 0 (surprise) hence we are i n s ome kind of trouble (a cannot b e zero)
unless the experimental erro rs allow
l
At any ra te some momentum (46%) must be in gluons . ( fp (x) -f (x) ) dx
nIt predicts
which is hard but not imposs ib le to reconcile with the experimental . 1 7 . I t also
f • 3l
predi cts vw2n /vw2p + 2 / 3 (hence probab ly wrong) by suppos ing that as x + l , a falls
fas ter than v .
Lec ture 33
Future tests of charged partons • quarks
So far no direct data is accurate enough to test or eliminate the idea that
charged partons are quarks . However further experiments can . We give two examples ,
neutrino s cat tering and polarized electron and pro ton s cattering . •
We shall discuss the inelas tic s cat te ring o f neutrinos from pro ton \-P + µ
plus anything in detail in the las t part of this course . We do not know weak
interaction theory as well as quantum electrodynamics but can use these experi -
ments themse lves to test i t (e , g , if interaction is not a point-like four fermion
in teraction s caling will fail in the form derived below) - but s upposing it is
all righ t for now, we suppose the coupling wi th quarks is (ii y µ (l+iy 5 ) v ) (Q'y ( l+iy 5 ) Q)
u
wi th the quark quan tum numbers from u quark to cos e c d + sin e c 8 where e c
2
is Cabibbo ' s angle , s in e c ""' . 06 , For simp licity o f our rough discuss ion he re
we take e c - o .
154 Photon-Hadron Interactions
•
2
M11 v i [ 4 P 11P vw2 - 4it o 11 vw 1-2i & µ va A P 0 q Aw3 J ( 33 . 2 )
( leaving out all terms proportional to q because they give small effects at
11
2
the µ , v end propo rtional to (mass µ ) ) . w1 , w2 are as in ele ctricity except
they con tain VV and AA (nearly equal) ; the new term w3 comes from interference VA .
ln the scaling region vW , 2MW 1 , vw3 scale to three functions f (x ) , £ 1 (x) , t 3 (x)
2 2
(x • -q /2 Mv ) . lf partons have spin 1/2 , t • 2 x t • For the quark model we ge t
2
2 1
f 1 v p - 2 (u + d cos e + s s in e )
2 2
f 1 vn
2 2
• 2 (d + u cos e + s sin B )
f 1vp • 2 (u + d cos e + ii s in e )
2 2
f l vn
-
• 2 ( d + u cos e + s s in B)
2 - 2
2 2
f 3v p � 2 (u - d cos e - s s in e)
f 3 vn
2 2
= 2 (d - u cos e - s sin e)
f 3 vp -2 (u - d cos e - 8 s in e )
2 2
•
f
vn •
2 2
-2 (d - u cos e - ii s in e) ( 33 . 3 )
3
f vn + s sin e)
2 2
_
f vn 4 (u cos e
•
l 3
- 2
f 1vp + +
vp - 2
£3 •4 (d cos e s sin e)
+ s sin B) ( 33 . 4)
- 2 - 2
f 3vn • 4 (u cos e
2
Negle cting sin 8
p p p p
4 u · f 1 v + f 3v 4 u - f 1v f v
3
(33 .5)
n n n n
4 d • f 1 v + f 3v 4 d - f 1v f v
3
Sum rules
(33 . 6)
the firs t is Adler ' s sum rule (not quark model dependent) , the thi rd was first
( 3 3 . 7)
the las t term can probably be neglected , it gives a very close check on the model .
Also
fv
n
3 f 3v
n •
2
2 ( - (d-d) + cos a (u-u) + 2
rtn e C s-8> >
- 6 COB
2 ep e 2
ec (f - f n ) + sin e ( (s-8> - (d-d) ) ( 33 . 8)
by sub traction with the previous relation can be used to eliminate the las t
term if there was any doub t .
allows us to ob tain u + u, d + d , ii + a
p p ep en
Knowledge o f f 1v , f 1v , f , f
If the electron beam and the pro ton is polarized in the deep inelas ti c
s cat tering we can obtain s till more information , this time about how the
partons are polarized in a polarized pro ton . For defini teness we shall assume
Fo r a particle at res t the s tate may be des cribed by saying the spin is in some
particular s pace direction 8. Relativis tically trans formed spin s tates can be
described by a uni t four-vector s µ ( the t ran s formed s is as though i t were a
polar vector) satis fying s µs µ • l and s p 0 (p momen t um of s tate) . When
µ µ µ
• •
taking matrix elements we can use s purs and sum over s tates if we replace the
usual p roj ection operator ' + m by the mo re comple te {l+m) (i-iy s � ) /2 to proj ect
into the spin s tate s . Thi s s is a pseudovector . The s pin s tate o f p in the
µ µ
J factor o f
v
2 2
wµ v - '\� < X I J v l p ' >*<X I J I P> 2 110 (Mx - (P + q) )
L µ
need not be the san� as the spin of p in the J factor ; but the quantity we want
µ
depends on s pin as a 2 x 2 matrix (dens ity matrix) one of two states in , one o f
two out . However a l l the information is con tained i f the diagonal is known fo r
all s tates , not only for p • p ' • +z or -z but also for p • p' • x • ( (+ z) + (-z) '/12
for example ( in fact , these three and p • p' • -x are enough in our case) . Thus
(33.9)
( 33 . 10 )
Let u s use f o r example the coordinate system where the pho ton has· pure z
component
q • (0 , -2 P x, 0 , 0) 2Mv • 4p 2 x
2
P • (£ , P, O , 0 ) -q
µ
Now f i rs t s uppose the helici ty of the pro ton is +1 thus s • (P , £, O , O ) /M .
µ
Parton Model 157
Fur ther le t
2
h (x) • No . of partons ( charge ) wi th helici ty + in a pro ton of hel . +
+
2
h (x) • No . of par tons
_
( charge) with he licity - in a p ro ton of he l . -
µ µ µ µ
we have w • (P , e , O , O) /m • xMS /m for hel . + and w • xMS /m for hel . -
1 1
We there fo re h ave
( 3 3 . 11)
The function gives 6 ( �-x) / 2Mv in the s caling limit . The firs t integral gives
the res ults we already know 2 MW • f (x) • h -+h and v W • xf (x) . The se cond
1 + - 2
in tegral gives
( 3 3 . 12 )
s
0 0
• w • (O , O , 1 1 0) for par ton with pol . + x and w
0
• -s
0
for pol . -x .
2
k (x) • No . of par tons ( charge) with pol . + x in a pro ton of po l . + x
+
with po l . - x in
2
k (x) • No , of partons ( charge ) a pro ton of po l . - x .
_
We have an expres sion o f the type ( 3 3 . 1 1 ) with h+ replaced b y k and h rep laced
+ _
by k from whi ch we derive
_
3 2 2
2M vG + 2!r v G • m(k (x) - k (x)H' x ( 3 3 . 13)
1 2 + _
3 2 2
therefore the function 2M vG + 2M�v G s cales .
1 2
2 2 3
Hence from ( 33 . 12 ) and ( 3 3 . 13) !r v G s cales and Mv G s cales and we wr i te
2 1
2 2
M vG 1 • g Cx) , Mv G • g (x) . These are the s caling p redic t ions for spin cases .
1 2 2
( 33 . 14)
( 33 . 15 )
h
+
+ h
_
• k +
+
k_ • f (x)
158 Photon-Hadron lnteractions
Of course , these new functions g 1 , g 2 cannot be obtained from f (x) because new
dis tributions are involved , the number of partons with various spins is s tates o f
various helicity , so that w e derive nothing b u t the s caling rules even if w e use
a special model like quarks , etc. However , in that model we can ob tain a
remarkable sum rule of Bjorken .
We can use our proton wave function (described by parton) not only to ob tain
predi ctions for deep inelastic scaling but like any o ther wave function we can
write the expectation value of various operators for i t . We have already done
that trivially when we wrote j ll as the operator L ks e iy and found the
to tal charge on the p ro ton to be expressed as /l� �::; �
- t d-d) - t (s -s dx .3
Now we will discuss the expectation of a less trivial operator , the axial S
We calculate both sides for a proton moving fast to the right wi th momentum P
and spin s 0 and sum on proton s tates via a spur operator correctly projecting. On
the left s ide we have the corresponding sum on quarks , thus
Total of sp l <j+m) (l-i y wf) y 5y µ ] ! (GA /GV) sp 1 ¢+M) (l -ir 5f> r 5r µ l L mwµ
-
•
•
summe d with weight + 1/2 for u, u; 1/2 for d, d • (GA /GV ) s0M/ 2 . Now the sum
on the left is exactly the integral over x of 2 g1 , (or 2 (g 1+g2 ) depending on
the s pin case) that we needed for deep inelastic ep s cat tering excep t the
weights in the latter case were 4/9 for u, u, 1/9 for d, d and 1/9 for s , s for
-
the proton (or 1/9 for u, u, 4/9 for d, d, 1/9 for s , 8 for the neutron) while
-
here we want 1/2 for u, u, 1/2 for d , d. But the di f fe rence of s1 for p roton
g
lp and neutron gln has weights 1 / 3 for u , u, 1 / 3 for d , d or 2 / 3 of what
we want . Hence we have for the two polarization cases
2 J
1
(glp - g ln ) dx •
1
-
3
() GA
-
Gv
( 33 .1 6)
[i t
and x cases , an equali ty whi ch would have to hold no matte r wh a t the wei gh ts
Ji 2
dx • 0 ( 3 3 . 18)
0
a res ult of ro tat ional s ymme t ry (angular momen t um conservation) .
The fi rs t relat ion is more remarkable and would tes t s i multaneously the
quark view , and the in terp re t ation of the weak current as y ( l+iv ) no t only for
µ 5
the elemen tary parti cles e , v , µ, v but also for the elemen tary cons ti tuent
e µ
quarks (or quark quantum numbers on spin 1/ 2 partons ) . Its ve rification , or
failure , would have a mo s t decis ive e f fect on the dire c tion o f future high
Lecture 34
research prob lem about angular momentum and part on wave funct ions ( func tions
transverse momenta kl. is no t known , and are very in teres ting theoretically . Can
momentum proper ties of parton wave functions ? For examp le , what res t rictions
if any on the wave function come from the fac t that the to tal angular momentum
of the proton is j us t 1/2? Or , s uppose the parton probab ility dis tribution f (x)
were known for a 6 o f helicity + 3/2 . What could you say about the dis trib ution
Dre ll has suggested another expe riment which can be predi cted by knowledge
of the parton dis tribution . Again we use the parton as quark model to des cribe
the idea, but it can of course be analyzed the same way in o ther representa tions
+ µ µ-
for par tons . The reaction is p + p + + any thin g . P ro ton p lus p roton (or
+ -
p + p) at very high s (e . g . P in CM is large ) makes a µ µ pair whose energy is
160
Tests of the Parton Model 161
The picture is
the µ pair is made from the annihi lation of a p arton in pro ton A and an ti -parton
in p ro ton B and the o ther partons j us t keep going on to make final hadrons , the
"any th ing" . By sui table cho ice o f the energy and momentum ( in z dire c t ion) o f
� " ( c -p ) / 2P
z
It will be eas ies t to meas ure the pair at large angles ( the angular dis tri
b ution is 1 + cos e ) so no great loss o f intens ity o ccurs here , and b ackgro und
2
hadrons which should move p rimarily in z di re ction are avo i ded . If 6 " 0 is
chosen howeve r , the momenta of the two µ 1 s give direct ly P x , P � be cause in s uch
1
a one-dimentional collis ion of re lativi s t i c particles the momen tum of the final
The cross section for th is is the c ross sect ion fo r relativis tic p air
4 2 2
annihilation ( e ( 1 + coa e ) / 16 c ) t imes the chan ce C o f finding the app rop riate
4 - 1 -
C • 9 1 2
-
2 1
-
(u (x ) u (x ) +u (x ) u (x ) ) + 9 ( d (x ) d (x )+d (x ) d (x ) ) +
2 2
1 1
1
+ 9 1
-
2
-
( s (x ) s (x )+s (x ) s (x2 ) )
1 ( 34 . 1)
high enough that Px 1 , Px2 are large compared to 1 GeV) . One might at f i rs t be
pro tons scatter via s trong interact ions so the pos s ib i li ty exi s t s tha t the parton
distribut ions are firs t dis turbed by " s t rong" interact ions be fore they annihilate
to make 11 ' s . But we a re coming to unde rs tand tha t the "s t rong in teractions "
are not so s t rong in th is sense - that in hadron-hadron col lis ions it is not
the f as t par tons that interact , b ut the wees . This accounts for low momen tum
avoid this wee interacting reg ion (interact ions via fini te energies , not o f
Lec ture 35
p p + u u- +
+ +
any thing ( con tinued)
Gronau has pointed out that with no further data than we already have we
( a • . 2 4) we have
ep
But the las t fac tor is j u s t f (x) , the 2MW1 for the pro ton and is known ,
easily se rve as a test of any parton mo del - and will be an interes ting experiment
to watch .
If B is not zero we ge t
In a s imilar way we could describe the pro ton antipro ton case Jl+P+µ+µ - +
anything . I t would be determined by a fac tor
( 35 . 2)
If finally , the functions u , ii etc . are determined for the pro ton , the
corresponding functions for the pion u ( x) , ii ( x) etc . could be ob tained through
x x
+
( difficult to get enough energy) w + p + µ µ - + x for now
1 s x )-s x ) - x ) s x ) )
+ 9 ( p ( l w ( 2 +s p ( l w ( 2 ( 35 . 3 )
e4
do
cID • e·i 2 2 2
( l+cos e ) ( 35 . 4)
4q
o ( e+e - + hadrons)
( 35 . 5 )
o ( e+e - + µ+µ - )
That is , the ratio approaches a constant , and the constant is the sum of
squares of parton charge summe d on each type that exists . If there are any
1 2
1 as strongly e .g . 4
spin 0 partons they would contribute 4 ( e for each spin zero
par ton) • We call D • 4 e / "Drell ' s constan t �·
"' 2 4 1 1 • 2/ 3
For example if partons are quarks we expect L... e i • 9 + 9+ 9 for D .
I f partons must carry integral charges , and be spin 1/2 , the sum mus t be a t least
1 64 Photon-Hadron Interactions
l. In the case o f quarks we would have the dif ficulties we have dis regarded
as to how a pair of opposi tely 1110 ving charges of ± 2 / 3 say can turn into only
hadrons of integral charge . There may be s ome way to get the required extra
quarks from a soft sea o f pairs , but one can perhaps legi timately question
whether the above arguments are entirely clear , involving as they do an assumption
that the parton s tates go entirely into hadron s tates . I s i t really consistent
to do this and omit s tates of quark quantum numbers ? I t would be mos t exc iting
if the ratio does come out 2 / 3 as well it migh t , for the theore tical ques tions
produced by such a simple answer would be very interes ting . I think it would
be worthwhile as an exercise beforehand to assume the result was really 2 / 3 and
see what paradoxes , if any , would then have to be resolved .
Aswe discussed p reviously theoretically ( lecture 5 ) , the total cross section
+ - 2 2
for o (e e � any hadrons) as a func tion of q ( q • 4El, E • CM energy) de te rmines
the vacuum expectation value of the product of currents
(q µ q v - o µv q ) a
2 2
<O / J µ (-q ) Jv (q) / O> • ( q0 ) p (q ) ( 35 .6)
with
2 2
CJ -
(4 TTe ) 2
--- p (q )
2q 2
2
( q µ q - q o \lV ) v (q )
2
F . T . <O j {J (l) J (2) } T j O>
v v
= ( 35 . 7)
µ
l 2
where dispersion theory told us Im( iv) - 2 p (q ) '
00
J
_.i.._
2
4TTe i (v(q ) -v(O) J •
2 2 cr ( s ) ds
-- - ( 35 . 8)
2 2
rr (4rre ) 4m 2 s -q
"
(l + above quantity) is the factor by which a photon propagator l/q 2 mus t be
multiplied for first order effect of virtual hadrons i .e .
2
CA _
_ _
,.� _
•
1
2 4TTe 2 i [ v ( q ) -v(O) ]
q
The low energy tes ts , like e f fect on Lamb shift depend j us t on the lowest
2
q or only on iv ' (O) that is os -1 ds . J
Today we can
begin to say something quite detailed about this , for in the
2 2
region ( 2m " ) to jus t above 1 GeV the cross section is dominated by the P , w , �
2 2
production . Perhaps p (q ) set tles down to D/6 TT for large q fairly soon .
Al though one migh t s uppose we migh t have to go decidedly above the nucleon
Tests of the Parton Model 1 65
pair creation , this may not be t rue ; soft parti cle produc tion may domina te at
all ene rgies and reach i ts asymptote soone r .
A . Cisneros has calculated the various hadronic con tributions t o iv ' (O) .
The con tributions from the p , w and + turn out to be ten times larger than
2
the large q contribution wider reasonable assump tions about the likely value of
2 2
Drell ' s cons tant and the value o f q at which p (q ) attains i ts asymptote .
+
Data from e e - intersecting rings can be used directly to ob tain the
con tribution from the p. The data is simply inserted in the formula
..
f
2
2
iv ' (O) .tl!L.l
1
• 2 2 2 dq (35 . 9 )
2
11 ( 411e ) 4m11 q
to ob tain
( 35 . 10 )
(What was used here was the data on the pion form factor which is almos t
everything there is in y (virtual ) + p + hadrons . The 411 inelas ticity is very
2 2
smal l even at q ( . 8 GeV) ) .
•
In the cas e of the w and + we assume VDM, which works wel l in this type
of process . In this case the contribution from the vector meson v • w o r + is
2
p (q ) ( 35 . 1 1 )
2
From the relation Im ( iv(q ) ) 2 1 p (q 2 ) we deduce
•
2
2 1 mv
i vv ( q ) - 2 --..,2.---,2.----
•
( 35 . 1 2 )
gv (q -mv ) -ir m
v v
For a very narrow resonance , which is a good approximation for w and + ,
2 2 2 2 2
we have p (q ) • mv g,,- 2 116 (q -mv ) the value o f iv ' (O) is simply
( 35 . 13)
-2 2
(This formula gives i v ' (0) 5 . 3 x 10 GeV- for the p, in good agreement with
p
•
what was ob tained above using the data directly . ) The contributions o f the w
( 35 . 14)
( 35 . 15 )
166 Photon-Hadron Interactions
2
We now evaluate the con tribution from the " tail" of p (q ) assuming it has
2 2
at tained i ts asymptotic value D/6n at q • q0
2
S.... . _D _ _L (35 . 16)
l2 i q 0
4 2
q
•
-� 2 2 2
This gives iv ' (O) . 84 x 10 Dq0 - If q 0 is as low as l GeV and D
t
-2 -2
is the quark value 2 / 3 we have . 56 x 10 GeV for iv' (O) , We believe the
t
actual value of the non -vector meson contribution not to be much larger than this ,
we ass ign to it 100% uncert ainty . Adding the various con tributions we get
finally
( 7 . 3 t 1 . 1) x 10
-2 2
iv' (0) • Gev -
hadrons
This corresponds to a correction to the magne tic moment of the muons o f
8
(5 , 5 ;t . 7 ) x 10 - in (g -2 ) /2 .
2 2 l •
2 2
For comparison , iv' (O) for muons is [ 4 n 15 M 1 - 15 , 3 x 10 - Ge v - ,
µ
so the hadronic contribution is about one half as large .
2 2 2
Away from q ""' O , the contributions to i (v( q ) - (0) ) from the tail of p ( q )
grow as the vector meson con tributions fal l , so the uncertainties are greater .
Ano ther expec t ation of intere s t from the parton model is the angular dis tri
+ -
bution o f hadronic products in high energy e e collisions .
,f;' hadrons
hadrons �
We expect that partons of spin 1/2 are produced with an ( l+cos 0) angular
2
dis tribution and that the final observed hadrons will have small transverse momenta
relative to the direction 6 in which the partons were pro duced at high enough
energy . Two bursts of oppositely moving hadrons are expected which will determine
the angle e . If there are also charged partons of spin 0 the angular dis tribution
will be
D 1 (l + cos e) + 2 D0 (1 - cos 0 )
2 l 2
1 2
where D 1 and D0 are Dre ll ' s cons tants for partons of spin 1 / 2 and spin 0
1 2
respectively . For quarks we expect of course only a (1 + cos 0 ) dis tribution .
2
Inelastic Scattering as Properties of Operators
Lecture 36
We now cons ider it in a more abstract way . Firs t we note that if the s tates
p and x are imagined to con tain their center of mass momentum factors , the
2 2
2 w6 (M - (p+q) ) would be unnecessary , for the class of s tates x ' now mean one
x
of mass Mx moving with momentum P . Thus
167
1 68 Photon-Hadron Interactions
or by completeness ,
O
J
hadrons
J
Such terms do no t ex:lst for the non-di.igonal matrix elemen ts of the product of
operators term < p / J ( -q ' ) J (q) / p> and we should like to define K as the limit
v µ
of this as p + p to ob tain a sui tab ly use ful defini tion o f matrix elements
of the pro duct operator. We can get the same res ult if we exp lici t ly s ub t ract
the contributions of the disconnected diagrams o f the type above . Hence we
write
K
µv
(q ) • < pj J v ( -q ) J µ (q ) ip > - <O / Jv (-q) J µ (q) j O><p / p> ( 36 . 1)
Then if the final p is changed to p ' slightly di fferent , the express ion
de f ined by ( 36 . 1) is cont inuous .
We would like to consider , for theoretical purposes , various o ther currents
beside electromagnetic - we can have vario...s su3 generalizat ions in an octet (or
nonet) (i . e . currents wi th different charge numbers ) and axial curren ts as well .
Thus we le t these indices (su 3 and axial) as well as the µ , v indices o f space
be contained in a s ingle letter A ( calling J µ (l) A ( l) e t c . ) to save wri ting•
in general arguments , and can always go back and insert indi ces at the end .
Thus i f A is any operator o f our allowed set ( for our examples , a vecto r or
axial vector local current in su 3 octet or poss ib ly s inglet) we wri te (obvious ly
the diagonal pro ton s tate can be generalized immediately too , but we leave chat
for the s tudent) .
And las t ly we omi t writing the l as t term for s impli city , b u t you mus t
2-1 , t
i s Fourier trans form s KBA (q) . We exp li ci t ly
i call the compon e n ts of
q • (v , Q) in the sys tem where the p roton i s a t res t (o the rs via re la tivis t i c
trans formation) . Thus I P > is at res t
•
�l x
2 < p ! B (-Q) x> z!
x
<x ! A (Q) j p >5 ( Ex -E -v)
P
( 36 . 3)
Since K gives the product ope rator we can find from i t the commuta to r as
We note , in our case (in the res t sys tem of the proton) , since the p r ot on
is the lowe s t mass of all s tates o f baryon number one (and A does not change
baryon number) that Ex > E fo r all x. Hence if v < 0 th e 6 fwiction is zero -
p
�A (v , Q) • 0 if v < 0 (36 . 4)
discus sed the s cat tering ampli tude for an incoming photon (virtual or real)
coupled to J (l) say A (l) to an outgoing one coupled to B is determined by the
operator
we discuss the effect of seagulls (if there are any , 6 (2 -l) type terms ) later
and omit them for a while .
( 36 8 )
.
(The superscript F indicates that the choice o f s ign of the imaginary part
for negative frequencies is taken according to the convention of Feynman in his
c
QED papers ; there is a different choice called causal amplitude , T BA ' which is
of ten very useful . )
we have the convolution of these . In the same way the second term is the
convolution of -i/ (v-i£) ( the F . T . of 9 ( t -t ) and K (-v ,
1 2 AB ..Q) .) Hence
( 36 . 9 )
t
v ' -v
l
Now use
+ i£
. p ,.!-
v -v
ho (v ' -v)
( 36 . lO)
where we have explicitly added possib le seagull terms oBA which is simply an
unknown ( finite order) polynomial in v , Q . Thus the scattering ampli tude is ,
except for a polynomial , given in terms of the commutator .
Negative �. is of course , not defined by experiment . However it can be
ob tained from measurements of the reaction with antiparticles (using A for A)
via the connection implied by ( 36 . 10) and the relation for commutators ( resulting
Inelastic Scattering as Properties of Operators 1 71
Lecture 37
�
T A( v , Q) • :
T A (v , Q) + i8 ( -v) CBA ( v , Q} ( 3 7 . 1)
c c * ;t
Thus (TBA ( v , Q} } • TBA (-v , -1.1)
�
T A (v , Q) • J CBA (v 1 ' Q) v �����: + 0BA ( 3 7 2)
.
c
The significance of T BA (which is j ust as good a way of des cribing scattering
as T ) in coordinate space can now be seen . (TC is given by an expression like
F
( 36 . 9) except that the sign of the +iE in the las t term is reversed , } In ( 3 7 . 1)
we have expressed things in terms of the commutator but the las t term can be
more s imply expressed as the product operator , from ( 36 . 5) and ( 36 . 6) we have
8 ( -v ) CBA( v , Q) • -KAB (-v , -Q) , the •
F . T . of A( l ) B ( 2 ) . Hence ( 37 . 1) says
- iT C • F . T . { B (2)A (l) } T - F .T . A ( l) B (2 )
for t 2 > t 1 w e have B (2)A(l) - A(l) B (2) • [B (2) , A(l) ) and for t 2 < t 1 , w e have
A(l) B (2) - A(l) B (2) •. 0
Hence the scatte ring amplitude ( causal)is the Fourier trans form of the
re tarded commutator + seagulls . By retarded commutator we mean the commutator
( 3 7 . 3)
diagonal elements , or lowest s t ates (so the produc t operator is 0 for v < 0) ,
The general defini tion of the Feynman amplitude is the time o rdered opera tor ( 36 , 8) ,
(37.5)
Cross ing
{T � (v , Q) } * • T� (-v , -Q)
(37 . 7 )
Imaginary Part
• O BA ( -v ,
/
oB (v , Q) -Q) • oAB (v , Q)
(3 7 . 9 )
Inelastic Scattering as Properties of Operators 1 7)
becomes upon Fourier t rans form for our diagonal element on the pro ton ,
..
1 C a b ( v , Q ) dv / 2 w <p ! J ax
J0 J
µ
µ
b
•(O , O) i p > ( 3 7 . 10)
we shall ask are ( 1) What limitations on the F . T . result from the fact that the
coamutator vanishes outside the light cone? ( 2 ) From experimental facts about
( )
usual way
( 37 . 1 4)
Here we will just do the special case m • O , and s ay wha t the case m2 ; 0
1 74 Photon-Hadron Interactions
-) •
C ( t ,x j 2 irk2 dk dcos e
3 e
-ikt ikRcos e
e + C,C,
(21') 2k
For f inite mass we obtain the same s ingulari ty on the light cone , zero
Thus we see that the commu tator for free particles is zero outs ide the ligh t
We have dis cussed the s ingularities of two fields . Also inst ructive is to
dis cuss the s ingularities from two currents . One such current might be in a free
an example we will leave out these gradients , e t c . and find the commutator for
where C ( 2 , l) is the free field coDDDu tator we worked out before . Clearly K
m
2
has the same s ingulari ty sgn ( t -t ) 6 ( s ) as noted before for C , but this time
2 1 12
multiplied by an operator function of two positions F ( 2 , l) cal led of ten s
b ilocal operator . This operator is needed , of course , only on the light cone .
s ingularities along the cone are modulated by a function of the dis t ance from
outside the light cone also , and non-zero only inside . An interes ting question is
how i t makes the t ransi tion as we cross the light cone , Some sort of j ump is
Inelastic Scattering as Properties of Operators 1 75
experiments indicate K(v , Q} satis fies B j o rken scaling - that is, it is only a
fwiction of E; • -q / 2Mv .
2
The mos t straightforward way to analyze this is to take the inverse trans form
o f K, using the Bj o rken limit and seeing what s ingularities it gives - then
concluding the character of the s ingularities , because they depend only on the
high v , Q limit are the same for the complete function as they are for the
Bj o rken limit . We will not try to be rigorous , because for one thing we really
do not know , experimentally , K for q 2 > 0 , nevertheless we can see what happens .
2 2 2
Let Q be along the z axis . -q • Q -v • (Q z+v} (Q -v} • 2Mv ( , hence for
z z
•
large v , Q z is nearly v in fact Q z v+M( . The Fourier inverse of a function
of ( only would look like
Je Je
1 i ( vt-Q z} iQ z ( t-z} -iM(t
-- z f ( E;} dvdQ 1
• -- e f ( E;} d(dQ z
(211 )
2 3 2
(211)
• s ( t} O (t-z)
J o (q2+2MvB} f ( B } dB
2
f (-q /2Mv} • 2Mv
Now omit for a moment the 2Mv (suppose we took the trans form of w2 } and
2
note for large v , q 2 that to a good enough approximation the 6 (q +2Mv8} csn be
replaced by 6 ( q +2MvB+� s 2 } . So we get for large enough v , ( q +2MvB+� s • (q+$P} 2 )
2 2 2
Now we can inser t , if we wish , a sgn(v} to keep the asymme try as is required
for a F . T . of a commutator, and for large v this is just sgn (v+BM} ( if the
proton is at res t} , Hence asymptotically
1 76 Photon-Hadron Interactions
2 2
F . T . sgn ( t ) o ( s ) • sgn (v) o ( q )
hence
Jf ( S ) dS e
F . T .W2 -
-iSP · xsgn ( t ) o ( t 2 -R2 )
that s caling shows that the s ingularities on the light cone of the current
cOllDDu tators are j us t of the same severity as they would be for free particle
fields .
This is , of course , what we expect from the parton interpretation for there
we assumed in the final statl!s x (which become the intermediate states of the
commutator) the partons can be considered as free . Thus there is no surprise
in the conclus ion , it is only that we wish to state what we have discussed
(scaling) in an abs tract and general way (as current commutator s ingularities)
with minimum references to a mode l . Although the s tatement s till seems to refer
to free particle fields , that is only a shorthand to wri ting 6 and o ' functions
on the light cone .
Each general property of partons we assumed said something more explicit
about the character of the singularity . For example , to say charged partons
are spin 1/2 is to say �he singularities are like those o or 6 ' functions charsc-
teris tic of free Dirac field commutators . In addition , vector and axial vector
results are related . I f , for example , one adds that partons are quarks certain
numerical relations are implied , as we have seen , between the singular parts of
the commutators o f various kinds of currents . (That is to say for example , our
Inelastic Scattering as Properties of Operators 1 77
� p
�
relations among f ep and f P , f etc . implied by expressing everything in terms
of the six functions u, u, d , d, s , s. )
Quarks , as free particles , have not been found . There are many questions
as to whether the detailed views of the parton model are correct for quarks . In
particular , the question whether a single par ton quark moving out in recoil need
or need no t take its non-integral quantum number with it .
In addition it is often ve ry use ful to see , when results have been obtained
from a mode l , j ust how much depends on the model and whether in fact the results
Lecture 38
Li gh t Cone Algebra
An interes ting ques tion answered by Fri tzsch snd Gell-Mann ( 19 71 Coral
Gables conference) is how the s caling results of the "partons as quarks " csn
be stated in a way which - at the end - doesn ' t involve quark wave functions
or ope rators at all .
We wish to s tate that the ligh t cone s ingulari ties are like those for free
quark (spin l/2 , su 3 triplet) c0Dm1utators . We mesn , of course , the largest
&' or & type of s ingularity , as we say the "leading" (in high frequency)
singularity . Firs t we see what the singulari ties are like if currents were
f
( Jus t as the propagator is < + m) / (p
2 2 2 2
- m ) instead of l/ (p - m ) so this
j + m comes likewise into the cODm1utator ( the real part of the propagator) .
Since we are looking for the leading singularity near the light cone Cm is
1 78
Light Cone A lgebra 1 79
2
� ( • ) and the mass term is smaller than the large gradient , thus
o f two curren ts each a b i l inear form in �. are exactly the same whe ther we
s uppose a) the fields obey the us ual anticommu ts tion relations at equal times
3
• <i2 ) �( i1 ) + �(x > • <i2 ) • 6 (i - i2 ) and � (i1 ) , � (i ) an ti commute as is
1 1 2
app ropriate to Fe rmi parti cles ; o r we suppose b) the spin 1/2 f i elds obey
3 -
commutat ion relations � ci2 > �Ci1 > - �cx1 H <x 2 > • 6 Cx 2 - x > and � <x H <i2 >
1 1
co1111u
1 te as is appro p ri ate to Bose part icles ,
This is very interes ting b e cause it says the "Bose quark" model which
to 11 partons as quarks " at higher energy . We can c a l l this parton model "partons
as Bose quarks . "
a
Next a current o f su 3 type a (described � ). ) is
•
a a
\I II
J (l) �(l) :>. y • ( l) (38 .2)
Now when we commute two currents , we find a s imple but tediously complicated
resu l t (see re ference for details) , To illustrate the idea , which is all we
intend to do here , we take the case of two electric currents , and also drop
several terms which would vanish i f we took spin averaged mat rix e lements , then
where
(38 . 4)
(2 , 1 on mutual light cone} an d :>. 6 is diag , ( 4/ 9 , 1 /9 , 1/9} • ;1. Y ;1. Y in our case .
Now we s t ill s ee the quark fields in ( 38 . 4) , but the idea now is to dis -
to arb i trary currents and including omi t ted terms of course} are gene rally valid .
180 Photon-Hadron Interactions
They give the ligh t cone s ingularities and de f ine , on the ligh t cone at leas t
a set of new operators :� ( 2 -1) . It is the matrix e lements of these ope rato rs
q uark ope rato rs are seen . Ins tead only some new bilocal operators are defined
(meaning depending on two poin ts ) . They are de fined only on the light cone by
[ Th is sys tem woul d be a true " algeb ra " if the p ropert ies o f the '.t' could
now be de f ined indepen den tly . Fo r example , an i de al s i tuation ( in fact a
comple te theory of the hadrons ) wo uld res u l t i f equations giving the commutators
o f s uch ·:P a could be given in terms o f J ' s and '.f' s themse lves . This does no t
commutators of two :ft s which have their variab les on the s ame ligh t ray
cone ) again can be expres sed a s a n '.f'. This re lation has been a l s o hypo thes ized ,
but i t is not much . I t gives predict ions for certain two-current inclus ive
+ -
reactions l ike e + p � e + µ + µ + any hadrons . They are , of course , j us t
those expe cted f rom the par ton model in the same condi tions . We discus s them
later . ]
for that we can use the result (ve ry s imilar to those we have already der ived ,
J d
4
x e
iq • X
6 ( x ) F ( x , x1 , x2 , x3 )
1 2
0
11
J - e
d t - iMf; t
F(t , O, O, t)
Iti
that is , i t involves on ly the integral along the ray t • z common to the plane
i v ( t -z)
t • z ( from e ) and the l ight cone .
Naturally every result o f this theory is also a result of the " par ton as
quark " theory for the latter is a model of the former ; but not eve ry parton
res ult can be der ived from ( 38 . 3) . It i s pos sib le , there fore , that many of
them are wrong and only ( 3 8 . 3 ) and no t the comp lete field model may survive .
Light Cone A lgebra 181
Lecture 39
Properties of Commutators in Momentum Space
C ( q) •
J
C ( u) P , V .
l
16 11
c -u >
2 2 --
4
d u
(211)
4
( 39 . 1)
;& o
(plus pieces from the light cone) , (C (q) is the F . T . of a commutator, )
In a s imilar vein , since the F . T . of
�
we see that C must have the form
1 82
Properties of Commutators in Momentum Space 183
C (q) •
J 2 d4
X( u) sgn ( qo -uo ) o ' ( (q-u) )
( 2 w)
� (16 wi ) ( 39 . 2)
plus light cone pieces , Of course we can say also in (39 . 1) that the C (u) of
the integrand is any function F(u) at all , and the C that comes out on the le ft
will be the F . T . of a function zero outside the light cone . I have not been
able to make much use of these observations .
In us ing the simple views above we mus t be careful of one point . The
func tion we 111U l tiplied C (x , t) b y , G ( x , t) , was defined as 1 ins ide the light cone
and 0 outs ide ; what is it exactly on the light cone ? It is indefinite there ,
ordinarily that is such a small region that it makes no difference in the integral
ove r space t ime of C (x , t) G(x, t) but in fact it makes a great uncertainty because
C ( x , t) has a 6 function singularity j us t where G(x , t) is poorly defined . We can
correctly straighten out our formulas for this effect in the following way , We
call Ca (q) the asymptotic C ( q) giving j us t the light-cone 6 s ingularities . Than
C (q) -Ca (q) has no light cone s ingularity and so equat ion ( 39 . 1) for example , holds
if C is replaced by c-c a on both s ides . One can then simplify or rearrange the
equation to obtain one like (39 . 1) b ut with some addi tional terms related to
the light cone behavior .
A far more s ub tle and useful observat ion was made by Dyson . He noticed ,
v > - M + J�� 2
and likewise
v < + M - � ( 3 9 . 3)
Dyson has proved that the ne cessary and suf ficient condi tion fo r C ( q) to
vanish in region S • s ( q) < q0 < s (q) and to have a Fourier trans form vanish
1 2
outs ide the light cone is that C ( q) can be wri tten as
C ( q) • J J: 4
d u s
2 2 2 2
s gn ( q0 -u ) 6 ( (q-u) -s ) t ( u , s )
0
( 39 . 4)
where � vanishes outs ide a region R , but is otherwiae arbi trary . The region R
1 84 Photon-Hadron Interactions
( 39 . 4)
It i s easy to see , by drawing hyperbolas that for our case the region R
(using 37 . 2 ) becomes
T c (q) �
1
2 11
J J
R
d 4u ds 2 � (u s 2 )
---'---=--�...._ ___
2 2
(q-u) -s +i£ (qo -uo )
+ a (q ) ( 39 . 5 )
One o f the di f fi cul ties i n using the Dyson representation i s that the func tion
� is no t unique , many can give the same C (q) . Another very s imilar repres en tation
has been derived (I sus p e c t not as rigorous ly as Dyson ' s ) by Deser , Gilbert and
Sudarshan especially for a problem like ours whe re C is a function only of the
two invariants q2 , v, it is
C (q 2 , v ) s � �
2 Mv
...
da
+l
dB H (a , S ) s gn (q +SM) 6 (q+Sp ) 2 -o )
0
( 39 . 6 )
0 - 1
Properties of Commutators in Momentum Space 185
de finition for o .
2
2MW (q , v )
1
• 2wv � �
m
do
+l
2
dBh (o , S ) s gn (v+MB ) 6 (q +2SMv -a ) ( 39 . 7)
0 -1
In addit ion , for our case , the asymme try in time for the coumutator becomes
the p roperty that
h (o , -B ) • -h (o , B ) ( 39 . 8)
Region I
2 2
First of all we have the s caling limit v + • , -q + • , -q / 2Mv • x finite .
Here 2MW 1 • f (x) a function of x only . Eq . (39 . 7) gives (sgn ( v+SM) • 1 for
1 86 Photon-Hadron Interactions
JJ
dB h (o , B ) 6 (-2Mvx + 2MvB-o ) do
0 -1
Now if we assume h (o , B) falls away rapidly enough with a only fini te a are
involved here , so as v + "' the a can be dropped in the 6 function and we have
h (o , x) do a function o f x as required
f (x) • f ..
h ( o , x) do ( 39 . 9 )
0
I believe i t was this argument with the Dyson representation which either
led Bjorken to his scaling hypothesis , or helped to confirm his suspicions of
i ts truth .
q 2 > 0 s caling
We can now determine how the function 2MW 1 (q , v )
2
We have here a bonus .
q + 00 such that
2 2 00 ,
+
f
should go in the posi tive q s caling region . Letting v
l 00
+ /2Mv • x ' is finite we have 2MW 1 • h (o , -x ' ) do which again scales ; but
even more , from the symmetry of h it sc�es to the same function fep (x ' )
-f ep (x ' )
2
2MW • q > 0
1
How can we see this remarkable result from the parton model? Can we
2
equally well derive the function for q > 0 by using physical arguments ? We
will explain qualitatively how it comes about .
For positive q we mus t be concerned in the commutator ( 36 . 1 ) with the
2
subtracted vacuum piece . In the vacuum we can make pairs , and at high l j us t
a pair o f partons
We are concerned with how this pair production probability is modified by the
presence of a proton
t? ·
At firs t one might think the modification mus t come through interaction of
Properties of Commutators in Momentum Space 187
the newly made partons with those in the proton - in interaction about which
we know lit tle except that it is fini te and is no t involved in our derivation of
2
f (x) for negative q But at high energy a mo re importan t effect is the
exclus ion principle - partons cannot be made if they are already present in
the pro ton (if Fermi s tatistics is assumed) . Thus in our diagram the proba
bility to produce a u to the left and u to the right at x' is 4/9 units (for
the charge is 2 / 3) . However , it is altered at those x by the chance , u (x) , that
a u parton is present in the proton . Thus we have a contrib ution proportional
to (4/9u (x' ) ; if the u p arton is moving forward it is (4/9 ) u (x ' ) . If a d par
ton pair is produced the probab ility is 1/9 , modi fied by the chance of finding
a d parton already in the proton , the contribution is (l/9 ) d (x ' ) , e tc . ) Thus
the entire contribution is 4/9 [u(x' )+ii (x ' ) ]+ l/9 [ d ( x ' )+d (x ' ) ] + l/9 [ s (x ' )+s(x ' ) ]
or f ep (x ' ) as require d .
On the other hand with the Bose statistics s ign the imaginary part o f
<O j JJ : O> i s negative and i t cannot be written a s a sum of positive probabilities
E j <O j J j x> I 2 but is minus that sum. This circumstance is the b as is of the
,
x·
usual proofs o f the relation of spin an d statistics ; spin one-half particles
canno t be consistently interpreted as Bose particles . The moat straightforward
interpretation of Bose quarks would mean the "Drell. Constant" defined for the
e+e - + any hadrons crose section would have the imposeible value -2/ 3 . Of
course a completely naive interp retation of quarks as Bose particles would also
lead to the wrong result that the wave function would be synmetrical under
the interchange of two protons . If quarks obey para-s tatistics both of these
problems would be removed .
188 Photon-Hadron Interactions
Region II
( 39 . 7) reads
2MW 1 • 2Mv �� m
da
+l
2
dB h (a , S) o (q +2Mv S -a )
0 -1
h (a , S) • k(a) / S ( 39 . 1 0)
2!ti 1 • 2Mv f m
k (a
a-q
� da
0
Thus
�
-q
i�0
m
a-q
da
�
gives the -q dependence of the totalmvirtual photon cross section . The limit of
g ( -q 2 ) as -q 2 gets large approaches k (a ) da .
0
Lec ture 40
Region III
2 2 2
v +
m , large -q but Mx � • 2Mv - (-q )
2
Finally we turn to region III :
2 2 2
kept finite . There we expect 2MW 1 (-q ) -y h (Mx -M ) 2Mv where h is a function
•
2 -Y
of M z _; only . The (-q ) 2 2
is typi fied by the elastic s cattering (Mx -M • 0)
x
2 2
and y is perhaps 4 or 5 . For x near 1 , f (x) ..... A ( l -x) y-1 and for large Mx -M •
2 2 .2 1
h (Mx -� ) goes as A (Mx2 -� ) y- •
Properties of Commutators in Mo111e11t11111 Space 189
J f
"' +l
2 2 (40 . 1)
2MW1 • 2Mv da dB O ( Mx -M +2Mv (B-l) -a )h (o , fl )
0 -1
Again a "natural" guess (neglect M 2 -� -a relative to 2 Mv (B-1) in the 6 function)
�
Mx -� . But again we can fix it up by a mor! comp li cated guess for the behavior
of h (o , B) near a 1 . Because , from ( 39 . 9 )
• do h (a , x) i s zero ss x + 1 (like
0
A (l-x) y - l ) , Therefore one way to get the answer would be to assume that h (a , B )
goes to zero near a • 1. For example s uppose
h (a , B) • D (o ) (1-B) y - l as fl + 1
l
"'
2 2 1
2MW1 • (2Mv) -y+ l (Mx -M -a) y - D (a ) da ( 40 .2)
2
hence of the correct form if ( call Mx -� • A)
J
A
h (A) • ( A -o ) y -l D ( a ) d11 (40 . 3)
0
This appears a t firs t satis factory ; h will have the right asymptotic behavior
AA y-l for large A if D(a) falls off fast enough with o, and j D (o ) da • A .
0
However , we will again be fooled in our qualitative e xpectations for h (A) from
this equation if we suppose that D(a) is a s imp le function . The operation of
integration in ( 40 . 3 ) is a very powerful smooth ing operation ; for no s imple
D(o) would we be lead to guess what we know is true - that h ( A ) shows a series
of peaks and valleys for small A , the resonance peaks , It is no t that (40 . 3)
1 90 Photon-Hadron Interactions
says resonance peak.a are impossib le - it is that (40 . 3) plus a guess that D (o )
i s not too complicated (i . e . does not vary i n a rapid and peculiar way ) fails
to do so . To see this take the case y • 4 and note that ( 40 . 3) can be solved as
1 d4
DO.) • 6
-:-:ii h ( >. ) (40 . 4 )
d >.
The fourth derivative of a resonance oscillates in a very special manner . (For
a much more intelligent way to deal with the resonance region see Cornwall ,
Corrigan and Norton , Phys . Rev . DJ 5 36 , ( 19 71) and the next section . )
To summar ize this inves tigation , the Dese r , Gilbert , Sudarshsn representation
is use ful as a formal tool to get dispersion relations , etc . It is , as of course
it ought to be , capable of representing what we know of the experimental data
with some function h (o , S ) . But it is disappointing in that its predictive
utility - by guessing "reasonable" behavior for h (o , S ) - is not very great first
because we have no phys ical picture o f what the function is , and second i t seems
like a function that behaves in a way as complicated as (if not more complicated
than) the data i tself .
It seems that h (o , S ) is not , in fact , a function that has much direct
physical s ignificance and is purely an artificial expression in a certain
mathematical form of the fact that the commutator vanishes outside the light
cone .
It might pay to try to find ano ther representation o f this fact in which
the kernel functions may have some direct physical representation .
_,,.,1 ., r
2MT 1 ) is
2 -l ( 40 . 5 )
2KT 1 • dS h (a , O) (q +2K,""'1< op (_) ) + , ...... ..
0 - 1
ii
... +l
2MW1 2 ( 40 . 6 )
• 2Mv d dB h (o , S ) sgn (v+Ml! ) O (q +2Mv-o)
0 -1
Properties of Commutators in Momentum Space 1 91
This is the causal amp l i t ude convention , in the Feynman conven t ion the
it sgn (v+MI!) is replaced s imply by i t . The "seagulls" are a polynomial in q 2 ,v .
This can be readi ly verified ( s imp ly n o t i ce tha t sgn (v+MS ) 6 (q2+2MvS-o) • s gn (v+MS )
a«v+MS ) 2 -E2) • ( 6 (v+MS -E) -6 (v+MS+E»l2E where w e have writ ten E • +(o+M2 S 2 -+Q2 ) 1 12) .
v (W1 _,. 0 as v + O b u t T1 does no t ) . To avoi d this use in our dispe rs ion relation
W/ v to ge t
2 2
(T 1 (q , v ) -T1 (q , O ) ) /v .
re lation)
2 ...
d (v ') 2
•
2'r
.Y...... J 0
(40 . 6a)
s een by dire c t s ub s t i tution o f ( 40 . 6 ) into (40 . 6a) , remembering that W(q 2 , -v ' ) •
-W (q 2 , v 1 ) , i . e . th a t h (o , I! ) •- h ( o , -S ) .
C ons i de r
J dv '
v ' - (v+i t )
o ci+2Ml!v ' -o) sgn (v '+MS) • -l
q 2 -o+2MSµ+iugnS
J dv '
v ' - (v+i c ) v
'
1 92 Photon-Hadron Interactions
W/ v ' is symme tric so we can set the range from 0 to wi th l / (v ' - (v+iE)) +
00
2
1/(-v ' - (v+iE ) ) -2v/ (-v ' 2+ (v+iE ) ) . From th is equation (40 . 6a) follows .
•
f ---
2
2
d (v 1---
) 2
W (q v ' )
(v ' ) - (v+iE) 2 2 •
for l < 0 •
2
For q • 0 these are of course the Kronig-Kramers formulas relating real and
imaginary parts o f the index of re fraction ( forward light scat tering) which
is necessary if signals are not to come out before they go into a block of
s cat tering material .
We try now once again (unsuccess fully as it wil l turn out ) to try to
ob tain some expected limi tations on 2MW 1 because we know it is a causal ( ze ro
outs i de the light cone in space) commutator . This time we shall use behavior
valid not only for large v but for any region . For example we know for the
elas tic s cattering if the p ro ton were a point charge we would have for v > O ,
2MW1 2Mv6 ( (p+q) 2 -M""2 )
• =
2
2Mv6 (q +2Mv ) . (To get the correct symme try for v
posi tive and negative this can be writ ten more correctly as 2MW 1 2 Mv (sgn(v+M) •
=f0
�
q 2 -µ
(40 . 7 )
for nega tive q 2 , i t can be done . We can see why this is , and also get a
better physical fee l for const ructing causal functions by the fo llowing con-
siderat ions . I t is eas iest to deal wi th the s cat tering functions T , fo r these
to be causal they must be the Fourier trans form of a retarded coumutator and
thus zero outs ide a forward light cone , Let a ( x , t) be such a function and A(q)
its F . T . and let b (x , t) be another such function , B (q) its F , T , Thus the con-
vo lution of a (x , t) and b (x , t) is obviously , by geometry , such a function (zero
outside forward ligh t cone ) and hence i ts F . T . or s imply A ( q ) B (q ) is again
satis factory (a causal scat tering function) .
We see that comb inations by multiplication (and addition of causal scatte ring
functions are causal scat tering funct ions . The s implest causal scattering
function is
[q2 - m2 + i£ sgn q 4 ] -l (40 . 8)
i
We can generalize this (multiply by e u · x in space time) to find for any four-
vector u
lJ
that
(40 .9)
is a causal function . Thus the e lastic scat tering function for p oin t p a rti cles
[ (q+p) 2 -!r2 + i£ sgn ( +M) )
-1
(40 . 10)
evidently is causal (as well as what you ge t by putting -q for q) , We can
multiply this by an express ion like ( 40 . 8) , we see
-l
2 2
[q -t11 +i£ sgn v ) [ (q+p) 2 -i/- + i£ sgn(v+M) ] -l
is causal . This is true for any m2 • µ and hence any superposition with weight
J p ( µ) dµ
Tl
•
1
(40 . ll)
q - µ+- i£ sgn v
2
v q 2+2Mv+ i£ sgn ( v+M)
is , by itself causal . To get the correct symme try for v one need only add the
corresponding expression with v replaced by -v ; we will suppose it is always
done and not write i t explicitly .
1 94 Photon-Hadron Interactions
(You might find it physically more satis factory and easier to in terpret
if the factor is considered as a form factor due to virtual particles like p ' s
at each vertex . Those at one vertex would contribute a factor
-1
w(µ) d µ
2 (40 . 12)
q -µ+ i£ sgnv
0
where µ is the mass squared of the virtual particle and w measures the weigh t
o f its contribution ; g i s causal . W e would then expect t o multiply ( 40 . 10) by
2 2
( g ( q ) ) , one factor for each coupling ; the result would still be causal ,
and possibly physically easier to unders t and . )
To get w 1 / v (commutator) from (40 . 11) w e need only take its imaginary par t
v
•f Pv + q -µ
2
p ( µ ) d µ " sgn(v-Hl) 6 (q -2Mv) +
0
f
2 1 ( 40 . 13)
+ ngnv6 (q -µ) p (µ ) dµ PV -2--
q +2Mv
0
( 40 . 14)
even if fairly exact . It is not easy to reverse the integral into accurate
knowledge of p (m) unless physical arguments ( P dominance etc . ) are availab le .
Thus again we are thrown back in this problem to unders tand the process
J0
1
can conclude , by (40 . 14) , that p (µ) dµ • O. Such a relation is called a
2 2
superconvergence relation -f (Q ) converges for large Q more rapidly than
Properties of Commutators in Momentum Space 1 95
4
the form woultl suggest . Again s ince f falls as ( l / Q2 ) we can conclude that
the moments �
µ np (µ) d11 vanish for n 0 , 1, .
2
• Alternatively , f can be
Lecture 41
The idea discussed at the end of the previous lecture is iDD11e diately
generalizable to scattering through an intermediate resonance , say of mas s Mx2 '
2
M -i/- • >. . (q2+ Mv ->. + i£ sgn (v +M) ) -1 .
2
call A point coupling would give scattering as
x
We can mul.tiply by any form factor, say with a p ( >. , 11) • We are thus representing
things by a s um of s channel resonances each of which has a square forir factor
l
f { >. , - > • J p ( i , \I ) dµ { 41 . 1)
-q + 11
+
p ( >. , µ ) d >.dp
q 2 -l!+i £ s gnv q 2+2 Mv-Hi£ sgn ( v+M)
"' "'
w1
-
v
•
jj p { >. 1 11) PV 22
q
11
-11
[ sgn {v+M) H q 2+2Mv->.) -s gn ( v-M) 6 ( q2 -2Mv ->.) ] +
0 0
For q2 < 0 the �a s� term vanishes and we have for v > 0 simply
w /v
1
•
j
11
..
f ( >. , -q
2 2
) 6 (q +2Mv->. ) d>. • f (q2+2 Mv , -q
2
) ( 41 . 4 )
0
a superposition of contributions for each e f fective mas s with a fo rm factor
2
f ( A , -q ) of form given in ( 41 . 1) .
Have we not gone in a complete circle ? Our original express ion giving
1 96 Photon-Hadron lnteractions
2
( for q > O , v > 0)
2
This looks j us t like (41 . 4) , The 6 is of course 6 (q +2Mv-A) so we interp re t
2
f ( A , -q ) as L
l< p l J (Q) l x> l 2 summed over s t ates of a given mass 2 • �+ A ,
2
(At first this would seem to make a function of Q , the space part of the
2 2 2
momentum trans fer s quare d , ins tead of q v -Q ; but it is the same because
=
2
the 6 function relates v to Q and it can be expressed either way . ) We see
2 2
f ( A , -q ) mus t be posi tive for -q > O ; of course , s ince the lowes t elas t i c
2
M
x M ( A • O) is separa ted from the con tinuum at Mx M+mn ( A th • 2 mn M+mn ) .
• a
2
The function f ( A , q ) and hence p (A , µ ) will have a 6 ( A) cont ribution and after
that the integral in (41 . 3) will s t art at an inelas tic thresho ld A to m
,
th
We may not have gone in a comp lete circle . Firs t we now know (a) that the
weigh t factor f( >., -q ) mus t be expressib le in the form (41 . 1 ) and (b ) we know
2
2
what the func tion looks like in the experimentall y unavailable region +q > 0
(see 41 . 3) and , of course , the scat tering funct ion (41 . 2 ) that goes with i t .
every causal funct ion must be exp ress ible as (41 . 2) and at the moment we do
no t think we can .
Since (41 . 2 ) is causal it mus t be express ible in the DGS form ( 39 . 6 ) . One
way to do this (suggested by Comwall , Corrigan and Nor ton , Phys . Rev . D3 5 36 ( 19 71) )
l
'"'l'---
2 • 2 -
2
2 q _,, q +2Mv-,
[ q +2Mva- AS- µ ( 1-B ) l � A
parts to prove
h (a , B) • ��
m m
0 0
Of course this can now be s impli fied . I f from h (a , B ) we could always find a
Properties of Commutators in Moment11111 Space 1 97
p ( >. , µ ) which would give this h we would have a proof (assuming the DGS rep re-
sentation is proved) that (41 . 2 ) is also a necessary form . We suspect it
cannot be done and i t is possible that (41 . 2 ) , although very physical , is not
the complete exp ression ; but o ther forms (o ther types of diagrams - or "channels "
o ther than s channels) might have to be added to it to ge t a complete repre-
sentation . This is a good problem .
<
2 2
W for all q in terms o f W for q O?
We have a form that suggests the quest ion a s to whether knowledge o f the
2
fact tha t W is causal , and knowledge of i ts value for negative q only, enables
us to find it for all q , v .
2
We are not now concerned with the practical fact
2
that knowledge of f ( >. , -q ) to a given limited experimen tal accuracy does no t
permi t discovering p ( .l. , µ ) very well and does not mathematically alone define
f ( >. , -q ) for q > O . Rather we suppose w1 perfectly known for q < 0 and ask
2 2 2
f
self energies , which can be defined in terms o f T as integrals perhaps (e . g .
2 2
T ( q ) d4 q/q ) . If T were un i que 1y de termine d b y W in the experimental 1 y
accessible region we might search for express ions for these integrals directly
in terms of this W at q < O . (Cottingham formula for self-energy . )
2
2 2
Given W(q , v ) in the availab le region (momentum like q ) how unique is
W(q , v) in the energy-like region ? Let W (q 2 , v ) and W (q 2 , v ) be two causal
2
a b
functions identical in the q < 0 region . Le t us study what is possible for
2
wd -
J J..
4
d u
0
ds
2
sgn (q o -uo ) o ( (q-u) -a
2 2 2
) t( u , s ) (41 . 6 )
4>
2 2
where i s non-zero ins ide the region where the hyperboloid (q-u) ..s does not
penetrate the region S of q space where we know Wd vanishes . Thus S is the
1 98 Photon-Hadron Interactions
< 5 u •
2
entire q 0 region . It is seen that every hyperboloid cuts unless O.
Thus the mos t general form fo r W is
d
m
1 •
2 2 2 2 2
d
- sgnv6 (q -s ) ' (s ) ds sgnv � (q ) ( 41 . 7)
0
where by de finition Hx) • 0 for x < O.
2
Thus complete knowledge o f w 1 (q , v ) in the experimentally available
q < 0 region permits definition o f w1 everywhere within an arbitrary constant ,
2
2 2
( independent o f v ) function of q for positive q • T is de termined also up
to the function
(41 .8)
Strictly this argument is not valid , gradients o f 6 func tions in space time
2
could be used in Dyson ' s theorem. More correctly , at each posi tive q the
function W is de termined by the behavior of W(q , v ) for q < 0 up to an unknown
2 2
2
polynomial in v , the coe fficients of which are arbitrary functions of q •
Physical arguments about large v asymptotic behavior would have to be used to
limit the degree of these polynomials . The s caling limit for
0 and the x <
fact that W is odd tells us that (41 . 7) must be of the form v' (q ) ( for v > 0)
2
Lecture 42
Elect romagnetic Self Energy
We now discuss s few places where knowledge of T would help us calculate the
electrodynamic p roperties of pro tons and neutrons . Since we have measured , in W ,
the electromagnetic coupling o f p ro tons w e migh t hope t o use the experimental
knowledge to determine the electromagnetic energy of pro ton and neutron and
their measured difference , to compare to experiment . As we shall see the hope
2
is , at presen t , frus trated be cause knowledge of W for q < 0 where it is avail-
able is not quite enough to determine the electromagnetic coupling everywhere
(T) - the arbi trary func tion of �(s) men tioned in the p revious lecture is not
de termine d . Since the answer for the p-n mass diffe rence is only one number we
are frustrated until we can find a theoretical or experimen tal way to determine
2
T unique ly - for example to determine t 1 (q ,0) of the dispersion relation (40 . 6a)
2
for q < O .
Be fore we dis cuss this by formal mathema tics let us see what we can expect .
As is well known , the self mass of a point spin 1 /2 particle diverges logarith-
mi cally . For the electron t. (m2 ) m z-;;-
2 Je 2 A2
• tn 2 where A is some upper cut off
m
for electrodynami cs if we replace the photon propagator
:..;, �q 1 99
200 Photon-Hadron Interactions
This Am2 is experimentally undefinab le . A s imilar infinity for the AM2 electro-
magnetic for the proton uould also - by itself not be observab le , b ut
• - 1 . 2 9 34 MeV
one could always say anything could happen - but now that we have some knowledge
of them we mus t answer more speci fi cally .
The divergence o ccurs from high frequencies and at first it was thought the
hadrons mi ght be soft at high frequency and the electromagne tic self energy con-
vergent . But now we know for inelas tic s cattering a t leas t they look like they
are made of point-like cona ti tuents . Does this point-like behavior mean the
energy mus t diverge ? Of course , the pro ton could divex ge and the neutron also-
only the di f ference need converge - but the difference in point-like s truc ture is
w 1p -Wln is also finite and point-like in the s caling limi t . lt is the coincidence
of the 6 -func tion of the photon prepagator and of the electron propagator which
makes the divergent self energy - and now we see the pro ton , the neutron , and
proton minus neutron all have s ingular behavior on the light cone so i t at
firs t looks like divergence is inevi tab le ,
Le t us es t imate how much . Since we are dis cuss ing the high energy behavior
we can use the i dea that proton•partons . Clearly the self energy diagram of
mos t importance at high energy in the s caling limit are when a photon is emitted
and absorbed by the same parton - therefore , as far as the divergent (tn A 2 )
part is con ce rned it is as i f each parton gets a shift in mass proportional to
Am 2 • m 2 e 2 tnA 2 /m 2 where e , mi are the charge and mass o f a parton . How
i i i i i
much does this change the mass of the p ro ton? We cannot honestly say . One
might try to calculate the change in E - P z � / 2 P by calculating the sum o f
•
2 2
the ch anges in £ -P • 9 � of each parton . We would find
2
There are obj e ctions to this (by the way it is also even mo re divergen t
since f "' l/ x and the x integral cannot be carried to 0) . The energy is not
ep
Electromagnetic Self Energy 201
j us t the sum of the kine tic ene rgies of the partons , interaction energies among
the partons are also invo l ved . Thi s is reflected in the dangerous formula con-
taining the mas s squared of a parton - a thing we said was meaningles s to
t.Aup to now . The dis t rib ut ion of partons is changed in f i rs t order also so
2
the total m change is not co rrec tly evaluated . ( I t might be though t we are
right to take < ljJ I AV l 11» for the per t urba t i on on the Hamil tonian us ing the old
AV • Am a•a is
wave funct ion - b u t
2 not the change in the Hamiltonian , only
of the Lagrangian - in the Hamiltonian there are many o ther e f f e c t s on the
interac t ion terms through 1 / /2; factors e t c . , so we have not computed the e f fect
o f the perturbation co rre ctly . )
It may well be that m 2 O , or is effectively z ero (a suggestion to me
•
i
due to Zachariasen) due to in terac tions (or as a matter o f principle) so the
2
logarithmically divergent perturbations A i depending on inA neve r arise . This
may be so , but we do not know. We mus t turn to more detailed q uan tit at i ve
analyses if we are to try to study this further .
Electromagnetic mass shifts come from the emission an d reabsorption o f
a virtual pho ton .
J�
of the photon . Hence by taking Fourier trans form
2 2 F 2
AM • 4ire
(211)
2 _
_l
q +i£ \,µ (q , v ) (42 . l )
We have written
so that
(42 . 2)
and suppose the imag inary part W • w1+[l-v 2 /q 2 )w2 -w1J/4 scales to f (x) in the
Bj o rken limi t .
Lecture 43
C o t t in gham formula
(43 . l)
where 4T • T
\l \l
W e mus t integrate this over a l l q2 bu t w e have seen T is
de te rmined by i ts behavior f o r q2 < 0 (where experiment can s a y some thing
ab out i t) so it is pos s ib le that maybe (43 . 1 ) can be wri t ten in terms of T for
that in the four-dimens ional integral the conto ur on v could be changed (withou t
iw ,
from the real line v • ......, to
m to the imaginary
n
axis v • 111 • ....... to � . (We show how i t i s done later . ) Suppos e i t i s t rue .
Now
(43 . 2 )
( Co t tingham formul a)
Now the quan t i t y is all righ t for -q 2 b u t is comp le tely unphys ical for v is
· J"
0
dC -\ {!.
-q
>
4
C -<i2>rc -l ,o> -!.]"(�=\+� - )
2
0
v'
1 1
� W(q2 , v ' ) dv ' ;t
2 v '.2
(43 . 5) J
This then succeeds in get ting an explicit formula for the self energy in
terms of in a region access ible to experiment . However , we are frus trated
2
W ( q , v)
2
by the unknown term T ( q , 0) .
to determine whether the self energy diverges . Le t us look first at the contri -
bution from the second term in
could come . Put q - 2
•
l�
..
vdv �� 2
+ 1 -1 -} (� W(x, v)
0
J i
.. 1
�v ; xf (x) dx
f
1
-q A ..q
This provides a cut-off for our v integral (of order t / 2 Mx) so the part
2
diverging as ln t. has a coefficient xf (x) dx . �
the ln A 2 term i f it
J f
However , other term in could also produce
2
T ( q , 0) a
2
only falls as C / ( -q 2 ) as -q 2 "• (If · + ( s ) ds/ (q 2 -. ) this C is • < s) ds . )
2
+ T(q ,0)
It is therefore possible that these divergences from T and W cancel and that the
self energy (or at least the proton -neutron difference) is finite and calculable .
There are two views we could take at presen t . We know of course the p -n
mass difference converges so let us talk about T and W for the p-n difference at
least . In principle T could be determined by experiment and is therefore defined
204 Photon-Hadron Interactions
(a) Equation (43 .5) with this T s t il l gives a logari thmi c divergence .
The reason is that our theories are wrong for high energy ; this and the
(b) The T is s uch tha t the integrals converge and agree with the expe ri-
that it is to some exten t arb i t rary - hence that our theory is not ab le
to calculate this mass difference precisely and mus t be " renormalized . "
cons tan t , corresponding to the elec tromagne tic mass dif ference of u
and d quarks , would suf fice to make all the hadron self energy dif ferences
As Z achari asen has sugges ted to me , the b e s t (mos t limit ing) thing to
le arn that ( a) mus t be so . Zachariasen has shown tha t all will be convergent
I believe thi s is a very good p roblem to work on . I myse l f haven ' t found
enough time wh ile preparing these notes to analyze it in a more e lemen tary or
It is in teres t ing tha t T (O , O) can be obt ained theore tically - the fo rward
Compton s catt2ring from a p ro ton off a real (on shell) pho ton . For Q + O , v + 0
we have very long wave length slow f ie lds to which , of course , the proton looks
1f •
e2
T ( O ,O) • -
F
Use DGS representation for T
n( a , B )
I
v 2 --0_2+2Mv B -a+ i£
Call E • ../ cr+i-f s+Q2 and no te the s ingularities are at v • Q -i£ , -(J.+i£
S ince E > SM, the poles below the axis are all for v
y----- - > O. The contour goes like
the dot ted line , it can evidently be ro tated to the imaginary axis .
Le cture 44
Express ion for self ene rgy in terms of W only
q
2
only , wi thout at the same time involving ourse lves wi th ano the r unknown
2
function T ( q , O ) ; and furthe r each of two parts is infinite and it is hard to
guess at the di fference . Perhaps we should abandon th is and take a las t look
at j us t the 6M2 exp ressed in terms of W { q v ) for pos i tive and ne gative q2 •
2
, It
is given by
W ( v , Q ) dQdv (44 . 1)
206 Photon-Hadron Interactions
F
T µv
. (J
(This is obtained by expressing TF in the form
K
IJV
v+v ' - i
dv '
_
2w
+
and ins erting into (42 , 1) . W ( v , Q ) depends only on Q , the magni tude of Q; and
we have d 3 Q • 4wQ 2 dQ) .
To study the possible divergence of the integral at leas t , we go to large
2
v and in fact to the s caling region -q /2Mv x, hence Q v+Mx . We can write • s
t.M2 •
2
dv::
1
JJ
m
0 -v /M
For large v the uppe r limit for x is kinematically 1 . 2MW approaches the
2
function f (x) for positive x in the s caling limit . Fo r negative x (posi tive q )
it approaches as we have seen -f (-x) ; note , however, that -x is not kinematically
limited to 1 .
Dynamical ly -x i s limi ted , W ( x , v) exists for x < -1 but for large v falls
J
rapidly until there is nothing left of o rder one .
m +l
• o.
I
The contribution of the scaling region gives dv f (x) dx
-1
But this only says that a divergence higher than logarithmic vanishes , a thing
Expanding the factor (v+tlx) I (2v+Mx) seems to give a term
� xf (x) ( dv /v) dx ;
we expect anyway .
Mx/2v leading to this is not all , we shall have to know to
2
order l /v how W differs from i ts Bjorken limi t for both positive and negative q •
Th is sums up the p roblem ; experiment can in p rinciple help us with the
2
approach to the limi t for q negative . But we shall have to rely on theory to
2
obtain the con trib utions to order l/v for positive q before we can decide
whether !!.� diverges according to present theories .
quarks , two u quarks and a d quark wi th total spin 1/2 wi th wave function
I P> • __l (2 uud - udu - duu) ( t H ) symme tri z ed •
16
(44 . 3)
pro ton this contributes AM /2Mp • 9a (we shall normali z e all mass squared changes
2
2
to 2Mp of the proton as a s cale for measuring a , the t rue change in M is then
2Mp a . )
b) An interaction ene rgy be tween pairs whi ch we take as p roportional
to the produc t of ch arges . The interaction must depend on the mutual spin
relation of the p air . Thus write S ( l+y ) i f spin is parallel and S ( l-y) if
spin is antiparallel ; this is S ( l+yP) where P i s the spin exchange operator .
We mul tiply by -2 for ud or us , +4 for uu and +l for dd , ds , s d , ss pairs ,
call this factor xi "
j
The electromagne tic sel f mass operator can there fore be wri tten
2
AM • 2 Mp { 4 a ( No . of u ' s ) + a (No . of d ' s ) + a (No . of s ' s ) +
I t is easy t o get the expe ctation of this operator for every s tate . Thus
on the proton the operator Bx alone does no t see the spins and gives (2 (4-2 -2 ) uud -
( 4 -2 -2 ) udu - (4 -2 -2 ) duu ) t t + / /6 which is zero . For the neutron , change a u to d ,
the coe fficient i s - 3 s o there i s a contribution - 3S . Next we s t udy SyxP
Lecture 45
Electromagne tic self mass , quark mo del (continued)
+
We can calculate the EM self masses for every particle of the 1/2 octet
as wss done in the previous lecture for the proton , the result is :
p • 9a + 6By
n • 6a - JS + JBy
+I: • 9a + 6 By
i:0 • 6a - JS - J By
I: • Ja + JS
:;: • Ja + JB
:; 0 • 6a - JS + JBy
A0 • 6a - JS + } By
Hence
tiM ti.il- / 2Mp
exp
.
p -n 3a + 3 5 + 3 By - 1 . 29 MeV - 1 . 21 MeV
-= :!: :!:
�o - .
3a - 6 B + lBy - 6.6 .7 - 6.7 .9
i: +- i: o 3a + 38 + ·i; By
=
- 3 . 06 - 3.64
oI: -I: - 3a - 6s - 23 �r
.
- 4 . 86 :!: .07 - 5 . 78
ci: +-i: - )-(p-n)-e 0.. :-) o • 0 :!: .7 + .s :!: .9
We have three constants for four maa s di fferences so we have the relation
(an su3 relation)
{I:+-I: -) - (p-n) - e 0- = ->· 0
which fits well . (We choose to use ti.il- for no extremely good reason , the relation
fits better with l!.M but there is li ttle to choose , i t is inside the experimental
error for ti.il- also . )
parti cles with fewer u quarks are heavier , i . e . more pos i tively charged b aryons
are lighter . The sign of the 8 term from electrostatic repulsion is pos itive
as expected . We find at traction of paral lel magne ts in an s s tate but repuls ion
of antiparalle l ( the magnets are on top of each other) which is correc t ; the
net repuls ion in the parallel case is less .
Con tinuing in the 5 6 su6 multiple t to the decime t and suppos ing the cons tan ts
are the same we can predict eve rything:
8i /2Mp (predicted)
•
++
8 12a + 12 8 (1-y) - 5 . 12 MeV
8+ = I: + . 9a - 6 . 00
• ;
80 - E o o 6 a - 38 (l+y) - 4 . 72
8- - I: n
- -
. .: 3a + 38 (l+y ) - 1.28
2
8M /2Mp (pred . ) 8M(pred . ) 8M(exp . )
:!:
° ++
8 -8 + 0.4 + 0.3 2 .9 .9
:!:
++
8 --8 + 3.8 + 2.9 7 .9 6.8
- 0.9 - o.7
++
8 +- 8
i: - - E + + 4.7 + 3.2 3.3 :!: 1 .5
- - - _::: o + 3. 4 + 2.2 4.9 :!: 2. . 0
We have two constants to fit with two parameters and have no prediction .
•
•
We ge t S ' . 14 , 3a -2 . 38 again confirming that a is negative . In fact a is
close to i ts value for the baryons wh ich is what FKR'e relativistic quark model
210 Photon-Hadron Interactions
would expect . Furthermore the 8 ' • 8 ( 1-y ) is nearly of the same order of
magni tude .
Finally we do the vec tor mesons . Here we need 8 ( 1 -y ) • b for paralle l spins
p •
+ S a + 2b
0
p Sa - 1_
2 b
s
(j)0 Sa - I b
*
K +• Sa + 2b
K*o• 2a - b
* *
The only data on 6M is that for K + - K o whi ch is -S . 7 ± 1 . 7 MeV o r
•
2
6m /2Mp -S . l ± l . S • 3a + 3b . I f 3a z -2 . 38 this gives 8 m - . 9 ± S , which is
very bad , it is o f the wrong s i gn . If the values for the octet baryons are
used we predict 6m2 /2M • ± 1 . 9 !
p
In this system we also have two o ther e f fects (discussed in Lecture lS)
a) the electromagne tic mixing matrix be tween p0 and w0 , wi th off diagonal term
p0 w0
)
3a - 3b /2 . b) The annihilation term between and which we found to be
0
p
0 �Sl
•
(j)0
6m MeV .
.17
J
6m - if /2
p
.. _'., .
o ++
l - A is 0.8 so only l i t t le is gaine d here . But the me s on s i tua tion i s
a l t e re d (be caus e o f the small w mas s ) . The cons tants come o ut fo r t h e pseudo-
ba ryons may also differ from 8' for pseudo s c alar me s ons be cause the s i ze of
the wave funct ion is so d i f fe rent s o the mean l/r d i f f e rs . The e l e c t r i c and
2
why S ' is so di f fe rent as the Am gives .
Why the value of Ja should d i f fe r in one case f rom the o ther is not
2
clear . Were it no t for expe riment l would dec ide 3a calcul a te d for Am / 2M is
p
the same f o r pseudos calar and pseudovec t o r mesons and is 0 . 6 o f that for
in ba ryons .
Eviden tly the naive theory does not work well , we do no t unders t an d
Al • 2 ma s s d i f f e rences
There is one observation that can b e made here whi ch s ugges ts tha t dynamic
calculations might be possible fo r some comb inat ions . No tice that a is the
self energy term, po s s ib ly involving high frequency behavior , b u t S (lty) is due
to mutual interac t i on and ought to show no divergence . Cert ain comb inations
*
i (A + A- - A
o
-
+
A )
2 1: 0
+
i: + i:- -
energy . Thi s ene r gy depends on th e p ro duc ts o f two curren t ope ra tors JJ each
2 1
In terms of lz these go as con s t an t , Iz , and Iz - 3 l (I+l ) respec tive ly .
t h e 61 • l 1 to
z
or terms are measured by differences p ropo r t i ona l I like
z'
2
l: - E the Lil 2 term 1 and is
-
But to z
+
or p-n . R e f fect is p ro po r tion al
electromagnetic J we re lll +1 Li l •
if cur ren t z • (ins tead of 6 I
z
• 0) , 1. We
in calcula ting this might converge rapidly at high v, provided partons are
quarks . For if all the current carrying fundamental operators (psrtons ) have
succession at h igh energy . Thus the virtual photon exchange diagrams like A
con tribute only to Al • 1 and are imposs ible for Lil • 2 , only B are allowed for
1\1 - 2
s t rong
interac tion
- parton
Baryon
the s t rong interactions are - i f the ini tial momen tum dis t r ibu tion of the
partons involves only s low ones so can the final s tate . But i n B, i f the
in te ract ion can pull the partons back together to give much o f a di ago nal
Lecture 46
7·
��
11
2
factor m / (q2 -m 2) at each pho t on co up ling so the mass di fference is
p p
m
4
(46 .1)
j
To o rde r zero
d q 1
4
2
4
p
(Am ) 0 • 4 2
m
2
-411e 3 --
(211 ) q
2
m 2 m
+ --\- .l.n �
m
(46 .2)
m
p 1T
2
In terms o f Am • Am /2m this i s 11 + - 11 ° • 4 . 1 MeV as comp a re d t o the experi-
resonances give les s than 2 0% , and the result is in good agreement with the
dat a .
baryons were calculated by Gross and Pagels , Phys . Rev . fil 1381 (1968) us ing
su 3for the magnetic momen ts , and GE and � varying like (l + q2 / . 71) -2 •
They ge t
t.M (elast ic) t.M(exp . )
p -n + 0 . 79 MeV - 1 .29 MeV
E+ - t 0 + 0 . 16 - 3 . 06
Eo - E - - 0 . 88 - 4 . 86
::: 0 - :: - 1 . 10 - 6 . 5 :t . 7
(Most o f this i s due t o the charge , since the re are factors o f q from Yµ4-ilYµ
in the magne tic part . ) Note that E + + E - - 2E 0 • +1 .54 MeV (elas tic) compared
to +1 . 8 experimental , as we expec ted this s hould be dominated by the elas t i c
term . (Decimet intermediate states give less than .1 and presumably higher
ones even less . )
Note that o the r combinations (o ther than t.I • 2) do not involve "a"
( see lectures 44 and 45 , "a" i s the EM self energy o f a quark) such as (p-n)
- ( !! 0 - :=: - ) • In this case one canno t show that these differences also involve
a product of currents s uch that each current does no t act on the same quark .
But the fact that "a" is no t involved suggests that we may be able to ge t at
this difference also by estimating ma trix elements to various known s tates
us ing , for example , the quark model . Howeve r , the e lastic term fails utterly
this time t.m(elastic) • +1 . 9 , t.Mexp . • 4 . 2 :t .7. Why? To explain this we
look at the contributions from the s caling region . Let u (x) , u (x) , d (x) , d (x) ,
-
vW Ix f (x) for p , n , :=: o and
2
s (x) , s (x) be as des c ribed in lecture 31 , then •
is
-
•
4
9 (u + u) + 91 (d + -d) + 1
9 (s + s-)
'+-
•
1
9 ( u +
-
u) + 94 (d + -d) 1 (s
9 + s- )
..
_o
- 9 u) + 94 a)
l l (s s- )
fe :: (u + (d + + 9 +
-
•
l
9 (u + u) + 91 (d
- + 94
+ d) (s
-
+ s) (46 . 3)
(The neutron is ob tained from the pro ton by replacing u by d and vice v.ersa .
The :=: 0 is like the neutron but s replaces u and vi ce vers a . The :=: i s like the
:0 but u replaces d and vice versa . ) Therefore the s caling function for (p-n) -
Electromagnetic Self Energy 215
0 - l - - -
( : - : ) is 3 ( u + u + s + s - 2d - 2d) which is no t necessa rily zero so
high frequencies can come in . In the model of valence quarks plus sea it is
Le cture 4 7
We now go on to dis cuss o ther e f fects involving two pho ton couplings .
s cat te ring is exac tly in the forward direct ion the s ca t tering ampl i tude is given
o.
2 2
)J V
by T ( q , v ) fo r q • We p revio usly mean t the average ove r pro ton s p ins ,
thus T is the spin averaged forward s c a t terin g , we could also meas ure for
special spin direct ions of the p ro ton . The imaginary part of the forward
!!2. • e
At
dt
-2
For energies from 2 to 7 GeV we ge t around 6 Gev • For energies from 8 t o
-2 2
16 GeV , A • 8 Gev is clos e r ; the re i s some s ign o f a quadratic te rm At + B t
wp at 9 GeV has A • 9, B • 2.5.) Thus pho ton di f fraction looks very much the
s ame as wo ul d be expected for hadrons except for the very much smaller cross
We now dis cuss the forward s c a t tering in mo re detail including spin e f fe cts .
(47 . 1)
as a spin ma t rix operating be tween spin s t ates of the p ro ton in the lab , system ,
The var io us meas ured quan t i t ies are expressed in terms of f and f as fol lows :
1 2
216 Photon-Hadron Interactions
The total cros s section is the imaginary part of the diagonal (in spin)
scattering
Im f 1 (v) v to t
'\N ( 47 . 2)
4 TT
This is known . The forward differential cross sect ion for unpolarized forward
scattering is
1
( 4 7 . 3)
16TT
The real part of f 1 can be ob tained from the imaginary part by a dispersion
relation , (Eq . (40 . 6a ) fo r q2 • O) where we use the fact that f 1 (o) -e 2 /M . �
e2
Re f 1 (v) ( 47 . 4 )
M
Th is has been evaluated (see Damashek and Gilman, Phy s . Rev . Dl 131 9 ( 19 70 )
or Buschhorn e t al . Phys . Le tt . 33B 241 ( 19 70) and the sum of the firs t two
terms of ( 4 7 . 3) is compared to the experimental (do /dt) to see how b ig the
0
las t term is . They agree wi thin errors for v from 2 . 5 to 17 GeV so the n l f j 2 /v
2
contribution is less than 10% over the entire range . (The con tribution is
greates t from the first term in (47 . 3) above about 5 GeV . The se cond is 15%
at 2 GeV and falls away at higher v) .
e2 �
O , f 2 (v) • � µ 2 v where µ is the
We also know , for small v , as v -
2M A A
anomalous part of the magnetic moment of the nucleon in nuclear magnetons .
At finite angles the asymme try parameter has been meas ured
E • --- (47 . 5 )
OJ_ Oii
+
where o1 and 011 are differential cross sections at fixed t for incoming
photons polari zed perpendicular or parallel to the plane of collis ion , For
t • 0 E mus t be zero , of course ; but wi thin the limits of experimental error
expects the ssme result . The q uestion is : With incident light in the z direction
with x polarization do more pho tons scatter at a small angle e � Q (transverse) /v
in the direction x or in the direction y ? From the poin t of view o f diffraction ,
currents generated by the incident wave must be adequate to produce the correct
forward scat tered wave to interfere with the incident wave to accoun t for the
loss of intens ity of this wave represented by the total cross section . These
currents are obviously limited to the spatial region of the pro ton , and so they
produce scattered waves in other directions , the usual eAt of diffraction from
the pro ton , j us t as in hadronic collisions . B ut these currents mus t make pure
x polarization at leas t in the forward direction , they are x directed currents .
In o ther direct ions at small angles we have the same intensi ty for x and y
deflection excep t for a COB 0 l ab "" 1 - 0.2 lab projection fo r X de flection ,
.
therefore i: ( 1 - cos > I 8iab . > e � ab _ 1 2 :::o -t/ 2 } "" + .03 for
= ( 1 + cos
l\ab . =
in the same energy region , unknown ones may be guessed from the quark model) ,
There is also a computable neutral pion exchange term
p x
The coupling o f two photons to a neutral pion is
p
known from the w0 + 2y decay .
All these calculations can be checked and controlled by fit ting the calculated
imaginary part o f the scattering to the nicely meas ured to tal cross section
o yN which shows the expected resonance bumps in this region o f energy .
tot
particle with no internal excited s ta tes . Such a term is cal led a Born term.
Thus we wri te T v (not averaged over spin directions o f the p ro ton ) as a S IDll
µ
from the Born term and the res t f rom o ther diagrams
T (47 . 6 )
µv
µ µ µ
vation) ; so v < x l p l p> • Q .< x l l Co) l p> , hence if v , Q go to zero together we
see that j matrix elements go to zero . A mo re rigorous (but harder to interpre t)
argument is given below .
We now compute the limit of TB as v , Q + O. If
A
µ is the anomalous
µA
moment in nuclear magnetons , the coupling of a photon is Y µ + 'iii (y 4 - 4Y µ ) so
µ
T v
B • (y +
µA
(y 4 - id.y ) ) ; it
II
+ +
� (y
v
+
µA
(y 11 4 - id.Y11 ) +
µ µ 2M µ 2p • q + q 2M
(47 . 7)
2
_ L
M
R
while the contribution from T s tarts as v 2 and thus we have as v + 0
£ 1 (o) 2
• - e /M
2 2 2
f 2 (v ) /v • f i (o) - e µ /211
• (47 . 8)
A
R
To show how T is small more formally (at least for the f 1 term) note
B
tha t T µ v total as well as T each separately sati s fy the gauge condi tion q µ T µ v O , a
R
h ence we mus t h ave q µ T µ R • O Now we can wri te T µv ( fo r the symmetric spin
•
v
averaged case at least) as a power series Tµ v ap µ p v + b (p µ q v + p v q µ ) +
a
2
c 8 IJV + .order q • There mus t be no po les like l/p • q in a , b , c (wilike T )JV
total
)J
B R
which has such poles coming from T ) . Now qµ T 0 requires •
µv
2 3
(p • q) P µ a + (p . q ) q µ b + q P µ b + qµ c + order q 0 •
discussion see Low , Phys . Rev . 96 1428 ( 1 954) and Gell-Mann , Phys . Rev. 2! 14 33
(1954) . )
Forward Compton scattering from non-relativis t4.c Schroedinger equation
The equation is (with first order relativistic corrections )
Hijl •
[ l {p - eA) • {p - eA)
-
+ -t + -t
- --..,.
1
.. (p+ • + e {l + µ + • ±
p)2 - - A) 0 11 +
2M BM"' 2M
+ +
The incident amplitude is -tA • e i , ±� • ive+ i and ±11 • i +k x e i . In the laboratory
p is zero , hence the leading term comes from the A·A term and is - (e 2 / 2M):i . :f
contributing to £ 1 • Next we have the term t . 8 operating in second order , two
diagrama with energy denominators -v and +v
220 Photon-Hadron Interactions
2 2
which combines with the previous term to change the (1 + µ )
A
to µ
A
we
therefore have
2
•
e .,.
Amp . ei
2M
Other Two - Current Effects
Lec ture 4 8
depending on whe ther the spins of the elec tron and p ro ton are paral lel o r anti-
Re lat ivis tically we can wri te ( Rm • Rydberg , µ ' µ magne tic momen t s of p and
p e
e, µ • Bohr magne ton)
0
2
+m
M')
-3 3
U + y a > t ""R -P ( 48 . 1)
-3
The factor ( 1 + m/M) comes f rom reduced mass co rrec tions to the Schroedinger
2 2
equation in ge tting j � (o) j ; 1 + 3a /2 is a modi f icat ion due to the Dirac
equation . The other factors t, -� 1' are all near one and are due to higher
o rder quantum electrodynamic corrections . They have been separated into three
factors for convenience of discuas ion . f collll!S from QED mo difica tions of the
221
222 Photon-Hadron Interactions
A B
� comes from the proton recoil diagrams of type B in which the fact that the
pro ton is not a point charge is included by using measured form fac tors . 'f' comes
from two pho ton exchange terms of the form C , D .
All these factors except 'P which depends on as yet unknown properties
of the proton have been calculated to very high accuracy - further 6 E is
measured to absurdly high accuracy . The cons tants like a, µp etc , are now
known well enough to determine 'f' to about four parts per million . Theo
retically , the deviation due to 1" is about this same order of magnitude .
As long as ';" remains uncertain we cannot use these accurate measurements and
calculations to improve our knowledge of the cons tants - or to put the prob lem
the other way if more accurate values of the other cons tants become available
that will tell us"P, an electromagnetic property of the pro ton - and we would
be challenged to calculate it , write -P • 1 + 6 . Let us see what i s involve d .
Obviously the proton coupling is to two currents . We separate the cases
that the photons are of low energy and momentum from the cases where they are
high . Where they are low , b inding of the electron in initial and intermediate
s tates etc . mus t be considered - but here for low v, Q, the proton acts , as we
have seen , like a particle of charge and magnetic moment and thus we can do
this part of the calculation . For definiteness we do it putting in the experi -
mental form factors once for each photon and integrate over all moments , call
this 61 • Write 6 • 6 1 + 6 2 • Now for high virtual momentum electron binding ,
even the electron mass can be neglected and we can imagine the electron and
p roton to be free and at rest before and after the scattering - thus our two
Other Two-Current Effects 223
current operator < p I J µJ \) I p> for proton in and out of same momentum (res t ) is
all that is involve d . Naturally s tates x other than pure proton Born states
are involved . Of course we have already counted some high momentum contri-
butions in 6 1 and we shall have to sub tract them, thus we shall have not j us t
elas tic is calculated from the Born
T but rather T - Tµvelastic where T
w w w
term alone with form factors . Call this T µv ' • T µv - Tµvelas tic . Then 6 2 is
proportional· to diagrams c , D with electron at res t initially and finally .
If Tµv ' is a y matrix on the proton spin it depends on the spin flip amplitude
for the proton - thus on the antisymme trical part of T v ' (proton at res t ) .
µ
"-f �)
62 2 T r (C µ v
:h=t:
f (l+y t ) y zy 5 ) T r C\ v ' f (l+y t ) y z y 5 ) (48 . 2 )
)
2 2
The factor l / (q ) is for the two pho ton propagators . The first trace
is the Compton scattering by the electron
c
• y \)
l l
µv yµ + yµ
� + 4- m � - 4
neglecting the momentum p and mass o f the electron . The second factor is the
coupling of two photons to the pro ton , which we do not know b ut which we are
discuss ing ; we need its antisymmetric part . In Lecture 33 we wro te its
imaginary part wµv in the form
2
where G 1 and c 2 are functions of q and v defining the imaginary part . Let
the complete scattering functions of which c and c 2 are the imaginary part
1
be called s 1 and s 2 • That is , we write a form for T j us t like the above
µv
for W excep t s , s 2 replace G , G 2; and c , G are Im s , Im s • Then by
µv 1 1 1 2 1 2
subs titution we can express 6 2 directly in terms of s , s • One gets ( see
1 2
Drell and Sullivan , Phys . Rev . 154 1477 (1 9 6 7 ) and C .N . Iddings , Phys . Rev .
1388 446 (1 9 65 ) ) •
... 2 .. J d4
� )
� 3 lc 2 q2
�
- }) sl cl , v > 3
+ �
2
q } s2 cl , vj�
224 Photon-Hadron Interactions
certainty can be estab lished since , for example , 2g , which is the dif ference
of up and down spin partons h+ (x) - h_ (x) (see Lec ture 33) cannot exceed the
sum h+ (x) + h (x) which is f (x) and is measured . Generally posi tivity of the
_
trace of Im Tµ v on any diagonal s tate limi t the size of G 1 , c2 in terms of w1 ,
w2 in some way . This problem should be pursued
And if the answer to the first ques tion is yes and there are unknown
functions brought in in the fixed q2 dispersion relations ? Then Cot tingham 1 s
scheme does not work and we shall have to use other methods o f analysis such
as the fixed Q dispersion relations to ob tain expressions on which we can apply
our partial physical understanding of pho ton hadron interactions - partons ,
scaling, quark mode l , etc . , - in conj unction with as many sum rules e tc . , that
we know (as wel l as possibly even the numerical value of the n-p mass differences)
to guide us as much as possible to calculate this quantity 6 2 as accurately as
possible , and with an honest estimate of the possible theoretical limits of
uncertainty .
Other Two-Current Effects 225
Lecture 4 9
Other two-current effects
I should j us t like to add a few miscellaneous remarks o f situations in
which the double operation of two currents is involved, namely , in the dis-
integration of pseudoscalar mesons . They all present interesting questions
for study ; this is mean t merely as an introduction . Of course matrix elements
of one J are involved in s ingle photon decays like "' ... 11y and we have already
µ
discussed them .
Two J ' s are obviously involved in two pho ton decays like 11° -> 2 y or
n° ... 2y . An honest calculation of either one o f these would be very interes ting
(one in which the validity o f the assump tions is backed up by considerably more
than the mere fact that the answer agrees with this one experiment) . What does
su 3 say? •
...!. (uii - dd ) , n° • ...!. (uii + d d - 2 sii) to get
Use the quark model 11°
12 16
the ratio o f amplitudes amp 11°/amp n° - ...!. ci - !-) /...!. ci + !. - 2!.) - 13. I t says
12 9 9 16 9 9 9
that the amplitude for 11 is /3 that for n , so the rate should be three times as
high . Experimen tally the widths are r ( 11° ... 2 y) • 7.2 t 1 . 0 eV , and the partial
width n ... 2y is r (n ... 2y) • 1.0 t .3 KeV . The ratio is 1/140 ins tead of l!
O f course , the reason for the abject failure is the very large mass
difference of w and n (mII /m � • 1/15 ) and we mus t be much more careful of
2
n
More sens ibly we should write M in its simplest relativis tic form which
for a pseudoscalar meson di sintegrating into 2y ' s of polarization e 1 , e ,
2
momenta k 1 , k2 is M a E vap e1µ e v k
•
µ 2 10 �9 •
Now the guess would be that a is determined by su 3 • This means M goes
mll 3
as Q2 or m2 Rate / Rate • 2 , a 2 /a 2 . _l_ This at leas t moves in
II n
.•.
n m 140
n
226 Photon-Hadron Interactions
2 2
11 11
a /a - 0.4 ins tead of 3 0 . •
This is not bad but (a) what of the remaining discrepancy and (b) why
2
should a be given by su
3 and not say a/m or a/m ? Eviden tly we have b een doing
too much comparison to experiment and too lit tle thinking . Can we reason out
from other things we know j ust how the 11 rate and 11 rate should be compared , or
With regard t o the mechanism o f the decay ever since it was s ugges ted
by Gell-Mann , Sharp and Wagner , Phys . Rev . Le tters ! 261 (1962) it is supposed
y 11
The p 11 11 coupling being known one can compare this wi th 11 + yy and p redict
the ratio r (11 + 1111y ) / r (11 + yy) with good s ucces s . Ge ll-Mann et al . go t . 25 ,
corresponding to di fferen t choices of how the coupling cons tants symmetry rules
depend on the masses of the states see Brown , Muncek and Singler , Phys . Rev .
Le tters 21 707 (1968) and Chan , Clavelli and Torgerson , Phys . Rev . 185 1 754
(1969 ) . The choice made in the latter paper fi ts very nice . ) This is inter
esting for at first sight the numerator is order o and denominator is order o2
so the o rder should be 137 but here many numerical cons tants accumulate to
overwhelm this factor and make the ratio nearly 1000 times smaller . For this
2
larger probab ili ties for larger m The experimen ts (Gormley) are so
1111
accurate to see this e ffect , even quantitatively , so there can be na doub t of
the mechanism in this case . It is likely therefore that no deep mys teries lie
However the n + 311 does present a challenge . The G pari ty of the 11 ' s
i s - , of the n is + s o the disin tegrat ion i s not allowed , s t rongly . Cons ider
+
n + 11°11 11 - , charge conj ugation (which we think is surely satisfied for decays
at this rate) (n and 11° are charge conj ugation + because they can go into 2y )
+
requires that the 11 11 mus t be symmetri c , they mus t be in an I • 0 or 2 s tate .
n decays is
Branching ratio %
- 0
n 11 11
+
11 23.1 ± 1.1
...
°
311 30 . 3 ± 1 . 1
... 11 11 -y
+
4.7 ± 0.2
... yy 38 . 6 ± 1
0
... 11 yy 3.3 :!: 1
G parity . Hence the final s t ate of three pions mus t have I • 1 . From this we
+ +
+ - °
can ge t an es timate of the ratio r (n 311° ) / r (n 11 11 11 ) . Combining three
symme tric 3, 1
antisymme tric 0
If we suppose the 311 are in their lowes t space wave , s wave and symmetri c ,
s ince they are Bose w e will have I spin 3 o r l ; EM permits only the 1 (but
by proper use of space s tates the skew-symme tric I • 1 s tates could come in) .
228 Photon-Hadron Interactions
(momen t um dependen t ) space s tate the o ther I = 1 can come in reducing the 3 / 2 ;
How do we cal culate the rate n + Jn ? Where does the in termediate photon
ac t ; can we gue ss wh i ch int e rme diate s t ates are mos t l ikely to be impo r t an t ?
A quan t i t ave e s t imate fo r this rate i s a p rob lem that n o dis cus s ion o f the
Lecture 50
HYfo theses in the Parton Model
We should now like to discuss what we can say about what the products
X would look like in deep inelas tic scattering e + p + e + X. lb.ere are some
measurements for certain definite final s tates X for small energies and low
q 2 of the virtual photon (see Berkelman , 1 9 71 Cornell conference ) . Mos t
of these can be unde rstood from direct extensions o f our theories for photon
(q2 • 0 ) reactions . We have already dis cussed pion production from virtual
pion exchange as yie lding information on the pion form factor . In addi tion
p production has been s t udied from virtual photons , with no s urprises - VDM
gives a fair account - see our theory discus sed in Lectures 16 to 2 1 where we
merely have to replace k2out by q 2 in the equations there yielding a fac tor
m 2 / (m ·2 -q 2 ) (q2 is negative ) relative to the q 2 • 0 case .
p p
It is necessary also to make an assumption of how the longitudinally
po larized photon (possible when q 2 � O) couples in relation to the longi tudinally
polarized p - it is assumed that these ampli tudes are re lated by a factor
m 2
e the extra factor q2 /m 2 is an ansatz made because gauge
m 2 q
2 p
-
p
invariance requires that J longitudinal vanishes as q 2 + O. lb.is may be valid
for no t too large q 2 , but of course if q2 becomes really very large the
229
230 Photon-Hadron Interactions
•
asamp tions about small p + p product ion may begin to fail .
2 2
What happens at large q , and how it ties on to small q , in every case
for very large v will be our present concern . We shall have to be guided by
theory and l will take this opportllll i ty to review the parton model and some
assumptions that can be made about i t . We shall lis t assmptions that we can
make - without today being sure of wh i ch are righ t and which wrong - jus t to
see what their consequences are wi th the hope tha t experiment may la ter make
the selection {e . g . are charged partons quarks ? } , The re fore in our list
some ass umptions will (perhaps } be incons is tent with o thers . The assumptions
facts - so one migh t be warned that if a particular assumpt ion neatly exp lains
some experimental fac t it may not really be a significan t confirmation for the
assumption migh t have been made with that fact in mind . Finally , little
effort will be made to derive one assmption from another - they are ce rtainly
not independent . This will there fore unfortunately not be s mathematical and
sound sys tem, but rather a lengthy "intui tive" or physical dis cussion .
Fo r a good discuss ion along the same lines see J . Bj o rken ' s pape r in
General Framework
by giving the amplitude for finding various numbers of field quan ta , or partons
P in the z direction (P + �> . The wave func tion is being des cribed in Fock
etc.
where � i s the ampli tude t o find no par tons (usually zero } ; � (p } is the
o 1 1
ampli tude there is one parton (of s uch and such a type , an index we are omit ting)
Hypotheses in the Parton Model 231
p2
which has moment um p 1 ; � is the amplitude there are two partons of momenta
2
pl ' etc . This can be written in other ways . For examp le let j VAC>
*
represent the vacuum s tate and ap the operator creating a parton of momentum
*
I p1 I P1P2
p . Then we can write the wave function s tate I �> F j VAC> where =
p* • �o · l +
;
�l (p ) a +
l l
f
is s ome function of the creation operators .
Then we make the fol lowing assumptions :
Al . The amp li tude to find a large P� on any parton falls rapidly with lj_
such that e ffectively we can in first app roximation cons ider all P1 are finite
(as P + "' ).
A2 . The "wave func tion" for longitudinal momentum o f order P , i . e . ft�
depends only on x .
This requires some complicating remarks to make its definition clear
+ "'
•
for there is , as P an ever increas ing contribution to � for small x .
Pn ·
More precisely consider the dens ity matrix. • • •
Let �m (p 1 p 2 pn ) be the amp litude
to find n partons of momenta p 1 to Then for example the dens i ty for one at
6 (k-p ) � d 3p • p (k)
i i
n i
That this depends only on k1 and x • k /p when x is finite as P + is the "'
L
L J �: (p 1 , p 2 , . . k ·P
assumption we wan t to make - with all i ts generalizations . E . g . the one
particle densi ty mat rix (like • •
pn ) �n (p 1p 2 . . k ' • ·
n)
p (k , k ' ) depends only on k1 ,kL , k /P;k � /P etc . , when x is fini te .
n
• • • •
.L
is a function of all of the momenta including those of finite momentum
(which we call wees ) · The s caling doesn ' t work for those - in fact the
mean number of particles rises with P so the wave function �n for any fixed
n falls with P ( like a power o f P) but the "relative wave function" e . g . , a
ratio in which on ly one part icle (or a finite number) is moved depends for
fini te x , p only on x, p •
� 1
A3 . In the wave function the ampli tude to find fini te longitudinal
232 Photon-Hadron Interactions
Lecture 5 1
Hypotheses in the Parton Model (continued)
A4 . To have continui ty between Al and A2 the mean number of partons o f
a given type f o r small x goes as dx/x and the wees go as dp � as P + m.
z
The number for p negative falls off rapidly so that for p • xP , finite x
z 2
negative there are no partona ; although for P z negative and finite there are
some (a fixed amount falling rapidly with negative p z ) .
(As a trivial example of the kind of behavior envisaged in the wee region
* *
consider a wave function like exp (E Ck 8k > l o> where ak creates a particle of
longitudinal momentum k and � varies as Ck • a/ (w-k ) w
3/2 with w • lfk 2+ 1 say ,
and a • constant . )
Ai. The behavior o f the wees i s nearly (as P + m completely) independent
of the distribution of the fast ( finite x) partona . Thia again is complicate d .
If we s t retch out the variable p z by , f or examp 1 e , de fi ni ng Y f.n ( • 'vP/ zZ + 1 GeV
2
0 1n2P
We have drawn a graph of the mean number of particles in dy . I f we
look near finite y , P + m we see the behavior of the wees ; i f we look at
finite 1n2P-y (finite x) we see the behavior of the s caling fas t ones ; in
between is a plateau with a finite densi ty of partons so the mean number of
partons rises as 1nP .
It is easy to understand the densi ty but how do we understand the wave
function ? This gives the amplitude for every configuration , which is a set
Hypotheses in the Parton Model 233
0 tn2P
How does this ampli tude vary with the posi tion of the dots ? I t behaves
like a wave function for a finite one dimensional drop of liquid of thickness
tn2P , The ampli tude is large only i f particles are mom o r less everywhere ,
with a uniform dens ity except near the surfaces at 1n2P and O . The behavior
at one surface is no t strongly inf luenced by what the configuration is at the
o ther surface - they are insulated from each other by the long ( ? , 1n2P is
never really very large } intermediary plateau ,
Mathemat ically we are dis cussing the solution of H� • E� for a state of
fixed P z (but we do no t know H, o f course } . I f � is the momentum operator
(e . g . like t P z a:8it etc . ) we wan t � � • P� . Cons ider then that � is an
( 5 1 . 1)
�
longer write VP z
and the behavior of the wees (except for normalization - total s trength of
wees ) is independent of the way it s tarted at fini te x . (In the wee region
interaction energies are comparable to kinetic energies . )
The behavior at this end is a solution of the equation
w� - o cs1 . 2 }
(Note the omission of P . ) In general the operator W does not have a zer.o
234 Photon-Hadron Interactions
eigenvalue if all boundary conditions are used - but here we relax the boundary
m, �
z +
condition of finiteness as p (It is like solving the Schroedinger
• •
equation Hw Ew for E 0 when it does not have this eigenvalue by relaxing
say the condition at r � m and thus studying open scat tering s tates approxi-
mately to which the real large r behavior will have to be attached - here we
mus t ultimately really at tack the finite x solution o f 2PWw m2 w . ) •
The assump tion A6 is not comple tely obvioua from field theory - for
there might be some long range direct e f fect of the fast partons on the s low
ones in principle . The choice is guided by experiment (which shows that the
righ t-moving products of hadronic collis ions depend only on the right-moving
ini tial colliding parti cle and not on what it collided with . To see how this
assump tion is uaed see Bl, and J . Benecke et al . Phys . Rev . ,!!! 2 159 (1969) .
The phys ical assumption Bl that we make later says speci fically that there is
!!:]_. Con tinui ty demands , since the wees are adj acen t to the sea (plateau
region ) that we also have that � (e . g . mean numbers and correlations o f
AS . The
probability there are no partons in a sufficiently large gap of
have a proton s tate and we ask for certain partons a , b , c , for y > y 1 and others
below y
2
s t <E-- Ay � a b c
• •I I• • •
y 1·
con tribution S 1 across the gap . It is evident that this "quantum numbers
carried across the gap " de fined as the quan tud numbers of the s tate minus
those to the righ t of the gap ( a , b , c) is j us t the sum of those to the left
same idea for distributions when two hadrons are colliding . Then it is the
quantum numbers of the hadron moving to the r ight minus the to tal quantum
comp letely and the gap widens s imply becauae P increaaes ; hence t,y .. .tn2P and
the ampli tude falls as p-Y . This was used in analyzing the p ro ton form factor
anything at all over a wide range of y up to the o ther b oundary (order .e.n2P
236 Photon-Hadron Interactions
away ) ay is kep t fixed as tn2P r ises . Probab ility goes as (x 2 tx1 ) -Y where
x 1 , x2 are the x values at each end of the gap ( used for asymptotic behavior
of deep inelastic scat tering near x l ) .
•
Hadron - Hadron Collisions at Extreme Energies
Lecture 52
(
For a hard collis ion suppose the momen ta of A , B are PA ' P B respec tively
)
in the z direction - for example take center of mass PA • P B • P . For a
,...____-__
finite z-velocity v trans format ion from this PA • fP , PB • t P with f • �( l _v ) / (l+v ) .
We only work with PA , P B or P + m .
The asymptotically incoming wave function will be , o f course , (i . e .
before "in teraction") , some kind o f product wave func tion o f A o f momen tum PA
to right and B of momentum P 8 to left , Technical problems arise here . In field
theory this cannot simply be a product o f the wave functions o f each particle
that we have been des cribing for that is not unique ( for example suppose A
contains a Fermion parton at momentum p , and B also contains one of the same
kind at the same momentum,but there cannot be two in the field as they are
*
Fermions ) . Thua if A is represented as a creation operator PA on the vacuum ,
237
238 Photon-Hadron Interactions
* *
�A • FA I VAC> which creates all our partons , and B by FB we can define the
incoming asymptotic wave function as
* *
There is some trouble in the wee region whe re F B FA do not co1111u1 te (note
* *
that no creation operators for fast partons p xP appear both in F and F8
....
A A
because A and B are moving in opposite directions) , Actually this is only
technical because we only want the state after the interaction . The prob lem
would arise only if we were quantitatively calculating the interactions ; but
now we wish to talk about how the wave f\Dlction looks after interaction hence
say after "interaction plus correction for overlap in defining the initial
s tate " , The overlap affects only the wees , but we shall assume the interaction
affec ts only the wees also . (By interaction we mean the e f fects o f the fact
* *
that although FA I VAC> and F8 1 VAC> are both eigenfunctions of H J �> • E l �> ,
* *
FAF8 1 VAC> is not . )
Following we s tate the assump tions we shall make regarding the interacting
wave function ,
.!!!_. Partons interact only if their relative four-momentum is finite ,
J
assuming they have some finite maaa o f order 1 GeV . Thia ia equivalen t to
the statemen t , i f y • i.n ( p z +1 + p z ) (in GeV) ( so y for negative P z ia j us t
2
We use this ass umption to get a t the wave function (in terms o f parton
distrib utions ) for the outgoing final s tate after interac tion . The dia tri-
bution in y of partons for the in s tates A and B have ranges of y small to
i.n2PA and - i.n2P8 to small respective ly . We put them toge ther smearing things
(effect of interaction) over a range &y • 1 . This smearing near y • 0 j oins
the posi tive and negative y regions (from A and B respectively) . Since these
regions were the same for both A , B (see A7) this can be done mos t simply by
Hadron-Hadron Collisions at Extreme Energies 239
j us t extending the common p lateau region from one to the other . The exact
dis tinguish parti cles having finite momenta in the exac t cen ter of mass .
Longi tudinal trans forma tions with a velocity v not too close to c leave the
dist ributions o f such part icles essen tially unchanged . (Ass ump t ion due to
C.N. Yang . )
In our applicat ion the particles are par tons , the t rans formation alters
2 .tn
1 l+v
the pos i tion of the ori gin of y by .tnf • (a fini te amo un t ) and the
l-v
assump tion s ays the dist ribut ion should look the s ame . Hence the smearing
j us t has the e f fect o f extending the p lateau region o f A smoo thly back into B .
plateau
A in
p lateau
of B B in
-4,._������-+-���- y
.tn2P 2
F INAL STATE
p la eau
We mi ght ask i f the smearing of AY, • 1 changes the d is t ribution of part ons
for y near .tn2P where they were de termine d entirely by A . (This region is
called the A fragmen tat ion re g ion , near -.tn2P the B fragmen tation region . )
B
But near A the dis trib ut ion already s a tis f ies ( in the sense that a li quid
Therefore we imagine that the final real hadron parti cles come from th e
is one whose dens i ty matrix looks like that for the A in the A single particle
fragmentation region near in2 P , or bette r if x • p / P , f o r x > 0) , like
(y
A 1
B alone in the B fragmentation region (y near -in2 / P / , i . e . for x < 0 ) and like
B
the universal plateau region in regions in be tween < l x l small and wee ) .
(/ )
with invariance under fini te velocity trans formations which applies to particles .
2 1
But technically the curve in the y space y • tn 1 P z +1 + P z fo r partons
migh t show a bump near y • 0 , a bump which moves when we make a finite
Lorentz trans formation , which is fine s ince wave functions need not be relativis tic
invariants (I thank F. Merritt for pointing this out) . But it mus t be a bump
so constructed to have no physical effect as a bump in the final real hadron
distribution C2 . I t is a "theoretical artifact" due to carelessness in finding
( J +pJ
the right normalization and def inition of var iables for the wave function . The
2 1
resul t cannot really be smooth in the s cale of y • in p z +1 for if so it
(J +p2)
would not be smooth (near y • O ) i f I had arbitrarily chosen to use
y ' • tn p z +1/ �
2
for dy/dy ' is not cons tant .
Lecture S J
Hadron-hadron col lisions at extreme energies (continued)
We now go on to describe what the products might look like in hadronic
col lisions (still leaving out diffraction dissociation ) . We have , of course ,
no quantitative way to ge t from the wave function des cribed in partons to the
wave function des cribed in outgoing real hadrons . But we shall simply assume
that in y space the relation of parton + hadrons is much like the relation
hadrons + partons described in A. Later on we shall have t o describe the
products expec ted from s t ates which unlike that in B J , have gaps in them , for
example , a s tate with j us t two partons at opposite ends of the y s cale ,
separated by 2P . We assume ( the "complement' of BJ) :
Cl . The hadrons which would result from the disintegration of a state
whose initial wavefunc tion consists of one parton a (or a few) go ing to the right
� a and one b parton going to the lef t at yb separated by a large gap
..Ya ::Yb • tn2P consists o f hadrons o f finite transverse momenta in three regions .
lbere are hadrons going to the r ight with y
near Ya depending on the character
of a; there are hadrons going to the left with y near yb characterized by
parton b and those in between distributed in a universal uniform plateau -
242 Photon-Hadron Interactions
It has been proposed that this "sea corresponding to a parton gap " be
i tself a gap . This will not be inconsis tent with what we will say next (C2 ) ,
and which we use in hadron collisions . But i t does no t seem reasonab le by
itself to me if hadrons make a universal sea as B2 supposes - for I think that
means that if there were any dis tribution of two lumps o f hadrons wi th a gap
between , they would make a sea of par tons , so if the re is no sea of partons
there is no separated lumps of hadrons , but there must be a hadron sea . At
any rate it is my s trong belief that there is in fact such a sea in this case
and not a gap in hadron momenta corresponding to the gap in parton momenta.
As A. Cisneros points out the two outgoing lumps carrying opposite
hadronic quantum numbers ( in the case e+e - + hadrons ) would genera te a dipole
s t rong current which would tend to radiate softer hadrons into low x . To pre
vent this radia tion becomes increasingly mo re di fficult as the energy increases ,
(as for two-body exchange reac tions ) . Any exclusive two-lump probability will
fall as a power of energy relative to the total inclus ive reaction in whi ch the
radiation , generating an intermediate plateau , is permit ted .
Now for a wave function as in B 3 we can imagine the various partons
disin tegrating more or less like in Cl but not really independently , those at
the ends of the y range determine the hadrons there , and tho se in the center
affecting the hadrons in the center , but in a universal manner independent of
y in thi s region . Thus we again ge t a hadron dis tribution like that in Cl
with three regions , but the plateau may be a new and different dis t ribution
but a sea nevertheless . (Whether the two plateau regions , the one in Cl
corresponding to an initial parton gap , and this one for wave function B J ,
are the same or no t i s a di fficult problem I have not yet been able t o decide .
We shall call the assump tion that they are equal C6 , see lecture 55 . ) Thia
assumption can be made by j us t repeating the wording o f Cl j us t changing the
name of the sea , o r it can be put in another totally equivalen t way .
Hadron-Hadron Collisions at Extreme Energies 243
parton into two or mo re , and these again are broken up e tc . , in a cas cade
fashion making profound changes , for example by filling in the gap in the
low P z region and creating some sort of parton plateau in the final outgoing
s tate "appropriate" to the initial s tate Cl .
The reason no such extensive modification is made in going from the
ini tial s tate in a hadron collision to the final appropriate state BJ is this .
The distribution o f the fas t (non-wee) partons in the init ial s tate already
s atis fies H� • E� so little dis turbance is worked there by the Hamiltonian .
Only in the mutually overlapping wee regions does the further action of H
( )
for the hadrons are on their mass shell and have a definite mass . We take
1.n (E+p z ) • +m +p 2
2 2
y • 1.n V/p1 z
'
�
+p z say in GeV ( change o f scale means j us t a
change in origin of y) :
CJ. Putting th i s a l l toge ther i t means that i n a hadron collision A + B +
to hadron wave functions . We shall not dis cuss them in de tail for hadron
collis ions is no t our main subject but give some examples . We assume ,
analogously to AB that
on the quantum numbers of A-C . Looked at from a Regge point of view this
whose a depends on the quantum numbers exchanged in the
-2 (l-a )
should go as s
t channel , which is the same as A-C . Thus a is identi fied wi th 2 ( 1-a ) (or
whatever the correct power law o f energy fall-o f f turns out to be) and we make
a contact with the theory of exclusive reactions . (The same goes if C is
Hadron-Hadron Collisions at Extreme Energies 145
n + p
2
two particles like of fixed total mas s not necessarily at resonance .
I do not know o f examples o f f resonance where this powe r law has been checked -
again we see the uni ve rsal p rin ciple that going to higher ene rgy does not l i f t
a resonance ever higher against "non-resonan t " background - tile latter can
always also be though t of as tails of o ther resonances . )
Again applying this to the case that C only is near the end o f its range ,
o f mass � say going t o proton W and pion . Th i s spills protons over a range
f rom x • (E -p ) / � to x • (E +e ) / � (as P + m ) , where E ' p are the energy
p p p p p p
and momentum of the proton in the res t frame o f the resonance . Although the
latter is less than one , there is a very small gap (of range . 9 8 to 1 for
Ha2 • 2 . 16) f ree of dif fract ion gene rated pro tons it is too small to isolate
experimentally . I f a gross plot is made for x not sufficiently near 1
c
various numbe rs of dissociat ion pro t ons are included and th e variation o f
•
numbers appears far from (1-x )
c
l-Za
dx
c This di f ficul ty does n o t arise for
pions when the incident par t i c le i s a pro ton .
Lecture 54
I should like to make a few coaments about our "conclus ion" from C2 .
First s ince the mean number of particles goes as dx/x in the small region and
of a given kind ( the multiplicity for tha t kind) rises logarithmically with
1nP or with lns . This is al so obvious in the area of the y plot whe re the
plateau expands logarithmi cally with s . But the plateau region is (statistically )
neutral , i ts average for any addi tive quantum number such a s charge , third
momentum e t c . , mus t be zero (becaus e if not dx/ x would give a ins dependent
value for one o f these fixed cons erved q uan tum numbers ) . We expect this from
J
Thus s uch integrals as
additive quan tum numbers) for the right -moving particles , by s imp ly adding
the to tal number for all for which x > 0 ( i . e . p > 0) in the CM system .
z
This number will vary from event to event , of course , but we want the s tatistical
expected mean ove r many events . Thi s "mean righ t q uan tum number " will approach
a cons tant as s + .. . Thus we can talk of the " righ t mean 3 -isospin " or the
" righ t mean s t rangenes s " . I t is evident that thes e mean right quantum numbers
mus t , under the ideas o f C2 , be the same as those o f the incoming righ t -moving
particle . The plateau region does not le t any quantum numbe rs s l ip through i t .
( I f for examp le we take a symme tric colli sion A +A then by overall quantum
number conservation and symme try the right quantum numbers ( and the left)
the energy , the momen tum (minus a cons tan t ) , the 3 isospin , s t rangenes s , baryon
NOTE : We show that , dis regarding quan ti ties of order l / P , the difference of
total energy E and to tal momentum P o f the par tic les moving to the right is a
2 2
+m ) /2Px (m is the mas s of the hadron) which is of order l/P and therefore
J.
(p
negligib le . The main cont ribution comes from x near zero where the dis tribution
of a particular type is cdp/ & , hence the con tribution to & - p of these is
Hadron-Hadron Collisions at Extreme Energies 247
c
f The integral g ives c .m
V
P.J. TIU , so D
�
p •O
z
is the sum of c ove r the transverse momenta and the types of hadrons
in the plateau . If the pla teaus are universa l , the cons tan t D is unive rsal
and may be eas ily calculated in te rms of already measured quan t i ties .
part i c les moving in i t s own direct ion . The " t rain A" has the quantum numbers
of the particle A and its ene rgy (by the conse rva tion of energy) but has l os t
a ce rta in momentum D in the inte rac t ion , it is held back a b i t by the in ter -
action , A and B each lose D to the othe r . ( Such a finite momen tum trans fer
is, of course , consis ten t and un ders tandable i f only wees inte rac t in the
collision . )
A2 to A6 , the to tal momentum o f the par tons is , o f course P , the total momentum
-Mil +p
2 2
r
Jip2
..L
of the s tate , but the total ene rgy E E • i s no t the to tal
i i
i
energy E "' P because of inte rac tion ene rgy wh i�h compensa tes the expe c ted
f in i te excess of E E above Ep )
i zi'
As a first s tep to describe these things formally , we are trying to
(where " in righ t " means having very large positive longi t udinal momentum P ,
and " in left" means -P) in terms o f outgoing hadron s tates - an element of
the s ma trix . O f course the mos t likely thing is tha t the two parti cles
outline our ideas . Formally this wave func tion can b e given i n terms o f the
*
ampli tude to f ind various outgoing hadrons . If c is the ( formal) operator
to represent the creation o f some kind o f a had ron (kin d , transve r se momentum
"'
p , and longi tudinal momentum p are indices of c ) we can represent s uch
1
*
s t a te s by X / VAC> where X is some operator fun c tion of the c . We have been
[ the exac t way the plateau of M cuts off for finite x is arb i trary ; i ts
choice affects the de fini tion of GL , GR defined late r , but the final operator
X is not dependent on this } . Next we write X as G G� where G is to modify
L R
*
the sea on the right ( for x > O. It involves creation operators c to add
particles to (and beyond) the plateau operator M , and annihillation operators
c to take particles out (which were put in by our arbitrary cho ice of how the
p lateau M is defined) but all these for x > 0 ( that is the meaning of the R) .
*
Likewise G L is an operator func tion of c , c only for x < O . The operators
L R
G , G commute s ince they contain operators o f different particles (some signs
mus t be adjus ted for Fermi particles ) . Thus we write
Comments : By assuming that the wee region is the same for each hadron and that
Hadron-Hadron Collisions at Extreme Energies 249
only wees in terac t have we not assumed that all to tal cross sect ions a or a
pp rrp
etc . are equal , clearly contrary to fac t ? I have no t though t this out c learly
but have always suppo sed that the part of the wave func tion which does inter-
act (which is always infinites imal compared to the part whe re they go pa s t
each o ther wi thout inte rac t ing) could s t il l have some normal ization re lated
to the total cross sect ion for that parti cular collision without being in-
cons is tent with o ther ideas . In the formal express ion above , for example ,
each G could carry a numerical coe f f icient g proper to A . This would make
A A
2 2
total cross sec tions proportional to g g , or as is said , factorizable . I t
A B
ma y b e that the previous assump tions d o no t imply that the t otal cross s e c tions
are necessarily equal , but rather perhaps that they are factorizab le . It would
a •
imply for examp l e , that
rr p VfO r;;-
pp \f rr rr
etc . We do not have any evidence on
transve rse momentum behavior (except to say that transverse momenta in hadronic
collis ions are limi ted , a res ul t taken direc tly from experiment ) . Obviously
lots o f in teres t ing theore tical ques t ions remain , s uch as what func tion is the
transve r se momentum d i s t r ibut ion , how does it differ for various values o f x ,
o r for rr ' s and K ' s ? How should exclus ive cross sections vary with t , e t c . ?
This entire realm o f phenomena has been le ft out o f our analys is , an excel lent
Lec ture 5 5
�e rms o f partons the re are terms giving the coupling o f partons with the
vector po tent ial of the quantum ele c t romagne tic field . We shall assume in
the spirit of minimum e lec t romagne tic coup ling that they couple in the simplest
way expec ted from the propagat ion operators via gauge invariance . That is we
assume :
minimum coupling operator . That is also the coupling that would be valid i f
for the present this will not concern us for we shall suppose par tons are
ei the r spin 0 or spin 1/2 . Al though we are in danger of no t having the mos t
gene ral case w e shall never theless explici tly next take the working hypo thes is
D2 . All the charged partons are of spin 1 /2 and hence couple through
the current operator e tly where _ e tl is the parton charge (ti is an index for
ll
the kind of parton) .
250
Final Hadronic States in Deep Inelastic Scattering 251
We have seen how the assumptions Al-A8 p lus these two Dl and D2 lead to
the s c a ling expectations for the deep inelas tic s ca t tering ( lecture 2 7 ) and
discuss what we can s ay about the p roduc ts in photon collis ions , in part i c u la r
we 2Mvx , Mv •
2
begin wi th the deep ine lastic ep scattering region q • - P •q
sys tem with the virtual pho ton purely spacelike q • (O , -2Px , 0 , 0 ) P • (P , P , 0 , 0 )
µ µ
q • -4P x , 2 Mv • 4P x . Then as a result o f our awsump tions the parton wave
2 2 2 2
�
-2Px
Before
Immediately
After Px
Coup ling
That is , one parton ( say type a) moving to the le f t , the remaining par tons
the partpp dist ribut ion 1 u s t des c r ibed , The total cross sect ion in this s cale
is proport ional to � e
a
2
n (x) so the dens i ty ma t rix per collision is a super-
0
pos i t ion o f cases of d i f fe rent types a of partons with weights , wa depending
es n (x) ;
2 2
on the charac ter o f the collis ion t h r ough x ;
a
w (x) • e
a na (x) /ta a
the sum of the weigh ts be ing 1 .
One obvious consequence of this and our othe r assumpt ions is that in this
sys tem the transve r se momenta will be limi ted an� ( fo r fixed x as we vary P ,
o r i f you like v ) .
n (night movers) we expe c t it behaves like the unive rsal M-plateau charac teris t i c
o f a wave funct ion like B 3 , which w e know f r o m hadron collis ions . For nega tive
n we are in the "plateau region of init ial par ton gap " (de fined in connection
252 Photon-Hadron Interactions
wi th Cl) , We have no t assumed these two p lateaus are the same so the coe fficient
of d11 / 11 need not be the same , If they are not we shal l have trouble defining
what happens in the trans ition region - it cannot go s imply as dp z / E for that
leads to the same coe f ficient for plus and minus 11 . We see however that this
ques tion is to some extent an arti fac t of our particular cho ice of coordinate
system . Note that the s ta te described he re as " immediately after coupling" is
an ini tial parton s tate ( in the sense discussed in the remark fol lowing the
discuss ion on Cl) - there s till mus t be interac tions from the Hamiltonian
be fore it becomes the "appropriate" outgoing wave function . This will produce
cascading of the le ft-moving parton into the gap smearing the wee region into
negative 11 and making large readj us tments for the righ t-moving sys tem also
(because , for f inite x; they are no longe r the correct solution of Hw • Ew
le f t and righ t . )
Assumption C5 , if it were right and this continui ty in dP z / E , would
seem to sugge s t that both plateaus fit togethe r . I am not sure of myself
here but shall put i t down as an explicit assumption which would , if i t is
true , remove all our difficulties - the dn / n region i s always universal ,
that of hadron-hadron collisions .
Final Hadronic States in Deep Inelastic Scattering 253
every ini tial pa rton distribution . This as sump t ion is at p resent , ve ry weakly
based and may eas ily be wrong - it is an in teres t ing conj e c ture .
We now make a more detailed discuss ion of our expe c tations for the le f t -
moving part icles . (I have profi ted grea tly from conversat ions wi th A . Cisneros
on these mat ters . ) Fo r these parti cle s a variable more convenien t than n (which
goes down to -x) is z • - n/x • -p 2 / Px = P • p/P • q the fraction that the left
moving part icle ' s momentum i s o f the to tal le f t -moving momen t um . S ince this
the pho ton i n the laboratory sys tem . I t is the p roper variab le for seeing
how the vir tual pho ton fragment s . Of course as v � m , x fixe� the d i s t ributions
If we could be sure that only a parton of type a came out (which by the
way , can be much more nearly done for neutrino scat tering - in the quark model ,
neutrino scat ter ing can lead to unique quarks to the le f t ) the dis tribu tion
left ( a ) and the adj acent wees from the hadron ( i f at all) - and these
latter are unive r sal and unaffected by the removal o f the parton a t x from
the proton . The la t ter does no t affect left-moving hadrons for its rela tive
relat ive probabili ties of producin g di f ferent kinds of partons a will depend
on x . The actual dis t r ibu tions D (x , z ) wil l b e the weigh te d average fo r each
part on
2
w (x) - e n ( x)
a a a / � e s 2n s (x)
s
dis cuss later a specific par ton mo del (quarks ) as wel l as some practi cal
que s t ions about the possible ext rac tion of the isolated D ( z ) from experimen t ,
a
as well as the possib i l i t ies o f finding the w (x) by special guesses ab out how
a
D ( z ) may behave . To me the possib ility of special funct ions charac teris t i c
a
o f each kind o f parton i s a very interes t ing possib i l i ty , an d one tha t could
actions .
These same funct ions D ( z ) wil l appear in certain other expe riments - for
a
examp le , of course , in deep neut rino p ro ton � µ + p roduc ts experiments . The
analys is is nearly the same as here excep t that the fundamen tal coup ling may
be differen t so al though n (x) are the same the weigh ts w (x) come out dif ferently .
a a
+ -
Again in the e e collis ion the assumption D2 says our ini t ial s tate
- 2
is j us t a pair , parton a and antiparton a with weight e • Thus , ass uming
a
2 R L R L
C6 the final s ta te would be
a a a a a a
E
e (D D- + D- D ) I M -p lateau> again producing
Ia
e
a
2
D (z)
a
I f , for examp le , in some expe riment we could be sure that a cert ain
parton a came out to the left say , then as we have seen we would expect tha t
number o f tha t number fo r all hadrons llK)Ving to the le f t ave raged over al l
or dete rmine in terms of experiment the quan t um numbers of the par tons . If
the s tate is n o t pure w e shall have to know some thing of the weigh ts w (x) t o
a
mak e this useful - but there are so many di f ferent kinds o f expe riments
possible that in principle the w (x) can be de termined as we ll as the ove rall
a
quan tum numbers of the partons .
The particles to the righ t under de ep ine las tic e-p s c a t tering come from
fragmen ta tion of a pro ton with one a parton of llK)mentum fract ion x remove d ,
say E (z) . They are evi den tly not ve ry fundamen tal . But it is clear
(p -a , x )
that the same kind of f inal s tate resul ts (on both s i des , le f t and righ t ) in
+
Dre ll ' s experiment p + p � µ + µ + any hadrons so the p ro ducts in this
Final Hadronic States in Deev Inelastic Scattering 255
NOTE : Our f inal hadron s t ate is acco rding to our ass ump tions
w ( ) DL E IM
R
a X
p lateau>
a (p-a , x)
a
exp ress ion is p robab ly impo s s ib le for one ope rator M always . Be caus e nothing
would seem to p reven t us from wri ting D 0 I M p lateau > whi ch woul d have to tal
R L
a $
quan tum numbers of two quarks (a + $ ) whi ch is impossible to write in terms
o f the hadron ope rato r s having only integral quan t um numbers . (I am indeb ted
to J. Mandul a for po inting this out . ) A va lid mathema t i cal repres en tation
The reade r should be warned that a number o f these s caling p redic tions
for special p roduc ts of reactions may only hold at much higher energies than
that at whi ch s caling for the total cross sec tion (vW and W ) sets in .
2 1
This warning res ults from theore t i cal experience wi th a numb e r o f . examp les
o f analogo us theo rems in non-relativi s t i c quan tum me chanis whe re the sum
works well before the individual terms do . This is because if cer t ain inter-
ac tions are dis regarded in wo rking the total p robab i l i ty by assuming ce rtain
s tates only are "en tered" subsequent interactions may not change the total
p robab i l i ty the s t ate was "en tered" b u t may redis t ribute that p rob ability
In the special case o f small x the predic tions are espe cially simple .
tribu tion j us t l ike that of a p ro ton with only a ve ry low x parton removed
and the wees dis turbed (by interact ion with the p lateau deve loping from the
left ) , Thus all the partons ot any s ub s t antial x are exac tly like that of
a proton - and we can expe c t the s ame dis t ribution of hadrons to come out
(at leas t for z >> x ) as do come out fo r a hadron colli s ion of a p rot on , s ay
256 Photon-Hadron Interactions
,,,;i F
R R
( p-a , x )
Hence for x small E
p
Next for small x all n (x) dx go as
a
Ca dx/ x where Ca is a cons tant , so
L RI
D
r p
F M-plateau>
the s ame as it does for a hadron collis ion . And the virtual pho t on also
fragmen ts in a un iversal way independent o f x . Since w e have ass ume d the low
on the part icle s t ruck by the pho ton (normalized to the total cross -sec tion for
collision , of course ) . A smal l x pho ton and a hadron behave j us t like the
collis ion of two hadrons , each fragments in its own charac teris tic way . That
Region of finite q
2 , v + •
For finite q, negative q2 we can s till use our sys tem o f coordinates in
affected by the pho ton Q ; that is only the wees are e f fe c ted . They are ,
the wee region . we cannot there fore predi c t what will happen there , but we
can note (a ) that it is the s ame for every hadron A , A + y + produc ts for
we have the s ame wees for eve ry hadron according to A6 , and (b ) the frag-
men tat ion o f the fini te x in the above sys tem is charac teris tic of partons
of sys tem A only , for only the wees are effec ted by the pho ton .
In consequence of (a) the produc ts on the left which can be desc ribed
z •
!'..:2.
in terms of p . q where P is the proton four-momen tum, p is tha t of a
product and q that of the pho ton , for finite z as v + • , is some kind of a
dis t r ibution D
2
y , qz ( z ) . The dist ribut ion clearly depends on q , the vir tual
Final H adronic States in Deep Inelastic Scattering 257
mass of the pho t on , because the comp licated interact ions of the wees depend
fragmen ts in the same way as it does for hadron col l i s ions . The se cons idera
•
2
tions hold for q 0 also , of course , but our coordinate sys tem is inconvenien t
fo r such a case .
We could also use the cen te r of mass sys tem for any finite q
2
in teracts only with the wee p artons o f the targe t p ro ton (or h adron A) . Th is
interact ion is complicated but p roduces the same dist rib ution for any hadron
2
fo r given q The hadron b ehaves as it always does where its wees are dis-
turbed whe ther by another hadron or by a photon . Formally our final hadron
s t ate is
L RI (5 5 . 2 )
D 2 F M-plateau>
y,q A
Thus as far as high energy inelas tic collis ions are concerned the
This of course makes a nice union with the idea o f ve c to r meson dominance ,
2
that a free pho ton ( q -0 ) h as a certain reas onab le prob ab i lity to be a vir tual
vector meson and as such woul d behave in hadron collis ions like a hadron . We
note now we shall � have to de termine with what p rob ability it looks like
2
a hadron and how this var ies with q , for in any event i t , as a whole , should
In the center o f mas s p ic ture (and also in the spacelike q figure ) there
are terms o f coup l ing in whi ch the pho ton f i rs t divides into par tons on the
way in , for example one fast one slow , and these s l ow partons interact or
annihilate with the wee partons o f the hadron . Thus the p i c ture that the
incomin g photon looks with s ome amplitude like partons itsel f is reins t ated .
such diagrams falls away and only the dire c t coup ling t e rm o f pho ton s c a t tering
258 Photon-Hadron Interactions
x
2
Con t inuity o f large q and small region
x As
Finally we shall match our finite
2
q region to our small region .
2
we have done before we shall suppose when v is very large and -q large but
2
-q /2Mv small the limi t may be taken in either order - i . e . , we can ge t the
+ m x
2
res ult e i ther from our fini te q , v fo rmula or from f ini te , but small ,
v + m formula .
Thus ( S S . 1 ) mus t agree with (5S . 2 ) for large q2 • This is easily done ,
2
the results agree if only we add the res ult : •
D , q 2 Dr for large q That is :
y
2 of
The fragmen tat ion p roduc ts o f a pho ton of large -q become independent
(We are in all casea normali zing to the total cross section
2 2
-q as -q rises .
Lecture 56
Partons as Qu&rks
We could now go on to discuss various models of what quantum numbers
partons carry , but we shall limit ourselves to one example , the one that is
most interes ting . The s tudent should try other examples , such as the Sakata
mode l , to see whether we can e lilDinate them by experiments now done , or proposed .
We shall suppose the charged partons come in six varieties , three plus
their antiparticles . The three called u , d , s carry the quantum numbers of the
three quark s tates (of the low energy quark jllO d el) . This we swmnarize by :
El . The charged partons are guarks . Mos t of our previous assumptions
were guided , or we thought they were guided by field theory or considerations
based on high energy experiments . This , of course , is not , it is an inspired
guess . But i t is also contrary to what can be true if the field theory is too
ordinary . For in such a theory there would be a base s tate o f quark number
one (and non-integral charge in a localized wave packet) and , in view of the
conservation of quark numbe r , some eigens tate of the sys tem of quark nuinber
one is expected . In other words we expect to see real particles with quark
quantum number s . They have not been seen . I t is possib le to imagine that
they have very large mass - but this makes it very hard to take all the
259
260 Photon-Hadron Interactions
previous assump t ions about all parton interact ions limi ted to the GeV region ,
etc . The re may be some way to reconc ile all this - i t is one o f the mo s t
anothe r unnecessary assumption that I will not use bu t I add to remind you
"physical quarks have high mas s " , go ahead - you s t i l l h ave theore ti cal work
may be wrong - one o f the mo s t impo rtant expe rimen tal j obs o f the future i s
discussed in lec ture 32 , come to the conclusion from experiment that in any
case :
E 3 . Neut ral partons als o exis t . What they may b e like we do not now
know , exce p t perhap s they may not be vecto r (b ecause the �.P degene racy is
not l i f t ed ) .
to re concile E l , E2 and the other assumption s we have expl icitly made about
A careful review o f our assump tions shows this . The re is poss ible doub t ,
re lat ive y) but this is only used to make the later assump t ions mo re plaus ible
conclus ion is in connec t i on with " righ t -moving quantum numbers " (discussed
in lec ture 54) for the parton function D(lR and whi ch should now be non-integral .
These numbers are defined s tatis t ically as the ave rage over all events -
al though each event mus t give inte gral values the average , of course , need no t
the mean numb e r o f baryon less antib aryons found on the righ t should ( at
The arguments lead in g to this conclus ion are no t obviated by the fac t
that the partons do not have integral quantum numbe rs . Imagine for a very
Partons as Quarks 26 1
large tnP tha t the quarks are dis t ributed in the ( f inal s t ate) wave funct ion
in a long plateau: ( the left end o f the plateau is gene ra ted by a long chain
* *
of cascades via terms l ike a a a f rom the ini tial s ingle quark)
v v
q Q 9
4J_ v
.o.\
11 K 11 11 p K 11
make legitimate hadron quan tum numbers . In do ing so they take combinations
of quarks over a finite range of y as il lus t rated above . (The overall total
triality mus t be zero , o f course , fo r the ini tial s t ate has zero triality ; in
our illus trat ion the initial s t ate has baryon numb e r one . ) We assume tha t
there is some non-zero probab ility pe r dy to p ick of 3 quarks (or 3 ant iquarks)
number ( and s t rangene s s ) becomes randomized and the central region of the
p lateau i s neu t ral on the ave rage in these variab les . Thi s means tha t the
r i gh t and le ft mean quantum numbers of the f inal hadrons approach a con s t ant
that depends as we have supposed on the ini tial righ t or left quark charac ter .
It doesn ' t make any d i f fe rence exac tly where you cut the plateau in deciding
rep resen t s u ch a . d i r e c t measure o f the suppo sed non-in tegral quan t um numb e r s
o f the parton quarks that we should s ay some words ab out i t s pos s ib le ve ri-
type of quark thrown to the left so the beauty o f the res ul t would be con fused
by h aving to first know the w (x) by one method or ano ther (see me tho ds below) .
a
On the other hand for x near 1 we have been le d by experiment (ratio o f
v W2n to v w2p ' lec ture 31) to suppose that only u quarks s urvive - hence i f
Another way t o insure pure quarks is by neut rino s cat t e ring which we dis cuss
be low .
it may be harder to form K ' s and still more di fficult to form baryons (becauae
of their masses ? ) . If this is the case we should need a very long plateau
indeed to ge t equilibrium in baryon number, although hyperon charge might be
easier and isospin eaaies t of all . In the energy ranges availab le to experi -
men t therefore I would expect the quantum number rules to work best for
isospin , next for hypercharge , and b aryon number las t (i . e . requiring the
largest energy) .
Tiie eas ies t would be isospin . For example we would expect that the
left-moving z isospin : t I ziNi ' the sum of each product hadron moving to the
left ( for ex�le in the system with qµ pure spacelike , or maybe the center of
mass) , will per collision , for a given x be given by
that is , i t will rise from zero for small x (where u (x) • u (x) etc . ) , never
rise above + 1/2 (or fall below - 1/2) and approach + 1/2 for x near 1 (where ,
we think u (x) dominates ) .
'nlis relation can serve as can a number of others of this type in one o f
two ways , In one case we migh t suppose u (x) , u (x) e tc . , already known from
some o ther method - such as neutrino scat tering explained in lecture 33 - or
analysis o f p + p + µ+ µ - + anything , or by perhaps one of the equations developed
further on . In that case (56 . 1 ) is a quantitative prediction to tes t the
consis tency o f all the ideas that partons are quarks . Alternatively it can
i tself be used to give further information on the six separate functions
u {x) , ii (x) e tc . , so that they may be separately de termined . niese could then
be compared to the results of other methods of getting them , but do not serve
directly as a check of the model .
However , even without complete separation of all the functions a relation
such as may be used to check the quark mode l becauae of the exis tence
�
(55 . 1)
1 l
of sum rules (Equation 3 1 . 2 ) such as
J
(u (x) - u (x) ) dx . 2 ,
0
(d (x) - d (x) ) dx . 1
horse or cart . There are a ve ry large numb er of such rel ations whose
s ome of the gene ral rela tions expe c ted from our theo ry and leave the choice of
the best way to use them o r comb ine them to compare to expe riment up to you .
Le c ture 57
Parton& as Quarks ( con tinued)
There is ano the r way to ins ure that one ge ts a pure quark o f one type
recoiling in the final s tate , and that is wi th deep inelas tic neu trino or
E4 . The weak interaction is via the current Q ' yµ (!±l- 5 ) Q where Q is the
Dirac operator for a u quark and Q' that for a "Cab ibbo quark " - a qu!-��
in teresting que s tion is , howeve r , bes ide the s cope o f our discus s ion ; after
i t is determined the same kinds o f ques tions and remarks will apply to
products gene rated by the aforemen tioned current operator . In all cases
f � -f;
p p
(j us t the s caled func tion 2Mw -2MW f o r an tineutrino s cattering on
1 3
protons ) was j us t pure ly u ( x) . Hence the same holds for the products , ie . ,
the probab ili ty of a product in w (determined as the app ropriate coe ffi cient
1
in the variation of cross section wi th labo ratory neut rino angle keeping q , v
f ixed) minus tha t in w for vp s ca t tering are purely the products for a u quark
3
264 Photon-Hadron Interactions
being kno cked backwards (forwards is a pro t on les s a d q uark with p robability
.06) . Thus by s t udy ing the products to the le f t in this comb ination we are
types . For example , f �P+f�p for neutrinos on p ro tons gives pure u quarks
and should give var ious p ro ducts with probab i l i ty D u ( z ) u (x) ; a fixe d dis
trib ution independen t o f x, the total cross section being u (x) . Again
f � -f �
p p
for neut rinos on pro tons gives nearly a pure d quark (cos ec
2
• . 94
d quark, s in ec •
2
. 0 6 s quark ) .
From these dis tributions the total quan tl.DD nl.DDb ers of the quarks can be
de te rmined . In this way alone cert ain models can b e e liminated , but o thers
can not . For example we cannot distinguish the quark model from the th ree -
triplet model where the re are 9 partons (and 9 ant ipar tons ) in s e ts of three .
A, B , and C each set having three s t ates like quarks wi th various integer
quant \DD numb ers . Thus what would come out in the expe riment on £ -f for
1 3
v p scat tering , where we expe ct a pure u quark in the quark mode l , woul d ,
1 / 3 integral and j us t equal t o the u value , f o r that is how the integer charges
A, B, C were chosen .
+ -
for (O the r experiments , s uch as e e + any thing or
corre lations of left and right s i de dis integra tions migh t dis t inguish these
models . )
There are a large number of predi ctions implic i t in the re la tion that
the dis tribut ion o f a given final hadron in the left (photon) direct ion ( for
convenience no t no rmal ized) is in gene ral given for deep inel as t i c ep s cat tering
by : (Equation 5 5 . 1 )
(5 7 . 1 )
Partons as Quarks 265
where D ( z ) e t c . , are the distributions of the produc t in ques tion for pure
u
up quarks , e t c . There are s ix func tions in general so it is di fficult to
cert ain comb inations of measu rements fewer functions are involved . We
In this way we see for w product ion there are real ly only
-
produces a w .
w w
o!! os!!
+ +
D - D -
(5 7 . 2 )
s s s
+
In fact i f we mea sure the number of w minus the number of w at a
D
w
\x , z ) - D
w
-
(x , z ) • A(z) [} (u(x) - u (x) ) - } ( d (x) -d (x) � (5 7 . 3)
+ -
where A ( z ) • D
w w
(x) - D ( z ) . Thus we expec t the dis tribution (probab i l i ty
u u
as a func tion o f z) to be the same for all x . As we vary x we can determine
4
9 -
(u (x) -u (x) ) -
1
� -
d (x) -d (x) ) , within a cons t an t . This is j us t as in 5 6 . 1 , but
we do no t have to measure ove r all z to in tegrate , and meas ure other parti cles
+
as we ll . A me re measuremen t of w and w- at some convenient z would be enough .
The abso lute coe f f icient can be de termine d in two ways , ei ther from the s um
a +
4
as 9 u (x ) + f
ep
(x) known function as x 1. Additional in forma tion would
come from the same experiment on the neutron , o f course , (we ge t 9 (d-d ) +
4 -
9 (u-u) ) .
1 -
+
The sum o f the number of w and w does not give us much that is new
about the dis t ributions , but we can roughly p redict its x dependence
w+ w- 11'+ w- � -
D (x, z) + D (x , z ) • (D ( z ) + D ( z ) )
u u
\? (u (x ) + u (x) ) +
266 Photon-Hadron Interactions
+9
1 -
(d (x) + d (x ) ) + 9
1
- o"
s
-+-""-- (s (x)
11 11
+
J
s (x) ) (5 7 . 4 )
Du + Du
The express ion i n curly b racke ts i s the same a s fep (x) except for the
coefficient of the las t term (which should be simply 9 1)
. Howeve r , that term
is probably small ( for not only should s , s be less than say u , u in a proton
but also u is enhanced by 94
relative to 9 1)
so the distribution of 11 + plus 11
is p robably nearly independent o f x , and if normalized to f ep (x) depends on
z only .
We can do similar things for the p roduc tion o f other particles , for
+
K
example K-mesons . Here there are six independent functions Da (z) , the
others are obtained by isospin reflection or charge conj ugation . (For example ,
+ -
Dl<u • D-
K
• D
Ko DK° The s tudent can verify that for v on pro tons
d d •
u
+
if we measure particles at a given z to the lef t , if N ( z ) is the number o f
K+ e t c . , w e find the following resul ts : The isotopic spin di fference depends
on one function , with the same x dependent coefficient as before :
0 {, 0 ) (
N+-N 0+N -N - • \D+u-Du0+Du-Du- ( z ) 9
4 1 (d (x) -d- (x
(u(x)-u- (x) ) - 9 ) >) ,
0
another comb ination (which does not require K and K° to be distinguished)
als o factorizes into one fun ction o f x, one o f z :
Partons as Quarks 267
ep
but the las t factor in curly b racke ts is like ly to be close to f (x) for any z .
(a meas ure of hypercharge , b ut unequally sens i t ive to s t r ange and non-s trange
survive , call them � . s respe ctively . Then a measure of K mes ons to the le f t
ttU + S e ad + s ii
+ 0
Number o f K Number of K
au + S s ttd + S s
- 0
Number o f K Number o f K
I n this , a s i n a l l cases , data o n vn gives add i tional info rmat ion, change
We have not di s cus sed the right dis t r ibutions , but the re are re lat ions
see that the obj e c t s p roduced at the right ( in the hadron fragmentat ion
region) are the s ame on bo th s ides for the pro ton but no t for the neutron -
for the en experiment we mus t observe the isospin re fle c ted p roducts . If
this is done , the relat ion holds f o r the part i al cro ss sections f o r any
Finally to b ring a l l the hypotheses w e have made ab out par tons together
en ep
into one list we note finally our sugge s t ion Chat the f /f rates s t rongly
suggests that when a pro ton has a quark near x • 1 and a rema inder of small
of the 56 mul tiplet in su language meaning our su only q ual i t atively , not
6 6
exactly quantitatively .
268 Photon-Hadron Interactions
the case that the part on is a quark , hence � in su 3 (no t an antiquark) and
near x • 1 than does the octet . We have already discussed the imp lications
for t o tal cross sections . It has imp licat ions for produc ts also , of course .
the the chance the baryon cont ains a fast quark near x • 1 of the s ame kind .
There are then many imp l ications for produc ts - and we have ment ioned some
that come from u (x) being larger than all the o thers as x � 1 . Th ere are
o the r s , of ano ther type , for example (Cisneros , private communication ) in the
+ -
e e � hadrons s ince uu has four t imes the probab ility of d d the chance of
producing a proton with x near l( in the cen ter o f mass ) and any thing else is
mesons . The analogous assumption is tha t when one quark takes mo s t of the
momentum it is of a type that the low energy quark model suppos es the meson
remainde r is CJ) wi th s pin 1 / 2 , p lus the charge conj ugate with the s ame
p robability ( for the charge conj ugate i t is Cl} near x • 1 and the remainder
is Q) .
We have built a very tall house of cards making so many weakly-b ased
conj e c tures one upon the other and a great deal may be wrong . (P robab ly
the weake s t is C6 - same p lateau for gap and hadron - but i f it we re wrong i t
does not alte r t h e thrus t o f any o f the others - j us t i n t h e ope rator expres sions
we shall have to be care ful to use the righ t plateaus . ) Neve rtheles s this is
the bes t guess I can make now - and we can try to us e them has working hypo theses .
that the regularities observed h ave some ent i rely di f feren t b as is or are ,
Partons as Quarks 269
in part , accidental . The es tabl ishment of evidence for the quark mode l
(and we have indicated very many ways - both in the last few lec tures as well
as earlier - (Llewellyn Smith ' s sum rule (Equa tion 3 3 . 6 ) , the spin sum rule
for g 1 p - gln e tc . ) by high energy experimen ts would confirm at once the reality
o f the regular ities interpre ted by the low energy quark model . This would
make firm a conj ecture of deepes t significance to understanding high energy
physics - the importance of quark quantum numbers .
Supposing for a moment this is done , then the next serious ques t ion will
become theoretical - what exactly is the relation of the quark qualities at
high energy and those at low energy . The "partons as quarks " model does no t
imp ly the low energy model ( i . e . why are the wave functions not more complicated
involving quark antiquark pairs ) nor vice vers a . At present their relation
would not be understood . To s tart working on this now will take a little
courage - you might was te your t ime - maybe partons as quarks will not be
confirmed . I f you do start , poss ibly one place to start might be to think
about low energy matrix elements like � � p + y in a fas t moving sys tem in
which all (or some) momen ta are of the order P so parton wave functions can
be used . (We have one relation of this kind in Bj orken ' s sum rule for g ,
Equation 33 . 16 . )
Finally i t should be noted that even i f our house o f cards survives and
proves to be right , we have not thereby p roved the exis tence of partons . The
final resul t of our considerations has been to describe the result of the
J j p> (for large v , -q 2
1J
ope ration o f a curren t on a proton s tate • 2Mvx) as
conmutation rules for quanti ties , curren ts , defined in general whe ther par tons
"exist " or no t) .
From this point of view the partons would appear as an unnecessary
scaffolding that was used in building our hous e of carda .
On the other hand, the partons would have been a use ful psychological
guide as to what relations to expe ct - and i f they continued to serve this
270 Photon-Hadron Interactions
way to produce o ther valid expe ctat ions they would of course be gin to become
" re al " , possib ly as real as any other theore tical s t ructure invented to
At any rate we shall see . It is good to have s ome thing to look forward
to.
Appendix A. The lsospin of Quark Fragmentatior
Products
APPENDIX A
The discuss ion (Lecture 56 ) lead ing to the i dea tha t addi tive
no ted that what ho lds fo r 3-isospin holds also for any o ther component
s uch as 1-isos pin or 2-iso spin (al though of course in practice they are
wh ich at firs t seems impos s ible - except that we have an inde f inite
ln this mo de l s uppose quarks carry only iso spin 1/2 and hadrons are
271
272 Photon-Hadron Interactions
makes hadrons
(pions )
of type
12
antiquark of index y and a quark o f index 6 form a n characterized by
vector V is then p roportional to the yo matrix element of the two-
by-two matrix o • V where o are the Pauli mat rices . Write this as < y l o V l o >
· .
This is o f course
(Al)
Having the amp litude ( in an su 2 invarian t fonn, o f course ) we can
ask many quest ions . The re lat ive probab ility o f finding any con figuration
P (V1 . . . v > • rr
N
(A2)
where pa ' P a are 2x2 dens i ty ma tr ices co rresponding to the s t ates ,
number o f pions , s ummi n g over the charac ter o f the res t . In fac t we
shal l do two cases ; s ummi ng ove r all to ge t the normaliza tion o f our
�� �
In any case a sum over an unobserve d means a
)
sum on V over 3 pe rpendi cular values , symbolized by ·· · (A · V ) ( B . V)
"""
i:...., (o ·V+1'r ) (a+a · +B) (o . +v) • Ja -o • +B (AJ)
v
'71 •
N
3 ab + (-l)
N
(A .B) (A4 )
N
For large N this is almos t exac tly J ab and so we shall divi de by
Pk (Vk) •+
3 ab I v 1 • . .
Tr E8 (a 0V1 } . . • (a ·Vk ) • • • (a · V } p (a ·V/ )
N a • • •
VN
except Vk
(cr · Vk
*
> . . �
. (cr · V 1
from which any question about the one particle dis tribution func tion can
now be answered. For example , sum over all possibili ties for Vk and check
normalization ( for large N) .
+
The mean pro b ab ili ty th e k t h meson i s 11 + is Pk Tl ob taine d b y se t t i ng
11
... 1
a ·V • � (a x+iay ) in
put cr V
(A5 ) . For • = l (a x -icry ) •
12
Thus the
12
mean isospin of the k th hadron is PkTl - Pk!!-
+
-icrx cry
• a term like (A5 ) with
... :t- *
a · Vk • • •
a · v k replaced by · · ·
(- r
o lus icry · · · cr x . No te first us ing
this is 2 a z and hence gives � t B/b ,
-l
only the one coming from p
a
likew ise the term in cr • A only contributes if cr is in the z direction .
We find
(A6 )
which is very smal l . Thus the plateau region has become neutral .
+
Pk 11 - P k11 can only avoid being small if either k is small (i . e . near
the S end) or near N (i . e . near the a end) . In the former case neglecting
terms of order 3 -N we have
(A7 )
F inally the total z-iso spin quantum number of all the pions to the
+
11
{P
k
the z-component quan tum number of the le f t -moving quark! The sum on k
sum is of the s ame orde r ; the same as if we s umme d the left term above
the pla teau we s top , it is only necessary that we s top the sum on k at
the quark h as total isos pin 1 / 2 , and not higher . We leave to the reade r
APPENDIX B
sect ions are the eas ies t tes t o f quark quan t um numb e rs , we give an
analysis here .
the Fermi cons tant and s the s quare of the center of mass energy . For
f
is
l
1 -
2 x (d + 3 u) dx
0
the fac t o r x coming b e cause the cross s e c tion varies w i th s . Fo r
a nucleon i s
2 76
Appendix B. A Test of Partons as Quarks 277
J
1
-
- x (d + u + 3
1 -
(d + u) ) dx .
0
the anti-neut rino cross sec tion is
a • f (a vp + avn ) - f x (d + u + t (d +
u} } dx
0
S ince d, u are pos itive , but undoub tedly less than d , u, we see tha t
Th e sum i s
J
1
• x (u + u + d + d) dx
a +a
t
0
J � J } J
1 1 1
x ( u + u + d + d ) dx + x (s + s) dx
ep en
x(f + f ) dx -
0 0 0
J
Experimen tally thi s in tegral is 0 . 31 so if we could forge t the
• • 0 . 74 .
- 4 9 -
integral x ( s+S' ) dx we wo uld have a+a 3 · 5 ( . 31) Rut s+s
mus t surely be le as t han u+ii and d+d and , when weighed by x, s urely
much less . It would be hard to manage to make inclus ion of the las t
tes t of our parton quark mode l : a+a canno t exceed O . 74 and ye t almo s t
65-66
Conservat i on o f current , 5 , 1 3- 1 5
279
280
c onserva t ion , 18
L i gh t cone algeb ra , 1 7 8- 1 8 1
P ion ab s o rp tion 6 3- 6 4
R Regge p o l e formula , 73
Resonances , 32-34
S ma t r ix , 4
S eagull s , 13
Ve c t o r mesons , 8 2 - 8 5 , 88-89
w W e e p a r t on s , 232 , 234