hep-th0207228/text
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: % On the construction of nonsingular Ekpyrotic models and
3: % cosmological density perturbations
4: % 
5: % RB, FF, ST
6: %
7: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
8: 
9: %\documentstyle[prd,twocolumn,eqsecnum,aps]{revtex}
10: \documentstyle[prd,eqsecnum,aps,epsf]{revtex}
11: %\documentstyle[prd,eqsecnum,aps]{revtex}
12: %\documentstyle[preprint,eqsecnum,aps]{revtex}
13: 
14: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
15: \renewcommand{\thefootnote}{\fnsymbol{footnote}}
16: \def\be{\begin{equation}}
17: \def\ee{\end{equation}}
18: \def\ba{\begin{eqnarray}}
19: \def\ea{\end{eqnarray}}
20: \newcommand{\beqa}{\begin{eqnarray}}
21: \newcommand{\eeqa}{\end{eqnarray}}
22: \newcommand{\bea}{\begin{eqnarray}}
23: \newcommand{\eea}{\end{eqnarray}}
24: \newcommand{\mpl}{M_{Pl}}
25: \newcommand{\mg}{M_G}
26: \newcommand{\k}{{\kappa}}
27: \newcommand{\lmk}{\left(}
28: \newcommand{\rmk}{\right)}
29: \newcommand{\lkk}{\left[}
30: \newcommand{\rkk}{\right]}
31: \newcommand{\lnk}{\left\{}
32: \newcommand{\rnk}{\right\}}
33: \newcommand{\zk}{z_k}
34: \newcommand{\calr}{{\cal R}}
35: \newcommand{\ch}{{\cal H}}
36: \newcommand{\bx}{{\bf x}}
37: \newcommand{\bfk}{{\bf k}}
38: \newcommand{\dphi}{\delta\varphi}
39: \newcommand{\ds}{\delta\sigma}
40: \newcommand{\ddphi}{\Delta\varphi}
41: \newcommand{\dds}{\Delta\sigma}
42: 
43: %%%%% singlefig %%%%%
44: \newcommand{\singlefig}[2]{
45: \begin{center}
46: \begin{minipage}{#1}
47: \epsfxsize=#1
48: \epsffile{#2}
49: \end{minipage}
50: \end{center}}
51: %
52: %%%%% figcaption %%%%%
53: \newenvironment{figcaption}[2]{
54:  \vspace{0.3cm}
55:  \refstepcounter{figure}
56:  \label{#1}
57:  \begin{center}
58:  \begin{minipage}{#2}
59:  \begingroup \small FIG. \thefigure: }{
60:  \endgroup
61:  \end{minipage}
62:  \end{center}}
63: %
64: 
65: %------------------------------
66: \def\beq{\begin{equation}}
67: \def\eeq{\end{equation}}
68: \newcommand{\gsim}{\mbox{\raisebox{-1.ex}{$\stackrel
69:      {\textstyle>}{\textstyle\sim}$}}}
70: \newcommand{\lsim}{\mbox{\raisebox{-1.ex}{$\stackrel
71:      {\textstyle<}{\textstyle \sim}$}}}
72: \newcommand{\square}{\kern1pt\vbox{\hrule height
73: 1.2pt\hbox{\vrule width 1.2pt\hskip 3pt
74:    \vbox{\vskip 6pt}\hskip 3pt\vrule width 0.6pt}\hrule
75: height 0.6pt}\kern1pt}
76: %------------------------------
77: 
78: 
79: \begin{document}
80: %\draft \twocolumn[\hsize\textwidth\columnwidth\hsize\csname
81: %@twocolumnfalse\endcsname
82: 
83:  
84: \title{On the Construction of Nonsingular Pre-Big-Bang and Ekpyrotic \\
85: Cosmologies and the Resulting Density Perturbations} 
86: 
87: \author{Shinji Tsujikawa$^{1)}$, Robert Brandenberger$^{2)}$ 
88: and Fabio Finelli$^{3)}$ } 
89: 
90: \address{1) Research Center for the Early 
91: Universe, University of Tokyo, Hongo, Bunkyo-ku, Tokyo 169-8555, 
92: Japan.\\
93: email: shinji@resceu.s.u-tokyo.ac.jp\\
94: 2) Institut d'Astrophysique de Paris, 98bis Blvd. Arago, F-75014 Paris,
95: France,
96: \\and\\
97: Physics Department, Brown University, Providence, RI 02912, USA.\\
98: email: rhb@het.brown.edu\\
99: 3) I.A.S.F. - Sezione di Bologna, C.N.R., Via Gobetti 101, 40129 Bologna,
100: Italy.\\
101: email: finelli@tesre.bo.cnr.it\\
102: [.3em]} 
103: \date{\today} 
104: \maketitle
105: \begin{abstract}
106: We consider the construction of nonsingular Pre-Big-Bang and Ekpyrotic
107: type cosmological
108: models realized by the addition to the action of specific higher-order terms
109: stemming from quantum corrections. We study models involving
110: general relativity coupled to a single scalar field with a
111: potential motivated by the Ekpyrotic scenario. We find that the inclusion of 
112: the string loop and quantum correction terms in the string frame makes it 
113: possible to obtain solutions of the variational equations which are
114: nonsingular and bouncing in the Einstein frame, 
115: even when a negative exponential 
116: potential is present, as is the case in the Ekpyrotic scenario. This
117: allows us to discuss the evolution of cosmological perturbations
118: without the need to invoke matching conditions between
119: two Einstein Universes, one representing the contracting branch, the second
120: the expanding branch. We analyze the spectra of perturbations 
121: produced during the bouncing phase and find that the spectrum of 
122: curvature fluctuations
123: in the model proposed originally to implement the Ekpyrotic scenario
124: has a large blue tilt ($n_{\cal R}= 3$).  Except for instabilities introduced 
125: on small scales, the result agrees with what is obtained by imposing 
126: continuity of the induced metric and of the extrinsic curvature across a 
127: constant scalar field (up to $k^2$ corrections equal to the constant energy 
128: density) matching surface between the contracting and the expanding 
129: Einstein Universes.  We also discuss nonsingular cosmological solutions 
130: obtained when a Gauss-Bonnet term with a coefficient suitably dependent on 
131: the scalar matter field is added to the action in the Einstein frame with a 
132: potential for the scalar field present.  In this scenario, nonsingular 
133: solutions are found which start in an asymptotically flat state, 
134: undergo a period of super-exponential inflation and end with a graceful exit.  
135: The spectrum of fluctuations is also calculated in this case.
136: \end{abstract}
137: 
138: \vskip 1pc \pacs{pacs: 98.80.Cq}
139: \vskip 2pc
140: %]
141: 
142: \baselineskip = 14pt
143: 
144: %%%%%%%%%%%%%%%%%%%%%%%%%%
145: \section{Introduction}                            
146: %%%%%%%%%%%%%%%%%%%%%%%%%%
147: 
148: There has recently been a lot of interest in cosmological scenarios in
149: which it is assumed that instead of emerging from an initial Big Bang
150: singularity, our Universe has resulted from an Einstein frame 
151: bounce which connects a
152: previous contracting phase with the present phase of cosmological
153: expansion. A lot of this interest has been fueled by string cosmology,
154: the attempt to merge string theory and cosmology. Pre-Big-Bang (PBB)
155: cosmology \cite{Veneziano:1991ek,Gasperini:1992em} 
156: (see \cite{Lidsey:1999mc,Gasperini:2002} for a comprehensive review) and the 
157: Ekpyrotic scenario \cite{Khoury:2001wf} are two well-known
158: models in which our present phase of cosmological expansion is
159: postulated to have emerged from a previous phase of cosmological
160: contraction \footnote{In PBB cosmology this statement is true from the point 
161: of view of the Einstein frame metric, in the Ekpyrotic scenario it is true 
162: from the point of view of the four space-time dimensional effective action 
163: which is used to describe the cosmology.}.  In both examples, however, the 
164: cosmological description in terms of an effective action breaks down at the 
165: bounce.  In the case of PBB cosmology this bounce corresponds to a region of 
166: high curvature where higher derivative and string corrections to the 
167: effective action will be important, in the case of the Ekpyrotic scenario 
168: the bounce occurs when two four space-time dimensional branes collide in a 
169: five dimensional bulk.
170: 
171: Models with a cosmological bounce potentially provide an alternative
172: to cosmological inflation in addressing the homogeneity problem
173: of standard cosmology and in yielding a causal mechanism of structure
174: formation, the latter since at times long before the bounce fixed
175: comoving scales of cosmological interest today will have been inside
176: the Hubble radius.\footnote{See 
177: refs.~\cite{Kallosh:2001ai,Kallosh:2001du,Enqvist:2001zk,Rasanen:2001hf,Felder:2002jk,Linde:2002ws}
178:  for critical arguments on the Ekpyrotic scenario.}
179: However, since in both PBB and
180: Ekpyrotic scenarios the Hubble parameter increases during the
181: collapsing phase, symmetry arguments such as those used originally
182: \cite{Press} to predict the scale-invariance of cosmological fluctuations
183: in inflationary cosmology would lead one to expect a blue
184: spectrum of curvature perturbations in these models, at least in
185: effective field theory models in 
186: which there is only one ``matter'' field. As outlined in Appendix A, in
187: PBB cosmology one expects a spectrum with spectral index $n=4$,
188: whereas in the Ekpyrotic scenario one expects $n=3$. In the case of
189: PBB cosmology, this heuristic prediction was confirmed 
190: \cite{BGGMV} by a general relativistic analysis (which is, however, 
191: subject to the caveats indicated below). In the case of the Ekpyrotic
192: scenario, there is a large disagreement in the results. Whereas
193: the works of 
194: \cite{Lyth:2001pf,Brandenberger:2001bs,Hwang:2001ga,Tsujikawa:2001ad} 
195: yield results in
196: agreement with the heuristic prediction (namely $n=3$), others
197: \cite{Khoury:2001zk,Durrer:2002jn} obtain a scale-invariant 
198: spectrum of adiabatic fluctuations (thus also casting
199: doubt on past results in the literature on the spectrum of
200: fluctuations in the PBB scenario).
201: 
202: The singularity of the effective action at the time of the bounce
203: makes it impossible to follow the evolution of the background
204: cosmology and of the resulting cosmological perturbations
205: rigorously \cite{Lyth:2001nv}. 
206: In much of the previous work, both in the context of PBB 
207: cosmology \cite{Deruelle:1995kd} and of the Ekpyrotic scenario 
208: \cite{Brandenberger:2001bs,Hwang:2001ga,Tsujikawa:2001ad,Durrer:2002jn}, 
209: the fluctuations were computed by
210: matching two Einstein Universes (the first representing the contracting
211: phase, the second the expanding phase) along a space-like surface
212: (representing the bounce region) and applying continuity of the
213: induced metric and of the extrinsic curvature across the surface
214: \cite{Hwang:1991an,Deruelle:1995kd}. As emphasized in \cite{Durrer:2002jn}, 
215: the result will depend on
216: how the matching surface is chosen \footnote{As emphasized already
217: in \cite{Brandenberger:2001bs} and \cite{Finelli:2001sr}, there is a 
218: consistency
219: check for proposed matching surfaces: when applied to the reheating
220: surface in inflationary cosmology, the correct result should emerge.
221: This does not happen with the prescription advocated in 
222: \cite{Durrer:2002jn}, nor
223: does it with the matching prescription of \cite{Khoury:2001zk} which is
224: not based on a geometric analysis (see also 
225: \cite{Martin:2001ue,Martin:2002ar} for
226: a criticism of the latter matching prescription).}. 
227: 
228: In the context of PBB cosmology, it was realized 
229: \cite{Gasperini:1996fu,Brustein:1997cv,Brustein:1997xn} (see
230: also \cite{Rey:1996ad,Foffa:1999dv,Cartier:1999vk}) that
231: higher derivative corrections (defined in the string frame)
232: to the action induced by inverse
233: string tension and coupling constant corrections can yield a
234: nonsingular background cosmology
235: \footnote{Construction of nonsingular cosmologies in pure Einstein
236: gravity by means of specific higher derivative terms is also
237: possible (see e.g. \cite{MB,BMS} and its application to
238: PBB cosmology in \cite{Brandenberger:1998zs,Easson:1999xw}).}. 
239: This then allows the
240: study of the evolution of cosmological perturbations without having
241: to use ad hoc matching prescriptions. The effects of the higher
242: derivative terms in the action on the evolution of fluctuations in
243: the PBB cosmology was investigated in \cite{Cartier:2001is}. It was found
244: that for low frequency modes, the spectrum of fluctuations is unaffected
245: by the higher derivative terms, and the result obtained is the same as
246: what follows from the
247: analysis using matching conditions between two Einstein Universes 
248: \cite{BGGMV,Deruelle:1995kd} joined along a constant scalar field
249: hypersurface. 
250: 
251: Since the Ekpyrotic scenario makes use of a negative exponential
252: potential for the scalar matter field, which leads to an extra
253: instability of the system, it is not clear that the higher derivative 
254: terms used in \cite{Gasperini:1996fu,Brustein:1997cv,Brustein:1997xn}
255: can in this case achieve a nonsingular cosmology. The first main
256: result of this paper is that, with suitably chosen coefficients,
257: the above mentioned terms are indeed sufficient to produce a
258: nonsingular cosmology.
259: 
260: In this paper, we add the same higher derivative terms used in 
261: \cite{Brustein:1997cv} to the action (which includes
262: a positive or negative potential for the scalar matter field)
263: and construct nonsingular bouncing cosmologies. At the
264: level of the effective action, our Lagrangian can be viewed as giving
265: both nonsingular solutions of modified PBB type (the modification
266: consisting of the addition of an exponential potential for the
267: dilaton), and also nonsingular Ekpyrotic solutions. 
268: The justification for adding these higher derivative terms is different
269: in the cases of modified PBB cosmology and in the Ekpyrotic
270: scenario. In the case of PBB cosmology, both the string
271: coupling constant and the curvature become large as the
272: dilaton increases, thus justifying the
273: inclusion of both higher derivative terms of the gravitational action
274: and of quantum corrections. In the case of Ekpyrotic cosmology (we have
275: the initial scenario of \cite{Khoury:2001wf} in 
276: mind in which a bulk brane impacts
277: our physical space-time orbifold fixed plane at the time of the bounce
278: and in which the dilaton and hence the string coupling constant are fixed),
279: the density and hence curvature at the bounce are large, thus justifying
280: including higher derivative terms. In addition, the brane collision is
281: a quantum mechanical process, thus justifying including loop corrections
282: in the action. Note that our method yields a way of constructing
283: a nonsingular bouncing Universe which works even in a spatially flat
284: Universe and is thus different from the constructions of 
285: \cite{Gordon:2002jw} in which a positive spatial curvature is 
286: used to generate a bouncing cosmology. 
287: Since tracing back the
288: spatial curvature into the very early Universe given the present date
289: leads - under the assumption that there was no period of inflation
290: after the bounce - to a highly suppressed curvature at early times,
291: our approach in obtaining a bouncing cosmology appears more realistic.
292: 
293: We follow the fluctuations through the bounce,
294: and study the spectrum of the resulting cosmological perturbations at
295: late times. In this analysis, no matching conditions at the
296: bounce are necessary. Note, however, that in principle the final
297: spectrum could depend on the frame in which the higher order
298: correction terms are introduced, and on the specific form of the
299: correction terms. In our nonsingular scenario discussed in Sec.IV,  
300: the correction terms are
301: defined in the string frame and we find
302: that the final spectrum of cosmological fluctuations on long wavelength
303: scales has a shape which agrees
304: with what is obtained when applying the matching conditions of 
305: \cite{Hwang:1991an,Deruelle:1995kd}
306: on a constant scalar field surface \footnote{Note that
307: up to terms of order $k^2$, the constant scalar field
308: surface and the constant energy density surface are identical, as
309: discussed in \cite{Brandenberger:2001bs}.} 
310: (the most physical choice of the
311: matching surface both in PBB and Ekpyrotic models). In particular, for
312: the Ekpyrotic model of \cite{Khoury:2001wf} rendered nonsingular by our
313: construction, we obtain a blue spectrum of the curvature perturbation 
314: with index $n_{\cal R}= 3$.
315: 
316: %%%%%%%%%%%%%%%%%%%%%%%%%%
317: \section{Single scalar field with an exponential potential}
318: %%%%%%%%%%%%%%%%%%%%%%%%%%
319: 
320: The Lagrangian considered in this paper can be used to describe both
321: a modified PBB model in which the dilaton has an exponential
322: potential as well as the Ekpyrotic scenario. Our Lagrangian describes
323: gravity plus a single scalar matter field $\phi$. In the case of
324: PBB cosmology, the physical frame is the string frame, and 
325: $\phi$ is the dilaton field. In the case
326: of the original version of the Ekpyrotic scenario \cite{Khoury:2001wf},
327: the physical frame is the Einstein frame since the dilaton is fixed, and
328: the field $\phi$ is related to the separation of a bulk brane from our
329: four-dimensional space-time orbifold fixed plane. In the case of
330: the second version of the Ekpyrotic scenario \cite{Khoury:2001bz} and in
331: the cyclic variant thereof 
332: \cite{Steinhardt:2001vw,Steinhardt:2001st}, $\phi$ is the modulus
333: field denoting the size of the orbifold (the separation of the two
334: orbifold fixed planes). 
335: 
336: We begin with the Lagrangian of the four-dimensional effective 
337: theory in the string frame, which is:
338: %
339: \be \label{effactions} 
340: {\cal S}_S \, = \, \int d^4x \sqrt{-g} e^{-\phi} 
341: \left[ \frac12 R+\frac12 (\nabla \phi)^2-V_S(\phi) \right]\,, 
342: \ee 
343: %
344: where $R$ is the Ricci scalar, $V_S(\phi)$ is the scalar field potential 
345: in the string frame. In this form, the action looks reminiscent of the action 
346: for PBB cosmology. Note that
347: $V_S(\phi)=0$ in the simplest version of the PBB scenario.  
348: We set the units such that $8\pi G \equiv 1$ with $G$ being 
349: a four-dimensional gravitational constant.   
350: Making a conformal transformation 
351: %
352: \be \label{conformal} 
353: \hat{g}_{\mu\nu}=e^{-\phi}g_{\mu\nu}\,, 
354: \ee
355: %
356: the action in the Einstein frame can be written as 
357: %
358: \be \label{effactione}
359: {\cal S}_E \, = \, \int d^4x \sqrt{-\hat{g}}\left[ \frac12 \hat{R}
360: - \frac14 (\hat{\nabla} \phi)^2-V_E(\phi) \right]\,, 
361: \ee
362: %
363: where 
364: %
365: \be \label{ekypo} 
366: V_E(\phi) \equiv e^{\phi}V_S(\phi) \,. 
367:  \ee 
368:  %
369: Introducing a rescaled field $\varphi=\pm \phi/\sqrt{2}$,
370: the action (\ref{effactione}) reads
371: %
372: \be \label{effactione2}
373: {\cal S}_E \, = \, \int d^4x \sqrt{-\hat{g}}\left[ \frac12 \hat{R}
374: -\frac12 (\hat{\nabla} \varphi)^2-V_E(\phi(\varphi)) \right]\,. 
375: \ee 
376: %
377: In this form, the action is seen to describe both the PBB
378: model in the Einstein frame, as well as the Ekpyrotic 
379: scenario \cite{Durrer:2002jn}.
380: 
381: The Ekpyrotic scenario is characterized by an exponential 
382: potential \cite{Khoury:2001wf} 
383: %
384: \be \label{einpoten} 
385: V_E = - V_{0} \exp 
386: \left(-\sqrt{\frac{2}{p}}\,\varphi\right)\,.  
387: \ee 
388: %
389: with $0 < p \ll 1$. The field $\varphi$ 
390: denotes the separation of two parallel branes. 
391: According to the Ekpyrotic scenario, the branes are 
392: initially widely separated but are approaching each other, 
393: which means that $\varphi$ begins near $+\infty$ and is
394: decreasing toward $\varphi = 0$.  In the PBB scenario, in contrast, the 
395: dilaton starts out from a weakly coupled regime with $\phi$ increasing
396: from $-\infty$.  
397: Thus, if we want the potential (\ref{einpoten}) to describe a
398: modified PBB scenario with a dilaton potential which is important
399: when $\phi \to 0$ but negligible for $\phi \to - \infty$, we 
400: have to use the relation $\varphi=-\phi/\sqrt{2}$ between the field 
401: $\varphi$ in the Ekpyrotic case and the dilaton $\phi$ in the PBB case.
402: 
403: Adopting the Friedmann-Robertson-Lemaitre-Walker (FRLW) metric 
404: $ds^2=-dt_E^2+a_E^2dx_E^2$ in the Einstein frame, 
405: the background equations are given by 
406: %
407: \be \label{basicein} 
408: 3H_E^2=\frac12\dot{\varphi}^2+V_E(\varphi)\,, \\
409: ~~~\ddot{\varphi}+3H_E\dot{\varphi}+V_E'(\varphi)
410: =0 \,, 
411: \ee 
412: %
413: where a prime denotes a derivative with respect to a cosmic time,
414: $t_E$. For the exponential potential (\ref{ekypo}) we have the following exact 
415: solution 
416: %
417: \be \label{ekysolution} 
418: a_E \propto (-t_E)^p,~~~H_E=\frac{p}{t_E},~~~ 
419: V_E=-\frac{p(1-3p)}{t_E^2},~~~ \dot{\varphi}=
420: \frac{\sqrt{2p}}{t_E} \,.  
421: \ee 
422: %
423: The solution for $t_E<0$ describes the contracting universe
424: in the Einstein frame prior to the collision of branes.
425: 
426: In the string frame the action is given by (\ref{effactions})
427: with potential 
428: %
429: \be \label{ekypos}
430: V_S= - V_{0} \exp \left[\left(\frac{1}{\sqrt{p}}-1\right) 
431: \phi\right]\,.  
432: \ee 
433: %
434: The FRLW metric in the string frame is described by
435: $ds^2=e^{-\phi}(-dt_S^2+a_S^2dx_S^2)$, which is connected to 
436: the quantities in the Einstein frame as 
437: %
438: \be \label{ta} 
439: dt_S=e^{-\varphi/\sqrt{2}}dt_E,~~~ 
440: a_S=e^{-\varphi/\sqrt{2}}a_E \,, 
441: \ee 
442: %
443: where we used the relation $\phi=-\sqrt{2}\varphi$.
444: Integrating the first relation gives
445: %
446: \be 
447: -(1-\sqrt{p})t_S=(-t_E)^{1-\sqrt{p}} \,. 
448:  \ee 
449:  %
450: Therefore the evolution of $a_S$ and $\phi$ in the string frame is
451: %
452: \be 
453: \label{ekysolution2}
454:  a_S \propto (-t_S)^{-\sqrt{p}},~~~ 
455: \phi=-\frac{2\sqrt{p}}{1-\sqrt{p}}\ln \left[ -(1-\sqrt{p})t_S\right] \,.  
456: \ee 
457: %
458: This illustrates the super-inflationary solution with growing dilaton 
459: from $\phi=-\infty$.
460: Note that singularities are inevitable in both frames as $t \to 0$.
461: We wish to analyze whether this singularity can be avoided 
462: by including higher-order corrections.
463: 
464: %%%%%%%%%%%%%%%%%%%%%%%%%%
465: \section{General actions and evolution equations}
466: %%%%%%%%%%%%%%%%%%%%%%%%%%
467: 
468: In this section  
469: we present the background and perturbed equations
470: in the case of a generalized action containing higher derivative terms. 
471: We write this action in the form \cite{Cartier:2001is,Hwang:1999gf}
472: %
473: \begin{eqnarray}
474:  S = \int d^4 x \sqrt{-g} \left[ \frac12 f(R, \phi) - \frac12 \omega 
475:  (\phi) (\nabla \phi)^2-V(\phi)+{\cal L}_c \right],
476: \label{lag}
477: \end{eqnarray}
478: %
479: where $f(R, \phi)$ is a function of the Ricci scalar $R$ and
480: a scalar field $\phi$.  $\omega(\phi)$ and $V(\phi)$ are general functions
481: of $\phi$.  The Lagrangian ${\cal L}_c$ represents the higher-order
482: corrections to the tree-level action. Both higher derivative
483: gravitational terms and terms involving $\varphi$ appear.
484: The action (\ref{lag}) applies not only to 
485: low-energy effective string theories, but also to effective action
486: approaches to Einstein quantum gravity and to scalar tensor theories, among
487: others.
488: 
489: As mentioned in the Introduction, our motivation for considering
490: the addition of higher derivative terms in the effective action is
491: to construct a nonsingular bouncing model and thus to overcome the 
492: singularity problem (``graceful exit problem'') which plagues both 
493: the PBB and the Ekpyrotic scenario. 
494: The higher-order contribution ${\cal L}_c$ can be written as the sum 
495: of the $\alpha'$ classical correction ${\cal L}_{\alpha'}$ and the quantum 
496: loop correction ${\cal L}_{q}$ \cite{Brustein:1997cv,Cartier:2001gc}. 
497: Both involve the same gravitational and scalar field terms, but are multiplied
498: by different powers of $e^{-\phi}$.  
499: 
500: The leading $\alpha'$ (string) correction to the
501: gravitational action we adopt is given 
502: by \cite{Gasperini:1996fu,Brustein:1997cv}
503: %%%%%%%%%%%%%
504: \begin{eqnarray}
505:  {\cal L}_{\alpha'} = -\frac12 \alpha' \lambda
506:   \xi(\phi) \left[ c R_{\rm GB}^2+ d 
507:  (\nabla \phi)^4 \right]\,,
508: \label{lagalpha}
509: \end{eqnarray}
510: %%%%%%%%%%%%%
511: where $\xi(\phi)$ is a general function of $\phi$ and $R_{\rm GB}^2 
512: =R^2-4R^{\mu\nu}R_{\mu\nu}+ R^{\mu\nu\alpha\beta}R_{\mu\nu\alpha\beta}$ is 
513: the Gauss-Bonnet term. The inverse string tension, $\alpha'$, is  set to unity.
514: The Gauss-Bonnet term has the property of keeping the order of the
515: gravitational equations of motion unchanged. It has been known since
516: the early days of string theory that this term arises as the lowest
517: string correction to the gravitational field equations in a string
518: theory background. At tree level (lowest order in $\hbar$), we have
519: $\xi(\phi) = -e^{-\phi}$. When applied to PBB cosmology, it turns out
520: that as $t \to -\infty$, the invariants $R_{\rm GB}^2$ and $(\nabla \phi)^4$
521: decay faster than the coefficient function $\xi(\phi)$ blows up. Hence,
522: the correction terms in the Lagrangian are unimportant for large negative 
523: values of $\phi$, but become important as the system approaches 
524: the strongly coupled region ($\phi \sim 0$).
525: 
526: Following \cite{Brustein:1997cv}, we take the higher $n$-loop correction
527: terms ${\cal L}_{q}$ in addition to the
528: tree-level term ${\cal L}_{\alpha'}$.
529: For the moment, however, we will keep $\xi(\phi)$ general.
530: We will give specific forms for $\xi(\phi)$ and ${\cal L}_{q}$ later.
531: 
532: %%%%%%%%%%%%%%%%%%%%
533:  \subsection{Background equations}
534: %%%%%%%%%%%%%%%%%%%%
535: 
536: Variation of the action (\ref{lag}) with respect to the scale factor,
537: the lapse function (then set to $1$ after the variation) and the
538: scalar matter field leads to 
539: the following background equations \cite{Cartier:2001is} 
540: %%%%%%%%%%%%%
541: \begin{eqnarray}
542: & & H^2 = \frac{1}{6F} \left( \omega \dot{\phi}^2+RF-f
543: +2V-6H\dot{F}+\rho_c \right)\,, \\
544: & & \dot{H}= \frac{1}{2F} \left(-\omega \dot{\phi}^2+H\dot{F} 
545: -\ddot{F}-\frac12 \rho_c-\frac12 p_c \right)\,, \\
546: & & \ddot{\phi}+3H\dot{\phi}+\frac{1}{2\omega}
547: \left( \omega_{\phi}\dot{\phi}^2-f_{\phi}+2V_{\phi}
548: -\Delta_{\phi} \right)=0 \,,
549: \label{back}
550: \end{eqnarray}
551: %%%%%%%%%%%%%
552: where $F \equiv \partial f/\partial R$ and $H \equiv \dot{a}/a$.
553: $\rho_c$, $p_c$, and $\Delta_{\phi}$ correspond to the higher-order
554: curvature and derivative corrections with stress-energy tensor  
555: $T^{\mu}_{\nu}=(-\rho_c, p_c, p_c, p_c)$. 
556: $\Delta_{\phi}$ comes from the variation 
557: of ${\cal L}_c$ with respect to $\phi$.  
558: For the tree-level $\alpha'$ correction (\ref{lagalpha}), one has 
559: %%%%%%%%%%%%%
560: \begin{eqnarray}
561: & & \rho_c =2\alpha' \lambda \left( 12c \dot{\xi}H^3
562: -\frac32 d\xi \dot{\phi}^4 \right)\,,
563: \label{corre1} \\
564: & & p_c \equiv -2\alpha' \lambda \left\{ 4c \left[ \ddot{\xi}
565: H^2+ 2\xi H(\dot{H}+H^2)\right]+\frac12 d\xi \dot{\phi}^4 \right\}\,, \\
566: & & \Delta_{\phi} \equiv -\alpha' \lambda \left[ 24 c \xi_{\phi}
567: H^2 (\dot{H}+H^2)-d \dot{\phi}^2(3\dot{\xi}\dot{\phi}
568: +12\xi \ddot{\phi}+12\xi\dot{\phi}H) \right]\,.
569: \label{alphacorre}
570: \end{eqnarray}
571: %%%%%%%%%%%%%
572: Note that taking into account quantum loop corrections provides 
573: additional source terms for $\rho_c$, $p_c$, and $\Delta_{\phi}$.  
574: We will discuss this issue in Sec.~IV.
575: 
576: %%%%%%%%%%%%%%%%%%%%
577:  \subsection{Perturbation equations}
578: %%%%%%%%%%%%%%%%%%%%
579: 
580: A perturbed space-time
581: metric has the following form for scalar perturbations 
582: in an arbitrary gauge (see e.g. \cite{MFB}, where the function
583: $A$ is denoted by $\phi$):
584: %%%%%%%%%%%%%
585: \begin{eqnarray}
586: ds^2 = -(1+2A)dt^2 + 2a(t)B_{,i} dx^idt 
587: +&a^2(t)[(1-2\psi)\delta_{ij}+2E_{,i,j}] dx^i dx^j\,, 
588: \label{pmetric}
589: \end{eqnarray}
590: %%%%%%%%%%%%%
591: where a comma denotes the usual flat space 
592: coordinate derivative. 
593: We introduce the curvature perturbation, ${\cal R}$, in the comoving gauge
594: \cite{Lyth:1985} 
595: %%%%%%%%%%%%%
596: \begin{eqnarray}
597: {\cal R} \equiv \psi+\frac{H}{\dot{\phi}}\delta \phi\,.
598: \label{metric}
599: \end{eqnarray}
600: %%%%%%%%%%%%%
601: The perturbed Einstein equations for the action (\ref{lag})
602: are written in the form \cite{Hwang:1999gf,Cartier:2001is} 
603: %%%%%%%%%%%%%
604: \begin{eqnarray}
605: \frac{1}{a^3Q} \left(a^3Q \dot{\cal R} \right)^{\bullet} -s 
606: \frac{\Delta}{a^2} {\cal R}=0\,,
607: \label{peinstein}
608: \end{eqnarray}
609: %%%%%%%%%%%%%
610: where
611: %%%%%%%%%%%%%
612: \begin{eqnarray} \label{Qs2}
613: Q &\equiv& \frac{\omega \dot{\phi}^2 +3I
614: (\dot{F}-4\lambda c \dot{\xi}H^2)-6\lambda d \xi 
615: \dot{\phi}^4}{\left( H + I\right)^2}\,,  \\
616: s &\equiv& 1+\frac{4\lambda c\xi 
617: \dot{\phi}^4 -16\lambda c \dot{\xi}\dot{H}I+ 8\lambda c 
618: (\ddot{\xi}-\dot{\xi} H)I^2} {\omega \dot{\phi}^2 
619: +3I(\dot{F} -4\lambda c \dot{\xi}H^2)
620: -6\lambda d \xi \dot{\phi}^4}\,.
621: \label{Qs}
622: \end{eqnarray}
623: %%%%%%%%%%%%%
624: with $I \equiv (\dot{F}-4\lambda c \dot{\xi}H^2)/ (2F-8\lambda c \dot{\xi} H)$.
625: 
626: Introducing a new quantity, \footnote{Note that $\Psi$ is the variable in
627: terms of which - for unmodified Einstein gravity - the action for 
628: fluctuations has the canonical form of a
629: free field action with time dependent mass (see e.g. \cite{MFB} where
630: the variable is denoted as $v$).}
631: $\Psi \equiv z{\cal R}$, with $z \equiv 
632: a\sqrt{Q}$, each Fourier component of $\Psi$ satisfies the second order 
633: differential equation 
634: %%%%%%%%%%%%%
635: \begin{eqnarray}
636: \Psi_k''+\left(sk^2-\frac{z''}{z}\right)\Psi_k=0\,,
637: \label{Psi}
638: \end{eqnarray}
639: %%%%%%%%%%%%%
640: where a prime denotes the derivative with respect to conformal time,
641: $\eta \equiv \int a^{-1}dt$.  In the large scale limit, $|sk^2| \ll |z''/z|$, 
642: eq.~(\ref{Psi}) is integrated to give 
643: %%%%%%%%%%%%%
644: \begin{eqnarray}
645: {\cal R}_k =C_k+D_k \int \frac{d\eta}{z^2}\,,
646: \label{Rk}
647: \end{eqnarray}
648: %%%%%%%%%%%%%
649: where $C_k$ and $D_k$ are integration constants. 
650: The curvature perturbation 
651: is conserved on super-Hubble scales as long as the second term in 
652: eq.~(\ref{Rk}) is not strongly dominating, as in the case of the single 
653: field, slow-roll inflationary scenarios.  
654: 
655: If the evolution of $z$ before the bounce is given in the form 
656: \footnote{Note that as long as the additional terms ${\cal L}_c$ in
657: the action are negligible, then $\gamma = p/(1 - p)$, and thus
658: $0 < \gamma \ll 1$ for the collapsing phase of Ekpyrotic cosmology,
659: and $\gamma = 1/2~(p = 1/3)$ for the collapsing phase of PBB.} 
660: %%%%%%%%%%%%%
661: \begin{eqnarray}
662: z \propto (-\eta)^{\gamma}\,,
663: \label{z}
664: \end{eqnarray}
665: %%%%%%%%%%%%%
666: the second term in eq.~(\ref{Rk})
667: yields $\int d\eta/z^2 \propto (-\eta)^{1-2\gamma}$.  Therefore curvature
668: perturbations can be amplified for $\gamma \ge 1/2$ on super-Hubble scales, 
669: while they are not for $\gamma<1/2$ \cite{Tsujikawa:2001ad}
670:  (Note that ${\cal R}_k \propto {\rm ln} 
671: (-\eta)$ for $\gamma=1/2$).  Whether this enhancement occurs or not depends 
672: on the time evolution of $z$, and therefore on the string
673: cosmological model.
674: 
675: We need to go to the next order solution of eq.~(\ref{Psi})
676: in order to obtain the spectrum of curvature perturbations.  If $s$ is 
677: a positive constant (as it will be in the asymptotic limits), 
678: the solution for $\Psi_k$ is expressed by the 
679: combination of the Hankel functions: 
680: %%%%%%%%%%%%%
681: \begin{eqnarray}
682: \Psi_k=\frac{\sqrt{\pi |\eta|}}{2}\left[ c_1 H_{\nu}^{(1)}
683: (x) +c_2 H_{\nu}^{(2)}(x) \right]\,,
684: \label{han}
685: \end{eqnarray}
686: %%%%%%%%%%%%%
687: where 
688: %%%%%%%%%%%%%
689: \begin{eqnarray}
690: x \equiv \sqrt{s} k |\eta|\,~~~~\nu \equiv \frac12 |1-2\gamma|\,.
691: \label{def}
692: \end{eqnarray}
693: %%%%%%%%%%%%%
694: The solution (\ref{han}) corresponds to the Minkowski 
695: vacuum state in the small-scale limit ($k \to \infty$).
696: 
697: We can expand the Hankel functions in the following 
698: form \cite{Hwang:2002ks}:
699: %
700: \begin{eqnarray} 
701: H_{\nu}^{(1, 2)} (x)=\sum_{n=0}^\infty \frac{1}{n!}
702: \left(-\frac{x^2}{4}\right)^n \frac{1}{\sin \pi \nu}
703: \left[ \left(\frac{x}{2}\right)^{\nu} \frac{\pm ie^{\mp i
704: \pi \nu}}{\Gamma(\nu+n+1)}+\left(\frac{x}{2}\right)^{-\nu} \frac{\mp 
705: i}{\Gamma(-\nu+n+1)}\right]\,.
706: \label{Hankel}
707: \end{eqnarray}
708: %
709: The curvature perturbation, ${\cal R}_k=
710: \Psi_k/z$, has two solutions which are proportional to 
711: $k^{\nu}|\eta|^{\nu-\gamma+1/2}$
712: and $k^{-\nu}|\eta|^{-\nu-\gamma+1/2}$, which follow from the 
713: first and second term in eq.~(\ref{Hankel}), respectively.
714: In the large-scale limit ($k \to 0$), the contribution of 
715: the second term dominates over the first one, 
716: thereby yielding the spectrum of the curvature perturbation as 
717: %%%%%%%%%%%%%
718: \begin{eqnarray}
719: P_{{\cal R}} \equiv \frac{k^3}{2\pi^2}\left| 
720: {\cal R}_k \right|^2 \propto k^{3-2\nu} \propto k^{n_{\cal{R}}-1}\,,
721: \label{PR}
722: \end{eqnarray}
723: %%%%%%%%%%%%%
724: in which case the spectral tilt is 
725: %%%%%%%%%%%%%
726: \begin{eqnarray}
727: n_{\cal{R}}-1=3-\left| 1-2\gamma \right|\,.
728: \label{ind}
729: \end{eqnarray}
730: %%%%%%%%
731: %%%%%%%%%%%%%
732: Note that we have $k^{-\nu}|\eta|^{-\nu-\gamma+1/2}=k^{-\nu}|\eta|^0$
733: for $\gamma<1/2$, in which case the constant mode $C_k$ in 
734: eq.~(\ref{Rk}) corresponds 
735: to the solution which comes from the second term in eq.~(\ref{Hankel}).
736: The Ekpyrotic scenario with a negative potential 
737: ($0<p<1/3$) belongs to this case ($\gamma<1/2$)
738: as we will show later. 
739: When $\gamma>1/2$ one has $k^{-\nu}|\eta|^{-\nu-\gamma+1/2}=
740: k^{-\nu}|\eta|^{1-2\gamma}$, which means that ${\cal R}_k$ grows 
741: before the graceful exit ($\eta<0$). We will discuss a 
742: string-inspired model that belongs to this case in Sec.~V.
743: The PBB scenario corresponds to the marginal case with 
744: $\gamma=1/2$.
745: 
746: Note that $s$ is exactly unity in eq.~(\ref{Psi}) when
747: the corrections ${\cal L}_c$ are not taken into account. 
748: In the presence of higher-order corrections (${\cal L}_c \ne 0$),  $s$ is 
749: generally a time-varying function, in which case the formula (\ref{ind}) can 
750: not be directly applied. 
751: Nevertheless it is still valid if $s$ is a slowly varying positive function.
752: 
753: In subsequent sections we shall apply the above general formulas 
754: to concrete string-inspired models. In Section IV we apply the
755: string loop and quantum corrections to the low energy effective action
756: in the string frame and find nonsingular bouncing cosmological solutions.
757: In Section V, we consider a situation with fixed dilaton and add a
758: Gauss-Bonnet term to the Einstein frame action. We find nonsingular
759: cosmological solutions which begin in an asymptotically flat state,
760: undergo a period of super-exponential inflation which terminates
761: with a graceful exit.
762: 
763: 
764: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
765: \section{Inclusion of higher-order corrections in the string 
766: frame \\
767: --dilaton-driven case} 
768: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
769: 
770: In the context of PBB cosmology, the natural frame to use
771: in order to define the correction term ${\cal L}_c$ in the
772: Lagrangian is the string frame. In this case, we should use
773: $f=e^{-\phi}R$ and $\omega=-e^{-\phi}$, i.e. 
774: %
775: \be \label{actionloop} 
776: {\cal S}_S \, = \, \int d^4x \sqrt{-g}\left\{ e^{-\phi} \left[ \frac12 
777: R+\frac12 (\nabla \phi)^2-V_S(\phi) \right]+{\cal L}_c\right\} \,,
778: \ee 
779: %
780: where $\phi$ corresponds to the dilaton. This Lagrangian was
781: suggested in \cite{Gasperini:1996fu,Brustein:1997cv}, and used
782: to construct nonsingular background cosmological solutions of
783: PBB cosmology in the absence of a potential for the dilaton.
784: Fluctuations in this model were studied 
785: in \cite{Cartier:2001is}
786: in the absence of a dilaton potential. 
787: 
788: In the following, we
789: extend these analyses to the case of a non-vanishing dilaton
790: potential. We will first construct nonsingular solutions which
791: in the Einstein frame correspond to nonsingular bouncing cosmologies.
792: We then study how the fluctuations evolve across the bounce and
793: compute the spectrum of fluctuations. The analysis in this section
794: thus applies immediately to the modified PBB scenario in which the
795: dilaton has a negative exponential potential. We can also apply
796: the results to the initial version \cite{Khoury:2001wf}
797: of the Ekpyrotic scenario. In
798: this case, the brane collision occurs at $\varphi = 0$, and since
799: thus the gravitational coupling constant does not change significantly
800: near the bounce, the difference in the role of the higher derivative
801: terms between the Einstein frame (the frame in which it appears
802: most logical to define the correction terms in the Lagrangian) and
803: the ``string frame'' (quotation marks used here because in the case
804: of the Ekpyrotic scenario $\varphi$ is not the dilaton, the dilaton
805: being fixed) is not expected to be significant. The application of
806: the results of this section to the version of the Ekpyrotic scenario
807: with moving boundary branes \cite{Khoury:2001bz} and to the cyclic
808: scenario \cite{Steinhardt:2001vw} is more problematic since in
809: this case $\varphi$ is the dilaton, $\varphi \to -\infty$ at the
810: bounce, and thus the difference in the evolution of models with
811: correction terms defined in the Einstein and string frames is
812: expected to be important. 
813: 
814: The reason why nonsingular solutions are possible in
815: the presence of the correction term ${\cal L}_c$, is that such a term
816: can lead to violations of the null energy condition (from the
817: perspective of an observer using unmodified Einstein equations). 
818: Thus, it is expected to lead to a successful graceful exit, in
819: the same way that introducing matter violating the null energy
820: condition allowed the construction of nonsingular bouncing models in
821: \cite{Hwang:2001zt,Peter:2002cn}. 
822: 
823: In this model the background equations 
824: are written as 
825: %
826: \beqa \label{motionse} 
827: 6H^2-6H\dot{\phi}+\dot{\phi}^2-2V_S=e^{\phi}\rho_c 
828: \,, \\
829: 4\dot{\phi}H-4\dot{H}-6H^2-\dot{\phi}^2+2\ddot{\phi}+2V_S=
830: e^{\phi}p_c \,, \\
831: 6\dot{H}+12H^2+\dot{\phi}^2-2\ddot{\phi}-6H\dot{\phi}
832: -2(V_S-V_S')=e^{\phi} \Delta_{\phi}\,. 
833: \label{motionse2} 
834: \eeqa 
835: %
836: The dilatonic corrections ${\cal L}_c$ are the sum of the tree-level 
837: $\alpha'$ corrections and the quantum $n$-loop corrections ($n=1, 2, 3, 
838: \cdots$), with the function $\xi(\phi)$ [see (\ref{lagalpha})] given by
839: %
840: \beqa 
841: \xi(\phi)=-\sum_{n=0} C_n e^{(n-1)\phi} \,,
842: \label{xifunction} 
843: \eeqa 
844: %
845: where $C_n$ ($n \ge 1$) are the coefficients of 
846: $n$-loop corrections with $C_0=1$.
847: In this case the source terms due to ${\cal L}_c$ on the right hand side of 
848: (\ref{motionse})-(\ref{motionse2}) are given by \cite{Cartier:2001gc}
849: %
850: \begin{eqnarray} \label{C_n}
851: \rho_c=\sum_{n=0} C_n \left\{\rho_c \right\}_n\,,~~~
852: p_c=\sum_{n=0} C_n 
853: \left\{p_c \right\}_n,~~~ \Delta_\phi=\sum_{n=0} C_n 
854: \left\{\Delta_\phi \right\}_n\,,
855: \label{sum}
856: \end{eqnarray}
857: %
858: where 
859: %
860: \begin{eqnarray}
861: \left\{ \rho_c \right\}_n &=& \alpha' \lambda
862: \dot{\phi} e^{(n-1)\phi} 
863: \left\{-24c(n-1)H^3+3d\dot{\phi}^3 \right\}\,, \\
864: \left\{p_c \right\}_n &=& \alpha' \lambda e^{(n-1)\phi} \left\{8c(n-1)H 
865: \left[(n-1)\dot{\phi}^2H+\ddot{\phi}H+ 2\dot{\phi}(\dot{H}+H^2) \right] 
866: +d\dot{\phi}^4 \right\}\,, \\
867: \left\{\Delta_{\phi} \right\}_n &=& \alpha' \lambda  e^{(n-1)\phi} 
868: \left\{24c(n-1)H^2(\dot{H}+H^2) -3d\dot{\phi}^2 
869: \left[4\ddot{\phi}+4\dot{\phi}H+ (n-1)\dot{\phi}^2 \right] \right\}\,,
870: \label{corre}
871: \end{eqnarray}
872: %
873: with $\lambda=-1/4$.  Following ref.~\cite{Cartier:2001is} 
874: we choose the coefficients 
875: as $c=-d=-1$.  Note that the above corrections include the $\alpha'$ 
876: corrections (\ref{corre1})-(\ref{alphacorre}), corresponding to $n=0$.  
877: 
878: It is also convenient to relate the Hubble parameter and its derivative in the 
879: Einstein frame with those in the string frame by using eq.~(\ref{ta})
880: %
881: \begin{eqnarray} 
882: H_E = e^{\phi/2}\left(H_S-\frac{\dot{\phi}}{2}\right),~~~~ 
883: \dot{H}_E = e^{\phi} \left( \dot{H}_S-\frac{\ddot{\phi}}{2}
884: +\frac12 \dot{\phi}H_S-\frac{\dot{\phi}^4}{4}\right)\,.
885: \label{relation}
886: \end{eqnarray}
887: %
888: Here the dots on the right hand side 
889: denote the time-derivative with respect to $t_S$. 
890: The energy density $\rho_E$ and the pressure $p_E$ in the Einstein frame are 
891: expressed as 
892: %
893: \begin{eqnarray}
894: \rho_E=3H_E^2\,,~~~p_E=-3H_E^2-2\dot{H}_E\,.
895: \label{relation2}
896: \end{eqnarray}
897: %
898: Once we know the evolution of the background in the string frame, 
899: it is easy to find the evolution of $H_E, a_E, \varphi=-\phi/\sqrt{2}$
900: and to check whether the null energy condition, $\rho_E+p_E>0$,
901: holds or not in the Einstein frame by using eqs.~(\ref{relation})
902: and (\ref{relation2}).  Note that in the absence of higher-order corrections
903: (${\cal L}_c=0$) one has $2\dot{H}_E=-(\rho_E+p_E)=-\dot{\phi}^2<0$.  
904: In this case once the contraction begins ($\dot{H}_E<0$) the Hubble 
905: parameter is {\it always} negative.  Therefore it is not possible to 
906: have bouncing solutions
907: required for the nonsingular Ekpyrotic scenario unless higher-order 
908: corrections ${\cal L}_c$ are taken into account.
909: 
910: %%%%%%%%%%%%%%%%%%%%%%%%
911:  \subsection{Background evolution} 
912: %%%%%%%%%%%%%%%%%%%%%%%%
913:  
914: In the absence of a negative exponential potential ($V_S=0$),
915: it was found in Ref.~\cite{Gasperini:1996fu} that curvature singularities 
916: can be avoided by taking into account higher-order corrections ${\cal L}_c$.
917: In this case we have nonsingular bouncing solutions
918: in the Einstein frame due to the violation of the null energy condition. 
919: We are interested in whether singularity avoidance is possible or not 
920: in the presence of the Ekpyrotic potential (\ref{ekypos}). Note that
921: since near the bounce $H^2 \sim t^{-2}$, higher curvature corrections
922: to the Einstein action will likewise be important in the presence of a
923: potential.
924: 
925: When $V_S \ne 0$ and ${\cal L}_c=0$ the background solutions are described 
926: by eqs.~(\ref{ekysolution}) and (\ref{ekysolution2}).  In the string frame 
927: the scale factor evolves super-inflationary with growing Hubble rate 
928: ($\dot{H}_S>0$).  We plot in Fig.~\ref{dilaton1} the evolution of 
929: background quantities both in the string and Einstein frames [see the case 
930: (i)].  The dilaton $\phi$ starts out from the weakly coupled regime 
931: $g^2_{string}\equiv e^{\phi} \ll 1$, corresponding to widely separated 
932: branes in the Ekpyrotic scenario, $\varphi=-\phi/\sqrt{2} \gg 1$.  
933: In the Einstein frame the Universe 
934: is contracting with a negative Hubble rate.  The solution inevitably 
935: meets a curvature singularity as $\phi$ grows toward the strongly coupled 
936: regime ($g^2_{string} \sim 1$).
937: 
938: Our first main finding is that with $V_S(\phi) \ne 0$ there exist nonsingular 
939: trajectories  in the presence of higher-order 
940: corrections (${\cal L}_c \ne 0$).  
941: Thus, the presence of the potential for the dilaton
942: does not prevent the higher derivative terms from being able to
943: smooth out the curvature singularity. The details
944: depend on the value of the power-law index, $p$.
945: When $p \ll 1$ the Ekpyrotic potential (\ref{einpoten}) is exponentially 
946: suppressed for $\varphi~\gsim~1$, in which case the dynamics of 
947: the system is hardly affected by the negative potential except for 
948: the region, $\varphi \sim 0$. However, in this region the higher
949: derivative terms play a crucial role. 
950: 
951: In our simulations, we have adopted the potential 
952: (\ref{einpoten}) for $\varphi>0$ and $V_E=0$ for $\varphi<0$. This
953: is in the spirit of the first version of the Ekpyrotic scenario
954: \cite{Khoury:2001wf} in which the potential vanishes at the brane 
955: collision, the bulk brane is absorbed by the orbifold fixed
956: plane via a small instanton transition, and there is no potential
957: left afterwards.   We show in Fig.~\ref{dilaton1} the dynamical 
958: evolution of the system for $p=0.1$. The case (i) is the one in
959: which only tree-level terms are present and in which singularity
960: avoidance is not possible.  
961: The case (ii) corresponds to the one where both tree-level and one-loop 
962: corrections are taken into account ($C_1=1.0$ and $C_2=0$).  Inclusion of 
963: one-loop corrections makes it possible to have nonsingular cosmological 
964: solutions.  In fact $\rho_E+p_E$ becomes negative around $t_S \sim 115$ in 
965: Fig.~\ref{dilaton1}, after which the Hubble parameter $H_E$ begins to grow.  
966: The Universe starts to expand once $H_E$ crosses zero.  Namely the 
967: violation of the null energy condition allows to have nonsingular bouncing solutions in the 
968: Einstein frame.  Nevertheless we should notice that the scale factors 
969: evolve super-inflationary both in the string and Einstein frames due to 
970: unbounded increase of $H_S$ and $H_E$, together with rapid growth of the 
971: field $\phi$.  Therefore we are faced with another problem, namely how 
972: to connect to the stage of a decreasing Hubble parameter.
973: 
974: If two loop terms are added
975: (keeping the previous tree-level and one loop terms)
976: phenomenologically more appealing nonsingular solutions can be
977: obtained.  When  $C_2$ is positive, the evolution of 
978: the system does not differ significantly compared to the case (ii).
979: However, it is possible to obtain a decreasing
980: Hubble rate if we take a negative value of $C_2$.
981: The case (iii) of Fig.~\ref{dilaton1} corresponds to the 
982: coefficients $C_1=1.0$ and $C_2=-1.0 \times 10^{-3}$.  
983: We find that the growth rates of the scale factor and of $\phi$ are slowed 
984: compared to the case (ii). 
985: We see that $\rho_E+p_E$ becomes positive and begins to 
986: decrease toward $+0$ after the short period of violations of 
987: the null energy condition.
988: Although this case does not correspond to the radiation-dominated Universe
989: after the graceful exit, it is possible to connect to it by taking into 
990: account the decay of the dilaton to radiation \footnote{If we were
991: to include production of radiation at a fixed time during the expanding phase,
992: we could use the well-known results on the constancy of ${\cal R}$ in
993: the expanding phase \cite{Lyth:1985,Brandenberger:1984,Bardeen:1983} to  
994: argue that the spectrum of fluctuations on large scales will be the same
995: as what is obtained in this paper. The crucial fact about our bounce is
996: that it is not symmetric in time (see Fig.~1).}. However, including  
997: radiation in the Ekpyrotic cosmology has some subtle points,
998: and we do not consider this problem in the present work. 
999: 
1000: We have checked that the addition of three loop terms with coefficients
1001: chosen to be of the order $10^{-7}$ (roughly the same hierarchy of
1002: coefficients between the two and three loop terms as between the
1003: one and two loop terms)  does not change the results of the two loop
1004: analysis in a significant way. With a coefficient of the three loop
1005: term of the order 1, the background solution ceases to be nonsingular.
1006: 
1007: We emphasize that we have nonsingular bouncing solutions 
1008: in the Einstein frame even in the presence of a negative exponential
1009: potential.  When $p \ll 1$ the potential is vanishingly small for 
1010: $\varphi \gg 1$, in which case the dynamics of the system is practically 
1011: the same as that of the zero potential discussed in 
1012: Ref.~\cite{Brustein:1997cv}.
1013: In this case the dilaton starts out from the low-curvature regime $|\phi| 
1014: \gg 1$, which  is followed by the string phase with linearly growing dilaton 
1015: and nearly constant Hubble parameter.
1016: During the string phase one has \cite{Gasperini:1996fu} 
1017: %
1018: \begin{eqnarray} 
1019: a_S \propto (-\eta_S)^{-1},~~~~
1020: \phi=-\frac{\dot{\phi}_f}{H_f} {\rm ln} (-\eta_S)
1021: +{\rm const}\,,
1022: \label{const}
1023: \end{eqnarray}
1024: % 
1025: where $\dot{\phi}_f \simeq 1.40$ and $H_f \simeq 0.62$.
1026: In the Einstein frame this corresponds to a contracting Universe with
1027: %
1028: \begin{eqnarray} 
1029: a_E \propto (-\eta_E)^{\dot{\phi}_f/(2H_f)-1}\,.
1030: \label{scaleein}
1031: \end{eqnarray}
1032: %  
1033: 
1034: On the other hand, we can consider the scenario where the negative Ekpyrotic 
1035: potential dominates initially but the higher-order correction becomes 
1036: important when two branes approach sufficiently.  Numerically we confirmed 
1037: that it is possible to have nonsingular solutions 
1038: (see Fig.~\ref{dilaton2}).  
1039: In the simulations we included the correction terms 
1040: of ${\cal L}_c$ only for $\varphi~\lsim~1$.  In this case the background 
1041: solutions are described by eq.~(\ref{ekysolution}) or (\ref{ekysolution2}) 
1042: before the higher-order correction terms begin to work.  Given this 
1043: background solution, one can obtain the spectra of curvature perturbations 
1044: analytically, as we will see in the next section.  The spectra depend on 
1045: whether the higher-order terms are always dominant or not relative to the 
1046: negative potential before the bounce.
1047: 
1048: %%%%%%%%%%
1049: \begin{figure}
1050: \begin{center}
1051: \singlefig{14cm}{dilaton1.eps}
1052: \begin{figcaption}{dilaton1}{14cm}
1053: The evolution of $H_S$, $H_E$, $a_S$, $a_E$, $\phi$, and
1054: $\rho_E+p_E$ with $c=-1$, $d=1$, $p=0.1$.
1055: We choose initial conditions $\phi=-20$, $H=5.0 \times 10^{-3}$.
1056: The cases correspond to (i) only
1057: tree-level correction terms but no higher-order 
1058: corrections ($C_1=C_2=0$), (ii) tree-level and one-loop corrections 
1059: present ($C_1=1.0, C_2=0$) (iii) tree-level and one- and two-loop 
1060: corrections present with $C_1=1.0$ and $C_2= -1.0 \times 10^{-3}$.
1061: \end{figcaption}
1062: \end{center}
1063: \end{figure}
1064: %%%%%%%%%%
1065: %%%%%%%%%%
1066: \begin{figure}
1067: \begin{center}
1068: \singlefig{9cm}{dilaton2.eps}
1069: \begin{figcaption}{dilaton2}{9cm}
1070: The evolution of $\varphi$ and $a_E$  
1071: with $c=-1$, $d=1$, $p=0.1$, $C_1=1.0$ 
1072: and $C_2= -1.0 \times 10^{-3}$.  In this 
1073: case we include the correction term ${\cal L}_c$ only for $\varphi<1$.  
1074: We choose initial conditions $\phi=-15$, $H=1.5 \times 10^{-3}$.
1075: Prior to the collision of branes at $\varphi=0$, the universe is slowly 
1076: contracting, which is followed by the bouncing solution 
1077: through higher-order corrections.
1078: \end{figcaption}
1079: \end{center}
1080: \end{figure}
1081: %%%%%%%%%%
1082: 
1083: We have also studied Ekpyrotic potentials with other values of $p$, and
1084: found that if the potential is negative, corresponding to $p<1/3$, then 
1085: singularity avoidance is possible for suitable choices of $C_1$ and $C_2$ 
1086: as in the case $p = 0.1$ shown in Fig.~\ref{dilaton1}.   
1087: In the case of $p>1/3$ the field $\phi$ climbs up a positive
1088: exponential potential due to the Hubble contraction term. 	 
1089: When $p~\gsim~1/2$ with $V_0=p(1-3p)$ we found that the field $\phi$ returns 
1090: back before it reaches the strongly coupled region, $\phi~\gsim~0$.  
1091: This is equivalent to the fact that two parallel branes do not approach 
1092: each other sufficiently.  In such cases the positive exponential potential
1093: makes the field bounce back before the higher-order correction 
1094: becomes important (this may be related to the instability discussed
1095: recently in \cite{Heard:2002dr}).  
1096: If we choose smaller values of $V_0$, it is possible to 
1097: have nonsingular bouncing solutions which are similar to 
1098: Fig.~\ref{dilaton1}.  This case corresponds to the one where the effect of 
1099: the positive potential is negligible compared to higher-order corrections, 
1100: in which case the background solutions are given by eq.~(\ref{const}).  When 
1101: the positive potential is dominant from the beginning, it is difficult to 
1102: obtain a solution where a successful graceful exit is realized by 
1103: higher-order corrections.
1104: 
1105: %%%%%%%%%%%%%%%%%%%%%%%
1106:  \subsection{Density perturbations}
1107: %%%%%%%%%%%%%%%%%%%%%%%
1108: 
1109: Let us proceed to the analysis of the evolution and 
1110: the spectra of density perturbations.
1111: We shall consider two cases: (i) the effect of the potential $V(\phi)$ is 
1112: always negligible relative to the correction term ${\cal L}_c$, and (ii) 
1113: the effect of the correction term ${\cal L}_c$ becomes important only 
1114: around the graceful exit ($\phi \sim 0$). Note that the second case
1115: is the physically more interesting one for applications to Ekpyrotic
1116: cosmology.
1117: 
1118: %%%%%%%%%%%%%%%%%%%%%%%%%%
1119:  \subsubsection{{\rm Case (i)}: $|V(\phi)| \ll |{\cal L}_c|$}
1120: %%%%%%%%%%%%%%%%%%%%%%%%%%
1121: 
1122: When the correction terms (\ref{corre1})-(\ref{alphacorre})
1123: always dominate relative to the exponential potential (\ref{einpoten}),
1124: the spectra of density perturbations are similar to the ones discussed in 
1125: ref.~\cite{Cartier:2001is}.  During the string phase with linearly growing 
1126: dilaton and nearly constant Hubble parameter with $\dot{\phi}_f \simeq 
1127: 1.40$ and $H_f \simeq 0.62$, we have a sufficient amount of inflation with 
1128: e-folds $N \equiv {\rm ln} (a/a_i) >60$ provided that the dilaton field 
1129: satisfies $|\phi| \gg 1$ initially \cite{Cartier:2001is,Cartier:2001gc}.  
1130: In this stage $Q$ defined in eq.~(\ref{Qs}) is proportional to $e^{-\phi}$ 
1131: by making use of eq.~(\ref{const}), thereby leading to 
1132: %
1133: \begin{eqnarray} 
1134: z \propto (-\eta_S)^{\gamma}\,,~~~{\rm with}~~~
1135: \gamma=-1+\frac{\dot{\phi}_f}{2H_f}\simeq 0.13\,.
1136: \label{gamma}
1137: \end{eqnarray}
1138: %
1139: Making use of the relation (\ref{ind}) which is valid for positive $s$, the 
1140: spectral tilt of the large-scale curvature perturbation is 
1141: %
1142: \begin{eqnarray} 
1143: n_{\cal{R}}-1=3-\left| 3- \frac{\dot{\phi}_f}{H_f} \right| 
1144: \simeq 2.26\,.
1145: \label{tilt1}
1146: \end{eqnarray}
1147: %
1148: 
1149: The evolution of the frequency shift $s$ is nontrivial (see Fig.~\ref{s}). 
1150: In the low curvature regime where the higher-order terms are not
1151: important, $s$ is positive ($s \simeq 1$), as in the usual
1152: PBB scenario.  It then changes sign and becomes 
1153: negative during a short transition from the low-curvature regime to the 
1154: string phase.  During the string phase, $\dot{\phi}$ and 
1155: $H_S$ are constant ($\dot{\phi} \simeq 1.40$ and $H_S \simeq 0.62$), 
1156: and $\xi \sim -e^{-\phi}$. The correction term on the
1157: right hand side of eq.~(\ref{Qs}) for $s$ dominates
1158: in this phase.
1159: It follows from eq.~(\ref{Qs}) that the $\phi$ dependence of the 
1160: leading term cancels out between $\xi(\phi)$ and $\omega(\phi)$, 
1161: and that hence $s$ is
1162: constant and negative until the graceful 
1163: exit.  In a stage with negative constant $s$, the solution of 
1164: eq.~(\ref{Psi}) can be written in the form 
1165: %
1166: \begin{eqnarray}
1167: \Psi_k=\sqrt{|\eta|} 
1168: \left[c_1 I_{\nu} (x) +c_2 K_{\nu} (x) \right]\,,
1169: \label{Psid}
1170: \end{eqnarray}
1171: %
1172: where $x$ and $\nu$ are given in (\ref{han}) (with $s$ replaced by the
1173: absolute value of $s$), and $I_{\nu}$ and $K_{\nu}$ are modified 
1174: Bessel functions, whose asymptotic solutions are $I_{\nu} \propto x^{\nu}$, 
1175: $K_{\nu} \propto x^{-\nu}$ for $x \to 0$, and $I_{\nu} \sim e^x/\sqrt{2\pi 
1176: x}$, $K_{\nu} \sim \sqrt{\pi/(2x)}e^{-x}$ for $x \to \infty$.  
1177: Then one reproduces the spectral tilt (\ref{tilt1}) in the large-scale limit 
1178: ($|sk^2| \ll |z''/z|$).  For small-scale modes curvature perturbations show 
1179: exponential instability due to negative frequency shift.  After the horizon 
1180: crossing ($|sk^2|~\lsim~|z''/z|$), curvature perturbations are frozen, 
1181: since $\gamma$ is smaller than $1/2$ in this case.
1182: 
1183: It was shown in ref.~\cite{Cartier:1999vk} that the ratio 
1184: $\dot{\phi}_f/H_f$ is required to lie in the range 
1185: $2 \le \dot{\phi}_f/H_f \le 3$ for the successful graceful exit in the presence 
1186: of other forms of higher-order $\alpha'$ correction.  Therefore the 
1187: spectral tilt lies in the range 
1188: %
1189: \begin{eqnarray} 
1190: 2 \le n_{\cal{R}}-1 \le 3\,,
1191: \label{tilt2}
1192: \end{eqnarray}
1193: %
1194: which is valid for large-scale modes ($|sk^2| \ll |z''/z|$).
1195: Therefore we have blue-tilted spectra as long as the correction ${\cal L}_c$
1196: dominates compared to the exponential potential.
1197: 
1198: %%%%%%%%%%
1199: \begin{figure}
1200: \begin{center}
1201: \singlefig{10cm}{s.eps}
1202: \begin{figcaption}{s}{10cm}
1203: The evolution of the frequency shift $s$
1204: for $c=-1$, $d=1$, $p=0.1$ with initial 
1205: conditions $\phi=-100$, $H=1.5 \times 10^{-3}$.  
1206: We include the quantum correction ${\cal L}_c$ from the beginning.  
1207: The shift $s$ is approximately constant (and negative) 
1208: during the string phase, which is followed by the 
1209: stage of decreasing curvature with positive $s$.  {\bf Inset}: The 
1210: evolution of $s$ in the case where the quantum correction is taken into 
1211: account only for $\varphi~\lsim~1$.  Note that $s$ rapidly changes sign 
1212: around the graceful exit.
1213: \end{figcaption}
1214: \end{center}
1215: \end{figure}
1216: %%%%%%%%%%
1217: 
1218: 
1219: %%%%%%%%%%%%%%%%%%%%%%%%%%
1220:  \subsubsection{{\rm Case (ii)} : $|V(\phi)| \gg |{\cal L}_c|$
1221:  {\rm but} $|V(\phi)|<|{\cal L}_c|$ {\rm for} $\varphi \sim 0$}
1222: %%%%%%%%%%%%%%%%%%%%%%%%%%
1223: 
1224: When the correction term ${\cal L}_c$ becomes important only around the 
1225: graceful exit ($\phi \sim 0$), the spectra of density perturbations 
1226: generated before the bounce are mainly determined by the exponential 
1227: potential.  In this case the evolution of the background can be 
1228: characterized by eq.~(\ref{ekysolution2}).  Then the quantity $Q$ in 
1229: eq.~(\ref{Qs2}) evolves as 
1230: %
1231: \begin{eqnarray}
1232: Q=\frac{2\dot{\phi}^2e^{-\phi}}{(2H_S-\dot{\phi})^2}
1233:  \propto (-\eta_S )^{2\sqrt{p}/(1-p)}\,.
1234: \label{Qstring}
1235: \end{eqnarray}
1236: %
1237: Therefore we find  
1238: %
1239: \begin{eqnarray} 
1240: z \propto (-\eta_S)^{\gamma}\,,~~~{\rm with}~~~
1241: \gamma=\frac{p}{1-p}\,,
1242: \label{gamma_po}
1243: \end{eqnarray}
1244: %
1245: and the spectral tilt for the curvature perturbation is 
1246: %
1247: \begin{eqnarray} 
1248: n_{\cal{R}}-1= \cases{
1249:          \frac{2}{1-p} & (for $0<p<1/3$)\,, \cr
1250:           \frac{4-6p}{1-p} & (for $1/3<p<1$)\,. \cr
1251:           }
1252: \label{tilt_po}
1253: \end{eqnarray}
1254: %
1255: This coincides with the result in the Einstein frame
1256: performed in 
1257: ref.~\cite{Lyth:2001pf,Brandenberger:2001bs,Hwang:2001ga,Tsujikawa:2001ad}.
1258: For very slow contraction with a negative Ekpyrotic potential
1259: ($p \ll 1$), one has blue tilted spectra with $n_{\cal{R}}-1=2$.  
1260: Since $\gamma$ is less than $1/2$ for $p<1/3$ (i.e., negative potential),
1261: curvature perturbations are not enhanced in the large-scale limit even in 
1262: the presence of the correction ${\cal L}_c$ around the graceful exit.  
1263: The simplest PBB scenario with zero potential corresponds to 
1264: $p=1/3$ and $\gamma=1/2$, in which case one has $n_{\cal{R}}-1=3$.
1265: In this case curvature perturbations evolve as ${\cal R}_k \propto
1266: {\rm ln}(-\eta)$ as found from eq.~(\ref{Rk}) with eq.~(\ref{z}).
1267: 
1268: We have solved the evolution equation (\ref{peinstein}) for 
1269: the cosmological fluctuations numerically. 
1270: Experience from studying fluctuations
1271: in inflationary cosmology teaches us that following the evolution
1272: equation for $\Psi$ instead of for the gravitational potential $\Phi$
1273: is less likely to be effected by numerical noise.  Since in a
1274: contracting Universe, one of the
1275: modes of $\Phi$ increases much more rapidly than the dominant mode of
1276: $\Psi$, we believe that it is advantageous to use $\Psi$ in our
1277: case as well.  In addition, from a more conceptual point of view, the variable 
1278: ${\cal R}$ is preferable since it is more closely related to $\Psi$ 
1279: in terms of which the action for cosmological fluctuations takes on its 
1280: canonical form. Note that $\Psi$ is also the good variable to use when 
1281: following cosmological fluctuation from inflation through reheating
1282: \cite{Finelli:1998bu}.
1283: In Fig.~\ref{speevolution} we plot the resulting evolution of the spectra of 
1284: curvature perturbations for several different frequencies.  The 
1285: higher-order correction ${\cal L}_c$ is included only when two branes 
1286: approach sufficiently, i.e., $\varphi~\lsim~1$.  We find that large-scale 
1287: modes ($k \ll 1$) are not enhanced as predicted by  eq.~(\ref{Rk}).
1288: In contrast, small-scale 
1289: curvature perturbations exhibit rapid growth around the graceful exit 
1290: ($\varphi \sim 0$).
1291: 
1292: There are two reasons for this instability. 
1293: The first is the fact that the frequency shift $s$ 
1294: becomes negative for a short period where the higher-curvature effect is 
1295: dominant (see the inset of Fig.~\ref{s}). As is obvious from 
1296: (\ref{Psi}),  an exponential instability
1297: for $\Psi$ is induced by negative $s$, which is stronger for larger $k$.  
1298: We expect this instability will become important for modes with $\vert s 
1299: \vert_{\rm max}\,k^2~\gsim~1$, where $\vert s \vert_{\rm max}$ is the maximal 
1300: negative value of the function $s$.  {}From the inset of 
1301: Fig.~\ref{s} the maximal absolute value of $s$ during the negative 
1302: branch is about $\vert s \vert_{\rm max} \sim 10^2$.  Hence, the 
1303: $s$-instability is expected to be important only for modes with 
1304: $k~\gsim~10^{-1}$.  By comparing runs with $s$ given by the general 
1305: formula and 
1306: runs with $s = 1$, we were able to determine numerically that the actual 
1307: cutoff value of $k$ below which the instability due to the $s$-term is 
1308: negligible is $k \sim 10^{-2}$.  Thus, we conclude that the main source of 
1309: the short wavelength instability of the fluctuation modes around the bounce 
1310: must be a second one, namely the nontrivial nature of the bouncing 
1311: background and its result on the quantity $z''/z$.
1312:  
1313: After the transition to the 
1314: expanding Universe, the curvature perturbation is nearly
1315: conserved as found in Fig.~\ref{speevolution}.
1316: 
1317: %%%%%%%%%%
1318: \begin{figure}
1319: \begin{center}
1320: \singlefig{11cm}{speevolution.eps}
1321: \begin{figcaption}{speevolution}{11cm}
1322: The evolution of the spectra of curvature perturbations, $P_{\cal R}$,
1323: for $c=-1$, $d=1$, $p=0.1$, $C_1=1.0$ and $C_2= -1.0 
1324: \times 10^{-3}$.  The initial conditions are chosen to be $\phi=-15$, 
1325: $H=1.5 \times 10^{-3}$.  We include the 
1326: higher-order correction ${\cal L}_c$ only for $\varphi~\lsim~1$.  The 
1327: curvature perturbation does not exhibit significant variation during the 
1328: contracting phase.  However small-scale modes are enhanced 
1329: around the graceful exit.
1330: \end{figcaption}
1331: \end{center}
1332: \end{figure}
1333: %%%%%%%%%%
1334: 
1335: We show in Fig.~\ref{spectrum} the spectra of curvature perturbations
1336: for $p=10^{-3}$ in the case where the correction ${\cal L}_c$ is included
1337: only for $\varphi~\lsim~1$.  We find that the numerical value of the 
1338: spectral tilt is $n_{\cal R}-1 \sim 2$ for $k~\lsim~10^{-4}$, 
1339: which coincides with the analytic estimation (\ref{tilt_po}).
1340: However this estimate is no longer valid for small-scale modes due to the 
1341: negative frequency shift and the instability around the graceful exit.  The 
1342: spectra are highly blue-tilted for $k~\gsim~10^{-4}$ as found in 
1343: Fig.~\ref{spectrum}.  This growth of small-scale fluctuations obviously 
1344: works as the gravitational back-reaction to the background evolution.  
1345: Although we did not consider the effect of the back-reaction here, it is 
1346: certainly of interest to investigate how the background evolution is 
1347: modified around the graceful exit. We have performed the simulations
1348: with various choices of time steps to make sure that the effects we
1349: find are not numerical artifacts. The spectra obtained are independent
1350: of the specific value of the time step.
1351: 
1352: %%%%%%%%%%
1353: \begin{figure}
1354: \begin{center}
1355: \singlefig{10cm}{spectrum.eps}
1356: \begin{figcaption}{spectrum}{10cm}
1357: The final spectra of the curvature perturbation, $P_{\cal R}$, and of
1358: the gravitational potential, $P_{\Phi}$,  in a simulation with $c=-1$, $d=1$, 
1359: $p=10^{-3}$, $C_1=1.0$ and $C_2= -1.0 \times 10^{-3}$ 
1360: with same initial conditions as in Fig.~\ref{speevolution}. 
1361: The superscript ``b'' denotes the quantities {\it before} the bounce 
1362: (at $t_S=300$), while ``a'' indicates the quantities {\it after} 
1363: the bounce (at $t_S=1000$).  
1364: We have included the quantum correction ${\cal L}_c$ for 
1365: values $\varphi~\lsim~1$.  The spectral tilt is $n_{\cal{R}}-1 \sim 2$ for 
1366: $k~\lsim~10^{-4}$, which agrees with the analytic estimation of 
1367: eq.~(\ref{tilt_po}).  For the modes $k > 10^{-4}$, the spectra are highly 
1368: blue-tilted, due to an instability of small-scale modes during the graceful 
1369: exit.  The fluctuations in $\Phi$ are nearly scale-invariant on large scales 
1370: before the bounce.
1371: \end{figcaption}
1372: \end{center}
1373: \end{figure}
1374: %%%%%%%%%%
1375: 
1376: 
1377: In Fig.~\ref{spectrum} we have also plotted the induced fluctuations
1378: of $\Phi_k^E$ in the Einstein frame, determined from the results for ${\cal 
1379: R}_k^S$ in the string frame, using the relation
1380: %
1381: \begin{equation}
1382: \Phi_k^E \propto \dot{\cal R}_k^S/k^2 \,. 
1383: \label{relphi}
1384: \end{equation}
1385: %
1386: This corresponds to eq.~(\ref{modi}) in Appendix B, which follows from the 
1387: relation (\ref{Phi}) in the Einstein frame.  Note that this relation is 
1388: valid in the absence of higher curvature corrections to the Lagrangian, and 
1389: will therefore be good at times long before and long after the bounce.  For 
1390: the negative Ekpyrotic potential ($0<p<1/3$), one has $0<\gamma<1/2$ from 
1391: eq.~(\ref{gamma_po}), in which case the second term in eq.~(\ref{dotPsi}) 
1392: completely vanishes.  In this case we have the relation (\ref{Phi3}), 
1393: namely 
1394: %
1395: \begin{eqnarray} 
1396: \Phi_k^E \propto H_{\nu-1}^{(1, 2)}/k \,.
1397: \label{Phi2}
1398: \end{eqnarray}
1399: %
1400: Note that $H_{\nu-1}^{(1, 2)}$ can be written by the 
1401: sum of two terms which are proportional to $(k|\eta_S|)^{\nu-1}$ and 
1402: $(k|\eta_S|)^{-\nu+1}$.  
1403: 
1404: Since $0<\nu<1/2$ for $0<\gamma<1/2$ (i.e., $0<p<1/3$), the term 
1405: proportional to $(k|\eta_S|)^{\nu-1}$ is the growing mode during the 
1406: contracting phase on large scales.  Therefore the spectrum of $\Phi_k^E$ 
1407: before the bounce can be estimated as 
1408: %
1409: \begin{eqnarray} 
1410: P_{\Phi}^{{\rm b}} \propto k^{2\nu -1} \propto k^{-2\gamma} \propto 
1411: k^{n_{\Phi}-1} \,,
1412: \label{PPhib}
1413: \end{eqnarray}
1414: %
1415: from which we have
1416: %
1417: \begin{eqnarray} 
1418: n_{\Phi}-1=-\frac{2p}{1-p} \,.
1419: \label{Phispe}
1420: \end{eqnarray}
1421: %
1422: Then we have a scale-invariant spectrum before the bounce
1423: for $p \sim 0$, as first pointed out in \cite{Khoury:2001zk}.  
1424: This agrees with our numerical result shown in 
1425: Fig.~\ref{spectrum}.
1426: 
1427: The term proportional to 
1428: $(k|\eta_S|)^{\nu-1}$, however, decays after the graceful exit as long as 
1429: $\nu<1$.
1430: The dominant mode in $\Phi_k^E$ long after the bounce 
1431: is described by the term $(k|\eta|)^{-\nu+1}$, in which case the spectrum 
1432: of $\Phi$ is written as 
1433: %
1434: \begin{eqnarray} 
1435: P_{\Phi}^{{\rm a}} \propto k^{3-2\nu} \,.
1436: \label{PPhia}
1437: \end{eqnarray}
1438: %
1439: {}From Fig.~\ref{spectrum} we find  that the spectrum of $\Phi$ 
1440: is blue-tilted with $n_{\Phi} \sim 3$ for $k~\gsim~10^{-10}$ 
1441: (small-scale modes for $k~\gsim~10^{-2}$ exhibit larger blue tilt with 
1442: $n_{\Phi}>3$).  This corresponds to the value $\nu \sim 1/2$ in 
1443: eq.~(\ref{PPhia}) after the bounce.
1444: 
1445: Our numerical calculations show that large-scale modes 
1446: with $k~\lsim~10^{-10}$ do not exhibit such a blue spectrum.  
1447: This can be understood that the term proportional to
1448: $(k|\eta_S|)^{\nu-1}$ which is dominant in the contracting phase
1449: does not become smaller than the one proportional to
1450: $(k|\eta_S|)^{-\nu+1}$ in the expanding branch for very small $k$, unless we 
1451: evolve the fluctuations until long after the bounce.  However it is 
1452: rather difficult to follow such large amount of time numerically.  In 
1453: addition the second term in eq.~(\ref{dotPsi}) is not numerically 
1454: negligible relative to the first term for these large-scale modes due to 
1455: the modification of the equation of state after the bounce [when 
1456: $\gamma>1/2$ the second term in eq.~(\ref{dotPsi}) is nonvanishing as found 
1457: by eq.~(\ref{re})].  Nevertheless, we expect that the term proportional to 
1458: $(k|\eta_S|)^{-\nu+1}$ in the first term in eq.~(\ref{dotPsi}) eventually 
1459: dominates long after the bounce, in which case the spectrum is given by 
1460: eq.~(\ref{PPhia}).  Therefore the final spectrum of $\Phi$ is not generally 
1461: scale-invariant.  The spectral index is dependent on the evolution of the 
1462: background after the bounce (i.e., $\gamma$).  In this sense including 
1463: radiation is necessary in order to evaluate the spectrum of $\Phi$ in 
1464: realistic cases where the solution connects to our Friedmann branch.
1465: 
1466: {}From Fig.~\ref{spectrdecay} we find that 
1467: the amplitude of $\Phi$ decreases after the bounce, thus showing
1468: that the dominant pre-bounce mode of $\Phi$ couples exclusively
1469: to the decaying mode of $\Phi$ after the bounce, as derived in
1470: \cite{Brandenberger:2001bs} using matching conditions on a
1471: constant scalar field hypersurface.
1472: 
1473: %%%%%%%%%%
1474: \begin{figure}
1475: \begin{center}
1476: \singlefig{10cm}{sdecay.eps}
1477: \begin{figcaption}{spectrdecay}{10cm}
1478: The evolution of the curvature perturbation, $P_{\cal R}$, and of
1479: the gravitational potential, $P_{\Phi}$ for the fluctuation mode
1480: corresponding to $k = 10^{-9}$. 
1481: The model parameters and initial conditions are the same as
1482: in Fig.~\ref{spectrum}.
1483: The amplitude of the gravitational potential near the bounce when 
1484: the higher derivative terms dominate cannot be trusted since 
1485: $\Phi_k^E$ is computed from ${\cal R}_k^S$ via 
1486: the equation (\ref{relphi}) which is only valid 
1487: in the absence of such higher derivative terms. 
1488: As follows from this plot, the dominant growing mode of $\Phi_k^E$ 
1489: during the period of contraction couples
1490: only to the post-bounce decaying mode. At the time of the bounce,
1491: the curve for $\Phi$ is dominated by numerical noise. However, since
1492: $\Phi$ is computed at each time separately from the value of ${\cal R}$
1493: at that time, this does not introduce numerical errors in the late
1494: time values of $\Phi$.
1495: \end{figcaption}
1496: \end{center}
1497: \end{figure}
1498: %%%%%%%%%%
1499: 
1500: Eqs.~(\ref{PR}) and  (\ref{gamma_po}) indicate that a scale-invariant 
1501: spectrum may be 
1502: obtained for $p=2/3$ for the modes which are enhanced during the bouncing 
1503: phase \cite{Finelli:2001sr}.  In order to obtain such a spectrum, the 
1504: exponential potential (positive in this case) is required to dominate the 
1505: higher-order term except around the graceful exit.  However we have found 
1506: that, for some likely initial conditions, the field $\varphi$ bounces back 
1507: toward larger $\varphi$ before the higher-order correction begins to work.  
1508: In language appropriate to Ekpyrotic cosmology, this means that the branes 
1509: never collide.  If the higher-order term always dominates compared to the 
1510: positive exponential potential, we have blue-tilted spectra (\ref{tilt1}).
1511: 
1512: 
1513: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1514:  \section{Inclusion of higher-order corrections in the Einstein
1515:  frame \\
1516:  --modulus driven case} 
1517: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
1518: 
1519: In this section we consider adding higher derivative terms
1520: defined in the Einstein frame. We add a Gauss-Bonnet term
1521: proportional to $R_{\rm GB}^2$ multiplied by a function of
1522: the modulus field $\varphi$ to the action. Such a term arises
1523: as the one loop correction in the context of orbifold compactifications
1524: of the heterotic superstring \cite{Antoniadis:1992rq}. Since the
1525: initial version of Ekpyrotic cosmology \cite{Khoury:2001wf} is
1526: based on an orbifold compactification of a theory dual to heterotic
1527: superstring theory, the correction terms used in this section are
1528: well motivated in the context of the scenario of \cite{Khoury:2001wf}.
1529: Indeed, it was found in Ref.~\cite{Antoniadis:1993jc} (see also
1530: \cite{Rizos:1993rt}-\cite{Alekseev:eh})
1531: that the inclusion of the Gauss-Bonnet 
1532: term coupled to a modulus field in the Einstein frame leads to 
1533: the possibility of obtaining nonsingular solutions. In the work of
1534: \cite{Antoniadis:1993jc}, the potential for the modulus field was
1535: taken to vanish. In this section we will include an 
1536: exponential potential.\footnote{The authors of 
1537: ref.~\cite{Alekseev:eh} analyzed nonsingular cosmological
1538: solutions in the presence of some positive potentials (not the  
1539: exponential potential).}
1540: More specifically, the correction Lagrangian we consider here
1541: corresponds to [in the notation of (\ref{lag}) and (\ref{lagalpha})]  
1542: $f=R$, $\omega=1$, $c=-1$, $d=0$,
1543: $\xi(\varphi)={\rm ln}[2e^{\varphi}\eta^4(ie^{\varphi})]$ with 
1544: $\eta(ie^{\varphi})$ being the Dedekind $\eta$-function 
1545: \cite{Antoniadis:1993jc}.  
1546: Here $\xi(\varphi)$ is approximately given by 
1547: %%%%%%%%%%%%%
1548: \begin{eqnarray}
1549: \xi(\varphi) \simeq -\frac{\pi}{3}
1550: \left(e^{\varphi}+e^{-\varphi}\right)\,.
1551: \label{xiapprox}
1552: \end{eqnarray}
1553: %%%%%%%%%%%%%
1554: The sign of $\lambda$ is chosen to be positive, which is 
1555: different from the one discussed in the previous 
1556: section. Note also that even though
1557: the coefficient $\xi(\varphi)$ becomes large at large brane
1558: separation (large negative values of the dilaton in the case
1559: of PBB cosmology), this increase is outweighed by the falloff
1560: of the curvature invariant, as in the case of the model considered
1561: in the previous section. Thus, in Ekpyrotic cosmology
1562: the correction terms are expected  to
1563: become important only in the high curvature region.  
1564: 
1565: Let us analyze the one-field system of a modulus $\varphi$, 
1566: keeping the dilaton fixed. We will consider solutions starting
1567: in an asymptotically flat region and beginning in the expanding
1568: branch. We have not found solutions which begin in a contracting
1569: phase and undergo a successful bounce. However, note that in
1570: the original Ekpyrotic scenario of \cite{Khoury:2001wf}, the scale
1571: factor on the orbifold fixed plane corresponding to our 
1572: four-dimensional space-time corresponds to an initially asymptotically 
1573: flat region, and is always expanding. Thus, the solutions found here might be applicable
1574: to a version of Ekpyrotic cosmology formulated entirely in terms of
1575: physics on the orbifold fixed plane. 
1576: 
1577: %%%%%%%%%%%%%%%%%%%%%%%%
1578:  \subsection{Background evolution} 
1579: %%%%%%%%%%%%%%%%%%%%%%%%
1580: 
1581: As was discussed in \cite{Antoniadis:1993jc}, when $V_E=0$ the PBB 
1582: singularity can be avoided for positive values of $\lambda$ when the
1583: $\alpha'$ corrections introduced above are taken into account.  
1584: The sign of $\lambda$ is crucial 
1585: for the existence of nonsingular cosmological solutions.
1586: For negative values of $\lambda$, the $\alpha'$ corrections do not help to 
1587: lead to a successful graceful exit, as was analyzed in 
1588: ref.~\cite{Toporensky:2002ta}. 
1589: 
1590: In the absence of the Ekpyrotic potential, 
1591: the background evolution for $t_E<0$ is given by \cite{Kawai:1998bn} 
1592: %%%%%%%%%%%%%
1593: \begin{eqnarray}
1594: a_E \simeq a_0,~~~H_E=\frac{H_0}{t_E^2},~~~
1595: \dot{\varphi}=\frac{5}{t_E},
1596: \label{backevo}
1597: \end{eqnarray}
1598: %%%%%%%%%%%%%
1599: where $a_0$ and $H_0$ ($>0$) are constants.
1600: The Gauss-Bonnet term leads to a  
1601: violation of the null energy condition ($\rho_E+p_E<0$)
1602: at sufficiently large curvatures and thus enables a graceful exit.
1603: If we start in an expanding branch (contrary to the
1604: spirit of PBB and Ekpyrotic cosmology), this leads to 
1605: a super-inflationary solution ($\dot{H}_E>0$) 
1606: until a ``graceful exit'' (see Fig.~\ref{modulus}).
1607: The Universe is initially 
1608: expanding very slowly  with a nearly constant scale factor.
1609: After the Hubble parameter reaches its peak value 
1610: $H_E=H_{\rm max}$, the system connects to a Friedmann-like Universe with 
1611: $H_E \simeq 1/(3t_E)$, $a_E \propto t_E^{1/3}$, and 
1612: $\varphi \propto -{\rm ln} t_E$.
1613:  
1614: If the Ekpyrotic potential is present, the situation is 
1615: quite different.  
1616: We have adopted the potential (\ref{einpoten}) 
1617: for $\varphi>0$ and $V_E=0$ for $\varphi<0$. Once again,
1618: we start in an expanding phase. Initially, the potential term
1619: is not important and the Universe evolves in a super-inflationary
1620: trajectory until a graceful exit after which the Hubble expansion
1621: rate begins to decrease. When $p<1/3$, corresponding to the case of a 
1622: negative exponential potential, then as $\varphi \to 0$, the potential becomes
1623: important and leads to a change in sign of $H_E$.
1624: We find that the system enters a stage of slow contraction 
1625: [see the case (ii) of Fig.~\ref{modulus}]. 
1626: Note that in Fig.~\ref{modulus} $\dot{H}_E$ changes the sign twice.
1627: After the negative Hubble peak, the Hubble rate begins to grow toward 
1628: $H_E \to -0$ 
1629: without changing sign. Then the system enters a very slow contracting phase
1630: with a nearly constant scale factor. In this stage the field $\varphi$ 
1631: evolves rapidly toward large negative values. 
1632: In the presence of negative exponential potential ($p<1/3$) 
1633: we have found that 
1634: the contracting stage eventually appears even when $V_0$ is small.
1635:  
1636: %%%%%%%%%%
1637: \begin{figure}
1638: \begin{center}
1639: \singlefig{15cm}{modulus.eps}
1640: \begin{figcaption}{modulus}{15cm}
1641: The evolution of $H_E$, $a_E$, $\varphi$, and 
1642: $\rho_E+p_E$ in the modulus-driven case 
1643: with $c=-1$, $d=0$.
1644: We choose initial conditions $\varphi=20$ and
1645: $H=5.087 \times 10^{-4}$.
1646: Each case corresponds to (i) $V_0=0$ with $p=0.1$,
1647: (ii) $V_0=0.01p(1-3p)$ with $p=0.1$,
1648: (iii) $V_0=p(1-3p)$ with $p=0.5$.
1649: \end{figcaption}
1650: \end{center}
1651: \end{figure}
1652: %%%%%%%%%%
1653: 
1654: When $p>1/3$ there exists a positive potential barrier as the field $\varphi$
1655: approaches zero. The case (iii) in Fig.~\ref{modulus} corresponds to 
1656: $p=1/2$ with $V_0=p(1-3p)$. The effect of the positive potential
1657: is important around $\varphi \sim 0$, which works to return the field back 
1658: toward larger $\varphi$. After the graceful exit the Hubble rate is always positive
1659: with slowly changing $\varphi$. The scale factor evolves as a power-law    
1660: ($a \propto t^p$) due to a positive exponential potential.
1661: The $p>1/3$ case provides us with reasonable nonsingular 
1662: cosmological solutions.
1663: Nevertheless we need to caution that  these nonsingular solutions
1664: are different from the bouncing ones where the contraction of 
1665: the universe occurs before the graceful exit.
1666: 
1667: One may argue that the bouncing trajectories may be found 
1668: by including the correction ${\cal L}_c$ only around $\varphi \sim 0$.
1669: However we have numerically found that this is not the case.
1670: The super-inflationary evolution characterized by eq.~(\ref{backevo}) is   
1671: typically required for the construction of nonsingular solutions 
1672: in the present scenario. 
1673: 
1674: %%%%%%%%%%%%%%%%%%%%%%%
1675:  \subsection{Density perturbations}
1676: %%%%%%%%%%%%%%%%%%%%%%%
1677: 
1678: When the Gauss-Bonnet term is dominant relative to the Ekpyrotic potential, 
1679: the spectra of density perturbations can be analyzed as in the case of
1680: the zero potential ($p=0$ or $p=1/3$).
1681: In this case the evolution of the background during the phase of 
1682: modulus-driven 
1683: inflation is given by eq.~(\ref{backevo}), thereby leading to 
1684: $\dot{\xi}(\varphi) \simeq
1685: -(\pi/3)\dot{\varphi} e^{\varphi} \propto (-t_E)^4$. 
1686: Making use of this relation together with 
1687: eq.~(\ref{Qs2}), we find the evolution of $Q$ and $z$ as 
1688: %%%%%%%%%%%%%
1689: \begin{eqnarray}
1690: Q \propto (-t_E)^2,~~~~z \propto (-t_E) \propto (-\eta_E)\,.
1691: \label{Qz}
1692: \end{eqnarray}
1693: %%%%%%%%%%%%%
1694: This means that $\gamma=1$ in eq.~(\ref{z}), in which case
1695: curvature perturbations are enhanced on super-Hubble scales during 
1696: super-inflation (${\cal R}_k \propto (-\eta_E)^{-1}$)
1697: due to the growth of the second term in eq.~(\ref{Rk})
1698: \cite{Kawai:1999pw}.  
1699: We show in Fig.~\ref{modu_evo} the evolution of 
1700: curvature perturbations in the case of zero potential ($p=1/3$) for two 
1701: different modes ($k=10^{-3}$ and $k=10^{-1}$). 
1702: We find that curvature perturbations are amplified before the graceful exit.
1703: 
1704: In order to obtain the spectral tilt of density perturbations, we have to 
1705: caution that the function $s$ defined by eq.~(\ref{Qs}) is a time-varying 
1706: function and is proportional to $(-t_E)$.
1707: Therefore the formula (\ref{ind}) can not be directly applied. 
1708: Instead one is required to consider the evolution 
1709: equation for curvature perturbations: 
1710: %%%%%%%%%%%%%
1711: \begin{eqnarray}
1712: \ddot{{\cal R}}_k+\frac{2}{t_E}\dot{{\cal R}}_k-\alpha
1713: \frac{k^2}{a_0^2} t_E {\cal R}_k=0\,,
1714: \label{dR_k}
1715: \end{eqnarray}
1716: %%%%%%%%%%%%%
1717: where $\alpha~(>0)$ is a constant that depends on $H_0$ 
1718: in eq.~(\ref{backevo}).  
1719: The solution of this equation is written in terms of the Bessel functions 
1720: %%%%%%%%%%%%%
1721: \begin{eqnarray}
1722: {\cal R}_k=(-t_E)^{-1/2} \left[c_1 J_{-1/3}(x)+c_2J_{1/3}(x)\right]\,,
1723: \label{Bessel}
1724: \end{eqnarray}
1725: %%%%%%%%%%%%%
1726: where $x \equiv \frac23 \sqrt{\alpha} \frac{k}{a_0}(-t_E)^{3/2}$.  
1727: Notice that this solution asymptotically approaches the 
1728: Minkowski vacuum for $x \to \infty$.  
1729: Since $J_{\pm 1/3} (x) \propto k^{\pm 
1730: 1/3}$ in the $x \to 0$ limit, the spectrum of large-scale curvature 
1731: perturbation is proportional to  
1732: $P_{{\cal R}} \propto k^{7/3}$.
1733: Therefore the spectral index is  
1734: %%%%%%%%%%%%%
1735: \begin{eqnarray}
1736: n_{\cal R}-1=\frac73\,,
1737: \label{PowerM}
1738: \end{eqnarray}
1739: %%%%%%%%%%%%%
1740: which is a blue-tilted spectrum.
1741: 
1742: In the absence of the Ekpyrotic potential, the evolution of the background 
1743: in the asymptotic future is given by $\varphi \sim \sqrt{3/2}\,{\rm ln} t_E$,
1744: $H_E \propto 1/(3t_E)$, and $a \propto t_E^{1/3}$.  Therefore one has 
1745: $z \propto t_E^{1/3} \propto \eta_E^{1/2}$ in eq.~(\ref{Rk}), 
1746: in which case curvature perturbations exhibit logarithmic growth,
1747: %%%%%%%%%%%%%
1748: \begin{eqnarray}
1749: {\cal R} \propto {\rm ln}~\eta_E \,.
1750: \label{log}
1751: \end{eqnarray}
1752: %%%%%%%%%%%%%
1753: This indicates that the second term in eq.~(\ref{Rk}), which we call 
1754: ``D-mode'', dominates even after the graceful 
1755: exit.  In the case where the D- mode decays after the graceful exit, 
1756: the surviving spectra observed in an expanding Universe should correspond 
1757: to the first term in eq.~(\ref{Rk}) [``C-mode''].  In the present model, 
1758: however, the D-mode survives in an expanding branch.  Therefore the 
1759: spectrum of the curvature perturbation during super-inflation can be 
1760: preserved even after the graceful exit.  In fact, the numerical value of the 
1761: final spectral tilts of ${\cal R}$ are found to be $n_{\cal R}-1 \sim 2.3$ 
1762: for the modes $k \ll 1$ (see Fig.~\ref{modu_spectra}). 
1763: This agrees well with the analytic result (\ref{PowerM}).
1764: 
1765: 
1766: %%%%%%%%%%
1767: \begin{figure}
1768: \begin{center}
1769: \singlefig{10cm}{modu_evo.eps}
1770: \begin{figcaption}{modu_evo}{10cm}
1771: The evolution of the spectra of curvature perturbations in the modulus-driven 
1772: case ($c=-1$ and $d=0$) for $(p, k)=(1/3, 10^{-1}), (1/3, 10^{-3}), (1/2, 10^{-1}),
1773: (1/2, 10^{-3})$.  
1774: We choose initial conditions $\varphi=23.888$ and
1775: $H=4.158 \times 10^{-4}$.
1776: Note that the $p=1/2$ case corresponds to the positive
1777: exponential potential while the zero-potential corresponds to $p=1/3$.  Around 
1778: the graceful exit curvature perturbations exhibit rapid growth especially 
1779: when the potential is positive $(p>1/3)$.
1780: \end{figcaption}
1781: \end{center}
1782: \end{figure}
1783: %%%%%%%%%%
1784: 
1785: %%%%%%%%%%
1786: \begin{figure}
1787: \begin{center}
1788: \singlefig{10cm}{modu_spectra.eps}
1789: \begin{figcaption}{modu_spectra}{10cm}
1790: The final spectra of curvature perturbations
1791: in the modulus-driven case ($c=-1$ and $d=0$) for $p=1/3$ and $p=1/2$.
1792: The initial conditions are the same as in Fig.~\ref{modu_evo}.
1793: When the positive potential is present ($p>1/3$), the amplitude of the spectrum is
1794: larger than in the case of the zero-potential. 
1795: We numerically find that the spectral tilt is 
1796: $n_{\cal R}-1 \sim 2.3$ for $k \ll 1$, 
1797: which agrees well with the analytic estimation, $n_{\cal R}-1=7/3$.
1798: \end{figcaption}
1799: \end{center}
1800: \end{figure}
1801: %%%%%%%%%%
1802: 
1803: When a positive Ekpyrotic potential is present ($p>1/3$), the dynamics of the 
1804: system is more unstable around the graceful exit.  This leads to the 
1805: violent growth of curvature perturbations when the field bounces back due to 
1806: the potential barrier.  This threatens the viability of the cosmological 
1807: perturbation theory around the graceful exit.  Nevertheless the perturbations
1808: are not singular as long as the background is smoothly joined to the 
1809: expanding branch.  We have numerically evaluated the power spectra of 
1810: ${\cal R}$ for the modes which left the horizon during super-inflation. 
1811: Although the amplitude is larger compared to the case of the zero-potential 
1812: ($p=1/3$), the final spectral tilts are similar for large-scale modes
1813: ($k \ll 1$), see Fig.~\ref{modu_spectra}.  Again the final spectra are 
1814: found to be blue-tilted.  We should also mention that the frequency 
1815: shift $s$ becomes negative for the Hubble rate which is larger than unity 
1816: around the graceful exit \cite{Kawai:1999pw}. 
1817: In this case the small-scale modes show
1818: exponential instability as we pointed out in the dilaton-driven case.
1819: The negative Ekpyrotic potential ($p<1/3$) is not worth 
1820: studying, since this case does not connect to the expanding branch as 
1821: analyzed in the previous subsection.
1822: 
1823: Finally, we should mention that we have neglected the effect of radiation 
1824: in all our analysis.  
1825: However this is expected to appear at some moment of time.
1826: This can also alter the final spectra of curvature perturbations due to the 
1827: dominance of the C-mode in eq.~(\ref{Rk}).
1828: We leave to future work about these realistic situations.
1829: 
1830: 
1831: %%%%%%%%%%%%%%%%%%%%%%%
1832:  \section{Discussions and Open Issues}
1833: %%%%%%%%%%%%%%%%%%%%%%%
1834: 
1835: We have studied the effects of higher derivative terms in the
1836: joint gravitational and matter action for theories motivated
1837: by Pre-Big-Bang and Ekpyrotic cosmology with a single scalar
1838: matter field with an exponential potential. Applied to
1839: PBB cosmology, our model corresponds to a theory with an
1840: exponential potential for the dilaton. In the language of
1841: the initial version of Ekpyrotic cosmology \cite{Khoury:2001wf}, 
1842: our scalar field is the modulus field
1843: corresponding to the separation of the bulk brane from our
1844: orbifold fixed plane, in the second version \cite{Khoury:2001bz}
1845: and in its cyclic version \cite{Steinhardt:2001vw,Steinhardt:2001st}
1846: the field is the radius of the extra spatial dimension. The
1847: higher derivative terms introduced are the leading string and
1848: quantum corrections to the low energy effective action of
1849: string theory.
1850: 
1851: When applying the correction terms in the string frame, and
1852: for suitable choices of the coefficients of the higher-order 
1853: corrections we find
1854: nonsingular cosmological solutions which in the
1855: Einstein frame correspond to bouncing Universes. We thus find that
1856: higher derivative terms can smooth out the singularities in PBB
1857: and Ekpyrotic cosmology and lead to a graceful exit (in the
1858: language of PBB cosmology) or a nonsingular bounce (in the
1859: language of Ekpyrotic cosmology). We have thus generalized the
1860: results of \cite{Gasperini:1996fu} to models with exponential
1861: scalar field potentials.
1862: 
1863: We have studied the evolution of fluctuations in our nonsingular
1864: bouncing cosmologies. This analysis is not plagued by the
1865: matching ambiguities inherent to analyses where the
1866: contracting and expanding cosmologies are matched across
1867: a singular space-time surface. For all potentials with $0 < p \ll 1$ we 
1868: find a blue spectrum of curvature fluctuations. The precise
1869: spectral index depends, as expected, on whether the higher
1870: derivative correction terms are important at times when the
1871: scales on which we compute the fluctuation spectrum exit the Hubble
1872: radius during the phase of contraction. 
1873: 
1874: If the higher derivative terms are not dominant when the scales
1875: exit the Hubble radius, the index of the spectrum agrees with what
1876: is obtained by applying the general relativistic matching conditions
1877: on a uniform density hypersurface 
1878: \cite{Brandenberger:2001bs,Hwang:2001ga,Tsujikawa:2001ad}. 
1879: The only difference is an instability of small-scale fluctuation modes during
1880: the bounce (see also \cite{Kawai:1999pw}) which leads to a further
1881: steepening of the spectrum. Our result implies that the growing
1882: mode of $\Phi$ during the contracting phase which is 
1883: scale-invariant for $0 < p \ll 1$ is effectively uncoupled with the
1884: dominant constant mode of $\Phi$ in the expanding phase, a result
1885: obtained in the context of matching conditions in \cite{BGGMV} (in
1886: the case of PBB cosmology) and in 
1887: \cite{Brandenberger:2001bs,Hwang:2001ga} for the Ekpyrotic scenario.
1888: If the higher derivative terms dominate when scales of interest
1889: exit the Hubble radius, then the spectrum is blue with a slope
1890: of $n_{\cal R}-1 \simeq 2.26$. Note that our result implies that it is
1891: the curvature fluctuation ${\cal R}$ (more precisely the variable
1892: ${\tilde \zeta}$ originally introduced by Bardeen in \cite{Bardeen:1988hy}
1893: and used in \cite{Brandenberger:2001bs}, which equals ${\cal R}$ up to
1894: terms which are suppressed by $k^2$ for large-scale fluctuations)
1895: which is effectively conserved for large-scale perturbations across
1896: the bounce.
1897:  
1898: We have also studied nonsingular cosmological models obtained 
1899: by adding a Gauss-Bonnet term (defined in the Einstein
1900: frame) multiplied by a suitably chosen
1901: function of the single scalar matter field in the problem (a
1902: modulus field). Once again, we have included an exponential
1903: potential for the modulus field. Although we do not find bouncing
1904: cosmologies, we find interesting nonsingular cosmological solutions
1905: which begin in an asymptotically flat region, undergo super-exponential
1906: inflation followed by a graceful exit to a phase with decreasing Hubble
1907: radius. In the presence of a negative exponential potential
1908: ($0 < p < 1/3$), the solutions reach a maximal radius and 
1909: begin to contract as the field crosses $\varphi=0$.
1910:  During this period of
1911: contraction, the Hubble
1912: parameter remains finite. Such solutions might be applicable to
1913: Ekpyrotic Universe models formulated in terms of physics on the
1914: four-dimensional orbifold fixed plane corresponding to our visible
1915: space-time.  
1916: When the potential is positive 
1917: ($p>1/3$), the modulus $\varphi$ bounces back around the brane collision 
1918: toward larger $\varphi$ due to the barrier of the positive potential.  
1919: Although singularities can be avoided around $\varphi \sim 0$, this 
1920: model does 
1921: not correspond to bouncing solutions where the contraction of the Universe
1922: occurs due to the Ekpyrotic potential before the graceful exit.
1923: 
1924: We have also studied the spectrum of curvature fluctuations ${\cal R}$
1925: in this modulus-driven cosmology. 
1926: When higher-order corrections are important before the bounce
1927: (as it must be for the existence of nonsingular solutions),
1928: one has $n_{\cal R}-1 \sim 7/3$. 
1929: This result is again
1930: in agreement with what can be obtained by neglecting the graceful
1931: exit and matching two Einstein Universes at a constant density
1932: hypersurface. 
1933: 
1934: Note that we have chosen to evolve the curvature perturbation 
1935: ${\cal R}$ {on comoving hypersurfaces}, and found that the spectral 
1936: index is given by 
1937: (\ref{tilt_po}) when the Ekpyrotic potential is dominant.  Note that this
1938: spectrum in the $0 < p </ 1/3$ case comes from the C-mode in eq.~(\ref{Rk}), 
1939: which is blue-tilted for very slow contraction ($0 < p \ll 1$).  Since the 
1940: large-scale D-modes are enhanced for $1/3<p<1$ during the 
1941: contracting phase, 
1942: the spectrum of ${\cal R}$ will be scale-invariant for $p \sim 2/3$ {\it 
1943: right after} the bounce \cite{Finelli:2001sr}.  
1944: 
1945: If we follow instead the gravitational potential $\Phi$ 
1946: in the longitudinal gauge, its spectral index generated during the 
1947: collapsing phase is estimated as eq.~(\ref{Phispe}), which is 
1948: different from that of ${\cal R}$ [see eq.~(\ref{tilt_po})].
1949: When $p \sim 0$, corresponding to a very 
1950: slow contraction, the growing mode (D-mode) of $\Phi$ is approximately
1951: scale-invariant.
1952: The authors of ref.~\cite{Khoury:2001zk,Durrer:2002jn} 
1953: claimed that a scale-invariant spectrum of the dominant post-bounce
1954: mode of $\Phi$ would inherit this scale-invariant spectrum.
1955: 
1956: However, we know that when 
1957: computed at late times long after the bounce, in an expanding universe, the 
1958: spectra of ${\cal R}$ and $\Phi$ must be identical. Thus, given
1959: our results concerning the spectrum of ${\cal R}$, we know that
1960: the spectrum of $\Phi$ long after the bounce cannot be scale-invariant.
1961: Our numerical simulations show that the contribution from the
1962: pre-bounce D-mode decays 
1963: after the system enters the expanding branch, and thus shows that
1964: the pre-bounce growing mode of $\Phi$ couples exclusively to the
1965: post-bounce decaying mode. This results in 
1966: a blue-tilted spectrum of $\Phi$ when evaluated long after the bounce 
1967: (see Fig.~\ref{spectrum}). 
1968: For very large-scales with $k~\lsim~10^{-10}$, we need to solve
1969: the equation of fluctuations up to sufficient amount of time
1970: in order to find the complete decay of the D-mode relative to the 
1971: C-mode. In addition the second term in eq.~(\ref{dotPsi}) is not numerically 
1972: negligible for very small $k$ when $\gamma$ is greater than $1/2$.  
1973: Nevertheless the term proportional to $(k|\eta_S|)^{-\nu+1}$ in the first 
1974: term in eq.~(\ref{dotPsi}) eventually dominates long after the bounce 
1975: ($\eta_S \to \infty$), thereby yielding the spectrum (\ref{PPhia}).  
1976: Therefore the spectrum of $\Phi$ long after the bounce is not generally 
1977: expected to be scale-invariant, whose spectral index depends on the 
1978: evolution of the background in an expanding branch.
1979: 
1980: Since near the bounce the magnitudes of the two modes of $\Phi$ and
1981: ${\cal R}$ differ by such a large ratio, we must worry about the
1982: possibility of numerical errors. In particular, if one were to follow
1983: the evolution equation for $\Phi$, it would be difficult to ensure that
1984: numerical noise does not lead to an artificial coupling between the
1985: pre-bounce growing mode and the post-bounce dominant (constant) mode.
1986: We have checked that our results do not seem to suffer from a similar
1987: problem by repeating the simulations with different values of the 
1988: time step, $\Delta t$.
1989: We did not find any dependence of the results within the range of time steps
1990: we have chosen ($10^{-5} \le \Delta t \le 10^{-3}$).
1991: 
1992: Let us compare our findings to results which have already appeared
1993: in the literature. As mentioned repeatedly, our results concerning
1994: the spectrum of fluctuations obtained in the classes of nonsingular bouncing
1995: Universe models considered in this paper agree with the results of
1996: \cite{Brandenberger:2001bs,Hwang:2001ga,Tsujikawa:2001ad} 
1997: obtained when removing the higher derivative correction
1998: terms (thus going back to a singular background) and matching the
1999: fluctuations on a constant scalar field matching surface. The
2000: results imply that the growing mode of $\Phi$ in the contracting
2001: phase does not source the post-bounce dominant mode of $\Phi$.
2002: Our results thus indicate that the conjecture of 
2003: \cite{Khoury:2001zk,Durrer:2002jn}, namely
2004: that the growing mode of $\Phi$ in the contracting phase (which in
2005: Ekpyrotic cosmology has a scale-invariant spectrum) should {\it generically}
2006: determine the amplitude and spectrum of the dominant mode of $\Phi$
2007: in the post-bounce phase, is not valid.  As emphasized in \cite{Martin:2001ue} 
2008: and \cite{Durrer:2002jn}, in the case of a singular background the spectrum 
2009: of fluctuations in the expanding phase depends sensitively on the details 
2010: of the matching conditions used.  Since we have only used one class of ways 
2011: to smooth out the singularity, the sensitive dependence on the matching 
2012: surface might not have been completely eliminated, but might find itself 
2013: reflected in a sensitive dependence of the final spectrum on the specific 
2014: form of the correction terms in the action.  We leave the study of this 
2015: issue to future work.
2016: 
2017: Our work indicates that it is difficult to obtain a scale-invariant spectrum 
2018: of curvature fluctuations for a single field PBB or Ekpyrotic cosmology.  
2019: However, in the case of Ekpyrotic cosmology there is the intriguing fact 
2020: that the growing mode of the gravitational potential $\Phi$ during the 
2021: phase of contraction has a scale-invariant spectrum.  To obtain a 
2022: scale-invariant spectrum of $\Phi$ and thus also of the curvature 
2023: fluctuation ${\cal R}$ at late times in the expanding phase, a suggestion 
2024: \cite{Khoury:2001zk,Durrer:2002jn} was to non-trivially connect the growing 
2025: mode of $\Phi$ during the contracting phase with the constant mode in the 
2026: expanding phase.  We have shown that this does not occur in the single 
2027: field case with our choice of correction terms to the action (needed to 
2028: obtain a nonsingular bounce).
2029: 
2030: Note that there are examples where a large growth of $\Phi$ during the
2031: phase of contraction persists after the bounce (see e.g. 
2032: \cite{Finelli:2001sr,Peter:2002cn,Gordon:2002jw}). A criterion for when
2033: this occurs has been proposed recently in \cite{Gordon:2002jw}. 
2034: The condition is that the relative variation of 
2035: ${\cal R}$ over a Hubble time scale should be appreciable,  
2036: i.e. the following relation 
2037: \cite{Gordon:2002jw}
2038: \be
2039: \frac{\dot {\cal R}}{H {\cal R}} \gg  1\,,
2040: \label{inequality}
2041: \ee
2042: should hold close to or right at the bounce. We use eq.~(\ref{Rk}) and 
2043: restrict to the case of an exponential potential, 
2044: in which case one has
2045: \begin{eqnarray}
2046: \frac{\dot {\cal R}}{H {\cal R}}=
2047: \frac{D_k (1-p) \eta}{p a^2 (C_k + D_k (-\eta)^{1-2\gamma}) } = 
2048: \cases{ \frac{D_k (1-p)}{p C_k} (-\eta)^{\frac{1-3p}{1-p}} 
2049: & for $0<p<1/3$ \,,
2050: \cr \frac{1-p}{p} & for $1/3<p<1$ \,.
2051: \cr } 
2052: \label{evidence}
2053: \end{eqnarray}
2054: The marginal case $p=1/3$ (important for PBB and also for Ekpyrotic
2055: scenario, in which the potential disappears close to the bounce) should be
2056: treated separately
2057: and leads to:
2058: \be
2059: \frac{\dot {\cal R}}{H {\cal R}} = \frac{2 D_k}{C_k + D_k \ln
2060: (-k\eta)}\,.
2061: \label{evidence2}
2062: \ee 
2063: 
2064: Surprisingly enough, both the results (\ref{evidence}) and
2065: (\ref{evidence2}) indicate that $\Phi$ could {\em never} match to 
2066: ${\cal R}$ nontrivially.
2067: For $p \le 1/3$, $\dot {\cal R}/H {\cal R} \rightarrow 0$ as
2068: $\eta \rightarrow 0$. For $1/3 < p <1$,  $\dot {\cal R}/{H {\cal
2069: R}} \rightarrow (1-p)/p \sim {\cal O}(1)$ as $\eta \rightarrow 0$. This
2070: latter case is interpreted as a variation of ${\cal R}$ rather than a
2071: change induced by $\Phi$. Interestingly enough, if one takes seriously the
2072: ratio (\ref{inequality}), the singularity at the bounce (i.e., if
2073: $\eta=0$ is reached or not) does not matter in the impossibility
2074: of matching $\Phi$ to ${\cal R}$.  
2075: 
2076: Recently several authors \cite{Peter:2002cn,Gordon:2002jw} considered 
2077: models of a bouncing Universe (realized in \cite{Peter:2002cn} by introducing 
2078: matter violating the weak energy condition and in \cite{Gordon:2002jw} by 
2079: making use of spatial curvature in the background metric) in which ${\cal 
2080: R}$ grows dramatically across the bounce and there is a coupling between 
2081: the growing mode of $\Phi$ in the contracting phase and the dominant mode 
2082: of $\Phi$ in the expanding phase.  In this case, it may be possible to 
2083: obtain a scale-invariant spectrum, as already realized in 
2084: \cite{Finelli:2001sr}.
2085: 
2086: Although we have concentrated on the density perturbation in the single field 
2087: scenario, the situation can be changed by taking into account a second scalar 
2088: field \cite{Finelli:2001sr,Notari:2002yc,Finelli:2002we}.  The system of 
2089: multi-component scalar fields generally induces isocurvature perturbations, 
2090: which can be the source of adiabatic perturbations.  In such a case the 
2091: relation (\ref{inequality}) could be satisfied, since isocurvature 
2092: perturbations act as source term for $\dot {\cal R}$ in addition to $\Phi$.  
2093: In fact the authors of ref.~\cite{Notari:2002yc} considered a specific 
2094: two-field system with a brane-modulus $\varphi$ and a dilaton $\chi$.  When 
2095: the dilaton has a negative exponential potential with a suppressed 
2096: Ekpyrotic potential for $\varphi$, the entropy ``field'' perturbation can 
2097: be scale-invariant if the model parameters are fine-tuned 
2098: \cite{Notari:2002yc}.  It was also pointed out in 
2099: ref.~\cite{Finelli:2001sr} that the quantum fluctuation of a light scalar 
2100: field (with a non-canonical kinetic term as studied in 
2101: \cite{Starobinsky:2001xq}) such as axion may lead to the flat spectra of 
2102: isocurvature perturbations.  If the correlation between adiabatic and 
2103: isocurvature perturbations is strong, adiabatic perturbations may be 
2104: scale-invariant.
2105: 
2106: We wish to stress that our work is not conclusive. In particular, in 
2107: order to fully evaluate the final power spectra, one should
2108: solve the equations of motion for fluctuations in a 
2109: nonsingular bouncing model including radiation.
2110: Important issues which should be 
2111: investigated further include:
2112: \begin{itemize}
2113: \item The final power spectra of the curvature perturbation
2114: are found to be blue-tilted for the nonsingular Ekpyrotic models 
2115: we have considered, which rely on specific higher derivative
2116: correction terms. Are there other correction terms
2117: to the action which are motivated by string theory, lead to 
2118: nonsingular bouncing scenarios and yield a flat spectrum even in the 
2119: single field case?  
2120: Perhaps toy bouncing models using exotic scalar fields or matter
2121: in refs.~\cite{Hwang:2001zt,Peter:2002cn} can be a good starting 
2122: point to construct viable nonsingular Ekpyrotic models. 
2123: \item We do not include the effect of radiation (or particles) which can be 
2124: efficiently produced near the bounce.  Particle production around
2125: the transition region is expected to be quite 
2126: efficient \cite{Tsujikawa:2001ud}, and this
2127: could lead to an additional instability of small-scale metric 
2128: perturbations.  This effect may also 
2129: non-trivially alter the nonsingular background evolution by the back-reaction 
2130: effect of created particles.  It is also required to include the radiation 
2131: after the bounce in order to evaluate the surviving spectra accurately, 
2132: although the coupling between the scalar field and radiation should be 
2133: chosen carefully in that case.
2134: \item It is of interest to study 
2135: the effect of isocurvature perturbations in the two-field system of 
2136: nonsingular Ekpyrotic scenarios.  In particular isocurvature perturbations 
2137: can be affected by the instability of the background near the bounce.  In 
2138: order to obtain the final spectra of adiabatic perturbations, we need to solve 
2139: the coupled equations of adiabatic and entropy perturbations through the 
2140: nonsingular bounce including radiation.  It is important to investigate 
2141: whether nearly scale-invariant spectra are obtained by conversion from 
2142: isocurvature to adiabatic perturbations.
2143: \end{itemize}
2144: 
2145: Our analysis  also applies to cyclic Universe models proposed in 
2146: ref.~\cite{Steinhardt:2001vw,Steinhardt:2001st} in which the
2147: bounce has been regularized by including higher-order corrections. Thus,
2148: our conclusions about the difficulty in obtaining a scale-invariant
2149: spectrum of fluctuations carry over to single field realizations of the cyclic
2150: scenario.
2151: In fact, we have done some simulations in the case of a simple 
2152: negative potential $V=m^2(\phi^2-\phi_c^2)$ for $|\phi|<\phi_c$,
2153: and found that the solutions can be nonsingular so long as
2154: the higher-order effect dominates around the graceful exit.
2155: Note, however, that the spectra of density perturbations
2156: will be the same as in the Ekpyrotic scenario. 
2157: 
2158: Recently, a preprint has appeared \cite{Mukherji:2002ft} in which
2159: in the context of a brane world scenario a nonsingular bouncing
2160: cosmology is obtained by considering the motion of a D3 brane as a
2161: boundary of a five dimensional charged anti de Sitter black hole.
2162: In this model, computed in linear theory, the spectrum of 
2163: gravitational wave fluctuations was shown not to be scale-invariant.
2164: This result supports the conclusions we have reached \footnote{We
2165: are grateful to the Referee for pointing out this reference.}.
2166: 
2167: 
2168: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2169: \section*{APPENDIX~A: HEURISTIC DERIVATION OF THE SPECTRUM OF FLUCTUATIONS}
2170: 
2171: In this Appendix we give a heuristic derivation of the spectral index
2172: of cosmological perturbations in the PBB and Ekpyrotic scenarios. This
2173: analysis is based on two key assumptions. The first is
2174: the assumption that the amplitude of the
2175: fluctuations when they exit the Hubble radius during the phase of contraction
2176: (in the Einstein frame) is given by the Hubble constant. This assumption
2177: is reasonable assuming that the fluctuations are quantum vacuum perturbations
2178: which freeze when their wavelength crosses the Hubble radius. 
2179: 
2180: The second assumption is that the ``physical magnitude'' of the fluctuations
2181: remains unchanged while the wavelength of the fluctuation is larger than
2182: the Hubble radius. This assumption is much less obvious, although at
2183: first sight this assumption may seem obvious based on causality, namely the
2184: fact that microphysics cannot influence physics on scales larger than
2185: the Hubble radius. However, in inflationary cosmology and in models
2186: with a contracting period such as the PBB and Ekpyrotic scenarios, the
2187: forward light cone (causal horizon) is much larger than the Hubble radius, 
2188: and the spatial coherence of background fields over scales of the forward
2189: light cone can lead to nontrivial effects on fluctuation modes on these
2190: scales, one of the most dramatic manifestations of this effect being
2191: the parametric amplification of super-Hubble (but sub-horizon) cosmological
2192: fluctuations during reheating in certain two-field inflationary models
2193: \cite{Taruya:1997iv}-\cite{Tsujikawa:2002nf}.
2194: Furthermore, the term ``physical magnitude'' of cosmological fluctuations 
2195: is not well determined.  On super-Hubble scales, the magnitude of the 
2196: density fluctuations depends sensitively on the coordinate system chosen.  
2197: It is possible to choose coordinate-invariant (gauge-invariant) variables 
2198: to describe the fluctuations, but there are many choices, and even in 
2199: single field inflationary models many of these gauge-invariant fluctuation 
2200: variables increase on super-Hubble scales (however, the increase between 
2201: initial Hubble radius crossing during inflation at $t_i(k)$ and final 
2202: Hubble radius crossing during the late time FRW cosmology at $t_f(k)$ is by 
2203: a factor which only depends on the ratio of the equations of state at the 
2204: two Hubble radius crossings).  This increase is a self-gravitational 
2205: effect.
2206: 
2207: In spite of the above caveats, let us proceed with the heuristic
2208: discussion of the amplitude of density fluctuations, applying it
2209: first to inflationary cosmology (exponential expansion to be
2210: specific). The quantity we wish to calculate is the mean square
2211: mass fluctuation on a scale $k$ when the corresponding wavelength
2212: enters the Hubble radius at final Hubble radius crossing $t_f(k)$.
2213: This quantity, denoted $|{{\delta M} \over M}(k, t_f(k))|^2$ is
2214: given by the power spectrum of fluctuations [see eq.~(\ref{PR})], and its
2215: $k$-dependence on the spectral index $n$ is given by
2216: \begin{equation}
2217: \left|{{\delta M} \over M}(k, t_f(k)) \right|^2 \sim k^{n-1} \,.
2218: \end{equation}  
2219: Based on the first assumption,
2220: \begin{equation}
2221: \left|{{\delta M} \over M}(k, t_i(k))  \right|^2 \sim H^2(t_i(k)) 
2222: \sim {\rm const.} \,,
2223: \end{equation}
2224: and using the second assumption we infer that
2225: \begin{equation}
2226: \left|{{\delta M} \over M}(k, t_f(k)) \right|^2 
2227: \sim \left|{{\delta M} \over M}(k, t_i(k)) \right|^2 
2228: \sim {\rm const.} \, ,
2229: \end{equation}
2230: and that hence the power spectrum is scale-invariant with an index $n = 1$.
2231: 
2232: PBB cosmology is characterized (in the Einstein frame) by a scale factor
2233: which scales as
2234: \begin{equation}
2235: a(t) \sim t^{1/3}\,,
2236: \end{equation}
2237: and thus
2238: \begin{equation}
2239: H(t) = {1 \over {3 t}} \,.
2240: \end{equation}
2241: The condition of the initial Hubble radius crossing during the
2242: period of contraction 
2243: \begin{equation}
2244: k a^{-1}(t_i(k)) = H(t_i(k))\,,
2245: \end{equation}
2246: leads to
2247: \begin{equation}
2248: t_i(k) \sim k^{-3/2}\,,~~~
2249: H(t_i(k)) \sim k^{3/2}\,,
2250: \end{equation}
2251: and thus, applying our two basic assumptions as in the case of inflationary
2252: cosmology, to
2253: \begin{equation}
2254: \left|{{\delta M} \over M}(k, t_f(k)) \right|^2 \sim k^3\,,
2255: \end{equation}
2256: which corresponds to a blue spectrum with index $n = 4$.
2257: 
2258: The analysis for Ekpyrotic cosmology is analogous. The only difference
2259: is that the value of $p$ is different, $0 < p \ll 1$ and hence
2260: \begin{equation}
2261: t_i(k) \sim k^{- 1/(1-p)}\,,~~~
2262: H(t_i(k)) \sim k^{1/(1-p)} \,,
2263: \end{equation} 
2264: and thus, taking $p = 0$ at the end,
2265: \begin{equation}
2266: \left| {{\delta M} \over M}(k, t_f(k)) \right|^2 \sim k^2 \,,
2267: \end{equation}
2268: which corresponds to a blue spectrum with index $n = 3$.
2269: 
2270: Obviously, given the caveats discussed at the beginning of this Appendix,
2271: the results for PBB and Ekpyrotic cosmology cannot be trusted without
2272: a fully relativistic analysis. The growth rates of cosmological
2273: fluctuations are very different in expanding and contracting cosmologies,
2274: and thus even given that the above heuristic analysis works in the
2275: case of inflationary cosmology, this does not mean it has to work for
2276: PBB and Ekpyrotic cosmologies. However, the results of our paper
2277: are in agreement with those derived from the heuristic analysis.
2278: 
2279: 
2280: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2281: \section*{APPENDIX~B: Analytic estimates for the gravitational potential}
2282: 
2283: Let us analyze the gravitational $\Phi$ in more details.  In the Einstein 
2284: frame, the gravitational potential $\Phi_k^E$ in the longitudinal gauge is 
2285: expressed in terms of ${\cal R}_k^E$ in the absence of 
2286: higher-order corrections \cite{Finelli:1998bu,Hwang:2002fp}: 
2287: %
2288: \begin{eqnarray} 
2289: \Phi_k^E=\frac{a_E^2\dot{H}_E}{k^2 H_E} \dot{R}_k^{E}\,.
2290: \label{Phi}
2291: \end{eqnarray}
2292: %
2293: The gravitational potential $\Phi_k^S$ in the string frame is
2294: related with the one in the Einstein frame as \cite{Hwang:re} 
2295: %
2296: \begin{eqnarray} 
2297: \Phi_k^S=\Phi_k^E+\frac{\delta F}{2F}=\Phi_k^E-\frac12 \delta\phi_k \,.
2298: \label{rel}
2299: \end{eqnarray}
2300: %
2301: Making use of eqs.~(\ref{ta}) and (\ref{relation}), we  find that 
2302: the curvature perturbation in the Einstein frame is exactly the same as 
2303: that in the string frame (i.e, ${\cal R}_k^E={\cal R}_k^S$).  
2304: Therefore the gravitational potential in the Einstein frame is expressed in 
2305: terms of $\dot{{\cal R}}_k^S$: 
2306: %
2307: \begin{eqnarray} 
2308: \Phi_k^E=\frac{a_S^2\left( \dot{H}_S-\ddot{\phi}/2
2309: + \dot{\phi}H_S/2-\dot{\phi}^4/4\right)} {k^2(H_S-
2310: \dot{\phi}/2)} \dot{\cal R}_k^S\,.
2311: \label{modi}
2312: \end{eqnarray}
2313: %
2314: Note that dots in eq.~(\ref{modi}) denote the time-derivative with respect 
2315: to $t_S$.  This is the equation which we solve numerically.
2316: 
2317: Taking note of the relation, $H_{\nu}'(x)=H_{\nu-1}(x)-(\nu/x)H_{\nu}(x)$, 
2318: one finds 
2319: %
2320: \begin{eqnarray} 
2321:  \dot{\cal R}_k^S=\frac{\sqrt{\pi}}{4a_Sz} \left[ 2\sqrt{s|\eta_S|}k 
2322:  \left(c_1 H_{\nu-1}^{(1)}(x) +c_2 H_{\nu-1}^{(2)}(x) \right) \pm 
2323:  \frac{1}{\sqrt{|\eta_S|}}\left(1-2\nu \mp 2|\eta_S| \frac{z'}{z}\right) 
2324:  \left(c_1 H_{\nu}^{(1)}(x) +c_2 H_{\nu}^{(2)}(x) \right) \right]\,,
2325: \label{dotPsi}
2326: \end{eqnarray}
2327: %
2328: where each sign corresponds to the case with $\eta_S>0$ and $\eta_S<0$,
2329: respectively. 
2330: When the evolution of $z$ is given by $z \propto |\eta_S|^{\gamma}$,
2331: we have
2332: %
2333: \begin{eqnarray} 
2334: 1-2\nu \mp 2|\eta_S| \frac{z'}{z} =1-2\nu- 2\gamma
2335: =1-2\gamma-\left|1-2\gamma \right|=
2336: \cases{ 0 & for $\gamma<1/2$ \,, 
2337: \cr 2(1-2\gamma) & for $\gamma>1/2$ \,.  \cr }
2338: \label{re}
2339: \end{eqnarray}
2340: %
2341: This term completely vanishes during the contracting  phase in the 
2342: Ekyprotic cosmology with $p<1/3$, since $\gamma$ is less than $1/2$.
2343: In this case the gravitational potential in the Einstein frame can be 
2344: expressed as
2345: %
2346: \begin{eqnarray} 
2347: \Phi_k^E=\frac{a_S \sqrt{\pi s |\eta_S|}\left( \dot{H}_S-\ddot{\phi}/2
2348: + \dot{\phi}H_S/2-\dot{\phi}^4/4\right)}{2z(H_S-\dot{\phi}/2)k} 
2349: \left(c_1 H_{\nu-1}^{(1)}(x) +c_2 H_{\nu-1}^{(2)}(x) \right)\,.
2350: \label{Phi3}
2351: \end{eqnarray}
2352: %
2353: This relation is used to estimate the spectrum of $\Phi_k^E$
2354: in Sec.~IV.
2355: 
2356:  
2357: 
2358: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2359: \section*{ACKNOWLEDGMENTS}
2360: We are grateful to Jai-chan Hwang, Daisuke Ida, Justin Khoury, Jerome Martin, 
2361: Patrick Peter, Alexey Toporensky, Gabriele Veneziano, David Wands
2362: and Jun'ichi Yokoyama for useful discussions.  
2363: F.F.  wishes to thank Cyril Cartier, 
2364: Ruth Durrer and Filippo Vernizzi for their invitation to visit the University 
2365: of Geneva and for stimulating discussions.  S.T.  is grateful for financial 
2366: support from the JSPS (No.  04942).  R.B.  wishes to thank the CERN Theory 
2367: Division and the Institut d'Astrophysique de Paris for their hospitality 
2368: and support during the time the work on this project was done.  He also 
2369: acknowledges partial support from the US Department of Energy under 
2370: Contract DE-FG02-91ER40688, TASK A.  
2371: Work similar to the study reported on here is being
2372: carried out by C. Cartier, R. Durrer and F. Vernizzi.
2373: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2374: 
2375: % references
2376: 
2377: \begin{thebibliography}{99}
2378: 
2379: %% superstring cosmology (review)
2380: 
2381: %\cite{Veneziano:1991ek}
2382: \bibitem{Veneziano:1991ek}
2383: G.~Veneziano,
2384: %``Scale factor duality for classical and quantum strings,''
2385: Phys.\ Lett.\ B {\bf 265}, 287 (1991).
2386: %%CITATION = PHLTA,B265,287;%%
2387: 
2388: \bibitem{Gasperini:1992em}
2389: M.~Gasperini and G.~Veneziano,
2390: %``Pre - big bang in string cosmology,''
2391: Astropart.\ Phys.\  {\bf 1}, 317 (1993)
2392: [arXiv:hep-th/9211021].
2393: %%CITATION = HEP-TH 9211021;%%
2394: 
2395: %\cite{Lidsey:1999mc}
2396: \bibitem{Lidsey:1999mc}
2397: J.~E.~Lidsey, D.~Wands and E.~J.~Copeland,
2398: %``Superstring cosmology,''
2399: Phys.\ Rept.\  {\bf 337}, 343 (2000)
2400: [arXiv:hep-th/9909061].
2401: %%CITATION = HEP-TH 9909061;%%
2402: 
2403: %\cite{Gasperini:2002}
2404: \bibitem{Gasperini:2002}
2405: M.~Gasperini and G.~Veneziano,
2406: % ``The Pre-Big Bang Scenario in String Cosmology,''
2407: arXiv:hep-th/0207130.
2408: %%CITATION = HEP-TH 0207130%%
2409: 
2410: %\cite{Khoury:2001wf}
2411: \bibitem{Khoury:2001wf}
2412: J.~Khoury, B.~A.~Ovrut, P.~J.~Steinhardt and N.~Turok,
2413: %``The ekpyrotic Universe: Colliding branes and the origin of 
2414: %the hot big  bang,''
2415: Phys.\ Rev.\ D {\bf 64}, 123522 (2001)
2416: [arXiv:hep-th/0103239].
2417: %%CITATION = HEP-TH 0103239;%%
2418:  
2419: %\cite{Kallosh:2001ai}
2420: \bibitem{Kallosh:2001ai}
2421: R.~Kallosh, L.~Kofman and A.~D.~Linde,
2422: %``Pyrotechnic Universe,''
2423: Phys.\ Rev.\ D {\bf 64}, 123523 (2001)
2424: [arXiv:hep-th/0104073].
2425: %%CITATION = HEP-TH 0104073;%%
2426: 
2427: %\cite{Kallosh:2001du}
2428: \bibitem{Kallosh:2001du}
2429: R.~Kallosh, L.~Kofman, A.~D.~Linde and A.~A.~Tseytlin,
2430: %``BPS branes in cosmology,''
2431: Phys.\ Rev.\ D {\bf 64}, 123524 (2001)
2432: [arXiv:hep-th/0106241].
2433: %%CITATION = HEP-TH 0106241;%%
2434: 
2435: %\cite{Enqvist:2001zk}
2436: \bibitem{Enqvist:2001zk}
2437: K.~Enqvist, E.~Keski-Vakkuri and S.~Rasanen,
2438: %``Hubble law and brane matter after ekpyrosis,''
2439: Nucl.\ Phys.\ B {\bf 614}, 388 (2001)
2440: [arXiv:hep-th/0106282].
2441: %%CITATION = HEP-TH 0106282;%%
2442: 
2443: %\cite{Rasanen:2001hf}
2444: \bibitem{Rasanen:2001hf}
2445: S.~Rasanen,
2446: %``On ekpyrotic brane collisions,''
2447: Nucl.\ Phys.\ B {\bf 626}, 183 (2002)
2448: [arXiv:hep-th/0111279].
2449: %%CITATION = HEP-TH 0111279;%%
2450: 
2451: %\cite{Felder:2002jk}
2452: \bibitem{Felder:2002jk}
2453: G.~N.~Felder, A.~Frolov, L.~Kofman and A.~Linde,
2454: %``Cosmology with negative potentials,''
2455: Phys.\ Rev.\ D {\bf 66}, 023507 (2002)
2456: [arXiv:hep-th/0202017].
2457: %%CITATION = HEP-TH 0202017;%%
2458: 
2459: %\cite{Linde:2002ws}
2460: \bibitem{Linde:2002ws}
2461: A.~Linde,
2462: %``Inflationary theory versus ekpyrotic / cyclic scenario,''
2463: arXiv:hep-th/0205259.
2464: %%CITATION = HEP-TH 0205259;%%
2465:  
2466: \bibitem{Press}
2467: W. H. Press, Phys. Scr. {\bf 21}, 702 (1980);
2468: %\cite{Press:gg}
2469: %\bibitem{Press:gg}
2470: W.~H.~Press,
2471: %``Galaxies May Be Single Particle 
2472: %Fluctuations From An Early, False Vacuum Era. (Talk),''
2473: {\it  In *Les Arcs 1981, Proceedings, Cosmology and Particles*, 137-156}.
2474: 
2475: \bibitem{BGGMV}
2476: %\cite{Brustein:1994kn}
2477: %\bibitem{Brustein:1994kn}
2478: R.~Brustein, M.~Gasperini, M.~Giovannini, V.~F.~Mukhanov and G.~Veneziano,
2479: %``Metric perturbations in dilaton driven inflation,''
2480: Phys.\ Rev.\ D {\bf 51}, 6744 (1995)
2481: [arXiv:hep-th/9501066].
2482: %%CITATION = HEP-TH 9501066;%%
2483: 
2484: %\cite{Lyth:2001pf}
2485: \bibitem{Lyth:2001pf}
2486: D.~H.~Lyth,
2487: %``The primordial curvature perturbation in the ekpyrotic Universe,''
2488: Phys.\ Lett.\ B {\bf 524}, 1 (2002)
2489: [arXiv:hep-ph/0106153].
2490: %%CITATION = HEP-PH 0106153;%%
2491: 
2492: %\cite{Brandenberger:2001bs}
2493: \bibitem{Brandenberger:2001bs}
2494: R.~Brandenberger and F.~Finelli,
2495: %``On the spectrum of fluctuations 
2496: %in an effective field theory of the  ekpyrotic Universe,''
2497: JHEP {\bf 0111}, 056 (2001)
2498: [arXiv:hep-th/0109004].
2499: %%CITATION = HEP-TH 0109004;%%
2500: 
2501: %\cite{Hwang:2001ga}
2502: \bibitem{Hwang:2001ga}
2503: J.~c.~Hwang,
2504: %``Cosmological structure problem in the ekpyrotic scenario,''
2505: Phys.\ Rev.\ D {\bf 65}, 063514 (2002)
2506: [arXiv:astro-ph/0109045].
2507: %%CITATION = ASTRO-PH 0109045;%%
2508: 
2509: %\cite{Tsujikawa:2001ad}
2510: \bibitem{Tsujikawa:2001ad}
2511: S.~Tsujikawa,
2512: %``Density perturbations in the ekpyrotic Universe and 
2513: % string-inspired  generalizations,''
2514: Phys.\ Lett.\ B {\bf 526}, 179 (2002)
2515: [arXiv:gr-qc/0110124].
2516: %%CITATION = GR-QC 0110124;%%
2517: 
2518: %\cite{Khoury:2001zk}
2519: \bibitem{Khoury:2001zk}
2520: J.~Khoury, B.~A.~Ovrut, P.~J.~Steinhardt and N.~Turok,
2521: %``Density perturbations in the ekpyrotic scenario,''
2522: arXiv:hep-th/0109050.
2523: %%CITATION = HEP-TH 0109050;%%
2524: 
2525: %\cite{Durrer:2002jn}
2526: \bibitem{Durrer:2002jn}
2527: R.~Durrer and F.~Vernizzi,
2528: %``Adiabatic perturbations in pre big bang models: 
2529: % Matching conditions and  scale invariance,''
2530: arXiv:hep-ph/0203275.
2531: %%CITATION = HEP-PH 0203275;%%
2532: 
2533: %\cite{Lyth:2001nv}
2534: \bibitem{Lyth:2001nv}
2535: D.~H.~Lyth,
2536: %``The failure of cosmological perturbation 
2537: %theory in the new ekpyrotic  scenario,''
2538: Phys.\ Lett.\ B {\bf 526}, 173 (2002)
2539: [arXiv:hep-ph/0110007].
2540: %%CITATION = HEP-PH 0110007;%%
2541: 
2542: %\cite{Deruelle:1995kd}
2543: \bibitem{Deruelle:1995kd}
2544: N.~Deruelle and V.~F.~Mukhanov,
2545: %``On matching conditions for cosmological perturbations,''
2546: Phys.\ Rev.\ D {\bf 52}, 5549 (1995)
2547: [arXiv:gr-qc/9503050].
2548: %%CITATION = GR-QC 9503050;%%
2549: 
2550: %\cite{Hwang:1991an}
2551: \bibitem{Hwang:1991an}
2552: J.~c.~Hwang and E.~T.~Vishniac,
2553: %``Gauge-invariant joining conditions for cosmological perturbations,''
2554: Astrophys.\ J.\  {\bf 382}, 363 (1991).
2555: %%CITATION = ASJOA,382,363;%%
2556: 
2557: %\cite{Finelli:2001sr}
2558: \bibitem{Finelli:2001sr}
2559: F.~Finelli and R.~Brandenberger,
2560: %``On the generation of a scale-invariant spectrum of adiabatic 
2561: %fluctuations in cosmological models with a contracting phase,''
2562: Phys.\ Rev.\ D {\bf 65}, 103522 (2002)
2563: [arXiv:hep-th/0112249].
2564: %%CITATION = HEP-TH 0112249;%%
2565: 
2566: %\cite{Martin:2001ue}
2567: \bibitem{Martin:2001ue}
2568: J.~Martin, P.~Peter, N.~Pinto Neto and D.~J.~Schwarz,
2569: %``Passing through the bounce in the ekpyrotic models,''
2570: Phys.\ Rev.\ D {\bf 65}, 123513 (2002)
2571: [arXiv:hep-th/0112128].
2572: %%CITATION = HEP-TH 0112128;%%
2573: 
2574: %\cite{Martin:2002ar}
2575: \bibitem{Martin:2002ar}
2576: J.~Martin, P.~Peter, N.~Pinto-Neto and D.~J.~Schwarz,
2577: %``Comment on 'Density perturbations in the ekpyrotic scenario',''
2578: arXiv:hep-th/0204222.
2579: %%CITATION = HEP-TH 0204222;%%
2580: 
2581: 
2582: 
2583: %\cite{Gasperini:1996fu}
2584: \bibitem{Gasperini:1996fu}
2585: M.~Gasperini, M.~Maggiore and G.~Veneziano,
2586: %``Towards a non-singular pre-big bang cosmology,''
2587: Nucl.\ Phys.\ B {\bf 494}, 315 (1997)
2588: [arXiv:hep-th/9611039].
2589: %%CITATION = HEP-TH 9611039;%%
2590: 
2591: %\cite{Brustein:1997cv}
2592: \bibitem{Brustein:1997cv}
2593: R.~Brustein and R.~Madden,
2594: %``A model of graceful exit in string cosmology,''
2595: Phys.\ Rev.\ D {\bf 57}, 712 (1998)
2596: [arXiv:hep-th/9708046].
2597: %%CITATION = HEP-TH 9708046;%%
2598: 
2599: %\cite{Brustein:1997xn}
2600: \bibitem{Brustein:1997xn}
2601: R.~Brustein and R.~Madden,
2602: %``Graceful exit in string cosmology,''
2603: Hadronic J.\  {\bf 21}, 202 (1998)
2604: [arXiv:hep-th/9711134].
2605: %%CITATION = HEP-TH 9711134;%%
2606: 
2607: 
2608: %\cite{Rey:1996ad}
2609: \bibitem{Rey:1996ad}
2610: S.~J.~Rey,
2611: %``Back reaction and graceful exit in string inflationary cosmology,''
2612: Phys.\ Rev.\ Lett.\  {\bf 77}, 1929 (1996)
2613: [arXiv:hep-th/9605176].
2614: %%CITATION = HEP-TH 9605176;%%
2615: 
2616: %\cite{Foffa:1999dv}
2617: \bibitem{Foffa:1999dv}
2618: S.~Foffa, M.~Maggiore and R.~Sturani,
2619: %``Loop corrections and graceful exit in string cosmology,''
2620: Nucl.\ Phys.\ B {\bf 552}, 395 (1999)
2621: [arXiv:hep-th/9903008].
2622: %%CITATION = HEP-TH 9903008;%%
2623: 
2624: %\cite{Cartier:1999vk}
2625: \bibitem{Cartier:1999vk}
2626: C.~Cartier, E.~J.~Copeland and R.~Madden,
2627: %``The graceful exit in string cosmology,''
2628: JHEP {\bf 0001}, 035 (2000)
2629: [arXiv:hep-th/9910169].
2630: %%CITATION = HEP-TH 9910169;%%
2631: 
2632: %\cite{Mukhanov:1991zn}
2633: %\bibitem{Mukhanov:1991zn}
2634: \bibitem{MB}
2635: V.~Mukhanov and R.~H.~Brandenberger,
2636: %``A Nonsingular Universe,''
2637: Phys.\ Rev.\ Lett.\  {\bf 68}, 1969 (1992).
2638: %%CITATION = PRLTA,68,1969;%%
2639: 
2640: %\cite{Brandenberger:1993ef}
2641: %\bibitem{Brandenberger:1993ef}
2642: \bibitem{BMS}
2643: R.~H.~Brandenberger, V.~Mukhanov and A.~Sornborger,
2644: %``A Cosmological theory without singularities,''
2645: Phys.\ Rev.\ D {\bf 48}, 1629 (1993)
2646: [arXiv:gr-qc/9303001].
2647: %%CITATION = GR-QC 9303001;%%
2648: 
2649: %\cite{Brandenberger:1998zs}
2650: \bibitem{Brandenberger:1998zs}
2651: R.~H.~Brandenberger, R.~Easther and J.~Maia,
2652: %``Nonsingular dilaton cosmology,''
2653: JHEP {\bf 9808}, 007 (1998)
2654: [arXiv:gr-qc/9806111].
2655: %%CITATION = GR-QC 9806111;%%
2656: 
2657: %\cite{Easson:1999xw}
2658: \bibitem{Easson:1999xw}
2659: D.~A.~Easson and R.~H.~Brandenberger,
2660: %``Nonsingular dilaton cosmology in the string frame,''
2661: JHEP {\bf 9909}, 003 (1999)
2662: [arXiv:hep-th/9905175].
2663: %%CITATION = HEP-TH 9905175;%%
2664: 
2665: %\cite{Cartier:2001is}
2666: \bibitem{Cartier:2001is}
2667: C.~Cartier, J.~c.~Hwang and E.~J.~Copeland,
2668: %``Evolution of cosmological perturbations 
2669: %in non-singular string  cosmologies,''
2670: Phys.\ Rev.\ D {\bf 64}, 103504 (2001)
2671: [arXiv:astro-ph/0106197].
2672: %%CITATION = ASTRO-PH 0106197;%%
2673: 
2674: %\cite{Gordon:2002jw}
2675: \bibitem{Gordon:2002jw}
2676: C.~Gordon and N.~Turok,
2677: %``Cosmological perturbations through a general relativistic bounce,''
2678: arXiv:hep-th/0206138.
2679: %%CITATION = HEP-TH 0206138;%%
2680: 
2681: %\cite{Khoury:2001bz}
2682: \bibitem{Khoury:2001bz}
2683: J.~Khoury, B.~A.~Ovrut, N.~Seiberg, P.~J.~Steinhardt and N.~Turok,
2684: %``From big crunch to big bang,''
2685: Phys.\ Rev.\ D {\bf 65}, 086007 (2002)
2686: [arXiv:hep-th/0108187].
2687: %%CITATION = HEP-TH 0108187;%%
2688: 
2689: %% Cyclic Universe
2690: %\cite{Steinhardt:2001vw}
2691: \bibitem{Steinhardt:2001vw}
2692: P.~J.~Steinhardt and N.~Turok,
2693: %``A cyclic model of the Universe,''
2694: arXiv:hep-th/0111030.
2695: %%CITATION = HEP-TH 0111030;%%
2696: 
2697: %\cite{Steinhardt:2001st}
2698: \bibitem{Steinhardt:2001st}
2699: P.~J.~Steinhardt and N.~Turok,
2700: %``Cosmic evolution in a cyclic universe,''
2701: Phys.\ Rev.\ D {\bf 65}, 126003 (2002)
2702: [arXiv:hep-th/0111098].
2703: %%CITATION = HEP-TH 0111098;%%
2704: 
2705: %\cite{Hwang:1999gf}
2706: \bibitem{Hwang:1999gf}
2707: J.~c.~Hwang and H.~Noh,
2708: %``Conserved cosmological structures 
2709: %in the one-loop superstring effective action,''
2710: Phys.\ Rev.\ D {\bf 61}, 043511 (2000)
2711: [arXiv:astro-ph/9909480].
2712: %%CITATION = ASTRO-PH 9909480;%%
2713: 
2714: %\cite{Cartier:2001gc}
2715: \bibitem{Cartier:2001gc}
2716: C.~Cartier, E.~J.~Copeland and M.~Gasperini,
2717: %``Gravitational waves in non-singular string cosmologies,''
2718: Nucl.\ Phys.\ B {\bf 607}, 406 (2001)
2719: [arXiv:gr-qc/0101019].
2720: %%CITATION = GR-QC 0101019;%%
2721:  
2722: %\cite{Mukhanov:1990me}
2723: \bibitem{MFB}
2724: V.~F.~Mukhanov, H.~A.~Feldman and R.~H.~Brandenberger,
2725: %``Theory Of Cosmological Perturbations. Part 1. 
2726: %Classical Perturbations. Part 2. 
2727: %Quantum Theory Of Perturbations. Part 3. Extensions,''
2728: Phys.\ Rept.\  {\bf 215}, 203 (1992).
2729: %%CITATION = PRPLC,215,203;%%
2730: 
2731: %\cite{Lyth:1984gv}
2732: \bibitem{Lyth:1985}
2733: D.~H.~Lyth,
2734: %``Large Scale Energy Density Perturbations And Inflation,''
2735: Phys.\ Rev.\ D {\bf 31}, 1792 (1985).
2736: %%CITATION = PHRVA,D31,1792;%%
2737: 
2738: %\cite{Hwang:2002ks}
2739: \bibitem{Hwang:2002ks}
2740: J.~c.~Hwang and H.~Noh,
2741: %``Identification of perturbation modes and 
2742: % controversies in ekpyrotic  perturbations,''
2743: arXiv:hep-th/0203193.
2744: %%CITATION = HEP-TH 0203193;%%
2745: 
2746: %\cite{Hwang:2001zt}
2747: \bibitem{Hwang:2001zt}
2748: J.~c.~Hwang and H.~Noh,
2749: %``Non-singular big-bounces and evolution of linear fluctuations,''
2750: Phys.\ Rev.\ D {\bf 65}, 124010 (2002)
2751: [arXiv:astro-ph/0112079].
2752: %%CITATION = ASTRO-PH 0112079;%%
2753: 
2754: %\cite{Peter:2002cn}
2755: \bibitem{Peter:2002cn}
2756: P.~Peter and N.~Pinto-Neto,   
2757: %``Primordial perturbations in a non singular bouncing Universe model,''
2758: arXiv:hep-th/0203013.
2759: %%CITATION = HEP-TH 0203013;%%
2760: 
2761: 
2762: 
2763: %\cite{Brandenberger:tg}
2764: \bibitem{Brandenberger:1984}
2765: R.~H.~Brandenberger and R.~Kahn,
2766: %``Cosmological Perturbations In Inflationary Universe Models,''
2767: Phys.\ Rev.\ D {\bf 29}, 2172 (1984).
2768: %%CITATION = PHRVA,D29,2172;%%
2769: 
2770: %\cite{Bardeen:qw}
2771: \bibitem{Bardeen:1983}
2772: J.~M.~Bardeen, P.~J.~Steinhardt and M.~S.~Turner,
2773: %``Spontaneous Creation Of Almost Scale - 
2774: %Free Density Perturbations In An Inflationary Universe,''
2775: Phys.\ Rev.\ D {\bf 28}, 679 (1983).
2776: %%CITATION = PHRVA,D28,679;%%
2777: 
2778: %\cite{Heard:2002dr}
2779: \bibitem{Heard:2002dr}
2780: I.~P.~Heard and D.~Wands,
2781: %``Cosmology with positive and negative exponential potentials,''
2782: arXiv:gr-qc/0206085.
2783: %%CITATION = GR-QC 0206085;%%
2784: 
2785: 
2786: %\cite{Antoniadis:1992rq}
2787: \bibitem{Antoniadis:1992rq}
2788: I.~Antoniadis, E.~Gava and K.~S.~Narain,
2789: %``Moduli corrections to gauge 
2790: %and gravitational couplings in four-dimensional superstrings,''
2791: Nucl.\ Phys.\ B {\bf 383}, 93 (1992)
2792: [arXiv:hep-th/9204030].
2793: %%CITATION = HEP-TH 9204030;%%
2794: 
2795: %\cite{Antoniadis:1993jc}
2796: \bibitem{Antoniadis:1993jc}
2797: I.~Antoniadis, J.~Rizos and K.~Tamvakis,
2798: %``Singularity - free cosmological 
2799: %solutions of the superstring effective action,''
2800: Nucl.\ Phys.\ B {\bf 415}, 497 (1994)
2801: [arXiv:hep-th/9305025].
2802: %%CITATION = HEP-TH 9305025;%%
2803: 
2804: %\cite{Rizos:1993rt}
2805: \bibitem{Rizos:1993rt}
2806: J.~Rizos and K.~Tamvakis,
2807: %``On the existence of singularity free solutions in quadratic gravity,''
2808: Phys.\ Lett.\ B {\bf 326}, 57 (1994)
2809: [arXiv:gr-qc/9401023].
2810: %%CITATION = GR-QC 9401023;%%
2811: 
2812: %\cite{Kanti:1998jd}
2813: \bibitem{Kanti:1998jd}
2814: P.~Kanti, J.~Rizos and K.~Tamvakis,
2815: %``Singularity-free cosmological solutions in quadratic gravity,''
2816: Phys.\ Rev.\ D {\bf 59}, 083512 (1999)
2817: [arXiv:gr-qc/9806085].
2818: %%CITATION = GR-QC 9806085;%%
2819: 
2820: %\cite{Easther:1996yd}
2821: \bibitem{Easther:1996yd}
2822: R.~Easther and K.~Maeda,
2823: %``One-Loop Superstring Cosmology and the Non-Singular Universe,''
2824: Phys.\ Rev.\ D {\bf 54}, 7252 (1996)
2825: [arXiv:hep-th/9605173].
2826: %%CITATION = HEP-TH 9605173;%%
2827: 
2828: %\cite{Kawai:1998bn}
2829: \bibitem{Kawai:1998bn}
2830: S.~Kawai and J.~Soda,
2831: %``Non-singular Bianchi type I 
2832: %cosmological solutions from 1-loop  superstring effective action,''
2833: Phys.\ Rev.\ D {\bf 59}, 063506 (1999)
2834: [arXiv:gr-qc/9807060].
2835: %%CITATION = GR-QC 9807060;%%
2836: 
2837: %\cite{Yajima:1999gk}
2838: \bibitem{Yajima:1999gk}
2839: H.~Yajima, K.~Maeda and H.~Ohkubo,
2840: %``Generality of singularity avoidance 
2841: %in superstring theory: Anisotropic  case,''
2842: Phys.\ Rev.\ D {\bf 62}, 024020 (2000)
2843: [arXiv:gr-qc/9910061].
2844: %%CITATION = GR-QC 9910061;%%
2845: 
2846: %\cite{Alexeyev:2000eb}
2847: \bibitem{Alexeyev:2000eb}
2848: S.~O.~Alexeyev, A.~Toporensky and V.~Ustiansky,
2849: %``The nature of singularity in Bianchi I cosmological 
2850: % string gravity  model with second order curvature corrections,''
2851: Phys.\ Lett.\ B {\bf 509}, 151 (2001)
2852: [arXiv:gr-qc/0009020].
2853: %%CITATION = GR-QC 0009020;%%
2854: 
2855: %\cite{Toporensky:2002ta}
2856: \bibitem{Toporensky:2002ta}
2857: A.~Toporensky and S.~Tsujikawa,
2858: %``Nature of singularities in anisotropic string cosmology,''
2859: Phys.\ Rev.\ D {\bf 65}, 123509 (2002)
2860: [arXiv:gr-qc/0202067].
2861: %%CITATION = GR-QC 0202067;%%
2862: 
2863: %\cite{Alekseev:eh}
2864: \bibitem{Alekseev:eh}
2865: S.~O.~Alexeyev, A.~V.~Toporensky and V.~O.~Ustiansky, 
2866: %``Non-Singular Cosmological Models 
2867: %In String Gravity With Constant Dilaton And 
2868: %Second-Order Curvature Corrections,''
2869: Class.\ Quant.\ Grav.\  {\bf 17}, 2243 (2000).
2870: %%CITATION = CQGRD,17,2243;%%
2871: 
2872: %\cite{Kawai:1999pw}
2873: \bibitem{Kawai:1999pw}
2874: S.~Kawai and J.~Soda,
2875: %``Evolution of fluctuations during graceful exit in string cosmology,''
2876: Phys.\ Lett.\ B {\bf 460}, 41 (1999)
2877: [arXiv:gr-qc/9903017].
2878: %%CITATION = GR-QC 9903017;%%
2879: 
2880: %\cite{Bardeen:1988hy}
2881: \bibitem{Bardeen:1988hy}
2882: J.~M.~Bardeen,
2883: %``Cosmological Perturbations 
2884: %From Quantum Fluctuations To Large Scale Structure,''
2885: DOE/ER/40423-01-C8
2886: {\it Lectures given at 2nd Guo Shou-jing Summer School on Particle Physics and Cosmology, 
2887: Nanjing, China, Jul 1988}.
2888: 
2889: %\cite{Notari:2002yc}
2890: \bibitem{Notari:2002yc}
2891: A.~Notari and A.~Riotto,
2892: %``Isocurvature perturbations in the ekpyrotic Universe,''
2893: arXiv:hep-th/0205019.
2894: %%CITATION = HEP-TH 0205019;%%
2895:  
2896: %\cite{Finelli:2002we}
2897: \bibitem{Finelli:2002we}
2898: F.~Finelli,
2899: %``Assisted contraction,''
2900: arXiv:hep-th/0206112.
2901: %%CITATION = HEP-TH 0206112;%%
2902:  
2903: 
2904: %\cite{Starobinsky:2001xq}
2905: \bibitem{Starobinsky:2001xq}
2906: A.~A.~Starobinsky, S.~Tsujikawa and J.~Yokoyama,
2907: %``Cosmological perturbations from multi-field inflation 
2908: %in generalized  Einstein theories,''
2909: Nucl.\ Phys.\ B {\bf 610}, 383 (2001)
2910: [arXiv:astro-ph/0107555].
2911: %%CITATION = ASTRO-PH 0107555;%%
2912: 
2913: %\cite{Tsujikawa:2001ud}
2914: \bibitem{Tsujikawa:2001ud}
2915: S.~Tsujikawa and H.~Yajima,
2916: %``Massive fermion production in nonsingular 
2917: %superstring cosmology,''
2918: Phys.\ Rev.\ D {\bf 64}, 023519 (2001)
2919: [arXiv:hep-ph/0103148].
2920: %%CITATION = HEP-PH 0103148;%%
2921: 
2922: %\cite{Mukherji:2002ft}
2923: \bibitem{Mukherji:2002ft}
2924: S.~Mukherji and M.~Peloso,
2925: %``Bouncing and cyclic universes from brane models,''
2926: arXiv:hep-th/0205180.
2927: %%CITATION = HEP-TH 0205180;%%
2928: 
2929: %\cite{Taruya:1997iv}
2930: \bibitem{Taruya:1997iv}
2931: A.~Taruya and Y.~Nambu,
2932: %``Cosmological perturbation with two scalar fields
2933: % in reheating after  inflation,''
2934: Phys.\ Lett.\ B {\bf 428}, 37 (1998)
2935: [arXiv:gr-qc/9709035].
2936: %%CITATION = GR-QC 9709035;%%
2937: 
2938: %\cite{Bassett:1998wg}
2939: \bibitem{Bassett:1998wg}
2940: B.~A.~Bassett, D.~I.~Kaiser and R.~Maartens,
2941: %``General relativistic preheating after inflation,''
2942: Phys.\ Lett.\ B {\bf 455}, 84 (1999)
2943: [arXiv:hep-ph/9808404].
2944: %%CITATION = HEP-PH 9808404;%%
2945: 
2946: %\cite{Finelli:1998bu}
2947: \bibitem{Finelli:1998bu}
2948: F.~Finelli and R.~H.~Brandenberger,
2949: %``Parametric amplification of gravitational 
2950: %fluctuations during reheating,''
2951: Phys.\ Rev.\ Lett.\  {\bf 82}, 1362 (1999)   
2952: [arXiv:hep-ph/9809490].
2953: %%CITATION = HEP-PH 9809490;%%
2954: 
2955: %\cite{Bassett:1999cg}
2956: \bibitem{Bassett:1999cg}
2957: B.~A.~Bassett and F.~Viniegra,
2958: %``Massless metric preheating,''
2959: Phys.\ Rev.\ D {\bf 62}, 043507 (2000)
2960: [arXiv:hep-ph/9909353].
2961: %%CITATION = HEP-PH 9909353;%%
2962: 
2963: %\cite{Finelli:2000ya}
2964: \bibitem{Finelli:2000ya}
2965: F.~Finelli and R.~H.~Brandenberger,
2966: %``Parametric amplification of metric fluctuations 
2967: %during reheating in two  field models,''
2968: Phys.\ Rev.\ D {\bf 62}, 083502 (2000)
2969: [arXiv:hep-ph/0003172].
2970: %%CITATION = HEP-PH 0003172;%%
2971: 
2972: %\cite{Bassett:2001jg}
2973: \bibitem{Bassett:2001jg}
2974: B.~A.~Bassett, M.~Peloso, L.~Sorbo and S.~Tsujikawa,
2975: %``Fermion production from preheating-amplified 
2976: %metric perturbations,''
2977: Nucl.\ Phys.\ B {\bf 622}, 393 (2002)
2978: [arXiv:hep-ph/0109176].
2979: %%CITATION = HEP-PH 0109176;%%
2980: 
2981: %\cite{Tsujikawa:2002nf}
2982: \bibitem{Tsujikawa:2002nf}
2983: S.~Tsujikawa and B.~A.~Bassett,
2984: %``When can preheating affect the CMB?,''
2985: Phys.\ Lett.\ B {\bf 536}, 9 (2002)
2986: [arXiv:astro-ph/0204031].
2987: %%CITATION = ASTRO-PH 0204031;%%
2988: 
2989: %\cite{Hwang:2002fp}
2990: \bibitem{Hwang:2002fp}
2991: J.~c.~Hwang and H.~Noh,
2992: %``Cosmological perturbations in a generalized gravity 
2993: %including tachyonic condensation,''
2994: arXiv:hep-th/0206100.
2995: %%CITATION = HEP-TH 0206100;%%
2996: 
2997: %\cite{Hwang:re}
2998: \bibitem{Hwang:re}
2999: J.~c.~Hwang,
3000: %``Cosmological Perturbations In Generalized
3001: % Gravity Theories: Formulation,''
3002: Class.\ Quant.\ Grav.\  {\bf 7}, 1613 (1990).
3003: %%CITATION = CQGRD,7,1613;%%
3004: 
3005: \end{thebibliography}
3006: 
3007: \end{document}
3008: 
3009: 
3010: