1:
2: % ****** Start of file apssamp.tex ******
3: %
4: % This file is part of the APS files in the REVTeX 4 distribution.
5: % Version 4.0 of REVTeX, August 2001
6: %
7: % Copyright (c) 2001 The American Physical Society.
8: %
9: % See the REVTeX 4 README file for restrictions and more information.
10: %
11: % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed
12: % as well as the rest of the prerequisites for REVTeX 4.0
13: %
14: % See the REVTeX 4 README file
15: % It also requires running BibTeX. The commands are as follows:
16: %
17: % 1) latex apssamp.tex
18: % 2) bibtex apssamp
19: % 3) latex apssamp.tex
20: % 4) latex apssamp.tex
21: %
22: %\documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
23: \documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
24:
25: % Some other (several out of many) possibilities
26: %\documentclass[preprint,aps]{revtex4}
27: %\documentclass[preprint,aps,draft]{revtex4}
28: %\documentclass[prb]{revtex4}% Physical Review B
29:
30: \usepackage{graphicx}% Include figure files
31: \usepackage{dcolumn}% Align table columns on decimal point
32: \usepackage{bm}% bold math
33:
34: %\nofiles
35:
36:
37: %\renewcommand{\theequation}{\thesection.\arabic{equation}}
38: \newcommand{\be}{\begin{equation}}
39: \newcommand{\ee}{\end{equation}}
40: \newcommand{\bear}{\begin{eqnarray}}
41: \newcommand{\eear}{\end{eqnarray}}
42: \newcommand{\ba}{\begin{array}}
43: \newcommand{\ea}{\end{array}}
44: \newcommand{\lae}{\begin{array}{c}\,\sim\vspace{-28pt}\\< \end{array}}
45: \newcommand{\gae}{\begin{array}{c}\,\sim\vspace{-28pt}\\> \end{array}}
46: \newcommand{\tg}{\widetilde{g}}
47: \newcommand{\ov}{\overline}
48: \def\vbr{$\vphantom{\sqrt{F_e^i}}$}% vertical brace for tables
49:
50: \def\aa{\alpha}
51: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
52: \font\fiverm=cmr5
53: % the stuff below defines \eqalign and \eqalignno in such a
54: % way that they will run on Latex
55: \newskip\humongous \humongous=0pt plus 1000pt minus 1000pt
56: \def\caja{\mathsurround=0pt}
57: \def\eqalign#1{\,\vcenter{\openup1\jot \caja
58: \ialign{\strut \hfil$\displaystyle{##}$&$
59: \displaystyle{{}##}$\hfil\crcr#1\crcr}}\,}
60: \newif\ifdtup
61: \def\panorama{\global\dtuptrue \openup1\jot \caja
62: \everycr{\noalign{\ifdtup \global\dtupfalse
63: \vskip-\lineskiplimit \vskip\normallineskiplimit
64: \else \penalty\interdisplaylinepenalty \fi}}}
65: \def\eqalignno#1{\panorama \tabskip=\humongous
66: \halign to\displaywidth{\hfil$\displaystyle{##}$
67: \tabskip=0pt&$\displaystyle{{}##}$\hfil
68: \tabskip=\humongous&\llap{$##$}\tabskip=0pt
69: \crcr#1\crcr}}
70: % The oldref and fig macros are for formatting
71: % references and figure lists at the end of the paper.
72: % If you type \oldref{1}Dirac, P.A.M. you will get
73: % [1] Dirac, P.A.M.
74: % Same goes for \fig except you get Figure 2.1
75: \def\oldrefledge{\hangindent3\parindent}
76: \def\oldreffmt#1{\rlap{[#1]} \hbox to 2\parindent{}}
77: \def\oldref#1{\par\noindent\oldrefledge \oldreffmt{#1}
78: \ignorespaces}
79: \def\figledge{\hangindent=1.25in}
80: \def\figfmt#1{\rlap{Figure {#1}} \hbox to 1in{}}
81: \def\fig#1{\par\noindent\figledge \figfmt{#1}
82: \ignorespaces}
83: %
84: % This defines et al., i.e., e.g., cf., etc.
85: \def\ie{\hbox{\it i.e.}{}} \def\etc{\hbox{\it etc.}{}}
86: \def\eg{\hbox{\it e.g.}{}} \def\cf{\hbox{\it cf.}{}}
87: \def\etal{\hbox{\it et al.}}
88: \def\dash{\hbox{---}}
89: % common physics symbols
90: \def\tr{\mathop{\rm tr}}
91: \def\Tr{\mathop{\rm Tr}}
92: \def\Im{\mathop{\rm Im}}
93: \def\Re{\mathop{\rm Re}}
94: \def\bR{\mathop{\bf R}{}}
95: \def\bC{\mathop{\bf C}{}}
96: %\def\Lie{\mathop{\cal L}} % fancy L for the Lie derivative
97: \def\partder#1#2{{\partial #1\over\partial #2}}
98: \def\secder#1#2#3{{\partial^2 #1\over\partial #2 \partial #3}}
99: \def\bra#1{\left\langle #1\right|}
100: \def\ket#1{\left| #1\right\rangle}
101: \def\VEV#1{\left\langle #1\right\rangle}
102: \def\gdot#1{\rlap{$#1$}/}
103: \def\abs#1{\left| #1\right|}
104: \def\pr#1{#1^\prime}
105: \def\ltap{\raisebox{-.4ex}{\rlap{$\sim$}} \raisebox{.4ex}{$<$}}
106: \def\gtap{\raisebox{-.4ex}{\rlap{$\sim$}} \raisebox{.4ex}{$>$}}
107: % \contract is a differential geometry contraction sign _|
108: \def\contract{\makebox[1.2em][c]{
109: \mbox{\rule{.6em}{.01truein}\rule{.01truein}{.6em}}}}
110: \def\slash#1{#1\!\!\!/\!\,\,}
111: \def\beq{\begin{equation}}
112: \def\eeq{\end{equation}}
113: \def\bea{\begin{eqnarray}}
114: \def\eea{\end{eqnarray}}
115: \def\half{\frac{1}{2}}
116: \def\aeq{\eeq}
117: \def\bq{\begin{quote}}
118: \def\eq{\end{quote}}
119: \def\pr{{\sl Phys. Rev.~}}
120: \def\np{{\sl Nucl. Phys.~}}
121: \def\pl{{\sl Phys. Letters~}}
122: \def\prl{{\sl Phys. Rev. Letters~}}
123: \def \Msol {M_\odot}
124: \def\GeV{\,{\rm GeV}}
125: \def\eV {\,{\rm eV}}
126: \def\Mpc{\,{\rm Mpc}}
127: \def\pc{\,{\rm pc}}
128: \def\half{\frac{1}{2}}
129: %% macros to produce the symbols "less than or of order of"
130: %% and "greater than or of order of" %
131: \def \lta {\mathrel{\vcenter
132: {\hbox{$<$}\nointerlineskip\hbox{$\sim$}}}}
133: \def \gta {\mathrel{\vcenter
134: {\hbox{$>$}\nointerlineskip\hbox{$\sim$}}}}
135: %% a few convenient (?) abbreviations: %
136: \def \endpage {\vfill \eject}
137: \def \endline {\hfill \break}
138: \def \etal {{\it et al.}\ }
139: \relax
140:
141: %\input math_macros
142: %\input prepictex
143: %\input pictex
144: %\input postpictex
145: \newdimen\tdim
146: \tdim=\unitlength
147: \def\bar{\overline}
148: \def\stpltsmbl{\setplotsymbol ({\small .})}
149: \def\tarrow{\arrow <8\tdim> [.3,.6]}
150: \def\moose#1#2#3{\tarrow from #1 to #2 \plot #2 #3 /}
151: \def\mooseb#1#2#3{\tarrow from #3 to #2 \plot #2 #1 /}
152: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
153: \begin{document}
154:
155:
156: \preprint{FERMILAB-Pub-06/010-T}
157:
158: %\vspace{1.5 in}
159:
160: \title{
161: Anomalies, Chern-Simons Terms\\
162: and Chiral Delocalization in Extra Dimensions }
163:
164:
165: \author{Christopher T. Hill}
166:
167: \email{hill@fnal.gov, }
168:
169: \affiliation{
170: {{Fermi National Accelerator Laboratory}}\\
171: {{\it P.O. Box 500, Batavia, Illinois 60510, USA}}
172: }%
173:
174:
175:
176: \date{January 18, 2006}% It is always \today, today,
177: % but any date may be explicitly specified
178:
179: \begin{abstract}
180: Gauge invariant topological interactions, such as the
181: $D=5$ Chern-Simons term, are required in models in
182: extra dimensions that split anomaly free representations.
183: The Chern-Simons term is necessary to maintain the overall
184: anomaly cancellations of the theory, but it
185: can have significant, observable, physical effects.
186: The CS-term locks the KK-mode parity to the
187: parity of space-time, leaving a single parity symmetry.
188: It leads to new processes amongst KK-modes
189: \eg, the decay of a KK-mode
190: to a 2-body final state of KK-modes.
191: A formalism for the effective interaction
192: amongst KK-modes is constructed, and the decay of a KK-mode
193: to KK-mode plus zero mode is analyzed as an example. We elaborate
194: the general KK-mode current and anomaly structure of these theories.
195: This includes a detailed study of the triangle diagrams
196: and the associated ``consistent anomalies'' for Weyl
197: spinors on the boundary branes.
198: We also develop the non-abelian formalism. We illustrate
199: this by showing in a simple way how a $D=5$ Yang-Mills
200: ``quark flavor'' symmetry leads to the $D=4$ chiral lagrangian of mesons
201: and the quantized Wess-Zumino-Witten term.
202:
203:
204:
205: \end{abstract}
206:
207:
208: \pacs{11.10.-z, 11.10.Kk, 11.15.-q, 11.25.Mj, 11.25.Uv, 11.30.Rd, 11.40.-q}
209: % PACS, the Physics and Astronomy
210: % Classification Scheme.
211: %\keywords{Suggested keywords}%Use showkeys class option if keyword
212: %display desired
213: \maketitle
214:
215: \section{\bf Introduction}
216:
217:
218:
219:
220: We explore the physics of a
221: Chern-Simons term in a model of a gauge theory
222: in a compactified extra dimension.
223: A Chern-Simons term
224: leads to new physical processes, \ie, new interactions
225: amongst the KK-modes of gauge fields,
226: characterized by a quantized coefficient.
227: The Chern-Simons term locks KK-mode parity
228: to the parity of space-time via the 5-$\epsilon$ symbol
229: permitting processes in which KK-parity is violated,
230: while KK-parity combined with $D=4$ space-time parity is conserved.
231: That is, KK-mode parity is a spurious symmetry and is independent
232: of space-time parity, until a Chern-Simons term appears, which
233: unites these into a single parity symmetry.
234:
235: The effects of Chern-Simons terms are intertwined with the anomalies
236: of matter fields. In the present discussion
237: matter fields will be localized on boundary branes.
238: Since the Chern-Simons term is a bulk interaction, it probes
239: physics away from the boundary (branes).
240: Taken together, however, the anomalies and Chern-Simons
241: terms produce gauge invariant amplitudes containing
242: new physics. Perhaps the most novel new physics is
243: the decay of
244: a single KK-mode into a two KK-mode final state. We study
245: the latter process in this paper as an explicit example of
246: the formalism.
247:
248:
249: It should not be surprising that
250: there is nontrivial physics associated with the CS term.
251: For example, in the case of $D=3$ QED, the CS term
252: gives the photon a mass \cite{schonfeld}.
253: Moreover, the $D=5$ Yang-Mills CS term for $SU(N)$ can be
254: deformed into a Wess-Zumino-Witten (WZW) term \cite{wess,witten},
255: of an $SU(N)_L\times SU(N)_R$
256: chiral lagrangian of mesons, under certain compactification
257: schemes \cite{zachos}
258: (we give a simple derivation of this in Section V of the present
259: paper \cite{hill4}).
260: Many essential physical processes in QCD, such as
261: $\pi\rightarrow 2\gamma$,
262: $\phi\rightarrow K\bar{K}$, and $\phi\rightarrow 3\pi$, etc.,
263: are controlled by the WZW term. In particular,
264: the pion parity in a chiral lagrangian is defined by the WZW term
265: via the 4-$\epsilon$ symbol
266: and would be a spurious symmetry without it. This is the
267: analogue of the fate of KK-parity
268: in the presence of the CS term in the $D=5$ bulk.
269:
270: Split fermion theories \cite{nima} are
271: compelling in that anomalous representations of chiral fermions occur
272: in a delocalized way in extra dimensions. Given that
273: the standard model involves flavor and chirality in a
274: nontrivial (non-vectorlike) way,
275: it would seem plausible that chiral delocalization
276: would occur if extra dimensions should exist. Chirality is then,
277: in a sense, emergent from the spatial localization in
278: extra dimensions. A key discriminant
279: for such a theory is, indeed, likely to be the KK-mode parity
280: violation through CS term interactions.
281:
282:
283: We study in detail
284: a $U(1)$ theory, but we also indicate
285: how things work in non-abelian theories as well.
286: Our goal is to develop a solid formalism for correctly
287: obtaining gauge invariant amplitudes. In the non-abelian case we
288: use this formalism to demonstrate how
289: the Wess-Zumino-Witten term, with its quantized coefficient, arises
290: by morphing a $D=5$ Yang-Mills theory of quark flavor into
291: a compactified $D=4$ theory, where it becomes a chiral
292: lagrangian of mesons, the ``mesons'' being the $A^5$ gauge fields
293: \cite{zachos}.
294:
295:
296: In $D=5$, for a $U(1)$ gauge theory, there exists a CS term of the form
297: $ \epsilon_{ABCDE}A^A \partial^B A^C \partial^D A^E $, and we'll consider
298: this operator to be part of a lagrangian of a
299: $D=5$ theory compactified to $D=4$ with boundary
300: branes. Under a gauge transformation
301: this term produces anomalies on the boundary branes, which
302: we'll refer to as the ``{\em CS anomalies}." Anomalies represent
303: nonconserved currents, \ie,
304: $\partial_Aj^A =$(anomaly).
305: Any gauge
306: theory would be {\em a priori} inconsistent with such anomalies since the
307: equation of motion, $\partial_A F^{AB} = j^B$, implies
308: $\partial_A j^A= 0 $, by the antisymmetry of $F^{AB}$.
309: We thus require some mechanism to cancel anomalies, and
310: this typically implies something like chiral matter fields attached
311: to different boundary branes that produce their own ``{\em matter anomalies}''
312: \cite{jackiw,adler,bardeen,jackiw2}.
313: The CS anomalies must then cancel against the matter anomalies \cite{skiba}.
314:
315:
316: The relevant fermionic anomalies on the boundaries
317: are the ``consistent anomalies,'' \ie, the form
318: that arises directly from Feynman diagrams.
319: We explicitly verify in detail
320: the results of Bardeen \cite{bardeen} for the consistent anomalies
321: of massless Weyl spinors, (and the massive ``left-right symmetric'' case)
322: arising from the effective action operators
323: in Appendix C. A key point, often confused in the
324: literature, is that
325: the Weyl spinor anomaly (a consistent anomaly) {\em is not
326: ``half the Dirac spinor axial current anomaly of QED,'' } \cite{adler}
327: (the ``covariant anomaly'').
328: The consistent and covariant forms of
329: the anomalies differ by the addition of a counterterm to the action that
330: makes the vector (electromagnetic) current conserved, even in the presence of
331: a background axial vector photon.
332: Remarkably (though perhaps not surprisingly), we find that this counterterm
333: is precisely ``boundary term''
334: that arises when we integrate out the fermions on the boundary branes in the
335: large fermion mass limit. In combination
336: with the CS anomaly, which has the form of
337: the consistent anomaly, this maintains the zero-mode gauge invariance,
338: and it shifts the lowest axial vector current anomaly into the covariant form.
339: We obtain the resulting tower of KK-mode covariant
340: current anomalies in Appendix B.
341:
342: Our formal problem, in part, is
343: to bring the full
344: effective action into a manifestly gauge invariant form.
345: Indeed, one of the beautiful aspects of gauge field theories
346: under compactification
347: is the manner in which the KK-mode mass terms are automatically
348: generated in a manifestly gauge invariant way. The modes $A_\mu^n$
349: become packaged together
350: with their longitudinal components $\partial_\mu A_5^n $, as
351: they acquire mass -- they become gauge invariant
352: ``Stueckelberg fields.''
353: Once we are assured that the action can be brought into Stueckelberg
354: combinations of massive fields,
355: then we are assured of unambigous Feynman rules that respect
356: gauge invariance for the massive fields.
357: We find that a Wilson line gauge transformation that brings
358: $A^5 =0$ in the bulk simultaneously packages
359: all of the massive KK-modes in the CS-term into
360: Stueckelberg form. This leads to a remarkably
361: compact formula for the CS-term part
362: of the effective action. In the case
363: of chiral fermions on the boundaries, the
364: triangle diagrams maintain the zero-mode gauge invariance.
365: A physical amplitude is thus a sum of the CS term contribution
366: in the bulk plus a boundary term that comes from the chiral
367: fermions on the branes.
368: In the case of delocalized chiral fermions with
369: a Wilson line mass term, and a $U(1)$ gauge group,
370: the boundary term arises from triangle diagrams involving the
371: fermions on the branes.
372:
373:
374: We consider
375: two distinct ways to implement matter fields.
376: The conceptually simpler method employs axions, $\phi_L$ and $\phi_R$, on
377: the $x^5=0$ ($L$) and $x^5=R$ ($R$) coupled
378: to $F\tilde{F}(0)$ and $F\tilde{F}(R)$ on
379: their respective boundary branes.
380: This construction involves
381: no triangle diagrams, yet it demonstrates, as a matter of
382: principle that the CS term
383: bulk interactions are physical. However, this theory
384: is somewhat less interesting because here
385: the zero mode photon itself becomes massive, by eating one linear
386: combination, $\phi_L+\phi_R$, of the axions. A massive Stueckelberg
387: photon field now appears explicitly in the CS term.
388: A second, orthogonal, combination $\phi_L-\phi_R$
389: remains as a physical uneaten axion.
390: We discuss this in Appendix A.
391:
392: We are mainly interested
393: in QED in $D=5$ with ``split chiral electrons''
394: on the boundary branes. Though QED
395: is a vectorlike theory, if we want to promote
396: it to $D=5$, yet maintain a naturally small electron mass
397: (compared to $1/R$, the compactification scale),
398: then we require chiral delocalization
399: \eg, split chiral electrons on the boundary branes.
400: We thus place $\psi_L$ on the left-brane $x^5=0$ and $\psi_R$
401: on the right-brane $x^5=R$. The electron mass term is then a
402: bilocal operator involving a Wilson line,
403: $m \bar{\psi}_L\exp(\int i\; dx^5 A_5)\psi_R +h.c.$ where
404: the integral runs from $x^5=0$ to $x^5=R$.
405: This is the technically natural setting for QED in $D=5$, and
406: it necessarily contains the CS-term due to the split
407: fermion representation.
408: The anomalies on the branes cancel the CS term
409: anomalies for a particular
410: choice of the coefficient of the CS term. Thus the
411: CS term is ``quantized.'' In this model the photon stays
412: massless, and the
413: full effective interaction consisting of
414: the CS term, together with the
415: fermion triangle loop contributions is now explicitly gauge invariant
416: for all modes \footnote{
417: Note, of course, that the branes can be explicitly
418: constructed by coupling a $D=5$
419: bulk electron to a kink plus antikink solition.
420: This would yield the brane chiral
421: electrons as domain wall fermions. Our branes are presently
422: imposed by hand, yet the present model can be viewed as a description
423: of domain wall fermions in the limiting case of a thin domain wall,
424: and a large fermionic Dirac mass in the bulk.}.
425:
426: The non-abelian theory can be developed along parallel lines.
427: It can also contain a $\tr(AdAdA) + ...$ CS-term, and boundary brane
428: chiral fermions.
429: We develop the non-abelian case to the point of establishing
430: the quantized coefficient of the CS-term. We then revisit the
431: problem of morphing a $D=5$ Yang-Mills theory of quark flavor into
432: the low energy Wess-Zumino-Witten term by showing how
433: the $\Tr(\pi d\pi d\pi d\pi d\pi )$ term arises with the Witten quantization
434: in a theory with flipped orbifold boundary conditions \cite{zachos}.
435:
436: Most technical details have been
437: relegated to a series of Appendices. The main text attempts
438: to give the sequential arguments in a more conceptual form.
439: We have included relevant background issues concerning anomalies
440: to enhance the reader's familiarity with some of the subtleties.
441:
442:
443:
444:
445: \section{The $U(1)$ Case: General Argument}
446:
447: Consider a $U(1)$ gauge theory
448: in $D=5$, with the covariant derivative, field strength and
449: kinetic term lagrangian density:
450: \bea
451: D_A & = & \partial_A - iA_A \; ,
452: \qquad F_{AB} = i[D_A, D_B]\; ,
453: %\nonumber \\
454: \qquad
455: L_0 = -\frac{1}{4\widetilde{e}^2} F_{AB}F^{AB}\; .
456: \eea
457: We define the gauge fields to have canonical
458: dimension for $D=4$, \ie, $A\sim M^1$,
459: where then $1/\widetilde{e^2}$ has dimension of $M^1$
460: (the e.t.c. is then $[A_i(\vec{x}),\dot{A_j}(\vec{y})] =
461: i\widetilde{e}^2\delta^4(\vec{x}-\vec{y})$).
462:
463: The theory admits
464: a Chern-Simons term, defined by
465: the local lagrangian density:
466: \bea
467: L_{CS} & = & c \;\epsilon^{ABCDE} A_{A}\partial_B A_C\partial_D A_E
468: = \frac{c}{4} \epsilon^{ABCDE} A_{A}F_{BC} F_{DE}
469: \eea
470: We can then define the non-compactified $D=5$ theory by the action
471: $S=S_0 + S_{CS}$ where:
472: \bea
473: \label{SCS}
474: S_0 & = & \int d^5x\; L_0\; , \qquad
475: S_{CS} = \int d^5x\; L_{CS} \; .
476: \eea
477: The variation of the action with respect to $A_A$
478: generates the equation of motion:
479: \beq
480: \widetilde{e}^2\frac{\delta S}{\delta A_A} = \partial_B F^{BA} - J^A = 0.
481: \eeq
482: We see that a conserved ``Chern-Simons current''
483: appears as the source term in the theory,
484: \beq
485: J^A = \frac{3c}{4}\tilde{e}^2 \epsilon^{ABCDE} F_{BC} F_{DE},
486: \qquad \qquad \partial^A J_A=0.
487: \eeq
488: The non-compactified action is readily seen to be gauge invariant,
489: provided we forbid surface terms, \ie, we
490: view all fields as approaching zero sufficiently
491: rapidly at infinity. Under a gauge
492: transformation:
493: \beq
494: A_A \rightarrow A_A + \partial_A \theta
495: \eeq
496: we see that $L_0$ is strictly invariant and we also have upon integrating
497: by parts:
498: \bea
499: \label{test}
500: S_{CS} & \rightarrow & S_{CS} + \frac{c}{4} \int d^5x\; \epsilon^{ABCDE}
501: \partial_{A}\theta \; F_{BC} F_{DE} \nonumber \\
502: & = & S_{CS} - \frac{c}{4} \int d^5x\; \epsilon^{ABCDE}
503: \theta \partial_{A} F_{BC} F_{DE}
504: %\nonumber \\
505: %& = &
506: \;\; = \; S_{CS}.
507: \eea
508:
509: The situation changes, however, when we compactify
510: the theory and must accomodate surface terms.
511: Let us compactify the fifth
512: dimension, $0 \leq x^5 \leq R$. We thus imagine that, located
513: at $x^5=0$ and $x^5 = R$, are surfaces (branes) denoted
514: respectively as $I$ and $II$, upon which we may choose to
515: apply various boundary conditions.
516:
517: For example, if we
518: apply the condition, $F_{\mu 5}|_I = F_{\mu 5}|_{II}= 0$, then we have
519: a ``magnetic superconducting parallel plate capacitor'' \footnote{This
520: is dual
521: to a the conventional electric superconducting parallel plate capacitor;
522: a magnetic superconductor is a confining phase of the theory since it admits
523: electric flux tubes that would confine electric charges}.
524: This corresponds to an orbifold since $F_{\mu 5} =0$
525: can be satisfied with the gauge choice,
526: $\partial_\mu A_5 =0$ and $\partial _5 A_\mu =0 $.
527: This, in turn, implies
528: that $A_5$ is an odd function on the extended interval, $0\leq x^5 \leq 2R$,
529: while $A_\mu$ is even, corresponding to the normal orbifold configurations
530: for a theory compactified on $S_1/Z_2$.
531: The advantage of stating boundary conditions
532: in the ``capacitor'' language is that they are
533: then manifestly gauge
534: invariant, \ie, orbifolding does not
535: break gauge invariance but, rather, corresponds to a particular gauge choice
536: for a physical capacitor system.
537:
538: If we now repeat the check of gauge invariance, we see
539: that there is a surface term generated on each of the branes
540: that follows from performing the integration by parts:
541: \bea
542: \label{test2}
543: S_{CS} & \rightarrow & S_{CS} + \frac{c}{4}\int_{II} d^4x\;
544: \theta(R)\; \epsilon^{\mu\nu\rho\sigma}
545: F_{\mu\nu} F_{\rho\sigma}(R)
546: %\nonumber \\
547: %& &
548: -
549: \frac{c}{4}\int_{I} d^4x\;
550: \theta(0)\; \epsilon^{\mu\nu\rho\sigma}
551: F_{\mu\nu} F_{\rho\sigma}(0)\;.
552: \eea
553: The theory is no longer gauge invariant since the action has
554: shifted by the two surface terms that take the form
555: of anomalies. We refer to these terms as the ``CS anomalies.''
556: If, however, the branes contain additional matter fields with anomalous
557: gauge currents, then they can cancel the above
558: CS anomalies
559: on the branes, and the overall theory, bulk plus branes,
560: becomes gauge invariant.
561:
562: Let us presently assume that we have arranged a generic
563: mechanism of cancelling the CS anomaly.
564: Since we are compactifying the theory on $0\leq x^5 \leq R$, it
565: is useful to put the CS term into a form that consists of
566: two terms, one that isolates $A_5$ and another that isolates $\partial_5$.
567: It is readily seen that $L_{CS}$ can be written as:
568: \beq
569: \label{CS0}
570: L_{CS} = \frac{3c}{4} \epsilon^{\mu\nu\rho\sigma} A_{5}F_{\mu\nu} F_{\rho\sigma}
571: +{c} \;\epsilon^{\mu\nu\rho\sigma} (\partial_{5}A_{\mu}) A_\nu F_{\rho\sigma}\;
572: .
573: \eeq
574: To write $L_{CS}$ in this form we have discarded
575: total divergences in the $D=4$ theory that do not affect
576: the physics.
577: In the compactified case our problem is to bring the CS term into a manifestly
578: gauge invariant form. We will see that this is possible for all massive
579: modes, but not for the zero
580: mode.
581:
582: Consider a Wilson line that emanates from, \eg, brane I, $ x^5 =0$,
583: toward an arbitrary point in the bulk, $x^5 = y$:
584: \beq
585: U(y) = \exp \left( i \int_0^y dx^5 A_5(x^5)\right)
586: \eeq
587: and we have:
588: \beq
589: \partial_y U = iA_5(y)U
590: \eeq
591: Using the Wilson line as a gauge transformation, we have:
592: \beq
593: \label{gt}
594: A_A \rightarrow A_A + iU^\dagger \partial_A U
595: \eeq
596: and we thus see that:
597: \beq
598: A_5 \rightarrow A_5(y) + iU^\dagger \partial_y U =
599: A_5(y) - \partial_y \int_0^y dx^5 A_5(x^5) = 0
600: \eeq
601: Assuming the cancellation of CS anomalies occurs with some
602: generic matter fields, this gauge transformation
603: thus annihilates the first
604: term of eq.(\ref{CS0}), and takes the CS term into the form
605: \footnote{The ``annihilation'' of the first term
606: of eq.(\ref{CS0}) can be seen
607: to occur in detail upon performing the gauge transformation
608: for any particular matter anomaly cancellation with the CS anomaly.
609: It involves, however,
610: both the first ($A^5$), as well as the second
611: ($\partial_5$), terms of eq.(\ref{CS0}). The second term yields
612: $\partial_5\partial_\mu A^5$ terms that must be integrated
613: by parts, and the cancellation is then just the full matter anomaly
614: and CS term anomaly cancellation.}:
615: \beq
616: \label{S1}
617: L_{CS} =
618: {c} \;\epsilon^{\mu\nu\rho\sigma} (\partial_{5}B_{\mu})
619: B_\nu F_{B\rho\sigma}
620: \eeq
621: where we define the gauge transformed $A_\mu$'s as $B_\mu$'s,
622: \beq
623: \label{Stu}
624: B_\mu = A_\mu - \partial_\mu \int_0^y A_5 dx^5; \qquad
625: F_{B\mu\nu} = \partial_\mu B_\nu - \partial_\nu B_\mu .
626: \eeq
627: Thus, the resulting Chern-Simons action takes the form:
628: \bea
629: \label{S2}
630: S_{CS} & = &
631: {c}\;\epsilon^{\mu\nu\rho\sigma} \int d^4 x \int_0^R dy\; (\partial_y B_{\mu})
632: B_\nu F_{B\rho\sigma} .
633: \eea
634: Eq.(\ref{S2}) is the form we desire. As we'll see
635: below, the $B_\mu$ field
636: will lead to gauge invariant (``Stueckelberg'') combinations for each
637: massive KK-mode in the compactified theory when we do the mode expansion.
638: The massless zero mode (photon) gauge field, however, will appear
639: explicitly in the result, and the full zero mode gauge invariance
640: requires the addition of the matter effects (\eg, the triangle diagrams).
641: Note the $\partial_y$ in the integrand which leads to the breaking
642: of KK-parity. This is associated with $\epsilon_{\mu\nu\rho\sigma}$
643: so that, overall, only the product
644: of KK-parity and space-time parity is conserved.
645:
646: We now turn to QED with chiral electrons
647: as a means of providing the anomaly cancellation
648: on the branes.
649: An alternative theory with axions on the branes is developed in Appendix A.
650:
651:
652:
653: \section{$D=5$ QED, Orbifold Compactification}
654:
655:
656: \subsection{Chiral Fermions}
657:
658:
659: Consider QED in $D=5$ on an orbifold
660: with periodic domain $0 \leq x^5 \leq R$. We place
661: electrons on the boundary branes located
662: at $x^5 =0$ and $x^5=R$. The electrons
663: are chiral, with $\psi_L$ ($\psi_R$) on the left-brane, $I$ at $x^5 =0$
664: (right-brane, $II$, at $x^5=R$).
665: These fermions have anomalies on their respective branes.
666: These matter anomalies will cancel the CS anomalies
667: provided the coefficient
668: $c$ takes on a special value dictated by
669: the fermionic anomalies. We first establish this special value
670: of $c$.
671:
672:
673:
674: \begin{figure}[t]
675: \vspace{4.5cm}
676: \special{psfile=branes12.eps
677: angle=0 hscale=60 vscale=60 hoffset=140 voffset=-60
678: % angle=-90 hscale=60 vscale=50 hoffset=0 voffset=240
679: }
680: \vspace{3.5cm}
681: \caption[]{
682: %\small
683: \addtolength{\baselineskip}{-.3\baselineskip}
684: Orbifold with split, anomalous fermions (electrons).
685: $\psi_L$ ($\psi_R$) is attached to the $D=4$ left-brane, $I$ (right-brane,
686: $II$).
687: Gauge fields propagate in the $D=5$ bulk, which has a compactification
688: scale $R$. The bulk contains a Chern-Simons term, and the branes
689: produce triangle diagram amplitudes in the effective action. The anomalies
690: from the Chern-Simons term cancel the anomalies from the triangle diagrams
691: on the respective branes so the overall theory is anomaly free. }
692: \label{dirac2}
693: \end{figure}
694:
695:
696:
697: This model has the same divergence structure
698: for the fermion loops as does ordinary QED in $D=4$.
699: The only nontrivial new bulk
700: interaction is the topological CS term. It is not
701: hard to verify that the topological CS term
702: is not subject to renormalization at the one
703: loop level, from diagrams involving internal
704: gauge fields in the continuum $D=5$ theory.
705: This non-renormalization happens because
706: the CS-term is a topological invariant.
707: This probably holds to all orders in the $D=5$ perturbation
708: theory, and is related
709: to the $D=4$ Adler-Bardeen non-renormalization theorem for the anomaly.
710: The CS term does renormalize the
711: bulk kinetic term with a quadratic divergence,
712: and it may induce additional non-topological
713: counterterms with factors of $\tilde{e}^2$. This is the usual
714: problem of a $D=5$ theory, so we imagine some kind of
715: UV cut-off, such as an overarching string theory as the
716: UV completion.
717:
718: The full
719: action of the theory is defined as,
720: \beq
721: S=S_0 + S_{CS} + S_{branes}
722: \eeq
723: where the bulk kinetic term action for the theory is:
724: \bea
725: S_0 & =& -\frac{1}{4\widetilde{e}^2}\int_0^R dy\int d^4 x F_{\mu\nu}F^{\mu\nu}
726: %\nonumber \\
727: %& & \;\;\;
728: -\frac{1}{2\widetilde{e}^2}\int_0^R dy\int d^4 x
729: F_{\mu 5}F^{\mu 5}.
730: \eea
731: The fermionic matter action on the branes is:
732: \bea
733: S_{branes} & = & \int_I d^4x \;\bar{\psi}_Li\slash{D}_L\psi_L
734: %\nonumber \\
735: %& &
736: + \int_{II} d^4x \;\bar{\psi}_Ri\slash{D}_R\psi_R
737: \eea
738: where:
739: \beq
740: D_{L\mu} = \partial_\mu - iA_\mu(x_\mu, 0)\; ,
741: \qquad \qquad D_{R\mu} = \partial_\mu - iA_\mu(x_\mu, R)
742: \eeq
743: and
744: \beq
745: \psi_L = \frac{1-\gamma^5}{2} \psi \; , \qquad
746: \qquad
747: \psi_R = \frac{1+\gamma^5}{2} \psi
748: \eeq
749: We have thus ``split'' QED into two chiral
750: theories living on distinct branes, $I$ and $II$.
751: The $\psi_L$ and $\psi_R$
752: chiral projections are key ingredients of
753: the theory.
754: This structure can, of course, come
755: about if there is a thin domain wall (kink) at $x^5=0$ and an
756: anti-domain wall (anti-kink) at $x^5=R$, where $\psi_L$ and $\psi_R$
757: are then the fermionic zero modes.
758: We presently have no bulk propagation of the fermions,
759: and this can be engineered
760: if the fermions have a very large Dirac mass in the bulk
761: away from the domain walls.
762:
763: The chiral electrons
764: have anomalies on their respective branes.
765: Consider a gauge transformation in the bulk:
766: \beq
767: \label{gt00}
768: A_A(x_\mu, y) \rightarrow A_A(x_\mu, y)
769: + \partial_A \theta (x_\mu, y)
770: \eeq
771: and we therefore have:
772: \bea
773: \label{shift0}
774: S_{branes} & \rightarrow &
775: S_{branes} + \int_I d^4x \;\bar{\psi}_L\gamma_\mu\partial^\mu\theta\psi_L (x_\mu, 0)
776: %\nonumber \\
777: %& &
778: + \int_{II} d^4x \;\bar{\psi}_R\gamma_\mu\partial^\mu \psi_R(x_\mu, R)
779: \nonumber \\
780: & \rightarrow &
781: S_{branes} -
782: \int_I d^4x \;\theta (x_\mu, 0) \partial_\mu J_L^\mu
783: % \nonumber \\& &
784: - \int_{II} d^4x \; \theta (x_\mu, R) \partial_\mu J_R^\mu
785: \eea
786: where $J_{L,R}^\mu =\bar{\psi}\gamma_\mu \psi_{L,R}$.
787: Note that this
788: can be induced by a gauge transformation
789: of the electrons on the branes,
790: \beq
791: \psi_L \rightarrow \exp(i\theta (x_\mu, 0))\psi_L \; , \qquad
792: \psi_R \rightarrow \exp(i\theta (x_\mu, R))\psi_R \;.
793: \eeq
794: At the quantum loop
795: level this transformation is anomalous, the currents
796: are not conserved. We can view
797: it as generating the
798: Noether terms on the branes of eq.(\ref{shift0}) through
799: the fermionic functional measure of the path integral
800: \cite{fujikawa}.
801: To proceed, we must determine the forms of the anomalies on
802: the branes, and this requires care.
803:
804:
805: \subsection{Consistent Anomalies on the Branes}
806:
807:
808: Consider the theory of a single Weyl spinor in $D=4$:
809: \beq
810: \label{tt0}
811: S = \int d^4x \; \bar{\psi}_L(i\slash{\partial}+\slash{A}_L)\psi_L
812: \eeq
813: The theory is anomalous and the anomaly is
814: unambiguously determined as:
815: \bea
816: \label{bardeen100}
817: \partial_\mu J_L^\mu & = & -\frac{1}{48\pi^2}
818: F_{A_L}^{\mu\nu}\widetilde{F}_{{A_L}\mu\nu}\; , \qquad
819: J^\mu_{L}=\bar{\psi}\gamma^\mu\psi_L\; ,
820: \qquad
821: \makebox{and} \qquad
822: \widetilde{F}_{\mu\nu} = \half \epsilon_{\mu\nu\rho\sigma}{F}^{\rho\sigma}.
823: \eea
824: Moreover, if we have a pair of Weyl spinors in $D=4$:
825: \beq
826: \label{t0}
827: S = \int d^4x \;\left[ \bar{\psi}_L(i\slash{\partial}+\slash{A}_L)\psi_L +
828: \bar{\psi}_R(i\slash{\partial}+\slash{A}_R)\psi_R \right]
829: \eeq
830: then we can treat these fields symmetrically, each as in eq.(\ref{tt0}).
831: The anomalies are given implicitly in Bardeen's paper
832: for the massive case, \cite{bardeen}, and
833: we confirm these results in Appendix C for the massless Weyl spinor
834: theory of eq.(\ref{tt0}:
835: \bea
836: \label{bardeen10}
837: \partial_\mu J_L^\mu & = & -\frac{1}{48\pi^2}
838: F_{A_L}^{\mu\nu}\widetilde{F}_{{A_L}\mu\nu}; \qquad
839: % \nonumber \\
840: \partial_\mu J_R^\mu = \frac{1}{48\pi^2}
841: F_{A_R}^{\mu\nu}\widetilde{F}_{A_R\mu\nu} \;\;\;\;\;
842: \eea
843: where:
844: \beq
845: J^\mu_{L}=\bar{\psi}\gamma^\mu\psi_L,
846: \qquad \qquad
847: J^{\mu}_{R}=\bar{\psi}\gamma^\mu\psi_R\; .
848: \eeq
849: If we construct the vector and axial vector currents,
850: $J=J_L+J_R$, and $J^5=J_R-J_L$, and we define
851: define $A_R=V+A$ and $A_L=V-A$, we have by simple algebra:
852: \bea
853: \label{bardeen11}
854: \partial_\mu J^\mu & = & \frac{1}{12\pi^2}
855: F_V^{\mu\nu}\widetilde{F}_{A\mu\nu}
856: % \nonumber \\
857: \qquad
858: \partial_\mu J^{5\mu} = \frac{1}{24\pi^2}\left(
859: F_V^{\mu\nu}\widetilde{F}_{V\mu\nu} + F_A^{\mu\nu}\widetilde{F}_{A\mu\nu}
860: \right)
861: %\;\;\;\;\;
862: \eea
863: This latter result is the form quoted in
864: eq.(44) of \cite{bardeen}. One should pay particular
865: attention to the coefficients in eq.(\ref{bardeen100}),
866: eq.(\ref{bardeen10}) and eq.(\ref{bardeen11}).
867:
868: This
869: form of the vector and axial-vector current anomalies
870: is known as the {\em ``consistent anomaly,''} as it is consistent with
871: the direct calculation of the triangle diagrams of the Weyl spinors.
872: As stated above, the form of the anomaly for a
873: theory containing {\em only} a single
874: pure left-handed Weyl fermion
875: is unambiguous, given by the first expression
876: in eq.(\ref{bardeen10}). There is no ambiguity when there
877: is only one gauge field $A_L$ and there is no nonvanishing
878: counterterm (a term of the
879: form $\epsilon_{\mu\nu\rho\sigma}A_L^\mu A_L^\nu \partial^\rho A_L^\sigma$
880: is zero)
881: that can be added to the theory to redefine the anomaly.
882:
883: The form in eq.(\ref{bardeen11})
884: is just the sum of left-handed and right-handed
885: Weyl fermion consistent anomalies, and is referred to as the ``left-right
886: symmetric
887: anomaly.'' However, taken together with both the
888: left-handed and right-handed Weyl spinors
889: and two gauge fields, $A_L$ and $A_R$, there is now an ambiguity
890: in the form of the anomaly. We now have the freedom
891: to introduce counterterms such as
892: $\epsilon_{\mu\nu\rho\sigma}A_L^\mu A_R^\nu \partial^\rho A_L^\sigma$, \etc,
893: and these can modify the form of the anomaly.
894: We can now force {\em the vector current to be conserved by
895: adding to the lagrangian a particular counterterm.}
896:
897: To see this, consider a term in the action of the form:
898: \beq
899: \label{WZ}
900: S' = \frac{1}{6\pi^2}
901: \int d^4x \; \epsilon_{\mu\nu\rho\sigma}A^\mu V^\nu \partial^\rho V^\sigma
902: \eeq
903: This term is unique, having even parity and nonvanishing.
904: Upon variation wrt $V$ or $A$ it
905: adds corrections to the vector and axial currents:
906: \bea
907: \label{bardeen12}
908: \frac{\delta S' }{\delta V_\mu} & = & \delta J^\mu =
909: -\frac{1}{3\pi^2}\epsilon_{\mu\nu\rho\sigma}A^\nu \partial^\rho V^\sigma
910: +\frac{1}{6\pi^2}\epsilon_{\mu\nu\rho\sigma}V^\nu \partial^\rho A^\sigma
911: \nonumber \\
912: \frac{\delta S' }{\delta A_\mu} & = & \delta J^{5\mu} =
913: \frac{1}{6\pi^2}\epsilon_{\mu\nu\rho\sigma}V^\nu \partial^\rho V^\sigma
914: %\nonumber \\
915: \eea
916: The full currents, $\tilde{J} = J + \delta J$,
917: now satisfy:
918: \bea
919: \label{bardeen13}
920: \partial_\mu \tilde{J}^\mu & = & 0 \;,
921: %\nonumber \\
922: \qquad \qquad \partial_\mu \tilde{J}^{5\mu} = \frac{1}{8\pi^2}\left(
923: F_V^{\mu\nu}\widetilde{F}_{V\mu\nu} + \frac{1}{3}
924: F_A^{\mu\nu}\widetilde{F}_{A\mu\nu}
925: \right).
926: \eea
927: This is called the ``{\em covariant}'' form of the anomaly.
928: The theory is now invariant, and operators transform
929: covariantly with respect
930: to the {\em vector} gauge symmetry.
931: We thus see that the coefficient of the $F_V\tilde{F}_V$ in the
932: divergence of the axial current now corresponds to
933: Adler's result in QED \cite{adler}. The vector current
934: is conserved even though there is an axial vector
935: background field.
936: Thus, Adler's coefficient of the anomaly \cite{adler} arises
937: as a mixture of the Weyl fermion (consistent) anomaly and the
938: counterterm. In fact, for the $F_V\tilde{F}_V$ part,
939: the Adler coefficient is $1/3$ $\times$ (triangle diagrams) plus
940: $2/3$ $\times$ (counterterm). These coefficients are often
941: confused in the literature when authors incorrectly assume
942: that the left-handed Weyl current
943: has an anomaly coefficient that, in magnitude, is half that of Adler's result.
944:
945: One might wonder what kind of UV completion theory
946: leads to this counterterm in the effective action.
947: In fact, as we see below (and in Appendix B), this counterterm is
948: just the Chern-Simons term expressed as $D=4$ effective interaction,
949: arising from our $D=5$ theory,
950: when we truncate on the zero-mode and first KK-mode.
951:
952: In our present situation, therefore,
953: a theory with the spatial delocalization
954: of the chiral fermions where anomalous
955: representations are placed on distinct branes in $D=5$,
956: dictates the use of the {\em consistent anomaly} on each brane.
957: We have the correspondence
958: $B_\mu(0) = A_L = V_\mu-A_\mu $ and $B_\mu(R) = A_R = V_\mu+A_\mu $, hence
959: the anomalies take the form on the branes:
960: \bea
961: \label{bardeen4}
962: \partial_\mu J_L^\mu & = & -\frac{1}{48\pi^2}
963: F^{\mu\nu}(0)\widetilde{F}_{\mu\nu}(0) \qquad\makebox{(brane I)}\nonumber \\
964: \partial_\mu J_R^\mu & = & \frac{1}{48\pi^2}
965: F^{\mu\nu}(R)\widetilde{F}_{\mu\nu}(R) \;\;\;\;\; \makebox{(brane II)}
966: \eea
967: From eq.(\ref{shift0}),
968: eq.(\ref{bardeen4}) the shift in the action
969: under the gauge transformation,
970: \bea
971: \label{fermions}
972: \psi_L(x_\mu) & \rightarrow & \exp(i\theta(0,x_\mu))\psi_L(x_\mu) \nonumber \\
973: \psi_R(x_\mu) & \rightarrow & \exp(i\theta(R,x_\mu))\psi_R(x_\mu)
974: \eea
975: thus takes the form:
976: \bea
977: S_{branes} & \rightarrow &
978: S_{branes}
979: %\nonumber \\ & &
980: +
981: \;\frac{1}{48\pi^2}\int_I d^4x \;\theta (x_\mu, 0)
982: F^{\mu\nu}\widetilde{F}_{\mu\nu}(0) \
983: %nonumber \\& &
984: -\;\frac{1}{48\pi^2}\int_{II} d^4x \; \theta (x_\mu, R)
985: F^{\mu\nu}\widetilde{F}_{\mu\nu}(R)
986: \nonumber \\
987: \eea
988: On the other hand,
989: the CS term produced, under
990: the gauge transformation, the result of
991: eq.(\ref{test2}) on the boundaries:
992: \bea
993: S_{CS} & \rightarrow & S_{CS} -\frac{c}{2} \int_I d^4 x \; \theta (x_\mu, 0)
994: F^{\mu\nu}\widetilde{F}_{\mu\nu}
995: %\nonumber \\ & &
996: +\frac{c}{2}\int_{II} d^4 x \; \theta (x_\mu, R)
997: F^{\mu\nu}\widetilde{F}_{\mu\nu}
998: \eea
999: (note the factor of $1/2$ from the introduction of $\widetilde{F}$).
1000: The cancellation of the
1001: anomalies therefore requires:
1002: \beq
1003: \label{coeff}
1004: c = \frac{1}{24\pi^2}
1005: \eeq
1006: Our discussion may have been somewhat tedious, but
1007: this result is essential to the correct implementation of
1008: the CS-term in a higher dimensional theory.
1009: This is one of several ways
1010: to obtain the quantization of the coefficent of
1011: the CS-term in $D=5$. With multiple copies
1012: of boundary fermions we can have integer multiples of $c$.
1013: We caution the reader
1014: that different configurations of boundary branes, or
1015: chiral fermions in the bulk, can lead to differing results
1016: for $c$ in different domains of the extra dimension
1017: (\eg, see the discussion
1018: in the non-abelian case below eq.(\ref{nonabcase})).
1019: For the configuration of a physical domain $[0,R]$ with a pair of
1020: boundary branes $I$ and $II$ the result of eq.(\ref{coeff})
1021: is the correct coefficient of the Chern-Simons term in the $U(1)$
1022: and non-abelian cases.
1023:
1024:
1025:
1026: \subsection{Mode Expansion}
1027:
1028: Let us now consider the compactification in flat space-time.
1029: For the orbifold (magnetic superconducting branes)
1030: $A_\mu$ is defined as an even function
1031: in the doubled interval $[0,2R]$
1032: and $A^5$ is odd. The physical extra dimension
1033: spans the interval $[0,R]$, which
1034: dictates the normalization
1035: of the fields. We perform a conventional mode expansion for
1036: the KK-mode tower of gauge fields:
1037: \bea
1038: \label{mode}
1039: A_\mu^0 (x,y) & = & \sqrt{\frac{1}{ R}}\widetilde{e} A_\mu^0 (x) \nonumber \\
1040: A_\mu (x,y) & = & \sum_{n=1}^\infty (-1)^n\sqrt{\frac{2}{ R}} \widetilde{e} \cos(n\pi y/R)A_\mu^n (x) \nonumber \\
1041: A_5(x,y) & = & \sum_{n=1}^\infty (-1)^{n+1}\sqrt{\frac{2}{ R}} \widetilde{e}
1042: \sin(n\pi y/R) A_5^n (x)
1043: \eea
1044: The sign conventions, $(-1)^n$, are designed so that the
1045: $A_\mu^n$ ($B_\mu^n$; see below) with $n$ odd
1046: couple with a positive sign to the axial current,
1047: $\bar{\psi}\gamma^5\psi$.
1048:
1049: The kinetic terms contained in $S_0 \equiv S_1+S_2$ become:
1050: \bea
1051: S_1 & = & -\frac{1}{4\widetilde{e}^2}\int_0^R dy\int d^4 x\; F_{\mu\nu}F^{\mu\nu}
1052: %\nonumber \\
1053: = -\frac{1}{4}\sum_{n}\int d^4 x\; F^n_{\mu\nu}F^{n\mu\nu}
1054: \eea
1055: \bea
1056: \label{massterm}
1057: S_2 & = & \frac{1}{2\widetilde{e}^2 }\int_0^R dy\int d^4 x\; F_{\mu 5}F^{\mu 5}
1058: %\nonumber \\
1059: = \half \sum_{n=1}M_n^2\int d^4 x \; B_\mu^n B^{n\mu}
1060: \eea
1061: where the $B^n_\mu(x)$ ( $A^n_\mu(x)$) without
1062: the argument $y$ are $D=4$ fields:
1063: \bea
1064: \label{gi}
1065: M_n & = & {n\pi}/{R}\; ;
1066: % \nonumber \\
1067: \qquad
1068: B_\mu^n = A_\mu^n + \frac{1}{M_n}\partial_\mu A_5^n\; ;
1069: % \nonumber \\
1070: \qquad
1071: F^n_{\mu\nu} \equiv \partial_\mu B^n_\nu - \partial_\nu B^n_\mu.
1072: \eea
1073: We thus observe that the gauge field mass term of eq.(\ref{massterm})
1074: is manifestly gauge invariant. That is, it is automatically
1075: expressed in terms of the $B_\mu^n$, which are
1076: ``Stueckelberg fields.'' The Stueckelberg fields
1077: are combinations of transverse and longitudinal gauge fields
1078: that are manifestly gauge invariant, \ie, if we
1079: shift $A^n_\mu \rightarrow A^n_\mu + \partial_\mu \theta^n$
1080: then we can also shift $A_5^n \rightarrow A_5^n -M^n\theta^n$ and
1081: we see that $B_\mu^n$ is invariant. The $B_\mu^n$ fields
1082: have the same mode expansion as the $A_\mu^n$ in eq.(\ref{mode}).
1083:
1084:
1085: The physical value of the electric charge
1086: follows by considering the
1087: zero mode component of a coupling $A_\mu J^\mu$,
1088: where $J^\mu$ is the vector current on the branes
1089: (sum of left current on $L$ and right current on $R$),
1090: and we see that:
1091: \beq
1092: e = \widetilde{e}/\sqrt{R} \equiv e_0
1093: \eeq
1094: Note that $e$ is dimensionless, since $\widetilde{e}$
1095: has dimensions of $M^{-1/2}$. Likewise,
1096: if we consider a transverse KK-mode coupling to a
1097: current on the brane, $A_\mu^n J^\mu $
1098: we see that the coupling
1099: differs by a normalization factor of $\sqrt{2}$, and we
1100: thus define:
1101: \beq
1102: e' = \sqrt{2}\widetilde{e}/\sqrt{R} = \sqrt{2} e \equiv e_n\;\;\;(n\neq 0)
1103: \eeq
1104: The $B^n_\mu$ couple as:
1105: \bea
1106: & & \bar{\psi}\gamma_\mu \psi_L\sum_n (-1)^n e_n B^{n\mu}
1107: + \bar{\psi}\gamma_\mu \psi_R\sum_n e_n B^{n\mu}
1108: \nonumber \\
1109: & = & \bar{\psi}\gamma_\mu \psi\left(\half \sum_n (1+(-1)^n)B^{n\mu} \right)
1110: + \bar{\psi}\gamma_\mu\gamma^5 \psi\left(\half\sum_n (1-(-1)^n)B^{n\mu} \right)
1111: \eea
1112:
1113:
1114: We now turn to the Chern-Simons term, eq.(\ref{S2}) with
1115: the quantized coefficient of eq.(\ref{coeff}).
1116: We substitute the mode expansion:
1117: \bea
1118: S_{CS} & = & \frac{1}{24\pi^2} \int_0^R dy\int d^4 x
1119: \;\epsilon^{\mu\nu\rho\sigma}
1120: (\partial_{y}B_{\mu}) B_\nu F_{\rho\sigma}
1121: \nonumber \\
1122: & \equiv & \frac{1}{12\pi^2}
1123: \sum_{nmk}\int d^4 x \;
1124: (e_ne_me_k) c_{nmk}(B^n_{\mu} B^m_\nu \widetilde{F}^{k\mu\nu})
1125: %\nonumber \\ & &
1126: \eea
1127: The structure constants, $c_{nmk}$, are determined by performing
1128: the wave-function
1129: overlap integrals in the bulk, which is straightforward:
1130: \bea
1131: \label{ccc}
1132: c_{nmk} & = & (-1)^{(k+n+m)} \int_0^1 dz\;
1133: \partial_z [\cos(n\pi z) ]\cos(m\pi z)\cos(k\pi z)\nonumber \\
1134: & = &
1135: \frac{n^2(k^2+m^2-n^2)\left[(-1)^{(k+n+m)}-1 \right]}{
1136: (n+m+k)(n+m-k)(n-k-m)(n-m+k)}
1137: %\nonumber \\
1138: \eea
1139: Note that in particular cases of interest to us
1140: these reduce to:
1141: \bea
1142: \label{ccc2}
1143: c_{nm0} & = & c_{n0m} =
1144: -\frac{n^2}{n^2-m^2}\left[(-1)^{n+m}-1\right]
1145: \nonumber \\
1146: c_{0nm} & = & c_{000} = 0 \nonumber \\
1147: c_{n00} & = &
1148: \left[1-(-1)^{n}\right].
1149: \eea
1150: The selection rules for KK-mode production and decay
1151: can almost be inferred from these
1152: results, but the effects of the matter fields must also
1153: be incorporated. For example, while the CS-term appears to allow
1154: a KK-mode decay to two zero modes, since $c_{n00}$ is nonzero, we actually
1155: find in Section IV
1156: that this is completely cancelled by the triangle diagrams in
1157: the $m^2 > M_n^2$ limit, (while it remains
1158: allowed in the case of axions on the branes). The decay of an odd (even)
1159: KK-parity mode to an even (odd) KK parity mode plus a zero mode,
1160: through the nonzero $c_{nm0}$ is, however, allowed when the
1161: triangle diagram effects are included.
1162:
1163: The effective action for the full theory
1164: can now be written.
1165: It is again convenient to reabsorb the coupling constants into the
1166: gauge fields, and write the effective action
1167: in the following compact form as an effective $D=4$ theory:
1168: \bea
1169: \label{effective2}
1170: S_{full} & = & \int d^4 x\;\bigl [ \;\bar{\psi}
1171: (i\slash{\partial}+ \slash{V} +
1172: \slash{{\cal{A}}}\gamma^5-m)\psi
1173: %\nonumber \\ && \qquad
1174: + \frac{1}{12\pi^2} \sum_{nmk}c_{nmk}B^n_\mu B^m_\nu
1175: \widetilde{F}^{k\mu\nu}
1176: \nonumber \\ & &
1177: %\!\!\!\!\!\!\!\!\!
1178: -\frac{1}{4e^2} F^0_{\mu\nu}F^{0\mu\nu}
1179: -\frac{1}{4e'{}^2}\sum_{n\geq 1} F^n_{\mu\nu}F^{n\mu\nu}
1180: +\sum_{n\geq 1} \frac{1}{2e_n^2}M_n^2 B_\mu^n B^{n\mu} \bigr ]
1181: %\nonumber \\
1182: \eea
1183: where:
1184: \bea
1185: V_\mu & = & \sum_{n\; even}B_\mu^n, \;\;\;\qquad
1186: {\cal{A}}_\mu = \sum_{n\; odd}{B}_\mu^n
1187: \eea
1188: and the photon zero mode is defined as:
1189: $A_\mu \equiv B_\mu^0.$
1190: Sums over even $n$ now
1191: include $n=0$, unless otherwise indicated.
1192: We caution that $S_{tree}$ must still be supplemented by the boundary
1193: brane matter effects \ie, the
1194: triangle diagram loops of Appendix C.
1195:
1196:
1197: In eq.(\ref{effective2}) we have supplemented the action with
1198: an electron mass term. The electron mass has to be viewed
1199: as arising from a bilocal bulk term in the parent theory of the form:
1200: \beq
1201: m\bar{\psi}_L(x_\mu,0) W \psi_R(x_\mu,R) + h.c., \qquad \qquad
1202: W = \exp(i\int_0^R A_5(x_\mu,x_5) dx^5)
1203: \eeq
1204: This is gauge invariant in the full $D=5$ theory owing
1205: to the Wilson line, $W$. When we
1206: perform the gauge transformation of
1207: eq.(\ref{gt}) the Wilson line becomes $W= U(0)W U(R)^\dagger = 1 $
1208: and the mass
1209: term goes into the Dirac form, $-m\bar{\psi}\psi$,
1210: as displayed in eq.(\ref{effective2}).
1211: The full set of gauge transformations on the branes thus
1212: form a $U(1)_L\times U(1)_R$ gauged chiral lagrangian.
1213: Since the $B_\mu^n$ axial fields are in the Stueckelberg form,
1214: there is no conflict with the gauge invariance of the
1215: electron mass term. Moreover, the vectorial gauge transformations, \ie, those
1216: with $\theta(0) = \theta(R)$
1217: commute with the electron mass term.
1218:
1219:
1220: The KK-modes with even (odd) $n$ must now
1221: be interpreted as vectors, with $J^P=1^-$ (axial-vectors,
1222: with $J^P=1^+$). The $\epsilon$ symbol
1223: has locked the internal KK-mode parity of the $x^5$ wave-functions
1224: in the bulk to the parity of space-time. Put another way,
1225: the CS-term is explicitly violating the notion of independent
1226: symmetries of KK-mode parity and space-time parity.
1227: All of the massive $B^n_\mu$ are seen to be gauge invariant
1228: in the sense of Stueckelberg fields,
1229: \ie, they appear in the gauge invariant combinations as written in
1230: eq.(\ref{gi}).
1231:
1232: Note that if we truncate the theory on the zero mode $B^0$ and first KK-mode,
1233: $B^1$, the CS term goes into the form:
1234: \beq
1235: \frac{1}{12\pi^2} c_{100}B^1_\mu B^0_\nu
1236: \widetilde{F}^{0\mu\nu} \; = \; \frac{1}{6\pi^2}\epsilon^{\mu\nu\rho\sigma}
1237: A_\mu V_\nu {\partial}_{\rho}V_{\sigma}
1238: \eeq
1239: where the $c_{111}$ term vanishes by antisymmetry.
1240: This is precisely the same form as the counterterm of eq.(\ref{WZ}),
1241: and correspondingly the full vector and axial vector currents have
1242: covariant anomalies as in eq.(\ref{bardeen13})
1243: Using a remarkable
1244: identity amongst the $c_{nmk}$ structure constants, obtained
1245: in eq.(\ref{dd}),
1246: the full current and anomaly structure for the tower
1247: of KK-modes is derived in Appendix B. This yields the {covariant} form
1248: of the anomaly for the tower of KK modes, and the vector zero-mode
1249: current is indeed conserved.
1250:
1251: \subsection{The Full Action Including Matter Effects}
1252:
1253: We now supplement the full tree action of eq.(\ref{effective2})
1254: with the contributions of the matter fields. We do
1255: this by integrating out the matter fields.
1256: The gauge field part of the full action
1257: now takes the form:
1258: \bea
1259: \label{effective3}
1260: S_{tree} & = & \int d^4 x\;\bigl [
1261: \frac{1}{12\pi^2} \sum_{nmk}\bar{c}_{nmk}B^n_\mu B^m_\nu
1262: \widetilde{F}^{k\mu\nu}
1263: \nonumber \\ & &
1264: %\!\!\!\!\!\!\!\!\!
1265: -\frac{1}{4e^2} F^0_{\mu\nu}F^{0\mu\nu}
1266: -\frac{1}{4e'{}^2}\sum_{n\geq 1} F^n_{\mu\nu}F^{n\mu\nu}
1267: +\sum_{n=0} \frac{1}{2e_n^2}M_n^2 B_\mu^n B^{n\mu} \bigr ]
1268: %\nonumber \\
1269: \eea
1270: where the ${c}_{nmk}$ have been replaced by the $\bar{c}_{nmk}$.
1271:
1272: In Appendix A we show
1273: that with axions as matter fields the CS term is unmodified:
1274: \beq
1275: \bar{c}_{nmk} = {c}_{nmk} \qquad \qquad \makebox{(axions)}.
1276: \eeq
1277: In this case the photon has acquired a mass, and
1278: we include a nonzero $M^2_0$ term. There is also a
1279: residual physical axion, $\phi^{(-)}$ and terms
1280: containing it must be included into eq.(\ref{effective3})
1281: from eq.(\ref{axionaction4}). The full
1282: mass matrix of KK-modes further
1283: involves small off-diagonal mixings of the massive
1284: photon with the $n$ even KK-modes, as displayed
1285: in eq.(\ref{axionaction4}). These off-diagonal mixings are
1286: negligible, but the diagonal photon mass in a unitary gauge
1287: makes the photon itself a Stueckelberg field and
1288: gauge invariance is protected by shifts in the
1289: $\phi^{(+)} = (\phi_L-\phi_R)/2 $ combination
1290: of the brane axions.
1291:
1292: The more interesting case of chiral fermions is studied in Appendix C.
1293: The simplest case is that of the decoupled fermion, in which
1294: $m$ is large compared to $M^a$, $M^b$ and $M^c$, and $a\rightarrow b+c$
1295: is the exclusive tree body decay mode mediated by the CS term and anomaly.
1296: We obtain for the boundary term
1297: an effective operator, ${\cal{O}}_3$, describing
1298: the 3-gauge boson amplitude of the triangle diagrams from Appendix C,
1299: in eq.(\ref{an10}) (with coupling constants restored):
1300: \beq
1301: \label{calo3}
1302: {\cal{O}}_3 = -\frac{1}{12\pi^2}\epsilon^{\mu\nu\rho\sigma}
1303: \sum_{nmk} (e_n e_m e_k)a_{nmk}B^n_\mu B^m_\nu\partial_\rho B_\sigma^k
1304: \eeq
1305: where:
1306: \beq
1307: \label{acoeff}
1308: a_{nmk} = \half (1 - (-1)^{n+m+k})(-1)^{m+k}
1309: \eeq
1310: Note that this has the form of the counterterm that mediates consistent
1311: and covariant anomalies when truncated on the lowest modes.
1312: In this case, we therefore have:
1313: \beq
1314: \bar{c}_{nmk} = {c}_{nmk} - a_{nmk} \qquad \makebox{(massive spinors)}
1315: \eeq
1316: hence we can write:
1317: \beq
1318: \label{barc}
1319: \bar{c}_{nmk} = \left[(-1)^{(k+n+m)} -1\right]
1320: \left(\frac{n^2(k^2+m^2-n^2)}{
1321: (n+m+k)(n+m-k)(n-k-m)(n-m+k)} + \half (-1)^{m+k}\right)
1322: \eeq
1323: By adding these terms
1324: we are decoupling the fermions
1325: and the effective action becomes purely bosonic.
1326:
1327:
1328: %\newpage
1329:
1330: \section{New Physics from the Chern-Simons Term}
1331:
1332:
1333: From the CS term in eq.(\ref{effective3}) we deduce
1334: the Feynman rule for a vertex as shown in Fig. 2
1335: for the process $B^a \rightarrow B^b + B^c$:
1336: \bea
1337: \label{effective5}
1338: T_{CS} & = & -\frac{ee'{}^2}{12\pi^2} \big[
1339: (-\bar{c}_{abc}+\bar{c}_{bac}+\bar{c}_{bca}-\bar{c}_{cba})[B]
1340: %\nonumber \\
1341: %& &
1342: + (\bar{c}_{acb}-\bar{c}_{cab}+ \bar{c}_{bca}-\bar{c}_{cba})[A]
1343: \big]
1344: \eea
1345: where $[A]$ and $[B]$ are (as in Appendix C):
1346: \beq
1347: [A] = \epsilon^{\mu\nu\rho\sigma}
1348: \epsilon^a_\mu \epsilon^b_\nu\epsilon^\gamma_\rho k^\sigma
1349: \qquad
1350: [B] = \epsilon^{\mu\nu\rho\sigma}
1351: \epsilon^a_\mu \epsilon^b_\nu\epsilon^\gamma_\rho q^\sigma
1352: \eeq
1353: Here we have used momentum conservation
1354: $ p = k + q$.
1355: and
1356: we have rescaled coupling constants back into the
1357: interaction of eq.(\ref{effective3}).
1358: As written, $q$ and $k$ are outgoing momenta, and
1359: give a factor of $+i$; we also have $+i$ from $e^{iS}$
1360: and $-1$ from the CS-term coefficient.
1361:
1362: \subsection{Decay of KK-mode to KK-mode plus $\gamma$}
1363:
1364:
1365: \begin{figure}[t]
1366: \vspace{4cm}
1367: \special{psfile=three.eps
1368: angle=0 hscale=40 vscale=40 hoffset=60 voffset=-20
1369: % angle=-90 hscale=60 vscale=50 hoffset=0 voffset=240
1370: }
1371: \vspace{1.5cm}
1372: \caption[]{
1373: %\small
1374: \addtolength{\baselineskip}{-.3\baselineskip}
1375: Chern-Simons term plus triangle diagrams yield a three body vertex
1376: describing the decay of KK-modes $a\rightarrow b + c$. In the text we study
1377: the special case where $c=0$ and $B^c$ is the photon $\gamma$.}
1378: \label{dirac2}
1379: \end{figure}
1380:
1381: As an example application of
1382: the formalism, let us now
1383: compute the tree approximation decay width of the $a$th KK-mode
1384: into the $b$th KK-mode plus a massless zero-mode (photon).
1385:
1386:
1387: For the $a$th KK-mode of 4-momentum $p_\mu$, polarization $\epsilon_\mu^a$,
1388: decaying to
1389: the $b$th mode of momentum $k_\mu$, polarization $\epsilon_\mu^b$
1390: and the $\gamma$
1391: of momentum $q_\mu$, polarization $\epsilon_\mu^\gamma$, eq.(\ref{effective3})
1392: leads to the Feynman rule for the 3-body vertex of Fig.(2):
1393: \bea
1394: \label{effective5}
1395: T_{CS} & = & -\frac{ee'{}^2}{12\pi^2} \big[
1396: (-\bar{c}_{ab0}+\bar{c}_{ba0}+\bar{c}_{b0a}-\bar{c}_{0ba})[B]
1397: %\nonumber \\
1398: %& &
1399: + (\bar{c}_{a0b}-\bar{c}_{0ab}+ \bar{c}_{b0a}-\bar{c}_{0ba})[A]
1400: \big]
1401: \eea
1402: The pure CS term using the coefficients $c_{nmk}$ yields
1403: \bea
1404: T_{CS} & = & \frac{ee'{}^2}{6\pi^2}
1405: \left[
1406: \left(\frac{a^2+2b^2}{a^2-b^2}\right)[B]-[A]
1407: \right]
1408: \eea
1409: Here we have used the pure $c_{nmk}$ structure constants
1410: of eqs.(\ref{ccc},\ref{ccc2})
1411: and we have assumed that $a+b$ is odd, as required for a nonzero result.
1412: Note that the KK-mode masses are $M_a=\pi a/R$, so the vertex
1413: rule can be written as:
1414: \bea
1415: \label{effective5b}
1416: T_{CS} & = & \frac{ee'{}^2}{6\pi^2}
1417: \left[
1418: \left(1 + \frac{3M_b^2}{M_a^2-M_b^2}\right)[B] -[A]
1419: \right]
1420: \eea
1421: As we have stressed, it is the combination of
1422: the matter fields and the CS-term that is gauge invariant.
1423:
1424: For the massive fields, the condition on the outgoing
1425: polarization $p^\mu\epsilon_\mu^a$
1426: is fixed (essentially by our gauge choice, $A^5=0$).
1427: This arises from the free field equation of motion of
1428: the Stueckelberg fields, $\partial_\mu F^{a\mu\nu}
1429: +M_a^2 B_\nu = 0$ whence $\partial_\mu B^\mu = 0$.
1430: However,
1431: for the massless zero mode (photon) there is no such restriction on
1432: the polarization.
1433: Correspondingly under the zero mode
1434: gauge transformation:
1435: \beq
1436: \label{shift}
1437: \epsilon^\gamma_\mu \rightarrow \epsilon^\gamma_\mu + \kappa\; q_\mu
1438: \eeq
1439: we see that the term $[B]$ is gauge invariant, while
1440: $[A]$ is {\em not} invariant, undergoing a shift with
1441: eq.(\ref{shift}).
1442:
1443: We now include the fermion triangle loops.
1444: The triangle diagrams in the large $m$ limit
1445: produce a simple expression for the amplitude
1446: that has been computed in eq.({\ref{calo3}), or eq.({\ref{an10})
1447: and thus contribute the $a_{nmk}$ coefficients
1448: of eq.(\ref{acoeff})
1449: to the $\bar{c}_{nmk} = {c}_{nmk}-a_{nmk}$ combination.
1450: For the
1451: process of interest, $B^a\rightarrow B^b + \gamma$ we see that
1452: the eq.({\ref{calo3}) contribution takes the form:
1453: \bea
1454: \label{effective5c}
1455: T_L+T_R & = & \frac{e e'{}^2}{6\pi^2}\left[[A]
1456: - \half(3(-1)^b -1) [B]\right] +{\cal{O}}(1/m^2)
1457: % \nonumber \\
1458: \eea
1459: We can check this result directly from the vertices
1460: in eqs.(\ref{final},\ref{final2}) where we take $c=0$ and $a+b$ is then
1461: odd. Hence,
1462: $f_{ab0} = 2$ and $g_{ab0} = -(3(-1)^b-1)$ and eq.(\ref{final})
1463: yields the above result.
1464:
1465:
1466: Thus, the resulting
1467: full Feynman vertex rule is the sum of eq.(\ref{effective5b})
1468: and eq.(\ref{effective5c}), and takes the form:
1469: \bea
1470: \label{effective6}
1471: \bar{T} \equiv T_{CS}+T_L+T_R & = &
1472: %\nonumber \\
1473: %& & \!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1474: %\!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1475: \frac{e e'{}^2}{2\pi^2}
1476: \left(\frac{M_b^2}{M_a^2-M_b^2} - \half((-1)^b-1)\right)[B].
1477: %\nonumber \\
1478: \eea
1479: We see that the $[A]$ term, which violated the electromagnetic
1480: gauge invariance, has miraculously cancelled, as indeed it must!
1481: We now have a result that is fully gauge invariant:
1482: the longitudinal component of the zero-mode,
1483: $\epsilon^\gamma_\mu \propto q_\mu$,
1484: decouples from the full amplitude, since
1485: $B\rightarrow 0$ when $\epsilon^c_\mu \rightarrow q_\mu$.
1486:
1487: The physical transition amplitudes for abnormal
1488: parity, $1^+$ [$a$ odd, $b$ even], and normal parity, $1^-$ [$a$ even,
1489: $b$ odd], $B^a$ decay thus take the compact forms:
1490: \bea
1491: \label{effective6}
1492: \bar{T}^+ & = &
1493: % \!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1494: %\!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1495: \frac{e e'{}^2}{2\pi^2}
1496: \left(\frac{M_a^2}{M_a^2-M_b^2} \right)[B]
1497: \qquad 1^-\rightarrow 1^+ + \gamma
1498: \nonumber \\
1499: \bar{T}^- & = &
1500: % \!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1501: % \!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1502: \frac{e e'{}^2}{2\pi^2}
1503: \left(\frac{M_b^2}{M_a^2-M_b^2} \right)[B]
1504: \qquad 1^+\rightarrow 1^- + \gamma
1505: %\nonumber \\
1506: \eea
1507: Several comments on the physical structure of the decay amplitude
1508: are in order.
1509: We see that if the massive gauge fields are
1510: nearby in mass, $M_a^2 \approx M_b^2$, then the amplitude
1511: is mainly dominated by the CS term, eq.(\ref{effective5b}).
1512: On the other hand, in the limit $M_b^2 << M_a^2$ the
1513: decay amplitude is dominated by a
1514: coherent superposition of the Chern-Simons term
1515: and the matter anomalies (triangle diagrams)
1516: of eq.(\ref{effective5c}). This does not mean that the partial width
1517: is predominantly governed by the $M_a^2 \approx M_b^2$ limit, since here
1518: the phase space is becoming small (in fact, we'll
1519: see that it is dominated by the decays to the lightest KK-mode
1520: longitudinal components, since those have the smallest ``decay constants,''
1521: \ie, the final state couples as $\sim k^\mu/M_b$ giving
1522: an $(M_a/M_b)^2$ enhancement).
1523: The smallest allowed
1524: difference $M_a^2 - M_b^2$ for large $a$, is given
1525: by $b=a-1$, whence the mass $M_a^2 = a^2\pi^2/R$
1526: and mass difference is
1527: $M_a^2 - M_{a-1}^2 \approx 2a\pi^2/{R^2}$.
1528: In the limit
1529: of large $a $ and $b=a-1$ we then have:
1530: \bea
1531: \label{effective6}
1532: \bar{T}^+ & \approx & \bar{T}^+ \approx
1533: % \!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1534: %\!\!\!\!\! \!\!\!\!\! \!\!\!\!\! \!\!\!\!\!
1535: \frac{ee'{}^2a}{4\pi^2}[B]
1536: \eea
1537: This growth of the amplitudes with large $a\approx b$ can
1538: be viewed as a short distance limit.
1539: The factor $a $ is yielding the power-law running of the coupling
1540: constant, $e'{}^2 a \equiv e'{}^2 (M_a) $.
1541: In the limit $M_b^2 << M_a^2 $ the decay amplitudes
1542: are suppressed, and by
1543: the anomaly (triangle diagrams) CS-term interference amplitude:
1544: \bea
1545: \label{effective6}
1546: \bar{T} & \rightarrow &
1547: \frac{ee'{}^2}{4\pi^2}
1548: \left((-1)^b+1\right)[B]
1549: \eea
1550: The odd-even effect here,
1551: is a consequence of the $D=5$ overlapping
1552: wavefunctions.
1553: It is remniscent of a chiral flip suppression as in,
1554: \eg, $\pi^\pm$ decay to $e\nu$, but it involves the particular
1555: matching of the CS term with the boundary anomaly in a
1556: way that makes the result somewhat opaque. This
1557: can presumably vary if a different theory is taken on the branes.
1558: For example, in the case of the brane axions as in Appendix A
1559: there is no brane triangle diagram contribution and this
1560: odd-even effect is washed out.
1561:
1562: Turning to the partial
1563: width calculation,
1564: the zero-mode gauge invariance implies, as usual,
1565: that in
1566: summing over final state $\gamma$ polarizations we can use
1567: the familiar:
1568: \beq
1569: \sum_\lambda \epsilon^\gamma_\mu(\lambda) \epsilon^\gamma_\nu(\lambda)
1570: = -g_{\mu\nu}
1571: \eeq
1572: and conveniently drop the (singular) $q_\mu q_\nu/q^2$
1573: terms.
1574:
1575: For the heavy vector meson polarizations the
1576: source-free equations
1577: of motion, eq.(\ref{eom}), imply that
1578: the $B_\mu^n$ (Stueckelberg) fields must obey the Lorentz gauge condition,
1579: $\partial^\mu B_\mu^n =0$
1580: (as usual,
1581: the outgoing and incoming fields are treated
1582: as freely propagating fields, so we ignore their anomalous
1583: source terms, which are part of the interaction generating
1584: the transition). Thus, the polarization sums for
1585: massive fields are:
1586: \bea
1587: \sum_\lambda \epsilon^a_\mu(\lambda) \epsilon^a_\nu(\lambda)
1588: & = & -\left(g_{\mu\nu}- \frac{p_\mu p_\nu}{M_a^2}\right)
1589: \nonumber \\
1590: \sum_\lambda \epsilon^b_\mu(\lambda) \epsilon^b_\nu(\lambda)
1591: & = & -\left(g_{\mu\nu}- \frac{k_\mu k_\nu}{M_b^2}\right)
1592: \eea
1593:
1594: Note that if we had axions as matter fields there
1595: would be no loop correction and the original vertex, $T_{CS}$,
1596: of eq.(\ref{effective5b}) would be the full result.
1597: However, then the zero mode photon would
1598: be a massive Stueckelberg field, and we would use the polarization sums:
1599: \bea
1600: \sum_\lambda \epsilon^\gamma_\mu(\lambda) \epsilon^\gamma_\nu(\lambda)
1601: & = & -\left(g_{\mu\nu}- \frac{p_\mu p_\nu}{m_\gamma^2}\right)
1602: \eea
1603:
1604: Squaring the amplitude and summing over the $b$ and $\gamma$
1605: polarizations and averaging over the $a$ polarization
1606: yields:
1607: \bea
1608: <\bar{T}^2> & = & \frac{1}{3}\left(\frac{e^3}{\pi^2}\right)^2
1609: \left(\frac{M_a^2}{M_b^2} \right)
1610: (M_a^2 + M_b^2) \qquad 1^-\rightarrow 1^++\gamma
1611: \nonumber \\
1612: <\bar{T}^2> & = & \frac{1}{3}\left(\frac{e^3}{\pi^2}\right)^2
1613: \left(\frac{M_b^2}{M_a^2}\right)
1614: (M_a^2 + M_b^2) \qquad 1^+\rightarrow 1^- +\gamma
1615: \eea
1616: where we have set $e'{}^2 = 2 e^2$.
1617: The former case is quite singular in the $M_b<< M_a$ limit.
1618: This arises from the decay of the transverse $1^-$ particle
1619: into the longitudinal $1^+$.
1620: The $M_b^2/M_a^2$ suppression in the $1^+\rightarrow 1^- +\gamma$
1621: is the analogue of a chiral suppression, such
1622: as in $\pi\rightarrow e\nu$.
1623:
1624: Thus, putting in phase space,
1625: the partial decay width for $a\rightarrow b + \gamma$
1626: in the $M_a^2 >> M_b^2$ limit, is:
1627: \bea
1628: \Gamma_{1^-\rightarrow 1^+\gamma} & = &
1629: \frac{2\alpha^3}{3\pi^3}\left( \frac{M_a^3}{M_b^2} \right)\; ,
1630: \qquad \qquad
1631: %\nonumber \\
1632: \Gamma_{1^+\rightarrow 1^-\gamma} =
1633: \frac{2\alpha^3}{3\pi^3}M_b \; .
1634: \eea
1635: In the limit $\Delta M = M_a - M_b<< M_a$ we have:
1636: \bea
1637: \Gamma_{1^\pm\rightarrow 1^\mp\gamma} & = &
1638: \frac{2\alpha^3}{3\pi^3}\Delta M \; .
1639: \eea
1640: The most conspicuous effect is the $M_a^2/M_b^2$ enhancement.
1641: This could be absorbed into $\alpha'{}^2$ factors ($\alpha' = 2\alpha$) and
1642: viewed as a power-law running of the coupling constants:
1643: \bea
1644: \Gamma_{1^-\rightarrow 1^+\gamma} & = &
1645: \frac{\alpha(0)\alpha'{}^2(M_a)}{6\pi^3}\left( \frac{M_a}{R^2 M_b^2} \right)
1646: \nonumber \\
1647: \Gamma_{1^+\rightarrow 1^-\gamma} & = &
1648: \frac{\alpha(0)\alpha'{}^2(M_a)}{6\pi^3}\left( \frac{M_b}{R^2 M_a^2} \right)
1649: \eea
1650: The point of writing this latter result
1651: is that there is nothing particularly pathological
1652: about the enhanced decay rates of superheavy KK modes
1653: through the CS term, compared to the usual
1654: pathologies of extra dimensions. The expected order of magnitude
1655: for such decays would have been $\sim \alpha(0) [\alpha'(M_a)]^2 M_a$
1656: and we see that typically $R^2M_b^2= \pi^2b^2 >> 1$. Thus,
1657: these decay widths are consistent with naive expectations.
1658:
1659: The quadratic growth of the
1660: width $\Gamma_{1^-\rightarrow 1^+\gamma} \propto M_a^2$
1661: is presumably related to the
1662: quadratic divergence of the two point function in the continuum
1663: $D=5$ theory where the Chern-Simons term plays the role of the
1664: interaction as in Fig.(2). While we have not performed
1665: the detailed analysis, this loop presumably satisfies
1666: a $D=5$ dispersion relation (a sum-rule in
1667: the $D=4$ effective theory), and yields the radiative
1668: corrections to the power-law growth of
1669: the coupling constant.
1670:
1671:
1672:
1673: \subsection{Zero Mode $+$ Zero Mode $\rightarrow $ KK-Mode Vanishes}
1674:
1675: We note that the CS-term contains the vertex
1676: describing $a\rightarrow 0 + 0$:
1677: \bea
1678: \label{effective13}
1679: T_{CS} & = & -\frac{ee'{}^2}{12\pi^2} \big[
1680: (-\bar{c}_{a00}+\bar{c}_{0a0}+\bar{c}_{00a}-\bar{c}_{00a})[B]
1681: %\nonumber \\
1682: %& &
1683: + (\bar{c}_{a00}-\bar{c}_{0a0}+ \bar{c}_{00a}-\bar{c}_{00a})[A]
1684: \big]
1685: \eea
1686: Consider the case in which $\bar{c}_{nmk} = c_{nmk}$
1687: and using $c_{a00} = (1-(-1)^a)$, $c_{0mk} = 0$
1688: we have the amplitude:
1689: \beq
1690: T_{CS}= -(1-(-1)^a) \frac{ee'{}^2}{12\pi^2} ([A]-[B] )
1691: \eeq
1692: This is the result for the axionic case.
1693:
1694: For fermionic
1695: matter fields we see from eq.({\ref{acoeff}) that the
1696: $a_{a00} = (1-(-1)^a)/2$, $a_{0a0} = a_{00a}=(-1)^a(1-(-1)^a)/2$,
1697: thus:
1698: \beq
1699: T_{L}+ T_R = (1-(-1)^a)\frac{ee'{}^2}{12\pi^2} ([A] - [B]),
1700: \eeq
1701: and, the combined amplitude is:
1702: \beq
1703: \label{cancel}
1704: T_{CS} + T_{L} + T_{R} = 0.
1705: \eeq
1706: There are no couplings of a single KK-mode to two zero-modes in the
1707: $m^2 > M_a^2$ limit. This is a consequence of gauge invariance
1708: of the zero mode.
1709:
1710: In the case of axions on the branes we would have a
1711: zero-mode $+$ zero mode $\rightarrow $ KK-mode vertex, but
1712: then the zero-mode is {\em not} massless. The cancellation
1713: of eq.({\ref{cancel})
1714: may be a general result for a massless zero-mode.
1715: We caution that we have proved it in the large $m$ case
1716: and the effects of massless fermions, or an off-shell
1717: zero-mode, may be non-zero. It would be of interest
1718: to explore the standard model with split fermions, \eg,
1719: in which the $U(1)_Y$ anomalies are delocalized,
1720: to examine if processes like $Z+\gamma \rightarrow B^a$ can occur
1721: via mixing, or the nonzero $Z$ mass.
1722:
1723:
1724:
1725: \section{The Non-Abelian Case}
1726:
1727: We now consider an $SU(N)$ gauge theory
1728: in $D=5$, with the covariant derivative:
1729: \beq
1730: D_A = \partial_A - iA_A \qquad A_A \equiv A_A^a\frac{\lambda^a}{2}
1731: \eeq
1732: The field strength and the
1733: kinetic term lagrangian density are:
1734: \bea
1735: G_{AB} & = & i[D_A,D_B]
1736: = \partial_A A_B - \partial_B A_A -i[A_A, A_B];
1737: \qquad
1738: L_0 = -\frac{1}{2\widetilde{g}^2} \Tr(G_{AB}G^{AB}),
1739: \eea
1740: and $1/\widetilde{g^2}$ again has dimension of $M^1$.
1741: A gauge transformation involves a local gauge rotation,
1742: $U = \exp(i\theta^a(x^A)\lambda^a/2)$ and the gauge
1743: field and covariant field strength transform as:
1744: \beq
1745: A_A\rightarrow U^\dagger iD_A U + A_A
1746: -\partial_A\theta^a\frac{\lambda^a}{2}
1747: - i\theta^a A_A^b \left[\frac{\lambda^a}{2},\frac{\lambda^b}{2} \right]
1748: + \; ...
1749: \eeq
1750:
1751: The Yang-Mills Chern-Simons term,
1752: also known as the ``second Chern character,''
1753: takes the form:
1754: \bea
1755: \label{CSterm0}
1756: {\cal{L}}_{CS} & = & c\;\epsilon^{ABCDE}
1757: \Tr \Bigl ( A_A \partial_B A_C \partial_D A_E
1758: %\nonumber \\
1759: %& & \!\!\!\!
1760: - \frac{3i}{2}A_A A_BA_C \partial_D A_E
1761: - \frac{3}{5}A_A A_B A_C A_D A_E \Bigr )
1762: \eea
1763: This can be rewritten in a convenient
1764: form involving gauge covariant field
1765: strengths,
1766: \bea
1767: \label{CS2}
1768: {\cal{L}}_{CS}
1769: & = &
1770: \frac{c}{4}\epsilon^{ABCDE}
1771: \Tr \Bigl (A_A G_{BC}G_{DE}
1772: %\nonumber \\
1773: %& & \!\!\!\!
1774: + i A_A A_B A_C G_{DE}
1775: - \frac{2}{5}A_A A_B A_C A_D A_E \Bigr ) ,
1776: \eea
1777: It should be noted that these expressions vanish
1778: for gauge groups having no $d$ symbols.
1779: (See \cite{zachos} for further discussion and references).
1780:
1781: We now include
1782: the Chern character into the action of a $D=5$
1783: theory
1784: as in the the QED case:
1785: \bea
1786: \label{SCS2}
1787: S_0 & = & \int d^5x\; L_0; \qquad
1788: S_{CS} = \int d^5x\; L_{CS} .
1789: \eea
1790: The variation of the action with respect to $A_A$
1791: again generates the equation of motion:
1792: \beq
1793: \widetilde{e}^2\frac{\delta S}{\delta A^a_A} =
1794: [D_B, G^{BA}]^a - J^{aA} = 0.
1795: \eeq
1796: and there is again a conserved Chern-Simons current
1797: appearing as the source term in the theory,
1798: \bea
1799: \label{current2}
1800: J^a_A & = & \frac{3c}{2}\; \epsilon_{ABCDE}
1801: \Tr(\frac{\lambda^a}{2}\{G^{BC},G^{DE}\}).
1802: \eea
1803: The current explicitly requires that $SU(N)$ possess a $d$-symbol,
1804: hence $N\geq 3$; and it is covariantly
1805: conserved, $[D^A,J^a_A\lambda^a/2]=0$.
1806:
1807: The CS-anomaly arising on boundaries under a gauge
1808: transformation is a more complicated expression. In anomaly
1809: matching to boundary matter fields it suffices to
1810: keep track only of the $\Tr(dAdA)$ terms.
1811: If we compactify the theory as
1812: we did in the QED case with boundary branes $I$ and $II$,
1813: we see that, under an infinitesimal gauge transformation,
1814: there are surface terms:
1815: \bea
1816: \label{test2ym}
1817: S_{CS} & \rightarrow & S_{CS} + c\int d^4x\;
1818: \theta^a(R)\; \epsilon^{\mu\nu\rho\sigma}
1819: \Tr(\frac{\lambda^a}{2}\partial_\mu A_\nu \partial_\rho A_\sigma)\; + \;...
1820: \nonumber \\
1821: & &
1822: \qquad \qquad -
1823: {c}\int d^4x\;
1824: \theta^a(0)\; \epsilon^{\mu\nu\rho\sigma}
1825: \Tr(\frac{\lambda^a}{2}\partial_\mu A_\nu \partial_\rho A_\sigma) + \; ...
1826: %\nonumber \\
1827: \eea
1828: The action has
1829: shifted by the two CS anomalies.
1830:
1831: We again consider the orbifold compactification
1832: with branes I and II now containing,
1833: respectively, quarks $q_L$ and $q_R$ transforming
1834: as $N$ under the $SU(N)$ gauge group. For our
1835: envisioned application we view $SU(N)$ to be
1836: the flavor symmetry, so we'll simply assume the quarks
1837: also carry an (ungauged) color index $N_c$.
1838:
1839: The anomalies on the branes of the chiral quarks are
1840: again required to be the {\em consistent} Yang-Mills anomalies.
1841: These are
1842: given in full form in Bardeen's paper (and we can infer
1843: the $\Tr dAdA$ terms from Appendix C). Written
1844: in terms of
1845: the chiral quarks on their respective boundary branes
1846: we have:
1847: \bea
1848: \label{bardeen20}
1849: \partial_\mu J^{a\mu}_L & = & -\frac{N_c}{24\pi^2}
1850: \epsilon^{\mu\nu\rho\sigma}
1851: \tr\left(\frac{\lambda^a}{2}\partial_\mu A_{L\nu} \partial_\rho A_{L\sigma} \right)
1852: + \; ...
1853: \qquad
1854: \nonumber \\
1855: \partial_\mu J_R^{a\mu} & = & \frac{N_c}{24\pi^2}
1856: \epsilon^{\mu\nu\rho\sigma}
1857: \tr\left(\frac{\lambda^a}{2}\partial_\mu A_{R\nu} \partial_\rho A_{R\sigma} \right)
1858: + \; ...
1859: \eea
1860: where:
1861: \beq
1862: J^{a\mu}_{L}=\bar{q}\gamma^\mu\frac{\lambda^a}{2}q_L,
1863: \qquad \qquad
1864: J^{a\mu}_{R}=\bar{q}\gamma^\mu\frac{\lambda^a}{2}q_R, \qquad
1865: \makebox{and} \qquad
1866: \widetilde{G}_{\mu\nu} = \half \epsilon_{\mu\nu\rho\sigma}{G}^{\rho\sigma}.
1867: \eeq
1868: Under the $SU(N)$ flavor gauge transformation the quarks undergo the
1869: transformation:
1870: \beq
1871: q_L \rightarrow \exp\left( i\theta^a(0)\frac{\lambda^a}{2} \right) q_L
1872: \qquad
1873: q_R \rightarrow \exp\left( i\theta^a(R)\frac{\lambda^a}{2} \right) q_L
1874: \eeq
1875: and the consistent anomalies produce the shifts on
1876: the branes:
1877: \bea
1878: \label{test3ym}
1879: S_{quark} & \rightarrow & S_{quark} - \frac{N_c}{24\pi^2}\int_{II} d^4x\;
1880: \theta^a(R)\; \epsilon^{\mu\nu\rho\sigma}
1881: \Tr(\frac{\lambda^a}{2}\partial_\mu A_\nu \partial_\rho A_\sigma)
1882: \nonumber \\
1883: & &
1884: \qquad \qquad +
1885: \frac{N_c}{24\pi^2}\int_{I} d^4x\;
1886: \theta^a(0)\; \epsilon^{\mu\nu\rho\sigma}
1887: \Tr(\frac{\lambda^a}{2}\partial_\mu A_\nu \partial_\rho A_\sigma) + \qquad ...
1888: %\nonumber \\
1889: \eea
1890: Thus, we conclude:
1891: \beq
1892: \label{nonabcase}
1893: c = \frac{N_c}{24\pi^2}
1894: \eeq
1895: This coefficient applies in the case we have constructed
1896: of two branes $I$ and $II$ bounding the physical interval $[0,R]$.
1897: One can consider a different construction in which the physical
1898: interval is extended to $[0,2R]$, but
1899: the branes $I$ and $II$ remain located at $x^5=0$ and $x^5=R$,
1900: and we use periodic boundary conditions on all fields. This physically
1901: corresponds to a kink $+$ anti-kink soliton with fermionic zero-modes on
1902: $S_1$. If we impose a parity
1903: symmetry on $[0,R] \leftrightarrow [R,2R]$ then
1904: we would have a CS term in the domain $[0,R]$
1905: with coefficient $c_1= -1/48\pi^2$ and an anti-CS term in the domain
1906: $[R,2R]$ with $c_2 = 1/48\pi^2$. In fact,
1907: without the parity symmetry, the domains can
1908: have CS terms with arbitrary coefficients, $c_1$ and $c_2$. The
1909: anomaly matching simply requires $c_1-c_2 = -1/24\pi^2$.
1910: For concreteness we use the orbifold with physical domain
1911: $[0,R]$ and $c_1 = -1/24\pi^2$ in the application below.
1912:
1913: Note that we can also introduce a Wilson line mass term for
1914: our separated quarks of the form:
1915: \beq
1916: \int d^4x\; m\;\bar{q}_L(0)P\exp\left(i\int_0^R A_5 \; dx^5 \right)q_R(R) + h.c.
1917: \eeq
1918: This will play the role of a constituent quark mass
1919: below when we derive the constituent chiral quark model,
1920: and $P\exp(-i\int_0^R A_5 dx^5) = \exp(2i\tilde{\pi}/f_\pi)$
1921: plays the role of the chiral field of mesons when we truncate
1922: on the zero-mode of the theory. In this way, the Yang-Mills theory
1923: of flavor can be morphed into an $SU(N)_L\times SU(N)_R$
1924: chiral lagrangian of mesons, and
1925: the CS term becomes the Wess-Zumino-Witten term, as developed
1926: using deconstruction in ref.(\cite{zachos}).
1927:
1928:
1929: The Chern-Simons term
1930: may again be written in a form, as in the $U(1)$ case,
1931: that separates the $A_5$ and $\partial_5$ terms,
1932: \bea
1933: \label{CS11}
1934: {\cal{L}}_{1} =
1935: \frac{c}{2}\Tr ((\partial_5A_\mu)K^\mu)
1936: +\frac{3c}{4}\epsilon^{\mu\nu\rho\sigma}\Tr(A_5
1937: G_{\mu\nu}G_{\rho\sigma}),
1938: \eea
1939: where we find \cite{zachos}:
1940: \bea
1941: K^{\mu} \equiv \epsilon^{\mu\nu\rho\sigma}
1942: \left(iA_\nu A_\rho A_\sigma + G_{\nu\rho}A_\sigma
1943: + A_\nu G_{\rho\sigma}\right).
1944: \eea
1945: In deriving this result, some total divergences in
1946: the $D=4$ subspace have been discarded, which play
1947: no role in the physics or in the anomaly matching,
1948: as in the $U(1)$ case.
1949:
1950: Again we can perform the gauge transformation that
1951: sets $A^5=0$ using the Wilson line $U(y)$ and defining
1952: the ``Stueckelberg fields'' $B_\mu$ as:
1953: \beq
1954: U(y) = P\exp (i\int_0^{y} A_5(x^5) dx^5) \qquad\qquad
1955: B_\mu = U^\dagger [iD_\mu, U].
1956: \eeq
1957: The Chern-Simons action thus becomes:
1958: \bea
1959: \label{CS20}
1960: {\cal{L}}_{1} & = &
1961: \frac{c}{2}\epsilon^{\mu\nu\rho\sigma}\int d^4x \int_0^R dy
1962: \Tr \left[\partial_y B_\mu\left(iB_\nu B_\rho B_\sigma + G_{\nu\rho}B_\sigma
1963: + B_\nu G_{\rho\sigma}\right)\right]
1964: \eea
1965: and the quark mass term goes into the
1966: Dirac form $m\bar{q}_L q_R+h.c. = m\bar{q} q $.
1967:
1968: As we did in the $U(1)$ case, we can pass to a KK-mode
1969: expansion and calculate the effective interaction
1970: amongst the KK-modes. Again, the KK-mode parity is now
1971: locked to space-time parity and the CS term yields new
1972: interactions, including 4-body amplitudes.
1973:
1974: As an example application of this formalism
1975: we turn presently to something slightly different
1976: and give a simple derivation of the chiral constituent quark
1977: model and the Wess-Zumino-Witten term \cite{wess,witten}
1978: from a Yang-Mills theory in $D=5$.
1979:
1980:
1981: \subsection{Application: The Wess-Zumino-Witten Term}
1982:
1983: Consider the theory truncated on the zero-mode $A^5$ field.
1984: All Stueckelberg gauge fields $B_\mu$ now become pure gauge fields:
1985: \beq
1986: B_\mu(x_\mu , y) = iU^\dagger(y) \partial_\mu U(y)
1987: \eeq
1988: and the field strength, $G_{\mu\nu}$, is now zero.
1989:
1990: The Wilson line, $U$, now contains only
1991: the zero mode $A^5$ and the path-ordering is
1992: no longer necessary because the $x^5$ wavefunction
1993: is factorized from the flavor orientation:
1994: \beq
1995: U(y) = \exp \left(i\int_0^y A_5 \; dx^5 \right)
1996: \eeq
1997: We identify
1998: the Wilson line extending between
1999: the two branes with a chiral field of mesons:
2000: \beq
2001: \label{chiral0}
2002: W = \exp \left(i\int_0^R A_5 \; dx^5 \right) \equiv \exp (2i\tilde{\pi}/f_\pi) ,
2003: \qquad \tilde{\pi} = \pi^a \lambda^a/2
2004: \eeq
2005: For the pseudoscalar octet, $f_\pi = 93$ MeV with this normalization.
2006: We remark that this completely fixes the normalization of
2007: our $\tilde{\pi}$ field. That is, the kinetic term,
2008: $\Tr(G_{\mu 5}G^{\mu 5}) \sim \Tr(\partial\tilde{\pi})^2$
2009: normalization determines $f_\pi$
2010: in terms of $R$ and $\tilde{g}$, but the Wilson line completely
2011: specifies the definition of $f_\pi$ in terms of the $A^5$ line integral
2012: which is all we need presently (\ie, we can just assume the kinetic terms
2013: are correctly normalized by the relationship between $\tilde{e}$
2014: $R$ and $f_\pi$).
2015:
2016: The $A^5$ wavefunction on
2017: the orbifold in the $x^5$ coordinate is $\propto \sin(\pi x^5/R)$,
2018: and using eq.(\ref{chiral0}) we can simply write:
2019: \beq
2020: U(y) = \exp (-2ih(y)\tilde{\pi}/f_\pi );
2021: \qquad h(y) = \half\big( 1-\cos(\pi y/R) \big)
2022: \eeq
2023: We can actually use any monotonic wave-function $h(y)$
2024: satisfying $h(0)=0 $ and $h(R) = 1$.
2025: Thus, expanding the pure gauge vector potential, we see:
2026: \beq
2027: \label{expand}
2028: B_\mu(x_\mu , y) = - 2\times
2029: \left(-\frac{1}{f_\pi}h(y)\partial_\mu \tilde{\pi} + i
2030: \frac{1}{f^2_\pi}h(y)^2
2031: [\tilde{\pi}, \partial_\mu \tilde{\pi}] + {\cal{O}}([\tilde{\pi},[ \tilde{\pi},
2032: \partial \tilde{\pi} ] ]\right)
2033: \eeq
2034: Since we now have pure gauge configurations,
2035: the only surviving term in
2036: the CS-term of eq.(\ref{CS20}) is the
2037: $\Tr ((\partial_y B_\mu) B_\nu B_\rho B_\sigma)\epsilon^{\mu\nu\rho\sigma} $
2038: term.
2039: We note that
2040: $\Tr (B_\mu B_\nu B_\rho B_\sigma)\epsilon^{\mu\nu\rho\sigma} =0 $
2041: by the cyclicity of the trace, which makes eq.(\ref{CS20}) easy
2042: to evaluate when we
2043: substitute the expression of eq.(\ref{expand}), and it takes the form:
2044: \bea
2045: \label{CS21}
2046: {\cal{L}}_{1} & = &
2047: \frac{8ic}{f_\pi^5}\epsilon^{\mu\nu\rho\sigma}\int d^4x \int_0^R dy\;
2048: (\partial_y h(y))\; h(y)^4\;
2049: \Tr \left[ [\tilde{\pi}, \partial_\mu \tilde{\pi}]
2050: \partial_\nu \tilde{\pi}\partial_\rho \tilde{\pi}\partial_\sigma \tilde{\pi}
2051: \right] + \; ...
2052: \eea
2053: where the ellipsis refers to higher powers in the $\tilde\pi$ field.
2054:
2055: Since $h(0) = 0$ and $h(y)=1$ we see that the integral over $y$
2056: is trivial -- the integrand is an exact differential and
2057: the result doesn't depend upon the particular shape of the wavefunction
2058: $h(y)$ ! For example, the zero-mode $A^5$ could have been a constant $x^5$
2059: whence, $h(y) = y/R$, and we would obtain the same result.
2060: The resulting expression is:
2061: \bea
2062: \label{CS21}
2063: {\cal{L}}_{1} & = &
2064: \frac{2N_ci}{15f_\pi^5}\epsilon^{\mu\nu\rho\sigma}\int d^4x
2065: \Tr \left[ \tilde{\pi} \partial_\mu \tilde{\pi}
2066: \partial_\nu \tilde{\pi}\partial_\rho \tilde{\pi}\partial_\sigma \tilde{\pi}
2067: \right]
2068: \eea
2069: Where we have removed the commutator in eq.(\ref{CS21}) using
2070: cyclicity of the trace, and we have
2071: substituted the coefficient, $c = 1/24\pi^2$. This
2072: agrees with Witten's famous result \cite{witten}.
2073: In ref.(\cite{hill4}) we develop the fully gauged WZW term
2074: from the present formalism.
2075:
2076: We can develop a matching
2077: to the QCD spectrum from a compactified
2078: $D=5$ Yang-Mills theory by considering the boundary conditions
2079: dual to an orbifold,
2080: $\epsilon^{ABCD5}F_{AB}|_I = \epsilon^{ABCD5}F_{AB}|_{II}=0$,
2081: equivalently, $F_{\mu\nu}|_I = F_{\mu\nu}|_{II}=0$.
2082: This implies a ``flipped'' or
2083: an ``anti-orbifold'' in which $A^5$ now
2084: has a zero-mode with a flat wave-function in $x^5$
2085: (and even basis functions on
2086: the $[0,2R]$ interval), while $A_\mu$ is now odd, beginning
2087: with a $J^P = 1^-$ massive mode, then a $J^P=1^+$ recurrence, etc.
2088: For an $SU(3)$ gauge group of flavor
2089: this, remarkably, has the spectrum corresponding to the QCD mesons.
2090: The Wilson line, $\int dx^5 A_5$, over the zero-modes forms the octet
2091: of {\em pseudoscalar} mesons (containing $\pi, K,\eta$), while the first
2092: $A^5$ KK-modes correspond to the $0^+$ octet ($a^0$). The $A^\mu$ KK-mode tower
2093: begins with the ($\rho$) vector meson octet,
2094: then the ($A^1$) axial-vector meson octet, etc. We expect
2095: the quantized coefficient of the CS term remains the
2096: same in this model, hence the
2097: WZW term remains the same.
2098: Here the CS term clearly exclusively becomes the
2099: WZW term, because the boundary conditions
2100: $F_{\mu\nu}|_I = F_{\mu\nu}|_{II}=0$
2101: prohibit any Stueckelberg field anomalies that might contribute
2102: to the WZW term.
2103:
2104:
2105:
2106:
2107:
2108: \section{Conclusions}
2109:
2110:
2111: This paper has investigated the physics of the Chern-Simons term
2112: in gauge theories in compactified extra dimensions.
2113: The main thrust is that Chern-Simons terms must occur
2114: in association with ``chiral delocalization,'' whereby anomalous
2115: chiral fermions are placed in different locations in a $D=5$ bulk (\ie, ``split''
2116: anomalies).
2117: We view chiral delocalization as a compelling attribute of extra dimensional
2118: theories, providing a rationale for the existence of flavor-chirality
2119: (non-vectorlike representations)
2120: as is seen in the standard model. We observe that even a vectorlike theory,
2121: such as QED, must become chirally delocalized
2122: when it is imbedded into $D=5$, to naturally protect the small
2123: electron mass. The electron mass becomes a Wilson line connected the
2124: chiral partners. In these cases the Chern-Simons term is inevitable.
2125: We study two models, (1) chiral fermions on opposing branes and
2126: (2) axions on branes.
2127:
2128: For bulk propagating gauge fields,
2129: The Chern-Simons term locks KK-mode parity to the parity of space-time.
2130: Indeed, KK-mode parity,
2131: if it is present, is a spurious symmetry, independent of space-time parity.
2132: The Chern-Simons terms blends these symmetries into a single
2133: surviving parity. This is the exact
2134: analogue of the pion parity for the Wess-Zumino-Witten term.
2135:
2136: Let us summarize how the analysis procedes in general. We begin
2137: in a $D=5$ gauge theory, compactified
2138: in $0 \leq x^5 \leq R$, with chirally delocalized fermions
2139: on the boundaries (branes).
2140: The theory contains a bulk-filling Chern-Simons term.
2141: The chiral fermions have a gauge invariant mass term that is bilocal,
2142: $\sim\bar{\psi}_L(x,0)W\psi_R(x,R)+h.c.$, and
2143: involves the Wilson line, $W=P\exp(i\int_0^R A_5dx^5)$
2144: that spans the bulk.
2145: A general gauge transformation in the bulk produces anomalies on the
2146: boundaries coming from the Chern-Simons term.
2147: Likewise, this gauge transformation produces anomalies, coming
2148: from the fermions on the boundaries. These anomalies take the consistent form,
2149: \ie, they are the direct result of the Feynman triangle loops for the
2150: fermions, and have the identical same form
2151: as the anomalies from the CS term (see Appendix).
2152: We demand that these anomalies cancel, and this fixes the coefficient of the
2153: CS term, generally to $c=1/24\pi^2$.
2154:
2155: We now rewrite the CS term into a form that displays separately $A_5$
2156: and $\partial_5$. We then perform a master gauge transformation that converts
2157: $A_5\rightarrow B_5 = 0$, and $A_\mu \rightarrow B_\mu$.
2158: This also sets the Wilson line spanning the bulk between
2159: the branes to unity. The massive components of
2160: the $B_\mu$ are now gauge invariant Stueckelberg fields, having ``eaten''
2161: their longitudinal degrees of freedom contained
2162: in the non-zero modes of $A_5$.
2163:
2164: Finally, we integrate out the fermions in the large $m$ limit. This
2165: produces effective interactions (the log of the Dirac determinant)
2166: on the boundaries. The form of
2167: this effective interaction is just Bardeen's counterterm \cite{bardeen}
2168: that maps consistent anomalies into covariant ones. We thus have an
2169: expression for total action, ${S}_{full}$, the sum of $S_{CS}$,
2170: the Chern-Simons term, and
2171: the boundary terms from the fermionic Dirac determinant,
2172: summarized by the matrix elements of ${\cal{O}}_3$.
2173: These are functionals of the Stueckelberg fields contained
2174: in the mode expansion of ${B}_\mu$. The new interactions involve
2175: the ``structure constants,'' $\bar{c}_{nmk}$, for the different
2176: KK-modes, $(n,m,k)$. We demonstrate
2177: that $S_{full}$ leads to
2178: new physical processes involving 3-body decay amplitudes
2179: amongst KK-modes. These processes violate naive KK-mode parity, but
2180: conserve the combined space-time and KK-mode parity.
2181: As an example of the formalism, we explicitly compute the decay
2182: widths for massive KK-modes into lighter KK-modes, plus the zero mode.
2183:
2184: The Chern-Simons term coefficient was determined at the outset by cancelling
2185: the anomalies of matter fields that are localized
2186: on branes. We are lead to a study of the massless and
2187: massive consistent anomalies of Weyl fermions. The relationship
2188: between the ``consistent'' and ``covariant'' anomalies
2189: involves a counterterm, (which is shown to be our boundary term,
2190: term truncated on the zero-mode and first KK-mode of a $D=5$ theory).
2191: The tower of KK-mode currents and their anomalies are determined
2192: by the ``structure constants,'' $c_{nmk}$, that appear in
2193: the Chern-Simons term in a mode expansion.
2194: The form of the covariant anomalies involves a miraculous identity
2195: amongst the $c_{nmk}$.
2196:
2197: We finally develop the non-abelian formalism, to the
2198: point of computing the coefficient of the Chern-Simons term.
2199: This is similar to the $U(1)$
2200: case, though we postpone the construction of the KK-mode
2201: effective lagrangian and derivation of the structure constants.
2202: As an example, however, we show how an $SU(N)$ Yang-Mills theory of
2203: quark flavor can be compactified into a low energy $SU(N)_L\times SU(N)_R$
2204: chiral langrangian, where the $A^{5n}$ gauge fields have become ``mesons.''
2205: In this case the Chern-Simons term built of KK-mode of gauge fields, becomes
2206: the Wess-Zumino-Witten term. We immediately obtain
2207: Witten's coefficient for this WZW term, and the formalism will be developed
2208: in a subsequent paper, yielding the fully gauged WZW term \cite{hill4}.
2209:
2210: There is large number of theories to which these considerations apply.
2211: These include various incarnations
2212: of Randall-Sundrum models, Little Higgs theories, and
2213: models of (anomaly) split fermion representations
2214: in extra dimensions. We further envision applications to
2215: string theory, and AdS-CFT QCD as well. The WZW term of gravitation in
2216: a split anomaly mode, \eg, in $D=6$ and $D=7$, would also be
2217: an intriguing application.
2218: An interesting candidate for further study
2219: is the standard model with left-handed fermions on brane $I$ and right-handed
2220: fermions on brane $II$. An $SU(2)\times U(1)$ theory
2221: is a subgroup of $SU(3)$ and
2222: does contain $d$-symbols, and a non-abelian CS term is present
2223: to cancel the delocalized consistent matter anomalies.
2224: In this theory the Higgs mechanism has to yield the Wilson line connecting
2225: the two branes, so the Higgs field is spread out in the bulk.
2226: This is remniscent of many ``Higgsless'' theories. It is of interest
2227: to explore the induced physics via the CS term in these models.
2228: It would likewise be interesting to explore an axi-gluonic
2229: QCD that can be constructed by splitting left-color
2230: onto $I$ and right-color onto $II$. Again, a quantized CS term will occur.
2231:
2232: The background geometry has been taken flat in the present discussion. The
2233: results may have limited sensitivity to the introduction of curvature.
2234: It would be interesting to reformulate the $U(1)$ model
2235: in a Randall-Sundrum geometry to test the ideas. Moreover,
2236: gravitational CS terms could be developed in a parallel manner.
2237:
2238: Originally, in beginning this analysis, we had hoped
2239: that the production
2240: of single KK modes through the CS term would
2241: be an available channel. For example, gluon fusion into
2242: a colored axi-gluon would be a spectacular LHC signature,
2243: or $e^+e^-\rightarrow \gamma+Z'$ might occur at a sufficiently
2244: energetic linear collider.
2245: The $U(1)$ case in the large $m_{electron}$ limit disallows such
2246: processes by the cancellation of the triangle diagrams with
2247: the CS-term for one KK plus two zero-mode vertices, as
2248: in the coefficients $\bar{c}_{mnk}$. However, the axion
2249: model shows that such processes can occur (though in association with photon
2250: mass). We have not explictly
2251: checked, however, that this process does
2252: not occur for massless zero-modes in Yang-Mills,
2253: or considered processes such as ``associated KK mode production''
2254: \eg, zero + zero $\rightarrow$ zero + KK.
2255: Thus, the collider physics implications of the Chern-Simons term remain to
2256: be investigated.
2257:
2258: The $U(1)$ $D=5$ continuum bulk theory consisting
2259: only of gauge kinetic terms and the Chern-Simons term, with
2260: matter restricted to the branes, is in itself
2261: an interesting system. The bulk topological
2262: theory is subject to some nonrenormalization constraints
2263: and it would be interesting to study its loop
2264: structure in $D=5$ further.
2265:
2266:
2267: %\newpage
2268: \appendix
2269:
2270: \noindent\section{Axions on the Branes}
2271:
2272: The simplest anomaly free model with the bulk filling
2273: CS term involves the incorporation of ``left" ($\phi_L$)
2274: and ``right" ($\phi_R$) axion fields confined to the
2275: respective boundary branes.
2276: The axions are coupled to the left- and right- anomalies,
2277: and they can freely shift to absorb the
2278: induced boundary anomalies from the CS term
2279: under a gauge transformation.
2280:
2281:
2282: The kinetic terms of the axions must
2283: be invariant under this shift, and this requires that they
2284: couple longitudinally to the
2285: left- and right-gauge fields on the respective branes. This
2286: further locks the axion shift
2287: to the gauge transformation. In the $D=4$
2288: effective theory, we then find that
2289: one linear combination of the axions, $\phi^{(+)}$,
2290: is eaten to become a longitudinal massive photon,
2291: where:
2292: \beq
2293: \phi^{(+)} = \half (\phi_L +\phi_R) \qquad
2294: \phi^{(-)} = \half (\phi_L-\phi_R)
2295: \eeq
2296: $\phi^{(-)}$ remains as a massless state in the spectrum, coupled
2297: longitudinally to the pseudovector KK-modes. This is the physical axion.
2298: The model exhibits the fact that the bulk CS term interactions
2299: are physical, and yields $\bar{c}_{nmk} = c_{nmk}$.
2300:
2301: Consider QED in $D=5$ on an orbifold
2302: with periodic domain $0 \leq x^5 \leq R$. The full
2303: action of the theory is,
2304: \beq
2305: S=S_0 + S_{CS} + S_{branes}
2306: \eeq
2307: where the gauge field kinetic term action for the theory is:
2308: \bea
2309: S_0 & =& -\frac{1}{4\widetilde{e}^2}\int_0^R dy\int d^4 x\; F_{\mu\nu}F^{\mu\nu}
2310: %\nonumber \\
2311: %& &
2312: -\frac{1}{2\widetilde{e}^2}\int_0^R dy\int d^4 x \;
2313: F_{\mu 5}F^{\mu 5} \\ \nonumber
2314: \eea
2315: The action $S_{CS}$ is defined in eq.(\ref{SCS}).
2316: On brane $I$ at $y=0$ ($II$ at $y=R$),
2317: we place an axion field $\phi_L(x^\mu)$ ($\phi_R(x^\mu)$), and we
2318: define the action:
2319: \bea
2320: \label{axionanomaly}
2321: S_{branes} & = & \half \int_I d^4x \;
2322: m^2(A_\mu(0,x_\mu) -\frac{1}{m}\partial_\mu \phi_L)^2
2323: %\nonumber \\
2324: %& + &
2325: + \half \int_{II} d^4x \;
2326: m^2(A_\mu(R,x_\mu) -\frac{1}{m}\partial_\mu \phi_R)^2
2327: \nonumber \\
2328: & & + \frac{c}{4m}\int d^4x\;
2329: \phi_L\; \epsilon^{\mu\nu\rho\sigma}
2330: F_{\mu\nu} F_{\rho\sigma}|_{I}
2331: %\nonumber \\
2332: %& & -
2333: -\frac{c}{4m}\int d^4x\;
2334: \phi_R\; \epsilon^{\mu\nu\rho\sigma}
2335: F_{\mu\nu} F_{\rho\sigma}|_{II}
2336: \eea
2337: This construction converts our $U(1)$ gauge
2338: theory into an effective chiral theory, \ie,
2339: with distinct axion ``chiralities'' living on the two distinct branes.
2340: The $m^2$ terms contain the axion kinetic
2341: terms, as well as longitudinal couplings
2342: to gauge fields on the respective branes, locking the $\phi$'s
2343: to the gauge fields under gauge transformations.
2344: Thus a gauge transformation in the bulk:
2345: \beq
2346: \label{gtB}
2347: A_A(x_\mu, y) \rightarrow A_A(x_\mu, y)
2348: + \partial_A \theta (x_\mu, y) ,
2349: \eeq
2350: implies, through the $m^2$ terms, a transformation
2351: of the axions on the branes:
2352: \beq
2353: \label{axionshift}
2354: \phi_L \rightarrow \phi_L + m\theta (x_\mu, 0);\qquad
2355: \phi_R \rightarrow \phi_R + m \theta (x_\mu, R).
2356: \eeq
2357: The CS term under this transformation
2358: generates anomalous terms on the branes,
2359: as in eq.(\ref{test2}). These anomalies
2360: are now cancelled by the matching
2361: shifts in the axion fields coupled to
2362: $F\widetilde{F}_I$ and $F\widetilde{F}_{II}$ as in eq.(\ref{axionanomaly})
2363: and the theory is gauge invariant.
2364:
2365: We can thus perform the Wilson line
2366: gauge transformation of eq.(\ref{gt}) with impunity.
2367: The brane action then becomes a functional
2368: of the Stueckelberg fields $B_\mu$:
2369: \bea
2370: \label{axionaction2}
2371: S_{branes} & = & \half \int_I d^4x \;
2372: m^2(B_\mu(0,x_\mu) -\frac{1}{m}\partial_\mu \phi_L)^2
2373: %\nonumber \\
2374: % & + &
2375: + \half \int_{II} d^4x \;
2376: m^2(B_\mu(R,x_\mu) -\frac{1}{m}\partial_\mu \phi_R)^2
2377: \nonumber \\
2378: & & + \frac{c}{4m}\int_I d^4x\;
2379: \phi_L\; \epsilon^{\mu\nu\rho\sigma}
2380: F_{B\mu\nu} F_{B\rho\sigma}|_{I}
2381: %\nonumber \\
2382: %& & -
2383: -\frac{c}{4m}\int_{II} d^4x\;
2384: \phi_R\; \epsilon^{\mu\nu\rho\sigma}
2385: F_{B\mu\nu} F_{B\rho\sigma}|_{II}
2386: \eea
2387: where all gauge fields are now of the form
2388: eq.(\ref{Stu}).
2389:
2390: It is useful to write $B_\mu(x^\mu,y)$ in terms of
2391: a zero mode, $A^0(x^\mu)$ which is independent of $y$,
2392: and the non-zero mode $x^5$ parity components,
2393: $B^{+}_\mu(x^\mu,y)$ and $B^{-}_\mu(x^\mu,y)$,
2394: which have
2395: non-zero KK-mode momentum. We have
2396: the $x^5$ parity assignments:
2397: \bea
2398: A^{0}_\mu(x^\mu,y) & = & A^{0}_\mu(x^\mu,R-y),
2399: \qquad
2400: %\nonumber \\
2401: B^{+}_\mu(x^\mu,y) = B^{+}_\mu(x^\mu,R-y),
2402: %\qquad
2403: \nonumber \\
2404: B^{-}_\mu(x^\mu,y) & = & -B^{-}_\mu(x^\mu,R-y).
2405: \eea
2406: Thus, on the branes we define:
2407: \bea
2408: B^{+}_\mu(x^\mu) & \equiv & B^{+}_\mu(x^\mu,0) = B^{+}_\mu(x^\mu,R),
2409: \qquad
2410: %\nonumber \\
2411: B^{-}_\mu(x^\mu) \equiv B^{-}_\mu(x^\mu,0) = -B^{-}_\mu(x^\mu,R).
2412: \eea
2413: The decomposition thus takes the form:
2414: \beq
2415: B_\mu(x^\mu,y) = A^0(x^\mu)+
2416: B^{+}_\mu(x^\mu,y) + B^{-}_\mu(x^\mu,y)
2417: \eeq
2418: We presently require only that
2419: $A^0(x^\mu)$, $B^{+}_\mu(x^\mu,y)$ and $B^{-}_\mu(x^\mu,y)$
2420: are orthogonal fields upon intergation over $y$ from $0$ to $R$.
2421:
2422:
2423: After performing the gauge transformation of
2424: eq.(\ref{gtB}), we have brought the field $A_5(x^\mu,y)=0$
2425: everywhere throughout the bulk. There remains a residual
2426: gauge transformation that we can do which maintains this
2427: gauge condition, which
2428: redefines
2429: the zero mode (photon) field:
2430: \beq
2431: \label{ggt}
2432: B^0_\mu(x^\mu) = A^0_\mu(x^\mu)+ \partial_\mu\theta(x^\mu)
2433: \eeq
2434: where:
2435: \beq
2436: \theta(x) = \frac{1}{{2}m}(\phi_L(x^\mu)+ \phi_R(x^\mu))
2437: \equiv \frac{1}{m}\phi^{(+)}
2438: \eeq
2439: We do not shift the $\phi^{(+)}$ and $\phi^{(-)}$ fields
2440: under this transformation. The bulk CS term, however, generates
2441: the surface anomalies under this gauge transformation,
2442: which cancel the $\phi^{(+)}$ anomalous coupling on the branes
2443: (this can be explicitly checked by substituting eq.(\ref{ggt})
2444: into the CS term and
2445: noting the integrand is an exact differential
2446: in $y$).
2447: This brings the photon field into the form of
2448: a massive Stueckelberg field
2449: and $\phi^{(+)}$.
2450: thus disappears from the action. The brane action
2451: now takes the form:
2452: \bea
2453: \label{axionaction3}
2454: S_{branes} & = & \int d^4x \;\left[ \half\left(\partial_\mu \phi^{(-)}\right)^2+
2455: m^2\left(B^0_\mu(x^\mu)+
2456: B^{+}_\mu(x^\mu) \right)^2
2457: %\right.
2458: %\nonumber \\
2459: %& & \qquad \qquad + \;\left.
2460: + m^2 \left(B^{-}_\mu(x^\mu)\right)^2 - mB^{-}_\mu(x^\mu)\partial^\mu
2461: \phi^{(-)}\right]
2462: \nonumber \\
2463: & &
2464: %\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
2465: + \frac{c}{m}\int d^4x\;
2466: \phi^{(-)}\; [
2467: F_{B^0\mu\nu} \tilde{F}_{B^0}^{\mu\nu} + 2 F_{B^0\mu\nu} \tilde{F}_{B^+}^{\mu\nu}
2468: %\nonumber \\
2469: %& & \qquad
2470: + F_{B^+\mu\nu} \tilde{F}_{B^+}^{\mu\nu}
2471: +F_{B^-\mu\nu} \tilde{F}_{B^-}^{\mu\nu}].
2472: \nonumber \\
2473: \eea
2474: This shows that the theory contains a residual massless
2475: axion, $\phi^{(-)}$ coupled to the anomalies, and
2476: longitudinally to the pseudovector KK-modes. This is
2477: remniscent of the $\pi^0$. It is not eaten since
2478: the KK-modes, $B^+$ amd $B^-$, acquire (large) masses
2479: from the bulk
2480: kinetic terms. There will also be a
2481: mass term induced for the photon of leading order order:
2482: $
2483: \approx m^2 \left[B^0(x^\mu)\right]^2
2484: $
2485: with corrections due to mixing with the KK-modes of order
2486: $m^4/M_{KK}^2 + ...$
2487:
2488: Let us now consider the orbifold compactification in flat space-time.
2489: The mode expansion follows the identical form as in the
2490: text, eqs.(\ref{mode}).
2491: The axion coupling to the KK-modes is obtained
2492: from substituting the mode expansions into eq.(\ref{axionaction3}).
2493: We define:
2494: \beq
2495: \tilde{m}^2 = \frac{2m^2\tilde{e}^2}{R}
2496: \eeq
2497: and we have the Stueckelberg field for
2498: the massive photon with our zero mode normalization:
2499: \beq
2500: B^0_\mu = A_\mu^0
2501: -\frac{1}{\sqrt{2}\tilde{m}}\partial_\mu \phi^{(+)}
2502: \eeq
2503:
2504: The full $D=4$ effective action, including the brane axion
2505: component of eq.(\ref{axionaction3}), together
2506: with the bulk gauge field kinetic terms and CS term, becomes:
2507: \bea
2508: \label{axionaction4}
2509: S_{axion} & = &
2510: %\nonumber \\ && \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
2511: \int d^4x \left[ \half\left(\partial_\mu \phi^{(-)}\right)^2
2512: +
2513: \tilde{m}^2\left(\frac{1}{\sqrt{2}}B^0+
2514: \sum_{n\;even}\!\! B^{n}_\mu \right)^2
2515: %\right.
2516: %\nonumber \\
2517: %& & \qquad \left.
2518: +\tilde{m}^2 \left(\sum_{n\;odd}B^{n}_\mu\right)^2
2519: \!\! - \tilde{m}B^{-}_\mu\partial^\mu
2520: \phi^{(-)}\right]
2521: \nonumber \\
2522: & &
2523: \!\!\!\!\!\!\!\!\!\!\!
2524: +\; \frac{c}{2\tilde{m}}\int d^4x\;\phi^{(-)}\;\epsilon^{\mu\nu\rho\sigma}
2525: \times
2526: \nonumber \\
2527: & &
2528: \;\;\;\;\; \left[\;
2529: e^2e' F^{B^0}_{\mu\nu} F^{B^0}_{\rho\sigma}
2530: + 2ee'{}^2 F^{B^0}_{\mu\nu}\sum_{n \;even} F^{n}_{\rho\sigma}
2531: %\right.
2532: %\nonumber \\
2533: %& & \!\!\!\!\!\!\!\!\!\!\!\!
2534: % + \left.
2535: + e'{}^3\sum_{n,m\; even} F^{n}_{\mu\nu} F^{m}_{\rho\sigma}
2536: +e'{}^3 \sum_{n,m\; odd} F^{n}_{\mu\nu} F^{m}_{\rho\sigma}\right]
2537: \nonumber \\
2538: & &
2539: \!\!\!\!\!\!\!\!\!\!\! +
2540: \int d^4x\;
2541: \left[
2542: %\frac{e^2e'}{12\pi^2} \sum_{n}c_{n00}B^n_\mu A_\nu \widetilde{F}_{A}^{\mu\nu}
2543: %\right.
2544: %\nonumber \\ &&
2545: %+\;
2546: %\frac{ee'{}^2}{12\pi^2} \sum_{nm}c_{nm0}
2547: %(B^n_\mu B^m_\nu \widetilde{F}_{A}^{\mu\nu}+ B^n_\mu A_\nu
2548: %\widetilde{F}^{m\mu\nu})
2549: %\nonumber \\ && \qquad\qquad \left.
2550: \;+\frac{1}{12\pi^2} \sum_{nmk}e_n e_m e_kc_{nmk}B^n_\mu B^m_\nu
2551: \widetilde{F}^{k\mu\nu}
2552: %\nonumber \\ & &\!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
2553: +\sum_{n}\left( -\frac{1}{4} F^n_{\mu\nu}F^{n\mu\nu}+
2554: \half M_n^2 B_\mu^n B^{n\mu}\right) \right].
2555: \nonumber \\ &&
2556: \eea
2557: This effective action is characterized by the presence of
2558: the physical axion, $\pi^-$, the photon mass term,
2559: $\tilde{m}^2 B^0_\mu B^{0\mu}$ and the pure CS term structure constants,
2560: $\bar{c}_{nmk} = c_{nmk}$.
2561:
2562:
2563:
2564: %\newpage
2565:
2566:
2567: \section{KK-mode Currents and Covariant Anomalies }
2568:
2569:
2570:
2571: We derive the currents
2572: of the theory of eq.(\ref{effective2}) by variation of the full action wrt
2573: $B^n_\mu$. The spinor currents are supplemented
2574: by current contributions from the Chern-Simons term:
2575: \bea
2576: \tilde{J}^{n}_\mu & = & \frac{\delta S}{\delta B^{n\mu}} =
2577: \bar{\psi}\gamma_\mu \psi|_{n\;even}
2578: +\bar{\psi}\gamma_\mu \gamma^5\psi|_{n\;odd} + J_{\mu}^{n\;CS}
2579: \eea
2580: where $J^{n\;CS}$ is the Chern-Simons current:
2581: \bea
2582: % \nonumber\\
2583: %&& \!\!\!\!\!\!\!\!\!\!\!\!\!\!\!
2584: J_{mu}^{n\;CS} = \frac{\epsilon_{\mu\nu\rho\sigma}}{12\pi^2}
2585: \sum_{mk}\bigl[(c_{nmk}-c_{mnk}+
2586: c_{kmn}-c_{mkn}) B^{m\nu }
2587: \partial^\rho {B}^{k\sigma} \bigr]
2588: %\nonumber\\
2589: \eea
2590: In what follows
2591: we use the fermionic current
2592: consistent anomalies as computed in \cite{bardeen}
2593: (Appendix C.4).
2594:
2595: We can now compute the current divergences.
2596: The consistent anomalies can be
2597: written in one compact formula for both axial
2598: and vector currents as (from eq.(\ref{bardeen11})):
2599: \bea
2600: \label{an}
2601: \partial^\mu J_\mu^n & = & \frac{1}{48\pi^2}
2602: \sum_{mk}\left( 1-(-1)^{n+m+k}\right){F}^m_{\mu\nu} \widetilde{F}^{k\mu\nu}
2603: \eea
2604: The $J^n_\mu$ are the axial currents
2605: for $n$ odd ($J^{(n\; odd)} = \bar{\psi}\gamma_\mu \gamma^5\psi $ )
2606: and vector currents for $n$ even
2607: ($J^{(n\; even)} = \bar{\psi}\gamma_\mu\psi $ ).
2608:
2609: Moreover, the divergence of the Chern-Simons current
2610: takes the form:
2611: \bea
2612: \partial^\mu J_{\mu}^{n\;CS} & = &
2613: \frac{1}{48\pi^2}\sum_{m,k}
2614: (c_{nmk}-c_{mnk}+ c_{nkm}-c_{knm})
2615: {F}^m_{\mu\nu} \widetilde{F}^{k\mu\nu}
2616: \eea
2617: where we have exploited the even symmetry under
2618: interchange of indices of $k\leftrightarrow m$
2619: to resymmetrize the summand.
2620: Thus, the full current divergence takes the form
2621: \bea
2622: \partial^\mu \tilde{J}^{n}_\mu & = &
2623: \frac{1}{24\pi^2}\sum_{m,k}
2624: d_{nmk}
2625: {F}^m_{\mu\nu} \widetilde{F}^{k\mu\nu}
2626: \eea
2627: where we have a remarkable identity:
2628: \bea
2629: \label{dd}
2630: d_{nmk} & = & \half \left[(1-(-1)^{n+m+k})+
2631: (c_{nmk}-c_{mnk}+ c_{nkm}-c_{knm})
2632: \right]
2633: \nonumber \\
2634: & = & \frac{3}{2}[(-1)^{n+m+k}-1]\frac{n^2(k^2+m^2-n^2)}{(k+m-n)
2635: (k+m+n)(k+n-m)(k-m-n)}
2636: \nonumber \\
2637: & = & \frac{3}{2}c_{nmk}
2638: \eea
2639: where the $c_{nmk}$ are defined in eq.(\ref{ccc}).
2640: Hence, we have the final result for the entire KK-tower
2641: of currents:
2642: \bea
2643: \partial^\mu \tilde{J}^{n}_\mu & = &
2644: \frac{1}{16\pi^2}\sum_{m,k}
2645: c_{nmk}
2646: {F}^m_{\mu\nu} \widetilde{F}^{k\mu\nu}
2647: \eea
2648: where $n$ even (odd) is a vector (axial vector) current.
2649: The conservation of the electromagnetic
2650: current is now manifest, since:
2651: \beq
2652: c_{0mk} = 0
2653: \eeq
2654: while the remaining $n\geq 0$ currents have anomalies.
2655:
2656: In particular, as a check of this result,
2657: if the theory contains only the $n=0$
2658: photon, $\gamma$, and the $n=1$ axial vector meson, $B$, we then have
2659: a conserved vector electromagnetic current and we find the
2660: divergence of the full axial vector current:
2661: \bea
2662: \label{vresult}
2663: \partial^\mu J_{\mu}^{5} & = &
2664: \frac{1}{16\pi^2}\left(c_{100}
2665: {F}_{\gamma\;\mu\nu} \widetilde{F}_{\gamma}^{k\mu\nu}
2666: + c_{111}
2667: {F}_{B\;\mu\nu} \widetilde{F}_{B}^{k\mu\nu}
2668: \right)
2669: \nonumber \\
2670: & = &
2671: \frac{1}{8\pi^2}
2672: {F}_{\gamma\;\mu\nu} \widetilde{F}_{\gamma}^{\mu\nu}
2673: +\frac{1}{24\pi^2}
2674: {F}_{B\;\mu\nu} \widetilde{F}_{B}^{\mu\nu}
2675: %\nonumber \\
2676: \eea
2677: The latter expression is Bardeen's result for the axial
2678: vector anomaly when he enforces a conserved vector current.
2679: The CS term is playing the role of Bardeen's counterterm, which
2680: brings the current divergences into the present form \cite{bardeen}.
2681:
2682: The general issue of anomaly consistency is that
2683: the full currents are sources of the gauge
2684: field equations of motion:
2685: \bea
2686: \partial^\mu F^0_{\mu\nu} & = & e\tilde{J}^0_\nu \nonumber \\
2687: \partial^\mu F^n_{\mu\nu} + M_n^2B_\nu^n & = & e'\tilde{J}^n_\nu \qquad (a\neq 0)
2688: \eea
2689: Since $\partial^\mu\partial^\nu F^0_{\mu\nu}=0$
2690: by antisymmetry of $F_{\mu\nu}$ we see that
2691: the full $\tilde{J}^0_\mu$ must be conserved, as indeed it is.
2692: Since the axial vector
2693: mesons have masses (and are Stueckelberg fields with imbedded
2694: longitudinal components) there
2695: is not an inconsistency here either, as the double divergence
2696: implies:
2697: \beq
2698: \label{eom}
2699: M_n^2\;\partial^\nu B_\nu^n = \partial^\nu \tilde{J}^n_\nu =
2700: \makebox{(anomaly)}_n.
2701: \eeq
2702: This is just the equation of motion of the longitudinal modes,
2703: the $A_5^n$ which are eaten by the $B_\mu^n$ fields.
2704:
2705: %\newpage
2706:
2707:
2708:
2709:
2710:
2711:
2712:
2713:
2714:
2715:
2716:
2717:
2718:
2719:
2720: \section{Triangle Diagrams }
2721:
2722: The present Appendix is schematic and we only quote
2723: the main results.
2724: A detailed description of these calculations
2725: is available elsewhere \cite{hill3}.
2726:
2727: The massless calculation yields a result equivalent to
2728: Bardeen's result for the consistent anomalies \cite{bardeen}.
2729: Bardeen performs, however, a
2730: massive spinor loop calculation and quotes the anomalies
2731: in $V\pm A$ form. All anomalies we obtain presently
2732: fully confirm Bardeen's result in both the massless and massive cases.
2733: We do, however, see a slight subtlety in the form of the
2734: pure current divergences expressed in the $V_L$ and $V_R$ forms
2735: obtained in the massive case (the current divergence are
2736: not the full anomaly in
2737: that case, since the $im\bar{\psi}\gamma^5\psi$
2738: term must be subtracted. We also obtain the effective operator
2739: description of the these currents used in the text.
2740:
2741: \subsection{Single Massless Weyl Spinor}
2742:
2743: Consider the action:
2744: \beq
2745: S_L = \int d^4x\; \bar{\psi}_L(i\slash{\partial} +\slash{V}_L)\psi_L
2746: \eeq
2747: where the guage fields:
2748: \beq
2749: V_{L\mu} = B^a_{L\mu} +B^b_{L\mu} + B^c_{L\mu}
2750: \eeq
2751: couple to the current:
2752: \beq
2753: J_{L\mu} = \bar{\psi}_L \gamma_\mu \psi_L
2754: \eeq
2755: and the components of
2756: $V_L$ have the respective masses $M^{a}$, $M^{b}$, $M^{c}$.
2757: We are compute the triangle loop with
2758: three distinct external fields, $B^a$, $B^b$ and $B^c$, these
2759: can be alternatively viewed as distinct momentum components of the
2760: single field $V$. If all three fields were identical (exact
2761: Bose invariance) the amplitude would vanish, since
2762: it would involve an operator $VVdV$ which is zero.
2763: It is the external momentum differences
2764: or flavor indices that distinguish these fields
2765: and allow non-zero operators such as $[B] \sim B^a B^b dB^c$
2766: and $[B] \sim B^a B^c dB^b$, \etc.
2767: In the massless Weyl fermion case of interest presently, we compute
2768: in a limit $M^a >> M^b \sim M^a \sim 0$. We can view
2769: this as an operator product expansion
2770: of the triangle diagrams in which the internal lines carrying
2771: $p^2 = M_a^2$ are treated as a short-distance expansion.
2772:
2773:
2774:
2775: \begin{figure}[t]
2776: \vspace{5cm}
2777: \special{psfile=triangle2.eps
2778: angle=0 hscale=40 vscale=40 hoffset=30 voffset=-100
2779: % angle=-90 hscale=60 vscale=50 hoffset=0 voffset=240
2780: }
2781: \vspace{4.5cm}
2782: \caption[]{
2783: %\small
2784: \addtolength{\baselineskip}{-.3\baselineskip}
2785: Bose symmetric triangle diagrams
2786: for $B^a(p)\rightarrow B^b(k)+B^c(q)$
2787: The external lines are on mass-shell, $p^2=M_a^2 $, $k^2=M_b^2$
2788: and $q^2=M_c^2$. The respective polarizations are $\epsilon^a_\mu$,
2789: $\epsilon^b_\mu$ and $\epsilon^c_\mu$. The internal momentum
2790: routing and integration momenta are chosen so that
2791: both diagrams have a common
2792: denominator. }
2793: \label{dirac2}
2794: \end{figure}
2795:
2796:
2797: With the particular choice of momentum
2798: routing in the figure, we have the following expression for the sum
2799: of the triangle diagram and its Bose symmetric counterpart,
2800: which have a common denominator:
2801: \bea
2802: % & &\!\!\!\!\! \!\!\!\!\!
2803: T & = & (-1)(i)^3 (i)^3 \int \frac{d^{4}\ell}{(2\pi)^{4}}
2804: \frac{N_1+N_2}{D} \nonumber \\
2805: N_1 & = & \Tr[\slash{\epsilon}_a L
2806: (\slash{\ell}-\slash{q})
2807: \slash{\epsilon}_c L(\slash{\ell})
2808: \slash{\epsilon}_b L(\slash{\ell}+\slash{k})]
2809: \nonumber \\
2810: N_2 & = & -\Tr[\slash{\epsilon}_a L
2811: (\slash{\ell}+\slash{k})
2812: \slash{\epsilon}_b L(\slash{\ell})
2813: \slash{\epsilon}_c L(\slash{\ell}-\slash{{q}})]
2814: \nonumber \\
2815: D & = &
2816: (\ell+k)^2(\ell^2)(\ell-q)^2
2817: \eea
2818: where,
2819: \beq
2820: p=k+q, \qquad L =\half(1-\gamma^5), \qquad R =\half(1+\gamma^5).
2821: \eeq
2822: The overall sign contains: $\times(i)^3$ (vertices; note
2823: that our vector potentials have the opposite sign to the
2824: conventions of Bjorken and Drell, hence flipping the vertex
2825: rule from $-i\gamma_\mu \rightarrow +i\gamma_\mu$)), $\times (i)^3$
2826: (propagators), $\times (-1)$ (Fermi statistics).
2827: In $N_2$ we've factored out an overall minus sign.
2828: Note that one must use extreme care to write the
2829: given correct cyclic ordering of
2830: the factors that make up the numerator, relative to
2831: the momentum routing signs \cite{hill2}. This affects the overall sign of
2832: the triangle loop with three gauge vertices (but is has
2833: no effect upon the $im\bar{\psi}\gamma^5\psi $ loop computed in the
2834: massive case).
2835:
2836: We unify the denominator using:
2837: \beq
2838: \frac{1}{ABC} = 2\int_0^1dy\int_0^y dz \frac{1}{(Az + B(y-z)+C(1-y))^3}
2839: \eeq
2840: The unified denominator becomes:
2841: \beq
2842: \frac{1}{D} = 2\int_0^1dy\int_0^y dz
2843: \frac{1}{( \ell^2 +2\ell\cdot(zk-(1-y)q)+zk^2 +(1-y)q^2 )^3}
2844: \eeq
2845: Shifting the loop momentum to a symmetric
2846: integration momenta, $\overline{\ell}$:
2847: \bea
2848: \ell & = & \overline{\ell} - zk+(1-y)q
2849: %\nonumber \\
2850: \eea
2851: the unified denominator becomes:
2852: \bea
2853: & & (\overline{\ell}^2 +z(1-z)k^2 +y(1-y)q^2
2854: %\nonumber \\ & & \qquad \qquad \qquad
2855: +2k\cdot qz(1-y))
2856: %\nonumber \\
2857: \eea
2858: (see the comment on shifting momenta below).
2859:
2860: We define the following vertex tensors :
2861: \bea
2862: \label{match}
2863: A & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^\mu_a{\epsilon}^\nu_b
2864: {\epsilon}^\rho_c k^\sigma \qquad \longleftrightarrow \qquad -i \bra{b,k;c,q }
2865: \epsilon_{\mu\nu\rho\sigma}B^{a\mu} B^{c\nu}\partial^\rho B^{b\sigma}\ket{
2866: a,p }
2867: \nonumber \\
2868: B & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^\mu_a{\epsilon}^\nu_b
2869: {\epsilon}^\rho_c q^\sigma \qquad \longleftrightarrow \qquad
2870: i \bra{ b,k;c,q } \epsilon_{\mu\nu\rho\sigma}B^{a\mu}B^{b\nu}
2871: \partial^\rho B^{c\sigma} \ket{a,p }
2872: \nonumber \\
2873: C & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^\mu_a{\epsilon}^\nu_b
2874: {k}^\rho q^\sigma \qquad \longleftrightarrow \qquad
2875: \half \bra{ b,k } F_{\mu\nu}^a \tilde{F}^{b\;\mu\nu} \ket{a,p }
2876: \nonumber \\
2877: D & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^\mu_a{\epsilon}^\nu_c
2878: {k}^\rho q^\sigma \qquad \longleftrightarrow \qquad
2879: -\half \bra{ c,q } F_{\mu\nu}^a \tilde{F}^{c\;\mu\nu} \ket{a,p }
2880: \nonumber \\
2881: E & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^\mu_b{\epsilon}^\nu_c
2882: {k}^\rho q^\sigma \qquad \longleftrightarrow \qquad
2883: \half \bra{ b,k;c,q } F_{\mu\nu}^b \tilde{F}^{c\;\mu\nu} \ket{ 0}
2884: \eea
2885: where we have indicated the corresponding
2886: operator matrix elements, $\bra{out }{\cal{O}}\ket{in} $
2887: and note:
2888: \beq
2889: \tilde{F}_{\mu\nu} = \half \epsilon_{\mu\nu\rho\sigma}F^{\rho\sigma}
2890: \eeq
2891: is the standard definition of the dual field strength.
2892:
2893: It is a well known ambiguity of the triangle loops in momentum
2894: space that shifting
2895: loop momenta can lead to residual terms, owing to the superficial
2896: linear divergence. Such terms can only be of the
2897: form $\propto [A] - [B]$, and they would yield
2898: an anomaly that does not respect Bose symmetry in
2899: the three $a$, $b$ and $c$ channels. For example,
2900: setting $\epsilon_a \rightarrow p_\mu$
2901: we obtain $\rightarrow -2[E] \sim -2F_L\tilde{F}_L $, while
2902: setting $\epsilon_c\rightarrow -q$ we obtain
2903: $\rightarrow +[C] \sim +F_L\tilde{F}_L$
2904: Imposing this symmetry on the triangle loops is a luxury we
2905: have only in the massless Weyl case, since we do not
2906: have to subtract
2907: $\bra{b,c}im\bar{\psi}_L\gamma^5\psi_R + h.c.\ket{0}$
2908: to obtain the anomaly.
2909: Imposing Bose symmetry on the anomaly as
2910: constraint on the calculation removes the surface term ambiguity.
2911: It turns out, however, for the particular
2912: momentum routing we have chosen the result for the anomaly
2913: is fully and nontrivially Bose symmetric,
2914: thus there is no surface term. Moreover,
2915: even the superficial log divergence (also $\propto [A] - [B]$)
2916: is cancelled, as we see below,
2917: and the triangle loops are UV finite.
2918:
2919:
2920: We now compute the triangle loops. Since we are mainly
2921: interested in a heavy KK mode decaying to low mass
2922: KK-modes, kinematically we have:
2923: \beq
2924: p = k+q \qquad M_b^2 = k^2 \approx 0 \qquad M_c^2 = q^2 \approx 0 \qquad
2925: M_a^2 \approx 2k\cdot q
2926: \eeq
2927: Hence the large $M_a^2$ limit corresponds to a symmetrical expansion
2928: in $k^2/2k\cdot q $ and $q^2/2k\cdot q$.
2929:
2930: For the large electron mass limit, $M_a^2 <<m^2$,
2931: we define the loop integrals with the usual Wick rotation
2932: on the loop energy
2933: $\bar\ell_0$ and a Euclidean momentum space cut-of $\Lambda^2$:
2934: \bea
2935: \int \frac{d^4\bar\ell}{(2\pi)^4}\frac{(1,\;\bar\ell^2)}{(\bar\ell^2 -m^2+i\epsilon)^3}
2936: & = & \left[ \frac{-i}{16\pi^2}\left( \frac{1}{2m^2 }\right),
2937: \frac{i}{16\pi^2}\left[\ln\left(\frac{\Lambda^2}{m^2}\right)
2938: - \frac{3}{2}\right]\right]
2939: \nonumber \\
2940: \int \frac{d^4\bar\ell}{(2\pi)^4}\frac{(1,\;\bar\ell^2)}{(\bar\ell^2 -m^2+i\epsilon)^4}
2941: & = &
2942: \left[ \frac{i}{16\pi^2}\left( \frac{1}{6(m^2)^2 }\right),
2943: \frac{-i}{16\pi^2}\left( \frac{1}{3(m^2)^2 }\right)\right]
2944: \eea
2945: The familiar Wick rotation is a counterclockwise rotation of the contour
2946: of the $\bar\ell_0$ integral in the complex plane. The rotation is
2947: clockwise to avoid the poles at $\pm\sqrt{\vec{l}^2+m^2}\mp i\epsilon$,
2948: in the resulting Euclidean integral. For us, $m^2$ is actually
2949: $m^2-z(1-y)M_a^2$ from the unified denominator.
2950: In the case of $M_a^2 >>m^2$ the
2951: the $\vec{\ell}^2$ integrals develop
2952: a cut structure. Nonetheless,
2953: the results are analytic functions of large
2954: $m^2$, and in the limit $M_a^2 >> m^2$
2955: we can simply replace $m^2\rightarrow -z(1-y)M_a^2$. In the
2956: massless spinor case the $zy$ integrals now acquire
2957: infrared singularities, but massive case and the
2958: the anomaly are infrared finite.
2959:
2960: The diagrams yield a superficial
2961: log divergence. With $k^2=q^2=0$ (the $\alpha_0$
2962: contribution) and upon doing the denominator unification integrals,
2963: this yields a finite result:
2964: \bea
2965: \label{an1}
2966: T_{\ln(\Lambda^2)}
2967: & = &
2968: \frac{2i}{16\pi^2}\;\int_0^1dy\int_0^y dz\;
2969: (2i)\big[(1-3z)A + (2-3y)B \big]\left[\ln\left(\frac{\Lambda^2}{-z(1-y)M_a^2 }\right)
2970: -\frac{3}{2} \right]
2971: \nonumber \\
2972: & = &
2973: -\frac{1}{24\pi^2}[A]+\frac{1}{24\pi^2}[B]
2974: \eea
2975: Note that $T_{\ln(\Lambda^2)}$ is Bose symmetric under interchange of
2976: the photon and $b$-KK mode (let $q\leftrightarrow k$, hence $M_b^2\rightarrow
2977: 0$,
2978: $\epsilon_a\leftrightarrow
2979: \epsilon_b$, and note $A\leftrightarrow B$). However,
2980: as mentioned above, this term cannot be the full result, because
2981: its anomaly is not Bose symmetric, and we indeed must keep
2982: residual finite terms.
2983:
2984: Note, as a check on
2985: the large $m^2$ case, that if the argument of
2986: the log, ${-\Lambda^2}/{z(1-y)M_a^2 }$ is replaced by ${\Lambda^2}/{m^2}$
2987: then $T_{\ln(\Lambda^2)} = 0$. The result is finite because:
2988: \bea
2989: \label{an2}
2990: \int_0^1dy\int_0^y dz\;
2991: \big[(1-3z)A + (2-3y)B \big]
2992: & = & 0
2993: \eea
2994: This furthermore implies that the imaginary part of
2995: the expression is vanishing.
2996:
2997: Combining all of
2998: terms of the Feynman diagrams
2999: yields the following full result for the triangle
3000: diagrams (see \cite{hill2} for the detailed calculation):
3001: \bea
3002: T & = &
3003: -\frac{1}{12\pi^2}[A] + \frac{1}{12\pi^2}[B]
3004: \nonumber \\
3005: & & +\frac{1}{4\pi^2}[Ck\cdot \epsilon_c + Ak\cdot q]\frac{I_b}{M_a^2}
3006: \nonumber \\
3007: & & -\frac{1}{4\pi^2}[C q\cdot \epsilon_c + Aq^2]\frac{I_c}{M_a^2}
3008: \nonumber \\
3009: & & +\frac{1}{4\pi^2}[D\epsilon_b\cdot k + B k^2 ]\frac{I_b}{M_a^2}
3010: \nonumber \\
3011: & & -\frac{1}{4\pi^2}[D\epsilon_b\cdot q + B k\cdot q ]\frac{I_c}{M_a^2}
3012: \nonumber \\
3013: & &
3014: -\frac{1}{4\pi^2\; M_a^2}[E](\epsilon_a\cdot k \; I'_b + \epsilon_a\cdot q \; I'_c)
3015: + {\cal{O}}(q^2,k^2).
3016: \eea
3017: The integrals
3018: $I_i$
3019: and $I'_i$ are infrared divergent in our expansion. The result
3020: is manifestly Bose symmetric only if we perform the unification
3021: integrals, $I_i$ and $I'_i$ with
3022: a Bose symmetric IR cut-off. For a particular choice of small
3023: IR cut-offs
3024: $x_i$, the leading log divergent terms are:
3025: \bea
3026: I_b & = & \int_0^{1-x_b}\int_0^y dz\;dy\;\frac{z(z-y)}{z(1-y)}
3027: = \half \ln(x_b) + k_b
3028: \nonumber \\
3029: I_c & = & \int_0^1\int_x^y dz\;dy\;\frac{(1-y)(z-y)}{z(1-y)}
3030: =\half \ln(x_c) + k_c
3031: \nonumber \\
3032: I'_{b}& = &\int_0^{1-x}\int_0^y dz\;dy\;\frac{2z-zy-z^2}{z(1-y)} = - \half \ln(x_b)
3033: +k'_b
3034: \nonumber \\
3035: I'_{c}& = & \int_0^1\int_x^y dz\;dy\;\frac{z+y-zy-y^2}{z(1-y)} = - \half \ln(x_c)
3036: +k'_c
3037: %\nonumber \\
3038: \eea
3039: Note that $I_b=I_c$ and $I'_b=I'_c$
3040: if $x_b = x_c$ and Bose symmetry is maintained.
3041: The physical cutoffs are of order
3042: $x_b \sim M_b^2/M_a^2$ and $x_c \sim M_c^2/M_a^2$,
3043: and $k$ and $k'$ are indeterminate.
3044: This can be replaced with a more physical procedure by
3045: resumming $k^2$ and $q^2$ into the denominators.
3046: The logarithmic IR singularities in, \eg, the $ q^2 =0$
3047: limit are presumably cancelled by collinear $\bar{\psi}\psi $
3048: propagation in the $B^c\rightarrow B^b+ \bar{\psi}+\psi$
3049: process, where $\bar{\psi}+\psi$ rescatter into a photon.
3050:
3051: Note, however, a final lemma that is relevant to the anomaly:
3052: \beq
3053: \label{sumI}
3054: I_a + I_b + I'_b + I'_c = 2
3055: \eeq
3056: which is an infra-red ``safe'' quantity.
3057:
3058: \subsection{Massless Weyl Spinor Anomaly}
3059:
3060:
3061: The amplitude we have just computed is:
3062: \beq
3063: T
3064: = \bra{b,c}T ...\;
3065: i\int d^4 x\; \exp(-ip\cdot x)\epsilon^{a}_{\mu}\bar{\psi}\gamma ^\mu
3066: \psi_L \; ... \ket{0}
3067: \eeq
3068: On the other hand, the amplitude we want is the matrix
3069: element of the current divergence:
3070: \bea
3071: W & = & \bra{b,c}T ...
3072: \int d^4 x\; \exp(-ip\cdot x)\partial_{\mu}\bar{\psi}\gamma ^\mu
3073: \psi_L \; ... \ket{0} \nonumber \\
3074: & = & \bra{b,c}T ...
3075: \int d^4 x\; (-\partial_{\mu}\exp(-ip\cdot x)) \bar{\psi}\gamma ^\mu \psi_L
3076: \; ... \ket{0}
3077: \nonumber \\
3078: & = & \bra{b,c}T ...
3079: \int d^4 x\; \exp(-ip\cdot x) ip_\mu \bar{\psi}\gamma ^\mu \psi_L
3080: \; ... \ket{0}
3081: \eea
3082: We thus obtain $W$ from $T$ by the replacement:
3083: \beq
3084: W = T(\epsilon^{a}_{\mu}\rightarrow p_\mu)
3085: \eeq
3086: Under this substitution we have:
3087: $[A]\rightarrow -[E]$, $[B]\rightarrow [E]$,
3088: $[C]\rightarrow 0$, $[D]\rightarrow 0$,
3089: and $\epsilon_a\cdot k \rightarrow k\cdot q$,
3090: $\epsilon_a\cdot q \rightarrow k\cdot q$.
3091:
3092: We thus obtain:
3093: \bea
3094: T & \rightarrow & W \; = \;
3095: \frac{1}{12\pi^2}[E] + \frac{1}{12\pi^2}[E] -
3096: \frac{1}{8\pi^2}(I_a + I_b + I'_b + I'_c)[E]
3097: \nonumber \\
3098: & = & -\frac{1}{12\pi^2}[E]\qquad
3099: \eea
3100: where we use eq.(\ref{sumI}).
3101: The result is infra red non singular.
3102: Note that we have the operator correspondence
3103: $[E]\rightarrow (1/4)F\tilde{F}$ from eq.(\ref{match}).
3104: Our result for the anomaly thus
3105: corresponds to the operator equation:
3106: \beq
3107: \label{resultL}
3108: \partial^\mu \bar{\psi}\gamma_\mu\psi_L =
3109: -\frac{1}{48\pi^2}F_{L\mu\nu}\tilde{F}^{\mu\nu}_L
3110: \eeq
3111: It is trivial to infer the right-handed current
3112: anomaly by flipping signs:
3113: \beq
3114: \partial^\mu \bar{\psi}\gamma_\mu\psi_R =
3115: \frac{1}{48\pi^2}F_{R\mu\nu}\tilde{F}^{\mu\nu}_R
3116: \eeq
3117: Defining $L=V-A$ and $R=V+A$ we can write this in the form:
3118: \bea
3119: \partial^\mu \bar{\psi}\gamma_\mu \psi & = &
3120: \frac{1}{12\pi^2}F_{V\mu\nu}\tilde{F}_A^{\mu\nu}
3121: \nonumber \\
3122: \partial^\mu \bar{\psi}\gamma_\mu\gamma^5\psi & = &
3123: \frac{1}{24\pi^2}(F_{V\mu\nu}\tilde{F}_V^{\mu\nu}
3124: +F_{A\mu\nu}\tilde{F}_A^{\mu\nu} )
3125: \eea
3126: which agrees
3127: with Bardeen's result for the left-right symmetric
3128: anomaly \cite{bardeen}.
3129:
3130: As a further check on the calculation,
3131: we can also examine the anomaly in the $B^c$ current,
3132: by letting
3133: $\epsilon_c\rightarrow -q$ (the minus
3134: sign occurs since $B^c$ is outgoing), and we take the $c$ field to be
3135: on-shell and massless, \ie.
3136: set $q^2=0$.
3137: Whence $[A] =[C]$ and $[B]=[D]=[E] =0$:
3138: \bea
3139: T & \rightarrow & -\frac{1}{12\pi^2} [C]
3140: \eea
3141: Using eq.(\ref{match}) this corresponds to eq.(\ref{resultL}),
3142: consistent with the $a$ channel result.
3143: Likewise, we can check the $B^b$ channel, and
3144: verify the same result.
3145: We can, furthermore, check the
3146: off-shell gauge invariance for $c$ identified
3147: with a photon, and $M_c^2 =0$,
3148: setting $\epsilon_c\rightarrow -q$ and examining
3149: the ${\cal{O}}(q^2)$ terms. These are found to cancel \cite{hill2}.
3150: This implies that the only non-gauge invariant
3151: part of the amplitude is the
3152: anomaly.
3153:
3154:
3155: \noindent
3156: \subsection{Finite Electron Mass}
3157:
3158:
3159: We now turn to the case
3160: of a finite, and large electron mass, where ``large''
3161: means in comparison to external momenta and masses.
3162: We carry out the analysis of the loops in
3163: the presence of the full electron mass term, with the couplings
3164: \beq
3165: \int d^4x\; \left( \bar{\psi}_L(i\slash{\partial} +\slash{V}_L)\psi_L
3166: +\bar{\psi}_R(i\slash{\partial} +\slash{V}_R)\psi_R
3167: -m(\bar{\psi}_L\psi_R
3168: + h.c.)\right)
3169: \eeq
3170: where we take separate $L$ and $R$ fields,
3171: $B_\mu^{a\;L,R}$:
3172: \beq
3173: V_{L\mu} = B^{aL}_\mu + B^{bL}_\mu + B^{cL}_\mu \qquad \qquad
3174: V_{R\mu} = B^{aR}_\mu + B^{bR}_\mu + B^{cR}_\mu
3175: \eeq
3176: Note that in comparison to the KK-mode normalizations used
3177: in the text we have:
3178: \beq
3179: B^n_{L\mu} = (-1)^n B^n_{\mu} \qquad \qquad
3180: B^n_{R\mu} = B^n_{\mu}
3181: \eeq
3182: We will implement this relationship subsequently, but presently
3183: we work in the independent and generic $V_L$, $V_R$ basis.
3184:
3185: We presently adopt an obvious generalized notation for vertices, \eg,
3186: \bea
3187: A^{LRL} & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^{L\mu}_a{\epsilon}^{R\nu}_b
3188: {\epsilon}^{L\rho}_c k^\sigma \;,
3189: \qquad\qquad
3190: A^{LRR} = \epsilon_{\mu\nu\rho\sigma}{\epsilon}^{L\mu}_a{\epsilon}^{R\nu}_b
3191: {\epsilon}^{R\rho}_c k^\sigma \qquad .\;.\;.
3192: \nonumber \\
3193: C^{LR} & = & \epsilon_{\mu\nu\rho\sigma}{\epsilon}^{L\mu}_a{\epsilon}^{R\nu}_b
3194: {k}^\rho q^\sigma \qquad .\;.\;.
3195: \eea
3196: and so forth.
3197:
3198: We have just computed the $LLL$
3199: ($RRR$) loops arising from the pure
3200: massless $\psi_L$ ($\psi_R$). In the case of a massive
3201: electron
3202: the $LLL$ ($RRR$) loops have the same
3203: numerator structure, but the denominator now
3204: contains electron mass terms:
3205: \bea
3206: D & = &
3207: [(\ell+k)-m^2][(\ell^2)^2-m^2][(\ell-q)^2-m^2]
3208: \eea
3209: This causes all of the previously computed $LLL$
3210: ($RRR$) terms to become suppressed
3211: in the large $m^2$ limit. For example, the $\alpha_0$ term
3212: previously computed for $m^2=0$
3213: now becomes:
3214: \bea
3215: \label{an3}
3216: T_{\ln(\Lambda^2)}
3217: & = &
3218: -\frac{1}{4\pi^2}\;\int_0^1dy\int_0^y dz\;
3219: \big[(1-3z)A + (2-3y)B \big]\left[\ln\left(\frac{\Lambda^2}{m^2-z(1-y)M_a^2}\right)
3220: -\frac{3}{2} \right]
3221: \nonumber \\
3222: & \longrightarrow &
3223: -\frac{M_a^2}{480\pi^2 m^2 }\left([A] - [B]\right),
3224: \eea
3225: and now vanishes in the large $m^2$ limit.
3226: All of the new terms of interest in the massive electron
3227: case arise from the numerator terms containing mass insertions.
3228: This represent mixing from $\psi_L$ to the $\psi_R$,
3229: and thus generates new vertices, such as $[A]^{LRL}$, \etc.
3230:
3231: We compute the triangle loops with a
3232: single pure left-handed $\epsilon^{aL}_\mu \gamma^\mu L$ vertex, carrying
3233: in momentum $p$, and again noting the
3234: the cyclic order in which numerator terms are written, we have:
3235: \bea
3236: % & &\!\!\!\!\! \!\!\!\!\!
3237: T_L & = & (-1)(i)^3(i)^3\int \frac{d^{4}\ell}{(2\pi)^{4}}
3238: \frac{N_1+N_2}{D} \nonumber \\
3239: N_1 & = &
3240: \Tr[\slash{\epsilon}_a L(\slash{\ell}-\slash{q}+m)
3241: (\slash{\epsilon}^L_c L+\slash{\epsilon}^R_c R) (\slash{\ell}+m)
3242: (\slash{\epsilon}^L_b L+\slash{\epsilon}^R_b R) (\slash{\ell}+\slash{{k}}+m)]
3243: \nonumber \\
3244: N_2 & = & -\Tr[\slash{\epsilon}_a L(\slash{\ell}+\slash{k}-m)
3245: (\slash{\epsilon}^L_b L+\slash{\epsilon}^R_b R)(\slash{\ell}-m)
3246: (\slash{\epsilon}^L_c L+\slash{\epsilon}^R_c R)(\slash{\ell}+\slash{q}-m)]
3247: \nonumber \\
3248: D & = &
3249: [(\ell+k)^2-m^2][\ell^2-m^2][(\ell-q)^2-m^2]
3250: \eea
3251: Note the sign flips in the momentum and $m$ terms in $N_1$
3252: and momenta in $N_2$, a consequence
3253: of having factored out an overall minus sign.
3254:
3255: We obtain the result (full details are available in \cite{hill2}):
3256: \bea
3257: % & &\!\!\!\!\! \!\!\!\!\!
3258: T_L & = &
3259: (-4im^2)\times 2\int_0^1 dy \int_0^y dz\; \int \frac{d^{4}\bar\ell}{(2\pi)^{4}}
3260: \nonumber \\ & & \left( \frac{-z[A]^{LRL} - y[B]^{LRL}
3261: +(1-z)[A]^{LLR} +(1-y)[B]^{LLR}
3262: +z[A]^{LRR} - (1-y)[B]^{LRR}}{
3263: (\bar\ell^2 +z(1-z)k^2 +y(1-y)q^2
3264: +2k\cdot qz(1-y) - m^2 )^3} \right)
3265: \nonumber \\
3266: \eea
3267: This result is negligible in the limit
3268: $k^2, \; 2k\cdot q,\; q^2 >> m^2$.
3269: However, in the limit of large $m^2$
3270: the result reduces to:
3271: \bea
3272: \label{TL}
3273: T_L & = & \frac{1}{24\pi^2}(
3274: [A]^{LRL} + 2[B]^{LRL}-2[A]^{LLR} -[B]^{LLR} -[A]^{LRR} + [B]^{LRR})
3275: \eea
3276: From this we can easily infer the result for a computation
3277: of the triangle loops with a single pure $\slash{\epsilon}_\mu^{aR}
3278: R$
3279: (right-handed) vertex:
3280: \bea
3281: T_R & = & -\frac{1}{24\pi^2}(
3282: [A]^{RLR} + 2[B]^{RLR}-2[A]^{RRL} -[B]^{RRL} -[A]^{RLL} + [B]^{RLL})
3283: \eea
3284: Combining these we have:
3285: \bea
3286: T_{L} +T_{R} & = & \frac{1}{24\pi^2}([A]^{LRL} + 2[B]^{LRL}
3287: -2[A]^{LLR} -[B]^{LLR} -[A]^{LRR} + [B]^{LRR}\nonumber \\
3288: & & -[A]^{RLR} - 2[B]^{RLR}
3289: + 2[A]^{RRL} +[B]^{RRL} +[A]^{RLL} - [B]^{RLL})
3290: \eea
3291:
3292: For KK-mode $B_\mu^n$ we have an $x^5$ wave-function
3293: parity of $(-1)^{n}$, and $B_{\mu L}^n = (-1)^{n}B_{\mu R}^n
3294: = B_{\mu}^n$.
3295: The KK-modes are normalized so that an axial vector
3296: (odd $n$) couples to $\bar{\psi}\gamma_\mu \gamma^5\psi$ with positive sign.
3297: Thus, we can write:
3298: \bea
3299: T_{L} +T_{R} & = & \frac{1}{24\pi^2}((-1)^{a+c}([A] + 2[B])
3300: -(-1)^{a+b}(2[A] +[B]) -(-1)^{a}([A] - [B])\nonumber \\
3301: & & -(-1)^{b}([A] + 2[B])
3302: + (-1)^{c}(2[A] +[B]) +(-1)^{b+c}([A] - [B])
3303: \eea
3304: This can be put into a compact final expression:
3305: \bea
3306: \label{final}
3307: T_{L} +T_{R} & = & \frac{1}{12\pi^2}(f_{abc}[A]^{} + g_{abc}[B])
3308: \eea
3309: where:
3310: \bea
3311: f_{abc} & = & \half((-1)^{a+c} -2(-1)^{a+b} -(-1)^{a}
3312: -(-1)^{b} +2(-1)^{c} +(-1)^{b+c}) \nonumber \\
3313: g_{abc} & = & \half(2(-1)^{a+c}-(-1)^{a+b} +(-1)^{a}
3314: -2(-1)^{b} +(-1)^{c} -(-1)^{b+c}).
3315: \eea
3316: Note that if $a + b + c $ is even, then $f = g =0$, which is
3317: the condition that a transition cannot occur!
3318: But, of course,
3319: the {\em condition that a transition can occur} is $a + b + c $ odd.
3320: When $a + b + c $ is odd, we can therefore write:
3321: \bea
3322: \label{final2}
3323: f_{abc} & = & -(-1)^{a}-(-1)^{b} +2(-1)^{c} \nonumber \\
3324: g_{abc} & = & (-1)^{a}-2(-1)^{b} +(-1)^{c}
3325: \eea
3326: Under $b\leftrightarrow c$ we have $A\leftrightarrow -B$ and
3327: thus $g_{abc}\leftrightarrow -f_{acb}$, which confirms Bose symmetry.
3328: Under the exchange
3329: $a\leftrightarrow b$ we have $B\rightarrow -B$ and
3330: $A\rightarrow A+B$ (since the $k$ in the $A$ vertex now becomes $-k-q$
3331: with the sign flip, since $a$ is incoming momentum $k+q$).
3332: Thus the vertex becomes:
3333: \beq
3334: T_{L} +T_{R}\rightarrow \frac{1}{12\pi^2}(f_{bac}[A]^{} + (f_{bac}-g_{bac})[B])
3335: \eeq
3336: and we immediately verify that $f_{bac} = f_{abc}$
3337: and $f_{bac}-g_{bac} =g_{abc}$.
3338: The amplitude is seen to be fully Bose symmetric (we leave the
3339: verification of $a\leftrightarrow c$ Bose symmetry to the reader).
3340:
3341: The vertex calculation can be represented by an operator of
3342: the form:
3343: \beq
3344: \label{triop}
3345: {\cal{O}}_3 = -\frac{1}{12\pi^2}\epsilon^{\mu\nu\rho\sigma}
3346: \sum_{nmk}a_{nmk}B^n_\mu B^m_\nu\partial_\rho B_\sigma^k
3347: \eeq
3348: where:
3349: \beq
3350: a_{nmk} = \half [1 - (-1)^{n+m+k}](-1)^{m+k}
3351: \eeq
3352: For the process $a \rightarrow b + c$
3353: the matrix element of ${\cal{O}}$ takes the form (we've multiplied
3354: by $+i$ from $e^{iS}$):
3355: \bea
3356: \label{an10}
3357: i\bra{a}{\cal{O}}\ket{b,c} & = & \frac{1}{12\pi^2} \big[
3358: (-a_{abc}+a_{bac}+a_{bca}-a_{cba})[B]
3359: %\nonumber \\
3360: %& &
3361: + (a_{acb}-a_{cab}+ a_{bca}-a_{cba})[A]
3362: \big]
3363: \nonumber \\
3364: \eea
3365: and we see that (for $a + b + c$ odd):
3366: \bea
3367: -a_{abc}+a_{bac}+a_{bca}-a_{cba} & = & g_{abc} \nonumber \\
3368: a_{acb}-a_{cab}+ a_{bca}-a_{cba} & = & f_{abc}
3369: \eea
3370:
3371:
3372:
3373: \subsection{Massive Left-Right Symmetric Anomaly}
3374:
3375:
3376: The current divergence,
3377: $\partial_\mu\bar{\psi}\gamma_\mu\psi_L$, is obtained by
3378: the replacement $\epsilon_\mu \rightarrow p_\mu$ in $T_L$.
3379: We thus have that
3380: $A \rightarrow -E$ and $B\rightarrow E$ and we make use of
3381: the operator correspondence eq.(\ref{match}):
3382: \bea
3383: \label{div1L}
3384: \partial_\mu\bar{\psi}\gamma_\mu\psi_L & = & \frac{1}{48\pi^2}(
3385: F^L_{\mu\nu}\tilde{F}^R_{\mu\nu}+F^R_{\mu\nu}\tilde{F}^R_{\mu\nu})
3386: \eea
3387: \bea
3388: \label{div1R}
3389: \partial_\mu\bar{\psi}\gamma_\mu\psi_R & = & -\frac{1}{48\pi^2}(
3390: F^L_{\mu\nu}\tilde{F}^R_{\mu\nu}+F^L_{\mu\nu}\tilde{F}^L_{\mu\nu})
3391: \eea
3392: We emphasize that this result is {\em not the anomaly}.
3393: To extract the anomaly, we note that the equations of motion
3394: yield the divergences of the spinor currents:
3395: \beq
3396: \partial^\mu \bar{\psi}\gamma_\mu \psi_L = -im(\bar{\psi}_L\psi_R
3397: -\bar{\psi}_R\psi_L) + \makebox{anomaly}
3398: \eeq
3399: \beq
3400: \partial^\mu \bar{\psi}\gamma_\mu \psi_R = -im(\bar{\psi}_R\psi_L
3401: -\bar{\psi}_L\psi_R) + \makebox{anomaly}
3402: \eeq
3403: We thus need to subtract the vacuum to 2-gauge field
3404: matrix element
3405: of the mass term, which is the operator $-im\bar{\psi}\gamma^5\psi$,
3406: to obtain the anomaly. The mass term yields
3407: a similar triangle diagram structure, and we define:
3408: \bea
3409: % & &\!\!\!\!\! \!\!\!\!\!
3410: M^5 & = &
3411: (-1)(i)^2(i)^3\int' \int \frac{d^{4}\ell}{(2\pi)^{4}}
3412: \frac{N_1+N_2}{D} \nonumber \\
3413: N_1 & = & (-i)(-im)\Tr[\gamma^5(\slash{\ell}-\slash{q}+m)
3414: (\slash{\epsilon}^L_c L+\slash{\epsilon}^R_c R) (\slash{\ell}+m)
3415: (\slash{\epsilon}^L_b L+\slash{\epsilon}^R_b R) (\slash{\ell}+\slash{{k}}+m)]
3416: \nonumber \\
3417: N_2 & = & (+i)(-im)\Tr[\gamma^5(\slash{\ell}+\slash{k}-m)
3418: (\slash{\epsilon}^L_b L+\slash{\epsilon}^R_b R)(\slash{\ell}-m)
3419: (\slash{\epsilon}^L_c L+\slash{\epsilon}^R_c R)(\slash{\ell}-\slash{q}-m)]
3420: \nonumber \\
3421: D & = &( \ell^2 +2\ell\cdot(zk-(1-y)q)+zk^2 +(1-y)q^2 -m^2 )^3
3422: \eea
3423: The result is (see \cite{hill2} for details):
3424: \bea
3425: M^5 & = & \bra{ 0} -im\bar{\psi}\gamma^5\psi \ket{b,c} = \frac{1}{24\pi^2}
3426: [2(E^{LL} + E^{RR})+ (E^{LR} + E^{RL})] \; ,
3427: \eea
3428: or, the operator correspondence:
3429: \bea
3430: im\bar{\psi}\gamma^5\psi \rightarrow
3431: -\frac{1}{48\pi^2}
3432: [F_L\tilde{F}_L + F_R\tilde{F}_R + F_L\tilde{F}_R] \; .
3433: \eea
3434: Forming the difference of the current
3435: divergence with $-im\bar{\psi}\gamma^5\psi$ we have:
3436: \bea
3437: \label{div}
3438: \partial_\mu\bar{\psi}\gamma_\mu\psi_L
3439: + im(\bar{\psi}_L\psi_R
3440: -\bar{\psi}_R\psi_L)
3441: & = & \frac{1}{48\pi^2}(
3442: F_L\tilde{F}_R+F_R\tilde{F}_R)
3443: -\frac{1}{48\pi^2}
3444: [F_L\tilde{F}_L + F_R\tilde{F}_R + F_L\tilde{F}_R]
3445: \nonumber \\
3446: & = & -\frac{1}{48\pi^2}
3447: F_L\tilde{F}_L \; .
3448: \eea
3449: Likewise:
3450: \bea
3451: \label{divR}
3452: \partial_\mu\bar{\psi}\gamma_\mu\psi_R
3453: + im(\bar{\psi}_R\psi_L
3454: -\bar{\psi}_L\psi_R)
3455: & = & - \frac{1}{48\pi^2}(
3456: F_R\tilde{F}_L+F_L\tilde{F}_L)
3457: +\frac{1}{48\pi^2}
3458: [F_L\tilde{F}_L + F_R\tilde{F}_R + F_L\tilde{F}_R]
3459: \nonumber \\
3460: & = & \frac{1}{48\pi^2}
3461: F_R\tilde{F}_R\; .
3462: \eea
3463:
3464:
3465: \newpage
3466: \subsection{Summary}
3467:
3468: \noindent
3469: {\bf Pseudoscalar Mass Term}:
3470: \bea
3471: im\bar{\psi}\gamma^5\psi \rightarrow
3472: -\frac{1}{48\pi^2}
3473: [F_{L\mu\nu} \tilde{F}_L^{\mu\nu}+F_{R\mu\nu} \tilde{F}_R^{\mu\nu}
3474: +F_{L\mu\nu} \tilde{F}_R^{\mu\nu}]
3475: \eea
3476:
3477:
3478: \noindent
3479: {\bf Consistent Anomalies}:
3480: \vskip 0.2in
3481:
3482: \noindent
3483: (1) Pure Massless Weyl Spinors ($p_i\cdot p_j >> m^2$):
3484: \bea
3485: \partial^\mu \bar{\psi}\gamma_\mu \psi_L & = &
3486: -\frac{1}{48\pi^2} F_{L\mu\nu} \tilde{F}_L^{\mu\nu}
3487: \nonumber \\
3488: \partial^\mu \bar{\psi}\gamma_\mu \psi_R & = &
3489: \frac{1}{48\pi^2} F_{R\mu\nu} \tilde{F}_R^{\mu\nu}
3490: \eea
3491:
3492: \noindent
3493: (2) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3494: \bea
3495: \partial^\mu \bar{\psi}\gamma_\mu \psi_L +im(\bar{\psi}_L\psi_R
3496: -\bar{\psi}_R\psi_L) & = & -\frac{1}{48\pi^2} F_{L\mu\nu} \tilde{F}_L^{\mu\nu}
3497: \nonumber \\
3498: \partial^\mu \bar{\psi}_R\gamma_\mu \psi_R + im(\bar{\psi}_R\psi_L
3499: -\bar{\psi}_L\psi_R) & = & \frac{1}{48\pi^2} F_{R\mu\nu} \tilde{F}_R^{\mu\nu}
3500: \eea
3501:
3502: \noindent
3503: (3) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3504: \bea
3505: \partial^\mu \bar{\psi}\gamma_\mu \psi_L & = &
3506: \frac{1}{48\pi^2}(
3507: F_{L\mu\nu} \tilde{F}_R^{\mu\nu} +
3508: F_{R\mu\nu} \tilde{F}_R^{\mu\nu} )
3509: \nonumber \\
3510: \partial^\mu \bar{\psi}\gamma_\mu \psi_R
3511: & = & -\frac{1}{48\pi^2} (
3512: F_{L\mu\nu} \tilde{F}_R^{\mu\nu}
3513: +F_{L\mu\nu} \tilde{F}_L^{\mu\nu})
3514: \eea
3515:
3516: \vskip 0.2in
3517:
3518:
3519: \noindent
3520: {\bf Consistent $L=V-A$ and $R=V+A$ Forms}:
3521: \vskip 0.2in
3522:
3523: \noindent
3524: (1) Pure Massless Weyl Spinors ($p_i\cdot p_j >> m^2$):
3525: \bea
3526: \partial^\mu \bar{\psi}\gamma_\mu \psi & = &
3527: \frac{1}{12\pi^2} F_{V\mu\nu} \tilde{F}_A^{\mu\nu}
3528: \nonumber \\
3529: \partial^\mu \bar{\psi}\gamma_\mu\gamma^5 \psi & = &
3530: \frac{1}{24\pi^2}( F_{V\mu\nu} \tilde{F}_V^{\mu\nu}
3531: + F_{A\mu\nu} \tilde{F}_A^{\mu\nu} )
3532: \eea
3533:
3534: \noindent
3535: (2) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3536: \bea
3537: \partial^\mu \bar{\psi}\gamma_\mu \psi & = & \frac{1}{12\pi^2} F_{V\mu\nu} \tilde{F}_A^{\mu\nu}
3538: \nonumber \\
3539: \partial^\mu \bar{\psi}\gamma_\mu \gamma^5\psi -2im\bar{\psi}\gamma^5\psi
3540: & = & \frac{1}{24\pi^2}( F_{V\mu\nu} \tilde{F}_V^{\mu\nu}
3541: + F_{A\mu\nu} \tilde{F}_A^{\mu\nu} )
3542: \eea
3543:
3544: \noindent
3545: (3) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3546: \bea
3547: \partial^\mu \bar{\psi}\gamma_\mu \psi & = &
3548: \frac{1}{12\pi^2} F_{V\mu\nu} \tilde{F}_A^{\mu\nu}
3549: \nonumber \\
3550: \partial^\mu \bar{\psi}\gamma_\mu \gamma^5 \psi
3551: & = & -\frac{1}{12\pi^2}( F_{V\mu\nu} \tilde{F}_V^{\mu\nu} )
3552: \eea
3553:
3554: \noindent
3555: {\bf Covariant Forms}:
3556: \vskip 0.2in
3557:
3558: \noindent
3559: Add a term to the lagrangian of
3560: the form $(1/6\pi^2)\epsilon_{\mu\nu\rho\sigma}A^\mu V^\nu
3561: \partial^\rho V^\sigma $. The currents are now modified to
3562: $\tilde{J} = J +\delta J$ and $\tilde{J}^5 = J^5 + \delta J^5$
3563: as described in the text.
3564:
3565: \noindent
3566: (1) Pure Massless Weyl Spinors ($p_i\cdot p_j >> m^2$):
3567: \bea
3568: \partial^\mu \tilde{J}_\mu & = & 0
3569: \nonumber \\
3570: \partial^\mu \tilde{J}^5_\mu & = &
3571: \frac{1}{8\pi^2}( F_{V\mu\nu} \tilde{F}_V^{\mu\nu}
3572: + \frac{1}{3}F_{A\mu\nu} \tilde{F}_A^{\mu\nu} )
3573: \eea
3574:
3575: \noindent
3576: (2) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3577: \bea
3578: \partial^\mu \tilde{J}_\mu & = & 0
3579: \nonumber \\
3580: \partial^\mu \tilde{J}^5_\mu - 2im\bar{\psi}\gamma^5\psi
3581: & = & \frac{1}{8\pi^2}( F_{V\mu\nu} \tilde{F}_V^{\mu\nu}
3582: +\frac{1}{3} F_{A\mu\nu} \tilde{F}_A^{\mu\nu} )
3583: \eea
3584:
3585: \noindent
3586: (3) Heavy Massive Weyl Spinors ($p_i\cdot p_j << m^2$):
3587: \bea
3588: \partial^\mu \tilde{J}_\mu & = & 0
3589: \nonumber \\
3590: \partial^\mu \tilde{J}^5_\mu
3591: & = & 0
3592: \eea
3593: The latter case is completely summarized by the fact
3594: that, for KK-modes,
3595: the three-gauge boson amplitude is described by the operator:
3596: \beq
3597: \label{triop2}
3598: {\cal{O}}_3 = -\frac{1}{12\pi^2}\epsilon^{\mu\nu\rho\sigma}
3599: \sum_{nmk}a_{nmk}B^n_\mu B^m_\nu\partial_\rho B_\sigma^k
3600: \eeq
3601: where:
3602: \beq
3603: a_{nmk} = \half [1 - (-1)^{n+m+k}](-1)^{m+k}
3604: \eeq
3605: This operator is equivalent to
3606: $(-1/6\pi^2)\epsilon_{\mu\nu\rho\sigma}A^\mu V^\nu
3607: \partial^\rho V^\sigma $
3608: when we truncate on the first two KK-modes,
3609: and identify $B^0 = V$ and $B^1=A$.
3610: Adding the
3611: $(1/6\pi^2)\epsilon_{\mu\nu\rho\sigma}A^\mu V^\nu
3612: \partial^\rho V^\sigma $ term cancels this quantity,
3613: completely cancels the triangle diagrams,
3614: and the resulting currents then have vanishing divergences.
3615:
3616:
3617:
3618:
3619:
3620: \newpage
3621: \vskip .2in
3622: \noindent
3623: {\bf Acknowledgments}
3624: \vskip .1in
3625: We especially thank
3626: Bill Bardeen and Bogdan Dobrescu for numerous
3627: helpful discussions. We also wish to thank Tao Han and
3628: Sherwin Love for helpful discussions in the early phase of this work.
3629: This work is supported in part by
3630: the US Department of Energy, High Energy Physics Division,
3631: Contract W-31-109-ENG-38, and
3632: grant DE-AC02-76CHO3000.
3633:
3634:
3635: \begin{thebibliography}{99}
3636:
3637:
3638: \bibitem{schonfeld}
3639: J.~F.~Schonfeld,
3640: %``A Mass Term For Three-Dimensional Gauge Fields,''
3641: Nucl.\ Phys.\ B {\bf 185}, 157 (1981);\\
3642: %%CITATION = NUPHA,B185,157;%%
3643: W.~Siegel,
3644: %``Unextended Superfields In Extended Supersymmetry,''
3645: Nucl.\ Phys.\ B {\bf 156}, 135 (1979).
3646: %%CITATION = NUPHA,B156,135;%%
3647: S.~Deser, R.~Jackiw and S.~Templeton,
3648: %``Topologically Massive Gauge Theories,''
3649: Annals Phys.\ {\bf 140}, 372 (1982)
3650: [Erratum-ibid.\ {\bf 185}, 406.1988\ APNYA,281,409
3651: (1988\ APNYA,281,409-449.2000)];
3652: %%CITATION = APNYA,140,372;%%
3653: S.~Deser, R.~Jackiw and S.~Templeton,
3654: %``Three-Dimensional Massive Gauge Theories,''
3655: Phys.\ Rev.\ Lett.\ {\bf 48}, 975 (1982).
3656: %%CITATION = PRLTA,48,975;%%
3657:
3658:
3659: \bibitem{wess}
3660: J.~Wess and B.~Zumino,
3661: %``Lagrangian Method For Chiral Symmetries,''
3662: Phys.\ Rev.\ {\bf 163}, 1727 (1967).
3663: %%CITATION = PHRVA,163,1727;%%
3664:
3665: \bibitem{witten}
3666: E.~Witten,
3667: %``Global Aspects Of Current Algebra,''
3668: Nucl.\ Phys.\ B {\bf 223}, 422 (1983).
3669: %%CITATION = NUPHA,B223,422;%%
3670:
3671: \bibitem{zachos}
3672: C.~T.~Hill and C.~K.~Zachos,
3673: %``Dimensional deconstruction and Wess-Zumino-Witten terms,''
3674: Phys.\ Rev.\ D {\bf 71}, 046002 (2005).
3675: %%CITATION = HEP-TH 0411157;%%\\
3676:
3677: \bibitem{hill4} C.~T.~Hill,
3678: ``Exact Equivalence of the
3679: $D=4$ Gauged Wess-Zumino-Witten Term
3680: and the $D=5$ Yang-Mills Chern-Simons Term,''
3681: (to appear).
3682:
3683: %\cite{Arkani-Hamed:1999za}49)
3684: \bibitem{nima}
3685: N.~Arkani-Hamed, Y.~Grossman and M.~Schmaltz,
3686: %``Split fermions in extra dimensions and exponentially small cross-sections
3687: %at future colliders,''
3688: Phys.\ Rev.\ D {\bf 61}, 115004 (2000);\\
3689: E.~A.~Mirabelli and M.~Schmaltz,
3690: %``Yukawa hierarchies from split fermions in extra dimensions,''
3691: Phys.\ Rev.\ D {\bf 61}, 113011 (2000);\\
3692: G.~C.~Branco, A.~de Gouvea and M.~N.~Rebelo,
3693: %``Split fermions in extra dimensions and CP violation,''
3694: Phys.\ Lett.\ B {\bf 506}, 115 (2001);\\
3695: W.~F.~Chang, I.~L.~Ho and J.~N.~Ng,
3696: %``Lepton universality, rare decays and split fermions,''
3697: Phys.\ Rev.\ D {\bf 66}, 076004 (2002);\\
3698: W.~F.~Chang and J.~N.~Ng,
3699: %``CP violation in 5D split fermions scenario,''
3700: JHEP {\bf 0212}, 077 (2002);\\
3701: H.~V.~Klapdor-Kleingrothaus and U.~Sarkar,
3702: %``Majorana neutrinos with split fermions in extra dimensions,''
3703: Phys.\ Lett.\ B {\bf 541}, 332 (2002);\\
3704: Y.~Grossman and G.~Perez,
3705: %``Realistic construction of split fermion models,''
3706: Phys.\ Rev.\ D {\bf 67}, 015011 (2003);\\
3707: B.~Lillie and J.~L.~Hewett,
3708: %``Flavor constraints on split fermion models,''
3709: Phys.\ Rev.\ D {\bf 68}, 116002 (2003);\\
3710: Y.~Grossman, R.~Harnik, G.~Perez, M.~D.~Schwartz and Z.~Surujon,
3711: %``Twisted split fermions,''
3712: Phys.\ Rev.\ D {\bf 71}, 056007 (2005);
3713:
3714: \bibitem{jackiw}
3715: J.~S.~Bell and R.~Jackiw,
3716: %``A Pcac Puzzle: Pi0 $\to$ Gamma Gamma In The Sigma Model,''
3717: Nuovo Cim.\ A {\bf 60}, 47 (1969).
3718: %%CITATION = NUCIA,A60,47;%%\\
3719:
3720:
3721: \bibitem{adler}
3722: S.~L.~Adler,
3723: %``Axial Vector Vertex In Spinor Electrodynamics,''
3724: Phys.\ Rev.\ {\bf 177}, 2426 (1969).
3725: %%CITATION = PHRVA,177,2426;%%
3726:
3727: \bibitem{bardeen}
3728: W.~A.~Bardeen,
3729: %``Anomalous Ward Identities In Spinor Field Theories,''
3730: Phys.\ Rev.\ {\bf 184}, 1848 (1969).
3731: %%CITATION = PHRVA,184,1848;%%\\
3732:
3733: \bibitem{jackiw2}
3734: See also: I.~S.~Gerstein and R.~Jackiw,
3735: %``Anomalies In Ward Identities For Three-Point Functions,''
3736: Phys.\ Rev.\ {\bf 181}, 1955 (1969).
3737: %%CITATION = PHRVA,181,1955;%%
3738:
3739:
3740: \bibitem{skiba} This can be studied in the
3741: context of deconstruction: W.~Skiba and D.~Smith,
3742: %``Localized fermions and anomaly inflow via deconstruction,''
3743: Phys.\ Rev.\ D {\bf 65}, 095002 (2002);
3744: C.~T.~Hill and A.~K.~Leibovich,
3745: %``Natural theories of ultra-low mass PNGB's: Axions and quintessence,''
3746: Phys.\ Rev.\ D {\bf 66}, 075010 (2002)
3747: %%CITATION = HEP-PH 0205237;%%
3748:
3749: \bibitem{fujikawa}
3750: K.~Fujikawa,
3751: %``Path Integral For Gauge Theories With Fermions,''
3752: Phys.\ Rev.\ D {\bf 21}, 2848 (1980)
3753: [Erratum-ibid.\ D {\bf 22}, 1499 (1980)].
3754: %%CITATION = PHRVA,D21,2848;%%
3755: %%Cited 849 times in SPIRES-HEP
3756:
3757: \bibitem{hill2} C. T. Hill and C. Zachos, Work in progress.
3758:
3759: \bibitem{hill3} C. T. Hill, ``Lecture Notes for Massive Spinor
3760: and Massless Spinor Triangle Diagrams,'' hep-th/0601155.
3761:
3762: \end{thebibliography}
3763:
3764: \end{document}
3765:
3766:
3767:
3768:
3769:
3770: