hep-th0307248/df.tex
1: \input harvmac
2: \overfullrule=0pt
3: \parindent 25pt
4: \tolerance=10000
5: 
6: %\draftmode
7: \input epsf
8: 
9: \newcount\figno
10: \figno=0
11: \def\fig#1#2#3{
12: \par\begingroup\parindent=0pt\leftskip=1cm\rightskip=1cm\parindent=0pt
13: \baselineskip=11pt
14: \global\advance\figno by 1
15: \midinsert
16: \epsfxsize=#3
17: \centerline{\epsfbox{#2}}
18: \vskip 12pt
19: {\bf Fig.\ \the\figno: } #1\par
20: \endinsert\endgroup\par
21: }
22: \def\figlabel#1{\xdef#1{\the\figno}}
23: \def\encadremath#1{\vbox{\hrule\hbox{\vrule\kern8pt\vbox{\kern8pt
24: \hbox{$\displaystyle #1$}\kern8pt}
25: \kern8pt\vrule}\hrule}}
26: 
27: \def\half{{\textstyle{1\over2}}}
28: \def\titlestyle#1{\par\begingroup \interlinepenalty=9999
29: \fourteenpoint
30: \noindent #1\par\endgroup }
31: \def\adag{a^{\dagger}}
32: \def\apm{{\alpha^{\prime}}}
33: \def\erf{{\,\rm erf}}
34: \font\cmss=cmss10
35: \font\cmsss=cmss10 at 7pt
36: \def\rlx{\relax\leavevmode}
37: \def\inbar{\vrule height1.5ex width.4pt depth0pt}
38: 
39: \def\half{{1\over 2}}
40:  \def\sub{\scriptscriptstyle}
41:  \def\m{{\mu}}
42:  \def\w{{\omega}}       
43:  \def\n{{\nu}}
44:  \def\ep{{\epsilon}}
45:  \def\d{{\delta}}
46:  \def\rh{\rho}
47:  \def\t{{\theta}}
48:  \def\a{{\alpha}}
49:  \def\T{{\Theta}}
50:  \def\frac#1#2{{#1\over #2}}
51:  \def\l{{\lambda}}
52:  \def\G{{\Gamma}}
53:  \def\D{{\Delta}}
54:  \def\g{{\gamma}}
55:  \def\s{{\sigma}}
56:  \def\ch{{\chi}}
57:  \def\b{{\beta}}
58:  \def\CA{{\cal A}}
59:  \def\CC{{\cal C}}
60:  \def\CI{{\cal I}}
61:  \def\CO{{\cal O}}
62:  \def\o{{\rm ord}}
63:  \def\Ph{{\Phi }}
64:  \def\L{{\Lambda}}
65:  \def\CN{{\cal N}}
66:  \def\p{\partial}
67:  \def\pslash{\p \llap{/}}
68:  \def\Dslash{D \llap{/}}
69:  \def\Mp{m_{{\rm P}}}
70:  \def\apm{{\alpha'}}
71:  \def\r{\rightarrow}
72:  \def\Re{{\rm Re}}
73:  \def\MG{{\bf MG:}}
74:  \def\al{\alpha'}
75:  \def\de{\partial}
76:  \def\db{\bar{\partial}}
77:  \def\we{\wedge}
78:  \def\lr{\leftrightarrow}
79:  \def\f {\frac}
80:  \def\ti{\tilde}
81:  \def\ap{\alpha}
82:  \def\pr{\propto}
83:  \def\mb{\mathbf}
84:  \def\ddd{\cdot\cdot\cdot}
85:  \def\no{\nonumber \\}
86:  \def\la{\langle}
87:  \def\lb{\rangle}
88:  \def\ep{\epsilon}
89:  \def\ddbp{\mbox{D}p-\overline{\mbox{D}p}}
90:  \def\ddbt{\mbox{D}2-\overline{\mbox{D}2}}
91:  \def\ov{\overline}
92:  \def\sq{\sqrt}
93:  \def\cl{\centerline}
94: 
95: 
96: \lref\AdPoSi{
97: A.~Adams, J.~Polchinski and E.~Silverstein,
98: ``Don't panic! Closed string tachyons in ALE space-times,''
99: JHEP {\bf 0110} (2001) 029
100: [arXiv:hep-th/0108075].}
101: %%CITATION = HEP-TH 0108075;%%
102: 
103: 
104: \lref\DabholkarIF{
105: A.~Dabholkar,
106: ``Tachyon condensation and black hole entropy,''
107: arXiv:hep-th/0111004.
108: %%CITATION = HEP-TH 0111004;%%
109: }
110: 
111: 
112: 
113: \lref\chicago{ J.~A.~Harvey, D.~Kutasov, E.~J.~Martinec, G.~Moore,
114: ``Localized tachyons and RG flows'', arXiv:hep-th/0111154.
115: %%CITATION = HEP-TH 0111154;%%
116: }
117: 
118: \lref\bega{
119: O.~Bergman and M.~R.~Gaberdiel,
120: ``Dualities of type 0 strings,''
121: JHEP {\bf 9907} (1999) 022
122: [arXiv:hep-th/9906055].}
123: %%CITATION = HEP-TH 9906055;%%
124: 
125: 
126: \lref\Va{
127: C.~Vafa,
128: ``Mirror symmetry and closed string tachyon condensation,''
129: arXiv:hep-th/0111051;
130: %%CITATION = HEP-TH 0111051;%%
131: A.~Dabholkar and C.~Vafa,
132: ``tt* geometry and closed string tachyon potential,''
133: JHEP {\bf 0202} (2002) 008
134: [arXiv:hep-th/0111155].}
135: %%CITATION = HEP-TH 0111155;%%
136: 
137: 
138: 
139: \lref\DaGuHeMi{
140: J.~R.~David, M.~Gutperle, M.~Headrick and S.~Minwalla,
141: ``Closed string tachyon condensation on twisted circles,''
142: JHEP {\bf 0202}, 041 (2002)
143: [arXiv:hep-th/0111212].
144: %%CITATION = HEP-TH 0111212;%%
145: }
146: 
147: 
148: \lref\GuHeMiSc{
149: M.~Gutperle, M.~Headrick, S.~Minwalla and V.~Schomerus,
150: ``Space-time energy decreases under world-sheet RG flow,''
151: arXiv:hep-th/0211063.}
152: %%CITATION = HEP-TH 0211063;%%
153: 
154: \lref\HoVa{
155: K.~Hori and C.~Vafa,
156: ``Mirror symmetry,''
157: arXiv:hep-th/0002222.}
158: %%CITATION = HEP-TH 0002222;%%
159: 
160: \lref\HoIqVa{
161: K.~Hori, A.~Iqbal and C.~Vafa,
162: ``D-branes and mirror symmetry,''
163: arXiv:hep-th/0005247.}
164: %%CITATION = HEP-TH 0005247;%%
165: 
166: 
167: 
168: \lref\HoKa{
169: K.~Hori and A.~Kapustin,
170: ``Duality of the fermionic 2d black hole and N = 2 Liouville 
171: theory as  mirror symmetry,''
172: JHEP {\bf 0108} (2001) 045
173: [arXiv:hep-th/0104202].}
174: %%CITATION = HEP-TH 0104202;%%
175: 
176: 
177: \lref\WittenBSFT{
178: E.~Witten,
179: ``On background independent open string field theory,''
180: Phys.\ Rev.\ D {\bf 46} (1992) 5467
181: [arXiv:hep-th/9208027];
182: %%CITATION = HEP-TH 9208027;%%
183: 
184: E.~Witten,
185: ``Some computations in background independent off-shell string theory,''
186: Phys.\ Rev.\ D {\bf 47} (1993) 3405
187: [arXiv:hep-th/9210065].}
188: %%CITATION = HEP-TH 9210065;%%
189: 
190: 
191: \lref\BSFT{
192: A.~A.~Gerasimov and S.~L.~Shatashvili,
193: ``On exact tachyon potential in open string field theory,''
194: JHEP {\bf 0010} (2000) 034
195: [arXiv:hep-th/0009103];
196: %%CITATION = HEP-TH 0009103;%%
197: 
198: D.~Kutasov, M.~Marino and G.~W.~Moore,
199: ``Some exact results on tachyon condensation in string field theory,''
200: JHEP {\bf 0010} (2000) 045
201: [arXiv:hep-th/0009148];
202: 
203: D.~Kutasov, M.~Marino and G.~W.~Moore,
204: ``Remarks on tachyon condensation in superstring field theory,''
205: arXiv:hep-th/0010108.}
206: %%CITATION = HEP-TH 0010108;%%}
207: 
208: 
209: \lref\DD{
210: P.~Kraus and F.~Larsen,
211: ``Boundary string field theory of the DD-bar system,''
212: Phys.\ Rev.\ D {\bf 63} (2001) 106004
213: [arXiv:hep-th/0012198];
214: %%CITATION = HEP-TH 0012198;%%
215: 
216: T.~Takayanagi, S.~Terashima and T.~Uesugi,
217: ``Brane-antibrane action from boundary string field theory,''
218: JHEP {\bf 0103} (2001) 019
219: [arXiv:hep-th/0012210].}
220: %%CITATION = HEP-TH 0012210;%%
221: 
222: \lref\Ho{
223: K.~Hori,
224: ``Linear models of supersymmetric D-branes,''
225: arXiv:hep-th/0012179.}
226: %%CITATION = HEP-TH 0012179;%%
227: 
228: 
229: 
230: \lref\ta{
231: T.~Takayanagi,
232: ``Holomorphic tachyons and fractional D-branes,''
233: Nucl.\ Phys.\ B {\bf 603} (2001) 259
234: [arXiv:hep-th/0103021];
235: %%CITATION = HEP-TH 0103021;%%
236: T.~Takayanagi,
237: ``Tachyon condensation on orbifolds and McKay correspondence,''
238: Phys.\ Lett.\ B {\bf 519} (2001) 137
239: [arXiv:hep-th/0106142].}
240: %%CITATION = HEP-TH 0106142;%%
241: 
242: \lref\DoMo{
243: M.~R.~Douglas and G.~W.~Moore,
244: ``D-branes, Quivers, and ALE Instantons,''
245: arXiv:hep-th/9603167.}
246: %%CITATION = HEP-TH 9603167;%%
247: 
248: \lref\DiDoGo{
249: D.~E.~Diaconescu, M.~R.~Douglas and J.~Gomis,
250: ``Fractional branes and wrapped branes,''
251: JHEP {\bf 9802} (1998) 013
252: [arXiv:hep-th/9712230].}
253: %%CITATION = HEP-TH 9712230;%%
254: 
255: \lref\DiGo{
256: D.~E.~Diaconescu and J.~Gomis,
257: ``Fractional branes and boundary states in orbifold theories,''
258: JHEP {\bf 0010} (2000) 001
259: [arXiv:hep-th/9906242].}
260: %%CITATION = HEP-TH 9906242;%%
261: 
262: 
263: \lref\TaUeo{
264: T.~Takayanagi and T.~Uesugi,
265: ``Orbifolds as Melvin geometry,''
266: JHEP {\bf 0112} (2001) 004
267: [arXiv:hep-th/0110099].}
268: %%CITATION = HEP-TH 0110099;%%
269: 
270: 
271: \lref\TaUet{
272: T.~Takayanagi and T.~Uesugi,
273: ``D-branes in Melvin background,''
274: JHEP {\bf 0111} (2001) 036
275: [arXiv:hep-th/0110200];
276: %%CITATION = HEP-TH 0110200;%%
277: T.~Takayanagi and T.~Uesugi,
278: ``Flux stabilization of D-branes in NSNS Melvin background,''
279: Phys.\ Lett.\ B {\bf 528} (2002) 156
280: [arXiv:hep-th/0112199].}
281: %%CITATION = HEP-TH 0112199;%%
282: 
283: \lref\DuMo{
284: E.~Dudas and J.~Mourad,
285: ``D-branes in string theory Melvin backgrounds,''
286: Nucl.\ Phys.\ B {\bf 622} (2002) 46
287: [arXiv:hep-th/0110186].}
288: %%CITATION = HEP-TH 0110186;%%
289: 
290: \lref\CoGu{
291: M.~S.~Costa and M.~Gutperle,
292: ``The Kaluza-Klein Melvin solution in M-theory,''
293: JHEP {\bf 0103} (2001) 027
294: [arXiv:hep-th/0012072].}
295: %%CITATION = HEP-TH 0012072;%%
296: 
297: \lref\GuSt{
298: M.~Gutperle and A.~Strominger,
299: ``Fluxbranes in string theory,''
300: JHEP {\bf 0106} (2001) 035
301: [arXiv:hep-th/0104136].}
302: %%CITATION = HEP-TH 0104136;%%
303: 
304: \lref\Gu{
305: M.~Gutperle,
306: ``A note on perturbative and nonperturbative instabilities of 
307: twisted  circles,''
308: Phys.\ Lett.\ B {\bf 545}, 379 (2002)
309: [arXiv:hep-th/0207131].
310: %%CITATION = HEP-TH 0207131;%%
311: }
312: 
313: \lref\Da{
314: J.~R.~David,
315: ``Unstable magnetic fluxes in heterotic string theory,''
316: JHEP {\bf 0209}, 006 (2002)
317: [arXiv:hep-th/0208011].
318: %%CITATION = HEP-TH 0208011;%%
319: }
320: 
321: \lref\HeKaLaMc{
322: S.~Hellerman, S.~Kachru, A.~E.~Lawrence and J.~McGreevy,
323: ``Linear sigma models for open strings,''
324: JHEP {\bf 0207} (2002) 002
325: [arXiv:hep-th/0109069].}
326: %%CITATION = HEP-TH 0109069;%%
327: 
328: \lref\EgSu{
329: T.~Eguchi and Y.~Sugawara,
330: ``D-branes in singular Calabi-Yau n-fold and N = 2 Liouville theory,''
331: Nucl.\ Phys.\ B {\bf 598} (2001) 467
332: [arXiv:hep-th/0011148].}
333: %%CITATION = HEP-TH 0011148;%%
334: 
335: 
336: \lref\OoOzYi{
337: H.~Ooguri, Y.~Oz and Z.~Yin,
338: ``D-branes on Calabi-Yau spaces and their mirrors,''
339: Nucl.\ Phys.\ B {\bf 477} (1996) 407
340: [arXiv:hep-th/9606112].}
341: 
342: 
343: \lref\RussoTF{
344: J.~G.~Russo and A.~A.~Tseytlin,
345:  ``Magnetic backgrounds and tachyonic instabilities in closed superstring  
346: theory and M-theory,''
347: Nucl.\ Phys.\ B {\bf 611}, 93 (2001)
348: [arXiv:hep-th/0104238].
349: %%CITATION = HEP-TH 0104238;%%
350: }
351: 
352: 
353: 
354: \lref\WittenYC{
355: E.~Witten,
356: ``Phases of N = 2 theories in two dimensions,''
357: Nucl.\ Phys.\ B {\bf 403}, 159 (1993)
358: [arXiv:hep-th/9301042].
359: %%CITATION = HEP-TH 9301042;%%
360: }
361: 
362: 
363: \lref\SuyamaGD{
364: T.~Suyama,
365: ``Properties of string theory on Kaluza-Klein Melvin background,''
366: JHEP {\bf 0207}, 015 (2002)
367: [arXiv:hep-th/0110077].
368: %%CITATION = HEP-TH 0110077;%%
369: }
370: 
371: 
372: 
373: \lref\TseytlinZV{
374: A.~A.~Tseytlin,
375: ``Closed superstrings in magnetic flux tube background,''
376: Nucl.\ Phys.\ Proc.\ Suppl.\  {\bf 49}, 338 (1996)
377: [arXiv:hep-th/9510041].
378: %%CITATION = HEP-TH 9510041;%%
379: }
380: 
381: \lref\TseytlinEI{
382: A.~A.~Tseytlin,
383: ``Melvin solution in string theory,''
384: Phys.\ Lett.\ B {\bf 346}, 55 (1995)
385: [arXiv:hep-th/9411198].
386: %%CITATION = HEP-TH 9411198;%%
387: }
388: 
389: \lref\wittenbh{
390: E.~Witten,
391: ``On string theory and black holes,''
392: Phys.\ Rev.\ D {\bf 44}, 314 (1991).
393: %%CITATION = PHRVA,D44,314;%%
394: }
395: 
396: \lref\tse{
397: J.~G.~Russo and A.~A.~Tseytlin,
398: ``Magnetic flux tube models in superstring theory,''
399: Nucl.\ Phys.\ B {\bf 461}, 131 (1996)
400: [arXiv:hep-th/9508068].
401: %%CITATION = HEP-TH 9508068;%%
402: }
403: 
404: 
405: \lref\piljin{
406: Y.~Michishita and P.~Yi,
407: ``D-brane probe and closed string tachyons,''
408: Phys.\ Rev.\ D {\bf 65}, 086006 (2002)
409: [arXiv:hep-th/0111199].
410: %%CITATION = HEP-TH 0111199;%%
411: }
412: 
413: 
414: \lref\MaMo{
415: E.~J.~Martinec and G.~Moore,
416: ``On decay of K-theory,''
417: arXiv:hep-th/0212059.
418: %%CITATION = HEP-TH 0212059;%%
419: }
420: 
421: \lref\Goa{
422: S.~Govindarajan and T.~Jayaraman,
423: ``On the Landau-Ginzburg description of boundary CFTs and special 
424: Lagrangian submanifolds,''
425: JHEP {\bf 0007}, 016 (2000)
426: [arXiv:hep-th/0003242].
427: %%CITATION = HEP-TH 0003242;%%
428: }
429: 
430: 
431: \lref\Gob{
432: S.~Govindarajan, T.~Jayaraman and T.~Sarkar,
433: ``On D-branes from gauged linear sigma models,''
434: Nucl.\ Phys.\ B {\bf 593}, 155 (2001)
435: [arXiv:hep-th/0007075].
436: %%CITATION = HEP-TH 0007075;%%
437: }
438: 
439: \lref\Goc{
440: S.~Govindarajan and T.~Jayaraman,
441: ``Boundary fermions, coherent sheaves and D-branes on Calabi-Yau  
442: manifolds,''
443: Nucl.\ Phys.\ B {\bf 618}, 50 (2001)
444: [arXiv:hep-th/0104126].
445: %%CITATION = HEP-TH 0104126;%%
446: }
447: 
448: 
449: \lref\HeMc{
450: S.~Hellerman and J.~McGreevy,
451: ``Linear sigma model toolshed for D-brane physics,''
452: JHEP {¥bf 0110}, 002 (2001)
453: [arXiv:hep-th/0104100].
454: %%CITATION = HEP-TH 0104100;%%
455: }
456: 
457: \lref\SuYa{
458: K.~Sugiyama and S.~Yamaguchi,
459: ``D-branes on a noncompact singular Calabi-Yau manifold,''
460: JHEP {\bf 0102}, 015 (2001)
461: [arXiv:hep-th/0011091].
462: %%CITATION = HEP-TH 0011091;%%
463: }
464: 
465: 
466: \lref\DoFi{
467: M.~R.~Douglas and B.~Fiol,
468: ``D-branes and discrete torsion. II,''
469: arXiv:hep-th/9903031.
470: %%CITATION = HEP-TH 9903031;%%
471: }
472: 
473: 
474: \lref\Vaq{
475: C.~Vafa,
476: ``Quantum Symmetries Of String Vacua,''
477: Mod.\ Phys.\ Lett.\ A {\bf 4}, 1615 (1989).
478: %%CITATION = MPLAE,A4,1615;%%
479: }
480: 
481: 
482: \lref\Bi{
483: M.~Billo, B.~Craps and F.~Roose,
484: ``Orbifold boundary states from Cardy's condition,''
485: JHEP {\bf 0101}, 038 (2001)
486: [arXiv:hep-th/0011060].
487: %%CITATION = HEP-TH 0011060;%%
488: }
489: 
490: 
491: \lref\CallanBC{
492: C.~G.~Callan, C.~Lovelace, C.~R.~Nappi and S.~A.~Yost,
493: ``String Loop Corrections To Beta Functions,''
494: Nucl.\ Phys.\ B {\bf 288}, 525 (1987).
495: %%CITATION = NUPHA,B288,525;%%
496: }
497: 
498: \lref\SenSM{
499: A.~Sen,
500: ``Tachyon condensation on the brane antibrane system,''
501: JHEP {\bf 9808}, 012 (1998)
502: [arXiv:hep-th/9805170].
503: %%CITATION = HEP-TH 9805170;%%
504: }
505: 
506: \lref\Martinec{
507: E.~J.~Martinec,
508: ``Defects, decay, and dissipated states,''
509: arXiv:hep-th/0210231.
510: %%CITATION = HEP-TH 0210231;%%
511: }
512: 
513: \lref\YHE{
514: Y.~H.~He,
515: ``Closed string tachyons, non-supersymmetric orbifolds and 
516: generalised  McKay correspondence,''
517: arXiv:hep-th/0301162.
518: %%CITATION = HEP-TH 0301162;%%
519: }
520: 
521: 
522: 
523: 
524: 
525: \baselineskip 18pt plus 2pt minus 2pt
526: 
527: \Title{\vbox{\baselineskip12pt \hbox{hep-th/0307248}\hbox{HUTP-03/A046}
528:   }}
529: {\vbox{\centerline{Evolution of D-branes} 
530: \centerline{Under Closed String Tachyon 
531: Condensation}}}
532: \centerline{Shiraz Minwalla and Tadashi Takayanagi }
533: 
534: 
535: \medskip\centerline{ Jefferson Physical Laboratory}
536: \centerline{Harvard University}
537: \centerline{Cambridge, MA 02138}
538: 
539: \vskip .1in \centerline{\bf Abstract}
540: 
541: We study the evolution of stable D-branes of ${\bf C/Z}_n$ and 
542: twisted circle
543: theories in the process of closed string tachyon condensation. 
544: We interpret 
545: the fractional branes in these backgrounds as type II branes wrapping 
546: (`blown up') cycles, and trace the evolution of
547: the corresponding cycles under tachyon condensation. We also
548: study RG flows of the corresponding ${\cal N}=2$ boundary conformal field
549: theories. We find flows along which fractional D-branes either 
550: disappear or evolve into other fractional D-branes, and other flows along
551: which bulk branes either disappear or evolve into stable branes. 
552: 
553: 
554: 
555: \noblackbox
556: 
557: \Date{July 2003}
558: 
559: \listtoc
560: \writetoc
561: 
562: 
563: \newsec{Introduction}
564: 
565: Consider a classical scalar field theory with action 
566: \eqn\actwc{S= \int d^dx {1\over 2 g^2} \left[(\p_\mu \ph)^2 +V(\ph)\right]}
567: where $V(\ph)$ takes the form shown in Fig. 1. A glance at this action 
568: carries a lot of information. For instance, it is immediately clear that
569: this system has at least two classically stable static 
570: backgrounds ($\ph= a$ and
571: $\ph=c$) and two unstable backgrounds ($\ph=0$ and $\ph=b$). The
572: spectrum around the unstable solution $\ph=0$ includes a tachyon; the end
573: point of tachyon condensation process is a gas of radiation about 
574: the stable vacuum $\ph=a$. 
575: \fig{Configuration space of scalar field theory with potential.}
576: {potential.eps}{3.0truein}
577: 
578: It is instructive to contrast our understanding of the model \actwc\ with
579: our current understanding of  string theory. We know a great deal 
580: about the properties of several particular classical solutions of string 
581: theory but lack intuition on the global structure of the configuration
582: space. It is not even clear whether all solutions (e.g. the Bosonic String)
583: are part of the same configuration space.
584: 
585: The recent study of closed string tachyon condensation has begun to address 
586: this issue. For example, Adams, Polchinski and Silverstein (APS) \AdPoSi\ 
587: have argued that unstable 
588: ${\bf C/ Z}_n$ 
589: orbifolds of type II theory decay to type II theory in flat space, a
590: process roughly analogous to the decay of $\ph=0$ to 
591: $\ph=a$ in the model \actwc\ described above (see also \DabholkarIF 
592: \Va \chicago \Martinec). 
593: Similarly, generalizations of this study indicate that twisted circles or 
594: Melvin compactifications of type II theory \tse\ 
595: decay to supersymmetric 
596: circle compactifications of flat space \DaGuHeMi\ 
597: (for recent discussions of closed string tachyon condensation
598: in the twisted circle see also e.g. \GuSt \RussoTF \SuyamaGD \Gu \Da). 
599: These decay processes access otherwise unexplored regions of the 
600: configuration space of string theory, and deserve to be studied in greater 
601: detail. D-branes have often proved to be useful probes in string theory and
602: the unstable backgrounds we study host a rich spectrum of stable D-branes.
603: In this paper we will study how these stable branes evolve in the process
604: of closed string tachyon condensation.\foot{See \MaMo\ for a discussion 
605: of similar issues in the context of nonsupersymmetric 
606: ${\bf C}^2 /{\bf Z}_n$.
607: orbifolds.}
608: 
609: Our study will focus largely on twisted circle theories. These rather 
610: versatile models appear in a two parameter family labeled by the 
611: size of the 
612: twisted circle $R$ and the angular twist $2\pi (1-{1\over n})$ in 
613: an auxiliary 2-plane,  upon traversing 
614: this circle . As we review in 
615: section 2 below, twisted circles reduce to ${\bf C/Z}_n$ orbifolds 
616: in the limit\foot{Intriguingly 
617: they also reduce to orbifolds of type 0 theory
618: in the limit $n \to \infty$; however the smoothness of this limit is open
619: to debate.} $R \to 0$ \TaUeo, and so may be thought of as a
620: desingularization or `blow up' of ${\bf C / Z}_n$ (in particular 
621: the ${\bf C/Z}_n$ fractional 0-brane may be `blown up' into a wrapped 
622: twisted
623: circle 1-brane). Consequently, twisted circle dynamics reduces to 
624:  ${\bf C /Z}_n$ dynamics upon taking the appropriate limit. 
625: 
626: We study the evolution of D-branes under tachyon condensation using two 
627: complementary approaches. Our first approach uses the remarkable fact that 
628: the Euclidean (more precisely Liouville) evolution of twisted 
629: circle models is approximately captured by an easily obtained and completely 
630: smooth interpolating geometry, a space that is similar in some respects 
631: to Taub Nut space. In order to take advantage of this fact, we
632: interpret the D-branes on twisted circle backgrounds as usual type II 
633: branes wrapping various geometric cycles of the twisted circle background. 
634: We then follow the evolution of these cycles to late Euclidean `time' in the
635: interpolating RG-flow geometry. We find that the cycles wrapped by twisted 
636: circle D-branes always evolve, under RG flow in one of three distinct ways.
637: \item{1.} In some cases the relevant cycle remains nonsingular through the 
638: process of RG flow and smoothly evolves into a nontrivial cycle in the IR 
639: geometry. This allows us to unambiguously
640: trace the evolution of the corresponding D-brane under tachyon condensation.
641: For example we conclude that  the twisted circle 3-brane evolves into the 
642: usual type II 3-brane (a similar result is true of the 2-brane of 
643: ${\bf C/ Z}_n$).
644: \item{2.} In other cases, the relevant cycle shrinks to zero size at a 
645: particular Euclidean time, and remains trivial through the  subsequent 
646: RG flow. This happens only at the 
647: fixed point (or the origin) of the orbifold.
648: In these cases (which 
649: includes the twisted circle fractional 1-brane and so the ${\bf C/ Z}_n$ 
650: fractional 0-brane) we conjecture the existence of flows along which the 
651: corresponding D-brane simply disappears at some finite time during RG flow. 
652: \item{3.} Most interestingly, in some cases the relevant cycle degenerates
653: at exactly one point in the full RG flow geometry (intuitively, this 
654: cycle is non vanishing both before and after the singular point $r=0$).
655: In such cases we conjecture the existence of two qualitatively distinct 
656: RG-flows (precisely which one is implemented depends on the details of the 
657: initial conditions). Along the first kind of flow the corresponding D-brane
658: disappears when its cycle degenerates. Along the second kind of flow
659: the D-brane evolves, through the singularity, and eventually wraps the 
660: corresponding smooth cycle in the supersymmetric space. For example, we 
661: conjecture that the twisted circle bulk 1-brane (resp the ${\bf C/Z}_n$ 
662: bulk 0-brane) can either disappear near the origin, 
663: or evolve into a 1-brane wrapping the 
664: supersymmetric circle (resp bulk 0-brane on ${\bf C}$) under the process of 
665: tachyon condensation. 
666: 
667: {~~~}
668: 
669: Geometrical considerations also lead us to conjecture 
670: the existence of flows in which certain bulk branes (e.g. a D1-brane
671: wrapped on a compactified circle in flat space) are created out of nothing, 
672: in the process of tachyon 
673: condensation. Finally the flows we have 
674: described above may be superposed to produce more complicated flows (e.g. 
675: one in which a fractional brane is disappears, while at the same time 
676: a bulk brane is created out of nothing).
677: 
678: We are able to confirm some of these geometrically inspired 
679: conjectures using a more rigorous approach. Recall that 
680: \Va\ and \DaGuHeMi\ were able to obtain exact results on closed string
681: tachyon condensation 
682: by studying ${\cal N}=2$ supersymmetric renormalization group flows in 
683: Gauged Linear 
684: Sigma Models (GLSMs). Introducing D-branes into the story 
685: corresponds to putting the GLSMs on the disk with appropriate boundary 
686: conditions and adding relevant boundary degrees of freedom 
687: (`Chan Paton factors') \HoIqVa \Gob \Ho. When the boundary 
688: conditions and interactions in question also preserve ${\cal N}=2$ 
689: supersymmetry, it is not difficult to construct RG flows that follow the 
690: evolution of these boundary conditions to late Euclidean times. 
691: Using this method, we are able 
692: to construct examples of all the flows described in items 1-3 above, 
693: including both kinds of flows described in item 3. As one check on the formal 
694: reasoning in this part of the paper, we present (following 
695: \AdPoSi\ and \GuHeMiSc\ in the absence of boundaries) an explicit solution 
696: to the 1-loop beta function equations that describes the ${\bf C/Z}_n$ 2-brane 
697: relaxing to the supersymmetric 2-brane in flat space.
698: 
699: As we have noted, D-branes disappear along some of the flows 
700: described above. Interestingly this disappearance occurs, roughly, via a 
701: process of  open-string tachyon condensation (even though the branes under 
702: study were stable in the UV). This open string instability is  triggered 
703: (at some finite Euclidean time) by the closed string tachyon condensation 
704: process (see sections 5 and 7 for more details).  
705: 
706: We have, so far, described the evolution of D-branes under generic bulk 
707: RG flows, along which twisted circles and ${\bf C/ Z}_n$ decay to flat space. 
708: As is well known, world sheet $N=2$ supersymmetry guarantees the 
709: existence of 
710: more fine tuned RG flows in both these models; for instance flows from  
711: ${\bf C}/{\bf Z}_n$ to ${\bf C}/{\bf Z}_m\ \ (n>m)$. It is not difficult 
712: to repeat the analysis described above on these fine tuned flows. As above 
713: we find the geometrically motivated conjectures as to the fate of these 
714: D-branes are always borne out by a more rigorous analysis of these 
715: flows. In 
716: particular, the flows describing the evolution of a single fractional 
717: D-branes 
718: are always of type described in item 2 above. Some multi-fractional branes
719: fall into the category of item 3; such branes either disappear or 
720: evolve into specific other fractional branes under closed string tachyon 
721: condensation. The story with bulk branes is unchanged. The analysis 
722: of these special flows is particularly useful as it permits a better 
723: understand of how geometrical information is encoded in corresponding 
724: boundary interactions due to open string tachyon fields.
725: 
726: 
727: 
728: This paper is organized as follows. In section 2 we review the closed
729: string backgrounds of interest to this paper, namely  twisted circles, 
730: ${\bf C}/{\bf Z}_n$  orbifolds and type 0 theory. In section 3 we review 
731: the tachyon condensation of twisted circle models \DaGuHeMi\ and  
732: derive a smooth interpolating Euclidean 4-geometry that approximately
733: captures this tachyon condensation, and study it in detail. 
734: In section 4 we review the D-branes in twisted circle backgrounds, and 
735: interpret them as usual D-branes wrapping nontrivial cycles in the twisted
736: circle background. We also demonstrate that these branes reduce, in the 
737: appropriate limits, to the D-branes of ${\bf C/Z}_n$ and type 0 theory. 
738: In section 5 we study the evolution of stable D-branes under tachyon 
739: condensation in the models introduced above from two different viewpoints. 
740: In section 6 we consider evolution of D-branes under closed
741: tachyon condensation via more general 
742: decay channels of the twisted circles and orbifolds.
743: In Appendix A we review the so called `interpolating 
744: orbifold' (which interpolates  between type II and type 0 string theory).
745: In Appendix B we give a detailed review of closed string tachyon 
746: condensation in ${\bf C/Z}_n$ and its description by gauged 
747: linear sigma model with some new results.
748: In Appendix C we compute the classical sigma model metric of gauged 
749: linear sigma model for the twisted circle. In Appendix D we
750: discuss D-branes in the Landau-Ginzburg (LG) models 
751: which are dual to our models.
752: In Appendix E we present 
753: a useful list of T-duality relations of twisted circle.
754: 
755: \newsec{Twisted Circles, ${\bf C}/{\bf Z}_n$ and Type 0 theory}
756: 
757: As we have explained in the introduction, in this paper we will study 
758: the evolution of stable D-branes in the decay of particular 
759: unstable closed string 
760: backgrounds. In this section we introduce these unstable backgrounds; 
761: Twisted Circles, ${\bf C}/{\bf Z}_n$ and type 0 theory. We will 
762: emphasize that two of these backgrounds, namely 
763: ${\bf C}/{\bf Z}_n$ and type 0 may be realized as specific 
764: limits of the Twisted Circle theory.
765: 
766: 
767: \subsec{Twisted Circles}
768: 
769: Consider type II theory in flat ten-dimensional space
770: $(\vec x\in {\bf R}^{6+1},y\in {\bf R},z\in {\bf C})$, 
771: quotiented by a circle identification
772: accompanied by a rotation in an auxiliary plane
773: \eqn\twisid{
774: (\vec x,y,z) \sim (\vec x, y+2\pi R,e^{2\pi i\zeta}z).
775: }
776: We will refer to such a space (or the non-trivial three-dimensional part of
777: it) as a ``twisted circle'' with radius $R$. 
778: Note that the particular case $\zeta=0$ corresponds to the usual  
779: supersymmetric circle compactification, while $\zeta=1$ is a circle 
780: compactification accompanied by a $2 \pi$ rotation in a transverse plane,
781: i.e. a Scherk-Schwarz compactification.
782: 
783: Twisted sectors of \twisid\ are strings that wind around the $y$ direction
784: and are simultaneously constrained to rotate in an arc on the $z$ plane. 
785: As the squared length of such strings grows like $r^2$, generic
786: twisted sector states are localized to the origin of the $z$ plane. The
787: exact spectrum of type II theory on twisted circles is not difficult to 
788: derive (\refs{\tse,\TseytlinEI,\TseytlinZV}; see \refs{\TaUeo, \DaGuHeMi} 
789: for a brief review.) and confirms these qualitative expectations. 
790: 
791: For technical reasons we will restrict ourselves in this paper to the 
792: study of \twisid\ with a discrete set of twists; $\zeta={(n+1)/n}$ with 
793: $n$ odd i.e.
794: \eqn\twisidn{
795: (\vec x,y,z) \sim (\vec x, y+2\pi R,e^{2\pi i{n+1 \over n}}z).
796: }
797: Note that $2 \pi n \zeta=2 \pi(n+1) \approx 0$. Consequently 
798: it is best to think of \twisidn\ as a
799: ${\bf Z}_n$ orbifold $({\bf C}\times {\bf S}^1)/{\bf Z}_n$
800: of a supersymmetric circle compactification 
801: \eqn\susycirc{
802: (\vec x,y,z) \sim (\vec x, y+2\pi n R, z),
803: }
804: of type II theory. States in the twisted sector of the ${\bf Z}_n$ 
805: orbifold are 
806: all localized on ${\bf C}$ while states in the untwisted sector 
807: are the usual
808: winding and Kaluza Klein modes on \susycirc\ and are delocalized. 
809: 
810: At large $R$ all twisted sector states are very massive and 
811: so  \twisidn\ is classically stable. On the other hand, 
812: it turns out that \twisidn\ is unstable at small $R$; more precisely, the 
813: spectrum includes at least one tachyon if and only if 
814: $2\apm (\f{n-1}{n})>R^2$; the 
815: mass of this state is $M^2=(R/\alpha')^2-2\alpha'(\f{n-1}{n})$. 
816: In this paper, 
817: we will be interested in \twisidn\ in its unstable regime. 
818: 
819: \subsec{${\bf C}/{\bf Z}_n$ from $R \to 0$}
820: 
821: In the limit 
822: $R \to 0$ \twisidn\ reduces to a simpler ${\bf Z}_n$ orbifold 
823: of \susycirc\ \TaUeo.
824: \eqn\twisidno{
825: (\vec x,y,z) \sim (\vec x, y, e^{2\pi i{n+1 \over n}}z);
826: }
827: i.e. the usual ${\bf C}/{\bf Z}_n$ orbifold in a plane transverse 
828: to the circle compactification. Note that the winding number of strings 
829: around the twisted circle in \twisidn\ is conserved; this conservation 
830: law modulo $n$ reduces to the `quantum symmetry' of 
831: the ${\bf Z}_n$ orbifold 
832: \twisidno. (Recall that the ${\bf Z}_n$ quantum symmetry of a ${\bf Z}_n$ 
833: orbifold \Vaq\ is generated by an element $h$ defined by the action 
834: \eqn\qs{h:|k\lb\to e^{\f{2\pi ik}{n}}|k\lb,} where 
835: $|k\lb$ is any state in the $k^{th}$ twisted sector).
836: 
837: 
838: In summary, the $R \to 0$, the IIA/B twisted circle model \twisidn\ 
839: reduces to IIA/B theory on ${\bf R}^7 \times {\bf C}/{\bf Z}_n$ times a
840: zero size circle, or (by T-duality) to IIB/A theory on ${\bf R}^8 \times 
841: {\bf C}/{\bf Z}_n$ \TaUeo.
842: 
843: \subsec{Type 0 from $n \to \infty$}
844: 
845: In the limit $n \to \infty$, IIA theory on \twisidn\ reduces to IIA theory
846: on a Scherk Schwarz circle, i.e. IIA theory modded out by 
847: $(-1)^{F_s} \times \sigma$ where $\sigma$ is the translation $y \to y+2 \pi
848: R$ and $F_s$ is the space time fermion number. This Scherk Schwarz
849: compactification of type II theory has been referred to as an 
850: `interpolating orbifold' \CoGu, as it interpolates between IIA theory on 
851: ${\bf R}^{10}$ (as $R \to \infty$) and 0B theory on ${\bf R}^{10}$ 
852: (as $R \to 0$). (See Appendix A for more details). Consequently, one may
853: obtain type 0B/A theory from the twisted circle of IIA/B theory by first 
854: taking  $n \to \infty$ and then taking $R \to 0$ \DaGuHeMi . 
855: 
856: 
857: 
858: \newsec{Closed String Tachyon Condensation in Twisted Circles}
859: 
860: The backgrounds described in section 2 are all unstable; their 
861: spectrum includes at least one closed string tachyon. In this section, first
862: we review the tachyon condensation\foot{Recall ${\bf C}/{\bf Z}_n$ 
863: and type 0 theories may be obtained as 
864: limits of twisted circle theories.} of twisted circle theories \DaGuHeMi . 
865: After that we derive and study the smooth 
866: `classical' Euclidean 4-geometry of the corresponding tachyon condensation 
867: process in detail. The intuition we gain from this study will
868: prove useful to us in section 5 below. 
869: 
870: \subsec{Decay of Twisted Circles in Gauged Linear Sigma Model Description}
871: 
872: In this subsection we will review a construction of a renormalization 
873: group flow that describes the tachyon condensation of twisted circles 
874: \DaGuHeMi. The construction of this renormalization group flow uses 
875: ${\cal N}=(2,2)$
876: gauged linear sigma models \WittenYC \HoKa\
877: (see \DaGuHeMi\ and references therein for 
878: all details) and is similar to Vafa's construction \Va\ of a 
879: flow for ${\bf C/Z}_n$ (reviewed in detail in Appendix B). 
880: 
881: 
882: Consider a gauged linear sigma model that contains two charged 
883: superfields $\Ph_1$ and $\Ph_{-n}$ that transform under $U(1)$ gauge 
884: transformations as 
885: \eqn\ft{
886: \Ph_1 \to e^{i\a} \Ph_1,\qquad
887: \Ph_{-n} \r e^{-in \a} \Ph_{-n},
888: }
889: together with an axionic field $P=P_1+iP_2$ 
890: whose gauge transformation (see \HoKa\ for general arguments) is
891: \eqn\ft{
892: P \to P + i \alpha.}
893: Note that 
894: $P_1$, the real part of $P$, effectively plays the role of a dynamical FI 
895: term. As in B.2, the low energy dynamics is described by a quantum corrected 
896: sigma model on the supersymmetric manifold of zero-energy configurations 
897: satisfying (D-term condition)
898: \eqn\zet{-\f{D}{e^2}=
899: |\ph_1|^2-n|\ph_{-n}|^2+k P_1=0,
900: }
901: modulo gauge equivalences, where $k$ is the coefficient of the kinetic 
902: term of the $P$ field (see \DaGuHeMi ) \foot{ More precisely, we study the 
903: gauged linear sigma model modded out by a discrete non anomalous
904: R symmetry (see B.3). This modding out imposes the type II GSO 
905: projection on the low energy sigma model.}.
906: 
907: At large values of $P_1$ (UV region), we use the
908: gauge freedom to set $\ph_{-n}$ to be real and positive, and then use
909: \zet\ to solve for $\ph_{-n}$ as $\ph_{-n} = \sqrt{(|\ph_1|^2 +k P_1)/n}$.
910: Plugging this into the classical action (see Appendix C for details)
911: we obtain the flat sigma model on the complex plane ${\bf C}$ 
912: (parameterized by $\ph_1$) times a
913: cylinder (parameterized by $P$) orbifolded by the unfixed ${\bf Z}_n$ 
914: gauge symmetry
915: \eqn\unfixexgt{
916: \ph_1 \r e^{{2 \pi i/ n}} \ph_1, ~~~P_2 \r P_2+{2 \pi i \over n}.
917: }
918: Consequently, at large $P_1$, dynamics reduces to that of 
919: a sigma model on the
920: twisted circle theory $({\bf C}\times {\bf S}^1)/{\bf Z}_n$
921: with a circle of radius 
922: $R_{UV}={\sqrt{\al k} \over n}$. Its metric is given by
923: \eqn\twim{ds^2= (d\rho)^2+\rho^2(d\theta)^2
924: +\f{k}{2}(dP_1)^2+\f{k}{2}(dP_2)^2,} where $\theta$ is the angle
925: of\foot{In Appendix C we denote this angle by $\theta_1$ just for notational
926: convenience. Both are the same.} 
927: $\ph_1$ and we defined $\rho=|\ph_1|$.
928: 
929: In \DaGuHeMi\ it is argued that $P_1$ may be regarded as a 
930: Liouville or scale 
931: direction, with $P_1 \r \infty$ corresponding to the UV.  
932: Consequently, the evolution of our target space in the $P_1$ 
933: mimics the flow of the energy scale from UV to IR as the change of $P_1$ 
934: from $P_1=\infty$ to $P_1=-\infty$. In this sense the coordinate 
935: $-P_1$ can be regarded as 
936: the real time $t$ after the Wick rotation. Consequently, the 
937: end point of tachyon condensation in the twisted circle theory is 
938: captured by the behavior of the sigma model as $ P_1 \r -\infty$. In this 
939: limit we obtain the cylindrical flat space ${\bf C}\times {\bf S^1}$ 
940: with radius $R_{IR}=\sqrt{\al k}$
941: \eqn\flatm{ ds^2= d\rho'^2+n^2\rho'^2 (d\theta)^2
942: +\f{k}{2}(dP_2-d\theta)^2,}
943: where we defined $\rho'=|\ph_{-n}|$.
944: Thus we conclude that the twisted circle
945: theory decays into a supersymmetric circle compactification of flat space.
946: 
947: Now consider the limit $k\to 0$; the twisted circle reduces to the orbifold 
948: ${\bf C}/{\bf Z}_{n}\times {\bf S^1}$, where the radius $R'$ of this 
949: ${\bf S}^1$
950: is $n$ times the radius $R$ of the twisted circle background \TaUeo. 
951: According to 
952: \flatm, the end point of this decay is ${\bf C}\times {\bf S^1}$, where 
953: the radius of ${\bf S}^1$ is given by $n R = R'$. Consequently, 
954: the decay of 
955: ${\bf C}/{\bf Z}_n \times {\bf S}^1$ 
956: replaces the orbifold by flat space, and leaves 
957: the ${\bf S}^1$ 
958: untouched, consistent with the conclusions of \AdPoSi\ \Va 
959: (for completeness we review tachyon condensation
960: in ${\bf C/ Z}_n$ in detail in Appendix B).
961: 
962: In the limit $n \to \infty$ of the twisted circle reduces to a 
963: ${\bf Z}_2$ orbifold of type 0 theory on a circle of radius ${\apm \over R}$
964: (see previous section). If the description of tachyon condensation 
965: commutes with $n \to \infty$ (and it is not clear that it does), 
966: it follows that the end point of tachyon condensation of the ${\bf Z}_2$ 
967: orbifold of type 0A/B theory on an ${\bf S}^1$ (reviewed in subsection 2.3) 
968: is simply type IIB/A theory on ${\bf R}^{10}$.
969: 
970: \subsec{A Construction of the Interpolating Liouville Flow}
971: 
972: As we have reviewed in the previous subsection, the entire process of
973: tachyon condensation of the twisted circle theory is described by a
974: conformal field theory. The CFT that captures the details of this tachyon 
975: condensation process may be constructed via a two step procedure. In the 
976: first step we plug the D-term equation \zet\ into the GLSM action, 
977: fix the $U(1)$ gauge and then integrate out the (massive) gauge field which 
978: appears quadratically in the Lagrangian. These manipulations are easy to 
979: perform and lead, in the sigma model limit 
980: $e \to \infty$ to a ${\cal N}=(2,2)$ sigma model
981: on a smooth four dimensional target space\foot{
982: Here we again set the angle of the complex scalar field 
983: $\ph_{-n}$ to zero by gauge fixing. We can also employ
984: a gauge invariant expression without gauge fixing, which 
985: looks more complicated. Since the result does not change,
986:  we will use the former expression. We thank K.Hori very much 
987: for comments on this point.} 
988:  (see Appendix C for details)
989: \eqn\metrico{ds^2=  d \rho^2+ d \rho'^2
990: +\f{2}{k}(n\rho' d \rho' -\rho d\rho)^2 + 
991:  {{k \over 2} \rho^{2}d\tilde{\theta}_1^2
992:  +{k \over 2}\rho'^2d\tilde{\theta}_2^2
993:  +\rho^2\rho'^2 
994:  (nd\tilde{\theta}_1+d \tilde{\theta}_2)^2
995:  \over 
996: \rho^2 + n^2 \rho^{'2}+{k \over 2}  }.}
997: The coordinates that appear in \metrico\ are related to those in \twim\ and 
998: \flatm\ by $\tilde{\theta}_1\equiv \theta-P_2$ and 
999: $\tilde{\theta}_2=nP_2$. \unfixexgt\ implies that $\tilde{\theta}_{1,2}$ 
1000: are periodic with periodicity $2\pi$, and that the radial parameters $\rho$ 
1001: and $\rho'$ run from $0$ to $\infty$. 
1002: 
1003: In the next (and unsatisfyingly implicit) step of this construction we 
1004: flow to the IR of this sigma model to obtain the desired CFT. 
1005: Unfortunately, we have not been able to find an exact description of the 
1006: resulting ${\cal N}=(2,2)$ SCFT. However, in this paper we will be interested 
1007: only in qualitative aspects of of \metrico.
1008: We will assume that, as in \HoKa, (and several earlier analysis of Calabi 
1009: Yau constructions, see for instance \WittenYC\ ), the asymptotic geometry 
1010: and topology contained in the exact SCFT match those of \metrico. In the
1011: rest of this section we will study the geometry of \metrico\ 
1012: (we refer to \metrico\ as the `classical' geometry of tachyon condensation).
1013: 
1014: \subsec{The Classical Geometry of Tachyon Condensation}
1015: 
1016: \fig{Geometry change under the closed 
1017: string tachyon condensation. The bubble of flat space appears at the origin 
1018: when $P_1=0$. Later it will expand and finally it covers the whole space.}
1019: {phase.eps}{3.0truein}
1020: 
1021: As we have argued above, \metrico\ describes a space that interpolates 
1022: between the twisted circle geometry and flat space. It is easy to see 
1023: this in detail. Rewriting the metric in terms of the coordinates 
1024: $\rho$, $\theta$, $P_1$ and $P_2$; see around \ft - \flatm\ for notation)
1025: we find that, in the UV limit ($P_1 \gg 0$) the interpolating space reduces 
1026: to the
1027: twisted circle \twim
1028: \eqn\metricone{\eqalign{
1029: ds^2&=(d\rho)^2+\f{k}{2}(dP_1)^2+\f{(\rho d\rho
1030: +\f{k}{2}dP_1)^2}{n(\rho^2+kP_1)}\cr
1031: &+\!\f{(n\rho^4+(nkP_1\!+\!\f{k}{2})\rho^2)(d\theta)^2
1032: \!-\!k\rho^2(d\theta)(dP_2)\!+\!\f{k}{2}((n+1)\rho^2\!+\!nkP_1)(dP_2)^2}
1033: {(n+1)\rho^2+nkP_1+\f{k}{2}}.\cr
1034: & \approx (d\rho)^2+\rho^2(d\theta)^2
1035: +\f{k}{2}(dP_1)^2+\f{k}{2}(dP_2)^2 ~~~~(P_1 \gg 0).\cr}}
1036: On the other hand, rewriting 
1037: the metric in terms of the coordinates $\rho'$, $\theta$, $P_1$ and 
1038: $\tilde{\theta_1}$, we see that the interpolating space reduces to flat
1039: space \flatm\ (i.e. supersymmetric circle) in the IR ($P_1 \ll 0$)
1040: \eqn\metrictwo{\eqalign{
1041: &ds^2=(d\rho')^2+\f{k}{2}(dP_1)^2+\f{(n\rho' d\rho'
1042: +\f{k}{2}dP_1)^2}{n\rho'^2+kP_1}\cr
1043: &+\!\f{\f{k}{2}(n\rho'^2\!+\!kP_1)(d\tilde{\theta}_1)^2
1044: +n^2\rho'^2(n\rho'^2\!+\!kP_1)d\theta^2+\f{kn^2}{2}\rho'^2
1045: (d\tilde{\theta}_1-d\theta)^2}
1046: {n(n+1)\rho'^2+kP_1+\f{k}{2}}\cr
1047: &\approx d\rho'^2+n^2\rho'^2 (d\theta)^2
1048: +\f{k}{2}(d\tilde{\theta}_1)^2
1049: .\cr}}
1050: 
1051: 
1052: In greater detail, \metrico\ represents the decay of the twisted circle 
1053: background via the  nucleation and growth of a bubble of supersymmetric 
1054: $S^1$ compactification. This bubble is nucleated at 
1055: `time' $P_1= 0$ (and at $\rho=\rho'=0$).  The radius $R_b$ of this 
1056: nucleated bubble (see Fig.2) grows in time 
1057: like $R_b\sim \sqrt{-kP_1}$ (this 
1058: diffusive growth\foot{
1059: In order to see this note that, in the far region $\rho'^2>> k|P_1|$ 
1060: \metrico\ reduces to the twisted circle, while in the opposite limit 
1061: $\rho'^2<< k|P_1|$ we get the flat space \flatm. $\rho'^2 \approx k|P_1|$
1062: (see Fig.2) 
1063: represents a transition between these regions. } 
1064: matches the behavior observed in \AdPoSi, \GuHeMiSc\ ). These results
1065:  precisely 
1066: show the expected behavior of the tachyon condensation 
1067: process (see Appendix
1068: B.1 for a general discussion).
1069: 
1070: In order to understand the transition between the twisted circle and
1071: supersymmetric geometry, it is useful to focus on the point at which 
1072: the bubble of supersymmetric circle is first nucleated, i.e. the 
1073: neighborhood of $P_1=\rho=\rho'=0$. \metrico\
1074: reduces to
1075: \eqn\metricflat{ds^2=d\rho^2+\rho^2 d\tilde{\theta}_1^2
1076: +d\rho'^2+\rho'^2 d\tilde{\theta}_2^2,}
1077: i.e. (smooth) polar coordinate of ${\bf R}^4$. 
1078: 
1079: It is interesting to trace the behavior of various cycles in the 
1080: twisted circle, as
1081: the space transforms itself from the twisted circle to flat space. The 
1082: twisted circle ${\bf S}^1_{A}$ defined by (see Fig.3)
1083: \eqn\defcc{{\bf S}^1_{A}:(\rho,\theta,P_2)=(0,s_1,s_1),\ \ 0\leq 
1084: s_1 \leq 2\pi/n,}
1085: which is  non contractible for $P_1 \gg 0$ (UV region), shrinks 
1086: and disappears at $P_1=0$. This is
1087: simply the angle circle $\tilde{\theta_2}$ of one of the two planes in 
1088: \metricflat. This cycle becomes trivial for $P_1<0$.  
1089: \fig{One cycles in twisted circle.}{twists.eps}{2.5truein}
1090: 
1091: It is also interesting to track the evolution of the supersymmetric circle
1092: ${\bf S}^1_{B}$ defined by
1093: \eqn\defthc{{\bf S}^1_{B}:(\rho,\theta,P_2)=(0,s_2,0),\ \ 
1094: 0\leq s_2 \leq 2\pi,}
1095: back in `time', from the IR to the UV. This cycle (see Fig.3)
1096: also vanishes at $P_1=0$; in fact it is simply the other angle 
1097: ${\tilde \theta}_1$ in \metricflat. 
1098: 
1099: In section 5 below we will be interested in the third cycle (see Fig.3)
1100: \eqn\deftc{{\bf S}^1_{C}:(\rho,\theta,P_2)=(a,b,u),\ \
1101:  0\leq u \leq 2\pi,\ a,b=constant}
1102: which winds $n$ times around the twisted circle in the UV $P_1 \gg 0$ and 
1103: winds once around the supersymmetric circle in the IR $P_1 \ll 0$. Note 
1104: that this cycle vanishes at the point $\rho=P_1=0$. 
1105: 
1106: \newsec{D Branes on Twisted Circles, ${\bf C}/{\bf Z}_n$ and Type 0}
1107: 
1108: The spectrum of D-branes on Twisted circles has been previously derived
1109: using a boundary state analysis \TaUet\ \DuMo. In this section
1110:  we will review the results, and provide a spacetime 
1111:  interpretation of these branes. 
1112: We will also describe how these branes reduce, in the appropriate limits, 
1113: to the branes of ${\bf C/Z}_n$ and type 0 theory.
1114: 
1115: \subsec{D Branes on Twisted Circles}
1116: 
1117: D branes on twisted circle backgrounds are of two sorts; those that live at
1118: a point on the twisted circle, and those that wrap around the twisted
1119: circle. 
1120: 
1121: A branes at a point on the twisted circle is best thought of
1122: as an array of branes on the covering space (see, for instance \piljin). 
1123: At low energies such branes behave much like their counterparts in type 
1124: II theory on flat space.
1125: 
1126: D branes that wrap the twisted circle are more interesting.
1127: Consider a 1-brane that wraps the twisted circle once.
1128: As in the case of twisted sector strings, such a brane is confined to the 
1129: origin of the $z$ plane. Classically, its mass is given simply by the usual 
1130: D1-brane tension times the minimum length of the twisted circle 
1131: $M={R \over g \apm}$. Of course such 1-branes actually appear in 
1132: degenerate one parameter family of solutions labeled by the Wilson line 
1133: $\theta=\int A \cdot dl \in [0, 2 \pi)$ along the brane.
1134: 
1135: As we have discussed above, the twisted circle may be regarded
1136: as a ${\bf Z}_n$ orbifold of \susycirc, a circle of radius $nR$. 
1137: The 1-brane of
1138: the previous paragraph is `S-dual' to a twisted sector string
1139: state of this orbifold; it is called a fractional brane. On the other hand
1140: a 1-brane wrapping the twisted circle $n$ times is the analogue of an
1141: untwisted sector string state, and is referred to as a bulk brane. 
1142: We will now argue that the bulk 1-brane may be thought of as a 
1143: collection of $n$ fractional 1-branes, with a distribution of Wilson
1144: lines. Refer to \DoMo \DiDoGo\ for original 
1145: literature on (fractional) D-branes 
1146: in supersymmetric orbifolds (e.g. ${\bf C}^2/{\bf Z}_n$).  
1147: 
1148: A collection of $n$ fractional 1-branes is described, at low energies, 
1149: by a $1+1$
1150: dimensional $U(N)$ gauge theory, with adjoint matter fields $Z$, $\psi$
1151: and $\phi$, where $Z$ represents the position of the brane in the $z$ plane, 
1152: $\psi$ is the corresponding fermion, and $\phi$ represents all other 
1153: massless fields 
1154: (the gauge field, fluctuations in $R^7$ plus corresponding fermions). In
1155: the vacuum describing $n$ separate 1-branes, these adjoint fields obey the 
1156: boundary conditions\foot{
1157: In the full string theory
1158: \eqn\bcf{ \chi(2 \pi R)= e^{{2 \pi i k(n+1) \over n}}C \chi(0) C^{-1}}
1159: where $k=n_z-n_{\bar z}+\half(n_{\psi}-n_{\bar \psi})$, where
1160: $n_z, n_{\bar z}, n_{\psi}, n_{\bar \psi}$ respectively represent the
1161: number of $Z$, $\bar Z$, $\psi$ and $\bar \psi$ open string oscillators
1162: excited in producing the open string field $\chi$. }
1163: \eqn\bc{\eqalign{ Z(2 \pi R)=& e^{{2 \pi i (n+1) \over n}}S Z(0) S^{-1}, \cr
1164:              \psi(2 \pi R)=& e^{{2 \pi i (n+1) \over 2n }}S 
1165:              \psi(0) S^{-1},\cr
1166:               \phi(2 \pi R)=& S \phi(0) S^{-1}. \cr }}
1167: where $S$ is a gluing $U(n)$ matrix representing the Wilson line on the 
1168: brane. Different values of $S$ represent different classical vacua of
1169: this system. 
1170: In particular $S=I$ represents a configuration of $n$  
1171: strings, 
1172: independently winding the twisted circle. On the other hand, when 
1173: $S$ is chosen as  the shift matrix, the $n$ 1-branes are effectively 
1174: strung together into a bulk brane; a single brane winding the circle $n$ 
1175: times. As the shift matrix has eigenvalues
1176: \eqn\eigenvalues{e^{2 \pi i k \over n},~~~ k= 1 \ldots n.}
1177: we conclude that bulk 1-brane is simply a collection of 
1178: $n$ fractional 1-branes with a relative collection of Wilson lines given by 
1179: \eigenvalues. 
1180: 
1181: Wrapped 3-branes on the twisted circle are rather similar to
1182: 1-branes. Consider a single space-filling brane in the twisted circle 
1183: background. Gauge fields on a collection of 
1184: $n$ such D3-brane, obey the boundary conditions
1185: \eqn\bco{\eqalign{ Z(y+2 \pi R, e^{{2 \pi i \over n}} z)=
1186: & e^{{2 \pi i (n+1) \over n}}C Z(y, z) C^{-1} \cr
1187:              \psi(y+2 \pi R, e^{{2 \pi i \over n}} z)=&
1188: e^{{2 \pi i (n+1) \over 2n }} C \psi(y,z) C^{-1} \cr
1189:               \phi(y+2 \pi R, e^{{2 \pi i \over n}} z)=&
1190: C \phi(y,z) C^{-1}\cr }}
1191: where the fields have the same meaning as in \bc\ and $C$ is the Wilson
1192: Line\foot{The pure gauge configuration corresponding to (for instance) 
1193: a U(1) Wilson lines is $(A_y, A_z, A_{\bar z})=({\theta \over 2 \pi R}, 0,
1194: 0)$.}.
1195: When $C$ is the shift matrix, the $n$ `fractional' 3-branes are strung 
1196: together into a single 3-brane winding the circle $n$ times, i.e. a bulk 
1197: 3-brane.  
1198: 
1199: Finally, it is easy to verify the absence of 2-branes that wrap the 
1200: twisted circle $m<n$ times, as no such brane is flat in the 
1201: covering space. It is, of course, not difficult to construct bulk
1202: 2-branes which wrap the twisted circle a multiple of $n$ times. 
1203: 
1204: 
1205: \subsec{D-Branes on ${\bf C}/{\bf Z}_n$}
1206: 
1207: We now study the D-branes described in the previous section, in
1208: the limit $R \to 0$. According to subsection 2.2, these branes must reduce 
1209: to D-branes of ${\bf C/Z}_n \times {\bf S}^1$. In this subsection we will
1210: explain how this works in detail. 
1211: 
1212: Consider a unit fractional 1-brane of the twisted circle theory, in the
1213: limit $R \to 0$. Such a brane wraps once around the twisted circle (of
1214: radius $R$), and so in some sense, wraps ${1 \over n}$ of the larger
1215: ${\bf S}^1$ factor in ${\bf C/Z}_n \times {\bf S}^1$, see subsection 2.2. 
1216: We will now argue
1217: that there are $n$ different flavors of this 1-brane. 
1218: 
1219: Consider two unit fractional 1-branes of the twisted circle theory, 
1220: with relative Wilson line $\theta$. Open string modes stretching 
1221: between these two branes are charged under the relative gauge field, and 
1222: so obey boundary conditions \bcf\ with $C \chi(0) C^{-1} = e^{i \theta} 
1223: \chi(0)$. When $\theta =0$, only open string modes with $k=0$ mod $n$ 
1224: ($k$ is defined in \bcf) survive the $R \to 0$ limit.  At generic values of 
1225: $\theta$ all modes are infinitely massive. However, in the neighborhood of 
1226: $2 \pi \theta={m \over n}$, a new set of modes become 
1227: massless, namely those 
1228: with $k=m$ mod $n$. Consequently, there are $n$ species of 1-branes that 
1229: can communicate with the $\theta=0$ brane; branes with 
1230: $\theta ={2 m \pi \over n}+ \CO(R)$ where $m$ runs between $0$ and $n-1$. 
1231: 
1232: These results are easier to visualize upon T-dualizing along 
1233: the ${\bf S}^1$ 
1234: factor of ${\bf C/Z}_n \times {\bf S}^1$. The stretched one branes of
1235: the previous paragraphs turn approximately into usual 0-branes on a circle 
1236: of size $R'={ \apm \over n R}$. Turning on a relative Wilson line $\theta$
1237: corresponds to separating the branes by $\Delta y= \theta \apm / 
1238: R={\theta n R'}$. In taking $\theta$ from 0 to $2 \pi/n$ we take 
1239: one of the 0-branes once around the circle; that process converts it into 
1240: a zero brane of another variety! Repeating this operation produces yet
1241: another zero brane, but, upon repeating the process $n$ times we revert to
1242: the initial state. 
1243: 
1244: Below we will find use for an operator $Q$ 
1245: whose action 
1246: is to transport a zero brane once around the circle (in T-dual language 
1247: $Q$ shifts the Wilson line by $e^{{i 2 \pi \over n}}$). 
1248: If the $n$ distinct
1249: 0-branes are denoted by $|i\rangle$, then clearly $Q |i 
1250: \rangle=|i+1 \rangle$.
1251: 
1252: In ${\bf C/Z}_n \times {\bf R}^8$ 
1253: (the formal result of taking $R$ to zero on the 
1254: twisted circle theory) the $n$ distinct species of 0-branes are completely 
1255: distinct; they cannot be continuously deformed into each 
1256: other; restated ${\bf C}/{\bf Z}_n$ has $n$ distinct fractional 
1257: 0-branes. However the operator $Q$ of the previous paragraph is well
1258: defined as $R \to 0$ and reduces to the generator $h$ 
1259: of the quantum symmetry \qs\
1260: of ${\bf C}/{\bf Z}_n$ in the limit\foot{This is easily seen; $Q$ generates 
1261: discrete translations, and so counts discrete moment or discrete windings
1262: in the T-dual picture.} and so we conclude that the fractional 0-branes are
1263: related to each other by the action of the quantum symmetry. It is well 
1264: known that ${\bf C}/{\bf Z}_n$ does indeed have $n$ fractional 0-branes with
1265: these properties. Of course the discussion of
1266: this section may be mimicked to derive similar results for 
1267: 2-branes of ${\bf C}/{\bf Z}_n$.
1268: 
1269: To summarize, we have found that the familiar fractional 0-branes (and
1270: 2-branes\foot{Note that a single D2-brane in ${\bf C}/{\bf Z}_n$ should be
1271: treated as a fractional D2-brane from the group theoretical viewpoint 
1272: of \DoMo \DiDoGo (belonging to one of $n$ irreducible representations of 
1273: ${\bf Z}_n$). However, its tension
1274: is not fractional but usual value. Thus
1275: we will call it just a D2-brane in this paper.}) 
1276: of ${\bf C}/{\bf Z}_n$ have a simple geometric
1277: interpretation upon `blowing up' the ${\bf C}/{\bf Z}_n$ singularity 
1278: into a twisted 
1279: circle; they correspond to 1-branes (resp 3-branes) wrapping the twisted 
1280: circle. This geometrical intuition will help us understand the fate of 
1281: fractional 0-branes during tachyon condensation, later in this paper.
1282: 
1283: 
1284: \subsec{D-Branes in Type 0 Theory}
1285: 
1286: In the limit $n \to \infty$, the IIA/B twisted circle turns into a 
1287: Scherk-Schwarz 
1288: circle of radius $R$. As we have reviewed in Appendix A, the T-dual 
1289: of the Scherk-Schwarz compactification of type IIA/B theory is 
1290: the ${\bf Z}_2$
1291: orbifold of 0B/A on $S^1$ of radius ${\apm \over R}$, where the 
1292: ${\bf Z}_2$ is generated by $(-1)^{F_L} \times \sigma'$ and  $\sigma'$ is 
1293: a shift across half the circumference of the circle. Of course, as $R \to
1294: 0$ this theory reduces locally to 0B/A in flat space. 
1295: 
1296: In the limit $n \to \infty$ the fractional 1-brane of the twisted circle is
1297: simply a D1-brane wrapping the Scherk-Schwarz circle. We will now
1298: investigate what happens to this D1-brane upon T-dualizing along this
1299: circle. 
1300: 
1301: Recall that flat space string theories with $N=(1,1)$ world sheet
1302: supersymmetry have two 0-brane boundary states with 
1303: different spin structures
1304: denoted by $|+\rangle $ and $|-\rangle$ (e.g. see \bega ). 
1305: These boundary states obey 
1306: \eqn\pmsta{\eqalign{&(-1)^{F_L} |+\rangle =-|-\rangle, \cr 
1307:                     &(-1)^{F_R} |+\rangle =-|-\rangle, \cr
1308: 		    &(-1)^{F_L} |-\rangle =-|+\rangle, \cr 
1309:                     &(-1)^{F_R} |-\rangle =-|+\rangle}}
1310: {}From \pmsta\ we conclude that 
1311: \item{1.} The only state invariant under $(-1)^{F_L}$ and $(-1)^{F_R}$ is
1312: $$|+\rangle - |-\rangle ;$$ 
1313: this is the 0-brane of type IIA theory.
1314: \item{2.} Any linear combination of 
1315: $$|+\rangle \   {\rm and} \ |-\rangle ,$$
1316: is invariant under 
1317: $(-1)^{F_L+F_R}$; these are the so called electric and  magnetic 0-branes
1318: of type 0A theory.  
1319: \item{3.} 
1320: $$|+ (x)\rangle ~~- ~~|-(x+\pi R) \rangle ,$$ (where the $(x)$ in 
1321: brackets refers to the position of the brane on the circle) is the only 
1322: combination of 0A zero branes invariant under the ${\bf Z}_2$ orbifold of 
1323: 0A on a circle of size $ 2 \pi R$, described earlier in this section.
1324: 
1325: With these facts in mind, it is natural to guess, and easy to verify, that 
1326: the T-dual of a 1-brane wrapping the thermal IIB circle is the 
1327: a configuration of electric and magnetic 0-branes described in item 3 
1328: above. 
1329: 
1330: \subsec{Dirac Quantization}
1331: 
1332: In this subsection we will study the charges of the fractional branes
1333: described in the previous subsection, and explain how they are consistent 
1334: with Dirac quantization. This subsection is a side branch away from the 
1335: main line of development of this paper. The reader eager to turn to the
1336: fate of D-branes under closed string tachyon condensation can skip to the
1337: next section.
1338: 
1339: Consider an action that describes the interaction of 
1340: 3-form and 7-form field strengths with 1-branes and 5-branes
1341: \eqn\acoftf{\eqalign{S&=C\int d^{10}x F_3^2+ e\int A_2 + g\int A_6 \cr
1342: &=C\int d^{10}x F_7^2+ e\int A_2 + g\int A_6 ,}}
1343: where $F_3=dA_2$, $F_7=d A_6$ and $F_7=*F_3$ and $A_2$ (resp $A_6$) is 
1344: integrated over the world volumes of the 1-branes (resp 5-branes).
1345: Dirac quantization imposes the constraint 
1346: \eqn\dccond{{eg \over C} = {2 n \pi},}
1347: on the charges in \acoftf. In particular, the minimally charged 1-brane
1348: (charge $e_0$) and 5-brane (charge $g_0$) obey 
1349: \eqn\mincp{{e_0 g_0 \over C}=2 \pi.}
1350: 
1351: As an exercise that will be useful later in this subsection, consider 
1352: the dimensional reduction of \acoftf\ (with minimally charged 1-branes
1353: and 5-branes) on ${\bf T}^2 \times{\bf  S}^1$ 
1354: where the torus has volume $V_2$ and the 
1355: circle has radius $R$. Let $A_1$ denote the reduction of $A_2$ along $S^1$, 
1356: let $A_4$ denote the reduction of $A_6$ along $T^2$. 
1357: The dimensionally reduced
1358: action that governing the interaction of particles (10 dimensional strings 
1359: wrapping $S^1$) and 3-branes (10 dimensional 5-branes wrapping the $T^2$) 
1360: with $A_1$ and $A_4$ is
1361: \eqn\dimredact{S=C (2 \pi R) V_2\int d^{7}x F_2^2+ e_0(2 \pi R)
1362: \int A_1 + g_0V_2\int A_4.}
1363: Note, of course, that, as a consequence of \mincp,  the charges in 
1364: \dimredact\ automatically saturate the 7-dimensional Dirac Quantization 
1365: constraint 
1366: ${(e_0 2 \pi R) (g_0V_2) \over C (2 \pi R) V_2   }=2 \pi$.
1367: 
1368: We now adapt the exercise of the previous paragraph to a compactification
1369: $({\bf T}^2  \times {\bf S}^1)/ {\bf Z}_n$ 
1370: of the twisted circle background 
1371: $({\bf C}  \times {\bf S}^1)/ {\bf Z}_n$. 
1372: For clarity consider a specific 
1373: case; let $n=3$ and let the $z$ plane in \twisidn\ be an `equilateral 
1374: 2-torus' (the complex plane modded out by translation vectors at an 
1375: angle ${2 \pi \over 3}$ to each other) of volume $V_2$. The 3-volume of
1376: this twisted circle background is $2 \pi R V_2$. The length of the smallest
1377: 1-cycle along the twisted circle is $2 \pi R$. The volume of the minimal 
1378: dual 2-cycle is $V_2$ for any nonzero $R$ and, as in the previous
1379: paragraph, the electric and magnetic charges of the dimensionally 
1380: reduced 7 dimensional theory automatically obey the Dirac quantization 
1381: condition\foot{These arguments explain the phenomenological 
1382: observation noted in \TaUet\ that 
1383: we cannot find any D-brane in the Twisted Circle which is 
1384: localized in the circle direction and 
1385: which
1386: becomes a fractional D-brane in the $R\to 0$ limit.
1387: Such a brane, if exists, will break the Dirac quantization rule.}
1388: . 
1389: 
1390: We now turn to the $R \to 0$ limit of this situation. As we have explained
1391: in subsection 2.2, the twisted circle reduces to 
1392: ${\bf T}^2 / {\bf Z}_3 \times {\bf S}^1$ where the radius of the 
1393: ${\bf S}^1$ is 
1394: $3 R$. Note that the minimal volume 2-cycle along the $T^2$ in this
1395: orbifold has volume $V_3/3$. Further, although the orbifold 
1396: circle has length 
1397: $3R$, the effective charge of the wrapped 1-brane (which reduces to the
1398: fractional 0-brane as $R \to 0$), continues to be $e_0 2 \pi R$ (that is
1399: why this object is a fractional rather than bulk brane) and the 
1400: effective 7-dimensional action governing the electromagnetic interactions
1401: between the 2-brane (i.e. `fractional' 
1402: 2-brane wrapping a cycle of volume $V_2/3$)
1403: and the fractional 0-brane is 
1404: \eqn\dimredact{S=C (2 \pi R) {V_2}\int d^{7}x F_2^2+ 
1405: {e_0 {2 \pi R }}
1406: \int A_1 + g_0{V_2 \over 3}\int A_4.}
1407: At first sight \dimredact\ appears to violate 
1408: Dirac Quantization (as ${(e_0 {2 \pi R }) (g_0{V_2 \over 3}) 
1409: \over {C (2 \pi R) V_2 }   }={2 \pi \over 3} \notin 2 \pi {\bf Z}$).  
1410: The resolution to this puzzle is rather instructive. The 2-branes 
1411: and fractional 1-branes of ${\bf T}^2 / {\bf Z}_3 \times {\bf S}^1$ 
1412: are certainly
1413: charged under untwisted RR sector potentials. However they are 
1414: equally charged under all twisted sector RR 2 form potentials 
1415: (which are massless for ${\bf T}^2 / {\bf Z}_3 \times {\bf S}^1$
1416:  but not for the 
1417: twisted circle theory at any nonzero $R$). The true phase in taking a 
1418: fractional 0-brane around the Dirac String attached to the
1419: 2-brane is the three times the naive result from \dimredact, restoring
1420: consistency. 
1421: 
1422: 
1423: \newsec{D-branes Under Tachyon Condensation}
1424: 
1425: In this section we turn to the main topic of this paper; an 
1426: investigation of the fate of D-branes under tachyon condensation. Here 
1427: we consider the evolution of D-branes only under bulk flows that 
1428: lead to a supersymmetric end point; we postpone the discussion 
1429: of fine tuned flows to the next section. We will 
1430: first state our conjectures, and then proceed, in the rest of this section,
1431: to provide evidences for them from two different viewpoints. 
1432: 
1433: {\bf Twisted circles} :  Any brane located at a 
1434: point on the twisted circle evolves into the corresponding brane at a point 
1435: on the supersymmetric circle. 3-branes that wrap the twisted circle once 
1436: unambiguously evolve into usual 3-branes that wrap the supersymmetric circle 
1437: once. All fractional 1-branes disappear in the process of tachyon 
1438: condensation. A bulk 1-brane (which wraps the twisted circle $n$ times) 
1439: either disappears (this is possible only near the origin) 
1440: or evolves into a bulk 1-brane that wraps the 
1441: supersymmetric circle once. 
1442: 
1443: 
1444: 
1445: {\bf Orbifold ${\bf C}/{\bf Z}_n$}: All $n$ varieties of space filling 
1446: 2-branes evolve into the usual space filling 2-branes in flat space. 
1447: Every fractional 0-brane disappears in the process of tachyon condensation.
1448: A bulk 0-brane either disappears 
1449: (again this is possible only near the fixed point) or 
1450: evolves into a usual 0-brane in flat space.
1451: 
1452: {\bf ${\bf Z}_2$ Orbifold of type 0A on ${\bf S}^1$}: 
1453: It should be emphasized that all results about type 0 theory are obtained 
1454: upon taking a potentially dangerous $n \to \infty$ limit. 
1455: With this caveat in mind, we find the following results. 
1456: A pair of electric and magnetic D(p+1)-branes wrapping 
1457: the circle evolve into a single Dp-brane in IIB theory. 
1458: A pair of diametrically separated electric 
1459: and magnetic 0-branes disappear in the process of tachyon condensation. 
1460: On the other hand, diametrically opposed 0A 2-branes 4-branes and 6-branes
1461: evolve respectively into single IIB 3-branes, 5-branes and 7-branes.
1462: 
1463: We will now proceed to provide evidence for these conjectures. 
1464: The rest of this section is organized as follows. In subsection 5.1 we 
1465: explain how the results for twisted circle theories follow from topological 
1466: properties of the interpolating geometry presented in the previous
1467: section. Results for ${\bf C/Z}_n$ and type 0 theory follow 
1468: from the appropriate limits (see section 2 and 3) of the twisted circle
1469: theory. In subsection 5.2 we comment on how our results are consistent with 
1470: charge conservation. In subsection 5.3 and 5.4, respectively, we provide
1471: independent evidence for our conjectures by studying the boundary condition 
1472: of world-sheet under the RG-flow in GLSMs. In appendix D we find further 
1473: confirmation of our results by an analysis of D-branes in the dual LG models.
1474: 
1475: \subsec{Geometrical Analysis}
1476: 
1477: Let us first consider the behavior of a fractional D1-brane
1478: under closed string tachyon condensation in the twisted circle theory
1479: from geometrical viewpoint. As we have explained in the previous
1480: section, a D1-brane that wraps the twisted circle once is called a 
1481: fractional D1-brane. In order to understand the fate of this 
1482: fractional brane under tachyon condensation, we can follow the
1483: evolution of this cycle as tachyon condensation proceeds.
1484: The geometry of the GLSM model was discussed in detail in section 3; 
1485: as we argued there, the twisted circle ${\bf S}^1_A$ vanishes at 
1486: $P_1=\rho=0$
1487: and cannot be extended beyond this point. Indeed, near that special
1488: point this cycle is simply an angle in a 2-plane in ${\bf R}^4$.
1489: Consequently a D1-brane wrapping the twisted cycle
1490: circle simply self annihilates at $P_1=\rho=0$. The induced metric 
1491: on the world volume of such an annihilating brane (along one possible 
1492: trajectory) is given by the everywhere smooth cigar (see Fig.4)
1493: \eqn\tcam{ds^2=\f{k}{4nP_1}(dP_1)^2+knP_1(ds_1)^2
1494: =(dr)^2+n^2r^2(ds_1)^2\ \ (r\equiv \sqrt{P_1 k/n}),} 
1495: \fig{The smooth cigar and wrapped D1-brane.}{d1cigar.eps}{3truein}
1496: 
1497: In contrast consider a bulk D1-brane. Such a brane wraps the twisted
1498: circle $n$ times at early `times', and so may be thought of as a
1499: brane wrapping the cycle ${\bf S}^1_{C}$ described in section 3
1500: above. Notice that  ${\bf S}^1_{C}$ is nontrivial in the UV ($P_1 \gg
1501: 0$) as well as in the IR. However this cycle degenerates at a point 
1502: along this RG flow, and so fall in the class described in item 3. in the 
1503: introduction. Correspondingly, we conjecture that a bulk 0-brane can 
1504: either disappear or evolve into an ordinary brane wrapping the 
1505: supersymmetric circle.\foot{
1506: Note also that ${\bf S}^1_{C}$ always has a non-zero radius except
1507: $P_1=0$. Moreover, we can slightly shift the position of ${\bf S}^1_{C}$
1508: from the origin and obtain a non-contractible one cycle. In such a situation
1509: we would expect the bulk brane to survive the process of 
1510: tachyon condensation.}
1511: 
1512: On a slightly different note, observe that the supersymmetric cycle 
1513: ${\bf S}^1_{B}$ describes the creation of a bulk zero brane 
1514: (of the supersymmetric background), correspondingly we expect to find 
1515: an RG flow that describes bulk 0-branes popping out of nothing. 
1516: 
1517: It is also possible to 
1518: extend our considerations to D3-branes. Since the
1519: world-volume for such a  brane fills all of space in the UV, it will  
1520: continue to do so in the IR; consequently the 3-brane of 
1521: the twisted circle evolves into a usual D3-brane in flat space after
1522: the closed string tachyon condensation. As we have explained in section
1523: 4.1, twisted circle D3-branes appear in a one parameter family of solutions
1524: labeled by the Wilson line around the twisted circle. It is not difficult 
1525: to trace the evolution of this Wilson line into the IR; see 
1526: subsection 5.3 below. 
1527: 
1528: 
1529: Finally we consider the behavior of D0-branes and D2-branes in the orbifold 
1530: ${\bf C}/{\bf Z}_n$. This is obtained by the small radius limit
1531: $R\to 0$ as we have explained\foot{ We believe this limit 
1532: is smooth as evidenced from e.g. the structure of GLSM \DaGuHeMi\ and the
1533: calculation of partition function \TaUeo.}
1534:  in section 3. 
1535: Thus we conclude that a fractional D0-brane will
1536: disappear after the tachyon condensation; a bulk D0-brane
1537: either disappears or evolves into a bulk D0-brane in flat space, while 
1538: a bulk D2-brane evolves into its supersymmetric counterpart (we will 
1539: explain in subsection 5.3, all $n$ ${\bf C/Z}_n$ D2-branes have the same 
1540: end point under
1541: tachyon condensation). The latter fact for a bulk D0-brane 
1542: is consistent with the results 
1543: in \AdPoSi. These conclusions will be confirmed by investigating 
1544: of the boundary interactions in the later subsections.
1545: 
1546: 
1547: \subsec{Charge Conservation}
1548: 
1549: The `disappearance' of a D-brane under tachyon condensation naively
1550: appears to be in conflict with charge conservation. Our geometrical
1551: description of this process makes it clear that, at least in the 
1552: Euclidean version of this process, this `paradox' is spurious. 
1553: 
1554: As we have described above, the world volume of the `disappearing' 
1555: fractional D1-brane is the everywhere smooth `cigar' \tcam. 
1556: This Euclidean D1-brane couples to the RR-2 form field
1557: $F_{\mu\nu}$ through the everywhere conserved current 
1558: \eqn\cur{\sqrt{G}J^{\mu\nu}=\int d^2\sigma \f{\de\xi^\mu}{\de\sigma^{\ap}}
1559: \f{\de\xi^\nu}{\de\sigma^{\beta}}\ep^{\ap\beta}\prod_{\lambda}
1560: \delta(\xi^{\lambda}(\sigma)-x^{\lambda}); }
1561: certainly charge conservation is nowhere violated. 
1562: 
1563: The disappearance of D-brane charge can be made to appear more puzzling 
1564: upon dimensionally reducing along the twisted circle. The fractional 
1565: D1-brane is charged under the gauge field obtained from the Kaluza-Klein 
1566: reduction of $F_{\mu\nu}$; this charge disappears at `time' $P_1=0$. 
1567: Of course `charge conservation is violated' only when the twisted circle 
1568: shrinks to zero size; precisely at this point the dimensionally
1569: reduced description degenerates (for instance the 9 dimensional coupling 
1570: blows up). The correct description is 10 dimensional, and
1571: manifestly charge conserving.
1572: 
1573: Of course all considerations that apply to D-strings wrapping the twisted 
1574: circle also apply to ordinary strings wrapping the twisted circle. 
1575: However, as we have explained behind, such strings belong to the
1576: twisted sector. Thus we conclude that twisted sector charges disappear
1577: in the process of tachyon condensation.\foot{In \MaMo\ another
1578: mechanism of the disappearance of the twisted RR-field
1579: was proposed in non-supersymmetric orbifold theories. It would be interesting
1580: to study the relation between this and our geometrical result.}
1581: 
1582: Recall that the reduction of the twisted circle to the
1583: orbifold is most conveniently carried out after a T-duality;
1584: consequently, it is instructive to examine the issue of charge
1585: conservation after T-dualizing. In \HoKa\ 
1586: Hori and Kapustin have demonstrated that
1587: the T-duals of gauge linear sigma models, including 
1588: 2d Black hole \wittenbh\ and our model defined by \ft,
1589: lead to super potentials that break
1590: translational invariance along the dual circle. Roughly speaking, the 
1591: potential may be thought of as a condensate (or plasma) of gravitons
1592: with positive and negative momentum, 
1593: explaining the lack of conservation of winding or
1594: twisted strings prior to T-dualizing. It seems possible that this T-dual 
1595: geometry should also be thought of as possessing a condensate of 0-branes
1596: and anti 0-branes, explaining the violation of 0-brane charge conservation.
1597: 
1598: 
1599: \subsec{GLSM analysis of ${\bf C}/{\bf Z}_n$ Boundary Flows}
1600: 
1601: In the previous subsections we have analyzed the evolution of the D-brane 
1602: spectrum in twisted circle theories (and hence ${\bf C/Z}_n$) utilizing 
1603: approximate geometry for Euclidean twisted circle tachyon condensation. 
1604: In this subsection we will check some of those results using more rigorous
1605: methods. Recall that twisted circle and ${\bf C/Z}_n$ 
1606: world sheet theories are 
1607: described by the UV of a GLSM (see section 3). In this section we will
1608: explicitly construct the GLSM boundary conditions that correspond 
1609: to certain D-branes in these backgrounds. We then analyze the evolution of 
1610: these boundary conditions under world sheet RG flow; the fate of the 
1611: D-brane in question is determined by the IR limit of these 
1612: boundary conditions.  
1613: 
1614: The analysis of this section makes crucial use of the ${\cal N}=2$ 
1615: supersymmetry in the world-sheet GLSM theory. Several authors (see 
1616: for instance \HoIqVa \Gob \Ho \HeMc \Goc \HeKaLaMc; see also \OoOzYi\ 
1617: for original arguments on boundary states) 
1618: have studied GLSMs on world sheets with boundaries, and determined
1619: conditions to ensure that the boundary conditions preserve 
1620: ${\cal N}=2$ B-type 
1621: supersymmetry, and we will utilize their results below.
1622: 
1623: 
1624: {\bf D2-branes}
1625: 
1626: As we have reviewed above, ${\bf C/Z}_n$ possesses $n$ different
1627: (fractional) 2-branes. We first consider the simplest of these 2-branes 
1628: (denoted by $|0\rangle$ in section 4 above). In the GLSM,
1629: this 2-brane corresponds to
1630: the following boundary condition \Ho\ assuming $\theta=0$
1631:  (see Appendix B for details of this GLSM)
1632: \eqn\bcdtt{{\cal D}_+ \Phi={\cal D}_- \Phi,\ \  \Sigma=\bar{\Sigma}, ~~~
1633: v_{01}=0,}
1634: where $\Phi$ denotes the chiral superfields whose lowest components 
1635: are $\phi_1$ and $\phi_n$, $\Sigma$ is the twisted chiral (vector)
1636: multiplet, and $v_{01}$ is the field strength. These boundary conditions 
1637: are not modified by RG flow\foot{From the sigma point of view 
1638: this follows as
1639: the beta function for the spacetime gauge field vanishes when evaluated on 
1640: the configuration $A_{\mu}=0$. This result also follows from the 
1641: Landau Ginzburg analysis, see Appendix D.}.
1642: 
1643: We now turn to $|m\rangle$, the remaining $n-1$ types of 
1644: D2-branes $(m\neq 0)$ in 
1645: ${\bf C/Z}_n$. Recall that $|m\rangle = h^m |0\rangle$ where $h$ generates
1646: the ${\bf Z}_n$ quantum symmetry of ${\bf C/Z}_n$. 
1647: In fact, the quantum symmetry acts
1648: very simply on the (UV of the) GLSM; it effectively shifts\foot{
1649: This fact can be easily understood in the Landau Ginzburg dual
1650: picture (see Appendix D). The shift $h:U\to U+\f{2\pi i}{n}$ of 
1651: quantum symmetry is equivalent to the shift of imaginary part of $t$ as
1652: is obvious in eq.(D.1).} 
1653:  the $\theta$ angle as
1654: $\theta \r \theta +2 \pi$. This shift is an exact symmetry for a GLSM
1655: on a compact world sheet without a boundary (because ${1\over 2 \pi}\int v$ 
1656: is integrally quantized in such a theory). However its action on boundary 
1657: conditions is nontrivial, (recall flux is not quantized for a 2d gauge
1658: theory on the disk). In particular in the UV of the GLSM RG flow,
1659: fractional instantons (configurations in which the phase of the vev of
1660: $\ph_n$ winds once on the boundary of the disk) violate this shift
1661: symmetry, and distinguish\foot{In fact the one
1662: point function of the bulk ${\bf C/Z}_n$ tachyon, on the disk, with boundary 
1663: conditions corresponding to the fractional brane $|i\rangle$ is given
1664: by (see \DoMo \DiGo \Bi)
1665: \eqn\bo{c_{m}={\cal N} \ e^{-2\pi i m/n}.} 
1666: This one point function
1667: corresponds to the GLSM partition function in the presence of a single 
1668: fractional instanton, which is clearly proportional to  
1669: $e^{-i\theta/n}$, consistent with the assignment $\theta = 2\pi m$ for
1670: the brane $|m\rangle$. }  
1671: the $n$ different D2-branes.  
1672: 
1673: On shifting $\theta$ as described above the boundary condition continues 
1674: to be ${\cal N}=2$ supersymmetric. In fact shifts in $\theta$ are the only 
1675: ${\cal N}=2$ supersymmetric boundary conditions generated by 
1676: marginal or relevant 
1677: boundary operators (spacetime gauge fields). As the $\theta$ angle is not 
1678: renormalized under RG flow, we conclude that the fractional branes
1679: $|m\rangle$ flow in the IR to the GLSM with Neumann boundary conditions and 
1680: $\theta= 2 \pi m$. However the fractional instantons that allowed one to
1681: distinguish between $|m\rangle$ in the UV are absent in the IR (because the 
1682: $\ph_1$ rather than $\ph_n$ has a vev in the IR). Consequently $\theta$ is 
1683: effectively periodic with periodicity $2 \pi$ in the IR, and all the $n$ 
1684: branes $|m \rangle$ evolve into the same bulk flat space D2-brane.
1685: 
1686: In order to develop intuition for these flows, we now study the 
1687: evolution of the spacetime gauge fields on the $m^{th}$ (fractional)
1688: D2-brane in the `classical' sigma model approximation. As we have 
1689: explained in Section 3, the world-sheet gauge field $v_0$ is easily 
1690: computed as a function of scale in this approximation. However $v_0$ is 
1691: simply related\foot{This is shown by the identification 
1692: of the boundary coupling 
1693: \eqn\binti{m \int dx^0 ({\sigma + {\bar \sigma} \over 2} - v_0),}
1694: with the usual sigma model gauge coupling $\int A^\mu(X)\de_0 X^\mu$ setting
1695: $\sigma=0$ in the Higgs branch. This gauge coupling comes from the 
1696: shift of the parameter 
1697: $\theta$ in the term $\f{\theta}{2\pi}\int v$. The $\sigma$ dependent term
1698: is also induced to preserve ${\cal N}=2$ supersymmetry (see \Ho). 
1699: } to the spacetime gauge field; 
1700:  $m v_{0}=A_\mu(X)\de_{0}X^\mu$ 
1701: \HoIqVa \Ho. Using these formulas we find that,
1702: for $r>0$ (in the UV region\foot{
1703: Under the RG-flow the FI parameter $r$ changes following (B.2) 
1704: as reviewed in the Appendix B. The UV limit
1705: corresponds to $r=\infty$, while IR limit to $r=-\infty$.}) 
1706: the flux is delta functionally localized,  
1707: \eqn\gfluxuv{F=-m \delta(\rho)\ d\rho\we d\theta.}
1708: On the other hand when $r<0$ 
1709: \eqn\gflux{F=-\f{2 m  n|r| \rho}{\left(n(n+1)\rho^2+|r|\right)^2}\ 
1710: d\rho\we d\theta.}
1711: Thus, as the RG-flow proceeds, the flux \gflux\ (whose integral is
1712: conserved) expands out with the bubble of flat space. 
1713:  In the IR limit the flux is completely
1714: diluted and the $n$ D2-branes are indistinguishable. 
1715: 
1716: This property of boundary RG-flow can also be seen in 
1717: the sigma model with boundary coupling 
1718: $\int_{\de\Sigma} A_{\mu}(X)\de X^\mu$. The one loop RG-flow
1719: equation \CallanBC\ leads to (in the weak curvature approximation)
1720: \eqn\rga{\f{\de A_\mu}{\de \lambda}=\nabla^\nu F_{\nu\mu}-\xi^\nu 
1721: F_{\mu\nu}+\nabla^\nu \phi\  F_{\mu\nu},} 
1722: where $\lambda$ is the logarithm of length scale (proportional to 
1723: $|r|$ in \gflux) 
1724: under the flow and 
1725: $\xi^\nu$ represents the freedom of coordinate transformations. 
1726: After we substitute the 
1727: results\foot{
1728: They are given by the metric $(ds)^2=\lambda(f(s)^2ds^2+s^2 d\theta^2)$
1729: and the parameter $\xi=\f{1}{2}sf(s)ds$ with zero dilaton \GuHeMiSc.} 
1730: of bulk RG-flow in \GuHeMiSc, we obtain the following 
1731: equation assuming that only $A_\theta$ is nonzero,
1732: \eqn\rga{\f{\de A_\theta}{\de \lambda}=\f{s}{f(s)\lambda}\ \de_{s}
1733: \Bigl(\f{1}{f(s)s}\de_{s}A_\theta+\f{1}{2}A_\theta\Bigr),}
1734: where we rescaled the coordinate as $s=\rho/\sqrt{\lambda}$ and defined
1735: the function $f(s)$ by $f(s)=(1+W(\f{1}{n}-1-s^2/2))^{-1}$
1736: ($W$ is the inverse function of $xe^x$). By setting the right-hand side of 
1737: \rga\ to zero, we find a solution (up to constant shift)
1738: \eqn\sola{
1739: A_{\theta}=A_{\theta}(0)\cdot e^{-\int^{s}_0\f{s'f(s')}{2}ds'}.}
1740: After transforming into the previous radial coordinate $\rho$
1741: we get the gauge flux
1742: \eqn\solf{F_{\rho\theta}=-\f{A_{\theta}(0)\cdot f(\rho/\sqrt{\lambda})
1743: \cdot \rho/\sqrt{\lambda}}{2\sqrt{\lambda}}
1744: \cdot e^{-\int^{\rho/\sqrt{\lambda}}_0\f{s'f(s')}{2}ds'}.}
1745: Since the function $f$ satisfies the bound $1/n<f<1$, it is obvious that
1746: the flux $F_{\rho\theta}$ value takes its the maximum at 
1747: $\rho\sim O(\sqrt{\lambda})$ as in \gflux\ 
1748: and the asymptotic behavior ($r\to \infty$)
1749: is similar to the solution to diffusion equation as
1750: $F_{\rho\theta}\sim e^{-\f{f\rho^2}{4\lambda}}$. 
1751: Thus it is completely smeared
1752: in the IR limit $\lambda\to \infty$ and we again confirmed the 
1753: diffusion property in a different picture.
1754: 
1755: It is instructive to analyze how the spectrum of open strings between 
1756: the fractional branes $|0 \rangle$ and $|m \rangle$ evolves as a function
1757: of RG scale. In the UV the relative (spacetime) field strength between these
1758: two branes is given by \gfluxuv\ (or equally, a Wilson line 
1759: \eqn\wilsonl{\int _{0}^{{2\pi \over n}} A_{\theta} d \theta= -{2 \pi m
1760: \over n},}
1761: in the `angle' direction). 
1762: Consequently, open strings stretching between these branes obey twisted 
1763: boundary conditions. For instance the (open string) tachyon field obeys
1764: \eqn\pro{
1765: T(z)=e^{-2\pi im /n}\ T(gz),}
1766: where the operator $g$ acts on the coordinate $z$ of ${\bf C}/{\bf Z}_{n}$ 
1767: as $g:z\to e^{2\pi i/n}z$ in agreement with \bcf. As the RG flow proceeds
1768: towards the IR, however, the flux enclosed within a circle of any finite 
1769: spacetime radius (and hence the effective Wilson line at that radius) 
1770: goes to zero (see \gflux). In other words the world-volume fields obey 
1771: twisted boundary condition like \pro\ for $|z|>> \sq{|r|}$, but satisfy 
1772: the untwisted one inside the bubble of flat space $|z|<< \sq{|r|}$.
1773: 
1774: \vskip .2in
1775: 
1776: {\bf D0-branes}
1777: 
1778: We will now study the evolution of D0-brane boundary conditions under 
1779: the RG flow. Unfortunately it appears difficult to find simple boundary
1780: conditions that both preserve ${\cal N}=2$ and also reduce in the UV to the
1781: D0-brane; however, we can circumvent this problem with a trick. 
1782: As we will explain below, it is straightforward to construct 
1783: ${\cal N}=2$ GLSM 
1784: boundary conditions that reduce in the UV to the $D2-{\bar D}2$ system.
1785: These boundary conditions may then be deformed (while preserving 
1786: ${\cal N}=2$
1787: supersymmetry) by turning on a vortex for the open string tachyon. The
1788: resulting boundary condition may be interpreted as a single D0-brane (the 
1789: $D2-{\bar D}2$ brane have disappeared via open string tachyon 
1790: condensation \SenSM). 
1791: Non renormalization of the boundary super potential 
1792: (guaranteed by ${\cal N}=2$ supersymmetry) allows us to follow the evolution
1793: of the boundary conditions as a function of RG scale, and determine the
1794: fate of fractional D0-branes under tachyon condensation. 
1795: 
1796: To be more concrete, let us consider the $D2 \bar{D}2$ system 
1797: in ${\bf C/Z}_n$ 
1798: where the $D2$ brane is of type $|0\rangle$ and the ${\bar D} 2$
1799: brane is of type $|m \rangle$ from the viewpoint of GLSM.
1800: As we have seen before, the action of ${\bf Z}_{n}$ quantum symmetry
1801: (or equally the shift of $\theta$ by $2\pi$) changes the types
1802: of 2-branes in UV limit. Since the shift of $\theta$ by $2\pi n$ is the 
1803: symmetry of GLSM with boundary in the orbifold limit, we have the 
1804: periodicity $|m+n \rangle=|m \rangle$ at the beginning of the RG-flow.
1805: 
1806: In order to construct boundary conditions that represent the $D2-{\bar D}2$
1807: system, we must introduce Chan Paton factors. This is conveniently achieved
1808: by introducing a new complex fermionic boundary variable $\eta$. It will 
1809: turn out (see below) that $\eta$ and its conjugate ${\bar \eta}$ obey 
1810: the conjugation relations $\{\eta,\bar{\eta}\}=1$; consequently
1811: quantization leads to a two state system. Operators on this Hilbert space
1812: are $2 \times 2$ matrices; as expected for the $D2 \bar{D}2$ system. 
1813: 
1814: Then the $D2 \bar{D}2$ system in ${\bf C/Z}_n$ 
1815:  is described by the GLSM with boundary 
1816: conditions \bcdtt, together with gauge invariant 
1817: boundary interactions (this can also 
1818: be seen as the 
1819: boundary string field theory \WittenBSFT \BSFT\ 
1820: for brane-antibrane system \DD)
1821: \eqn\BSFT{
1822: S_{\de\Sigma}=\int_{\de\Sigma} d\tau d\theta d\bar{\theta} \
1823: \bar{\Gamma}e^{-mV}\Gamma
1824: +\int_{\de\Sigma} d\tau d\theta\ \Gamma T(\Phi)+(h.c.),}
1825: where $\Gamma$ is a charge $m$ boundary fermionic chiral superfield\foot{
1826: In the UV limit this charge means that the ${\bf Z}_n$ orbifold 
1827: action acts as
1828: $\Gamma\to\Gamma\ e^{2\pi i/m}$. Thus the tachyon followed the projection
1829: $T\to T\ e^{-2\pi i/m}$. This fact agrees with \pro .}
1830: whose 
1831: lowest  component is $\eta$ 
1832: \eqn\bdyf{\Gamma=\eta+\theta F-i\theta\bar{\theta}\partial_{\tau}\eta ,}
1833: and $V$ is the GLSM vector superfield whose lowest component is given by
1834: $(v_0-\f{\sigma+\bar{\sigma}}{2})\theta {\bar \theta}$. 
1835: In components, the first term in \BSFT\ is simply 
1836: \eqn\detint{
1837: -m \int {\bar \eta} \eta (v_0 -{\sigma + {\bar \sigma} \over 2})
1838: \sim -m \int \f{\sigma_3}{2} (v_0 -{\sigma + {\bar \sigma} \over 2}),}
1839: the analogue of \binti\ on a single D2-brane. Note, in particular, that 
1840: $v_0=\de_{\tau}X^\mu A_{\mu}(X)$\foot{The scalar field $\sigma$, which is 
1841: in the same ${\cal N}=2$ multiplet as $v_\mu$, is taken to be zero
1842: since we are interested in the Higgs branch. For the analysis
1843: of the Coulomb branch in similar GLSMs see \MaMo.} 
1844: where $A_{\mu}$ is the relative gauge field in the $D2 \bar{D}2$ system.
1845: The second term in \BSFT\ represents the boundary deformations 
1846: corresponding to the open-string tachyon field $T(\Phi)$, where $T$ is a
1847: complex scalar.
1848: 
1849: It turns out that the GLSM with boundary conditions \bcdtt\ and boundary 
1850: interactions \BSFT\ preserves ${\cal N}=2$ supersymmetry provided 
1851: that the tachyon 
1852: $T$ is a holomorphic function of $\Phi$. The simplest solution to this 
1853: condition is $T=0$; this represents a $D2-{\bar D}2$ system in
1854: the UV, and flows to a usual 
1855: $D2-{\bar D}2$ system in the IR. In particular, the 
1856: open string tachyon, which is protected against quantum corrections 
1857: owing to a non-renormalization theorem, remains zero throughout the flow. 
1858: 
1859: {~~~}
1860: 
1861: {\it Flows from bulk 0-branes into bulk 0-branes}
1862: 
1863: We now consider boundary conditions with the open string tachyon turned on. 
1864: We first consider the case $m=0$ (i.e. 
1865: the 2-brane and anti 2-brane are of the same variety).
1866: In this situation $\Gamma$ is uncharged; consequently gauge invariance of 
1867: \BSFT\ requires that $T(\Phi)$ should be gauge invariant (uncharged). 
1868: The simplest example is given as follows
1869: \eqn\stak{
1870: T(\Phi_{-n},\Phi_{1})=T_{1}\Phi_{-n}(\Phi_{1})^n,}
1871: where the coefficient $T_{1}$ corresponds to the strength of the
1872: open-string tachyon field. In the UV limit the field $\Phi_{-n}$ has a large
1873: expectation value and can be replaced by $\sq{r}/n$, while $\Phi_{1}$ can be
1874: regarded as a coordinate of the orbifold ${\bf C}/{\bf Z}_n$. Thus the 
1875: tachyon field is expressed\foot{Note that this 
1876: field configuration satisfies the projection \pro\ 
1877: setting $m=0$.}
1878: \eqn\takw{
1879: T(z)=\f{T_{1}\sq{r}}{n}z^n.}
1880: 
1881:  In the deep UV $T(z) = k z^n$ where $k$ diverges; 
1882: and so represents a bulk D0-brane \ta\ at the origin of the orbifold. 
1883: (The tachyon field \stak\ has winding number $n$ and so would have
1884: represented $n$ D0-branes in flat space. A bulk 0-brane on ${\bf C/Z}_n$ is 
1885: precisely a collection of $n$ covering space 0-branes modded out by the 
1886: projection, explaining our identification).
1887: 
1888: On the other hand, in the IR limit, the field $\phi_{1}$ has 
1889: an expectation value  $\phi_{1} \approx \sqrt{|r|} $ 
1890: , and another field 
1891: $\phi_n$ is identified with $z$ the coordinate in flat space. 
1892: The tachyon field configuration \stak\   
1893: \eqn\takz{
1894: T(z)=T_{1}\sq{|r|} z ,}
1895: has unit winding number and so may be identified with a D0-brane in the 
1896: flat space. Thus we conclude that a bulk D0-brane evolves into an
1897: ordinary D0-brane the closed string tachyon condensation, consistent 
1898: with the results of subsection 5.1. 
1899: 
1900: This analysis is easily extended to more complicated 
1901: tachyon configurations.
1902: By following the RG flow in the presence of the polynomial 
1903: valued tachyon field
1904: \eqn\takcom{
1905: T(\Phi_{-n},\Phi_{1})=\sum_{l=0}T_{l}(\Phi_{-n}(\Phi_{1})^n)^l,}
1906: we conclude that a configuration of $n$ separated bulk D0-branes 
1907: (located at the zeros of $\sum_{l=0}T_{l}w^l=0$)  evolve into $n$ 
1908: separated D0 branes in flat space.  
1909: 
1910: {~~~}
1911: 
1912: {\it Flows from bulk 0-branes into nothing}
1913: 
1914: We can also choose the following tachyon field for the same kind of
1915: $D2-\bar{D}2$ system setting $m=n$ (remember the periodicity 
1916: $| m+n \lb=| m \lb$ in the UV limit)
1917: \eqn\stakbn{
1918: T(\Phi_{-n},\Phi_{1})=T_{1}(\Phi_{1})^n.}
1919: Even though the starting point is the same as before (a bulk D0-brane),
1920: the final product is different. Since the tachyon field takes a 
1921: large constant value and is completely condensed in the end, 
1922: the bulk D0-brane disappears in this process. Note also that this only 
1923: happens near the origin ($z=0$) because
1924: it is 
1925: impossible to shift the location of the D0-brane away from the origin
1926: as in \takcom. These results are completely consistent with those 
1927: obtained from the geometrical consideration in section 5.1.
1928: 
1929: {~~~}
1930: 
1931: {\it Flows from fractional 0-branes into nothing}
1932: 
1933: We now turn to the study of tachyon condensation on the more general 
1934: system consisting of a D2-brane $| m \lb_{D2}$ and an anti D2-brane
1935: $|\bar{0}\lb_{D2}$. In this situation $\Gamma$ has charge $-m$, and 
1936: charge conservation demands that $T(\ph)$ have charge $m$. 
1937: The simplest example of such a tachyon field is
1938: \eqn\takfr{
1939: T(\Phi_{-n},\Phi_{1})=T_{1}(\Phi_{1})^{m}.}
1940: In the UV the tachyon field $T(z)\sim z^{m}$ may be interpreted as 
1941: representing $|m  \lb_{D0},|m-1\lb_{D0},\ddd,|1\lb_{D0}$
1942: (this follows from an analysis of the RR charges of this configuration, 
1943: see \ta). On the other hand, in the IR limit $\phi_1 \approx \sqrt{|r|}$ and
1944: the tachyon field is simply constant (and this constant value increases
1945: without bound as we flow to the deep IR). Thus the open string tachyon 
1946: is completely condensed; all open string degrees of freedom vanish in this
1947: limit; consequently the fractional 0-brane has simply disappeared.
1948: This matches the results of section 5.1 again.
1949: 
1950: {~~~}
1951: 
1952: {~~~}
1953: 
1954: {\it 0-brane production from nothing}
1955: 
1956: Finally, let us consider the case of $m=-n$ 
1957: (the same kind of D2 and antiD2). 
1958: Then an allowed tachyon field is
1959: \eqn\takflux{
1960: T(\Phi_{-n},\Phi_{1})=T_{1}\Phi_{-n},}
1961: which shows the initial nothing state will become a D0-brane
1962: in the flat space after the closed string tachyon condensation. 
1963: Once again, this also matches with the previous geometrical argument
1964: and corresponds to the cycle ${\bf S}^1_{B}$.
1965: 
1966: \subsec{GLSM Analysis of Twisted Circle Boundary Flows}
1967: 
1968: As we have described in section 3, the decay of twisted circle theories 
1969: via closed string tachyon condensation could easily be understood utilizing 
1970: a GLSM. The low energy limit of the GLSM is a CFT with a Liouville factor;
1971: the Liouville direction may (roughly) be thought of as a scale or Euclidean
1972: time. In this subsection we will study the same GLSM on a world sheet with 
1973: a boundary, imposing boundary conditions that preserve 
1974: ${\cal N}=2$ B-type world sheet
1975: supersymmetry. The boundary conditions we study will always be Neumann in 
1976: the Liouville ($P_1$) direction (this is required for 
1977: ${\cal N}=2$ world sheet 
1978: supersymmetry, \SuYa \EgSu); and in most examples we study the boundary 
1979: interaction will also be independent of $P_1$. As in the previous
1980: subsection, ${\cal N}=2$ supersymmetry guarantees these 
1981: features persist in the 
1982: low energy CFT; it is then a simple matter to follow the boundary
1983: conditions, and hence trace the D-branes from $P_1 \gg 0$ (twisted circle) 
1984: to $P_1 \ll 0$ (supersymmetric circle). As the arguments of this
1985: subsection closely mimic those of the previous subsection, we will be very 
1986: brief. 
1987: 
1988: The decay of a twisted circle D3-brane is described by a D4-brane in the 
1989: GLSM of section 3. The twisted circle D3-brane with zero Wilson line is 
1990: described by Neumann boundary conditions on $\ph_1$, $\ph_n$, 
1991: $P_1$ and $P_2$, together with the conditions $v_1=v_{01}=0$ on the
1992: boundary (these conditions preserve ${\cal N}=2$ B-type world sheet
1993: supersymmetry). Turning on a Wilson line corresponds to adding the boundary 
1994: interaction \binti. The UV Wilson line ${ 2 \pi m \over n}$ 
1995: ($m$ in \binti\ varies continuously between $0$ and $n$) evolves 
1996: into\foot{Taking the zero radius limit (see 
1997: subsection 4.2) we conclude that all ${\bf C/ Z}_n$ fractional
1998: 2-branes map into the bulk flat space 2-brane, in agreement with the 
1999: previous subsection.} the 
2000: IR Wilson line ${ 2 \pi m}$. 
2001: 
2002: As in the previous subsection, it is possible to construct the twisted
2003: circle fractional $D1$ brane ($D2$ brane on including 
2004: the Liouville direction) via $D3 {\bar D}3$ ($D4 {\bar D}4$ on including
2005: the Liouville direction) tachyon condensation. $P_1$ independent boundary 
2006: deformations of the $D4 {\bar D}4$ are identical to \stak\ for a bulk 
2007: D-brane and \takfr\ for a fractional D-brane. The arguments and conclusions
2008: of the previous subsection carry through almost unmodified, once again
2009: confirming the geometrical reasoning of subsection 5.1. In particular, we
2010: find that the bulk D1-brane in the twisted circle either disappears or 
2011: evolves into an ordinary 
2012: D1-brane in flat space after the closed string tachyon condensation. On the 
2013: other hand, the fractional D1-brane disappears from the spectrum. 
2014: 
2015: 
2016: Finally we should point out that it is also possible to  construct large
2017: classes of ${\cal N}=2$ boundary interactions which depend on $P$. 
2018: For example, we can take 
2019: \eqn\ptab{
2020: T(P,\Phi_{-n},\Phi_{1})=T_1(\Phi_{1})^{m+q}e^{-qP}.}
2021: These are expected to lead to other boundary states of the GLSM 
2022: after the boundary RG-flow\foot{Here we should consider relevant
2023: perturbations $q<\f{k}{4(n-1)}$ or marginal one $q=\f{k}{4(n-1)}$.
2024: }. We will not investigate the
2025: relevance of these D-branes to our closed string tachyon condensation.
2026: 
2027: \newsec{Evolution of D-branes under More General RG Flows}
2028: 
2029: Before we conclude this paper, we would like to apply our method of 
2030: analyzing D-brane spectrum to more general RG flows. In particular,
2031: let us consider\foot{We are very grateful to A. Adams and R. Gopakumar
2032: for suggesting an investigation of this case to us.}
2033:  the RG Flows from 
2034: ${(\bf C\times S^1)/Z}_{n}$ to ${\bf( C\times S^1)/Z}_{m}\ \ (n>m)$. We 
2035: further
2036: assume that $n$ and $m$ are coprime. Then we can construct gauged linear
2037: sigma models which describe these RG-flows in the same way as before. 
2038: This is defined by just replacing the
2039: chiral superfields $\Phi_{1}$ in \ft\ with the charge $m$ superfields
2040: $\Phi_m$ in the model discussed in section 3.1. If we take the small radius
2041: limit $k\to 0$, then it reduces to the RG-flow from 
2042: ${\bf C/Z}_{n}$ to ${\bf C/Z}_{m}$, which is also realized as a gauged 
2043: linear sigma model in the same way \Va.
2044: 
2045: Now let us consider the evolution of D-branes under 
2046: these RG-flows\foot{Even though we use the terminology of 
2047: D-branes in the twisted circle theory, 
2048: we can also get the equivalent result for the D-branes in the
2049: orbifold.}. For simplicity, we consider the explicit example of 
2050: $(n,m)=(5,3)$, 
2051: though its generalization is obviously straightforward. Again we can apply 
2052: two different methods. 
2053: 
2054: First, we analyze the geometrical picture 
2055: as in section 5.1. We can compute the classical metric as in Appendix C 
2056: and indeed get smooth four geometry. In the UV limit it 
2057: approaches the metric
2058: of ${(\bf C\times S^1)/Z}_{n}$ given by
2059: \twim, while in the IR limit it becomes the metric of 
2060: ${(\bf C\times S^1)/Z}_{m}$
2061: \eqn\cnm{ds^2=d\rho'^2+\f{n^2}{m^2}\rho'^2d\theta^2
2062: +\f{k}{2}\Bigl(\f{d\theta}{m}-dP_2\Bigr)^2.} Note that the angle 
2063: $\theta(\equiv \theta_m)$ has the periodicity $2\pi$ and also we have
2064: the ${\bf Z}_{n}$ identification 
2065: $(\theta,P_2)\sim (\theta+\f{2\pi m}{n},P_2+\f{2\pi}{n})$.
2066: 
2067: Consider a fractional D1-brane wrapped on the twisted circle 
2068: ${\bf S}^1_A$ in the UV 
2069: defined in the same way as in section 3.3 (see Fig.5).
2070: After the RG flow, it again disappears as we can see from \cnm\ explicitly.
2071: However, a bound state of two fractional D1-branes wraps a cycle 
2072: equivalent to ${\bf S}^{1}_{D}$ (equivalence upto the shift of
2073: the UV trivial cycle ${\bf S}^{1}_{B}$). It follows from \cnm\ that 
2074: the cycle ${\bf S}^{1}_{D}$ degenerates at $\rho'=0$, 
2075: but then evolves into the twisted circle of ${(\bf C\times S^1)/Z}_{m}$ upon
2076: subsequent evolution. Thus we predict that even though 
2077: a single fractional D1-brane disappears under tachyon condensation, 
2078: a bound state of two such branes can either evolves into
2079: a fractional D1-brane or disappear upon closed tachyon condensation. 
2080: On the other hand, if we consider a bulk D1-brane wrapped on 
2081: ${\bf S}^{1}_{C}$, it will simply evolve into a bulk D1-brane in the IR 
2082: twisted circle as in the previous case of $m=1$.
2083: 
2084: \fig{Various cycles in the twisted circle ${(\bf C\times S^1)/Z}_{5}.$}
2085: {d1decaymn.eps}{2.5truein}
2086: 
2087: 
2088: Next let us understand the above result by employing 
2089: the analysis of ${\cal N}=2$ supersymmetric 
2090: boundary interaction. The tachyon configurations
2091: \eqn\tnmdo{T=\ T_1 \Phi_3,\ \ \ T_1 (\Phi_3)^2,\ \ \ T_1 (\Phi_3)^2 \Phi_{-5},
2092: \ \ \ T_1 (\Phi_3)^5 (\Phi_{-5})^3, \ \ \ T_1 (\Phi_3)^5, \ \ \
2093: T_1 (\Phi_{-5})^3,  }
2094: respectively represent
2095: \item{1.} The disappearance of a single fractional D0-brane.
2096: \item{2.} The disappearance of the bound state of two fractional D0-branes.
2097: \item{3.} The evolution of a bound state of 2 fractional D0-branes into 
2098:           a single fractional D0-brane.
2099: \item{4.} The evolution of a bulk D0-brane into a bulk D0-brane.
2100: \item{5.} The disappearance of a bulk D0-brane. 
2101: \item{6.} The creation of a bulk D0-brane out of nothing.
2102: 
2103: Note the striking match between geometrically inspired conjectures and 
2104: our explicit constructions of boundary RG flows. Indeed, the tachyon fields
2105: \tnmdo\ explicitly 
2106: encode the information of the lattice vectors (see Fig.5) via
2107: the `translation rule'
2108: $\Phi_m\lr {\bf S}^{1}_{A}$ and $\Phi_{-n}\lr -{\bf S}^{1}_{B}$. 
2109: This obviously allows us to see the results of D-brane evolutions under the
2110: general $(n,m)$ RG flow\foot{The evolution of D3-branes can also be 
2111: discussed in the GLSM picture. A D3-brane is parametrized by the 
2112: Wilson line $\f{2\pi l}{n}\ \ (0\leq l< n)$ in the UV limit. 
2113: After the RG flow, the periodicity becomes small as $0\leq l< m$ by the 
2114: radius change.}
2115: .
2116: 
2117: 
2118: 
2119: 
2120: \newsec{Discussion}
2121: 
2122: In this paper we have studied the evolution of D-branes under closed string
2123: tachyon condensation in non-supersymmetric orbifold and twisted 
2124: circle backgrounds. Our analysis employed two complementary methods. 
2125: First, we used the gauge linear model description of the twisted
2126: circle tachyon condensation process to derive a smooth four dimensional
2127: geometry that approximately describes the evolution of the twisted circle
2128: in an energy scale (roughly Euclidean time) direction. Interpreting 
2129: the branes of the twisted circle theory
2130: as usual type II branes wrapping nontrivial cycles, we traced the evolution
2131: of these cycles as the tachyon condensed. Our second method involved 
2132: the analysis of RG flows in gauged linear sigma models on the disk; 
2133: we used world sheet ${\cal N}=2$ supersymmetry to follow the evolution of 
2134: boundary interactions (sometimes using ideas from boundary string field 
2135: theory) into the IR.  Both methods yielded the same result; 
2136: localized D-branes disappear (or 
2137: evolve other fractional D-branes in general RG flows) as the closed string
2138: tachyon condenses, while D-branes that are not tied to the fixed point 
2139: either disappear or evolve into usual D-branes in flat space. 
2140: 
2141: Our study suggests several directions for future study. First, twisted
2142: circle tachyon condensation is only approximately described by the four
2143: dimensional
2144: Euclidean geometry studied in this paper. The exact process is captured by 
2145: flowing the sigma model based on this approximate
2146: geometry to the IR. It would be intensely interesting  
2147: to find an exact description of the resultant ${\cal N}=2$ CFT. Indeed 
2148: Hori and Kapustin \HoKa\ were able to solve a related problem in the simpler 
2149: setting of 2d Black hole. 
2150: Such a description would permit detailed understandings 
2151: of several 
2152: aspects of the process of tachyon condensation. 
2153: 
2154: Secondly, we have only studied tachyon condensation as a function of 
2155: the Liouville direction (roughly Euclidean time). It is certainly important
2156: to understand the Lorentzian analogue of this process (the evolution of this
2157: background as a function of time). It is not clear to us that the
2158: Euclidean and Lorentzian processes are even qualitatively similar; for instance
2159: the geometry on equal `time' ($P_1$) cycles of our smooth 4-geometry 
2160: \metrico\ degenerates at the slice $P_1=0$. It would be very interesting to 
2161: understand this issue better\foot{See \GuHeMiSc\ for a recent discussion on
2162: the connection between world sheet RG flow and time evolution.}.
2163: 
2164: Our second approach to this problem also highlights several interesting
2165: connections. As we have explained the world sheet RG flow with certain 
2166: boundary conditions (for instance boundary conditions corresponding to
2167: the fractional 0-brane of ${\bf C/Z}_n$) leads to world sheet theory with 
2168: no boundary in the IR. As we have explained in subsection 5.3, there is a
2169: sense in which closed string tachyon condensation triggers open string
2170: tachyon condensation on an initially stable brane. Intuitively, it is not 
2171: difficult to understand why this happens. For instance, 
2172: the fractional 0-brane of ${\bf C/Z}_n$ may be `blown up' into a D1-brane
2173: wrapping a twisted circle. This circle (non contractible in the UV)
2174: degenerates in the process of the closed string tachyon condensation. The
2175: degeneration of this cycle allows the $D1$ brane circular loop to self
2176: annihilate via open string tachyon condensation (analogous to 
2177: $D1 \bar{D}1$ annihilation). As is remarkable even in the general RG flow
2178: analysis in section 6, we can obtain the same conclusions 
2179: from our two different viewpoints. This might be mathematically 
2180: regarded as a novel correspondence
2181: (analogous to McKay correspondence\foot{See also \MaMo \YHE\ for 
2182: similar discussions in four dimensional unstable orbifolds.})
2183: between geometrical structure and 
2184: algebraic one under `time' evolution.
2185: 
2186: Another interesting direction of investigation concerns the limit 
2187: $n \to \infty$ of the models studied in this paper. As we have explained in 
2188: Section 2, in this limit the twisted circle reduces to an orbifold of 
2189: type 0 theory. If the study of tachyon condensation commutes with $n \to
2190: \infty$, the results of this paper determine the fate of several 
2191: type 0 D-branes under the process of closed string tachyon condensation. As 
2192: the type 0 tachyon is delocalized, type 0 tachyon condensation is 
2193: particularly interesting, and deserves further study.
2194: 
2195: Finally, in this paper we have considered configurations consisting of a
2196: fixed number of D-branes, in the classical limit $g_{string} \to 0$. In
2197: this limit D-branes probe (and react to) the closed string geometry, but do
2198: not back react on it. New interesting phenomena may occur in other limits;
2199: for instance if the number of branes is taken to infinity. We leave such
2200: investigations to future work.
2201: 
2202: 
2203: \vskip 0.4in
2204: 
2205: \centerline{\bf Acknowledgments}
2206: 
2207: \vskip .1in
2208: 
2209: We would like to thank J. David, T. Eguchi, M. Gutperle, 
2210: Y.Hikida, A. Iqbal, T. Kawano, Y. Michishita, A. Sen, E. Silverstein, 
2211: A. Strominger,  H. Takayanagi, D. Tong, N. Toumbas, C. Vafa,  S. Yamaguchi, 
2212: P. Yi and especially  A. Adams, R. Gopakumar, M. Headrick and K. Hori for 
2213: several extremely useful discussions. We would particularly like to thank 
2214: M. Headrick for disabusing us of a potentialy embarrassing misconception. The
2215: work of SM was supported  in part by DOE grant DE-FG03-91ER40654
2216: and a Harvard Junior Fellowship. The work of TT was
2217: supported in part by DOE grant DE-FG03-91ER40654.
2218: 
2219: 
2220: \appendix{A}{Type 0 theory and Interpolating Orbifolds}
2221: 
2222: Type 0A/B theory may be regarded as an orbifold by $(-1)^{F_s}$ ($F_s$ is
2223: the spacetime fermion number) of  type IIA/B theory. This is most easily 
2224: seen at the level of the torus partition
2225: function. The projection by $(-1)^{F_s}$ retains only states in the
2226: NS-NS or RR sectors of the theory, i.e. states with fermion 
2227: boundary conditions
2228: $(-,-)$ or $(+,+)$ on the $\sigma$ cycle of the torus. It now remains to 
2229: add in the twisted sectors; as the net result of this process is symmetric 
2230: (i.e. modular invariant) 
2231: between the two cycles of the torus, the final partition function thus 
2232: contains only $(+,+)$ and $(-,-)$ along the $\tau$  cycle as well and 
2233: the resulting partition function is that of type 0A/B.
2234: 
2235: Recall, that type 0A and type 0B theory are T-dual to each other; 
2236: under T-duality $\psi_9 \r - \psi_9$, and so $(-1)^{F_l+F_r} \r
2237: -(-1)^{F_l+F_r}$ in the Ramond sector (this is most clearly seen on 
2238: expressing $(-1)^{F_l+F_r}$ in terms of bosonized fermions),
2239: interchanging type 0A and type 0B.
2240: 
2241: We now move onto the consideration of a slightly more involved orbifold
2242: of type IIA/B theory; the orbifold by $(-1)^{F_s} \times \sigma$ where
2243: $\sigma$ represents a shift by $2 \pi R$ in the $z$ direction \CoGu.
2244: As we now review, in the $R \to 0$ limit this so called `interpolating 
2245: orbifold' is equivalent to type 0B theory on ${\bf R}^{10}$; it
2246: interpolates  between IIA theory on 
2247: ${\bf R}^{10}$ (as $R \to \infty$ ) and
2248: 0B theory on ${\bf R}^{10}$ (as $R \to 0$). It is clear that the projection
2249: by $(-1)^{F_s} \times \sigma$ retains only fermions of half odd momentum
2250: and bosons of integral momentum around the $z$ circle. At the level of the 
2251: torus partition function, the sum over windings of the $\tau$ cycle over 
2252: the spacetime circle (parameterized by the winding number $w_{\tau}$) is
2253: weighted by the additional factor $(-1)^{w_t}$ in the $(-,+)$ or 
2254: $(+,-)$ sectors. Once again, the symmetry of the twisting procedure 
2255: between the $\sigma$ and the $\tau$ cycles demands that the summation over 
2256: windings in the $\sigma$ direction be weighted by $(-1)^{w_s}$ when the
2257: fermions have $(-,+)$ or $(+,-)$ boundary conditions in the $\tau$
2258: direction. This reverses the GSO projection, setting 
2259: $(-1)^{F_L}=(-1)^{F_R}=1$, (instead of the usual $(-1)^{F_L}=(-1)^{F_R}=-1$)
2260: in sectors of odd winding. 
2261: 
2262: In summary, this orbifold retains 
2263: only fermions of half odd momentum and bosons of integral momentum. 
2264: Furthermore, the GSO projection is reversed in sectors of odd winding.
2265: It is now clear that the orbifold of IIA theory by $(-1)^{F_s} \times \sigma$
2266: reduces, as $R \to 0$, to 0B theory in 
2267: noncompact ${\bf R}^{10}$. In this limit all the states of non-zero 
2268: momentum around the circle are lifted out of the spectrum and this  
2269: includes all the fermions (since by the antiperiodic boundary 
2270: conditions they have half-integral momenta). In the sector with 
2271: zero-momentum along the circle,  
2272: the only effect of the boundary conditions is to reverse the GSO projection 
2273: for states of odd winding number $w$.
2274: Thus when $w=2w'$ ($w$ is the winding number) the spectrum consists of the
2275: (NS+,NS+) and (R+,R-) sectors, and for $w=2w'+1$ the (NS-,NS-)
2276: and (R-,R+) sectors. This is approximately the spectrum in the
2277: zero-momentum, winding number $w'$ sector of type 0A theory on a circle of
2278: radius $2R$. This equality becomes exact in the limit $R \to 0$, where it is
2279: more appropriate to describe the theory as 0B in infinite flat space.
2280: 
2281: We now describe the T-dual of the interpolating orbifold. T-duality
2282: is easiest to understand in terms of states rather than partition functions.
2283: As usual, T-duality
2284: interchanges momentum and winding. The T-dual theory is most usefully 
2285: thought of as a theory with radius ${\apm \over 2R}$. Sectors of odd winding
2286: are all fermions, sectors of even winding are all bosons. In addition, 
2287: states of integral momentum have the IIB/A GSO projection, while states of 
2288: half odd momentum have the reversed GSO projection. In the limit that
2289: $R \to 0$, this theory is approximately type 0B/A theory on a circle of
2290: radius ${\apm \over 2R}$. At finite $R$ the model is an orbifold of
2291: type 0B/A theory by $(-1)^{F_L} \times \sigma'$ where $\sigma'$ is
2292: a shift by  ${2 \pi \apm \over 2R}$ in the $z$ direction. The projection
2293: associated with this orbifold ensures that states with the usual (IIB)
2294: GSO projection have integral momentum; states with the opposite 
2295: GSO projection
2296: have half odd momentum. Twisted sectors then produce the rest of the
2297: spectrum, as may be seen, as usual, by demanding symmetry of the partition
2298: function under $\sigma \leftrightarrow \tau$.
2299: 
2300: 
2301: 
2302: \appendix{B}{Decay of ${\bf C}/{\bf Z}_n$ }
2303: 
2304: \subsec{Review of Tachyon Condensation in ${ \bf C/Z}_n$}
2305: 
2306: Recall that type II theory on ${\bf C}/{\bf Z}_n$ has tachyons 
2307: in its twisted
2308: sectors. Adams Polchinski and Silverstein (APS) argued that the tachyon
2309: condensation process in this background nucleates a bubble of flat space at
2310: the tip of the cone \AdPoSi; the bubble then grows without bound. 
2311: 
2312: Let us consider the actual process of tachyon condensation in more detail. 
2313: Let the initial state of the system be given by the ${\bf C}/{\bf Z}_n$
2314: background perturbed by $\int d^2z ~ \ep~ T$ (in $0-$picture), 
2315: where $T$ is the most relevant 
2316: twisted sector operator. The subsequent evolution of the system under 
2317: worldsheet RG flow has two distinct stages. The first stage (the stage
2318: of linear growth) is characterized by the condition $\ep \ll 1$. In this 
2319: stage $\ep$ simply grows with RG scale at a rate determined by the 
2320: $2(1-\delta)$, where $\delta$ is the holomorphic scaling dimension of $T$; 
2321: no other operators are turned on. Eventually $\ep$ grows to unit order, and 
2322: begins to mix in a nonlinear fashion with all other operators. At this 
2323: point the spacetime metric (hence target space geometry) begins to evolve
2324: with RG scale, and we enter the second stage of the RG flow. 
2325: According to the APS conjecture, in this stage a large bubble of flat two 
2326: dimensional space is nucleated about the tip of (what was) the ${ \bf
2327: C/Z}_n$ cone, and this bubble then grows without bound. 
2328: Once the size of the bubble is large in string unit, the further evolution 
2329: of the system is accurately described by the 1-loop RG equations of the 
2330: worldsheet nonlinear sigma model. 
2331: 
2332: In order to substantiate this picture, APS first computed the metric on the 
2333: moduli space of D0-branes in the ${\bf C}/{\bf Z}_n$ background perturbed by 
2334: $\int d^2z \ep T$, where $\ep \ll 1$. They found that the 0-brane moduli 
2335: space is ${\bf C}/{\bf Z}_n$ with the singularity smoothened out at 
2336: length scale $\ep \sqrt{\apm}$. This demonstrates that D-branes already 
2337: see the first stage of the RG flow in terms of a growing 
2338: bubble of flat space. Once the size of this bubble becomes large in string
2339: units, a fundamental string must see the same geometry as the D0-brane;
2340: consequently this computation provides evidence for the conjecture
2341: described above. In order to further substantiate their picture, 
2342: APS went on to find an approximate solution
2343: \foot{An exact solution with the same property was
2344: later found in \GuHeMiSc.} to the 1-loop worldsheet 
2345: RG equations on the string
2346: worldsheet (valid once the flat space bubble is large in string units) 
2347: corresponding to this bubble steadily expanding in a sea of 
2348: ${ \bf C/Z}_n$. 
2349: 
2350: In summary, according to the APS picture, D-branes see the tachyon
2351: condensation\foot{Most recent studies of the subject utilize worldsheet RG 
2352: flow (or Liouville flow) to study tachyon condensation. A recent discussion 
2353: of the connection between world sheet RG flow and the physically more
2354: interesting time evolution, may be found in \GuHeMiSc.} process as an
2355: always expanding bubble of flat space. Closed strings see a two stage 
2356: process. In the first stage, the tachyon grows but all other operators 
2357: including the metric are unaffected, and (geometrically speaking) nothing 
2358: happens. Eventually, however, (when the coefficient of the tachyon is 
2359: of order unity, which is also when the D-brane bubble becomes string scale)
2360: the tachyon backreacts onto the geometry, producing a string scale bubble
2361: of flat space, which then proceeds to grow in a diffusive fashion. 
2362: 
2363: Following up on the work of APS, Vafa \Va\ presented an exact construction 
2364: of the RG flow on the string worldsheet by examining the low energy
2365: behavior of a particular gauged linear sigma model. We will review 
2366: this construction below.
2367: 
2368: \subsec{Tachyon Condensation using the Gauged Linear 
2369: Sigma Model Construction}
2370: 
2371: In this subsection we review Vafa's construction of the RG flow describing
2372: ${\bf C/Z}_n$ decay utilizing a two dimensional ${\cal N}=(2,2)$ $U(1)$ 
2373: gauge theory (so called gauged linear sigma model (GLSM) \WittenYC\ ). 
2374: 
2375: Consider a gauge linear sigma model with action 
2376: \eqn\act{S={1\over 2 \pi}
2377: \int d^2\s  \left[d^4\t \left({ \bar \Ph_1} e^V \Ph_1
2378: +{ \bar \Ph_{-n}} e^{-nV} \Ph_{-n}
2379: -{1 \over 2 e^2} |\Sigma|^2\right)+{\rm{Re}}\int d^2\ti{\t}\ t\Sigma
2380: \right],}
2381: where the two charged chiral superfields $\Ph_1$ and
2382: $\Ph_{-n}$ transform under $U(1)$ gauge transformations as in \ft, 
2383: $V$ represents a vector multiplet and its lowest 
2384: component is a gauge field $v_{\mu}$ (its 
2385: field strength is defined to be $v_{01}$); the twisted chiral superfield
2386: $\Sigma$ is its superfield-strength.
2387: The complex valued constant\foot{Actually $r$ is renormalized and runs as
2388: a function of energy scale as we will see later.} 
2389:  $t=r+i\theta$ parameterizes both  the Fayet-Ilipoulos term $r\int d^4\t V$ 
2390: and the theta term $\f{\theta}{2\pi} \int v_{01}$.
2391: 
2392: The quantum theory generated by \act\ is super renormalizable; 
2393: in fact it is easy to verify that all correlators are rendered
2394: finite after a simple one loop renormalization of the FI term
2395: (see section 3, \WittenYC)
2396: \eqn\fit{r_0(\Lambda_0)=r(\Lambda) +(n-1) \ln 
2397: \Bigl({\Lambda_0 \over \Lambda}\Bigr). }
2398: Here $r_o$ is the bare FI term defined at the cut off scale $\Lambda_0$, 
2399: and $r$ is the finite FI term associated with the finite renormalization 
2400: scale $\Lambda$. 
2401: 
2402: \subsec{The Sigma Model Limit}
2403: 
2404: Consider the low energy dynamics of \act. In Wess-Zumino gauge,
2405: the terms involving the auxiliary gauge field $D$ are
2406: \eqn\dterms{{1 \over 2 \pi} \int d^2\s
2407: \left( {D^2 \over 2 e^2}+D(|\ph_1|^2-n|\ph_{-n}|^2 +r) \right),}
2408: where $\ph_1$, $\ph_{-n}$ are the lowest components of
2409: $\Ph_1$, $\Ph_{-n}$ respectively. At energies small compared to $e$
2410: we may restrict attention to fluctuations on
2411: the supersymmetric manifold of zero-energy configurations satisfying
2412: \eqn\ze{-\f{D}{e^2}=|\ph_1|^2-n|\ph_{-n}|^2+r=0.}
2413: The moduli space of classical vacua is parameterized by $\ph_1$ and
2414: $\ph_{-n}$ constrained according to \ze\ modulo gauge transformations,
2415: so that the dynamics is that of a one complex dimensional supersymmetric 
2416: sigma model. 
2417: 
2418: \subsec{GSO Projection}
2419: 
2420: We have described above how the GLSM reduces at low energies, to a 
2421: sigma model. We will now describe an exact ${\bf Z}_2$ 
2422: symmetry of the GLSM that 
2423: flows at low energies to $(-1)^{F_L}$ on the sigma model (similar remarks 
2424: apply to $(-1)^{F_R}$). This symmetry may then be used to impose a type II
2425: `GSO projection' on the GLSM.
2426: 
2427: Classically, the gauge theory
2428: has a discrete chiral symmetry given by $\psi_-  \r - \psi_-$,
2429: $v_\mu \r - v_{\mu}$  $\lambda_+ \r -\lambda_+$
2430: (all other fields are invariant; $v_\mu$ and $\lambda$ are the gauge boson
2431: and the spinor respectively in the gauge multiplet, and $\psi_-$ represent
2432: the negative chirality components of all matter fermions - this a special
2433: example of the left moving R symmetry described, for instance, in
2434: section 2 of \WittenYC.) is a symmetry of
2435: the gauge theory Lagrangian; we call the operator that generates this
2436: symmetry $(-1)^{F_L}$. This change of variables induces a Jacobian
2437: $(-1)^{{n-1 \over 2 \pi} \int v_{01}}$ in the quantum path integral; this
2438: Jacobian is trivial if and only if $n$ is odd.
2439: Thus $(-1)^{F_L}$ is non-anomalous in
2440: the quantum theory if $n$ is odd, as we assume in this paper.
2441: 
2442: 
2443: 
2444: \subsec{The Sigma Model in the UV}
2445: 
2446: Let us start with the UV limit $r=\infty$.
2447: We can use the
2448: gauge freedom to set $\ph_{-n}$ to be real and positive, and then use
2449: \ze\ to solve for $\ph_{-n}$ as $\ph_{-n} = \sqrt{(|\ph_1|^2 +r)/n}$.
2450: We then plug this solution into the kinetic terms 
2451: \eqn\kintsf{
2452: {1 \over 2 \pi}
2453: \int d^2\s ( -\CD^\m \ph_1 \CD_\m \ph_1 -\CD^\m \ph_{-n} \CD_\m \ph_{-n}),
2454: }
2455: (where $\CD_\m \ph_1$, $\CD_\m \ph_{-n}$ are the usual covariant
2456: derivatives) and classically integrate out\foot{When $r$ is very large the
2457: classical approximation is valid as (according to \ze) $\ph_{-n}$ is also
2458: very large in this limit, and so the gauge boson is very 
2459: massive, by the Higgs
2460: mechanism - recall the gauge theory is free in the UV). } 
2461: the gauge boson $v_\m$ (see
2462: subsection B.7 for the details).
2463: We find that the dynamics in this limit is governed by the flat sigma model
2464: $S={1\over 2 \pi} \int d^2\s \left(
2465: \p_\m \ph_{1} \p^\m { \bar \ph_1} \right)$.
2466: Note, however, that our gauge choice leaves unfixed a
2467: residual ${\bf Z}_n$ group of gauge transformations, generated by
2468: \eqn\unfixexgt{
2469: \ph_1 \r e^{{2 \pi i/ n}} \ph_1,
2470: }
2471: consequently, the low energy physics of our GLSM in the UV limit
2472: is the supersymmetric sigma model on ${\bf C}/{\bf Z}_n$.
2473: 
2474: \subsec{Deviations from the UV Fixed Point}
2475: 
2476: Next we consider the deformation of the UV sigma model as $r$ decreases
2477: away from infinity, but is still very large $r \gg 1$. As the GLSM has
2478: ${\cal N}=2$ worldsheet supersymmetry for all values of $r$, 
2479: the corresponding
2480: deviation in the sigma model must correspond to adding a linear combination 
2481: of the chiral operators to the worldsheet Lagrangian. The correct deformation
2482: turns out to be 
2483: \eqn\twd{\int d^2z\ e^{-{r + i \theta \over n}} V_{T},}
2484: where $V_{T}$ is the lowest dimensional tachyon vertex
2485: operator in the first twisted sector. In order to see this identification 
2486: more clearly, note that, when $r \gg 1$, \act\ has fractional instantons
2487: (vortices in which the vev of $\ph_n$ winds at infinity \WittenYC). 
2488: Any Euclidean 
2489: correlation function in the theory \act\ may be represented as a sum over 
2490: sectors containing arbitrary numbers of instantons and anti instantons; 
2491: each instanton must also be integrated over the world sheet. On flowing to
2492: low energies, these fractional instantons become twist fields (recall 
2493: that the phase of  $\ph_1$ changes by ${2 \pi \over n}$ on circling a
2494: fractional instanton). The sum over all numbers and positions of
2495: instantons and anti-instantons is thus achieved by perturbing the low
2496: energy sigma model by \twd, where the coefficient of $V_T$ is the action of
2497: a single instanton. 
2498: 
2499: \subsec{Classical RG flow to the IR}
2500: 
2501: In this subsection we will compute the sigma model metric corresponding to 
2502: ${\bf C/Z}_n$ RG flow, in the classical approximation 
2503: described in subsection
2504: 3.2 above.
2505: 
2506: First we assume that $r$ is positive and solve the D-term condition \ze\
2507: as follows
2508: \eqn\phii
2509: {\ph_{1}=\rho\ e^{i\theta_{1}},\ \ \ \ph_{-n}=
2510: \sq{\f{\rho^2+r}{n}}e^{i\theta_n},}
2511: where the $U(1)$ gauge transformation \ft\ acts as $(\theta_n,\theta_1)
2512: \sim (\theta_n-n\ap,\theta_1+\ap)$.
2513: Then the kinetic terms of the scalar field action become
2514: \eqn\kine{
2515: L=(\partial_{\mu}\rho)^2+\rho^2(\de_{\mu}\theta_1-v_\mu)^2
2516: +\f{\rho^2}{n(\rho^2+r)}(\de_\mu \rho)^2
2517: +\f{\rho^2+t}{n}(\de_{\mu} \theta_n+nv_{\mu})^2.}
2518: The equation of motion of the gauge field in the sigma model limit
2519: $e\to \infty$
2520: leads to
2521: \eqn\vectorf{
2522: v_\mu=\f{\rho^2\de_\mu\theta_1-(\rho^2+r)\de_\mu\theta_n}
2523: {(n+1)\rho^2+nr}.}
2524: After we integrate out the vector field, we get
2525: \eqn\sigmaone{
2526: L=\Bigl(1+\f{\rho^2}{n(\rho^2+r)}\Bigr)(\de_{\mu}\rho)^2
2527: +\f{\rho^2(\rho^2+r)}{n((n+1)\rho^2+nr)}(\de_\mu\theta)^2,}
2528: where we have defined the gauge invariant field
2529: $\theta=n\theta_1+\theta_n$ ($\theta\sim \theta+2\pi$).
2530: In the UV limit $r\to \infty$ (or $r>>\rho$)
2531: this sigma model corresponds to the metric of ${\bf C}/{\bf Z}_n$
2532: \eqn\uvm{
2533: ds^2=d\rho^2+\f{\rho^2}{n^2}d\theta^2.}
2534: In the region $\rho>>r$ we obtain the metric of ${\bf C}/{\bf Z}_{n+1}$
2535: \eqn\metp{
2536: ds^2=\f{n+1}{n} d\rho^2+\f{\rho^2}{n(n+1)}d\theta^2.} As we argue in section
2537: 4.2, the appearance of ${\bf C}/{\bf Z}_{n+1}$ is due to our classical 
2538: approximation and the quantum correction should modify 
2539: it such that asymptotic
2540: geometry is given by ${\bf C}/{\bf Z}_n$.
2541: 
2542: Next we consider the case of negative $r$. The solution to the D-term
2543: condition is given by
2544: \eqn\dtm{
2545: \ph_{1}=\sq{n\rho'^2+|r|} e^{i\theta_1},\ \ \
2546: \ph_{-n}=\rho' e^{i\theta_n}.}
2547: By using the same method as before we obtain the gauge field
2548: \eqn\vectorm{
2549: v_\mu
2550: =\f{(|r|+n\rho'^2)\de_\mu\theta_1-n\rho'^2\de_\mu\theta_n}
2551: {n(n+1)\rho'^2+|r|},}
2552: and the sigma model metric
2553: \eqn\sigmatwo{
2554: ds^2=(1+\f{n^2\rho'^2}{|r|+n\rho'^2})d\rho'^2
2555: +\f{\rho'^2(n\rho'^2+|r|)}{n(n+1)\rho'^2+|r|}d\theta^2.}
2556: In the IR limit $r\to -\infty$ (or $|r|>>\rho'$) 
2557: we obtain the metric of ${\bf C}$
2558: \eqn\irm{
2559: ds^2=d\rho'^2+\rho'^2d\theta^2.}
2560: On the other hand, in the far region $\rho'>>|r|$, the metric looks like
2561: ${\bf C}/{\bf Z}_{n+1}$
2562: \eqn\flat{
2563: ds^2=(n+1)d\rho'^2+\f{\rho'^2}{n+1}d\theta^2.}
2564: 
2565: \subsec{Qualitative Properties of the Exact Flow}
2566: 
2567: The evolution of the sigma model metric behind occurs in two stages. When  
2568: $r>0$, the metric remains almost constant. More precisely ${\bf C/Z}_n$ 
2569: is deformed to ${\bf C / Z}_{n+1}$ at very large distances; in \sigmaone\
2570: this deformation propagates inwards. This propagation is clearly reversed
2571: by quantum corrections (a bubble of ${{\bf C/Z}_n}$ immersed in 
2572: ${\bf C / Z}_{n+1}$ grows rather than shrinks in worldsheet RG flow 
2573: \AdPoSi, \GuHeMiSc.), and is an artifact of our approximation; in the
2574: quantum corrected RG flow we expect the metric to stay constant for $r>0$.
2575: 
2576: As $r$ turns negative, however, \sigmatwo\ describes a bubble of flat
2577: space which is nucleated at $\rho=0$. This bubble then expands out
2578: in a diffusive manner; its radius grows like 
2579: $R \sim \sqrt{r} \sim  \sqrt{ (n-1) \ln \Lambda}$, in rough agreement with 
2580: the one loop sigma model growth discussed in \AdPoSi,
2581: \GuHeMiSc. Consequently, the evolution of the sigma model described behind 
2582: (together with the minor expected quantum corrections) is summarized in 
2583: Fig. 6, and is in perfect agreement with the APS picture reviewed in B.1. 
2584: 
2585: \fig{Change of geometry under closed string tachyon in
2586: ${\bf C}/{\bf Z}_{n}.$}{orbd.eps}{6truein}
2587: 
2588: 
2589: 
2590: \appendix{C}{Classical Metric in Twisted Circle GLSM}
2591: 
2592: As in Appendix B, let us compute the classical metric of
2593: the gauged linear sigma model in section 4.3, which has
2594: the kinetic terms
2595: \eqn\kinem{\eqalign{
2596: &L_{kin}=\f{k}{2}(\de_{\mu}{P_1})^2+\f{k}{2}(\de_{\mu}{P_2}+v_\mu)^2
2597: +(\de_{\mu}\rho)^2+\rho^2(\de_{\mu}\theta_1+v_{\mu})^2 \cr
2598: &\ \ \ \ +(\de_{\mu}\rho')^2+\rho'^2(\de_{\mu}\theta_n-nv_{\mu})^2, \cr}}
2599: and the D-term condition $n\rho'^2-\rho^2=kP_1$ (or \zet).
2600: Just in the same way as in B.7 we can integrate out the 
2601: gauge field. After we impose the gauge fixing condition
2602: $\theta_n=0$ from the beginning (just for simplicity)
2603: \eqn\vmum{
2604: v_\mu=-\f{\rho^2 \de_{\mu} \theta_1+\f{k}{2}\de_{\mu}P_2}
2605: {\rho^2+n^2\rho'^2+\f{k}{2}},}
2606: we get the sigma model metric
2607: \eqn\sinem{\eqalign{
2608: &ds^2=(d\rho)^2+(d\rho')^2+\f{k}{2}(dP_1)^2 \cr
2609: &\ \ +\!\f{\rho^2(n^2\rho'^2\!+\!\f{k}{2})(d\theta_1)^2\!-\! k\rho^2
2610: d\theta_1 dP_2\!+\!\f{k}{2}(\rho^2\!+\! n^2\rho'^2)(dP_2)^2}
2611: {\rho^2+n^2\rho'^2+\f{k}{2}}. \cr }}
2612: First we assume $P_1>0$ then we have ($\rho\equiv |\ph_1|$)
2613: \eqn\mets{\eqalign{
2614: ds^2&=(d\rho)^2+\f{k}{2}(dP_1)^2+\f{(\rho d\rho
2615: +\f{k}{2}dP_1)^2}{n(\rho^2+kP_1)}\cr
2616: &+\!\f{(n\rho^4+(nkP_1\!+\!\f{k}{2})\rho^2)(d\theta_1)^2
2617: \!-\!k\rho^2(d\theta_1)(dP_2)\!+\!\f{k}{2}((n+1)\rho^2\!+\!nkP_1)(dP_2)^2}
2618: {(n+1)\rho^2+nkP_1+\f{k}{2}}.\cr}}
2619: In the UV limit $\rho<<k P_1$ we obtain
2620: \eqn\metss{
2621: ds^2=(d\rho)^2+\rho^2(d\theta_1)^2+
2622: \f{k}{2}(dP_1)^2+\f{k}{2}(dP_2)^2,}
2623: which is equivalent to the twisted circle ${\bf C\times S^1}/{\bf Z}_n$
2624: after we take the remained ${\bf Z}_n$ gauge equivalence
2625: $(\theta_1,P_2)\sim (\theta_1+\f{2\pi}{n},P_2+\f{2\pi}{n})$
2626: into account. On the other hand in the opposite limit $\rho^2>>k P_1$
2627: we get
2628: \eqn\mettt{
2629: ds^2\simeq \f{n+1}{n}(d\rho)^2+\f{n}{n+1}\rho^2(d\theta_1)^2
2630: +\f{k}{2}(dP_1)^2+\f{k}{2}(dP_2)^2-\f{k}{n+1}d\theta_1 dP_2,}
2631: which may be described by a twisted circle
2632:  ${(\bf C\times S^1)}/{\bf Z}_{n+1}$. This geometry
2633: is also argued to be an artifact of the classical approximation. If we
2634: take the quantum correction into account, the asymptotic 
2635: region should always 
2636: be the previous twisted circle 
2637: ${(\bf C\times S^1)}/{\bf Z}_{n}$.
2638: 
2639: Finally in the case of $P_1<0$ the result is given by 
2640: ($\rho'\equiv |\ph_{-n}|$)
2641: \eqn\metf{\eqalign{
2642: &ds^2=(d\rho')^2+\f{k}{2}(dP_1)^2+\f{(n\rho' d\rho'
2643: +\f{k}{2}dP_1)^2}{n\rho'^2+kP_1}\cr
2644: &+\!\f{(n^2\rho'^2\!+\!\f{k}{2})(n\rho'^2\!+\!kP_1)(d\theta_1)^2
2645: \!-\!k(n\rho'^2\!+\!kP_1)(d\theta_1)(dP_2)\!
2646: +\!\f{k}{2}(n(n+1)\rho'^2\!+\!kP_1)(dP_2)^2}
2647: {n(n+1)\rho'^2+kP_1+\f{k}{2}}.\cr}}
2648: If we take the limit $\rho'^2<<k |P_1|$, then the metric becomes
2649: \eqn\metpo{
2650: ds^2=d\rho'^2+n^2\rho'^2 d\theta_1^2+\f{k}{2}(dP_2-d\theta_1)^2,}
2651: and thus we get the flat space
2652: ${\bf C\times S^1}$. This gives the bubble of flat space. It is also easy to
2653: see that
2654: the opposite limit $\rho'^2>>k |P_1|$ corresponds to
2655: ${(\bf C\times S^1)}/{\bf Z}_{n+1}$ as before.
2656: 
2657: 
2658: \appendix{D}{D-brane Spectrum in dual LG Theory}
2659: 
2660: 
2661: \subsec{LG Theory dual to GLSM}
2662: 
2663: In \HoVa\ Hori and Vafa 
2664: have argued that a gauge linear sigma model with $p$ different
2665: matter chiral multiplets is exactly dual to a Landau Ginzburg model
2666: of $p+1$ interacting twisted chiral multiplets. The first of these
2667: twisted chiral multiplets is simply the gauge multiplet $\Sigma$.
2668: The remaining twisted chiral multiplets are obtained as follows. Each of
2669: the chiral multiplets of the GLSM may be written as $\ph_i=e^{\psi_i}$,
2670: where the imaginary part of $\psi_i$ is periodic with period $2 \pi$.
2671: We now, roughly, T-dualize this GLSM with respect to each of these periodic
2672: variables. The dual angular coordinate pairs up with the real part of $\psi$
2673: and the original fermions to form a twisted chiral multiplet $Y_i$\foot{This
2674: follows from the fact that T-duality reverses the sign of the left moving
2675: part of the angular variable; this exchanges the left moving part of
2676: $\psi_i$ with its complex conjugate, but has no such effect on the right
2677: moving part.}. While the imaginary part of $Y_i$ is a gauge invariant and
2678: supersymmetrized version of the T-dual of the
2679: imaginary part of $\psi $, it turns out that the real part of the twisted
2680: chiral multiplet 
2681: is given by $Y_i+{\bar Y_i}  =2 {\bar \ph_i}e^{2 Q V} \ph_i$
2682: (see eqns 3.18-3.19, \HoVa, for more details). The twisted superpotential
2683: for these fields that follows from the usual T-duality arguments is
2684: $\Sigma (\sum_{i} Q_i Y_i -t)$ where $Q_i$ are the charges of the fields
2685: $\ph_i$. In addition to this classical contribution, the twisted chiral
2686: potential also receives instanton (vortex) contributions; as shown in 
2687: \HoVa\ the exact quantum corrected twisted superpotential is
2688: the classical piece plus $\sum_{i}e^{Y_i}$.
2689: 
2690: The kinetic terms for these twisted chiral 
2691: multiplets $Y_i$ and $\sigma$ cannot
2692: be computed exactly. However, classically, at low energies ($e \to \infty$)
2693: integrating the $\sigma$ field out simply results in the constraint
2694: $\sum_{i} Q_i Y_i +t=0$ on the $Y_i$ fields. This constraint, the mirror
2695: of the GLSM $D$ term equation, is expected to be exact at low energies.
2696: Consequently, the low energies, we have the Landau Ginzburg theory of
2697: a $p-1$ twisted chiral multiplets ($Y_i$ subject to the constraint described
2698: above) and the twisted chiral superpotential $\sum_{i}e^{Y_i}$.
2699: 
2700: Now let us apply this procedure to the GLSM \act\ for
2701: ${\bf C}/{\bf Z}_n$ in appendix B and
2702: its Liouville version (for twisted circle) in section 3.1. 
2703: Performing the mirror transformation
2704:  we obtain the following LG potential for the former model \Va\
2705: \eqn\sp{
2706: W=e^{-Y_1}+e^{-Y_2}=e^{-nU}+e^{-t/n-U},}
2707: where $Y_1$ and $Y_2$ are dual to $\ph_{1}$ and $\ph_{-n}$ satisfying
2708: the relation $-Y_1+nY_2=t$. In other words, we can write
2709: $Y_1(=nU)=|\ph_{1}|^2+i\phi_{1},\ \ \
2710:  Y_2=|\ph_{-n}|^2+i\phi_{-n},$
2711: where the phases $\phi_{1}$ and 
2712: $\phi_{-n}$ are dual to $\theta_1$ and
2713: $\theta_{n}$.
2714: The periodicity of these fields 
2715: $Y_{1,2}\sim Y_{1,2}+2\pi i$ leads\foot{Here
2716: we used the constraint $-Y_1+nY_2=t$ and thus the periodicity of $U$ is
2717: not $2\pi i/n$ but $2\pi i$.} to the identification
2718: $U\sim U+2\pi i$.
2719: Note that we can see the symmetry ${\rm Im}t(=\theta)\to{\rm Im}t+2\pi i$ 
2720: by considering
2721: the accompanied field redefinition $U\to U-2\pi i /n$.
2722: 
2723: Now let us consider the change of geometry 
2724: from the viewpoint of the LG theory.
2725: Before the tachyon condensation $(r\to \infty)$, we obtain the simple
2726: potential
2727: \eqn\LGUV{
2728: W=e^{-nU},}
2729: and this has the ${\bf Z}_n$ quantum symmetry 
2730: \eqn\qsLG{
2731: h:U \to U+2\pi i/n,}
2732: which corresponds to the previous quantum symmetry \qs\ in GLSM.
2733: Indeed this potential is the same as the large radius limit ($k\to\infty$) 
2734: of the model 
2735: \HoKa\ and can be
2736: identified with the orbifold ${\bf C}/{\bf Z}_n$.
2737: The closed string tachyon condensation is represented by the second
2738: term of \sp\
2739: and
2740: breaks this symmetry. This term represents
2741: the fractional
2742: instanton (vortex) effect in GLSM and is consistent with our arguments 
2743: around \twd . The
2744: final state after tachyon condensation can be regarded as the flat space
2745: ${\bf C}$ as the potential $W\sim e^{-U}$ shows in the $t\to -\infty$ limit.
2746: 
2747: 
2748: It is also possible to perform the mirror transformation for the model
2749: discussed in section 3.1
2750: by using the method in \HoKa . Then we obtain the LG potential \DaGuHeMi\
2751: \eqn\LGM{
2752: W=e^{-nU}+e^{-Y_P /n-U},}
2753: where
2754: the fields $Y_P$ denote the dual twisted chiral
2755: field of $P$. One may ask 
2756: how we can see the (dual of)
2757: twisted identification of ${\bf C}/{\bf Z}_{n}\times {\bf S}^1$ 
2758: for the large
2759: $P_1$.
2760: This is understood as follows. Because we have the 
2761: periodicity $2\pi$ for each of
2762: the imaginary parts of $Y_P,Y_1$ and $Y_2$ with the 
2763: constraint $-Y_1+nY_2=Y_P$, we
2764: find not only
2765: the periodicity $U\sim U+2\pi i{\bf Z}$ and $Y_p\sim Y_p+2n\pi i{\bf Z}$
2766: but also the ${\bf Z}_n$ identification
2767: \eqn\twi{
2768: (U,Y_P)\sim (U+2\pi i/n,Y_P-2\pi i).}
2769: This projection \twi\ just corresponds to the twisted identification (see
2770: Appendix E).
2771: Because of this monodromy the true radius is 
2772: $\ti{R}_{UV}=\f{n\sqrt{\al}}{\sqrt{k}}$.
2773: Thus it is obvious that we obtain the twisted circle in the UV limit.
2774: On the other hand, in the IR limit the dominant term becomes the second
2775: term and thus we get the flat space\foot{
2776: If we consider $P_1$ takes a large negative value,
2777: then we should use the good
2778: variables $Y_1,Y_p$ as $W=e^{-nY_1+Y_p}+e^{-Y_1}$ and thus
2779: we have no such twisted
2780: identification ($\ti{R}_{IR}=\sqrt{\f{\al}{k}}$).}.
2781: 
2782: \subsec{Description of D-branes in LG Theory dual to Orbifold}
2783: 
2784: In this subsection we will investigate the D-brane spectrum 
2785: of ${\bf C/ Z}_n$ in the dual Landau Ginzburg model \LGUV. 
2786: We will use the constraints from ${\cal N}=2$ supersymmetry 
2787:  \HoIqVa \Goa\ in what follows.
2788: 
2789: 
2790: We concentrate on D1-branes which preserve the A-type supersymmetry 
2791: (A-brane). The general arguments of \HoIqVa \Goa, demonstrate that 
2792: such D-branes have a (middle dimensional) Lagrangian sub manifold as 
2793: their world-volume; further the imaginary part of the superpotential is 
2794: constant along this world-volume. The first condition is satisfied by 
2795: any curve in the $U$ plane. The second one is crucial and is less easily
2796: satisfied. We first consider the 1-branes that are mirror (T-dual) to the 
2797: GLSM D2-branes $|m\lb_{D2}$ of subsection 4.1. Note that
2798: we T-dualize along the Im$U$ direction to map between these pictures, 
2799: consequently the branes we are after are localized in the $Im U$
2800: direction. They are given by \foot{Even though
2801: one can consider other possibilities of curved world-line, they do not
2802: satisfy the boundary conformal invariance in the UV limit.}
2803: \eqn\lgdd{
2804: {\rm{Im}}U=\f{2\pi im}{n},}
2805: where the numbers $m=0,1,\ddd,n-1$ label the types of 
2806: D1-branes\foot{One may note that the supersymmetric
2807: condition allows the half-integer value of $m$. There are related
2808: to integer $m$ by the operation $(-1)^{F_L}:U\to -U$. Since the
2809: boundary state is invariant under GSO projection $\f{1+(-1)^{F_L}}{2}$,
2810: we should add the brane $m+n/2$ to $m$ in \lgdd.
2811: This complication does not change our arguments later and we will not 
2812: explicitly mention it below.}. We denote this
2813: world-volume \lgdd\ by $\gamma_{m}$ (see Fig.7 below).
2814: Indeed the ${\bf Z}_{n}$ quantum symmetry action
2815: $h:{\rm Im}U \to {\rm Im}U+2\pi i/n$ 
2816: generates the other D1-branes in the same way as in GLSM.
2817: 
2818: \fig{The LG description of D2-brane.}{LGD2.eps}{2.5truein}
2819: 
2820: We now turn to a discussion of D1-branes that are dual to D0-branes. 
2821: The direct determination of such branes is complicated by the following
2822: fact. The Landau Ginzburg model that reduces to 
2823: ${{\bf C/Z}_n}$ has vanishing kinetic term (this is the $k\to \infty$ limit
2824:  in \HoKa). The Liouville potential, is, consequently effectively 
2825: infinite when Re$U$ is negative, and is effectively zero 
2826: when Re$U$ is positive. For any finite value of the the LG kinetic term 
2827: (if $k$ in \HoKa\ is not strictly infinite), the wrapped 1-branes of the 
2828: Landau Ginzburg model \sp\ are attracted to, and stuck at, the 
2829: strong coupling region. Strictly in the $k \to \infty$ limit this 1-brane 
2830: is free to wander in the bulk. Thus we do not expect to find any ${\bf
2831: C/Z_n}$ brane, localized in the $ReU$ direction, using the usual 
2832: Landau Ginzburg techniques (which work when the LG kinetic term is 
2833: nonsingular). Nonetheless it is not difficult to build a 
2834: qualitative picture of
2835: the states we expect. 
2836: 
2837: As discussed in section 5, GLSM 0-branes may be obtained from 
2838: GLSM $D2-\overline{D2}$ by the open string tachyon condensation. 
2839: A bulk D0-brane is a vortex on a D2 and an anti D2-brane of the same sort; 
2840: consequently it is natural to conjecture that the world volume of such a
2841: brane is a circle along the Im$U$ direction (see the left-hand side of
2842: Fig.8). Indeed it has the correct
2843: D0-brane charge after the T-duality and is obviously movable away from the
2844: fixed point in the $k \to \infty$ limit\foot{
2845: One may think that this configuration of a D0-brane breaks the A-type
2846: supersymmetry because it does not satisfies the condition of
2847: superpotential mentioned before. However, our non-compact orbifold
2848: corresponds to the limit of vanishing the kinetic term of the
2849: Liouville theory as we have noted. Thus the Liouville
2850: potential can be neglected for Re$U>0$ and it will keep the A-type
2851: supersymmetry. This is also equivalent to the fact that the dilaton
2852: is constant in the limit $k\to\infty$.}.
2853: On the other hand, we can generate a fractional D0-brane
2854: $|m\lb_{D0}$ from a D2 $|m+1 \lb_{D2}$ and a $\overline{D2}$
2855: $|\bar{m} \lb_{\bar{D2}}$. In this case we argue that its world-volume
2856: after the open string tachyon condensation is (approximately)
2857: given by (see the right-hand side of Fig.8)
2858: \eqn\lgdez{
2859: \{U=x\ |x<0\}\cup \{U=\f{2\pi im}{n}+i\theta\ |0\leq
2860: \theta\leq \f{2\pi}{n}\}\cup
2861: \{U=y+\f{2\pi i}{n}\ |y<0\}.}
2862: It seems reasonable that annihilation (open string tachyon condensation)
2863: of the $|\bar{m} \lb_{\bar{D2}}$ occurs only for
2864: positive Re$U$ because the existence of the Liouville potential; thus this
2865: open string tachyon condensation leaves behind a remnant stuck at $U=0$; 
2866: this remnant is the T-dual of the fractional 0-brane. Note, of course,
2867: that it winds a fraction of the Im$U$ circle. 
2868: 
2869: \fig{The LG description of D0-brane.}{LGD0.eps}{4truein}
2870: 
2871: 
2872: \subsec{Computation of RR-charges}
2873: 
2874: To check these interpretations, let us compute the coupling to
2875: the RR-sector ground state by using the information of superpotential.
2876: 
2877: In the papers \HoIqVa \Goa\  D1-branes in
2878: the LG theory
2879: of $W=\Phi^{k+2}$, which is equivalent to the ${\cal N}=2$ minimal 
2880: model in the
2881: IR limit, were
2882: analyzed and
2883: we can use this result by performing the change of the variable
2884: $\Phi\to e^{-U}$.
2885: 
2886: First we start with the D1-brane given by \lgdd . The couplings to
2887: the $k^{th}$ twisted RR-sector ground state 
2888: (${\cal N}=2$ chiral state) can be expressed \foot{
2889: Here the sector $k=0$ corresponds to the untwisted sector and its
2890: coefficient is divergent due to the infinite volume. Below we can
2891: omit this sector because it does not contribute to
2892: the present calculations.} by the following `boundary state'
2893: \eqn\bdd{
2894: |m \lb_{D2}=\sum_{k=0}^{n-1}\f{c^{0}_{k,m}}{N_k}|e^{-kU}\lb,}
2895: where we can compute the coefficient $c^{0}_{k,m}$ as follows
2896: \eqna\cofb
2897: $$\eqalignno{
2898: c^{0}_{k,m}
2899: &=\int_{\gamma_{m}}dUe^{-kU}e^{-W} \cr
2900: &=e^{-2\pi im k/n}\int^{\infty}_{-\infty} dx e^{-kx}e^{-e^{-nx}}=
2901: \f{e^{-2\pi im k/n}}{n}\ \Gamma(k/n).\cr}$$
2902: The normalization $N_k$ of the RR-sector ground state $|e^{-kU}\lb$
2903: can also
2904: determined by using the intersection number (see the later arguments and
2905: also \HoIqVa (p.50) for more detail)
2906: \eqn\ints{
2907: N_k=\la \ov{e^{-kU}}|e^{-kU}\lb=\f{2\sin (\pi k/n)}{n}
2908: \Gamma(k/n)^2.}
2909: Thus we get the normalized couplings
2910: \eqn\lgdtb{
2911: c_{k,m}(=\f{c^{0}_{k,m}}{\sq{N_k}})=
2912: \f{1}{\sq{2n\sin(\pi k/n)}}e^{-2\pi im k/n}.}
2913: This result matches 
2914: exactly with the D2-brane
2915: boundary state in
2916: the orbifold theory \ta\ including the normalization.
2917: 
2918: Next we turn to D1-branes which are dual to D0-branes in GLSM.
2919: Their couplings to the RR-sector ground states $c'_{k,\ap}$ do not change
2920: under open string tachyon condensation 
2921: and thus we can compute it as follows
2922: \eqn\lgdzb{
2923: c'_{k,m}=c_{k,m+1}-c_{k,m}=
2924: \sq{\f{2\sin(\pi k/n)}{n}}e^{-\pi i(k/n+1/2)}\cdot e^{-2\pi im k/n}.}
2925: Thus we again find the exact agreement with the result of the
2926: boundary states in
2927: the orbifold theory \DiGo\ \Bi\ up to an irrelevant phase factor.
2928: Note also that the intersection numbers $I_{m,m'}$
2929: (or equally open string
2930: Witten index Tr$_{R}(-1)^F$ \DoFi) of these D1-branes are
2931: given by the overlap of the two boundary state with the
2932: insertion\foot{
2933: In the case of ${\cal N}=2$ minimal model we should take
2934: $(-1)^{F_L}=ie^{\pi ik/n}$ and the index is slightly different from our
2935: model.} of
2936: $(-1)^{F_L}=ie^{\pi ik}$ as follows
2937: \eqna\indo
2938: $$\eqalignno{
2939: \la m|(-1)^{F_L}|m' \lb_{D0}
2940: &=\sum_{k=1}^{n-1}ie^{\pi ik/n}\cdot \f{2\sin(\pi k/n)}{n}
2941: \cdot e^{\f{2\pi i}{n}(m-m')} \cr
2942: &=\delta_{m,m'+l}-\delta_{m,m'+l+1},\cr}$$
2943: and indeed reproduce the quiver theory 
2944: result\foot{This is easy to see from the fact that the ${\bf Z}_n$
2945: projection is given by $e^{\f{2\pi i}{n}(m-m')}\cdot 
2946: e^{\f{2\pi i(n+1)}{n}s_1}=1$,
2947: where $s_1=\pm \f{1}{2}$ is the (spacetime) spin of fermionic
2948: ground states in ${\bf C}/{\bf Z}_{n}$.}.
2949: 
2950: \subsec{Speculations on D-branes in the LG Model}
2951: 
2952: Consider the LG dual to the ${\bf C/Z}_n$ 2-brane $|0\rangle$ described in 
2953: subsection 4.2. As we have described above, the Landau Ginzburg description
2954: of this state is a 1-brane located at Im$U$=0 (assuming $\theta=0$). 
2955: In fact Im$U$=0 continues to 
2956: preserve ${\cal N}=2$ supersymmetry even upon turning on the tachyon
2957: superpotential in \sp; so the mirror to $|0\rangle$ evolves into the mirror
2958: of a bulk 2-brane in flat space, in agreement with our previous results.
2959: 
2960: The situation is not as simple for other 2-branes in \lgdd. On turning on
2961: the tachyon contribution to \sp, the world line of these branes begins to
2962: bend inwards towards Im$U$=0 (see Fig. 9) (this follows from the condition
2963: that the imaginary part of the superpotential remain constant). However, as
2964: the closed string tachyon is localized, at any finite $r$ the effect of
2965: tachyon condensation should be restricted to the region 
2966: $Re$U$ \approx \sqrt{|r|}$. 
2967: Consequently the world line of the 1-brane may be
2968: expected to curve back up to its initial value (see the right end of
2969: Fig. 9). This configuration naively breaks ${\cal N}=2$ 
2970: worldsheet supersymmetry,
2971: but, as in subsection D.2, this breaking may vanish in the $k \to \infty$
2972: limit. If these speculations are indeed true, they would also explain
2973: the disappearance of the fractional 0-branes \lgdez\ under tachyon 
2974: condensation. However we leave the verification (or otherwise) of these
2975: speculations to future work.
2976: 
2977: \fig{The D2-brane in LG description after the RG-flow.
2978: 	We show the world-volume of both D2-branes $|m \lb_{D2}$ and
2979: $|0 \lb_{D2}$. Even though the former D2-brane will be deformed
2980: for $-|r|/n(n-1)<$Re$U<|r|/n$, where the Liouville potential
2981: $e^{-U}$ dominates the superpotential, it does not change
2982:        in the asymptotic region Re$U>|r|/n$. Finally we can see that
2983: in the limit $r\to -\infty$ all these fractional
2984: 	D2-branes will become the same usual D2-brane in flat space.
2985:        (In this picture we assumed $\theta=0$ for simplicity.)}
2986: {D1RG.eps}{3truein}
2987: 
2988: 
2989: \appendix{E}{Useful T-duality Relations}
2990: 
2991: Here we would like to summarize the various T-dual expressions
2992: of the twisted circle $({\bf C}\times {\bf S^1})/{\bf Z}_n$ model.
2993: We denote the ${\bf Z}_n$ projection
2994: by $g=\exp(\f{2\pi i(n+1)}{n}J_{12})$ and its dual ${\bf Z}_n$
2995: quantum symmetry \foot{Note that this includes a
2996: particular example of $g=(-1)^{F_S}$ and $h=(-1)^{F_L}$ for ${\cal N}=2$
2997: (see also \bega, \CoGu, \AdPoSi ).}
2998:  by $h$ \qs . The $1/n$ shift is also
2999: represented by $\sigma_{1/n}$ and its dual operator by
3000: $\tilde{\sigma}_{1/n}$. Then we obtain the following equivalent
3001: descriptions
3002: \eqna\tdualr
3003: $$\eqalignno{
3004: &{\rm{IIA(B)\ \ \  Melvin\ \  Model\ \  (radius}}\ \  R\ \  {\rm{and}} \ \
3005: \zeta(=qR)=(n+1)/n) &\tdualr a\cr
3006: & \ \ \ \simeq
3007: {\rm{IIA(B)\ \ on\ \  }} [{\bf C}\times {\bf S^1}\ \
3008: ({\rm{radius}}\ \  nR)]/(g\cdot \sigma_{1/n})\ \ \
3009: {\rm{(=twisted\ \  circle\ \ )}}  &\tdualr  b\cr
3010: & \ \ \ \simeq {\rm{IIA(B)\ \  on\ \  }} [{\bf C}/{\bf Z}_n
3011: \times {\bf S^1}\ \ ({\rm{radius }}\ \ R)]/
3012: (h\cdot\ti{\sigma}_{1/n})  &\tdualr c\cr
3013: & \ \ \ \simeq
3014: {\rm{IIB(A)\ \  Melvin\  Model\ \  (radius\ \  1/R\ \
3015:  and}}\ \  \beta\al R=(n+1)/n)  &\tdualr  d\cr
3016: & \ \ \ \simeq {\rm{IIB(A)\ \  on\ \  }} [{\bf C}\times {\bf S^1}\ \
3017: ({\rm{radius }}\ \  1/nR)]/(g\cdot \ti{\sigma}_{1/n})  &\tdualr e\cr
3018: & \ \ \ \simeq {\rm{IIB(A)\ \  on\ \  }} [{\bf C}/{\bf Z}_n
3019: \times {\bf S^1}\ \ ({\rm{radius }} \ 1/R)]/
3020: (h\cdot \sigma_{1/n}), &\tdualr f \cr }$$
3021: where the parameter $\beta$ corresponds to the B-field flux T-dual to the 
3022: twisted parameter $q=\zeta/R$ (see \tse \TaUeo).
3023: These are all explained as T-duality relations.
3024: 
3025: Finally let us consider the relation to LG theory. As we have seen, the UV
3026: region ($Y_p\to\infty$) of the
3027: LG theory \LGM\ is equivalent to the twisted circle. This is
3028: understood in the above picture as follows. This LG theory is defined by
3029: the superpotential \LGUV , which is equivalent to
3030: the orbifold ${\bf C}/{\bf Z}_n$, and by the twisted
3031: identification \twi\
3032: \eqn\wlg{
3033: {\rm{typeIIA(B)\ on}}\   [{\bf C}/{\bf Z}_n
3034: \times {\bf S^1}\ \ ({\rm{radius }}\  1/R)]/
3035: (g'\cdot\sigma_{1/n}),}
3036: where the ${\bf Z}_n$ action $g'$ denotes the shift $U\to U+2\pi i/n$.
3037: If we take T-duality in the angular direction of ${\bf C}$, then
3038: we have the projection $h$ instead of $g'$.
3039: Therefore this model is indeed
3040: equivalent to the expression \tdualr{f}.
3041: 
3042: \listrefs
3043: 
3044: \end
3045: