1: %\documentclass[prd,showpacs,preprint,aps,nofootinbib,grouprefs]{revtex4}

2: \documentclass[aps,prd,showpacs,preprint,onecolumn,floatfix,nofootinbib]{revtex4}

3: %\bibliographystyle{apsrev}

4: 

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

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

7: \newcommand{\ba}{\begin{eqnarray}}

8: \newcommand{\ea}{\end{eqnarray}}

9: \newcommand{\bc}{\begin{center}}

10: \newcommand{\ec}{\end{center}}

11: \newcommand{\vs}{\vspace*{3mm}}

12: \newcommand{\dis}{\displaystyle}

13: \usepackage{amsmath}

14: \usepackage{amstext}

15: \usepackage{amsthm}

16: \usepackage{amssymb}

17: 

18: %\numberwithin{equation}{section}

19: 

20: %\pagestyle{plain}

21: \topmargin 0.2cm

22: %\textheight23.5cm

23: %***************************************************

24: %\addtolength{\oddsidemargin}{12mm}

25: %\addtolength{\evensidemargin}{12mm}

26: %\addtolength{\textwidth}{-24mm}

27: %%***************************************************

28: %\addtolength{\topmargin}{11mm}

29: %\addtolength{\footskip}{11mm}

30: %\addtolength{\textheight}{-22mm}

31: %***************************************************

32: %\renewcommand{\baselinestretch}{1.57}

33: %***************************************************

34: %\tightenlines

35: %\textwidth16cm

36: %\setlength{\oddsidemargin}{0cm}

37: %\setlength{\evensidemargin}{0.4cm}

38: %\setlength{\jot}{0.3cm}

39: %

40: 

41: 

42: 

43: 

44: 

45: \begin{document}

46: 

47: 

48: \preprint{MZ-TH/04-16}

49: %\title{Proper time flow equation for gravity}

50: %\author{Alfio Bonanno}

51: %\affiliation{ INAF -  Osservatorio Astrofisico di Catania,

52: %%Via S.Sofia 78, I-95123 Catania, Italy\\

53: %INFN, Via S. Sofia 64, I-95123 Catania, Italy}

54: %\author{Martin Reuter}

55: %\affiliation{Institut f\"ur Physik, Universit\"at Mainz\\

56: %Staudingerweg 7, D-55099 Mainz, Germany}

57: 

58: \begin{center}

59: 

60: 

61: 

62: {\LARGE \textsc{Proper Time Flow Equation for Gravity}}

63: 

64: 

65: 

66: \vspace{1.4cm}

67: {\large A.~Bonanno}\\

68: 

69: \vspace{0.7cm}

70: \noindent

71: \textit{INAF -  Osservatorio Astrofisico di Catania,

72: Via S.Sofia 78, I-95123 Catania, Italy\\

73: INFN, Via S. Sofia 64, I-95123 Catania, Italy}\\

74: 

75: \vspace{0.7cm}

76: {\large M.~Reuter}\\

77: 

78: \vspace{0.7cm}

79: \noindent

80: \textit{Institute of Physics, University of Mainz\\

81: Staudingerweg 7, D-55099 Mainz, Germany}\\

82: 

83: \end{center}

84: 

85: \vspace*{0.6cm}

86: 

87: 

88: \date{\today}

89: 

90: \begin{abstract}

91: We analyze a proper time renormalization group equation for Quantum Einstein Gravity

92: in the Einstein-Hilbert truncation and compare its predictions to those of the

93: conceptually different exact renormalization group equation of the effective average action. We employ a smooth infrared

94: regulator of a special type which is known to give rise to extremely precise critical

95: exponents in scalar theories. We find perfect consistency between the proper time and the average action

96: renormalization group equations. In particular the proper time equation, too, predicts the existence of a non-Gaussian fixed point

97: as it is necessary for the conjectured nonperturbative renormalizability of Quantum Einstein Gravity.

98: \end{abstract}

99: %\pacs{11.10.Hi,04.60.-m,11.15.Tk,04.50.+h}

100: \pacs{11.10.Hi,04.60.-m,11.15.Tk}

101: 

102: \maketitle

103: 

104: \section{\label{sec:level1}Introduction}

105: Recently a lot of work went into the exploration of the Wilsonian renormalization group 

106: (RG) flow of Quantum Einstein Gravity 

107: \cite{mr,dou,oliver1,frank1,oliver2,souma,percaper,frank2,litimgrav} 

108: and its possible impact on black holes and 

109: cosmology 

110: \cite{bh,cosmo1,cosmo2,elo,esposito,h1,h2}.

111: By definition, Quantum Einstein Gravity (QEG) is the conjectured quantum field

112: theory of the spacetime metric whose bare action is (infinitesimally close to) an ultraviolet (UV) 

113: attractive non-Gaussian fixed point of the RG flow. As outlined by S.Weinberg  long ago \cite{wein}

114: this theory would be nonperturbatively renormalizable or ``asymptotically safe" and, most probably, 

115: predictive and internally consistent even at the shortest sub-Planckian distances. Most of the 

116: existing RG calculations  were performed within the framework of the exact RG equation pertaining to the 

117: effective average action \cite{avact,avactrev,ym}, 

118: applied to a truncated space of action functionals (theory space).

119: It turned out that the UV fixed point necessary for asymptotic safety does indeed exist within the 

120: Einstein-Hilbert \cite{souma,oliver1,frank1} and the $R^2$-truncation \cite{oliver2}, and 

121: detailed analyses of the reliability of those approximations \cite{oliver1,oliver2} revealed that the 

122: fixed point found is unlikely to be a truncation artifact. But clearly it is necessary 

123: to collect further evidence for its existence at the exact level, both within the effective 

124: average action formalism (by generalizing the truncation) and with conceptually  independent 

125: methods. 

126: 

127: In fact, a conceptually independent support of the asymptotic safety conjecture

128: comes form the work of Forg\'{a}cs and Niedermaier \cite{max} who were able to analyze the path integral 

129: over the metrics with two Killing vectors exactly and to show that at least this sector is asymptotically 

130: safe. In view of recent progress \cite{ajl} on causal, {\it i.e.} Lorentzian dynamical triangulations

131: \cite{amb} one also might hope that it will soon be possible to compare the analytical results

132: to Monte Carlo data. 

133: 

134: In the present paper we are going to analyze the non-Gaussian fixed point of QEG using the ``proper time

135: renormalization group" which has been extensively used for other theories recently 

136: \cite{pt,wf1,litim,sm}.

137: While this formalism is less complete and systematic than the exact RG equation of the average action

138: \cite{ym}, it has had quite spectacular successes in high precision calculations of critical exponents

139: \cite{wf1}. Even though, at the technical level, the proper time method is reminiscent of the 

140: average action approach, it is well known \cite{litim} that the proper time RG equation is {\it not} a special

141: case of the exact RG equation of the average action; it is conceptually independent in this sense. 

142: The approach we are going to use implements a certain RG improvement of one-loop 

143: perturbation-theory, employing an infrared (IR)-regularized

144: proper time representation of the corresponding one-loop determinants. We shall derive a flow equation

145: for the scale dependent gravitational action from it whose structure is quite different from that of the 

146: exact RG equation of the average action, for every possible choice of its built-in cutoff operator 

147: ${\cal R}_k$ \cite{avactrev}. Apart from the conceptual independence the main motivation for the present

148: analysis are the remarkably precise critical exponents which have been obtained for the Wilson-Fisher

149: fixed point with this method \cite{wf1}. In fact, we shall be mostly interested in the universal properties of the 

150: non-Gaussian fixed point of QEG,  $ (g_\ast, \lambda_\ast)$, in particular its critical exponents $\theta'$ and $\theta''$ and the 

151: product $g_\ast \lambda_\ast$.  

152: 

153: \section{The proper time flow equation}

154: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

155: The ultimate goal of the RG approach would be the calculation of the integral 

156: $Z=\int {\cal D}\gamma_{\mu\nu} \exp(-S[\gamma_{\mu\nu}])$ over all metrics

157: $\gamma_{\mu\nu}$ where $S$ is  an arbitrary bare action, invariant under general coordinate

158: transformations. We are going to employ the background 

159: field formalism and start by writing $\gamma_{\mu\nu}$ as a background 

160: $\bar{g}_{\mu\nu}$ plus a (not necessarily small)

161: fluctuation: $\gamma_{\mu\nu}=\bar{g}_{\mu\nu}+h_{\mu\nu}$. Furthermore, we gauge fix the path integral 

162: by adding gauge fixing and ghost terms to $S$. We use an a priori arbitrary gauge fixing term 

163: $S_{\rm gf}[h;g]$ of the background-type, meaning that it is invariant under a {\it combined}

164: transformation of $h_{\mu\nu}$ and $\bar{g}_{\mu\nu}$. This property will guarantee that the 

165: effective action becomes invariant under general coordinates transformation. Thus, denoting the Faddeev-Popov

166: ghosts by $C^{\mu}$ and $\bar{C}_\mu$, and their action by $S_{\rm gh}[h,C,\bar{C};\bar{g}]$, the path

167: integral reads $Z=\int{\cal D}h{\cal D}C{\cal D}\bar{C} \exp ( -S[\bar{g}+h]-S_{\rm gf}-S_{\rm gh})$. Next

168: we couple $h_{\mu\nu}$ to a source, Legendre transform $\ln Z$ in the usual way, and thus obtain the 

169: effective action $\Gamma[\bar{h};\bar{g}]$. Here $\bar{h}_{\mu\nu}\equiv \langle h_{\mu\nu} \rangle$ is the expectation

170: value of the fluctuation. The corresponding expectation value of the complete metric will be

171: obtained by $g_{\mu\nu}= \langle \gamma_{\mu\nu} \rangle \equiv \bar{g}_{\mu\nu} +\bar{h}_{\mu\nu}$. It is

172: customary to consider $\Gamma$ a functional of $g$ and $\bar{g}$ rather than $\bar{h}$ and $\bar{g}$,

173: and to define $\Gamma[g,\bar{g}]\equiv \Gamma[\bar{h}=g-\bar{g}; \bar{g}]$. As always in the background

174: formalism \cite{mr,adler,ym} the ``ordinary" effective action, the generating functional of one-particle

175: irreducible Green's functions, $\Gamma[g]$, 

176: is obtained by setting $\bar{h}=0$ or $\bar{g}=g$: $\Gamma[g]=\Gamma[g,g]\equiv\Gamma[\bar{h}=0;\bar{g}]$.

177: 

178: As in every field theory, the one-loop correction to the effective action is given by a set 

179: of determinants: 

180: \be\label{1}

181: \Gamma_1[g,\bar{g}] = \frac{1}{2}{\ln \det} \; \widehat{S}^{(2)} -{\ln \det} \; S_{\rm gh}^{(2)}

182: \ee

183: Here we set $\widehat{S}[\bar{h};g]\equiv S[\bar{g}+\bar{h}] +S_{\rm gf}[\bar{h}; \bar{g]}$

184: and $\widehat{S}^{(2)}$ denotes the matrix of the second functional derivatives of 

185: $\widehat{S}$ with respect to $\bar{h}_{\mu\nu}$, and likewise for the ghosts.

186: 

187: In order to evaluate $\ln \det \Omega ={\rm Tr}\ln \Omega$ for $\Omega = \widehat{S}^{(2)}$ and 

188: $S_{\rm gh}^{(2)}$ we use the proper time representation ${\rm Tr} \ln \Omega = -\int_0^\infty (ds/s) \; {\rm Tr }\exp (-s\Omega)$.

189: We regularize the short-distance singularities of this expression by introducing a UV cutoff $\Lambda$ 

190: which suppresses the contributions to the $s$-integral with $s \lesssim \Lambda^{-2}$. Furthermore, in 

191: order to set up a Wilsonian RG equation, we also introduce a IR cutoff $k$ which suppresses the contributions due to 

192: large proper  times $s\gtrsim k^{-2}$. In principle one could employ a sharp proper time cutoff which amounts

193: to replacing $\int_0^\infty ds$ with $\int_{1/\Lambda^2}^{1/k^2} ds$. 

194: In this paper we use instead a more general class of 

195: proper time cutoffs.

196: %\footnote{We do not perform the rescaling $k^2\rightarrow mk^2$ as in \cite{sm}}. 

197: %The cutoff used in the 

198: %present paper coincides with that used in \cite{wf1}}.

199: One writes 

200: ${\rm Tr} [\ln \Omega]_{\rm reg} = -{\rm Tr} \int_0^\infty f_k(s) \exp (-s\Omega)$ where the shape of the function

201: $f_k(s)$ is chosen such that it interpolates smoothly between $f_k(s)\approx 0$ for $s\gg k^{-2}$ and $f_k(s)\approx 1$

202: for $s\ll k^{-2}$, and similarly in the UV. Actually the UV cutoff will play no role in that follows since we shall 

203: concentrate on the $k$-derivative of the regularized determinant. Introducing the ``RG-time" $t=\ln k$ we have

204: \be\label{2}

205: \partial_t {\rm Tr}[\ln \Omega]_{\rm reg} = -{\rm Tr}\int_0^\infty\frac{ds}{s}\; \partial_t f_k(s) \exp (-s\Omega)

206: \ee

207: For an appropriately chosen $f_k(s)$ the scale derivative $\partial_t f_k(s)$ is independent of $\Lambda$.

208: Hence, at the level of (\ref{2}), we may safely perform the limit $\Lambda \rightarrow \infty$. 

209: 

210: %In the present invistigation we are going to use the following one-parameter familty of smooth 

211: For actual calculations  we shall use the following one-parameter family of smooth 

212: cutoffs $f_k\equiv f_k^{m}$ which has been extensively used in the literature \cite{pt},

213: in particular in high precision calculation 

214: of the critical exponents\footnote{We do not 

215: perform the rescaling $k^2\rightarrow mk^2$ as in \cite{sm}.} \cite{wf1}:

216: \be\label{3}

217: f_k^m(s) = \frac{\Gamma(m+1,{\cal Z} sk^2) -\Gamma(m+1,{\cal Z} s\Lambda^2)}{\Gamma(m+1)}

218: \ee 

219: Here $m$ is an arbitrary real, positive parameter which controls the shape of the $f_k^m$ in the 

220: interpolating regions, and $\Gamma(\alpha,x)=\int^\infty_x dt \; t^{\alpha-1} e^{-t}$ denotes the incomplete 

221: Gamma-function. Furthermore, ${\cal Z}$ is a constant (actually a matrix in field space) which 

222: has to be adjusted in the following way.      

223: If the field of type ``$a$" has a kinetic term $-Z_a D^2$, we set ${\cal Z} =Z_a$ for this field 

224: to make sure that the eigenvalues of $-D^2$ are cut off precisely at $k^2$, rather than $k^2/{\cal Z}_a$ \cite{avactrev}.

225: The scale derivative of (\ref{3}) is given by 

226: \be\label{4}

227: \partial_t f_k^m(s) = -2\;\frac{1}{m!}\; ({\cal Z} s k^2)^{m+1} \exp (-{\cal Z} sk^2) 

228: \ee

229: which is indeed independent of $\Lambda$. Inserting (\ref{4}) into (\ref{2}) and performing the integral

230: we obtain

231: \be\label{5}

232: \partial_t {\rm Tr} [{\rm \ln} \Omega]_{\rm reg} = 

233: 2 \; {\rm Tr} \Big ( \frac{{\cal Z}k^2}{\Omega +{\cal Z}k^2} \Big )^{m+1} 

234: \ee

235: For $m$ integer, in the limit $\Lambda \rightarrow \infty$, Eq.(\ref{3}) yields the following explicit 

236: representation:

237: \be\label{6}

238: f_k^m(s) = \exp (-{\cal Z} sk^2 ) \sum_{\mu =0}^{m} \frac{1}{\mu!} \; ({\cal Z} sk^2)^{\mu}

239: \ee

240: 

241: If we regularize the determinants in (\ref{1}) with a proper time cutoff then $\Gamma_1$ and, as a consequence, 

242: the complete one loop effective action becomes $k$-dependent. We denote it by $\widehat{S}_k[g,\bar{g}]\equiv

243: \widehat{S}[\bar{h};\bar{g}] + \Gamma_1[g,\bar{g}]_{\rm reg}$. Its scale derivative is given by

244: \be\label{7}

245: \partial_t \widehat{S}_k[g,\bar{g}] = -\frac{1}{2} {\rm Tr} \int_0^\infty \frac{ds}{s}

246: \; \partial_t f_k^m(s) [\exp(-s\widehat{S}^{(2)}) -2\exp(-s{S}^{(2)}_{\rm gh}) ]

247: \ee 

248: 

249: Up to this point we performed a standard one-loop calculation, with the classical action $\widehat{S}$

250: appearing on the RHS of (\ref{7}), and the quantum corrected $\widehat{S}_k$ on its LHS. 

251: Now we perform a RG-improvement of this one-loop calculation by feeding-back $\widehat{S}_k$ into the RHS

252: of (\ref{7}). This leads to the following ``proper time RG equation" for the functional $\widehat{S}_k$:

253: \be\label{8}

254: \partial_t \widehat{S}_k [g,\bar{g}] = -\frac{1}{2} {\rm Tr} \int_0^{\infty} \frac{ds}{s}\; 

255: \partial_t f_k^m (s) [\exp(-s\widehat{S}^{(2)}_k)-2\exp (-s{S}^{(2)}_{\rm gh} ) ]

256: \ee

257: For a detailed discussion of flow equations of this type we refer to the literature \cite{pt,wf1}.

258: Note that only {\it after} the derivatives implicit in $\widehat{S}_k^{(2)}$

259: have been performed one can set $g=\bar{g}$ in order to obtain the ``ordinary" running action 

260: $S_k[g]\equiv \widehat{S}_k[g,g]$ since these derivatives are to be performed at fixed 

261: $\bar{g}$. 

262: 

263: The above flow equation could be generalized to allow for an RG evolution in the ghost

264: sector. Starting from the standard one-loop effective action 

265: $\Gamma_1[g,\bar{g},\xi,\bar{\xi}]_{\rm reg}$

266: in presence of nonvanishing ghost expectation values 

267: $\xi=\langle C \rangle$ and $\bar{\xi} = \langle \bar{C} \rangle$,

268: and following the same steps as above, we arrive at

269: \be\label{+}

270: \partial_t \widetilde{S}_k [g,\bar{g},\xi,\bar{\xi}]

271: =-\frac{1}{2} \; {\rm STr} \int_0^{\infty} \frac{ds}{s}\;

272: \partial_t f_k^m(s)  \exp(-s\widetilde{S}^{(2)}_k) 

273: \ee

274: The functional

275: $\widetilde{S}_k [g,\bar{g},\xi,\bar{\xi}]$ is a generalization of $\widehat{S}+{S}_{\rm gh}$ at the tree-, and

276: $\widehat{S}+{S}_{\rm gh}+\Gamma_1[g,\bar{g},\xi,\bar{\xi}]_{\rm reg}$ at the one loop level,

277: respectively. In (\ref{+}), $\widetilde{S}^{(2)}_k$ is a supermatrix involving derivatives with respect to

278: $g$, $\xi$ and $\bar{\xi}$, and the supertrace takes care of the relative minus sign between the graviton and

279: the ghost contributions. In the present paper we are not going to use the generalized flow equation (\ref{+}).

280: 

281: \section{The $\beta$-functions of $g$ and $\lambda$} 

282: The RG equation (\ref{8}) describes a flow on the infinite dimensional space of functionals $\widehat{S}[g,\bar{g}]$.

283: We can try to obtain nonperturbative solutions (RG trajectories) by truncating this theory space.

284: In the following we shall consider the Einstein-Hilbert approximation \cite{mr} whose truncated theory space is

285: 2-dimensional. The corresponding ansatz for $\widehat{S}_k$ is parametrized by a $k$-dependent Newton constant

286: $G_k$ and cosmological constant $\bar{\lambda}_k$:

287: \be\label{9}  

288: \widehat{S}_k[g,\bar{g}] =

289: 2\kappa Z_{Nk} \int d^d x\sqrt{g} \{ -R(g) +2\bar{\lambda}_k \}

290: +\kappa^2 Z_{Nk} \int d^d x\sqrt{g} \bar{g}^{\mu\nu} ( {\cal F}_\mu^{\alpha\beta}g_{\alpha\beta})

291: ({\cal F}_\nu^{\rho\sigma}g_{\rho\sigma})

292: \ee

293: The first term on the RHS of (\ref{9}) is the familiar Einstein-Hilbert action in $d$-dimensions with the

294: $k$-dependent couplings. We wrote $G_k\equiv \bar{G}/{Z}_{Nk}$ for the running Newton constant, with a 

295: fixed constant $\bar{G}$, and introduced the convenient abbreviation $\kappa = (32 \bar{G})^{-1/2}$. The

296: second term of (\ref{9}) is the gauge fixing term for the background version of the harmonic coordinate 

297: condition \cite{adler,mr}, with gauge fixing parameter $\alpha=1$. The operator ${\cal F}_\mu^{\alpha\beta}$

298: is given by ${\cal F}_\mu^{\alpha\beta}= \delta_\mu^\beta\bar{g}^{\alpha\gamma}\bar{D}_{\gamma}

299: -\frac{1}{2}\bar{g}^{\alpha\beta}\bar{D}_\mu$ where the covariant derivative $\bar{D}_\mu$ is constructed from 

300: the background metric. Note that ${\cal F}_\mu^{\alpha\beta}\bar{g}_{\alpha\beta}=0$ so that the gauge fixing 

301: term drops from (\ref{9}) upon setting $g=\bar{g}$. The ghost Lagrangian pertaining to this ansatz \cite{mr}

302: is $\propto \bar{C}_\mu {{\cal M}^\mu}_\nu C^\nu$ with the kinetic operator ${\cal M}\propto S_{\rm gh}^{(2)}$

303: given by 

304: \be\label{10}

305: {\cal M}{[g,\bar{g}]^{\mu}}_\nu = \bar{g}^{\mu\rho}\bar{g}^{\sigma\lambda}\bar{D}_\lambda

306: (g_{\rho\nu} D_{\sigma}+g_{\sigma\nu}D_\rho)-\bar{g}^{\rho\sigma}\bar{g}^{\mu\nu}{\bar D}_\lambda

307: g_{\sigma\nu}D_\rho

308: \ee

309: 

310: Next we insert the truncation ansatz into the flow equation (\ref{10}). As a result, it boils down to a coupled 

311: system of ordinary differential equations for the two functions ${Z}_{Nk}$ and $\bar{\lambda}_k$.

312: They are obtained by performing a derivative expansion of the traces on the RHS of (\ref{8}) and then comparing 

313: the coefficients of the two invariants $\int\sqrt{g}$ and $\int\sqrt{g} R$ on both sides of the equation.

314: After having performed the second functional derivatives in $\widehat{S}^{(2)}_k$ we may set $\bar{h}=0$.

315: This leads to 

316: %\be\label{pp}

317: $\widehat{S}^{(2)}_k{[g,g]^{\mu\nu}}_{\rho\sigma} = 2\kappa^2 Z_{Nk}

318: [-{ K^{\mu\nu}}_{\rho\sigma}D^2+{U^{\mu\nu}}_{\rho\sigma}]$

319: %\ee

320: where\\

321: \be\label{11}

322: { K^{\mu\nu}}_{\rho\sigma}= \frac{1}{4}[\delta^\mu_\rho\delta^\nu_\sigma+\delta^\mu_\sigma\delta^\nu_\rho-g^{\mu\nu}g_{\rho\sigma}]

323: \ee

324: and 

325: \ba\label{12}

326: &&{U^{\mu\nu}}_{\rho\sigma} = 

327: \frac{1}{4}[\delta^\mu_\rho\delta^\nu_\sigma+\delta^\mu_\sigma\delta^\nu_\rho-g^{\mu\nu}g_{\rho\sigma}](R-2\bar{\lambda}_k)

328: + \frac{1}{2}[ g^{\mu\nu}R_{\rho\sigma}+g_{\rho\sigma}R^{\mu\nu}]\nonumber\\[2mm]

329: &&-\frac{1}{4}[\delta^\mu_\rho {R^{\nu}}_\sigma+\delta^\mu_\sigma {R^{\nu}}_\rho+\delta^\nu_\rho {R^{\mu}}_\sigma

330: +\delta^\nu_\sigma {R^{\mu}}_\rho]

331: -\frac{1}{2}[{{{R^{\nu}}_\rho}^\mu}_\sigma + {{{R^\nu}_\sigma}^\mu}_\rho]

332: \ea 

333: It is sufficient to retain the terms proportional to $\int \sqrt{g}$ and $\int\sqrt{g} R$ from the

334: derivative expansion of the traces $\rm Tr \; exp (\cdots) $ appearing in (\ref{8}). They can be obtained 

335: straightforwardly by means of standard heat kernel techniques. Since the calculation is similar to

336: (but simpler than) the one in \cite{mr} we omit the details here. Also the subsequent 

337: $s$-integration can be performed easily in terms of Gamma functions. Equating the result to 

338: the LHS of the flow equation,

339: \be\label{13}

340: \partial_t \widehat{S}_k[g,g] =2\kappa^2 \int d^dx \sqrt{g} 

341: [ -R(g)\partial_t Z_{Nk} +2\partial_t (Z_{Nk}\bar{\lambda}_k)]

342: \ee 

343: we obtain two differential equations of the form $\partial_k Z_{Nk}=\cdots$ and $\partial_k(Z_{Nk}\bar{\lambda}_k)=\cdots$ .

344: It is convenient to rewrite them in terms of 

345: the dimensionless Newton constant $g(k)\equiv k^{d-2}G_k\equiv k^{d-2}Z_{Nk}^{-1}\bar{G}$

346: and the dimensionless cosmological constant $\lambda(k)\equiv k^{-2}\bar{\lambda}_k$. This leads to the following

347: system of equations:

348: \begin{subequations}

349: \ba\label{bb}

350: &&\partial_t g = \beta_g(g,\lambda) \equiv [d-2+\eta_N]g\\[2mm]

351: &&\partial_t \lambda = \beta_\lambda (g,\lambda)

352: \ea

353: \end{subequations}

354: The anomalous dimension $\eta_N \equiv -\partial_t {\rm ln} Z_{Nk}$ is given by

355: \be\label{15}

356: \eta_N=8(4\pi)^{1-\frac{d}{2}}\Big [ \frac{d(7-5d)}{24}\; (1-2\lambda)^{\frac{d}{2}-m-2}-\frac{d+6}{6}\Big ]g \; 

357: \frac{\Gamma(m+2-\frac{d}{2})}{\Gamma(m+1)}

358: \ee

359: and the beta-function of $\lambda$ reads

360: \be\label{16}

361: \beta_\lambda= -(2-\eta_N)\lambda +4(4\pi)^{1-\frac{d}{2}}\Big [\frac{d(d+1)}{4}\;(1-2\lambda)^{\frac{d}{2}-m-1}-d\Big ]g 

362: \frac{\Gamma(m+1-\frac{d}{2})}{\Gamma(m+1)}

363: \ee

364: The equations (15) with (\ref{15}) and (\ref{16}) are much simpler than their counterparts from the

365: average action. In particular they do not contain terms (proportional to $\eta_N$) stemming from the 

366: differentiation of the ${\cal Z}$-factors (see ref.\cite{mr}), and they allow for an explicit evaluation

367: of the threshold functions.

368: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

369: \section{universal quantities}

370: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

371: We have calculated the non-trivial fixed point $(g_\ast,\lambda_\ast) \not = 0$ in $d=4$, 

372: implied by the simultaneous vanishing of $\beta_g$ and $\beta_\lambda$ in (15), 

373: and the critical exponents $\theta'$, $\theta''$ \cite{oliver1,frank1} associated to the stability 

374: matrix for the non-trivial fixed point. The results are depicted in Tab.\ref{t1} for various 

375: values of the cut-off parameter $m$. The most important result is that the proper time

376: RG equation, too, predicts a non-Gaussian fixed point. It exists for any value of $m$.

377: The nonuniversal coordinates $\lambda_\ast$ and $g_\ast$ show a significant $m$-dependence, and it is

378: impressive to see how this $m$-dependence cancels out in the product $\lambda_\ast g_\ast$ which 

379: is universal in an exact calculation \cite{frank1}. For every value of $m$, the stability 

380: matrix $(-\partial_i \beta_j)$, $i,j \in \{g,\lambda \}$, has a pair of complex conjugate 

381: eigenvalues $\theta' \pm i \theta''$.  

382: \begin{table}

383: \caption{

384: Fixed point values and critical exponents for various values of the 

385: cutoff  parameter $m$. Note the plateau behavior of the universal 

386: quantities $\theta'$, $\theta''$, and  $\lambda_\ast g_\ast$ as the value of $m$ increases. \label{t1}} 

387: \begin{ruledtabular}

388: \begin{tabular}{cccccc}

389: $m$& $g_\ast$ & $\lambda_\ast$ &  $\lambda_\ast g_\ast$& $\theta'$& $\theta''$ \\

390: \hline

391: 3/2& 0.763 & 0.192 & 0.147 & 2.000 & 1.658\\

392: 2& 1.663 & 0.118 & 0.138 & 1.834 & 1.230\\

393: 3& 1.890 & 0.066 & 0.125 & 1.769 &1.081 \\

394: 4& 2.589 & 0.046 & 0.119 & 1.750 &1.001\\

395: 5& 3.281 & 0.035 & 0.115 & 1.742 & 0.959\\

396: 6& 3.970 & 0.028 & 0.113 & 1.737 &0.934\\

397: 10& 6.718 & 0.016 & 0.108 & 1.729 & 0.886\\

398: 40& 27.271 & 0.0038 & 0.103 & 1.722 & 0.840 \\

399: \end{tabular}

400: \end{ruledtabular}

401: \end{table}

402: Remarkably, those critical exponents and the  

403: product $\lambda_\ast g_\ast$ tend to constant values as $m$ increases; 

404: they form a ``plateau". The universal

405: quantities $\theta'$,$\theta''$, and $\lambda_\ast g_\ast$

406: show excellent stability properties 

407: for this class of regulators, and these values are

408: in complete agreement with the values found in the framework of 

409: the exact flow equation  

410: for the effective average action for gravity \cite{oliver1,frank1}. 

411: In fact, the differences between the proper time and the average action

412: RG equations are of the same order of magnitude as the residual scale dependence 

413: which is present in each of the two approaches and which is due to the truncated

414: theory space.

415: \section{conclusions}

416: We have presented a nonperturbative flow equation for gravity 

417: which employs a smooth cutoff function in the proper time

418: integration. The specific cutoff function employed in this paper   

419: is known to give rise to particularly accurate critical exponents 

420: in several scalar theories. The main result of our analysis is to have shown that

421: the non-Gaussian fixed point exists already in a simple improved 1-loop

422: calculation, an approach less sophisticated than that of the exact RG

423: equation.  

424: 

425: We have then shown that the universal quantities 

426: determined by the non-Gaussian fixed point are in very good agreement with the results from

427: the average action 

428: \cite{oliver1,frank1,oliver2}, and in particular, that they show a very weak dependence

429: on the regulator.  Our result can be taken as  

430: a further, conceptually different indication that the fixed point is rather ``robust" and should survive also

431: in a more general truncation.

432: \bibliography{pt}

433: 

434: \end{document}

435: