• Nenhum resultado encontrado

Braz. J. Phys. vol.32 número1

N/A
N/A
Protected

Academic year: 2018

Share "Braz. J. Phys. vol.32 número1"

Copied!
23
0
0

Texto

(1)

Sewing String Tree Verties with Ghosts

LeonidasSandoval Junior

Department ofMathematis

CentrodeCi^eniasTenologias

UDESC,UniversidadedoEstadodeSantaCatarina,Brazil

Reeivedon13November,2001

Itisshownhowtosewstringvertieswithghostsattreelevelinordertoproduenewtreeverties

usingthe GroupTheoreti approahtoStringTheory. It isthenveriedtheBRSTinvarianeof

thesewnvertexandshownthatithastheorretghostnumber.

I Introdution

Inthe early daysof String Theory, oneway to obtain

amplitudesforthesatteringofanarbitrarynumberof

strings was by using the fatorization property, what

meansthat thesatteringamplitudeofN stringsmay

beinterpretedasthesatteringamplitudesofasmaller

numberofstringssewntogether. Thismadeitpossible

tobuildtheN stringsatteringamplitudebyknowing

the expression for three string sattering amplitudes.

Even thoughthey were veryingenious and suessful,

those alulations didn't take into aount the ghost

strutureoftheverties,andthatiswhatisdonehere.

In[1℄,itwasshownhowtosewtreevertieswithout

ghosts using the Group Theoreti approah to String

Theory[2℄in ordertoobtainanew,ompositevertex.

Following the same proedure, we shall alulate the

sattering amplitude of N strings taking aount the

ghoststruture.

Weshallstartwithashortreviewofhowtosewtree

vertieswithoutghosts. Whatwemust dois sew two

legsoftwoverties,onelegfromeahvertex. Whatwe

havein thebeginningare twovertiesV

1 andV

2 with

N

1 andN

2

legs, respetively(Fig. 1).

Figure1. Individualverties.

Wenow sew leg E from V

1

withthe adjointof leg

F from V

2

. What we have now is the substitution of

thetwosewnlegsbyapropagator(Fig. 2). Whenthis

propagatoris writteninparametriform,itisan

inte-grationof oneof the variables (in order to anel one

spurious degreeof freedom) and aonformalfator P

whihontainstermsofL

n

'satingonlegE only.

Figure2. Sewnverties.

SotheresultingvertexV

(alledtheomposite

ver-tex)hasthegeneriform

V

=V

1 PV

y

2

; (1)

wherethehermitianonjugateofV

2

isforthesewnleg

F onlyand

P = Z

dxP (2)

wherexisasuitablevariable. Inwhatfollows,weshall

oftenwrite P insteadofP,alling attentiontothe

in-tegrationwhenneessary.

Whenthetwovertiesare sewntogether, we

iden-tify legs E and F. We also identify theKoba-Nielsen

variable z

E

with oneof the Koba-Nielsen variables of

vertexV y

2

, and theKoba-Nielsen variable z

F

we

iden-tify with one of the Koba-Nielsen variables of vertex

V

1

. In[1℄, this identiation is made in the following

way: z

E

maybeidentied withz

F 1 orz

F+1 , and z

F

maybeidentiedwithz

E 1 orz

E+1

. Sotherearefour

possibleombinations: a) z

E =z

F 1 ; z

F =z

E 1 ; b)

z

E =z

F+1 ; z

F =z

E+1 ;)z

E =z

F 1 ; z

F =z

E+1 ;d)

(2)

II Osillator ase

Thetwooriginalvertiessatisfysomeoverlapidentities

and so shall do the omposite vertex. One partiular

overlap identity is given by onsidering the operator

Q i

withonformalweightd=0dened by[2℄

Q i

(

i )=

1

X

n= 1

n6=0 1

p

n

i

n (

i )

n

+ i

0 ln

i +

i

0 :

(3)

where i

n

arebosoniosillatorswithommutation

re-lations

[ i

n ;a

j

m ℄=

Æ ij

Æ

n; m

; n;m6=0; (4)

[ i

0 ;a

j

n

℄=0; 8n : (5)

Theoverlapidentityisgivenby

V

Q i

(

i ) Q

j

(

j )

: (6)

Beause we areonsidering the adjointof leg F in

vertexV

2

,wemustseewhattheadjointoftheseoverlap

identities are. First,by thedenition of Q i

,wehave

that

Q iy

(

i )= Q

i

(

i ) =Q

i

(

i

): (7)

So,theadjointoftheoverlapequationsisgivenby

Q i

(

i ) Q

j

(

j )

V y

=0: (8)

We now take the overlap identity onsidering the

eets of the operator Q i

(

i

) on the vertex V

1 on a

generilegiandonleg E 1

(Fig. 3):

Figure3. OverlapidentityforV

1 .

V

1

Q i

(

i ) Q

E

(

E )

=0: (9)

We may then insert the unit operator 1 = PP 1

andmultiplybyP withoutalteringtheresult:

V

1 PP

1

Q i

(

i ) Q

E

(E)

P : (10)

SinetheonformaloperatorP atsonlyonleg E,

it will haveno eet onQ i

(

i

). Inorder to ompute

the eet of P on Q E

(

E

), we must know that, for

a onformal transformation V ating on a onformal

operatorR (z)ofweightd,

VR (z)V 1

=

dVz

dz

d

R (z): (11)

SineQ E

(

E

)hasonformalweightd=0,wehave

P 1

Q E

(

E

)P =Q E

P 1

E

; (12)

and(Fig. 4)

Figure4. OverlapidentityforV1P.

V

1 P

Q i

i Q

E

(P 1

E )

=0: (13)

The seond term in theexpression aboveis faing

now leg F of vertex V y

2

, orbest, its Hermitian

onju-gate. ConsideringthattheHermitianonjugateofQ i

(givenby(7)),wethenhavethefollowingoverlap

iden-titybetweenlegsiandF:

V

1 P

Q i

(

i ) Q

F

( P 1

F )

=0: (14)

Weanthenmakeaylingtransformationinorderto

obtaintheorretfatorforanarbitraryleg j(j 6=F)

ofvertexV y

2

. Theonlytermthatwillbeaetedisthe

termdependingonlegF:

V 1

j V

F Q

F

P 1

E

V 1

F V

j =

Q j

V 1

j V

F P

1

E

: (15)

Doingthis, theoverlapidentity fortheomposite

ver-texV

(Fig. 5)anbewrittenas

Figure5.OverlapidentityforV

.

V

Q i

(

i ) Q

j

(V 1

j V

F P

1

E )

=0; (16)

whih is the overlap equation between two arbitrary

legsiandj oftheompositevertexV

.

1

(3)

But the overlap identity for the omposite vertex,

sinenoneoflegs i orj involvesthepropagator,must

begivenby

V

Q i

(

i ) Q

j

(

j )

=0 (17)

and soin order for theequation wehaveobtainedfor

the overlapof the omposite vertex V

to be true we

musthave

V

F P

1

V 1

E

=1)P 1

= 1

V 1

F V

E

(18)

whihimpliesthat thepropagatorisgivenby

P =V 1

E V

F

: (19)

Inordertogiveanexpliitexpressionforthe

prop-agator,wewillnowhoose

i

tobeoftheform

i =V

1

i

z=z z

i

: (20)

This hoie is alled the\simple yling" [1℄ and it is

the one that simplies our alulations the most. In

thishoie, thepropagatorisgivenby

Pz= 1

z +z

F z

E

; (21)

orin termsoftheL E

n

operators 2

,

P=e (z

F z

E 1)L

E

1

( 1) L

E

0

e L

E

1

e L

E

1

: (22)

This form works for all hoies for the omposite

vertexdisussedbefore 3

. Thetruepropagatorisgiven

by expression (22)integrated overasuitable variable.

Choosingthisvariabletobes=z

F z

E

,wethenhave

P = Z

0

1

dsP = 1

L E

1 ( 1)

L E

0

e L

E

1

e L

E

1

; s=z

F z

E :

(23)

Before going any further,wemust disuss another

aspetofthetheorythatdependsonthepartiularway

in whih thelegs areidentied during thesewing

pro-edure. Let us onsider the more general ase of an

arbitraryyling V

i

. Thiskindof ylingmaydepend

onotheroordinatesthatarenotz

i

. Asanexample,let

ussupposethatweareidentifyingoordinatez

E ofleg

E with oordinate z

F 1

of vertex V y

2

and oordinate

z

F

of leg F with oordinate z

F 1

of vertex V

1 . The

overlapidentitybetweenlegsiandE 1onvertexV

1

is

V

1 h

Q i

V 1

0i z

Q (E 1)

V 1

0(E 1) z

i

(24)

where we are alling V 1

0j

the yling transformation

onlegj (j =1;:::;N). Theseylingtransformations

maydependontheotherlegs. Asanexamplewhenthis

happens, wetake another hoie of the yling

trans-formations

i

that isnot astrivialas(20)but givesa

simplerformulaforthepropagator. Thishoieisgiven

by[1℄[3℄

i =V

1

0i z=

(z

i+1 z

i 1 )

(z

i+1 z

i )

(z z

i )

(z z

i 1 )

(25)

whihisthetransformationthattakesz

i 1 ,z

i andz

i+1

to1,0and1,respetively. Itsinverseisgivenby

V

0i z=

z

i 1 (z

i z

i+1 )z+z

i (z

i+1 z

i 1 )

(z

i z

i+1 )z+(z

i+1 z

i 1 )

: (26)

Intheaseoftheylinggivenby(25),theyling

forlegE 1willdependonlegE,whihisnotpresent

in the omposite vertex. On this vertex, the overlap

betweenlegE 1andanarbitrarylegj reads

V

h

Q (E 1)

V 1

0(E 1) z

Q j

V 1

0j z

i

: (27)

This overlap equation involves terms that depend on

legE,whihisnon-existentintheompositevertexV

.

Theorretoverlapsshouldbegivenby

V

h

Q (E 1)

V 1

E 1 z

Q j

V 1

j z

i

; (28)

where the yling transformations V 1

E 1 and V

1

j do

notdepend onlegsE orF. So,in ordertorestore the

orret yling transformation for the omposite

ver-tex, a onformaltransformationmust be madeon the

yling transformationson vertexV

1

. These aregiven

by

V 1

0j

!C

1 V

1

0j

; j6=E ; (29)

where

C

1 =

N1

Y

i=1 V

1

i V

0i

: (30)

In this denition,weonsider impliitthat the

trans-formationV 1

E V

0E

=1sinetheonformal

transforma-tionsonlegE willnotbepartoftheompositevertex

andsoneednotbemodied.

Consideringthegeneralase,wehavethatthe

over-lap equation forV

1

(Fig. 6) that will lead to the

or-retompositevertexisnowobtainedfromtheoriginal

overlap

2

Otherformsforthispropagatoraregivenby[1℄:

P = e L

E

1 =s

( 1) L

E

0

s 2L

E

0

e L

E

1 =s

P = ( 1) L

E

0 L

E

1 =s

s 2(L

E

0 L

E

1 =s)

e (L

E

1 L

E

1 )=s

:

3

(4)

Figure6. OverlapidentityforV

1 .

V

1

Q i

(

0i ) Q

E

(

0E )

=0; (31)

where

0i =V

1

0i

z and

0E =V

1

0E

z. Byinserting

on-formaltransformation(30),weobtain(Fig. 7)

Figure7. Overlapidentityfor V1C 1

1 .

V

1 C

1

1

Q i

(

i ) Q

E

(

E )

=0: (32)

Insertingnowthepropagator,weobtain(Fig. 8)

Figure8. OverlapidentityforV

1 C

1

1 P.

V

1 C

1

1 P

Q i

(

i ) Q

E

(P 1

E )

=0 (33)

and we expet the omposite vertex to have a

dier-entform(givenshortly)thanin(1)inordertoamount

fortheontributionsoftheonformaltransformations.

The seond termofthe overlapis nowfaingleg F of

vertexV y

2

sothatwehavethefollowingoverlapbetween

legs iandF:

V

1 C

1

1 P

Q i

(

i ) Q

F

( P 1

E )

=0: (34)

We arenowfaingtheonformaltransformationC

F 4

,

dened by

C

F =V

1

F V

0F

(35)

whihisneessaryinordertohange

0F !

F .

Intro-duingthistransformationweobtain(Fig. 9)

Figure9. OverlapidentityforV1C 1

1 PCF.

V

1 C

1

1 PC

F

Q i

(

i ) Q

F

(V 1

0F V

F P

1

E )

=0:

(36)

Making aylingtransformation from legF toleg

j,wethenobtain(Fig. 10)

Figure10. OverlapidentityforV1C 1

1 PCFV

y

2 .

V

1 C

1

1 PC

F V

y

2

Q i

(

i ) Q

j

(V 1

0j V

F P

1

E )

=0:

(37)

One again, a onformal transformation must be

in-trodued beause of theyling transformationsV 1

0j .

Thisisdenedby

C

2 =

N

2

Y

i=1

i6=F V

1

i V

0i

: (38)

sothatwenowhave(Fig. 11)

Figure11. OverlapidentityforV1C 1

1 PCFV

y

2 C

1

2 .

4

(5)

V

1 C

1

1 PC

F V

y

2 C

1

2

Q i

(

i ) Q

j

(V 1

j V

F P

1

E )

=0: (39)

Theompositevertexmustbedened intermsofthenewylingtransformationsand soitmustnowinlude

theonformaltransformationsthatperformthishange. So,itwillnowbedenedby

V

=V

1 C

1

1 PC

F V

y

2 C

1

2

: (40)

Consideringthis, theoverlapidentityfortheompositevertexV

anbewrittenas

V

Q i

(

i ) Q

j

(V 1

j V

F P

1

E )

=0: (41)

Sinetheorretoverlapidentityfortheompositevertexisgivenby

V

Q i

(

i ) Q

j

(

j )

=0 (42)

wemusthave

V

F P

1

V 1

E

=1 (43)

whihimpliesoneagainthat

P =V 1

E V

F

: (44)

Fortheylingtransformation(25),itisonlyneessarytodoonformaltransformationsonlegsE 1,E,E+1,

F 1,F and F +1,depending onthepartiular waythevariables assoiated withthese legs are identiedwith

thevariables assoiated tolegs E andF. Inthis partiularexample(whih isasea seenbefore), theonformal

transformationsaregivenby

V 1

0(E 1) !C

1 V

1

0(E 1)

; C

1 =r

L E 1

0

; (45)

V 1

0(F 1) !C

2 V

1

0(F 1)

; C

2 =t

L F 1

0

(46)

where

r= (z

F 1 z

E 2 )

(z

F 1 z

E 1 )

(z

E 1 z

E )

(z

E 2 z

E )

; t= (z

E 1 z

F 2 )

(z

E 1 z

F 1 )

(z

F 1 z

F )

(z

F 2 z

F )

: (47)

Fortheyling(25),thepropagatorsobtainedforthefourpossibleombinationsdisussedbeforearegivenby

a) P

a =a

L E

0

; b) P

b =e

L E

1

a L

E

0

e L

E

1

; ) P

=e

L E

1

b L

E

0

; d) P

d =b

L E

0

e L

E

1

; (48)

where 5

a= (z

E+1 z

E 1 )(z

F+1 z

F 1 )

(z

E+1 z

F 1 )(z

F+1 z

E 1 )

; b= (z

E+1 z

E 1 )(z

F+1 z

F 1 )

(z

E+1 z

F+1 )(z

F 1 z

E 1 )

: (49)

Thetruepropagatorsareobtainedwhenweintegratetheexpressionsabovemultipliedbyasuitableonstant. The

resultsare:

a) P

a =

Z

1

0 daa

L E

0 1

= 1

L E

0

; b) P

b =

Z

1

0 dae

L E

1

a L

E

0 1

e L

E

1

=e L

E

1 1

L E

0 e

L E

1

; (50)

) P

=

Z

1

0 dbe

L E

1

b L

E

0 1

=e L

E

1 1

L E

0

; d) P

d =

Z

1

0 dbb

L E

0 1

e L

E

1

= 1

L E

0 e

L E

1

: (51)

Itisnowneessarytoverifytheeet ofthegaugetransformationsC

1 andC

2

ontheompositevertexasgiven

byformula(40). Weshall doit byverifyingtheeetof C

1

onvertexV

1

. Inorderto dothis weneedtheexpliit

expressionforthebosoniosillatorvertexV

1

,givenby[1℄[4℄

V

1 =

N

Y

i=1 i

h0j !

exp 2

6

6

4 1

2 N

1

X

i;j=1

i6=j 1

X

n;m=0

i

n D

nm V

1

0i V

0j

j

m 3

7

7

5

(52)

5

TheoeÆientaanbeonnetedwiththeoeÆientinreferene[1℄bya=

(6)

whereV 1

0i

andV

0j

areylingtransformationsinvolvinglegE andtheosillators i

n

haveommutationrelations

givenby(4, 5). MatriesD

nm

()aredenedin thefollowingway[3℄:

D

n0 () =

1

p

n [(0)℄

n

; (53)

D

nm () =

r

m

n 1

m!

m

z m

[(z)℄ n

z=0

; (54)

D

00 () =

1

2 ln

d

dz (z)

z=0

(55)

andhavethefollowingmultipliationproperty:

D

nm (

1

2 )=

1

X

p=1 D

np (

1 )D

pm (

2 )+D

n0 (

1 )Æ

0m +Æ

0n D

0m (

2

): (56)

InordertoalulatetheeetsoftheonformaltransformationC

1

ontheseosillatorswemustmakeuseofthe

followingonformaloperator[2℄

P i

(

0i )=

1

X

n= 1

n6=0 p

jnj i

n (

0i )

n 1

+ i

0 (

0i )

1

(57)

whihhasonformalweightone,whatmeansittransformslike

P i

(

0i )=

d

0j

d

0i P

j

(

0j

): (58)

Anosillator i

n

(n1)anbeexpressedintermsofthisonformaloperatorinthefollowingway:

i

n =

1

p

n I

0i =0

d

0i (

0i )

n

P i

(

0i

): (59)

AtingonitwiththeonformaltransformationC

1

,wehave

C

1

i

n C

1

1 =

1

p

n I

0i=0 d

0i (

0i )

n

d

d

0i V

1

i V

0i

0i

P i

(V 1

i V

0i

0i

): (60)

Makingnowahangeofvariables

0i !

i

, wehave

C

1

i

n C

1

1 =

1

p

n I

i=0 d

i (V

1

0i V

i

i )

n

P i

(

i

): (61)

Expanding(V 1

0i V

i

i )

n

intermsofD

nm

()matriesandP i

(i)intermsoftheosillators,weobtain

C

1

i

n C

1

1 =

1

X

p= 1

p6=0 r

p

n I

i =0

d

i "

p

nD

n0 V

1

0i V

i

(

i )

p 1

+ 1

X

m=1 r

n

m D

nm V

1

0i V

i

(

i )

m p 1 #

i

p

+ 1

p

n I

i=0 d

i "

p

nD

n0 V

1

0i V

i

(

i )

1

+ 1

X

m=1 r

n

m D

nm V

1

0i V

i

(

i )

m 1 #

i

0

: (62)

Performingtheintegrationswethenobtain

C

1

i

n C

1

1 =

1

X

m=0 D

nm V

1

0i V

i

i

m

: (63)

Usingthesameproessfor i

0

, weobtain

C

1

i

C 1

= i

(7)

Using these transformation properties, the multipliation rules of matries D

nm

() (equation (56)) and the

property[3℄

D

nm

()=D

mn (

1

); (65)

weanshowthat theeetoftheonformaltransformationC

1

onvertexV

1

isgivenby

V

1 C

1

1 =

N

Y

i=1 i

h0j !

exp 2

6

6

4 1

2 N1

X

i;j=1

i6=j 1

X

n;m=0

i

n D

nm V

1

i V

j

j

m 3

7

7

5

; (66)

d

i.e. theeet ofC

1

on vertex V

1

is to hange V 1

0i !

V 1

i

andV

0j !V

j

thus eliminatingthedependene of

theylingtransformationsofeverylegexeptlegEon

thelatter. Thesame anbe doneto obtainthe eet

ofonformaltransformationC

2

onvertexV y

2

,withthe

same results. So, the eet of these transformations

isto eliminatefromthe ylingtransformationsofthe

ompositevertexalldependeneonthesewnlegsEand

F.

III Introdution of ghosts

Wenowintrodueghostsinthevertexsothatwhatwe

must sew now are two verties with someghost

vari-ablesattahedto them,i.e. wewillbeonsideringthe

physialverties[4℄whihhavetheorretghost

num-ber. In this ase, in addition to satisfying the

over-lapidentitieswiththeonformaloperatorQ i

,thetwo

physial verties and the omposite vertex must also

satisfy someoverlapidentities with the onformal

op-eratorsb i

and i

,givenby[4℄

b i

(

i ) =

1

X

n= 1 b

i

n (

i )

n 2

; (67)

i

(

i ) =

1

X

n= 1

i

n (

i )

n+1

(68)

where i

n and b

i

n

are ghost antiommuting osillators

withantiommutationrelations

f i

n ;b

j

m g=Æ

n; m

: (69)

These operators have, respetively, onformal weights

2and 1,whatmeansthattheytransformlike

b i

(

i ) =

d

j

d

2

b j

(

j

); (70)

i

(

i ) =

d

j

d

i

1

j

(

j

): (71)

TheoverlapidentitiesforavertexV withthese

op-eratorsaregivenby

V "

b i

(

i )

d

j

d

i

2

b j

(

j )

#

=0; (72)

V "

i

(

i )

d

j

d

i

1

j

(

j )

#

=0: (73)

Weshallbeworkingherewithoverlapidentitiesfor

the physial vertex U [4℄, whih hasthe orretghost

number,insteadof theoverlapidentities for vertexV.

Thephysialvertexisgivenby[4℄

U =V N

Y

i=1

i6=a;b; N

X

j=1 1

X

n= 1 e

ij

n b

j

n

(74)

where a;b; are any three legs of the vertex and the

matrixe ij

n

isgivenby

1

X

n= 1 e

ij

n L

j

n =V

1

j

zi V

j

(75)

where the yling transformations are now dened on

the omplete generators L i

n

of the onformal algebra

of the bosoni osillators and of the ghost osillators.

These vetorse ij

n

havethefollowingproperty:

V

z

i =V

N

X

j=1 1

X

n= 1 e

ij

n L

j

n

: (76)

Inordertoderivetheoverlapidentityforthe

physi-alvertexU,wemustmultiplytheoverlapidentityfor

(8)

V "

b i

(

i )

d

j

d

i

2

b j

(

j )

# 0

B

N

Y

k =1

k 6=a;b; N

X

l=1 1

X

n= 1 e

k l

n b

l

n 1

C

A

=0; (77)

V "

i

(

i )

d

j

d

i

1

j

(

j )

# 0

B

N

Y

k =1

k 6=a;b; N

X

l=1 1

X

n= 1 e

k l

n b

l

n 1

C

A

=0 (78)

andpassitthroughtheoverlapidentities,obtaining

U "

b i

(

i )

d

j

d

i

2

b j

(

j )

#

=0; (79)

U "

i

(

i )

d

j

d

i

1

j

(

j )

#

+V N

X

p=1

p6=a;b; ( 1)

p N

Y

k =1

k 6=a;b;

k 6=p N

X

l=1 1

X

q= 1 e

k l

q b

l

q 1

X

n= 1 "

e pi

n (

i )

n+1

d

j

d

i

1

e pj

n (

j )

n+1 #

=0: (80)

d

From(80)weanseethat therewill bean

anoma-lous term in the i

overlap of the physial vertex U

unless both legs i and j are preisely those legs (a, b

or ) that do not have any ghosts attahed to them.

These ghosts whih are attahed to all the other legs

areresponsiblefortheanomalousterms.

III.1 Analysis of the ghost number

Before going any further, it is neessary to make

some onsiderationson theghost numberof the

om-positevertex. Asweshallseeshortly,intheasewhere

weperformthesewingwith ghostsinluded,usingthe

physialverties,theresultingompositephysial

ver-tex will nothave the orret ghost number unless we

insertsomeextraghostsinvertexU

1

beforethesewing

takesplae. Consideringthis,weshalldenethe

om-positevertexto begivenby

U

=U

1 GPU

y

2

(81)

where Garesomeextraghoststhatwill beintrodued

inordertomakevertexU

havetheorretghost

num-berandP isthepropagator(initsintegratedform).

Wemustnowanalyzetheghostnumberofthe

om-posite vertexand of itsparts in orderto alulate the

ghost number that theextra ghosts G must have. In

order todothis, weshallusetheghostnumber

opera-tor N

gh

. Fora vertex with N legs, theghost number

operatorisdenedby

N gh

= N

X

i=1 1

X

n= 1

i

n b

i

n 1

X

n=2 b

i

n

i

n !

: (82)

The reason why the ghost number operator is a sum

fromi=1toi=N isbeausethere areN vauathat

willannihilatetheoperatorsorrespondingtoeahone

ofthem. Whenatingonthephysialvertex,this

oper-atorgivesaghostnumberN,whatistheorretghost

numberforatreevertexwithN legs.

Intheaseoftheompositevertex,ithasN

1 + N

2 2

legs(beauseitdoesnothavelegsEandF,whihhave

beensewntogether)andsoitmusthaveghostnumber

(N

1 +N

2

2). Forthisvertex,theghostnumber

oper-atorN gh

anbedividedintotwoparts:

N gh

=N

gh

1 +N

gh

2

(83)

where

N gh

1 =

N1

X

i=1

i6=E 1

X

n= 1

i

n b

i

n 1

X

n=2 b

i

n

i

n !

; (84)

N gh

2 =

N

2

X

i=1

i6=F 1

X

n= 1

i

n b

i

n 1

X

n=2 b

i

n

i

n !

: (85)

Thisghostnumberoperator willhavethefollowing

ef-fetontheompositevertex:

U

N

gh

=U

N gh

1 +N

gh

2

=(N

1 +N

2 2)U

: (86)

Given formula (81)for the omposite vertex, we then

have

U

N

gh

= U

1 N

gh

1 GPU

y

2 +U

1 h

G;N gh

1 i

PU y

2

+ U

1 GPU

y

N gh

(9)

Inordertoalulatethis, wemustpaysome

atten-tion to terms one and three of the righthand side of

theexpressionabove. Weknowthat

U

1

N gh

1 +N

gh

E

=N

1 U

1

; (88)

U

2

N gh

2 +N

gh

F

=N

2 U

2

(89)

where

N gh

E =

1

X

n= 1

E

n b

E

n 1

X

n=2 b

E

n

E

n

; (90)

N gh

F =

1

X

n= 1

F

n b

F

n 1

X

n=2 b

F

n

F

n

: (91)

Taking theHermitian onjugate ofequation (89) only

onlegF,weobtain

U y

2 N

gh

2 +N

gh

F y

U y

2 =N

2 U

y

2

: (92)

SineforN gh

F y

(andforanyarbitraryghostnumber

operator)

N gh

F y

= N

gh

F

+3 (93)

we then obtain, substituting (88), (92) and (93) into

equation(87),

U

N

gh

=(N

1 +N

2 3)U

1 GPU

y

2 +U

1 h

G;N gh

1 i

PU y

2 U

1 N

gh

E GPU

y

2 +U

1 GPN

gh

F U

y

2

: (94)

PassingN gh

E

throughtheextraghostsG,wethenobtain

U

N

gh

= (N

1 +N

2 3)U

1 GPU

y

2 +U

1 h

G;N gh

1 i

PU y

2

U

1 h

N gh

E ;G

i

PU y

2 U

1 GPN

gh

E U

y

2 +U

1 GPN

gh

F U

y

2

: (95)

Wemustnowrememberthat, inthe ompositevertex, weidentify everyoperator onleg E with operators on

legF sothat N gh

E =N

gh

F

. Doingthis,thelasttwotermsin (95)anelandweobtainthefollowingresult:

U

N

gh

=(N

1 +N

2 3)U

1 GPU

y

2 +U

1 h

G;N gh

1 +N

gh

E i

PU y

2

: (96)

d

ThefatthatU

hasghostnumberN

1 +N

2

2then

impliesthat

h

G;N gh

1 +N

gh

E i

=G; (97)

i.e. theextraghoststhatmustbeintroduedinvertex

U

1

musthaveghostnumber1. 6

There is an innite number of ombinations of

ghosts that haveghost number1. Weould haveany

linear ombination of ghosts of the type b, bb, bbb,

et. butitwillprovetobesimplertohooseGtobea

ombinationofb 0

sonlysothatwemayrepresentitas

G= N1

X

i=1 1

X

n= 1

i

n b

i

n

(98)

where i

n

aretheoeÆientsofthelinearombination.

Inordertodeterminetheorretlinearombination,we

mustuse someotheronditions,likeBRST invariane

ofthesatteringamplitude. Thisweshallseenext.

III.2 BRST invariane

Wemustnowimposethatthesatteringamplitude

obtainedfromtheompositevertexisBRSTinvariant

and hek whether this ondition is strong enough to

determineG. Thesatteringamplitude[1℄isobtained

byatingwiththeompositevertex

U

=U

1 GPU

y

2

(99)

on a ertain number of physial states

(j

1 ij

2 i:::j

N

i)andthenbyintegratingoverall

vari-ablesz

i

(i=1;:::;N

1 +N

2

;i6=E;F):

W = Z

N

1 +N

2

Y

i=1

i6=E;F dz

i U

1 GPU

y

2 j

1 ij

2 i:::j

N

i: (100)

6

(10)

P is the propagatorin its integrated form and G are

theextraghoststobeinsertedinU

1 .

TheationoftheBRSThargeQonthissattering

amplitudeisgivenby

WQ= Z

N

1 +N

2

Y

i=1

i6=E;F dz

i [U

1

;Q℄GPU y

2 j

1 ij

2 i:::j

N i

+ Z

N1+N2

Y

i=1

i6=E;F dz

i U

1

[G;Q℄PU y

2 j

1 ij

2 i:::j

N i+

Z

N1+N2

Y

i=1

i6=E;F dz

i U

1

G[P;Q℄U y

2 j

1 ij

2 i:::j

N i

+ Z

N1+N2

Y

i=1

i6=E;F dz

i U

1 GP

h

U y

2 ;Q

i

j

1 ij

2 i:::j

N

i: (101)

d

Therstandthirdtermswillresultintotalderivatives

that givezerowhen oneintegratesoversomevariables

[1℄ and so what remain are just the seond and third

terms.

The ommutator P is given by a pure onformal

transformation, and it is a funtion of the generators

L E

n

(n= 1;:::)only. AstheBRSThargeommutes

withallL E

n 's,i.e.

L E

n ;Q

=0 (102)

wehave

[P;Q℄=0: (103)

Consideringnowthat

b i

n ;Q

=L i

n

; (104)

wethenhave,forGgivenby(98),

U

1

[G;Q℄=U

1 N1

X

i=1 1

X

n= 1

i

n L

i

n

: (105)

InorderforthesatteringamplitudeW tobeBRST

invariant, expression (105) must be zero or a total

derivative(thatanbeintegratedouttobeomeanull

surfaeterm). At thesametime, wewantthese extra

ghoststo plae(talkingintermsofthesimpleyling)

aghoston oneof the legsin U

1

that donot haveany

ghostsattahedtothem. Ifwenowrememberproperty

(76),wesee that we ansatisfythese onstraintsin a

niewaybyhoosingGtobegivenby

G=( 1) N1+a

N1

X

j=1 1

X

n= 1 e

aj

n b

j

n

(106)

wherea(a6=E)isoneofthelegsofvertexU

1

thatdoes

nothaveghosts attahed to it. Inserting these ghosts

invertexU

1

,wehave

U

1 G=V

1 N

1

Y

i=1

i6=a;b; N

1

X

k =1 1

X

m= 1 e

ik

n b

k

m ( 1)

N1+a N

1

X

j=1 1

X

n= 1 e

aj

n b

j

n =V

1 N

1

Y

i=1

i6=b; N

1

X

j=1 1

X

n= 1 e

ij

n b

j

n

: (107)

Usingformula(106)fortheextraghostsG,wethenhave

U

1

[G;Q℄ = U

1 ( 1)

N1+a N

1

X

j=1 1

X

n= 1 e

aj

n

b j

n ;Q

=( 1) N1+a

U

1 N

1

X

j=1 1

X

n= 1 e

aj

n L

j

n

= ( 1) N1+a

V

1

z

a N1

Y

i=1

i6=a;b; N1

X

j=1 1

X

n= 1 e

ij

n b

j

n

(108)

(11)

d

Inreferenes[5℄and[6℄,theextraghostshavebeen

plaedin the propagator. Although this anbedone,

thereisnowayoneanderiveaformulafortheghosts

in the propagatorfor a generalyling. In that ase,

theextraghostsmustbederivedandBRSTinvariane

hastobehekedforeahpartiularyling. Also,the

resultingompositevertexobtainedinthat aseis not

similarinitsghoststruturetoanordinarytreevertex,

althoughithastheorretghostnumber.

III.3 Overlap identities

Wemustnowusetheoverlapidentitiestodetermine

thepropagatorthatsatises them. Inorder todothis

weshallstartwithvertexU 0

1

,whihisthevertexwith

ylingtransformationsV 1

i

whihinvolvelegE.

Con-sideringequations(79)and(80),theoverlapidentities

forvertexU 0

1

betweenanarbitraryleg iandlegE are

givenby(Fig. 12)

Figure12. OverlapidentityforU 0

1 .

U 0

1 "

b i

(

0i )

d

0E

d

0i

2

b E

(

0E )

#

=0; (109)

U 0

1 "

i

(

0i )

d

0E

d

0i

1

E

(

0E )

#

+V 0

1 N

1

X

p=1

p6=a;b; ( 1)

p N

1

Y

k =1

k 6=a;b;

k 6=p N

1

X

l=1 1

X

q= 1 e

k l

q b

l

q 1

X

n= 1 "

e pi

n (

0i )

n+1

d

0E

d

0i

1

e pE

n (

0E )

n+1 #

=0: (110)

Theextraghostsmust then be insertedin vertex U 0

1

so that theomposite vertex will havetheorretghost

number. Multiplyingexpressions(109)and(110)bytheextraghostsG(givenby(106))andpassingthemthrough

theoverlapsweobtain(Fig. 13)

Figure13. OverlapidentityforU 0

1 G.

U 0

1 G

"

b i

(

0i )

d

0E

d

0i

2

b E

(

0E )

#

=0; (111)

U 0

1 G

"

i

(

0i )

d

0E

d

0i

1

E

(

0E )

#

+U 0

1 ( 1)

N1+a 1

X

n= 1 "

e ai

n (

0i )

n+1

d

0E

d

0i

1

e aE

n (

0E )

n+1 #

+V 0

1 N1

X

p=1

p6=a;b; ( 1)

p N1

Y

k =1

k 6=a;b;

k 6=p N1

X

l=1 1

X

q= 1 e

k l

q b

l

q !

0

N1

X

j=1 1

X

n= 1 e

aj

n b

j

n 1

A

1

X "

e pi

n (

0i )

n+1

d

0E

d

0i

1

e pE

n (

0E )

n+1 #

(12)

Theseondandthird termsof equation(112)anbeombinedsothat itbeomes

U 0

1 G

"

i

(

0i )

d

0E

d

0i

1

E

(

0E )

#

+V 0

1 N

1

X

p=1

p6=b; ( 1)

p N

1

Y

k =1

k 6=b;

k 6=p N

1

X

l=1 1

X

q= 1 e

k l

q b

l

q !

1

X

n= 1 "

e pi

n (

0i )

n+1

d

0E

d

0i

1

e pE

n (

0E )

n+1 #

=0: (113)

Atthispoint,wemustintrodueonformaltransformationsofthetypeofC

1

,givenby(30)inordertohaveat

the endtheorret ylingtransformations fortheomposite vertex. Inorder todo thisweneedto usematries

E

nm

(),dened by[5℄

E

nm ()=

1

(m+1)!

m+1

z m+1

"

(z) n+1

z z

1 #

z=0

(114)

whihhavethefollowingproperties:

1

X

t= 1 E

rt (

1 )E

ts (

2 )=E

rs (

1

2

) ; r;s;t= 1;0;1; (115)

E

rn

()=0 ; r= 1;0;1 ; n2; (116)

1

X

p= 1 E

np (

1 )E

pm (

2 )=E

nm (

1

2

) ; n;m 1; (117)

1

X

p=2 E

np (

1 )E

pm (

2 )=E

nm (

1

2 )

1

X

r;s= 1 E

nr (

1 )E

rs (

2 )Æ

sm

; n;m2: (118)

TheationoftheoperatorC

1

ontheghostsb i

n

anthenbealulatedinthefollowingway: rstwewriteb i

n in

termsofanintegralovertheonformaloperatorb i

n (

i )

b i

n =

I

0i =0

d

0i (

0i )

n+1

b i

(

0i

): (119)

ThenweinserttheoperatorC

1 :

C

1 b

i

n C

1

1 =

I

0i=0 d

0i (

0i )

n+1

d

d

0i V

1

i V

0i

0i

2

b i

V 1

i V

0i

0i

: (120)

Afterahangeofvariables

i =V

1

i V

0i

0i

wehave 7

C

1 b

i

n C

1

1 =

I

i=0 d

i V

1

0i V

i

i

n+1

d

d

i V

1

0i V

i

i

1

b i

(

i

): (121)

UsingmatriesE

nm

(),wethenmayexpand

0i

in termsof

i

. Ifwealsoexpandb i

(

i

),wethenobtain

C

1 b

i

n C

1

1 =

1

X

m= 1 1

X

p= 1 I

i =0

d

i E

nm (V

1

0i V

i )(

i )

m+1

b i

p (

i )

p 2

: (122)

Performingtheintegrationwethenhave

C

1 b

i

n C

1

1 =

1

X

m= 1 E

nm (V

1

0i V

i )b

i

m

: (123)

Using(123)inequations(111)and(113)andmultiplying(111)by(d

i =d

0i )

2

and(113)byd

i =d

0i

,wethen

have(Fig. 14)

7

(13)

Figure14. OverlapidentityforU 0

1 GC

1

1 .

U 0

1 GC

1

1 "

b i

(

i )

d

E

d

i

2

b E

(

E )

#

=0; (124)

U 0

1 GC

1

1 "

i

(

i )

d

E

d

i

1

E

(

E )

#

+V 0

1 C

1

1 N

1

X

p=1

p6=b; ( 1)

p N

1

Y

k =1

k 6=b;

k 6=p "

N

1

X

l=1 1

X

q;t= 1 e

k l

q E

qt V

1

0l V

l

b l

t #

1

X

n;m= 1 "

e pi

n E

nm V

1

0i V

i

(

i )

m+1

d

E

d

i

1

e pE

n E

nm V

1

0E V

E

(

E )

m+1 #

=0: (125)

Beforegoingfurther,somewordsmustbesaidabouttheeetsof C 1

1

onvertexU 0

1

withtheextraghostsG.

Thisisgivenexpliitlyby[4℄[5℄

U 0

1

G =

N1

Y

i=1 i

h0j !

exp 2

6

6

4 N1

X

i;j=1

i6=j 1

X

n=2 1

X

m= 1

i

n E

nm V

1

0i V

0j

b j

m 3

7

7

5

1

Y

r= 1 N

1

X

i=1 1

X

s= 1 E

rs (V

0i )b

i

s

N

1

Y

i=1

i6=b; N

X

j=1 1

X

n= 1 e

ij

n b

j

n

: (126)

MakinguseofmatriesF

nm

(),denedby[5℄

F

nm ()=

1

(m 2)!

m 2

z m 2

(

[(z)℄ n 2

z (z)

1 )

z=0

(127)

wemayalulate inasimilarwayaswedidfortheb i

n

ghoststheeetofC

1

onthe i

n

ghosts,obtaining

C

1

i

n C

1

1 =

1

X

m=2 F

nm (V

1

0i V

i )

i

m

: (128)

Usingthistogetherwiththeproperty

F

nm

()=E

mn (

1

) (129)

andequation(123),wemaythenshowthattheresultofatingwithC 1

1 onU

0

1 Gis

U 0

1 GC

1

1 =

N1

Y

i=1 i

h0j !

exp 2

6

6

4 N1

X

i;j=1

i6=j 1

X

n=2 1

X

m= 1

i

n E

nm V

1

i V

j

b j

m 3

7

7

5

1

Y

r= 1 N1

X

i=1 1

X

s= 1 E

rs (V

i )b

i

s

N1

Y

i=1 N

X

j=1 1

X

n;m= 1 e

ij

n E

nm (V

0j V

j )b

j

m

(14)

Soweansee that inthis asetheationofC 1

1 onU

0

1

Gisnotjust to hange V

0i !V

i

. Beauseof thepeuliar

nature ofe ij

n

,ittransformsas

e ij

n !

1

X

m= 1 e

ij

n E

nm (V

0j V

j )b

j

m

: (131)

Onlyinonepartiulargroupofylingtransformations(asweshallseelater)willthisbejustequivalenttohanging

V

0i !V

i

. WeshallallfromnowonU 0

1 GC

1

1 U

1 andV

0

1 C

1

1 V

1

. ThealulationforvertexC

F U

y

2 C

1

2

willbe

similar totheonewehavejust madeforU 0

1 GC

1

1 .

Havingdonethis,wemustinsertthepropagatorP intotheoverlapidentities(124)and (125)inthesameway

asintheasewithnoghosts. Butnowwemusttakeextraaresinetherearetermsdependingonb E

q

intheseond

termofequation(125). Usingequation(123)asaguideline,wehave

P 1

b E

q P =

1

X

t= 1 E

qt (P)b

E

t

(132)

sothattheresultof insertingP intooverlaps(124)and(125)is(Fig. 15)

Figure15. OverlapidentityforU

1 P.

U

1 P

"

b i

(

i )

d

d

i P

1

E

2

b E

P 1

E

#

=0; (133)

U

1 P

"

i

(

i )

d

d

i P

1

E

1

E

P 1

E

#

+V

1 P

N1

X

p=1

p6=b; ( 1)

p N1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N1

X

l=1

l6=E e

k l

q E

qt (V

1

0l V

l )b

l

t +e

k E

q E

qt (V

1

0E V

E P)b

E

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm (V

1

0i V

i )(

i )

m+1

d

E

d

i

1

e pE

n E

nm (V

1

0E V

E )(

E )

m+1 #

=0: (134)

TheoperatorsoftheoverlapequationsarenowfainglegF ofvertexU y

2

. Inordertoobtaintheoverlapidentities

forthisleg,wemustnowidentifytheoperatorsoflegE withtheonesof legF, whihareadjointoperators:

b E

t !b

F

t y

; b E

!b F

y

; E

! F

y

: (135)

First,asb F

and F

areonformaloperatorswithweights2and 1,respetively,wehave

b F

y

(

F

) = b

F

(

F ) =

d

d

F

F

2

b F

(

F )=(

F )

4

b F

(

F

); (136)

F

y

(

F

) =

F

(

F ) =

d

d

F

F

1

F

(

F ) (

F )

2

F

(

F

): (137)

Then, forb F

n and

F

n

,weobtain

F

n y

=

F

n

=

F

n

; (138)

b F

y

= b

F

b F

(15)

Then,wemustalsomakethehange

d

E

d

i

1

e pE

n (

E )

n+1

=

d

F

d

i

1

d

E

d

F

1

e pE

n (

E )

n+1

=

d

F

d

i

1 1

X

m= 1 e

pE

n E

nm V

1

E V

F

(

F )

m+1

: (140)

So,theoverlapequationsbeome

U

1 P

"

b i

(

i )

d

d

i P

1

E

2

P 1

E

4

b F

P 1

E

#

=0; (141)

U

1 P

"

i

(

i )+

d

d

i P

1

E

1

P 1

E

2

F

P 1

E

#

+V

1 P

N1

X

p=1

p6=b; ( 1)

p N1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N1

X

l=1

l6=E e

k l

q E

qt V

1

0l V

l

b l

t +e

k E

q E

qt V

1

0E V

E P

b F

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm V

1

0i V

i

(

i )

m+1

d

F

d

i

1

e pE

n E

nm V

1

0E V

F

(

F )

m+1 #

=0: (142)

Wearethen faingtheonformaltransformationC

F

that takes

F into

0F

. Insertingthis transformationwe

obtain(Fig. 16):

Figure16. OverlapidentityforU

1 PC

F .

U

1 PC

F "

b i

(

i )

d

d

i P

1

E

2

P 1

E

4

dV 1

0F V

F P

1

E

d P 1

E 2

b F

V 1

0F V

F P

1

E

#

=0; (143)

U

1 PC

F "

i

(

i )+

d

d

i P

1

E

1

P 1

E

2

dV 1

0F V

F P

1

E

d P 1

E

1

F

V 1

0F V

F P

1

E

#

+V

1 PC

F N1

X

p=1

p6=b; ( 1)

p N1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N1

X

l=1

l6=E e

k l

q E

qt V

1

0l V

l

b l

t e

k E

q E

qt V

1

0E V

E P V

1

F V

0F

b F

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm V

1

0i V

i

(

i )

m+1

d

0F

d

i

1

e pE

n E

nm V

1

0E V

0F

(

0F )

m+1 #

=0: (144)

WearefaingnowvertexV 0y

2

. Thisvertexsatisesthefollowingoverlapidentity[4℄:

1

X

s= 1 E

rs (V

0F )b

F

s V

0y

2

= V 0y

2 N2

X

i=1 1

X

s= 1 E

rs (V

0i )b

i

s

(16)

Usingequation(117),wehave

E

qt (V

1

0E V

E P V

1

F V

0F )b

F

t =

1

X

u= 1 E

qu (V

1

0E V

E P V

1

F )E

ut (V

0F )b

F

t

: (146)

All terms on b F

t

with t 2 get annihilated by the onjugate vauum j0i

F

, while we may use identity (145) to

substitutethetermsinb F

r

, r= 1;0;1. Doingthis,equation(144)beomes

U

1 PC

F "

i

(

i )+

d

d

i P

1

E

1

P 1

E

2

dV 1

0F V

F P

1

E

d P 1

E

1

F

V 1

0F V

F P

1

E

#

+V

1 PC

F N

1

X

p=1

p6=b; ( 1)

p N

1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N

1

X

l=1

l6=E e

k l

q E

qt V

1

0l V

l

b l

t +

N

2

X

l=1

l6=F e

k E

q E

qt V

1

0E V

E P V

1

F V

0l

b l

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm V

1

0i V

i

(

i )

m+1

d

0F

d

i

1

e pE

n E

nm V

1

0E V

0F

(

0F )

m+1 #

=0: (147)

Equations(143)and(147)aretheoverlapidentities betweenlegsiandF.

In orderto obtainthe overlapsbetweenleg i of vertex U

1

and anarbitrary leg j of vertexU y

2

, we must now

performaylingtransformationthatwilltaketheoperatorsfromlegF tolegj. Theeetofthistransformation

on F

P 1

V 1

E V

i

i

is

V 1

j V

F

F

V 1

0F V

F P

1

E

V 1

F V

j =

dV 1

0j V

F P

1

E

dV 1

0F V

F P

1

E !

1

j

V 1

j V

F P

1

E

: (148)

Then, wemustalsowrite

d

0F

d

i

1 1

X

m= 1 e

pE

n E

nm V

1

0E V

0F

(

0F )

m+1

=

d

0j

d

i

1 1

X

m= 1 e

pE

n E

nm V

1

0E V

0j

(

0j )

m+1

: (149)

Doingthis,overlapequations(143,147)beome(Fig. 17)

Figure17. OverlapidentityforU1PCFV 0y

2 .

U

1 PC

F V

0y

2 "

b i

(

i )

d

d

i P

1

E

2

P 1

E

4

dV

1

0j V

F P

1

E

d P 1

E !

2

b F

V 1

0j V

F P

1

E

3

5

=0; (150)

U

1 PC

F V

0y

2 2

4

i

(

i )+

d

d

i P

1

E

1

P 1

E

2 dV

1

0j V

F P

1

E

d P 1

E !

1

F

V 1

0j V

F P

1

E

3

(17)

+V

1 PC

F V

0y

2 N1

X

p=1

p6=b; ( 1)

p N1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N1

X

l=1

l6=E e

k l

q E

qt V

1

0l V

l

b l

t +

N2

X

l=1

l6=F e

k E

q E

qt V

1

0E V

E P V

1

F V

0l

b l

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm V

1

0i V

i

(

i )

m+1

d

0j

d

i

1

e pE

n E

nm V

1

0E V

0j

(

0j )

m+1 #

=0: (151)

TheoperatorsarenowfaingtheghoststhatsurroundvertexU 0y

2

(likeinequation(74)):

N

2

Y

k =1

k 6=d;g;h 0

B

N

2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q +

1

X

q= 1 e

k F

q b

F

q 1

C

A

: (152)

sothat we mustinsertthese ghosts intothe expressionsfor theoverlaps. Before doingthat, wemustnotie that

theextraghostsatingonvertexV 0y

2

haveat theirleftboththeonformaltransformationC

F

andthepropagator

P sothatwemustrstpassthemthroughin ordertoreahvertexU

1 :

PC

F N2

Y

k =1

k 6=d;g;h 0

B

N2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q +

1

X

q= 1 e

k F

q b

F

q 1

C

A

= N

2

Y

k =1

k 6=d;g;h 0

B

N

2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q 1

X

q;t= 1 e

k F

q E

qt (V

1

0F V

F P

1

)b F

t 1

C

A PC

F

: (153)

Then,identifying legsEand F,wehavethefollowingexpressionfortheghosts:

N2

Y

k =1

k 6=d;g;h 2

6

4 N2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

b E

t 3

7

5

P : (154)

WemaynowpassitthroughtheonformaltransformationC

1

,obtaining

N2

Y

k =1

k 6=d;g;h 2

6

4 N2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

V 1

E V

0E

b E

t 3

7

5

P : (155)

This isnowfaingvertexV 0

1

,whihsatisesthefollowingoverlapidentity[4℄:

V 0

1 1

X

s= 1 E

rs (V

0E )b

E

s = V

0

1 N1

X

i=1

i6=E 1

X

s= 1 E

rs (V

0i )b

i

s

; r= 1;0;1: (156)

Usingthisidentity,wemaythenwritetheextraghostsas

N

2

Y

k =1

k 6=d;g;h 2

6

4 N

2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q +

N

1

X

l=1

l6=E 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

V 1

E V

0l

b l

t 3

7

5

: (157)

PassingitbakthroughC

1

wethenhave

N

2

Y

k =1 2

6

4 N

2

X

l=1 1

X

q= 1 e

k l

q b

l

q +

N

1

X

l=1 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

V 1

E V

l

b l

t 3

7

5

(18)

Wenowinserttheseghostsintoexpressions(150)and(151). Wedosobymultiplyingthemby(158)andpassing

itthroughthersttermoftheoverlaps. Whatweobtainis(Fig. 18)

Figure18. OverlapidentityforU1PCFU 0y

2 .

U

1 PC

F U

0y

2 "

b i

(

i )

d

d

i P

1

E

2

P 1

E

4

dV

1

0j V

F P

1

E

d P 1

E !

2

b F

V 1

0j V

F P

1

E

3

5

=0; (159)

U

1 PC

F U

0y

2 2

4

i

(

i )+

d

d

i P

1

E

1

P 1

E

2 dV

1

0j V

F P

1

E

d P 1

E !

1

F

V 1

0j V

F P

1

E

3

5

+U

1 PC

F V

0y

2 N

2

X

p=1 ( 1)

p N

2

Y

k =1

k 6=d;g;h

k 6=p 2

6

4 N

2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q +

N

1

X

l=1

l6=E 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

V 1

E V

l

b l

t 3

7

5

1

X

n= 1 "

1

X

m= 1 e

pF

n E

nm V

1

0F V

F P

1

V 1

E V

i

(

i )

m+1

+

d

d

i P

1

E

1

P 1

E

2 dV

1

0j V

F P

1

E

d P 1

E !

2

e pj

n V

1

0j V

F P

1

E

n+1 3

5

+V

1 PC

F V

0y

2 N

1

X

p=1

p6=b; ( 1)

p N

1

Y

k =1

k 6=b;

k 6=p 1

X

q;t= 1 2

6

4 N

1

X

l=1

l6=E e

k l

q E

qt (V

1

0l V

l )b

l

t +

N

2

X

l=1

l6=F e

k E

q E

qt V

1

0E V

E P V

1

F V

0l

b l

t 3

7

5

1

X

n;m= 1 "

e pi

n E

nm (V

1

0i V

i )(

i )

m+1

d

0j

d

i

1

e pE

n E

nm (V

1

0E V

E )(

0j )

m+1 #

N2

Y

k =1

k 6=d;g;h 2

6

4 N2

X

l=1

l6=F 1

X

q= 1 e

k l

q b

l

q +

N1

X

l=1

l6=E 1

X

q;t= 1 e

k F

q E

qt V

1

0F V

F P

1

V 1

E V

l

b l

t 3

7

5

=0: (160)

We are then faing the last term of this omposite vertex: the onformal transformation C

2

on vertex V 0y

2 .

Insertingitintoequations(159)and(160), weobtain(Fig. 19)

Figure19. OverlapidentityforU1PU y

(19)

U 1 PU y 2 " b i ( i ) d d i P 1 E 2 P 1 E 4 dV 1 j V F P 1 E d P 1 E ! 2 b F V 1 j V F P 1 E 3 5

=0; (161)

U 1 PU y 2 2 4 i ( i )+ d d i P 1 E 1 P 1 E 2 dV 1 j V F P 1 E d P 1 E ! 1 F V 1 j V F P 1 E 3 5 +U 1 PV y 2 N2 X p=1 ( 1) p N2 Y k =1 k 6=d;g;h k 6=p 1 X q;t= 1 2 6 4 N2 X l=1 l6=F e k l q E qt (V 1 0l V l )b l t + N1 X l=1 l6=E e k F q E qt V 1 0F V F P 1 V 1 E V l b l t 3 7 5 1 X n= 1 " 1 X m= 1 e pF n E nm V 1 0F V F P 1 V 1 E V i ( i ) m+1 + d d i P 1 E 1 P 1 E 2 dV 1 j V F P 1 E d P 1 E ! 2 e pj n V 1 j V F P 1 E n+1 3 5 +V 1 PV y 2 N1 X p=1 p6=b; ( 1) p N1 Y k =1 k 6=b; k 6=p 1 X q;t= 1 2 6 4 N1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t + N2 X l=1 l6=F e k E q E qt V 1 0E V E P V 1 F V l b l t 3 7 5 1 X n;m= 1 " e pi n E nm (V 1 0i V i )( i ) m+1 d j d i 1 e pE n E nm (V 1 0E V E )( j ) m+1 # N2 Y k =1 k 6=d;g;h 1 X q;t= 1 2 6 4 N2 X l=1 l6=F e k l q E qt (V 1 0l V l )b l t + N1 X l=1 l6=E e k F q E qt V 1 0F V F P 1 V 1 E V l b l t 3 7 5

=0 (162)

wherewehavealledC

F U 0y 2 C 1 2 U y 2 andC F V 0y 2 C 1 2 V y 2 .

WemustnowextrattheghostsfromvertexU

1

in theseondtermofequation(162):

U

1

P = V

1 N1 Y k =1 k 6=b; 1 X q;t= 1 2 6 4 N1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t +e k E q E qt (V 1 0E V E )b E t 3 7 5 P = V 1 P N 1 Y k =1 k 6=b; 1 X q;t= 1 2 6 4 N 1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t +e k E q E qt (V 1 0E V E P)b E t 3 7 5 : (163) Identifying b E t withb F t

andpassingnowtheseghoststhroughtheonformaltransformationC

F

,weobtain

U 1 PC F =V 1 PC F N1 Y k =1 k 6=b; 1 X q;t= 1 2 6 4 N1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t e k E q E qt (V 1 0E V E P V 1 F V 0F )b F t 3 7 5 : (164)

Usingnowoverlapidentities (145)forvertexV y

2

andinsertingtheonformaltransformationC

2

,wethenobtain

(20)

Substituting (165)intoequation(162)wethenobtaintheoverlapidentitiesbetweenlegsiandj: U " b i ( i ) d d i P 1 V 1 E V i i 2 P 1 V 1 E V i i 4 d d i P 1 V 1 E V i i 2 b j V 1 j V F P 1 V 1 E V i i #

=0; (166)

U " i ( i )+ d d i P 1 V 1 E V i i 1 P 1 V 1 E V i i 2 d d i P 1 V 1 E V i i 1 j V 1 j V F P 1 V 1 E V i i # +V N 1 Y k =1 k 6=b; k 6=p 1 X q;t= 1 2 6 4 N 1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t + N 2 X l=1 l6=F e k E q E qt V 1 0E V E P V 1 F V l b l t 3 7 5 N 2 X p=1 p6=d;g;h 1 p N 2 Y k =1 k 6=d;g;h k 6=p 1 X q;t= 1 2 6 4 N 1 X l=1 l6=E e k F q E qt V 1 0F V F P 1 V 1 E V l b l t + N 2 X l=1 l6=F e k l q E qt (V 1 0l V l )b l t 3 7 5 1 X n;m= 1 " e pF q E nm V 1 0F V F P 1 V 1 E V i ( i ) m+1 + d d i P 1 V 1 E V i i 1 P 1 V 1 E V i i 2 dV 1 j V F P 1 E d P 1 E ! 2 e pj n E nm (V 1 0j V j ) V 1 j V F P 1 V 1 E V i i m+1 3 5 +V N1 X p=1 p6=b; ( 1) p N1 Y k =1 k 6=b; k 6=p 1 X q;t= 1 2 6 4 N1 X l=1 l6=E e k l q E qt (V 1 0l V l )b l t + N2 X l=1 l6=F e k E q E qt V 1 0E V E P V 1 F V l b l t 3 7 5 1 X n;m= 1 " e pi n E nm (V 1 0i V i )( i ) m+1 d j d i 1 e pE n E nm V 1 0E V j ( j ) n+1 # N2 Y k =1 k 6=d;g;h k 6=p 1 X q;t= 1 2 6 4 N1 X l=1 l6=E e k F q E qt V 1 0F V F P 1 V 1 E V l b l t + N2 X l=1 l6=F e k l q E qt (V 1 0l V l )b l t 3 7 5

=0: (167)

Ifwenowimposethat thesearetheorretoverlapequationsbetweenlegs iandj oftheompositevertexwe

thenmusthave:

V 1 j V F P 1 E = j (168)

what xesthepropagatoras

P =V 1

E V

F

(169)

whihis thesameform ofthepropagatorforthebosonipart,butnowwiththeylingsdened ontheomplete

generatorsL i

n

. Theoverlapequationsnowread

U " b i ( i ) d j d i 2 b j ( j ) #

=0; (170)

Imagem

Figure 2. Sewn verties.
Figure 3. Overlap identity for V
Figure 12. Overlap identity for U 0 1 .  U 0 1 " b i ( 0i )  d 0E d 0i  2 b E ( 0E ) # = 0 ; (109) U 0 1 "  i ( 0i )  d 0E d 0i  1  E ( 0E ) # +V 0 1 N 1 X p=1 p6=a;b; ( 1) p N 1 Y k =1 k 6=a;b; k 6=p N 1 Xl=1 1 Xq= 1 e k lq b l q 1 Xn= 1 " e p
Figure 14. Overlap identity for U 0 1 GC 1 1 . U 0 1 GC 11 " b i ( i )  d E d i  2 b E ( E ) # = 0 ; (124) U 0 1 GC 11 "  i ( i )  d E d i  1  E ( E ) # + V 01 C 11 N 1 X p=1 p6=b; ( 1) p N 1 Y k =1 k 6=b; k 6=p " N 1 Xl=1 1 Xq;t= 1 e k lq E qt
+5

Referências

Documentos relacionados

used in designing Damavand plasma position ontrol system.. This model is used for

widen the TCAqs apabilities in hardware and software, inluding new VME stations and speial.. GPIB ards that will allow for the reation of loal

on nitrogen gas, normally used as optial alibration method for eletron density measurements.. in Thomson sattering was realized, and the intensity of both detetors was

The results of experiments on Alfv en wave urrent drive and plasma heating in the TCABR tokamak.. are analyzed with the help of a numerial ode for simulation of the diusion of

spheromak is studied using the priniple of minimum rate of energy dissipation..

ratio (S/N) during beam emission spetrosopy of the Li beam injeted into the plasma, speially if.. density utuation measurements

The analysis of the energy and partile balane in the system plasma-relativisti runaway.. beam in TCABR, whih takes into aount only the ollisional mehanism of the heat

The inrease of toroidal magneti eld leads to gradients in the mean plasma radial.. proles and the onset of