1: % The $U(1)$s in the Finite $N$ Limit of Orbifold Field Theories

2: 

3: \documentclass[12pt]{article}

4: \usepackage{epsfig,epic,eepic}

5: \usepackage{amsfonts,amsmath}

6: \usepackage{hyperref}

7: \usepackage{hangcaption}

8: 

9: \renewcommand{\bottomfraction} {0.7}

10: % Set equation numbering to (1.1) format

11: \numberwithin{equation}{section}

12: 

13: \newcommand{\smallfrac}[2] {{\textstyle{\frac{#1}{#2}}}}

14: \newcommand{\tinybox}[1] {\mbox{\tiny {#1}}}

15: \newcommand{\s}[1] {{\scriptstyle{#1}}}

16: 

17: \newcommand{\eqneq}{\hspace{-6pt}=\hspace{-6pt}}

18: \newcommand{\tr} {\mbox{tr}}

19: \newcommand{\diag} {\mbox{diag}}

20: \newcommand{\offdiag} {\mbox{offdiag}}

21: \newcommand{\N} {{\cal N}} % Number of SUSYs

22: \newcommand{\Z} {{\mathbb Z}} % The set of integer numbers

23: \newcommand{\One} {{\bf 1}} % The unit matrix

24: 

25: \newcommand{\Ident} {{\cal I}}

26: \newcommand{\Trace} {{\cal T}}

27: \newcommand{\phicl}[1][] {{\phi^{\text{c}}_{#1}}}

28: 

29: % General definition for inserting figures as math symbols

30: \newcommand{\eepicFig}[2] {\hspace{-6pt}\raisebox{#1}{\input{#2.eepic}}\hspace{-6pt}}

31: \newcommand{\epsFig}[2] {\raisebox{#1}{\psfig{figure=#2.eps}}}

32: 

33: % Basic propogators figures

34: \newcommand{\PFund} {\eepicFig{0pt}{P_Fund}}

35: \newcommand{\PFundI} {\eepicFig{0pt}{P_Fund_I}}

36: \newcommand{\PFundA} {\eepicFig{0pt}{P_Fund_A}}

37: \newcommand{\PFundB} {\eepicFig{0pt}{P_Fund_B}}

38: \newcommand{\PIdent} {\eepicFig{-1pt}{P_Ident}}

39: \newcommand{\PIdentII} {\eepicFig{-1pt}{P_Ident_II}}

40: \newcommand{\PIdentAA} {\eepicFig{-1pt}{P_Ident_AA}}

41: \newcommand{\PIdentIA} {\eepicFig{-1pt}{P_Ident_IA}}

42: \newcommand{\PIdentAB} {\eepicFig{-1pt}{P_Ident_AB}}

43: \newcommand{\PTrace} {\eepicFig{-1pt}{P_Trace}}

44: \newcommand{\PTraceII} {\eepicFig{-1pt}{P_Trace_II}}

45: \newcommand{\PTraceAA} {\eepicFig{-1pt}{P_Trace_AA}}

46: \newcommand{\PTraceSqr} {\eepicFig{-1pt}{P_Trace_Sqr}}

47: \newcommand{\PPhoton} {\eepicFig{-1pt}{P_Photon}}

48: 

49: % Basic vertices figures

50: \newcommand{\VFund} {\eepicFig{-12pt}{V_Fund}}

51: \newcommand{\VFundTrace} {\eepicFig{-12pt}{V_Fund_Trace}}

52: \newcommand{\VAdjoint} {\eepicFig{-12pt}{V_Adjoint}}

53: \newcommand{\VAdjointX} {\eepicFig{-12pt}{V_AdjointX}}

54: \newcommand{\VQuark} {\eepicFig{-12pt}{V_Quark}}

55: \newcommand{\VPhoton} {\eepicFig{-12pt}{V_Photon}}

56: \newcommand{\VBifund} {\eepicFig{-12pt}{V_Bifund}}

57: \newcommand{\VBifundX} {\eepicFig{-12pt}{V_BifundX}}

58: 

59: % s,t,u channel figures

60: \newcommand{\sChannelFund} {\eepicFig{-16pt}{sChannel_Fund}}

61: \newcommand{\sChannelFundX} {\eepicFig{-16pt}{sChannel_FundX}}

62: \newcommand{\tChannelFund} {\eepicFig{-16pt}{tChannel_Fund}}

63: \newcommand{\uChannelFund} {\eepicFig{-16pt}{uChannel_Fund}}

64: \newcommand{\sChannelPhoton} {\epsFig{-16pt}{sChannel_Photon}}

65: \newcommand{\tChannelPhoton} {\epsFig{-16pt}{tChannel_Photon}}

66: \newcommand{\uChannelPhoton} {\epsFig{-16pt}{uChannel_Photon}}

67: 

68: % Propogator vertex figures

69: \newcommand{\PTraceVAdjoint} {\eepicFig{-12pt}{P_Trace_V_Adjoint}}

70: \newcommand{\PTraceVAdjointX} {\eepicFig{-12pt}{P_Trace_V_AdjointX}}

71: \newcommand{\PTraceVBifund} {\eepicFig{-12pt}{P_Trace_V_Bifund}}

72: \newcommand{\PTraceVBifundX} {\eepicFig{-12pt}{P_Trace_V_BifundX}}

73: 

74: % Propogator one loop corrections figures

75: \newcommand{\POneLoopFund} {\eepicFig{-8pt}{P_One_Loop_Fund}}

76: \newcommand{\POneLoopOO} {\eepicFig{-16pt}{P_One_Loop_OO}}

77: \newcommand{\POneLoopXO} {\eepicFig{-16pt}{P_One_Loop_XO}}

78: \newcommand{\POneLoopOX} {\eepicFig{-16pt}{P_One_Loop_OX}}

79: \newcommand{\POneLoopXX} {\eepicFig{-16pt}{P_One_Loop_XX}}

80: \newcommand{\POneLoopBifund} {\epsFig{-16pt}{P_One_Loop_Bifund}}

81: \newcommand{\POneLoopBifundTrace} {\epsFig{-16pt}{P_One_Loop_Bifund_Trace}}

82: \newcommand{\PBifundOneLoop} {\eepicFig{-2pt}{P_Bifund_One_Loop}}

83: \newcommand{\PBifundOneLoopTrace} {\eepicFig{-2pt}{P_Bifund_One_Loop_Trace}}

84: \newcommand{\PBifundOneLoopYukawa} {\epsFig{-16pt}{P_Bifund_One_Loop_Yukawa}}

85: 

86: % Vertex one loop correction figures

87: \newcommand{\VOneLoopOOO} {\eepicFig{-24pt}{V_One_Loop_OOO}}

88: \newcommand{\VOneLoopOXO} {\eepicFig{-24pt}{V_One_Loop_OXO}}

89: \newcommand{\VOneLoopXOX} {\eepicFig{-24pt}{V_One_Loop_XOX}}

90: \newcommand{\VOneLoopXXX} {\eepicFig{-24pt}{V_One_Loop_XXX}}

91: 

92: % Vacuum bubble diagrams (zero point functions)

93: \newcommand{\VacuumOneLoop} {\eepicFig{-20pt}{Vacuum_One_Loop}}

94: \newcommand{\VacuumOneLoopTrace} {\eepicFig{-20pt}{Vacuum_One_Loop_Trace}}

95: \newcommand{\VacuumOneLoopPhoton} {\epsFig{-20pt}{Vacuum_One_Loop_Photon}}

96: \newcommand{\VacuumFiveLoopPhoton} {\epsFig{-38pt}{Vacuum_Five_Loop_Photon}}

97: 

98: \newcommand{\PTwoLoopNonPlanar} {\epsFig{-26pt}{P_Two_Loop_Non_Planar}}

99: \newcommand{\TwoPointOneLoopOX} {\eepicFig{-14pt}{Two_Point_One_Loop_OX}}

100: 

101: \begin{document}

102: 

103: %\begin{titlepage}

104: \thispagestyle{empty}

105: 

106: % Preprint

107: \begin{flushright}

108: TAUP-2624-2000\\

109: {\tt hep-th/0003235}

110: \end{flushright}

111: 

112: % Title

113: \title{The $U(1)$s in the Finite $N$ Limit of Orbifold Field Theories}

114: \bigskip

115: \begin{center}

116: {\LARGE\bf

117: The $U(1)$s in the Finite $N$ Limit of Orbifold Field Theories

118: }

119: \end{center}

120: 

121: % Author

122: \author{Ehud Fuchs}

123: \bigskip

124: \begin{center}

125: {\bf Ehud Fuchs}\footnote{\tt udif@tau.ac.il}

126: \end{center}

127: 

128: % Address

129: \begin{center}

130: {\em School of Physics and Astronomy\\

131:   Beverly and Raymond Sackler Faculty of Exact Sciences\\

132:   Tel Aviv University, Ramat Aviv, 69978, Israel

133: }

134: \end{center}

135: 

136: % Abstract

137: \bigskip

138: \begin{abstract}

139: We study theories generated by orbifolding the $\N=4$ super conformal

140: $U(N)$ Yang Mills theory with finite $N$, focusing on the r\^ole of the

141: remnant $U(1)$ gauge symmetries of the orbifold process.

142: It is well known that the one loop beta functions of the non abelian $SU(N)$

143: gauge couplings vanish in these theories.

144: It is also known that in the large $N$ limit the beta functions vanish to all

145: order in perturbation theory.

146: We show that the beta functions of the non abelian $SU(N)$ gauge couplings

147: vanish to two and three loop order even for finite $N$.

148: This is the result of taking the abelian $U(1)$ of $U(N)=SU(N)\otimes U(1)$

149: into account.

150: However, the abelian $U(1)$ gauge couplings have a non vanishing beta

151: function. Hence, those theories are not conformal for finite $N$.

152: We analyze the renormalization group flow of the orbifold theories,

153: discuss the suppression of the cosmological constant and tackle the

154: hierarchy problem in the non supersymmetric models.

155: \end{abstract}

156: 

157: % Date

158: \vfill

159: March 2000

160: 

161: %\end{titlepage}

162: \setcounter{page}{0}

163: \newpage

164: \setcounter{footnote}{0}

165: 

166: \thispagestyle{empty}

167: \tableofcontents

168: \setcounter{page}{0}

169: \newpage

170: 

171: \section{Introduction}

172: 

173: Supersymmetric Conformal Field Theories (SCFT) in the large $N$ limit

174: have been extensively studied and are very well understood.

175: Both the hierarchy problem and the cosmological constant problem are solved

176: in SCFT.

177: Unfortunately, we live in a non-supersymmetric non-conformal finite $N$ world.

178: Orbifolds of SCFT give us an opportunity to study non-SCFT using our

179: knowledge of SCFT and, hopefully, without loosing all the properties

180: of SCFT.

181: %The differences between the original SCFT and the orbifold theory appear

182: %as $\frac{1}{N}$ corrections.

183: In this paper, we analyze orbifolds of SCFT with finite $N$, focusing

184: on the r\^ole of the $U(1)$ gauge symmetries that the orbifold process

185: leaves us with.

186: %We claim that for finite $N$ it is essential to take the $U(1)$s into

187: %account in order to keep as much of the properties of SCFT as possible.

188: 

189: The large $N$ limit was first introduced by 't Hooft \cite{'tHooft:74}

190: who taught us that planar diagrams dominate the amplitudes of $U(N)$

191: gauge theories in the large $N$ limit.

192: He also noticed the analogy between the topologies of Feynman diagrams

193: and the topologies of strings of the dual string model.

194: 

195: More recently, Maldacena conjectured \cite{Maldacena:98} that there is

196: a correspondence between type IIB string theory on $AdS_5\times S^5$ and

197: four dimensional $\N=4$ $U(N)$ SCFT.

198: In the large $N$ limit it is a correspondence between IIB supergravity

199: and $\N=4$ SCFT. (For a review and references see \cite{Aharony:99}.)

200: 

201: In ``The Wall of the Cave'' \cite{Polyakov:98} Polyakov suggested that

202: non supersymmetric non conformal field theories should be described by

203: type 0 string theory.

204: %%\footnote{Our paper is just another brick in the wall.}

205: The problem with type 0 string theory is that it has a tachyon in the

206: closed string sector.

207: Klebanov and Tseytlin showed in \cite{Klebanov:98} that the coupling of

208: the tachyon to the R-R fields shifts the effective mass of the tachyon

209: and can cure its instability.

210: 

211: In type 0B there is a doubling of the R-R sector.

212: Specifically, the five-form field strength $F_5$ is unconstrained, giving

213: rise to electric and magnetic D3 branes.

214: In \cite{Klebanov:99} the field theory living on $N$ electric and

215: $N$ magnetic D3 branes was first analyzed. It is an $SU(N)\otimes SU(N)$

216: non supersymmetric theory. The gauge coupling one loop $\beta$ function

217: is zero and the two loops $\beta$ function vanishes in the large

218: $N$ limit, suggesting that in the large $N$ limit this is a

219: non-supersymmetric conformal field theory.

220: 

221: We noticed that the two loop $\beta$ function also vanishes for finite

222: $N$ if a diagonal

223: $U(1)$ gauge field with a matching $U(1)$ scalar is included in the model.

224: This observation was the trigger to this paper.

225: 

226: The inclusion of the $U(1)$ fields makes the $SU(N)$ two loop $\beta$

227: function vanish. However, the $\beta$ function of the $U(1)$ gauge is non

228: vanishing already at one loop. Hence, the theory is not conformal

229: for finite $N$.

230: 

231: The $U(N)\otimes U(N)$ model is a $\Z_2$ orbifold of $\N=4$ $U(2N)$

232: super Yang Mills where $\Z_2$ is in

233: the center of the $SU(4)_R$ symmetry \cite{Klebanov:99,Nekrasov:99}.

234: This leads to the possibility that including the $U(1)$ fields in general

235: $\N=4$ orbifolds makes the two loop $\beta$ function vanish.

236: 

237: Orbifolds in the AdS/CFT correspondence where first considered in

238: \cite{Kachru:98}. In string theory the orbifold acts on the

239: $SO(6)\sim SU(4)$ isometry group of $S^5$. In field theory the 

240: orbifold acts on the $SU(4)_R$ symmetry.

241: 

242: Na\"{\i}vely, one expects that the orbifolds will have no effect on the other

243: symmetries of the theory.

244: On the string theory side, this means that the isometry group of $AdS_5$,

245: i.e. $SO(4,2)\sim SU(2,2)$, remains intact.

246: On the field theory side, this means that the conformal group

247: $SO(4,2)$ is not broken, leading to a conformal field theory.

248: 

249: However, the na\"{\i}ve expectation is not realized.

250: The one loop $\beta$ function of the gauge coupling does

251: vanish \cite{Lawrence:98}, But the higher loop corrections vanish only in

252: the large $N$ limit \cite{Vafa:98,Bershadsky:98}.

253: The source of the large $N$ requirement in orbifolds is not obvious from

254: the field theory point of view since the original $\N=4$ theory is conformal

255: also for finite $N$.

256: 

257: We claim that taking

258: ``The $U(1)$s in the finite $N$ limit of Orbifold Field Theories''

259: into account is required for the understanding of the orbifolded theories.

260: The $U(1)$s can be ignored by setting their couplings to zero,

261: yet the vanishing of the two loop $\beta$ functions when the $U(1)$s are

262: taken into account signifies their r\^ole in the orbifold theories.

263: 

264: Orbifold theories with finite $N$ were already analyzed in the literature

265: but without taking the $U(1)$ factors into account.

266: In \cite{Frampton:98}, the conditions for the canceling of the two loop

267: $\beta$ functions were considered.

268: In \cite{Frampton:99} it was suggested that softly broken conformal

269: symmetry could solve the hierarchy problem.

270: In \cite{Csaki:99} the RG flow of the orbifold theories was analyzed.

271: 

272: We start our analysis of finite $N$ theories by presenting in

273: section \ref{sec:Unitary} the

274: double line notation for finite $N$. We find a subset of diagrams that have

275: no subleading corrections in $N$, and entitle them as ``calculable''.

276: 

277: In section \ref{sec:Orbifolds} we present the double line notation for

278: orbifold theories.

279: We claim that it is natural to choose all the coupling in the orbifold

280: theory equal

281: and introduce the concept of a natural line on which all the couplings

282: are equal and related to the original $\N=4$ coupling.

283: We prove the vanishing of the $\beta$ function up to three loops for

284: finite $N$ on the natural line.

285: Our proof is based on the proof for the large $N$

286: limit \cite{Vafa:98,Bershadsky:98}

287: combined with the fact that all diagrams up to three loops are ``calculable''.

288: 

289: In subsection \ref{sec:Hierarchy}  we discuss the scalar mass corrections

290: that vanish up to three loops for most of the scalars. This helps to solve

291: the hierarchy problem in orbifolds that do not have $U(1)$ scalars.

292: In subsection \ref{sec:Cosmological} we discuss the vacuum bubble diagrams

293: that vanish up to four loops. This could have solved the cosmological

294: constant problem if it were not for the running of the $U(1)$ couplings.

295: However, it still leads to a suppressed cosmological constant relative to

296: general non supersymmetric field theories.

297: In subsection \ref{sec:Anomaly} we discuss the $U(1)$ anomalies

298: in the chiral orbifolds.

299: 

300: Since the orbifold theories are not conformal for finite $N$,

301: an analysis of the renormalization group flow is in order.

302: This is done in section \ref{sec:RGFlow}.

303: We start with the easiest case when the orbifold projection leaves us

304: with an $\N=2$ supersymmetry. In this case the $U(1)$ fields are decoupled

305: from the $SU(N)$ fields leaving the $SU(N)$ theory conformal,

306: but strictly speaking, those theories are not conformal because of

307: the running of the $U(1)$ couplings.

308: 

309: For the $\N=1$ orbifolds we use the arguments of \cite{Leigh:95} to analyze

310: the manifold of fixed point. We find only the fixed line found in

311: \cite{Kachru:98} when the $U(1)$ fields decouple and show that the natural

312: line flows to the fixed line in the IR.

313: 

314: For the non-supersymmetric orbifolds, the lack of any non renormalization

315: theorems limits our results to what we can directly calculate.

316: We calculate the effective scalar potential to one loop order in an attempt

317: to check the validity of the orbifold theory.

318: We also calculate the $\beta$ functions to determine the RG flow

319: of the model.

320: 

321: In section \ref{sec:Summary} we summarize our results and 

322: discuss the prospects of generalizing the proof of the vanishing

323: of the $\beta$ function to all orders in perturbation theory.

324: We also point out

325: some open issues and related topics not pursued in this paper.

326: 

327: \section{The Unitary Group in Double Line Notation}

328: \label{sec:Unitary}

329: 

330: \subsection{Elementary Group Theory}

331: 

332: We start with a short presentation of the unitary group in order to

333: introduce the double line notation. The double line notation

334: introduced by 't Hooft \cite{'tHooft:74} gives the leading order behavior

335: in $N$. We present a notation that gives exact results including

336: subleading terms in $N$. Our notation closely resembles Cvitanovi\'c's

337: birdtracking notation \cite{Cvitanovic:76,Cvitanovic:84}.

338: 

339: The unitary group $U(N)$ is the group of unitary transformations on a

340: vector (quark) $q$ with $N$ complex components, leaving

341: $\bar{q}{q}=\delta^i_j q_i q^j$ invariant. The Kronecker delta is the

342: projection operator (propagator) of the defining (fundamental)

343: representation

344: \begin{eqnarray}

345: \s{i}\PFund\s{j} &=& \delta^i_j \ .

346: \label{eqn:PFund}

347: \end{eqnarray}

348: 

349: All invariant tensors can be constructed by products of Kronecker deltas.

350: We are mainly interested in the adjoint representation, since all

351: matter in the $\N=4$ SYM model is in this representation.

352: The adjoint representation is constructed from a quark-antiquark

353: state. There are two invariant tensors for the quark-antiquark

354: state, the identity $\Ident$ and the trace $\Trace$

355: \begin{eqnarray}

356: \s{a}\PPhoton\s{b} \Rightarrow \left\{

357: \begin{array}{rcl}

358: \Ident &=& ^{j_1}_{i_1}\PIdent^{j_2}_{i_2}

359:     = \delta^{i_1}_{i_2}\delta^{j_2}_{j_1} = \delta^{ab} \\

360: \Trace &=& ^{j_1}_{i_1}\PTrace^{j_2}_{i_2}

361:     = \delta^{i_1}_{j_1}\delta^{i_2}_{j_2} \ .

362: \end{array}\right.

363: \end{eqnarray}

364: Where $a,b=1\ldots N^2$ or in other words $a=(^i_j)$.

365: The eigenvalues of the trace matrix can be calculated using

366: the trace tensor equation

367: \begin{eqnarray*}

368: \Trace^2 = \PTraceSqr = N \PTrace =  N\Trace \ .

369: \end{eqnarray*}

370: The roots of the equation are $\lambda_1=N$ and $\lambda_2=0$.

371: With each root we can associate a projection operator

372: $P_i=\frac{\Trace-\lambda_j\Ident}{\lambda_i-\lambda_j}$

373: \begin{eqnarray}

374: P_{SU(N)} = \frac{\Trace-N\Ident}{0-N} &=& \PIdent-\smallfrac{1}{N}\PTrace \ ,

375: \label{eqn:PTraceless}\\

376: P_{U(1)} = \frac{\Trace-0\Ident}{N-0} &=& \smallfrac{1}{N}\PTrace \ .

377: \label{eqn:PTrace}

378: \end{eqnarray}

379: Those projection operators are orthonormal, $P_iP_j=\delta_{ij}P_j$, and

380: complete, $\sum P_i=\Ident$,

381: giving us the $SU(N)$ and the $U(1)$ propagators.

382: 

383: The generator of the defining representation (the quark-antiquark

384: gluon vertex) is

385: \begin{eqnarray}

386: (T^a)^i_j &=& \VQuark = c\VFund \ .

387: \label{eqn:VFund}

388: \end{eqnarray}

389: where $c$ is an overall normalization set by the Dynkin index of the

390: fundamental representation

391: \begin{eqnarray}

392: \tr[T^aT^b] &=& C(F)\delta^{ab} = \nonumber\\

393: c^2\POneLoopFund &=& C(F)\PIdent \\

394: &\Rightarrow& c^2=C(F) \ .\nonumber

395: \end{eqnarray}

396: 

397: The adjoint generator is

398: \begin{eqnarray}

399: (G^a)^{bc}=if^{abc} =

400:    \raisebox{12pt}{$\s{a}$}\mathop{\VPhoton}_b\raisebox{12pt}{$\s{c}$} =

401:    \sqrt{C(F)}\left(\VAdjoint-\VAdjointX\right) \ ,

402: \label{eqn:AdjGen}

403: \end{eqnarray}

404: where we choose the convention that indices are marked counterclockwise.

405: The antisymmetric form of the adjoint generator is required in order to

406: satisfy the Lie algebra for the fundamental generators,

407: \begin{eqnarray}

408: \left[T^a,T^b\right] &=& if^{abc}T^c \ ,\nonumber\\

409: \tChannelFund-\uChannelFund &=& \sChannelFund-\sChannelFundX \ .

410: \end{eqnarray}

411: It is easy to check that \eqref{eqn:AdjGen} also satisfies the Lie algebra for

412: the adjoint generators,

413: \begin{eqnarray}

414: [G^a,G^b]&=&if^{abc}G^c \ ,\nonumber\\

415: \tChannelPhoton-\uChannelPhoton &=& \sChannelPhoton \ .

416: \label{eqn:AdjAlgebra}

417: \end{eqnarray}

418: 

419: In order to calculate Feynman diagrams with gauge invariant external sources

420: we also need the one loop vacuum bubble diagram

421: \begin{eqnarray}

422: \VacuumOneLoopPhoton=

423: \VacuumOneLoop-\smallfrac{1}{N}\VacuumOneLoopTrace=

424: N^2-1=d(G) \ .

425: \label{eqn:vacuum}

426: \end{eqnarray}

427: 

428: Now we have the tools to calculate the group

429: factor of any Feynman diagram. Just replace each vertex with the two

430: vertices in \eqref{eqn:AdjGen}, each $SU(N)$ propagator with \eqref{eqn:PTraceless}

431: and each $U(1)$ propagator with \eqref{eqn:PTrace}, and sum all the

432: diagrams. For a Feynman diagram with $V$ vertices and $P$ $SU(N)$

433: propagators one needs to sum $2^{V+P}$ diagrams. The number of

434: diagrams one needs to sum can be reduced by using the fact that the

435: $U(1)$ propagator decouples from the adjoint vertex \eqref{eqn:AdjGen}

436: \begin{eqnarray}

437: \PTraceVAdjoint-\PTraceVAdjointX=0 \ .

438: \label{eqn:U1Decoupling}

439: \end{eqnarray}

440: Consequently, we can replace each $SU(N)$ propagator with the identity

441: propagator $\Ident$.\footnote{

442: Digressing to non commutative geometry, we point out that in the double

443: line notation it is manifest that the $U(1)$ in non commutative geometry

444: does not decouple from the adjoint vertex because we need to add

445: different phases to each diagram\cite{Bigatti:99}

446: \begin{eqnarray}

447: e^{ip_1\wedge p_2}\PTraceVAdjoint-e^{ip_2\wedge p_1}\PTraceVAdjointX \neq 0\ .

448: \label{eqn:U1NonComm}

449: \end{eqnarray}

450: }

451: 

452: The calculation of the Dynkin index for the adjoint representation

453: comes from the group factor of the one loop correction to the two point

454: function

455: \begin{eqnarray}

456: &\POneLoopOO-\POneLoopXO-\POneLoopOX+\POneLoopXX = &\nonumber\\

457: &2N\left(\PIdent-\smallfrac{1}{N}\PTrace\right) &

458: \label{eqn:PAdjOneLoop}\\

459: &\Rightarrow C(G)=2NC(F) \ .&\nonumber

460: \end{eqnarray}

461: We see that although we used $U(N)$ propagators, we actually calculated

462: the Dynkin index for the $SU(N)$ propagator. This is a result of the

463: $U(1)$ decoupling \eqref{eqn:U1Decoupling}. The Dynkin index for the

464: $U(1)$ propagator is zero.

465: 

466: It is useful (and easy) to calculate the group factor of the one loop

467: correction to the three point function

468: \begin{eqnarray}

469: &\VOneLoopOOO-3\times\left(\VOneLoopOXO-\VOneLoopXOX\right)-\VOneLoopXXX

470: =& \nonumber\\

471: &N\left(\VAdjoint-\VAdjointX\right) \ . &

472: \label{eqn:VAdjOneLoop}

473: \end{eqnarray}

474: The "$3\times$" stands for the three possible permutation of each

475: diagram. Each permutation results in a different diagram, so "$\times$"

476: can not be treated as the multiplicity of the diagram, but this does not

477: matter since the term in the brackets is zero anyway.

478: Equation \eqref{eqn:VAdjOneLoop} tells us that the group factor of the

479: one loop correction to the vertex \eqref{eqn:AdjGen} is

480: $NC(F)=\frac{1}{2}C(G)$.

481: 

482: \subsection{'t Hooft Large $N$ Limit}

483: 

484: Before proceeding to higher loop diagrams, we wish to recall 't Hooft

485: results for the large $N$ limit \cite{'tHooft:74}.

486: Using the double line notation we can get the $N$ dependence of a

487: Feynman diagram with adjoint fields from topological considerations.

488: A connected diagram with

489: $V=V_3+V_4$ vertices, $E=\frac{1}{2}(3V_3+4V_4)$ edges (propagators)

490: and $F$ faces (closed lines in the double line notation) has a

491: group coefficient proportional to

492: \begin{eqnarray}

493: g_{YM}^{V_3+2V_4}N^{F}=\lambda^{E-V}N^{\chi} \ ,

494: \label{eqn:planar}

495: \end{eqnarray}

496: where $\chi\equiv V-E+F=2-2g$ is the Euler characteristic and $g$ is the

497: genus of the surface defined by the double line diagram with all the

498: faces shrunk to a point.

499: Each Feynman diagram is translated into a number of diagrams in

500: the double line notation which can have different genera.

501: The leading $N$ contribution comes from the double line

502: diagrams with the minimal genus. The 't Hooft limit is defined by taking

503: $N$ to infinity while leaving $\lambda=g^2N$ fixed.

504: 

505: \subsection{``Calculable'' Diagrams}

506: 

507: For some diagrams we can calculate not only the leading $N$ contribution,

508: but the exact $N$ dependence using \eqref{eqn:AdjAlgebra},

509: \eqref{eqn:PAdjOneLoop} and \eqref{eqn:VAdjOneLoop}.

510: We will refer to diagrams that can be thus calculated

511: as ``calculable'' (all diagrams are calculable but the ``calculable'' ones

512: are easily so).

513: We now show generally that 

514: {\em any ``calculable'' $L$-loop Feynman diagram of adjoint fields has a

515:   group factor proportional to

516: }

517: \begin{eqnarray}

518: g_{YM}^{V_3+2V_4} N^{L-1}(N^2-1)=\lambda^{E-V}(N^2-1) \ .

519: \label{eqn:Lloop}

520: \end{eqnarray}

521: {\em  with no subleading corrections in $N$.

522: }

523: Here $L$ is the number of loops in the Feynman diagram. It can be defined as

524: the number of momentum loops needed to be integrated over in the vacuum

525: bubble diagram.

526: The factor of $(N^2-1)$ comes from the one loop vacuum

527: bubble diagram \eqref{eqn:vacuum} that has $L=1$.

528: For every two

529: 3-point vertices and for every 4-point vertex we add to the diagram

530: we get one more loop, therefore $L=1+\frac{1}{2}V_3+V_4$.

531: 

532: When calculating the group factor of the diagram, each time

533: \eqref{eqn:PAdjOneLoop} or \eqref{eqn:VAdjOneLoop} is used, a factor of $N$

534: is added and a loop is removed. \eqref{eqn:AdjAlgebra} does not change neither

535: the power of $N$ nor the number of loops. Hence, after using

536: \eqref{eqn:PAdjOneLoop} or

537: \eqref{eqn:VAdjOneLoop} $L-1$ times, we get the one loop vacuum bubble

538: diagram \eqref{eqn:vacuum} with a factor of $N^{L-1}$.

539: 

540: We assumed that each four point vertex has the group structure of two

541: 3-point vertices. This is not true in general, but it is true for

542: the $\N=4$ model.

543: 

544: Equation \eqref{eqn:planar} tells us that the leading order

545: contribution in $N$ comes from the diagram with the minimal genus.

546: Equation \eqref{eqn:Lloop} tells us that for ``calculable'' diagrams the

547: exact $N$ dependence is $(N^2-1)$.

548: The two equations can match only for diagrams with genus zero.

549: Hence we conclude that

550: {\em non-planar ``calculable'' diagrams have a vanishing group factor.

551: }

552: For example the non planar two loop correction to the photon is

553: ``calculable'' which means that

554: \begin{eqnarray}

555: \PTwoLoopNonPlanar=0 \ .

556: \label{eqn:NonPlanar}

557: \end{eqnarray}

558: The first ``incalculable'' vacuum diagram is the five loop vacuum

559: bubble diagram

560: \begin{eqnarray}

561: &\VacuumFiveLoopPhoton=&

562: \label{eqn:5Loops}\\

563: &2C(F)^4(N^4+12N^2)\VacuumOneLoopPhoton=

564: 2C(F)^4(N^4+12N^2)d(G) \ .& \nonumber

565: \end{eqnarray}

566: It is ``incalculable'' because in each loop there are four vertices and

567: we can not change this with the use of \eqref{eqn:AdjAlgebra}.

568: We calculated the group factor of this ``incalculable'' diagram by

569: summing up all the $2^V=256$ double line diagrams.

570: 

571: The first ``incalculable'' propagator diagram is the four loop diagram

572: obtained by cutting out a propagator from \eqref{eqn:5Loops}.

573: The first ``incalculable'' three vertex diagram is the three loop

574: diagram obtained by cutting out a vertex from \eqref{eqn:5Loops}.

575: 

576: \section{Orbifolds in the Double Line Notation}

577: \label{sec:Orbifolds}

578: 

579: \subsection{The Orbifold Process}

580: 

581: The orbifold of $\N=4$ SYM is defined by a discrete subgroup

582: $\Gamma$ of the global $R$ symmetry group $SU(4)_R$.

583: The action of the orbifold on the gauge group $U(|\Gamma|N)$ is

584: defined by the $\gamma$ matrices

585: \begin{eqnarray*}

586: g\in\Gamma &:& g\rightarrow\gamma_g \ ,

587: \end{eqnarray*}

588: where $\gamma_g$ are $(|\Gamma|\times|\Gamma|)\otimes\One_{N\times N}$

589: matrices with $\gamma_1=\One_{|\Gamma|\times|\Gamma|}\otimes\One_{N\times N}=

590: \One_{|\Gamma|N\times |\Gamma|N}$.

591: The orbifold breaks the gauge group into $U(N)^{|\Gamma|}$.

592: For simplicity we assumed that all irreducible representations of $\Gamma$

593: are one dimensional, namely that $\Gamma$ is abelian.

594: For a general discrete group $\Gamma$ with irreducible representations

595: labeled $r_i$,

596: the orbifold breaks the gauge group $U(\sum_i d_i N)$

597: into the $\bigotimes_i U(d_i N)$ gauge group where $d_i=\dim r_i$.

598: 

599: The cancellation of tadpoles in string theory imposes that the

600: representation of $\Gamma$ has to be regular \cite{Vafa:98},

601: meaning $\tr[\gamma_g]=0$ $\forall g\neq 1$. The regularity of $\Gamma$

602: guarantees the cancellation of the $SU(N)$ gauge anomalies, though it is not

603: a necessary condition.

604: 

605: The spectrum of the orbifold theory is defined by the projection operator

606: \cite{Bershadsky:98}

607: \begin{eqnarray}

608: P=\frac{1}{|\Gamma|}\sum_{g\in\Gamma}

609:   r_g\otimes\gamma_g^{\dagger}\otimes\gamma_g \ ,

610: \label{eqn:OrbifoldProj}

611: \end{eqnarray}

612: where $r_g$ is the representation of the projected field under

613: the $SU(4)_R$ symmetry group. The projection leaves only fields that

614: are invariant under the orbifold $\Gamma$, i.e.

615: \begin{eqnarray*}

616: A_{\mu}=\gamma_g^{\dagger}A_{\mu}\gamma_g && \forall g\in \Gamma \ ,\\

617: \phi_I=\gamma_g^{\dagger}(r^6_g)^J_I\phi_J\gamma_g && \s{I,J=1\ldots6} \ ,\\

618: \psi_I=\gamma_g^{\dagger}(r^4_g)^J_I\psi_J\gamma_g

619: && \s{I,J=1\ldots4} \ .

620: \end{eqnarray*}

621: 

622: For the double line notation we split the fundamental propagator

623: \eqref{eqn:PFund} into $|\Gamma|$ parts

624: \begin{eqnarray}

625: \delta^i_j &=& \delta^k_l\delta^{i_k}_{j_l} 

626:                    \ \ \ \ \s{k,l=0\ldots|\Gamma|-1} \ ,\nonumber\\

627: \s{i}\PFund\s{j} &=& \diag(\s{i_0}\PFundI\s{j_0},

628: \s{i_1}\PFundA\s{j_1},\s{i_2}\PFundB\s{j_2},\ldots) \ .

629: \end{eqnarray}

630: The action of the projection operator \eqref{eqn:OrbifoldProj} on the

631: adjoint propagator depends on

632: the group element $g$ under which the propagator transforms,

633: \begin{eqnarray}

634: %\lefteqn{\PIdent-\smallfrac{1}{|\Gamma|N}\PTrace \rightarrow } && \nonumber\\

635: g_0=1:\ 

636:     \delta^l_k(\Ident-\smallfrac{1}{|\Gamma|N}\Trace)

637:         &\eqneq&

638:     \diag(\PIdentII-\smallfrac{1}{N}\PTraceII, \nonumber\\

639:    &&\phantom{\diag}\ \PIdentAA-\smallfrac{1}{N}\PTraceAA,\ldots) \ ,\\

640:     \delta^l_k(\smallfrac{1}{|\Gamma|N}\Trace)

641:         &\eqneq&

642:     \diag(\smallfrac{1}{N}\PTraceII,\smallfrac{1}{N}\PTraceAA,\ldots) \ ,\\

643: g_{k'}=e^{\frac{2\pi i}{|\Gamma|}k'}:\ 

644:     \delta^l_{k+k'}(\Ident-\smallfrac{1}{|\Gamma|N}\Trace)

645:         &\eqneq&

646:     \offdiag_{k'}(\PIdentIA,\PIdentAB,\ldots) \ ,

647: \label{eqn:BiFundProp}

648: \end{eqnarray}

649: where $k'=1\ldots|\Gamma|-1$.

650: The $\Trace$ propagator is multiplied by a factor of $|\Gamma|$

651: because

652: $\delta^l_k\delta^k_l\delta^{i_k}_{j_l}=|\Gamma|\delta^{i_k}_{j_l}$.

653: We see that fields that are invariant under $SU(4)_R$ are in adjoint

654: representations of 

655: ${\displaystyle\left(\vphantom{1^1}SU(N)\otimes U(1)\right)}^{|\Gamma|}$,

656: while non-invariant fields are in bifundamental representations.

657: $\offdiag_{k'}$ means that the nonzero elements are shifted $k'$ places

658: off the diagonal.

659: 

660: In \eqref{eqn:BiFundProp} we assumed that the orbifold group is

661: $\Gamma=\Z_{|\Gamma|}$. For other groups we would have a different

662: representation for $g_{k'}$ and would get the bifundamental

663: fields in different permutations.

664: 

665: The embedding of $\Z_{|\Gamma|}$ in $SU(4)_R$ can be specified using four

666: integer weights $(k_1,k_2,k_3,k_4)$ describing how $\Z_{|\Gamma|}$ operates

667: on the fundamental representation ${\bf 4}$ of $SU(4)_R$. Because we

668: are interested in subgroups of $SU(4)$ and not of $U(4)$, we require

669: $k_1+k_2+k_3+k_4=0 \mod |\Gamma|$.

670: Consequently, the embedding of $\Z_{|\Gamma|}$ is

671: parameterized by three integer numbers. The transformation of the

672: antisymmetric representation ${\bf 6}$ is described by the

673: six integers $k_{(i,j)}=k_i+k_j$, where $(i,j)$ are $(^4_2)$ unordered

674: pairs.

675: 

676: To describe a general orbifold group, not necessarily abelian, we can

677: use quiver diagrams \cite{Douglas:96} with a node for each irreducible

678: representation of $\Gamma$ ($|\Gamma|$ nodes for abelian groups).

679: The diagrams have

680: four directed links from each node describing the four fermions $(\psi)$,

681: four directed links going into each node $(\bar{\psi})$,

682: three undirected (double directed) links from each node describing the

683: six real scalars (three complex scalars) and one undirected link going

684: from each node to itself describing the gauge field in the adjoint

685: representation.

686: 

687: The effect of the orbifold on the three vertex \eqref{eqn:AdjGen} is

688: described by three integers that must satisfy $k_1+k_2+k_3=0 \mod |\Gamma|$

689: (a closed loop in the quiver language), since the vertex is

690: $SU(4)_R$ invariant. If we choose $k_1=-k_2\equiv k$, $k_3=0$,

691: we get the bifundamental generator

692: \begin{eqnarray}

693: (T^{a_l})^{i_l}_{j_l}\delta^{j_{l+k}}_{i_{l+k}} &-&

694: (T^{a_l})^{j_l}_{i_l}\delta^{i_{l-k}}_{j_{l-k}}= \nonumber\\

695: \sqrt{C(F)}\left(

696: \raisebox{12pt}{$^{i_{l+k}}_{i_l}$}\mathop{\VBifund}_{a_l}

697: \raisebox{12pt}{$^{j_{l+k}}_{j_l}$}\right. &-& \left.

698: \raisebox{12pt}{$^{i_l}_{i_{l-k}}$}\mathop{\VBifundX}_{a_l}

699: \raisebox{12pt}{$^{j_l}_{j_{l-k}}$}

700: \right) \ .

701: \label{eqn:BiGen}

702: \end{eqnarray}

703: We can read from \eqref{eqn:BiGen} that an adjoint in $SU(N)_l$ couples

704: to a couple of bifundamentals, $(N_l,\overline{N}_{l+k})$ and

705: $(N_{l-k},\overline{N}_l)$. For $k_1,k_2,k_3\neq0$ we get vertices of three

706: bifundamental. Those vertices exist only for some orbifolds, those that

707: have a triangle with all vertices on different nodes of the quiver diagram.

708: 

709: It is obvious that the $U(1)$ factors do not decouple any more,

710: \begin{eqnarray}

711: \PTraceVBifund-\PTraceVBifundX \neq 0 \ .

712: \label{eqn:U1Coupling}

713: \end{eqnarray}

714: The $|\Gamma|$ $U(1)$ factors of the orbifold theory are not independent,

715: as the sum over the $U(1)$ charges for each field is zero.

716: This is a consequence of the decoupling of the $U(1)$ from the original

717: $U(|\Gamma|N)$ theory \eqref{eqn:U1Decoupling}.

718: 

719: The Dynkin index of the bifundamental representation is

720: \begin{eqnarray}

721: &\POneLoopBifund-\smallfrac{1}{N}\POneLoopBifundTrace =

722: N\left(\PIdent-\smallfrac{1}{N}\PTrace\right) & \nonumber\\

723: &\Rightarrow C^{SU(N)}(B)=NC(F) \ ,&

724: \label{eqn:POneLoopBiFund}\\

725: &\smallfrac{1}{N}\POneLoopBifundTrace =

726: N\left(\smallfrac{1}{N}\PTrace\right) & \nonumber\\

727: &\Rightarrow C^{U(1)}(B)=NC(F) \ .&

728: \label{eqn:POneLoopBiFundU1}

729: \end{eqnarray}

730: The $N$ dependence of the $U(1)$ Dynkin index comes from the multiplicity

731: of the bifundamental representation.

732: Diagrams with one twisted vertex are projected out in the orbifold process

733: \eqref{eqn:OrbifoldProj}. Diagrams with two twisted vertices will give the

734: Dynkin index of the anti-bifundamental representation $C(\bar{B})=NC(F)$.

735: One can see that we no longer have the luxury of ignoring the trace

736: propagator $\Trace$.

737: 

738: The second Casimir $(T^aT^a)^i_j=C_2(F)\delta^i_j$ of the bifundamental

739: representation is calculated from the one loop correction to the

740: bifundamental propagator,

741: \begin{eqnarray}

742: &\PBifundOneLoop-\smallfrac{1}{N}\PBifundOneLoopTrace

743:  = \smallfrac{N^2-1}{N}\PIdentIA & \nonumber\\

744: &\Rightarrow C^{SU(N)}_2(B)=\smallfrac{N^2-1}{N}C(F) \ ,&

745: \label{eqn:PBiOneLoop}\\

746: &\smallfrac{1}{N}\PBifundOneLoopTrace

747:  = \smallfrac{1}{N}\PIdentIA & \nonumber\\

748: &\Rightarrow C^{U(1)}_2(B)=\smallfrac{1}{N}C(F) \ .&

749: \label{eqn:PBiOneLoopU1}

750: \end{eqnarray}

751: 

752: The gauge group is 

753: $\displaystyle\left(SU(N)\otimes U(1)\right)^{|\Gamma|}$ meaning that for

754: every $SU(N)$ second Casimir there is a $U(1)$ second Casimir. The

755: contribution of the one loop bifundamental propagator to the Feynman

756: diagram will alway be of the form

757: \begin{eqnarray}

758: \left(g_N^2 \smallfrac{N^2-1}{N} + g_1^2 \smallfrac{1}{N}\right)C(F) =

759: \left(g_N^2 N + \frac{g_1^2-g_N^2}{N} \right)C(F) \ ,

760: \label{eqn:C2}

761: \end{eqnarray}

762: where $g_N$ is the $SU(N)$ gauge coupling and $g_1$ is the $U(1)$

763: gauge coupling.

764: From \eqref{eqn:C2} we see that the $U(1)$ factor can be neglected in the

765: large $N$ limit, in the 't Hooft limit it is suppressed by a factor of

766: $\frac{1}{N^2}$.

767: We also see that if we choose $g_N=g_1$, the subleading corrections in

768: $N$ are canceled.

769: This is the natural choice since we originally had a $U(N)$

770: symmetry that was split by the RG flow to $SU(N)\otimes U(1)$.

771: 

772: \subsection{The Natural Line}

773: 

774: The orbifold theory has $2|\Gamma|$ gauge couplings.

775: In the space of gauge couplings we choose the two dimensional

776: manifold parametrized by $(g_N,g_1)$ for which all

777: the $SU(N)$ couplings are the same and all the $U(1)$ couplings are

778: the same. It is the natural choice since we originally had a

779: $U(|\Gamma|N)$ symmetry.

780: For the $\Z_{|\Gamma|}$ orbifold this manifold has a

781: $\Z_{|\Gamma|}$ symmetry and all the RG equations have a

782: $\Z_{|\Gamma|}$ symmetry. Accordingly, if we start in this manifold, we

783: will stay in it.

784: We also choose the Yukawa couplings to be equal to the gauge coupling

785: and the quartic couplings to be equal to the gauge coupling squared.

786: The RG flow can make those couplings different.

787: 

788: In the space of couplings, we choose to start the RG flow from a point

789: on a one dimensional manifold (line) parametrized by

790: the coupling $g$ to which all the couplings are equal at some renormalization

791: scale $\mu_N$. $g$ can be related to the coupling of the original $\N=4$

792: theory.

793: This is the natural submanifold to choose because of the $\N=4$ origin

794: of the orbifold theory and we will refer to this submanifold as the

795: {\em natural line}.

796: 

797: In view of the AdS/CFT correspondence, $g$ can be related to the string

798: coupling $g^2\sim g_s$.

799: There are two scales in the field theory orbifold, the regularization scale

800: $\Lambda$ and the renormalization scale $\mu_N$. In the AdS/CFT correspondence

801: the regularization scale is related to the string scale

802: $\Lambda\sim \frac{1}{\sqrt{\alpha'}}$, and the renormalization scale is

803: related to the $AdS_5$ fifth coordinate $\mu\sim U=\frac{r}{\alpha'}$.

804: Our model is not conformal, therefore we do not expect an AdS geometry.

805: The renormalization scale $\mu_N$ where all the couplings are equal, should

806: be related to some unique $U_N$ in the new geometry. From dimensional

807: consideration it should also be somehow related to the string scale.

808: The regularization scale in field theory is not physical, but we have to

809: take it into consideration when we discuss the hierarchy problem and the

810: cosmological constant problem.

811: 

812: In (\ref{eqn:POneLoopBiFund}-\ref{eqn:PBiOneLoopU1}) we calculated diagrams

813: obtained by orbifold projections of the one loop diagram

814: \eqref{eqn:PAdjOneLoop}.

815: We found out that their group factor is proportional to $g^2N$ with no

816: subleading corrections on the natural line.

817: The only other non trivial orbifold projection of \eqref{eqn:PAdjOneLoop} is

818: \begin{eqnarray}

819: \PBifundOneLoopYukawa=N\PIdentIA \ .

820: \end{eqnarray}

821: The vertices in this diagram can only be Yukawa vertices and

822: since the Yukawa couplings are equal to the gauge couplings on the natural

823: line, this diagram also has a group factor of $g^2N$.

824: 

825: The orbifold projections of the one loop

826: correction to the three point vertex \eqref{eqn:VAdjOneLoop}

827: also have a group factor proportional to $g^2N$ with no subleading corrections

828: in $N$, when all the couplings are equal to $g$.

829: ``Calculable'' diagrams were defined as diagrams that can be calculated

830: using \eqref{eqn:PAdjOneLoop}, \eqref{eqn:VAdjOneLoop} together with

831: \eqref{eqn:AdjAlgebra}. Hence, we conclude that

832: {\em ``calculable'' diagrams of orbifold theories have no subleading

833:   corrections in $N$ on the natural line.

834: }

835: The group factor of those diagrams is the same as in \eqref{eqn:Lloop}.

836: 

837: \subsection{Vanishing of the $\beta$ Functions}

838: 

839: In \cite{Vafa:98,Bershadsky:98} it was shown that in the large $N$

840: limit the correlation

841: functions of the orbifold theories coincide with those of $\N=4$.

842: This leads to the vanishing of the $\beta$ function of the

843: orbifold theories to all orders in perturbation theory in the large

844: $N$ limit.

845: We want to generalize the proof for finite $N$, but not to all orders,

846: only to orders for which all diagrams in that order are ``calculable''.

847: 

848: The proof in \cite{Vafa:98,Bershadsky:98} was for planar diagrams with

849: all external legs attached to the same boundary.

850: The fact that ``calculable'' diagrams have no subleading corrections

851: on the natural line leads to the conclusion that non planar ``calculable''

852: diagrams have a vanishing contribution.

853: For example, all orbifold projections of \eqref{eqn:NonPlanar} will vanish

854: on the natural line.

855: If the external legs are attached to different boundaries there are

856: several possibilities \cite{Vafa:98}

857: \begin{itemize}

858: \item For two point functions, each leg is attached to a boundary of

859:   itself. The color indices of the leg are traced, meaning that they

860:   are $U(1)$ legs. This is the source of the running of the $U(1)$

861:   coupling constant

862: \begin{eqnarray*}

863: \TwoPointOneLoopOX \ .

864: \end{eqnarray*}

865: \item For three point functions, one of the external

866:   legs must be attached to a boundary of itself. This is the source of the

867:   running of the Yukawa coupling of the $U(1)$ fields.

868: \item For four point functions, the previous argument does not

869:   apply and we have $SU(N)$ diagrams.

870:   Those diagrams have a different group structure from that of the original

871:   $\N=4$ quartic couplings. In orbifolds with at least $\N=1$ supersymmetry,

872:   the perturbative non renormalization of the superpotential guarantees that

873:   new quartic couplings will not be generated. In $\N=0$ orbifolds, new

874:   quartic couplings are generated as we shall see in the next section.

875: \item Five or more point functions can not be generated because we are dealing

876:   with renormalizable field theories.

877: \end{itemize}

878: 

879: Two point functions are ``calculable'' up to three loops, and three

880: point functions are ``calculable'' up to two loops, hence there are no

881: subleading corrections in $N$ on the natural line and the corresponding

882: diagrams coincide with the $\N=4$ ones. We conclude that

883: {\em on the natural line the two point functions of the non abelian fields

884:   have a zero $\beta$ function up to three loops

885:   and the three point functions of the non abelian fields have a zero

886:   $\beta$ function up to two loops.

887: }

888: 

889: The orbifold theories on the natural line are not finite because the

890: $\beta$ functions of the $U(1)$ couplings are non-zero already at one loop,

891: which may cause the orbifold theories to flow away from the natural line.

892: To see the flow of the $SU(N)$ couplings one has to calculate the second

893: derivative of the coupling, since the first derivative (the $\beta$

894: function) is zero on the natural line. Since the first derivative is

895: a function of the $U(1)$ coupling, the second derivative will depend

896: on the $U(1)$ $\beta$ function.

897: 

898: \subsection{The Hierarchy Problem}

899: \label{sec:Hierarchy}

900: 

901: Generally in field theory, the scalar

902: two point functions diverge quadratically leading to scalar masses of the

903: order of $\Lambda^2$, where $\Lambda$ is some cutoff scale. To keep the 

904: scalars light, mass counterterms must be very fine tuned. This is called the 

905: hierarchy problem.

906: 

907: In supersymmetric theories, the mass of the scalars is protected by

908: the non renormalization of the superpotential, solving the hierarchy

909: problem.

910: In non supersymmetric orbifolds, a mass counterterm is not needed for most of

911: the scalar fields, at least up to three loops, because of the vanishing of

912: the scalar two point function.

913: This is again a result of the matching between the $\N=4$ diagrams and

914: orbifold diagrams for the scalar two point functions.

915: This could have helped to solve the hierarchy problem for non supersymmetric

916: theories, but this matching does not work for the $U(1)$ scalars which are

917: diverging already at one loop.

918: 

919: The problem of the diverging $U(1)$ scalar mass is not general to all $\N=0$

920: orbifolds. There are $\N=0$ orbifolds with no scalars in the adjoint

921: representation and hence no $U(1)$ scalars, e.g., the $\Z_4$ orbifold

922: with weights $(1,1,1,1)$.

923: But in those theories the $U(1)$ gauge symmetries have anomalies as is

924: discussed is subsection \ref{sec:Anomaly}

925: 

926: The divergence of the $U(1)$ scalar mass and the cancellation of the other

927: scalar masses are demonstrated in subsection \ref{sec:N0} for the $Z_2$

928: non supersymmetric orbifold, using the effective potential formalism.

929: 

930: The mass of the $U(1)$ scalar depends on the regularization scheme.

931: If we claim that field theory is related to string theory by the AdS/CFT

932: correspondence, the scheme we should choose is adding massive

933: fields corresponding to the massive open strings between D3 branes in the

934: orbifolded Type IIB string theory.

935: Those massive fields will act as a cutoff.

936: We can hope that in this scheme the $U(1)$ scalars will be massless.

937: 

938: In \cite{Frampton:99} it was already suggested that non supersymmetric

939: conformal field theories with a softly broken conformal symmetry may

940: solve the hierarchy problem. We suggest that the $U(1)$ couplings behave

941: as naturally occurring soft symmetry breaking terms of the conformal

942: symmetry in the sense that the flow of the $U(1)$ couplings induces the

943: flow of the other couplings. The ``soft breaking'' parameter is

944: $\frac{1}{N}$.

945: 

946: In \cite{Csaki:99} the $U(1)$ factors were not taken into account resulting in

947: a mass term for all scalar fields suppressed by a factor of $\frac{1}{N}$.

948: There it was suggested to solve the hierarchy problem by choosing a

949: {\em very large} but finite $N$.

950: Notice that when the $U(1)$ factors are taken into account the scalar mass

951: vanish, but only at the renormalization scale $\mu_N$. At other scales

952: we will get $\Lambda^2$ contributions again.

953: 

954: \subsection{The Cosmological Constant}

955: \label{sec:Cosmological}

956: 

957: The vacuum energy in field theories is generally of the order of

958: $\Lambda^4$, where $\Lambda$ is some cutoff scale. In field theory

959: the constant shift of the vacuum energy is unobservable. In general

960: relativity the vacuum energy plays the r\^ole of the cosmological

961: constant. The cosmological constant is expected to be very small while

962: $\Lambda^4$ is very large. This is called the cosmological constant problem.

963: 

964: The contributions to the cosmological constant come from the zero point

965: functions (vacuum bubble diagrams). To one loop order, bosons and fermions of

966: the same mass have equal and opposite contributions to the cosmological

967: constant.

968: In orbifold theories the cosmological constant vanishes to one loop because

969: the number of bosons and fermions is the same.

970: In non supersymmetric orbifolds this would not be true without taking the

971: $U(1)$ factors into account.

972: 

973: We can go further than that in the loop expansion.

974: We can use the vanishing of the vacuum energy in the $\N=4$ theory to

975: conclude that the vacuum bubble diagrams of orbifold theories vanish on the

976: natural line up to four loops. (In five loops we have the ``incalculable''

977: bubble diagram \eqref{eqn:5Loops}.) This leads to a suppression of the

978: cosmological constant by a factor of $g^8$.

979: 

980: In \cite{Kumar:98} it was suggested that

981: the cosmological constant in orbifold theories vanishes in cases where one

982: has a non supersymmetric fixed line. In our finite $N$ non supersymmetric

983: models we have found no fixed line, therefore we do not expect that the

984: cosmological constant would vanish.

985: 

986: \subsection{$U(1)$ Anomalies}

987: \label{sec:Anomaly}

988: 

989: The regularity of the orbifold guarantees the cancellation of the $SU(N)$

990: gauge anomalies, but not of the $U(1)$ anomalies. The $U(1)$ gauge symmetries

991: are anomalous in the chiral orbifolds.

992: 

993: $Z_{|\Gamma|}$ orbifolds are not chiral if their weights are of the form

994: \begin{eqnarray}

995: (k_1,-k_1,k_2,-k_2) \ .

996: \label{eqn:NonChiral}

997: \end{eqnarray}

998: This means that $\N=2$ orbifolds are not chiral,

999: $\N=1$ orbifolds are chiral and $\N=0$ orbifolds can be either chiral or 

1000: not chiral.

1001: 

1002: In the chiral orbifolds all the $U(1)$s are anomalous and there are no

1003: non anomalous combination except for the trivial $U(1)$ which is completely

1004: decoupled from the rest of the fields. An example of an anomalous orbifold

1005: is given in subsection \ref{sec:N1}.

1006: 

1007: It was shown in \cite{Ibanez:98} that those $U(1)$ anomalies cancel by

1008: a generalized Green-Schwarz mechanism. Fayet-Illiopoulos terms are generated

1009: for the anomalous $U(1)$ gauge fields giving them mass of the order of the

1010: string scale.

1011: 

1012: From the form of the non chiral orbifold \eqref{eqn:NonChiral} it can be

1013: seen that in non chiral orbifold there are scalar $U(1)$s. This means that

1014: in the orbifold theories with no anomalies there will be scalars that can

1015: acquire mass as was discussed in subsection \ref{sec:Hierarchy}.

1016: 

1017: \section{The Renormalization Group Flow of Orbifold Theories}

1018: \label{sec:RGFlow}

1019: 

1020: \subsection{General $\beta$ Functions}

1021: 

1022: Now we want to analyze the RG flow from the natural line on which the

1023: couplings reside

1024: at the renormalization scale $\mu_N$. We have the freedom to choose the

1025: coupling we start with, so we can rely on perturbation theory by choosing

1026: small $g$. We are also taking all the masses to zero at the renormalization

1027: scale $\mu_N$.

1028: %In the non supersymmetric theories, the fields can acquire masses, so our

1029: %analysis will be valid only as long as the energy scale is much larger

1030: %than those masses.

1031: 

1032: For our analysis we calculate the gauge coupling $\beta$ function

1033: up to two loop order because the first loop correction vanishes.

1034: The Yukawa coupling $\beta$ function is calculated up to one loop.

1035: We do not need to calculate the quartic coupling $\beta$ function because

1036: the quartic coupling does not participate in the evolution of the gauge

1037: and Yukawa couplings at the orders we are looking at.

1038: The quartic coupling $\beta$ function will be calculated for the $\N=0$

1039: orbifold using the effective action.

1040: 

1041: The gauge coupling beta function for a product gauge group, up to two loops,

1042: depends on the gauge couplings $\{g\}$ and the Yukawa coupling matrices $Y$

1043: \cite{Machacek:83},

1044: \begin{eqnarray}

1045: \beta_{g_k}(\{g\},Y) &\eqneq& \frac{d g_k}{d \log \mu}

1046:             = \beta_{g_k}^{(1)}(g_k)+\beta_{g_k}^{(2)}(\{g\},Y) \ \ ,\\

1047: \beta_{g_k}^{(1)}(g_k) &\eqneq& -\frac{g_k^3}{(4\pi)^2}\left [

1048: \frac{11}{3}C_2(G_k)

1049: -\frac{2}{3}\sum_{\tinybox{fermions}}C(F_k)-\frac{1}{6}\sum_{\tinybox{scalars}}{C(S_k)}

1050:                     \right] \ ,\ \ \ \ \ \\

1051: \beta_{g_k}^{(2)}(\{g\},Y) &\eqneq& -\frac{g_k^3}{(4\pi)^2}\left [

1052: \frac{34}{3}\frac{g_k^2}{(4\pi)^2}C_2(G_k)^2 \right. \nonumber\\

1053: &&-\sum_{\tinybox{fermions}}{\left(

1054:             \sum_{l\in^{\tinybox{gauge}}_{\tinybox{groups}}}{2\frac{g_l^2}{(4\pi)^2}C_2(F_l)}

1055:            + \frac{10}{3}\frac{g_k^2}{(4\pi)^2}C_2(G_k) \right)C(F_k)}

1056:        \nonumber\\

1057: &&-\sum_{\tinybox{scalars}}{\left( \sum_{l\in^{\tinybox{gauge}}_{\tinybox{groups}}}{2\frac{g_l^2}{(4\pi)^2}C_2(S_l)}

1058:            + \frac{1}{3}\frac{g_k^2}{(4\pi)^2}C_2(G_k) \right)C(S_k)}

1059:        \nonumber\\

1060: &&\left.+\frac{1}{(4\pi)^2}Y_4(F)

1061:                     \right] \ ,

1062: \end{eqnarray}

1063: where $G_k$ is the adjoint representation of the $k$th gauge field,

1064: $F_k$ is the representation of the fermions under the $k$th gauge group,

1065: $S_k$ is the representation of the scalars under the $k$th gauge group,

1066: $(Y^S)^{F^1}_{F^2}$ is the Yukawa coupling matrices representing the

1067: coupling between a scalar $S$ and two fermions $F^1,F^2$,

1068: and $Y_4(F)$ is the Yukawa coupling contribution defined as

1069: \begin{eqnarray}

1070: Y_4(F)\delta^{ab}&=&\mathop{\tr}_{\tinybox{fermions}}\sum_{\tinybox{scalars}}

1071:                        Y^SY^{\dagger S}T^aT^b \ .

1072: \end{eqnarray}

1073: The summations are over Weyl fermions and real scalars.

1074: The Dynkin index of the fundamental representation is normalized to

1075: $C(F)=\frac{1}{2}$.

1076: 

1077: The one loop $\beta$ function for the Yukawa coupling is \cite{Machacek:84}

1078: \begin{eqnarray}

1079: \lefteqn{\beta_{Y^S}=\frac{d Y^S}{d \log\mu}=} \nonumber\\

1080: &&\frac{1}{(4\pi)^2}

1081: \left[\vphantom{\sum_{k\in^{\tinybox{gauge}}_{\tinybox{groups}}}}

1082: \frac{1}{2}\left(Y^{\dagger S'}Y^{S'}Y^S+Y^S Y^{S'}Y^{\dagger S'}\right)

1083: + 2Y^{S'}Y^{\dagger S}Y^{S'}

1084: + Y^{S'}\tr \left(Y^{\dagger S'}Y^S\right) \right.\nonumber\\

1085: &&\phantom{\frac{1}{(4\pi)^2}}\left.

1086: {} - 3\sum_{k\in^{\tinybox{gauge}}_{\tinybox{groups}}}

1087:   \left(\vphantom{Y^{S'}}g_k^2 C_2(F^1_k)Y^S+Y^S C_2(F^2_k)\right) \right] \ .

1088: \end{eqnarray}

1089: The first two terms are the scalar loop corrections to the two fermion legs.

1090: The third term is the one point irreducible scalar correction.

1091: The fourth term is the fermion loop correction to the scalar leg.

1092: The last two terms are the gauge bosons loop corrections to the two fermion

1093: legs.

1094: 

1095: In the following subsections we analyze orbifold theories with

1096: $\N=2,1$ and $0$ supersymmetries.

1097: 

1098: \subsection{$\N=2$ Orbifolds}

1099: 

1100: The orbifold leaves an $\N=2$ supersymmetry if $\Gamma\subset

1101: SU(2)\subset SU(4)_R$. The simplest case is the $\Z_2$ orbifold with

1102: weights $(1,1,0,0)$ that leaves the following matter content,

1103: \begin{eqnarray*}

1104: \begin{array}{l|c@{}c@{}c@{}c|c}

1105:  & \multicolumn{1}{c@{\ \otimes}}{SU(N)}

1106:  & \multicolumn{1}{c@{\ \otimes}}{SU(N)}

1107:  & \multicolumn{1}{c@{\ \otimes}}{U(1)}

1108:  & \multicolumn{1}{c|}{U(1)}

1109:  & SU(2) \\

1110: \hline

1111: V_{N_1} & G & 1 & 0 & 0 & 1 \\

1112: V_{N_2} & 1 & G & 0 & 0 & 1 \\

1113: V_{1_1} & 1 & 1 & 0 & 0 & 1 \\

1114: V_{1_2} & 1 & 1 & 0 & 0 & 1 \\

1115: \hline\vphantom{\overline{N}^1}

1116: H^1_2 & N & \overline{N} & \phantom{-}1 & -1 & 2

1117: %H^2_1 & \overline{N} & N & -1 & \phantom{-}1 & 2

1118: \end{array}

1119: \end{eqnarray*}

1120: where $V$ and $H$ are the vector and hyper multiplet of $\N=2$.

1121: The $SU(2)$ is a global symmetry that is a remnant of the original $SU(4)_R$

1122: symmetry. There is also the usual $SU(2)_R\otimes U(1)_R$ symmetry.

1123: 

1124: The $U(1)$ charges are written up to a normalization factor.

1125: When calculating the group factor of Feynman diagrams, we use

1126: the double line notation as in \eqref{eqn:POneLoopBiFundU1}

1127: and \eqref{eqn:PBiOneLoopU1}.

1128: 

1129: The non-renormalization theorem of $\N=2$ guarantees that there can only

1130: be one loop corrections to the perturbative $\beta$ function.

1131: The different gauge bosons can only interact through the bifundamental

1132: hypermultiplet and those interactions only occur at two loop order.

1133: Consequently, the different gauge bosons do not interact.

1134: 

1135: We choose

1136: all the $SU(N)$ couplings to be the same and all the $U(1)$ couplings to

1137: be same, and then there are only two independent couplings, with the $\beta$

1138: functions

1139: \begin{eqnarray}

1140: \beta_{g_N} &=& 0 \ ,\nonumber\\

1141: \beta_{g_1} &=& 2\frac{g_1^3N}{(4\pi)^2} \ .

1142: \end{eqnarray}

1143: If we start on the natural line, we have $g_N=g_1$ and we get a theory that

1144: is not finite because of the running of the $U(1)$ coupling constant.

1145: However, we can choose $g_1=0$ and get a finite theory.

1146: In other words, since the $U(1)$ gauge couplings are IR free,

1147: we can say that the $U(1)$ decouples in the IR, and the theory is IR finite.

1148: 

1149: For a general $Z_{|\Gamma|}$ orbifold we have a $|\Gamma|$ dimensional

1150: manifold of fixed points parameterized by the $|\Gamma|$ $SU(N)$ gauge

1151: couplings.

1152: 

1153: \subsection{$\N=1$ Orbifolds}

1154: \label{sec:N1}

1155: 

1156: The orbifold leaves an $\N=1$ supersymmetry if $\Gamma\subset

1157: SU(3)\subset SU(4)_R$. The simplest case is the $\Z_3$ orbifold with weights

1158: $(1,1,1,0)$ \cite{Kachru:98} having the following matter content,

1159: \begin{eqnarray*}

1160: \begin{array}{l|c@{}c@{}c@{}c@{}c@{}c|c}

1161:  & \multicolumn{1}{c@{\ \otimes}}{SU(N)_1\!}

1162:  & \multicolumn{1}{c@{\ \otimes}}{\!SU(N)_2\!}

1163:  & \multicolumn{1}{c@{\ \otimes}}{\!SU(N)_3\!}

1164:  & \multicolumn{1}{c@{\ \otimes}}{U(1)_1\!}

1165:  & \multicolumn{1}{c@{\ \otimes}}{\!U(1)_2\!}

1166:  & \multicolumn{1}{c|}{\!U(1)_3\!}

1167:  & SU(3) \\

1168: \hline

1169: V_{N_1} & G & 1 & 1 & 0 & 0 & 0 & 1 \\

1170: V_{N_2} & 1 & G & 1 & 0 & 0 & 0 & 1 \\

1171: V_{N_3} & 1 & 1 & G & 0 & 0 & 0 & 1 \\

1172: V_{1_1} & 1 & 1 & 1 & 0 & 0 & 0 & 1 \\

1173: V_{1_2} & 1 & 1 & 1 & 0 & 0 & 0 & 1 \\

1174: V_{1_3} & 1 & 1 & 1 & 0 & 0 & 0 & 1 \\

1175: \hline\vphantom{\overline{N}^1}

1176: \Phi^1_2 & N & \overline{N} & 1 & \phantom{-}1 & -1 & \phantom{-}0 & 3 \\

1177: \Phi^2_3 & 1 & N & \overline{N} & \phantom{-}0 & \phantom{-}1 & -1 & 3 \\

1178: \Phi^3_1 & \overline{N} & 1 & N & -1 & \phantom{-}0 & \phantom{-}1 & 3

1179: \end{array}

1180: \end{eqnarray*}

1181: where $V$ and $\Phi$ are the vector and chiral multiplets of $\N=1$.

1182: The $SU(3)$ is a global symmetry remnant of the original $SU(4)_R$

1183: symmetry. There is also the usual $U(1)_R$ symmetry.

1184: The superpotential is\footnote{

1185:   The entire Lagrangian is normalized by a factor of $\frac{1}{C(F)}$.

1186:   This is due to the use of the notation $\Phi\equiv T^a\Phi^a$ that

1187:   contributes a factor of $C(F)$. For example,

1188:   $\tr(D_{\mu}\Phi D^{\mu}\Phi)=\tr(T^aT^b)D_{\mu}\Phi^a D^{\mu}\Phi^b=

1189:   C(F)D_{\mu}\Phi^a D^{\mu}\Phi^a$.

1190: }

1191: \begin{eqnarray}

1192: \sqrt{2}h\sum_{k=1}^3

1193: \tr\left([\Phi^k_{k+1},\Phi^{k+1}_{k+2}]\Phi^{k+2}_k\right) \ .

1194: \end{eqnarray}

1195: The trace here stands for taking the singlet representation under all gauge

1196: groups and the $SU(3)$ global symmetry group.

1197: 

1198: Generally we can have different $h_k$ for the three summands.

1199: However, for the sake of simplicity and naturalness, we choose

1200: $h_k=h$, $g_N^k=g_N$ and $g_1^k=g_1$.

1201: The RG flow will not alter this choice.

1202: Before the orbifolding, $h=g$ would have yielded an $\N=4$ theory.

1203: 

1204: The Yukawa terms in the Lagrangian are (in components)

1205: \begin{eqnarray}

1206: &&\sqrt{2}g_N\sum_{k=1}^3\left(\phi^k_{k+1}\bar{\psi}^{k+1}_k\lambda^{N_k} +

1207:   \phi^{k-1}_k\bar{\psi}^k_{k-1}\lambda^{N_k}\right) + \\

1208: &&\sqrt{2}g_1\sum_{k=1}^3\left(\phi^k_{k+1}\bar{\psi}^{k+1}_k\lambda^{1_k} +

1209:   \phi^{k-1}_k\bar{\psi}^k_{k-1}\lambda^{1_k}\right) +

1210: \frac{h}{\sqrt{2}}\sum_{k=1}^3\phi^k_{k+1}\psi^{k+1}_{k+2}\psi^{k+2}_k +

1211: \text{h.c.} \nonumber

1212: \end{eqnarray}

1213: where the trace over the gauge and global indices is implicit.

1214: The coupling of the first two terms has to be equal to the gauge

1215: coupling as a consequence of supersymmetry. The third term comes from the

1216: superpotential.

1217: 

1218: There are non vanishing triangle anomalies for the $U(1)$ gauge symmetries

1219: in this theory. For example

1220: \begin{eqnarray}

1221: U(1)_1^2U(1)_2: && \sum Q_1^2Q_2=-3N^2 \ .

1222: \end{eqnarray}

1223: To study the cancellation of those anomalies we need to study the effective

1224: theory of the orbifolded string theory.

1225: It not sufficient to study the orbifolded $\N=4$ field theory.

1226: It was shown in \cite{Ibanez:98} that there is a generalized Green-Schwarz

1227: mechanism that cancels those anomalies.

1228: Fayet-Illiopoulos terms are generated to the anomalous $U(1)$s. Those FI

1229: terms give a mass to the $U(1)$ gauge fields of the order of the string scale,

1230: which means that in the limit $l_s\rightarrow 0$ the $U(1)$s are decoupled

1231: from the effective field theory.

1232: 

1233: Next, we will calculate the RG flow of this theory with the $U(1)$ gauge

1234: fields despite it being inconsistent because of the $U(1)$ anomalies. 

1235: The purpose of the calculation is to show that the theory reaches a fix point

1236: which is stable under fluctuations of the form of the $U(1)$ fields.

1237: 

1238: We can use the Leigh-Strassler arguments \cite{Leigh:95} to check whether 

1239: there is a manifold of fixed points.

1240: From $\N=1$ SUSY arguments we know \cite{Shifman:86} that the $\beta$

1241: functions have the form

1242: (when absorbing a factor of $4\pi$ into the couplings),

1243: \begin{eqnarray}

1244: \beta_h &\eqneq&

1245:     h\left(-3+\sum\left(d(\Phi)+\smallfrac{1}{2}\gamma\right)\right)

1246:   = \smallfrac{3}{2}h\gamma \ ,\nonumber\\

1247: \beta_{g_N} &\eqneq& -\frac{g_N^3}{1-g_N^2C_2(G)}

1248:     \left(3C_2(G)-\sum C(R)(1-\gamma)\right)

1249:   = \frac{-3g_N^3N\gamma}{1-g_N^2N} \ ,\\

1250: \beta_{g_1} &\eqneq& -\frac{g_1^3}{1-g_1^2C_2(G)}

1251:     \left(3C_2(G)-\sum C(R)(1-\gamma)\right)

1252:   = 3g_1^3N(1-\gamma) \ .\nonumber

1253: \label{eqn:BetaForm}

1254: \end{eqnarray}

1255: The denominator of $\beta_{g_N}$ is zero at $g_N^2N=1$,

1256: but for small couplings it is smooth and positive.

1257: 

1258: There is a linear relation between the first two equations, so setting

1259: the three $\beta$ functions to zero gives us two conditions on the three

1260: couplings. This yields a fixed line at $\gamma=0$, $g_1=0$.

1261: 

1262: This fixed line, however,  is not the natural line $g_N=g_1=h$ from which we

1263: want to start the RG flow.

1264: We have already shown that on the natural line $\beta_h$ vanishes up to two

1265: loop order and $\beta_{g_N}$ vanishes up to three loop order.

1266: The form of the $\beta$ functions \eqref{eqn:BetaForm} tells us that

1267: if one $\beta$ function vanishes then so does the other.

1268: It is tempting to speculate that on the natural line both $\beta_h$

1269: and $\beta_{g_N}$ vanish to all orders in perturbation theory.

1270: 

1271: By an explicit calculation of the $\beta$ functions we can parameterize the

1272: fixed line $\gamma(g_N,g_1,h)|_{g_1=0}=0$ and check whether the natural

1273: line flows to the fixed line in the IR.

1274: The $\beta$ functions of the gauge couplings up to 2-loop order and the

1275: Yukawa coupling up to 1-loop order are

1276: \begin{eqnarray}

1277: \beta_h &=& \frac{6}{N}\frac{h}{(4\pi)^2}\left(N^2h^2-N^2g_N^2-(g_1^2-g_N^2)\right) \ ,\nonumber\\

1278: \beta_{g_N} &=& -12\frac{g_N^3}{(4\pi)^4}\left(N^2h^2-N^2g_N^2-(g_1^2-g_N^2)\right) \ ,

1279: \label{eqn:N1Beta}\\

1280: \beta_{g_1} &=& 3\frac{g_1^3N}{(4\pi)^2}

1281:                 -12\frac{g_1^3}{(4\pi)^4}\left(N^2h^2-N^2g_N^2-(g_1^2-g_N^2)\right) \ .\nonumber

1282: \end{eqnarray}

1283: The $\beta$ functions are of the expected form \eqref{eqn:BetaForm}, giving

1284: a consistency check of our calculations.

1285: The first two $\beta$ functions vanish on the two dimensional manifold

1286: defined by the relation

1287: \begin{eqnarray}

1288: h^2=g_N^2+\frac{g_1^2-g_N^2}{N^2} \ .

1289: \end{eqnarray}

1290: The natural line $g_1=g_N=h$ is obviously on this manifold.

1291: The relation between $h, g_N$ on the fixed line $(g_1=0)$ is

1292: \begin{eqnarray}

1293: h^2=(1-\frac{1}{N^2})g_N^2 \ .

1294: \label{eqn:hgRatio}

1295: \end{eqnarray}

1296: In the large $N$ limit the fixed line coincides with the natural line.

1297: 

1298: In figure \ref{fig:N1Flow}

1299: \begin{figure}[b]

1300: \begin{center}

1301: \setlength{\unitlength}{0.00016667in}

1302: \begin{picture}(25000,14000)

1303: %\put(0,0){(0,0)}

1304: %\put(25000,14000){(1,1)}

1305: \put(0,0){\psfig{figure=N1Flow.eps}}

1306: \put(14800,13600){$g_1^2$}

1307: \put(23000,10600){$g_N^2$}

1308: \put(24000,9600){$g_1^2$}

1309: \put(24000,6000){$h^2$}

1310: \end{picture}

1311: \end{center}

1312: \isucaption{The RG flow of the $\N=1$ orbifold theory from the natural line

1313:   $g_1^2=g_N^2=h^2=0.01$ to the fixed line in the IR.

1314:   $g_1^2,g_N^2,h^2$ are plotted as a function of $g_1^2$.

1315:   The graph was plotted for $N=5$ and gives the expected

1316:   $h/g_N$ ratio at $g_1=0$.}

1317: \label{fig:N1Flow}

1318: \end{figure}

1319: we plot the numerical solution of the $\beta$ functions \eqref{eqn:N1Beta}.

1320: We plot $g_1^2, g_N^2, h^2$ as a function of $g_1^2$

1321: and not as a function of the energy scale $\mu$.

1322: This is permissible because $g_1$ is a monotonic increasing function of

1323: $\mu$ ($\beta_{g_1}$ is positive definite).

1324: 

1325: The solution demonstrates how the natural line flows to the fixed line

1326: in the IR.

1327: In the IR $(g_1=0)$ we get the expected ratio \eqref{eqn:hgRatio} between

1328: $h$ and $g_N$. This fix line is stable under fluctuations in $g_1,g_N,h$.

1329: This consolidates the assertion that this orbifolded string theory is

1330: described in the IR by an effective field theory on a fixed line.

1331: 

1332: \subsection{$\N=0$ Orbifolds}

1333: \label{sec:N0}

1334: 

1335: For the non supersymmetric orbifolds we will focus on the $\Z_2$ orbifold

1336: with weights $(1,1,1,1)$. 

1337: The $\Z_2$ is in the center of $SU(4)_R$, as can be seen from the weights

1338: that do not break the $SU(4)$ symmetry.

1339: We will specify which of the essential results are general and which

1340: are specific to this case.

1341: Applying the $\Z_2$ orbifold to a Type II string theory

1342: reproduces the Type 0 string theory.

1343: The field theory we are describing lives in Type 0B string theory on $N$

1344: dyonic (electric-magnetic) D3 branes \cite{Klebanov:99}.

1345: The theory has the following matter content,

1346: \begin{eqnarray*}

1347: \begin{array}{l|c@{}c@{}c@{}c|c}

1348:  & \multicolumn{1}{c@{\ \otimes}}{SU(N)}

1349:  & \multicolumn{1}{c@{\ \otimes}}{SU(N)}

1350:  & \multicolumn{1}{c@{\ \otimes}}{U(1)}

1351:  & \multicolumn{1}{c|}{U(1)}

1352:  & SU(4) \\

1353: \hline

1354: A^{\mu}_{N_1} & G & 1 & 0 & 0 & 1 \\

1355: A^{\mu}_{N_2} & 1 & G & 0 & 0 & 1 \\

1356: A^{\mu}_{1_1} & 1 & 1 & 0 & 0 & 1 \\

1357: A^{\mu}_{1_2} & 1 & 1 & 0 & 0 & 1 \\

1358: \hline\vphantom{\overline{N}^1}

1359: (\psi^I)^1_2 & N & \overline{N} & \phantom{-}1 & -1 & 4 \\

1360: (\psi^I)^2_1 & \overline{N} & N & -1 & \phantom{-}1 & 4 \\

1361: \hline

1362: \phi^{IJ}_{N_1} & G & 1 & 0 & 0 & 6 \\

1363: \phi^{IJ}_{N_2} & 1 & G & 0 & 0 & 6 \\

1364: \phi^{IJ}_{1_1} & 1 & 1 & 0 & 0 & 6 \\

1365: \phi^{IJ}_{1_2} & 1 & 1 & 0 & 0 & 6

1366: \end{array}

1367: \end{eqnarray*}

1368: where $I,J=1\ldots 4$ are the $SU(4)$ fundamental indices.

1369: The fermions are Weyl spinors and the scalars are real.

1370: 

1371: Because the scalars are in the adjoint representation there can

1372: only a be Coulomb branch and the gauge group

1373: can maximally break into its Cartan subalgebra $U(1)^{2N}$.

1374: Therefore, there is no spontaneous symmetry breaking of the $U(1)$

1375: gauge fields specified in the table above.

1376: This is not true for all $\N=0$ orbifolds because in general, we can

1377: have scalars in the bifundamental representation,

1378: for example, $\Z_5$ with weights $(1,2,3,4)$ \cite{Kachru:98}.

1379: On the other hand there are $\N=0$ orbifolds with no scalars at all

1380: in the adjoint representation,

1381: for example, $\Z_5$ with weights $(1,1,1,2)$.

1382: 

1383: The Yukawa terms in the Lagrangian are

1384: \begin{eqnarray}

1385: \frac{Y_N}{2}\sum_{k=1}^2\varepsilon_{IJKL}

1386: \tr\left(\phi^{IJ}_{N_k}(\psi^K)^1_2(\psi^L)^2_1\right) +

1387: \frac{Y_1}{2}\sum_{k=1}^2\varepsilon_{IJKL}

1388: \tr\left(\phi^{IJ}_{1_k}(\psi^K)^1_2(\psi^L)^2_1\right) \ .\ 

1389: \end{eqnarray}

1390: The $U(1)$ scalars couple to the theory only through the $Y_1$

1391: Yukawa coupling.

1392: 

1393: The quartic terms in the Lagrangian are

1394: \begin{eqnarray}

1395: \frac{\lambda}{4}\tr\left(\

1396: \left[\phi^{IJ}_{N_1},\phi^{KL}_{N_1}\right]^2 +

1397: \left[\phi^{IJ}_{N_2},\phi^{KL}_{N_2}\right]^2

1398: \right) \ .

1399: \label{eqn:quartic}

1400: \end{eqnarray}

1401: This is the classical scalar potential.

1402: The $U(1)$ scalars do not participate in this potential because they are

1403: abelian.

1404: 

1405: We have a classical moduli space

1406: that can be parametrized by the diagonalized scalar {\it vev} matrices

1407: \begin{eqnarray}

1408: \begin{array}{r@{}c@{}l}

1409: (\phicl[N_k])^{IJ}&=&

1410:   \displaystyle\diag\left((y_k^{IJ})^1,\ldots,(y_k^{IJ})^N\right)

1411:   - \smallfrac{1}{N}\sum_{i=1}^N (y_k^{IJ})^i \cdot\One_{N\times N} \ ,\\

1412: (\phicl[1_k])^{IJ}&=&

1413:   \displaystyle\smallfrac{1}{N}\sum_{i=1}^N (y_k^{IJ})^i\equiv

1414:      \smallfrac{1}{N}\tr(y_k^{IJ}) \ .

1415: \end{array} \s{k=1,2}

1416: \end{eqnarray}

1417: 

1418: Since there is no supersymmetry, nothing protects the fields from

1419: acquiring a mass by quantum corrections.

1420: There are also new quartic scalar coupling terms that can

1421: appear in the renormalization process \cite{Csaki:99}, for example,

1422: \begin{eqnarray}

1423: \lambda'

1424: \tr\left(\varepsilon_{IJKL}\phi^{IJ}_{N_1}\phi^{KL}_{N_1}\right)

1425: \tr\left(\varepsilon_{IJKL}\phi^{IJ}_{N_2}\phi^{KL}_{N_2}\right) \ .

1426: \label{eqn:newQuartic}

1427: \end{eqnarray}

1428: Those quantum corrections will in general lift the classical moduli.

1429: 

1430: In order to analyze the behavior of the theory we use the

1431: Coleman-Weinberg effective potential \cite{Coleman:73} as was done in 

1432: \cite{Zarembo:99,Tseytlin:99} for this model without the $U(1)$ factors.

1433: The ``zero loop'' (tree level)  effective potential comes from expanding

1434: the classical potential around the classical {\it vev}s, i.e. setting

1435: $\phi=\phicl+\phi'$.

1436: There are no masses in the original Lagrangian, but in the effective

1437: Lagrangian the fields acquire masses from the scalar {\it vev}s.

1438: The eigenvalues of the mass matrices are

1439: \begin{eqnarray}

1440: \mu_{\tinybox{gauge}}^{ij}(\phicl)^2&=&g_N^2|y^i-y^j|^2 \ ,\nonumber\\

1441: \mu_{\tinybox{scalar}}^{ij}(\phicl)^2&=&\lambda|y^i-y^j|^2 \ ,\\

1442: \mu_{\tinybox{fermion}}^{ij}(\phicl[1],\phicl[2])^2&=&

1443:   \left|Y_N\left(

1444: \left(y_1^i - \smallfrac{1}{N}\tr(y_1)\right) -

1445: \left(y_2^j - \smallfrac{1}{N}\tr(y_2)\right)\right)\right. \nonumber\\

1446:  && \left.{}+Y_1\left(\smallfrac{1}{N}\tr(y_1) -

1447:               \smallfrac{1}{N}\tr(y_2)\right)\right|^2 \ .\nonumber

1448: \end{eqnarray}

1449: The $SU(4)$ indices are implicit,

1450: where $|y|^2=\frac{1}{2}\varepsilon_{IJKL}y^{IJ}y^{KL}$

1451: and $\mu_{\tinybox{fermion}}$ has two $SU(4)$ indices related to the

1452: $SU(4)$ indices of the fermion to which the mass couples.

1453: Therefore, $\mu_{\tinybox{fermion}}$ has also spinor indices.

1454: 

1455: The one loop effective potential includes the tree level scalar potential

1456: \eqref{eqn:quartic}, renormalization counterterms, and

1457: the one loop correction to the effective potential,

1458: \begin{eqnarray}

1459: V^{\text{1-loop}}_{\text{eff}} &\eqneq& \phantom{{}-{}} 2\sum_{i,j=1}^N\left[

1460:  V\left(\mu_{\tinybox{gauge}}^{ij}(\phicl[1])^2\right) + 

1461:  V\left(\mu_{\tinybox{gauge}}^{ij}(\phicl[2])^2\right)

1462: \right] \nonumber\\

1463: && {}+ 6\sum_{i,j=1}^N\left[

1464:  V\left(\mu_{\tinybox{scalar}}^{ij}(\phicl[1])^2\right) + 

1465:  V\left(\mu_{\tinybox{scalar}}^{ij}(\phicl[2])^2\right)

1466: \right] \\

1467: && {}- 8\sum_{i,j=1}^N\left[

1468:  V\left(\mu_{\tinybox{fermion}}^{ij}(\phicl[1],\phicl[2])^2\right) +

1469:  V\left(\mu_{\tinybox{fermion}}^{ij}(\phicl[2],\phicl[1])^2\right)

1470: \right] \ , \nonumber\\

1471: V(\mu^2)&\eqneq&\frac{1}{2}\int \frac{d^4p}{(2\pi)^4}

1472:            \ln\frac{p^2+\smallfrac{1}{2}\mu^2}{p^2} \nonumber\\

1473:  &\eqneq& \frac{2\pi^2}{2(2\pi)^4}\left(

1474:     \frac{1}{4}\Lambda^2\mu^2

1475:     + \frac{1}{16}\mu^4 \left(

1476:       \ln\frac{\smallfrac{1}{2}\mu^2}{\Lambda^2}-\frac{1}{2}\right)

1477:     + O\left(\frac{\mu^6}{\Lambda^2}\right)

1478:   \right) \ .\ 

1479: \end{eqnarray}

1480: 

1481: The $\Lambda^2$ term provides the one loop correction to the scalar masses.

1482: As can be seen from the effective potential, the scalar mass is corrected 

1483: by a gauge boson loop, a scalar loop and a fermion loop.

1484: The $\Lambda^2$ term is

1485: \begin{eqnarray}

1486: (V^{\text{1-loop}}_{\text{eff}})_{\Lambda^2} = \frac{\Lambda^2}{64\pi^2}

1487:   \sum_{i,j=1}^N

1488: \begin{array}[t]{l}

1489: \displaystyle\vphantom{\sum}

1490:   (2g_N^2+6\lambda)\left(|y_1^i-y_1^j|^2 + |y_2^i-y_2^j|^2\right) \\

1491: \displaystyle\vphantom{\sum}

1492: - 16Y_N^2\left|

1493:   \left(y_1^i - \smallfrac{1}{N}\tr(y_1)\right) -

1494:   \left(y_2^j - \smallfrac{1}{N}\tr(y_2)\right)\right|^2 \\

1495: \displaystyle\vphantom{\sum}

1496: - 16Y_1^2\left|\smallfrac{1}{N}\tr(y_1) -

1497:                   \smallfrac{1}{N}\tr(y_2)\right|^2 \ .

1498: \end{array}

1499: \end{eqnarray}

1500: 

1501: On the natural line, where all the couplings are equal to $g$ we get

1502: \begin{eqnarray}

1503: (V^{\text{1-loop}}_{\text{eff}})_{\Lambda^2} &=& -\frac{\Lambda^2}{4\pi^2} g^2N^2

1504:  \left|\smallfrac{1}{N}\tr(y_1) - \smallfrac{1}{N}\tr(y_2)\right|^2 \nonumber\\

1505:  &=& \frac{m^2}{2}\left(\phicl[1_1]-\phicl[1_2]\right)^2 \ ,

1506: \label{eqn:U1mass}

1507: \end{eqnarray}

1508: which generates a quadratically divergent tachyonic mass for the diagonal

1509: $U(1)$ scalar.

1510: The mass of the $U(1)$ scalar comes from the fermion loop

1511: and can be set to zero by choosing $Y_1=0$ as was done in \cite{Tseytlin:99}.

1512: Otherwise, we should include a renormalization counterterm for the

1513: scalar mass.

1514: Using this counterterm we can set the renormalized mass to zero.

1515: It would have been much more elegant if there were some underlying mechanism

1516: that naturally sets this scalar mass to zero as discussed in subsection

1517: \ref{sec:Hierarchy}.

1518: 

1519: The logarithmic term in the effective potential provides the one loop

1520: correction to the quartic coupling $\lambda$.

1521: It is sufficient to compute this term when only one component out of the

1522: ${\bf 6}$ of $SU(4)$ is nonzero.

1523: On the natural line the logarithmic term is

1524: \begin{eqnarray}

1525: \left(V^{\text{1-loop}}_{\text{eff}}\right)_{log(\Lambda)} &=& 

1526: -\frac{8g^4}{256\pi^2}\left(

1527: 6\left(\tr(\phicl[N_1]^2)-\tr(\phicl[N_2]^2)\right)^2 \right.\nonumber\\

1528:  && \phantom{-\frac{8g^4}{256\pi^2}}

1529:     -8N\left(\tr(\phicl[N_1]^3)-\tr(\phicl[N_2]^3)\right)

1530:       (\phicl[1_1]-\phicl[1_2])

1531: \nonumber\\

1532:  && \phantom{-\frac{8g^4}{256\pi^2}}

1533:     -12N\left(\tr(\phicl[N_1]^2)+\tr(\phicl[N_2]^2)\right)

1534:       (\phicl[1_1]-\phicl[1_2])^2

1535: \nonumber\\

1536:  && \phantom{-\frac{8g^4}{256\pi^2}}\left.{}\vphantom{{1^1}^1}

1537:     -2N^2(\phicl[1_1]-\phicl[1_2])^4

1538: \right) \ .

1539: \label{eqn:Logarithmic}

1540: \end{eqnarray}

1541: There is no $\tr(\phicl[N]^4)$ term. This is a manifestation of the fact

1542: that the one loop $\beta$ function for the original quartic coupling

1543: \eqref{eqn:quartic} vanishes on the natural line.

1544: In models with scalars in the bifundamental representation this vanishing

1545: occurs only if the $U(1)$ factors are taken into account.

1546: 

1547: The logarithmic term is of the form $(\phicl)^4\ln(\phicl)^2$. The quartic

1548: coupling is the fourth derivative of the effective potential with respect to

1549: the scalar fields, $\lambda'=\frac{d^4 V_{\text{eff}}}{d{\phicl}^4}$.

1550: This derivative diverges at the origin $(\phicl=0)$, requiring a definition of

1551: a renormalized coupling away from the singularity at some arbitrary

1552: renormalization scale $M$,

1553: \begin{eqnarray}

1554: \lambda'=\left.\frac{d^4 V_{\mbox{eff}}}{d{\phicl}^4}\right|_{\phicl=M}

1555:   =C_{\lambda'}+c_{\lambda'}\frac{3g^4}{4\pi^2}\left(

1556:     \ln\frac{\smallfrac{g^2}{2}M^2}{\Lambda} +\frac{11}{3}\right) \ ,

1557: \end{eqnarray}

1558: where $C_{\lambda'}$ is the renormalization counterterm, and

1559: $c_{\lambda'}$ is a constant depending on the $\phicl$ we differentiate

1560: with respect to, calculated in \eqref{eqn:Logarithmic}.

1561: Since $\lambda'$ did not exist in the original Lagrangian we would like

1562: to require $\lambda'=0$.

1563: We are not allowing arbitrary couplings, therefore the renormalization scale

1564: $M$ is the renormalization scale $\mu_N$ defined for the natural line.

1565: Finally, the effective potential for the new quartic couplings is

1566: \begin{eqnarray}

1567: V_{\text{eff}}=c_{\lambda'}\frac{g^4\phicl^4}{32\pi^2}\left(

1568:   \ln\frac{\phicl^2}{\mu_N^2}-\frac{25}{6}\right) \ .

1569: \end{eqnarray}

1570: 

1571: The first term in \eqref{eqn:Logarithmic} is the term found in

1572: \cite{Tseytlin:99}. This term leads to a repulsive potential between

1573: {\it vev}s of scalars of the same type. For example taking

1574: $y_1^1=-y_1^2=\rho$ leads to a repulsive potential of the form

1575: $\rho^4\ln\rho^2$. Notice also that there are flat directions in

1576: \eqref{eqn:Logarithmic}, for example $y_1^1=y_2^1=-y_1^1=-y_1^2=\rho$.

1577: 

1578: The other terms in \eqref{eqn:Logarithmic} give a potential to the

1579: diagonal $U(1)$ scalar. Because of the opposite sign this potential is

1580: attractive at short distances. At long distances this potential is repulsive,

1581: but at the scale it becomes so, higher loop terms should be taken

1582: into account.

1583: 

1584: Assuming that the choice of taking all the masses to zero on the natural

1585: line is consistent, we go ahead and calculate

1586: the explicit $\beta$ functions for the gauge couplings to two loop order

1587: and for the Yukawa couplings to one loop order,

1588: \begin{eqnarray}

1589: \beta_{g_N}&\eqneq&-\frac{g_N^3}{(4\pi)^4}\left(-24N^2g_N^2 - 8(g_1^2-g_N^2)

1590:   + 24N^2Y_N^2 + 24(Y_1^2-Y_N^2)\right) \ ,\nonumber\\

1591: \beta_{g_1}&\eqneq&\frac{11}{3}\frac{g_1^3N}{(4\pi)^2} \nonumber\\

1592:   &&-\frac{g_1^3}{(4\pi)^4}\left(-8N^2g_N^2 - 8(g_1^2-g_N^2)

1593:   + 24N^2Y_N^2 + 24(Y_1^2-Y_N^2)\right) \ ,\\

1594: \beta_{Y_N}&\eqneq&\frac{Y_N}{(4\pi)^2}\left(-6Ng_N^2 - \frac{6}{N}(g_1^2-g_N^2)

1595:   + 6NY_N^2 + \frac{2}{N}(Y_1^2-Y_N^2)\right) \ ,\nonumber\\

1596: \beta_{Y_1}&\eqneq&\frac{Y_1}{(4\pi)^2}\left(-6Ng_N^2 - \frac{6}{N}(g_1^2-g_N^2)

1597:   + 2NY_N^2 + 4NY_1^2 + \frac{2}{N}(Y_1^2-Y_N^2)\right) \ .\nonumber

1598: \label{eqn:N0beta}

1599: \end{eqnarray}

1600: As expected, the $\beta$ functions for $g_N$ and $Y_N$ vanish on the natural

1601: line where all the couplings are the same. Notice that the $\beta$ function

1602: for $Y_1$ also vanishes on the natural line.

1603: Moreover, $Y_1=Y_N\Rightarrow \beta_{Y_1}=\beta_{Y_N}$, meaning that if

1604: the Yukawa couplings start the same at the renormalization scale, they will

1605: stay the same, at least to one loop order.

1606: 

1607: It would be interesting to look for fixed points where all the $\beta$

1608: functions are zero.

1609: If we take all the $\beta$ functions to the first non vanishing order,

1610: then the only fixed point is the trivial fixed point where all the couplings

1611: are zero. If we use the $\beta_{g_1}$ that we calculated to two loop order

1612: we get several non trivial fixed points. There is one fixed point on

1613: the natural line,

1614: \begin{eqnarray}

1615: \frac{g_N^2}{(4\pi)^2}=

1616: \frac{g_1^2}{(4\pi)^2}=

1617: \frac{Y_N^2}{(4\pi)^2}=

1618: \frac{Y_1^2}{(4\pi)^2}= \frac{11}{48N} \ .

1619: \label{eqn:N0FixPoint}

1620: \end{eqnarray}

1621: This fixed point, however, is inconsistent, because we did not evaluate all

1622: the $\beta$ functions to the same order and ignored the quartic couplings.

1623: The fixed point is also not very interesting because it is a non

1624: stable fixed point.

1625: All the same, the calculations give the order of magnitude of the range of

1626: validity of our calculations. We can trust the lowest order perturbative

1627: expansion as long as $\frac{g^2N}{(4\pi^2)}\ll\frac{11}{48}$.

1628: 

1629: In figure {\ref{fig:N0Flow}

1630: \begin{figure}[b]

1631: \begin{center}

1632: \setlength{\unitlength}{0.00016667in}

1633: \begin{picture}(32000,13000)

1634: %/put(0,0){(0,0)}

1635: %/put(32000,13000){(1,1)}

1636: \put(0,0){\psfig{figure=N0Flow.eps,width=2.5in}}

1637: \put(16000,0){\psfig{figure=N0FlowZoom.eps,width=2.5in}}

1638: \put(12400,11800){$g_1^2$}

1639: \put(15000,9600){$g_N^2,Y_N^2,Y_1^2$}

1640: \put(15200,800){$g_1^2$}

1641: \put(30800,12200){$g_N^2$}

1642: \put(31200,7200){$g_1^2$}

1643: \put(27800,0){$Y_N^2,Y_1^2$}

1644: \end{picture}

1645: \end{center}

1646: \isucaption{The RG flow of the $\N=0$ theory from the natural

1647:   line $g_1=g_N=Y_1=Y_N=0.01$.

1648:   The graph was plotted for $N=5$.

1649:   In the IR we get a free theory.

1650:   The plot on the right is a zoom on the natural line.

1651: }

1652: \label{fig:N0Flow}

1653: \end{figure}

1654: we plot the numerical solution of the $\beta$ functions.

1655: We start the RG flow from a point on the natural line.

1656: In the IR all the couplings flow to zero. 

1657: The $\beta$ functions \eqref{eqn:N0beta} were calculated assuming that all

1658: the fields are massless, so we can trust them only as long as the

1659: energy scale is larger than the fields masses. Still, if we get an IR free

1660: theory when ignoring the masses, we will get an IR free theory also with

1661: the masses taken into account.

1662: 

1663: Zooming on the natural point we see that the $SU(N)$ gauge coupling has

1664: a local minimum on the natural line while the Yukawa couplings have

1665: a local maximum. This can be calculated directly from the second derivative

1666: of the couplings (first derivative of the $\beta$ function).

1667: 

1668: It is possible that higher loop calculations will give more interesting

1669: results than the flow to the trivial fixed point.

1670: This can be seen by analyzing the $\beta$ functions around

1671: the unstable fix point \eqref{eqn:N0FixPoint}.

1672: There seem to be solutions in which all or some of the couplings flow

1673: to infinity in the IR.

1674: 

1675: \section{Summary and Discussion}

1676: \label{sec:Summary}

1677: 

1678: We analyzed the finite $N$ limit of orbifold field theories.

1679: For finite $N$ the remnant $U(1)$ factors from the orbifolding procedure

1680: should be considered. 

1681: The $U(1)$ factors seem to cancel the $\frac{1}{N}$ contributions in Feynman

1682: diagrams \eqref{eqn:C2}, at least up to three loops.

1683: This encourages us to study the orbifolded theory on the natural line where

1684: all the couplings are equal.

1685: 

1686: The simplest case to analyze is of the $\N=2$ orbifolds, but it turns out that

1687: it is too simple. There is no interaction between the different gauge

1688: groups due to the non renormalization theorem.

1689: The $SU(N)$ couplings are finite and the $U(1)$ couplings are IR free.

1690: The theory on the natural line is not finite, but we can easily make it

1691: finite by choosing zero $U(1)$ couplings or by taking the IR limit.

1692: 

1693: The case of $\N=1$ orbifolds is more interesting because here the $U(1)$

1694: factors do affect the rest of the gauge couplings. Still, the final

1695: conclusions are the same as in the $\N=2$ case. The theory on the natural

1696: line is not conformal, but in the IR limit it flows to a point on the fixed

1697: line.

1698: 

1699: In the case of $\N=0$ orbifolds all hell breaks loose.

1700: The fact that we get the $\N=0$ theory from the $\N=4$ SCFT leads to

1701: ``miraculous'' cancellation of Feynman diagrams up to three loops on

1702: the natural line,

1703: yet this does not seem to be enough. Quantum effects can generate new

1704: terms that did not exist in the original theory, like the mass term

1705: of the $U(1)$ scalar \eqref{eqn:U1mass}.

1706: The theory seems to be inconsistent unless we add a mass term to it,

1707: but adding a mass term is inconsistent with our attempt to look at

1708: the orbifolded theory exactly as it emerges from the orbifolding process.

1709: 

1710: The scalar mass diverges polynomially, meaning that it is scheme depended.

1711: We hope that there is a regularization scheme in which all the polynomial

1712: divergences cancel out.

1713: The alleged scheme might be defined from the AdS/CFT correspondence

1714: by adding massive fields that correspond to massive open strings

1715: between D3 branes in the orbifolded type IIB string theory.

1716: 

1717: There is also a problem of renormalization scheme dependence related to our

1718: claims on the cancellation of 3-loop Feynman diagrams.

1719: All the explicit calculations

1720: of the $\beta$ functions we performed where scheme independent, but the 3-loop

1721: Feynman diagrams are scheme dependent. Again, we are not sure in what scheme

1722: our claims are valid, but we assume that such a scheme exists.

1723: 

1724: Another property of the $\N=0$ theories is that we can no longer assume

1725: that the $U(1)$ factors decouple in the IR, leaving us with an

1726: $SU(N)^{|\Gamma|}$ theory. 

1727: Our analysis shows that, at least for small couplings, the entire

1728: theory is IR free. Moreover, if we start with the same Yukawa couplings

1729: for the $SU(N)$ scalars and the $U(1)$ scalars,

1730: then our calculation shows that at least to one loop order,

1731: they will remain the same.

1732: 

1733: The five loop bubble diagram \eqref{eqn:5Loops} hindered us from claiming

1734: that the cancellation of the Feynman diagrams continues to all orders.

1735: There are a few hints that this cancellation might survive to all orders,

1736: based on the fact that the $\N=4$ theory is finite to all orders.

1737: 

1738: The first hint is that the proof that the correlation functions of

1739: orbifold theories coincide

1740: with those of $\N=4$ can be generalized to non planar diagrams.

1741: The proof is valid for abelian orbifolds,

1742: at least as long as all the external legs are on the same face.

1743: In \cite{Bershadsky:98} it was shown, that non planar

1744: diagrams in the orbifold theories are different from the $\N=4$ diagrams,

1745: using as example, the non planar diagram \eqref{eqn:NonPlanar}.

1746: However, in eq. (13) there is a missing $\gamma_3$ factor.

1747: After adding it, there is a

1748: match in the non planar diagram \eqref{eqn:NonPlanar}

1749: between the orbifold and $\N=4$ theories, at least for abelian

1750: orbifolds.

1751: 

1752: The second hint comes from dividing the Feynman diagrams of each order

1753: into subsets defined by their $N$ dependence.

1754: For example, five loop bubble diagrams can be divided into two subsets,

1755: ``calculable'' diagrams with a group factor of $N^4d(G)$ and diagrams with

1756: subleading $N$ terms (like the diagram in \eqref{eqn:5Loops}).

1757: 

1758: We know that the $\N=4$ theory is finite to all orders independent of

1759: the coupling $g$, meaning that there is a cancellation of the Feynman

1760: diagrams at every order. The finitude of the $\N=4$ theory also does

1761: not depend on $N$ meaning that there is a cancellation of the Feynman 

1762: diagrams in each subgroup defined above.

1763: 

1764: We can use this cancellation to claim that the five loop diagrams in the

1765: first subset must have zero contribution to orbifold theories.

1766: However, we can not make the same claim for the second subset because

1767: in that subset there are different diagrams with different $N$

1768: dependence.

1769: 

1770: We have only analyzed orbifolds from the field theory point of view.

1771: It would be interesting to find a manifestation of the $U(1)$ running

1772: coupling constants in string theory orbifolds and in brane configurations. 

1773: The brane configuration for the $\N=2$ theory, for example,

1774: is a set of $|\Gamma|$

1775: NS5 branes on a circle with $N$ D4 branes stretched between them.

1776: The decoupling of the $U(1)$ factors can be seen directly from the brane

1777: configuration \cite{Witten:97}.

1778: 

1779: In the AdS/CFT correspondence it would be interesting to find string states

1780: corresponding to operators with $U(1)$ factors. The correspondence was

1781: done in \cite{Oz:98} for orbifolds in the large $N$ limit. We do not know

1782: how to generalize it for finite $N$, but it might be possible to use our

1783: knowledge in field theory to gain some insight of the string theory.

1784: 

1785: In addition to the conformal $SO(4,2)$ symmetry, the original $\N=4$ theory

1786: has an $SL(2,\Z)$ symmetry. It would be interesting to analyze

1787: the effect of the orbifold on this symmetry.

1788: The $SL(2,\Z)$ symmetry acts on the coupling constant and it is not clear

1789: whether it has any meaning in non conformal theories.

1790: In \cite{Bergman:99} it was suggested that $SL(2,\Z)$ is also a symmetry

1791: of the type 0B string. If this is true, then

1792: it would be interesting to investigate the

1793: manifestation of the $SL(2,\Z)$ symmetry in the non supersymmetric

1794: non conformal ``type 0'' field theory.

1795: 

1796: To summarize, orbifolds of the $\N=4$ field theory give

1797: us non supersymmetric non conformal theories with very interesting features.

1798: The exquisite cancellation of the vacuum bubble diagrams in those theories

1799: up to at least four loop order suggests that

1800: the cosmological constant is very small in those theories.

1801: Consequently, the cosmological constant problem could be solved in

1802: the orbifolded theories.

1803: We do not claim that the cosmological constant vanishes completely in

1804: those theories because the running of the $U(1)$ factors stimulate the

1805: running of the bubble diagrams.

1806: 

1807: For finite $N$ the $U(1)$ factors also behave as soft symmetry breaking terms

1808: of the conformal symmetry in the sense that the flow of the $U(1)$

1809: couplings induces the flow of the other couplings.

1810: The ``soft breaking'' parameter is $\frac{1}{N}$.

1811: The conformal symmetry is broken explicitly,

1812: but it is broken by terms that occur naturally in the orbifold process.

1813: We propose this soft symmetry breaking term for solving the hierarchy problem

1814: as suggested in \cite{Frampton:99}.

1815: 

1816: \subsection*{Acknowledgments}

1817: 

1818: I would like to thank Amihay Hanany, Vadim Kaplunovsky, Shimon Yankielowicz

1819: and especially Jacob Sonnenschein for many useful discussions.

1820: I would also like to thank Angel Uranga for pointing out to me the

1821: anomalies in the chiral orbifolds.

1822: 

1823: %\newpage

1824: \bibliography{hep}

1825: \bibliographystyle{utcaps}

1826: 

1827: \end{document}

1828: