1: \documentstyle[12pt,epsf]{article}

2: 

3: 

4: \renewcommand{\topfraction}{0.8}

5: \renewcommand{\bottomfraction}{0.8}

6: \renewcommand{\baselinestretch}{1.3}

7: % shortcuts

8: \newcommand{\bmat}{\left(\begin{array}}

9: \newcommand{\emat}{\end{array}\right)}

10: \def\NPB#1#2#3{Nucl. Phys. B{#1} (19#2) #3}

11: \def\PLB#1#2#3{Phys. Lett. B{#1} (19#2) #3}

12: \def\PLBold#1#2#3{Phys. Lett. {#1B} (19#2) #3}

13: \def\PRD#1#2#3{Phys. Rev. D{#1} (19#2) #3}

14: \def\PRL#1#2#3{Phys. Rev. Lett. {#1} (19#2) #3}

15: \def\PRT#1#2#3{Phys. Rep. {#1} C (19#2) #3}

16: \def\MODA#1#2#3{Mod. Phys. Lett.  {#1} (19#2) #3}

17: %\def\mb#1{\mbox{\boldmath $#1$}}

18: \def\Deq#1{\mbox{$D$=#1}}

19: \def\Neq#1{\mbox{$N$=#1}}

20: \def\preal{{\rm Re\,}}

21: \def\pim{{\rm Im\,}}

22: \def\ds{\displaystyle}

23: \def\yzero{\smash{\hbox{$y\kern-4pt\raise1pt\hbox{${}^\circ$}$}}}

24: \def\p{\partial}

25: \def\a{\alpha}

26: \def\b{\beta}

27: \def\g{\gamma}

28: \def\d{\delta}

29: \def\p{\tilde p}

30: \def\Om{\Omega}

31: \def\om{\omega}

32: \def\th{\theta}

33: \def\vt{\vartheta}

34: \def\vphi{\varphi}

35: \def\-{\hphantom{-}}

36: \def\ov{\overline}

37: \def\s2{\frac{1}{\sqrt2}}

38: \def\wh{\widehat}

39: \def\wt{\widetilde}

40: \def\oh{\frac{1}{2}}

41: \def\beq{\begin{equation}}

42: \def\eeq{\end{equation}}

43: \def\beqa{\begin{eqnarray}}

44: \def\eeqa{\end{eqnarray}}

45: \def\tr{{\rm tr \,}}

46: \def\Tr{{\rm Tr \,}}

47: \def\diag{{\rm diag \,}}

48: \def\IF{\relax{\rm I\kern-.18em F}}

49: \def\II{\relax{\rm I\kern-.18em I}}

50: \def\IP{\relax{\rm I\kern-.18em P}}

51: \def\inbar{\vrule height1.5ex width.4pt depth0pt}

52: \def\IC{\relax\hbox{\kern.25em$\inbar\kern-.3em{\rm C}$}}

53: \def\IR{\relax{\rm I\kern-.18em R}}

54: \def\hm{\relax{n_H}}

55: \def\vac{|0 \rangle}

56: \def\vm{\relax{n_V}}

57: \def\cc{{\cal C}}

58: \def\ck{{\cal K}}

59: \def\ci{{\cal I}}

60: \def\cu{{\cal U}}

61: \def\cg{{\cal G}}

62: \def\cn{{\cal N}}

63: \def\cm{{\cal M}}

64: \def\cp{{\cal P}}

65: \def\ct{{\cal T}}

66: \def\cv{{\cal V}}

67: \def\cz{{\cal Z}}

68: \def\ch{{\cal H}}

69: \def\cf{{\cal F}}

70: \def\tv{\tilde v}

71: \def\Dsl{\,\raise.15ex\hbox{/}\mkern-13.5mu D} %this one can be subscripted

72: %\def\IZ{Z\kern-.5em  Z}

73: \def\IZ{Z}

74: \def\id{{\rm I}}

75: %\def\cp#1{\relax\ifmmode {\IP\kern-2pt{}_{#1}}\else $\IP\kern-2pt{}_{#1}$\fi}

76: 

77: % draw box with width #1pt and line thickness #2pt

78: \newcommand{\drawsquare}[2]{\hbox{%

79: \rule{#2pt}{#1pt}\hskip-#2pt%  left vertical

80: \rule{#1pt}{#2pt}\hskip-#1pt%  lower horizontal

81: \rule[#1pt]{#1pt}{#2pt}}\rule[#1pt]{#2pt}{#2pt}\hskip-#2pt%  upper horizontal

82: \rule{#2pt}{#1pt}}% right vertical

83: 

84: % Young tableaux

85: \newcommand{\fund}{\raisebox{-.5pt}{\drawsquare{6.5}{0.4}}}%  fund

86: \newcommand{\Ysymm}{\raisebox{-.5pt}{\drawsquare{6.5}{0.4}}\hskip-0.4pt%

87:         \raisebox{-.5pt}{\drawsquare{6.5}{0.4}}}%  symmetric second rank

88: \newcommand{\Yasymm}{\raisebox{-3.5pt}{\drawsquare{6.5}{0.4}}\hskip-6.9pt%

89:         \raisebox{3pt}{\drawsquare{6.5}{0.4}}}%  antisymmetric second rank

90: \newcommand{\antifund}{\overline{\fund}}

91: \newcommand{\bYasymm}{\overline{\Yasymm}}

92: 

93: 

94: 

95: 

96: \topmargin

97: -1.5cm

98: \textwidth

99: 15.5cm

100: \textheight

101: 24cm

102: \oddsidemargin

103: 0.7cm

104: \evensidemargin

105: 1.2cm

106: 

107: \begin{document}

108: 

109: 

110: %----------------------------------------------------------------------%

111: %  numbering equations with section number

112: %----------------------------------------------------------------------%

113: \makeatletter

114: \@addtoreset{equation}{section}

115: \makeatother

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

117: %----------------------------------------------------------------------%

118: %  title page

119: %----------------------------------------------------------------------%

120: \pagestyle{empty}

121: %\vspace*{1.0in}

122: \rightline{IFT UAM-CSIC/03-27}

123: \rightline{\tt hep-th/0307196}

124: \vspace{0.5cm}

125: \begin{center}

126: \LARGE{From UV/IR mixing to closed strings\\[20mm]}

127: \large{

128: Esperanza L\'opez\\[5mm]}

129: \small{

130: Departamento de F\'{\i}sica Te\'orica C-XI

131: and Instituto de F\'{\i}sica Te\'orica  C-XVI\\[-0.3em]

132: Universidad Aut\'onoma de Madrid,

133: Cantoblanco, 28049 Madrid, Spain\\

134: Esperanza.Lopez@uam.es\\[4mm]}

135: \small{\bf Abstract} \\[7mm]

136: \end{center}

137: 

138: \begin{center}

139: \begin{minipage}[h]{14.0cm}

140: 

141: It was shown in hep-th/0301099 that the leading UV/IR

142: mixing effects in noncommutative gauge theories on D-branes

143: are able to capture information about the

144: closed string spectrum of the parent string theory.

145: The analysis was carried out for D-branes on nonsupersymmetric

146: $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds of Type IIB.

147: In this paper we consider D-branes on twisted circles 

148: compactifications of Type II string theory. We find 

149: that the signs of the leading UV/IR mixing effects know about the 

150: mass gap between the lowest modes in NSNS and RR closed 

151: string towers. Moreover, the relevant piece of the field theory 

152: effective action can be reproduced purely

153: in the language of closed strings. Remarkably this approach

154: unifies in a single structure, that of a closed string exchange

155: between D-branes, both the leading planar and nonplanar

156: effects associated to the absence of supersymmetry.

157: 

158: 

159: 

160: 

161: 

162: 

163: 

164: 

165: 

166: \end{minipage}

167: \end{center}

168: \newpage

169: %----------------------------------------------------------------------%

170: %  Resetting of counters

171: %----------------------------------------------------------------------%

172: \setcounter{page}{1}

173: \pagestyle{plain}

174: \renewcommand{\thefootnote}{\arabic{footnote}}

175: \setcounter{footnote}{0}

176: %----------------------------------------------------------------------%

177: %  Paper begins

178: %----------------------------------------------------------------------%

179: 

180: \section{Introduction}

181: 

182: The study of noncommutative field theories has attracted much interest

183: in the last years because noncommutativity is expected to capture basic 

184: aspects of the sort distances behavior of gravity \cite{Doplicher:1994tu}. 

185: This point of view is supported by the central role that D-branes 

186: play in the matrix model description of M-theory \cite{Banks:1996vh},

187: since D-branes provide a fundamentally noncommutative description of 

188: space-time \cite{Witten:1995im}. 

189: 

190: In order to explore the consequences of noncommutativity, most works 

191: have considered the simple deformation of $\mathbf{R}^n$

192: \beq

193: [x^\mu , x^\nu]=i \theta^{\mu \nu} \, ,

194: \label{noncom}

195: \eeq

196: where $\theta^{\mu \nu}$ is an antisymmetric matrix with constant entries.

197: The reason to focus in (\ref{noncom}) is that explicit calculations 

198: can be done for field theories living on such spaces. (\ref{noncom}) 

199: implies uncertainty relations in space-time and therefore 

200: the nondecoupling of ultraviolet and infrared degrees of freedom.

201: This new behavior has drastic consequences for the dynamics of the

202: theory and in particular for its nonplanar sector: It was shown in 

203: \cite{Minwalla:1999px} that UV/IR mixing leads

204: to the appearance of new infrared divergences in nonplanar graphs, 

205: whose origin is in the integration to high momenta in loops. 

206: 

207: The relations (\ref{noncom}) are naturally realized on the world-volume

208: of D-branes in a constant B-field background, where 

209: $\theta^{\mu \nu} \sim 1/B_{\mu \nu}$ \cite{Douglas:1997fm}. Nonplanar 

210: field theory diagrams can then be related to nonplanar string diagrams.

211: This suggests an important role of closed strings in the understanding 

212: of UV/IR mixing \cite{Minwalla:1999px,Arcioni:2000bz}. 

213: More generically, it is natural to expect that if 

214: (\ref{noncom}) captures relevant aspects of quantum gravity its effects 

215: will know about the closed string sector (although the decoupling of

216: closed strings {\it does not} fail in the noncommutative field theory 

217: limit \cite{Gomis:2000bn,Liu:2000qh}). In this line, a remarkable 

218: relation between UV/IR mixing effects and {\it properties} of the closed 

219: string spectrum was uncovered in \cite{Armoni:2003va}. 

220: The leading IR divergences due to UV/IR mixing strongly modify 

221: the dispersion relations of the theory and in some cases render the 

222: perturbative vacuum unstable \cite{Minwalla:1999px}. 

223: A one to one correspondence

224: between these noncommutative instabilities and the presence of 

225: closed string tachyons in the parent string theory was obtained

226: for gauge theories on D-branes 

227: at nonsupersymmetric $\mathbf{C}^3/\mathbf{Z}_N$ orbifold singularities.

228: 

229: ${\cal N}=4$ noncommutative $U(1)$ at finite temperature

230: presents a non-trivial but mild version

231: of UV/IR mixing, where no infrared divergence develops.

232: In spite of that, for temperatures bigger than a critical $T_c$ 

233: excitations of tachyonic nature appear in the system 

234: \cite{Landsteiner:2001ky}. A closely related 

235: theory arises on D3-branes wrapped

236: along a Scherk-Schwarz circle. It can be seen that noncommutative 

237: instabilities set up well below the threshold for the appearance of 

238: closed string tachyons, invalidating a direct relation between both

239: phenomena. A motivation of this paper is to determine whether the leading

240: UV/IR mixing effects know about properties of the closed string spectrum 

241: in this more general case.

242: 

243: The paper is organized as follows. In section 2 we summarize the 

244: results obtained in \cite{Armoni:2003va} for noncommutative 

245: $\mathbf{C}^3/\mathbf{Z}_N$ quiver theories. In section 3 we consider 

246: Type II string theory compactified on twisted circle backgrounds 

247: \cite{Russo:2001na}, which are generalizations of a Scherk-Schwarz 

248: compactification. The closed string spectrum contains tachyons for

249: sufficiently small radius of the circle. In section 3 and 4 we analyze 

250: Type IIB D3-branes wrapped along the twisted circle. In section 5 we 

251: study Type IIA D2-branes transverse to the twisted circle. The field 

252: theory limit on D2-branes and D3-branes corresponds to

253: the string regimes with and without closed string tachyons respectively.

254: The sign of the leading UV/IR mixing effects determines if the associated 

255: correction will tend to destabilize or not the perturbative vacuum.

256: For the gauge theories on both D2- and D3-branes we find that these 

257: signs are correlated with the $($mass$)^2$ difference between the 

258: lowest NSNS and RR modes in each winding sector. This will 

259: be our proposal on how to generalize the results of \cite{Armoni:2003va}.

260: Notice that for $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds the RR zero-point 

261: energy is zero and thus the sign of the gap indicates the presence or not 

262: of closed string tachyons. However for twisted circle compactifications 

263: the mass of closed strings, both in NSNS and RR sectors, is shifted by a 

264: nonzero winding energy.

265: 

266: We include in section 2 a proposal for deriving, purely in the language 

267: of closed strings, the piece of the gauge invariant effective action 

268: responsible for the instabilities \cite{Armoni:2001uw}. This 

269: provides a very direct relation between UV/IR mixing effects and closed 

270: strings. A second goal of this paper is to extend this derivation to the 

271: gauge theories on D2- and D3-branes at twisted circle backgrounds.

272: We will show in sections 4 and 5 that this is indeed

273: possible. All the leading UV/IR mixing effects in nonplanar 

274: graphs, as well as the leading planar corrections, have the structure 

275: of a closed string exchange. The possibility to unify planar 

276: and nonplanar effects in a single structure shreds light on  

277: the puzzling different behavior of planar and nonplanar sectors.

278: We present our conclusions in section 6.

279: In an appendix we include the derivation of some useful field theory

280: results.

281: 

282: 

283: \section{Open Wilson lines versus closed strings}

284: 

285: %%%%%%%%%%%%%%%%%%%%%%%%

286: \begin{figure}

287: \centering

288: \epsfxsize=3.5in

289: \hspace*{0in}\vspace*{0in}

290: \epsffile{channel.eps}

291: \caption{\small Schematic representation of a non-planar graph.}

292: \label{channel}

293: \end{figure}

294: %%%%%%%%%%%%%%%%%%%%%%%%%

295: 

296: Non-planar graphs in noncommutative field theories posses an

297: intrinsic UV cutoff in the form of phases whose argument oscillates 

298: rapidly for high loop momenta. When the momentum exchange in 

299: the non-planar channel vanishes (see Fig. 1), these phases trivialize 

300: and the regularization stops being effective. This implies 

301: $\Lambda_{np} \sim {1 \over P \theta}$ and, as a consequence, the

302: translation of UV into IR behavior. In its most extreme form, it 

303: converts potential UV divergences into IR divergences 

304: \cite{Minwalla:1999px}. 

305: 

306: The leading IR divergences modify the dispersion relations of 

307: noncommutative field theories as follows 

308: \beq

309: E^2= {\vec p}^2 - c \; {g^2 \over \p^2} \, ,

310: \label{disp}

311: \eeq

312: where $g$ is the (dimensionless) coupling constant, 

313: $\p^\mu=\theta^{\mu \nu} p_\nu$ and $c$ is a model dependent constant. 

314: For non-commutative gauge theories $c\sim N_b-N_f$, with $N_b$ and $N_f$ 

315: the number of bosonic and fermionic degrees of freedom in the adjoint 

316: representation. Supersymmetric theories have a softer UV behavior and 

317: thus the leading IR divergences are absent: $c=0$ 

318: \cite{Matusis:2000jf}. When $N_b>N_f$, (\ref{disp}) turns the low 

319: momentum modes unstable \cite{Ruiz:2001hu,Landsteiner:2001ky}. 

320: 

321: In order to explore a possible connection between noncommutative instabilities 

322: and closed string tachyons, it proved useful in \cite{Armoni:2003va} to 

323: study a family of theories with a sufficiently rich phenomenology. This 

324: was provided by noncommutative D-branes on $\mathbf{C}^3/\mathbf{Z}_N$ 

325: orbifold backgrounds. 

326: The orbifold acts with twist $(a_1,a_2,a_3,a_4)/N$ and $(b_1,b_2,b_3)/N$ 

327: on $SO(6)$ spinors and vectors respectively. The integers $a_\alpha$ are 

328: subject to $\sum a_\alpha=0({\rm mod} N)$ and are related to

329: $b_l$ by $b_1=a_2+a_3$, $b_2=a_1+a_3$, $b_3=a_1+a_2$.

330: The gauge theory on $n$ D3-branes placed at the fixed 

331: point of the orbifold has gauge group $G=\otimes_{i=1}^N U(n_i)$, where 

332: $\sum n_i=n$. The coupling constants of all gauge group factors coincide.

333: The matter content is given by 

334: $(\fund_i,\antifund_{i+a_\alpha})$ Weyl fermions and $(\fund_i,

335: \antifund_{i+b_l})$ complex scalars. Both the gauge theory 

336: on the D3-branes and the closed string spectrum are supersymmetric 

337: if at least one $a_\alpha=0({\rm mod} N)$. 

338: 

339: Turning on a B-field background on two of the spatial directions of the 

340: D3-branes will render the world-volume noncommutative, {\it i.e.} 

341: $[x^1,x^2]=i \theta$. In the generic nonsupersymmetric case the 

342: noncommutative gauge theory 

343: presents a complicated pattern of UV/IR mixing effects. The leading 

344: infrared contribution to the nonplanar polarization tensor was calculated 

345: in \cite{Armoni:2003va}. As can be seen 

346: from Fig. 2, it affects only $U(1)_i \in U(n_i)$ degrees of 

347: freedom\footnote{In the absence of noncommutativity the theories we are 

348: considering have generically mixed anomalies, that will lead to the

349: massification of the $U(1)$'s through a Green-Schwarz

350: mechanism \cite{Ibanez:1998qp}. When $\theta \neq 0$ the situation is 

351: more subtle: the $U(1)$'s become massive and decouple only for $\p=0$ 

352: \cite{Armoni:2002fh}, while for $\p \neq 0$ the anomaly vanishes

353: \cite{Martin:2000qf}. We will always consider

354: the latter case.}

355: and is non-diagonal in group labels indices. However the linear 

356: combinations $B^{(k)}_\mu= {1 \over \sqrt{N}} \sum e^{2 \pi i {jk \over N}} 

357: {\rm Tr} A^{(j)}_\mu \,$ diagonalize it, with the result

358: \beq

359: \Pi^{\mu \nu}_k= 

360: \epsilon_k \; {g^2 \over \pi^2} {\p^\mu \p^\nu \over \p^4} \, .

361: \label{pior}

362: \eeq

363: The quantities $\epsilon_k$, which play an analogous role to $c$ in 

364: (\ref{disp}), have a simple expression in terms

365: of the orbifold twist parameters 

366: \beq

367: \epsilon_k = 2 \left( 1-\sum_{\alpha=1}^4 

368: \cos {2 \pi a_\alpha k \over N} +\sum_{l=1}^3 

369: \cos {2 \pi b_l k \over N} \right) \, .

370: \label{ep}

371: \eeq 

372: 

373: Remarkably, these quantities can be rewritten in terms of the 

374: masses of four low lying closed string modes in the NSNS $k^{th}$ twisted 

375: sector of the orbifold background

376: \beq

377: \epsilon_k=-16 \prod_{\alpha=1}^4 \sin {\pi \alpha' m_\alpha^2 \over 2} \, .

378: \label{emass}

379: \eeq

380: These masses satisfy the following properties: {\it i)}

381: among them is the lowest mode in the NSNS $k^{th}$ twisted sector;

382: {\it ii)} $-1 \leq \alpha' m_\alpha^2 < 2$; {\it iii)} only one

383: of the $m_\alpha^2$ can be negative (see \cite{Armoni:2003va} for details). 

384: This implies a direct relation between the sign 

385: of $\epsilon_k$ and the presence or not of tachyons in the $k^{th}$ twisted 

386: sector of the parent string theory: $\epsilon_k>0$ if the $k^{th}$ twisted

387: sector contains tachyons; $\epsilon_k=0$ only if the $k^{th}$ twist

388: is supersymmetry-preserving; 

389: $\epsilon_k<0$ when the twisted sector

390: is nonsupersymmetric but does not include tachyons. 

391: This last situation does not

392: arise in $\mathbf{C}^n/\mathbf{Z}_N$ orbifolds with $n<3$, and was the reason 

393: to consider orbifolds of $\mathbf{C}^3$. 

394: 

395: 

396: %%%%%%%%%%%%%%%%%%%%%%%%

397: \begin{figure}

398: \centering

399: \epsfxsize=2.5in

400: \hspace*{0in}\vspace*{0in}

401: \epsffile{nonplanar.eps}

402: \caption{\small Non-planar contribution to the polarization tensor.

403: $i,j$ are group label indices and $a,b$ gauge group indices.}

404: \label{nonplannar}

405: \end{figure}

406: %%%%%%%%%%%%%%%%%%%%%%%%%

407: 

408: Let us come back now to the origin of UV/IR mixing, which is in the

409: uncertainty relations derived from (\ref{noncom}): 

410: $\Delta x^1 \Delta x^2 \geq \theta$ in our case. This phenomenon is 

411: present in any noncommutative theory, independently whether 

412: there are UV or IR divergences. It would seem then that the natural degrees 

413: of freedom of noncommutative theories are those that made this behavior 

414: evident. Contrary to this expectation, noncommutative field theories 

415: are formulated using local fields. 

416: This puzzle is solved by studying the effective action of the theory.

417: It has been shown both for scalar \cite{Kiem:2001pw} and  

418: gauge theories \cite{VanRaamsdonk:2001jd,Armoni:2001uw,Kiem:2001dm}

419: that the 1-loop nonplanar effective action, including contributions from all 

420: the N-point functions, can be rewritten in terms of straight open Wilson line 

421: operators \cite{Ishibashi:2000hs,Rey:2000hh,Das:2000md,Gross:2000ba}. 

422: This result was extended 

423: in \cite{Kiem:2001du} to the 2-loop effective action of scalar theories. 

424: Straight open Wilson line operators exhibit the desired behavior

425: since their momentum, $p$, is correlated with their transversal extent

426: $\p$,

427: \beq

428: {\widetilde W}(p)\, = \, {\rm Tr} \int d^4 x \; P_\ast 

429: \left(e^{i\, g \int_0^1 d \sigma 

430: \, {\tilde p}^\mu A_{\mu}(x+{\tilde p}\, \sigma)}\right) 

431: \ast e^{i p x} \, .

432: \label{w}

433: \eeq

434: For the orbifold gauge theories above, the gauge invariant piece

435: of the 1-loop effective action containing (\ref{pior}) has the

436: simple expression \cite{Armoni:2003va}

437: \beq

438: \Delta S = {1 \over 2 \pi^2} \sum_{k=0}^{N-1} \epsilon_k

439: \int {d^4 p \over (2 \pi)^4} \; {1 \over {\tilde p}^4} \; 

440: W^{(N\!-k)}(p) \, W^{(k)}(-p) \, ,

441: \label{invaction}

442: \eeq

443: where $W^{(k)} = {1 \over \sqrt{N}}\sum_{j=1}^N e^{2 \pi i \, {jk \over N}} 

444: {\widetilde W}^{(j)}$, and ${\widetilde W}^{(j)}$ denotes (\ref{w}) with

445: the vector field belonging to the $j^{th}$ gauge group factor.

446: 

447: Several facts suggest the interpretation of (\ref{invaction}) in terms

448: of a closed string exchange between D-branes. (\ref{invaction}) seems 

449: to know about the different closed string twisted sectors since the 

450: quantities $\epsilon_k$ measure an independent property of each sector. 

451: In addition, closed string modes in the $k^{th}$ twisted sector couple to 

452: linear combinations of field theory operators such as those that define 

453: $B_\mu^k$ and $W^k$ \cite{Douglas:1996sw}. Moreover, it has been 

454: shown in \cite{Das:2000ur,Okawa:2000sh,Liu:2001ps} that closed strings 

455: couple to straight Wilson line operators on noncommutative D-branes.

456: We should notice that closed strings can only couple to gauge invariant 

457: operators and the open Wilson lines provide the basic set of such operators in 

458: noncommutative gauge theories \cite{Gross:2000ba}. However, the condition 

459: of gauge invariance only determines the separation between the start and end 

460: point of the Wilson line but not the path.

461: 

462: Reviewing \cite{Armoni:2001uw,Armoni:2003va}, we will propose now a direct 

463: derivation of (\ref{invaction}) in terms of a closed strings. Our motivation 

464: is 

465: twofold. We have mentioned two aspects of UV/IR mixing, the first associated 

466: to the physics of the non-planar sector and the appearance of an extreme

467: infrared behavior, the second associated to the question of what are

468: the natural noncommutative excitations and the role of open Wilson line 

469: operators. Closed strings provide a connection between these two

470: aspects since they couple naturally to open Wilson lines and, as we will

471: see, the IR divergent term $1/\p^4$ in (\ref{invaction}) can be

472: related to a closed string propagator.

473: Our second goal is to obtain an alternative understanding of the 

474: $\epsilon_k$'s and its relation to closed string tachyons.

475: 

476: In the absence of B-field, scalar closed string modes couple to the

477: brane tension at leading order in $\alpha'$: $T \sim 

478: {\rm Tr} \, {\mathbf 1}/\alpha'^2$. When $B\neq 0$ the trivial field 

479: theory operator ${\rm Tr} \,{\mathbf 1}$ gets promoted

480: to the open Wilson line operator (\ref{w}). Hence, at leading order in 

481: $\alpha'$, the contribution to the D3-brane effective action from the 

482: emission, propagation and posterior absorption of a scalar closed string 

483: mode $\varphi$ in the $k^{th}$ twisted sector is  

484: \beq

485: \Delta S= |D_\varphi|^2 \int {d^4 p \over (2 \pi)^4} \, 

486: W^{(N-k)}(p) W^{(k)}(-p) \, f({\tilde p},u) \, ,

487: \label{staction}

488: \eeq

489: with $D_\varphi$ a numerical factor proportional to the disk amplitude 

490: of $\varphi$ with no open string insertion. The function $f({\tilde p},u)$

491: denotes the closed string propagator

492: \beq

493: f({\tilde p},u)= {\alpha'}^{-{d+2 \over 2}} \int {d^d v \over

494: (2 \pi)^d} {e^{i v u} \over v^2+{\tilde p}^2+(2 \pi \alpha' m_\varphi)^2} \, .

495: \label{f}

496: \eeq

497: $d$ is the number of dimensions

498: transverse to the D-brane where the twisted field $\varphi$ can propagate: 

499: $d=0,2,4,6$ depending on the particular $\mathbf{C}^3/\mathbf{Z}_N$ 

500: orbifold. We have defined $v=2 \pi \alpha' p_\perp$, with $p_\perp$ the 

501: transversal momentum to the D-brane. $u$ has been introduced in order to have 

502: a well defined

503: closed string propagator, it has the interpretation of an infrared regulator 

504: from the point of view of the field theory. In the denominator we have used 

505: the relation between open ($\eta$) and closed ($g$) string metrics 

506: $g^{-1}=\eta^{-1}\!-\! \theta \eta \theta/(2 \pi \alpha')^2$ 

507: \cite{Seiberg:1999vs}, and discarded

508: terms suppressed by two $\alpha'$ powers; $m_\varphi$ is the mass of $\varphi$.

509: The factor of $\alpha'$ in front of the integral can be

510: obtained just by dimensional analysis. 

511: 

512: From (\ref{staction}) we want to extract

513: a contribution to the noncommutative field theory effective action.

514: We need to take the limit $\alpha' \rightarrow 0$. In this limit $f$ 

515: diverges due to the negative $\alpha'$ power in front of the integral. 

516: However we should notice the following.

517: If we had done the $\alpha' \rightarrow 0$ limit of the standard annulus 

518: diagram associated to each nonplanar N-point function, 

519: we would had obtained a result ${\cal O}({\alpha'}^0)$ whose leading

520: IR contribution should reproduce the corresponding term in (\ref{invaction}). 

521: This has been checked for the 2-point function in 

522: \cite{Bilal:2000bk,Gomis:2000bn}. The question is then whether

523: we can directly define an ${\cal O}({\alpha'}^0)$ contribution

524: from (\ref{staction}), regarding as artifacts other $\alpha'$ powers.

525: We observe that $\p$ acts as an infrared regulator for the integral in 

526: (\ref{f}). Therefore, for $\p \neq 0$, we can expand the integral in powers 

527: of $(2 \pi \alpha' m_\varphi)^2 \sim \alpha'$ to the desired order.

528: At ${\cal O}({\alpha'}^0)$ we obtain

529: \beq

530: f({\tilde p},u)|_{{\cal O}(\alpha'^0)}\, \sim \, \int {d^d v \over

531: (2 \pi)^d} {e^{i v u} \over (v^2+{\tilde p}^2)^{{d \over 2}+2}} 

532: \rightarrow_{u \rightarrow 0} {c \over \p^4} \, ,

533: \label{ffin}

534: \eeq

535: where $c$ is a constant that depends on the dimension $d$ and whose

536: value we will not need to determine. After removing the field theory 

537: infrared regulator $u$ we recover $1/\p^4$, independently of $d$. 

538: The possibility to relate 

539: $1/\p^4$ to a closed string propagator does not mean that the 

540: decoupling of closed strings fails in the noncommutative field theory 

541: limit, since the IR singularities do not have kinetic part. Hence 

542: they do not force the introduction of additional degrees of freedom.

543: 

544: In (\ref{staction}) we have considered the exchange of a single

545: closed string mode. Any closed string mode able to couple to $W^k$

546: will contribute the same $1/\p^4$ up to a numerical factor

547: depending on $D_\varphi$ and $m_\varphi$. The coefficients $\epsilon_k$

548: that appear in (\ref{invaction}) will be thus a collective effect of the 

549: closed string towers. 

550: However, we know from (\ref{emass}) that $\epsilon_k$ is determined

551: by a finite set of low lying string modes and, moreover, its sign 

552: just depends on the presence of tachyons. To reconcile these

553: two facts, notice that modes in the NSNS and RR sectors contribute 

554: with opposite sign to the exchange between D-branes. 

555: In addition, when the theory is supersymmetric we have 

556: $\epsilon_k=0$, implying that the contribution from both

557: towers must cancel. This suggests to consider $\epsilon_k$ as 

558: a measurement of the misalignment between the NSNS and RR towers. 

559: It seems then consistent that the presence of tachyons, 

560: which can only belong to the NSNS sector, is linked to the sign of the 

561: mentioned misalignment given that the lightest RR modes are massless. 

562: 

563: We would like to end this section with a comment on the cases  

564: with scalars in the adjoint representation (this 

565: situation only arises if some $b_l=0$). 

566: It was shown in \cite{Das:2000ur,Okawa:2000sh,Liu:2001ps} that in those 

567: cases closed string modes couple to generalized Wilson line operators 

568: \beq

569: {\widetilde W}(p,v)\, = \, {\rm Tr} \int d^4 x \; P_\ast 

570: \left(e^{i\, g \int_0^1 d \sigma \big(

571: \, {\tilde p}^\mu A_{\mu}(x+{\tilde p}\, \sigma) + 

572: v_a \phi_a(x+ {\tilde p} \sigma) \big)}\right) 

573: \ast e^{i p x} \, ,

574: \label{Ws}

575: \eeq

576: where $v_a$ label as before the momentum in the directions transverse to 

577: the D3-branes and $\phi_a$ denote the adjoint (real) scalars. In this 

578: case the closed string derivation of the effective action we have 

579: proposed suggests to extent (\ref{invaction}) to

580: \beq

581: \Delta S = {1 \over 2 \pi^2 c} \sum_{k=0}^{N-1} \epsilon_k

582: \int {d^4 p \over (2 \pi)^4} {d^d v \over (2 \pi)^d }

583: \; W^{(N\!-k)}(p,v) \, W^{(k)}(-p,-v) {e^{ivu} \over 

584: (v^2+\p^2)^{{d \over 2}+2}} \, , 

585: \label{actsc}

586: \eeq

587: with $c$ the same constant as in (\ref{ffin}). Adjoint scalars, as gauge 

588: bosons, get pole-like IR corrections to their self-energy. They can be 

589: obtained from (\ref{actsc}) by expanding $W^k$ to linear order in 

590: ${\rm Tr} \phi_a$ using\footnote{The integral (\ref{ffin}) for $d=2,4,6$

591: can be performed explicitly resulting in all cases proportional to 

592: ${u^2 \over \p^2} K_2(u \p)$,

593: with $K_2$ a modified Bessel function. Using the properties

594: of Bessel functions, it is now easy to derive (\ref{ffins}).}

595: \beq

596: \int {d^d v \over (2 \pi)^d}  \,

597: { e^{i v u} \over (v^2+{\tilde p}^2)^{{d \over 2}+2}} \; v_a v_b

598: \rightarrow_{u \rightarrow 0} {c \over 2 \p^2} \delta_{ab} \, .

599: \label{ffins}

600: \eeq

601: The relative factor $1/2$ 

602: between (\ref{ffin}) and (\ref{ffins}) is important to correctly 

603: reproduce the field theory results, as we will see later.

604: This support the usefulness of the closed string point of view

605: to understand UV/IR mixing effects, since the generalization of

606: the open Wilson line operators to include the adjoint scalars

607: (\ref{Ws}) is not dictated by gauge invariance. 

608: 

609: 

610: \section{Strings on twisted circles}

611: 

612: In this and the following sections we will explore whether the 

613: remarkable connection between UV/IR mixing and closed strings found for 

614: D-branes at $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds extends to more general 

615: cases. We will consider D-branes in Type II string theory compactified on 

616: twisted circle backgrounds. These are $\mathbf{R}^9 \times \mathbf{S}^1$ 

617: spacetimes where shifts along the circle are combined with rotations on 

618: several 2-planes \cite{Russo:2001na}. We will focus on the case 

619: where the rotations act on three planes and coincide with those 

620: that define a $\mathbf{C}^3/\mathbf{Z}_N$ orbifold

621: \beq

622: (y,z^l) \rightarrow (y +2 \pi R, e^{2 \pi i{b_l \over N}} z^l) \, ,

623: \label{tc}

624: \eeq 

625: where $y$ denotes the coordinate along the circle, $z^l$ the complex 

626: coordinates of $\mathbf{C}^3$ and $b_l$ are the twist parameters of 

627: the previous section. For $R \rightarrow \infty$ we recover

628: Type II string theory on $\mathbf{R}^{10}$. For $R=0$ the Type IIA(B)

629: twisted circle compactification reduces, after T-duality, to the associated 

630: $\mathbf{C}^3/\mathbf{Z}_N \otimes \mathbf{R}^4$ Type IIB(A) orbifold

631: model \cite{Takayanagi:2001jj}. 

632: 

633: The spectrum of closed strings on these backgrounds was derived in

634: \cite{Russo:2001na}. World-sheet bosons and fermions with indices transverse 

635: to the three 2-planes behave as in an ordinary circle compactification,

636: except for a different quantization condition on $p_y$.

637: Invariance of the wavefunctions under (\ref{tc}) requires 

638: \beq

639: p_y= {1 \over R}\left( m - \sum_l {b_l J_l \over N} \right) 

640: \;\; , \;\;\;\;\;\; m \in \mathbf{Z} \, ,

641: \label{py}

642: \eeq

643: where $J_l$ are the angular momenta on the three 2-planes.

644: The identification (\ref{tc}) links the winding number along the 

645: circle, $w$, with the boundary conditions for the world-sheet fields 

646: with indices along the 2-planes. For a given $w$, and for

647: the restricted set of examples we are considering, the left and 

648: right moving modes of these fields behave as those in the $k^{th}$ 

649: twisted sector of the associated $\mathbf{C}^3/\mathbf{Z}_N$ orbifold, 

650: with $k=w ({\rm mod} N)$. It is useful to define 

651: \beq

652: N'_k=N_k-c_k \; , \;\;\;\;\;\; {\bar N}'_k={\bar N}_k-{\bar c}_k \, , 

653: \eeq

654: with $N_k$ (${\bar N}_k$) the oscilator contribution

655: to $L_0$ (${\bar L}_0$) and $c_k$ (${\bar c}_k$) the zero-point energy of 

656: the left (right) modes in the NS or R $k^{th}$ twisted sector of the 

657: orbifold model. The mass spectrum of closed strings in the

658: twisted circle background (\ref{tc}) is then given by

659: \beq

660: M^2= {2 \over \alpha'} (N'_k+{\bar N}'_k) + {1 \over R^2}  

661: \left( m - \sum_l {b_l J_l \over N} \right)^2 + 

662: {w^2 R^2 \over \alpha'^2} \, ,

663: \label{spectrum}

664: \eeq

665: subject to the level matching constrain

666: \beq

667: N'_k-{\bar N}'_k= 

668: w \left( m - \sum_l {b_l J_l \over N} \right) \, .

669: \label{lm}

670: \eeq

671: 

672: $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds generically have tachyons in the 

673: twisted sectors ($k \neq 0$). Therefore the associated twisted circle 

674: backgrounds may also have tachyons for appropriate radius, provided that

675: $w \neq 0({\rm mod}N)$.

676: Since the second term in (\ref{spectrum}) always gives a positive

677: contribution to the mass, let us consider the case it vanishes.

678: This implies: {\it i)} the level matching condition reduces to that of 

679: the related $\mathbf{C}^3/\mathbf{Z}_N \otimes \mathbf{R}^4$ orbifold model; 

680: {\it ii)} $\sum_l {b_l J_l \over N} \in \mathbf{Z}$, which imposses 

681: the orbifold projection.

682: Under these conditions, the first term in (\ref{spectrum})  

683: reproduces the masses of the orbifold $k^{th}$ twisted sector. 

684: The lowest mode in each winding sector of the twisted circle 

685: compactification is contained among this set. 

686: When $m_k^2<0$, with $m_k$ the mass of 

687: the lowest mode in the $k^{th}$ twisted sector of the associated 

688: orbifold \footnote{See \cite{Armoni:2003va} 

689: for the expression of 

690: $m_k^2$ in terms of the twist parameters $b_l$.},

691: the twisted circle compactification will contain tachyons for

692: radii

693: \beq

694: {R \over \sqrt{\alpha'}} < k^{-1} \sqrt{\alpha' |m^2_k|}\, ,

695: \label{thres}

696: \eeq

697: where we have taken $w=k$ in order to minimize the winding energy.

698: 

699: We will be interested in studying field theories arising on

700: noncommutative D-branes in these backgrounds. 

701: We will consider Type IIB D3-branes wrapped along the twisted circle 

702: and Type IIA D2-branes transverse to the twisted circle 

703: \cite{Dudas:2001ux,Takayanagi:2001aj}.

704: The intrinsic scale of the D3-brane theory is set by the Kaluza-Klein 

705: masses $m \sim 1/R$. Hence the

706: natural field theory limit for the D3-branes consist in sending  

707: $\alpha'\rightarrow 0$ keeping $R$ fixed. In this regime the 

708: winding energy dominates (\ref{spectrum}) and, 

709: as can be seen from (\ref{thres}), there are no tachyonic modes. 

710: Contrary, the natural scale of the D2-brane theory is $R/\alpha'$, 

711: governing the mass of strings with both ends on 

712: the brane and non-zero winding along the circle. The interesting field 

713: theory limit in this case is achieved by keeping $R/\alpha'$ fixed

714: as $\alpha'\rightarrow 0$. From (\ref{thres}) we observe

715: that this corresponds to the regime with closed string tachyons.

716: Therefore we obtain an optimal setup to explore the relation

717: between the leading UV/IR mixing effects and properties of the

718: closed string spectrum. Notice however that the noncommutative field 

719: theory is not able to resolve the value of $R$ for which tachyonic modes 

720: appear, since any $R \sim \sqrt{\alpha'}$ maps to $R=0$ in the field theory

721: limit appropriate for the D3-brane and 

722: $R/\alpha'=\infty$ for the D2-brane. 

723: 

724: \section{Wrapped D3-branes}

725: 

726: In this section it will be convenient to view the previous twisted circle 

727: backgrounds as freely acting orbifolds of 

728: $\mathbf{R}^9 \times \mathbf{S}^1_{R'}$, with $R'=N R$. We denote the 

729: identification (\ref{tc}) by $t$; it satisfies $t^N=1$. 

730: 

731: We place $n$ D3-branes on $\mathbf{R}^9 \times \mathbf{S}^1_{R'}$ wrapped 

732: along the circle and localized at $z^l=0$. The gauge theory on the D3-branes

733: is ${\cal N}=4$ $U(n)$. After performing the orbifold projection,

734: only modes whose wavefunction is left invariant survive

735: \beq

736: Z=\gamma_t \,( t \, Z )\, \gamma_t^{-1} \, ,

737: \label{orCP}

738: \eeq

739: where $Z$ is any of the ${\cal N}=4$ fields. The matrix

740: $\gamma_t$ represents the action of the orbifold on the Chan-Paton 

741: indices. Since it must satisfy $\gamma_t^N=1$, it can be taken

742: to be diagonal with eigenvalues $e^{2 \pi i j/N}$ appearing 

743: with multiplicity $n_j$ ($\sum n_j=n$), as for

744: $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds. Let us decompose the field 

745: $Z$ in components $Z_{ij}$, transforming as $(\fund_i,\antifund_j)$ 

746: under the subgroup $\otimes U(n_j)$ which commutes with $\gamma_t$. 

747: The condition (\ref{orCP}) translates into a restriction on the allowed 

748: momenta $p_y$ analogous to (\ref{py})

749: \beq

750: p_y= {1 \over R}\left( m - {i-j + \sum_l b_l J_l \over N} \right) 

751: \;\; , \;\;\;\;\;\; m \in \mathbf{Z} \, .

752: \label{pD}

753: \eeq

754:  

755: \begin{table}

756: \begin{displaymath}

757: \begin{array}{|c|c|c|}

758: \hline

759: A^{ij}_{\mu \nu} & \psi^{ij}_\alpha & \phi_l^{ij} \\

760: \hline

761: {1 \over R}\big( m - {i-j \over  N} \big) &

762: {1 \over R}\big( m - {i-j + a_\alpha \over  N} \big) &

763: {1 \over R}\big( m - {i-j + b_l\over  N} \big) \\

764: \hline

765: \end{array}

766: \end{displaymath}

767: \caption{Field content of the theory and associated 

768: $p_y$. $A^{ij}_{\mu \nu}$ are vector fields,

769: $\psi^{ij}_\alpha$ ($\alpha=1,..,4$) Weyl fermions and 

770: $\phi_l^{ij}$ ($l=1,2,3$) complex scalars. $a_\alpha$ are the 

771: $\mathbf{C}^3/\mathbf{Z}_N$ twist parameters for

772: $SO(6)$ spinors, as in section 2.}

773: \end{table}

774: 

775: \noindent

776: Hence the theory on the wrapped D3-branes has the degrees of freedom of 

777: ${\cal N}=4$ $U(n)$ compactified on a circle of radius $R$, with 

778: both supersymmetry and gauge symmetry broken by the different 

779: fractional momenta along the circle; see Table 1.

780: 

781: We turn on now a constant B-field along the two non-compact spatial

782: directions of the D3-brane, rendering its world-volume noncommutative.

783: Since the theory behaves at high energies as ${\cal N}=4$ Yang-Mills, 

784: there will be no UV divergence and thus no associated IR divergence. 

785: But given that supersymmetry is broken at the scale $1/NR$, we expect 

786: finite UV/IR mixing effects \cite{Landsteiner:2001ky}. 

787: From now on we will take the point of view of the dimensionally 

788: reduced theory and will focus on the zero-modes. The only 

789: fields with zero-modes on the circle are: $A_\mu^{ii}$, 

790: the gauge fields associated to $\otimes U(n_i)$, 

791: $\psi_\alpha^{ii+a_\alpha}$ and $\phi_l^{ii+b_l}$. This is the same set 

792: of fields that would survive the associated $\mathbf{C}^3/\mathbf{Z}_N$ 

793: orbifold projection. A complete treatment should of course include all 

794: fields and their Kaluza-Klein modes. The reason to focus on zero-modes is 

795: that the leading UV/IR mixing terms are simpler, and we expected them to 

796: have an stronger effect, on the massless 3d fields.

797: 

798: As we have already mentioned, the nonplanar 2-point functions 

799: project on $U(1)$ degrees of freedom and mix different group labels 

800: (see Fig. 2). We normalize the Lie algebra generators of each $U(n_i)$ 

801: such that ${\rm Tr} \, t_r t_s= \delta_{rs}$, with $t_0={1 \over \sqrt{n_i}} 

802: \mathbf{1}$. We take the external legs of the nonplanar 2-point functions 

803: to correspond to ${\rm Tr}A_\mu^{ii}=A_\mu^{ii\, (0)} \sqrt{n_i}$,

804: and analogously for adjoint scalars \footnote{We will discard backgrounds 

805: with some $a_\alpha=0$, since the associated orbifold and twisted circle 

806: backgrounds never contain closed string tachyons. Hence we will not

807: encounter adjoint spinors

808: with zero-modes.}. It is easy to see that with this choice of

809: normalization for the external legs, the nonplanar 2-point functions are

810: independent of the ranks $n_i$. We can then proceed

811: with the evaluation as if our gauge group would be $U(1)^N$.

812: The 2-point functions for the zero-modes in that case are calculated 

813: in detail, at leading order, in an appendix.

814: 

815: Let us introduce some convenient notation. We define 

816: $\Pi_{ij}^{(1)}$ by $\Pi_{ij}^{\mu \nu}= {\p^\mu \p^\nu \over \p^2}

817: \, \Pi_{ij}^{(1)}$, where $\mu=1,2,3$ are the three noncompact

818: directions, and $\Pi_{ij}^{(2)}=\Pi_{ij}^{yy}$. All other components of 

819: the polarization tensor for zero-modes are zero. The matrix 

820: $\Pi_{ij}^{(3)}$ will refer to the self-energies of adjoint scalar 

821: zero-modes when present (this requires that some $b_l=0$),

822: $\Sigma_{ij}^{ab}=\delta^{ab} \, \Pi_{ij}^{(3)}$. Using this compact 

823: notation and (\ref{apinp1})-(\ref{apinp3}), the nonplanar

824: 2-point functions are given by

825: \beq

826: \Pi^{(M)}_{ij}= {g^2 \over 2 \pi^2 \p} \, c_{ij} \sum_\rho (-1)^{F_\rho} 

827: \! \int_0^\infty d l \, N_{{\bar \nu}_\rho}(Rl) \, f^{(M)}({\tilde p}l)  \, .

828: \label{pift}

829: \eeq

830: The index $\rho$ runs over all degrees of freedom transforming in the

831: $(\fund_i,\antifund_j)$ and $(\fund_j,\antifund_i)$ representations, 

832: and $F_\rho$ is their fermion number. The exchange of an adjoint field

833: in the loop contributes twice that of a bifundamental, thus $c_{ii}=1$ 

834: while $c_{ij}=1/2$ for $i \neq j$. The function $N_{{\bar \nu}_\rho}$ 

835: is given in (\ref{N}). It represents the generalization of the Bose-Einstein 

836: and Fermi-Dirac distributions to fields with 

837: general fractional momenta along a circle,

838: $p_y=(m-{\bar \nu}_\rho)/R$. The functions $f^{(M)}$ are

839: \beq

840: f^{(M)} (x) = 

841: \sin x +b^{(M)} x \cos x \, , 

842: \eeq

843: with $b^{(1)}=-1$, $b^{(2)}=1$ and $b^{(3)}=0$. 

844: 

845: The integrals in (\ref{pift}) can be easily evaluated 

846: recalling (\ref{dis}) and performing a Poisson resummation

847: \beq

848: \int_0^\infty dl N_{{\bar \nu}_\rho}(R l) \sin \p l  =  \pi T \sum_{k=0}^{N-1} 

849: \cos \big( 2 \pi {\bar \nu}_\rho k \big) \, N_{\nu_k} (T \p) \, , 

850: \label{poisson}

851: \eeq

852: where we have defined $\nu_k=k/N$ and $T=1/2 \pi R N$. 

853: $\int dl N_{{\bar \nu}_\rho} l \cos \p l$ can be directly obtained

854: from the above expression, using that $l \cos \p l={\partial \sin \p l  

855: \over \partial \p}$. Substituting these results, together

856: with the explicit expression of $N_{\nu_k}$ (\ref{N}), we arrive at 

857: \beq

858: \Pi^{(M)}_{ij}= g^2 T^2  \;

859: \sum_{k=0}^{N-1} M_{ij;k} \;  F_k^{(M)}(T {\tilde p}) \, ,

860: \label{polt}

861: \eeq

862: with 

863: \beq 

864: F^{(M)} (x) =  {1 \over \cosh 2 \pi x -\cos 2 \pi \nu_k}

865:  \left[\, {\sinh 2 \pi x \over 2 \pi x }+ b^{(M)} \,

866: {1-\cos 2 \pi \nu_k \cosh 2\pi x \over \cosh 2 \pi x -\cos 2 \pi \nu_k}

867: \, \right]    \, ,

868: \label{sums}

869: \eeq

870: and $M_{ij;k} = c_{ij} \sum_\rho (-1)^{F_\rho} \cos 2 \pi {\bar \nu}_\rho k$. 

871: From the data of Table 1, an straightforward 

872: calculation gives 

873: \beq

874: M_{ij;k} = \epsilon_k \, \cos \big( 2 \pi (i-j) \nu_k \big) \, .

875: \label{mij}

876: \eeq

877: Remarkably the quantities $\epsilon_k$ coincide with those 

878: resulting from the diagonalization of the polarization tensor for 

879: the $\mathbf{C}^3 / \mathbf{Z}_N$ quiver theories analyzed in section 2, 

880: eq. (\ref{ep}).

881: However the matrices $\Pi_{ij}^{(M)}$ are not yet in a diagonal form. 

882: Notice that their $ij$ entry depends only on 

883: $(i-j)\, {\rm mod} N$. Any matrix with this

884: property can be diagonalized by a set of orthogonal vectors 

885: $(e_k)_j={1 \over \sqrt{N}} e^{2 \pi i j k/N}$. In the new basis (\ref{polt})

886: reads  

887: \beq

888: \Pi^{(M)}_k=  g^2 T^2 \, N \; \epsilon_k \; 

889: F_k^{(M)} (T \p) \, .

890: \label{pid}

891: \eeq

892: 

893: The sign of the 2-point functions is governed by $\epsilon_k$, as it 

894: was the case for the $\mathbf{C}^3 / \mathbf{Z}_N$ quiver theories,

895: because the functions $F_k^ {(M)}$ are always positive. 

896: This can easily be shown by direct inspection of (\ref{sums}).

897: For $k \neq 0$, $F_k^{(M)}$ is finite as $\p \rightarrow 0$. 

898: The case $k=0$ is special. $F_0^{(M)}$ diverges for small $\p$ since

899: this sector contains the IR divergences of the noncompact 4d theory.

900: However, $\epsilon_0=0$ because the uncompatified limit is supersymmetric. 

901: As $T\rightarrow \infty$, $F_k^{(M)}$ tends to $1/2 \pi T \p$.

902: In this limit (\ref{pid}) reproduces the $1/\p \,$ IR divergences of 

903: the $\mathbf{C}^3 / \mathbf{Z}_N$ quiver theory on a D2-brane. 

904: Therefore at some point of the flow to small radius

905: (\ref{pid}) will dominate the dispersion relations for low momentum

906: modes, and noncommutative instabilities will appear.

907: 

908: 

909: %%%%%%%%%%%%%%%%%%%%%%%%

910: \begin{figure}

911: \centering

912: \epsfxsize=2.5in

913: \hspace*{0in}\vspace*{0in}

914: \epsffile{planar.eps}

915: \caption{\small Planar contribution to the 2-point function of adjoint 

916: fields. $i,j$ are group label indices and $a,b,c$ gauge group indices.}

917: \end{figure}

918: %%%%%%%%%%%%%%%%%%%%%%%%%

919: 

920: The analysis of the critical radius at which some modes become unstable

921: is quite involved. It strongly depends on which fields we are considering, 

922: vector or scalar, on the sector $k$ and on the particular brane

923: configuration, characterized by the ranks $n_i$. We will not go through 

924: this complicated phenomenology, because what interests us is the structure 

925: behind these effects. However for completion, we include below the

926: leading planar contributions to the 2-point functions. 

927: 

928: We reduce again to the zero-modes of gauge bosons and adjoint scalars. 

929: As can be seen from Fig.3, in this 

930: case there is no mixing among the different group labels. The planar

931: correction for the 3-dimensional gauge bosons is absent; for the others

932: fields we have (see (\ref{apip}))

933: \beq

934: {\hat \Pi}^{(2)}_{ii}= 2 {\hat \Pi}^{(3)}_{ii}=- {g^2 \over \pi^2} \, 

935: \sum_\rho c_\rho \, (-1)^{F_\rho} \! 

936: \int_0^\infty d l \, N_{{\bar \nu}_\rho}(Rl) 

937: \, l   \, .

938: \label{pi}

939: \eeq

940: The summation in $\rho$ is over the degrees of

941: freedom transforming in the $(\fund_i,\antifund_j)$ and 

942: $(\fund_j,\antifund_i)$, for any $j$. $c_\rho=1$ for

943: a field in the adjoint representation, while $c_\rho=1/2$ for

944: a bifundamental.

945: The external legs will be normalized now such that ${\hat \Pi}_{ii}^{(2)}$

946: couples to ${\rm Tr}(A_y^{ii})^2$, which is appropriate for a planar graph,

947: and analogously for ${\hat \Pi}_{ii}^{(3)}$.

948: With this normalization, the 2-point functions will

949: include an explicit dependence on the ranks $n_j$. We obtain

950: \beq

951: {\hat \Pi}^{(2)}_{ii}= -{2 g^2 T^2 \over \pi^2 } \;

952: \sum_{k=0}^{N-1} \epsilon_k  \left( \sum_{j=1}^N \, n_j\,\cos 

953: \big(2 \pi (i-j) \nu_k \big) \right) Z_2[\nu_k] \, ,

954: \label{planar}

955: \eeq 

956: with $Z_m[x]$ a generalized Riemann Zeta function.

957: 

958: It is worth stressing that the result of evaluating the integrals 

959: in (\ref{pift}) was given again in terms  of functions 

960: $N_{\nu_k}$, with their argument changed from $Rl$ to $T \p$.

961: A similar observation, in the context of noncommutative 

962: thermal field theories, was done in \cite{Fischler:2000fv}.

963: Since $T \sim 1/R$, they suggested that there must be some

964: hidden winding states in the noncommutative theory.

965: In \cite{Arcioni:2000bz} it was argued that these mysterious states

966: were closed strings winding around the temperature circle

967: of a parent string theory. They noticed that an infinite

968: set of closed strings must play a role, very much as in

969: ordinary field theory limits of string theory. This raised the question 

970: about the usefulness of the closed string picture. 

971: In the next section we will present a very explicit realization  

972: of these ideas along the lines of section 2. In particular

973: we will show that the closed string picture is indeed useful to 

974: understand the structure of both (\ref{pid}) and (\ref{planar}).

975: 

976: 

977: \section{Closed string exchange}

978: 

979: 

980: Let us analyze what type of Wilson line operators couple to 

981: closed strings in twisted circle backgrounds. It is useful to consider 

982: first D3-branes wrapped along an ordinary circle of radius $R'=NR$.

983: Closed strings couple to the straight open Wilson line operators (\ref{Ws})

984: on noncommutative D-branes. These operators depend on the 

985: components of the gauge field along the noncommutative directions and on 

986: adjoint scalars, associated to the position of the branes in transverse 

987: dimensions. However this must be generalized when one direction on the

988: D-brane is compact. T-dualizing, the D3-brane is converted

989: to a D2-brane localized on the circle. The zero-mode of $A_y$ describes 

990: the positions of branes on the T-dual circle and thus plays a similar role 

991: to the adjoint scalars. Hence in order to be consistent with T-duality, 

992: we should include a dependence on $A_y$ in the exponent of the Wilson line.

993: 

994: The explicit construction of the open Wilson line operators that are

995: gauge invariant under the 4d gauge transformations is very involved.

996: However we will not need to know their precise expression. We are only 

997: interested on the coupling of closed strings to the zero-modes of gauge 

998: fields and scalars, since this will be enough to reproduce the results 

999: of the past section. With this restriction, we propose the following 

1000: modification of (\ref{Ws})

1001: \beq

1002: {\widetilde W}_{p_y=0}  = 2 \pi R' \, {\rm Tr} \int d^3 x \; P_\ast 

1003: \left(e^{i\, g \int_0^1 d \sigma 

1004: \, \big( {\tilde p}^\mu A_{\mu}(x(\sigma))+

1005: v_a \phi_a(x(\sigma))+ 2 \pi R' s  A_y (x(\sigma))\big)}\right) 

1006: \ast e^{i p x} \,+ \, ...  \, ,

1007: \label{wsw}

1008: \eeq

1009: where $A_\mu(x)={1 \over 2 \pi R'} \int dy A_\mu(x,y)$ and the same 

1010: for $A_y$ and $\phi_a$; we have denoted $x(\sigma)=x+\p \sigma$. 

1011: The dots indicate terms depending on modes with $p_y\neq 0$, necessary 

1012: to make ${\widetilde W}$ 4d gauge invariant. The term written explicitly 

1013: above is however gauge invariant under 3d gauge transformations. 

1014: As in section 2, $v=2 \pi \alpha' p_{\perp}$, with $p_\perp$ the momentum 

1015: on the transverse directions of the associated closed string mode. 

1016: $s \in \mathbf{Z}$ is its winding number. Notice that $2 \pi R's= 

1017: 2 \pi \alpha' p_{\bar v}$, with 

1018: $p_{\bar v}=s R'/ \alpha'$ the momentum on the dual T-circle.

1019: The factor $2 \pi R'$ in (\ref{wsw}) 

1020: comes from the trivial integration along the circle.  

1021: 

1022: The interpretation of the twisted circle backgrounds (\ref{tc}) as

1023: freely acting orbifolds of $\mathbf{R}^9 \otimes \mathbf{S}^1_{R'}$,

1024: divides the closed string spectrum (\ref{spectrum}) in twisted and

1025: untwisted sectors. The untwisted sector is composed of modes

1026: with $w=0({\rm mod}N)$. The $k^{th}$ twisted sector ($k=1,..,N-1$) 

1027: contains modes with $w=k({\rm mod}N)$. 

1028: %From the point of view of the 

1029: %initial $\mathbf{S}^1$ circle, of radius $R'=RN$, modes in $k^{th}$ 

1030: %twisted sector require fractional winding number $w N^{-1}=\nu_k+s$. 

1031: These modes will couple to linear combinations of field theory operators 

1032: determined by $\gamma_t$, the matrix that represents the action of the 

1033: orbifold on the Chan-Paton indices \cite{Douglas:1996sw}. 

1034: Since $\gamma_t$ for the twisted circle compactifications coincides 

1035: with that of $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds, we should consider 

1036: the same linear combinations as in that case. This suggests that an scalar 

1037: closed string mode with $w=k({\rm mod}N)$ will couple to

1038: \beq

1039: W^{(k)}= {1 \over \sqrt{N}} \sum e^{2 \pi i j \nu_k} 

1040: {\widetilde W}^{(j)}(\nu_k) \, ,

1041: \eeq

1042: where ${\widetilde W}^{(j)}(\nu_k)$ is

1043: the operator (\ref{wsw}) with the fields belonging to the

1044: $j^{th}$ gauge group factor and $s$ substituted by

1045: $(\nu_k+s)=w N^{-1}$.

1046: 

1047: In the noncommutative field theory limit \cite{Seiberg:1999vs}, taken at 

1048: fixed radius, the on-shell condition a closed string mode

1049: $\varphi$ with $w=k({\rm mod}N)$ is

1050: \beq

1051: {\tilde p}^2+y^2+ (2 \pi R')^2 (\nu_k+s)^2+ 8 \pi^2 \alpha'

1052: (N'_k+{\bar N}'_k)=0 \, ,

1053: \label{spsw}

1054: \eeq

1055: where we have used (\ref{spectrum}) and neglected terms suppressed by 

1056: two powers of $\alpha'$. 

1057: We have now all the ingredients to derive the contribution to the field 

1058: theory effective action from a closed string exchange, following the

1059: analysis of section 2. Generalizing

1060: (\ref{staction}) and (\ref{actsc}) to the present case, we obtain from the 

1061: emission, propagation and posterior absorption of $\varphi$ by the brane 

1062: \beq

1063: \Delta S = {h_\varphi \over 2 \pi R'} \; 

1064: \sum_{s=-\infty}^{\infty}  \int {d^3 p \over (2 \pi)^3} \, 

1065: {d^d v \over (2 \pi)^d} \; { W^{(k)}(P) \; W^{(N-k)}(-P)   \over 

1066: \Big(\,v^2+{\tilde p}^2 + (2 \pi R')^2(\nu_k+s)^2 \, \Big)^{{d \over 2}+2}} 

1067: \, e^{iyu} \, ,

1068: \label{saction}

1069: \eeq

1070: where $P\equiv(\p,v,\nu_k+s)$, $d$ is the number of transversal 

1071: directions where the closed string can propagate. $h_\varphi$ depends on 

1072: the disk amplitude with one insertion of $\varphi$

1073: and on the last term in (\ref{spsw}). 

1074: 

1075: All scalar closed string modes contribute to the field theory

1076: effective action as in (\ref{saction}), only differing in the value of 

1077: $h_{\varphi}$. Since $W^{(k)}$ (\ref{spsw}) carries no momentum along

1078: the circle, it can only couple to closed string modes with $p_y=0$. 

1079: In section 3 we saw that for these modes, the last term

1080: in (\ref{spsw}) reproduces the spectrum in the $k^{th}$ twisted

1081: sector of the related $\mathbf{C}^3/\mathbf{Z}_N$ orbifold.

1082: We will assume that the disk amplitudes with 

1083: one insertion of these modes and no open string insertions also coincide 

1084: for orbifolds and twisted circles backgrounds.  

1085: With this hypothesis, after taking into account the effect

1086: of the whole closed string tower, $h_\varphi$ should be promoted to the same 

1087: quantities that govern the effective action of D3-branes in 

1088: $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds (\ref{actsc}): 

1089: $\epsilon_k / 2 \pi^2 c$, with $c$ given in (\ref{ffin}) and (\ref{ffins}).

1090: 

1091: We are ready to derive the contributions to the 2-point functions predicted

1092: by the closed exchange technique. We study first the non-planar contributions.

1093: They are obtained by expanding the Wilson line operators to first order

1094: in the fields. The linear combinations that 

1095: diagonalize $\Pi_{ij}^{(M)}$ appear in a natural way: 

1096: $B_M^{(k)}={1 \over \sqrt{N}} \sum e^{2 \pi i j \nu_k} {\rm Tr} Z_M^{(j)}$,

1097: with $Z_M^{(j)}=(A_\mu,A_y,\phi_a)^{jj}$. In order to compare with 

1098: previous results, it is important to take into account that the fields

1099: in the exponent of (\ref{wsw}) are $Z^{(j)}(x)={1 \over 2 \pi R} 

1100: Z^{(j)}_{n=0}(x)$, with $Z^{(j)}_{n=0}$ the zero-modes for which

1101: the 2-point functions were calculated in the past section. 

1102: From (\ref{saction}) we obtain then an

1103: expression for $\Pi^{(M)}_k$ analogous to (\ref{pid}), with the functions

1104: $F_k^{(M)}$ replaced by

1105: \beqa

1106: F_k^{(1)st} & = & 4 R'^2 \, \sum_{s=-\infty}^{\infty} 

1107: {\p^2 \over \Big( \, {\tilde p}^2 + (2 \pi R')^2(\nu_k +s)^2 \, \Big)^2} 

1108: \label{pi1} \, ,\\

1109: F_k^{(2)st} & = & 4 R'^2 \, \sum_{s=-\infty}^{\infty} 

1110: {(2 \pi R')^2 (\nu_k+s)^2 

1111: \over \Big( \, {\tilde p}^2 + (2 \pi R')^2(\nu_k +s)^2 \, \Big)^2} \, , \\

1112: F_k^{(3)st} & = & 2 R'^2 \,  \sum_{s=-\infty}^{\infty}

1113: {1 \over {\tilde p}^2 + (2 \pi R')^2(\nu_k+s)^2} \, . \label{pi3} 

1114: \eeqa

1115: For the origin of the relative factor $1/2$ in $F_k^{(3)st}$

1116: see (\ref{ffins}). Using (\ref{dis})-(\ref{N}), it is easy to show that 

1117: $F_k^{(M)st}=F_k^{(M)}$.

1118: $\Pi_k^{a\mu}=\Pi_k^{ay}=0$ because the associated integrand is an odd 

1119: function of $y$. $\Pi_k^{y\mu}=0$ is obtained by summing the contributions 

1120: from the $k^{th}$ and $(N-k)^{th}$ sectors. 

1121: 

1122: To derive the planar contribution to the 2-point functions from 

1123: (\ref{saction}), we expand one Wilson line operator to second order in

1124: the fields while retaining from the second one only the leading term. 

1125: Due to the structure of (\ref{saction}),

1126: we get only contributions diagonal in the 

1127: group label indices, as was shown to be the case in the field theory 

1128: analysis (see Fig. 3). The leading term in the expansion of $W^{(k)}$ is

1129: proportional to ${1 \over \sqrt{N}} \sum e^{2 \pi i j \nu_k} n_j \, 

1130: \delta^{(3)}(p)$. Since the closed string propagator is finite as 

1131: ${\tilde p}\rightarrow 0$ while the gauge field $A_\mu$ appears always 

1132: multiplied by corresponding ${\tilde p}^\mu$ factors, the result for 

1133: ${\hat \Pi}^{\mu \nu}_{ii}$ is zero. For the other components we get exactly

1134: (\ref{planar}).

1135: 

1136: The contribution to the 1-loop effective action from the vacuum

1137: diagrams, $\Gamma$, can be evaluated using that 

1138: \beq

1139: {\partial \Gamma \over \partial \beta}= - {V_3 \over 2}

1140: \, \sum_\rho (-1)^{F_\rho} \int 

1141: {d^3 l \over (2 \pi)^3} N_{{\bar \nu}_\rho} \, l \, ,

1142: \eeq

1143: where $\beta= 2 \pi R$ and $V_3$ represents the volume of the three 

1144: noncompact dimensions. The integration is over the euclidean 

1145: 3-momentum and $\rho$ runs over all degrees of freedom of the theory. 

1146: We obtain

1147: \beq

1148: \Gamma = {V_4 \, T^4 \over \pi^2} \, \sum_{i,j} \, n_i \, n_j\, 

1149: \sum_k \epsilon_k \, \cos \big(2 \pi (i-j) \nu_k \big)  \, Z_4[\nu_k] \, ,

1150: \eeq

1151: with $V_4=2 \pi R \, V_3$. Retaining only the first term in the 

1152: expansion of both Wilson line operators, one can check that 

1153: (\ref{saction}) also reproduces 

1154: $\Gamma$. It was argued in \cite{Armoni:2003va} that $\Gamma$ for the 

1155: $\mathbf{C}^3/\mathbf{Z}_N$ quiver theories of section 2, could be 

1156: recovered from (\ref{invaction}) in an analogous way. However this implied

1157: identifying ${1 \over \p}|_{\p=0}$ with an UV cutoff $\Lambda$, 

1158: and introduced certain ambiguity. This problem does not arise

1159: in the theories treated here, since they are UV finite.

1160: 

1161: Let us summarize our results. The closed string picture

1162: captures both the leading planar and non-planar terms which will give 

1163: rise in the limit $T \rightarrow \infty$ to the UV and IR divergences 

1164: linked to the absence of supersymmetry. Moreover, it offers an 

1165: explanation for why the same coefficients $\epsilon_k$ appear in

1166: $\mathbf{C}^3/\mathbf{Z}_N$ orbifolds and twisted circle backgrounds.

1167: Twisted circle backgrounds, for sufficiently large radius, 

1168: cure the problem of closed string tachyons by shifting the spectrum 

1169: with a positive winding energy. However this affects equally 

1170: NSNS and RR sectors. Since NSNS and RR sectors contribute with 

1171: opposite sign to (\ref{saction}), this shift is not perceived by 

1172: the field theory. The sign of the leading UV/IR mixing effects is 

1173: still given by that of the gap between the lowest NSNS and RR modes.

1174: 

1175: The nonplanar functions that we calculated in this and the past 

1176: section are regular in $\p$. From the string point of view, this

1177: crucially depends on being in a regime where winding energy

1178: dominates the closed string spectrum. Instabilities arising from

1179: finite UV/IR effects can be expected to have a qualitatively different 

1180: nature from those associated to IR divergences. It is then an open 

1181: question whether the one to one correspondence between tachyons

1182: and noncommutative instabilities found for $\mathbf{C}^3/\mathbf{Z}_N$ 

1183: orbifolds, might hold in more general IR divergent cases. The fact that 

1184: the nonplanar polarization tensor (\ref{pid}) develops a pole-like 

1185: behavior at $R=0$, value which maps to the string regime with tachyons,

1186: supports this possibility. 

1187: The twisted circle backgrounds provide us with another example of a

1188: field theory related to an unstable string regime: that on the 

1189: world-volume of a D2-brane placed transverse 

1190: to the circle. In the next section we will analyze the leading

1191: UV/IR mixing effects in this theory.

1192: 

1193: \section{Transversal D2-branes}

1194: 

1195: 

1196: We consider $n$ Type IIA D2-branes extending in the 

1197: directions transverse to (\ref{tc}) and localized at $z^l=y=0$. 

1198: It is convenient to picture the brane configuration on the $\mathbf{R}^{10}$

1199: covering space. It corresponds to an infinite array of D2-brane

1200: sets placed at intervals $2 \pi R$ on the $y$ direction.

1201: Let us denote open strings stretching between the $i^{th}$ and

1202: $j^{th}$ sets by $(i,j)$. The action (\ref{tc}) relates $(i,j)$ 

1203: strings with $(i+1,j+1)$, but does not impose any projection

1204: on each of them separately. Thus the spectrum on the branes will be 

1205: maximally supersymmetric \cite{Dudas:2001ux}. We are interested in 

1206: taking the field theory limit such that the masses of the winding 

1207: strings are keep finite, {\it i.e.} $\alpha'\rightarrow 0$ with 

1208: $R/\alpha'=r$ fixed. The field content on the branes will

1209: consist of an infinite set of multiplets $\Psi_w$ 

1210: with masses $m_w= w r$. Each multiplet $\Psi_w$

1211: transforms in the adjoint of $U(n)$ and contains 3 complex scalars 

1212: $\phi_{l,w}$, 4 Weyl fermions $\psi_{\alpha,w}$

1213: and, for $w\neq 0$, a massive vector $A_{\mu,w}$, while for

1214: $w=0$ a massless vector plus a real scalar $A_{\mu,0}$, $\varphi_0$.

1215: The scalars $\phi_{l,0}$, $\varphi_0$ are associated to fluctuations 

1216: of the branes in the $z^l$ and $y$ directions. The previous field theory 

1217: limit corresponds to the string regime with tachyonic modes, as can be 

1218: seen from (\ref{thres}).

1219: 

1220: In this section we will focus on the evaluation of the polarization tensor. 

1221: We turn on a B-field on the two spatial directions of the D2-branes.

1222: We will analyze first what the closed string picture predicts for the 

1223: leading UV/IR mixing effects. As before, closed strings will couple to 

1224: open Wilson line operators. The explicit construction of these operators 

1225: is again involved. For our purposes, it will be enough to know few terms

1226: in their expansion 

1227: \beq

1228: W^{(w)}(p)= \delta_{w,0} \, (2 \pi)^3 \delta^{(3)} (p) {\rm Tr} {\mathbf{1}}  

1229: + i g \p^\mu {\rm Tr} A_{\mu,w}(p)+... \, .

1230: \label{Ww}

1231: \eeq

1232: Given that the identity operator carries no winding number, it can only

1233: appear in the expansion of $W^{(0)}$.

1234: 

1235: Closed string modes that couple to (\ref{Ww}) have winding number 

1236: $w$. Since the D2-branes break translational invariance along the $y$ 

1237: direction, their momentum $p_y$ is a free parameter. There is however 

1238: a restriction. D2-branes located at $z^l=0$ do not break rotational 

1239: invariance on $C^3$, and thus their vacuum state carries no 

1240: angular momentum. Neither do the fields $A_{\mu,w}$. Therefore (\ref{py}) 

1241: implies $p_y=m/R$ with $m \in \mathbf{Z}$. 

1242: 

1243: In the noncommutative field theory limit that we are considering in this

1244: section, the on-shell condition for a closed string mode $\varphi$ with 

1245: $w=k({\rm mod}N)$ and $p_y=m/R$ reads

1246: \beq

1247: {\tilde p}^2+y^2+ {4 \pi^2 m^2 \over r^2} + 

1248: 8 \pi^2 \alpha'(N'_k+{\bar N}'_k)=0 \, ,

1249: \label{spsw2}

1250: \eeq

1251: where again we have neglected terms suppressed by two $\alpha'$

1252: powers. Using the same reasoning as in the past sections,

1253: we obtain the following contribution to the transversal part of the

1254: polarization tensor from a closed string exchange

1255: \beq

1256: \Pi^{\mu \nu}_w \sim h_\varphi \; r^{-1} g^2   {\p^\mu \p^\nu \over 

1257: \Big(\,{\tilde p}^2 + {4 \pi^2 m^2 \over r^2} \, \Big)^2} \, .

1258: \label{strexch}

1259: \eeq

1260: Notice that the dependence on $\p$ is that proper of a 4d theory.

1261: This arises as follows. The definition of the closed 

1262: string propagator includes now a factor $1/2 \pi R$, because the 

1263: brane is transversal to the circle. In the $\alpha'\rightarrow 0$ 

1264: limit that keeps finite the mass of winding strings, $R$ absorbs 

1265: one $\alpha'$ factor and gives raise to the $r^{-1}$ factor in

1266: (\ref{strexch}). As a result, the counting of 

1267: $\alpha'$ powers is as for the D3-brane case.

1268: 

1269: The combination $2 \pi m/r$ in (\ref{strexch}) plays an analogous role 

1270: to the winding energy in (\ref{saction}). There is however a crucial

1271: difference; in the latter case, the winding energy could not be zero if 

1272: $w \neq 0({\rm mod}N)$. Contrary, $m=0$ is an allowed value for any 

1273: $w$. It is clear that $m=0$ will dominate the small $\p$ limit of 

1274: the sum in (\ref{strexch}). Let us focus then on this value. 

1275: 

1276: The quantity $h_\varphi$ depends on the last term in (\ref{spsw2}). 

1277: When $m=0$, the associated closed string mode has $p_y=0$.

1278: As we have already explained, the last term in (\ref{spsw2}) reproduces 

1279: then the spectrum of the $k=w ({\rm mod} N)$ twisted sector of the related

1280: $\mathbf{C}^3 /\mathbf{Z}_N$ orbifold.

1281: $h_\varphi$ depends also on a numerical factor, $D_\varphi^{tc}$,

1282: which can be extracted from the disk amplitudes 

1283: of $\varphi$ with boundary conditions on the D2-brane.

1284: However we can not take now the brane in its vacuum state, since

1285: this restricts us to $w=0$. Let us consider the disk amplitude with one 

1286: insertion of $A_{\mu,w}$ on its boundary. This should reproduce 

1287: $D_\varphi^{tc}$ times the second term in (\ref{Ww}). 

1288: As in the past section, we will assume that $D_\varphi^{tc}$ coincides 

1289: with the analogous factors from the orbifold case of section 2. 

1290: With that hypothesis, and after taking into 

1291: account the contribution from all closed string modes to 

1292: (\ref{strexch}), $h_\varphi$ will be promoted to the same

1293: quantities $\epsilon_k$ that appeared in the previous

1294: sections. Hence, at small $\p$, the closed string picture suggests

1295: the following result for the polarization tensor

1296: \beq

1297: \Pi^{\mu \nu}_w \sim \epsilon_k \; r^{-1} g^2   {\p^\mu \p^\nu \over 

1298: {\tilde p}^4} \, .

1299: \label{strexch2}

1300: \eeq

1301: with $k=w({\rm mod}N)$.

1302: 

1303: We will like to add a comment. Comparison with (\ref{Ws}), for $w=0$, 

1304: implies that the expansion of $W^{(0)}$ at leading order in the fields

1305: should include a term $\sim \alpha' p_y {\rm Tr} \, \varphi_0= {m \over r} 

1306: {\rm Tr} \, \varphi_0$, where $\varphi_0$ is the massless scalar of

1307: the $\Psi_0$ multiplet. For $W^{(w)}$, with $w \neq 0$, we expect that 

1308: this is promoted to a term $\sim {m \over r}{\rm Tr} A_{L,w}$, 

1309: where $A_{L,w}$ denotes the longitudinal component of the massive 

1310: gauge field $A_{\mu,w}$. Using this and the closed string exchange

1311: technique, we could try to derive a contribution to the longitudinal 

1312: piece of the polarization tensor. However the string derivation 

1313: predicts no contribution from $m=0$, and thus we have ignored it.

1314: This is consistent with the field theory analysis below.

1315: 

1316: We will confirm now the proposal (\ref{strexch2})

1317: from a direct field theory calculation. The appearance

1318: of IR divergences could have surprised us, since the field content 

1319: of the D2-brane theory is supersymmetric. Notice that 

1320: the spectrum on D3-branes, both at orbifold singularities

1321: and twisted circles, was not supersymmetric.

1322: The coefficients $\epsilon_k$ were essentially counting there the number 

1323: of bosonic minus fermionic degrees of freedom from the point of view of

1324: the field theory. 

1325: Contrary to those cases, supersymmetry is broken by interactions

1326: on the D2-brane. Thus we need to know how the Feynman rules 

1327: get modified.

1328: 

1329: Let us come back to image of the D2-brane configuration on the

1330: covering space. We will denote as $Z_{(i,j)}$ a field arising from an 

1331: string stretching between the $i^{th}$ and $j^{th}$ D2-brane sets.

1332: Before imposing (\ref{tc}), our gauge theory is $U(\infty)$

1333: broken to an infinite product of $U(n)$ factors by the separation of

1334: the branes. A field $Z_{(i,j)}$ transforms in the fundamental 

1335: representation of the

1336: gauge group factor $i^{th}$ and antifundamental of the $j^{th}$.

1337: Therefore the only terms that can appear in the lagrangian of the 

1338: theory form closed paths in the direction $y$; for example

1339: \footnote{These terms contain also derivatives. We are

1340: being schematic for simplicity.}

1341: \beq

1342: {\rm Tr} \; {\bar Z}_{(i,j)} Z_{(j,i)} \; , \;\;\;

1343: {\rm Tr} \; Z^1_{(i,j)}Z^2_{(j,k)}Z^3_{(k,i)} \, .

1344: \eeq

1345: All the kinetic terms are canonically normalized and the

1346: three- and four-point vertices depend on a single coupling constant

1347: $g$. 

1348: 

1349: Invariance under (\ref{tc}) implies

1350: \beq

1351: Z_{(i,j)}= \omega_Z^i \; Z_{(0,j-i)} \, ,

1352: \label{fields}

1353: \eeq

1354: with $\omega_{\phi_l}=e^{2 \pi i {b_l \over N}}$, 

1355: $\omega_{\psi_\alpha}=e^{2 \pi i {a_\alpha \over N}}$

1356: and $\omega_{A_\mu}=1$. From now on we will take the fields 

1357: $Z_{(0,w)}$ as reference and we will label them simply as $Z_w$.

1358: They form the multiplets that we defined in the beginning of

1359: the section as $\Psi_w$.

1360: 

1361: In order to obtain the polarization tensor for the vector fields

1362: $A_{\mu,w}$, we have to evaluate the graphs in Fig. 2 and 3.

1363: We need to know the propagators and vertices of the theory in terms 

1364: of the fields $Z_w$. 

1365: Since ${\bar Z}_{(i,j)} Z_{(j,i)}= \omega_Z^{-w} {\bar Z}_{-w} Z_w$,

1366: with $w=i-j$, and the field $Z_{(i,j)}$ is canonically normalized,

1367: the propagator for the field $Z_w$ will carry an additional factor

1368: $\omega_Z^w$. For constructing the graphs in Fig. 2 and 3, two 

1369: vertices are relevant

1370: \beq

1371: V_1= g_1 \; {\rm Tr} \; {\bar Z}_{w_1} A_{\mu,w} Z_{w_2} \;\; , \;\;\;\;\;

1372: V_2=- g_2 \; {\rm Tr} \; {\bar Z}_{w_1} Z_{w_2} A_{\mu,w} \, ,

1373: \eeq

1374: with $w_1+w_2+w=0$ and $g_1$, $g_2$ the coupling constant

1375: associated to each vertex. Using (\ref{fields}), it is easy to see 

1376: that $g_1=g \, \omega_Z^{-w_2}$ and $g_2=g \, \omega_Z^{w_1}$. The 

1377: different powers of $\omega_Z$ associated to the two vertices may 

1378: result surprising. However we should notice that the two 

1379: vertices come from different configurations in the covering

1380: space, see Fig. 4.

1381: 

1382: %%%%%%%%%%%%%%%%%%%%%%%%

1383: \begin{figure}

1384: \centering

1385: \epsfxsize=4.2in

1386: \hspace*{0in}\vspace*{0in}

1387: \epsffile{chain.eps}

1388: \caption{\small Interaction vertices $V_1$ and $V_2$ in the covering space.}

1389: \label{chain}

1390: \end{figure}

1391: %%%%%%%%%%%%%%%%%%%%%%%%%

1392: 

1393: 

1394: 

1395: The non-planar graph of Fig. 2 arises when the two vertices

1396: are $V_i$, $V_j$ with $i \neq j$ ($i,j=1,2$). 

1397: Using the rules explained above, we obtain the following

1398: contribution to the non-planar polarization tensor from a field 

1399: $Z$ running in the loop

1400: \beq

1401: {1 \over 2}\big( \omega_Z^{w}+\omega_Z^{-w} \big) \, \Pi^{\mu \nu}_Z \, ,

1402: \label{phs}

1403: \eeq 

1404: where $\Pi^{\mu \nu}_Z$ would be the analogous result when 

1405: $\omega_Z=1$. This simpler situation corresponds to D2-branes placed 

1406: transversally to an ordinary circle, and is T-dual to D3-branes wrapping 

1407: an ordinary circle of radius $r^{-1}$. The evaluation of 1-loop integrals 

1408: is most convenient in the T-dual picture. It results in a piece which is 

1409: $r$-independent and reproduces the uncompactified

1410: $r^{-1} \rightarrow \infty$ result, and a part that is UV finite and $r$ 

1411: dependent. The former one gives rise to an IR divergence

1412: and dominates in the limit of small $\p$. Since this is the IR

1413: divergence of a 4d theory, we have\footnote{In addition to the 

1414: graphs in Fig. 2, there are 1-loop diagrams

1415: with the insertion of a four-point vertex contributing to the 

1416: nonplanar polarization tensor. The same considerations above

1417: can be applied to them with similar results.}

1418: $\Pi^{\mu \nu}_Z \sim r^{-1} g^2 \, {\p^\mu \p^\nu  \over \p^4}$, 

1419: with $r^{-1} g^2$ the 4d coupling constant. Summing now to 

1420: $Z=(\phi_l, \psi_\alpha, A_\mu)$ and having into account the phases

1421: in (\ref{phs}), the parameters $\epsilon_k$

1422: reappear and we recover (\ref{strexch2}). 

1423: 

1424: It is also interesting to evaluate the planar piece of the 

1425: polarization tensor, Fig. 3. It arises when both vertices are $V_1$ 

1426: or $V_2$. In these cases, the dependence on the phases $\omega_Z$ 

1427: cancels. Since the spectrum of the theory is

1428: supersymmetric, adding the contributions from all fields will give a

1429: vanishing result. For the same reason the 1-loop vacuum 

1430: energy of the theory is zero. Both results are predicted by the closed

1431: string picture. Notice that the vacuum diagrams can only come from the

1432: leading term in the expansion of $W^{(0)}$, and the planar 

1433: part of the polarization tensor from higher terms in the 

1434: expansion of $W^{(0)}$, which we have not written explicitly in (\ref{Ww}). 

1435: Since $\epsilon_0=0$, we obtain no contribution.

1436: 

1437: 

1438: \section{Conclusions}

1439: 

1440: The main motivation of this paper was to answer some of the open questions 

1441: left in \cite{Armoni:2003va}, where a direct relation between noncommutative

1442: instabilities and closed string tachyons was found for a family of gauge 

1443: theories on D3-branes at $\mathbf{C^3}/\mathbf{Z_N}$ orbifolds. 

1444: In this section we summarize our results.

1445: 

1446: The first question was whether the relation between noncommutative

1447: instabilities and closed string tachyons holds in other examples. A useful

1448: tool to study this problem was a proposal to derive UV/IR mixing 

1449: effects purely in the language of closed strings. This point of

1450: view supported the distinction of two separate pieces in the leading 

1451: UV/IR mixing effects: their functional dependence on $\p$ and 

1452: their sign. The sign determines if the leading UV/IR mixing effects 

1453: will tend to destabilize or not the theory. The functional 

1454: dependence on $\p$ can be finite or divergent in the IR depending if

1455: supersymmetry is or not restored at some high scale.

1456: 

1457: The closed string derivation suggested that the sign of the leading UV/IR 

1458: mixing effects is linked to the misalignment between the NSNS and RR 

1459: towers of the parent string theory. In order to check this proposal, 

1460: we have analyzed in this paper twisted circle compactifications of

1461: Type II string theory. The NSNS and RR towers are shifted in the

1462: same amount by a positive winding energy. Tachyonic modes

1463: appear only for twisted circles of sufficiently small radius.

1464: We have studied Type IIB D3-branes wrapped on the twisted circle

1465: and Type IIA D2-branes transversal to it. We showed that in both cases

1466: the UV/IR mixing signs are governed by the $($mass$)^2$ difference 

1467: between the lowest NSNS and RR modes, and thus does not perceive the 

1468: additional winding energy. 

1469: 

1470: The appropriate field theory limits to D2- and D3-branes correspond to 

1471: string regimes with and without tachyons respectively. Although the signs of 

1472: the leading UV/IR mixing effects were not able to distinguish both regimes, 

1473: their functional dependence on $\p$ was, which we could related to a closed 

1474: string propagator. We found IR finite corrections for

1475: D3-branes and IR divergent ones for D2-branes. 

1476: Therefore the example of the D2-branes provides further support for

1477: a correspondence between closed string tachyons and noncommutative 

1478: instabilities associated to IR divergences. 

1479: For D3-branes a weaker but still interesting result holds.

1480: UV/IR mixing effects are still governed by the lowest modes in the associated

1481: closed string sector. Moreover destabilizing UV/IR mixing effects are absent

1482: for those sectors related to closed strings which are stable for any

1483: radius of the twisted circle. 

1484: 

1485: For the noncommutative gauge theory on D3-branes both at orbifolds

1486: and twisted circle backgrounds, the coefficients that governed the leading 

1487: UV/IR effects could be directly derived from the matter content of the 

1488: theory. It is worth stressing that this is not the case for the theory

1489: on the transversal D2-branes, in which the spectrum is supersymmetric

1490: while supersymmetry is broken by interactions. Yet the leading UV and IR 

1491: behavior of the field theory can be understood in terms of closed 

1492: strings. 

1493: 

1494: Since the severe consequences of UV/IR mixing were first recognized in 

1495: \cite{Minwalla:1999px}, the different behavior of the planar and 

1496: nonplanar sectors of the theory has been a puzzling fact. 

1497: The planar sector can have UV divergences, while

1498: the nonplanar translate UV into IR behavior. Cancellation of the

1499: leading UV divergences for supersymmetric theories implies the

1500: absence of the leading IR divergences. 

1501: This suggests that planar and non-planar terms are not disconnected but 

1502: they are part of a unified structure. In this line, we have shown that 

1503: the gauge invariant effective action containing the leading UV/IR

1504: mixing effects, reproduces also the leading contribution to

1505: the planar graphs. 

1506: The 1-loop vacuum graphs and planar 2-point functions can be 

1507: obtained from the actions (\ref{saction}) and (\ref{strexch}) for the 

1508: D3- and D2-brane theories respectively, when at least one of the 

1509: open Wilson line operators is substituted by its component in the 

1510: identity. 

1511: 

1512: For the gauge theory on the D3-branes, both planar and nonplanar pieces are 

1513: finite. We can then smoothly send the noncommutativity parameter $\theta$

1514: to zero. The action (\ref{saction}) does not trivialize in this limit. 

1515: In the absence of noncommutativity it might look surprising that the 

1516: closed exchange technique is able to reproduce some pieces of the 

1517: effective action. This however points towards the fact that the  

1518: field theory limit retains relevant information about closed strings. 

1519: We regard this as a nontrivial implication of our results. 

1520: 

1521: In \cite{Armoni:2001uw} it was shown that the closed exchange technique

1522: could also reproduce the logarithmic IR divergences of 4d noncommutative

1523: gauge theories. However no string interpretation was given of the 

1524: coefficients governing these divergences. It will be very interesting to 

1525: determine whether a relation with closed string properties also holds for 

1526: subleading UV/IR mixing effects. 

1527: 

1528: 

1529: 

1530: \vspace{5mm}

1531: 

1532: {\centerline {\large \bf Acknowledgments}}

1533: 

1534: \vspace{3mm}

1535: 

1536: I would like to thank C. G\'omez, K. Landsteiner and 

1537: specially A. Uranga for useful discussions.

1538: 

1539: \vspace{5mm}

1540: 

1541: 

1542: \section{Appendix}

1543: 

1544: Let us consider a field with momentum $p_y=(m-{\bar \nu})/R$ 

1545: ($m \in \mathbf{Z}$) along a circle of radius $R$, with ${\bar \nu}$

1546: arbitrary. We define $N_{\bar \nu}$ as

1547: \beq

1548: \sum_{m=-\infty}^\infty {1 \over (m-\bar \nu)^2 + R^2 p^2}=

1549: {\pi \over R \, p} \, N_{\bar \nu}(R \, p)  \, .

1550: \label{dis}

1551: \eeq

1552: For bosons with integer moding along the circle (${\bar \nu}=0$),

1553: $N_0=1+2 n_B$, with $n_B$ the Bose-Einstein distribution; for 

1554: fermions with half-integer moding (${\bar \nu}=1/2$), 

1555: $N_{1 \over 2}=1-2 n_F$, with $n_F$

1556: the Fermi-Dirac distribution. For arbitrary $\bar \nu$ we can

1557: consider ${1 \over 2} (-1)^F (N_{\bar \nu}-1)$, with $F$ the fermion 

1558: number, the 

1559: generalization of $n_B$ and $n_F$. The sum (\ref{dis}) can be

1560: evaluated in a closed form, with the result

1561: \beq

1562: N_{\bar \nu}(x) = {\sinh 2 \pi x \over \cosh 2 \pi x -

1563: \cos 2 \pi {\bar \nu}}  \, .

1564: \label{N}

1565: \eeq 

1566: In general ${1 \over 2}(-1)^F (N_{\bar \nu}-1)$, contrary to $n_{B,F}$,

1567: is not always positive and thus it can not have the interpretation

1568: of occupation numbers. This is not a problem, since $y$ will be 

1569: always an spatial direction for us.

1570: 

1571: We focus next in a 4d euclidean noncommutative $U(1)$ theory with 

1572: one direction compactified on a circle. The matter content 

1573: will consist of two real scalars in the adjoint

1574: representation: $\varphi$, with arbitrary ${\bar \nu}$, and $\phi$, 

1575: with ${\bar \nu}=0$ and couplings as if obtained 

1576: from the dimensional reduction of the gauge field from 5 to 4 dimensions. 

1577: We want to derive the contribution to the polarization tensor of the gauge 

1578: field and self-energy of $\phi$, from the 1-loop exchange of $\varphi$. 

1579: We are only interested in the contributions that arise from high loop 

1580: momenta. These are given by

1581: \beqa

1582: {\bar \Pi}^{AB} &=& 4g^2 {1 \over 2 \pi R} \sum_m \int {d^3l \over (2 \pi)^3} 

1583: {2 L^A L^B-\delta^{A B}L^2 \over L^4} \sin^2  \label{pigen}

1584: {\p .l \over 2} \, , \\ 

1585: {\bar \Sigma} &=& - 4g^2 {1 \over 2 \pi R} \sum_m \int {d^3l \over (2 \pi)^3} 

1586: {1 \over L^2} \sin^2 {\p .l \over 2} \, , 

1587: \label{inte}

1588: \eeqa

1589: where the indices $A,B$ run over the non-compact directions

1590: $\mu=1,2,3$ and the circle; $L=(l_\mu,l_y)$, with $l_y=(m-{\bar \nu})/R$, 

1591: and the external momentum is $P=(p_\mu,p_y)$.

1592: For $p_y \neq 0$, (\ref{pigen}) does not fulfills the Ward identities

1593: and subleading contributions must be taken

1594: into account. In the following we limit ourselves to $p_y=0$.

1595: 

1596: Using (\ref{dis}), we obtain

1597: \beqa

1598: {\bar \Pi}^{\mu \nu} &=& 2 g^2 \int {d^3 l \over (2 \pi)^3} \left[ \left(

1599: {N_{\bar \nu} \over l}- {d N_{\bar \nu} \over dl} \right) 

1600: {\hat l}^\mu {\hat l}^\nu -{N_{\bar \nu} \over l} \delta^{\mu \nu} \right] 

1601: \sin^2 {\p.l \over 2} \, , \\

1602: {\bar \Pi}^{yy} & = & 

1603: 2 g^2\int {d^3 l \over (2 \pi)^3}{d N_{\bar \nu} \over dl} 

1604: \sin^2 {\p.l \over 2} \, , \\

1605: {\bar \Sigma} & = &  - 2 g^2\int {d^3 l \over (2 \pi)^3}{N_{\bar \nu} \over l} 

1606: \sin^2 {\p.l \over 2} \, ,

1607: \eeqa

1608: where ${\hat l}^\mu=l^\mu / l$. Using $\sin^2(x)={1 \over 2}(1-\cos 2x)$,

1609: we can separate the planar and nonplanar pieces in the previous expressions.  

1610: The integration over the angular variables

1611: can be performed in an straightforward way, with the following results

1612: for the non-planar part

1613: \beqa

1614: \Pi^{\mu \nu} &=& {\p^\mu \p^\nu \over \p^2} {g^2 \over 2 \pi^2 \p}

1615: \int_0^\infty dl N_{\bar \nu} \, ( \sin \p l - \p l \cos \p l ) \, , 

1616: \label{apinp1}\\

1617: \Pi^{yy} &=& {g^2 \over 2 \pi^2 \p}

1618: \int_0^\infty dl N_{\bar \nu} \, ( \sin \p l + \p l \cos \p l ) \, , \\

1619: \Sigma &=& {g^2 \over 2 \pi^2 \p}

1620: \int_0^\infty dl N_{\bar \nu} \, \sin \p l \, .

1621: \label{apinp3}

1622: \eeqa

1623: The planar part is

1624: \beq

1625: {\hat \Pi}^{\mu \nu}=0 \, , \;\;\;\;\; 

1626: {\hat \Pi}^{yy} =2 {\hat \Sigma} = 

1627: -{g^2 \over \pi^2} \int_0^\infty dl N_{\bar \nu}\, l \, .

1628: \label{apip}

1629: \eeq

1630: 

1631: The expressions above apply when the field circulating in the loop

1632: is a real scalar. For a general field with fractional momentum

1633: ${\bar \nu}$, we should multiply (\ref{apinp1})-(\ref{apip}) by

1634: the number of degrees of freedom of the field times $(-1)^F$,

1635: with $F$ its fermion number. When there are several adjoint scalars,

1636: the self-energies are $\Sigma^{ab}=\delta^{ab} \Sigma$ with

1637: $\Sigma$ as in (\ref{apinp1})-(\ref{apip}). When there are several

1638: group factors, the non-planar 2-point functions will mix different

1639: group labels if there are bifundamental fields running in the loop.

1640: Each bifundamental degree of freedom gives half the contribution 

1641: in (\ref{apinp1})-(\ref{apinp3}). Since (\ref{apinp1})-(\ref{apip})

1642: only depend on the components of $p_\mu$ along the noncommutative

1643: plane, the Wick rotation along the additional noncompact direction to 

1644: Minkowski signature is trivial.

1645: 

1646: 

1647: \bigskip

1648: 

1649: 

1650: 

1651: \begin{thebibliography}{99}

1652: 

1653: %\cite{Doplicher:1994tu}

1654: \bibitem{Doplicher:1994tu}

1655: S.~Doplicher, K.~Fredenhagen and J.~E.~Roberts,

1656: ``The Quantum structure of space-time at the Planck scale and quantum 

1657: fields,''

1658: Commun.\ Math.\ Phys.\  {\bf 172} (1995) 187

1659: [arXiv:hep-th/0303037].

1660: %%CITATION = HEP-TH 0303037;%%

1661: 

1662: %\cite{Banks:1996vh}

1663: \bibitem{Banks:1996vh}

1664: T.~Banks, W.~Fischler, S.~H.~Shenker and L.~Susskind,

1665: ``M theory as a matrix model: A conjecture,''

1666: Phys.\ Rev.\ D {\bf 55} (1997) 5112

1667: [arXiv:hep-th/9610043].

1668: %%CITATION = HEP-TH 9610043;%%

1669: 

1670: %\cite{Witten:1995im}

1671: \bibitem{Witten:1995im}

1672: E.~Witten,

1673: ``Bound states of strings and p-branes,''

1674: Nucl.\ Phys.\ B {\bf 460} (1996) 335

1675: [arXiv:hep-th/9510135].

1676: %%CITATION = HEP-TH 9510135;%%

1677: 

1678: %\cite{Minwalla:1999px}

1679: \bibitem{Minwalla:1999px}

1680: S.~Minwalla, M.~Van Raamsdonk and N.~Seiberg,

1681: %``Noncommutative perturbative dynamics,''

1682: JHEP {\bf 0002} (2000) 020

1683: [arXiv:hep-th/9912072].

1684: %%CITATION = HEP-TH 9912072;%%

1685: 

1686: %\cite{Douglas:1997fm}

1687: \bibitem{Douglas:1997fm}

1688: M.~R.~Douglas and C.~M.~Hull,

1689: ``D-branes and the noncommutative torus,''

1690: JHEP {\bf 9802} (1998) 008

1691: [arXiv:hep-th/9711165].

1692: %%CITATION = HEP-TH 9711165;%%

1693: 

1694: %\cite{Arcioni:2000bz}

1695: \bibitem{Arcioni:2000bz}

1696: G.~Arcioni, J.~L.~Barbon, J.~Gomis and M.~A.~Vazquez-Mozo,

1697: ``On the stringy nature of winding modes in noncommutative 

1698: thermal field  theories,''

1699: JHEP {\bf 0006} (2000) 038

1700: [arXiv:hep-th/0004080].

1701: %%CITATION = HEP-TH 0004080;%%

1702: 

1703: %\cite{Gomis:2000bn}

1704: \bibitem{Gomis:2000bn}

1705: J.~Gomis, M.~Kleban, T.~Mehen, M.~Rangamani and S.~H.~Shenker,

1706: ``Noncommutative gauge dynamics from the string worldsheet,''

1707: JHEP {\bf 0008} (2000) 011

1708: [arXiv:hep-th/0003215].

1709: %%CITATION = HEP-TH 0003215;%%

1710: 

1711: %\cite{Liu:2000qh}

1712: \bibitem{Liu:2000qh}

1713: H.~Liu and J.~Michelson,

1714: ``Stretched strings in noncommutative field theory,''

1715: Phys.\ Rev.\ D {\bf 62} (2000) 066003

1716: [arXiv:hep-th/0004013].

1717: %%CITATION = HEP-TH 0004013;%%

1718: 

1719: %\cite{Armoni:2003va}

1720: \bibitem{Armoni:2003va}

1721: A.~Armoni, E.~Lopez and A.~M.~Uranga,

1722: ``Closed strings tachyons and non-commutative instabilities,''

1723: JHEP {\bf 0302}, 020 (2003)

1724: [arXiv:hep-th/0301099].

1725: %%CITATION = HEP-TH 0301099;%%

1726: 

1727: %\cite{Ruiz:2001hu}

1728: \bibitem{Ruiz:2001hu}

1729: F.~R.~Ruiz,

1730: ``Gauge-fixing independence of IR divergences in non-commutative U(1),  

1731: perturbative tachyonic instabilities and supersymmetry,''

1732: Phys.\ Lett.\ B {\bf 502}, 274 (2001)

1733: [hep-th/0012171].

1734: %%CITATION = HEP-TH 0012171;

1735: 

1736: %\cite{Landsteiner:2001ky}

1737: \bibitem{Landsteiner:2001ky}

1738: K.~Landsteiner, E.~Lopez and M.~H.~Tytgat,

1739: ``Instability of non-commutative SYM theories at finite temperature,''

1740: JHEP {\bf 0106} (2001) 055

1741: [arXiv:hep-th/0104133].

1742: %%CITATION = HEP-TH 0104133;%%

1743: 

1744: %\cite{Russo:1995ik}

1745: \bibitem{Russo:2001na}

1746: J.~G.~Russo and A.~A.~Tseytlin,

1747: ``Magnetic flux tube models in superstring theory,''

1748: Nucl.\ Phys.\ B {\bf 461} (1996) 131

1749: [arXiv:hep-th/9508068];

1750: %%CITATION = HEP-TH 9508068;%%

1751: %\cite{Russo:2001na}

1752: %\bibitem{Russo:2001na}

1753: J.~G.~Russo and A.~A.~Tseytlin,

1754: ``Supersymmetric fluxbrane intersections and closed string tachyons,''

1755: JHEP {\bf 0111} (2001) 065

1756: [arXiv:hep-th/0110107].

1757: %%CITATION = HEP-TH 0110107;%%

1758: 

1759: %\cite{Armoni:2001uw}

1760: \bibitem{Armoni:2001uw}

1761: A.~Armoni and E.~Lopez,

1762: ``UV/IR mixing via closed strings and tachyonic instabilities,''

1763: Nucl.\ Phys.\ B {\bf 632} (2002) 240

1764: [arXiv:hep-th/0110113].

1765: %%CITATION = HEP-TH 0110113;%%

1766: 

1767: %\cite{Matusis:2000jf}

1768: \bibitem{Matusis:2000jf}

1769: A.~Matusis, L.~Susskind and N.~Toumbas,

1770: ``The IR/UV connection in the non-commutative gauge theories,''

1771: JHEP {\bf 0012} (2000) 002

1772: [arXiv:hep-th/0002075].

1773: %%CITATION = HEP-TH 0002075;%%

1774: 

1775: %\cite{Ibanez:1998qp}

1776: \bibitem{Ibanez:1998qp}

1777: L.~E.~Ibanez, R.~Rabadan and A.~M.~Uranga,

1778: ``Anomalous U(1)'s in type I and type IIB D = 4, N = 1 string vacua,''

1779: Nucl.\ Phys.\ B {\bf 542} (1999) 112

1780: [arXiv:hep-th/9808139].

1781: %%CITATION = HEP-TH 9808139;%%

1782: 

1783: %\cite{Armoni:2002fh}

1784: \bibitem{Armoni:2002fh}

1785: A.~Armoni, E.~Lopez and S.~Theisen,

1786: ``Nonplanar anomalies in noncommutative theories and the Green-Schwarz  

1787: mechanism,''

1788: JHEP {\bf 0206} (2002) 050

1789: [arXiv:hep-th/0203165].

1790: %%CITATION = HEP-TH 0203165;%%

1791: 

1792: %\cite{Martin:2000qf}

1793: \bibitem{Martin:2000qf}

1794: C.~P.~Martin,

1795: ``The UV and IR origin of non-Abelian chiral gauge anomalies on  

1796: noncommutative Minkowski space-time,''

1797: J.\ Phys.\ A {\bf 34} (2001) 9037

1798: [arXiv:hep-th/0008126].

1799: %%CITATION = HEP-TH 0008126;%%

1800: 

1801: %\cite{Kiem:2001pw}

1802: \bibitem{Kiem:2001pw}

1803: Y.~Kiem, S.~J.~Rey, H.~T.~Sato and J.~T.~Yee,

1804: ``Open Wilson lines and generalized star product in nocommutative 

1805: scalar  field theories,''

1806: Phys.\ Rev.\ D {\bf 65} (2002) 026002

1807: [arXiv:hep-th/0106121];

1808: %%CITATION = HEP-TH 0106121;%%

1809: %\cite{Kiem:2001dk}

1810: %\bibitem{Kiem:2001dk}

1811: %Y.~Kiem, S.~J.~Rey, H.~T.~Sato and J.~T.~Yee,

1812: ``Anatomy of one-loop effective action in noncommutative scalar field  

1813: theories,''

1814: Eur.\ Phys.\ J.\ C {\bf 22} (2002) 757

1815: [arXiv:hep-th/0107106].

1816: %%CITATION = HEP-TH 0107106;%%

1817: 

1818: %\cite{VanRaamsdonk:2001jd}

1819: \bibitem{VanRaamsdonk:2001jd}

1820: M.~Van Raamsdonk,

1821: ``The meaning of infrared singularities in noncommutative gauge theories,''

1822: JHEP {\bf 0111}, 006 (2001)

1823: [hep-th/0110093].

1824: %%CITATION = HEP-TH 0110093;%%

1825: 

1826: %\cite{Kiem:2001dm}

1827: \bibitem{Kiem:2001dm}

1828: Y.~j.~Kiem, Y.~j.~Kim, C.~Ryou and H.~T.~Sato,

1829: ``One-loop noncommutative U(1) gauge theory from bosonic worldline  

1830: approach,''

1831: Nucl.\ Phys.\ B {\bf 630} (2002) 55

1832: [hep-th/0112176].

1833: %%CITATION = HEP-TH 0112176;%%

1834: 

1835: %\cite{Kiem:2001du}

1836: \bibitem{Kiem:2001du}

1837: Y.~j.~Kiem, S.~S.~Kim, S.~J.~Rey and H.~T.~Sato,

1838: ``Anatomy of two-loop effective action in noncommutative field theories,''

1839: Nucl.\ Phys.\ B {\bf 641} (2002) 256

1840: [arXiv:hep-th/0110066];

1841: %%CITATION = HEP-TH 0110066;%%

1842: %\cite{Kiem:2001fn}

1843: %\bibitem{Kiem:2001fn}

1844: %Y.~Kiem, S.~Lee, S.~J.~Rey and H.~T.~Sato,

1845: ``Interacting open Wilson lines in noncommutative field theories,''

1846: Phys.\ Rev.\ D {\bf 65} (2002) 046003

1847: [arXiv:hep-th/0110215].

1848: %%CITATION = HEP-TH 0110215;%%

1849: 

1850: %\cite{Ishibashi:2000hs}

1851: \bibitem{Ishibashi:2000hs}

1852: N.~Ishibashi, S.~Iso, H.~Kawai and Y.~Kitazawa,

1853: ``Wilson loops in noncommutative Yang-Mills,''

1854: Nucl.\ Phys.\ B {\bf 573}, 573 (2000)

1855: [hep-th/9910004].

1856: %%CITATION = HEP-TH 9910004;%%

1857: 

1858: %\cite{Rey:2000hh}

1859: \bibitem{Rey:2000hh}

1860: S.~J.~Rey and R.~von Unge,

1861: ``S-duality, noncritical open string and noncommutative gauge theory,''

1862: Phys.\ Lett.\ B {\bf 499} (2001) 215

1863: [arXiv:hep-th/0007089].

1864: %%CITATION = HEP-TH 0007089;%%

1865: 

1866: %\cite{Das:2000md}

1867: \bibitem{Das:2000md}

1868: S.~R.~Das and S.~J.~Rey,

1869: ``Open Wilson lines in noncommutative gauge theory and tomography of 

1870: holographic dual supergravity,''

1871: Nucl.\ Phys.\ B {\bf 590} (2000) 453

1872: [arXiv:hep-th/0008042].

1873: %%CITATION = HEP-TH 0008042;%%

1874: 

1875: %\cite{Gross:2000ba}

1876: \bibitem{Gross:2000ba}

1877: D.~J.~Gross, A.~Hashimoto and N.~Itzhaki,

1878: ``Observables of non-commutative gauge theories,''

1879: hep-th/0008075.

1880: %%CITATION = HEP-TH 0008075;%%

1881: 

1882: %\cite{Douglas:1996sw}

1883: \bibitem{Douglas:1996sw}

1884: M.~R.~Douglas and G.~W.~Moore,

1885: ``D-branes, Quivers, and ALE Instantons,''

1886: arXiv:hep-th/9603167.

1887: %%CITATION = HEP-TH 9603167;%%

1888: 

1889: %\cite{Das:2000ur}

1890: \bibitem{Das:2000ur}

1891: S.~R.~Das and S.~P.~Trivedi,

1892: ``Supergravity couplings to noncommutative branes, open Wilson lines 

1893: and generalized star products,''

1894: JHEP {\bf 0102} (2001) 046

1895: [arXiv:hep-th/0011131].

1896: %%CITATION = HEP-TH 0011131;%%

1897: 

1898: %\cite{Okawa:2000sh}

1899: \bibitem{Okawa:2000sh}

1900: Y.~Okawa and H.~Ooguri,

1901: ``How noncommutative gauge theories couple to gravity,''

1902: Nucl.\ Phys.\ B {\bf 599} (2001) 55

1903: [arXiv:hep-th/0012218].

1904: %%CITATION = HEP-TH 0012218;%%

1905: 

1906: %\cite{Liu:2001ps}

1907: \bibitem{Liu:2001ps}

1908: H.~Liu and J.~Michelson,

1909: %``Supergravity couplings of noncommutative D-branes,''

1910: Nucl.\ Phys.\ B {\bf 615} (2001) 169

1911: [arXiv:hep-th/0101016].

1912: %%CITATION = HEP-TH 0101016;%%

1913: 

1914: %\cite{Seiberg:1999vs}

1915: \bibitem{Seiberg:1999vs}

1916: N.~Seiberg and E.~Witten,

1917: ``String theory and noncommutative geometry,''

1918: JHEP {\bf 9909} (1999) 032

1919: [arXiv:hep-th/9908142].

1920: %%CITATION = HEP-TH 9908142;%%

1921: 

1922: %\cite{Bilal:2000bk}

1923: \bibitem{Bilal:2000bk}

1924: A.~Bilal, C.~S.~Chu and R.~Russo,

1925: ``String theory and noncommutative field theories at one loop,''

1926: Nucl.\ Phys.\ B {\bf 582} (2000) 65

1927: [arXiv:hep-th/0003180].

1928: %%CITATION = HEP-TH 0003180;%%

1929: 

1930: %\cite{Takayanagi:2001jj}

1931: \bibitem{Takayanagi:2001jj}

1932: T.~Takayanagi and T.~Uesugi,

1933: ``Orbifolds as Melvin geometry,''

1934: JHEP {\bf 0112} (2001) 004

1935: [arXiv:hep-th/0110099].

1936: %%CITATION = HEP-TH 0110099;%%

1937: 

1938: %\cite{Dudas:2001ux}

1939: \bibitem{Dudas:2001ux}

1940: E.~Dudas and J.~Mourad,

1941: %``D-branes in string theory Melvin backgrounds,''

1942: Nucl.\ Phys.\ B {\bf 622} (2002) 46

1943: [arXiv:hep-th/0110186].

1944: %%CITATION = HEP-TH 0110186;%%

1945: 

1946: %\cite{Takayanagi:2001aj}

1947: \bibitem{Takayanagi:2001aj}

1948: T.~Takayanagi and T.~Uesugi,

1949: %``D-branes in Melvin background,''

1950: JHEP {\bf 0111} (2001) 036

1951: [arXiv:hep-th/0110200].

1952: %%CITATION = HEP-TH 0110200;%%

1953: 

1954: %\cite{Fischler:2000fv}

1955: \bibitem{Fischler:2000fv}

1956: W.~Fischler, E.~Gorbatov, A.~Kashani-Poor, S.~Paban, P.~Pouliot and J.~Gomis,

1957: ``Evidence for winding states in noncommutative quantum field theory,''

1958: JHEP {\bf 0005} (2000) 024

1959: [arXiv:hep-th/0002067].

1960: %%CITATION = HEP-TH 0002067;%%

1961: 

1962: 

1963: 

1964: 

1965: 

1966: \end{thebibliography}

1967: 

1968: \end{document}

1969: 

1970: 

1971: 

1972: 

1973: 

1974: 

1975: 

1976: 

1977: 

1978: