Professional Documents
Culture Documents
(Frontiers in Physics) R. D. Field-Applications of Perturbative QCD-Addison-Wesley (1989)
(Frontiers in Physics) R. D. Field-Applications of Perturbative QCD-Addison-Wesley (1989)
Perturbative QeD
R. D. Field
Department of Physics
University of Florida
1\11 fights reserved. No part of this publication may be reproduc ed, stor d In <I r trioval
' y tom, or transmitted in any form or by any means, electronic, mechanical, photo-
copying, recording, or otherwise, without the prior writton permission 01 th publi hor.
Printed in the United States of America . Published simultan ou Iy In nn, do .
Field , Richard D.
Applications of perturbative QCD / Richard D. r lolel.
p. cm. -(Frontiers in physics; v. 77)
Includes index.
1. Quantum chromod ynamics . 2. P rtllrb Ilion (011111111111 dylll II IIh: ,)
3. Hadron interactions . I. Title. II . erlo . .
QC793.3.Q35F54 1989 539.7'2- dc 19 89-138
I I::lN 0-201 1-1 ::> !:>-3
Frontiers in Physics
David Pines/Editor
Volumes of the Series published from 1961 to 1973 are not officially num-
bered. The parenthetical numbers shown are designed to aid librarians and
bibl iographers to check the completeness of their holdings.
Titles published in this series prior to 1987 appear under either the W.
A. Benjamin or the Benjamin/Cummings imprint; titles published since 1986
appear under the Addison-Wesley imprint.
iii
Iv ,·',,,,,11 ,'1/1 nl'hYIII'I'
(:17) It. 1'. Jo'Cy llll lll1l Phuto II ad rOil IlI tt' r actio ll s, 1!)7£
(:IH) I';' It. Caialli(" lI o CO lllbill ittori c::l i.l IH.1 1l ' 1I0 rlllali'li1.ti o ll ill Q uall turn
Fi eld Theory, 1973
(:\D) : . B. Field , H. Arp , T he Redshift Cont roversy, 1973
and J . N. B ahcal!
(/10) D. Horn Hadron Physics at Very High Energies, 1973
F . Zachariasen
(1 I ) S. Ichimaru Basic Principles of Plasma Physics: A Statistical
Approach, 1973 (2nd printing, with revisions,
1980)
(12) G. E . Pake The Physical Principles of Electron Paramagnetic
T. L. Estle Resonance, 2nd Edition, completely revised,
enlarged, and reset, 1973 [cf. (9)-lst edition]
Volumes published from 1974 onward are being numbered as an integral part
of the bibliography.
13 R. C . Davidson Theory of Nonneutral Plasmas, 1974
11 S. Doniach Green's Functions for Solid State Physicists, 1974
E. H. Sondheimer
15 P. II. Frampton Dual Resonance Models, 1974
16 S. K. Ma Modern Theory of Critical Phenomena, 1976
17 D . Forster Hydrodynamic Fluctuations, Broken Symmetry,
and Correlation Functions, 1975
48 A . B. Migdal Qualitative Methods in Quantum Theory, 1977
49 S. W. Lovesey Condensed Matter Physics: Dynamic Correlations,
1980
50 L. D. Faddev Gauge Fields: Introduction to Quantum Theory,
A. A. Slavnov 1980
51 P. Ramond Field Theory: A Modern Primer, 1981 [cf. 74-
2nd edition]
52 R. A. Broglia Heavy Ion Reactions: Lecture Notes Vol. I: Elast.;c
A. Winther and Inelastic Reactions, 1981
53 R. A. Broglia Heavy Ion Reactions: Lecture Notes Vol. II, in
A. Winther preparation
54 H. Georgi Lie Algebras in Particle Physics: From Isospin to
Unified Theories, 1982
55 P. W. Anderson Basic Notions of Condensed Matter Physics, 1983
56 C. Quigg Gauge Theories of the Strong, Weak, and Electro-
magnetic Interactions, 1983
57 S. I. Pekar Crystal Optics and Additional Light Waves, 1983
58 S. J. Gates, Superspace or One Thousand and One Lessons in
M. T . Grisaru, Supersymmetry, 1983
M. Rocek,
and W . Siegel
r,ll II N ( :11 1111 S." .. i SI II'Iil l' 1.1. , A 1/t4'l II li N illld '1'1" ,1 1' li.'III'('!"' "
t.a l.io II.'! , WH·I
(iO (; . ( ;, I{o.'!.'! (; ralld UII IIit-d Tlwol'k.'!, 1!)H4
(; 1 S. W. !.ovI'sey Conde nsed Malll'r Physics: Dyn amic Corre la-
tions, 2nd Editio n, J986
(i2 1'. II. Frampto n Gauge Field Theories, 1986
li:1 .J. I. Katz High Energy Astrophysics, 1987
(;4 T. .I . Perbel Experimental Techniques in High Energy Physics,
1987
(i!) T. Ap plequist, Modern Kaluza-Klein Theories, 1987
A. Chodos, and
P. G. O. Freund
fii i G. Parisi Statistical Field Theory, 1988
()7 It C. Richardson Techniques in Low-Temperature Condensed
E. N. Smith Matter Physics, 1988
(iK .J. W. Negele Quantum Many-Particle Systems, 1987
II. Orland
til) E. Kol b The Early Universe, 1989
M. Turner
'to E. Ko lb The Early Universe : Reprints, 1988
M. 'Il lrller
'f I V. llarger Collider Physics, 1987
H. ,I . N. Phillips
'I: ~ T. ' I ;~iim a Computational Plasma Physics, 1989
7: 1 W. Kruer The Physics of Laser Plasma Interactions, ]988
'/fl I '. Ha mond Field Theory: A Modern Primer 2nd ed ition, 1989
[cf. 51-1st edition]
7r, B. F. Hatfield Quantum Field Theory of Point Pa rt icl('s and
Strings, 1989
7(i P. Soko lsky Introduction to Ultrahigh Energy Cosmic Ray
Physics, 1989
77 R. Field Applications of Perturbati w t~ C I) , I!)H!)
/
Editor's Foreword
vii
I )11 rI II/: I hI ' pa st dl 'cal ll' , pI '1'1 llri 1111 lVI ' 1111 11 11111111 t'iIlIlIlIOtiYII:IIIII('S «,,1( 'I)) It I:{ l 'IIII "W'ti
as a 11I;\iOI' Ik ld of study fo r physit'l. Is worlwll: III 1'!1·IIWIII.:lI'Y pal'l.lI'lI- pllysl('s; il lia s
sigll i lk:l llt applications in nUlI'r suh IIl'Id.-; of physi ('s, illeludillg heavy iOIl \'ollisioIlS ,
('os iliology, alld l'('lativislic astr·ophysi('s. An illtl'oductory account of QeD is thus
('spI'l'ially tim<'iy and l~i c k Field is unusually w ell-qualified to write such an account.
'Ihgether wilh Wchard Fcynman at Caltech he developed many of the basic ideas of
{lCI>, while during the past decade he has lectured on these and subsequent devel-
OPIIII' lIts ill summer schools and in one- and two-semester courses at both Caltech
,lIld the University of Florida. I share his hope that his book will benefit theorists
alld ('xpcrimc ntalists at every level of experience, from the graduate student with a
h('ginning understanding of Feynman diagrams to the experimentalist who wishes
to carry out his own QCD calculations. As a sometime tennis pal, it is a special
pleasure to welcome him to the ranks of FIP authors.
David Pines
I il'ha lla, Illino is
Man ' h, IH89
/
Preface
I have attempted to provide a book at the level of the first volume of Bj o rk,, "
and Drell but on perturbative quantum chromodynamics (Q e D) ra th er I.h ","
quantum electrodynamics. The book contains very little form al field th e ry. I
start with the Feynman diagrams of QeD and attempt to t each QeD by I j Ilg
calculations. In doing the calculations in this book the reader will no t Ill y
develop calculational skills and learn mathematical techniques, bu t T ho pe
gain an appreciation and understanding of the theory that m ight be m iss >d
in a formal field theory course. I believe that by working through this boo k a
student with a rudimentary understanding of Feynman diagrams will develo p
the skills and understanding necessary to perform individual research and will
be able to contribute to the field of perturbative QeD. Furthermore, I feel th e
book should be beneficial to both theorists and experimenters. Hopefu lly wit,J 1
this book students interested in experimental physics can learn to perf0 1"11 I
their own QeD calculations.
QeD is a precise and complete theory of quarks and gluons wh ich pil I'
ports to be an ultimate explanation of all strong interaction experim en ts 1\1,
all energies, high and low. There are many reasons to hope and expec t it t.o
be right . The question is, is it indeed right? Mathematical complexity haN,
so far, prevented quantitatively testing its correctness. The primary obst. ruc-
tion is the fact that the fundamental quarks and gluons of QeD apparelltl y
cannot be isolated as free particles, but are always confined within had roli H
by strong forces not amenable to treatment by perturbative m ethods . N v-
ertheless, because QeD is an asymptotically free theory, interaction forc H
become weak at small distances (large energies) and calculations using per-
turbation theory and Feynman diagrams are possible. Unfortunately, m ost,
processes involve both low and high energy aspects, and one must separat
the low energy pieces, which are not calculable by perturbative methods, frorn
the high energy perturbative parts. The nonperturbative (low energy) pieces
are parameterized, taken from data, or a model is built to describe the regim e.
xi
" .. ,'" "
II, w O\lld tILl c' (Jilly 0 111' 1"('1'1.'/(' ('u lllp l ,rl Ho lI wlt,lI d Id , t 1,0 " p rovc' " (l( :1
correct. . Il owevr r, " tr\l( ~ tes t.s or (l('rt.llri> ll t.iv(' Qel)" 0 1'1,(' 11 1.111'11 0 111, 11I('I'(' ly as
tests of t.he a utho rs' I V('rII('SS ill pnrH,llIderizill g t.h e II np ' rtllri>alivc un cal-
culable part of th e prob lem a nd not as actual tests of Q '0 . G r a t ca.re mu st
be taken in examining the sensitiv ity of predictions to th e uncalculable parts
of the problem. The belief that perturbative QCD is correct comes from the
fact that the theory correctly predicts the approximate behavior of a wide
class of experiments. It provides an understanding of why the "naive" parton
model works so well (weak interaction forces at short distances) and predicts
deviations from the parton model that are seen experimentally.
In Chapter 2 and 3 of this book the QCD perturbative predictions for
electron-positron annihilations are examined. One can learn a great deal about
perturbative QCD by studying electron-positron annihilations. Since this pro-
cess involves no color in the initial state, it provides an excellent "theoreti-
cal laboratory" in which to develop the tools of perturbative QCD. Calcula-
tions will be performed using two different regularization schemes; the "mas-
sive gluon scheme" and dimensional regularization. There are lessons to be
learned in both schemes and by comparing the results one can see clearly
those quantities that are regularization scheme dependent and those that are
not . Throughout the book we will always compare the "QCD" result with the
"naive" parton model.
Deep inelastic lepton nucleon scattering will be covered in Chapter 4
and Chapter 5 covers the large-mass muon pair production in hadron-hadron
collisions. Here we will use the techniques learned in Chapters 2 and 3 and
again all calculations will be performed using the two regularization schemes.
Chapter 6 is a bit more formal and will cover in more detail renorm alization
and the running coupling constant of QCD which is only briefly discussed in
the introduction in Chapter 1.
Chapter 7 covers applications of perturbative QCD to the prod uction
of particles and "jets" in hadron-hadron collisions. Much has been learned
about QCD by studying hadron-hadron collisions that involve large transverse
momentum or result in the production of a large mass object .
In Chapter 8 I present several other applications of perturbative QCD.
Here I do not go into quite as much detail as in the previous chapters . However,
after working through Chapters 2-7 the reader should be able to perform
the calculations presented in Chapter 8. At the end of each chapter I have
provided a modest reading list and references . My list is not as extensive as,
for example, the book by Quigg and I refer the reader to his book for a more
complete reading list.
I have provided a set of appendices containing information that should
be useful in working through this book . They contain formulas that I have
collected through the years starting when I was a Ph.D . st udent of Professor
J.D. Jackson at Berkeley in 1971.
/ 11 111
'I'hi " hoo k i" It II (j 'l l.p,l ow l.l, o f' 1.111' " 1111111 01 ' 1' 1-w hl/o l 1<-<'1.111" '/1 I ).!,ItV'· 1\.1, 1. 11
.Jo>l lll ( 1 ~)7~), n IIld,'r ( 1\)7\)), 11.1101 SLJ\ • ( I USG) Itlld II. 1I, 1' 11.dIl U I. (' (,O II Hi(' I }!,,' v,·
Id, CA LTEe II (J 980) and ILL th e UIlivcrsity of Fl o rid a ( 108'1) . 'I'h . C J\ I.' I' 1':<:II
'''l in,,' was t.wo se mcst I'S in length a nd .r was ab le t.o CO VC I' all til ' Ill ateri a l ill
( :hn.pl.ers 1 through 6 of this book. T he F lor id a course was o ne se l11 c!> l.r r ill
1!'I1 ~ l.h a nd I covered only selected topics throughout the boo k. T he h o k iN
Imil.ah le for a one or two semester special topics course in high -energy phY Nicl'
o r it can be used selectively to supplement a course on relati vi st ic qll ltlltlllil
IIl c('.ha nics. In addition the book can be used for reference and selr-s tlldy by
hol.h th eorists and experimenters.
I. could not have written this book without the six years of c 11 11.\) 01'11
I.ive work I did with Professor Richard Feynman at CALTECH . Ma ll Y of 1,)1 ('
di sc ussions I have presented here are an outgrowth of our work or a r l UI (1 11 1.
~ ro wth of things I learned from Professor Feynman . I am very th a.1I1 fill 1'01'
t he pri vilege and opportunity of working with Richard Feynman. I ,till II.IHO
g rateful to Geoffrey Fox and Stephan Wolfram both of whom taught 111 i.I lo t.
while I was at CALTECH and both have made contributions to thi s b ,1<.
I would like to thank the teaching assistants of the course I taught at 'A L-
TECH (1980), Rajan Gupta and Steve Otto. They provided soluti ons fo r a ll
the homework problems I assigned throughout the course and th ereby '0 11 -
I. ri buted greatly to this book.
I am grateful to the Aspen Center for Physics for the warm hosp it a lity
~ hown to me during the summer of 1987 where I wrote the first three cha.p ters
of this book. I also appreciate very much the hospitality shown to me by It
Schrieffer , S. Ellis, and A. Mueller at the Institute for Theoretical Physi('H
at Santa Barbara where I attended a six months workshop on "QCD a lld its
Applications" in Spring of 1988. I learned a lot at the workshop th a t h Ip ed
me to write this book.
I would like to thank my tennis pals (Aimee, Amanda, Dee Dec, G 'O I' JJ;",
Howie, Jason, Jeff, Jimmie, Joyce, Lorna, Mike, Mike, Pierre, Rick, a lld n,OIl )
for helping me keep my priorities straight.
Finally, I am grateful to my wife Jimmie for taking good care of I'll,' Il lid
our three children, Jason, Aimee, and Amanda, while I worked on til is boo k.
R.D. Field
Florida, FaJl 19813
/
Contents
xv
Vl I ' OnH 'llI l
n,:s V (' I' I,( 'X ( :W ' I' ('c loi O Il /l /' 1
fi .'1 'I'h(' n.lIllllill f!, ( ;ollplill g CO ll st allt
n,r> ' I'll<' Beta FIlIlct.i o ll
(U i Th e R enorm aliz al.i oll G roup Eqll ati on
Probl ems '2 f):~
Further Reading '2:)H
References '2:)1'
Introduction to QeD
1
In 1954 Yallg alld MillHI ("O Il Htrllrl.!'d 11. t1I1 '(HY 1l,lIlIl oII;O Il t-< t o Qllltlltltlil 10:1('("-
trodynamics (QED) but for a. $YH tCIlI ill whi ch til(' pa rticl es carry 1l10re than
one "charge." The generalization to two chargc$ Ill cdiated by a vector parti -
cle results in a theory that is inconsistent with observations. However, with
three charges mediated by vector particles obeying exact SU(3) symmetry
one arrives at a theory that seems to describe the strong interactions. In this
book we will learn about the nature of this theory by performing calculations.
However, let us first qualitatively examine the behavior of a theory with three
charges.
In the theory of Quantum Chromo dynamics (QCD) each quark carries a color
charge that can take on three values; R=red, G=green, B=blue. When the
field quantum (gluon) is emitted the quark color mayor may not change giving
nine ways of coupling a gluon between an initial and a final quark. Eight of
the gluons form a "color" SU(3) octet,
gl= RG g2 =RB g3 =GR
g4=GB gs=BR g6=BG
1 --
g7 = .j2 (RR - GG)
1 - - -
g8 = J6 (RR+ GG - 2BB) (1.1.1)
The singlet gluon couples equally to all quarks, and is independent of the
octet. Since there is no good reason to include it, it is usually ignored. The
octet of gluons and the three quark colors are assumed to form an exact SU(3)
color symmetry.
In QED, the strength of the coupling between two quarks is given by
(1.1.3)
where el and e2 are the electric charges of the two quarks in units of e (i. e.,
e,. = ~, ed = - ~, etc. ) and a is the fine structure constant,
(1.1.4)
/
II ~Hl(:1) (lf(:',I(lr :1
B B
B
c,cz=hk)(-i) =:
Figure 1.1 Color factors for the exchange of a single gluon between two blue,f) ,
quarks. The overa.ll color factor is the product of the factors for each vertex.
In QCD the situation is similar. The strength of the strong interaction cou-
pling for the exchange of a single gluon between two colored quarks is
where Cl and C2 are the color coefficients that can be deduced from (1.1.1)
an d
a.=9;/47r, (1.1 .6)
with 9. being the strong coupling constant. The extra 1/V2 at each vertex is
historical.
Following the discussion in Chapter 2 of the book by Halzen and Martill ~
let us consider, for example, the color factor for the interaction between tw
quarks of the same color, say B, as shown in Fig. 1.1. Out of the eight glUOIlH
only the one containing BE can contribute, 98 in (1.1.1). The color factor is
thus
(1.1.7)
(1.1.8)
R R R R
Figure 1.2 Color factors for the excha.nge of a. single gluon between two red, R,
quarks. The overa.ll color fa.ctor is the sum of the factors for each diagra.m .
B e
C1C2=(.fs)(-.ii)=-t C.C2=(+I)(-I)=-1
Figure 1.3 Color factors for the excha.nge of a. single gluon between a. blue, B , a.nd
a. blue-bar, lJ, qua.rk. The overa.ll color factor is the sum of the color factors for ea.ch
dia.gram.
Here all the colors occur equally so it is sufficient to consider just one, for
example BE interactions as in Fig. 1.3, and then multiply by three. For BE
interactions there are three possible gluon exchanges, gs, gs and g6, yielding
I ,'
(I.I , I())
T Ilt' Illinus signs come from the antiparticle vertices . Just as is t.h . caS(' III
q l'; () , <tlltiquarks have opposite charge to quarks.
As can be seen from (1.1.7) and (1.1.10) the color force can be both r(' plIl
/< IVC and a ttractive, respectively. Presumably, states with the largest (n cglll i 'III' )
w lor factors are the most tightly bound. For a color octet qij state si ngle glllOIl
4'xchange is repulsive with the color factor
(1.1.11)
or a ll the two body channels, the color singlet qij state is the most a l,l.r ll('
t.ive. Similarly, single gluon exchange in the color singlet three quark Nt.H,t." ill
st.rongly attractive with the color factor
-2. (J.I.I ~)
Since this analysis is based on single gluon exchange it cannot be taken too H(,
riously. Multiple gluon exchange and the triple-gluon vertex have been ign red
in a regime where lowest-order perturbation theory is surely not appli ·ahl e.
Nevertheless, on the basis of this naive single gluon exchange analysis, OIl('
might think it plausible that qij and qqq states only exist as color singl ts.
Color factors are generally not computed explicitly from the gluons ill
(1.1.1). It is easier to manipulate the SU(3) color matrices, T a , whi ch sati Hry
(1.1.1 :1)
where fabc are the antisymmetric structure constants of SU(3) given ill A p
pendix D . The quark-gluon vertex in Fig. C.3 is given by
-ig. 1';j 11-" ( I . I. 1'1)
where the indices i and j correspond to the quark colors and take on vallH'N
from 1 to 3 while "a" corresponds to the gluon color states and takes on va llll'H
from 1 to 8. The eight 3 x 3 SU(3) matrices, T a , are given in (D.1.3). TIH'
color factors in (1.1.7) and (1.1.8) are given by
(1. l. I.'))
The starting point of this book will be the Feynman rules for QCD given in
6 CI'l\l't('f I 1111.,011,1< tin"
e e
(0)
e e
Q Q
( b)
Q Q
Q Q
(c)
Q Q
Figure 1.4 (a) Lowest order vacuum polarization correction to the electric charge.
(b) Lowest order correction to the strong coupling due to a virtual quark-antiquark
pair. (c) Lowest order correction to the strong coupling due to a pair of virtual
gluons.
Appendix C. Given these rules we will perform many calculations and learn
about perturbative QCD by doing perturbative QCD. We will encounter both
infrared and ultraviolet divergences and develop schemes for handling them .
We will begin in Chapter 2 by studying the perturbative QCD corrections to
e+ e- annihilations. Pedagogically this is a good place to start. It is a good
"theoretical laboratory" to develop the tools necessary in analyzing other
processes. We will not examine in detail the ultraviolet divergences whose
renormalization results in the running coupling constant of QCD, a.(Q2),
until Chapter 6. For now I will simply outline the behavior of the effective
coupling in QCD and contrast it with the QED case.
/
I ,' "
III (,>10:1) 1.1" , Ht "'''/-" th or 1.11<' cO llplill g lillii'l l, h. , d"I."I' llllllI'd ,'x P"I'II""ld" t1l y
" 111'1' "'" 111 1d, w,' d, 'cl( l,' to "X IH'l' illH'lIt u ll y 1I11'1t.H III·' · 1.111' Hl.l'( ' lI /1, l.h o f "" '(' I,ro
1" "" .11,1.1' 1I1I," I·ll,r l.i o li s by H(·:r. I.I,('I'i li g a ll 1' ll'cl. l'OIi off 11.11 ,,1('C tl'OIi !tlld (,O IlI(lIl.l'ill).l,
III! 1" 11 1111, wit h :I. I.h ('O l'l'l. ica l ('alr lllati o li . The l'a.l.e 1'0 1' (' 1(' (, 1,1'0 11- (,1('(' 1.1'0 11 HI'I I,t
l 'I IIIf'. 1/1 Id('cd,('d by th e pr('sc nce of va IIIIIn p o litri~al,i o n li agmili s lik,' thl\l,
111,,\'\1 11 III Fi g. 1.'1 . 0 111' imrrt di atcly runs into tro ubl e, however, sinc(' th ' O il. :
, I, ,1 1" " loo p ('{ rr('d io n to th e photo n propagator diverg s like 10g( A) , wh"rI '
\ III 1111 "1" 1I11,!'!wio\ct. cutoff th a t can be a rbitrar ily la rge. In purti culHr , I,lrl'
1"111111)1, mdl' l' hubbl e co ntribution is
( I .' , I )
/I I I "
' I'lrc procedure for handling the ultraviolet divergences like those appeari ll ~ il l
/1('1 2) in (1.2.1) is called renormalization, One defines an experimenta l ckd,n c
charge, a, by the large distance behavior of the electric potential (Th o"IpHO Il
limit)
( I. VI)
wh ich experimentally is about 1/137, All results of calculations ar now ('X
pressed in terms of the experimental coupling, a, rather than the uno bs rvahl,·
bare coupling, ao. In terms of a, the effective coupling is given by
(1.2 .7)
where the quantity (B(Q2) - B(O)) is now independent of the artificial ultra-
violet cutoff A, The cutoff A is now sent to infinity while holding a constant.
From (1.2.1) and (1.2.7) we see that the large q2 behavior of the effecLiv
coupling is giv /I hy
2 _ 2 a
a QED (Q ) = a eff( Q ) = 1_ (a / 37r) log(Q 2/ m~ )' ( l.2 .8)
(0 ) ~pholon
02
~gluon
02
( C) ® ~gluon
02
~gluon
02
Figure 1.5 (a) Illustration of how vacuum polarization in QED will shielda p OAltiv('
bare charge. (b) The same shielding as in (a) but for a red change in QCD. (c) ShOWN
bow in QCD a red charge can radiate away its color via a RB gluon and b CCOlII( '
blue.
I
I
I
QQEO(Q
2
)1I
N I
g /
o
-., C\.. /
/
~ /
E /
- /
G.>
.!:
c /
o
u
/
./
"asymptotic
freedom" - 0
Figure 1.6 Illustration of the behavior of the QED and QCD effective coupling
constants as a function of the energy scale, Q2. In QED the effective coupling,
Cl'QED(Q2) , is small a.t small Q2, but becomes large a.t large Q2 (i.e., short dis-
tances). In QCD the effective coupling is large at small Q2 (i.e., large distances)
where confinement occurs, but decreases to zero at large Q2 (asymptotic freedom) .
\
Fig. 1.6. This means that for QeD, perturbation theory should work well at
high Q2 (short distances) but break down at small Q2 (large distances) where
a.(Q2) becomes large and presumably confines quarks within hadrons.
The nature of the QeD coupling constant a.(Q2) takes a bit of getting
used to. In QED it is easy to define what one means by the charge of an
electron e. It is related to the large distance behavior of the electric potential
(Thomson limit). One cannot do this for QeD since the Q2 - 0 limit of
a.(Q2) cannot be calculated by perturbation theory. Instead we had to choose
an arbitrary point JJ and define a, to be the effective coupling at that point,
a, == a.(JJ2). (1.2.14)
It, however, does not matter which point JJ one chooses. If one chooses instead
the point fi then the two couplings are related (to lowest order) by
III
(1.2 . 1())
rhll ll , 1.11(' " I' <tI" pa rameter ill the theory is no t a(J.1.2) or J.1.'2 but ra.t her a 11I 1I.'lH
rll rd, ' ,A, that is ind ependent of J.1.2 and is given to this o rd er by
2 1 2
10g(A ) = -=------(
aa J.1.
2) + 10g(J.1. ). (1.2 . 17)
wlt.h
( I. ~!.I \I)
II, is interesting to notice that (1.2.18) becomes infinite at Q2 A 2 whi (' h
m l'l'esponds to a distance of about 0.5 Fermi for A ::::: 500 MeV. (This i.. (' x
1,1't'lIIely crude, however, since (1.2 .18) is not applicable when a.(Q 2) beco nll'H
large. )
The physical reason for the behavior of a.(Q2) is illustrated in Fi g. 1. 5.
Quark-antiquark vacuum polarization shields the color charge as was th e CH.se
ill QED. However, now since the source can radiate charge (i.e., change froll l
I'cd to blue by emitting a red-blue-bar gluon), the charge is no longer located
a.t a definite place in space. It is diffusely spread out due to gluon emi ss iOIl
a nd absorption. As one increases the Q2 of the incoming gluon probe, I,h r by
looking at smaller and smaller spatial distances, it becomes less likely to find
the "charge" (red in Fig. 1.5) , This latter effect is stronger than the fo nliN
and the effective charge thus appears weaker and weaker as the Q 2 of till'
probe increases,
In Chapter 6 we will see that once we have calculated an observabl e Il sin .u;
Feynman diagram techniques with a coupling g. defined according to
then the correct procedure is to simply replace a. by the effective couplill tJ;
a. (Q2) in (1.2.18). By this procedure one automatically incorporates a n itp
propriate infinite set of diagrams into each calculation. It should be rea.l
ized, however, that in leading order the perturbative parameter A is arbitrary.
Changing the parameter A corresponds to adding a constant C as follows:
2 471'
a.(Q ) = (Jo [log(Q2jA2) + C]
471' 471'C
(1.2.21 )
= (Jo log(Q2 j A2) - (Jo log2(Q2 j A2) + ' ",
The constant C produces changes at order a. 2(Q2) . To leading order (1.2.18)
and (1.2.21) are equivalent . In a similar manner changing ones definition of
IZ C llllpl'('1 I 11I1,IOdll l l,loll
Q Q2 OLa o.
2 4 0.327 0.247
4 16 0.252 0.196
10 100 0.193 0.155
31.6 1,000 0.149 0.124
50 2,500 0.137 0.115
100 10,000 0.121 0.103
500 250,000 0.096 0.084
(1.2.22)
To order 0La,
2 2 [ 1306-38nJ ]
o.(Q )=OLQ(Q) 1- 41l' 33-2nJ oLQ(Q2)loglog(Q2/A2) . (1.2.23)
dll ' ,"1' likl ' It p o w"r (I" I !II ~ /(i') IT lal,iv(! t.o 1.111' Iof!, II ri till lIi c' cOI'n'(' I.iiJll Ill'
1" '1I.lIrlHlI.i V(' QC I) . '1'111'1'( ' is 11 0 flYSt. ' lIIaLi way Lo ('0111 PIII.(' Lilh' ,· " II i)l;llI'r
t.W IIl I. " ('0 I'I'Cc t. i ti S alld t.ll ey illtro du cc a furth er un cf! rl.aillt.y ill d ·1.(' rIIlillill )l; /I.
'1'111' llit.llat.i OIl is b est summarized by th e limiLs1
wiI.1I (y~ given in (1.2.23). Table 1.1 lists the value of the leadi ng o rd r QCI
,·tr, ·(' t.ivc coupling , O'LO, and the order O'to effective coupling in (1.2.23) V(~ rH II H
=
(J fo r A 200 MeV and nf 4. =
1'1'0 b lerns
1.1. Using the gluons in (1.1.1) show that the color factor
Vqiqj = ~ (PEX - ~) ,
where PEX is a color exchange operator with the property that pJ;;X = I I
for states symmetric in color exchange and PEX -1 for states an tisy III =
metric in color exchange.
1.. 2. Using the gluons in (1.1.1) or the color matrices in Appendix D compute t.1i\'
color factors for the following quark-antiquark, quark-quark, and tlne qll ,trk
states: (qq)[l], (qq)[S]' (qq)[3·], (qq)[6]' (qqq)[1], and (qqq)rsJ .
1.3. Using the commutation and anticommutation relations for the color matri r ."
in Appendix D,
[Ta,n] = iJabcTc,
{Ta, n} = ~c5ab + dabcTc,
verify that
1.6. Using nf =
4 and A = 200 MeV, compute the value of Q where the leading
order effective QCD coupling is equal to the leading order effective QED
coupling in (1.2 .8).
Further R eadin g
F.E. C lose, An Illtrodu ctioll to Q unrks flnd PltrtOIlS, A ·ad mi· I'r ss, N w
York,1979.
R.P. Feynman, "Gauge Theories," Les Houches, Session 29, North-Holland
Publishing Company, 1977.
S. Gasiorowicz and J .L . Rosner, "Hadron Spectra and Quarks," Am. J.
Phys. 49 (10),954 (1981).
F . Halzen and A.D. Martin, Quarks and Leptons: An Introductory Course
in Modern Particle Physics, John Wiley and Sons, Inc., 1984.
Chris Quigg, Gauge Theories of the Strong, Weak, and Electromagnetic
Interactions, Frontiers in Physics, The Benjamin-Cummings Publishing Com-
pany, Inc., 1983.
P. Ramond , Field Theory: A Modern Primer, Frontiers in Physics, The
Benjamin-Cummings Publishing Company, Inc., 1981.
References
1. C.N. Yang and R.L. Mill, Phys. Rev. 96, 191 (1954) .
2. F. Halzen and A.D. Martin, Quarks and Leptons: An Introductory Course
in Modern Particle Physics, John Wiley and Sons, Inc., 1984.
3. D.W . Duke and R.G. Roberts, Physics Reports 120, 275 (1985) .
4. W. De Boer, SLAC-PUB-4428, invited talk at the xth Warsaw Sym-
posium on Elementary Particle Physics, Kazimierz, Poland, May 25-29,
1987.
/
C IIAP'I' En 2
Electron-Positron Annihilations:
Total Rate to Hadrons
15
10 <;1. 111'(,,-" ' 1',.t,loI II ld,,' III. (, I ('
The vacuum ill Q '0 i~ all cX LI'(' III('ly ('olilp li cal.('d oi>j,·(' t,. Virt.ual pairs of
all existing elementary particles a.r co nsta.ntly bing cr atcd and til n die
after living the time allowed them by th e un certainty principle. The tim elike
virtual photon that is produced when a high energy electron and positron
annihilate provides an excellent probe of the vacuum . This virtual photon
will couple to any elementary particle that carries electric charge and if its
invariant mass, Q, is greater than the mass of the particle-antiparticle pair it
can transform the pair from virtual to real. When the pair are leptons (e. g.,
j1-+ j1--) they fly out and can be detected experimentally. When the pair consists
of, for example, a red quark-antiquark pair a remarkable phenomena occurs.
At early times after the quark and antiquark are created short distance QCD
dynamics dominates and the R quark and R antiquark interact only weakly
with each other since the strong coupling 0'. is small (i. e., asymptotic freedom).
In the center-of-mass frame the quark and antiquark begin to separate each
carrying momentum::::; Q/2. At later times the distance between the Rand R
color charges becomes large and the color forces between them increases. The
color force between the separating color charges becomes larger and larger as
time and distance increases until the vacuum breaks down and additional RR
quark pairs are created as illustrated in Fig. 2.1. These new quark-antiquark
pairs then combine with each other and with the original quark and antiquark
producing only color neutral (actually color singlet) hadrons in the final state.
Experimentally the original quark and antiquark can not be directly detected.
What is detected is the multitude of outgoing hadrons whose energies sum to
give Q. By examining and studying these outgoing hadrons we learn about
the quark and antiquark that initiated the process. Clearly from the point
of view of perturbative QCD we are only going to be able to describe the
early time (short distance) part of the phenomena where the QCD coupling is
small. Perturbative QCD can tell us little about the later time (long distance)
aspects of the phenomena. We will describe this long time, large distance,
behavior in terms of a simple parton "fragmentation" model.
One can learn a great deal about QCD perturbation theory by studying
electron-positron annihilations. This process provides an excellent "theoreti-
cal laboratory" in which to develop "tools" that can be used in performing
other QCD calculations, In this chapter we will study the maximum inclu-
sive process e+ e- --+ hadrons and compare it with the purely electromagnetic
reaction e+ e- --+ p+ j1-- , The annihilation of an electron and a positron into
hadrons contains no color in the initial state and we know from a theorem due
to Kinoshita!, Lee, and Nauenberg 2 that, at any order of perturbation the-
ory, if we sum (i.e., integrate) over all quark and gluon final states the result
is finite even for massless quarks and gluons. However, the total cross section,
for example, for the emission of a single real gluon, e+e- --+ qijg, is infinite.
This infinity is canceled by virtual gluon corrections leaving a finite total cross
section . To compute the finite corrections it is necessary to develop a regular-
ization procedure that will control the infinities that occur in the individual
real and virtual contributions. The two contributions are then added together
/
'.1 .1 Th(1 Nlllv I' ,tOil I' ,tOil Mml,,1 17
Figure 2.1 Illustration of the phenomena that occurs when a red quark-antiquark
pair is produced by a large invariant mass virtual photon in electron-positron a nlli-
hila tions . As time increases and the color charges begin to separate and the color
force between them increases. Eventually new quark-antiquark pairs are "pulJcd"
out of the vacuum and combine with each other and with the initial quark and
an tiquark to form color neutral (i.e., color singlet) hadrons in the final state.
and the sum is finite and independent of the regularization scheme. Before W{'
proceed with regularization schemes let us examine the naive parton mod ·J
expectations for e+e- - hadrons.
The amplitude for a virtual photon to "decay" into a 1'+1'- pair as shown in
Fig. 2.2a is given by
(2.1.1)
where Pi and Si are the momenta and spins, respectively, of the outgoing spin
1/2 fermions and where I have used the Feynman rules given in Appendix C. If
we take the absolute value squared of this amplitude and sum over final state
1M C hllpl.l1I ' '1'1)1,111 lI id,,, f... "I"
= e 2 tr(jY'fI'PI IV)
= 4e 2[(PI)I'(P2)v + (PI)v(P2)1' - gl'v PI 'P2], (2 .1.2)
where the masses of the particles have been neglected . The differential cross
section for the electromagnetic process e+ e- - t p.+ p.- shown in Fig. 2.2b is,
from Appendix A, given by
d(1
(2.1.3)
( b)
Figure 2.2 (a) Amplitude for the "decay" of a virtual photon into a Il+ Il - pair.
(b) Amplitude for an e+ e- pair to annihilate and produce a Il+ Il- pair by tilt>
production and "decay" of a single virtual photon.
=
dO em d cos Oem d¢em, where Oem is the scattering angle between the outg
ing J.l+ and the incoming e+. Integrating over the angles Oem and ¢cm giv I)
(2.1.14)
Replacing the J.l+ J.l- pair with a quark-antiquark pair as in Fig. 2.3 gives
41r 0 2
O'(e+e- - qij) = (3)"3 Q2e~, (2.1.15)
which is identical to 0'( e+ e- - J.l+ J.l-) except for the factor of 3 which comes
211
q, PI' 5,
Figure 2.3 Born amplitude for the "decay" of a virtual photon into a quark-
antiquark pair.
from the sum over the three quark colors and the factor e~, where eq is the
charge of the quark q in units of the electric charge e. Assuming that quarks
turn into hadrons with unit probability one arrives at the famous parton model
prediction for the ratio of the total cross section e+ e- -+ hadrons to the total
cross section for e+ e- -+ p+ p-. Namely,
Re+e- == u(e+e- -+ hadrons)ju(e+e- -+ p+p-)
n,
=3L>~;' (2.1.16)
i=1
where the sum is over all quark pairs that can be produced at the given
center-of-mass energy of Q.
It is convenient to visualize this process as an e+ e- pair which annihilates
producing a virtual photon, 'Y. which then "decays" into either a muon pair
. or a quark-anti quark pair (Fig. 2.2a or Fig. 2.3, respectively). From this point
of view we arrive at (2.1.16) by considering the two-body differential decay
rate,
1 -2 ~
dW = 2Eem IMI a-'R2 , (2.1.17)
cr'R2 = (21r)3(2Et)
d:Pl d=jJ2 ( )4 4(
(21r)3(2E2) 21r f> q -
)
PI - P2 . (2.1.18)
t,l I,ll(' dc('ay prodlldH 7" /llld IJ~ Illlbj 'ct t t l , ' (,OIH; trll,illl. I.hll.l. )JI I lJ~ If ,
Wltl'I'I' II is th' '1 -III ()IIII '1I1.11I11 or the virtual photOll, 'Y • . 1111. ~ gl'n.ti ll g OV('I' loll, '
:\ IIIOIII<!JItUITI of particl e '2 yi Ids
J 3 4
d p2 6 (q - PI - P2) = 6(Q - E1 - E2)' (2 . 1. 1D)
wll ere particles 1 and 2 have been taken to be massless and 01 alld 4) , lire'
I~ ilgl es with respect to an arbitrary choice of axis. Integrating over 0, 1I.lId ti"
yields
('2 . 1.'21)
The final integration over El is accomplished using the conservation of " 'rgy
Ii-function ,
(2.1 .22)
EldEI6(Q - E - E ) = El = ~
J 2E2
since for massless particles El = Q /2. Hence,
1 2 2Q 4'
(2.1.'2;l)
R2 = Jd R 2 =~, 6
871'
(2.1.2'1)
and since the matrix element does not depend on any of the variabkH of
integration the total decay rate is
1 -2 1-2
W = 2Q IMI R2 = 1671'QIMI , (2.1.'2f))
where 1M 12 is the appropriate spin averaged matrix element squared. For till'
"decay" of a virtual photon into a muon pair as shown in Fig. 2.2a we havl'
where gl'v is the metric defined in Appendix B. As I will explain in the next
22 ( ·It Ipt.l'l :.! '1'01." Hilit 1111 /' It '
secti on fo r a real pho toll til(' rcpl lL("(' IIH'1I1. ill ('2 . 1.'27 ) iH ('qUiV ltl(· 1I1. 1.0 SlIllIlIlill g
over the physical ( i . e., 17mls llc 1'se ) pola ri zatio n sl. a.l.('s . Fo r a virtu a l ph oto n
the replacement in (2 .1.27) results in a combin a tion of th e tra nsverse and
longitudinal cross sections which I will label by the subscript L. Namely,
2
=
IMI~ -L/J/J(P2;PI) 8e PI ·P2· = (2.1.28)
Combining (2.1.25) and (2.1.28) and using (2.1.5) gives
WerE -+ J-l+ J-l-) = aQ. (2.1.29)
Similarly,
WerE -+ qij) = 0'0 = 3ae~Q, (2.1.30)
and the ratio R = WerE -+ qij)/W(ri; -+ J-l+J-l-) is the same as that arrived
at in (2.1.16). I will call the rate the Born term, 0'0, and the amplitude
Ao = -iee q u(p2, S2)rI'V(PI, St}t/J(A), (2.1.31)
shown in Fig. 2.3 the Born amplitude.
Before proceeding to calculate the QCD corrections to the Born cross section
in (2.1.30) I would like to discuss techniques for summing over photon and
gluon polarization states. For the case of a real massless photon (or gluon)
propagating along the i-axis with 4-momentum kl' given by
(2.2.1)
(2.2.2)
T he repl ace ili elit ill (2.2 . 1G) wo ri H for [ til ILrbit.mry 1111 II ai )('r o f ('xl.<'I' ll(tl [('al
massless photons. It a lso wo rk s for t.lae cas' of 0 11. (' rcal exl. l'II a l IIl asslcss gillo n.
However, it does not work fol' proce, ses with two or more ex tern a l gluons in
which the triple gluon vertex is present . In this case the replacement in (2 .2.16)
leaves extraneous unphysical spin states which can be removed by including
diagrams with a fictitious "ghost" particle (see Appendix C). The couplings of
this ghost particle are adjusted to precisely remove the unphysical spin states
which arise from the replacement in (2.2.16). We will see an example of this
when we study the process 9 + q -. 9 + q in Chapter 7.
For a massive boson with 4-momentum given by (2 .2.1) (in this case
k5 = k~ + M2) one also has the longitudinal polarization state
(2.2.17)
It is easy to show that the three polarization states of a massive boson are
correctly summed over by tlie replacement
and
(2.3.2)
where I have used the Feynman rules from Appendix C and where
Pa = P2 + P3, (2.3.3)
Pb = Pl + P3, (2.3.4)
/
~.3 II hi (lIu"" " .111 If. 0 11
AR = (-ieeq)fL )
Figure 2.4 Amplitudes for the three-body "decay" of a virtual photon into a qua.rk-
antiquark pair and a gluon.
(2.3.14)
ii d:P1 d:P2
2E1 2E2
= (411")(211")
4
E dE E dE
1 1 2 2
11 -1
d
z, (2.3.17)
where z = cos 812 is the relative angle between particle 1 and particle 2 (we
can pick the i -axes in the direction of particle 1). The integration over z can
be carried out using
o(Q - E1 - E2 - E3) _ c[( )2]
- u q - PI - P2 - P3
2E3
= 0[Q2(1- Xl - X2 + ~X1X2(1- z))] ,(2.3.18)
resulting in
(2.3.19)
so that
(2.3.20)
At this point we cannot integrate the phase-space factor d 2 R3 any further since
/ ."
II", 1I1 !1.I.ri x d (' III(' II1. ill (:l .:1 .7) wi ll depell d ill g<' II (' r ld Oil J; I 1I.lId J;~. (:() lIlh illill~~
(:, :1 7) Ilild (2.:3 .£0) giWH 1.11<' difrercnti al de 'a.y rat(!
Q -2
dW = 32(271")3 IMI dXldx2,
wh<'l'e the subscript E refers to the fact that virtual photon polarizatioli ll tnl., '
have been summed by using the replacement in (2.1.27). The first I. ' 1'111 ill
(:l.:l.23) is given by
g2e 2e 2
5 22 = =
IARI~ ~ tr(Ta Ta) tr(12rv'Par/J'Ar/J'Parv)
Pa
g2e 2e 2
= ~(4)16 (2PI'Pa P2'Pa - PI'P2 Pa 'Pa)
Pa
= 32e2e~g; (:!)
_ 32 2 2 2 (1 - X2)
- e eqg. (1 - xI)' (2 .:1 .2'1 )
2 2 2 (1- xt)
= 32e eqg. (1 _ X2)' (2 .3.29)
and
CII \,pl,I1I 'J. ' 1'01.111 Il ld,,, liII (, I f'.
= 3'2c 2 e 2 q; ( (I - ~ - :2 - ~) . (2.3.30)
q. . XI)(I - X2) (1 - 3:d (I -X2 )
The three pieces Su, Sn and S 12 are gauge depend ent with the results above
holding only for the Feynman gauge (which is implemen ted by using (2.1.27)
to sum the gluon polarization states) . The sum
Sll + S22 + S12 = IAR + BRll
_ 2 2 2
2
Xl
+ X 22 ()
- 32e eqg. (1 _ xI)(l _ X2)' 2.3.31
is, of course, gauge invariant. Combining (2.3.22) with (2.3.23) and (2.3.31)
yields
1 dl7E 2a. xi + X~ (2.3 .32)
170 dXldx2 - 37T (1- xd(l - X2)'
where the Born term 170 is given in (2.1.30) and the strong coupling is
(2.3.33)
The order a. correction to the Born term total cross section, 170, in (2.1.30)
from the emission of a real gluon, l7(real), is arrived at by integrating (2.3.32)
over the kinematically allowed range of Xl and X2 . The boundary of the allowed
phase-space region for massless quarks and gluons is given by the implicit
solution of
stu = 0, (2.3.34)
or in terms of Xi
(1 - xd(l- x2)(1 - X3) = O. (2.3 .35)
The resulting triangular region
o ~ Xl ~ 1, (2.3.36)
1- Xl ~ X2 ~ 1, (2.3.37)
is shown in Fig. 2.5. The integral of (2.3.32) becomes
2 a.
I7E(real) = - 3
7T
17 0
10
11 1
dXl
I-Xl
dX2 (
1-
Xl2
Xl
+ X22
)(
1 - X2
)' (2.3.38)
and we are left with a problem that will occur over and over as we perform
perturbative QeD calculations. The differential cross section in (2.3 .32) di-
verges as Xl or X2 goes to 1 and l7(real) is infinite. The origin of this divergence
is clear. Consider, for example, the invariant tin (2.3 .12) and (2 .3.24) . In the
massless limit we have
(2.3.39)
where 823 is the angle between P2 and P3 and E2 and ware the energy of the
outgoing quark and gluon, respectively. The amplitude squared in (2.3.24)
\
and hence the differential cross section diverges as t - 0 which occurs when
/
2.:1 II IIJ ( 1.1U1I 1<:111 IlIi 011
X3
Figure 2.5 (a) Dalitz plot for the "decay" of I. virtual photon with invariant mass
Q into a massless quark, a.ntiquark, a.nd gluon ea.ch carrying fractional energy
Xi =
2EdQ. The shaded area is the allowed kinematic region. (b) Same as (a) except
now the gluon is given a fictitious mass, mg, a.nd f3 = mi/Q2.
the energy of the gluon goes to zero (w .... 0, "soft divergence") or when the
outgoing quark and gluon become parallel (cos 023 .... 1, "parallel divergence").
The first type of divergence is referred to as an infrared divergence (it occurs
\
:W ( ) "II pl.,'. :.l ' 1',,1 III It 111." 10" " I (' , 1, ,"1' " 11 "
in the limit 1".q -+ 0 a lld 111" I 0 , wll<' l't, "' /1 II lI d 111 " ilrc ti l!' I!, III O Ii a lld qll a rk
masses, respec tivel y) , whil e t.I H' f>e 'o lld if> r(' f'N rt'd 1.0 as a Ill ass f> ill g ul a ri t.y (it
occurs as mq -+ 0 wiLli 111,g 0). =
In order to proceed we must decide on some wa.y of reg ul a rizin g the in-
frared and mass singularities. In addition , it better be t ru e th a t no experimen-
tal observables depend on the manner in which we perform th e regul ariz ation.
Before we proceed to examine regularization schemes it is instructive to inte-
grate (2 .3.32) over the region
(2 .3.40)
(2 .3.41)
where m is an arbitrary mass (for example a pion mass) . T he result is the
total cross section for producing a real gluon whose invariant mass with both
the quark and antiquark is alwa.ys greater th an m 2 ,
1 fJ
umb'; -+- qijg) =
l fJ
-
dX1
11-fJ
I-x,
d X2
(d
d u:
Xl X2
)
' (2.3.42)
l fJ
1 fJ
-
.' dX1
ll-fJ
dX2
I-Xl
(1 + xt)
(1 - X2)
_l
-
fJ
l
-
fJ
d
Xl
ll-fJ d
I-Xl
X2
(1 + X2)
(1 - xI)
= -~ 10g(j3) - ~ , (2 .3.44)
where terms which vanish in the limit j3 -+- 0 have been dropped . The 10g(j3)
term arises from the integral
l - fJ 1
-1-- dX2 = 10g(xI/j3) . (2.3.45)
l I-Xl - X2
where again I have dropped terms that vanish in the limit j3 -+- 0 and I have
used
r l
-
fJ
10g(Xl) dXl = (log(xt) dX1 _ 10g(xI) dX1 r1
(:u .tI )
wl",r, · I, It(' dilogaril,ltlll fllll (,l, ioll , Li~(x), is defi ll ed ill ApJl( 'lIdi x I';. COII,i>illillll,
,, ' f IIl H :\11<1 using (£.2.12),
(2.:LtlH)
p, IV" S
2
•
O'mCrE -+ qijg 37r 0'0 { 2log2()
) = 2a. {J + 3 log ()
{J + 37r + '5}
2 '
where again terms that vanish in the limit {J -+ 0 have been dropp d.
The cross section in (2.3.50) has a term that diverges like log2 ({J) as (3 • ()
which comes from the region in which both Xl and X2 approach 1. ]n addit.ioll ,
I.here is a term that diverges like log({J) and there are terms that a r fillil,, ·
IIH (J -+ O. As Q increases this cross section increases like log(Q). The log:l(fj)
I,('rlll behaves like log2 (Q) at large Q, but the coupling constant in (1.2 . 18)
hehaves like l/log(Q) resulting in a net log(Q) dependence of (2 .3.50) as Q
III'comes large.
The virtual corrections in Fig. 2.6 and Fig. 2.7 have the same final state illi
does the Born term in Fig. 2.3. These amplitudes must be added and th<'11
sq uared as follows
(2.4.2)
Av =
Figure 2.6 Virtual gluon corrections to the Born amplitude in Fig. 2.3.
where
Po = Pl - 1: , (2.4.3)
Ph = P2 + 1:, (2.4.4)
and
q = Po + Pb· (2.4.5)
Combining this with the Ao amplitude in (2.1.31) and integrating over the
4-momentum, 1:, of the virtual gluon gives
4
= 30"0 g.
2 2( .)
-J
J 4
d 1: N(Pt.P2, 1:, q)
(211')4 (Pl _ 1:)2(p2 + 1:)21:2' (2.4.6)
(2.4.7)
and where O"v(virtual) refers to the order a. virtual contribution arising from
/
:l," Vi rt u,,) ()II" " , (J" rrr \'tlu" li 3 '1
Figure 2,7 Virtual gluon corrections to the Born amplitude in Fig. 2.3.
the vertex correction. The color factor tr(TaTa) = 4 has been included. Fur
thermore, 7J in (2.4.7) is a gauge parameter and comes from using
i [ k/.lkv]
- k 2 g/.lv + 7J--;;2 (2.4 .8)
=
for the gluon propagator (7J 0 is the Feynman gauge, 7J -1 is the Land au =
gauge, see Appendix C).
To evaluate O'v(virtual) we make use of the Feynman parameterizations
given in Appendix C. In particular we use
1 [1 1
ab = } 0 dy -:-[a-y-+-:-b~(1-_-y""7)=p , (2.4 .9)
and set
a = (P1 - k)2 = k 2 - 2P1 ·k, (2.4.10)
b = (P2 + k)2 = k 2 + 2p2 ·k, (2.4.11)
34
so that
ay + 6(1 - y) = k ~ - 2k ·py, (2.4 .12)
with
Py = YPI - (1 - Y)P2. (2.4.13)
This yields
4
.
y(vIrtual ) 4 2( ')/ d k t d N(Pl,P2,k,q) (2.4.14)
U = 3uO 2g. -z (271")4 10 Y (k2 _ 2py .k)2p·
The integration over k involves a liP and a 1/k4 piece. Namely,
1 q2 q2
pN(Pl,P2,k,q) = k2N1(Pl,P2,k,q)+ k4N2(Pl,P2,k,q), (2.4.15)
with
8 Pl·k P2 ·k P2·k - Pl·k k2
N 1(Pl,P2,k,q)=-2+ 4 +(4+27]) 2 +7]2' (2.4.16)
q q q
and
(2.4.17)
We now use
_1__ fl dx 2x
(2.4.18)
e2d - 10 [ex + d(l - x)]3'
and
_1__ fl dx 6x(1- x)
(2.4.19)
2 2
e d - 10 [ex + d(l- x)]4'
with
e=k 2 -2py.k, (2.4.20)
2
d= k , (2.4.21 )
so that
ex + d(l- x) = k 2 - 2xpy·k = (k - Xpy)2 - x 2p;
= J{2 - C, (2.4.22)
where
(2.4.23)
and
(2.4 .24)
Shifting from an integral over k to one over J( using k = J( + XPy yields
{
'2 J; '1 ~ NI(1.; . /\ IJ'/'/,) GJ; (I - J: )q ~N-( k . /,' I .f/ljI)} .
[/\' ~ cr' -I [/,' :/ C'J"
('1,.11. '2 . .. )
'1'10 " Hllift d numerator NJ is evaluated using the fact that
Pl,k P2· k 1/(2 1 2 (1 )
-+----xy -y
q4 8 q2 4 '
P2 · k - Pl·k 1
2 -+ 2'x, (V I. '27)
q
(V 1. '2H)
where terms odd in J( are dropped since they contribute nothing to th(~ illl.. ·
J.\ra.l over J(. Symmetric integration requires an average over the dir cl,iO Ii Hor
/\" f
I which is equivalent to the substitution
(2.'1 .'2!J)
III addition,
P~ = -y(l _ y)q2, (2 .4.:)0)
\ (2.4 .31 )
HO that the shifted numerators become
a nd
N 2(k -+ I< + XPy) -+ -TJ UJ(2 - x 2y(1 _ y)q2) ,
where I have rearranged several of the terms in N 1 .
The integral over the momentum J( is performed using (C.2.5),
d4J( (J(2)R = i( _l)R+M C R - M +2 r(R + 2)r(M - R - 2)
J
(271')4 [J(2 - C]M 1671'2 reM)
(2.1.3 /1)
yielding
The second term comes from last term of Nl in (2.4.32) and is ultraviolet
divergent like log(I<) as I< -+ 00. In the Landau gauge (i.e ., TJ = -1) this
term is absent but one is still left with the infrared divergences that occur in
the x and y interactions of the first term.
This vertex correction is not only infinite, it is gauge dependent. To get
p
)
p -k
a gauge independent result one must add the other order 0", corrections in
(2.4.1). The self-energy correction shown in Fig. 2.8 is given by
(2.4.36)
where again I have used the gluon propagator in (2.4.8), but I have not yet
included the color factor. It will turn out that E(p) is proportional to., so
that one can write
E(p) -iJE.= (2.4.37)
The contribution to the total "y. - qq rate from the virtual self-energy am-
plitudes in Fig. 2.7 is then
1
ab =
J 1
d:e [a:e + b(1 - :e)F'
(2.4.39)
and
1 Jd 2(1-:e) (2.4.40)
ab 2 = :e [a:e + b(1 _ :e)]3'
from (C.2.3) with
(2.4.41 )
/
C = -p2x(1 - x).
'I 'II(' ractors in (2.4.36) become
E = (_. 2)
19s
{11 J(271')4 [-2(1 - x) - CF7](1 + x)]
0
d
x
d
4
]{
[]{2 -
\ + 1 J 0
1
dx
d4](
(271')4
7](1 - x)[](2
[](2 -
+ 4x2p2]}
cp . (2.'1.4 8)
As was the case with the vertex corrections uv(virtual), the self-energy co rrec-
tions are gauge dependent and in general, contain both infrared and ultravioit't
divergences.
At this point we cannot proceed without a procedure for regul a rizin g
the ultraviolet divergences in uv(virtual) and us(virtual). If we hand le th 'sc
divergences correctly we will find that these ultraviolet divergences cancel (l. lId
that the sum
u(virtual) = uv(virtual) + us(virtual), (2 .'1 .'19)
is gauge invariant and contains only infrared divergences. Ifwe then regul a riz('
these infrared divergences together with the infrared divergences in u(real) w('
will find that the sum u(real) +u(virtual) is finite and contains no divergcnc Ii
of any kind .
We will consider two regularization schemes in this book, the "massive gluon
scheme,,3,4 and dimensional regularization 5,6 . In the "massive gluon scheme"
(MG) one regularizes the real and virtual corrections by giving the gluon a
fictitious mass, mg. This scheme breaks gauge invariance and is applicable
only if the triple-gluon vertex does not play a role in the process. For the
. I."dI'OIl H
process being considel' d til - triple glll Ol1 vc'rt.cx is 11 01, prl'll<' llt. NO Uli ,.. III -th d
is acceptable.
For the case of a m assive gluon, the different.ial cross section in (2 .3.32)
becomes
(2 .5.1)
d(T
dXI
= 2a. (To
311'
{I1 +- XlxI 10g(XI(1- XI)) _ ~_1_
2 1 - Xl
/3
1 1 /3(2 - xI) 1 /32 }
(2.5.3)
+ 2XI + 2+ (1 - xI)2 + 2 (1 - XI)3 '
where I have dropped some terms that vanish in the limit /3 -+ O. The first
term in (2.5.3) comes from
1;~" 1 ~ X2 dX2 = log «Xl + /3~1 - XI)) -+ log ( XI(l; XI)). (2.5.4)
Care must be taken in not dropping the f3 and f32 terms in (2.5.3) for these
terms give finite contributions when integrating over Xl. For example,
1 f3 /3
1 o
-
( 1- Xl )
2 dXI = 1,
d(T
dXI
= 2a. (To
311'
{I1 +- xI 10g(XI(1- XI)) _ ~_1_
Xl 21 - Xl
/3
1
+-XI + -21 + -6(1
5 - xI) } . (2.5.6)
2 4
Integrating over Xl using
1- f3 1
1 o
1 - f3
- - dXI = -log(/3),
1 - Xl
1
(2.5.7)
1 o
--log(l - Xl) dXI =
1- Xl
-t log2 (/3) , (2.5.8)
/
, ,I Vi ,I,111i1 <iI"II" ( :0 1'11" I, ""Ii M<: Hd .. IIIIM :w
I I !,
1
1 II J' '/I' A
I IO/ol,(J' I ) rlJ:, _ ;.
() I J' I :1 'I
lo
__I
I - XJ
10g( 1 - XI) dX I =- log2(,LI) + t. (:Ui , 10 )
1, 1,1,
1, 11 II , liS expected from (2,3,50). contains a log2«(3) and a 10g«(3) d iv,' r.u;"lI rf'
,IIA / ' • () ,
wlH!re
L
C(k) = L-P' (2 . G . ~)
Itlld where L is large. For small k, C(k) is approximately equal to one. bllt. at.
III.rge k it behaves like 1/k 2 causing the integrals to converge.
In the massive gluon scheme the infrared divergences are handled by r\ '
pl acing 1/k 2 by 1/(k 2 - m~) in the gluon propagator. Both the infrar d Iwd
1IItravioiet divergences can be treated simultaneously by the replacem ent.
1
k2 ---+ -
1£ 2
mg
de
(P _ £)2 ' (2 .6 .:\)
fo r the gluon propagator. With this replacement the self-energy correction ill
(2.4.48) becomes
-
~(p)
• 2
= (lg,)
1£
m
2
de Jo
t dx J(2'/1')4
dK 4
2(1 - x)[-2(1 - x)]
[K2 _ C]3 ' (2,6.4)
g
f and j{ gives
-
'EMG(P) = (-g')-62
2
1 7r
1
1
0
1
dx2(1-x)log
( - p 2 X -I-
-P
2
X
1)
oJ
+ mg2
2 1 2
= (-g')167r2Iog(Ljmg), (2 .6.6)
where terms that vanish in the limit m~j L -+ 0 have been dropped. From
(2.4.38) we arrive at
(2.6.7)
for the Feynman gauge. Similarly, the replacement (2.6.3) and setting TJ = 0
gives
20'' 0"0
O"v (virLu al) = -
37r
1L 11 11
m 2
9
de
0
dy
0
dx
With q2 > 0 the denominators in (2.6.11) can vanish over the range of the
x and y integrations. We can avoid this difficulty by requiring that q2 be
spacelike
(2.6.12)
and after we have performed the integrals we can analytically continue to the
" II (1IIIwlik,' ) rqr; io ll . Th,' illtqr; mhl /\.1'(' t,'di o Ii H Iillt Hl.r·/I,i/o\ ld,l<lI·wll rd . ' I'hl'
II II dt IIi
21'l'. { ~ }
" , ( VII I.lI ltI)
2
= ~11l' (]'o - log ((3) - 310g({3) -
- , -
27 - 3211' 2 + log(L/1II!,) ,
(~ . (j . 1 :1)
(2.fi . I /I)
11 11 (,? )I,ivl' II by (2 .6.12). Combining this with (],s(virtual) in (2 .6.7) !l: iV('H
\ 1111 11 cOIILa ins only infrared divergences. The dependence on th e ultrll.vi o l(' l,
, 111,,,11' /, has dropped out.
(]'MC ( vIrtual
. ) 20.
= S(]'o { -log 2()
{3 - (3 - 2'7
3 log() - 3211'2 + 11' 2} . . '
11' tlm ehke
(2.7.'1)
with
(2 .7.5)
a nd Q2 =
q2 > 0 (i.e., q2 timelike). Furthermore, because of (2.4.1) I need
only keep the real part. The 11'2 in (2.7.4) comes from the analytic continuati o n
from the spacelike to the timelike region of q2. Combining this with the result
in (2.5.11) for the real gluoll corrections, we arrive at
(]'Mc(real)
.
+ (]'JlfG(vrrtual )
=
20.
311' (]'o
{11'2
-3 + 5 - 27 - 3211'2 + 11' 2}
42 C ll ll pl'''1 'J. ' )'lItlll 11 101,/1 1111 /, ) /'
(2.7.6 )
Both the log2(,B) and log(,B) terms cancel out in the sum leavin g a finite resu lt
in the limit mg -+ O. Thus, the total rate for a virtual photon to decay into
partons has the following perturbation series
e+e-
O"tot =0"0
(1 +-;+
a. )
.... (2.7.7)
(2 .7 .8)
where a.(Q2) = 411" /(fJo log(Q2 / A2)) is the familiar running coupling constant
in (1.2.18) and A is the QCD perturbative parameter that sets the scale.
dW = _1_IMI2 d 2N - 2R (2 .8.1 )
2Ecm 2,
where the two-body phase-space factor
dN - I dN-lp
d - R2 = (211")N-If~EI) (211")N-I(~E2) (27r)N 8 (q - PI - P2),
2N 2 N (2 .8 .2)
however,
(2.8.4)
replaces (2.1.20). Here the angles (it, .. . ()N-2 are angles with respect to the
axes in N -1 dimensions. If N = 4 (2.8.4) reduces to (2.1.20) with ()2 in (2 .8.4)
/
', H It "/I l ( lIli OIl 1': '111 ,,11' Ill! , d 'nll ' ll
the phi H.lI gk, (/1 1. 'I'h.. 1I1/l.t.l'ix e1 CIIICll t in ( ~ . H . I ) d()<'H II Ot. dqll'lI d 011
I' P III )!,
t I"'H" 1I.nglcs a.nd 1I,I'y GiL li hi' ill t grated ou t by I' ' p<,a,tcd ww f til(' f01' 1I1111 11.
1".o
n
sm () d(} = v' - rr (t(n'2 ++ 1~ )) ,
7r (~ . H . f) )
= 2N- 3 r( ~ - 1)
J dN - lp1
2E1 7r
(N-2)/2
r(N-2)
E N- 3dE
1 1,
(2 .H.(j)
r( ~ - 1)
R2 = J d
2N-2
R2 = 2N-1 7rN/2-1
1
r(N _ 2)
QN- 4,
(2 .8.8)
and
(2.8 . 1I)
with
(2 .8.1 2)
being dimensionless. The "dimensional regularization mass", rnD, plays a n
analogous role to the gluon mass , rn g , in the massive gluon regularizatio n
scheme. Combining (2.8 .9) and (2 .8.10) with (2.8.8) gives
or
0'0
2
= 3ae q Q
r(2 + '2') (~2
r(2 + f) 47rmb
),/2 ' (2.8 .14)
where
(2.8 .15)
Of course, (2.8 .14) reduces to (2.1.30) when { = o.
To evaluate the three-body decay of a virtual photon into a quark, an-
tiquark, and a gluon we proceed in an analogous manner. The three-body
differential decay rate in N dimensions is
dW = _1_IMI2
2Ecm
d3N - 3R
3,
(2.8 .16)
where
dN-PI d N -P2
(271")N-I(2Et) (271")N-1(2E2)
N
d -P3 ( )N 6N( ) (2.8.17)
(271")N-1( 2E3) 271" q - PI - P2 - P3 .
(2.8.18)
To integrate over the directions of PI and P2 one makes use of (2.8.4) and
(2.8.6) which results in
dN-Pl d N -P2 2N - 3 71"N-2 N-3 N-3
11 2E1 2E2 = f(N _ 2) El dEl E2 dE2
1 1 dz(1 - z2)N/2-2,
-1
(2.8.19)
where as in (2.3 .17) I have chosen the axis to be in the direction of particle 1
with z = cos 0 12 , where 0 12 is the relative angle between particle 1 and particle
2, and all decay particles have been taken to be massless. Changing to the Xi
variables in (2.3.9) and using (2.3.18) gives
1 -1
1 dz (1 _ z2)N/2-2 6(Q - E1 - E2 - E 3) = 2(1- z2)N/2-2,
2E3 XIX2Q2
(2.8.20)
so that
d R3
2
=
Q2 (Q 2 )
16(271")3 471"
f 1 (1 -
f(2 + {)
z2) f/2
-4-
f
Xl dXI
f
X2 dx 2, (2.8.21)
where
(2.8.22)
Equation (2.8.21) reduces to (2.3.20) when N = 4 and as before we can-
not integrate d 2R3 any further without knowing the matrix element since, in
general, it will depend on Xl and X2.
In N = 4 + { dimensions the matrix element squared for the process
/
III analogy with the QED case in (2.8.10) we define a dimension I 88 stroll ll.
t"t )llpling, gs, in terms of the N dimensional coupling, gN, and th e I1l clSIi, 1110 ,
ItS follows
or
2
QeD _ gN a,
(2 .8.1())
aN = 471" - (m1)f/2'
where a, = g;/471" is dimensionless. Combining (2 .8.16), (2 .8.2 1), and (2.8 .2:1)
\
(2 .8.27)
20', (Q 2
) (/2 1
O"DR(real) =-3 0"0
71"
-4
7I"m D
2
r (2 +
f)2
11
o
11dX1 x~
I-Xl
dX2 x~
(1 ~ z
2)f/2
F(X1' X2), (2.8 .29)
where z is defined in (2.8.22) and where "DR" denotes the dimensional reg-
ularization scheme and "real" implies the production of a real gluon by ill
process ,. --+ qijg. The nested integrals in (2.8 .29) are painful to evaluate. It
is easier to "decouple" the integrals by defining a variable v such that
x2=1- vX 1, (2.8.30)
and to express everything in terms of Xl and v since both run from 0 to 1.
, J.ndIOIi M
(2 .8.31)
lTDR(real) = -3
20:.
lTo
(Q
-4
2 ) f/2
2
1
(f)
7r 7rm D f 2 + "2
11 dXl xHl- xd f
/
2
1l dv V
f
/
2
(1- vy/2 xlF(Xl, v), (2.8.32)
It is necessary to keep both the ( and (2 term in (2.8.33) since the integrations
in (2 .8.29) will produce 1/( and 1/(2 factors. With the use of (E.1.12),
l
r
10 d X XR-l(l_ X)M-l = f(R)f(M)
f(R + M) ,
(2834)
..
it. it; st.raightforward but tedious to insert (2.8.33) into (2.8.32) and perform
th e i lIt.cg rati ons. The result is
lTDR(real) = 20:.
37r lTo {8 + -;1[ (Q
(2 4log 47rm'b
2
)
+ 4/E - 6]
/
• I Virl,lIal ChlOIl (:o ll.,d 1111 I HI ~' , II fl lIl lI '17
(2.8.:18)
III analogy with the massive gluon scheme O"DR(real) contains both a log
Hqllared and a log divergence as mD -+ O.
r J l
+ 10 dx (27r)N
d N [( 71(l-x)
[4[(2jN+4x p2]
[/{2 - C]3 '
2
(2 .9. 1)
where C is given by (2.4.45) and where (2.4.46) and (2.4.47) are replaced by
'Ya(P - ¥ha -+ (2 - N), (1 - x), (2.9.2)
¥(P -I/)¥
k4 -+ -
1
k2
J.
p (1 - x) +
2,
k4
([(2
N + x 2p 2) , (2.9 .3)
(2.9.7)
-18 e ll I,Pt..,, :.! ' 1'01,,,1 11.01,,' I", (' 1(,- . 1111<1'"11"
where N = 4+( a IlJ !). is the dilll t' lI ~ i () III ('~H ('o uplill g ill (~ . H,25) , In the dill1 CIl-
sional regularization sch me th s If- 11 rgy co rr 'ct.io ll s vall is h when TI = - 1
(Landau gauge). Therefore wh en working in this scheme one usually chooses
the Landau gauge and ignores self-energy corrections. Namely,
EDR(Landau gauge) = O. (2 .9.8)
Furthermore EDR in (2.9.7) is zero for any value of 1] provided p2 = 0 (on
shell) and ( > O. This is a subtle point, but one can show that with the
dimensional regularization scheme it is consistent to set EDR = 0 for any 1]
and any € provided p2 = 0 (i.e., on shell).
In N dimensions the vertex correction in (2.4.6) becomes
. )
O' v( virtual = '3 O'02 gN
4 2 (
-t
')J dNk N(PI,P2,k,q)
(271')N (PI _ k)2(p2 + k)2k2' (2.9.9)
with
where N = 4+€ and Py and C are given by (2.4.30) and (2.4.31), respectively.
The integrals over 1(, x, and yare performed using (2.9.4), (2 .9.5) and (2.8.34)
with the result
. 20'. Q ) -2 (/2
r(1 - ~) r2(1 + ~)
O'DR(vutual) = 371' 0'0 ( 471'm1
reI + €)
(- (~ + ~ - (~1:~;)) ,(2.9.14)
where 0'0 is given by (2 .8. 14) and where mD is the dimensional regularization
/
, , 10 (hd,'r (Y. ( :11'1'1 '( I. "" ~ I) Ie . r li nllO'
III , MM t.loat ari scl:l fr O lO1 (' .K.' 0) n.lld (as was th e ('UIlI' III til(' IIl l1SMivI' ,11, 1110 11
1 101 ' 111 <' ) 1 have t a k ' lI
2
Q = _q2 > 0
( /. 1' ., q2 space/ike) in order to avoid singularities within the regions o f ill t 'g ra
1.1011 . l ~x pal1ding in powers of ( gives
.
O"DR(vutual) = 311"
20's
0"0
{
- (2
8 1
+ -; [
-410g ( 411"-2 b
m
Q 1- 4,E + ) 6
1.11 the dimensional regularization scheme the order O's virtual corrections a.re
from (2.9.17),
(2.10 .1)
.
O"DR(vlrtual) = 20's
-0'0
311"
{8
-2" + -1[
((
(Q
-410g --2- ) - 4'E
411"m D
2
+ 6]
7T~
-- 8
fi
71" ~} I (2. LO .2)
I,i lli clike
where now Q2 = q2 > 0 (i. e., q2 tim elike) and where I ha ve only kept the
terms that are real.
Combining this with the result in (2 .8.38) for the real gluon corrections,
+IOg2(~)
471" m D
+(2IE-3)log(~)
471"171D
2 771"2 57}
+,E - 3, E - 6 + 6" ' (2.10.3)
gives
Problems
2.1. Use 1p.1" + 'i''''p. = 2gp." and the cyclic property of traces to show that
tr [~'p.] =4ap.
tr[~Ip.V,,,] =4 (ap.b" + a"bp. - a·b gp.,,),
where a and b are arbitrary 4-vectors.
2.2. Calculate the differential cross section, dO' j d0. cm , and the total cross section,
O'tot, to order 0'2 for the electromagnetic process e+ e- -+ I· -+ J.l+ J.l- .
Neglect the masses of the electrons and muons.
2.3. Calculate the differential cross section, dO'jd0. cm , and the total cross section,
O'tot, for the fictitious process e+ e- -+ S~ -+ J.l+ J.l-, where So is a mass-
less scalar boson (spin 0). Neglect the masses of the electrons and muons.
Compare the answer with the results from problem 2.2.
llz = J 6
d Rz
1
= -.
871'
'61,,
I ,) , Consider the matrix element squared for a process with on e real xkr ll ll i
photon which can be written in the form
2.6. Assuming massless particles, integrate the three-body phase-space factor ill
(2.3.8),
d9 _ d:PI d:P2
'R3 - (271')3(2EI) (271')3(2E2)
2.7 . Show that the differential cross section for the process IE -+ qijg in Fig. 2.4
lJ2 ( :hll'I'I., ' r 'l 'I'ol,ld 1111 I", ftll (' I "
where (10 is the Born term in (2.1.30) and Xi = 2E;/Q, with Ecm =Q=
EI + E2 + E 3 .
2.8. Suppose that in the "decay" of a virtual photon into a quark, antiquark, and
a gluon we are only interested in the case where the gluon-quark and gluon-
antiquark invariant mass is larger than the mass of a pion, m". . Perform the
integral
1 fJ
jl-fJ dX2 (d ),
(1m(rE --+ qijg) =
1fJ -
dXl
I-x}
d (1;
Xl X2
where f3 = m;r/Q2 and where the differential cross is calculated to order
a •. Keep the exact result and show that it reduces to (2.3.50) in the limit
f3 --+ o.
2.9. Show that the cross section
1 - Xl - 13
1- 13 - Xl :S X2:S 1
- Xl
'
O"MG(virtual) = -2a.
0"0
{ 2 - - 7 271"2 }
-log (13) - 3Iog(f3) - - - - ,
371" 2 3
- -2 -2
where 13 = m;/Q and Q = _q2 >0 (spacelike).
2. 14. Integrate the two-body phase-space factor for massless particles in N s r ac"
time dimensions,
R _ jd2N-2R _ 1 r(1
2 - 2 - 871" r(1 +
+ ~)
f)
(Q 2)</2
471" '
2 r(2+~)(Q2)f/2
0"0 = 3ae q Q r(2 + f) 471" m b '
where N = 4+ f and a is the dimensionless electromagnetic couplin g in
(2.8.11 ).
2. 16. Integ rate til tllr c- bo dy pil ase s pac(" fa.d or fo r IlIllHS iesH pltrti les in N s pilce-
time dimensions,
2.17. Using
1 2 71'2 2
r ( 1+()=1-rE(+2(rE +6)( +" ',
in (E.1.18) show that
2
r (1 + ~) 1 1 2 2 2
3) = 1 +
r( 1 + '21'
'nE (+ 4s(6 (E - 771' ) ( +"',
A
and
r(I-~)r2(1+~) 1 1 2 22
r(1 + () = 1 + '2rE (+ 48(6rE - 71' ) ( + . .. .
2.18 . Show that in N spacetime dimensions that the differential cross section for
the process rf -+ qijg is given by
dUE _ 2a 8 2 Q ( Q2
-ae ---
)£ F(X1,X2) X1£X2£(1-
--
z2)£/2
dX1dx2 - 71' q 47rmb r(2 + () 4 '
where F(X1' .2:2) is given in (2.8.24) and z = 1 - 2(1 - Xl - X2)/(X1X2).
Perform the integral
/
where
-2
Q = -q 2 > 0 (spacelike).
I"nrt her Reading
J.D. Bjorken and S.D. Drell, Relativistic Quantum Mechanics, McG raw-
1Ii11 , New York, 1965.
R.P. Feynman, Photon-Hadron Interactions, Benjamin-Cummings Pllb
lishing Company, Reading, MA, 1972.
A. H. Mueller, P1Jysics Reports 73, 237 (1981).
References
Electron-Positron Annihilations:
Fragmentation Functions
and Jets
56
/
III 1.11 1' 1)I',' vi o lis (' h''Id.",· 1'1" 1-l 1.lld i,'d th ( to ta.l ('I'OHH ,.... (' I.io ll 1'0 1' 1.11,' 'I.llllilril ll
11,, " of 11 11 d ('(' l.ro ll lI11d 'I. po;;il.roll illto hadro llH, " I i ' IIiLdroll s. WI' 1'11 ' 1'1 '
11111. ,'011 (,('1'11 d a bo llt tll(! details of th e fin a l sta tc so 10ll g as it was hadrnni c.
lI 11w('vl'l', thcrc a re ma ny interesting questions a bo ut th e co nfi g ur a ti o ll or filial
"I lit" hadro ns. For example, before data became ava ilable so me physi cists hI'
1I.'v,'d t.h a t the final hadrons in e+e- annihilations would emerge sph ri cn.lly
IIV IIIIIII'I.ri c. Others argued that the final hadrons would "remember" the dir0.(,
110 11 for the initial quark and antiquark in Fig. 2.1 that initiated the prO('I'RH .
'!' Ir.,y conjectured that the final state would consist of a collection or had I' lil t;
• 1I 11,'d a "j et" moving in the direction of the initial quark and a j et of ha.d rOIi H
1I1" vi ng in the direction of the initial antiquark . We now know that .ids ,' xil-l l.
H ll d a.re a result of the asymptotic freedom property of QeD. Thc hadro ll H ill
" ,w I, have small transverse momentum relative to their parent qu a rkfl di,"'('
1.," " and the sum of their longitudinal momentum roughly gives th e P"I'I 'III,
q ll ll.rk momentum. The precise definition of a jet is arbitrary and dep'!lIdH Oil
1.1 1(' experiment, but jets are the manifestation of quarks (and gluon s) ill 1.11('
"" 11 1 world. Jets are as close as we can get experimentally to "seeing" qllarkH
II lId gluons. We study jets to learn about the underlying quarks and gIIl OIl H.
The maximum inclusive process e+ e- -+ hadrons examined in the pI' -
vio us chapter is an example of an observable where the infrared singularities
!.Il a t arise when one integrates over the final state partons are completely can-
CI'led by the virtual corrections. As Q2 increases 0'.(Q2) becomes sm a ll and
I.h c QeD result approaches the naive parton model prediction. One might g t
I.h c impression that since 0'.(Q2) becomes small at large Q2 (i.e ., asympto ti c
freedom) that all QeD perturbative calculations approach the naive pa rto ll
Ill odel predictions as Q2 -+ 00. As we shall see in this chapter this is definitely
1I 0t the case. In the previous chapter we integrated over all the phase space or
the final state partons. In this chapter we will ask questions about the na tllrl'
or final state. We will see that, for example, all the infrared singularities cl l')
not cancel when computing the quark inclusive cross section, e+ e- -+ q + X .
As a consequence of this the QeD results for the single hadron inclusive crOI-lH
Hection, e+ e- -+ h + X, differs more and more from the naive parton Ill odd
ex pectations as Q2 becomes large even though 0'.(Q2) becomes small . Oefo l'(,
procf'eding to compute parton differential cross sections in perturbative Q 'I)
we examine the expectations of the naive parton model.
In the naive parton model it is assumed, for example, that when the quark
in Fig. 2.1 separates from the antiquark that each fragments or cascades into
a collection (or "jet") of hadrons, each having small transverse momentum
relative to the parent quarks direction. The number of hadrons of type h with
energy fr actioll
(3 .1.1)
per dz is described by the "fragm entation function" D;(z, Q2), where q is the
initiating quark . Energy conservation implies that
1
L JOr
zD;(z, Q2) = 1. (3.1.2)
All h
The single particle inclusive cross section is given by
where
_ (+ _
(J.lJ.I) = + _) _ 411" 0: 2
U u e e ---> J.I J.I -"3 Q2 ' (3.1.4)
'" 11 1
~
All h 0
d(Jh
"2zTiQ
Z
2
) dz = e+e-
(Jtot , (3 .1.5)
where (J~~ e- is the total e+e- --->hadrons cross section and the ratio R in
(2.1.16) is
(3.1.6)
In general the inclusive cross section can be a function of the center-of- mass
energy Q, but in the naive parton model the fragmentations functions are
assumed to scale (i. e., depend only on the scaling variable z),
(Parton model). (3.1.7)
Neglecting mass effects, the probability of finding, for example, a 10 GeV
pion in a 20 Ge V jet is assumed to be the same as the probability of finding a
20 GeV pion in a 40 GeV jet. Furthermore, zD:(z) is assumed to approach
a constant as z -+ O. This implies that the mean multiplicity of particles of
type h emerging from the parent quark q,
Rank = 3 2 I
(dc) (cb) (ba) Meson States
•••
VVV
dd
Quark-Antiquark pairs
cc bb
Original
quark
flavor "a"
Figure 3.1 Hierarchy of mesons formed when an initial quark of flavor "a" combi ll ·s
with an antiquark from a produced quark-antiquark pair, "bb", forming the meson
of rank 1. The resulting quark of flavor "b" then combines with an antiquark froOi
another produced quark- antiquark pair forming the meson of rank 2 and so on.
model due to R.P. Feynman and myselfl (called the FF parameterization) ca.n
be used to parameterize the non-perturbative aspects of quark jets. It is 11 01,
meant to be a theory. It is simply a parameterization that incorporates many
of the expected features of fragmentation. The model assumes that quark j ta
can be analyzed on the basis of a recursive principle. The ansatz is based on
the idea that a quark of flavor "a" separating from an antiquark and hav-
ing some momentum Po in the i-direction creates a color field in which new
quark-antiquark pairs are produced. Quark "a" then combines with an anti-
quark, say "b", from the new pair bb to form a meson of flavor ab leaving the
remaining "b" flavor quark to combine with further antiquarks. A "hierarchy"
of mesons is thus formed of which ab is first in "rank", be is second in rank, cd
is third in rank, etc. , as illustrated in Fig. 3.1. The "chain decay" ansatz as-
e,
sumes that, if the rank 1 meson carries away momentum from a quark jet
of flavor "a" and momentum Po, the remaining cascade starts with a quark
of flavor "b" and momentum PI = Po - e,
and the remaining hadrons are
distributed in prec i:wly th e IHlIl1 . WIly ILll tilt' hadroll H whi ch ('O llI e from a. jet
originated by a quark of fl avor "b" with 1110trlCIltUilI PI . H is furt.h er assutrl ed,
as in the naive parton model, th a t a ll distributions scale. T h comp lete j et
can then be described by the fun ction f(.,,) defin ed by,
f('7)d." = the probability that the first
hierarchy (rank 1)
meson leaves fractional momentum." to the remaining
cascade, (3.1.9)
and is normalized so that
(3 .1.10)
The probability of having hierarchy sequence with the n-th meson having
momentum ~n is
00
where "'i = =
P;f Pi - 1 and ~i Pi-l - Pi .
If we now define the single particle distribution, F(z), as
F(z)dz = the probability of finding a meson (independent
of hierarchy) with fractional momentum z
within dz in a jet, (3 .1. 12)
then F(z) must satisfy the integral equation
d."
1
1
F(z) = f(l - z) + - f(.,,)F(z/.,,) . (3.1.13)
z ."
This equation arises because the meson might be first in rank (with probability
f(l - z )dz) or if not , then the first rank meson has left a momentum fraction
1] with probability f(1])d1], and in this remaining cascade the probability to
find z in dz is F( Z/77)dz/." .
The integral equation for F(z) in (3.1.13) can easily be solved by taking
Fourier transforms. If we define
and
(3 .1.15)
dT}
wl,cre
¢(z) = a(z) +
J ry g(7J)a(z/7J), p .I.' I)
10 zF(z)dz = 1. (3 .1.21)
(3.1.28)
The first term arises because the ab meson state might be of first rank (but
only if a = q, hence the delta function Oqa) with probability f(1- z) times the
chance, f3b, that the first new pair is of the required type. The second term
occurs if the ab meson is not of first rank. In this case, the first pair might
be cc (with probability f3c) and leave a momentum fraction 7J to the cascade
of quark c (with probability f(7J» in which we find the state with probability
p:b(z/7J)dz/7J.
If we now define an "average quark flavor" by
nJ
(3.1.34)
Thus, if we know the distribution of mesons, F(z), from quarks disregarding
flavor, then those of flavor ab occur in the "average quark" jet with proba-
bilities, f3af3b. Only the contribution of the first rank quark differs from the
average. Comparing (3.1.30) and (3.1.33) and using (3.1.34) gives
p;b(z) = oqaf3b/(1 - z) + f3af3b F(z), (3.1.35)
/
w lWI'I '
",( :: ) F( z ) - / (1 - z),
111 til(' probability of find ill f it Il ICSOn of rank hi gher t han onc at z. Fo r t,l1I'
,I,nic(' of /(17) in (3.1. 26 )
zF( z) = (d + 1)(1 - z)d+l . (3.1.:17)
The probabilities, Pqab(z) , are related to the fragmentat io n fLl 1I/'1,ioll H,
f);; (::),
by
wlt('rc rzr, is the probability that the hadron h contains the qu a rk-itlll.iqll a ril
- +
Ill,al.e ab o For example, r: d = 1 and r~il = r~d =
0 0
t. Insertin g (3. 1.:\5) illl.!)
(:\. 1.38) gives
(3. UlJ)
where
(3. 1.'10)
(3.1.'11)
For example,
A~+ = (3 A"'o -
u -- 1(3
2 A u'" =0
Ad+ = 0 A"'O
u -- 1(3
2 A~ = (3
0 - (3. 1.'12)
A;+ = 0 A; = 0 A'"$
=0
B"'+ = (32 B"'O = (32 - = (32
B'"
yielding
where
(3 .1.'15)
is the "m ean quark" charge. The hadrons in the j et carry an average total
charge, e q - e(q) , equal to the charge of the quark plus a correction, - e (q) ,
proportional to the deviation of the probability of productions of new pairs
from the SU(N) value of liN. For the case of electric charge we have
(3.1.46)
where I have, for the sake of this discussion used (3.1.29) and assumed {3c =
{3b = O. In this case, the average of the total electric charge of a u, d, and s
quark jet is
(Qu) = 1 - {3 = 0.6,
(Qd) = -{3 = -0.4,
(Q&) = -{3 = -0.4, (3 .1.47)
where I have taken {3 = 0.4 . The hadrons do "remember" something about
t he electri c charge of the quark from which they originated, although the
knowledge is not precisely the charge of the quark .
T hey a re, of course, QCD perturbative corrections that modify this simple
parton model picture. For example, the diagrams in Fig. 2.3 give rise to gluon
jets and break t he scaling assumption in (3 .1.7). We now proceed to examine
some of these corrections.
Equation (2.5.6) gives the differential cross section (in the massive gluon
scheme) for producing a quark in the process r* -+ qijg carrying fractional
energy x ,
~ duXfG = 20:&
Uo dx 371'
{1
+ x 2 10 (X(l- x)Q2)
1- x g m~
_~_l_+~x+~} . (3 .2.1)
2 1- x 2 2 x<1
1 01
.IIIt {daX~
dl,
(; + (aA! (;(virtlliLl) _ lV " 0"0)
7r
6(1 - l:)} elJ: = O. (:U .:I)
in the differential cross section, dO"~fG 1dx. We 'first turn dO"~, 01ilJ:
I 1I 111 'l' i,rO Il f)
lid" /I p"re "+ function" whose integral over x vanishes as in (3 .2.3). Na lllc'l y,
fl (dO"AfG) dx = o.
Jo dx +
1'111'11 we write a new differential cross section according to
- - -1 (dO"q)
-1dO"q_ - + 0'. 1'+'-"(1)
u - X , (:1. :l .il)
0"0 dx 0"0 dx + q
w lll'l'<'
(3 .:l .G)
f',",ctio ns" are discussed in Appendix E. In general, they are defined accordillJ.!;
to
1
(F(x ))+ == lim {F(X)O(l - x - (3) - 6(1- x - (3) f -f3 F(Y)dY } . (3 .2./))
f3-0 Jo
vor x < 1 - f3, (F(x))+ = F(x), but the integral over x of (F(x))+ vani sll t'H .
From (3.2.1) it is easy to see that
-1 (dO"XfG)
0"0
--
dx
-_ -Pq_qg(x)
0'.
+ 27r
log (2
Q Img2) .+.-(
+ O'.fMG q x),
•
(3 .2.9)
where
(3.2 .10)
and
0'.
.+.-q(. )- 20'.
f MG •
X 3 -
{(I + x 2) (log(1-
7r
1 -x
x)) + + 11+-xx Iog.( x ) 2
( 1r~:3 - iIr, ) 6( 1 - :I:)
}. (:3 .2.11)
11 Pq~qg(X) dx = o. (3 .2.12)
and
1+ x2 11"2 5
1
1
o 1- x
log(x) dx = --3 + -4' (3.2.14)
The quark differential cross section in (3.2.5) with (d(Tq / dx)+ given by (3.2.9)
contains both. real and virtual corrections.
From the definition (3.2.8) we see that the two "+ functions" appearing
in (3.2 .11) are given by
1 1
(1 _ x)+ = 1 _ x 0(1 - x - (3) + log«(3) 0(1 - x - (J), (3.2 .16)
and
1 d(TbR
-
(To
--
dx
_ 20',
-(X2(1
-? - x)Q2) </2
311" 411" m b r(2 +~)
1 11 0
dvv'-'-(I -V2X
)-'- F( X,V,
)
(3.2 .18)
where xF(x, v) is given by (2.8 .33) with Xl = X and N = 4 + (. Integrating
over v yields
where I have dropped some of the terms that vanish as i -+ o. Using the
/
cI"
I' I'/I,III-lion
r (1 + t )
------'------"---<- = J -I- ~ 'YI:J (+ 1 ( 2
T6 2'YE - 7r
2)
C
2
+ ... , (:J .2 .20)
r (I + C)
w,' a rrive at
1
- -derb-R=20'.
- {I+X:!
- - - +I+x2
-- 2 [Iog (X 2(I-X)Q2) + 'YE ]
ITo dx 37r 1- x C 1- x 47rmb
-1 (derb
-d- R ) =-2
a. Pq-+qgx)log
( (2 2) +as!DRq(x),
Q/mD e+e-
ero x + 7r '
where Pq-+qg(x) is the same function that appears in (3.2.9) and
e+e-( ) _ 20'.
a. f DR,q x - -3
{(I
+ x 2) (log(l -
1
X)) + 2(11 + x 2) Iog()
X
7r -x + -x
2
3
-- I 3 5 (27r
--x+-+ - - 6) 8(I- x ) }
2 (1 - x)+ 2 2 3
Equa ~ions (3.2.1) and (3.2 .21) illustrate a feature in common with many of
the calculations we will be performing in this book. The qua rk cross scctiollti
are regularization scheme dependent. The coefficient of the logeQ2) tcrm is
universal (scheme independent) but the "little f" functions in (3.2.9) and
(3.2. 22) are different and depend on the scheme. Nevertheless these "littl e 1"
fun ctions have a unique integral. By construction,
11 + -
o fbR,q (x)dx = 11
0
+ -
flfG,q(x)dx
Furthermore, equations (3.2 .9) and (3.2.22) contain explicitly the gluon maNN
= O. (3.2.21j)
produ cing a gluon ill til e p rO(,( ~HI; ,. ql/Y ('arryi ll g f .. ,,(" t.i o ll a.! (' lI cr gy x. In
this case g g
- 1 da
- -_ -1 (da - ) + a. J c + e - u, ( x. . ) , (3.2.25)
ao dx ao dx + 9
where a.I; + 0- = a.I: +0- = a./7r. In the massive gluon scheme one gets
-1
ao
(da~G)
-d-
x +
= 2 (a.)
-
27r
Pq--+gq(x) log (2
Q /mg2) + 2a ./MG
0+ 0-
'
),
g(x (3.2.26)
where
(3.2.27)
or
(3.2.28)
and
0+0- (
a. IMG,g x) 37r {2( 1 + ( 1- x) 2) (log(x»)
= 2a. -x- + - 2x - 47 )} .
8(x (3.2.29)
Here the "+ functions" are defined with respect to the point x = O. In
addition , it is easy to show that
The next step is to embed the parton cross sections within the desired hadronic
process. What is measured experimentally in e+ e- annihilations is the proba-
bility of observing an outgoing hadron from the process e+ e- --->
carrying a certain fraction z of the available energy,
---> hadrons ,*
(3.3 .1)
/ 111/ C hll,!,!.,·,':1 I",n/"IO .. Inl 01" "'\lndioll " 1111(1 .IdH
' l'hi H illclll~iV(~ Hi'I!.; I,· II Idl'Oli nOHS I'll' ·tion is I" ' 1 1\t(~ d to 1,/1(" !lnl"toll difr"rt 'lIl,iul
" " 1-111 sec tions ac "dill !': to
L [1 XD~,qi(X)dx = 1, (3.3.7)
All h 10
(:3.3.8)
This together with (3 .2.30) a nd (3.2.31) insures th a t the norm aliza tion con-
dition in (3.1.5) holds,
C(z ) = A * B ==
1z
1 dy
- A(z/y) B(y)
y
= 11 -
z
dy
y
A(y) B(z/y), (3.3.11)
and
D'Ns( z , Q2) = D~,NS * (1 + ;; Pq_ qg log(Q2 /m~) + a. f 1.: fq) , (3 .3.13)
11I111i, ill which th(' dllll"IIIII " II II I rqJ, lIl ari;\aliu ll niltHH gOdl to ~C'J'(), thi N cIiv,'r
Illu st be a ll (\rl,irn. I' ~ , ,,1' 1,1... Wily we have dOllc th(' cn,lcll illt.i II . Fo r ('X IIIII
/', "11 (' "
JlI" , WI' have d ivi ded the olHw rvahle D~s(z, Q 2) into two tcrlrl s, IJ~ NS(X) Ilnd
I'" ' (/11(11) log(Q2/ m;) , T his latter term diverges as mg -+ 0 but D~s(z , Q'2)
111111/1, remain finite. This means that D~ NS(x) must also diverge as mg -+ 0 in
NllI'h a way that the product is finite. The function D~ NS(z) must, thereror',
"/IV" the form '
h
DO,NS( -h
z)=Do,NS* (a. 2
1+ 27rPq-+qglog(mg/A 2
)+ ... ) , (3 .3. 15)
* (a,
t il'
-II h 2 2 )
DO,NS( Z) = Da,NS 1 + 27rPq-+qg 10g(A /mg) + , .. , (3 .a.IH)
wlu' rc A is a mass scale that is related to the size ofhadrons and where D~, N'<;
III lillite in the limit mg -+ O. The mass singularities are "factored" ofl' illto
II{i ,NS as follows
The mass singularity log(m;) has been absorbed into the unknown function
j)~,NS which is assumed to be well behaved in the limit mg -+ O. The scale
1\ can be taken to be the perturbative parameter in (1.2.18) which sets th e
value of O',(Q2) at a given scale Q. At this order the "factorization of the mass
Hingularities" is a trivial property of logarithms, It has been shown, howevcr ,
that to all orders of perturbation theory, one can factor out and absorb th esc
Hingularities into the functions Do,q and D O,g.3-6
Alternatively we can write
and where 11 is the "renormalization point" used to define the strong coupling,
a . == a. (J.l2) , as discussed in Chapter 6. The scale J.l is available to us and
can be used to "factor off" the mass singularities. From this point of view
(3.3.16) can be thought of as simply setting 11 = A for convenience, since the
scale J.l is arbitrary in any case. On the other hand, we will see in Chapter
Figure 3.2 Illustration of the "decay" of a quark, q, with invariant mass ..ji into a
gluon, g, and another quark carrying a fraction %2 of the parent quark's E + pz. The
outgoing quark has transverse momentum, kT, relative to the i-axis d.:!fined along
the direction of the initial quark.
6 that choosing the renormalization point I' as the "separation mass" will
allow us to compute the Q2 dependence of the fragmentation functions using
"renormalization group equations" (RGE).
Since DO.NS(Z) in (3.3.16) is unknown we still cannot calculate the ob-
servable D'Ns(z, Q2) from perturbation theory. These functions contain in-
formation on the way in which partons turn into hadrons which is clearly
a non-perturbative phenomena. We will, however, be able to calculate the
change in D'Ns(z, Q2) as we change Q2. Since 0'. (Q2) logQ2 is of order one,
we must first sum all terms of the form [0'.(Q2)logQ2]".
There are two ways to sum the leading log contributions. In Chapter 6 we
will sum them using "renormalization group equations" (RGE). Here we use
brute force and explicitly sum the leading log contributions. Let us begin by
/
, IIl l1illill g th ',,· • If f/II dOIl I)I.. dirr(' n'lIti ul <: ros~ ~(,\ t i () 11
drr 20'. XT + x~
(:1.'1 . 1)
0-0 dx 1dx2 - 371" (1 - Xl )(1 - X2 )'
I, ()i ll yet a nother point of view. This time we introduce the li ght cone va r iH.b k ~
In the region where kj./Q2 = z(1 - z)i is small this cross section can be
app roximated by the first term. The approximation
1 ( do- ) 20', 1 + z2
0-0 dzdi LPA = 371" (1 _ z)i' (3 .4 .1 0)
is known as the "leading pole" approximation (LPAf- l l and when integra ted
over i produce", the leading log singularity. The virtual corrections ca n be
included by writing (3.4.10) in terms of a "+ function" as follows
do-, )
-1 ( - O'.~) Pq_qg(z),
= --, (3.4 .11 )
0-0 dzdt LPA 271"t
where Pq_qg(z) is the same as (3.2.10). Namely,
Integrating the leading pole (3.4.11) over i gives the leading log (LL) piece of
(0 )
q
h
~~.::,
••• '------.=-.= ~-7
(b)
•
y,O h
Figure 3.3 (a) Illustration of the case where an initial quark produced by the
"decay" of a virtual photon of invariant mass, Q, emits n gluons and has its invariant
mass degraded from t1 to tc (Q2 ~ t1 ~ t2 ~ ... ~ tn ~ t c) whereby it subsequently
"fragments" into hadron, h. (b) Square of the amplitude for the process in (a) in the
"leading pole" approximation. In an axial gauge interference terms do not contribute
to leading order and the cross section takes on a simple ladder form .
-1
Uo
(dU)
-d
z LL
= 11 -1 (
f3 Uo
du -)LPA dt-= -2a. Pq_qg{z) 10g(Q2 /m g2 ),
-dzdt 11"
(3.4.13)
where I have taken {3 = m~/Q2 and where, for the moment, I have ignored
the t dependence of a,.
We will interpret (3.4 .11) as the probability that a quark of invariant mass
Vi propagates and "decays" into a quark and gluon carrying z and (1 - z),
respectively, of its (E + pz). We would like to construct final states with, for
example, n gluons by multiplying the LPA probability by itself n times (i. e.,
independent emission) as illustrated in Fig. 3.3. Unfortunately, if we examine
the contributions to the LPA probability from the three terms in (2.3.23) we
/
I lid
I.he interference term does not contribute at leading order. In an axial gaug "
th e sum over gluon polarization states is replaced by
8 12 = 8C (1 -2
- X2)
LPA
--+
° , (3.4.23)
where C is given by (3.4.17). In this gauge (and in any axial gauge) the cross
section for th e emiss ioll s of n glliOlls ill I.Ilt' LI'A a pprox ill 1f1t.i oll h ilS 1.1 1(' simpl
ladder structure shown in Fig. 3. 3b anJ tli(' pro bability of (' Illillili g n glu olls
becomes the product of each emiss io n,
( dZ 1 . .. dZdu/u? , )
n dh . . . dtn LPA
= [0,(:J) Pq-qg( zd] [0, (t,2) Pq_ q9(Z2 )]
211't1 211't2
. .. [o,(t,n) Pq_qg(zn)] , (3.4 .24)
211'tn
where Pq_qg(z) is given by (3.4 .12) and where the invariant masses are ordered
according to
(3.4.25)
or
(3.4.26)
The interference terms are non-leading.
It is easy to generalize the LPA cross section in (3.4.24) to include all
possibilities. The LPA probability for a parton of type j (quark, antiquark, or
gluon) with invariant mass Vi to propagate and "decay" into partons of type
h a nd h carrying fractions Z and (1- z), respectively, of its (E + pz) is given
by
o,(t)
211't Pj-hh(z), (3.4.27)
The "splitting function" Pg_qq(z) will be derived in Chapter 4 but the cal-
culation of Pg_gg( z ) will be left as an exercise for the reader. The LPA cross
section (or probability) of any final state configuration of quarks, antiquarks,
and gluons is given by the product of the probabilities for each individual
emission . Thus
where Pj(z) are the appropriate Pj-hh(z) functions and where the invariant
masses are ordered according to (3.4 .25) .
It is now easy to obtain the precise form of the leading log terms in the
perturbation series and to sum these terms to all orders. For simplicity I will
take the non-singlet fragmentation function in (3.3.14) which corresponds to
1', ( )II Iptl ,. a ,",h '" ' 111 .. 1 fI/I 1"II II dao ll II,IIcI .I (1t./I
IIII' LPA cross t;(,('I,i ) 11 ill (:I 11. :lt1) alld I will ill cl u d(~ ill t1H' HII I II It II Hll bpr()(" 'HH
III which the fi nal q ll l\l'k ill Fig . 3.:3 has a n invar ia nt II I as!; Rq ll Hr ' el grclltl'l" tllhll
II lId cor responds to no emitted gluons. The next term involves on glli OIl
"lIlission
1 dO' ;;10:.(t) ,
--d = Pq ..... qg( Z) -- , dt,
0'0 z ic 27rt
wilne
o:.(t) = ,8010g(t/A2) ,
with ,80 given by (1.2.19). To integrate (3.4.33) we change vari ables t
NO that
d" o:.(t)
(3.4 .36)
di = 27ri'
I~nd the integral becomes
(3.4.37)
1 dO'
--d d = Pq ..... qg(zt)Pq ..... qg(Z2)
0'0 z1 z2
;;10:(t d
- ,- dit
tc 27rt1
0:(t2) ,
- ,- dt2 .
tc 27rt2
,;;il (3.4.4 0)
After changing variables to " we are left with the "nested" integral
(3 .4.4 1)
The factor of 1/2 is crucial and arises because of the way the integrals are
"nested" (the area of the triangle formed by dividing a square along its diag-
onal is 1/2 the al'ea of the square) . The series now becomes
D'Ns(z, Q2) =D8,NS + "Pq ..... qg * Dg,NS
+ t,,2 Pq..... qg * Pq..... qg * D8,NS + . . . . (3.4.42)
It is easy to see that the series is that of an exponential. The n-th term involves
the nested integral
1" 1'"
o
dk 1
dD'lvs(z, Q2) _ P
d", - q-qg * DhNS (Q2) , (3.4.46)
or
(3.4.47)
where
(3.4.48)
and
(3.4.49)
Equation (3.4.47) is the usual Altarelli-Parisi 12 form of the equation that gov-
ern the Q2 evolution of the non-singlet fragmentation function D'Ns(z, Q2).13
The solution of (3.4.47) is (3.4.44). At leading log order the "little f" functions
do not contribute.
In (3.4.44) we have succeeded in separating the perturbative part, e"Pq - qg *,
(i.e., the gluon ladder in Fig. 3.3b) from the non- perturbative part, D'Ns(z, te),
which described the shaded blob in Fig. 3.3b. The fragmentation function
D'Ns(z, te) involves momenta scales less than the cut-off te . Previously we
called this function D~ NS{z) or Dg NS{z) and we were not careful in its def-
inition. The parton mo'del fragment~tion functions in (3.1.42) can be used to
parameterize the uncalculable D'Ns(z, tc) functions.
It is not difficult to deduce the equations that govern the Q2 dependence
of the individual quark and gluon fragmentation functions. From (3.3.12) we
have
Dqh (z, Q2) =Do,q
-h
* (a.
1 + 271" Pq_ qg log (Q2/ A)
2)
/
1\1
Figure 3.4 Illustrates that the leading order rate of change of the quark fragmen-
llltion function, D:(z,Q2), with respect to " is generated by the two terms th at
rise when the quark radiates a gluon. The observed hadron can originate from th .
r 'sulting quark jet or the resulting gluon jet. Similarly the leading order rate of
('hange of the gluon fragmentation function , D;(z,Q2), with respect to " is ge n-
nated by two terms. The incident gluon can produce a quark-antiquark pair th at
fragment into the observed hadron or it can radiate a gluon which fragments into
the observed hadron.
+ -Do,g
h
* (0', 22)
2", Pq_ 9q log( Q / A ) (3 .4.50)
and similarly
where the mass singularities have been absorbed into the unknown D~ fun c-
tions and the "little r functions have been dropped since they produce cor-
rections beyond leading order. The summation runs from 1 to 2nJ since it
includes both quarks and antiquarks .
Differentiating (3.4.50) and (3.4.51) with respect to T defined in (3.4.48)
a nd keeping only leading order gives
(3.4.53)
which because of (3.4.49) can be written as
dD;(z, Q2) h 2 h 2
dK = Pq- qg * Dq (Q ) + Pq- gq * Dg (Q ), (3.4.54)
h( Q") 2n,
dDg z, -
dr;; = '.LJ
" h 2 h 2
Pg-->qq * DqJQ ) + Pg-->gg * Dg (Q ), (3.4.55)
j=l
and is illustrated in Fig. 3.4. These coupled differential equations are solved
by defining a "singlet" fragmentation function
n,
D~(z, Q2) =L [D;;(z, Q2) + D~Jz, Q2)] , (3.4.56)
i=l
(3.4.58)
and
Pq-->qg(Y) 2n J Pq-->gq(y») .
P(y) =( (3.4.59)
Pg-->qq(Y) Pg-gg(y)
The non-singlet satisfies the equation,
dD'Ns(z, Q2) _ P
d ,..
- q-->qg * DhNS (Q2) . (3.4.60)
~, I V " H
,.
J) N.o.;{ Z,
2 - II
Q ) = DO.NS + 2'11" log(Q
(\' . 2/ A2) -Do
h - II (
,Ns* Pq-+ qg+a.Do, Ns * fq , 3.IJ .G:I
)
where P =
P q-+ qg , I =
I q , and l' = 10g(Q'2 / A'2) . The terms in (3.4.65) co ntain
Ili g a factor of l' are process independent but may depend on the regul a riza ti oll
H('\leme. They arise from the mass singularities. The "constant terms" (i. c.,
f(y), g(y) , etc. ) are, in general, process and regularization scheme depen-
dent. To compute the rate of change of D'Ns with respect to 10g(Q'2) we use
pert urbation theory (in a.) to deduce how D'Ns changes with a. but then we
IIiUSt include the fact that a. itself depends on 10g(Q'2) according to
da. '2
- dT = boa. + bi a. 3 +"', (3 .4 .66)
where bo = f3o/(4'11") with f30 given in (1.2.19) and where bi will be gi ven in
Chapter 6. Since QCD is a renormalizable theory a change in the ultrav iolet
cut-off, M elJt , is related to a change in the coupling a •. If one neglects qua rk
masses then the lagrangian in QCD is scale invariant so that changing Q a nd
Mcu t together makes no change. Therefore, a change in Q can be rela ted to
it certain change in a. and one arrives at a type of "renormalization group"
equation,
(3 .4 .67)
Inserting (3.4.65) and (3.4.66) into (3.4.67) and keeping terms to order a. 2
gives
h 2)
ciDNS(z, Q = 0', P * D'N S(Q2), (3.4 .69)
cir 27r
which is the same as (3.4.47) .
2 4 0.327 0
4 16 0.252 0.063
10 100 0.193 0.127
50 2,500 0.137 0.210
100 10,000 0.121 0.238
500 250 ,000 0.096 0.314
Suppose we keep only the first two terms in (3.5 .1), then from (E.3.7) we
/
11 :1 C h ll p kr :1 1"1 /1 11.\1 111 1 " " "'"1 11 1,11 1" 14 1\,,,01 .I d ~
, . ' •• I. h at flS Z -+ I,
li t ( ~ O~)
D~ S ( :' Q2)
NS ~ ' 0
.::11 + I\, 10g(1 - z ).
No Ill a tter how small I\, is, 10g(1 - z ) becomes large as z -+ so th at. 0 11('
IIIII HI. include higher and higher terms in ,.., . This behavior a rises b C<W H(' o f
1.111 \ 1/(1- y)+ term in Pq-+qg(y). Perhaps we can find another funct ion , / )/1 (y)
wll,1I the same singularity structure as y --+ 1, but which we can exp li citl y do
III(' slim
exp("'PA*) D'Ns == R* D'Ns · (:J .1).:\)
II. is not hard to find such a function . Consider the function
PA(y) = ~ (-IO:(y») + .
This fu nction has the property that its moments are
Since PA(Y) and pq-+q(y) have the same singularity structure as Y --+ 1 (i .e.,
-log(y) l-y) , PA(Y) is well behaved and one can remove the "+ function s"
"V
leaving
where IE is Eulers constant (see Appendix E). We can now expand exp(I\:PA *)
in (3.5.8) and keep only the first two terms. This gives
D'Ns(z, Q2) = R * D'Ns(Q6) + I\:R * PA * D'Ns(Q6) + 0(1\:2). (3.5.11)
Defining
i5'Ns(Z, Q2) == R * D'Ns(z, Q6)
-1
-
z
1
dy Dh ( / Q2) (-log(y)) ~1<-1
Y
NS Z y, 0 r (~)
31\:
, (3.5 .12)
we arrive at
h
DNS(z,Q 2) = DNS(z,Q-)+,..
-h ., 11
z
- h (
-dy DNS
y
2 ) PA(Y)+O ( I\: 2), (3.5.13)
z/y,Q
where PA(y) is given by (3.5.10) and K is defined in (3.4.35). Using this equa-
tion one can calculate D'Ns(z, Q2) from a main term, i5'Ns, and a correction
term that is proportional to 1\:. Equation (3.5.13) is not exact but is very ac-
curate even for,.. as big as 1/3 (see Table 3.1). Of course, if the first correction
term is not small then one can calculate the order ,..2 correctiOll term as well.
We see that in order to compute D'NS(z,Q2) we need to know the frag-
mentation function at a reference momentum Q5, D'Ns(z',Q5), for values of
z' in the range z :S z' :S 1. In addition, we need to know I\: which is calculated
from the knowledge of a.(Q2) which in turn depends on the QeD parameter
A. Thus, given D'Ns(z, Q5) the one parameter A governs the Q2 evolution.
The Q2 evolution of the singlet and gluon fragmentation functions is a bit
more complicated. As can be seen from (3.4.61) D~(z, Q2) depends on both
D~(z, Q6) and D;(z, Qij). Similarly D;(z, Q2) depends on both D;(z, Q5) and
D: (z, Qij). The manner in which this mixing occurs is governed by the matrix
P(y) in (3.4.59). Equation (3.5.8) is not valid in this case since the matrices
P A and P A do not commute. In this case we write (3.4.57) as
where
P A _- (aqP0 O
(3.5.15)
with
'1
Po(y) = (-log(y))+ (3 .5.16)
and
(3 .5.17)
/
DASP
SlAC-lBl
07.4 GeV
6 12 GeV
• 27.4-31.6 GeV
J TASSO
• 35.0-36.6 GeV
,0' o
.o
Figure 3.5 Data on the Q2 dependence of the charged pion fragmentation function
compared with QeD predictions (taken from Ref. 14).
(3.5.18)
where
(3.5.19)
with PA(y) g iv ' lI ill p .!>. !» Bile!
PAg(Y) = Pg_gg(y) - ClgPO(y)
y I i-y
= 6{ - '- + - - + - - + y(l -
1 - Y log(y) y
y)
(3.5 .22)
+" ""1
2nf 1 d -h
L.. - y Dv(z/y,
- Q 2 )Pg-qij(y), (3.5.23)
j=l Z y
where
(3.5 .24)
-h
The function D; is new and is computed by averaging Df(a, z, Q2) over a.
Namely,
(3.5.25)
p .I) .2G )
Jo'/)r quarks,
(:3 .5.29)
(3 .5.30)
(3 .5. 31 )
For example, if D~(z,Q5) = D;(z,Q5) then at higher values ofQ2 th e glu ofl
fragmentation function would fall off with a higher power of (1- z) as z -+ ] .
The difference in the power would be given by
(3 .5. 32)
At large Q2, gluon fragmentation functions are softer than the correspondill g
qu ark fragmentation function.
(0)
--+-"-:--+---i....--- • • • •
(b)
Figure 3.6 (a) Illustrates the case where an initial quark of invariant mass Vi
emits n unresolvably soft gluons until its invariant mass has been degraded to Vic.
This configuration of a quark and its accompanying soft gluons is defined as an
outgoing "final state" quark. (b) Illustration of the case where one hard gluon is
emitted followed by the emission of unresolvably soft gluons.
In the real world, on the other hand, quarks do not have an infinite amount
of time to radiate because at some stage non-perturbative effects take over
and hadrons are formed. We will use both an invariant mass cut-off, tc = I'~,
and a cut-off in z. Par tons with an invariant mass less than I'c are allowed to
propagate freely to the final state with no additional radiation. Also, we will
consider partons with z < Zc as unresolvably soft.
Given a quark with a maximum possible invariant mass of Vip, we must
decide whether it will emit a resolvable gluon or not. Let 1rq(tp, t c ) be the prob-
ability that a quark of maximum invariant mass Vip emits only unresolvably
soft gluons (i.e., z > 1 - zc) until its invariant mass has been degraded to
Vic. This situation is illustrated in Fig. 3.6a. To compute 1r'l(tp, t c) we must
integrate (3.4.24) over the appropriate regions and sum over all possible gluon
emissions. Namely,
/
till
(3 .6. 1)
W lil ' l't' the invariant m asses are ordered according to (3 .4.25) a nd
(:I.G .:.!)
l 'Ii.· nested integrals over the invariant masses are the same as in (3.'1 . I\~I ) ,
(3 . 6 . ~)
which m eans that the integration over the soft gluon region can be arrived a t
follows
IlH
(3.6.6)
with
iNS(O) = O. (3 .6.7)
We now have
(3 .6 .8)
or
(3 .6.!J)
with
(3.6.10)
Given that a quark does decide to produce resolvable radiation then we m ust,
produce correctly its invariant mass. Its invariant mass is given by (3.4 .24)
modified by possible subsequent emissions. We must calculate the invar iant
mass distribution for the situation shown in Fig. 3.6b in which one gluon is
"hard" and all the rest are soft. We will label this distribution by 3 q (t). Ac-
t ually we already know this distribution because we know that if we integra te
3 q(t) over t from tc to tp we must get 1 - 7rq(tp, tc). Namely,
(3 .6.11)
.t.O .1.. 1, 'I'h.· MOld.•' ( ) III" M" t,II " d fHl
or
(:3.6 .12)
(0)
(b)
Figure 3.7 (a) Illustration of the "decay" of a virtual photon of invariant m ll.8R
Q into a quark with maximum possible invariant mass =
Vip Q and an antiquark.
(b) Quark 1 in (a) subsequently "decays" into a gluon and quark 2 of maxim ul1I
invariant mass Vip = Vib where tl is the invariant mass squared of the gluon and
quark 2.
(3.6. 17)
or
(3.6.18)
with
(3.6.19)
a nd
(3.6.20)
(3.6 .21)
Step (i) proceeds as before but using (3.6 .18). Step (ii) is also the same except
now the t-distribution is given by
,::,.() __ d7rj(t,tc)
~J t - dt' (3.6.22)
Having determined that parton i is to radiate, and selected its invariant mass
t step (iii) is modified slightly. It becomes
(iii') The decay products hand h (j -+ h +h) and the value of z is determined
according to (3.4.27).
The partons it and h are allowed to evolve as was parton i, and the cas-
cade continues until all partons have chosen to generate no further resolvable
radiation with t > te '
Parton showers generated in this manner can be viewed almost classically.
"Virtual" partons will live for a time, ~T, determined by the degree to which
they are off-shell,
~T '" 1/~E '" (E + 1P'i)lt '" pit, (3.6.23)
a nd then "decay", where p and t are the magnitude of the momentum and
invari ant mass squared, respectively, of the parton. The time for the emission
of a gluoll is roughly lip, so as long as
lip ~ pit (3.6.24)
the independent emission (LPA) assumption is accurate. Equation (3.6.24)
is satisfied except perhaps at early times where t '" Q2. The development
of a parton shower is illustrated in Fig. 3.8. At early times (short distances)
invariant masses are large (on the order of Q). Here interferences are important
and gluons may be emitted at large angles where the LPA formula is not
accurate. One can correct for this by using (at this stage) the complete order
IX$ formula in (3.4.9). At later times the invariant masses of the partons are
smaller (because they are ordered as in (3.4.25)) and the angles of emission
are small. In this region one is justified in assuming independent emission,
interferences are not important and the LPA approach correctly describes the
development. At even longer times the invariant masses become small and
comparable to A. Here IX$(t) is large, non-perturbative effects dominate, and
the hadrons are formed. Presumably in this region interference effects are also
important. Fig. 3.9 shows an e+e- event at Q = 100GeV generated from a
QCD parton-shower Monte Carlo Model.
/ 11 .1
PARTON SHOWER
I
l>
o
::0
q o
Z
(f)
"
o
::0
s::
Invariant moss t
) >
large small t
time ) . ~
short time long time
distance )
short distance large distancE?
In addition to the Monte Carlo methods discussed in the previous section, one
can perform analytic calculations using the leading pole formula in (3.4.31) .
For example, suppose we are interested in the small z behavior of quark and
gluon jets. Let us define
PJ(tp , t e , zo) = the probability that quark with maximum invariant
mass, Vip, emits n gluons (as in Fig. 3.6) each
of which has its (k o + kz) ~ zoQ/2 while the quark
invariant mass is degraded to Vie. (3.7.1)
,
•.1
.
=
Figure 3.9 An event at Q 100 GeV genera.ted from a QCD parton-shower Monte
Carlo model. The event conta.ins a. ee, dil, and s:; quark pair together with 14 gluons
resulting in 47 particles with 32 charged (l!'0 = 7, l!'+ = 10, l!'- = 12, K+ = 6,
KO = 2, K- = 3, ~ = 5, lIiJ = 1,1'- = 1).
Similarly P~(tp, t e • zo) is the probability that a gluon emits n gluons each of
which has (k o + kz) ~ zoQ/2. We will consider only the case where Zo ~ 1 so
that we can approximate (3.4.12) and (3.4.30) by
(3.7.2)
/
\I <: 11111'1,/' " :
(:1 .7.:1)
I ""p(·(·t.ively, wher (I" II. lld (/. 1,11.1'(' th e same as in (3.5.17) . Usin g th e LPA e r O NN
where the invariant masses are ordered according to (3.4.25) and wI! "(' (I tI "
li nd a = a g for a quark and gluonjet, respectively. The range of th e z illl.l'l!, l'lI h(
I ii deduced from relationship
(tQ)
(:l .7,1i)
liO that
(3 .7.6)
The nested integrals over the invariant masses are the same as (3 .6.3) ,
1
tp
tc
a(td
-2-dt1
71't1
1t1
tc
a(t2)
-2-dt2 . . .
71't2
1 tc
tn
-
1
a(t n )
-2-dtn
71'tn
= {"P dK1 {"1 dK2 ... ("n-1 dKn = h:~ ,
(3 .7.7)
10 10 10 n.
where Kp is defined by (3.6.4). Namely,
2
Kp = /30 log{a.(tc)/a.(t p )}. (3 .7.8)
T he z integrations are done by changing variables to
Y; = log(z;), (3 .7,!J)
so that
dYi = dz;/z;, (3.7.10)
and
log [ZO/(Z1Z2 .. . zn)] = Yo - Y1 - Y2 - .. . - Yn . (3.7 .11)
With this change of variables,
In -
1
- 1 dZ
1 11
dZ 2 ... 11
10
Zo Z1 ZO / Z1 Z2 ZO /( Z1 Z 2 ... Z n_d Zn
= 0 10
1 Yo
dY1
Y O -Y1
dY2 ...
Y O-l'1- .. ·-l'n-1
dYn , (3.7.12)
which for n = 2 is
(3 .7.13)
and for n = 3 is
where h(p) are the modified Bessel functions of the first kind. Thus,
so that
exp (2 Jan.p 10g(1/ ZO»)
P(tp, t e , zo) - -O . (3.7.22)
Za- /47rJaK p 10g(1/zo)
The average number of gluons with (ko + kz) ~ zoQ/2 emitted by the jet
/
11'1
I I ",IVI 'II by
) ~~ ' o lll,.(t p , lc, zo)
( N)
P(lp , l e, zo)
'1'111 ' 1IIIII1erator can be wl'ilt. II
(3.7 .27)
<: luon jets are expected to have, on the average, a higher multiplicity and a
"softer" (i.e., less high momentum) distribution of particles than quark j ets.
+ -
We have seen an example of an observable, such as utQ/ in (2.7 .8), that
contains no infrared divergences. Both the log squared and the singl I g
canceled out when real and virtual corrections were added. \Ve have also
seen an example of a cross section, such as du h /dz in (3.3.6) that contains (t
single logarithmic divergence of the form 0'$(Q2) log(Q2) . Since this term is o r
order 1, all terms of the form [O's(Q2) log(Q2)]n were summed to arrive a t a
meaningful result . There are also observables, like U m in (2.3.50) that contain
double logarithmic divergences of the form 0'$log2(Q2). Here again one must
sum all terms of the form [0'5 log2( Q2)]n in order to obtain a meaningful result .
An interesting observable of this type is S( 0) which is illustrated in
Fig. 3.10 and defined by
S(O) = the probability that the quark created in the decay
,*
-+ qij is diverted from its initial direction (opposite
Figure 3.10 illustrates the case where the quark crea.ted by the "decay" of a
virtual photon of invariant mass Q is diverted by an angle 8 from its initial direction
(opposite the antiquark) by the emission of n gluons.
_ A
z-t Plane
0.0 """'-L~LL4~'"""""-'~~~--"'----'"
0.0 A-
1.0
t
=
Figure 3.11 Shows the z - i plane, where z 1 - x. The solid curve is zt = (} 2/ 1
and Rl is the region of integration for 8(8). The complementary region R2 is tt. t'
range of integration for T(8) and 8(8) + T(8) = 1.
Fig. 3.2 and to arrive at what will turn out to be the leading double logarithlll
we can use the LPA approximat.ion in (3.4.10),
1 ( d(1' ) 20', 1 + z2
(1'0 dzdi LPA = 311" (1 _ z)i' (3.8 .7)
This term dominates since the range of integration for S(O) is from (3.4 .6)
given by
k2 02
Q; •
= z(l-z)t $ 4' (3.8.8)
a .1l , 1111111 1i 1l f{ 1)0,,111., 1,,,,,.,, ,,1.111 11 11 lUI)
(T1(O»LDLA = 2a,
311' Jl.92
r 1
d,t
t
'
r 1
JiJ..
2~z
Z
4 .,
1 2
_- -2a'
- 21 -".. di log (0-:: )
311' .1.92 t 4t
•
(3 .8.15)
(S1 (6»LDLA = 1 -
2a, 2(?)
311' log O~ /4 , (3 .8.16)
where the subscript LDLA refers to the leading double log approximations
that we have made and the subscript 1 means that we have computed only
the single gluon emission contribution .
The quark transverse momentum spectrum can be computed easily since
from (3 .8.5)
~ dO' q (Q k 2 ) _ dS(kf) __ dT(kf)
(3.8.17)
0'0 dkf 'T - dkf - dk}
/
101
which is valid provided k,/;,/Q2 ~ 1 and which becomes infinitely larg(' i\H
q. O. The single gluon emission result in (3.8.18) does not giv (\ good
II pproximation to the correct answer obtained by summing over a ll I oHs ihk
/" llI on emissions.
The leading double logarithm summation is a bit more difficult to <i (' <iIl Ct '
I.hall the leading log summation. The LDLA result for n gluon rni fiH ioll H i i
)',iven by
1 [20'$
(S,,(O))LDLA = n! 2(2 )] n ,
- 371' log 0 /4 (:UUII)
Equation (3,8.16) was just the first two terms in the exponential series. J 5, LG
The k1' distribution in (3.8,18) which diverged at k1' = 0 becomes
-
1 (d<1)
dk2
40's 1 .,., {2a.
= --3 k 2 log(kT/Q-) exp --3 log (k1'/Q)
2 2 2},
<10 l' LOLA 71' l' 7r
(~3 , 8.22)
which now vanishes at k1' = O. The exponential factor,
F(k}) = exp { - 23~ log2(k}/Q2)} , (3 .8.2:1)
converts the single gluon result in (3.8.18) to the correct answer in (3 .8.2~) iii
sometimes referred to as the "Sudakov form factor" .17 When properly incilld d
the k1' distribution vanishes as k1' ---+ O. Once one allows for an infinite am Uili.
of gluon emission, it becomes very unlikely (in fact, infinitely unlikely) thaI.
the quark in Fig. 3.10 remains at k1' = 0 and is not deflected. The situatioll
becomes more complicated when effects due to the running coupling constant
as (k 1' ) are included. I refer the interested reader to the literature. 18 - 25
Problems
3.1. Show that (3.1.23) together with the normalization condition 011 J(T!) given
3.3. Show that the integraJ equation (3. 1.23 ) gu arantees that zF( z ) ~ R as
z---+ O
3.6. Show that Pqab(z) given by (3 .1.30) satisfies (3 .1.35) provided F(z) = F(z)-
/(1 - z).
3. 7. Show that as z -> 1 the fragmentation function F(z) in (3.1.23) approaches
/(1 - z ) and hence, Dqh(z) ,. . ., Ahq/(1 - z).
z---+l
3.8. W h at are the fragmentation functions DuK+ (z), DuK- (z), D$K+ (z), and
D!,< - (z) in the FF parameterization? (Assume that /3c = /3b = 0.) Us-
ing t he result from problem 6 show that D!f+(z)ID~+(z) -> (1- /3)1/3
as z -> 1. Da t a on the production of J{+ and 71'+ mesons in proton-proton
collisions a t large transverse momentum, PT, show that as PT becomes large
the J{ + 171'+ ratio a pproaches 1/2. What does this tell us about /3?
1 {I + x2
( I-x + = (l-x)+ +~6(1-x) }.
+ x2)
3.13. Using the definition of "+ functions" in (3 .2.8) verify (3.2.16) and (3.2.17).
I
10!1 ( :II ,.p l. •.• .1 FIII/IIII Itl lI l "II 1,'1111.1,111 11 ' h lld .1 01,,.
.I . I S. Let
1= 11 dv v~(l- v)~xF(x) v),
where xF(x, v) is given by (2.8.33) with Xl = x. Show that
1- f2(1+~) {1+X2 ~+!_l__ ~x+~+ . .. }
- f(l + f) 1- X f 21 - X 2 2 '
where some terms that vanish as f -+ 0 have been dropped . Is it ncccRl'l a.ry 1,1)
include the terms of order f2 in the expression for xF(x, v)? Use tllis fI ·'HoIl.
to verify (3.2 .19) . Also, verify (3.2.21), (3 .2.22), and (3.2.23).
3.19. Using the dimensional regularization scheme, DR, show that the differe n-
tial cross section for producing a gluon in the process /,. -+ qijg carr yin ~
fractional energy X is given by
(
1 ()
f2+2" a
11
dvv ~2 (1-v) ~2xF(x,v), -
where F(x , v) is arrived at from (2.8.24) by making the replacements XI =
1 - X + xv and X2 = 1 - vx . Perform the integral and verify tha t
1
0'0
(dlTbR) (as)
~ + = 2 271' Pq ..... gq(x) log(Q 2 /7nD)
2 e+e-
+ 2asiDR,g (x ),
3.21. Show that (3 .2.30) and (3 .2.31) together with (3 .2.1 2) in sures that th e nor-
malization condition in (3 .3.9) holds at leading order and at order a •.
3.22. Verify (3.4.4), (3.4.5), (3.4.6), and (3.4.7). Derive (3.4.9) from (3.4.1) using
z = Z2 defined in (3.4.7) and i defined in (3.4.8) .
3.23. Using the axial gauge defined in (3.4.18) choose n = q- P2/X2 and derive
(3.4.21), (3.4.22), and (3.4.23).
3.24. Derive the gluon "splitting function", Pg_gg(z), in (3.4.30).
3.25. Compute the order ,.3 term in the series in (3.4.42) and convince yourself
that the series exponentiates as stated in (3.4.44).
3.26. Show that, if instead of behaving as given in (3.4.34) a. were constant, then
the series for D'Ns(z, Q2) in (3.4.44) would be
3.27. Show that Ptl.(Y) in (3.5.9) is equivalent to (3.5.10), where 'YE is Eulers
constant,
1 00
e- X log(x) dx = -"YE.
3.29. Suppose that at Qo = 2 Ge V that the quark and gluon fragmentation func-
tions are equal and given by,
Show I.hat
/
wh(:r '
What is D;( z , Q'2)1 D~( z , Q2) at Q = 1000 GeV, z = 0.9, if A = 200 M\,Y
and nf = 4? What does this imply about gluon jets compared to QII ;trk jcl,,'r
.I.:to. Calculate the explicit form of "YNS(ze) in (3.6.6) . What is the numcrica.l va.lll('
of "YNS(Ze) for Ze = 0.01? In the LPA what is the numerical va.lll e or tilt'
probability that a quark produced by a virtual photon in an e+e- a.llllihil ll.-
tion at a center-of-mass energy of Q = 10 GeV emits only unrcsolvah ly 1'0 11.
gluons until its invariant mass has been degraded to Vic , if fie = 2 (;"Y ,
=
Ze = 0.01, A 200 MeV, and nJ =4 (consider only the non-sill!,;kl. pnol l
F'urthel' Reading
S.D . Ellis, "Jets and Quantum Chromodynamics," Lectures prese nt d a t.
I.he NATO Advanced Study Institute on Quantum Flavordynamics , Quanl.lIl1l
Chromodynamics, and Unified Theories, Boulder, 1979, ed. by ICT. M ahall
I.happa and J . Randa, (Plenum Press, 1980).
A. H. Mueller, Physics Reports 73, 237 (1981) .
References
10 . R.Odorico, NucJ. P/,ys . Dl!W , 189 ( 1982) . ibid Pity . Le/'/' . 1180,151
(1982) . ibid Nucl. Pllys. D228, 381 (1983) .
11. K. Konishi, A. Ukawa and G . Veneziano, Nucl. Phys. B157, 45 (1979) .
ibid Phys. Lett. 80B, 259 (1979) .
12. G. Altarelli and G. Parisi, Nucl. Phys. B126, 298 (1977).
13. J.F. Owens, Phys. Lett. 76B, 85 (1978).
14. R. D. Field and S. Wolfram, Nucl. Phys. B213, 65 (1983).
15. T. Kinoshita and A. Ukawa, Phys. Rev. D13, 1573 (1976). ibid Phys.
Rev. D15, 1596 (1977). ibid Phys. Rev. D16, 332 (1977).
16. J. Frenkel and J.C. Taylor, Nucl. PllYS. D109 , 439 ( 1976). ibid Nucl.
Phys. B116, 185 (1976).
17. V. Sudakov, Sov. Phys. JEPT 3, 65 (1956).
18. Yu. 1. Domshitzer, D.L. Dyakonov, and S.1. Troyan, PllYs. Lett. 78B,
290 (1978).
19. W.J. Stirling, Nucl. Phys. D145, 477 (1978) .
20. G. Parisi and R. Petronzio, Nucl. Phys. Dl54, 427 (1979).
21. C.Y . Lo and J .D. Sullivan, Phys. Lett. 86B, 327 (1979).
22. A.II. Mueller, Phys. Rev. D20, 2037 (1979).
23. J .C. Collins, Phys. Rev. D22, 1478 (1980).
24. II.F. Jones and J. Wyndham, Nucl. Phys. B176, 466 (1980).
25. S.D. Ellis and W.J. Stirling, Phys. Rev. D23, 214 (1981).
' II i\ 1" I' I '~ H ,I
107
The first evidence th a t th(' PI' tOil It a.s >4 ,,111-1 1,1'11 (' ('111'(' (, :IIII t' 1'1' 0 111 d"(' 11 ill<'la.!:i ti c
electron proton scaUering (e + ]> --+ (' + .\). II. i ~ pro ha.bly tlt e 111 0. t .tudi ed
QCD process. Not only does this process a ll ow for tests of pcrt.urbat ive QC D
but also provides us with knowledge on the di stributi on of pa rtons (q ua rks and
gluons) within the proton. It is important to understand the precise definition
of these distributions for they will be used in later chapters to make predictions
of other experiments.
That is, the net number of each kind of quark is just the number one arrives
at in the simple non-relativistic quark model.
If we define
D == 11 ;rd(x)({;v, (4.1.7)
II Illy
Q+Q+G'= 1,
wll Pre
Q=U+D+S+C+ .... (~ . I . I 0)
III the total fraction of momentum carried by qua.rks and
Q=U+D+S+C+ ... , ( 1\ . I. I I)
If! the total fraction carried by antiquarks. The dist.ributions of partons withill
II. neutron are arrived at using isospin symmetry, which implies that
Gn-+u(x) = G'p .... d(X) = d(x), (1 . 1.1 2)
Gn-+d(X) = Gp-+u(x) = u(x), (4 .l.J:J)
Gn-+.(x) = Gp ..... (x) = Sex) , (4. 1.11\)
G'n-+g(X) = Gp-+g(X) = g(x), (4 .1.1 5)
d ,c.
In the naive parton model, knowledge of the deep inelastic structure fun c-
ti ons for electron, neutrino, and antineutrino scattering constra.ins th e t/.( x),
d(x), u(x), d(x), sex), sex) , c(x) and e(x) distributions. For example,
4
F;P(x) = -x[u(x) +u(x) + c(x) + e(x)]
9
1 -
+ gx[d(x) + d(x) + sex) + .5(X»)' (1 .1.16)
4 -
F;n(x) = gx[d(x) + d(x) + c(x) + e(x)]
+ ~x[lt(x) + u(x) + s(x) + s(x»), (~ . J.1 7)
Figure 4.1 Collision of a virtual photon, i* , and a nucleon, N, in the parton mod .1.
The photon interacts with a parton carrying momentum fraction, e, of the nucleon.
which as _q2 = Q2 _ 00 =
and p.q Mv - 00 gives
_q2 Q2
e=--=
2P·q
--=x,
2Mv
(4 .1.45)
and
F;P(x) + F2n(x) 5
(4.1.47)
xF;P(z) + zF;n(z) = 18'
Both (4.1.46) and (4.1.47) are in accordance with data3 . Furthermore, the low
Quark Distributions Gp-q (x)
0.8 .....-...,--.------,---.------,-.......:........-:---,----.....--...........-...
0.6
--
)(
)(
0.2
Figure 4.2 Quark distributions within the proton (taken from Ref· 4).
e- X
P~
(0)
P
p.
Nucleon
X X
(b) (c)
Figure 4.3 (a) Deep inelastic electron nucleon scattering, e+p -+ e+X, is related to
the (b) virtual photon nucleon total cross section, "y. p -+ X, which can be express ·d
in terms of the (c) virtual photon quark total cross section, "y. q -+ X.
The square of the invariant amplitude for the inelastic scattering of an electron
off a nucleon shown in Fig. 4.3a is given by
-2 e4
IMI = Q4LI'VWI'V' (4.2.1)
where
(4.2.2)
is the square of the virtual photon 4-rnornentum and where the spin averaging
factors have been absorbed into the tensors Ll'v and WI'V' which describe
the structure of the leptonic and hadronic vertices, respectively. The leptonic
tensor is identi cal to (2. 1. 2) H lld is gi veil hy
p, = G)' (4.2.9)
(4.2 .11)
and
p.q = Mv, ( 4.2.12)
where M is the nucleon mass. The differential cross section in the laboratory
/
The structure functions It\ and {IV:! are related to the virtu a l ph t O il
IIl1 cleon total cross section shown in Fig. 4.3b . The cross section fo r ti ll' lI il
II000 plion of a virtual photon with helicit.y A is given by
* 4~2a
0').(, N) = --CI'(A)Cv(A)TVl'v)
qlab
* .
(-1 .:U '1)
where CI'(A) is the polarization 4-vector of the virtual photon wit.h bcli ·il.y >.
1I,Il d qlab is the magnitude of the 3-l110mentum of the virtual photon ill t,iJ('
I;tboratory frame. To write this in an invariant form we notice th at th nu x
factor in (A.3 .15) implies
Mqlab = [(p.q):! + Q2J\121~) (4 .2.15)
which gives
4~2a
0">.(,* N) = cl,(A)C~(A)nll"" (4 .2.1 6)
vv2 + Q2
We will also write this in the form
(4.2. 17)
where
4~2a
O"l'v(,* N) = TVl'v, (4 .2. 18)
vv2 + Q2
If we sum the virtual photon polarization states by the replacement
where I have labeled this cross section with the subscript ~ to denote that
(4.2.19) has been used. This cross sectioll is a combination of the transverse
and longitudinal cross sections.
The cross section fo r abso rp tio n of t.r UIl 5Vcrs{' pll o(, ns (A = ± I ) pro pa-
gating along the i -axis with 4-momcnt ull1 given by
qO)
o ( //
0 )
qJ.l =( 0 = 0 ' (4.2.21 )
q3 VI/2 + Q2
is arrived at by using
(4.2.22)
and computing
(4.2.23)
The result is
"N) 471"20'
O'T (I = Vl/ 2 + Q2 IVl . (4.2.24)
The cross section for absorption of longitudinal (or scalar) photons can
be arrived at by using
,.(A ~ ~ V~q'
0) (D ~ ~ (VV'C') , (4.2.25)
IL _
( 1)~ -
(P . q) ~ ) - !
--2-
'I
,
II I
Q
a = ,Af JI/2 + Q2'
IIH ilig
(1'('\ = 0) = a (PI' - P:,lJ q,,) , ( 1\ . '2 .:1:1)
'1-
with a given above the projection in (4 .2.16) becomes
2
(1' ('\ = O)(:P = O)WI''' = a (PI' - : / 'II') (p" - :~q 'I,,) W,III
= 2
a pl'P" WI''' , ( /1.~UI\)
Hlnce WI''' satisfies (4.2.4) and (4.2.5). It is easy to see that
2
a pI'P"WI''' = -Wl + (1 + ;:) W2 , (4.2 .35)
=J :::a Q2 { - W+ (1 + ;: ) W2} ,
1 (4. 2,:36)
where
( 4.2.39)
and
( 4.2.40)
We are going to be interested in the region Q"2 large and x fixed, where x is
the scaling variable in (tJ .1.tJ 5) . Nam ely,
Q2 Q2
x------ (4.2.41)
- 2P'q - 2MI/'
so that
v2 V
1+ -Q2 ----+ --, (4.2.42)
Q2 large 2111 X
;r fixed
and
(4.2.43)
= 8 ~2
7r~O:X
[O"E(,* N) + O"L( ,* N)], (4.2.44)
The structure functions Fl and F2 are related t.o TVl and v TV2 according to
Fl(x, Q2) == MWl(x, Q2), (4.2.47)
F2(X, Q2) == VW2(X, Q2). ( 4.2.48)
and it is convenient to define
(4.2.49)
and
F 2(x, Q3) = ..!:.F2(x, Q2). (4.2.50)
x
The longitudinal and transverse st.ructure funct.ions are defined by
FL(X, Q2) = F2(x, Q2) _ 2xFl (x, Q2), (4.2.51 )
F1'(x, Q2) = 2Fl (x, Q2), (4.2.52)
respectively, which can be written as
FL(x, Q2) = F2(X, Q2) _ Fl(x, Q2), (4.2.53)
F1'(x, Q2) = Fl(x, Q2), (4.2.54)
/ J 10
~ I , ~
.I I . ( J ' ,(j ) - - /-d J; ,Q ). ( 1.£.1)1»
x
" '''1111 ( /1.2.14) and (1. 2.11) w(! see th a t
.) (Q 2
Fdx, Q-) = 2 - 8 " ) O'L(-Y* N)
7l'-QX
= (-;-)
7l'-Q
P"Pv O',.A"Y* N) , (C! .5(»
HI that
(4.2.58)
lil('its ures the ratio of the longitudinal to the transverse virtual photon nu cl eO li
I,utal cross section. It is also useful to define
(4.2.59)
with
F2(X, Q2) = 2 (8~:X) (O'T("Y* N) + O'd,* N» . ( 4.2.60)
FE(X, Q2) = (8 ~2
7l'-QX
) O'~CJ* N),. (4 .2 .61)
then
FE(X, Q2) = ~Fl(X, Q2) - tF 2(X, Q2), (4. 2.62)
F 2 (x, Q2) = Fdx, Q2) + ~Fdx, Q2), (4 .2.63)
Fl(x , Q2) = FE(X , Q2) + tFL(X, Q2) , ( 4.2.61)
and
(4.2 .65)
We can now express the virtual photon nucleon total cross section in
terms of the virtual photon parton total cross section as shown in Fig. 4.3c .
For example, the structure function F(x, Q2) can be decomposed as follows
(4.2.66)
Fr;(x, Q 2 ) = 11:z:
dy (0) ) ~ (
-Gp_q(y:FE
Y
Z, Q
2)
and
=
1 :z:
1 -dYG(o)
Y
p-q ( Y) ( -Q2'8
1r~OfZ
' ( '}.* q) ,
) ) O'E (4.2.70)
(4.2.71)
with
A
O'L
(
r * N) = 4z2 ' ( r * q) ,
Q2 PI"PII 0'1"11 (4.2.72)
where the virtual photon quark total cross section is defined in analogy to
(4.2.17),
(4.2.73)
The factor of 2 in (4.2.71) is due to the fact that in (4 .2.23) we defined UL as
the average of the two longitudinal helicities rather than the sum of the two.
In the naive parton model one only considers the Born term r* q -+ q in
Fig. 4.1 when computing the virtual photon parton total cross section. This
Born term gives
(4.2.74)
and
(4.2.75)
provided that one neglects quark masses. This implies from (4.2.70) that
:Fr;(x, Q2) = e; G~':!.q(x), (4.2.76)
and from (4.2.71) that
(4.2.77)
so that
F 2 (x,Q2)= FAx) = e~ G~':!.q(x), ( 4.2.78)
Fl(x, Q2) = Fl(X) = e~ G1~q(x) , (4.2.79)
/
/
gluon, qCJ
gluon, qo
1.'lgure 4.4 Leading order diagrams for gluon production via the "Compton" pro-
'('RB 'Y* + q -+ q + g, where q..., and pq are the 4- momentums of the initial virtu al
l'I. oton and quark, respectively, and p~ and qg are the 4-momentums of the outgoing
quark and gluon, respectively.
ILJld
(4.2 .80)
In the naive parton model .1'1(Z) and .1'2(Z) are equal and measure the quark
distributions as given in (4.1.16) and (4.1.17). The longitudinal structure fun -
tion vanishes provided Q2 is large enough to neglect "higher twist" terms t/l1lt.
hehave like M2 /Q 2 •
dUt(_ "
- s, t) 1- -2 IM(.
= 64 7r8Pem "IE + q - q + 9 )12
dt
2
Z
= 167rQ4 1-(
M "IE* + q -+ q + g)1 2 , (4.3.1)
where I have used (A .3.17) a nd ill thi s case (A .3 .1 I ) beco mes
, ,2 (§ + Q3) 2 Q4
SPcm = 4 - 4 z 2' (4.3.2)
z --~-~
- ..,. (4.3 .3)
2pq .q-y s + Q-
The invariants are given by
s = (q-y + pq)2 = (p~ + qg)2 = _Q2 + 2q-y.pq, (4.3.4)
Pq - q-y )2 = (pq - qg )2 = - 2q-y ·Pq ,
t' = (' .1 (4 .3.5)
it = (qg - q-y)2 = (Pq _ p~)2 = -2q-y .qg, (4.3.6)
with momentum conservation giving
q-y + Pq = P~ + qg, (4.3.7)
so that
s + [ + it + Q2 = 0, (4.3.8)
where
q2..,.,
Q2 -_ _
(4.3.9)
where the 4-momentum of the initial quark, Pq, final quark, p~, gluon, qg and
vi rtual photon, q-y, are shown in Fig. 4.4.
The subscript 1: in (4.3.1) is to signify that (4.2 .19) has been used to sum
the incoming virtual photon polarization states, and the superscript. q labels
the subprocess 7* + q --> q + g . The amplitude squared is given by
IMCli; + q --> q + gW =e2e~g; H
8 {_~ _ ~ _ 2Q2(s + f + Q2)}
S t st ' (4.3.10)
where the factor of 4/3 is the color factor tr(T aT a)/3 as discussed ill Appendix
D and the factor 1/2 is to average over initial state quark spins. The differential
cross section is thus given by
t) dt, (4.3.12)
where
[min = 0, ( 4.3.13)
/
f- + f-
gluon,Qg gluon,QQ
Figure 4.5 Leading order diagrams for photon-gluon "annihilation" via the proccss
r· + 9 -- q + q, where q-y and qg are the 4- momentums of the initial virtual photon
.\IId gluon, respectively, and pq and qq are the 4-momentums of the outgoing quark
IIl1d antiquark, respectively.
( 4.3 .1 4)
As was the case for e+e- annihilations in (2.3.37) this integral is divergent and
we cannot proceed without choosing a regularization scheme. The differential
cross section in (4.3.11) becomes infinite as i - 0 and this vanishin g of i
occurs for the same reasons as shown in (2.3.39) (i.e., w - 0 or 023 - 0) .
One must also correct the naive parton model by including the possi bi lit.y
t.hat a gluon in the initial proton can produce a quark-antiquark pair whi ch
the virtual photon then couples to as shown in Fig. 4.5. The differential crOBS
section for the subprocess ,. + 9 - q + if is given by
diTg 1-
:; (s, i)
dt
= 64 1rSPcm
' 2 1M (,i: + 9 - q + ifW
z2 - 2
= 161rQ4IMbi: + 9 - q + if)1 , (4 .3.1 5)
where in this case
Q2 Q2
z--------
- 2qg .q-y - s + Q2'
(4.3.16)
and
(4.3.17)
(4.3.18)
('1.:\ 1 I)
with (4 .3.8) and C4.3.9) also holding. As ill (~ . :$ . I) th ' 8 11 hsc ri pt. },; is 1.0 Hif.( 11 II Y
that the polarization states of the incoming virtual photon have b ( ' II S llllllll l'd
using (4.2.19). The superscript 9 labels the subprocess 'Y' + [J - ; q + 'I . 'I'h.
amplitude in Fig. 4.5 squared is given by
1Mb; + 9 -+ q + iJW =e2e~g; H
il i 2Q2Cil+i+Q2)}
8{ -;c+-::-+ . , (4 . 3 . ~.W
t u ttt
where the factor of 4/8is the color factor trCT aT a)/8 as discussed in App cnJi x
D and 1/2 is to average over the two spin states of the initial state glu II
(considered as massless). The differential cross section in (4.3.15) is thus
ir~(8) = Jf
rimiD dirg ,
dtE (8, i) dt, (4.3.22)
t nu.x
and is infinite since i min = O. Again the origin of these divergences are th
sam e as in (2 .3.39) and we cannot proceed without choosing some scheme for
reg ularizing the infrared singularities.
3 1 2.::3 - z 2 1 2 2.:: 3 - Z4 }
- 21 - z + z + 1 + f3 (1 _ .::)2 + 2f3 (1 _ z )3 I
(.1.'1.'1)
h. I' IIII' Illb I "MG" refers to the m assive gluon schem e and
f3 = mz/Q2,
111,( I ~Iddined by (4.3 .3) . Terms that give no contribution to th e inlq .(f "
• ' I III the limit f3 --+ 0 have been dropped. The logarit.hmi c l 1'111 ill 1,11('
ltd III (lOllS section comes from the integral
+ 2Q~) ., A
dt = - - 2--
S _Q2/ z 2.::{I-z) (1-z)
3 Q2 Q2
=-- +2- (1.4 .7)
2z(l-z) z'
n lld
-mgQ
2 2j- fJ Z Q 2/(1-Z) dE
.,,- =---
(1- z)
Q~.
.,
(1 .1.8)
2
-Q'/z t z
This parton subprocess must be "embedded" in the experimentally 01>
IlI' rved reaction ,. + N --+ X as shown in Fig. 4.3c. For example,
(1.'1 .0)
(4.4 .14)
(4.4.15)
where 0'0 is given by (2 .1.30). Substituting (4.4.4) into this equation yields
where DIS refers to deep inelastic scattering and where the {3 and {32 terms
in (4.4.4) have been replaced by a eS-function contribution similar to what we
did in (2.5.6) .
It is interesting to compare this with the e+e- result from (2.5.6),
:1 I I :) }
I- - l : -f I h( 1 .1') . (,1,11. 1 )
'2 I - J; '2 '2 ,I
1': '1l1lLtions (4.4.16) a lld ('1.'1 . 17) a rc very simil ar but not qllit , t,I) ' t;;tJlII'. 'I'll('
Ild,c·gral over x of (4.IJ .17) was accomplished using (2.5.7) - (2 .5. 10) with tI, (·
/I ·IlIdt. in (2.5.11) that
2
"
(uMG(real»e+e- = -2a.
U o { log 2 ((3) + 3Iog(!3) - -7r + 5 } .
37r 3
Tlac' integration of (4.4.16) can also be performed using (2.5.7) - (2 .5. 10) nlld
ylc ·ldt; the slightly different result,
2
(o-MG (real»DIS = 2a. { ., . 27r
37r Uo log-({3) + 310g({3) + 3 + 2
}
,
with {3 given by (4.4.5) and Q'2 = _q2 > 0 (i.e., spacelike). The virt.ual glliOIl
,olltributions in Fig. 4.6 are given by (2.7 .1). Namely,
7r 2 • spacc likc
(4.'1.20)
II lid , as was the e+ e- case, the total
2
" ( )" . )
( UMG real + uAlG(vlrtual) DIS = 2a. { 27r'2
37r Uo 3 + 2- '72 - 327r }
'
(4.IJ .2 1)
opposite sign from the e+e- case in (2.7 .6). In this case the perturbati on s riC's
ha.s the form
DIS = Uo ( 1 - -a. + ... ) ,
U tot
7r
co mpared with
(4.IJ . 2;~)
"q ( .
U AIG vlftua
I) + 1
o
1 d' q
UMG,E d
d
Z
Z =
as
--Uo,
7r
(4.4 .24)
we write
1{d"q
1 U AIG E
dz '
a.
+ (o-XI G(virtual)+-;-uo)6(1-=)
}dz=O, (4.4.25)
and defin e "+ [un ctions" jlls t. as we did ill (3 .:2 .1\) and (:I .L.f») . NaJll<'ly,
(4.4.30)
where
Pq_qg(z) =
4(1 + z2)
3 z +
1- (4.4.31)
3 1
-2(I-z)+ +z+l- 3+4'3) 8(1-z) }.
(2if'2 (4.4.32)
As before the "little f" function in (4.4.30) depends on the process and the
regularization scheme chosen. However since (4.4 .28) must hold in any regu-
larization scheme, the integral
where z = x/y and where the superscript. q signifies that this is the contribu-
tion to FE from the subprocess 'Y. + q --4 q + g.
The differential cross section for scattering longitudinal photons via the
subprocess 'Yi + q --4 q + 9 can be calculated using the projection itechnique
/
dtTl ' 1\ _
' (s, l, - 64 ~ , ~-
-d
t 7r.~ J! e»l
1 - (-y. + q
('1Q2z ~ ) (Pq)"(pq),,IM ---->
~
q + 9)1 ,111
( 11 .11 .:17)
(I1 A .:18)
('1.4 .39)
so that
?
F 2 (x, Q-) = F:E(x, Q-)., + '2Fdx,
3 ?)
Q- , ( 4.4.43)
11ad~'DIS(Z) dz =
= - :: + 11 a.f;,DIS (z) dz +~ 11 a s fI,DIS (z) dz = 0, (4.4.46)
is not .
For the initial state gluon subprocess ,. + 9 -> q + ij in Fig. 4.5 we cannot
regul arize by taking q~ = rn~ because then the incoming gluon could actually
decay into a m assless quark-antiquark pair. To regulate this process we take
t he incom ing gluon slightly off-shell and spacelike, q~ = -rn~ . In this case the
differen ti al cross section in (4.3.21) becomes
dO"E
Ag _ 7raa.e q
2 z2 { uA t
A
2Q?~ 'A 2
- A
dt
Q4
- 2 ~
t
+ -:u: - + - tii
A-(t + u + Q )
2rn~ + u + mg)
+ ~(t A 2 2 2
- Q mg
A ( 1
72 1 - -;:-;:-
+ ""2 4) } , (4.4.47)
tu u t t u.
where some terms that vanish as rn~ -> 0 have been dropped. From (4.4 .15)
we arrive at
1 dU~1G) _ as ~. 2 2 , g,DI S
- 2-2 Pg- qij (",) log(Q /tn g) + 2adJl1G,:E(")' (4.4 .51 )
¥
(-
lTo
-d-
z DIS 7r
/
wil,h
I ,1/ ·',a( ~ ) -- l., ( ~ ~ + (1. -
~ ~
,. ) ~ )
~ ,
II lId
III ('va luating the integral of (4.4.47) over i it is conveni ent to usc
,1, =_Qz"
tu -
(;+~),
t 11
/ 1() that
Ilitegrating the first term over i and the second over it gives
(-1.4 .56)
where
(4.4.59)
giving
( 4.4 .(0)
smce
(4.4.61 )
Equation (4.4 .60) contains no di v )"gellt tC' 1"II15 a lld ,;111 C'a:-,i 11' 1)1' i II I,{'gra 1,('<1
over i giving
( 4.4.(2)
which from (4.4.15) gives a contribution to the longitudinal stru ct ure fun ction
of
(4.4.(3)
F2 ( x, Q2) 211 Y
= eq
x
-dy (C(o)
p-.;.q ()
y + c(o) ( ))
p-.;.q Y
w li ore the two- hody "Iill tH ' tip 1( ' ( ' i'lIc tOI' d~ N - ~U~ iH", iIII iIII.I' 1,0 ('L .H.'L ). N :MII!' I
2 d N - 111:1 dN - 1])4 N N
d'l. N- R 2 = (21T)N - l('LE:J) ('L1T)N - l(2E,J) (21T) ft (P3 +]J4 - PI - ]I~ ). (/l.fU )
III t, grating over P4 yields
J N 1
d - P4 8 (1)3
N + P4 - PI - P2) = 8(E3 + E4 - El - E2)' (1 .0.:1)
Now if we let y = cos 013 , where 013 is the scattering angle between p 1trt,icl( 'H
1 and 3 then
L 21T(N-2)/2 N" " N/" "
d N - P3 = r(N/2 _ 1) 1'3 -~ dP3 (1 - y~) ·-~dy .
where
(1 .5.G)
and
Pl'P2 = Vi Pcm, (1 .5.7)
with
P~m = [8 - (1111 + 1112)2][8 - [1111 - 1112)2]/(4.5). (4.5.8)
Combining (4.5.4) and (4.5.1) yields
dB- , ' 1 (P~m)N-3 - 2 (1 _ y2)N/2-2
dy (s, t) = 3211'8 Pcm IMI 2N-41TN/2-2r(N/2 _ 1)' (4.5 .n)
Pcm
'f
= '12 v'ss, (4 .5. 11)
, Q2
t = - 2z (1 - V), (4 .5. 12)
Q2
Z = S + Q2' (4.5.13)
'q 2 Z 1 - z).Q?)f/2
~ I
C1'DR(z, Q ) = 321TQ2 ( 41TZ 2< r(1 + (/2)' (4.5.15)
1:1.1\ ( :1. 11,1'1.(' " 4 I)(" 'P 1""111 I. k S, J\ 1.l, , · ,' i 'l)~
where
(1\ .5. 1G)
with N = 4 + (.
In N = 4 + f dimensions the matrix element squa.red is given by
4( 4z2 + 4yz - 4z + y2 - 2y + 5) 4( 4z 2 - 4z + y2 + 3)
{ ~----~------~--~--~+
(1 - y)(1 - z) (1 - y)(1 - z)
f
+
4z2 + (-4y - 4)z + y2 + 2y +1
f
2} , (4.5.17)
(l-y)(l-z)
where a~ED and a~CD are the N-dimensional couplings in (2.8.11) and
(2.8.26), respectively. The evaluate the integral I in (4.5.16) we use
rCA) r(t B + 1)
1 -1
1 ( 2 A
1 - y) - - dy
yB
l-y . -
_
-
r(A+tB+l)
(B odd)
_ rCA) r(~B + t)
-
r (A.+ Z1B+ 21) (B even) . (4.5 .18)
_q
u DR
(
E Z, Q
2) _ 1671'a~EDa.e~z
- 2
(Q 2(1_ z»)'/2 r(1 +~)
., ---':-.-~
. 3Q z471'mb [(1 + f)
I + z2 2 4z2 - 8z + 5 z2 + 2z - 1 }
{- - - + 2{I-z)
l- z f
+ 2(I-z)
f+ '" ,
(4.5.19)
where the mass mD comes from using (2.8.26) and a. is now dimensionless.
Using the N-dimensional analogue of (4.4.15),
1 du Q2 "q ()
- QED u 4.5 .20
Uo d z 871' 2a N e~z(1 +~)
where the factor of (1+~) comes from the Born term matrix element evaluated
in N spacetime dimensions
' 1'lIii'< nUl b O IlI( II\.I· .. d wd.11 !.I1f· (, I (. r('s ult. ill (:1.'2 . 11) ),
wh ich again is similar but not the same. If we expand the r fUll ·t.ioliR ill
('1.5.22) according to
r(I+-21 f) 1 1 2 ,,')
f(1 + f) =1+ 2/E f + 16(2/E - 7I"~) f~ +" ', ('1 .:; .'2'1)
~ dU'1m,E)
( ITo dz
= 2a.
31l'
{I1+- z2 (Q 2 z
[lOg (1 -"Z))
z471"ntb
+ /E]
DIS
1 +-
+- z2 2 3 -1- - z + 3
-- - ( t:' ,)~)
} ,1.;).'2
+ ...
l-z f 21-z
where terms that vanish as f -+ 0 have been dropped .
Integrating (4.5.22) over z yields
,
(ITDR(real))DIs - 371"
_2a. (Q )</2r(I-~)r2(1+
ITo 471" b
2
r(1 + f)
~)
ln
{ !f2 _ ~f + 13}
2
. (1 .5.:,W)
The virtual gluon contributions in Fig. 4.6 are given by (2.9.14) . Name ly,
(uDR(virtual))DIS
2
= 3as71" ITo
(Q
--"
)</ 2 2
r(I-~) 1'2(1 +~)
I'() 2
41n11b 1+ f
8 -6 - 8 }
{ --+
f2 f spacelike '
(4.5.27)
so that
This is the same result arrived at for the massive gluon scheme in (4.4.21). If
we define "+ functions" as in (4.4.30) we get from (4 .5.25) and (4.5.28)
1 d,q
ITDR,E ) _ a. 2 '2 q,DIS
-ITo ( dz - -2
1l' Pq .... qg(z) 10g(Q linD) + a.fDR '" (z),,~
( 4.5.29)
+
where Pq-+ qg( z) is given I y (11 -" .:11) ami
o'.
q DIS() 20'. {(I + z ")
/ DR E Z = - 2 -
(l og(l
1
- Z)) J + =~ I ()
- -1- - og Z
I 'iT - z + - z
2
- -3 1 -z+3- (7T-+3 ) }
8(1- z)
2(1-z)+ 3
+ -Pq-+qg(z)
o'.
27T
{2- + ,<
{
-log(47T) }
, ( 4.5.30)
'I'll con tribution to til 10 ll gi tudillalstructure function in ('1.5.63) is 1.1 1(' SO llie '
lI ill ce no divergences arc encountered, therefore (4.4.65) beconl 's
(IJ.!).:I
-r (
J""2 x, Q2) -e
_ q 211
x
dy
Y
(C(O)
p-+q (y) p-+ij (.»)
+ c(O) Y
where z = x/yo
The quark distribution are defined from the F3 structure function as III
(4.1.16). Namely,
n,
F 2(x, Q2) == I>~; (C~~q;(x, Q2) + C~:!.ij; (x, Q2») , (4.6.1)
i=1
where the superscript (2) refers to the F2 structure function and where T/.J is
the number of quark flavors. From (4.4.67) or (4.5.38) we see that
Os p
+ 27r q-+qg () I (Q2/ m 2 ) + Os fq.DIS(~)}
Z og 2 ~.
(4.6.2)
where z = x/y and where the mass m is the dimensional regularization mass
mD or the gluon mass mg depending on the scheme. The "little f" functions
are regularization scheme dependent and are given by (4.4.45) and (4.4.66)
in the massive glue scheme and by (4.5.33) and (4.5.37) in the dimensional
regularization scheme. If we define the convolution notations as in (3 .3.11)
then (4.6.2) becomes
(2) ( x,
G p-+q Q2) p-+q *
= G(O) (1 + 211'
a. Pq-+qg Iog (Q2/ 111 2) + a., jq,DIS)
2
As explained in (3.3.17) the log(m 2 ) divergences are absorbed into the un-
known distributions G(O) leaving
(2) (
G p-+q x, Q2) -_ G(O)
p_q * (a.
1 + 211' Pq- qg
.(
log Q
2/ A2) + ad2q,DIS)
+ riC
Gp_q0)
* (a.
211' Pg-qij Iog'(Q2/A2) + a. j9,DIS)
2 , (4.6.4)
which is analogous to (3.4.50) for the fragmentation functions . All leading log
terms of the form lase Q2) loge Q2»)n are then summed in precisely the same
way we did in Chapter 36 . The result for the "non-singlet" distribution
GNS(X, Q2) = Gp_q(X, Q2) _ Gp_if(X, Q2), (4.6.5)
is the same as (3.4.44). Namely,
GNS(X, Q2) = exp(h:Pq_ qg *) GNS(t c ), (4.6.6)
where", is as defined in (3.4.35),
2
'" = /3olog{a.(t c )/a.(t)}, (4.6.7)
(4.6.8)
or
(4.6.9)
smce
d", a.(Q2)
(4.6.10)
dr 211'
/
1\ Q~ D " "dollt • 't, ,\, LII. Fllnd Ollfl
~
d
""9(.'Qe, g~ 2"
f N
~(t)
- =L +
dK N j_' ~\Y.
Figure 4.7 illustrates that the leading order rate of change of the quark distribu-
tion, GN-+q(x,Q2), with respect to " is generated by two terms. Quarks can loose
momentum by radiating a gluon and additiona.1 quarks are generated when a gluon
splits into a quark-antiquark pair. Similarly, the leading order rate of change of th
gluon distribution, GN-+g(x,Q2), with respect to" is generated by two terms. Glu-
ons can loose momentum by radiating a gluon and additional gluons are produced
by Bremsstrahlung off of quarks and antiquarks.
with
(4.6.11)
Here I have dropped the superscript (2) since the "little f' functions in (4 .6.4)
do not contribute to the leading order Q2 evolution of the distribution fun c-
tions.
As illustrated in Fig. 4.7 it is easy to see that the evolution of the quark
and gluon distributions are governed by7
(4.6.12)
( 4.6.13)
where Pg -+ qq ( z) is given in (4.4.52) and
4 1 + (1 - z)2
Pq-+gq(z) = Pq-+qg (1- z) = 3 z ' (4.6.14)
140 ·/l1\1'1.(1r "
e-
e- e-
lo} .~ + .-~
p p
9 q
p
e-
e-
'j'
e-
lb} .~ q Y + 9 q
~
p 9 p
Figure 4.8 Illustrates how in deep inelastic electron proton scattering the leading
log contributions from the subprocess ;- +q --+ q+g and ;. +g --+ q+il are summed
to form the Q2 dependent quark distribution, Gp_q{x, Q2). Similarly, the leading
log contributions from the subprocess ;. +il -+ il+g and ;. +g --+ il+q are summed
to form the Q2 dependent antiquark distribution, Gp_q(x, Q2).
and
z
Pg_gg(z) = 6 [ (1- z)+
l-z
+ -z- + z(1 - z) +
(ll-i n ,)
12
1
6(1 - z) .
(4.6.15)
Fig. 4.8 shows how the leading logarithm contributions from the subprocesses
,. + q - q + 9 and ,. + 9 _ q + ij sum to form Gp_q(x, Q2) and similarly
the leading logarithm contributions from ,. + ij - ij + 9 and ,. + 9 - ij + q
sum to form Gp_lf(x, Q2).
The coupled equations in (4.6.12) and (4.6.13) are solved in a manner sim-
ilar to the fragmentation function evolution equations in (3.4.52) and (3.4.53).
One defines a "singlet" distribution,
nJ
which together with the gluon distribution satisfies a matrix equation similar
to (3.4.59),
( 4.6.17)
lit I Iif' 1'I I I'Iliai fl 11I1.io ll ~( I V" II hy
('I. (I . I H)
11 Pq-+qg( Z) dz = 0, (4.G.22)
(4.G.26)
so that the total momentum of all the partol1s with the proton is indep endent
of Q2. If we normalize
(4 .G.27)
u+d Quarks Gluon
1.0
N-o
-
)(
Q.
(,!)
)(
0.1
0 .01 L.......J..--I._..L.....l.L..L...L_..L........J...a....L....J...L.-.....
0.0 0.2 0.4 06 08 00 02 04 06 08 1.0
x x
Figure 4.9 Shows the Q2 dependence ofthe quark, xGp_u+d(X, Q2), and the gluon
distribution, xGp_,(x, Q2), within the proton.
as in (4.1.9), then it will remain equal to 1 at all other Q2 values. Figure 4.9
shows the Q2 dependence of the quark and gluon distributions from a recent
fit by Duke and Owens 8 with A = 0.2 GeV.
,1.7 Q2 Evoilltioll MOluent Method
I':qu a ti on (4.6.8) and (1 .6. 17) can be written in terms of the momcnts or til('
pllr t,on distribution as follows:
(1 .7. 1)
II,lI d
(tJ .7.'2 )
where
(4.7.'1)
a nd
An -
_ (A;:S 2n'A~ij)
, (1 .7.5)
A~q A~g
where M!:s, M~ and Al~ are the moments of the non-singlet, sing\ Land
gluon distributions, respectively. The "anomalous dimensions" An ar g'VCII
by
An
NS
= Jor1 z n-l Pq ..... qg(z) dz = 34 [ -;-12 + n (n 1+ 1) - 22: -
j='2 J
1
=- , .
(4 .7.(» n 1
1 2+ n + n)'2 ,
gq _
An -
1 o
1
Z
n-1
Pq--+gq(z) dz - 3
_ 4
2 + n + n '2
n
(.,
n~ - 1
(tI .7.7)
and
qij _
An -
1 o Z
n-1 .,. _ 1
Pg ..... qij(z) d~ - 2 nn+
( 1)( n+ 2)'
.
(4 .7.8)
Agg -
n -
r
Jo
1
zn-1p
g..... gg
(.,.)
~
d~ -3
'"' -
[_! +6
2
n(n - 1)
+ -,--_...,.2....,--_:-:-
(71 + 1)(n + 2)
-22:-=--9
n 1 1 71 ,
j='2 J
1 . (4.7.9)
Equations (4.7.1) and (4.7.4) give three independent Q'2 evolution equations
HIli
( A~o 0)
A;;-
RA R-1
n·
(4.7.13)
(4.7.14)
which gives
(4.7.20)
(4.7.21)
/
I 'l r.
II"JlII ('1 .7. 11 ) itlld (" .7 I ·.~) WI ' tH' I' 1.11 1\1,
6
[19 M2(Q') - "'; MJ(Cl)] =
[log(Q~/A2)/log(Q2/A2 )] dt [196M~(Q~) _ n; Mi (Q6)] I (" .7.12)
wllI 'l'('
I llId
[Mi(Q2) - M!(Q2)] =
d NS
[log(Q~/A2)/log(Q2/A2)] 2 [Mi(Q6) - M!(Q~)], (" .7.25 )
wltcre
dl{S = _2.A;'s = 2. 16 = 32 . (4 .7.26)
Po - Po 9 3(33 - 211J)
Using the notation defined in (4.1.6) - (4.1.11) for the total fraction of mo-
IlI cntum carried by the various quarks and solving equations (4 .7.22), (4 .7.24)
an d (4.7.25) yields
2 16 d+
. C(Q ) = (16 + 311J) - B ( 2,
(4 .7.20)
where
A = Q(Q~) - Q(Q~), (4 .7.30)
and
(= log(Q6/ A2 )jlog(Q 2 /A2). (4 .7.32)
Equation (4.7.24) is equivalent to (4 .6.26) and I have taken
Q(Q2) + Q(Q2) + C(Q2) = Q(Q6) + Q(Q6) + C(Q6) = 1. (4.7.33)
Since dl{s and dr are positive (provided 211J < 33) (4 .7.27), (4 .7.28), and
Momentum Carried by Quarks and Gluons
I.O.----r-.......----.-----..:---r-.....---.-"""T'""-"
E
-;:,
c
Q)
E
o
~
c
o
"0
~
-
CL
o
c
-o
u
o
~
lJ...
0 .0 L-..1.--&-.:::::J::::::::.i--.L.-....&...---,--,-----t...::::::.J
0.0 0.5 1.0
~ = log(Q!/A~/log ((fIA2)
Figure 4.10 Shows the Q2 dependence of the total fraction of the proton 'll
momentum carried by quarks, Q(Q2), antiqua.rks, Q(Q2), and gluons, G(Q2), if
Q(Q~) = 1 and Q(Q~) = G(Q~) = O. The momentum fractions are plotted verSUA
(= log(Q~/A2)/log(Q2/A2) which is equal to 1 when Q = Qo and goes to 0 as
Q -+ 00. The wavy line marks the boundary between the perturbative (left) and the
=
non-perturbative (right) regions (for A 200 MeV) . The dotted line corresponds to
the point where Q = G = 0.47 and Q = 0.06.
Q(Q2) -+
Ql_oo
~ n,
2 (16 + 3nJ)
- + ~ (~)
n,=4 2 7 (4.7.34)
(1\ .7.I(R)
III'
Thus,
)/S log(M,;vS(Q2)) = G(n) -log(log(Q2 / A2)), (4.7.55)
"
where
NS _ 2 A NS
d
n =-/30"' (4.7.56)
and
C(n) = )/S log(M,;vs(Q6)) + log(log(Q6/ A2 )), (4.7.57)
"
is independent of Q2. This means that
(4.7.59)
At this order of perturbation theory this implies that a plot of log (.A1,;vs (Q2»
versus log(Mf'S(Q2» is a straight line with a slope given by a ratio of the
anomalous dimensions
dnNS 4NS
• n
d NS - ANS' (4.7.60)
l l
and is independent of the perturbative parameter, A, and the number of quark
flavors, n J.
We can approximate the integral equations in (4.6.6) and (4.6.18) by using the
convolution method presented in Chapter 3 for the fragmentation functions.
In analogy to (3.5.13) the non-singlet structure function is given by
where
Illid
PLl(Y) = -4 -+-[1
y2 + -2-
3 1- Y log(y)
+ (-32 - 2,E) 8(1 - y) ] I ('IJU)
where K is defined in (4.6.7). The individual quark and gluon distributi oli AlI.n '
given by
('I .H.'I)
(4 .8.5)
where
- 2
Cp_i(a, x, Q ) = }'"
rt ydy Cp_i(xjy, Qo) (_log(y»a,,-l
real{)
2
, (11.8 .6)
and
PLlq(Z) = PLl(Z), (4 .8.7)
with PLl(z) given in (4.8.3) and
PLlg(Z) =6 [ -Z1
- Z
+ -og1
1( ) + 1 - Z + z(1 -
z z
z)
Furthermore a q and a g are given in (3 .5.17). The quantity G(XI Q2) is CO III -
(4.8 .0)
in (4.8.4) and ('1.8 .5) a ll wall ensy (/C't(' l"Iliin n.ti oll of til(' In.JW~ J; beli avior o f
G(x, Q2) since
(4 .8. 10)
If at Q5 we have
(4.8.11)
then
G(a,x,Q2)=A r(p+1) (-log(x»P+al< ~ C(1_x)PHa(Q2»,
r(P + 1 + a,,) x->1
( 4.8.12)
where
( 4.8.13)
For quarks
(4.8.14)
(4.8.15)
As in the fragmentation case, the gluon distribution becomes steeper than the
quark distributions and the difference is governed by the ratio 9
(4.8.16)
(4.9.2)
iu a,IIY SC h CII H' , Wi ' IIIIV"
2) (X Q2)] clx -
l('(~) (X Q2 ) -
'1 1 - Q '
dp-+ q , -
Th is m eans that the net number of quarks in the proton (4.6.19) is not a frcct d
by t he higher order corrections that produced the "little functi ons. This is r
a nice feature and it is one of the reasons that we use the :F2(x, Q2) o b. 'c rv a hl( ~
to define the quark distributions.
However, we could define "quark distributions" from t.he :FI(x, Q2) Rt.rII r
t ure. Namely,
n,
:Fl(X, Q2) == 2>~i (G~2qi (X, Q2) + G~~ifi(X , Q2») ,
i =l
where the superscript (1) refers to the :Fl structure function. In this Ca!>(
(4. 9.1) becomes
which means that G~2q(x, Q2) is also scheme dependent . However, we can-
not actually calculate G~2q or G~:!.q. If we are given G1:!.q or G~2q a t oll e
value of Q2, say Q5, we can calculate it at higher values of Q2 from the evolu -
tion equations which, at leading order, do not involve the "little fun ctio ns. r
We now define the observable :F2( X, Q2) in (4.6.1) to be our "reference di stri-
butions." That is, we define quark distributions Gp_q(x, Q2) according to
(4 .9.7)
Given Gp-+q(x, Q5) we can calculate Gp-+q(x, Q2) from the Q2 evolution equa-
tions. With this definition equation (4 .9.5) becomes
(4.9.11)
The f1!1 functions are related to the longitudinal function h given by (4.4.41)
and (4.4.64) . The f1!1 functions do not depend on the regularization scheme
and tell us how much the G~:lq(x, Q2) distributions differ from the "refer-
ence distributions" Gp->q(x, Q2). In the naive parton model equation (4.2.77)
holds and there is no difference between G~:lq and Gp->q. In QeD they dif-
fer at order a. and (4.9.9) tells us how to calculate G~:lq(x,Q2) in terms of
Gp->q(x, Q2).
The longitudinal structure function FL(X, Q'2) is given by
:h(x, Q2) =a.(Q2) 11 ~ {t e~i (Gp->q; (y, Q2) + Gp->q;(Y, Q2)) !1,DIS (z)
The sum runs from 1 to 2n, since it includes both quarks and antiquarks.
Equation (4.9.13) allows us to calculate the longitudinal structure function
FL{x, Q2) in terms of the structure function F2 (x, Q2) and the gluon distri-
bution Gp->g(x, Q2).
The integral of the observable RDIS(x,Q'2) defined in (4.2.58),
RD1S(x, Q2) = FL{x , Q2)/(FL{x, Q'2) + F2(X , Q2»)
= RfIS(x, Q2)/(1 + RfIS(x, Q2)), (4.9.15)
/ I r.:1
It.11d
('I .lL I (»)
Itre easy to estilllat i ll QCD. 111 part ic ular,
(H). 17)
where
Fi(Q:?) = fa1 Fi(x , Q2) dx, ('I .\). I H)
a.l 1
::fl'DIS(z)dz = ;; ~, ('I .fI . I !))
r 1
a. io zh
g,DIS _ a. 1
(z)dz - 211" 6' ('I .H.20)
The cross section in (4.2.13) for electron nucleon scattering can be expressed
as
(4.10.1)
where
y == viE, (4.10.2)
is the inelasticity parameter satisfying
o~ y ~ 1, (4.10.3)
and where
:F1(x, Q2) = 2Fl(X, Q2), (4.10.4)
:F2(X, Q2) = F2(X, Q'2)lx. (4.10.5)
In addition, a term proportional to M I E has been neglected in (4.10.1).
The corresponding cross section for neutrino and antineutrino nucleon
scattering is given by
d 2 (T(II,il)N G 2 ME [ .,
dxdy = F 7r F2(x, Q~)(l - y) + F1(x, Q2)xy2
(4.10.8)
j
where the sums are over the appropriate quark and anti quark flavors shown
in (4 .1.18) - (4.1.25). Fermions contribute to :F3(X, Q2) with a positive sign
while antifermions contribute with a negative sign. The distributions satisfy
/ 11. 10 Nnlll,dllli NII~ 1"011 ,' , 11,1,1,111 II~
(1 + CI'.J;/,J)/ ~' )
G~ ) 1/(J: ,q'J ) - :N _,/(Q'.!) *
+ GN_g(Q'2) * (a.g· DfS ) , (11 . 10 . 11 )
itn d similarly
where again convolution notation has been used. The "little f" fun cl.i OIlH /1' 1'
regularization scheme dependent as usual and are calculated from th ' virt.1I111
W-boson subprocesses W* + q --+ q + 9 and W* + 9 --+ q + ij. Th e "liWe J"
functions in (4.10.11) and (4.10.12) are identical to those computed [rolll tI",
virtual photon subprocess ,. + q --+ q + 9 and ,. + 9 --+ q + ij. The fun ctioll n
is new and in the massive gluon scheme it is given by
2
q,DIS( )_20'0{(1 2) (log(l-Z)) 1+z (_2l ('>'))
0'. f MG 3 Z - 3 +z
'11'
1
-z ++ 1
-z og -
_ ~
2(1-z)+
1 +3z _ (211'2 +~) 6(1- z)} .
3 4
(4 .10 .14)
where
O'.tl.if,DIS(Z) = 0'. (if,DIS(z) _ f~,DIS(z)) = -O'.JI,DIS(z)
= -0'. 4.
- -32z, ( 4.10.18 )
211'
O'.tl.jf,DIS (z) = 0'. (gf,DIS (z) - n,DIS (z)) = _ ex.fl,DIS (z)
These "little tlf" functions are unique anu ind ependent of th e regul arizati on
scheme.
For neutrino nucleon scattering (4.10.9) becomes
:F~N (x, Q2) =2GN-+d(X, Q2) + 2GN-+.(X, Q2)
+ 2GN-+u(X, Q2) + 2GN-+c(X , Q2), (4 .10.21)
and (4.10.10) becomes
:F~N (x, Q2) =2GN-+d(Q2) * (1 + o:.tlfj,DIS)
+ 2GN-+s(Q2) * (1 + O:stl!1,DIS)
- 2GN-+n(Q2) * (1 + o:.tl!1 ,DIS)
- 2GN-+c(Q2) * (1 + O:stlf3,DIS), (4.10.22)
where I am using convolution notation. The structure functions (4.1.26) and
( 4.1.27),
:FQ(X, Q2) == ~(:F2(X, Q2) + :F3(X, Q2», (4.10.23)
:FA(x, Q2) == ~(:F2(X, Q2) _ :F3(X, Q2», (4.10.24)
which in the naive parton model project out the quark and antiquark distri-
butions now give
:Fr:t (x, Q2) =GN--+d(Q2) * (2 + o:.tlf~,DIS)
+ GN-+.(Q2) * (2 + o.tl!1,DIS)
+ GN--+u(Q2) * (-o:.tlfJ,DIS)
+ GN--+c(Q2) * (-O:stlf~,DIS), (4.10.25)
or
:FvN(X, Q2) = 2GN-+d(X, Q2) + 2GN-+.(X, Q2)
+
1-
x
1 dy [GN-+d (y,2
Y
Q ) + GN-+. ( Q )JCl'stlf3'q DIS (z)
y,2
+ 11
x
dy [GN-+ u(y, Q2)
y
+ GN-+c(V, Q2)JO:stl!1,DIS(z), (4.10.26)
where z = x/Yo Similarly,
:F~N(x, Q2) =GN-+u(Q2) * (2 + O:stljj,DIS)
+ GN-+c(Q2) * (2 + O:stl!1,DIS)
+ G N d(Q2) * (_o:.tl!1,DIS)
-+
or
-11 x
dy [GN_d(y, Q2)
y
+ GN_.(y, Q2)la.~J1,D1S (z), (" . 10 .~8)
where z = x/Yo At order a., FQ(x, Q2) receives a contribution fr III 11,11 I. i
quarks and FA(X, Q2) receives a contribution from quarks. If w 1I1111tiply
(1.10.25) and (4.10.27) by x and integrate we see that
{1 q,DIS( _ 10 a. _ (
a. Jo z~f3 z)dz - -9 271" - -0.177 a.. 4.10 ,;$:3)
we have
F'LN (x, Q'2) =;:2GN_d(Q2) * (a,flJ1,D[ S)
+ 2G N_.(Q2) * (a.flli ,D Is)
+ 2GN-+u(Q2) * (a,flli,DIs )
+ 2GN-+c(Q2) * (a,flll'DIs)
+ 8GN-+II(Q2) * (a,flll'DI5), (4.10.36)
where again convolution notation is being used. Comparing with (4 .10 .21)
shows that
FIN (x, Q2) =F!iN (Q2) * (a.ll,Dls)
+ 8GN_g(Q2) * (a.II'DI5). (4.10.37)
If we substitute (4.4.41) and (4.4.64) for II and J1, respectively, then we
arrive at
where FL{x, Q2) = xFL{x, Q2) and F2(x, Q2) = XF2(X, Q2) . This result is the
same as the electron scattering case in (4.9.13) except now
av = 8. (4.10 .39)
Integrating over x as in (4.9.18) gives
II lId 1,11 g ill II 1., ' 1'111 11 1\ 11 ' III IW Id(' lIti ('al. Jf w 1I ('g l( 'r l, 1.11( ' S l. l'I lll }J,<' n. lld (' 11 11.1' 111
(, 0 111, ' nt withill 1,1,,· 1I11r1 <'O Il 1.11('11 1.0 ord er 0'., we h av< ~
Fe: (Q2) -l
- 18 ' (4 . 10 .1111)
F~N(QZ)
H.lld
R~N (Q2) = R~N (Q2), (1 .10. /15)
where Rz is defined in (4 .9.17).
Problems
4.1. Starting with the general form of the hadronic tensor given in (4 .2.6 ),
is given by
and substituting into (4.2.16) one projects out the longitudinal cross section
4.3. Using (4.2.3) for tlte lepto llic ten:sor I.J IIII iLlld (4.2.<l) for loll<" It a,dl"nllic 1.(·lIsor
Willi compu t e Lilli Willi a nd verify (1.2.J 3).
4.4. Show that for massless partons that the differential cross sections for the
"Compton" subprocess IE
+ q -. q + g in Fig. 4.1 a nd the "annihilation"
subprocess + g -. q + if in Fig. 4.5 are given by
IE
dcTt ( A A) = 7rQQ.e~z2 16 {52 + P - 2Q2 ,u }
di s, t Q4 3 _ si .
and
tmin
A
7rQQ. e~ z 16 {1+Z2
_ _ _..2.- - - - log ((1-z)Q2)
Q2 3 1- z z'2m~
3 1
-21 _ z + Z + 1 + 4"5 6(1 - z)
}
,
where some terms that vanish in the limit. mg -+ 0 have been dropped and
z=Q 2/(5+Q'2).
4.7. Integrate do-1rG ,Eldz given in (4.4.16) over z,
1/(1+.13) do-9
o-(real) =
1o
d!; d z,
z
/ 1(11
Add the virtual gluon cross section given in (4.4.20) and show t.ll a l,
DIS
O"tot = 0"0
( as
1 - -:;- + . .. ) .
,*
4.8. Prove that in the limit of massless partons that the leading order Jon,.,;itlldi II lI,)
cross section it L ( q -+ q) vanishes.
q,DIS( ) _ 2a. 2
a. j L Z - 311" z.
4.1 0. Show that the differential cross sections for the "annihilation" subproccl;"
,f + 9 -+ q + if in Fig. 4.5 is given by
_dO"_~~_b(s,j\ = _1I"_a_a_se-:-,~,-z_2 2
dt 'J Q4
{f2 + it 2 - ~(Q2 + m;)stil
tmin = -lngz, ?
2
t max = -Q /z.
~
iT~G d z , Q2)
,
=; -m'z
-Q'/z
9
d'g
_0"_::;
dt
di
and
9
a L(z ,Q )
2
= 1faa.e~z2
Q4
4z21°
Q2 0
".
2s dl
-Q-/z
show that
1 (p" )1+< 1
• ~e~n~l~_~~~~~_~
a =327rs Pern (47r)</2r(1+1)
1-1
1 dy (1- y2y/2IMf,
where Pern and P~rn are the initial and final state center-of-mass momentum,
respectively, and N =
4 + f, y =
COS(013), and
1 da"
&=
1 -1
d dy.
Y
/
~ . 11 . Sho w th " t fOl III I1HH I" "H 1" \lllll l/< ill N spa cc tilll (, dilll \' II Hio Il H tI,(, IlI ft l.li x .. I,·
m c nts flqll :U·(·t! fill 1,1", " ( :') llIpI.OIl " ~ ubpro ccss ill Fig. ~ . ~ a. lId t it e " a.lIl1ihdll
lion" subpro '"H" ill I,' i).;. 11. :; a.re g ivcn by
and
11(1-
-1
y2y/2_1_ dy
1- Y
::::11 (1-
_1
y2y/2_1_., dy:::: ~
1 - y~ {
fy ,
11(1-
-1
y2)</2L dy
l-y
::::11 (1- y2y/2~
-1 1-y~
dy:::: 2
{(1+{)
fy ,
and that
1 -1
1 (1 - y2)</2_y_ dy::::
1- Y
2
{( 1 + ()
f
y,
(1 - y2)'/2 ~ dy::::O,
where
1 -1
1
1- Y
2<f2(1 + *)
fy:::: f(1 + ()- .
4.16. Using (E.1.12) show that
where
fz :::: r (1 - ~) r (1 + ~) .
4.17. Show that (or massless partons in N spacetime dim e nsio n" til e IlIa,lrix ·1(:-
ments squared for the "Compton" subprocess in Fig. 1.4 is give n by
( ~ diTbR)
0'0 dz DIS
=200. (1 - Z)Q2)) </2 f(1 + ~)
311' z411'mb f(1 + £)
1+Z2 ~_~_1__ Z+3+ ... }.
{ I-z £ 21-z
Integrate this over z and show that
q
(dOO'dDR 200 (Q 2 )</2
(o-DR(real))DIS =
1o
1 )
z DIS
dz = 3 • 0'0
11'
-4 2
11'm D
f(1- ~f2(1 + ~)
f(1+£)
{!£2 _~ + 132 + ... }.
t
Combine this with the virtual corrections in (1.5.27) and verify that
w li C'r •
and
3
-- 1 -z+3- (-+3
1 1 "8(1-
2 )z}
)
2(1-z)+ 3
Verify that
4.19. Show that for massless partons in N spacetime dimensions the ma trix ('I, .
ments squared for the "annihilation" subprocess in Fig. 4.5 is given by
(-
1 -dUbR)
- a.
= 2-Pq_ gq () . 2/ nlD)
Z log(Q
:l
+ 20"' / .g,DJS(
DJ1 E -~ ) ,
0"0 dz DIS 271" '
where
_ 1('2
Pq_ gq ( Z ) -"2 Z + (1. - Z
)2) ,
and
11 Pq_qg(z) dz =0,
11 z(Pq_qg(z) + Pq_gq(z)) dz =0 ,
dNq = 0,
dr
and
d -
dr(Q + Q + G) = 0,
where r = log( Q2 / A2) and
Nq = 11 (G p_ q(x,Q'2)-G p_ q(x,Q2))dx,
is the net number of quarks of flavor qi within the proton and Q, Q, and G
are the total proton momentum fractions carried by quarks, antiquarks and
gluons, respectively.
,I II I N"lItrilio NtH 1"011 .'t'IIUm It 1'. 1/17
A NS-lI
n -
Z1l I I' 'l -
o
( 7)d~_4
q9.... '" - '3
_! +
[ 2
1
n(n + 1)
_ 2
j; ~l
n
j ,
(~) d ~ -_ 4 2' + n + 71
2
A ngq = io[1 Zn-1pq-+gq.. 3" 71(71 2 - 1) ,
3 [-! 6
+ 2
71(71-1)
+ 2
(71+1)(71+2)
- 2
i=2 j
t !- !n
9 f
1
4.22 . Show that regardless of the total fraction of the protons momentum carri d
by quarks, Q( Q6), antiquarks, Q( Q5), and gluons, G( Q6), at the reference
momentum, Qo, that the following is true asymptotically
where 71 J is the number of quark flavors and the momentum fraction s arc
normalized so that
4.23. Using the leading order Q2 evolution equations plot log(M:'S(Q2)) versll~
log(MfS (Q2» for 71 = 2,3,4 and f' = 2,3,4. Also plot [AI,;,S (Q2)) - J /d;,' 8
versus log(Q2) for n = 2,3,4 with A = 200rvleV and Qo = 2GeV.
4.24. Verify that
d. _ [1 . jq,DISC_) d- - 0
io[1 as jq,DIS(_)
2,MG - Z - io aS 2,DR ~ '" - ,
where ftf/i and N:g~s are given by UAA5) and (4.5.33), respectively.
Show that this ins ures th a t
where
G~~q(X, Q2) =Gp_ q(Q2) * (1 + Q.n,DIS)
+ Gp _ g (Q2) * (Q.J~ ,DJS) ,
and
4.25. Show that to order Q. the longitudinal structure function is given in terms
of F2(x,Q2) and the gluon distribution G p_ g (X,Q2) by
a = a v = 8,
for deep inelastic neutrino and antineutrino scattering. Verify that
-FL (2) Q. [8-
Q = 27r 9F2(Q~)
? 1
+ (jaG(Q 2)1,
where
F;(Q2) = 11 Fi(X, Q 2)dx,
Further Reading
G. Altarelli, "Partons in Quantum Chromodynamics," Physics Reports
81, 1 (1982).
A . H. Mueller, Physics Reports 73, 237 (1981).
W .B. Atwood, J.D. Bjorken, S.J. Brodsky, and R. Stroynowuri, "Lectures
on Lepton Nucleon Scattering and Quantum Chrol11odynamics," ed. B. A.
/
.JIIJr~ , alld
D. 11.11('11 1'1" , IIl1ldllIlI HI'1" H()stull Publishcl'H, ID 8~ .
A.J . Buras , n~'v . l\1ud . l'/ly s. ti2, j99 (1980) .
Chris Quigg, GilligI' 'I'lwo/,;c::i of the Strong, Weak, al/d Elcc/,/,ol/wglld.;(
Illt cractions, Frontiers ill Physics, The Benjamin-Cummings Publishing CO lli
!lallY, Inc., 1983.
References
I. See, for example, Chris Quigg, Gauge Theories of tJle Strol/g, We 'lI/c ,
and Electromagnetic Interactions, Frontiers in Physics, Th e Ikllj1l1llill
Cummings Publishing Company, Inc., 1983.
2. R.P. Feynman, Photon-Hadron Interactions, Benjamin, Reading , M A,
1972.
3. J. Dress, in Proceedings of the 1981 International Symposium on L 'p-
tons and Photon Interactions at High Energies, edited by W . Pf ii ,
Physikaliches Institut Universitiit Bonn, Bonn, 1981.
4. R.D. Field and R.P. Feynman, Phys. Rev. D15, 2590 (1977).
5. G. Altarelli, R.K. Ellis, and G. Martinelli, Nucl. Phys. B143, 521 (1978) .
Erratum B146, 544 (1978).
6. A.B. Carter and C.H. Llewellyn-Smith, Nucl. PJ1YS. B162, 397 (1980) .
7. G. Altarelli and G. Parisi, Nucl. Phys. B126, 278 (1977).
8. D. Duke and J. Owens, PllYS. Rev. D30, 49 (1984).
9. D.J. Gross, Phys. Rev. Lett. 32,1071 (1974).
10. G. Altarelli and G. Martinelli, Phys. Lett. 76B, 89 (1978).
'II J\ PTI ~ H [)
170
/
III 1!)70 Dr II l wei VII II I ,J" V"I"p h l it model ro r til' prodll cti o n o f' III !lHHiw
I"ptoll pairs ill IlildrO Il 11 1... 11'0 11 co llisiolls. In the trlodel a qu a rk 1'1'0 111 O l" ~ of 1,11('
ill cident hadrons a ll IIi II i1 al, '/3 with an antiquark in the ot lterill cid 'III, lI adru ll
producing a virtual photon which in turn "decays" into a massive lepl,o ll pn.ir .
III the previous chapter a spacelike virtual photon served as a prob ' of 1.1 1<'
Htructure of the proton. Here we use as a probe the timelike virtual pll o toll
tllat is produced and subsequently "decays" into a J-l+ J-l- pair. J-Jere til e ro l,'
of the 4-momentum transfer squared, Q2, of the scattered electroliR ill <I (.t' p
inelastic scattering is played by the mass squared of the muon pni 1' , /1-/ ~ ,
Large-mass muon pair production in hadron- hadron collision provid eH 1IIIIIIy
interesting tests of perturbative QCD. In particular, QCD predidfl H i ~l\ ld, '
deviations from the naive parton model predictions both in the rate 1'( r 11111 0 11
pair production and in their transverse momentum spectrum.
In the naive parton model large mass muon pairs are produced in proton-
proton collisions via the subprocess q + ij -+ /'* -+ J-l+ + J-l- . The experim ent a l
cross section is expressed in terms of the parton subprocess as follows
(5 .1.1)
where Gp--+q(xa)dxa is the probability of finding a quark with momen tum
(5 . 1. 2)
and Gp--+q(Xb)dxb is the probability of finding an antiquark with momelltunl
(5.1 .:!)
where PA and PB are the momentum of the initial two protons as showll ill
Fig. 5.1. It is convenient to define the dimensionless variables
T = /If:! / s, (5.1.4)
and
(5 .1.5)
where M is the mass of the muon pair and where s is the external proton-
proton center-of-mass energy squared
= (PA + PB )2 = 2Pc2m,
s (5 .1.6)
Pcm = hIS, (5.1.7)
and s is the internal parton parton center-of-mass energy squared
s = (Pq + pq)2 = 2pq,pq . (5 .1.8)
5. 1 'I'h<' N I Vtl I' r t llll Mod,,1 172
(b) p + p..... y* + X
o y*
WA
Figure 5.1 (a) Illustration of the collision of hadron A (momentum PA) with
hadron B (momentum PB) producing a p+ p- pair via quark-antiquark annihilation.
(b) The production of a virtual photon in proton-proton collisions, p + p --+ .,.* + X
is described in terms of a 2-t0-2 parton subprocess in which one incoming parton
has momentum xaPA and the other has momentum XbPB.
(5. 1.1 2)
si nce in this case the muon pair in Fig. 5.1a has no transverse mOlllcul,lIl11.
Equation (5.1.11) implies
XL=Xa-Xb, (5 . 1.1 :1)
where
XL == 2PL!vs, (5. 1.11\ )
a.nd (5.1.12) gives
XE
2
= XL2 + 4T,2 (5 . L.1 5)
where
XE == 2E/VS. (5 .1.16)
T he total cross section for a quark and anti-quark to annihilate into a muon
pair, q + if -+ 1'+ + 1'- is given by
, _ + _ _ 1 47ro:-., e q2
u(q + q -+ l' l' ) = Uo = 3' 3M2 ' (5 .1.17)
which is the same as the e+e- -+ 1'+ 1'- cross section computed in (2.1.14)
except for the color factor 1/3 and the replacement Q2 = M2, where M is the
virtual photon invariant mass with
(5 .1.18)
or
f = 1. (5 .1.19)
The color factor of 1/3 comes from our convention of averaging over initi al
state color. In this case there are three color states so the color factor is
3U) U) = ~. From (5.1.10) and (5.1.13) we see that Xa and Xb are completely
specified in terms of T and XL according to
(5 .1.20)
(5.1.21)
and (5.1.1) gives
dUDY 2 47ro: 2 1
- dd (s, M = 9 ~12 ( (5 .1.22)
T XL
,XL)
11 Xa + Xb ) Pqq(x a , Xb),
r,,' (:111011 1 ': 'lIi ~l ioll li lld 11111, 11 1 ,11.11 1". ( 11111111 1 1'7'1
Y=_ 1
2 log
(E+PL)
E - PL . (5.1.26)
In this case,
We now consider the possibility that the initial quark or antiquark can radiate
/
171S C hAp f" r 1\ M IHIIl I' I 1',11.1;0. Lilli
gluon, qo
"5
Figure 5.2 Leading order diagrams for the quark-antiquark "annihilation" subpro
cess q + q -+ 'Y. + g.
a gluon before annihilating into a virtual photon as shown in Fig. 5.2. The
differential cross section for the subprocess q + ij - "Y. + g is
dO-by (- i\ 1
- d - s, t, = 64 - -2
IM( q + q- - "Y • + g)12
t ~sPcm
1 - _ • 2
= 16~s2IM(q + q - "Y + g)1 . (5.2.1)
In this case,
- -2 1 -2
S Pcm = 4S . (5.2.2)
and the amplitude squared is given by
IM(q + ij - "YE + g)12 =e2e;g~ H
! 2M 2 (M -2 -i-u)} ,
s{ !-+-+ (5.2.3)
t u til
where 4/9 is the color factor and 1/4 is the spin averaging factor. As before
the subscript ~ refers to the use of (4.2.19) to sum the photon polarization
states. Except for these color and spin factors this amplitude is the same as
IM("YE + g - q + ij)12 in (4.3.21) with the replacement Q2 - _M2 since in
this case the invariant mass of the virtual photon is timelike,
M2 = q;, (5.2.4)
rather than spacelike as in (4.3.9). In this case the invariants are given by
S = (Pq + pqi, (5 .2.5)
t- = (q-y - pq) 2 , (5.2.6)
U = (Pg - pq)2, (5.2.7)
~.~ 0 1111111 1-:1111..1011 anti IlIlli,,1 ~H.I, O lllt)lll 176
Q,Pq
t- +
(5.2.10)
where
i min = 0, (5.2.11)
and
i max = M2 - S = -(1 - f)s, (5.2.12)
with f defined in (5.1.5) . This integral diverges for precisely the same reason
as the deep inelastic scattering integral (4.3.22) and again we cannot proceed
without a regularization scheme.
We must also correct the naive parton model by including the "Compton"
subprocess q + 9 - "1. + q shown in Fig. 5.3. The differential cross section for
this subprocess is
duhy (A '"
di s, t J =
1 IM(
1611'8 2 q + 9 - "1
•
+ q)12 , (5.2.13)
7
IM(g + 1f
which except for the color factor of 4/24 and the spin averaging facto r of J/~ ip.;
I.hc same as IM(-yi;+q --+ q+g)j2 in (4.3.10) with the replacement Q2 _ 111 ".
The invariants in (5 .2.14) are defined by
s = (Pq + qg)2 , (5" U fl)
~ 2
t = (q-y - pq) , (5 .2. 1(1)
u = (q-y _ qg)2, (.').2 . 17)
where M2 = q~ is the virtual photon mass squared. Substituting (5. 2. JI1) illto
(5. 2.13) gives
diT~y (s
dt
j\ =
' ')
7rQQ$e~
s2
!
3
{_is _~t + 2M2(s +!st - M2)} ' (5 .2. 18)
(5 .2.19)
is again infinite with i min given by (5.2.11) and we cannot proceed without a
regularization scheme.
(5.3.1)
The integration of this over i as defined in (5.2.10) is a bit more difficult than
the other cases we have considered so far. One must be careful to keep the
exact form for imin and i max . Namely,
with
fJ = 7n 2 /M 2 ,
g (5 .3.3)
where + is defined by (5.1.5) . The result is
(5.3.4 )
where t max and tmin are given by (5.3.2). In arriving at this result it is conve-
nient to use the same trick as in (4.4.54) . We write
so that
ut 2M2(M2 - t - u) -
- + - +
tutu -
(8
--
(1 + +2)
t (1 - +) -1
)
+
(8
--
(1 + +2)
u (1 - f) -1
)
.
(5.3.6)
Integrating the first term over t and the second over u gives
(_~(I+f2) _ )
tmiD
I
2 .
t mu t (1 _ f)
A
A
1 + f2)
[(
1 dt-
A
A_
A A ]
(5.3.8)
l'
,
(I':U ••
3 12)
/
170 MIIII" I ' .. 11'" ,, 111 11,111"
where t.h e q CO rr('H pOlld li t." 1,111' CO Ilt.rihutio ll 1'1'0 111 th e "allll ihil at io ll " H1Ihpr!)
C<'HH a nd the lililittl of illt.("~ rll, ti II 1\.1' d tc rl11ined frOll1 (5 .:1.11) usi1l g th l' f'ltd,
t.hat 0 :S T:S 1. The dOll b lc Jifrcrcutial cross sec tio n is [eta,tcd to do- jelt' hy
dUg 0' dUg
dM~~/q+ij-+J-I+J-I-+g)= 37rM2 :r(q+ii-+,* +g ), (:J .:1.I :1)
where the factor of 0'/(37rM2) comes from integrating over th e nili OIl I HI,II'
angular distribution . Integrating over i yields
where 0'0 is the parton model Born cross section given in (5.1.17) . C II I()II,ri III'.
(5.3 .13) and (5.3.4) we see that
(
1
0'0
dU'J.1G)
df DY
2 20'.
37r
{I + T,2
1- T
log ((1 -:: f)21\.12) _ 2(1 _ f)
T2m~
f 1-2..,fo log(tmin/tmax)
" 2
df = ~ _ 1 2(2)
10 1- f 12 og
1- 2..,fo log(,8f2/(1 - T)2) ,
+ 1
o
1 '
-T
dT, (5.3.16)
where i min , i max are given by (5.3.2). Integrating (5.3.15) over f from Tmin = 0
to Tmax = 1/(1 + ..JiJ)2 ~ 1 - 2ViJ gives
The virtual corrections shown in Fig. 5.4 are the same as the e+ e- ca.
in (2.7.3) but with Q2 = q~ = M2 (q~ timelike) . Namely,
'
( O'MG ( rea1) + O'MG
, vtrtua1)) DY == as I q 0'0 = 20's
(. DY
37r 0'0 {7r-
., 7 - 3
- 2' 27r
2
+ 7r 2}
= 20'. 0'0 {47r2 _~}
37r 3 2
1\ :1 hd r (Y, '(jrr~ l tl " ~ M ( l • c I..,.,." 180
• • •
1'; qy
q.Pq
l+~ ~. •
S
q.Pq q.Pq s q.Pij
+
q.Pq s -q.Pij
so that
DY
(Ttot = (To { 1 + (811"
""9 - 311"
7 ) a, + . ..} , (5.3.24)
or
(Tfor = (To (1 + 2.05a, + ...). (5.3.25)
I t! I ( Jh '1'1. ...
'I'lIt' ord er n. (. ITndlllll ll 1.11 1.1", 1)1',,11 YII,II pro '88 IU '( vny 111,1'.0;(' alld IHm i t.i V( ,' ,!I ,
' I'h e coe Hi ci nl. of 1.11(' II , 1.""111 il-l 1-10 la rge that o ne mll st w rry abo ll!. th., CO li
t.ribu tion from til · n " ~ 1I.lId hi gh I' order terms. Can we b li cv th fil'l;1. I.WIl
I. rms in (5 .3.24 ) ad'tI'J'I.tcly reproduces the sum for a, ~ l/a or IN? Fo r
n. = 1/3 the second term represents a 66% increase over the Born t. rnal
Actually the situation is better than it looks. If we keep track of I.h r 71'''
that comes from the analytic continuation from the spacelike (deep i IId nsl.ic
scattering) to the timelike (Drell-Yan) region of q2 we have
DY
O'tot = 0'0 ( 1 - O,045a, + la,
2 7+1.'") ) (5 .3.27)
7
=0'0exp(2;a s ) {1+ (2; - 3 71')a,+ ... }
(5.3.31)
where IfY is given by (5 . ~ . 23) a.nd we can now d fin c "- fllll r t.i li S" jll s ~ }\Ii
we did in Chapter 3 and in C h ap~ e r 4. Namely,
(5 .3.32)
where
, 4 (1 + f2) (5.3.33)
P q--+ qg (T)=3" I-f +'
is the same function as (4.4.31) and
where
11 o:.fJll/ (f) df = 0, (5.3.35)
is valid regardless of the scheme. We now write a new differential cross section
that includes both real and virtual corrections
~
d,q
(7M~.DY = ~
(d (7M~.DY
,q) + o:.IDY 8(1- f), (5 .3.36)
(70 dT (70 dT q
+
and inserting this and the Born term into (5.3.12) yields
where f = T/(XaXb) and the superscript q refers to the contribution from the
"annihilation" subprocess q + if -+ r* + g.
For the initial state gluon "Compton" subprocess, 9 + q -+ r* + q, shown
in Fig. 5.3 we regularize by taking the gluon off mass-shell q~ = -m~ as we
did in Chapter 4. The differential cross section in (5.2.18) becomes
daJ)y , ' _ 7ro:o:.e~ ~{_i ~
, (s, t) - '2 3 ' '
_ + 2(M2 - m~)(5 + i
"
- M2 + m~)
dt sst st
- M 22(1
mg 52 + i2I)} ' (5.3.38)
/
I I-I:J 1\1,"11, I' ll , 1'1 "dll' j 1111
III Lhi s cas t ill' l illli t,,1 01 IId"'p,rlt tI O Il (5.:.! .J9) bee IIl e
M '2 m g2 2
tllli ll = - -A- = -mgT ,
A
(5.:I.:W)
S
i max = M2 - S = -M2(1- 1' )11', (5 .:3.'1 0)
I\.nd integration over i yields,
dt _ S
j
imi n
2
mg. t'2 - - M2 . (5 .3.4 2)
tma.x
- 21 + T, - 3
2T~
A"} . (5 .3.1 5)
Inserting (5 .3.43) into (5 .3.12) gives the gluon contribution to dO'ldT. Namely,
dO'DY(
g 4
2) _...!:a
S dM2 S, M - 9 M2
eq 2211 11 dXa d
...!.!.. (0) ( ) (0) ( )
G p-+ q Xa G p-+g Xb
T Xa T/X 4 Xb
where
(5.4.4)
with N = 4 + { and
i = -!(s -
2
M2)(1 - y) = -~(1
2
- f)(1 - y), (5.4.5)
U = -!(s - M2)(1
2
+ y) = -~(1
2
- f)(1 + y). (5.4.6)
(5.4.8)
where the mass mD comes from using (2.8.26) and (l'. is now dimensionless.
Using the N-dimensional analogue of (5.3.14),
(5.4.9)
/
MIIIIII "11' " II,dull. (III
where the facLor o f ( i I ~ ) ('0 1lH'1:I frolll evaluaLillg Lh e 11 0 1'11 ter lll ill N di llll' lI
!lions,
(G .-1 . IO)
yields the result
( ~0"0 d&bR)
df DY
=2 20:.
311"
(1- f)2.\12) </2 r(1 + ~)
f411"m'b r(1 + ()
1+r22 (l-r)}
{ I - f ~+ 1+( ,
(f> .tJ • I J )
r (1- 2() = 1+
1
2'E (+ "6 + IE 2) (+
'18 (11"2 2 ... , (5.1.1 -1)
~ (reaI) + O"DR
( O"DR ~ (. vutuaI) DY == 0:. /DY
q 0"0 = 20:.
311" 0"0 { 3
411"2 - 27 } ' (5A . IG)
in the limit ( --4 0, which is the same result as in (5.3.19). We now define
"+ functions" and arrive at
~q
1
0"0 (dO"DR,DY
df
) _ 0:. ~
+ - 2211"Pq-+ qg (r)log(M /mD)
2 2
+ 20:.fDR
q,DY ~
(r), (5.4 .16)
-1 (dB-bR)
-d~ a. Pg..... qq (~)l
= -2 T og (M 21 mD
2) + a. fgDY(~)
DR T, (5.4.18)
Uo T DY 11"
where Pg ..... qq(f) is given by (5.3.44) and
If we combine the "annihilation" term with the "Compton" term and also
include terms with the initial two partons interchanged, then the muon pair
(or "Drell-Yan") cross section becomes (for one quark flavor)
s dUDY (s M2)
dM2'
= 411" a2e~
9 M2
11 11
r
dX a
Xa r/x¢
dXb
Xb
-(0) ~o) -(0) -(0) )
{ ( Gp ..... q(Xa)Gp ..... q(Xb) + Gp--+q(xa)Gp--+q(Xb)
[( ufor Iuo) 8(1 - f) + ;; 2Pq..... qg ( f) log(M2 I A2) + 2a./q,DY (f)]
+ (~o) -(0) ~o) ~o)
Gp ..... q(xa)Gp ..... g(Xb) + G p--+ g(xa)G p..... q(Xb)
)
If W 1I 0 W dl'jilll ' " lh ,, 11 Villi " qu ark Jistriblll.i 11 /:1 by 1.11(' parto ll ,"oliol
,tV.,
forll1ul a in (5 . 1.:30) WI ' Ii
(5 .S.:l)
(5.5 .5)
and
(5 .5.6)
The "convolution notation" defined in (3.3.11) is being used and also I have
used the unexponentiated form of ufof in (5.3.24),
ufof = [1 + - (811'
--
Uo 2 9
1
- - -
311'
7) a. ][ + -12 (811'-9 - -311'7)]
1 a. + 0 (2)
a. . ( 5.;)" .7)
These "Drell-Yan" parton distributions are scheme dependent since the "I it-
tie f" functions are scheme dependent. We now express them in terms of our
reference distributions from deep inelastic scattering
(5.5 .8)
as in (4.9.7), then from (4 .9.1) we see that
DY
Gp--+q(x, M 2) =Gp--+q(M 2 ) * [ 1 + a.D.fqDY + '12 as (811'
"'9 - 7 )]
311'
and
where
a.D.ff Y (z) = a. (fXf~Y (z) - fXf~~; (z))
q,DY()
= a. (fDR z - fq,DIS())
DR,2 Z
(5.5.15)
/ LSf) ' h n.Jli. r
(0)
p~
-
:~.
,,.Q
y
• ..(
+
+
Q 9 \,.-
P
P~Q 9
:t/,.+ I
, ,.+
(b) •• +
Q Y \
\IA--
P
F igure 5.5 Illustrates how in the production of large-mass muon pairs in hadron-
hadron collisions the leading log contributions from the subprocess q + q -+ " . + 9
and ij + 9 -+ ,," + ij are summed to form the M2 dependent quark distribution ,
Gp_q(x, M2), where M is the mass of the muon pair. Similarly, the leading log
contributions from the subprocess q + q -+ 9 +,," and 9 + q -+ q +,," are summed
to form the M2 dependent antiquark distribution, Gp_q(x, M2).
dCTDY 2 )
( S dM2 (x, M ) .
leadmg order
= 4?1"
9M2
a2
1
r
1
dX a DIS
-
Xa
2
Pqq (Xa, T/X a, M ).
(5.5.16)
This is illustrated in Fig. 5.5 where the leading logarithm contributions from
t he subprocesses q+ij -1" + 9 and q+ 9 -1" + q sum to give Gp_q(x , M 'i)
and Gp_q(x, M2) just as they did in Fig. 4.8. However, at order a. , the Drell-
Van cross section becomes
11 11
dM order a. CTO dM leading order
+ 4?1"
9M2
.
a2
1 111
dX a dXb pDIS(
r Xa .,./z:. Xb
99 Xa,Xb ,
M2) ~fDY( T" )
a, 9
411'(1'21
+ - - --
9M2 r
1
dX a
Xa
11 tlJ:b [>D / S (
r/ x. Xb
fig x", X, . I" M ~ ) "f/))'()
n. '-' II T, (5 .5.17)
(5.5.18)
and
n,
p~IS(Xa,Xb,Q2) = L:e~; [G p .... q;(xa'Q2)Gp .... g (Xb,Q2)
;=1
(5.5.19)
respectively.
The Drell-Yan "K-factor" is a complicated function and is defined to be
the ratio of the order a. result to the leading order result. Namely,
(s!k=.(s M2»
K (s M2) = dIll" order a. (5.5.20)
DY, (!k=.( M2»)
s dM' S, leading order
ILlid il. is (5 .:L2!J) I.hlli. WI ' ,1111 II lid 11 M" for (Jp~'. III thiN (liNe
"g
I\.Dy(s,M 2 _
)- {11 11
T
-dX
Xa
a
T/:t:,.
-dXb Pqg
Xb
DIS (xa,xb,M)
2
a.t:.!gDY (r' )
+11 11T
dXa
Xa T/:t:,.
dXb PgIS(Xa,Xb,M2) o:.t:.!J>Y(f)}
Xb
/{1 1~a D~ 2}
Xa Pqij (xa,Tjxa,k!) ,
T (5.5 .27)
(5.5 .28)
where f = rj(XaXb) and where t:.j[>Y is the same as (5.5.11) but with th
b-function contribution removed. In particular,
Af-DY(')
a • .u q T
= 2a.
37r
{(I + T
-2) (log(l-
1_ f
f») + + ~2 (1 _1f)+ _ 3 _ 2r- }.
(5.5.29)
As can be seen in Fig. 5.6 the term Kby{s, M2) is small except when r
approaches one. The "+ function" terms in (5.5.29) cause these corrections
to be arbitrarily large near the kinematic boundary. This can be seen clearly
in the "Drell-Yan" structure functions defined in (5.5.9) and (5.5.10). The
/1 ,/\ 'j 'lI Dr II Y II " /\ FN tor" 102
II II
Drell- Yon K- factor
3 8
..
.. KOY
.....
2 ...... .. .....
....... ..... ... . ...
O ----..,-.~ -:-.:..
---------
._. -. -. --- ........
-I
0,0 0.2 0,4 0.6 0.8
./T
Figure 5.6 Drell-Yan "K-factor", KDY, computed with A = 200 MeV at Va =
27.4 GeV plotted versus ,fT. The qua.rk, gluon, and c5-function contributions a.re
shown sepa.rately with KDY =
KJ)Y + Kl)Y + Kl)Y.
1.8
Q =10 GeV
N-..
0 1.6
- )(
:::J
t 1.4
0. /
(9 /
NO -.......
1.2 "
/
/'
"""""
-
>- t
)(
:::J
., ,,"
."
00.
1.0 ,. .,"
.,.,"
(9
Figure 5.7 Ratio of u-quark probability distributions defined in the Drell- Ya.1I
process. Gf.!u(x. Q2). to the "reference" distribution defined from the 7"2 structUr('
function in deep inelastic lepton-hadron scattering. Gp":"u(x. Q2). at Q = 10 GcV .
The solid (dashed) curves correspond to defining the Drell-Van quark distributiollR
without (with) the total cross section afar
explicitly removed.
Z _ 1. The solid curve in Fig. 5.7 shows the ratio G~.!u(z. Q2)/Gp ..... u (x. Q2)
=
at Q2 100 GeV 2 .
Instead of defining the "Drell-Yan" distribution functions according to
(5.5.3), it is perhaps better to explicitly remove tTer. Suppose we define the
"Drell-Yan" structure distributions according to
8
dtTDY
dM2 (8, M ) =
2 DY
tTtot
11 T
dz o ~DY
Zo P'f (zo, r/za, M ),
2
(5.5.31)
•. , '1'1,,· 1)1,.,11 VII " " /1 1,'/111.", " 104
with
(5.5.32)
(5.5.35)
sIn ce
(5.5.36)
With this definition (at least at this order of perturbation theory) the number
of quarks in the proton is the same for the "Drell-Yan" as for the reference
distributions.
The extraction of ufoi' in (5.5.31) removes most of the c5-function con-
tribution (Llf~Y still has a small c5-function term) and as long as one stays
away from the x -+ 1 region the structures functions, CDY (x, Q2) in (5.5.32)
do not differ greatly from the deep inelastic scattering reference distributions
G(x, Q2) (dashed curve in Fig. 5.7). The major difference between the leading
order and the order 0:. Drell-Yan muon pair cross section lies in the multi-
plicative factor
(5.5.37)
which for 0:.(JI12) = 1/3 is about a factor of 2 as shown in (5 .3.30). This can
be compared to the e+e- result in (2.7.8) of
uf~e- luo = (1 + 0.3180:.(Q2) + .. .), (5.5.38)
which for 0:. (Q2) = ~ is only a factor of 1.11 (i.e., about a 10% effect).
Experimentally the Drell-Yan "/{-factor" is roughly a factor of 2 which
j ()I'I <J II I'pt.",
II.}J; rees well wit.h t.11( ' ",,1,-, " . ",,·dI Ct. illil t o~(' t,)l('r witli t,)1(' ('X I)(l lH'llt ill t.. ·d ",,· ri. '11
III (5.5.37)8. Thi H 11111'/' (1 '1 '.~ t.o 1)(' II gr(,at t.riumph for P ' rt.llriJativ(' QCD . 'I'll('
,'x perimental Drdl- Vltll 'IH';t.H llrCmellts can be interpreted as a. IIH';UillI'C 'III. ,.iI.
of (J'e,r. This measurcm cnt is as significant as the experimcntal mcas ur(' IlIi'lit
of the order a. correction to (J'i~e- in (5.5.38) . The Drell-Yan eXpe rillH'.iI.lI.1
results deviate from the naive parton models in the precise mann er PI' did,('d
by perturbative QCD.
As we have seen, the "little f" functions are process depend cnt . 'l'hi H
means that the "K-factor" differs from process to process. Here we hav(' oill y
considered the Drell-Yan case. The "K-factor" must be recalculated fo r ,'v.'ry
process of interest, For example, the "K-factor" for quark-quark H 'aUt'r ill )!,
(i.e ., the subprocess q+q -+ q+q) has not yet been calculated. 11 w('V('r, tli."
exponential, exp(21ra.j3), in (5.5.37) arises from the 1r2 that came 1'1'0 111 th. ,
analytic continuation from the spacelike (deep inelastic scattering r(' f(' r('II ('I'
distributions) to the timelike (Drell-Yan) region of q2. Hence, we might (,X I)('c t
that whenever we try to describe a timelike (TL) process in terms of th e (11'.'"
inelastic (space\ike) structure functions we will have a correction term of tli. ,
form
TLj (J'o
(J'tot = exp (21r 2 )
-aa.(Q) (1 + Ba.(Q 2 ) + .. -), (5.5 .~19)
where the coefficient B must be calculated for each process under considera-
tion and it will depend on the choice of variable Q.
In QCD, transverse momentum of partons can arise in two ways. Firstly, ill ,
for example, a proton beam, quarks are confined in the transverse directi II
to within the proton radius. Therefore, from the uncertainty principle, th ey
must have some transverse momentum. This momentum called primordi a l,
is intrinsic to the basic proton wave function. It involves small Q2 valu es
and cannot be calculated from perturbation theory. It must be consid er d
unknown by bounded, falling off like an exponential or Gaussian in transverse
momentum, k T . This primordial parton transverse momentum "smears" th e
parton model Born term in (5.1.22) causing the muon pairs to have anon-zero
but bounded transverse momentum.
Secondly, in QCD, one expects an "effective" kT of partons in protons
due to the hard Bremsstrahlung of gluons which can be calculated pertur-
batively if the momentum transfers are large. For example, the perturbative
component of the transverse momentum of muon pairs is generated by the
2-to-2 constituent subprocess q + if -+ r* + q ("annihilation") in Fig. 5.2 and
q + 9 -+ r* + q ("Compton") in Fig. 5.3. The cross sections for these processes
10(1
(5 .6.1)
(5.6.2)
where I have used (5.3.13) and where the subscripts A and C refer to the
"annihilation" and "Compton" subprocesses, respectively. These subprocesses
must be embedded in the proton-proton collision reaction as in (5.3.8) . For
example, for the annihilation contribution we have
(5.6.3)
where
Pq =XaPA, (5.6.4)
Pq =XbPB, (5.6.5)
and
S =(PA + PB?, (5 .6.6)
t =(PJ.IJ.I - PA )2, (5 .6.7)
U =(PJ.IJ.I - PB)2, (5 .6.8)
which together with (5.2.5), (5.2.6), and (5.2.7) give
= XaXbS,
i; (5.6.9)
i - M2 = xa(t - M2), (5.6.10)
it - M2 = Xb(U - M2). (5.6 .11)
The condition
(5.6.12)
implies that
(5.6.13)
(5.6.14)
with
Xl = -(u - M 2 )/s = Hx} + 4T)teY, (5.6.15)
X2 = -(t - M 2)/s = ~(x} + 4T)t e- Y, (5.6.16)
where
XT = 2[JT/VS, (5.6.17)
where PT is the transverse momentum of the muon pair and y is their rapidity
/
11)7 C I, ",pt" " , 1\ MIIHII 1'1111 1',01111 1'1.1011
I' M d ·fill ed ill (I) . I .~W) . III t.lII M ('I ','l(', Il owever,
J.2
' I:: -- x T2 + x L2 + 4r I
(5 .n.1H)
I'! ' places (5.1.15) and
(5,6. 1fl)
rep laces (5 .1.27) and
(5 .6.20)
replaces (5.1.13). Equations (5 .6.10) and (5.6.11) become
i= -Xa SX 2 + M2, (5 .(U I)
U = -XbSXl + M2, (5.(l.'n)
and the cross section for producing muon pairs of mass M, rapidity V, II l1d
dMdUDY
Y PT
( 2 )
2d d 2 S, M , y, PT -
-1 1
X::"B
(0) (
. dXaGp-+q Xa ) G p-+
(0) ( )
q Xb
XaXb )
( Xa - Xl
(.!. dMdO-A2dt,(s,l)), (5.6 .23)
7r
where
min Xl - T
Xa = ---, (5 .6.24)
1- X2
is determined from (5 .6.14) by setting Xb = 1 and Xb is related to Xa as given
in (5.6.13). Summing leading log corrections converts G(O)(x) into G(x , M 2)
so that (5 .6.23) becomes
(5 .6.25)
(5.6.27)
Xa - Xl
so that
u = - ~ { (1 - f) + [( 1 - f? - 4p} / s] ! } . (5.6.30)
If we ignore the scale breaking of the structure functions in (5.6.26) (i. e.,
neglect t he M2 dependence of P(Xa, Xb, M2» then,
2 2 dUA 2
M Pr d d d 2 (x,AJ ,y,pr) = o:.FA(T,y,Xr), (5 .6.31)
T y Pr
The perturbative contributions in (5.6 .25) are shown in Fig. 5.8 together
with data on PP --> J-l+ J-l- +X at M = 8.4 GeY and W = v's = 27.4 Gey9. The
perturbative contributions are absolutely normalized and agree roughly with
the data at large transverse momenta. They, however, have the wrong shape
and diverge at Pr = O. We will see in the next section that this divergence
occurs at every order of perturbation theory but disappears in the sum. In
addition, nonperturbative effects at smalipT regularize this singularity leaving
a smooth transverse momentum distribution lO •
The soft, nonperturbative, primordial transverse momentum of the par-
tons within the initial protons produce a net transverse momentum, kr, of,
for example, the initial qij system that can be parameterized by a gaussian
f(k})
1
= -1rU (_k2)
42 exp 4; , (5.6.33)
qu q
y=o
M=8.4 GeV
W=27.4 GeV
(p2) = 1.9 GeV 2
T data
10-10
10-1IL-_ _L -_ _ _L-_--'L-_--:-'~-~
0.0 5.0
where the subscript S refers . to "smeared" and where I have included the
virtual corrections ov(s, M2) that contribute only at the point where kT = Pr.
Equation (5.6.35) can be rewritten in the form
O's(s, M2, y,p~) = J O'p(s, M2, y, q~) [f«Pr - iT )2) - f(p~)] d 2qT
This is a rather large value for the primordial transverse momentum of par-
tons within the proton. It means that at the energy in Fig. 5.9 about 0.9 Gey2
of the muon pair mean p} is due to the primordial motion and about 1.0 Ge y2
arises from the hard QCD subprocesses. We shall find out in the next section,
however, that summing all the leading double logarithmic terms causes the
perturbative contributions to vanish at PT = 0 and it is this corrected pertur-
bative cross section that should be "smeared." Doing so results in a slightly
smaller value of the primordial kT.
Actually the fit in Fig. 5.9 cannot by itself be viewed as a success of per-
turbative QCD. One could have fit the same data with just the Gaussian in
(5.6.33) provided 0'9 = 0.677 GeY. The test of the presence of the perturbative
component to the transverse momentum of muon pairs comes from examin-
ing the energy or mass dependence of the muon pair PT spectrum. Figure 5.10
shows that as the energy increases (at fixed At/) the spectrum becomes flatter.
This change in the muon pair transverse momentum spectrum is a general
PP .. fL+ fL- + X
dCT /dMdyd2PT
y=O
M=8.4 GeY
W= 27.4 GeY
(0 2) = 1.9 Gey2
r,- iiata
\
\
\
10-8
\
\
\
\
\
~\motion only \
\
\
\
(k~)prl_dlal \
- 0.92 Gey2 \
\
\
\
,
\
Figure 5.9 The distribution in transverse momentum, PT, of muon pairs, ",+".-
= ..;; =
produced in PI' collisions at W 27.4GeV compued with the leading order
perturbative QCD result "smeared" with a Gaussian primordial transverse momen-
tum spectrum with (kT)P-q = 600 MeV (solid curve). The dashed curve is the
primordial transverse momentum spectrum with no perturbative QCD terms (taken
from Ref. 9).
feature of QeD resulting from the lip} behavior of the perturbative contri-
butions. As we will see in more detail later , the mean value of p} has the form
IU L"'K,, 'I " Vt'fH(l MOIII (' lI t UIII MII O II I'llrn 211 2
pp- fL+fL- + X
M=8GeV
y= 0
<k2> = 0.9 GeV 2
T primordial
-W=53GeV
--~W=27.4 GeV(x3.9)
_ ••••• W= 19,4 GeV(X 18.3)
10-8
..,
,....--,
~
<.9
'-
.D
:t
L....-..J
~
a.
10-9 ..
N
"0
>-
"0
~
..
"0
'- W=19.4:
b
10-10 ··
"0
·
~
··
···
10- 11
0 2 4 6 10
PTGeV/c
Figure 5.10 Energy dependence of the large transverae momentum "tail" predicted
for the production of muon pairs in pp collisions from the leading order QCD sub-
processes. The leading order perturbative QCD result has been "smeared" with a
Gaussian primordial transverse momentum spectrum with {k T }p-9 600 MeV. =
(neglecting logarithms),
(Pf}p oc a,(M 2 ) (pf)max, (5.6.41)
/
MII . II. 1'111, 1',uolll , I,11) 1I
wl, cr'
1. ) "III. - l's [( 1 -
(7"1' r
)2 - XL
.2]
at fixed values of the dimensionless variables rand yor r and XL. Prcs llll ilthl y
the primordial component is independent of the energy and the of til 1111 10 11
pair mass so that the resulting mean p} of the muon pairs has the form
(p}) ~ (P})primordial + C 0'.(M 2) s, (5.6.01 01)
1
or
(5. 6.45 )
at fixed rand y, and C 1 and C 2 are approximately constant and can be
calculated from (5 .6.25).
(5.6. 17). It woul d ap pear th a t t he no ~ ~ect i II (5 .7. 1) di v(' rg"s like Lh)~. as
PT -+ O. However, this is not quite correct sin ce
I x}
x
min
a
- Xl =4 --- '
1- X2
(5 .7.4)
and
X2
xminx _ T _ ! __T_ (5 .7.5)
a 2 -41 - X2
'
so that the integral over Xa gives a logarithmic divergence as XT -+ O. In
particular,
(5.7.6)
,M2
,V)) {4(1', 2},
d(J'DY 2 d(J'DY 1
d d d 2
T Y PT
(s, M , y,PT) = ( dT(s
T Y Born
-3 "2log(s/PT)
11' PT
(5.7.12)
where
(
d(J'DY
dT(s,M ,V)
2) 411'(1'2 DIS
= 9M2 Pqq (xa,xb,M),
2
(5.7.13)
T Y Born
(5 .7. 1-1)
radiation damping occurs and the spectrum goes rapidly to zero . U111'0 1'1.11
nately at center-of-mass energies less than 200 GeV this perturbativ peak iH
washed out by the primordial transverse momentum smearing as in (5 .6.:31) .
However, the peak should become evident at very high energy.
In the naive parton model the weak intermediate W± bosons are produced ill
proton-proton collisions via the subprocess qi + iii W±. The experiment a l
-)0
U(qi + ijj -)0 W±; 05) = ~s IM(q; + ijj -+- W±W 8(05 - Ala,), (5.8.4)
where IMI2 is the spin averaged matrix element squared and Alw is the mass
of the vector boson . The Feynman rules for the couplings of W bosons to
quarks given in Appendix C yields
IM(u + de -+- w+W = H 8grv o5 , (5.8.5)
where 9w is the dimensionless weak coupling
2 GFMa, 71"0'
9w ' = 1n2 - -2- , (5 .8.6)
VL. XW
and Xw is the square of the sine of the Weinberg angle, Xw = sin 2 (Ow) and
G F is the Fermi constant. Here we sum over the three allowed helicity states
of the W± boson (A = 0, ±1) by using (2.2.18). The factor of 1/3 is the color
factor that arises from our convention of averaging over the initial color states
and summing over the final color states and the factor of 1/4 comes from
averaging over the quark spins. Furthermore, d8 is the rotated state
dB = d cos(Oe) + s sin(Oe), (5.8 .7)
where Oe is the Cabibbo angle. Combining (5.8.4) and (5.8.5) gives
u(u + dB - W+j s) = i7l"g~ 6(8 - Ma,), (5.8.8)
or
(5.8.9)
and
(5.8.10)
The W- cross sections are similar but with all particles replaced by their
respective antiparticles. The experimentally observable cross section in (5.8.1)
becomes
u(p+p- W± +Xjs) = J J
dX a dXb
(5.8.12)
where rw is defined according to
rw = Mev/ s . (5.8.13)
The joint probabilities Pqqw-J: are given by
PqqW+(x a, Xb) = [Gp-+u(Xa)Gp-+d(Xb) + Gp-+d(Xa)Gp-+u(Xb)] cos2(Oe)
+ [Gp-+u(xa)Gp-+.(Xb) + Gp-+.(xa)Gp-+ u (xb)]sin2(Oe), (5.8.14)
2117 Gh ptl1r /) Mil l'" I' , I',", hll t UII
gluon, qo
,.
5
u(p+p-+Z o +Xjs)=2v21rGFTZ
In
a
111 TZ
dz o
-PgqZ(Zo,TZ
Zo
/ zo), ( 5 .8.16)
where
TZ = Mils, (5.8.17)
and where the joint probability PqqZ is given by
Pqqz(zo, Z6) = [Gp....." (Z 0) Gp..... a (Z6) + Gp ..... a(Zo)Gp....." (Z6)] a- ~zw + ~z~]
+ [Gp ..... d(zo)Gp ..... izb) + Gp ..... J(Zo)Gp ..... d(Zb)
+ Gp ...... (zo)Gp ..... '(Zb) + Gp..... ,(zo)Gp...... (Z6)] [~ - izw + ~z~]. (5.8.18)
In QeD we must also consider the possibility that the initial quark or
anti quark can radiate a gluon before combining to form a W± or ZO boson as
shown in Fig. 5.11. For massless partons the differential cross section for the
.11 Yo t or BOHO/l 1'1I"lu d u/II 208
+
W-.qw
I ,,
qj.Pq ,,
,,
,
S gluon.qQ
Figure S.12 Leading order diagrams for the "Compton" subprocess qi+9 - W:!:+
qj .
(5.8.22)
pa rt n Ill od 'I for llllJl " III (fl.lt I :l) h" (,O IlH'H
Xa TW TW Ix. Xb
Xa TW TW Ix. Xb
+ v 27rGF3"1]1 -dXa]l
/0 dXb DIS
-
2)
DY
Pifgw±(Xa,xb,Mw a. t1 f g (rw),
A
TW Xa Twl x • Xb
(5.8.26)
where TW =
rw/(XaXb), rw =
Mar/s, and where t1ffY and t1ffY are
the same functions that appear in the Drell-Yan case, (5.5.11) and (5.5.12),
respectively. The joint probabilities qijW± are given in (5.8.24) and (5.8.25)
and the new joint probabilities qgW± and ijgW± are given in terms of the
lUi VodOi 11 ')/<011 I', n dll l !.lUll 21 II
a nd as in ti l(' Drl ·1! \'1111 111 111' (,,1 11111111.11 or t lar c t e n liS,
(5 .IUI!)
The a-function terril is id nti cal to the DreIl-Yan case and is givell by
w±
., 6 9
1\
W
± = --
O"tot
0"0
+ -0$
371"
+ ...
DY 9
= 1 + o$Iq + 371" O. + . ..
2
= 1+ (8; + 371") o$(Ma,) + .. . , (G .!Uh)
/ {1
1 dXa DIS 2 }
(5.8 .38)
TW Xa Pqqw±(Xa,rw/Xa,M) ,
and
(5 .8.39)
where D.j[>Y and D.ffY are given by (5 .5.29) and (5.5.12), respectively.
As in the Drell-Yan case the "annihilation" and "Compton" differential
cross sections in (5.8.19) and (5 .8.22), respectively, are responsible for the
transverse momentum of the produced W± bosons. Everything stated in Sec-
tions 5.6 and 5.7 about the transverse momentum of large mass muon pairs
is also true for the production of W± and ZO bosons in hadron- hadron col-
lisions. For example, at order 0$ , the transverse momentum spectrum of ~V±
mesons produ ced a t psc udora pidity, '!}, is given by
dU w±
- dd2 (S,Y,PT)
Y PT
= 11 ( )
x:;>;n
dX a
• XaXb
Xa - Xl
DIS 2
Pqijw±(xa,xb,Mw)
(1dU' ,A (s, ,t)
--
7r dt
' )
(5.8.40)
where the "annihilation" and "Compton" differential cross sections are given
in (5 .8.19) and (5.8.22), respectively, and
min Xl - rw
Xa = 1- ' (5.8.41 )
X2
and
(5.8.42)
(5.8.44)
where
(5.8.45)
with
Xa = JTW" eY ,
Xb = JTW" e- Y . (5.8.46)
C it 11,),1''' 1 MIlIOIII' ,II I 'II,,,,"j,II"
Probl'm:-
Xa =Vr eY ,
Xb =Vr e- Y ,
where
y = ~ log (~ ~ :~ , )
is the rapidity of the muon pair. Also verify that
du du du
-d
y
= XE-
d XE
= (xa + xb)-d
XL
'
q + ij -> ,*
5.2 . Show that the differential cross section for the "annihilation" subproccHH
+ g in Fig. 5.2 is given by
do-(,
dtS,t)
"= 1raa.e~~{j2+it2+2M2S}
s29 tit '
,*
5.3. Show that the differential cross section for the "Compton" subprocess q+g
+ q in Fig. 5.3 is given by
->
where M2 is the mass of til e virtual piloton, ,., a /ld till' PlLI'tO/I S :I.re ta.ke /l
to be massless .
5.4. Show that the differential cross section for the "annihila.tion" s ubprocess
q + if ---. ,E + g in Fig. 5.2 is given by
dit " 71'aa.e~ 8 { it i
di(s,t)= s2 9 t+~
2
+ 2(M2 + m g
)(M2 + m - 2 it - i)
, 9
itt
_M2M; C~ + ~2)}'
where M and mg are the masses of the virtual photon, ,., and gluon, re-
spectively, and the quarks are taken to be massless. Use (A.3.10) to verify
that in this case the maximum and minimum values of i are given by
1
1-2~ log(,8f2 1(1- f)2) ,
+ o
1 '
- r
dr,
Integrate this from fmin = 0 to f max = 1/(1 + ,fiJ)2 ~ 1- 2,fiJ and show
that
(O'MG(real»DY = 2a. { 2
371' 0'0 log (,8) + 3Iog(,8) + 71'-?} ,
where,8 = m~/M2.
/ 2 16 (;hn,pi,flr n Mil"" I' ,1', ,,01'1 11, 011
where
, 4(11-f
Pq-+ qg (r)=3
+ f2)
+'
and
5.8. Show that the differential cross section for the "Compton" subprocess q+g ->
1; + q in Fig. 5.3 is given by
dO- , i\ _ 7raa.e~ 1
, (s, t J - '2 3 '
{_f _
~
'
dt sst
2(M2 - m 2)(o9 + i - M2
+ g,
+ m 92)
o9t
1 diT~G) a. ,
( -0'0 d'r = -2 Pg-+qq(r)
DY 11'
where
ri .1I V(,II,or Iln'HIlI I "otlll' I,illil 21(1
and
where
5.11 . Show that in N spacetime dimensions the matrix element squared for the
"annihilation" subprocess q + ij -> IE
+ 9 is given by
IM(q + ij -> I~ + g)12 = 1611"2Q'~ED Q'~CD e~ ~ (1 + ~)
2( f2y2 - 2fy2 + y2 + f2 + 2f + 1) 2 }
{ (1 - f)2 (1 - y2) + (1 - y2) ( ,
1=11 -1
dy (1 - y2) f/2 1M1 2 = 1611"2Q'~ED Q'~CD e~
2
8(1 f)2 J2 (1+!){ 1+f 8 4(}
'9 + 2 r(1 + () (1- f)2 -; + 1 + ( .
Using (5.4.3) and (5.4.9) show that
( ~ dUbR)
(To df dy
=2 2Q'$ (M2(1 - f)2) f/2
311" f411"m1
r(1+~){I+f2 2 (I-f)}
---+~-.-:..
r(1 + () 1 - f ( 1+ ( .
2 17 (; 1. 1"1'1".1' /, Mil"" " ", ",, 10111< I, 1111
where
Pq--+qg(T)
A 4(1 + f2) +'
="3 1- f
and
5.13. Show that in N spacetime dimensions the matrix element squared for th e
"Compton" subprocess q + 9 - , ; + q is given by
( ~ dO-~R)
ITodr dy
= as'
271"
(M~(1-2iV) f/2
r471"m D
-ITo1 (do-
dr
bR )
--,- DY
= -P9.-qij
a.
271"
(')
r log (2/ 2) +a s ,9DRDY(,)
M mD r,
where
and
5.14. Substitute G~!:q(x, M2) and G~!:ij(x, M2) defined in (5.5.5) and (5.5.6),
respectively, into (5.5.3) and show that to order as (5.5.3) is the same as
(5.5.1) provided
DY/ ITo
IT tot = 1 + 0'. I qDY + ... ,
where
2a. (1
3 71" +T
'2) (log(1 -
l- T
'
f») +
that appears in the expression for t::..jf Y given in (5.5.29) results in the
behavior
where G~!:q(x, Q2) are defined in (5.5 .9) and Gp.-q(x, Q2) are the reference
distributions defined in deep inelastic scattering.
lh a.t ,t PP ('!1I 11 II I I. h.· " x plf 'lm io ll ro r 1'11:)1' IJ, i V('1I III (:').,') . ~!) ) 1'( 'I·" dl,/'l III I. h,·
be h a.v ior
G' /) )' ( ' ) ~ )
p- q X, (~ ----+ 1 + -
2
a .(Q
--
) Iog (1 - x )
Gp--+ q(x , Q2 ) x--+l 71" '
where G~!:q(x, Q2) are defined in (5 .5.9) and Gp--+q(x, Q 2) a re the reference
distributions defined in deep inelastic scattering.
5. 17. Substitute 8~!:q(x, M2) and 8~!:q(x, M2) defined in (5.5 .33 ) an d (5.5. :14),
respectively, into (5.5.32) and show that to order a. (5.5.31) is th e sa.lIl<· a"
(5.5.1) provided
DY/ lTo
lT tot = 1 + a. I qDY + ... ,
where
a.lqDY = a. (871"
9 - 7) •
371"
Show that with this definition of "DreIl-Yan" parton distribution fun ct io liH
that
r 1
io (~DY ~DY
Gp--+q'(x, M 2 ) - Gp--+q,(x, 2 )
M) dx =
11 (Gp--+q,(X, M2) - Gp_q,(X, M2)) dx,
so that (to order a.) the net number of quarks in the proton is the same for
the "Drell-Yan" distributions as for the reference distributions.
TI .H V('ri,(lr 11 0110 11 1', ,,oII1 I IiOll 2 2 (1
wh ere
x[ - r
x min _
a 1 _ Xl '
-
XaX2 - r
Xa - Xl
S = XaXb8,
i = -Xa 8X2 + M2,
U = -Xb8XI + M2,
1 2 .1
Xl = 2"(XT +4r)2eY ,
1 2 .1
X2 = 2"(XT + 4r)'e- Y ,
XT = 2PT /..;s.
5.19. Show that if one neglects the M2 dependence of the structure functions so
that
Gp...... q, (x, M2) =Gp ...... q,(x),
Gp ...... g(x, M2) =Gp ...... g(x),
then the contributions from the "annihilation" and "Compton" subprocess
jr\Fig. 5.2 and Fig. 5.3, respectively, can be written in the form
2 2 dUA 2
M PT d d d 2 (8,M ,y,PT) =cx.FA(r,y,XT),
r y PT
2 2 due 2
M PT d d d 2 (8,M ,y,PT) =cx.Fe(r,y,XT),
r y PT
where FA( r, y, XT) and Fe( r, y, XT) are functions of the dimensionless vari-
ables r, y, and XT. Determine FA(r,y,xT) and Fe(r,y,xT).
5.20. Show that the contribution from the "annihilation" subprocess q + if -+
jJ+ jJ- + 9 to the muon pair invariant cross section is given by
•21 MII IIII I' I I',"d ll dltl ll
where tit · joi nt 'Iii 1','" lnhi lity is gi ve n ill (5,5 ,'15) fw d
T = M~ / Ii,
xT =2PT / VS,
, 1
t =2'S(T - XaXb + V),
'11 =.!:.S(T - XaXb - V),
2
V2 =(T - XaXb)2 - XfXaXb = (XaXb - T+)(XaXb - L),
_ )2
T± =41 ( XT ± XT ,
2 +4 T.
-2 =xT
xT
Verify that the "annihilation" differential cross section is given by
dO-A _ 8a2a,e~ {(i - M2)2 + ('11 _ M2)2}
dM 2di - 27 M 2s2 iil .
Show that if one neglects the M2 dependence of the parton distribution
functions so that
then
where
l+ ____
T2 X2)
T_
( x~xl 2XaXb .
Show that in the limit XT -> 0 that
5.21. Show that the transverse momentum spectrum ill (5.7.14) vanishes at PT 0 =
and behaves like llpf at large PT. Also show that the distribution peaks
.H VI ' I to, 11"1" '" 1' ,"0111 1 U"" 2:l2
and Xw is the square of the sine of the Weinberg angle, Xw = sin 2 (Ow) and
G p is the Fermi constant. Show that in the naive parton model
O'(P+P- W
±
+X;s) = -271"g?v
3-
11 dX a
- PqqW±(Xa,rw/Xb),
s TW Xa
where rw = Mw / s and the joint probabilities PqqW± are given in (5.8.14)
and (5.8.15).
5.23. Show that in the naive parton model the total cross section for producing ZO
bosons in proton-proton collisions is given by
5.24. Verify that for massless partons the differential cross section for the "annihi-
lation" subprocess q + if - W± + 9 is given by
lI,iv C II by
+ vIn27rGr;;, 111 11
TW
-dXa
Xa TW/Xa
dXb DIS 2 DY "
-Pqqw±(xa,xb,Mw)2a.A.fq
Xb
(rw)
+ vIn27rGr;;, - 111 11
dX-a
Xa TW /Xa
TW
dXb
- DIS (
- PqgW
Xb
2 )
± x a , Xb, Mw a.A.fgDY( rw
" )
1dx
+ vIn27rGr;;,11 - -a11 dXb
- - PqgW± a 2) a.A.fgDY( rw,
DIS ( X , Xb, lvfW " )
TW Xa TW/Xa Xb
where fw =
TW /(XaXb), rw =
Mf.., / s, and where A.ffY and A.ffY arc
the same functions that appear in the Drell-Yan case, (5.5.11) and (5.5.12) ,
respectively, and the joint probabilities qqW±, qg W±, and qg W± are given
in terms of the deep inelastic scattering reference distributions according to
(5.8.24), (5.8.25), and (5.8.27)-(5.8.30), and where
Further Reading
G. Altarelli, "Partons in Quantum Chromodynamics," Pllysics Reports
81, 1 (1982) .
R. K. Ellis, "An Introduction to the QCD Parton Model", Fermil ab
preprint, FNAL-CONF-88/60-T, 1988.
References
1. S.D. Drell and T.M. Van, Phys. Rev. Lett. 25, 316 (1970) .
2. J. Kubar-Andre and F. Paige, Phys. Rev. D19, 221 (1979).
3. G . Altarelli, R.K. Ellis, and G. Martinelli, Nucl. Phys. B143, 521 (1978).
Erratum B146, 544 (1978).
4. G. Parisi, Pilys. Lett. gOD, 295 (1980).
5. G. Curci and M. Greco, f hys. LeU. 92D, 175 (1980).
6. H.D. Politzer, Nucl. Phys. B129, 301 (1977) .
7. C.T. Sachrada, Phys. Lett. 73B, 185 (1978) .
8. F. Khalafi and W.J . Stirling, Z. Phys. C18, 315 (1983).
9. D. Antreasyan et ai., Phys. Rev. Lett. 48,302 (1982).
10. G . AltareHi, G. Parisi, and R. Petronzio, Phys. Lett. 76B , 351 (1978).
ibid 76B, 356 (1978).
11. C. Quigg, Rev. Mod. Phys. 49, 297 (1977).
/ ----------~---------------------
Renormalization and
the Running Coupling
Constant in QCD
225
In this chapter we will compute the effective co uplill g ill Q 'D , 0'. (Q 2), us-
ing the same dimensional regularization techniques developed in the previolls
chapters. An infinite set of ultraviolet divergent terms will be chosen and
used to define an effective coupling. Normally renormalization is presented as
a procedure in which one introduces counter terms to the Lagrangian thereby
removing the ultraviolet divergences. There is a very nice discussion of the
field theoretic approach to renormalization in the book by Pierre Ramond .1 I
will be less formal here and simply consider renormalization as a procedure in
which one expresses experimental observables in terms of other experimental
observables. Once this is done the theory should be finite and independent of
any cut-offs used during the intermediate stages.
First we consider the quark loop corrections to the gluon propagator shown
in Fig. 6.1,
q - 2
IIlJv;ab(q) - -gNn, tr(TaTb)
J dNk tr['lJ¢Jv(¥ - ()]
(21r)N (k _ q)2k2 ' (6 .1.1)
/ k-q
a,k
13, k-q
k
" ,.->--... ....
fL,a / " lI,b
lOOOYOOO. (_I) ,OO~OOOJ
q \ q
, /
I
with
K = k-zq, (6.1.9)
C = Q2 z (1- z), (6 .1. 10)
and
Q2 = _q2 > O. (6.1.11)
We now shift the integration from k - K + zq arriving at
9 2
II"II;ob(q) = -9N C,6ab
JJdz
dNK N(k - K + zq)
(211')N [K2 - C)2 , (6.1.12)
228 C la l\,pt,'r 6 1t.'IIM .. , ,iiI'. ~till ll
(6.1.14)
has been made. The integration over f{ is performed using (C.2.6) with the
result
gN
2 J dNf{ 1 _ -ig; ~
(271")N [f{2 _ Cp - 1671"2 £ h
,[., _, ]f/2
x(l x) , (6.1.15)
where
(6.1.16)
and
N=4+c (6.1.17)
The mass mD comes from replacing the N-dimensional coupling gN with the
dimensionless coupling g. defined in (2.8.25). Similarly,
2
gN
J dNf{ f{2 _ -ig; ~
(271")N [f{2 _ C)2 - 1671"2 £
(4 ++ £)
2 £ fK Q
2 [,
x(1
_
x)
]1+f/2
. (6.1.18)
So the complete integration over f{ is given by
q ( ) _ -ig; ( ) 16
IIpv;ab q -1671"2 C,6 ab ~ fK
1
(gp/lQ2+ qpq/l) fo [X(I_X)]1+ f /2 dx. (6.1.19)
(6.1.20)
where
(6.1.21)
wit r
and IK and Ix are given by (6.1.16) and (6 .1.21), respectively. It is impo rtallt
that II~v;ab is proportional to 6ab as this is the same color factor appear-
ing in the Born term expression for the gluon propagator. Also th e facto r
(g/JvQ2 + q/Jqv) insures that II~v;ab conserves current . Namely,
q/J II~v;ab = 0, (6. 1.:.ltl)
I and
Ir = ~CJ c~v) .
1e
(6 .J. 2(j)
Here I have labeled the l/f singularity by the subscript UV to rem ind Wi
that it is an ultraviolet divergence. It arises from the integrations over J( in
(6.1.15) and (6.1.18).
The gluon loop in Fig. 6.1 gives a contribution of the form
(6.1.32)
we arrive at
(6.1.33)
with
+_ [2(2+f)2 (1l-2()] (6.1.34)
!g - -CA 4«1 + f)(3 + f) + 4«(1 + () ,
and
r 9
= CA
4f(1 + f)
(6.1.35)
o
Clearly something is wrong since II~v;ab does not conserve current. It con-
t ains a term proportional to (gl-'vQ2 - ql-'qv) as well as a current conserving
term proportional to (gl-'vQ2 + ql-'qv) . This is the first time we have had to
deal with this situation, but because of our choice to use a covariant gauge
(rather than, for example, an axial gauge) we have introduced spurious gluon
polarization states. These spurious states must be removed by computing a
"ghost" contribution using the ghost- ghost-gluon couplings in Appendix C.
Alternatively, we could have computed with an axial or "ghost free" gauge
but it is usually much easier to use the simple Feynman gauge and add in the
ghost contribution. The ghost loop contribution in Fig. 6.1 has the form of
(6.1.30),
o 2
III-'v;ab(q) = -gNCAbab
JJ dx
dNf{
(2;r)N
N(k --+ f{ + xq)
[f{2 _ CF ' (6 .1.36)
Remember to insert a ( -1) for the ghost loop. Integrating over f{ and x gives
. 2
II~v;ab(q) = - ll:;2bab (91-'vQ2 + ql-'qv) ![(!x!d
(6.1.38)
wit h
+_ CA (6 .1.:10)
10 - 4f(3+f)'
_ -CA
10 = 4f(1 + f) . (6 .1.'10)
The ghost contribution also has both a current conserving and a clln lit
non-conserving term and the non-conserving term exactly cancels the 11 0 11
conserving term in (6.1.35). The final result for the true gluon contribllti II iH
the sum
(G . I .'II )
where
(6.1.'12)
and
I 9Poie -- -Q,C
3 A (~) . (6.1.43)
fUV
(6.1.44)
where
+ f) - C A (10 + 3f) _ (1C - Q,C ) ~ + ... , (
f 3-- f q+ I g-- CJ 4(2f(1+f)(3+f) 3 J 3 A f 6.1.45
)
and
(6 .1.46)
... ( -i8~b2 gjJv ) Z3 = (-i8~~ gjJv ) + ( -i8:c2 gwx ) lla,6;cd(Q) ( -i8~b2 g,6v )
= (-iD aq g/IV) (1 +
b
2
g;? 1f(lxh) .
16n~
(6.1.47)
2:1~ C hllpt.(· .. Ii 11 ""0 '·'" db"Ii",,,,
so that to order g;
Z2 =1+ - g;- h (_p2
--"
)</2 r(1-~)r2(I+~)
---'"---::!'-c.'-----'----"<.L (6.2 .5)
1671"2 471"mb f(1 + f)
,i~::) ~ (T,T,) {; + ~,
where
h = (T,T,) (6.2.6)
and
Hole = (Tb Tb) (~)
(Uy
. ) (6 .2.7)
k
--=---~---:- (- ig, >'II T~l)
fL,a
(6.3.5)
so that
(6.3.6)
where C A is defined in (6 .1.29). Except for the color factor this is the same
calculation we did in Chapter 2. Nevertheless, it is instructive to repeat the
calculation in a slightly different manner. This time let's try the Feynman
parameterization in (C.2.2). Namely,
(6.3.8)
where
(6.3.9)
and
(6.3 .10)
/
((>.:I . I I )
wit.h
(6.3. 12)
and q2 is the virtual gluon 4-momentum squared which we take as spacelik
The shifted numerator has the form
N I1 (k -+ J< + P1Z2 - P2Z3) =A1 "A/I1P1 + A2 P2/11P'.l + A3 P1/11P2
+ A4 P2//lP1 + As J(2 1 /l, (6 .3. 1;1)
where I have used
t,-
~//l/~
t,-
= }_\er~'T.(f31er/I11f3 -+
[(2
filerl/ller = (2 -N N) Au'2 l/l' (6.3. 1II)
The Dirac equation for massless quarks guarantees that
and
P2V(P2, S2) = 0, (6 .3.16)
so that the surviving terms in the shifted numerator are
N I1 (k -+ J< + P1Z2 - P2Z3) = (A4Q2 + AsJ( 2h/l' (6 .3 .17)
where
(6 .3.18)
(6 .3.19)
In arriving at (6.3.17) I have used the fact that in the numerator of the integral
over J<,
P2//l'A -+ -2P1 'P21/l = Q21/l' (6.3 .20)
For the A4 term the integration over J( is performed using
2
gN
J dNJ( 1 -ig; 1 "'-1
(211')N [J<2 _ C)3 = 1611'2Q2 2'(=2=3) 2 fl(, (6.3.21)
with fl( given by (6.1.16) . There are no ultraviolet (UV) divergences in the
A4 term, only infrared divergences that arise from the integrals over Zl and
Z2. For example,
Jot Jor -
1 Z2
dZ2 dZ 1 Zl(Z2Z3)~-1
230 C it l "till' Ii UI ' ll(jJ' lIl .th,II.1. O il
(6.3.22)
where Ix is given in (6.1.21) and wh ere I removed the nested integrals by the
change of variable Z1 = v(l - Z2~ J.lle A4 term gives
(6.3.23)
with
(6.3.24)
The final form of 1~1) comes from expanding in powers of f and dropping
terms that vanish as f ---- O. Furthermore, the 1/f2 and 1/f singularities are
labeled with an IR to remind us that they are infrared singularities.
The A5 term in (6.3.17) produces an ultraviolet singularity. The integral
over J{ is given by
d Nr.-' -za• .2• ! (4)
J
2 _ _1\_ r.--2
1\ _ f/2_
aN (211")N [J{2 _ C]3 - 1611"22(Z2 Z3) + fVV 11K, (6.3.25)
dZ 1
J dNJ{ A5J{2 _ -ig; (1)
(211")N [J{2 _ C]3 - 1611"2 IxlK 15 , (6.3.26)
where
1~1) = -(2 + f) ______2_ + 1 + .... (6.3 .27)
f(l + f) fVV
r;(r 1) _
I-' -
a; ( 1C
1611"2 -2" A + TbTb ) "II-' IxlK Ivv
(1) (1»)
+ IIR , (6.3.28)
where
(1) _ _ _2_
1vv - fUV
, (6.3.29)
and
8
(1) _
fIR - - -2- + -8 - 8. (6.3.30)
fIR fIR
r(2) _ _
I-' -
g2
N
(-!..CA)
2
J(211")N
dNJ{ F a /I /3(Pl - k, -Q,P2 + k) rah/3
(P1 - k)2(p2 + k)2k2 '
(6331)
..
wh er till' F e Ylllllll1l !-(II.II I1;" hils b"( ' 11 II se d and 1.1 ... I.l'Ipl .. Jl, III OIl VI ' l't('X f ltdo l'
Fa /J {3 (PI, P2, TI:1) i · giV(' I' ill (fl. l .28) . Th e co lo r factor :U'iS('N 1'1'0111
I
= -2 CATa ' (6 .:U· )
Proceeding as before we arrive at
:;:;(2)
r I' (i) 2 Jor
= - g;' -2 CA
1
dZ 2 Jor
1
-
Z2
dZ 1
NJl.(k + PIZ2 -
J
where the shifted numerator in this case has the same form
dN/(
(271')N
-+ /(
[/(2 - Cj3
P2 Z3)
,
(6.3.3:l )
(6.3.34)
with
A4 = 2(z~ - Z2 - Zl + ZlZ2 + 1), (6 .3.35)
A __ 4(3 + f)
(6 .3.36)
s- (4+f)'
As was the case for ~l), the A4 term produces no ultraviolet divergences. [n
particular,
2
29 N
r
Jo
1
dZ 2 J
r
o
1
-
Z2
dZ 1
J dN /( A4Q2 _ -ig; (2)
(271')N [/(2 _ Cj3 - 1671'2 fdK 14 , (6.3 .37)
with
12) _ 2(3f + 8) -+ ~ _ 9 + ... (6.3.38)
4 - f(1 + f)(2 + f) fIR '
dZ 2
t-
Jo
z
,
dZ 1
J d N /( AS/(2 _ -ig;
(271')N [/(2 _ C']3 - 1671'2 I;eIK 15
(2)
, (6.3.39)
where
l2) = -4(3 + f) -+ -6 + 7 + ' .. (6.3.40)
s 1'(1 + 1')(2 + f) fUV
- (2)
The comp lete res ult for 1',. b co mes
-(2) _
f Il -
g; (1
1671"2 2"CA "ill
) .
If( Ix (luv
2) + IIR(2» ) , (6.3.4J)
with
(2) _ _ _6_
1uv - , (6 .3.42)
fUV
and
(6.3.43)
Again I have arbitrarily absorbed all the constant terms in 1~2) and 1~2) into
(2)
1IR '
Combining (6.3.28) and (6 .3.41) we see that the complete order 0; vertex
correction is given by
r = r<1) + r<2)
2
= 1:~2 IK Ix { (-~CA + Tb Tb) Ig~ + ~CA Ig~}
2
+ 1:~2 IKlx {(-!CA + TbTb) II(~) + ~CA li~?}. (6.3.44)
Only the ultraviolet divergences are included into Zl. One could, however,
also choose to include some of the constant terms.
r
The quantity in (6.3.44) contains both IR and UV poles. It is not obvious
that one can handle simultaneously both IR and UV poles with dimensional
regularization. Increasing the number of space-time dimensions, N = 4 + 10,
above four (10 > 0) improves the convergence of, for example, (6 .3.2) at low
/ 2:111
Figure 6.3 Shows the bare qqg vertex together with the leading oro ; r vertex cor-
rections, Zb the leading order quark self-energy corrections, Z2, a.nd the leadiug
order corrections to the gluon propagator, Z3.
We have computed, to order g~, all three corrections to the "bare" coupling
210 <J hltpter () lI oll(JI'llI al !loll I.Ifl lI
90 shown in Fig. 6.3. From (G.1.48), (6.2.5), a nd (6 .3.t1 7) W ' Il('e tim!'
95 (_q2) </2
Zl =1+ 16 1r-') fr -1rmb
4" h, (6.4.1)
96 ( _])3 ) </2
(6.4 .2)
Z2 = 1 + 161r Z fr 41rm1 12,
96 ( _qZ ) </2
Z3 = 1 + 161r1rH1b
2 fr -4" fa, (6.4.3)
where
(6.4.4)
and
- f iPo le + h·t,
f ,· - (6.4.5)
where
2
Hole = (GA + TbTb) (6.4.6)
(UV'
or
Z2VZ3 (6.'1 . 10)
fl. = Zl 90·
From (6.4 .1), (6.4 .2) and (6.4 .3) we see that, to order 96,
Z2yfZa
Zl
g5
=1+ 16 11'2 fr 41Tmb
(Q 2
)£/2 (h+'2/a-I1,
1 )
(6.4 .11 )
fr ( - Q2- ) </2
= 1 + -i [ log (Q2)
--.,- + IE ] +" ', (6.4.16)
41Tmb 2 41Tmb
and
where
(30 = !]CA - ~CJ = 11- ~nJ' (6.4.1 8)
since C J = n J 12 and for three colors C A = 3 and
h = 2h1 - 2h2 - h3, (6.4.19)
are the constant terms of our choice. Equation (6.4.15) now becomes
{~(3o+h+"-}
=- ~; {(30 D+ log(Q2/ m b) + IE -log(41T)] + h + .. -}, (6.4.20)
21\2 (:I 'II,pt"r H 1l( ~ II(III1II1I 1.I\, I, IOIl
where terms that vallish as (; -> 0 have been dropped . '1'1,,;1'" ('() I'I'I' cti liS ro rm
a geometric series of the form
a_(Q2) = ao (1 + a oBQcD(Q2) + a oBQcD(Q2)a oBQCD(Q2) + ...)
ao
2 ' (6.4 .21)
1 - aOBQcD ( Q )
so that
1 1 2
2) = - - BQCD(Q ). (6.4.22)
a. (Q ao
The function BQCD(Q2) diverges as f -+ 0, however, the bare coupling ao is
not an experimental observable. The idea behind renormalization is to express
experimental observables in terms of other experimental observables. 'Ve must
decide on what we mean by the "experimental QCD coupling," a_, and then
express the effective coupling, a. (Q2), in terms of it. In QED the effective
coupling is expressed in terms of the fine structure constant a :::::: 1/137. The
fine structure constant, on the other hand, is defined as the value of the
effective coupling in the Thompson limit (Q2 -> 0).
In QCD BQCD (Q2) diverges as Q2 ---+ 0 so we cannot define aeff( Q2) in
=
terms of its value at Q2 O. Instead we choose some value of Q2, say Q2 J.l2, =
and define the "experimental QCD coupling" to be
a. == a_(J.l2). (6.4.23)
We then express all observables in terms of a.(J.l2). From (6.4.22) we see that
when we express the effective coupling in terms of a.(J.l2) we get
1 1 ( 2 2)
a_(Q2) = a_(J.l2) - BQCD(Q ) - BQCD(J.l ) , (6.4.24)
T hi s ITI a nli tl lld, I •• U/J ) 11. 1101. It HCiHI.rat c fUII ·l. io ll o f (\ . ( I L ~ ) H.ll d I I.'}. , h il I. ill
st ad can be wril,l,(, 11 11M II rllll r l.i n o f a s in g Ie p aran l<'i.<' r , /I., tl l<I.(, is ill <i <,pend(, 1I1.
o f J.l2 a nd given by
~ 4~ 2
log(k) = - fJoo.$
a (?)
J.l~
+ log(J.l ). (6.1 .29)
where the mass scale A is a parameter that must be determined exp rilll(' l1
tally. It is the one parameter of perturbative QCD .
The choice of constant terms, h, to include in BQCD(Q'2) in (6.tl.:W) iff
arbitrary and a matter of convention. The choice does not effect th e Q2 de p" 11
dence of the effective coupling 0.$ (Q2), but it affects the value of the QC I) )111.
rameter A as determined experimentally. In the "minimal subtraction" scl H' IIIt '
(MS) no constant terms are absorbed into the effective coupling. On th e oLlII' 1'
hand, in the "modified minimal subtraction" scheme (MS) one absorbs tit
constant terms
IE -log(4~), (6 .4 .3] )
t hat arise from dimensional regularization into the effective coupling. In prac-
tice, at order o.~, this corresponds to making the replacement
(6.4.32 )
where 0.0 is the bare QCD coupling and 0$(Q2) is the running coupling in
(6.4.30).
---+ -
/30
- (6 .5.2)
£ -+0 4~ '
where
r = log(Q21/.t 2 ). (6 .5.3)
Differentiating (6.4.13) with respect to r gives
da.(Q2) 2 dBQCD(Q2)
dr = ao dr . (6.5.4)
(6 .5.6)
yielding
1 _ 1/30.22
2) - - ( 2) + -4 log(Q IJ.l ), (6.5.7)
a. (Q a. J.l 1T
or
a.(Q2) = as (J.l2) , (6.5.8)
1 + a.(J.l2)~ log(Q2 1J.l2)
which is the same as (6.4.26). Equation (6.5.5) is traditionally written in the
form
da.(r) _ /3( ()) (6.5.9)
dr - a. r ,
with the condition
(6.5.10)
The "beta function" /3(a.(r)) governs how the effective coupling a.(r) de-
pends on r. It can be expanded in a power series in a.(r) as follows:
-/3 (a.(r)) = bo a. 2(r) + b1 a. 3(r) + b2 a.,4(r) +" ', (6.5.11)
where the first term is given in (6.5.5),
bo = ::/
(6.5.12)
= bo n. ~ (Q :.1 )( 1+ bb as (2)
cia. (Q 'l ) I b2 'J 2 )
- ciT Q + b a.-(Q )+ ... , (6 .5. l fl)
o o
and rather th an truncating the series (b n = 0 n 2 2) we set
bn = bo ( bo
bi )n ' (6.5.16)
_ da~~Q2) = boas2(Q2) f
n=O
(:Ia s(Q2»)n
0
bo a. 2 (Q2)
(O ,f). 17)
- 1 - ~as(Q2)'
To order a. 3 (Q2) equation (6.5 .17) and (6.5.15) are equivalent, but th ' la t., ·..
expression provides a very nice "definition" of the QeD effective couplillg.
Solving for a.(Q2) we get
a
(1 b 1)
l
.(Q2) i
- - - - - - da -T (6 .5.18)
a.(1'2) boa. 2 b5 a. •- ,
or
1 1 2) + bb2i log ( a. Q-?)/ a.(p.-)
? ) = 10g(Q-?/W),
?
(6 .5. 19)
bo a. (p. 0
and we drop terms higher than order ala. In this case (6.5.23) becomes
471" [f3110gI0g(Q2/A2)]
(6.5.25)
= f3o log(Q2/A2) 1- 136 log(Q2/A2) ,
where 130 and 131 are given by (6.4.18) and (6.5.14), respectively.
The mass scale, A, is the one parameter ofperturbative QCD, however, its
definition is not unique. The precise definition of A is a matter of convention.
For example, the A appearing in the leading order expression for a. (Q2) in
(6.5.24) is not the same as the A in the higher order expression in (6.5.23).
Formally A is defined by
Since QCD is a theory of massless qual-ks and gluons we were forced to intro-
duce the scale J.l. The QCD coupling was defined to be the effective coupling
at the scale J.l,
(6.6.1)
Since physical observables cannot depend on our choice of J.l, a change in J.l
must be compensated by a charge in the coupling a •. Consider the case of a
dimensionless quantity such as Re+e- in (2.1.16). In general the observable
I
(H .6 .'L )
a nd the couplill g fI'. ill (6.6.1) but not 011 the renormalization point II ,
R(r,a.,J.I) = R(r, a.) . (6 .6.3)
The independence of Ron J.I strongly constrains its functiona.l form and implies
that the total derivative with respect to J.I is zero. Namely,
J.I
2 dR(r, a.) _
d 2
J.I
-
('1~
uJ.l
2 oa.~) R(
J.I J'l 2 + J.I J'l" J'l
uJ.l - ua.
) -
r, a. -
°, (6 .6 .tI)
where the first term accounts for the J.I dependence through r and the sec lid
term accounts for the dependence 011 a •. If we define the "beta fUll cti oll "
according to
., 00. 00. (
(l (a.) == J.I - OJ.l'1 = 010g(J.l2), O.U.fI)
then (6.6.4) becomes
R = R e+e- = e+e- /
(1tot (10 = 1 + -7r1 o. + .. .. (6 .6.12)
(6 .6. 15)
where A;:s are the moments of Pq_qg(z) given in (4.7.6) and M~O) are the
moments of the un calculable non-perturbative quark distributions G~1(x).
We know that the left-hand side of (6.6 .15) cannot depend on mass m since,
in fact, we believe that the result is finite in the limit m ~ O. This means
that M~O) must have a perturbation series of the form
(6.6.17)
The mass singularity log(m 2 ) has been absorbed into the non- pert.urbative
unknown distributions .M~O), but at the expense of introducing an apparent
dependence on the renormalization point Ji.. However, this is a great advan-
tage since we can use a renormalization group equation to sum the leading
logarithms. In this case the RCE is more complicated since a change in o. is
not by itself sufficient to compensate for a variation of J-L.
Formally, we can arrive at the more general RC E by considering the rela-
tionship between an unrenormalized observable, FUN, as computed from the
appropriate Feynman diagrams using bare quantities and a cut-off mass m
/
• 'III
(6.6. 18)
where Z is in general made up of several renormaliza tion factors as in (6.4.10) .
The left-hand side of (6 .6. 18) is clearly independent of p. so that
2 0 2 oa. 0 p.2 OZ)
( P. op.2 + P. op.2 oa. +Z op.2 Z F(T,a.) = 0, (6 .6. 19)
where T is given in (6.5.3) . The additional term takes care of the p. depend 11 <:('
with (6.6.8), (6.6.9), and (6.6.10) still holding. The T dependence of F is again
a consequence of the running coupling constant a.( T) but the form is morc
complicated and one needs to know I (a.(T)) as well as a.(T). The quantiti('H
F (0, a.(T)) , I (a.(T)), and j3(a.(T)) can each be expanded in a pertul'b a ti II
series as follows:
I(a.(T))=ao a.(T)+al a. 2 (T)+"" (6 .6.i:l)
F (0 , a.(T)) = 1 + Bl a.(T) + " ', (6 .6.2/1)
- j3(a.(T)) = bo a. 2(T) + b1 a. 3 (T) +"', (C).(j . 2!) )
where the first two series depend on the process being considered.
As an example, consider again the moments of the non-singlet quark diH
tribution
(6 .6.2G)
r·(T) I (a) }
C;:s(T,a.) = C;:s (O,a.(T)) exp { lex, ;(a) da , (6 .6.27)
where In( a. (r)) give n by (6.6.23) wit.h
ao = - 8~ I~n) , (6 .6 .28)
_ 1 (n)
a1 - - 3271'2 11 , (6 .6.29)
and
NS NS a.(r) ,
C n (0, oAr)) = 1 + En 4;-"" . . .. (6.6.30)
From (6.6.26) and (6.6 .27) we see that the change in M/;S(Q2) with respect
to a.(r) is given by
S
dM~S(Q2) _ ('n(a.(r)) 1 dC: (0,a 8 (r))) AI Ns (Q2)
da.(r) - ,8(a.(r)) + CfjS(O, a.(r)) da.(r) n .
(6.6.31)
Inserting the leading order expressions for In (a.( r)), and
(6 .6.31) results in
C:
s (0, a.( r)) into
(6.6.32)
or
(6.6.33)
where
2
'" = ,80 log{a.(Q5)/a.(Q2)}. (6.6.34)
[a.(Q2)/a.(Q5)]d =
n
[1- :; a.(Q2)log(Q2/Q6)f
n
Ii th a t th Il: I'; 11 11 ,1 III .·n,·d. ,JIIIIIIII -J t.h illfinit - I> 'ri. ·/< of ten I III [cv. log(cj') I"
that we expli ·it.l y HIIIIIIII< ' d ill C haptcr 3 for the fraglll clll.;t1,ioll fUli ct ioWi 1I,IId
in C hapter 4 C r th( ' parton distributions.
Including the lIext order in (6.6.31) results in
~ (Q2)dM: + B NS ) a s(Q2»)
S
"'.
(Q2)
da.(Q2)
= ('bn)
2{30
+ (h
n n 471"
M NS (Q2)
n ,
(6 6 3(1)
.. , "
where
"V(n) (3 "V(n)
/1 1/0
hn = 2{30 - 2{36 . (G .G.'10)
where
"BnNS =
L.l. -
BNS,B
n
_ BNS,A
n· (6.6.'15)
Since dn and h n are independent of the process, only the B;: S correctiolls
remain and are of order a. (Q2) when comparing one process with anot h r.
The coefficients 6.B;:s are simply related to the "little f" functions we ha.ve
encountered previo usly. For in st,a ll ce ,
1 dy
G~s(x, Q2) =
1
x
-G~s(x, Q2) {6(1- z ) + a .(Q2)6.f( z )} ,
Y
(6 .6.46)
The coefficients liBn are simply the moments of the "little 1" functions. In
leading order both Mf:S,A(Q2) and Mf:S,B(Q'2) satisfy (6.6.36).
Problems
6.3. Show that the quark loop corrections to the gluon propagator shown in
Fig. 6.1 are given by
. 2
II~,,;ab(q) = -1~;2 Dab (g/1"Q'2 + q/1q,,) fxfJ( f q ,
where
and
where
where
_ r(l- ~)r2(1 + ~) (~)(/2
IKlx- r(l+t) 41r1111 '
and
- CA [(5 + 2t)(2 + t) +--
2 - t] -1
I1 - 2(1+t)(3+t) l+t t'
_ C [ -(2 + t) 7] ~
h - A 2(1 + t)(3 + t) + 2(1 + t) t'
o
II~v;ab ()
q ig; Dab (I1 q~qv + h
= 161r gw Q2) IK l .x,
2
where
_ r(l- ~)r2(1 + ~) (~) (/2
IK Ix - r(l + t) 41rm1 '
and
6.9. Using the Dirac equation for massless par tons show that
and that
U(Pl, 5d1O'(Pl - ~hJ(P2 + nV(P2, 52) =
U(Pl,5d [-2P21/-1Pl - 211/-1Pl
+2P21/-1~ + (N - 2)h/-l~] V(P2, 52) .
Use this to show that the integral in (6.3.2),
reduces to
J (271")N (PI - k)2(p2 + k)3k2 '
where
J(= k - PIZ2 + P2Z1,
q = PI + P2,
Q2 = _q2,
C = Q2 Z2Za ,
1 = ZI + Z2 + Za,
A4= -(2 + f)Z2Za - 2z 1,
A __ (2+f)2
5- (4+f)'
and N = 4+f.
6.10. Show that
2
[1 dZ 2 [1-Z dZ 1 (z2za)f-1 = ; Ix,
Jo Jo f
where 1 = ZI + Z2 + Za and
/ I II ' I'h" I! " '""'II' dl... d, " " (1;'''lp 1':<111 ,.1,101,
6. 11. P.... fo l'lII t.11t' illl"'fI, ' 101 II VI' I' k ill (6.:1.2) itlld v{: l'ii'y 1.I, It!,
-( J) _
f,..
!I:
- I G11"~
(_ 1' )
2CA+Tbn I~'
[(I - ~ )r ~ ( 1 + ~)
f(1+ f )
( ~)
411"m
</2 ( _ _
D
2 _~+ ~-8+ .. .),
fUV fiR fIR
reduces to
J (211")N (Pl - k)2(P2 + k)2k2 '
where
J( = k - PlZ2 + P2Zl,
q = Pl + P2,
Q2 = _q2,
C = Q2 Z2Z3 ,
1 = Zl + Z2 + Z3,
A4 = 2(Z2 + Z3 - Z2Z3),
A __ 4(3 + f)
5- (4+f)'
and N = 4+c
6.13. Perform the integral over k in (6.3.31) and verify that
(-.!L)
b
</2 ( _ _ + ~ _
411" 1n
6
fUV fIR
2 + ...) ,
+nn (-++~-8+
fIR fIR
...)} .
6.15. Using
and
) 2
11pole = ( CA+nn-
fUV
,
Ifo1e =(nn)_2_,
fUV
Pole _(3C
f 3 - - J - A -2
-5C) -,
4 3 fUV
i,,; ind ' P(!II<I"III. oj 1I, .. jI·'";IIlI liliza.tio n poillt It rell LIly cI ,nic<' or !,I'e IIpp.·,.
l.im it A . UHill K 110111 ddillil.lOl1 o f A, show thatth leadi ll g ol'd'" 'XIH 'ssioll for
the QeD o llplill K, n.(Q~), in (6.4.30) corresponds to t.h e choice A =
6.18. Show that the general solution of the RGE equation
6.19. Show that the general solution of the more general RGE equation
{l
aB
(T) "Y(a) }
F(r, a.) = F(O, a.(r)) exp a, f3(a/a.
Using
2()
'Yn (a. ( r ) ) = - 811"1 'Yo(n) 1 (n)
a.(r) - 3211"2 'Y1 a. r +"',
where
(n)
dn=~'
2fio
and that to order 0'. 2 the solution is
6.21. Show that the B;:S terms in the expansion of the "coefficient function" in
(6.6.30) correspond to corrections of order 0'.2 in the Q2 evolution of the non-
singlet moments, but correspond to corrections of order a. when comparing
quark or gluon distributions defined in two different processes as in (6.6.44).
Further Reading
G. Altarelli, "Par tons in Quantum Chromodynamics," Pllysics Reports
81,1 (1982).
J .D. Bjorken and S.D. Drell, Relativistic Quantum Mechanics, McGraw-
Hill, New York, 1965.
A.J. Buras, Rev. Mod. PllYS. 52, 199 (1980) .
Chris Quigg, Gauge Theories of the Strong, lVeak, and Electromagnetic
Interactions, Frontiers in Physics, The Benjamin-Cummings Publishing Com-
pany, Inc., 1983.
P. Ramond, Field Theory: A Modern Primer, Frontiers in Physics, The
Benjamin-Cummings Publishing Company, Inc., 1981.
References
2. G . ' I. II OO I'I.II,lId !'vi . V. 'It.lil ltll, Nllc/. Pliys. 014 , I i'll) (1!)7~) .
3. D.J. Gross ami I" . Wil czek, Phys. Rev. Lett. 30 , J3-13 (J973).
4. H.D. Politzer, Phys. Rev. Lett. 30, 1346 (1973) .
5. W. Caswell, Phys. Rev. Lett. 33, 244 (1974) . D.R.T . Jones, Nuc/ .
Phys. B75, 531 (1974).
6. C. Callan, Phys. Rev. D2, 1541 (1970).
7. K. Symanzik, Comm . Math. P1lYs. 18, 227 (1970).
The Production of
Particles and Jets
in Hadron-Hadron
Collisions
260
/
( 1I ' lI pl.<-r 7 II nol' " 11 11 /1 01, " 11 Coll i lio ll "
III th e ea rly d aY H of ti l(' (>11,1' 1.0 11 IH odei IIl a llY I.lt l'ori HI.H 1\" lil·wd tl l(\.l. tI, (' o lil y
way to stu dy til e (·o IlHl.i t ll l: lltH wit hi n nucl eons W <I.'i hy pI'do l'lll ili g ek ep il l(' lltH
t ic electron a nd n lItrin o scatteri ng experi ments. lI acl roll- had ro n co lli sio li H
were considered by m a ny to be on a different foo tin g. A famous theor ist 0 11 ('"
remarked that, "you cannot learn about the insides of a watch by colli d ing two
watches together." On the contrary, however , history has shown th a l mu ch
can be learned about the constituent nature of nucleons from hadron- had rOIl
collisions that involve a large momentum transfer or that prod uce a la rp;('
mass object. In these cases short distances are being probed a nd th ' hnHir
constituent subprocesses dominate. At least for that portion of th pro hl l' lI l
that involves short distances the parton approach is applicable a nd ill ad di
tion we can employ QCD perturbation theory. In this chapter we will (:Xll lll i,ll'
some perturbative QCD applications to hadron-hadron collisio ns.
(7 . 1.:1)
where
XT = 2pT/Js, (7 . l.tJ)
and
Ie == tan(B c /2). (7. 1.5 )
In arriving at these expressions the masses of all the constituents a nd ti l('
incident hadrons have been neglected and PT is the transverse m omenlllill of
constituent c (and constit uent d provided the incident partons are pa ra ll el
to the incident hadrons). The invariants of the "internal" constitu ent 2- to-2
subprocess are defined by
7. 1 Si n",1 1\/1<1 J) uhl I'hot,o ll I', w lll ( UOII 262
(0) y
(b)
Figure 7.1 (a) Amplitudes for quark-antiquark annihilation into a photon and a
gluon, q + q .... 'Y + g. (b) Amplitudes for quark-antiquark annihilation into two
photons, q + q .... 'Y + 'Y.
(7.1.7)
it == (Pc - Pb)2 = -2 PC'Pb = ZbU, (7.1.8)
where the momentum fractions Za and Zb are given by
Pa = ZaPA, (7.1.9)
and
(7.1.10)
Momentum and energy conservation of the constituent 2-to-2 subprocess,
Pa + Pb = Pc + Pd, (7.1.11)
/
2 03 Gil pt!'r 7 II ",lI u lI II nd 1'0 11 '011111101111
A---~..------:;
• a b 4-_--
8
• -
insures that
s +i + it = 0, (7.1.12)
which means that
(7.1.13)
or
(7.1.14)
and
(7.1.15)
with
(7 .1.16)
(7 .1.17)
Furthermore,
2 tu iit
PT = - =-;-. (7.1.18)
8 8
t Pb )- (7 .1.20)
Xa = "2 XT
1 (170 + Td1) (7 .1.25)
(7.1.27)
(7.1.28)
(7.1.29)
(7.1.30)
(7 . 1.:ltl)
(7.1 .35)
where
Xa X2
Xb = (7 .1.36)
Xa - Xl
S = XaXbS, (7 .1.37)
and
-
t = -2S XaXT Te, (7.1.38)
(7 .1.41)
where Xa and Xb are given by (7.1.25) and (7.1.26), respectively, and sand i
by (7.1.27) and (7.1.28), respectively2 .
Summing the leading logarithmic contributions from gluon Bremsstrahlun g
converts the "bare" structure functions in (7.1.35) and (7.1.40) into th e Q'I.
dependent QCD structure functions GA .... a(x, Q2) and GB .... b(X , Q2) . T il !it'
"renormalization group" improved stru cture fun ctions sl\tili fy th ' Q2 evolu-
tion equations derived in Chapter 4 and, [or example, (7.1. 35) becomes
(7.1.42)
Here it is not clear what to use for the variable, Q, that determines the
scale at which the structure [unctions G(x, Q2) are sampled and also sets the
strength of the strong interaction coupling, a •. At leading order all choices for
Q that increase with the parton-parton center-of-mass energy are equivalent.
Different choices affect the beyond leading order terms. The best choice would
be the one that minimized the higher order terms, thus providing us with an
accurate approximation. However, this cannot be decided without knowing
the beyond leading order terms which in most cases are beyond ones ability
to calculate. Some guesses at the best choice include
Q2 = 4p~, (7.1.43)
and
28iu _
Q2 = _---;:-_
8 + i2 + 2 ·
2 it (7.1.44)
In most cases the choice of Q2 does not make much difference since it always
appears in conjunction with the QCD parameter A (i. e., Q2/ A) and the value
of A is still somewhat uncertain.
The differential cross section for the constituent subprocess q + q -> r + g
shown in Fig. 7.1 is given by
diT ,
di ( 8 , i) =
7raa.e; 4
82 92
(itt + ~i) , (7 .1.45)
which is the same as (5.2.9) except now M2 = O. The differential cross section
for the purely electromagnetic subprocess q + q -> r + r in Fig. 7.1b is given
by
diT "
2
7ra e: 1 (.u i)
-;:-(8, t) = - ' 2 - -3 2 -::- + -:: - , (7.1.46)
dt 8 t u
and it identical to (7.1.45) except the former has a color factor of 4/9 and the
latter a color factor of 3/9 = 1/3 and a. is replaced by a e;.
The ratio of these
two cross sections is thus given by
diT/di(qq -> rr) 3 a 2
, =--e, (7.1.47)
diT/dt(qq -> rg) 4 a. q
which leads to several interesting predictions for pion-proton collisions. If we
naively assume that the dominant contribut.ions to the production of single
and double photons at high PT are the annihilation terms uu -> rg and
/
2U7 C ia ,pt.,, ', II ,h,," 11 ".1" ,11 ( :lIlI i iOll'
uu -+ 'Y'Y, res p('<"I.iv.'l y, I," /I I' ('o llisi ns Iwd til llllll i hil lt.l.i o ll !.t' !'IIiS tid • 'Yfl
a nd dd -+ 'Y'Y, resp.·('tivl'iy. for 7f+]J co llisions we co nci lld e that
E du/d:J p (7f - V -+ n + X) 3 1 a
3 - -a 2
etA (7 . 1.'18)
Edu/d p(7f - p -+ 'Y + X) 4 a3 3 a.
and
3
Edu/d p(1r+p -+ n + X) _ 3 a 2 _ 1 a
3
Edu/d p(1r+p -+ 'Y + X) - 4 a 3 ed - 12 a.
(7.1.'19)
(7 . I .!i'))
which is the same as (5.2.18) but with M2 = O. 'When summed ovcr qUllrt H
and antiquarks the Compton subprocess contribute equally to 1r+ p a lld 1r ])
collisions.
An interesting source of double photons that contributes equally to 7f1 7)
and 1r-P collisions is gluon-gluon annihilation into two photons, 99 -+ 'Y'Y .
This process proceeds through a quark loop as shown in Fig. 7.4b . Thcr/'
are actually six amplitudes that must be summed. They correspond to til!'
gluons and photons attaching at various points around the qua.rk loop. Th.,
differential cross section can be arrived at from the light-light scatte rin g crOHS
section, 'Y'Y -+ 'Y'Y, first computed by De Tollis in 1965 3 . The res ult is giv 'n
by4
7. 1 ~1 1I/l.1 Ilild Ooublll Ph otoH 11111.111 1 tlI'1I 268
(0) y q
q ,.. gluon
S
(b)
,.
S
Figure 7.3 Amplitudes for the "Compton" production of a photon via the subpro-
cess (a) q + 9 - l' + q and (b) q + 9 - l' + q.
d (f'
- , (gg ->
, 0 '2a, 2
n; s, i) = -7rS
8'2 ~ eq
(nJ ) 2 2
dt ~ •
• =1
+(
'2
S ~U
'2)
log2( -slii.) + 2
('
S i ')
U log( -slii.)
)2
2
+ (£2 ~ ii. ) (log2(ilii.) + 7r2) + 2 (i ~ ii.) log(ilii.») 2]
1 [('2 + £2) (' _ i\ ('2 + '2)
+2 S u2 Iog2( -sit) + 271og( -sl£) + S i2 U log2( -slii.)
/
260 e h pt('r 7 II ",111111 II lid ... " Collllllooll
gluen y
Figure 7.4 The production of a single photon (a) and two photons (b) via g)uon-
gluon annihilation which proceeds through a quark loop.
2
+ 2 (s -::- i) log( -s/u) + (t"2 ~ u ) (log2(i/u) + 11'2) + 2 (i ~ u) log(i/u)]
t s S
+ 11'22 [(s2+£2)1
u' 2
(_'/1\ (8_£»)2
og s t J + u,
(7.1 .51\)
(7.2.1)
(7 .2.5)
/
2 71
Ratios
1.0 ~-..,-":"':"'=''-'F'----r----'----'r--.----.
w: 27.4 GeV
10
PrCGeV)
Figure 7.5 Beam ratios and double to single photon ratios in 7rp collisions at W =
..;s = 27.4 GeV and (J = 90°: (7r+ /7r-)p -- l' + X (solid dots), (7r+ /7r-)p -- n + X
(solid squares), 7r-P -- (nh) + X (up pointing triangles), 7r+P -- (nh) + X
(down pointing triangles). Also shown are the naive estimates: i(ed/eu)2 (dashed
line), i(a/a,)e; (dot-dashed line), i(a/a.)e~ (dotted line) (taken from Ref. 4) .
i == (Pc - = =
Po)2 -2 Pc·Po zot/zc, (7.2 .6)
u == (Pc - Ph)2 = -2 Pc ·Ph = ZbU/Ze. (7.2 .7)
The connection between the internal and external invariants comes from
Po = ZoPA, (7.2.8)
Ph = ZhPB, (7.2.9)
Ph = ZcPe, (7.2.10)
where constituent and hadron masses are neglected. The 2-to-2 scattering
constraint
s +i + u = 0, (7.2.11)
h "trigger"
A-----.--4
• ~-'----8
implies that
(7.2.12)
where
1
Zl = -u/ = lZT Th '
S (7.2.13)
Away
Jet
Figure 7.7 Illustration of the four jet structure resulting from a beam hadron
(entering from the left along the dotted line) colliding with a target hadron (entering
from the right along the dotted line) in the center- of-mass frame: two jets (collection
of particles moving in roughly the same direction) with large transverse momentum ,
PT, and two jets with small PT that result from the break-up of the beam and t arg t
hadron.
which together with
_ 2 sin{Oem) d3 p
dO em d XT - E ' (7.2.18)
87r XT
gives
(7 .2.19)
(7.2.20)
For the limit on Xa one sets both Zc = 1 and Xb = 1 and solves for Xa givin g
min Xl
Xa = ---. (7.2 .22)
1- X2
The invariant cross section for the production of a constituent parton can be
arrived at from (7.2.20) by setting
Dc(ze, Q2) = 15(1 - =c), (7 .2.23)
and using
(7 .2.24)
(7.2.25)
with
(7.2.26)
and x:;,in given in (7.2.22) . This is the same formula we arrived at in (7.1.42) .
As illustrated in Fig. 7.8 the "bare" quark distributions, C(O)(x), in (7.2.1)
have been replaced by the "renormalization group improved" distributions
C(x, Q2) that satisfy the Q2 evolution equations in Chapter 4. In so doing
we have in effect summed a set of leading log contributions. Similarly, the
"bare" fragmentation function, D(O)(x), have been replaced by the Q2 depen-
dent functions, D(z, Q2) and thus summing the leading log corrections to the
fragmentation functions . If we compute the constituent differential cross sec-
tion, dfTjdi, to order ex; then (7.2.20) and (7.2 .21) are correct to leading order.
The seven parton-parton differential cross sections that contribute are given
by 5
d. (1
~
-;::(ab --+
• ~ ~ _
cd, s, t) - -;:;;-IM(ab
'!rex. 2 - --+ cd)1 ,
2
(7.2.27)
dt s~
where
(7.2 .28)
(7.2 .29)
h
P--d2;::: h p_-<~;:
(0)
p
+
p
-
p--c~ h P --<J2!i;: h
(b)
p--.--.. p---~.
h h p_-<~;: h
(e)
p-----
+
P--j.....
-
Figure 7.8 (a) Illustration of how the leading log parts of the subprocess q + (I
9 + q + q sum to give the Q2 dependent fragmentation functions, D~ (z, Q2). (I»
Illustration of how the leading log parts of the subprocesses q + q -+ q + q + gAlle!
q + 9 -+ q + q + q sum to give the Q2 dependent quark probability dis tributioll R.
Gp_q(x, Q2). (c) Illustration of how the leading log parts of the subprocesses q+q
q + q + 9 and q + 9 -+ ~ + 9 + 9 sum to give the Q2 dependent gluon probability
distribution, Gp_g(x, Q ).
-_ -2 4 S + U- t + U- -2 2 12
8 U- 2) 2
IM(qiqi -+ qiqi)1 = '9 ( i2 +~ - 27 sf ' (7 .2.30)
-
IM(qiqi
_
-+
2
gg)1 = 27
32(il ui+i - 38(il~
2 2
+P) ,
)
2
(7 .2.31)
-
IM(gg -+
-2
qiqi)1
1(il ili+f2) - 83 (il 82+f2) '
=6
2 2
(7.2.32)
7: L /I "'~(' ' / 'I'II I1 NV( 'I'H(' MIIllI !l lItlll1i M ' ·' ''IHI 27G
-
IM(qig -+ qig)1
2= -91( t.l~+S2
its
)
+ ( '/t'2 j2I' S:.! ) , (7.2 .:13)
-
IM(gg -+ gg)1
2
= -29 ( iti
3 - -::)' - 7,)
us - -:-2
si ) . (7 .2.34)
s- t- u
We will not compute all of these cross sections here, however, it is instructive
to examine closely the subprocess gq -+ gq. This is the first time we have
dealt with two external gluons which causes some extra complications. The
amplitudes for gq -+ gq are shown in Fig. 7.9. We will first examine gauge
invariance for the QED and QCD case.
In the QED process ,q -+ ,q only the amplitudes A. and B in Fig. 7.9 are
present and
(7.2.35)
where (j.I(Ad and (v(A2) are the polarization 4-vectors of the incoming and
outgoing photons with helicities A1 and A2, respectively, and
'22-(
(A. + B) j.lV = -le equ P2,S2
) [,v(11+r/1hl
(
Ij.I(12-r/dl v ] (
)2 + ( )?
l )
U P1,Sl .
P1 + q1 P2 - q1 -
(7.2.36)
Gauge invariance requires that if we make the replacement
(7.2.37)
then
(A. + B)j.lv (qt}j.I<(A2) = o. (7.2 .38)
With this repl acement the factor in brackets in (7.2.36) becomes
__ ' 2- (
(A.+B) j.lV- ) [Tb T alv(p1+r/1hj.l
19.UP2 , S2 ( )?
P1 + q1 -
/
277
9 q
A = t-
q
t
5
9 q
B =
9 q
S
9 q
Figure 7.9 Amplitudes A, Band C for elastic qluon-quark scattering, g+q - g+q ,
and the amplitude G for elastic "ghost"-quark scattering.
+ Ta Tt'p<P2
P2 - q1
-/1hll
] U(Pl,SJ), (7.2.45)
where Ta and Tb are the SU(3) color matrices given in Appendix D which
satisfy
(7.2.46)
7 .'1. L I""~<l ' , 'rll,II Hv('I' " M OIlI!I II I,ll III M,I/'''''II 278
(7.2.47)
and
(7.2.48)
The first term in (7.2.47) vanishes as in the QED case but the second term does
not and gauge invariance does not hold. Of course this presents no problem
since the third amplitude in Fig. 7.9 involving the triple-gluon coupling will
restore the gauge invariance. From Appendix C we see that
(7.2.52)
when there is more than one external gluon. For one external gluon we show d
in Chapters 2 that this replacement (equivalent to the Feynman gauge) giv s
the correct sum over the two physical helicity states. The unphysical states
cancel out . With two or more external gluons these unphysical polariza ti II
states do not cancel out and one does not arrive at the correct gauge invarialll.
cross section by the use of (7.2.57).
One way to obtain the correct cross section is to insist that the polariill
tion states tl'(At} and t,,(A2) are physical (i.e., transverse) . This is a CCO lI1
plished by the projection
~ nl'k" + n"kl'
L: tl' ()A t"*( A) = - [
gl''' - (n.k) +
n2kl'k,,]
(n.k)2 , (7 .2.!)H)
yielding
(7.2.63)
where
(7.2.61\)
(7.2 .65)
(7 .2.66)
(7.2 .67)
(7.2.68)
7.'1. I.",.,. . , ' 1'1'11 11 \/(11' I' M 0 1111111 1.11 IIi M III/ 'I ll 28 11
2 DC. = --
44
{~ . (-U
i J Lv? - ,- ) ,
t
(7.2.(9)
giving
IA + B + CI 2 = - 9 4(8 8U+U 2 2
)
+ 2 ( 1 - 8U)
i2 - 1, (7.2.70)
which is identical to (7 .2.33) . In (7.2.64)-(7 .2.69) the first factor is the color
factor and the second factor arises from spin averaging . For example, the color
factor for IAI2 is given by
1
Htr(TaTbTbTa) = 2 4 [tr(TaTbTaTb) - i/abctr(TaTbTc)]
= 1
2 4 ( - 1;8bb 38aa) +t
_
-
8 ( 1
24 -12
+ 43) _- 2
9' (7.2.71)
and the color factor for ICI 2 is given by
H /abc!ab ctr(TcT e) = 2~ ~ /abc/abc
= -214~ (Fc)ab(Fc)ab = 214~ tr(FcFc)
_ 1 3 c _ 1 24 _ 1
- 242 Vee - 242" - 2' (7.2.72)
Alternatively, we could use the Feynman gauge projection in (7.2.57) t~
get her with the ghost-quark scattering diagram in Fig. 7.9. This gives
for the square of amplitude C where the subscript E signifies that (3.2.57)
has been used to sum the gluon polarization states. Comparing with (7.2.67)
we see that
ICI~ = ICI 2 - ~~
1
-2 • (7.2.74)
t-
The use of (7.2.57) has produced the spurious contribution
18'11
-2 i2 ' (7.2.75)
IGI2 = ~~ :2 tr(12rfiArf2)
1 (-'11)8
-2t2 (7.2.76)
where as before the first factor is the color factor (same as for ICI 2 ) and the
second factor arises from spin averaging (treating the ghost as a scalar). To
arrive at the correct gauge invariant result we must subtract (7.2.76) from
(7.2.73) . Namely,
(7 .2.77)
/
281 hAp t 1 "1 If ,It 1111 II ,111111 ;olli i II
L'
L h 2
oq,(z.a)
(0) q'
q - ..--0:::: ~ h
N ~
~
GN_q(x..rJ)
(b)
(e)
Figure 7.10 Illustration of the universal nature of the ~uark probability distribu-
tions, GN-+q(X, Q2), and the fragmentation functions, Dq (z, Q2) . At leading order
in QeD the same functions describe: (a) deep inelastic lepton-nucleon scattering,
(b) hadron production in e+e- annihilations, and (c) particle production at large
transverse momentum in hadron-hadron collisions.
and the two methods are equivalent. The extra (-1) for the ghost contribution
is analogous to the (-1) that one must insert for a ghost loop.
As illustrated in Fig. 7.10 (7.2.20) together with the seven cross sections
(7.2.27)-(7.2.34) can be used to predict the rate of large transverse momen-
tum meson production in hadron-hadron collisions from the structure func-
tions, G(z, Q2), determined in deep inelastic lepton-hadron scattering and the
7.2 LftrK ' 1'11\1\ Vor (. MIllIl f1 I1f.nlll Mf1riOIIlI 282
o
-I
~
N
-2
~ Theory {-pp-"..+x
~ -3 A a O.4GeV/c -pp _ y+X
..c
.=- -4
rb
...... -5 ........
.g ....., ....~. . . .
w -6
.........
8'
Q
-7
·-...,t •
......, ..
............
...............
",
-8
-9.~~~~-7~~~~~~~--~~__~~~~
2.0 3.0 4.0 5.0 6.0 10 II 12 13 14
PT
Figure 7.11 Predictions of leading order perturba.tive QeD for the production of
large transverse momentum pions in proton-proton collisions, pp -+ 11" + X, com-
pared with data at W = VB = 19.4 and 53 GeV. Also shown a.re the corresponding
predictions for large transverse momentum photons, pp -+ 'Y + X (taken from Ref.
6).
wil er' 1" f' i f! til(' 1>1 11 1.,111 I" 11I' II V,' ,',w II O'"Cltlll1. Fi~ , 7 , 11 " I, OW H 11 QC ;I) pr('
dictio ll for pp - . /I (I I .\' 111. H. ' (~ lIter- of- Ina', lI ' rg;y or I UA a lld 53 ;.' V
a nd OCM = =
90° ulli ll g (7.<! .:W) and A 0.4 Gey6 . To co mpa.re with data til<'
lead ing order formul a ill (7 .2.20) has been "sm eared" over the intrin 'ic (.raWl-
verse momentum of the parton within the initial hadron, (kr)"_q , and ov ' r
the transverse momentum of the hadrons within the outgoing jets, (kr)q _ h.
Smearing is an important effect in hadron-hadron collisions due to th e "trig-
ger bias" which selects configurations in which the initial partons are a lready
moving toward the trigger 7 ,
Another interesting application of the leading order formula in (7 .2.20)
is the prediction of large pr particle ratios in 1Tp and PP collisions. l' ig. 7. I<!
shows the predictions for 1T-P and pp interactions compared with data at. <!OO
GeY and OCM = 90° 8 , Although in some cases the agreement is not exnd , 1.1" ,
systematics of the data are described correctly. The proton conta in s III )'I ' II
than d quarks and since u-quarkjets contain (at high z) more 71'+ th an 71' - , I I('
expects and sees a 71'- /71'+ ratio considerably less than one in pp collisions . For
71'- p collisions, on the other hand, one predicts a 71'- /71'+ ratio slightly grea.ter
than one. This is because the d (and u) quark in the 71'- has on the average
a higher momentum than the u-quark within the proton which at 90 degre s
results in a slight excess of 71'- over 71'+ at large Pr . The data shows a ']1' - /']1'+
ratio that is slightly greater than one but not quite as large as the predict ion .
In PP collisions one can only produce large Pr J{- mesons by scattering
partons out of the proton "sea" or by producing them as non-leading particles
in the quark jets. A very small J{- / J(+ ratio for PP collisions is therefor
predicted. On the other hand, J{- mesons can be produced directly in 1T- ]J
collisions through the scattering of the valence it quark within the ']1' - so thaI.
the ]{- /]{+ ratio here should not be as smalL This is indeed what is observed
in Fig. 7.12b.
While the 71'- /71'+, J{- / J{+ and J{- /71'- ratios are predicted to be q uit(,
different for 71'- and proton beams, the J(+ /71'+ ratio is predicted to be (~f!
senti ally identical for the two beams. This is because neither the 71' - Or I,Iw
proton contain any valence strange quarks. In both reactions la rge ]Jr h' +
mesons arise from the fragmentation of u-quark jets. The f(+ /']1'+ ra ti o f I'
both reactions simply measures the extra difficulty of producing a S8 pair ov ' I'
a dd pair during the fragmentation process and should be the same for hath
reactions. The data in Fig. 7,12d confirm this.
Theoretically it is easy to define the ''jet'' cross section, It is the cross section
for producing a parton (quark or gluon) and is computed to leading order by
(7.2.25) together with the parton-parton cross sections in (7.2.27)- (7 .2 .34). As
7.:1 .I I, 1' 11 .. 1,, ( I. Oil 2tH
3.0
Particle Ratios + 17"- beam ~ p beam
2.0
1.0
0 .5
0.2
1.0
0.5
.,.~.
0.2
.
/
0.1
0 .05 :
2 4 6 0 2 4 6
PT(Gev/c)
Figure 7.12 Comparison of data on the large transverse momentum, PT, trig-
ger particle ratios for lr- p and pp collisions at a laboratory beam momentum of
200 GeV Ic and Bern = 90 0 with leading order QCD predictions. The solid and dashed
curves are the predictions for pp and 11"- p, respectively, and should only be com-
pared with data above about 3 Ge V I c. The dotted curves are extrapolations of the
low PT pp data (taken from Ref. 8) .
can be seen in Fig. 7.13 the 'jet" cross section is predicted to be considerably
larger than the single particle cross section 9 . The single particle trigger always
comes from a parton carrying more momentum (typically 10-15% more) than
h Allt r 7 " , h OH " , h Ull ( loll . (11111
, , , , , ,
u(pp- Jet + X l/U(pp-1T o + X)
W=500 GeV
10" W=53 GeV 8~m.
=900
W= 19.4 GeV
.. ,."
,. ", .".
.. -.. - " ,," ."".
......
"" "" " .""
./
-
/
.. I / / /
··· I I I I
./
·· I I I
I
I
I I I I
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Figure 7.13 Leading order QeD predictions for the ratio of "jet" production (ill
this case parton production) at large transverse momentum, PT, to the produ tiO Il
of a single lr°
at the same PT in proton-proton collisions. The predictions At W
..;s= 19.4, 53, and 500GeV are plotted versus XT = 2PT/W (taken from Ref. 6) ,
the trigger particle. Furthermore, the probability that one partid car ri,'/'!
almost all the momentum of the parent parton is quite small (only IL few
percent) . These two effects combine to give the u(Pp ~ "jet" + X )/U(1'11
11"0 + X) ratio seen in Fig. 7.13. Experimentally it is difficult to defin n "jet ,"
One can never be sure that one has included the desired hadrons and reje t, d
to undesired ones. Nevertheless, as can be seen in Fig. 7.14 , the early Q D
predictions of a large jet cross section are in rough accord with data 6 .
It is an interesting consequence of the leading order formulas in (7.2 .20)
and (7.2.25) that the single particle and "jet" cross sections are not very
sensitive to the QeD perturbative parameter A. One might think that since
the parton-parton differential cross sections are proportional to 0'~(Q2) that ,
for example, the jet rate would depend strongly on A. Fig. 7.15 shows that
this is not the case. The reason is due to the compensating effect of the parton
distributions, G(z, Q2). If A is small (i.e., 0', small) then du/diissmall bu t the
7.:1 .I"j, " "H i llel Ilil 286
3~)~----~----r---~----~
pp- "Jet" + X
W= 19.4 GeV
2
• E260
• E395
BCM::::: 90°
............
....-.
~
~ 0
\ ,
~ 0,
~
"-
\ ,,
.D
::l
..........
-I
,
~
a.
\\
~ -2
Single ,
b
'C Tf Rote \
W
'-...-' -3
Q
'0\
\
0'
o
-4
\ ,
'0,
-5 °
PP - Tfo + X
\
\
\
0,
-6~J~----~----~----~~'~
234 5 6
Pr GeV/c
Figure 7.14 Data from the first Fermilab experiments to confirm the leading order
prediction of perturbative QeD of a large "jet" cross section at high transverse
momentum in proton-proton collisions (taken from ReJ. 1!).
probability of finding a large x partons within the initial hadrons is large since
G(x, Q2) has not changed much from its reference value, G(x, Q~). On the
other hand, if A is large (i.e., a. large) then du/di is large but the probability
of finding partons at large x is small since G(x, Q2) will now have evolved far
away from its reference value. This means that data on the cross section for
/ 287 'II lit r 7 II ,hilI' II ,j",n ( loll .. (1111
2
10010 {EdO'"/d"p [f'blGev·)} versus PT
8.eM=90· pp - JET +X
0
A a 0.8 GeV
-2 A a 04 GeV
A a 0.05 GeV
-4
-6
-8 W a 500
-10
-12
-14
0 300
PT GeVlC
Figure 7.15 Shows the insensitivity of the leading order QeD predictions for th
production oflarge transverse momentum "jets" (in this case parton,,) to the precise
value of the perturbative parameter, A.
As we have seen in the previous sections in leading order QCD. mesons are
produced at large transverse momentum in hadron-hadron collisions as the
Hard Scattering Event
Jet 3
"high PT jet"
Jet 2 Jet I
"target jet" "beam jet"
Hadron Hadron
A
--- a b
8
Jet 4
"high PTjet"
Figure 7.16 Illustration of the naive parton model in which high transverse mo-
mentum particles are produced by a single, hard, large-angle elastic scattering of
partonsj one from the beam hadron and one from the target hadron. The large-PT
event consists of four jetsj two with large transverse momentum resulting from the
fragmentation of the outgoing partons and two with small transverse momentum
resulting from the break-up ofthe beam and target hadron as illustrated in Fig. 7.7.
result of a hard parton-parton collision; one parton from the incoming beam
hadron and one from the target hadron. The resulting elastic parton-parton
scattering produces two outgoing partons which subsequently "fragment" into
jets of hadrons producing the four "jet" event structure (two large PT jets,
a beam jet and a target jet) shown in Fig. 7.16. The invariant cross section
A + B - h + X has the form given in (7.2.20), where d&/di is the hard
scattering constituent differential cross section, a + b - c + d, and where one
must sum over all parton types (i.e., qq - qq, qg - qg, gg - gg, etc.).
The variables sand frr are the center-of-mass energy squared and transverse
momentum, respectively, for the hard constituent subprocess.
The effects of soft and collinear gluon emissions off the incoming partons
are treated by assigning the proper Q2 dependence to the parton structure
/ ZHt)
filII ·ti OIl S, GA _" .(J·, (J l ) , 'lIlId ll d y, soft. a ll d co llill ·I\.r g ill 11 1 II n 'III HH tl"ll ldllll l(
off.' the o utgoill g plirto ll H rn lil ltH ill a. Q2 depend 'li ce of tli (' fmg lll ~' "t. lit.i o ll
fun ct io ns, D~'( z , q 'L ). ' I'll<' (J'l dependence of the stru cture a nd fragllielif.;lti oll
functions is given by pert ur bation theory. Since, to leadin g o rder, o ne n cd
consider only the case where the gluon radiation is soft or collin ear, o ne is
still left with an effective 2-to-2 subprocess and a four j et stru cture simil ar to
t hat for the naive parton model shown in Fig. 7.16.
On the other hand, equation (7.2.20) is not completely satisfactory. A
successful description of single particle production at high PT requires t.hlit.
one include the effects of the "intrinsic" transverse momentum, kT , f t.Il1'
partons within the initial hadrons. In hadron-hadron collisions the in IIlill ~~
partons cannot be taken to be parallel with the incoming hadrons. III fli r t ,
phenomenologically {kT} appears to be quite large. Feynman, Fielcl, a nd Fox ll
found an effective {kT} of 0.85 GeV in hadron-hadron collisions. D aWol l ' 0 1
the rapid fall with PT of the constituent parton-parton cross section , 0 11 (' hllH1\
"trigger bias" in single particle experiments. One preferentially selec ts V 'litH
in which the incoming two partons are moving toward the trigger. Simil a rl y,
the outgoing hadrons are not produced precisely parallel to the outgoing par-
tons . One way to include the effects of both the transverse momentum spread
of the hadrons within the outgoing "jets" and the intrinsic kT of the partons
within the initial hadl"Ons is by "smearing" the basic equation (7.2.20) . T he
intrinsic transverse momentums are assumed to be distributed as a ga ussia n
(or an exponential) independent of both Q2 and the process and one then
integrates over these kT values.
However, in QCD, a large effective intrinsic transverse moment um is un-
derstood in terms of noncollinear gluon Bremsstrahlung. As illustra tecl ill
Fig. 7.17a, single particle triggers tend to come from the fragmenta tion of a
large PT parton that is balanced on the "away-side" by several lower mO lllCII
tum partons. The same bias effect occurs for "small aperture" j et (calol'ill H'
ter) triggers. By small aperture calorimeters I mean observations th a t HIIII I
the total transverse energy, ET, in a small region ('"'-'I steradian in the 1I1,{' r
of-mass frame) of phase space. Such a calorimeter roughly conta ins a.1I t.I l('
hadrons from a single parton ''jet.'' Clearly, the use of (7.2 .20) plus S lll l~ [lr
ing is a crude way to include the effects of non collinear gluon emiss ioll s. '1'111'
leading order formula does reasonably well in estimating the m agni t ud of ti l('
single particle cross section or the small aperture jet cross secti on , but is not.
very good if one is interested in more detailed questions as to the overall evellt.
structure. For one thing, (7 .2.20) plus smearing still treats the problem as a ll
effective 2-to-2 subprocess, whereas the subprocess in Fig. 7.17a is act ua ll y
a 2-to-4 subprocess (qq -+ qqgg) . In order to make further pl"Ogress a nd to
answer more detailed questions about high PT events one needs a method of
calculating (or approximating) the effects of 2 -+ N subprocesses. On e mu st
include the effects of the noncollinear emission of gluons off the initi a l a nd
final partons.
One way to proceed is to extend the QCD "leading pole" Mont 'a r!
(a) Single Particle or
Small Aperture Calorimeter
Proton Proton
) (
Proton Proton
Trigger
Figure 7.17 (a) Illustration of the trigger bias effect for a single particle or a small
aperture calorimeter experiments in hadron-hadron collisions. The large transverse
momentum, PT, trigger is obtained by combining a 2-t0-2 hard scattering of smaller
PT with additional transverse momentum gained by the incoming quarks through
the emission (or Bremsstrahlung) of gluons. (b) Illustration of the trigger bias that
occurs for large aperture calorimeter experiments. Large transverse energy, Eq, is
produced by the Bremsstrahlung of many lower energy gluons in addition to the
hard 2-t0-2 scattering.
-hOle"
h.O ___ ~:x,) Hadron
8, --- -~
B,
"Type 4· Hadrons: "Type 2" Hadron, :
All hadrons from final All haelrons from ,n,hal
par tons D,. Oz •.. . elc portons 8,. 82. . . . plus
hadrons from "hole" h, .
illustrated in Fig. 7.18 final state partons have timelike invariant masses anti
are allowed to radiate until their masses are below the cut-off
tial partons have spacelike invariant masses with maximum (negative) value
t:
= (1-':)2. Ini-
taken to be t!:.in = _Q2 (here one usually takes Q2 = 4p}. These partons
are evolved from an initial value of t~ = _(jJ~)2 up to this maximum. MuJ-
tiparton cross sections are approximate but all gluon emissions are retained
including the (infinite sum of) divergent soft and collinear configurations. One
is, however, not restricted to the collinear case as exact kinematics are main-
tained at each emission 12 .
Present day QCD Montc- 'a.rl o mod I ~ ror had rOIl -hnd l'O ll ('() lli ~ i ns COII-
tain many approxim a tions and shoul d not be ta ken too s' ri II sly lJ . At th e
parton level the "leading pole" approxima tion can be off by a si:tab le a mount
for wide angle Bremsstrahlung plus there are other theoretical uncertainties.
The most uncertain ingredient in the models, however, is the "hadroniza-
tion" phase (the algorithm used to turn the outgoing partons into hadrons).
At present we really do not know the correct way to fragment a collection
of partons into hadrons. There are a variety of suggestions in the literature.
The simplest is to fragment each parton independently in the hadron-hadron
center-of-mass frame according to the Field-Feynman prescription discussed
in Chapter 3. Clearly this method cannot be precisely correct . In hadron-
hadron collisions there are many soft quanta for which the FF prescription
is inadequate and inappropriate. Furthermore, when there are many nearby
quarks and gluons, one cannot consider their fragmentation as independent.
Presumably, it is color separation that is responsible for the fragmentation .
Nevertheless, with the FF hadronization scheme one can keep track of
where the outgoing hadrons come from. As is illustrated in Fig. 7.18, type
1 hadrons are those arising from initial partons AI, A 2 , •• • etc., plus the
hadrons arising from the fragmentation of the "hole" h A • Similarly, type 2
hadrons arise from the initial partons B I , B 2 , .. • etc., plus the "hole" hB.
Outgoing partons Ci and Di fragment into hadron of type 3 and type 4,
respectively. I have been careful not to call, for example, type 3 hadrons a
"j et ." Type 3 hadrons may form several "jets" or no ''jets'' or type 3 and type
1 hadrons may conspire to form one "jet." The precise definition of a "jet" is
at the discretion of the experimenter.
Although the parton-shower Monte-Carlo models are a crude attempt to
reproduce nature, they do contain in an approximate manner many of the
features expected from QCD. As the momentum transfer of the hard parton-
parton scattering subprocess, Q, increases the QCD parton-shower models
deviates more and more from the naive parton model. This is due to the fact
that the probability of radiating a hard gluon off an initial or final state parton
increases logarithmically as Q increases (see Fig. 7.19) . In addition, in the
QCD parton-shower approach there are dynamical cOl'l'elations between the
outgoing high PT jets (hadrons of type 3 and 4) and the low PT "background"
jets (hadrons of type 1 and 2) . Large PT outgoing jets means that a large
Q2 subprocess occurred which in turn stimulates a large amount of gluon
Bremsstrahlung in the initial state. In the QCD approach it is incorrect to
consider a high PT event as a superposition of two high PT jets on top of a
"minimum bias" background (i.e., ordinary low PT event). The great majority
of minimum bias events contain no hard scattering, whereas essentially all
large PT events occur as the result of a hard parton-parton scattering.
With the QCD parton-shower Monte-Carlo one can investigate many as-
pects of large PT hadron-hadron collisions that could not be estimated by
smearing the leading order equation in (7 .2.20) . One can , for example, ex-
amine more detailed questions as to the overall event shape and the manner
/
203
G luon Emission
Figure 7.19 Illustration of gluon emission. The probability of emitting a hard gluon
(one with PT > tl) is given by P(PT > tl) oc a.(Q2)log2(Q2ftl 2) and increa.,es
like log(Q2) which causes the QCD parton-shower Monte Carlo model (Fig. 7.18)
to devia.te more and more from the naive parton model (Fig. 7.16) as the transverse
momentum, PT, of the constituent subprocess, a + b -+ c + d, increases.
d
-5 5
pxGeV
(b) Triggered
2 PJ. 3GeV
pyGeV EJ. 10.26 GeV
d
-4 2 4
,,' PxGeV u
"
9"/ QI
-2
d
Px GeV
-5
PyGeV
-5
Figure 7.20 The transverse momentu~ projections for some typical events pro-
duced from a QCD parton-shower Monte Carlo model: (a) untriggered event at
hadron-hadron center-of-mass energy W = 24 GeV, (b) and (c) global transverse
energy triggered events at W = 24 and 540 Ge V, respectively. Partons are denoted
by dashed lines (gluon8) and solid lines (quark8) (taken from Ref. 14).
Cluster Multiplicit
5.-~'-~~-'~r-+-'-~-'
E~in=O.4GeV 17]1<1
E ~aCh >10 GeV
4
UA2 Data
O~~~~~--~~-L~~~
Figure 7.21 Average number of "clusters" having .E,-(each) > 10 GeV verlu" the
global transverse energy, ET resulting from a QCD parton-shower Monte Carlo
model (solid curve) and the naive parton model (dashed curve) for pp collisions at
..;; = 540 GeV. The data are from UA2 16 • (Figure taken from ReI. 15)
(those with transverse energy greater than ET(hot». Hot cells are combined
to form a "jet" if they lie within a "distance" d = ..;l1rp + 11¢J2 < do from
each other, with the jet direction being the vector sum of the momenta of
each cell in the jet. Cold cells (those with ET < PT(hot» are added to a jet
if d < do or if the angle of the cold cell relative to the jet is less than 45 0 and
the relative PT is less than 1 GeV.
In QCD, one encounters a multitude of event topologies and the choice of
trigger (or even the manner in which one bins the data) preferentially selects
certain topologies over others. Fig. 7.23 shows the percentage of events with
N "UA1-jets" (do = 1,PT(hot) = 2.5 GeV) with ET(each) > 15GeV for cells
of size l1TJl1¢J = 0.2 x 150 with ITJI < 2.5 versus the transverse energy of the
leading jet, ET(jet). The data are from the UA1 group17 and are in qualitative
agreement with the Monte-Carlo.
207 " .t II II I II .hull ( :011 M CIII"
(c) (d)
0.5
-)(-~
••
0.0 L..--'---'-_L..--'---'----JL--....L.---L..---JL--....L.-....a:::a:::o
o 23450 2345
Nc.! Nc.!
Figure 7.22 Probability of finding in a given event Net clusters having Er(each) >
10 GeV resulting from a QCD parton-shower Monte Carlo model (dashed lines a.nd
crosses) and the naive parton model (dashed lines and open circles) for fJp collisions
at.;s = 540 GeV with global transverse energy in the range: (a) 60 < Er < 80 Ge V,
(b) 80 < Er < 100 GeV, (c) 100 < Er < 120 GeV, and (d) 120 < ET < 160GeV.
The data (solid lines and solid dots) are from UA2 16 (taken from Ref. 15).
UAI- jet Probability 17] 1 < 2 .5
100
>15GeV
!t
each
ET
x
x
f i
I
x
-
(/)
JiL.
C
x
<1>
>
W
-
~
0
0
50
9
~ {
30 40 50 60 70
E T - jet (GeV)
Figure 7.23 Percentage of events with Njet "UAI-jets" with Er(each) > 15GeV
versus the transverse energy ofthe leading "UAI-jet", ErUet), resulting from a QCD
parton-shower Monte Carlo model (solid curves) for pp collisions at ..;s =
540 Ge V,
with Nj et = I solid dots, Nj et =
2 crosses, Nj et =
3 open circles, and Nj et 4 =
open squares. The data are from UAll 7 with Nj et =
I large solid dots, Nj et 2 =
=
large crosses, and Nj et 3 large open circles (taken from Ref. 15).
0
II
-9-
<1
~
<1
c 10
-
c:>
~
A
.W
...
V
Figure 7.24 Average transverse energy flow relative to the leading (highest ET)
cluster for global transverse energy in the range 2,000 < ET < 3, 000 Ge V in pp
collisions at.;s = 40,000 GeV resulting from a QCD parton-shower Monte Ca.r1o
model. Each bin gives the average transverse energy in 6,,6t/J = 0.1 x 180 0 with
16t/J1 ::; 90 relative to the leading cluster (i.e., toward side). The shaded region are
0
data from UA1 17 at W = 540 GeV and ET(jet) > 35 GeV (taken from Ref. 15).
About one half of the transverse energy and multiplicity flow plateau in
Fig. 7.24 and Fig. 7.25 arises from initial state gluon Bremsstrahlung. The
QeD parton-shower model predicts about 22.8 charged particles in t1,.,!l.q, =
7." MOIII, !tOO
Multiplicity Flow
35
2< ET< 3 TeV E;'in = 1.0GeV
1')1< 4
cell size :
30
6~6.=O.lX6°
0
0 I UAI DATA
CD 25
X
0
II 20
--e-
<l
~
<l 15
c
A
'V
III
~ 10
Z
V
Figure 7.25 Average particle multiplicity flow relative to the leading (highest ET)
cluster for global transverse energy in the range 2,000 < ET < 3,000 GeV in pp
collisions at Va = 40,000 Ge V resulting from a QCD parton-shower Monte Carlo
=
model. Each bin gives the average number of particles in 6,.,61/> 0.1 x 180 0 with
161/>1 :5 90 0 relative to the leading cluster (i.e., toward side). The shaded region are
data from UA1 17 at W =
540GeV and ET(jet) > 35GeV (multiplied by a factor
of 1.75 to convert from charged particle multiplicity to the total multiplicity) (taken
from ReI. 15).
1 X 3600 • This is about a factor of 3.5 times higher than the charged particle
:1()1 ()I,It,p!,!" 'f 11".11111' Il lI d,,," ( 1,,1I1t/IUII "
d II siLy o f 1\,1)0111, 0.11 f' p, ·c l., ·d 1',' III lI1inimum bias '·V(',oI.H It l. thiH (J ll t·rp;y. Thi H
cnh a nc menL I' ti" , pl at" /tll ov('(' Illinillium bi as is III HO ohst,rved by tilt' UA I
group at 51.0 e V, Ii )wcver , a t the larger E1' val li es of t.1t SS til ell' ct.
is predicted to be even greater . The large transverse energy and multipli city
plateaus arise because we are demanding large ET and hence a la rge Q 2.
The large Q2 induces a lot of initial and final gluon Bremsstrahlung and thi s
produces a large number of "plateau" hadrons.
At this state QCD Monte-Carlo models are still quite crude and only qu a l-
itative comparisons with data are warranted. Nevertheless much has a ll' ady
been learned. Existing data on the production of multi-jets with large trll.lIH
verse energy are in good qualitative agreement with the expectations o f Qe D
and deviate greatly from the predictions of the naive parton modcl. Ov('('
all the hadron collider data show strong qualitative evidence for QCD whi ('1i
strengthens the evidence from data on e+ e- annihilations and deci in laHti c
lepton-hadron scattering.
Problems
7.1. Show that the differential cross section for the "annihilation" subprocess
q + if --+ 1 + 9 in Fig. 7.1a is given by
dO- ,
di (s, i) =
7ro:o:.e~ 4
82
(ft i) '
9 2 7 +:a
and for the purely electromagnetic subprocess q + if --+ 1 + 1 in Fig. 7.1 b is
given by
d~ (8, i) = 7r~:e: -3 2 (~ + !) .
1
dt stu
Also show that the differential cross section for the "Compton" subprocess
q + 9 --+ 1 + q is given by
7.3. Consider the inclusive process A+B --+ c+d+X shown ill Fig. 7.2. Starting
from
show that the double differential cross section is give n hy
du
EC 3 d (A+B-+c+d+X;S,PT,Oc,Od)=
d Pc Yd
(0) (0) 1 dfT . i\
+ b -+ c + d, s , t"
A
where
S
A
= S 2 (
-xT 2+ -Te + -Td) ,
4 Td Te
t=
A S
-:tXT
2 ( Te)
1 + Td '
U=-~xf(l+~).
1
Xa + Td1 )
1
= 2 XT
(
Tc '
Xb = !XT (Te + Td ) ,
with T;. =
tan(Od2) and XT =
2PT/../i. Integrate this expression over Od
and show that the single particle inclusive cross section is given by
du 111.0.
E
d3p (A+ B -+ C+X;S,PT,Oc) = -
7r X:l1ft
dX a
G A-+a(Xa)GB-+b(Xb) ( ) (a
A
Xa - Xl dt
where
S= XaXbS,
A S
t= -2" X,.XT Te,
A S 1
u= -2" X,.XT Tc'
and
7.4. Consider the inclusive process A +B -+ h + X shown ill Fig. 7.6. Starting
from
s how t llitt
Edu / d~l,)(!1 I /J It + X; S, l)'/', 0" .. )
0
~1 11..0 dXb G~2.a(Xa)G}J2.b (Xb)Dt(Ze)
1
.• dXa
7r x~n ll' X bnlD
1 da- '
- ----;::(ab ---> cd ; s,t),
Ze dt
where
X2 Xl
Ze = -+-,
Xb Xa
".min _ Xa X 2
"'b -
Xa - Xl
min Xl
Xa = ---,
1- X2
s = XaXbS ,
i= Xat,
it = XbU,
where
Xl = -U/S = tXT/Th,
X2 = -tis = tXT T h ,
and
XT = 2PT /..;s,
Th = tan(Oem/2) .
Let Dg e(Ze) = 0(1 - ze) and show that the resulting formula is equivalc nt
to (7.1.41).
7.6. Repeat the calculation in (7.2.36), (7.2.38), and (7.2.39) for the case in wh ich
the quark in the subprocess
massless quarks,
---> ,q ,q
is massive. Show that , as is t ruc for
and
7.7. For the reaction 9 + q ---> 9 + q in Fig. 7.9 show that for massless par tons
provided
11"1,,, III " Il :H)4
and
provided
7.9. Using the axial gauge projections in (7.2.61) and (7.2.62) compute the dif-
ferential cross section for the "Compton" subprocess gq -+ gq and for the
"annihilation" subprocesses q + ij -+ gg and gg -+ q + ij. These later two
subprocesses differ only by spin and color factors.
Further Reading
G. Altarelli, "Partons in Quantum Chromo dynamics," Physics Reports
81,1 (1982).
F. Halzen and A.D. Martin, Quarks and Leptons. An Introductory Course
in Modern Particle Physics, John Wiley & Sons, Inc., 1984.
Donald Perkins, Introduction to High Energy Pllysic, The Advanced Book
Program , Addison-Wesley Publishing Company, Inc. , 1986.
R. K. Ellis, "An Introduction to the QCD Parton Model", Fermilab
preprint, FNAL-CONF-88/60-T, 1988.
P. Bagnaia and S. D. Ellis, "CERN Collider Results and the Standard
Model", University of Washington preprint, 1988.
References
Other Applications of
Perturbative QeD
306
307 ( l it '1'1.111' II () II IIII A Pll lk".I,11l1l j
In t his chapter wo wi ll l'xltill in se veral ot h ' l' lI.pp li('II.l.i oWi o f' pc rtlll'b :tt iv('
QC D. ] will no t b' ;tlll to include every irnp rta llt a.pplicati on. I have a.t
temp ted to pick app li cati ons that cover a broad ra ll ge but my choice is PI' j -
udiced somewhat by my own work.
The decays of heavy quarkonium states provide many interesting tl'NI.H of'
QCD . For sufficiently high quark mass the decay rates can be co rn plIl.('d ill
terms of the wave function at (or near) the origin from perturba ti on t1 J('OI'Y. III
the simplest approach one neglects spin dependence and relativistic · (I' 'cl./-1 I.ll d
computes the decay amplitudes to order 0:' •. The lowest order decay Ch;UlIl n! iN
determined from the fact that while both C-even and C-odd states can g ill l.o
three gluons, only C-even states can couple to two gluons. In addition J - I
states cannot decay into two on-shell massless gluons. Thus the ]pc 1-- =
(Is) state should decay primarily into three gluons. It also decays into two
gluons and a photon, but this mode is suppressed by 0:'/0:' • . To leading order
f(ls -+ gg,) 36 2 0:
--e - (8.1.1 )
f(ls -+ ggg) - 5 q 0:. '
and is independent of the wave function at the origin l . Computing the nex t
order correction to (8.1.1) and comparing with data allows for a determinat ioll
of 0:'.2. Existing experiments can measure only high energy photons and t he
total rate f(ls -+ gg,) is estimated from the calculated photon sp ectrum .
The "Born term" matrix element for the decay of a Is quarkoniu rn st;ti.('
into two massless gluons and a photon (Fig. 8.1a) yields a cross sect ion (1I0 J'
malized to unity) given byl
dO' X2)2 (1- Z,)2]
-0'1 dz,dxI = (71'2 1- 9)
[(1 - XI)2
(X2Z,)2
+ (1-
(XIZ,)2
+~-~
(XIX2)2 '
(8 . 1.~)
where XI ,2 = 2EI ,2/W are the fractional energies of the two gluons and z-y
2E-y/W is the fractional energy of the photon and W is the t ot al cen t J'-o f'
mass energy (i.e ., the mass of the quarkonium state) . For massless gluon s th e
range of values of Xl and are z,
o ::; z, : ; 1, (8.1 :1)
1- z, ::; Xl ::; 1, (8 . J .'1)
with
Xl + X2 + z, = 2, (8 . 'I .G)
resulting in the familiar triangular Dalitz plot shown in Fig. 8.2a. To hi gh
accuracy (8.1.2) can be approximated by a constant,
II. I 11 ..~ 10" 1)1'1 Y 30 6
( b)
Figure 8.1 (a) Born term diagram for the deca.y of the upsilon into two ma.ssless
gluons and a photon, T(ls) -+ gg-y. (b) Parton-shower diagram for the decay of
the upsilon . The initial two gluons are allowed to radiate additional gluons which in
turn can Bremsstrahlung more gluons producing two parton showers (or "jets").
(8.1.6)
which means that the decay is given primarily by phase-space alone. Integrat-
ing (8.1.6) over Xl yields a photon spectrum given by
1 du
--d =
U Z-y
111- z ..,
2dx1 = 2z-y, (8.1.7)
( 0)
Z.,
0
0
X,
( b)
I
1_ 82
Zy
0
0 28 I 1+8 2
X,
Figure 8.2 Allowed "Dalitz plot" decay regions for the decay of a Is quarkonilHlI
state of mass W into two gluons of energies Xi = 2Ei /W and a photon of energy
z.., = 2E..,/W for the case (a) where both gluons are massless and (b) where 011 '
gluon has a mass 6 = mg/W.
TCls)-+ gg)'
-L dCT /
;-
;1' -- , ......
\
CT dz)' / \
1.0 / \
/
/
/
,
\
\
/
/
I
, \
\
/
I
0
0 0.2 0.4 0.6 0.8 1.0
Z)'
Figure 8.3 Spectrum of the direct photon energies z-y 2E-y /W in the decay of =
a Is quarkonium state of mass W = 10GeV into a photon and two gluons. The
solid curve is the "Born term" matrix element result for massless gluons and the
dotted curve is the pure phase space ("2z-y") approximation. The dashed curve is
the prediction of a QCD parton-shower Monte Carlo model with A = 0.2 GeV and
Jl.c = 0.45 GeV .
We can estimate the invariant mass of a gluon "jet" by the use of the
parton-shower Monte Carlo approach discussed in Chapter 3. In this case,
the initial gluons produced by the heavy quarkonium decay are allowed to
Bremsstrahlung gluons until their invariant masses have been degraded to
some cut-off mass I-'c. The invariant masses of radiated partons are kinemati-
cally constrained to be less than those oftheir parents with the difference being
converted into the transverse momentum of the emitted partons. The radiated
gluons may themselves radiate more gluons or produce quark-antiquark pairs,
producing a shower of par tons as shown in Fig. 8.1b. The emission of the par-
tons is treated independently using the simple "leading pole" approximation
to QCD and exact kinematics are maintained. The invariant mass spectrum
of one of the initiating gluons in the decay 1(1s) -+ ggr resulting from the
/
31 'hn.pt.,r H Of 10 , A\•••1, /, 0""
J.. dO" 3
0" d8
0 .5 1.0
8 =mg/W
Figure 8.4 Invariant mass distribution of one of the two initiating gluons in til!"
decay of a Is quarkonium state of mass W =
10 GeV into two gluons and II. photon
resulting from a QCD parton-shower Monte Carlo model with A = 0.2 GeV and
Pc = 0.45 GeV.
(8.2.2)
with
(8.2.3)
and
(8.2.4)
/
:11:1 ( ;It \,pl.('I' H (Hil,,, API" ( 'II,l.luH
I ,'
1.!Ji = EJ -j == E'ern)
I (8.2 .!))
and
Ipil = Ipfl = Pcm, (8.2.6)
so that in the center-of-mass frame (8.2.2) becomes
(:~) point
(R .'2 .7)
with
f3 = Pcm/ Ecm. (H . ~ . H)
For a static target the form factor is simply the Fourier transform of th e (" 11 111')1;1'
distribution, p(r),
(8 .'2 .!))
where terms of the order (mass)2/Q2 have been neglected . This eq uali ( II
is illustrated in Fig. 8.6 where x and (1 - x) are the fraction of th e pio ll
momentum carried by the quark and antiquarks, respectively. Th e fUll 'li oll
<fJ(x, Q2) is related to the pion wave function and gives the amplitude fol'
finding the quark carrying the fractional momentum x and the anliqll arlc
carrying the fractional momentum 1 - x within the pion. All the so ft glll OIl
contributions (divergent mass singularities) of the form [0', log( Q2 / m)2]71 ha.v\'
been absorbed (summed) into <fJ(x, Q2) and generate its Q2 dependence (Il()e
Fig. 8.6b). The function T(x, y, Q2) can be considered as the "hard" scatlel'i li P;
amplitude for a parallel qij pair of total momentum P hit by a virtual ph l II ,
,*, of momentum q to end up as a parallel ij pair of momentum pI = P + q.
The amplitude T(x, y, Q2) has a well defined perturbation expansion or lh '
form
T(x, y, Q2) = 0'3(Q2)TB(X, y, Q2){ 1 + a,(Q2)T2(x, y, Q2) + ... } . (8.2. 1:,\)
The amplitude <fJo( u, P), for finding a quark and antiquark with m O Ifl t' 1l
tum uP and (1 - u)P, respectively, within a pion of momentum P is r Inl('d
to the pion wave function and has 16 components, u>. (uP)vJ.l «1 - u )P), wh ell
8.~ 'I'll P in ll 1"0 111 1 I" (~O l 314
p pi
7T 7T
P pi
7T 7T
(C) y*
X
+ +---
(I-X)
Figure 8.6 (a) Illustration of the calculation of the pion form factor. The "wave
function", rP(x, Q2), gives the amplitude for finding a quark of fractional momentum
x and an antiquark with fractional momentum I-x within a pion, where Q2 _q2 is =
the 4-momentum squared transferred by the virtual photon, 'Y • • The shaded region is
the "hard scattering" amplitude . (b) illustrates how the soft gluon mass singularities
are absorbed into rP(x, Q2) and generate its Q2 dependence. (c) Illustrates the "hard
scattering" part of diagram (b) which can be expanded in a perturbation series.
summed over the quark and anti quark helicities A and p. This amplitude can
be written as a four by four matrix as follows:
CPo(u, P) = E uA(uP)ii/'«1 - u)P)fA/'
A,/,
31 ~ C h pt." II Ot.h"l A!lplie t.iOIlIl
y.
p , ?q~P'-Ppi
~>I!===:::i@ >
7T 7T
(vP')
=L
initial
helicities
fi!1
helicities
_---+-:--¥00"/£L--'-"'-:
V\,((I-U) Vu,((I-V) P')
A,A' !-,-,I-':'" r
Figure 8.7 Illustration of the construction of a spin 0- pion from a spin ~ + qua.rk
and antiquark.
= J(I-~u
yI2
- ~
rs¢o(u)P = rs¢o(u)P/v2, (8 .2 . 11)
where the factor J(1- u)u coming from the normalization of the spin rl:l
has been absorbed into the function ¢o(u). Terms proportional to (mass)! P
and kT / P have been ignored. In this limit isospin symmetry is exact giving
¢o(u) = ¢o(l- u). The rs arises because the quark and antiquark have oppo-
site intrinsic parity and we demand that f+- = - f-+ to insure a 0- state.
As illustrated in Fig. 8.7 (8.2.14) allows us to write the amplitude for the in-
teraction of a virtual photon r*, with a pion, r*1I" ---. 11", in terms of a trac
over a fermion loop. In particular we can write
(8.2.15)
with
A(U,V,Q2) = !tr(rs/'Brs"B'), (8.2.]6)
8.2 1'11<1 Pio ll 1"01111 ,,',u'tor 3 16
Figure 8.8 One of the four diagrams contributing to the Born amplitude for '1'* 'II' -+
'11' . The other diagrams correspond to allowing the gluon to interact at the left of the
photon vertex and allowing the photon to also couple to the antiquark.
where Band B' depend, of course, on the details within the shaded blob
shown in Fig. 8.7. The function <Po(u) is an unknown. However, the integral
of <Po( u) over u is related to the pion decay constant, f.,.
In general, the coupling of a virtual photon to a pion is given by
Amp(r*1I' - 11') = ie.,(P'IJ!MIP)
= ie.,F.,(Q2)(p+)(It + ie.,F_(Q2)(p_)(It, (8.2.17)
where
(8.2.18)
and
(8.2.19)
are the only two independent vectors that can be constructed from the initial
pion momentum, P, and the final pion momentum, P', and the 4-momentum
transfer q. The quantity e., is the electric charge of the pion. Conservation of
the electromagnetic current, J!M, demands that F_(Q2) be identically zero
and hermiticity of J!M means that the pion form factor, F.,(Q2 = _q2), is
real.
We define the Born contributions to the pion form factor as the sum of
terms of order a. not including any contributions of the form a,[a.log(Q2 /m2)]n,
:lJ 7 ( hll.ptor Ii Ot hl1. AI'pIlCII.L\()1II1
whi ch are, however, t '(' /Ill i 'ally flhe same orde r Il.Ij tv • . 'I'here /.l.1' four t; rmll o f
order Q. contribut.in g. Olle diagram is shown in Fig. 8.8 and its contl'ibuti n to
the /,*71" -+ 71" amplitude is given by (8.2.15) with A(u,v,Q 2) = A Bl (u,v,Q2),
where
ABl(U,V,Q2) = ~ [itr(TaTa)] ieqg;
tr [/'5f'/'P,(rf + ufhet/'5f/'p,] /[(q
+ uP)2e], (8.2 .20)
I
where pI = P + q and k = (1 - v)P - (1 - u)P and where eq is the Ch lll'g(!
of the quark. The term in brackets is the familiar color factor
itr(TaTa) =~. (8 .2.2 1)
Evaluating the trace yields
(8 .2.22)
where the quarks, gluons and the 71" have been taken to be massless (i .f'. ,
p 2 = p I2 = 0) and where the strong coupling constant, g$ is related to Ct. us
usual by Q$ = g;/471".
The diagram with the gluon interacting with the quark before it does
with the virtual photon is given by (8.2.22) with Pet replaced by P~ and the
interchange of u and v. The remaining two Born graphs are arrived at by
allowing the photon to couple to the antiquark. This means that one sums
over quark charges e q leaving e,... Thus, the complete Born amplitude is given
by (8.2.15) with A(u,v,Q2) = AB(U,v,Q2) and
2 4 ie,..1671"Q$ I
3Q2(1- u)(l- v) (P + P )et.
AB(u, v, Q ) = (8.2.23)
where
(8.2.25)
The Born term result is not very significant since at the next order and
in higher orders there are soft gluon divergences (mass singularities) of the
form Q$(Q2)[Qs(Q2)log(Q2/ m 2)]n, where m is the mass used to regularize
the divergence. All these terms are of the same order and must be summed
to give the complete order Q$(Q2) result. These gluon mass singularities are
summed and absorbed into the unknown function 4>o( u) in (8.2.24) resulting
in a "renormalization group improved" wave function, 4>(x, Q2), similar to the
procedure used for the quark distributions, G(x, Q2), in Chapter 4.
B.2 '1'1,,' !' IOU F'JlIII I" !'tor 318
(0)
u x
q
q
(I-u) (I-X)
(b)
cp(x,Q2) = 1T)
+ 1T)
+ •••
Figure 8.9 (a) The quantity V9f-9f(U, x) represents the leading log forward am-
plitude for a quark and antiquark of fractional momentum u and 1 - u, respectively,
to obtain fractional momentum x and 1 - x after interacting with a gluon . (b) Il-
lustrates how the "wave function" iP(x, Q2) is built up from an infinite number of
"soft" gluon exchanges.
The leading log sum is most easily performed in an axial gauge where
the divergent terms take on the ladder structure as shown in Fig. 8.9. One
can define a kernel, V9f-9f( u, oX), which represents the amplitude (leading log
part) for a quark with fractional momentum u and antiquark with (1 - u) to
become a quark with fractional momentum oX an antiquark with (1 - oX) by
/
:JJt) C h ll,p l,,' , II (HII," Apll l n ti,looII I
th e ex '1lIlI l " o f 11 ~ I' 1011 :tH HhoW II i ll I"i y; . 8.D/I.. 'I'll(' c() lIlpld(' HlllIl , 1/) (!,:, CJ ~ )'
is arrived a t ill a 1I 1/\,IlIl Cr Hilllil ;\r t th e III th d ill Chapl.t'r :1 !llld is g iwll hy 5
A *B = 11 dz A(z)B( z ), (R .:2.'I,7)
Vqq-+qq(U'X)=~{I-U
3 I-x
(1+_1_)
u-x +
B(u-x)
The quantity K, in (8.2.26) arises from integrals over the loop momentum wit.h
the running coupling constant at the vertices as in Chapter 3 and is giv n by
(8 .2.30)
with 130 defined in (1.2.19). From (8.2.26) we see that </I (x , Q2) satisfies th e
following differential equation,
(8.2 .3 1)
The complete order a. 2 expression for the pion form factor is then given by
(8 .2.33)
where A;;s turns out to be the same non-singlet anomalous dimensions given
in (4.7.6),
(8.2.34)
Jro m
1
2 _ 4(2n + 3) 2
¢n(Q ) - (n + 1)(n + 2) dx Cn (2x - 1)¢(x, Q ), (8.2.35)
Given the wave function, ¢(x, Q6), at some Q6 one can use (8.2.38) and
(8.2.40) to compute it at any other Q2 provided one knows the perturba-
tive parameter A.
One can make use of the simple factorizing form for T( x, y, Q2) and write
the order a s (Q2) form factor as
(I-x)
Q2) 12 (8.2.41 )
:l 2 1 ( ;II " pl,,· , Ii (>l.I. fl ' I\ l'pl lo 11 1111 11 1
(8 .2. tJ 2)
Except for n = 0 all t erms die off like fractional powers of a. (Q2)/l"I',(Q;,)
log(Q6/A)/log(Q2/A2) (e.g. d1 = 0.0427 with nj = 4). This m a ns I.h ll l. lUi
Q2 becomes large F" (Q2) approaches the first term in (8.2.43)
for n =0
2
¢In(Qo) = {Co0 (8.2 .tJ G)
for n > 0
For this choice ¢lex, Q2) in (8 .2.40) does not evolve with Q2 (remember A[is
0) and is simply given by
¢lex, Q2) = Co x(l- x) . (8.2.t17)
In this case F,,(Q2) is equal to its asymptotic value at all Q2. Namely,
F (Q2) - 4
" - 7r 3
i a.(Q2) Ie 12
Q2 0,
( .2 .tJ8)
(8.2.49)
For pions this constant can be determined from the weak decay amplitucl
71" -+ /1V . Nam cly,
¢o = J3 J" (I" ~ 93 Mc V) . (8.2.50)
Co = Y3 j", (8.2.51)
and
(8 .2.52)
For other choices of ¢(x, Q6) this value is approached (very slowly) as Q2 -+
00. If ¢(x, Q2) is more peaked at x = 1/2 than (8.2.45) than the limit is
approached from below. If, on the other hand, ¢( x, Q6) is less peaked at
x = 1/2 than (8.2.45), then the limit is approached from above. In either case
the approach is quite slow.
Problems
8.1. Show that the "Born term" double differential cross section for the decay of
a quarkonium Is state into two massless gluons and a photon is given by
1 d(T -
____ 1 [(1 - Xl):! + (1 - X2)2 +.o...,-_-,--,=-
(1- z,)2]
(T dz,dxl - (71"2 - 9) (X2Z,)2 (XlZ,)2 (XlX2)2'
where Xl,:! = 2El,2/W are the fractional energies of the two gluons and
z, = 2E, /W is the fractional energy of the photon and W is the mass of
the quarkonium state. Show that to a high degree of accuracy this differential
cross section can be approximated by
8.2. Integrate the double differential cross section in (8.1.2) over the allowed range
of Xl and verify (8.1.8).
8.3 . Show that the point (or Mott) cross section for the scattering of electrons off
a structureless target is given in the center-of-mass frame by
(:~) point
where
f3 = Pcm/ Ecm.
8.4. Show that, in general, the coupling of a virtual photon to a pion is given by
where
2 4 167r
TB(U,V,Q )='3Q2(1-u)(1-v)'
8.7. Show that the asymptotic behavior of the pion form factor is given by
Further Reading
G. Altarelli, "Partons in Quantum Chromodynamics," Physics Reports
81, 1 (1982) .
F. Halzen and A.D. Martin, Quarks and Leptons. An Introdu ctory Cour'
in Modern Particle Pllysics, John Wiley & Sons, Inc., 1984.
References
1. P.B. MacKenzie and G.P. Lepage, Phys. Rev. Lett. 47, 1244 (1981).
2. S.J. Brodsky, G.P. Lepage and P.B. MacKenzie, Phys. Rev. D28, 228
(1983).
3. R.D. Field, Phys. Lett. 133B, 248 (1983) .
4. S.J. Brodsky and G.P. Lepage, Phys. Lett. 87B, 359 (1979). ibid Phys.
Rev. Lett. 43,545 (1979). Erratum ibid 43, 1625 (1979).
5. M.K. Chase, Nucl. Phys. B167, 125 (1980). ibid B174, 109 (1980) .
S-Matrix Form"u las
325
In this book I foll ow close ly th ' Wllvcllti o ll s of Qui g ,.1 wili cli I S !.h e Halli e as
Bjorken and Drell 2 except for th e norm a liz a tio n of Dira.c s pill o n.;,
S{Ja = 8{Ja - 4
i(2:rr)48 (p{J - Pa) M{Ja / VIJ.. (2E i ), (A.I.1)
where Q' and f3 label the initial and final states and the product of factors
(2Ei) is over both initial and final states.
The transition probability for the decay process Q' --+ (1,2, ... , n) is given by
(2:rr)4 2 rrn 3
d pi 4
dW{Ja = 2Ea IM{Jal ;=1 (2:rr)3(2E
i
) 8 (Pi + P2 + ... + Pn - Pa). (A.2.1)
dW
{Ja
= _1_
32:rr2
1M
{Ja
12 Pcmdncm
M;' (A .2.2)
For an initial state Q' = (1,2), where particle number 1 is the projectile and
number 2 is the target, and a final state f3 = (3,4, . .. , n) the differential cross
section is
(A.3.1)
/
3 27
by
du (0)
-dr. S,
1 2 P~m
= -64 -
1M{3Ot,
12 (A .3.2)
Hem 7r S Pem
PI + P2 = P3 + P4, (A .3.6)
:1 2 !:I i\ PP" lIri i x i\ .' Mn,l,ri x I"", <I lid II/I
which guaranlccs l il al
~
S +l + u = ""2
L..- rni . (A .3.7)
i=1
The invariant s is the square of the total energy in the center of mass frame
and is also given by
s = rni + rn~ + 2rn2(El)LAB . (A.3.8)
The invariant t is the 4-momentum transfer squared and is given by
t = tmin + 2PcmP~m(cos(Jcm -1), (A.3.9)
where
tmin = (Efm - E~m)2 - (Pcm - P~m)2, (A.3 .10)
where the initial and final center-of-mass momenta are given by
E cm
_ (s + mi - rn~)
(A.3.13)
1 - 20 '
E cm _ (s + m~ - m~)
(A.3.14)
2 - 20
The flux factor in (A.3.1) is given by
References
1. C . Quigg, Gauge Theories of the Strong, Weak, and Elec tromagll cLic
Interactions, Frontiers in Physics, The Benj a min-C ummin gs P ubli shill ll>
Company, Inc., 1983.
2. J .D . Bjorken and S.D . Drell , R elativistic Qu antum Mechanics, M(' Craw
Hill, New York, 1965.
Metric and Matrices
330
n.l Metric (N II dilll('IISiolls)
n
The mel ri c
9,," ~ (j °° ° - 1 0
- 1
-1
(lJ . 1. I)
or ° 0
(/J . 1.:1)
(D . 1.'1)
(B.4.1)
(i I , :1) , (/1.'1 'L)
wh ere 12 is t li 2 x 2 lIlIil, IIl lttrix a lld 1.11\' Pa.uli IIlal.l'i n 's, (T" s,\I,isfy
{Ui OJ}
J = 28ij J (11 .'1 .:1)
UiUj = i fi j kUk J (lJ .'1 .'1 )
and are represented by
It is useful to define
and
(lJ .'1 .7)
where
( /U ,H)
and
(D.4 ,9)
It is easy to see that
(IJ.4. IO)
In N space-time (one time, N -1 space) dimensions the Dirac m atric s sat. isfy
(lJ ..'). I )
where IN is the unit matrix in N dimensions and J.l and v run from 0 to N - I.
In addition,
1J.11J.1 =N
IN , (fl .5.2)
1J.110:1J.1 = (2 - N)ro:, (1J .5.:j)
1J.110:1{31J.1 = 4go:/3 IN + (N - 4)r0:1/3, (IJ .:H)
1J.110:1/31>.1J.1 = -21>.1/310: - (N - 4)r0:1{31).· (n.5.5)
1
Furthermore, the trace of IN is defin ed as follows :
A free spin- 1/2 particle of mass m and spin s is described by the spinor u(p, s)
which satisfies the Dirac equation
(j - m)u(p, s) = 0, (B .6.1)
while the adjoint spinor
u(p, s) == ut(p, sho, (B .6.2)
satisfies
u(p, s)(j - m) = O. (B.6 .3)
The spin 4-vector s satisfies
s·p= 0, (B.6.4)
s2 = -1, (B .6.5)
and in the rest frame of the particle has the form
(B.6 .6)
(/J ,(i . II )
In forming the absolute value squared of the scattering ampliLud 0 11 1' (' 11 1'11 1111
ters the Hermitian conjugate which can be expressed as
[u(p', s')f u(p, s )]t = u(p, s)f u(p' , s'), ( IJ .n, I !,)
where
-
f = ,of t 10 · (fl.G . IG)
Some examples are
= IJI.'
7J1. (D.6. 17)
= iiJl.II UJl.II' (D .G . 18)
(i,5) = if5, (1J .6. IU)
-~-~
U;.(p)
~
= JE+ m
-2)1171
((t: )t)
E+m X;.
(D .G .2:l)
where
X-! = (~) (B .(). 25)
References
External Lines:
For particles in the initial or final state one writes the following factor :
(a) Spin Zero Boson .......... 1
(b) Spin One Boson ........... fl'(..\)
Here fl'(..\) is the polarization 4-vector for a boson with helicity ..\. For
the case of a massless spin one boson propagating along the i-axis
with 4-momentum kl' given by
(C.l.l)
(C.1.2)
,,(A~O)~ ~ (D (C.1.5)
((.'. 1. 0 )
Vertex Factors:
For each intersection of three (or more) lines at one point th ere is a. vcr(,('x
factor which depends on the structure of the interaction Lagrangia n. SO IlI ('
examples are shown in Fig. C.l.
(a) Three scalar boson vertex
-ig . 1.1 0)
(b) ee-y vertex
( . I. 11)
Here e is the charge of the electron and the fine stru cturc CO II ~ t, i UlI .
a = e 2 /( 471').
(c) Charged spin zero boson-photon vertex
,,
11',
,,
'.------
/
- ig
/
;'f
/
,/
-iey,.,.
P~" 1"",
;'I
<II
,
,,
'1/ ,.,.
Feynman Parameterizations:
/
( :,.1 I,', '.y II II Jill I 11111 .. « ' I)' ,I
r(R + M)
r(R)r(M)
11
O
elx
x R - 1 (l - x)M - J
[ax + b(1 - x) ]IHM'
(C.2. 1J)
External Lines
(a) Gluons .......... (I'(.~)
Here cl'().) is the polarization 4-vector for a gluon with heli ciLy ,\ i tS
in (C.1.2). It satisfies the orthogonality condition
( .:l.I)
and
(·k =0 ( ' . :l. ~ )
a,p. b,JI
gluon IVVVV\IV\J)MNVv -i 80b [gplI+ TJ q~~I1]/q2
q
a b
ghost ------~----- -i 80b / q 2
q
Here I have taken the quark mass to be zero. The indices i and j
correspond to quark color and run from 1 to 3.
(b) Gluons (covariant gauge)
-i 6a b [(91'11 - q,.qll/q2) + j3 q,.qll/q2]
(G.3.4)
q2 + if
or
-i 6a b (91'11 + 1J qlJqll/q2)
(G.3 .5)
q2 + if
Here 1J == j3 -1 is an arbitrary gauge parameter (1J = 0 is the Feynman
gauge, 1J = -1 is the Landau gauge) . The indices a and b correspond
to the gluon color and run from 1 to 8.
(c) Ghost
(G.3 .6)
Vertex Factors:
The QCD vertex factors shown in Fig. C.3 are as follows:
(a) Quark-quark-gluon
. Ta
- t g.'Y1J ij (G.3.7)
<),;1 F y lllll llil "Ul liN Q( J)I - :I 1I4 :1
Triple -gluon
C. II
~
PZ
b.JJ-
PI
0).
Ghost-ghost - gluon
c/',
,
.. p
,~b.JJ-
;"
,
C,
Four-gluon
-ig~ f•• Ie. (~.,~O' -QMgl'-")
The indices i and j correspond to the quark color and run from 1 to :.I
and a corresponds to the gluon color and runs from 1 to S. The eight,
3 x 3 SU(3) matrices, T a, handle the color and will be discussed 10
Appendix D. The QeD coupling constant is given by
g2
a =-2.. (C.3 .S)
41r
(b) Triple-gluon coupling
V>'l.Iv(Pl,P2,P3) = -g8 'abc F>'l.Iv(Pl,P2,P3), (C.3 .9)
wh I' C
Le = (~~ ) L
Lu = (~) L (C.4.l)
where
dlJ =d cos(Bc) + s sin(Bc),
SIJ = S cos(Bc) - d sin(Bc), (C.4.2)
I
e
~ -ieY
fL
Wi
... - ...
...... ... fL -igw YfL (1 - 15)
_ ZOfL
....... .... - ...
ZO
... fL -igz)jL [Re(I+Y5)+Le( 1- Y5)]
........... ........
2
Re =2 sin 8w
2
L e = 2 sin ew-I
Figure C.4 Feynman rules for the coupling of leptons to photons and to W ± a.nd
ZO Bosons in the Weinberg-Salam model.
Vertex Factors:
The lepton-boson couplings shown in Fig. CA are as follows:
(a) eel vertex
(GAA)
q
~ - ieq e)'jL
q
u w±
...... -
.......... ~ -ig wYfL 0- Y5)
AgZYfL [Rq(I+Y5)+LqU-Y5)]
2
Rq =-2eq sin 8w
2
Lq = T3-2eqSin 8w
Figure C.S Feynman rules for the coupling of quarks to photons and to W± and
ZO Bosons in the Weinberg-Salam model.
1 ~
-
j2
(,'v M ,;
", (G.'1 .8)
a nd
MZ
2
= Mw/(l
2
-
2
XW)' (C A .!)
(d) eeZ o vertex
( .'1 . 10)
with
Re=2xw,
Le = 2 Xw - 1. «,' ,'1 . 1 1)
The quark-boson couplings shown in Fig. C.S are as ro ll o w ~ :
(e) qq-y vertex
(r..:. tI .1'l)
(f) ude W or CSe W vertex
(G'. '1 . I:1)
where the dimensionless weak coupling gw is given in (CA.S) .
(g) qqZO vertex
( '.'1 . 1'1)
with
Rq = -2eqxw,
Lq = T3 - 2eqxw, ( ~' . '1 . 1 5)
where T3 is the third component of "weak-isospin" of the qUttri , 1/ ,
and e q is its electric charge (measured in units of the elect ron li n.rg(·
e and where Xw is the square of the sine of the Weinb erg a.ngle.
References
1. D.J. Gross and F. Wilczek, Phys. Rev. D8, 3633 (1973). ibid D9 , 980
(1974).
2. H.D. Politzer, Physics Reports 14C (1974).
3. S. Weinberg, Phys. Rev. Lett. 31, 494 (1973) .
4. S. Weinberg, Phys. Rev. Lett. 19, 1264 (1967).
5. C. Quigg , C;w gc ')'/l cori cs of (,II(' S(,r o ng, 11\I(,l1.k, H lld /';/c'c/ f()1I1i 1 ~ 1I (' (, ic
Interactions, Frontiers in Physi cs, The BelljFl.l1lill -C ullllllill gs Pllblishing
Company, Inc., 1983 .
/
SU(3) of Color
D.l Structure Constants and Color Matricc
(l T,=i 0
1
1 -z
Tl =-
2
n' 0
0
0
0
n'
1
T3 =-
2
0 n'
0
-1
0
1
T4 =-
2 (;
0
0
0
n,
1
T5 =-
2 (; ~}
0
0
0
T6 = Ie n
2 ~
0
0
1
G ~} 0q
0 0
1 1
T7 =- 0 Ts = 2..;3 1 (D .L3)
2 -2
0
The structure constants d abe are defined according to
{Ta, Td = i 8ab + dabe Te· (D.L4)
The T a matrices satisfy
FaFa = 3, (O . U I )
tr(F aFb) = 3 Dab, (0 . 1. :2:2 )
faeddbed = 0, (1) . 1.:2:1)
FaDa = 0, (0. 1.:2 /1)
tr(FaDb) = 0, (1 .1. :25)
daebdbed= ~ Oab, (D. 1. 4(;)
DaDa = ~, (D. I .27)
tr(DaDb) = ~Oab, (D .l .28)
= i~ fab e,
tr(F aFbF e) ( D. 1. 29)
tr(DaFbFe) = ~dabe , (O. U O)
tr(DaDbFe) = i~ Jabe, (I) . I .:11 )
tr(DaDbDe) = - t d(jb e, ( /). 1.:12)
:l I)Z ApJlt:lldi x () SU(J) oj (;01111
(0) Tr (To To )= 4
I I
(b) 8x3 Tr(To To )= '6
k
b
(c)
(d)
Figure D.l Color factors for the square of the amplitudes (a) 1'* -+ q + q, (b)
9+ q ...... q + 1', (c) q + q ...... q + q, and (d) 9 + q ...... 9 + q.
(D .1.33)
D.2 EXa1l1p,l e -<ol()l" J:iad,o.'s
As is th e case with particle Sp illS, w ch ose to SlIln over th e fili a l ('.( I r !ita!.('"
and average over initial color stat . T he co lo r ract r r r th e sq II [1,1'(' r 1.i1( '
1'* -+ qijg amplitude shown in Fig. D.la is given by
(1) .:2 , I)
SInce
Oaa = 8.
Similarly, the color factor for the square of the gq -+ 1'* q amp litlld e Hhc WII ill
Fig. D.lb is given by
I 1 T aij T jai -- 24
83 1 t (T T ) _ 1
r a a - '6' ( IJ ,'.U )
where the factor of 1/24 arises from the initial state color ave ra.gill g.
The color factor for the square of the qq -+ qq amp litude Hil owlI ill
Fig. D.lc is given by
H TijTJiTklT~k = i tr(TaTb)tr(TaTb)
=i OOab) (~Oab)
_110 _2 (D .V I)
- 94 aa - 9'
The color factor for the square of the gq -+ gq amplitude shown ill Fig , O. l<l
is given by
~ ~ fabc!bad Tij Tji = i4 fabc!bad tr(Te T d)
~ Oed fabefbad
1
= 24
1 1
= 48 fabc!bae = 48 tr(F aF a)
- 1 3 (aa _ 1 (1) .'2 .0)
- 48 u - 2'
where (D.1.8) and (D.1.22) have been used.
References
354
/
AI'p (' "ti, x I'! . I"" 1101 F"". I,III "
Gamma Functi on
5
4
3
2
-4 2 3 4
-I
-2
-3
-4
-5
-6
lim f(X +
( f(x+b)
y)) = x a- b (E .l.11)
"'-+00
In addition,
f(B)f(M)f(C - B)f(C + M - A - B)
(E.l.13)
f( C + M - A)f( C + M - B)
where the latter is valid provided Re(C) > 0, Re(B) > 0, and Re(C + M -
A - B) > O.
The first and second derivatives of the gamma function evaluated at x = 1
are given by
r'(l) = -"IE, (E.l.14)
f" (1) = "IE 2 + ! 11"2 , (E.l.15)
respectively, where "IE is Eulers constant which satisfies
and
"IE = -1 00
e-
X
log(x)dx. (E.1.17)
From (E.1.14) and (E.1.15) we have the following Taylor series expansion of
r(l + E),
f(l + E) = 1 - "lEE + 2"1 ( "IE 2 11"2) E2 + ...
+ (; (E.1.18)
Liz(x) Re[Li,z(x)]
• -t ...... ..........-
.
·-l~d' --.lo.-
z
= til .'
,..
2. "'i1'
-4 -, -z h
\
2 , 4
\
\
-I \
\
\
\
-2 \ 1m [Li 2(x)]
\!.-i
,, .-109 (x)
-,
",,,
Figure E.2 Sketch of the behavior of Li2{X) ver"u.9 x .
E.2 Dilogarithm
or
L12
o t
-1%
log(l - t) dt,
· ( X ) -- (E .2.2 )
· ( X ) --
L12 -
11 o
log(l - xt) d t,
t
(E.2 .3)
Strictly speaking "+ functions" are distributions that are well behaved only
when convoluted with a smooth function that vanishes sufficiently rapidly as
x -+ 1. They have the property that
11 (F(x»+ dx =0 (E.3.1)
where
B(y) = 0 for Y::; 0 (E.3 .3)
B(y) = 1 for y > O. (E.3.4)
Two important "+ functions" are
( 1)
1- x +
= lim {_1_ B(l- x - (3)
1- x
fJ ..... O
+ log((3) 6(1- x - (3)} (E.3.5)
and
(
log(l - x»)
1- x +
= lim
fJ=O
{logl(~ ~ x) B(l - x - (3) + ~ log2((3) 0(1 - x - (3)}
(E.3.6)
/
10 4
9
8
(0) (r:'x ). 3
2
(b) (I,~IK-K») +
11oJII3l
7
6
5
4
3 -3
2 -4
I -5
0 -6
-I -7
-109(13)
-2 -8
-3 -9
-4
0.0 0.2 0.4 0.6 0 .8
-I<i)
1.0 .0 0.2 0 .4 0 .6 0 .8 1.0
II II
Figure E_3 Sketch ofthe behavior of "+ functions" l/(l-x)+ and (log( I - x)/(J - T) )
versus x.
which are illustrated in Fig. E.3. When convoluted with a well behn.v d flln
tion G(y) one arrives at
1 z
1 dy G(zjy) = G(z) log(l _ z) +
y (1 - y)+
11
z
dy [G(z/y) - yG(z)]
y (1 - y)
(£'.:.1.7)
and
+ 1Y
z
1 dy [G(zjy) - yG(z)] log(l _ y) .
(1- y)
(E .3. )
1 o
1
dx x(l-
~
x)C~2)(2x -
~
1)C;"')(2x - 1) =
{
(n
0
+ 2)(n + 1)
4(2n + 3)
n=m.
(EA.2)
The Gegenbauer polynomials can be deduced from the following recursive
relation:
(EA .3)
where
c~t\z) = 1, (E.4.4)
C~ t )(z) = 3z. (EA.5)
It is useful to note that 1/( 1 - x) can be expanded in terms of Gegenbauer
polynomials as follows:
_1_ =
I-x
f 2(2n + 3) C~t\2x _ 1).
n:o(n+l)(n+2)
(E.4.6)
/
Ind
:Ifl l
302 ludex
Deep inelastic scattering, 126, 176, 187, .F.F pararn cteri»atioll, 5!)
190, 209, 251 Field amplitude, 24.0
Delta function contribution, 191 Final state, 57, 76
Diagonalizing, 144 Fine structure constant, 242
Differential cross section, 326 Flavor dependence, 61
Dilogarithm, 31, 357 Flux factor, 115
Dimensional regularization mass, 43, Fourier transform, 60, 313
67, 71, 242 Fragmentation function, 58, 61, 63, 69-
Dimensional regularization, 37, 42, 66 70, 78, 80-82, 84, 86, 138, 140-141,
Dimensional regularization, 132 148-149, 186, 248, 251, 274, 282, 289
Dimensional regularization, 138, 184, Fragmentation, 16
186, 226, 238 Fragmentation, 58
Dimensionless coupling, 43, 45
Dimensionless variable, 174
Gamma function, 43, 355
Dimensionless weak coupling, 206, 346 Gauge dependent, 75
Dirac algebra, 332 Gauge invariance, 23
Dirac equation, 235, 276 Gauge invariance, 276
Dirac matrices, 332 -333 Gauge invariance, 276
Dirac spinors, 326 Gauge invariance, 317
Double logarithmic divergences, 97 Gauge parameter, 33
Double logarithms, 204 Gegenbauer moments, 320
Double photons, 266 Gegenbauer polynomials, 320, 359
Drell-Yan Model, 171, 176 Geometric series, 242, 245
DY distributions, 251 Ghost contribution, 230
DY structure functions, 191 Ghost loop, 230, 281
Ghost particles, 24
Ghost particles, 342
Effective coupling, 6-8, 226, 240, 244 Ghost-quark scattering, 280
Eigenvalues, 144 Gluon distributions, 140, 142, 149-150,
Eisenstates, 335 152
Eisenvectors, 335 Gluon fragmentation function, 69, 87
Electromagnetic current, 316 Gluon jet, 96
Eulers constant, 46, 84, 356 Gluon jet, 97
Event shapes, 292 Gluon loop, 229
Event topology, 295 Gluon polarization states, 279
Evolution equations, 151 Gluon propagator, 8, 226
Experimental charge, 8 Gluon-gluon fusion, 269
Exponential series, 101, 181 Gluons,2
External lines, 240
Hadronic tensor, 114
Hadronic vertex, 154
Factorization of the mass singularities, Hadronization, 292
71 Helicity basis, 335
Fermi constant, 154, 206, 346 Hermitian conjugate, 335
Fermion loop, 315 Hierarchy, 59
Feynman diagrams, 248 Higher twist, 12, 121
Feynman gauge, 23, 33, 39, 75, 229, Higher-order effective coupling, 244
232, 279, 342 Hole, 292
Feynman parameterization, 33, 226, Hot jet core, 298
234 Hypercharge, 63
In ci llsivl' Hi Il KI.. " a,droll ( ' IiIIH ,W,
I,i(l ll , ()~) I,.. pl.o ll 1I<IHIl II ( ' 0 11 pli "14", :11 :)
Ind c pClld (, ll1. t' lll isHio ll , 711, !J:l L.. pl.\) lli(' 1," 11 140 1' , 1 1:1
In f ra red d i v rge ll 'c , 2~) , 97, 235 Lig ht ('0 11 \' va,ri ll.hkN, 7:1
Infra red sill g lli arities, 57, 121 Light-lig ht HC;l.l.kri Il K, :l()7
Initial state color, 173 Litllc f rllll ct io ll s, (j7 7(), 7H HO, I:!H
Initial state quark s pins, ] 22 129, 136, 138- 1:)9 , 150- 15 1, 1:':', I HIi
Interference term, 75 ] 87, 195,25 1
Intrinsic parity, 315 Longitudillal c ro,'s scctiO Il , ll ei
Intrinsic transverse momentum, 283, Longitudill al 1I1 0 1l) CIlI.IIIIl 'o Il H,' rvill.i ' "I ,
289 264
Invariant amplitude, 326 Longitudill al p hOtO Il ':), 128, 1:11
Invariant differential , 265 Longitudin al s tate, 338
Invariants, 327 Longitudin al s trllct llrc flill c ti o n, I I H
Isoscalar target, no, 157 Longitudinal s tru cturc rllll cl. io ll, 1'211,
Isospin symmetry, 63, 109, 315 132,136, 152, 157- 158