1: \documentclass[12pt]{article}

2: \input epsf.sty

3: \topmargin -.5cm

4: \textheight 21cm

5: \oddsidemargin -.125cm

6: \textwidth 16cm

7: 

8: 

9: 

10: \newcommand{\omg}{\omega}

11: \newcommand{\ff}{F}

12: \newcommand{\eps}{\epsilon}

13: \newcommand{\ra}{\rangle}

14: \newcommand{\la}{\langle}

15: \newcommand{\T}{\chi_{T}(k)}

16: \newcommand{\Tm}{\chi_{T}(k')}

17: \newcommand{\Cn}{{\cal C}_n}

18: \newcommand{\vp}{\varphi}

19: \newcommand{\ve}{\varepsilon}

20: \newcommand{\tl}{\wt\lambda}

21: \newcommand{\hl}{\wh\lambda}

22: 

23: \newcommand{\vv} {\bar v}

24: \newcommand{\uu} {\bar u}

25: \newcommand{\K}{{\rm K_1}}

26: \newcommand{\Kt}{{\rm \widetilde K_1}}

27: 

28: 

29: \newcommand{\B}{b'}

30: \newcommand{\C}{c'}

31: \newcommand{\bB}{\bar b'}

32: \newcommand{\bC}{\bar c'}

33: \newcommand{\Bu}{B_{\vec u}}

34: \newcommand{\VV}{{\cal V}}

35: \newcommand{\BB}{{\cal B}}

36: \newcommand{\II}{{\cal I}}

37: \newcommand{\GG}{{\cal G}}

38: \newcommand{\FF}{{\cal F}}

39: \newcommand{\HH}{{\cal H}}

40: \newcommand{\MM}{{\cal M}}

41: \newcommand{\CC}{{\cal C}}

42: \newcommand{\OO}{{\cal O}}

43: \newcommand{\QQ}{{\cal Q}}

44: \newcommand{\PP}{{\cal P}}

45: \newcommand{\EE}{{\cal E}}

46: \newcommand{\LL}{{\cal L}}

47: \newcommand{\lll}{\langle\langle}

48: \newcommand{\rrr}{\rangle\rangle}

49: \newcommand{\square}{\Box}

50: \newcommand{\half}{{1\over 2}}

51: \newcommand{\wt}{\widetilde}

52: \newcommand{\wh}{\widehat}

53: \newcommand{\wc}{\check}

54: \newcommand{\wb}{\bar}

55: \newcommand{\bd}{\bar{\rm D}}

56: \newcommand{\RR}{{\cal R}}

57: \newcommand{\NN}{{\cal N}}

58: \newcommand{\TT}{{\cal T}}

59: 

60: \newcommand{\be}{\begin{equation}}

61: \newcommand{\ee}{\end{equation}}

62: \newcommand{\ben}{\begin{eqnarray}\displaystyle}

63: \newcommand{\een}{\end{eqnarray}}

64: \newcommand{\refb}[1]{(\ref{#1})}

65: \newcommand{\p}{\partial}

66: \newcommand{\sectiono}[1]{\section{#1}\setcounter{equation}{0}}

67: \renewcommand{\theequation}{\thesection.\arabic{equation}}

68: %\renewcommand{\theequation}{\arabic{equation}}

69: 

70: 

71: 

72: \begin{document}

73: {}~

74: \hfill\vbox{\hbox{hep-th/0207105}%\hbox{MRI-P-020703}

75: }\break

76: 

77: \vskip .6cm

78: 

79: 

80: \centerline{\Large \bf

81: Time Evolution in Open String Theory

82: }

83: 

84: 

85: \medskip

86: 

87: \vspace*{4.0ex}

88: 

89: \centerline{\large \rm 

90: Ashoke Sen }

91: 

92: \vspace*{4.0ex}

93: 

94: \centerline{\large \it Harish-Chandra Research

95: Institute}

96: 

97: \centerline{\large \it  Chhatnag Road, Jhusi,

98: Allahabad 211019, INDIA}

99: 

100: \centerline {and}

101: \centerline{\large \it Department of Physics, Penn State University}

102: 

103: \centerline{\large \it University Park,

104: PA 16802, USA}

105: 

106: \centerline{E-mail: asen@thwgs.cern.ch, sen@mri.ernet.in}

107: 

108: \vspace*{5.0ex}

109: 

110: \centerline{\bf Abstract} \bigskip 

111: 

112: We discuss a general iterative procedure for constructing time dependent

113: solutions in open string theory describing rolling of a generic tachyon

114: field away from its maximum. These solutions are characterized by two

115: parameters labelling the initial position and velocity of the tachyon

116: field, but one of these parameters can be removed by using time

117: translation symmetry.  The Wick rotated version of the resulting one

118: parameter family of inequivalent solutions describes a one parameter

119: family of boundary conformal field theories, each member of which is

120: related to the boundary conformal field theory describing the original

121: D-brane system by a nearly marginal deformation. We apply this technique

122: to construct a time dependent solution on a D-brane in bosonic string

123: theory which can be interpreted as the creation of a lower dimensional

124: brane during the decay of an unstable D-brane. 

125: 

126: \vfill \eject

127: 

128: \baselineskip=16pt

129: 

130: \tableofcontents

131: 

132: 

133: \sectiono{Introduction and Summary} \label{s1}

134: 

135: Time dependent solutions in string theory have received attention 

136: recently, both in closed string theory\cite{closed} and in open 

137: string 

138: theory\cite{0202210,0203211,0203265,0204203,0205085,0205098,0206102,0207004}.

139: (For early studies of open string tachyons see \cite{old}.)

140: In

141: particular in 

142: open string 

143: theory a 

144: special class of time 

145: dependent solutions were constructed which describe the rolling of the 

146: tachyon field on a non-BPS D-brane\cite{9806155,9809111}, or a 

147: brane-antibrane pair away from the 

148: maximum of the potential. Possible applications of these solutions to 

149: cosmology have been discussed in \cite{cosmo}.\footnote{For earlier 

150: attempts to 

151: apply the rolling tachyon solution to cosmology, see \cite{earl}.}

152: 

153: One common feature of the solutions associated with the rolling of an open 

154: string tachyon is that after removing the trivial 

155: parameter of the solution associated with time translation, the solutions 

156: are characterized by one parameter labelling the energy of the

157: solution.

158: This is precisely what one

159: would expect in a scalar field theory

160: with standard kinetic term, where a general spatially 

161: homogeneous solution 

162: will be characterized by the initial position and velocity of the scalar 

163: field, and one of these two parameters can be removed by using time 

164: translation invariance which allows us to set either the initial position 

165: or the initial velocity to zero.\footnote{In a multi-well potential there 

166: can be more than one

167: family of inequivalent solutions labelled by the energy.}

168: However this is somewhat surprising in 

169: open string field theory where the interaction terms have infinite number 

170: of derivatives.\footnote{Of course while the analysis of 

171: \cite{0203211,0203265} generates 

172: a one parameter family of time dependent solutions, it does not rule out 

173: the existence of other solutions and so it is still conceivable that the 

174: full theory has more solutions.}

175: 

176: The analysis of \cite{0203211,0203265} 

177: was facilitated by the fact that the rolling 

178: tachyon solution was related by Wick rotation to a boundary conformal 

179: field theory (BCFT) that was obtained from the original D-brane system 

180: by an 

181: {\it exactly marginal} deformation. The deformation parameter was related 

182: to 

183: the parameter labelling the inequivalent solutions. We cannot hope that 

184: this will be true for 

185: more general tachyonic states which might be present on a generic D-brane 

186: system, {\it e.g.} tachyons on an intersecting brane 

187: system\cite{9704006,9703217}.\footnote{Possible application of rolling

188: tachyon on 

189: intersecting D-brane system to cosmology has been emphasized in

190: \cite{inter}.} Thus 

191: the question arises: is it still possible to construct a one 

192: parameter family of inequivalent solutions which describe the rolling of a 

193: generic 

194: open string tachyon away from its maximum? 

195: 

196: This is the question we address 

197: in the paper, and show that under certain generic conditions, the answer 

198: to this question is in the affirmative. In particular once these 

199: conditions are satisfied, we have a one parameter family of inequivalent 

200: solutions, which are related by Wick rotation to a one parameter family of 

201: euclidean BCFT's. However, these BCFT's are not related to each other by a 

202: a marginal deformation. Instead, each of them is related to the BCFT 

203: describing the original unstable D-brane system by a nearly marginal 

204: deformation. The nearly marginal operator, however, is 

205: different for 

206: different solutions.

207: 

208: Unfortunately, although our analysis establishes the existence of a one 

209: parameter family of rolling tachyon solutions for a generic open string 

210: tachyon, unlike in the cases discussed in \cite{0203211,0203265},

211: generically the 

212: deformed BCFT's are 

213: not exactly solvable. As a result, we cannot give an explicit construction 

214: of 

215: the energy-momentum tensor by Wick rotating the boundary state of the 

216: deformed theory. Nevertheless, the success of this approach in generating 

217: explicit

218: solution in special cases where the deformation is exactly 

219: marginal\cite{0203211,0203265}, as well as in generating general 

220: class of solutions in 

221: $p$-adic string theory\cite{mo-zw} leads us to believe that this approach 

222: has a more 

223: general validity.

224: 

225: %The rest of the paper is organised as follows. 

226: 

227: In section \ref{s2} we 

228: illustrate our 

229: method of constructing the rolling tachyon solution by studying the 

230: example of a 

231: scalar field theory with standard kinetic term and potential $V(\phi)$ 

232: with a maximum at $\phi=0$. The idea is to relate the 

233: rolling tachyon solution via Wick rotation to the equation of motion for a 

234: scalar field with potential $-V(\phi)$. We can now solve for the periodic 

235: motion of this scalar field around $\phi=0$ using perturbation 

236: theory where we let the period of oscillation depend on the amplitude, and 

237: solve for the period and the orbit as a perturbation series in the 

238: amplitude\cite{golds}. Once we have obtained a solution this way we can 

239: inverse Wick 

240: rotate the solution to find a rolling tachyon solution. The amplitude of 

241: oscillation in the Wick rotated theory labels the initial

242: value of the 

243: rolling tachyon.

244: 

245: In section \ref{s3} we show how under certain conditions this procedure

246: can be generalized to open string field theory to construct rolling

247: tachyon solutions associated with a generic tachyon field. 

248: We also generalize this construction to the case where multiple tachyons

249: roll simultaneously beginning with arbitrary initial position and

250: velocity. We reinterpret

251: this construction in section \ref{s4} in terms of boundary conformal field

252: theory. In particular we show that in the Wick rotated theory, the family

253: of solutions labelled by the initial position of the tachyon field 

254: correspond 

255: to a family of BCFT's, and each member of this family

256: is

257: related to the BCFT describing the original D-brane by a nearly marginal

258: deformation. 

259: 

260: In section \ref{s5} we consider a specific example -- that of

261: a D-$p$-brane in bosonic string theory with one direction compactified on 

262: a circle of radius $R$,

263: -- and consider the rolling of the $(p-1,1)$ dimensional tachyon, obtained

264: by taking the first momentum mode of the $(p,1)$ dimensional tachyon along

265: the circle. We show how in the Wick rotated theory we can construct a one

266: parameter family of boundary conformal field theories, each related to the

267: original D-$p$-brane by a nearly marginal deformation, and how the inverse

268: Wick rotation of these BCFT's generate the family of rolling tachyon

269: solutions.  Unfortunately these BCFT's are not exactly solvable and hence

270: we cannot compute the analytic expressions for the energy-momentum tensor 

271: associated with these solutions.

272: However at a particular value of the radius, $R=\sqrt 2$, the family of 

273: BCFT's become related by a marginal deformation and are exactly solvable. 

274: As a result we can compute the time evolution of the energy momentum 

275: tensor explicitly by taking the inverse Wick rotation of the boundary 

276: states associated with these BCFT's. We find that during the course of 

277: time evolution there is non-trivial flow of energy density along the 

278: compact direction, and at a certain finite value of 

279: time, the energy density at a particular location on this circle

280: blows up. More specifically, the energy density as a function of the

281: coordinate $y$ along the compact circle approaches a delta function

282: singularity as we approach this critical time.

283: As we pass this  critical time, there is an apparent loss of energy

284: density at the location of the singularity, with the total amount of

285: energy lost being equal to the total initial energy of the system. Thus

286: just after the critical time the energy density averages to zero. If we

287: naively continue the

288: formula for energy density beyond this critical time, the energy density

289: gradually evolves to zero

290: everywhere. We suggest a natural interpretation of this as the creation of

291: a codimension one lump from the decay of the original brane.

292: 

293: In section \ref{s6} we generalize the construction to the case of 

294: superstring theory. In particular we consider rolling tachyon solution 

295: that should describe the creation of a D-$(p-1)$ -- $\bar{\rm D}$-$(p-1)$ 

296: pair due to the decay of a non-BPS D-$p$-brane wrapped on a circle. As in

297: the case of bosonic 

298: string theory, we can construct 

299: this solution in the Wick rotated theory as a nearly marginal deformation 

300: of the original D-$p$-brane conformal field theory. However, this BCFT is 

301: not 

302: exactly solvable, and hence we cannot explicitly construct the 

303: space-time dependent energy 

304: momentum tensor associated with this solution.

305: In section \ref{s7} we discuss 

306: possible generalization of this method to the study of closed string

307: tachyon condensation. Although 

308: part of the argument can be generalized to the case of closed strings, the 

309: main obstacle to finding the solution in closed string theory arises 

310: due to our inability to solve the equations of motion of the graviton and 

311: the dilaton field in the background of rolling tachyon.

312: We conclude in section \ref{s8} with a few remarks.

313: 

314: \sectiono{Example in Scalar Field Theory} \label{s2}

315: 

316: In this section we shall discuss the general method for constructing time 

317: dependent solutions in the context of a scalar field theory, which we 

318: shall 

319: generalize to string field theory in the later sections. We begin 

320: with the action of a scalar field $\phi$ in $p+1$ dimensions with standard 

321: kinetic term and potential $V(\phi)$:

322: \be \label{e1}

323: S = - \int d^{p+1} x [ \p^\mu\phi \p_\mu\phi + V(\phi)]\, .

324: \ee

325: We shall further assume that $V(\phi)$ has a maximum at $\phi=0$, with

326: \be \label{e2}

327: V''(0) = -m^2\, .

328: \ee

329: We want to study time dependent solution of this equation determined by a

330: given initial condition. For simplicity let us consider spatially

331: homogeneous field configurations. In this case the solution is

332: characterized by two parameters, -- the initial position and velocity of

333: $\phi$. For definiteness we shall further restrict to solution with

334: total energy $E< V(0)$. In this case $\p_0\phi$ vanishes at some instant

335: of time when $V(\phi)=E$. We shall take this to be the origin of $x^0$. 

336: Thus the solution is now characterized by only one parameter $\lambda$, -- 

337: the value of $\phi$ at $x^0=0$. We shall be interested in the solution 

338: where $\lambda$ is small but not infinitesimal. 

339: 

340: 

341: The equation of motion is:

342: \be \label{e3}

343: \p_0^2 \phi + V'(\phi) = 0\, .

344: \ee

345: It is of course straightforward to (numerically) integrate this equation, 

346: but 

347: this cannot be generilized to string field theory where the equations of 

348: motion contain infinite number of derivatives. 

349: We shall follow an indirect method that can be generalized in the context 

350: of 

351: string field theory. The basic idea is the same as the one followed in 

352: \cite{0203211,0203265}, namely we make a Wick rotation $x^0=i x$, and 

353: write eq.\refb{e3} as

354: \be \label{e4}

355: \p_x^2 \phi - V'(\phi) = 0\, .

356: \ee

357: If $\phi=f(x)$ is a solution of eq.\refb{e4}, then $\phi=f(-i x^0)$ will 

358: be a solution to \refb{e3}. Thus the aim is now to solve eq.\refb{e4} 

359: with the boundary condition $\phi=\lambda$, $\p_x\phi=0$ at $x=0$. 

360: This clearly can be thought of as a motion of a particle in potential 

361: $-V(\phi)$. Since $V(\phi)$ has a maximum at $\phi=0$, $-V(\phi)$ has a 

362: minimum at $\phi=0$, and for small $\lambda$, solution to the equation of 

363: motion 

364: will oscillate around zero. 

365: The period of oscillaton $T\equiv 2\pi/\omg$ is in general a function of 

366: $\lambda$.\footnote{An analytic expression of 

367: $T$ is

368: given by

369: $$ T = \sqrt 2 \, \int_{\phi_1}^{\phi_2} {d\phi

370:  \over \sqrt{V(\phi)-V(\lambda)}}\, ,

371: $$ 

372: where $\phi_1$ and $\phi_2$ are turning points where 

373: $V(\lambda)=V(\phi_1)=V(\phi_2)$. The initial value $\lambda$ of $\phi$ 

374: can be identified with either $\phi_1$ or $\phi_2$. However, we shall not 

375: use this formula.} Thus the 

376: solution can be expanded as

377: \be \label{e9}

378: \phi(x; \lambda) = \sum_{n=0}^\infty a_n \cos(n\omg x)\, .

379: \ee

380: 

381: We shall now solve \refb{e4} using perturbation 

382: theory\cite{golds}.

383: If we write

384: \be \label{e5a}

385: V(\phi) = -{1\over 2} m^2 \phi^2 +  V_{int}(\phi)\, ,

386: \ee

387: then \refb{e4} takes the form

388: \be \label{e5b}

389: (\p_x^2 + m^2) \phi -  V_{int}'(\phi) = 0\, .

390: \ee 

391: $ V_{int}(\phi)$ is of order $\phi^n$ with $n\ge 3$. Thus

392: for small $\lambda$ the solution behaves as

393: \be \label{e10}

394: \phi = 

395: \lambda\cos(mx) + \OO(\lambda^2)\, . 

396: \ee

397: Comparing this with \refb{e9} we get

398: \be \label{e11}

399: a_1 = \lambda + \OO(\lambda^2), \qquad

400: \omg = m + \OO(\lambda), \qquad 

401: a_n = \OO(\lambda^2) \quad \hbox{for} \quad n=0, n\ge 2\, .

402: \ee

403: {}From this we see that we can trade in the parameter $\lambda$ for 

404: the coefficient $a_1\equiv\hl$, and rewrite \refb{e11} as 

405: \be \label{e8}

406: a_1=\hl,  \qquad

407: \omg = m + \OO(\hl), \qquad

408: a_n = \OO(\hl^2) \quad \hbox{for} \quad n=0, n\ge 2\, .

409: \ee

410: We now substitute \refb{e9} into eq.\refb{e5b} to get

411: \be \label{e12}

412: \sum_{n=0}^\infty \Big({n^2\omg^2}-m^2\Big) \, a_n \cos(n\omg x)

413: = - V_{int}'\Big( \sum_{n=0}^\infty \, a_n \cos(n\omg x)\Big)\, .

414: \ee

415: The right hand side of this equation contains terms quadratic and higher 

416: order in $a_n$. Thus we can solve the equations iteratively as follows. 

417: The zeroeth order approximation is taken to be

418: \be \label{e13}

419: \omg=m, \qquad a_n=0 \quad \hbox{for} \quad n=0, \quad n\ge 2\, .

420: \ee

421: Also $a_1$ is set equal to $\hl$ to all orders.

422: We 

423: substitute the $k$-th order results for the $a_n$'s and $\omg$ on the 

424: right 

425: hand side of \refb{e12} to compute the coefficient of $\cos(n\omg x)$ 

426: for every $n$. 

427: Comparing this with the coefficient of $\cos(n\omg x)$ on the left hand 

428: side, 

429: we determine the $(k+1)$th order values of the $a_n$'s for $n=0$ and $n\ge 

430: 2$. On the other hand equating the coefficient of the $\cos(\omg x)$ 

431: terms on 

432: both sides, we determine the value of $\omg$ to $(k+1)$th 

433: order.\footnote{Actually this determines $\omg$ to order $\hl^k$, but 

434: since the coefficients $a_n$ are of order $\hl$ or higher, this determines 

435: the solution to order $\hl^{k+1}$.} This is 

436: possible since $a_1$ is set equal to $\hl$, and we are determining all the 

437: coefficients in terms of $\hl$.

438: 

439: This gives a solution to eq.\refb{e4}. Given this, we can now generate a 

440: solution to eq.\refb{e3} by making the substitution $x=-ix^0$. This gives

441: \be \label{e14}

442: \phi(x^0) = \sum_{n=0}^\infty a_n(\hl) \cosh(n\omg(\hl) x^0)\, ,

443: \ee

444: where the coefficients $a_n$ are the same as the ones determined by the 

445: previous method. This gives a family of solutions characterized by one 

446: parameter $\hl$. $\hl$ determines the initial value $\lambda$ of $\phi$, 

447: with the precise relation between $\hl$ and $\lambda$ being given by:

448: \be \label{e15}

449: \lambda = \phi(0) = \sum_{n=0}^\infty a_n(\hl)\, .

450: \ee

451: 

452: \begin{figure}[!ht]

453: \leavevmode

454: \begin{center}

455: \epsfysize=5cm

456: \epsfbox{fig1.eps}

457: \end{center}

458: \caption{The plot of a particular time dependent solution in a scalar 

459: field theory with potential $-\phi^2/2 + \phi^3/3$ for the choice

460: $\hl=.5$. This gives $\phi(x^0=0)\simeq 0.598122$. The top curve is the 

461: result of direct numerical integration with this boundary condition, while

462: the bottom curve is obtained 

463: by perturbative techniques discussed here with 60 harmonics. As we can see 

464: the two curves 

465: coincide for $-\tau/2 < x < \tau/2$ where $\tau$ is the period 

466: of oscillation, but the perturbation theory breaks down beyond this 

467: range.} \label{f1}

468: \end{figure}

469: 

470: 

471: For example, if we take the $\phi^3$ field theory with potential $-{1\over

472: 2}\phi^2 + {1\over 3}\phi^3$, then $V'_{int}(\phi) = \phi^2$. Using the

473: first order solution $\phi(x^0)=\hl\cos(\omega x^0)$ on the right hand

474: side of \refb{e12}, we get the solution to second order in $\hl$: 

475: \be \label{esecond}

476: a_1 = \hl, \qquad a_0 \simeq {1\over 2} \hl^2, \qquad a_2 \simeq -{1\over

477: 6}

478: \hl^2, \qquad \omega \simeq 1\, .

479: \ee

480: Note that $\omega$ remains constant at 1 since the Fourier expansion of

481: $(\hl\cos x)^2$ does not contain a term proportional to $\cos x$.

482: Substituting \refb{esecond} on the right hand side of \refb{e12} and

483: comparing coefficients of $\cos(\omega x)$ on both sides we get the

484: equation for $\omega$ at next order:

485: \be \label{eyyy1}

486: (\omega^2-1) \hl \simeq -{5\over 6} \hl^3 \qquad \to \qquad \omega^2

487: \simeq 1

488: -{5\over 6} \hl^2\, .

489: \ee

490: The coefficients $a_n$ to order $\hl^3$ are given by:

491: \be \label{eyyy2}

492: a_1 = \hl, \qquad a_0 \simeq {1\over 2} \hl^2, \qquad a_2 \simeq -{1\over

493: 6}

494: \hl^2, \qquad a_3 \simeq {1\over 48} \hl^3\, .

495: \ee

496: The explicit expression for the solution becomes complicated at higher

497: order, but we can evaluate it numerically. 

498: As illustrated in Fig.\ref{f1},

499: explicit numerical analysis in this $\phi^3$ field theory shows that this 

500: procedure generates the time dependent solution in this theory, describing 

501: rolling of $\phi$ from the maximum towards the local minimum of 

502: $V(\phi)$, quite accurately for half a

503: period of oscillation in either direction of $x^0$. 

504: Beyond this range the expansion \refb{e14} 

505: diverges and we must take recourse to analytic continuation in order to 

506: study the solution. 

507: 

508: \sectiono{Solution in String Field Theory} \label{s3}

509: 

510: We now consider some specific D-brane system in bosonic string theory with 

511: a tachyonic mode. Associated with the zero momentum tachyon state, there 

512: is a boundary operator $V_T$ of dimension $h<1$. In $\alpha'=1$ unit, the 

513: associated tachyon 

514: field has mass$^2$

515: \be \label{e2.1}

516: (h-1) \equiv -m^2\, .

517: \ee

518: If we displace the tachyon field $T$ a distance $\lambda$ away from its 

519: maximum, and let the system evolve in time, then to leading order in 

520: $\lambda$ the solution is given by 

521: $\lambda\cosh(mx^0)$.  In the Wick rotated theory this corresponds to 

522: $\lambda\cos(mx)$.

523: The corresponding solution of the linearizerd 

524: equations 

525: of motion in string field theory is given by

526: \be \label{e2.2}

527: \lambda\, \cos(mX(0)) \, V_T(0)c_1|0\ra\, .

528: \ee

529: This is a BRST invariant state of $L_0$ eigenvalue 0. 

530: 

531: 

532: The full string field theory equation of motion is given by:

533: \be \label{e2.2a}

534: Q_B |\Psi\ra + |\Psi*\Psi\ra = 0\, ,

535: \ee

536: where $|\Psi\ra$ is an open string state of ghost number 1, and

537: $Q_B$ is the BRST 

538: charge. We shall try to find a solution to eq.\refb{e2.2a} in the Siegel 

539: gauge

540: \be \label{e2.2b}

541: b_0|\Psi\ra = 0\, ,

542: \ee

543: where the equation takes the form:

544: \be \label{e2.2c}

545: L_0|\Psi\ra = -b_0 |\Psi*\Psi\ra\, .

546: \ee

547: Our goal will be to first construct a solution to \refb{e2.2c} that

548: extends \refb{e2.2} to higher orders in $\lambda$, and then carry

549: out

550: an inverse Wick rotation to  construct a time dependent solution to

551: eq.\refb{e2.2c} 

552: describing the rolling of the tachyon away from its maximum.

553: We begin by defining:

554: \be \label{e2.3}

555: |\phi_0\ra = \cos(\omg X(0)) \, V_T(0)\, c_1|0\ra\, ,

556: \ee 

557: where $\omg$ is a number that will be determined 

558: below.

559: We choose a set of linearly independent basis states $|\phi_r\ra$ ($0\le 

560: r < \infty$) in a

561: subspace $\HH_1$ of first quantized open string states, 

562: satisfying the following conditions:

563: \begin{enumerate}

564: \item Each $|\phi_r\ra$ has ghost number 1.

565: \item Each $|\phi_r\ra$

566: is symmetric under $X\to -X$, and carries $x$-momentum which 

567: is an

568: integer multiple of $\omg$.

569: \item All $|\phi_r\ra$ for $r\ge 1$ are orthogonal to $c_0|\phi_0\ra$.

570: \end{enumerate}

571: $\HH_1$ by definition is the subspace spanned by the basis states

572: $\{|\phi_r\ra, r\ge 0\}$ satisfying these properties, and

573: we shall look for a solution $|\Psi\ra$ of 

574: eq.\refb{e2.2c} in this subspace. We shall also require that when we 

575: express the solution $|\Psi\ra$ as a linear combination of the basis 

576: states $|\phi_r\ra$, the 

577: coefficient of $|\phi_0\ra$ in $|\Psi\ra$ is given by $\hl$; this in fact 

578: will be the definition of the parameter $\hl$. For later use, we shall 

579: define:

580: \be \label{edefpri}

581: |\chi\ra' = |\chi\ra - \NN^{-1} \la\phi_0|\chi\ra \, c_0|\phi_0\ra, \qquad 

582: \NN \equiv \la \phi_0|c_0|\phi_0\ra\, ,

583: \ee

584: for any state of $|\chi\ra$ ghost number 2. Physically $|\chi\ra'$ gives 

585: $|\chi\ra$ with its component along $c_0|\phi_0\ra$ removed.

586: 

587: The construction of the solution $|\Psi_{(n)}\ra$ to order $\hl^n$

588: proceeds as follows.

589: We define

590: \be \label{e2.4}

591: |\Psi_1\ra = \hl |\phi_0\ra\, ,

592: \ee

593: and  $|\Psi_p\ra$ iteratively by using the relation:

594: \be \label{e2.5}

595: |\Psi_p\ra = -{b_0\over L_0} (|\Psi_{p-1}*\Psi_{p-1}\ra)' + \hl 

596: |\phi_0\ra\, .

597: \ee

598: We repeat \refb{e2.5} $(n-1)$-times 

599: to compute $|\Psi_n\ra$. 

600: If we define

601: \be \label{edeffn}

602: f_n(\omg,\hl) = \NN^{-1} \, \la \phi_0| \Psi_{n-1} * \Psi_{n-1}\ra\, ,

603: \ee

604: then we have

605: \be \label{e2.6}

606: (|\Psi_{n-1}*\Psi_{n-1}\ra)' = |\Psi_{n-1}*\Psi_{n-1}\ra - f_n(\omg,\hl) 

607: c_0 |\phi_0\ra\, .

608: \ee

609: We now define $\omg_n$ as the solution of the equation:

610: \be \label{e2.9}

611: \hl (\omg_n^2 - m^2)+

612: f_n(\omg_n,\hl) = 0\, ,

613: \ee

614: and take the $n$-th order solution $|\Psi_{(n)}\ra$ to be

615: \be \label{e2.11a}

616: |\Psi_{(n)}\ra = |\Psi_n\ra_{\omg=\omg_n}\, .

617: \ee

618: Note that $\omg$ at $n$th order is determined {\it after we have 

619: computed $f_n(\omg,\hl)$ by repeating the iterative procedure $(n-1)$ 

620: times}. 

621: Since $f_n(\omg,\hl)$ is of order $\hl^2$, \refb{e2.9} gives 

622: $\omg_n=m+\OO(\hl)$.\footnote{Actually from the definition of $\phi_0$ 

623: and 

624: conservation of $X$-momentum, it follows 

625: that $|\phi_0*\phi_0\ra$ contains states with $X$-momentum $\pm 2\omg$ 

626: or $0$, but does not contain a state of momentum $\pm\omg$. Thus 

627: $|\phi_0*\phi_0\ra$ does not contain any component along $c_0|\phi_0\ra$. 

628: As a result the order $\hl^2$ contribution to $f_n(\omega,\hl)$ vanishes,

629: and 

630: $f_n(\omega,\hl)=\OO(\hl^3)$. Thus $\omg=m+\OO(\hl^2)$. Generalising this 

631: argument we can see that 

632: $f_n(\omega,\hl)$ must be an odd function of $\hl$, and hence $\omg$ must

633: be 

634: an even function of $\hl$.}

635: 

636: We shall now show that $|\Psi_{(n)}\ra$ determined this way 

637: satisfies:

638: \be \label{e2.11}

639: L_0 |\Psi_{(n)}\ra = - b_0 |\Psi_{(n)}*\Psi_{(n)}\ra + 

640: \OO(\hl^{n+1})\, ,

641: \ee

642: and hence satisfies the string field theory equations of motion to order

643: $\hl^n$.

644: To do this, we

645: first note that \refb{e2.5}, \refb{e2.6} for $p=n$ gives

646: \be \label{e2.7}

647: L_0|\Psi_n\ra = -b_0 |\Psi_{n-1}*\Psi_{n-1}\ra +   

648: f_n(\omg,\hl)|\phi_0\ra  + \hl (\omg^2 - m^2) |\phi_0\ra\, ,

649: \ee

650: where we have used the fact that

651: \be \label{e2.8}

652: L_0 |\phi_0\ra = (\omg^2 - m^2) |\phi_0\ra\, .

653: \ee

654: Eq.\refb{e2.9}, \refb{e2.7} now give 

655: \be \label{e2.10} 

656: [L_0 |\Psi_n\ra + b_0 |\Psi_{n-1}*\Psi_{n-1}\ra]_{\omg=\omg_n}=0\, .

657: \ee 

658: We shall now show that 

659: \be \label{e2.12}

660: |\Psi_p\ra = |\Psi_{p-1}\ra + \OO(\hl^p)\, ,

661: \ee 

662: for any $p\le n$.

663: In that case if we replace $|\Psi_{n-1}\ra$ on 

664: the right hand side of \refb{e2.10} by $|\Psi_n\ra$, the net error will be 

665: of order $\hl^{n+1}$, since $\Psi_{n}$ itself is order $\hl$. 

666: This, together with \refb{e2.11a}, would establish \refb{e2.11}.

667: 

668: In order to establish \refb{e2.12} we use proof by induction.

669: We use \refb{e2.5} to get:

670: \be \label{e2.13}

671: |\Psi_{p+1}\ra - |\Psi_{p}\ra 

672: = -{b_0\over L_0} (|\Psi_{p}*\Psi_{p}\ra - 

673: |\Psi_{p-1}*\Psi_{p-1}\ra)'\, .

674: \ee

675: Now suppose that \refb{e2.12} holds for some value of $p$. In that case 

676: the 

677: right hand side of \refb{e2.13} 

678: is of 

679: order 

680: $\hl^{p+1}$, since

681: $|\Psi_{p-1}\ra$ itself is of order $\hl$. Since \refb{e2.12} holds 

682: for $p=1$, this proves \refb{e2.12} for 

683: all 

684: $p\le n$.

685: 

686: Note however that for the above argument to go through, we need to ensure 

687: that $b_0(|\Psi_{p}*\Psi_{p}\ra -

688: |\Psi_{p-1}*\Psi_{p-1}\ra)'$ does not have a state with $L_0$ eigenvalue 

689: of order $\hl$, since this will give an additional power of $\hl$ in the 

690: denominator from the operation of $1/L_0$ and will destroy the counting of 

691: powers of $\hl$ that we have used. It is precisely due to this reason that 

692: we have removed the component of $|\Psi_{p}*\Psi_{p}\ra$ along 

693: $c_0|\phi_0\ra$ in this expression, since $c_0|\phi_0\ra$ has $L_0$ 

694: eigenvalue 

695: $(\omg^2 - m^2)\sim \hl$. For our analysis to hold it is important that

696: besides $|\phi_0\ra$,  

697: there are no other states appearing in the computation of 

698: $|\Psi_{p}*\Psi_{p}\ra$, which has $L_0$ eigenvalue of order $\hl$ or 

699: less.

700: 

701: This requirement can be reformulated purely in terms of matter sector 

702: vertex operators as follows.

703: The possible problematic states appearing in the expression for

704: $|\Psi_{p}*\Psi_{p}\ra$ are ghost number 2 states which are not

705: annihilated by $b_0$, and has $L_0$ eigenvalue 0 in the $\hl\to 0$ limit.  

706: Such states are of the form: 

707: \be \label{ef1} 

708: O(0)\, c_0 c_1 |0\ra\, , 

709: \ee

710: where $O$ is a matter sector vertex operator whose conformal weight

711: approaches 1 as $\hl\to 0$. As long as the theory does not contain any

712: operator of this type which can appear in the operator product of

713: $\cos(\omg X) V_T$ with itself (other than $\cos(\omg X) V_T$ itself,

714: whose effect has already been taken into account) the analysis given above

715: remains valid.  Matter operators of the form $\p X$ of dimension 1, which 

716: could cause potential problems, do not do so since the $X\to -X$ symmetry 

717: prevents their appearance in 

718: the operator product of $\cos(\omg X) V_T$ with itself.

719: 

720: The situation simplifies if not even $c_0|\phi_0\ra$ appears in the 

721: computation of $|\Psi_p*\Psi_p\ra$ for $\omega=m$.

722: In this case we see from \refb{edeffn} that

723: $f_n(m,\hl)$ vanishes identically for arbitrary $\hl$. Thus 

724: eq.\refb{e2.9} can be solved by setting $\omg=m$. 

725: The cases analysed in refs.\cite{0203211,0203265} are of this type. 

726: 

727: Once we have obtained the solution in the Wick rotated theory, we can 

728: perform an inverse Wick rotation $x=-ix^0$ to obtain a time dependent 

729: solution in the Lorentzian theory. This amounts to replacing $\cos(\omg 

730: X)$ by $\cosh(\omg X^0)$, and the oscillator $\alpha_m$ of $X$ by 

731: $-i\alpha^0_m$ of $-iX^0$. Performing these operations on $\Psi_{(n)}$ we 

732: can generate a time dependent solution of the equations of motion to order 

733: $\hl^n$. The constant $\hl$ labelling the solution parametrizes the 

734: initial 

735: condition on the tachyon field. Of course, we expect the  series expansion 

736: to be valid only for a limited range of $x^0$, and for $x^0$ beyond this 

737: range we need to obtain the result via analytic continuation.

738: 

739: Since the discussion so far has been somewhat abstract, we shall now give 

740: explicit form of the solution to order $\hl^3$. We have:

741: \be \label{eexp1}

742: |\Psi_1\ra = \hl |\phi_0\ra\, ,

743: \ee

744: \be \label{eexp2}

745: |\Psi_2\ra = \hl |\phi_0\ra - \hl^2 \, {b_0\over L_0} \, 

746: |\phi_0*\phi_0\ra\, .

747: \ee

748: No subtraction is needed to this order since $|\phi_0*\phi_0\ra$ does not 

749: have a component along $c_0|\phi_0\ra$ due to $X$-momentum conservation. 

750: To next order\footnote{We could ignore the $\OO(\hl^4)$ term in 

751: $|\Psi_3\ra$, coming from the $*$-product of ${b_0\over L_0} \,

752: |\phi_0*\phi_0\ra$ with itself, in computing the solution to order

753: $\hl^3$.} \ben \label{eexp3}

754: |\Psi_3\ra &=& \hl |\phi_0\ra - \hl^2 \, {b_0\over L_0} |\phi_0 * 

755: \phi_0\ra \nonumber \\ 

756: && + {b_0\over L_0} \bigg( \hl^3 |\phi_0\ra * {b_0\over L_0} \,

757: |\phi_0*\phi_0\ra

758: + \hl^3 {b_0\over L_0} \, |\phi_0*\phi_0\ra * |\phi_0\ra + f_3(\omg,\hl)

759: c_0 

760: |\phi_0\ra \bigg) \nonumber \\ \cr

761: && + \OO(\hl^4)

762: \een

763: where

764: \ben \label{eexp4}

765: f_3(\omg,\hl) &=& -\NN^{-1} \hl^3 \la \phi_0| \bigg( |\phi_0\ra * 

766: {b_0\over 

767: L_0} \, |\phi_0*\phi_0\ra + {b_0\over L_0} \, |\phi_0*\phi_0\ra * 

768: |\phi_0\ra \bigg)\nonumber \\

769: &=& -2\, \NN^{-1} \hl^3 \la \phi_0 * \phi_0 | {b_0\over L_0} |\phi_0 * 

770: \phi_0\ra

771: \, .

772: \een

773: This can be related to the four point amplitude $A^{(4)}$ in string field 

774: theory involving four 

775: external states $|\phi_0\ra$. Taking into account the sum over $s$, $t$ 

776: and $u$ channel diagrams, and a factor of 2 coming from each vertex since 

777: the three point coupling in string field theory is accompanied by a factor 

778: of $1/3$ instead of $1/6$, we have\footnote{Throughout this paper 

779: $A^{(4)}$ will 

780: denote the amplitude in the euclidean string (field) theory.}

781: \be \label{eexp5}

782: A^{(4)}(\omg) = 12 \, \la \phi_0 * \phi_0 | {b_0\over L_0} |\phi_0 * 

783: \phi_0\ra \, .

784: \ee

785: Thus \refb{eexp4} can be written as

786: \be \label{eexp6}

787: f_3(\omg,\hl) = - {1\over 6} \NN^{-1} \hl^3 A^{(4)}(\omg)\, .

788: \ee

789: $\omg_3$ is then the solution to the equation

790: \be \label{eexp7}

791: (\omg_3^2 - m^2) = {1\over 6} \NN^{-1} \hl^2 A^{(4)}(\omg_3)\, .

792: \ee

793: To leading order we can set $\omg_3=m$ on the right hand side of 

794: \refb{eexp7} and get

795: \be \label{eexp8}

796: \omg_3 = \sqrt{ m^2 + {1\over 6} \NN^{-1} \hl^2 A^{(4)}(m)}\, .

797: \ee

798: $|\Psi_3\ra$ given in \refb{eexp3}, evaluated at $\omg=\omg_3$, then 

799: gives the solution $|\Psi_{(3)}\ra$ to order $\hl^3$.

800: 

801: This concludes our discussion on generating the time dependent solution of 

802: string field theory, describing the rolling of a tachyon away from its 

803: maximum. We shall end this section by mentioning three generalizations of

804: this analysis:

805: \begin{itemize}

806: \item We have discussed the case where the tachyon begins rolling from an 

807: initial configuration where its time derivative vanishes and the field is 

808: displaced from its maximum. In this case the total energy of the system is 

809: less than that at the maximum. We could also consider the case where we 

810: begin with a configuration where the tachyon field is at its maximum and 

811: has a small velocity. The leading order solution in this case 

812: is proportional to $\sinh(\omg x^0)$ with $\omg=m$. As discussed in 

813: \cite{0203211,0203265}, 

814: the full solution in this case can be found by making the replacement:

815: \be \label{ereplace}

816: x^0 \to x^0 + i \pi/(2\omg), \qquad  \hl \to - i \hl\, ,

817: \ee

818: in the solution we have obtained earlier.

819: \item 

820: The method discussed here 

821: can also be used to generate time dependent solutions describing 

822: oscillation of positive 

823: mass$^2$ fields about the minimum of their potential. In this case there 

824: is no need for Wick rotation. For a field $\phi$ of mass $M$, we take the 

825: first 

826: order solution to be

827: \be \label{ep1}

828: \hl\, \cos(\omg X^0(0)) \, V_\phi(0)c_1|0\ra\, ,

829: \ee

830: where $V_\phi$ is the vertex operator for the zero momentum scalar. 

831: At leading order, $\omg=M$.

832: We can now follow the iterative procedure described in this section to 

833: generate the 

834: solution to arbitrary power of $\hl$, and determine $\omg$ as a function 

835: of $\hl$ to that order. The $\hl$ dependence of $\omg$ will represent 

836: the anharmonicity of the oscillator due to the interaction terms in the 

837: string field theory action.

838: 

839: \item One could consider generalizing the construction to the case where

840: many tachyon fields roll simultaneously. In general, if we have $n$ scalar

841: fields then for a two derivative action we have $2n$ initial conditions.

842: We shall now show that even in string field theory we can construct a $2n$

843: parameter family of solutions describing the rolling of $n$ tachyons.

844: To leading

845: order the solution in the Wick rotated theory is:

846: \be \label{egena1}

847: |\Psi\ra = \sum_{i=1}^n \hl^{(i)} |\phi_0^{(i)}\ra\, ,

848: \ee

849: where

850: \be \label{egena2}

851: |\phi_0^{(i)}\ra = \cos(\omega^{(i)} X(0)+\vp^{(i)}) V_T^{(i)}(0)

852: c_1|0\ra\, .

853: \ee

854: $V_T^{(i)}$ is the vertex operator of the $i$th tachyon with

855: mass$^2=-m_{(i)}^2$, $\{\hl^{(i)}\}$ and $\{ \vp^{(i)}\}$ 

856: are the $2n$ parameters labelling the solution, and to leading order

857: $\omega^{(i)}=m^{(i)}$. We

858: shall assume that the $m^{(i)}$'s are incomensurate.

859: We

860: now generate the higher order solutions by iterating

861: \refb{egena1} as in the case of a single tachyon field, keeping

862: the coefficient of $|\phi^{(i)}_0\ra$ held fixed at $\hl^{(i)}$ to all

863: orders. The analog

864: of $|\chi\ra'$ for a state $|\chi\ra$ of ghost number 2 is defined as the

865: projection of $|\chi\ra$ in the subspace orthogonal to the one spanned by

866: the states $|\phi^{(i)}_0\ra$. Thus comparison of the coefficients of

867: $c_0|\phi^{(i)}_0\ra$ terms in the equation of motion determines

868: $\omega^{(i)}$ as a function of the $\hl^{(j)}$'s, whereas the

869: comparison of the other terms in the equation of motion determine the

870: component of $|\Psi\ra$ in the subspace orthogonal to the one spanned by

871: the $c_0|\phi^{(i)}_0\ra$'s. At the end of the computation we need to make

872: the replacement $X\to -iX^0$, and $\vp^{(i)}\to -i \theta^{(i)}$ where

873: $\theta^{(i)}$ are taken to be real parameters.

874: 

875: For this procedure to work we need to ensure that the operator product of

876: the operators $\cos(\omega^{(i)} X + \vp^{(i)}) V_T^{(i)}$ in the matter

877: sector does not generate another operator of dimension $=\OO(\hl^{(i)})$

878: outside the set $\{\cos(\omega^{(i)} X + \vp^{(i)})V_T^{(i)}\}$. The

879: generic dangerous operators are $\sin(\omega^{(i)} X +

880: \vp^{(i)})V_T^{(i)}$ and $\p

881: X$. The fact that $\sin(\omega^{(i)} X + \vp^{(i)})V_T^{(i)}$ is not

882: generated can be seen as follows.

883: If in a correlation

884: function we have one insertion of $\sin(\omega^{(i)} X + \vp^{(i)})$ and

885: multiple insertions of $\cos(\omega^{(j)} X + \vp^{(j)})$ for different

886: $j$, then we can evaluate this by writing the sines and cosines as sum of

887: exponentials. 

888: As long as the $\omega^{(j)}$'s are incomensurate,\footnote{This

889: restriction on $\omega^{(j)}$ means that the points in the parameter space

890: where perturbation theory fails are dense in the full parameter space, but

891: the situation here is no different from that in perturbation theory in

892: ordinary Hamiltonian system.} a correlator involving

893: $\exp(\pm i (\omega^{(j)}X + \vp^{(j)}))$ will be non-zero only if the

894: $\pm\omega^{(j)} X$ in the exponent cancel pairwise. As a result

895: the accompanying $\pm\vp^{(j)}$ also cancel pairwise. It is now easy to

896: see that for each term in the correlator involving the

897: $e^{i(\omega^{(i)} x + \vp^{(i)})}$ in $\sin(\omega^{(i)} x + \vp^{(i)})$,

898: there is an equal and opposite contribution involving the 

899: $e^{-i(\omega^{(i)} x + \vp^{(i)})}$ term in $\sin(\omega^{(i)} x +

900: \vp^{(i)})$, obtained by reversing the signs of the exponents of all the

901: other terms in the correlator.\footnote{There are many ways to pair terms

902: so that they cancel; here we only mention one of them.}

903: These two terms cancel. Thus such a

904: correlator vanishes, implying in turn that the operator product of

905: $\cos(\omega^{(j)} X + \vp^{(j)})$ does not contain a term proportional to

906: $\sin(\omega^{(i)} X + \vp^{(i)})$.

907: 

908: To see the absence of $\p X$ in the operator product of $\cos(\omega^{(j)}

909: X + \vp^{(j)})$, we consider a correlation function with one insertion of

910: $\p X$ and multiple insertions of various factors of $\cos(\omega^{(j)} X

911: + \vp^{(j)})$ on the boundary of the world-sheet. Since $\p X$ can be

912: thought of as the restriction of a bulk holomorphic current to the

913: boundary, this contribution can be expressed as a sum of poles of $\p X$

914: at the location of the various insertions of $\cos(\omega^{(j)} X +

915: \vp^{(j)})$. The residue of the pole at the insertion of

916: $\cos(\omega^{(k)} X + \vp^{(k)})$ is given by  

917: $\omega^{(k)}\sin(\omega^{(k)} X +

918: \vp^{(k)})$.

919: Thus

920: \ben \label{epp1}

921: &&\bigg\langle \p X(z) \prod_{s=1}^N  \cos(\omega^{(j_s)}

922: X(t_s)  + \vp^{(j_s)})\bigg\rangle \nonumber \\

923: &\propto&\sum_{r=1}^N {\omega^{(j_r)}\over (z-t_r)} \bigg\langle

924: \sin(\omega^{(j_r)}

925: X(t_r) + \vp^{(j_r)})

926: \prod_{s\ne r}  \cos(\omega^{(j_s)}

927: X(t_s) + \vp^{(j_s)})\bigg\rangle \, ,

928: \nonumber \\

929: && \qquad 1\le j_s \le n   

930: \een

931: The result vanishes by previous

932: argument. This, in turn, shows that the correlator of $\p X$ with a set of

933: $\cos(\omega^{(j)} X + \vp^{(j)})$ vanishes and hence we do not generate

934: $\p X$ in the operator product of $\cos(\omega^{(j)} X + \vp^{(j)})$.

935: 

936: Of course there may be other special dimension 1 operators which may be

937: generated in the operator product of $\cos(\omega^{(j)} X + \vp^{(j)})

938: V_T^{(j)}$. In that case the perturbative procedure outlined here for

939: generating solutions of string field theory equations of motion breaks

940: down.

941: 

942: \end{itemize}

943: 

944: \sectiono{Boundary Conformal Field Theory Description} \label{s4}

945: 

946: In this section we shall describe the BCFT associated with the solutions

947: constructed in the previous section. Since in the euclidean theory the

948: final solution is obtained by iterating an initial solution of the form

949: \be \label{e3.1} 

950: \hl \cos(\omg X(0)) \, V_T(0)\, c_1|0\ra\, , 

951: \ee 

952: we

953: might expect that this corresponds to a BCFT that is obtained by

954: perturbing the original BCFT by a boundary term

955: \be \label{e4.1} 

956: \tl \, \int dt

957: \cos(\omg X(t))\, V_T(t)\, , 

958: \ee 

959: where $\tl=\hl+\OO(\hl^2)$ and $t$ is a parameter labelling the boundary

960: of the world-sheet. In order for \refb{e4.1} to describe

961: a BCFT the $\beta$-function of the theory must vanish. Since $\cos(\omg

962: X)\, V_T$ has conformal weight $(\omg^2 + h)=(\omg^2-m^2+1)$, the

963: $\beta$ function for the coupling $\tl$ has the form: 

964: \be \label{e4.2}

965: \beta_{\tl}=(\omg^2-m^2)\tl + g(\omg, \tl)\, , 

966: \ee 

967: where

968: $g(\omg,\tl)$ represents higher order (in $\tl$) contribution to the

969: $\beta$-function. Thus the vanishing of the $\beta$ function 

970: requires:\footnote{See \cite{zam} for a discussion on this. For

971: application of the $\beta$-function equation to the study of tachyon

972: condensation, see \cite{bsft,0010247}.} 

973: \be \label{e4.3} 

974: (\omg^2-m^2)\tl + g(\omg, \tl) =0\, . 

975: \ee 

976: This is equivalent to

977: the equation \refb{e2.9} in string field theory. Of course, in order to

978: show that the perturbed theory is conformal we must ensure that the

979: $\beta$-functions associated with all other operators also vanish. To this 

980: end

981: we note that if $O$ denotes a boundary operator of dimension $h_O$, and if

982: we add this operator with coefficient $\lambda_O$ in the perturbed theory,

983: then the $\beta$-function of the coupling $\lambda_O$ is given by 

984: \be \label{e4.4} 

985: \beta_O = (h_O -1) \lambda_O + g_O(\tl, \lambda_O, \ldots)\, 

986: ,

987: \ee 

988: where $\ldots$ denote the coupling constants associated with the other

989: operators added to the theory, and $g_O$ is the contribution to $\beta_O$

990: from higher order quantum corrections. Since $g_O$ is of order $\tl^2$ or

991: higher, as long as $(h_O-1)$ is of order 1, we can make \refb{e4.4} vanish

992: by choosing a $\lambda_O$ of order $\tl^2$. This procedure breaks down if 

993: $(h_O-1)$ is of order $\tl$ or higher for some operator $O$ other than 

994: $\cos(\omg X) V_T$ (which has already been discussed in eq.\refb{e4.3}). 

995: Thus we require that the operator product of $\cos(\omg X) V_T$ with 

996: itself does not generate an operator of dimension $1+\OO(\tl)$ other

997: than $\cos(\omg X) V_T$. 

998: This is 

999: identical to the condition derived in the previous section for 

1000: obtaining 

1001: a solution of the string field theory equations of motion in a 

1002: perturbation expansion in $\tl$.

1003: 

1004: We shall now describe explicit computation of $\omg$ to order $\tl^2$.

1005: For this we can set $\omg=m$ in the second term in \refb{e4.3}. This 

1006: gives

1007: \be \label{egive}

1008: \omg^2 = m^2 - \tl^{-1} g(m,\tl)\, .

1009: \ee

1010: Thus

1011: we need to compute $g(m,\tl)$.

1012: For this computation we can pretend 

1013: that the $X$-coordinate is

1014: compactified with period $2\pi/m$. In that case $\cos(mX)V_T$, being a

1015: dimension 1 operator, represents a massless scalar field $\vp$ in this

1016: auxiliary

1017: theory. Let $\VV(\vp)$ denote the tree level effective potential for this

1018: scalar field 

1019: obtained after integrating out all the other fields. 

1020: (For $\omg\ne m$, there is also a 

1021: `mass term' ${1\over 2} (\omg^2-m^2)\vp^2$ coming from the 

1022: dimension of 

1023: the operator $\cos(\omg X)V_T$.)

1024: Then

1025: by the usual correspondence between the

1026: equations of motion of

1027: string theory and the condition of conformal invariance, we can compute

1028: the $\beta$-function $g(m,\tl)$ as

1029: \be \label{e5.4}

1030: g(m,\tl) = \VV'(\tl)\, ,

1031: \ee

1032: for suitably normalized $\VV$.

1033: In order to compute $\VV(\vp)$ we simply compute the tree level amplitude

1034: involving the massless scalars $\vp$ in this auxiliary string theory.

1035: First of

1036: all it is clear that there is no $\vp^3$ term in $\VV(\vp)$, since the

1037: three

1038: point amplitude of the scalar field described by the vertex operator

1039: $\cos(mX) V_T$ vanishes due to $X$-momentum conservation. Thus the leading

1040: contribution to $\VV(\vp)$ comes at the quartic order. The coefficient of 

1041: this term can be

1042: identified as $-(\NN^{-1}/4!)$ times the on-shell four point amplitude 

1043: $A^{(4)}$ 

1044: of the

1045: auxiliary scalar

1046: field at zero momentum:

1047: \be \label{e5.4a}

1048: \VV(\vp) = -(\NN^{-1}/4!) \, A^{(4)} \, \vp^4 + \OO(\vp^6)\, ,

1049: \ee

1050: where $\NN$ is the BPZ norm of the state $\cos(mX(0))V_T(0)|0\ra$ in the 

1051: matter sector.\footnote{The simplest way to keep track of the 

1052: normalization factor 

1053: $\NN$ is to first do the computation by assuming $\cos(mX)V_T$ to be a 

1054: normalized operator, and then account for the normalization by 

1055: replacing $A^{(4)}$ by $\NN^{-2} A^{(4)}$ and $\tl$ by $\NN^{1/2}\tl$ in 

1056: \refb{egg2}.}

1057: Thus we get, from \refb{e5.4},

1058: \be \label{egg1}

1059: g(m,\tl) = -(\NN^{-1}/6) \, A^{(4)} \tl^3 + \OO(\tl^5)\, ,

1060: \ee 

1061: and hence from \refb{egive}

1062: \be \label{egg2}

1063: \omg^2 = m^2 + {\NN^{-1}\over 6} \, A^{(4)}\, \tl^2 + \OO(\tl^4)\, .

1064: \ee

1065: This of course agrees with eq.\refb{eexp8} derived from string field 

1066: theory.

1067: 

1068: At this stage we should mention one subtle point. 

1069: Interpreting 

1070: the perturbed theory \refb{e4.1} as a deformed BCFT in the sense of

1071: renormalization group flow makes sense strictly 

1072: if the operator $\cos(\omg X) V_T$ is a relevant perturbation, {\i.e.} if 

1073: $\omg^2< m^2$. {}From \refb{e4.3}, \refb{e5.4} we see that this requires 

1074: $g(\omg, 

1075: \tl)$, or equivalently $\VV(\vp)$, to be positive near $\vp=0$. In the 

1076: analysis of 

1077: string field theory in section 

1078: \refb{s3} we did not encounter such a constraint. If $\VV$ turns out to be 

1079: negative, we can still find a solution in string field theory by choosing 

1080: $\omg^2-m^2$ to be positive. This will correspond to a net potential for 

1081: the auxiliary scalar field

1082: $\vp$ which has a minimum at $\vp=0$ and a maximum at a nearby point, and 

1083: the non-trivial solution of \refb{e4.3} will correspond to the maximum of 

1084: the potential. Thus even for negative $\VV$, we should expect a family of

1085: BCFT labelled by $\tl$, although members of this family are not obtained

1086: as deformation of the original BCFT by a relevant perturbation. Rather,

1087: the original BCFT is obtained as a relevant deformation of each member of

1088: this family.

1089: 

1090: When $\omega^2< m^2$ so that the perturbing operator is relevant, it is

1091: more natural to treat $\omega$ rather than $\lambda$ as the independent

1092: parameter. We simply deform the theory by the operator $\cos(\omega X)

1093: V_T$, and consider the boundary CFT to which the theory flows in the

1094: infrared. This will give a family of boundary CFT's labeled by $\omega$.

1095: Inverse Wick rotation of these boundary CFT's then give us the family of

1096: time dependent solutions corresponding to different initial conditions.

1097: 

1098: As special examples of the general class of perturbed BCFT's discussed 

1099: above, we can consider the case where

1100: $g(m, \tl)$ vanishes for arbitrary $\tl$, and as a 

1101: result \refb{e4.3} is satisfied for $\omg=m$ for all $\tl$. In this case 

1102: the 

1103: operator $\cos(mX)\, V_T$ is an exactly marginal operator. The cases 

1104: analysed in refs.\cite{0203211,0203265} are of this type.

1105: 

1106: Given a BCFT obtained by perturbing the euclidean theory by the operator

1107: \refb{e4.1}, we can construct a BCFT by the inverse Wick rotation of this

1108: perturbed theory. This will correspond to adding a perturbation

1109: \be \label{e4.5}

1110: \tl \cosh(\omg X^0(0)) \, V_T(0)\, c_1|0\ra\, ,

1111: \ee

1112: to the world-sheet theory with Minkowski signature space-time. The 

1113: boundary state\cite{boundary} associated with this BCFT, containing

1114: information about the 

1115: time evolution of the energy-momentum tensor\cite{em-tensor,0203265}, can

1116: be

1117: obtained from Wick 

1118: rotation of the boundary state associated with the euclidean theory. Time

1119: evolution of the sources of other massless closed string fields, {\it

1120: e.g.} the dilaton, the anti-symmetric tensor field $B_{\mu\nu}$, and

1121: various Ramond-Ramond fields, can also be extracted from the boundary

1122: state.

1123: 

1124: To see more explicitly how we can extract the energy-momentum tensor from

1125: the boundary state, we recall that\cite{0203265} if part of the

1126: boundary state 

1127: involving oscillators of level (1,1) acting on the ghost number 3 vacuum 

1128: has the form

1129: \be \label{e5.17}

1130: \int d^{26} k \, [\wt A_{\mu\nu}(k) \alpha^\mu_{-1}

1131: \bar\alpha^\nu_{-1} 

1132: + \wt B(k) (\bar

1133: b_{-1}

1134: c_{-1} + b_{-1} \bar c_{-1})] (c_0+\bar c_0)c_1\bar c_1 |k\ra\, ,

1135: \ee

1136: then the energy momentum tensor $T_{\mu\nu}$ is given by:

1137: \be \label{e5.18}

1138: T_{\mu\nu}(x) = K (A_{\mu\nu}(x) + \eta_{\mu\nu} B(x) )\, ,

1139: \ee

1140: where $A_{\mu\nu}(x)$ and $B(x)$ are Fourier transforms of $\wt

1141: A_{\mu\nu}(k)$ and $\wt B(k)$ respectively, and $K$ is a $\tl$-independent

1142: constant. 

1143: In the Wick rotated theory the energy momentum tensor computed using this

1144: procedure will have the form:

1145: \be \label{eem1}

1146: T_{\mu\nu}(x, \vec x) = \sum_{n=0}^\infty T^{(n)}_{\mu\nu}(\vec x)

1147: \cos(n\omega x)\, ,

1148: \ee

1149: where $\vec x$ denotes the spatial coordinates of the original theory.

1150: Then after Wick rotation we have:

1151: \ben \label{eem2}

1152: T_{00}(x^0, \vec x) &=& - \sum_{n=0}^\infty T^{(n)}_{xx}(\vec

1153: x)\cosh(n\omega

1154: x^0)\, , \nonumber \\

1155: T_{0i}(x^0, \vec x) &=& i \sum_{n=0}^\infty T^{(n)}_{xi}(\vec 

1156: x)\cosh(n\omega x^0)\, , \nonumber \\

1157: T_{ij}  &=& \sum_{n=0}^\infty T^{(n)}_{ij}(\vec

1158: x)\cosh(n\omega x^0)\, .

1159: \een

1160: Since in the Wick rotated theory the sum over $n$ in \refb{eem1} is

1161: expected to converge

1162: and generate a finite energy-momentum tensor, we expect that after Wick

1163: rotation the sum over $n$ in \refb{eem2} will converge for sufficiently

1164: small $x^0$ and $\tl$. We

1165: do expect the sum to diverge for large $x^0$, and checking

1166: whether the results for small $x^0$ can be analytically continued beyond

1167: the range of convergence or not, we shall have to determine whether during 

1168: the time

1169: evolution the energy momentum tensor hits a real singularity at a

1170: finite value of $x^0$. 

1171: 

1172: In principle, we should also be able to extract the energy-momentum tensor

1173: from the solution of the string field theory equations of motion

1174: constructed in the previous section, since this solution is expected to

1175: have all the information about the deformed boundary CFT. At present

1176: however it is not known how to do this.

1177: 

1178: \sectiono{Explicit Example} \label{s5}

1179: 

1180: In this section we shall discuss an explicit example of the construction 

1181: described in the previous section. The system that we shall consider is a 

1182: D-$p$-brane of bosonic string theory, with one direction wrapped on a 

1183: circle of radius $R>1$. 

1184: This has the usual tachyonic mode 

1185: of mass$^2=-1$, whose time evolution was discussed in \cite{0203211}. But 

1186: this also has another tachyonic mode of mass$^2$

1187: \be \label{em.1}

1188: R^{-2}-1\equiv -m^2

1189: \ee

1190: coming from 

1191: the 

1192: first momentum mode of the standard tachyon along the circle. If we denote 

1193: by $y$ the coordinate along the circle, then the matter sector vertex

1194: operator 

1195: associated with this tachyonic mode (with zero momentum along the

1196: non-compact directions) can be taken to be:

1197: \be \label{e5.1}

1198: V_T = \cos(Y/R)\, .

1199: \ee

1200: We shall discuss the rolling of this tachyon 

1201: when it is displaced from its maximum. As usual we denote by $x$ the Wick 

1202: rotated time coordinate. The conformal field theory 

1203: associated with the 

1204: (Wick rotated) rolling tachyon solution is obtained by perturbing the 

1205: original free conformal field theory by the boundary operator

1206: \be \label{e5.2}

1207: \tl\,  \int dt \, \cos(\omg X(t)) \, \cos(Y(t)/R)\, .

1208: \ee

1209: Eq.\refb{e4.3} now takes the form:

1210: \be \label{e5.3}

1211: (\omg^2 + R^{-2} - 1)\tl + g(\omg,\tl) = 0\, .

1212: \ee

1213: To leading order the solution to this equation is 

1214: $\omg=m=\sqrt{1-R^{-2}}$. 

1215: {}From eq.\refb{egg2} we see that the computation of $\OO(\tl^2)$ 

1216: correction to $\omega$ 

1217: requires us to compute $A^{(4)}$ involving four external states associated 

1218: with the matter sector vertex operator $\cos(m X) \cos(Y/R)$.

1219: This is the task to which 

1220: we shall now turn.

1221: 

1222: Expressing the scalar field vertex operator as

1223: \be \label{e5.5}

1224: \cos(mX) \cos(Y/R) = {1\over 4} (e^{imX}+e^{-imX}) \, 

1225: (e^{iY/R} + e^{-iY/R})\, ,

1226: \ee

1227: we can relate the four point amplitude $A^{(4)}$ of the scalar field to 

1228: the 

1229: Veneziano amplitude involving external states associated with 

1230: matter sector vertex operators $e^{\pm i m X} e^{\pm i Y/R}$.

1231: Since the right hand side of eq.\refb{e5.5} has four terms, there are

1232: altogether $4^4=256$ terms involved in the computation of the four point

1233: amplitude $A^{(4)}$. However all but 36 of them

1234: vanish by $X$- or $Y$-momentum conservation. 12 of the non-vanishing terms

1235: are each given by ${1\over 4^4} V(s=-4, t=0)$, and each of the rest 24

1236: terms is given by ${1\over 4^4} V(s=-4R^{-2}, t = -4(1 - R^{-2}))$, where

1237: $V(s,t)$ is the Veneziano amplitude

1238: \ben \label{e5.6}

1239: V(s,t) &=& 2\, \bigg( {\Gamma(-1-s) \Gamma(-1-t)\over \Gamma(-2-s-t)}

1240: + {\Gamma(-1-s) \Gamma(-1-u)\over \Gamma(-2-s-u)}

1241: +{\Gamma(-1-u) \Gamma(-1-t)\over \Gamma(-2-u-t)} \bigg), \nonumber \\ \cr

1242: && \qquad \qquad u=-4-s-t\, .

1243: \een

1244: Thus we get

1245: \ben \label{e5.7}

1246: A^{(4)} &=& {1\over 4^4} \, ( 12 V(s=-4, t=0) + 24 V(s=-4 R^{-2}, 

1247: t=-4(1-R^{-2})) \nonumber \\

1248: &=& -{3\over 32} \, (R^2 - 2) [ 4 R^{-2} \ff'(4R^{-2}) / \ff(4R^{-2}) + 

1249: \ff(4 

1250: R^{-2}) - 1]\, ,

1251: \een

1252: where

1253: \be \label{e5.8}

1254: \ff(x) = \Gamma(1-x) \, \Gamma(1+x) = {\pi x\over \sin(\pi x)}\, .

1255: \ee

1256: Using eqs.\refb{egg2}, \refb{e5.7}, and the fact that $\NN=1/4$ for the

1257: operator \refb{e5.5}, we now get

1258: \be \label{e5.10}

1259: \omg^2 = 1 - R^{-2} - {1\over 16} (R^2-2)  [ 4 R^{-2} \ff'(4R^{-2}) / 

1260: \ff(4R^{-2}) + \ff(4

1261: R^{-2}) - 1]\tl^2 + \OO(\tl^4)\, .

1262: \ee

1263: This determines the value of $\omg$ to order $\tl^2$ for which 

1264: perturbation by the vertex 

1265: operator $\cos(\omg X) \cos(Y/R)$ generates a boundary CFT. The time 

1266: dependent solution describing the rolling tachyon is then 

1267: determined by the inverse Wick rotation $X=-iX^0$ of this solution. In 

1268: particular by inverse Wick rotating the boundary state associated with the 

1269: perturbed euclidean BCFT, we can determine the energy-momentum tensor 

1270: associated with the time dependent solution.

1271: 

1272: Unfortunately for generic $R$ the perturbed BCFT is not solvable, and 

1273: hence 

1274: we cannot explicitly compute the boundary state associated with this BCFT. 

1275: For $R\to\infty$ the amplitude given in eq.\refb{e5.7} vanishes. This is 

1276: consistent with the fact that in this limit the perturbing operator 

1277: $\cos(X)$ is exactly marginal and hence the corresponding $\beta$-function 

1278: must vanish. The associated conformal field theory has been

1279: discussed in detail in \cite{marginal} and its inverse Wick rotation has

1280: been used

1281: in \cite{0203211,0203265} to describe the rolling of a spatially

1282: homogeneous tachyon field on the D-brane. 

1283: $A^{(4)}$ also vanishes at $R=1$ where the perturbing

1284: operator is $\cos(Y)$. There is another value of $R$

1285: where $A^{(4)}$ vanishes, 

1286: namely at $R=\sqrt 2$.\footnote{Note that $\ff(4R^{-2})$ diverges at 

1287: $R=\sqrt 2$ and so in evaluating \refb{e5.10} for $R=\sqrt{2}$ we need to 

1288: carefully take the limit. Nevertheless the right hand side of \refb{e5.10} 

1289: can be shown to vanish at $R=\sqrt 2$.}

1290: Thus one might expect that the corresponding perturbation becomes exactly 

1291: marginal at this point. To see that this is indeed the case, let us note 

1292: that at $\omega=1/\sqrt 2$, the perturbing operator is

1293: given by:

1294: \be \label{e5.11}

1295: \tl \, \int dt \, \cos(X(t)/\sqrt 2) \, \cos(Y(t)/\sqrt 2)

1296: = {\tl\over 2} \, \Big[\cos\Big((X(t) + Y(t))/\sqrt 2\Big) + 

1297: \cos\Big((X(t) 

1298: -

1299: Y(t))/\sqrt 2\Big)\Big]\, .

1300: \ee

1301: This corresponds to perturbing the theory by a sum of two commuting 

1302: exactly marginal 

1303: deformations, and hence represents a marginal deformation. The associated 

1304: BCFT is exactly solvable. In fact, defining

1305: \be \label{e5.12}

1306: Z^1 = (X+Y)/\sqrt 2 \, , \qquad Z^2 = (X-Y)/\sqrt 2\, ,

1307: \ee

1308: we see that the net boundary state is given by the product

1309: \be \label{e5.13}

1310: |\BB\ra_{Z^1} \otimes |\BB\ra_{Z^2} \otimes |\BB\ra_{c=24} \otimes 

1311: |\BB\ra_{ghost}\, .

1312: \ee

1313: $|\BB\ra_{c=24}$ denotes the boundary state associated with the 24 free 

1314: scalar fields, and $|\BB\ra_{ghost}$ denotes the boundary state associated 

1315: with the ghost fields. These are given by their free field expressions. On 

1316: the 

1317: other hand the relevant part of the boundary states $|\BB\ra_{Z^1}$ and $ 

1318: |\BB\ra_{Z^2}$, associated with the perturbed BCFT associated with the 

1319: scalar fields $Z^1$ and $Z^2$, can be read out from the results of 

1320: \cite{marginal,9811237,0108238,0203265}. If we consider for simplicity the

1321: case of 

1322: the 

1323: D-25-brane, the part of the boundary state, relevant for the computation 

1324: of the energy-momentum tensor, is proportional to:

1325: \ben \label{e5.14}

1326: && \bigg[f\Big((iX(0)+iY(0))/\sqrt 

1327: 2\Big) 

1328: - {1\over 2} (\alpha^X_{-1} + \alpha^Y_{-1}) (\bar\alpha^X_{-1} + 

1329: \bar\alpha^Y_{-1}) g\Big((iX(0)+iY(0))/\sqrt 2\Big) \bigg] \nonumber \\

1330: && \bigg[f\Big((iX(0)-iY(0))/\sqrt 2\Big)

1331: - {1\over 2} (\alpha^X_{-1} - \alpha^Y_{-1}) (\bar\alpha^X_{-1} -

1332: \bar\alpha^Y_{-1}) g\Big((iX(0)-iY(0))/\sqrt 2\Big) \bigg] \nonumber \\

1333: && \bigg[1-\alpha^i_{-1}\bar\alpha^i_{-1}\bigg]  \, \bigg[1-\bar 

1334: b_{-1}

1335: c_{-1} - b_{-1} \bar c_{-1}\bigg] (c_0+\bar c_0)c_1\bar c_1 |0\ra\, .

1336: \een

1337: Here $\alpha^i_{-1}$, $\bar\alpha^i_{-1}$ for $2\le 

1338: i\le 25$ are the 

1339: oscillators associated with the spectator bosons $X^i$,

1340: $\alpha^X_{-1}$, $\bar\alpha^X_{-1}$, $\alpha^Y_{-1}$, $\bar\alpha^Y_{-1}$ 

1341: are the oscillators associated with the scalars $X$ and $Y$, and $b_n$,

1342: $c_n$, $\bar b_n$, $\bar c_n$ are the oscillators associated with the

1343: ghost fields $b$, $c$, $\bar b$, $\bar c$.

1344: The functions $g(x)$ and $f(x)$ have been given in \cite{0203211,0203265}:

1345: \be \label{e5.15}

1346: f(x^0)={1\over 1 + e^{x^0} \sin(\tl\pi/2)} + {1 \over

1347: 1 + e^{-x^0} \sin(\tl\pi/2)} - 1\, , \qquad

1348: g(x^0) = \cos(\tl\pi) +1 - f(x^0)\, .

1349: \ee

1350: Note that these formul\ae\ differ from that of \cite{0203211,0203265} by a 

1351: replacement $\tl\to \tl/2$, whose origin can be traced to the explicit 

1352: factor of $1/2$ multiplying $\tl$ on the right hand side of \refb{e5.11}.

1353: 

1354: We can now make an inverse Wick rotation $X=-iX^0$ to get the boundary 

1355: state in the Minkowski signature theory:

1356: \ben \label{e5.16}

1357: && \bigg[f\Big((X^0(0)+iY(0))/\sqrt 

1358: 2\Big) 

1359: - {1\over 2} (-i\alpha^0_{-1} + \alpha^Y_{-1}) (-i\bar\alpha^0_{-1} + 

1360: \bar\alpha^Y_{-1}) g\Big((X^0(0)+iY(0))/\sqrt 2\Big) \bigg] \nonumber \\

1361: && \bigg[f\Big((X^0(0)-iY(0))/\sqrt 2\Big)

1362: - {1\over 2} (-i\alpha^0_{-1} - \alpha^Y_{-1}) (-i\bar\alpha^0_{-1} -

1363: \bar\alpha^Y_{-1}) g\Big((X^0(0)-iY(0))/\sqrt 2\Big) \bigg] \nonumber \\

1364: && \bigg[1-\alpha^i_{-1}\bar\alpha^i_{-1}\bigg]  \, \bigg[1-\bar 

1365: b_{-1}

1366: c_{-1} - b_{-1} \bar c_{-1}\bigg] (c_0+\bar c_0)c_1\bar c_1 |0\ra\, .

1367: \een

1368: Comparing \refb{e5.16} with \refb{e5.17} we get,

1369: \ben \label{e5.19}

1370: A_{00} &=& {1\over 2} \, [g((x^0 + iy)/\sqrt 2) f((x^0 - i

1371: y)/\sqrt

1372: 2)  + g((x^0 - i 

1373: y)/\sqrt 

1374: 2) f ((x^0 + iy)/\sqrt 2)]\, , \nonumber 

1375: \\

1376: A_{0y} &=& A_{y0} = {i\over 2}  [g((x^0 + iy)/\sqrt 2) f((x^0 - i 

1377: y)/\sqrt 

1378: 2)  - 

1379: g((x^0 - i 

1380: y)/\sqrt

1381: 2)  f ((x^0 + iy)/\sqrt 2)]\, , \nonumber

1382: \\

1383: A_{yy} &=& -  {1\over 2} \,  [g((x^0 + iy)/\sqrt 2)f((x^0 - i

1384: y)/\sqrt

1385: 2) + g((x^0 - 

1386: i 

1387: y)/\sqrt

1388: 2)  f ((x^0 + iy)/\sqrt 2) ]\, , \nonumber

1389: \\

1390: A_{ij} &=& - \, \delta_{ij} \, 

1391: f ((x^0 + iy)/\sqrt 2)  f((x^0 - i

1392: y)/\sqrt

1393: 2)

1394: \, , \nonumber

1395: \\

1396: B &=& -  f ((x^0 + iy)/\sqrt 2)  f((x^0 - i

1397: y)/\sqrt

1398: 2)\, .

1399: \een

1400: Using \refb{e5.18}, \refb{e5.15} and \refb{e5.19} we get

1401: \ben \label{e5.20}

1402: T_{00} &=& K (A_{00} - B) = {K\over 2} (\cos(\tl\pi)+1) [f ((x^0 + 

1403: iy)/\sqrt 

1404: 2) + f((x^0 - i

1405: y)/\sqrt

1406: 2)]\, , \nonumber \\

1407: T_{0y} &=& K A_{0y} = {i K\over 2}  (\cos(\tl\pi)+1) [f((x^0 - i

1408: y)/\sqrt

1409: 2) - f ((x^0 +

1410: iy)/\sqrt

1411: 2) ] \nonumber \\

1412: T_{yy} &=&  K (A_{yy} + B) = -{K\over 2} (\cos(\tl\pi)+1) [f ((x^0 +

1413: iy)/\sqrt

1414: 2) + f((x^0 - i

1415: y)/\sqrt

1416: 2)]\, , \nonumber \\  

1417: T_{ij} &=& K(A_{ij} + B \delta_{ij})

1418: =  - 2 K\, \delta_{ij} \, f ((x^0 +

1419: iy)/\sqrt

1420: 2) \, f((x^0 - i

1421: y)/\sqrt

1422: 2)

1423: \een

1424: All other components of $T_{\mu\nu}$ vanish.

1425: {}From these expressions one can easily verify the conservation laws:

1426: \be \label{e5.21}

1427: \p_0 T_{00} - \p_y T_{0y} = 0, \qquad \p_0 T_{0y} - \p_y T_{yy} = 0\, .

1428: \ee

1429: If we consider a general D-$p$-brane with $p<25$, then $T_{ij}$ associated

1430: with the $(25-p)$ transverse directions vanish. Furthermore the

1431: non-vanishing components of $T_{\mu\nu}$ are each multiplied by a delta

1432: function involving the transverse coordinates. The $\tl$ independent

1433: constant $K$ is computed by requiring that for $\tl=0$, $T_{00}$ should

1434: reproduce the tension $\TT_p$ of a D-$p$-brane. This gives

1435: \be \label{ektp}

1436: K = {\TT_p\over 2}\, .

1437: \ee

1438: 

1439: {}From \refb{e5.15} we get

1440: \be \label{e5.22}

1441: f((x^0\pm i y)/\sqrt 2) = 

1442: {1\over 1 + e^{x^0/\sqrt 2} e^{\pm iy/\sqrt 2} \sin(\tl\pi/2)} + {1 \over

1443: 1 + e^{-x^0/\sqrt 2}  e^{\mp iy/\sqrt 2} \sin(\tl\pi/2)} - 1\, .

1444: \ee

1445: Thus $f((x^0\pm iy)/\sqrt 2)$ and hence

1446: the energy momentum tensor hits a singularity at

1447: \be \label{e5.23}

1448: x^0 = \sqrt 2 \, \ln \bigg|{1\over \sin(\tl\pi/2)}\bigg| \equiv x^0_c,

1449: \qquad y = \cases {\sqrt 2 \pi\, , \qquad \hbox{for} \quad \tl>0\, , \cr

1450: 0\, , \qquad \, \, \quad \, \hbox{for} \quad \tl<0\, .}

1451: \ee

1452: In particular the energy-density at the singular point approaches 

1453: $\infty$. Thus we cannot naively evolve the system beyond this time using

1454: the 

1455: classical open string field equations.

1456: 

1457: \begin{figure}[!ht]

1458: \leavevmode

1459: \begin{center}

1460: \epsfysize=8cm

1461: \epsfbox{fig2.eps}

1462: \end{center}

1463: \caption{The contour plot of the time evolution of the energy density

1464: $T_{00}$ computed from eq.\refb{e5.20}, \refb{e5.22} for $\tl=-0.1$. The

1465: horizontal axis

1466: is the $y$ axis, the vertical axis is the time ($x^0$) axis, and the

1467: shading at a

1468: given point $(y,t)$ shows the value of $T_{00}$, with darker shade

1469: representing higher energy density. We see from this that the $T_{00}$

1470: hits a singularity at $x^0\simeq 2.62$, $y=0$. At this space-time point, 

1471: the system

1472: loses a net amount of energy equal to the total initial energy of the

1473: system.  Immediately after this

1474: singularity the energy density becomes negative around $y=0$, as shown by 

1475: the lighter regions, and eventually the energy density approaches zero

1476: everywhere.} \label{f2} \end{figure}

1477: It is instructive to examine the nature of the singularity in a little 

1478: more detail. Using

1479: eqs.\refb{e5.20}, \refb{e5.22} we get the total energy of the system to 

1480: be:

1481: \be \label{etote}

1482: \EE = \int_0^{2\pi\sqrt 2} dy \, T_{00} = 2\pi\sqrt 2 K 

1483: (1 + \cos(\tl\pi))\, ,

1484: \ee

1485: for $x^0< x^0_c$. If we

1486: naively 

1487: continue the expression for $T_{00}$ to $x^0>x^0_c$, we find 

1488: that the total energy vanishes. In order to see the origin of this 

1489: `violation' of energy conservation, we can examine the form of the 

1490: singularity at $x^0 = x^0_c$, 

1491: $y=0$ (concentrating for definiteness on the $\tl<0$ case.) Defining

1492: \be \label{defth}

1493: \Theta \equiv e^{(x^0 - x^0_c)/\sqrt 2} = -e^{x^0/\sqrt 2}

1494: \sin(\tl\pi)\simeq 1

1495: \ee

1496: near the singularity, the singular part of $T_{00}$ near $(1-\Theta)\sim y

1497: \simeq 0$ is given 

1498: by:

1499: \be \label{esingpart}

1500: (T_{00})_{singular} = K(1 + \cos(\tl\pi)) \, {(1-\Theta) \over 

1501: (1-\Theta)^2 + {1\over 2} y^2} \simeq \, {\rm sgn}(1-\Theta) \, \pi

1502: \sqrt{2} \, K

1503: (1 + \cos(\tl\pi)) \, 

1504: \delta(y)\, .

1505: \ee

1506: Since $(1-\Theta)$ changes sign as $x^0$ passes from below to above

1507: $x^0_c$, there is a net loss of energy at $y=0$ at $x^0=x^0_c$,

1508: The amount of energy lost at $y=0$ as $x^0$ passes from below $x^0_c$

1509: to above $x^0_c$ is $2\pi \sqrt{2} \, K (1 + \cos(\tl\pi))$, -- same

1510: as the total 

1511: energy of the system. As a result

1512: the average of the energy 

1513: density in the rest of the system must vanish after the critical time

1514: $x^0_c$, as has already been seen in explicit calculation. If we

1515: continue the 

1516: formula for $T_{00}$ beyond this critical time, the system eventually

1517: evolves to a configuration with zero 

1518: energy density everywhere.\footnote{Of course, near the singularity

1519: $x^0=x^0_c$, $y=0$, the back reaction of closed string fields will be

1520: important, and a realistic analysis of the dynamics must include this

1521: effect.} There is an apparent violation of energy conservation at

1522: $x^0 = x^0_c$, but

1523: the most conservative interpretation of this is simply that we create a 

1524: codimension 1 D-brane at $y=0$, with a delta function source of energy

1525: that is not visible in the formula \refb{e5.20}. The total energy stored

1526: at $y=0$ does not 

1527: agree with the tension of a codimension 1 D-brane; in particular the ratio

1528: of the total energy of the system to that of a codimension 1 D-brane is

1529: given by:

1530: \be \label{eratio}

1531: {1\over \sqrt 2} (1 + \cos(\pi\tl))\, .

1532: \ee

1533: However the final state

1534: could represent a 

1535: state of the codimension 1 D-brane with non-trivial time dependent tachyon

1536: configuration, with the excess (deficit) energy stored in the 

1537: tachyon field. Fig. \ref{f2} shows the time evolution of $T_{00}$ given in

1538: eqs.\refb{e5.20}, \refb{e5.22} for a specific value of $\tl$.

1539: 

1540: 

1541: 

1542: \sectiono{Time Dependent Solutions on D-branes in Superstring 

1543: Theories} \label{s6}

1544: 

1545: The analysis of the previous sections can be easily generalized to the 

1546: case of superstring theories. Let us consider a generic D-brane system 

1547: with a tachyonic mode, and let the zero momentum tachyon vertex operator 

1548: have the form $e^{-\phi} V^{(-1)}_T$ in the $-1$ picture\cite{FMS}, where

1549: $\phi$ 

1550: denotes the bosonized ghost field and $V^{(-1)}_T$ denotes the matter part 

1551: of the vertex operator. If $V^{(-1)}_T$ has dimension $h<1/2$, then the 

1552: tachyon has mass$^2=(h-{1\over 2})\equiv -m^2$ and at the linearized 

1553: level the time dependent tachyon solution takes the form $\tl\cosh(m 

1554: x^0)$. In the Wick rotated theory this corresponds to $\tl\cos(mx)$, and 

1555: amounts to perturbing the BCFT by a term

1556: \be \label{e6.1}

1557: \tl \int dt\, d\theta \cos(m X(t,\theta)) V^{(-1)}_T(t,\theta)\, ,

1558: \ee

1559: where $t$ denotes the coordinate labelling the world-sheet boundary, 

1560: $\theta$ is the fermionic coordinate on the superspace, and we 

1561: consider $X$ and 

1562: $V_T^{(-1)}$ as 

1563: superfields. Although to lowest order in $\tl$ this represents a marginal 

1564: deformation, in general once higher order corrections are taken into 

1565: account this no longer represents a marginal deformation. We remedy this 

1566: by modifying \refb{e6.1} to 

1567: \be \label{e6.2}

1568: \tl \int dt \, d\theta \cos(\omg X(t,\theta)) V^{(-1)}_T(t,\theta)\, ,

1569: \ee

1570: where $\omg$ is a constant. The full $\beta$-function associated with 

1571: the coupling $\tl$ now takes the form:

1572: \be \label{e6.3}

1573: \omg_{\tl} = (\omg^2 - m^2)\tl + g(\omg,\tl)\, ,

1574: \ee

1575: as in \refb{e4.2}. We determine $\omg$ for a given $\tl$ by 

1576: demanding that $\beta_{\tl}$ vanishes. This gives:

1577: \be \label{e6.4}

1578: (\omg^2 - m^2)\tl + g(\omg,\tl) = 0\, .

1579: \ee

1580: This determines $\omg$ as a function of $\tl$. Thus for example the 

1581: leading correction to $\omg$ is given by

1582: \be \label{el1}

1583: \omg^2 \simeq m^2 - g(m, \tl) / \tl\, .

1584: \ee

1585: The $\beta$-functions associated with the other operators can be made to 

1586: vanish as in section \ref{s4} provided the product of \refb{e6.2} 

1587: with itself does not generate integral of any 

1588: superfield of dimension $1/2+\OO(\tl)$ other than the superfield 

1589: $\cos(\omg

1590: X(t,\theta)) V^{(-1)}_T(t,\theta)$ itself.

1591: 

1592: For an explicit example we can consider the case analogous to the one 

1593: discussed in section \ref{s5} for open bosonic string theory. We take a 

1594: non-BPS D-$p$-brane with one direction compactified on a circle of radius 

1595: $> {\sqrt 2}$. Let $y$ denote this compact direction. In this case the 

1596: first momentum mode of the tachyon along the circle, described by the 

1597: wave-function $\cos(y/R)$, can be thought of as a scalar field living

1598: in $((p-1)+1)$ 

1599: dimensions with mass$^2$:

1600: \be \label{e6.5}

1601: {1\over R^2} - {1\over 2} \equiv -m^2\, .

1602: \ee

1603: Displacing this tachyonic mode from its maximum and letting it roll 

1604: amounts to displacing the full $(p+1)$ dimensional tachyon from its 

1605: maximum by an amount proportional to $\cos(y/R)$ and let it roll. Since 

1606: the sign of the displacement is different for different values of $y$, we 

1607: expect that the time evolution of the tachyon will produce a kink-antikink 

1608: pair on the D-$p$-brane world volume, centered around $y=\pm\pi R/2$.

1609: 

1610: We shall study the time evolution by viewing this particular mode of the 

1611: tachyon as a scalar field in $((p-1)+1)$ dimensions.

1612: The vertex operator of this zero momentum tachyon in the $-1$ picture is 

1613: $e^{-\phi}\cos(Y/R)\otimes \sigma_1$

1614: where $\sigma_1$ is the Chan-Paton factor\cite{9808141,9809111}.

1615: Thus in order to construct the Wick rotated version of the rolling tachyon 

1616: solution, we need to perturb the BCFT by the operator

1617: \be \label{e6.6}

1618: \tl \int dt d\theta \cos(\omg X(t,\theta)) \cos(Y(t,\theta)/R)\otimes 

1619: \sigma_1\, .

1620: \ee

1621: If we expand the superfields $X(t,\theta)$ and $Y(t,\theta)$ as

1622: \be \label{e6.7}

1623: X(t,\theta) = X(t) + \theta\psi_x(t), \qquad

1624: Y(t,\theta) = Y(t) + \theta\psi_y(t),

1625: \ee

1626: where $\psi_x$ and $\psi_y$ are the world-sheet fermionic partners of the 

1627: $X$ and the $Y$ fields respectively, 

1628: then the perturbation can be expressed as

1629: \be \label{e6.8}

1630: -\tl \int dt [\omg \sin(\omg X(t)) \cos(Y(t)/R) + R^{-1} 

1631: \cos(\omg X(t)) \sin(Y(t)/R)]\otimes \sigma_1\, .

1632: \ee

1633: As in section \ref{s5}

1634: we can relate the computation of $g(m,\tl)$ appearing in 

1635: eq. \refb{el1} to leading order in $\tl$

1636: to the computation of the four point amplitude of four 

1637: on-shell external states, each described the the vertex operator 

1638: $e^{-\phi} \cos(m X) \cos(Y/R)\otimes \sigma_1$. This in turn will 

1639: determine the leading 

1640: correction to $\omg$ via eq.\refb{el1}. 

1641: We shall not carry out the detailed computation here. 

1642: 

1643: We can consider the special value of the radius $R=2$. In this case 

1644: $m=1/2$, and to leading order when we set $\omg=m$ the perturbation 

1645: \refb{e6.6} takes the form:

1646: \ben \label{e6.9}

1647: &&\tl \int dt d\theta \cos(X(t,\theta)/2) \cos(Y(t,\theta)/2)\otimes 

1648: \sigma_1 \nonumber \\

1649: &=& {\tl\over 2} \int dt \, d\theta \, [ \cos(Z_1(t,\theta)/\sqrt 2) + 

1650: \cos(Z_2(t,\theta)/\sqrt 2) 

1651: ]\otimes \sigma_1\, ,

1652: \een

1653: where

1654: \be \label{e6.10}

1655: Z_1 = (X+Y)/\sqrt 2, \qquad Z_2 = (X-Y)/\sqrt 2\, .

1656: \ee

1657: Writing

1658: \be \label{e6.11}

1659: Z_i(t,\theta) = Z_i(t) + \theta \psi_i(t)\, ,

1660: \ee

1661: we can rewrite the perturbation as

1662: \be \label{e6.12}

1663: -{\tl\over 2\sqrt 2} \int dt [\psi_1(t) \sin (Z_1(t)/\sqrt 2)

1664: + \psi_2(t) \sin (Z_2(t)/\sqrt 2)]\otimes \sigma_1\, .

1665: \ee

1666: Following the results of \cite{9808141,0003124} one can show that each of 

1667: the two terms 

1668: in \refb{e6.12} represents an exactly marginal deformation. Unfortunately 

1669: however 

1670: these two terms do not commute, instead they anti-commute due to the 

1671: presence of the fermion fields $\psi_i$. As a result the complete 

1672: perturbation is not exactly marginal and the $\beta$-function does receive 

1673: higher order corrections. Thus unlike in the case of section \ref{s5}, we 

1674: cannot solve the theory exactly in this case.

1675: In particular,

1676: we cannot obtain an analytic expression for the 

1677: boundary state and compute the energy momentum tensor associated with 

1678: this solution.

1679: 

1680: To see this more explicitly we fermionize  and rebosonize following the 

1681: procedure given in refs:\cite{9808141,0003124}. On the boundary the 

1682: relevant part of these 

1683: relations take the form:

1684: \be \label{e6.13}

1685: e^{iZ_i(t)/\sqrt 2} = {1\over \sqrt 2} (\xi^i(t) + i \eta^i(t)) \otimes 

1686: \tau_i\, ,

1687: \ee

1688: \be \label{e6.15}

1689: \psi_i(t) \eta_i(t) = -{i\over \sqrt 2} \p \phi_i(t)\, ,

1690: \ee

1691: where $\xi_i$ and $\eta_i$ are Majorana fermions, $\phi_i$ are scalars and 

1692: $\tau_i$ are Pauli matrices providing the cocycles needed for 

1693: the bosonization. In terms of the bosonic field $\phi_i$ the perturbation 

1694: \refb{e6.12} can be written as:

1695: \be \label{e6.16}

1696: {i\tl\over 4\sqrt 2} \int dt [\p_t\phi_1 \otimes \tau_1\otimes \sigma_1

1697: + \p_t\phi_2 \otimes \tau_2 \otimes \sigma_1]\, .

1698: \ee

1699: Since $\tau_1\otimes \sigma_1$ does not commute with 

1700: $\tau_2\otimes\sigma_1$, the above perturbation can be thought of as 

1701: switching on a pair of non-commuting Wilson lines, one along $\phi_1$ and 

1702: the other along $\phi_2$, each being of magnitude $\tl/4\sqrt 2$. As a 

1703: result the 

1704: potential does not vanish. Instead we get a contribution proportional to

1705: the square of the commutator, which in this case will be equal to $C\tl^4$ 

1706: for 

1707: an approprite positive constant $C$. This gives a contribution to 

1708: $g(m,\tl)$ equal 

1709: to $4 C\tl^3$. Eq.\refb{el1} then gives:

1710: \be \label{e6.17}

1711: \omg^2 = {1\over 4} - 4 C \tl^2\, .

1712: \ee

1713: This correction to the value of $\omg$ shows that \refb{e6.9} does not 

1714: represent an exactly marginal deformation.

1715: 

1716: If we had naively ignored the issue of non-commutativity of the two 

1717: operators in \refb{e6.16} and treated this as an exactly marginal 

1718: deformation, then we would end up with an expression for the energy 

1719: momentum tensor similar to that in \refb{e5.20}, with $f(x^0)$ given 

1720: by\cite{0203265}

1721: \be \label{e6.18}

1722: f(x^0) =  {1\over 1 + e^{\sqrt 2 x^0} \sin^2(\tl\pi/2)} + {1 \over

1723: 1 + e^{-\sqrt 2 x^0} \sin^2(\tl\pi/2)} - 1\, .

1724: \ee

1725: {}From this we see that $f((x^0\pm iy)/\sqrt{2})$ has singularities at 

1726: $x^0=\ln(1/\sin^2(\tl\pi/2))$, $y=\pm \pi$, and the energy density blows 

1727: up at these space-time points. Thus energy flows into these points from 

1728: the rest 

1729: of the system. These are precisely the points where we expect the 

1730: codimension 1 brane-antibrane pairs to form, and the nature of these 

1731: singularities has precisely the same delta function form as discussed at 

1732: the end of section \ref{s5}.\footnote{The analysis of RR charge density

1733: shows that RR charge also accumulates at these points.} Thus one might be

1734: tempted to interpret

1735: this as the creation of brane-antibrane pair.\footnote{Production of

1736: brane-antibrane pair from the decay of a higher dimensional brane has also

1737: been discussed in a different approximation in \cite{0204203}.} However,

1738: once we take into 

1739: account the non-marginality of the perturbation \refb{e6.16} this result 

1740: will be modified. In fact, if the perturbation \refb{e6.12} had been

1741: exactly marginal, and the energy momentum tensor had been given by

1742: \refb{e5.20}, \refb{e6.18}, we would have an inconsistency. To see this,

1743: note that following arguments similar to those in section \ref{s5} we

1744: would conclude in this case that the total initial energy of the system is

1745: deposited into a codimension 1 brane-antibrane pair, each carrying an

1746: energy density

1747: equal to:

1748: \be \label{e6.19}

1749: {1\over \sqrt 2} \, \TT_{p-1} \, (1 + \cos(\tl\pi))\, ,

1750: \ee

1751: where $\TT_{p-1}$ is the tension of a BPS D-$(p-1)$ brane. Thus for a

1752: generic $\tl$, \refb{e6.19} differs from the tension of a BPS

1753: D-$(p-1)$-brane. 

1754: Unlike in the case of bosonic D-branes, in this case we

1755: cannot attribute the excess (deficit) energy to the energy stored in the

1756: tachyon

1757: field, since there are no tachyons on a BPS D-brane. (The excess energy

1758: can still be attributed to other fields on the

1759: BPS D-brane, but there is no way a brane can have energy density lower

1760: than its tension.) Thus we would be led to an inconsistency, as there will

1761: be no interpretation for the final state for $(1+\cos(\pi\tl))< \sqrt{2}$.

1762: Thus

1763: it is just as well that \refb{e5.20}, \refb{e6.18} does not give the

1764: correct formula for the stress tensor.

1765: 

1766: \sectiono{Tachyon Dynamics in Closed String Theory} \label{s7}

1767: 

1768: In this section we briefly discuss possible extensions of the method 

1769: outlined in this paper to the study of tachyon dynamics in closed string 

1770: theory. 

1771: Given a zero 

1772: momentum

1773: tachyon field described by a dimension $(h,h)$ vertex operator $V_T$, we 

1774: try to construct a solution in the Wick rotated theory ($x^0\to i x$) by 

1775: perturbing the orginal CFT by 

1776: \be \label{e8.1}

1777: \tl \int d^2 z \, \cos(\omg X(z,\bar z)) V_T(z, \bar z)\, .

1778: \ee

1779: For $\omg^2 = 4(1-h)$ this would describe a marginal deformation to

1780: leading order in $\tl$. 

1781: However since in general higher order contribution to the $\beta$-function 

1782: does not vanish, we do not fix $\omg$ at the beginning. The 

1783: $\beta$-function associated with the perturbation \refb{e8.1} then takes 

1784: the form:

1785: \be \label{e8.2}

1786: \beta_{\tl} = 2\, \Big({1\over 4} \omg^2 + h - 1\Big)\, \tl + 

1787: g(\omg,\tl)\, .

1788: \ee

1789: We now adjust $\omg$ for a given $\tl$ such that \refb{e8.2} vanishes. 

1790: Thus this perturbation describes a conformal field theory.

1791: 

1792: The main obstruction to carrying out this program arises from the fact

1793: that

1794: in this case the operator product of \refb{e8.1} with itself does generate

1795: other dimension (1,1) operators, -- namely the vertex operators associated

1796: with the zero momentum graviton and dilaton fields.  (The source for the 

1797: dilaton field manifests itself in the form of a change in the central 

1798: charge induced by the perturbation.) Thus the rolling

1799: tachyon soution acts as a source for the zero momentum graviton and the

1800: dilaton fields, and {\it a priori} there is no guarantee that we shall be

1801: able to solve the equations of motion for these fields perturbatively in

1802: the parameter $\tl$. (See \cite{mukherji} for an analysis of this 

1803: problem in non-polynomial closed string field theory\cite{non-pol}.) This

1804: problem of course is not specific to string 

1805: theory, and even in the case of a normal scalar field theory coupled to 

1806: the 

1807: graviton and the dilaton, there is no guarantee that given a time 

1808: dependent solution involving the scalar field, we can solve for the 

1809: graviton and the dilaton field perturbatively in the parameter labelling 

1810: the solution of the scalar field equations. However, for tachyons 

1811: localized on a 

1812: subspace of the full space-time\cite{local}, it may be possible to solve 

1813: the 

1814: equations of motion of the graviton and the dilaton fields as a 

1815: perturbation expansion in the deformation parameter.

1816: 

1817: \sectiono{Conclusion} \label{s8}

1818: 

1819: In this paper we have discussed a general method for constructing time

1820: dependent solutions, describing the rolling of a tachyon on an unstable

1821: D-brane system, in cubic open string field theory. We have also provided a

1822: description of these solutions as Wick rotated version of euclidean

1823: boundary conformal field theories. The family of time dependent solutions

1824: associated with different initial conditions correspond to a family of

1825: boundary conformal field theories, each of which is related to the

1826: original D-brane system by a nearly marginal deformation. The construction

1827: can be easily generalized to the case of unstable D-brane systems in

1828: superstring theory.

1829: 

1830: In general these deformed boundary conformal field theories are not

1831: exactly solvable, and hence we cannot find explicit analytic expressions

1832: for the time dependence of the energy momentum tensor by Wick rotating the

1833: boundary state associated with the boundary conformal field theory. 

1834: However in some special cases the family of boundary CFT's are related by

1835: an exactly marginal deformation and are exactly solvable. This allows us

1836: to compute the time evolution of the energy momentum tensor explicitly for

1837: arbitrary initial condition on the tachyon. One such example discussed in

1838: this paper is the case of a bosonic D-$p$-brane wrapped on a circle of

1839: radius $\sqrt 2$, and we consider the first momentum mode of the tachyon

1840: along the circle. Pushing this particular tachyonic mode away from its

1841: extremum corresponds to an initial condition where the tachyon field has

1842: different signs in different parts of the circle, and hence we might

1843: expect that the time evolution of the configuration may produce lower

1844: dimensional D-branes. We explicitly construct the conserved energy

1845: momentum tensor associated with this solution and show that the time

1846: evolution of the system does indicate the creation of a codimension 1

1847: brane. 

1848: 

1849: One would like to generalize this to describe creation of a lower

1850: dimensional brane-antibrane pair due to the decay of an unstable D-brane

1851: in superstring theory.

1852: But unfortunately the situation here is more complicated as the associated

1853: family of conformal field theories are not related to each other by

1854: marginal deformation and are not exactly solvable. We hope to return to

1855: this problem in the near future.

1856: 

1857: We conclude with two remarks:

1858: 

1859: \begin{itemize}

1860: \item The solutions constructed here are obtained in {\it classical} open

1861: string field theory where we ignore coupling to closed string fields. This

1862: is expected to be a good approximation in the string coupling $g_s\to 0$

1863: limit. Once we switch on $g_s$, two effects will have to be taken into

1864: account in the study of the solution: classical backreaction of closed

1865: string fields on the solution, and the possibility of the decay into

1866: quantum closed string states. The time scale over which these effects will

1867: affect the solution is expected to go to infinity in the $g_s\to 0$ limit,

1868: but at present we do not know the precise dependence of this time scale on

1869: $g_s$.

1870: 

1871: Another parameter that is important in determining the effect of closed

1872: string coupling to the solution is $\tl$, or equivalently, the total

1873: energy

1874: density of the configuration. We would expect that for small energy

1875: density, the effect of coupling to the closed strings will be small, but

1876: again we do not know the precise dependence.

1877: 

1878: \item In the study of closed string tachyon condensation\cite{local},

1879: renormalization group (RG) flow has played a useful role. 

1880: The approach taken here is somewhat different: we interpret the complete

1881: time dependent solution for a specific initial condition as a single

1882: conformal field theory (after Wick rotation). Different inequivalent

1883: initial conditions will generate different boundary conformal field

1884: theories.

1885: 

1886: In fact it is possible to argue that at least for open string tachyon

1887: condensation, RG flow can never describe classical time

1888: evolution

1889: in a strict sense. RG flow takes us from a BCFT with higher boundary

1890: entropy to one of lower boundary entropy\cite{bent}. In the space-time

1891: language this corresponds to interpolating between a D-brane of higher

1892: energy to a D-brane of lower energy. It is clear that classical time

1893: evolution can never do that; due to conservation of energy it must take us

1894: from a D-brane of higher energy to one of lower energy plus other stuff.

1895: As a result, the RG analysis can at most give some qualitative information

1896: about time evolution in open string theory.

1897: 

1898: \end{itemize}

1899: 

1900: \medskip

1901: 

1902: {\bf Acknowledgement}:

1903: I would like to thank R.~Brower, R.~Gopakumar, J.~Minahan, S.~Minwalla, 

1904: N.~Moeller, L.~Rastelli and B.~Zwiebach for 

1905: useful discussions, and B.~Zwiebach for critical comments on the

1906: manuscript.

1907: This work was supported in part by a grant 

1908: from the Eberly College 

1909: of Science of the Penn State University. I would also like to acknowledge 

1910: the hospitality of the Center for Theoretical Physics at

1911: MIT, and a grant from the NM Rothschild and Sons Ltd

1912: at the Isaac Newton

1913: Institute where part of this work 

1914: was done.

1915: 

1916: 

1917: \begin{thebibliography}{99}

1918: 

1919: \bibitem{closed}

1920: G.~T.~Horowitz and A.~R.~Steif,

1921: Phys.\ Lett.\ B {\bf 258}, 91 (1991);

1922: %%CITATION = PHLTA,B258,91;%%

1923: G.~Veneziano,

1924: %``String cosmology: The pre-big bang scenario,''

1925: arXiv:hep-th/0002094;

1926: %%CITATION = HEP-TH 0002094;%%

1927: F.~Larsen and F.~Wilczek,

1928: Phys.\ Rev.\ D {\bf 55}, 4591 (1997)

1929: [arXiv:hep-th/9610252];

1930: %%CITATION = HEP-TH 9610252;%%

1931: G.~T.~Horowitz and D.~Marolf,

1932: JHEP {\bf 9807}, 014 (1998)

1933: [arXiv:hep-th/9805207];

1934: %%CITATION = HEP-TH 9805207;%%

1935: I.~I.~Kogan and N.~B.~Reis,

1936: Int.\ J.\ Mod.\ Phys.\ A {\bf 16}, 4567 (2001)

1937: [arXiv:hep-th/0107163];

1938: %%CITATION = HEP-TH 0107163;%%

1939: J.~Khoury, B.~A.~Ovrut, P.~J.~Steinhardt and N.~Turok,

1940: Phys.\ Rev.\ D {\bf 64}, 123522 (2001)

1941: [arXiv:hep-th/0103239];

1942: %%CITATION = HEP-TH 0103239;%%

1943: J.~Khoury, B.~A.~Ovrut, N.~Seiberg, P.~J.~Steinhardt and N.~Turok,

1944: Phys.\ Rev.\ D {\bf 65}, 086007 (2002)

1945: [arXiv:hep-th/0108187];

1946: %%CITATION = HEP-TH 0108187;%%

1947: P.~J.~Steinhardt and N.~Turok,

1948: arXiv:hep-th/0111030;

1949: %%CITATION = HEP-TH 0111030;%%

1950: N.~Seiberg,

1951: arXiv:hep-th/0201039;

1952: %%CITATION = HEP-TH 0201039;%%

1953: V.~Balasubramanian, S.~F.~Hassan, E.~Keski-Vakkuri and A.~Naqvi,

1954: %``A space-time orbifold: A toy model for a cosmological singularity,''

1955: arXiv:hep-th/0202187;

1956: %%CITATION = HEP-TH 0202187;%%

1957: L.~Cornalba and M.~S.~Costa,

1958: %``A New Cosmological Scenario in String Theory,''

1959: arXiv:hep-th/0203031;

1960: %%CITATION = HEP-TH 0203031;%%

1961: N.~A.~Nekrasov,

1962: %``Milne universe, tachyons, and quantum group,''

1963: arXiv:hep-th/0203112;

1964: %%CITATION = HEP-TH 0203112;%%

1965: J.~Simon,

1966: %``The geometry of null rotation identifications,''

1967: JHEP {\bf 0206}, 001 (2002)

1968: [arXiv:hep-th/0203201];

1969: %%CITATION = HEP-TH 0203201;%%

1970: E.~Kiritsis and B.~Pioline,

1971: arXiv:hep-th/0204004;

1972: %%CITATION = HEP-TH 0204004;%%

1973: H.~Liu, G.~Moore and N.~Seiberg,

1974: %``Strings in a time-dependent orbifold,''

1975: arXiv:hep-th/0204168;

1976: %%CITATION = HEP-TH 0204168;%%

1977: arXiv:hep-th/0206182;

1978: %%CITATION = HEP-TH 0206182;%%

1979: S.~Elitzur, A.~Giveon, D.~Kutasov and E.~Rabinovici,

1980: %``From big bang to big crunch and beyond,''

1981: arXiv:hep-th/0204189;

1982: %%CITATION = HEP-TH 0204189;%%

1983: L.~Cornalba, M.~S.~Costa and C.~Kounnas,

1984: arXiv:hep-th/0204261;

1985: %%CITATION = HEP-TH 0204261;%

1986: O.~Aharony, M.~Fabinger, G.~T.~Horowitz and E.~Silverstein, 

1987: %``Clean time-dependent string backgrounds from bubble baths,''

1988: arXiv:hep-th/0204158;

1989: %%CITATION = HEP-TH 0204158;%%

1990: B.~Craps, D.~Kutasov and G.~Rajesh,

1991: arXiv:hep-th/0205101;

1992: %%CITATION = HEP-TH 0205101;%%

1993: S.~Kachru, X.~Liu, M.~B.~Schulz and S.~P.~Trivedi,

1994: arXiv:hep-th/0205108;

1995: %%CITATION = HEP-TH 0205108;%%

1996: D.~Birmingham and M.~Rinaldi, 

1997: arXiv:hep-th/0205246;  

1998: %%CITATION = HEP-TH 0205246;%%

1999: A.~Lawrence,

2000: arXiv:hep-th/0205288;

2001: %%CITATION = HEP-TH 0205288;%% 

2002: V.~Balasubramanian and S.~F.~Ross, 

2003: %``The dual of nothing,''

2004: arXiv:hep-th/0205290;

2005: %%CITATION = HEP-TH 0205290;%%

2006: M.~Fabinger and J.~McGreevy,

2007: arXiv:hep-th/0206196;

2008: %%CITATION = HEP-TH 0206196;%%

2009: R.~Kallosh, L.~Kofman, A.~D.~Linde and A.~A.~Tseytlin,

2010: Phys.\ Rev.\ D {\bf 64}, 123524 (2001)

2011: [arXiv:hep-th/0106241];

2012: %%CITATION = HEP-TH 0106241;%%

2013: G.~T.~Horowitz and J.~Polchinski,

2014: arXiv:hep-th/0206228;

2015: %%CITATION = HEP-TH 0206228;%%

2016: C.~M.~Chen, D.~V.~Gal'tsov and M.~Gutperle,

2017: arXiv:hep-th/0204071;

2018: %%CITATION = HEP-TH 0204071;%%

2019: M.~Kruczenski, R.~C.~Myers and A.~W.~Peet,

2020: JHEP {\bf 0205}, 039 (2002)

2021: [arXiv:hep-th/0204144];

2022: %%CITATION = HEP-TH 0204144;%%

2023: S.~Roy,

2024: arXiv:hep-th/0205198;

2025: %%CITATION = HEP-TH 0205198;%%

2026: N.~S.~Deger and A.~Kaya,

2027: arXiv:hep-th/0206057;

2028: %%CITATION = HEP-TH 0206057;%%

2029: J.~E.~Wang,

2030: arXiv:hep-th/0207089.

2031: %%CITATION = HEP-TH 0207089;%%

2032: 

2033: \bibitem{0202210}

2034: M.~Gutperle and A.~Strominger,

2035: %``Spacelike branes,''

2036: arXiv:hep-th/0202210.

2037: %%CITATION = HEP-TH 0202210;%%

2038: 

2039: 

2040: 

2041: \bibitem{0203211} 

2042: A.~Sen, 

2043: %``Rolling Tachyon,'' 

2044: arXiv:hep-th/0203211. 

2045: %%CITATION = HEP-TH 0203211;%%

2046: 

2047: \bibitem{0203265}

2048: A.~Sen,

2049: %``Tachyon matter,''

2050: arXiv:hep-th/0203265.

2051: %%CITATION = HEP-TH 0203265;%%

2052: 

2053: \bibitem{0204203}

2054: K.~Hashimoto,

2055: arXiv:hep-th/0204203.

2056: %%CITATION = HEP-TH 0204203;%%

2057: 

2058: \bibitem{0205085}

2059: S. Sugimoto, S. Terashima

2060: arXiv:hep-th/0205085.

2061: %%CITATION = HEP-TH 0205085;%%

2062: 

2063: \bibitem{0205098}

2064: J.~A.~Minahan,

2065: %``Rolling the tachyon in super BSFT,''

2066: arXiv:hep-th/0205098.

2067: %%CITATION = HEP-TH 0205098;%%

2068: 

2069: \bibitem{0206102}

2070: A.~Ishida and S.~Uehara,

2071: %``Gauge fields on tachyon matter,''

2072: arXiv:hep-th/0206102.

2073: %%CITATION = HEP-TH 0206102;%%

2074: 

2075: \bibitem{0207004}

2076: K.~Ohta and T.~Yokono, 

2077: arXiv:hep-th/0207004

2078: %%CITATION = HEP-TH 0207004;%%

2079: 

2080: \bibitem{old}

2081: K.~Bardakci, 

2082: Nucl.\ Phys.\ B {\bf 68}, 331 (1974)

2083: %%CITATION = NUPHA,B68,331;%%

2084: Nucl.\ Phys.\ B {\bf 70}, 397 (1974);

2085: %%CITATION = NUPHA,B70,397;%%

2086: Nucl.\ Phys.\ B {\bf 133}, 297 (1978);

2087: %%CITATION = NUPHA,B133,297;%%

2088: K.~Bardakci and M.~B.~Halpern,

2089: Phys.\ Rev.\ D {\bf 10}, 4230 (1974);

2090: %%CITATION = PHRVA,D10,4230;%%

2091: Nucl.\ Phys.\ B {\bf 96}, 285 (1975).

2092: %%CITATION = NUPHA,B96,285;%%

2093: 

2094: \bibitem{9806155}

2095: O.~Bergman and M.~R.~Gaberdiel,

2096: %``Stable non-BPS D-particles,''

2097: Phys.\ Lett.\ B {\bf 441}, 133 (1998)

2098: [arXiv:hep-th/9806155];

2099: %%CITATION = HEP-TH 9806155;%%

2100: %``Non-BPS Dirichlet branes,''

2101: Class.\ Quant.\ Grav.\  {\bf 17}, 961 (2000)

2102: [arXiv:hep-th/9908126].

2103: %%CITATION = HEP-TH 9908126;%%

2104: 

2105: 

2106: 

2107: \bibitem{9809111}

2108: A.~Sen,

2109: %``Type I D-particle and its interactions,''

2110: JHEP {\bf 9810}, 021 (1998)

2111: [arXiv:hep-th/9809111].

2112: %%CITATION = HEP-TH 9809111;%%

2113: 

2114: 

2115: \bibitem{cosmo}

2116: G.~W.~Gibbons,

2117: %``Cosmological evolution of the rolling tachyon,''

2118: arXiv:hep-th/0204008;

2119: %%CITATION = HEP-TH 0204008;%%

2120: M.~Fairbairn and M.~H.~Tytgat,

2121: %``Inflation from a tachyon fluid?,''

2122: arXiv:hep-th/0204070;

2123: %%CITATION = HEP-TH 0204070;%%

2124: S.~Mukohyama,

2125: %``Brane cosmology driven by the rolling tachyon,''

2126: arXiv:hep-th/0204084;

2127: %%CITATION = HEP-TH 0204084;%%

2128: A.~Feinstein,

2129: %``Power-law inflation from the rolling tachyon,''

2130: arXiv:hep-th/0204140;

2131: %%CITATION = HEP-TH 0204140;%%

2132: T.~Padmanabhan, arXiv:hep-th/0204150;

2133: %%CITATION = HEP-TH 0204150;%%

2134: A.~Frolov, L.~Kofman and A.~A.~Starobinsky,

2135: arXiv:hep-th/0204187;

2136: %%CITATION = HEP-TH 0204187;%%

2137: D.~Choudhury, D.~Ghoshal, D.~P.~Jatkar and S.~Panda,

2138: arXiv:hep-th/0204204;

2139: %%CITATION = HEP-TH 0204204;%%

2140: X.~Li, J.~Hao and D.~Liu,

2141: arXiv:hep-th/0204252;

2142: %%CITATION = HEP-TH 0204252;%%

2143: G.~Shiu and I.~Wasserman,

2144: arXiv:hep-th/0205003;

2145: %%CITATION = HEP-TH 0205003;%%

2146: T.~Padmanabhan and T.~R.~Choudhury,

2147: %``Can the clustered dark matter and the smooth dark energy arise from the  

2148: %same scalar field?,''

2149: arXiv:hep-th/0205055;

2150: %%CITATION = HEP-TH 0205055;%%

2151: L.~Kofman and A.~Linde,

2152: %``Problems with Tachyon Inflation,''

2153: arXiv:hep-th/0205121;

2154: %%CITATION = HEP-TH 0205121;%%

2155: M.~Sami,

2156: %``Implementing Power Law Inflation with Rolling Tachyon on the Brane,''

2157: arXiv:hep-th/0205146;

2158: %%CITATION = HEP-TH 0205146;%%

2159: M.~Sami, P.~Chingangbam and T.~Qureshi,

2160: %``Aspects of tachyonic inflation with exponential potential,''

2161: arXiv:hep-th/0205179;

2162: %%CITATION = HEP-TH 0205179;%%

2163: T.~Mehen and B.~Wecht,

2164: arXiv:hep-th/0206212.

2165: %%CITATION = HEP-TH 0206212;%%

2166: 

2167: \bibitem{earl}

2168: G.~R.~Dvali and S.~H.~Tye, Phys. \ Lett. \ B {\bf 450}, 72 (1999)

2169: [arXiv:hep-ph/9812483];

2170: %%CITATION = HEP-TH 9812483;%%

2171: C.~Acatrinei and C.~Sochichiu,

2172: %``A note on the decay of noncommutative solitons,''

2173: arXiv:hep-th/0104263;

2174: %%CITATION = HEP-TH 0104263;%%

2175: S.~H.~Alexander, Phys. \ Rev. \ D {\bf 65}, 023507 (2002)

2176: [arXiv:hep-th/0105032];

2177: %%CITATION = HEP-TH 0105032;%%

2178: G.~R.~Dvali, Q.~Shafi and S.~Solganik,

2179: [arXiv:hep-th/0105203];

2180: %%CITATION = HEP-TH 0105203;%%

2181: C.~P.~Burgess, M.~Majumdar, D.~Nolte, F.~Quevedo, G.~Rajesh and 

2182: R.~Z.~Zhang, JHEP {\bf 0107}, 047 (2001)

2183: [arXiv:hep-th/0105204];

2184: %%CITATION = HEP-TH 0105204;%%

2185: A.~Mazumdar, S.~Panda and A.~Perez-Lorenzana,

2186: %``Assisted inflation via tachyon condensation,''

2187: Nucl.\ Phys.\ B {\bf 614}, 101 (2001)

2188: [arXiv:hep-ph/0107058];

2189: %%CITATION = HEP-PH 0107058;%%

2190: C.~P.~Burgess, P.~Martineau, F.~Quevedo, G.~Rajesh, and R.~J.~Zhang,

2191: JHEP {\bf 0203}, 052 (2002)

2192: [arXiv:hep-th/0111025];

2193: %%CITATION = HEP-TH 0111025;%%

2194: S.~Sarangi and S.~H.~Tye, 

2195: %``Cosmic string production towards the end of brane inflation,''

2196: arXiv:hep-th/0204074;

2197: %%CITATION = HEP-TH 0204074;%%

2198: S.~Corley and D.~A.~Lowe,

2199: Phys.\ Lett.\ B {\bf 528}, 139 (2002)

2200: [arXiv:hep-th/0108178].

2201: %%CITATION = HEP-TH 0108178;%%

2202: 

2203: \bibitem{9704006}

2204: E.~Gava, K.~S.~Narain and M.~H.~Sarmadi,

2205: %``On the bound states of p- and (p+2)-branes,''

2206: Nucl.\ Phys.\ B {\bf 504}, 214 (1997)

2207: [arXiv:hep-th/9704006].

2208: %%CITATION = HEP-TH 9704006;%%

2209: 

2210: \bibitem{9703217}

2211: A.~Hashimoto and W.~I.~Taylor,

2212: %``Fluctuation spectra of tilted and intersecting D-branes from the  

2213: %Born-Infeld action,''

2214: Nucl.\ Phys.\ B {\bf 503}, 193 (1997)

2215: [arXiv:hep-th/9703217].

2216: %%CITATION = HEP-TH 9703217;%%

2217: 

2218: \bibitem{inter}

2219: J.~Garcia-Bellido, R.~Rabadan and F.~Zamora,

2220: %``Inflationary scenarios from branes at angles,''

2221: JHEP {\bf 0201}, 036 (2002)

2222: [arXiv:hep-th/0112147];

2223: %%CITATION = HEP-TH 0112147;%%

2224: R.~Blumenhagen, B.~Kors, D.~Lust and T.~Ott,

2225: %``Hybrid inflation in intersecting brane worlds,''

2226: arXiv:hep-th/0202124;

2227: %%CITATION = HEP-TH 0202124;%%

2228: K.~Dasgupta, C.~Herdeiro, S.~Hirano and R.~Kallosh,

2229: %``D3/D7 inflationary model and M-theory,''

2230: Phys.\ Rev.\ D {\bf 65}, 126002 (2002)

2231: [arXiv:hep-th/0203019];

2232: %%CITATION = HEP-TH 0203019;%%

2233: N.~Jones, H.~Stoica and S.~H.~Tye,

2234: %``Brane interaction as the origin of inflation,''

2235: arXiv:hep-th/0203163.

2236: %%CITATION = HEP-TH 0203163;%%

2237: 

2238: \bibitem{mo-zw}

2239: N.~Moeller and B.~Zwiebach, 

2240: arXiv:hep-th/0207107.

2241: %%CITATION = HEP-TH 0207107;%%

2242: 

2243: 

2244: \bibitem{golds}

2245: H.~Goldstein, Classical Mechanics, Addison Wesley.

2246: 

2247: \bibitem{zam}

2248: A.~B.~Zamolodchikov,

2249: %``Renormalization Group And Perturbation Theory Near Fixed Points In 

2250: %Two-Dimensional Field Theory,''

2251: Sov.\ J.\ Nucl.\ Phys.\  {\bf 46}, 1090 (1987)

2252: [Yad.\ Fiz.\  {\bf 46}, 1819 (1987)].

2253: %%CITATION = SJNCA,46,1090;%%

2254: 

2255: \bibitem{bent} 

2256: I.~Affleck and A.~W.~Ludwig, 

2257: Phys.\ Rev.\ Lett.\ {\bf 67}, 161 (1991);  

2258: %%CITATION = PRLTA,67,161;%%,

2259: P.~Fendley, H.~Saleur and N.~P.~Warner,

2260: Nucl.\ Phys.\ B {\bf 430}, 577 (1994)

2261: [arXiv:hep-th/9406125];

2262: %%CITATION = HEP-TH 9406125;%%

2263: J.~A.~Harvey, D.~Kutasov and E.~J.~Martinec, 

2264: arXiv:hep-th/0003101.  

2265: %%CITATION = HEP-TH 0003101;%%

2266: 

2267: 

2268: \bibitem{bsft}

2269: A.~A.~Gerasimov and S.~L.~Shatashvili,

2270: %``On exact tachyon potential in open string field theory,''

2271: JHEP {\bf 0010}, 034 (2000)

2272: [arXiv:hep-th/0009103];

2273: %%CITATION = HEP-TH 0009103;%%

2274: D.~Kutasov, M.~Marino and G.~W.~Moore,

2275: %``Some exact results on tachyon condensation in string field theory,''

2276: JHEP {\bf 0010}, 045 (2000)

2277: [arXiv:hep-th/0009148];

2278: %%CITATION = HEP-TH 0009148;%%

2279: %``Remarks on tachyon condensation in superstring field theory,''

2280: arXiv:hep-th/0010108;

2281: %%CITATION = HEP-TH 0010108;%%

2282: D.~Ghoshal and A.~Sen,

2283: JHEP {\bf 0011}, 021 (2000)

2284: [arXiv:hep-th/0009191].

2285: %%CITATION = HEP-TH 0009191;%%

2286: 

2287: \bibitem{0010247}

2288: S.~Dasgupta and T.~Dasgupta,

2289: JHEP {\bf 0106}, 007 (2001)

2290: [arXiv:hep-th/0010247].

2291: %%CITATION = HEP-TH 0010247;%%

2292: 

2293: 

2294: \bibitem{boundary} 

2295: J.~Polchinski and Y.~Cai, 

2296: %``Consistency Of Open Superstring Theories,'' 

2297: Nucl.\ Phys.\ B {\bf 296}, 91 (1988); 

2298: %%CITATION = NUPHA,B296,91;%%

2299: C.~G.~Callan, C.~Lovelace, C.~R.~Nappi and S.~A.~Yost,

2300: Nucl.\ Phys.\ B {\bf 460}, 351 (1996)[arXiv:hep-th/9510161];

2301: %%CITATION = HEP-TH 9510161;%%

2302: M.~Li,

2303: Nucl.\ Phys.\ B {\bf 460}, 351 (1996)

2304: [arXiv:hep-th/9510161];

2305: %%CITATION = HEP-TH 9510161;%%

2306: O.~Bergman and M.~R.~Gaberdiel,

2307: Nucl.\ Phys.\ B {\bf 499}, 183 (1997)

2308: [arXiv:hep-th/9701137].

2309: %%CITATION = HEP-TH 9701137;%%

2310: 

2311: \bibitem{em-tensor}

2312: M.~B.~Green and M.~Gutperle,

2313: Nucl.\ Phys.\ B {\bf 476}, 484 (1996)

2314: [arXiv:hep-th/9604091];

2315: %%CITATION = HEP-TH 9604091;%%

2316: P.~Di Vecchia, M.~Frau, I.~Pesando, S.~Sciuto, A.~Lerda and R.~Russo,

2317: Nucl.\ Phys.\ B {\bf 507}, 259 (1997)

2318: [arXiv:hep-th/9707068];

2319: %%CITATION = HEP-TH 9707068;%%

2320: P.~Di Vecchia and A.~Liccardo,

2321: arXiv:hep-th/9912275.

2322: %%CITATION = HEP-TH 9912275;%%

2323: 

2324: 

2325: 

2326: \bibitem{marginal}

2327: C.~G.~Callan and I.~R.~Klebanov,

2328: Phys.\ Rev.\ Lett.\  {\bf 72}, 1968 (1994)

2329: [arXiv:hep-th/9311092];

2330: %%CITATION = HEP-TH 9311092;%%

2331: C.G.~Callan, I.R.~Klebanov, A.W.~Ludwig and J.M.~Maldacena,

2332: Nucl.\ Phys.\ {\bf B422}, 417 (1994)

2333: hep-th/9402113;

2334: J.~Polchinski and L.~Thorlacius,

2335: Phys.\ Rev.\ {\bf D50}, 622 (1994)

2336: hep-th/9404008.

2337: %%CITATION = PHRVA,D50,622;%%

2338: %%CITATION = NUPHA,B422,417;%%

2339: 

2340: \bibitem{9811237}

2341: A.~Recknagel and V.~Schomerus, 

2342: %``Boundary deformation theory and moduli

2343: %spaces of D-branes,'',

2344: Nucl.\ Phys.\ {\bf B545}, 233 (1999)

2345: hep-th/9811237.  

2346: %%CITATION = NUPHA,B545,233;%%

2347: 

2348: \bibitem{0108238}

2349: M.~R.~Gaberdiel and A.~Recknagel,

2350: %``Conformal boundary states for free bosons and fermions,''

2351: JHEP {\bf 0111}, 016 (2001)

2352: [arXiv:hep-th/0108238].

2353: %%CITATION = HEP-TH 0108238;%%

2354: 

2355: \bibitem{FMS}

2356: D.~Friedan, E.~J.~Martinec and S.~H.~Shenker,

2357: Nucl.\ Phys.\ B {\bf 271}, 93 (1986).

2358: %%CITATION = NUPHA,B271,93;%%

2359: 

2360: \bibitem{9808141}

2361: A.~Sen,

2362: JHEP {\bf 9809}, 023 (1998)

2363: [hep-th/9808141]; JHEP {\bf 9812}, 021 (1998)

2364: [arXiv:hep-th/9812031].

2365: %%CITATION = HEP-TH 9812031;%%

2366: %%CITATION = HEP-TH 9808141;%%

2367: 

2368: 

2369: \bibitem{0003124}

2370: J.~Majumder and A.~Sen,  

2371: JHEP {\bf 0006}, 010 (2000) 

2372: [arXiv:hep-th/0003124].  

2373: %%CITATION = HEP-TH 0003124;%%

2374: 

2375: 

2376: \bibitem{mukherji}

2377: S.~Mukherji and A.~Sen,

2378: %``Some all order classical solutions in nonpolynomial closed string field 

2379: %theory,''

2380: Nucl.\ Phys.\ B {\bf 363}, 639 (1991).

2381: %%CITATION = NUPHA,B363,639;%%

2382: 

2383: \bibitem{non-pol}

2384: B.~Zwiebach,

2385: Nucl.\ Phys.\ B {\bf 390}, 33 (1993)

2386: [arXiv:hep-th/9206084];

2387: %%CITATION = HEP-TH 9206084;%%

2388: arXiv:hep-th/9305026;  

2389: %%CITATION = HEP-TH 9305026;%% 

2390: Annals Phys.\ {\bf 267}, 193 (1998)  

2391: [arXiv:hep-th/9705241].  

2392: %%CITATION = HEP-TH 9705241;%%

2393: 

2394: \bibitem{local}

2395: A.~Adams, J.~Polchinski and E.~Silverstein,

2396: %``Don't panic! Closed string tachyons in ALE space-times,''

2397: JHEP {\bf 0110}, 029 (2001)

2398: [arXiv:hep-th/0108075];

2399: %%CITATION = HEP-TH 0108075;%%

2400: A.~Dabholkar,

2401: %``On condensation of closed-string tachyons,''

2402: arXiv:hep-th/0109019;

2403: %%CITATION = HEP-TH 0109019;%%

2404: J.~G.~Russo and A.~A.~Tseytlin,

2405: %``Supersymmetric fluxbrane intersections and closed string tachyons,''

2406: JHEP {\bf 0111}, 065 (2001)

2407: [arXiv:hep-th/0110107];

2408: %%CITATION = HEP-TH 0110107;%%

2409: A.~Dabholkar,

2410: %``Tachyon condensation and black hole entropy,''

2411: Phys.\ Rev.\ Lett.\  {\bf 88}, 091301 (2002)

2412: [arXiv:hep-th/0111004];

2413: %%CITATION = HEP-TH 0111004;%%

2414: C.~Vafa,

2415: %``Mirror symmetry and closed string tachyon condensation,''

2416: arXiv:hep-th/0111051;

2417: %%CITATION = HEP-TH 0111051;%%

2418: J.~A.~Harvey, D.~Kutasov, E.~J.~Martinec and G.~Moore,

2419: %``Localized tachyons and RG flows,''

2420: arXiv:hep-th/0111154;

2421: %%CITATION = HEP-TH 0111154;%%

2422: A.~Dabholkar and C.~Vafa,

2423: %``tt* geometry and closed string tachyon potential,''

2424: JHEP {\bf 0202}, 008 (2002)

2425: [arXiv:hep-th/0111155];

2426: %%CITATION = HEP-TH 0111155;%%

2427: Y.~Michishita and P.~Yi,

2428: %``D-brane probe and closed string tachyons,''

2429: Phys.\ Rev.\ D {\bf 65}, 086006 (2002)

2430: [arXiv:hep-th/0111199];

2431: %%CITATION = HEP-TH 0111199;%%

2432: J.~R.~David, M.~Gutperle, M.~Headrick and S.~Minwalla,

2433: %``Closed string tachyon condensation on twisted circles,''

2434: JHEP {\bf 0202}, 041 (2002)

2435: [arXiv:hep-th/0111212];

2436: %%CITATION = HEP-TH 0111212;%%

2437: S.~k.~Nam and S.~J.~Sin,

2438: %``Condensation of localized tachyons and spacetime supersymmetry,''

2439: arXiv:hep-th/0201132;

2440: %%CITATION = HEP-TH 0201132;%%

2441: S.~P.~De Alwis and A.~T.~Flournoy,

2442: %``Closed string tachyons and semi-classical instabilities,''

2443: arXiv:hep-th/0201185;

2444: %%CITATION = HEP-TH 0201185;%%

2445: S.~J.~Sin,

2446: %``Tachyon mass, c-function and counting localized degrees of freedom,''

2447: arXiv:hep-th/0202097;

2448: %%CITATION = HEP-TH 0202097;%%

2449: R.~Rabadan and J.~Simon,

2450: %``M-theory lift of brane-antibrane systems and localised closed string  

2451: tachyons,''

2452: JHEP {\bf 0205}, 045 (2002)

2453: [arXiv:hep-th/0203243];

2454: %%CITATION = HEP-TH 0203243;%%

2455: A.~Basu,

2456: %``Localized tachyons and the g(cl) conjecture,''

2457: arXiv:hep-th/0204247.

2458: %%CITATION = HEP-TH 0204247;%%

2459: 

2460: \end{thebibliography}

2461: 

2462: \end{document}

2463: 

2464: \bye

2465: 

2466: