Pre-geometrical field theory of the open string

Volume 215, number 1

PHYSICS LETTERS B

8 December 1988

P R E - G E O M E T R I C A L FIELD T H E O R Y OF T H E O P E N STRING Mihoko M. NOJIRI and Shin'ichi NOJIRI Department o(Physics, Kyoto University, Kyoto 606, Japan Received 10 August 1988

We propose a gauge invariant, background independent string action, which contains open and closed string fields and no kinetic terms. The kinetic term is generated through the condensation of the string fields, which is the solution of the equations of motion, we solve the equations and show that the action is classically equivalent to the open string action proposed by Hata et al.

There are several problems related to closed strings in open string field theory. We now point out two of them: (a) In the light-cone gauge formulation, a closed string field must be added as an elementary field since the pure open string field theory is inconsistent [ 1 ]. On the other hand, in the Lorentz covariant formulation, the closed string field arises as loop effects of the open string field [4,5]. (b) Though a space-time-metric-independent formulation of closed string field theory was proposed [ 6 ], that of the open string field theory of Hata et al. [7] is not known yet ~1. The above problems have a close connection with each other. The first problem, (a), suggests the existence of an open string action which has a larger gauge symmetry which includes general covariance and contains also a closed string field as an elementary field. We will obtain the light-cone gauge theory and the covariant theory by fixing the gauge symmetry of the action in different ways. On the other hand, the second problem, (b), arises probably because the covariant theory does not contain, as an elementary field, a closed string field which determines the geometry of space-time and because the theory has no stringy general coordinate invariance [ 10]. Therefore both of the above two problems will be solved by considering the system of open strings coupled with closed strings. In this paper, we give an answer to the

~ The background independent formulation of Witten's open string field theory [8] is already given in ref. [9].

0370-2693/88/$ 03.50 © Elsevier Science Publishers B.V. ( North-Holland Physics Publishing Division )

second problem and discuss the first one ~2 Hata and Nojiri proposed a new transformation on the open string field with a closed string functional parameter [ 11 ] in the formulation of string field theory proposed in ref. [ 7 ] : ~ c ~ = [r] - [ rq~].

( 1)

Here r is a closed string functional parameter. We denote an open string field which is given by the transition from a closed string field A by [A ] and the product of a closed string fieldA and open string field q~ by [A q~] [ 12 ] ~3. The structure of the transformation and its algebra are those of the stringy general coordinate transformation known in the closed string field theory [10]. The gauge invariant open string action: ~2 In this paper, we follow the notation of refs. [7,10-12 ] and the discussion given here is based on the theory of Hata et al. [4,6,7,10]. The one-loop scattering amplitude in covariant closed string field theory of Hata et al. obtained by applying the conventional canonical quantization violates unitarity since the integration over the moduli parameter covers the fundamental region infinitely many times. Recently, however, Hata showed that the conventional one is inapplicable since the interaction vertex is non-local in the time coordinate [ 2 ]. By applying Hayashi's theory [3] of hamiltonian formulation for field theories with non-local interactions, he found that the resulting one-loop amplitude coincides with that in light-cone gauge string field theory. The same result was also obtained by modifying the string field theory action and the BRS transformation order by order in h to recover the BRS symmetry violated by the path-integral measure. ~3 We denote open string fields by q~, hu, .4 .... and closed string fields by A, B ..... J.

99

Volume 215, number 1

PHYSICS LETTERSB

So=~'QB~+~g~'clo*~+ 2gZclJ'~oq~oclJ

(2)

is, however, not invariant under the transformation ( 1 ) and the variation of the action (2) is given by

8~.So=-QBr.{-Cqo)+½g(ClJT)}- --QBr'T.

(3)

Here (@) is a closed string field which is given by the transition from an open string field @ and we denote the product of two open string fields • and T, which is a closed string field, by (q~T) [12]. Eq. (3) tells that the action (2) is invariant if Qur is proportional to hi?. This indicates that the action is invariant under only the global part of the transformation. Recently the authors proposed a string action, which contains both open and closed string fields and has not only the gauge invariance of the open string field theory [7 ] but also stringy general coordinate invariance. We now explain this action briefly. We start with the following pre-geometrical action:

S~'~=g2{4~.Cl)o~o~+½J.(clJ~)+J,J.A}.

(9)

(~q)) + 2J,A =0.

(10)

Let {~o, Jo, A0} be a solution of eqs. ( 8 ) - ( 1 0 ) and we express the fields q~, J, A as the solution plus fluctuation:

• --+~o+~,

J--'Jo+J,

A-,Ao+A.

(11)

Then we define QOpC,, QC~OSe~,q~, T, ( ~ ) and [J] as follows: Q°pcnc/~g2{(__ )101+10o0oo00

+ ~o oq~oo~ + ¢booq~o~o + ½[Jo~] }, q~, T - = g { ¢ , o

(12)

T ° ¢ ' o + ( - )~ "b + ' (/'o q~o o T

+ ( _ )I,~l+ I~'lq~ooq~o T},

(13)

( ~ ) =- - g ( ~ o ) ,

(14)

[J] = - g [ J ~ o ] ,

(15)

Qctoscdj= 2ng2jo , j .

( 16 )

-Ao qb° 05+ ½[ J ~ ] }, 8op,+j= 0,

(Q°p~n) 2= (Q~L°~cd)z = 0.

invariant

under

the

gauge

8~"+~=g2{-qOo~oA+~oA°~

8~,~A = _ ~g2 ( A ~ ) ,

(5)

8~?"": 8~?""q)=-g[rq>],

8[?"~J= 4rigJ* r, 6~CA = 47rgA• r,

(6)

8 q"A = 2~g2a* J.

(7)

Note that we only need the o-product for open string fields, the ,-product for closed string fields and the (¢bT)-product ([Jcb]-product) in the action (4) and the transformations ( 5 ) - (7). The action (4) is background independent like in the case of the pure closed string [ 6 ]. We obtain the action with kinetic terms by considering the condensation. The equations of motion of (4) are given by 2q)o q~oq)+ [ J ~ ] =0,

(8)

(17)

Eq. ( 17 ) allows us to regard these operators Qop~nand Qdo+~d as BRS charges. Now after the redefinition ( 11 ), the pre-action (4) and the gauge transformations ( 5 ) - ( 7 ) are rewritten as follows:

S=So +g J" T+ ( 1/70J'Q°l°SedA + gZJ, J.A o + gZJ* J'A,

preJ=0, 8)~rc: 8)]r'-'(;/:)=8.1

100

J,J=0,

Here I~l is O(1) if q~ is Grassmann even (odd). These definitions ( 1 2 ) - ( 1 6 ) reproduce the same properties as those proved in ref. [7], i.e. (5.73a), (5.73b), (5.73c) and also eqs. (5), ( 7 ) - ( 9 ) in ref. [ 12],usingeq. (5.73d)in ref. [7] andeqs. (6), (10), (30) in ref. [ 12]. We can also show the nilpotency of QOpen and Q~lo+~d:

This action is transformations 8pre. o.

(4)

8 December 1988

( 18 )

80: 8o~=Q°P~nA+g~*A-A*~ _ g 2 { qbo ~)oA --

q~oAoq~+Ao q)° q~}

+ig-[J~], 6oJ=0,

8oA=ng(A) +g2 (q)A), 8c:

(19)

8~q~= [r]-g[rclo],

8,.J= (2/g) Q"l°~dr+ 2~gJ* r, 8~A=4~g{ Ao ,r + A , r},

(20)

Volume 215, number 1

PHYSICS LETTERS B qSo = Q j o ~o,

8~: 8 iq)= 8 t J = 0 ,

8,A = Q~~°~aa+2ng~J*a.

(21 )

The transformation 80 corresponds to the gauge transformation of open string field theory [ 7 ] and 8~ to the stringy general coordinate transformation [ 10 ], i.e. eq. ( 1 ). 8~ is an unfamiliar transformation which is important in the discussion of gauge fixing. In this paper, we now solve the equations of motion ( 8 ) - (10) and, after that, we show that the action in eq. (18) is classically equivalent to that in eq. (2). Eq. (10) is already solved in the case of the flat background [6] and in the case of the curved background [13]. Hereafter we consider a solution Jo which gives eq. (16) in the flat background. Using ( 12 ) - ( 16 ), we rewrite eqs. ( 8 ) and ( 10 ) as follows: QB qbo - 2g~o * ~o = 0, g(~o) +

(1/rc)Q~Ao=O.

(22) (23)

A solution {~o, Ao} should have vanishing string "length" parameter c~= 0, or else, the condensation of {@o, Ao} breaks the conservation of the string "length" parameters. There is a subtlety in the limiting procedure when the string "length" parameter ~ goes to zero [ 6 ]. We note, however, that an adequate procedure gives the correct properties of products, transition, etc. For example, the product of two string fields with vanishing string "length" parameter should be defined so that it gives the commutator of the corresponding vertex operators [ 14 ]. Let qSo be an arbitrary string functional with ghost number - 1 and infinitesimal string "length" parameter c~= e. Analysis of Neumann coefficients give the following equations [ 14 ]:

~oo q~o~< qs, 7~+0( ~/2), ~,~oac (1/e+O(1))Tr(.(zO~+O(e),

(~o O) ~c (0) +O(~'/:),

+O(E'/2).

(25)

we obtain ~o* q~=O(~),

q~* ~u = O ( ~ ) .

(24)

Here ~c is the FP ghost on the string [ 7 ]. In particular, if q3o has a form as follows:

(26)

Here ~o is another arbitrary string functional with ghost number - 1 and (o is the zero mode of the FP anti-ghost [ 7 ]. Eqs. (25), (26) and the nilpotency of the BRS charge eq. (17) tell that a solution ofeq. (22) is given by • o = lim QB(O ~to.

(27)

E~0

We can easily confirm that this solution q~o gives the non-vanishing *-product and open-closed transition through eqs. ( 1 3 ) - ( 1 5 ) , by using one simple example,

~o = (N/g)c_2 [0)6(p)O(O!--E).

(28)

Here N is a finite normalization constant. Eqs. (25), (26) and (27) tell q)o. q~=0 for an arbitrary field q) in Fock space. Due to this property, there remain less ambiguities in the limiting procedure when string "length" parameters vanish [ 14 ]. Using the solution in eq. (27), eq. (23) is rewritten as

Ql~tglim(g°Tu°)+lA°} ,~o

(29)

Here we used the following equation [ 12 ]: (Qz, q~) = Qs(@).

(30)

By solving eq. (29), we obtain Ao = -~zg lim (Jo 7Jo) +B.

(31)

E -*0

Here B is a closed string functional which satisfies the "on-shell" condition

@;B=O.

~o*Ooc (1/e+O(1))g.(O)O+O(e),

[ ~ o J ] zc [J]

8 December 1988

(32)

We now show that the action ( 18 ) or (4) is classically equivalent to the action (2) proposed by Hata etal. [7]. First we expand the closed string field J and A with respect to the anti-ghost zero mode (o into J=-

(o~j + ~./, A = -(o~.t + ~(,.

(33)

Using the gauge transformation (20) and (21), we 101

Volume 215, number l

PHYSICS LETTERS B

choose the following gauge condition: qzj = ~u, = 0 .

(34)

Second we note that J*J'Ao=Ao*J'J in the action (18) diverges due to the tachyon ~4. By regularizing this divergence by letting the string "length" p a r a m eter oe finite (o~= E:~ 0), we obtain

Ao*j=l

~-J+O(1).

(35)

Using eqs. ( 3 4 ) and ( 3 5 ) , the action ( 1 8 ) is rewritten as follows:

8 December 1988

theory [ 18 ]. His discussion is based on the cubic action [9 ] a n d associativity anomalies [ 15,17]. Although the discussion given in this paper is sometimes analogous to theirs, the relation is not clear at present. The authors would like to thank our colleagues at K y o t o U n i v e r s i t y for valuable discussions. This work is partially supported by the G r a n t - i n - A i d for Scientific Research form the Ministry o f Education, Science and Culture (#62790115 ). One o f the authors (S.N.) is i n d e b t e d to the J a p a n Society for the Prom o t i o n o f Science for financial support.

S=So -g¢aj T v,+ ( 1/7t)~sL~A + g2E - 2 0 j ~ j - } - g 2 ~jO,I¢)A-~'O( E ) .

(36)

Here we e x p a n d T in eq. ( 3 ) with respect to the antighost zero mode:

T= -eoT¢,+ T~.

(37)

After the redefinition e-'0.,-~0j

(38)

and letting e go to zero, we obtain,

S=So + g2 ~.,~j.

(39)

The action (39) is obviously equivalent to that in eq. (2). We p r o p o s e d a gauge invariant and b a c k g r o u n d ind e p e n d e n t string action, which contains b o t h open and closed string fields. We showed that this action is classically equivalent to that p r o p o s e d by H a t a et al., by choosing the gauge condition (34). A n o t h e r gauge condition might give the light-cone string field theory but we need further investigation [ 14 ]. A subject similar to that given here was discussed by Strominger and others [ 15-19] on the basis o f W i t t e n ' s action [8]. Strominger constructed closed string states in terms o f open string oscillators [ 17 ] and p r o p o s e d an action describing closed string field "4 See eq. (37) in ref. [12].

102

References [1]M. Kaku and K. Kikkawa, Phys. Rev. D 10 (1974) 1110, 1823. [2] H. Hata, Kyoto University preprints KUNS 929, 930 (1988). [31C. Hayashi, Prog. Theor. Phys. 10 (1953) 533. [4] H. Hata, K. ltoh, T. Kugo, H. Kunitomo and K. Ogawa, Phys. Rev. D 35 (1987) 1356. [5] E. Witten, Nucl. Phys. B 276 (1986) 291. [6] H. Hata, K. Itoh, T. Kugo, H. Kunitomo and K. Ogawa, Phys. Lett. B 175 (1986) 138. [7] H. Hata, K. Itoh, H. Kunitomo and K. Ogawa, Phys. Rev. D34 (1987) 2360. [81E. Witten, Nucl. Phys. B 268 (1986) 253. [9] G.T. Horowitz, J. Lykken, R. Rohm and A. Strominger, Phys. Rev. Lett. 57 (1986) 287. [ 10] H. Hata, K. ltoh, T. Kugo, H. Kunitomo and K. Ogawa, Phys. Rev. D 35 (1987) 1318. [ 11 ] H. Hata and M.M. Nojiri, Phys. Rev. D 36 (1987) 1193. [ 12 ] M.M. Nojiri and S. Nojiri, Prog. Theor. Phys. 79 (1988) 284. [13] K. Kikkawa, M. Maeno and S. Sawada, Phys. Lett. B 197 (1987) 524. 114] M.M. Nojiri and S. Nojiri, in preparation. [ 15] G.T. Horowitz and A. Strominger, Phys. Left. B 185 (1987) 45. [ 16] A. Strominger, Phys. Len, B 187 (1987) 295. [ 17] A. Strominger, Phys. Rev. Lett. 58 (1987) 629. [ 18] A. Strominger, Nucl. Phys. B 294 (1987) 93. [ 19 ] Z. Qiu and A. Strominger, Phys. Rev. D 36 ( 1987 ) 1794.