KAON P RO DUCTION IN p —p COLL I S I ONS

Squaring (1), averaging over spin, and performing the The inelasticity of the off-shell Ap scattering is
I&hase-space integration, we obtain for the total K- —
accounted for in (2a) by a correction factor viz. , the
production cross section at threshold ratio of the final to initial Ap c.m. momentum. This
factor, which multiplies the elastic 1tp cross section,
approaches unity at the higher energies but is important
near threshold.
The cross section is estimated using an eGective-range
approximation for o, , t(rt). Results, using efFective-range
I, 2It, 2a parameters obtained from both hypernuclei data and
AP scattering data, s are shown in Fig. 2. The lowest-
(e —rt
&/& energy IC-production data known to us occurs at
nl a., (n)dV, (2b) Eteb —2.807 GeV/c, ' which is well outside the region of
4 snx validity for this calculation. Further experimental work
is needed in the region of Pl b 2.40 GeV/c.
W = 2N yrrtsr+ e = 2 (p'+Hdtv')'"
It is clear that the above analysis also applies to the
= 235tv (1+Tt~b/2Mtv) t", (2c) process pp-+ pZE; however, since the square of the
pZE' coupling const:ant is an order of magnitude less
where a, , t(rt) is the singlet (triplet) total Ap cross section than that for pftE, ' the pred'icted cross section near
at total c.m. kinetic energy tt, and p is the c.m. momen- threshold is probably too small ( 0.2 ttb) for experi-
tum of the initial protons. ' mental veriYication.
' G. Alexander, O. Benary, U. Karshon, A. Shapira, G. Yeku-
R. Engelmann, H. Filthuth, A. Fridman, and B. Schiby,
tieli,
6 Equation {2a) is formally diRerent from Kq. (8a) of Ref. 1 Phys. I.etters 19, 715 (1966).
because relativistic kinematics are used here; also, we include a W. J. Fickinger, E. Pickup, D. K. Robinson, and E. O. Salant,
factor, 4{q/mz)'", erroneously omitted from Fq. (8b) of Ref. 1. Phys. Rev. 125, 2082 (1962).

PH YSICAL REVIEW VOLUME 179, NUMBER 5 25 MARCH 1969

Asymptotic Sum Rules at Infinite Momentum*

J. D. BJORKEN
Stanford Linear Accelerator Center, Stanford University, Stanford, California
(Received 30 September 1968)

By combining the qo ryo~i —

method for asymptotic sum rules with the P &cc method of Fubini and
Furlan, we relate the structure functions 5'2 and 5'& in inelastic lepton-nucleon scattering to matrix elements
of cpmmutators of currents at almost equal times at infinite momentum. We argue that the infinite-momen-
tum limit for these commutators does not diverge, but may vanish. If the limit is nonvanishing, we predict
vW&(v, q') —+ fs(v/q') and Wq(v, q') ~
fi( /q') vas v and q' tend to ~. From a similar analysis for neutrino
prpcesses, we conclude that at high energies the total neutrino-nucleon cross sections rise linearly with
neutrino laboratory energy until nonlocality of the weak current-current coupling sets in. The sum of r P
and l p cross sections is determined by, the equal-time commutator of the Cabibbo current with its time
derivative, taken between proton states at infinite momentum.

L INTRODUCTION techniques of current algebra. ' In particular, Cornwall

' 'NELASTIC lepton-nucleon scattering at high-
- - momentum transfer is a very direct means of probing
small-distance nucleon structure. Rejecting this fact
is the profound state of theoretical ignorance on what,
even qualitatively, can be expected in this process. '
Some small inroads have been recently made using the
* Work supported by the U. S. Atomic Energy Commission.
'For a good example, see J. Bjorken, in Proceedings of the
"
International School of Physics "Enrico Fermi, Coarse 41, edited
by J. Steinberger (Academic Press Inc. , New York, 1969};
Stanford Linear Accelerator Center Report Xo. SLAC-PUB-338
(unpublished).
and Xorton' have written down a large class of asymp-
totic sum rules, valid at large g', for inelastic electron
scattering. Of these, the sum rule of Callan and Gross4
relating an asymptotic integral over electron scattering
cross sections to a piece of the commutator of electro-
magnetic current with its time derivative is of special
interest. The purpose of this paper is to discuss such sum
rules in a slightly different language that of the inhnite-
' J. D. Bjprken, Phys. Rev. 163, 1767 (1967), and references
quoted therein.
3
J. M. Cprnwall and R. E. Norton, Phys. Rev. 177, 2584 (1969).
4 C. G.
Callan, Jr. , and D. Gross, Phys. Rev. Letters 21, 311
{1968).
 1548 J. D. BJORKEN

momentum method. We show that the electron scatter- with a similar expression for F~ or 8'2. However, the
ing data is related in a direct way to matrix elements of content of these results is more succinctly discussed in
electromagnetic current commutators at infinite nucleon terms of (1.1)—(1.4).
momentum. '»' In particular, we find that the structure Although a straightforward generalization of these
functions W2(q', v) and Wi(q', v) describing inelastic relations to di6erent currents and momentum states is
scattering7 tend to simple limits for large q'. not dificult, what is not straightforward is the inter-
pretation of the almost equal-time commutators appear-
lim vWp(q' v) =MF2( q'—
/Mv), (1.1) ing in (1.3) and (1.4). In particular, the spectrum of in-
fixed
termediate "frequencies" co = —
oo, y/q&
q'/M v is bounded above,
lirn MWi(q', v) = Fi( q'/M—v), (1.2) corresponding to at most the intermediate energy ap-
g2~oo, Vip& fixed propriate to the single-nucleon Z diagram (see I'ig. 1).
with Assuming that the limit (1.1) and (1.2) is nontrivial
—z 00
(nonvanishing), it will be most interesting to construct
F,(co) —
= Fr(to) = lim
Pz ~to
dv. since T d'x models with the kind of asymptotic behavior expressed

&&&P*ll:J*(x, /Pp),
p

. J*(0)jIP.&, (13)
in (1.1)—(1.4). This, however, is beyond the scope of this
paper.
In Sec. II, a simple derivation of the result is given.
F2(&o) i Section III remedies the swindle perpetrated on the
— —F, (~) = — lim d7 slnQ)7 d S reader in Sec. II, by providing a more honest deriva-
GO tion. In Sec. IV, we attempt a generalization to an arbi-
trary kinematical situation. In Sec. V, we apply the
&((P I I J (x,r/Pp) J (0)DIP ) (1.4) same method to neutrino reactions and find that vp and
—
where co= q'/Mv. The existence of these limits (aside vp total cross sections should rise linearly with energy.
from the Brandt-Sucher disease') is guaranteed by a The sum of vp and vp cross sections is determined by the
finite value of the integral appearing in the Callan-Gross equal-time commutator of the Cabibbo current with its
sum rule. Although the present data' appear to indicate first time derivative. Section VI summarizes our
that Ii2 is nonvanishing at q'~ 1—2 BeV2, it is still pos- conclusions.
sible that Ii2 — + 0 and the infinite-momentum commu-
ta, tors in (1.3) and (1.4) vanish in the limit. In such a II. SIMPLE DERIVATION OF THE
case the content of this paper is empty. ASYMPTOTIC LIMIT
Sum rules such as Cornwall and Norton have written
down' may be obtained by taking the sine transform of
The inelastic scattering cross section from an un-
polarized nucleon may be written' as
(1.3) and (1.4) and expanding both sides in a power
series in r. For example, for n= 1, 3, 5-
Iq'I "+' " dv dQdE' 4E' sin4(2tH)
dto to"Ft(to) = lim Wi(q', v)
M vn+2
&&LW~(q', v) cos'(-', ll)+2Wt(q', v) sin'(-', 8)$, (2.1)
= lim ( —i)"/2 where E', E' is the energy of incident and scattered elec-
Pz~oo

8"J,(x, t)
P n tron, 0 is the scattering angle of electron, q2 equals
4EE' sin'(-,'8), v equal—
s q P/M= (E E'), P is
momentum of target nucleon, g is the momentum of
the-
I, Bf"
, J. 0 I',
]=p
virtual photon, and

g=1 3 5 7
I'„— — I'„—— 8'2 q2 V

~ S. Fubini and G. Furlan, Physics 1, 229 {1965).

e R. Dashen and M. Gell-Mann, in ProeeeChlgs of the Third Coral
Gcbles Conference on Symmetry I'rincip/es ut High Energy, edited
by B. Kursunoglu, A. Perlmutter, and I. Sakmar (W. H. Freeman g PP' g2 p
and Co. , San Francisco, 1966). g
' See, e.g., T. de Forest and J. Walecka, Advan. Phys. 15, 1
(1966); and Kq. (2.2).
—
I'p

~ ~ (P J.(0) ~) J.(0)1»(2~)'&'(P. P
8 R. Brandt and
J. Sucher, Phys. Rev. Letters 20, 1131 (1968);
and (to be published).
= I I &~ I
—— q)
n
'E. Bloom, D. Coward, H. DeStaebler, J. Drees, J. Litt, G.
Miller, L. Mo, R. Taylor, M. Breidenbach, J. Friedman, G.
Hartman, H. Kendall, and S. Loken, report to Fourteenth Inter-
national Conference on High Energy Physics Vienna, 1968
(unpublished) and numerous private communications which we (2.2)
gratefully acknowledge.
 g ULEMA AY gprN»E Mo

function
g 1 ilg] e=l1 LICtcoll
y diagram.
~
P
q

a matri»
n all
.elements ( I.. . . p), .an I
average ov
Ig„„—
lied. Now d consl er t e limit. of q2 )
(2.2) as po~ q0 '
~

~+~
q2/~&) Notice that g'
,.
2
(timelike) in this
limit. ;,'8(.,)(ply„(*),~,(o)jl»
( hoosing ++0~ v+ 0 we fiild M
+polynomials in q, (3 1)
P'Pf Pp
W, +h, ;Wr
I z~
d s ere, as always in thisis Paa er an average over nucleon
M2 '
s in is implie d . A ccor
ording to the dictum of Harariri 10
x -"-"(pll ~, (.,p.~/p. ),~, (o))l», n unsubtracted ~ ~

commutator
1.1 and (1.2) and the assumption
ion that the
van i shes outsi 'd e thee light cone,
spacelike.
dispersion relation in v,
w le T1 req uires one subtraction, provi e i g,
1 "d2" ImT2(p', q2)
—8„28, F2(00) + h@F)(u)) T2=- (3.2)
2
7l 0 v v —z6
— '"(pILJ'(», rlp0), ~ (o)jl p) T2(2, q2) = T2(o, q2)
= lim d'x ~
— 2' "dp" ImT&(2, q2)
m d'x d7- sincor p V
2 —v2 —ze

x (pllJ, (», /po) J (o)&I» ( «) ( 4)

We choose to express th T in
d =(— g'/M. ). Us gth«ct th t
e;
i terms of the variables g2

~ ~

nes F1 and F2 for positive co as well as

'( )
e therefore we let co= g —ImT;(2, q2) = W;(p,
propria' tee for inelastic scatterin
ttering . Then we get q ) (3.4)
we obtain
F,(co) =F2(a)) = lm ' Ck) "W2(co', q2)
gz oo
Tr(~, q') = T2(~, q—
x(pIP*(, /po), J*(o)ll p ) ((u) 0) (2.5) 0 6& Gl

and ' d(o"W2(s)', q')

T2(Gl&q ) = —2 CO
(3.6)
p GD

0
We now take t h e l'imi't ggp ~i~ q fixed and P, tem-
&O) (2 6) porarily fixed. In this limit
sired result given in (1.3) and (1.4). Both
g p ositive; notice
t e c h
between t e tra nsverse and ongi u in
Pp
l have notice es
bee perpet a ted in letting ~ —or. ere ~ P2 4

0.
Section III is devoted to providing suc h fi
' d(o'
III. JUSTIFIICATION OF THE RESULTS T2 ~+ W2(co', q2) .
o. ~ o p.oovide a better derivation o the
t p reced-
d-
ing resu lt s, w e consider the covarian curr - o-1 H. Harari,', Phys.
Ph . Rev. I.etters 17, 1303 ((963)
 1550 D BJORKEN
On the other hand,d from (3.1), which implies

—w, &
/3
/

Iqol —w, &

dv

/
/

(3.12)
+polynomial
In other words,
imp imp
[J„(x,o),J„(0)j
4
do~(P I
lim doe'2wi((u' q'-) & ~, (3.13)

d'x(P f8J„(x,t)/itt,
I
J,(0)]IP)~=o+
g=o (3 &) lim I
q'I
dG)
Wo((v', q') & ~, (3.14)
+polynomial
an d because &I/i annd RI/2 are positive i't
Specia izingg
find
topandv/ 0 and
*
in thee s direction, we follows that for cu 2 I I
) se
semidefinite,

' dho"Wi(a&, q')

P
do&(PI[a
do+ .
J.(x,. t)/at, . J.(0)
0 P), =,o+polynomial jl, 1
y, ((u) = lim
G7 CO
/2
-Q 0Q (3.15)

= y uation exists through-

lim s'q'Ti(~, q')+sPo'
lane, barring ex rem

J.(0)jlP, =,
' '
behavior of I4'» in the limi . im ve im,
d ~(PI [W, (x, t)/at, 4
M doe'owo(oi', qo)
y, (oo)= hm Iq IM' (3.16)
,J,(0) P
3f throug ou ne. These results are t en suffi-
jD 2
dc' t to guarantee the existence o
= lim iq2
—
IVo(oo', q') . -time commuta or
q2-+ OQ

OO xed, cv = —gp
qo/Po ——i qo/Po
a assume that
Hereafter, wee sshall I

fixed.
4d /2
'
lim lq'I Wo(co', qo) pp doe"Wo((v, q')

«"(~"+l~'I)
= lim 2lqol
oo

—
d
W (.', q') & . (3.10) ' do~ "W, (a)', qo)-
+~v 2't(~, q')—
' (~"+ ~'I)— I
For fixe q, ion in v converges,s iif the Harari
h In terms of Fi an
and Ps, we find, from ( . 11, (1.2), and
'
(3.1),
.I hkd
scattering is so large a Gross integral does I',I'; d( "P,(~')
T'*
not exis; s umstance wou oversaturate
o 'l(~"+l~'I)

+'"" ~, ."~I.
I

sum ru es from curren More

1ikely is t e . If the Callan-Gross in- 1 4d "F,(~')-
1.3 and (1.4) necessan y ex I

b f.
of t n
definitions in terms o e se n ~'&PI[J'(x, /&), J, (0)jIP)
"
sections that in our limit
o
p
e ~

+polynomia .
2

Wo 5
+ q''l&&+ ) (r,
(3 11) &967 . The connection bee tween
w S' 8' and gg, o-~ is

'tion o
7 1nternuti n~l ~~mpos~~m
at High Energies (Stanfor i
o. g
osium oe Electron an
a, . Hand, in Proceedings of the
Notice t h a t F& and Fi, Eqs. 1.
and gTt in the limit: 4~ n g p
and (1..4),, are proportiona
1 3 an rtiona
4 ' li+"/I 'I) '

o o.
&&
&
179 SUM RULES AT I NFI NITE MOMENTUM
In the limit, consider the case of two diferent SV(3)&(QU(3) cur-
rents (I?I,I?2) sandwiched between spin-zero hadron
P;P,.
~ ~'3~ 3
states (pl, pg) of the same parity. We use the notation of
Bander and Bjorken":
go

TI(oo, g2) -+ polynomial. (3.19)

Pl+$1 ~ P2+g2 q

P= pl+ p2,
The existence of the commutator in (3.18) is guaranteed
P2 Pl A A~ (4. 1)
by the existence of an inverse Laplace transform of
(3.18). Having taken the limit ~I?2~ etc. , we may ~~, Q = qI+I?2,
continue (3.18) into the cut 4o plane, and obtain p=P ~
Q t=+2 8=+.Q —g 2 g
2

4
Col"F2(oo') d~"PI(4o) and take the limit
v
p C0 OJ
—M CO GO E pig p2, — —
-+co—
, — Qo 1 goo Q —0
such that
= lim i d'x
~= — —Q'/I
Qo/Po= (4.3)
.
P» cy

Xe'" (P P;(x, /Po), &;(0)jlP) remains finite. Encouraged by the reasonableness of this
(3 20)
(
limit in the special case of Secs. I—III, we Ilssmglle that
Upon taking the imaginary part of this relation, we it exists in this case as well.
reproduce (2.5) and (2.6). This justifies the short deriva- Under these circumstances, the new general invari-
tion given in Sec. II. ants 8 and t tend to a finite limit:

IV. GENERALIZATION
The preceding analysis can be generalized to arbitrary
currents and momenta of the states. As an example, we The covarjant an1p]itude ~ „s*tends;n the limit to

M ee= —(2gr) gi(4ollo12) I" d4x e'" 8(xo)(P2~ L j„(x),j„e(0)j ~P, )

-+ —(22r) 22(2E) d'x Cte-&~«~'"& lim (P„p»(Lj (xt) &e(0)j~P p )
P»~x

——22(2gr)'
& d'x (P*P21 ~ fj„(x,r/E), j„e(0)j P,P»)(+p»ynomla}s)
~

This last expression is a function of y2~, piL, and co alone. where we introduce the variable

2—
Upon writing out M„„~e in invariants (suppressing in-
dices nP) = —8/ =(p„+m, ) —(p„+, ). (4g)
M„„=P„P„AI+(P„Q,+P,Q„)A2+(P„Q, Pg„)A2-When these limits (4. 7) are inserted into (4.6), we find
+ (P„A„+P„A„)A 4+ (P„h„P.h„)A in the asymptotic infinite-momentum
+ (Q.~.+ Q.~.) Ao+ (Q.~.—Q.~.) A I limit

+Q„Q.A 2+ 14 „A,A g+g„.A lo, (4.6) 8„8„(8„g,+ 8„2t„) (8„1t„—

8„rt„)
we see that A4, A~, A6, and A7 would have to tend to ~l ~2
(Q') Itg in order that the limit be nonvanishing and
6nite. We consider this unlikely, but cannot exclude it.
Here we assume that in the limit these A; do not
+n, n.Ps+~, ~.F2+ g, PIo, (4 &)
contribute.
%e write where 8„= (1,1,0,0), It„= (1,0,0, 0).
Q'e are now free to identify various combinations of
A, e —1
1
F; e(ol, t, e), 2=1, 2, 3, — & these form factors in terms of the almost-equal-tin1e
(4 ?) currcllt colllllllltatols at lllflllltc 111OIllelltuIII. UsIIlg ol 'L

QoA, e~0, i=4, 5, 6, 7

A, e—1P; (te, o)o, e i=9, 10 "M. Bander and J. Hjorken, Phys. Rev. 1?4, t?04 (1968).
 BJORKEN
j tp inulca e t ansverse compoonents,

pI"&
00
2Fs
s~E ~e M-- --',
p aP

P, &+F0 e
~. ~0, aP
(4. 10)
& —F3~
M.o ~~-- e ~ (g, Q p0 ~ — 'jEI0 cI
M

R,e call

3fpv 2(2 )0 dsx

0
dre I .(„„/E), ~„s(0))I PI ) p. - . (4.11)
(+poiynoInlals)
invariant f unctions p~ ca
~
1 can be determine in arguments in c I$T that under our assump
ec.
terms o e current corre l ationn functions which
ntral role. Simi ar y, a r, )s% r'(~)'
.
'
sum rules, whose rig ht-h — annd side i ol ( ),
commutators
'
of,
J„withh 80"
8 „, can b obt "J d (5.3)
nd, 4. 11) in Inverse 2

S CO~ce TIICSC ale lndepeI1Clent w. --'0,

( '),
because in this case
e do not know what to doo with
and shall not pursue them
wi
..these results,

V. 5EUTRINO PROCESSES
f., h„h„,
n
od cIng the variables co=
the total cross section coming from
E))M')
.
' ~~ constant.
= —q ' Mv and
ec

(G'3EE1
x=v/E,

If we write the analog of

o (2.1) for the process 0„+I' ~ dco dx I

2Ir
p++hadrons asS1, 13,14
&(I (1—X) F,(co)+-', X'coFI(co)+-,-'XX 1 —-I,
X)coF0(co) )

dcoLF, (co)+-'coFI(co)+-s, coF0(co)]. (5 4)

EE' dQdE' dl g'I dI

0 cos'(00)+2WI sins(08)
(E+E')
0

neutrino energy.
the behavior o
'
sss section is pre dieted
ic
h'
ec
'

e coe

t e ual time an d a t, in fi t o
.
to rise linearly

t
M
his we take various corn of(
in the go, , —+ ll 1 ith th Ch
~ ~

leading to (2.5):
with kinematics as in ec. II, and
"dT
lim d'x
+&~co
e" 9'ILJ.(x), i'(0) jim) — 2~
M &« '"'(J'ILi. (x, r/E) i'(0)3 & = 1~,
Ws —g„,WI —
"' Pa P
'W, +, (5.2) m
~00
d $
dT

Fs(~)-
is t-hee C;Lbibl&o current, "'' iit follows from the
xe '"'(I' Li.(x r/E), i*to
(o)l I)= F()—
where j„ I

dT
lim
Ph . Rev.
T. D. Lee and C. N. Yang,, Phys. ev. 126, 2239 (1962).
"S.Adler, Phys. Rev.
-" (&IL~.(, /E), ~,'(0)&l~ =--;,
"Our normalization is j„=p» &&
179 SUM RULES AT I NFI NI TE MOMENTUM 1553

Substituting (5.5) into (5.4) we find, upon extending the ca integration to to,
GME
— '"'C(r),
dT
e (5.6)
4m. p — 2~
where

C(.) = l m d *P.Ilk'. (x, /~), j.'(o)3 Ej.(x-;/&),j'(o) j+ltTj*(x, /&), ~.'(0)hl&. ) (5 7)

An interesting result is obtained upon taking the sum of antineutrino and neutrino cross sections. By crossing
symmetry,
G'Afar
47' G'3' BC(r)
trtot + &tot ltd
Tc(r) ( t)
— 2~ 4x 87 & p

ds
»m ( —t) —
O'*IlE~j.(x,&)/&&, j'(0) j
aj, (x, l)/~l, j*'(o)3+-tL&j.(x,&)/&l, j„'(0)1 &.), , ~
(5.8)

Therefore, we predict not only that v p and vp total cross given in (2.5) and (2.6). Provided the Callan-Grosst
sections depend linearly on energy, but that the sum of integral is finite,

Ws(v, g') ( Do,

the total cross sections is determined by the equal-time
commutator of the Cabibbo current with its time lim
q2l ~~ J
q'/ —
dp
(6.3)
I

derivative at infinite momentum.

The Iinear rise of cross sections predicted here would we have shown that these commutators are not infinite,
be cut o6, were there an intermediate boson 8' ex- but may be zero (or ambiguous). The hypothesis that
changed, with the cutoff at E Ms'/Mv. Data from these commutators are indeed finite is equivalent to the
the deep-mine cosmic-ray neutrino experiments" ' are prediction
as yet inconclusive; however, a linear rise of neutrino MWt - Pi( —g'/cVr ), (6 4)
cross sections up to 10—100 BeV is not inconsistent with q2 ~oo

the data. "

(6.5)
VI. CONCLUSIONS
By combining the go
—+ i~
asymptotic limit with the Under similar assumptions, total vp and vp cross sec-
infinite-momentum. method, we have shown that in a tions are predicted to rise linearly with laboratory neu-
certain limit, the inelastic electron scattering structure trino energy. Of particular interest is the behavior of the
functions sum of cross sections, dependent, according to (5.8),
—
= Z, ( —qr/Xv) only on the equal-time commutator of the Cabibbo
mW, (v qr) (6.&) current with its time derivative, evaluated between
q&-moo, -q2ll3&=os fixed
J t( —q'/Mv) nucleon states at infinite momentum.
lim L( —g')W, (v, g') —Wi(v, g')$=- An extension of this technique to more general kine-
q&-+oo, co fixed
matical conditions, presented in Sec. IV, is possible, but
(6 2) by itself does not seem to lead to further insight into the
are directly related to Fourier transforms of almost- nature of this limit. A more physical approach into what
equal-time commutators at infinite nucleon momentum, is going on is, without question, needed.
"M. Menon et al. , Can J. Phys. 46, S344 (2968).
't F. Reines el al. , Can J. Phys. 46, S350 (1968). ACKNOWLEDGMENTS
~8 For neutrino-scattering
from a neutron, the term in (5.8)
proportional to f J,j„tj
changes sign, and the others are equal to
the corres ondin terms for the proton, in the approximation of
%e thank R. Norton, R. P. Feynman, members of
the Stanford Linear Accelerator Center (SLAC) theory
ignoring DS= 1 transitions. Thus the deep-mine experiments
measure mainly the diagonal commutators LJ', tg and LJ„J,tg J group, and the SLAC electron-scattering group for use-
ful and enjoyable conversations.
on1y.