1: \documentclass[twocolumn,eqsecnum,aps]{revtex4}
2: %\documentclass[preprint,aps]{revtex4}
3: %
4: %
5: %
6: %
7: \usepackage{graphicx}
8: %\topmargin -0.5in
9: %\textheight 8.5in
10: %\textwidth 6in
11: %\evensidemargin .125in
12: %\oddsidemargin .125in
13:
14: %\input boxedeps.tex
15: %\SetRokickiEPSFSpecial %% dvips by Tom Rokicki
16: %\ShowDisplacementBoxes
17: %\HideDisplacementBoxes
18: \newcommand{\sit}{\sigma_{T}}
19: \newcommand{\g}{\gamma}
20: \newcommand{\sil}{\sigma_{L}}
21: \newcommand{\D}{\overline{D}}
22: \newcommand{\da}{\dagger} % symbol for Hermitian conjugate dagger
23: \newcommand{\be}{\begin{equation}}
24: \newcommand{\eq}{\end{equation}}
25: \newcommand{\I}{{\sf I}}
26: \newcommand{\Tr}{{\rm \, Tr \!}} %symbol for trace
27: \newcommand{\integer}{\cal Z} % symbol for set of integers
28: \newcommand{\newl}{l} % dimensionless lambda
29: \newcommand{\newtau}{t} % dimensionless tau
30: \newcommand{\firstpaper}{{\mathbf I}} % symbol for our first paper
31: \newcommand{\klink}{P^+_{\rm link}} % sum of link momenta (heavy quark)
32: \newcommand{\dklink}{K_{\rm link}} % sum of discretised link momenta
33: \newcommand{\kmax}{{P_{\rm max}}} % maximum link momenta (heavy quark)
34: \newcommand{\dkmax}{K_{\rm max}} % maximukm discretised link momenta
35: \newcommand{\dm}{{\cal M}} % coefficients of dispersion
36:
37: \begin{document}
38:
39: % PRD setup..
40: %
41: %\baselineskip5pt
42: %\def\btt#1{{\tt$\backslash$#1}}
43: %\begin{document}
44: %\draft
45: \title{$SU(N_c \to \infty)$ Lattice Data and Degrees of Freedom
46: of the QCD string.}
47: \author{S. Dalley}
48: \affiliation{Department of Physics, University of Wales Swansea,
49: Singleton Park,
50: Swansea SA2 8PP, United Kingdom}
51: %%\date{\today}
52: \preprint{SWAT/05/453}
53: \pacs{}
54: %{\tt$\backslash$\string pacs\{\}} should always be input,
55: %even if empty.}
56: %\narrowtext
57:
58: %\end{center}
59: \begin{abstract}
60: Lattice simulation data on the critical temperature
61: and long-distance potential, that probe the
62: degrees of freedom of the QCD string, are critically reviewed.
63: It is emphasized that comparison of experimental or $SU(N_c)$ lattice
64: data, at finite number of colors $N_c$, with free string theory can be
65: misleading due to string interactions. Large-$N_c$
66: extrapolation of pure lattice gauge
67: theory data, in both 3 and 4 dimensions,
68: indicates that there are more worldsheet degrees of freedom than the purely
69: massless transverse ones of the free
70: Nambu-Goto string. The extra variables are consistent
71: with massive modes of oscillation that effectively contribute like
72: $c \approx 1/2$
73: conformal degrees of freedom to highly excited states.
74: As a concrete example, the highly excited spectrum of the
75: Chodos-Thorn relativistic string in $1+1$ dimensions is analyzed,
76: where there are no
77: transverse oscillations.
78: We find that the asymptotic density of states for this
79: model is characteristic of a $c=1/2$ conformal worldsheet theory.
80: The observations made here should also constrain the backgrounds of
81: holographic string models for QCD.
82:
83:
84: \end{abstract}
85:
86: \maketitle
87:
88: %\newpage
89: %\baselineskip .2in
90:
91: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
92: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
93: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
94:
95:
96:
97:
98: \section{Introduction}
99: \label{intro}
100:
101: It is well-known that the long-wavelength properties of confining gauge
102: theories can be
103: modelled quite accurately by thin, structureless relativistic strings
104: oscillating transverse to themselves. The prototypical example is the
105: Nambu-Goto model \cite{nambu}, with action proportional to worldsheet area.
106: At shorter wavelengths,
107: additional properties or degrees of freedom are likely to play a role.
108: A related
109: and popular
110: suggestion has been that the string is always thin and structureless,
111: but that the additional effects can effectively be seen
112: as motion in a curved higher dimension
113: \cite{maldacena,polyakov}.
114: One would like to characterize any deviations from structureless
115: strings in spacetime in order to determine what, if any, string theory exactly
116: describes confining gauge theories at all length scales.
117: In this paper, data from lattice simulations of $SU(N_c)$
118: pure gauge theories will be used as a guide.
119: It is only quite recently that such data have become accurate and
120: extensive enough to be useful in this regard. Based on this analysis,
121: a concrete
122: suggestion is made for additional degrees of freedom that must be included
123: at short wavelengths ---
124: longitudinal modes of oscillation similar
125: to those of the Chodos-Thorn massive string \cite{chodos}.
126:
127: The gauge theory data used are the asymptotic
128: density of states in the
129: hadron spectrum, the (related) Hagedorn temperature, and the
130: long-distance groundstate potential energy
131: of a winding string or pair of heavy sources. These probe the short and
132: long wavelength regimes respectively. The existence in gauge theory
133: of a Hagedorn temperature in particular ---
134: which is to be distinguished from the
135: deconfinement temperature in general --- is an important indication that a
136: string description exists on all length scales.
137: It is not the intention to provide here a
138: complete historical review of lattice data;
139: only the most accurate and relevant results have been selected.
140: Data from 3 as well as 4
141: spacetime dimensions are used, since pure gauge theory seems to be described by
142: a string theory in both cases.
143: Data for fixed, finite $N_c$, which are often used to
144: address such questions, can be misleading if compared with a free string
145: model. It is shown that string interactions of strength $1/N_c$ can mask
146: the effects of
147: additional worldsheet degrees of freedom. To identify the latter
148: unambiguously, one must extrapolate data to $N_c=\infty$.
149: Large-$N_c$ results are shown to
150: indicate a Hagedorn temperature below that of the
151: Nambu-Goto free string, equivalent to adding an extra effective
152: conformal degree
153: of freedom with central charge $c \approx 1/2$. However, the
154: ``Luscher-term'' \cite{luscheretal,luscher},
155: in the asymptotic $1/l$ expansion of the groundstate
156: energy of a gauge string with minimal
157: allowed length $l$, seems to be consistent with the Nambu-Goto result
158: and not to depend on $N_c$. These two statements are consistent if
159: the additional worldsheet degrees of freedom, beyond those of Nambu-Goto,
160: are massive. Only at energies large compared to their mass is a
161: conformal approximation valid.
162:
163: In the second part of the paper, it is suggested that massive longitudinal
164: oscillations are responsible for the extra
165: degrees of freedom. This is hardly an original observation. However,
166: new evidence is presented that such oscillations have roughly the right
167: number of degrees of freedom to account for the facts displayed in the first
168: part of the paper. The Chodos-Thorn massive relativistic string
169: is essentially the Nambu-Goto model with additional massive pointlike
170: insertions of energy-momentum on the string. Although this
171: model is difficult to solve in greater than two dimensions,
172: the asymptotic spectrum is found exactly in two dimensions, where only
173: longitudinal oscillations remain.
174: It is found that these are equivalent to those
175: of a $c=1/2$ worldsheet field theory.
176: Whether this is coincidence and we are barking up the wrong tree
177: remains to be seen. In any event, the lattice data strongly constrain any
178: purported string theory of QCD.
179:
180:
181: \section{$SU(N_c \to \infty)$ data}
182:
183: \subsection{Stringy observables}
184:
185: Let us first briefly review the physical observables used later.
186: The boundstate spectra of string theories
187: generically have a density of states,
188: $\rho(M)$ per unit interval of mass $M$, that grows exponentially
189: for large $M$.
190: If, on the worldsheet of the {\em free} string, there is a conformal field
191: theory of central charge $c$ representing physical oscillations (not
192: counting those elimated by reparameterization invariance etc.), then
193: as $M \to \infty$ \cite{vafa}
194: \be
195: \rho(M) \propto M^{-(3+D_\perp)/2}
196: {\rm exp} \left(\frac{M}{T_H}\right) \ ,
197: \label{exp}
198: \eq
199: where
200: \be
201: T_H = \sqrt{\frac{3 \sigma}{c \pi}} \ ,
202: \eq
203: $\sigma$ is the string tension, and $D_\perp$ is the effective number of
204: dimensions for transverse oscillations.
205: Note that the same asymptotic density of states can be expected even
206: if the worldsheet theory is massive, once the excitation energies
207: are far above the relevant mass scale. The theory is effectively
208: conformal in this regime.
209: As is well-known, the canonical partition function of the free string
210: gas diverges above the Hagedorn temperature $T > T_H$ \cite{hagedorn}.
211: The physics of this point
212: is a second order phase transition \cite{parisi}
213: --- we assume the power law corrections in
214: eq.~(\ref{exp}) are such that the internal energy does not diverge at $T_H$ ---
215: driven by the entropy of strings.
216: The transition is also
217: signalled by the vanishing of the mass $E_c(1/T)$ of the string
218: that winds once around the compact Euclidean time direction of circumference
219: $1/T$ in the
220: finite-temperature partition function \cite{olesen}.
221: All these statements have an analogue in confining gauge theories.
222: The (real) hadronic spectrum does rise exponentially \cite{broniowski}
223: and there are phase
224: transitions in pure gauge theory
225: at which winding Polyakov loops get a VEV \cite{polyakov2}.
226: When those transitions
227: are second order, the Polyakov loop mass vanishes and one may assume
228: this is a Hagedorn transition at $T=T_H$. If the gluon entropy is more
229: important than the string entropy, which tends to be the case for larger
230: $N_c$ \cite{1storder,bern,teper05},
231: a first order transition at $T_c<T_H$ will occur first.
232: However, it has been shown by Teper and Bringoltz \cite{bringoltz}
233: that one may study the approach to $T_H$ in the metastable superheated phase
234: above $T_c$;
235: in fact, this phase should become stable when $N_c \to \infty$.
236:
237: Another measure of the degrees of freedom of string theory results
238: from the asymptotic $1/l$ expansion \cite{luscheretal} of the groundstate
239: energy $E_o(l)$ of an open string with
240: endpoints at fixed separation $l$:
241: \be
242: E_o(l) = \sigma L + {\rm const.} -\frac{\pi\tilde{c}}{24 l} + \cdots \ .
243: \label{open}
244: \eq
245: The coefficient of the ``Luscher-term'',
246: with $\tilde{c} = c - 24h$, is related
247: to the Casimir energy of free massless fields \cite{luscher}
248: of central charge $c$
249: and the lowest dimension $h$ of primary field that propagates on the
250: worldsheet. There is a similar expression for a closed
251: string \cite{deF} that winds once around a compact spatial direction of
252: circumference $l$:
253: \be
254: E_c(l) = \sigma L + {\rm const.} -\frac{\pi\tilde{c}}{6 l} + \cdots \ .
255: \label{closed}
256: \eq
257: The analogues of these configurations in
258: pure gauge theory are the potential between heavy sources in the fundamental
259: represenation of $SU(N_c)$ and the Polyakov loop mass. As well as the
260: $l \to \infty$ limit, we will also be interested in the minimum allowed
261: $l$, since $E_c(1/T_H) = 0$ can be taken to define the Hagedorn temperature.
262:
263:
264: \subsection{Why $3 \neq \infty$}
265:
266: The original connection between string worldsheets and confining gauge
267: theories was made by 't Hooft \cite{hoof},
268: in the case of weak gauge coupling $g$
269: expansion, and by Wilson \cite{wilson},
270: in the case of strong gauge coupling expansion.
271: In both cases, the string coupling $g_s$, by which we mean the (logarithm
272: of the) coupling to the worldsheet topological invariant
273: \be
274: \int dx^0 dx^1 \ R \sqrt{- G} \ ,
275: \label{top}
276: \eq
277: is $1/N_c$. Here, $G= \det G_{\alpha \beta}$, where the
278: induced worldsheet metric is
279: \be
280: G_{\alpha \beta} = \frac{\partial X^\mu}{\partial x^\alpha}
281: \frac{\partial X_\mu}{\partial x^\beta} \ ,
282: \eq
283: $R$ its Gaussian curvature, and $X^\mu$ the
284: worldsheet embedding co-ordinate in spacetime of dimension $D$,
285: $\mu \in \{0,\cdots D-1\}$, and $\{x^0,x^1\}$ the intrinsic
286: coordinates on the worldsheet.
287: Eq.~(\ref{top}) governs the splitting and joining interactions of strings.
288: In the modern
289: era also, the ADS/CFT correspondence \cite{maldacena}
290: and its non-conformal non-supersymmetric
291: generalisations
292: are between $SU(N_c)$ gauge theories and
293: fundamental string theory with coupling $g_s \propto 1/N_c$ in the $N_c \to
294: \infty$ limit.
295: If we are trying to phenomenologically
296: determine the rest of the worldsheet action that
297: must be added to (\ref{top}) in gauge theories,
298: the presence of string interactions greatly
299: complicates matters. Most results for string theories
300: have been derived to leading orders in the string coupling expansion.
301: In particular, the spectrum of string states is known exactly for several
302: string actions, but only for free strings.
303:
304: The Nambu-Goto model adds
305: to the worldsheet action the area term:
306: \be
307: - \sigma \int dx^0 dx^1 \sqrt{-G}.
308: \label{NG}
309: \eq
310: By lightcone gauge fixing of reparameterization invariance and
311: Fourier decomposition of transverse coordinates ($i \in \{ 1 , \cdots D-2\}$)
312: for open strings
313: \be
314: X^i = X^{i}_0 + P^i x^0 + {\rm i} \sum_{n \neq 0}
315: \alpha_{n}^{i} {\rm e}^{-{\rm i} n x^0} \cos nx^1 \ ,
316: \eq
317: free strings ($g_s = 0$)
318: in the quantized model have mass spectrum operator
319: given by a sum of harmonic oscillators \cite{thorn}
320: \be
321: M^2 = 2 \pi \sigma \sum_{n=1}^{\infty}
322: \alpha_{-n}^{i} \alpha_{n}^{i} \ ,
323: \label{spec}
324: \eq
325: \be
326: [\alpha_{n}^{i}, \alpha_{m}^{j}] = n \delta_{nm} \delta^{ij} \ .
327: \eq
328: The Nambu-Goto model is a classical conformal field theory in two dimensions
329: with $c= D-2 = D_\perp$, $h=0$. Therefore, it predicts
330: \be
331: \tilde{c} = D-2 \ ,\ T_H = \sqrt{\frac{3 \sigma}{ (D-2) \pi}} \ .
332: \eq
333: In particular
334: \begin{eqnarray}
335: T_H & \approx & 0.98 \sqrt{\sigma} \ (D=3) \ ,\label{d3}\\
336: T_H & \approx & 0.69 \sqrt{\sigma} \ (D=4) \ .\label{d4}
337: \end{eqnarray}
338: Although this model is not Lorentz invariant for $D \neq 26$, the action
339: should probably be understood as the long distance approximation
340: of a more general one
341: describing gauge theory strings.
342: In principle one can include further dimensionful interactions
343: between the transverse degrees of freedom $X^i$
344: or add further degrees of freedom.
345: This procedure can be made formal through the $1/l$
346: expansion \cite{luscheretal}, and there is some indication that
347: consistency problems can be fixed order-by-order \cite{polchinski}.
348: If the additional degrees of freedom
349: are massive, they could be integrated out systematically
350: to leave higher order corrections to the Nambu-Goto action.
351:
352: At first sight, it appears that the simple Nambu-Goto
353: model, in the case of free strings, is in excellent agreement with
354: data on the observables mentioned above in $SU(3)$ gauge theory (the case
355: relevant for QCD). Firstly, recent precision lattice data on the value
356: of the Luscher coefficient,
357: indicate that $\tilde{c} = D-2$ \cite{luscher2,sommer}. Indeed,
358: this also appears to be the case
359: for other gauge groups, including
360: $Z_2$ \cite{caselle,wako}, $SU(2)$ \cite{lucini,wako}, and
361: $SU(N_c \to \infty)$ \cite{meyer}, suggesting
362: the $1/N_c$ string interactions do not affect the simple Casimir effect
363: argument for the coefficient of the Luscher term.
364: In principle, it is possible that there are additional massless degrees
365: of freedom,
366: $c > D-2$, but still $\tilde{c} = D-2$. Although this cannot technically
367: be ruled out, it seems an unlikely coincidence and, moreover,
368: the groundstate is usually not a scalar if $h>0$.
369: More generally,
370: the result would be consistent with additional dimensionful
371: worldsheet interactions or additional massive degrees of freedom, since
372: these should not contribute to the long-distance $l$ properties of the string.
373:
374: Pure lattice gauge theory simulations of the Polyakov loop correlators
375: enable accurate determinations of the transition temperature, with
376: the loop expectation value as order parameter.
377: For $D=3$, $SU(3)$ pure lattice gauge theory the transition
378: is 2nd order and agrees with the $D=3$ Hagedorn temperature of the
379: simple Nambu-Goto model (\ref{d3}) \cite{lattice,teper93}.
380: For $D=4$ and $SU(3)$, the deconfinement transition temperature $T_c$
381: is slightly below the Nambu-Goto
382: Hagedorn temperature (\ref{d4}) \cite{1storder}.
383: But it is weakly first
384: order, so the actual gauge theory Hagedorn temperature is probably slightly
385: higher than $T_c$.
386: Although it is difficult to study the density
387: of hadron states directly in lattice gauge theory, except in some effective
388: theories \cite{prl},
389: this data can be extracted from experiment. The meson resonance
390: spectrum shows a clear exponential rise.
391: A study by
392: Dienes and Cudell \cite{dienes}
393: establised that the best fit to the data was for
394: $c= D_{\perp}$ = 2, in agreement with the Nambu-Goto model.
395:
396: \begin{figure*}
397: $\displaystyle T_c/\sqrt{\sigma}$\hspace{5pt}
398: \raisebox{-1.75in}{\includegraphics[width=4in]{fig1.eps}}\\
399: \hspace{0.5in}$\displaystyle 1/N_{c}^{2}$
400: \caption{The variation of the transition temperature $T_c$, in
401: units of the zero-temperature
402: string tension $\sigma$, with the number of colours $N_c$ for pure
403: $SU(N_c)$ gauge theory in $D=3$ dimensions. Open circles \cite{teper93}
404: and
405: filled circles \cite{liddle} are from different simulations.
406: The linear fit is to the 2nd
407: order transitions at $N_c = 2,3, 4$, where we identify $T_c = T_H$. NG
408: indicates the free-string Nambu-Goto prediction.
409: \label{fig1}}
410: \end{figure*}
411:
412: \begin{figure*}
413: $\displaystyle T_c/\sqrt{\sigma}$\hspace{5pt}
414: \raisebox{-1.75in}{\includegraphics[width=4in]{fig2.eps}}\\
415: \hspace{0.5in}$\displaystyle 1/N_{c}^{2}$
416: \caption{Open circles show the variation of the deconfinement
417: temperature $T_c$, in units of the zero-temperature
418: string tension $\sigma$, with the number of colours $N_c$ for pure
419: $SU(N_c)$ gauge theory in $D=4$ dimensions \cite{teper05,1storder}.
420: The solid bars represent the range of Hagedorn temperatures obtained
421: in ref.\cite{bringoltz} by different fits to the Polyakov loop mass.
422: The straight line fit is to these and the 2nd order $SU(2)$
423: transition, where we identify $T_c = T_H$.
424: NG indicates the free-string Nambu-Goto prediction.
425: \label{fig2}}
426: \end{figure*}
427:
428: \begin{figure*}
429: $\displaystyle \ln t$\hspace{5pt}
430: \raisebox{-1.75in}{\includegraphics[width=4in]{fig3.eps}}\\
431: \hspace{0.5in}$\displaystyle M_t/\sqrt{\sigma}$
432: \caption{The mass $M_t$ of the $t^{\rm th}$ glueball in the spectrum
433: of pure $SU(2)$ (open circles) and $SU(\infty)$ (filled circles) gauge
434: theory in $D=3$ dimensions \cite{teper98}.
435: The lowest 19 states are shown. The straight
436: lines are to guide the eye only.
437: \label{fig3}}
438: \end{figure*}
439:
440: Given this evidence, one might be led to conclude that the string theory
441: relevant for QCD is just the Nambu-Goto one. Of course, this cannot be true
442: because the model is not even consistent in 3 and 4 dimensions. The spectrum
443: of the $SU(3)$ heavy-source potential shows clear
444: deviations from the Nambu-Goto
445: prediction for small separations \cite{morning}.
446: Therefore it would be suprising if the Hagedorn temperature were correctly
447: predicted.
448: Results for other gauge groups, however, imply a more
449: complicated picture. For $D=3$ pure gauge theory,
450: the transition is second order for $N_c = 2,3,4$ (or at worst
451: very weakly first order for $SU(4)$, so the Hagedorn point is
452: very close by), and first order for higher $N_c$ \cite{teper93,liddle,holland}.
453: The corresponding temperatures are plotted versus $1/N_c$ in Figure~1.
454: If we are correct in identifying the $N_c = 2,3,4$ points as Hagedorn
455: transitions, there is clearly a strong dependence of $T_H$ on $N_c$.
456: For $D=4$ pure gauge theory, the transition is 2nd order only for $SU(2)$
457: \cite{teper05}.
458: However, Teper and Bringoltz \cite{bringoltz}
459: have recently been able to follow the Polyakov
460: loop mass into the metastable phase above the 1st order transitions
461: for $N_c > 2$, identifying the Hagedorn temperature
462: from $E_c(1/T_H) = 0$. Figure~2 shows again that $T_H$ depends upon
463: $N_c$.
464: Recalling that this temperature in a free resonance gas
465: is dictated by the asymptotic density of states, it is natural to conclude
466: that the $1/N_c$ string interactions alter the spectrum so as to reduce the
467: density of states; they will also give states a width, but this is not usually
468: noticed in a lattice simulation. In fact,
469: this thinning of the spectrum
470: can be seen already in the low-lying glueball spectrum.
471: Figure~3 compares the lowest glueball masses for $SU(2)$
472: and $SU(N_c \to \infty)$
473: in 3 dimensions \cite{teper98}. For $\rho(M)$ to decrease, the mass shift
474: of the ${\rm t}^{th}$ glueball should increase with $M_t$.
475: A similar qualitative
476: effect can be discerned in the spectrum
477: for $D=4$ \cite{meyer2}, although fewer accurate
478: glueball masses are available in this case.
479:
480: The results for the $SU(3)$ Hagedorn temperature
481: thus appear to be accidentally close to those for the
482: free Nambu-Goto string. The strings of an $SU(3)$ gauge theory are not free,
483: but have $1/N_c$ interactions which mask worldsheet effects
484: not accounted for by the Nambu Goto model.
485: To compare like-with-like, one should
486: only use free string formulas when comparing with the $N_c \to \infty$
487: limit of gauge theory. What do such extrapolations predict? In pure
488: gauge theory, one expects physical observables to be expandable
489: as an aymptotic series in $1/N_{c}^2$ \cite{hoof}.
490: If one naively extrapolates the Hagedorn temperatures
491: of pure gauge theory to $N_c = \infty$, using fits $A + B/N_{c}^2$, one finds
492: \begin{eqnarray}
493: T_H & \approx & 0.83 \sqrt{\sigma} \ (D=3) \label{3D} \\
494: T_H & \approx & 0.63 \sqrt{\sigma} \ (D=4) \label{4D} \ .
495: \end{eqnarray}
496: Note that, unlike the papers from which the data is sourced,
497: the fit is only to the second-order transitions.
498: Since they arise from a separate
499: physical mechanism, there is no
500: reason for the first order points to be analytically related to them.
501: In general,
502: there is also no theoretical reason why one should truncate the fit at a simple
503: $1/N_{c}^{2}$ correction; this is dictated only by the paucity of data.
504: Given that the fit is taken down to $N_c=2$, corrections at $O(1/N_{c}^{4})$
505: are likely to be the largest source of
506: error in the numbers (\ref{3D})(\ref{4D}).
507: Comparing these with eq(\ref{exp}), in both cases this predicts an effective
508: $c \approx (D-2) + \frac{1}{2}$ worldsheet theory contributing to
509: highly excited states. Given that the Luscher coefficient seems
510: unaffected by string interactions and agrees with the Nambu-Goto
511: result, whatever degrees of freedom give rise to the ``$+\frac{1}{2}$''
512: must be massive, although effectively conformal at high energies.
513:
514: Given the volatility of numerical lattice results, this neat picture could
515: be upset, of course. There is already some inconsistency in the $D=3$
516: results for transition temperatures, since the extrapolation of
517: 2nd order transitions at $N_c = 2,3,4$ seems to lie below the first
518: order transitions at $N_c = 5,6$ (also shown on fig~1), which cannot be right.
519: A study of the superheated phase at higher $N_c$ would be useful
520: in this respect. The first order transitions
521: for $D=4$ and $N_c > 2$, on the other hand, lie correctly below the
522: ``Hagedorn line'' in fig~2.
523: While the ``$+\frac{1}{2}$'' may vary in size in the light of further
524: simulations, it is clear that present data are accurate enough
525: to demonstrate further string degrees of freedom with approximately
526: this contribution.
527:
528:
529: \section{Longitudinal string oscillations}
530:
531: \subsection{Spectrum}
532:
533: In this section, we present evidence that massive longitudinal string
534: oscillations may account for the shift in the effective
535: central charge in the highly excited spectrum. Longitudinal
536: degrees of freedom
537: are obviously not a new idea. In a sense, the Nambu-Goto model already
538: posseses them if $D \neq 26$. Both the covariant quantization \cite{goddard}
539: and the original Polyakov formulation (Liouville theory)
540: \cite{polyakov3} lead to an extra $c=1$ (massless)
541: degree of
542: freedom corresponding to longitudinal oscillation. This is too much; it will
543: add to the Luscher coefficient; it will produce too many degrees of
544: freedom for the asymptotic spectrum (in the Polyakov case, the Liouville
545: zero mode produces a continuous spectrum).
546: But massive longitudinal modes may be expected. For example,
547: they occur naturally in the
548: Nielsen-Olesen vortex solution of the Abelian Higgs model \cite{nielsen}.
549: Also the Polchinski-Strominger effective string action indicates that
550: the Liouville field gets a mass \cite{polchinski}.
551: Here,
552: a specific model with longitudinal oscillations wiil be analysed --- the
553: excited spectrum of the Chodos-Thorn \cite{chodos}
554: massive relativistic string.
555: Unfortunately, this model is difficult to solve above 2 dimensions. But the
556: longitudinal modes of interest in are present in 2 dimensions
557: and the asymptotic spectrum can be found exactly in this case.
558: Assuming that coupling
559: of longitudinal and transverse degrees of freedom in higher dimensions
560: does not drastically alter things, this will provide a measure of the
561: number of longitudinal degrees of freedom.
562:
563: \begin{figure*}
564: \raisebox{-1.75in}{\includegraphics[width=4in]{fig4.eps}}\\
565: \caption{Open string with particle insertions ($N=4$ in this example).
566: \label{fig4}}
567: \end{figure*}
568:
569: The action generalises the Nambu-Goto one (\ref{NG}) by the insertion
570: of massive particle degrees of freedom on the string
571: \be
572: - \int dx^0 dx^1 \left( \mu \sqrt{\partial_{0} X^\mu
573: \partial_{0} X_\mu}+
574: \sigma \sqrt{-G} \right) \ .
575: \eq
576: In principle, the mass distribution could be continuous along the string,
577: but we will consider the case when it is discrete. Such a string is likely
578: to have the power-law fall-off of high energy scattering amplitudes
579: characteristic of particle field theory, which does not occur with conventional
580: strings in general \cite{gross}.
581: In addition to transverse oscillations, there are now oscillations
582: associated with motion of the point masses along the length of the
583: string.
584: A semi-classical analysis of the free string spectrum of this
585: theory in $D=2$ dimensions
586: was performed by Bardeen {\em et al.} \cite{bardeen}.
587: In the sector with pointlike
588: insertions at positions $X_j$, $j \in \{1, \cdots, N\}$,
589: in the interior of an open string, and insertions at each end
590: $X_0$, $X_{N+1}$ (see Figure~4), they found the lightcone gauge Hamiltonian
591: \be
592: P^- = \sum_{j=0}^{N+1} \frac{\mu^2}{2 P^{+}_{j}} + \sum_{j=0}^{N}
593: \sigma |X^{-}_{j} - X^{-}_{j+1}| \ .
594: \label{ham}
595: \eq
596: In two dimensions, the entire lightcone momentum of the string is carried by
597: the insertions
598: \be
599: P^+ = \sum_{j=0}^{N+1} P^{+}_{j} \ .
600: \eq
601: Finding the
602: normal modes of classical solutions in the $\mu \to 0$ limit, Bardeen {\em et
603: al.}
604: then impose Bohr-Sommerfeld
605: quantization conditions to obtain a spectrum of masses
606: \be
607: M^2 = 2 \pi \sigma \sum_{n=1}^{N+1} n (l_n + {\rm const.}) \ ,
608: \label{sc}
609: \eq
610: where $l_n \in \{0,1,2, \cdots \}$.
611: Note that the massless limit $\mu \to 0$ is {\em not} the
612: Nambu-Goto model, provided we allow points on the string, where there were
613: masses, to move at the
614: speed of light. Although the actions are the same, the Hamiltonians are
615: not \cite{bardeen}.
616:
617: An important question is whether one should include sectors of different
618: $N$ independently in the spectrum (\ref{sc}).
619: In the $\mu \to 0$ limit, the classical solutions of the equations
620: of motion for an open string with $N$ insertions contain
621: the solutions for $N' < N$ insertions also. This happens because a subset
622: of the normal-mode motions occur with some of the insertions at coincident
623: points. For example, one of the $N=1$ classical solutions has the `interior'
624: point always attached to one or other end of the open string,
625: $X_1 = X_0$ or $X_1 = X_2$. Such a motion is geometrically indistinguishable
626: from the classical solution for $N=0$. Therefore, one might expect that
627: in this case the entire spectrum can be obtained by taking $N \to \infty$
628: in eq.(\ref{sc}).
629: If one compares this spectrum with eq.(\ref{spec}) and recall that
630: there are no transverse oscillations for $D=2$, the longitudinal
631: oscillations in this model as $N \to \infty$ appear to
632: asymptotically give
633: the same spectrum as a physical $c=1$ worldsheet degree of freedom
634: (like one extra transverse dimension). However,
635: once a mass $\mu$
636: is introduced, however small, the indistinguishability is lost; the
637: total mass of the insertion is significant. For this reason, the sectors
638: for each, fixed $N$
639: give rise to distinguishable states in the spectrum, resulting in many
640: more states than just the $N \to \infty$ sector. In fact, if all $N$ sectors
641: are allowed, the presence
642: of modes with $l_n =0$ would lead to each energy level in the spectrum
643: being infinitely degenerate in the $\mu \to 0$ limit
644: (neglecting the constant in (\ref{sc})).
645:
646: The resolution of this difficulty lies in the fact that
647: one expects the semi-classical spectrum (\ref{sc}) to be a
648: valid description
649: of the full quantum theory only for large
650: quantum numbers. To gain a better understanding of this, one can
651: perform a different kind of semi-classical analysis motivated by
652: two dimensional large-$N_c$ gauge theory. The expression (\ref{ham}) is
653: isomorphic to the lightcone Hamiltonian of two-dimensional large-$N_c$
654: gauge theory minimally coupled to fundamental and adjoint matter particles
655: at the endpoints
656: and interior points respectively. The origin of the linear string potential is
657: then the Coulomb force in two dimensions. Introducing the
658: lightcone wavefunction
659: $\phi_N(x_0, x_1, \cdots, x_{N+1})$ for the sector with $N$ interior points,
660: with $x_j P^+ = P^{+}_{j}$,
661: this linear potential between bosonic particles $j$ and $j+1$ becomes
662: \cite{numerical}
663: \begin{eqnarray}
664: && \frac{\sigma}{4\pi P^+}
665: \int_{0}^{x_j+x_{j+1}} \ \frac{(x_j +y)(x_j+2x_{j+1}-y)}{
666: (x_j-y)^2 \sqrt{y x_j x_{j+1} (x_j+x_{j+1} -y)}}
667: \nonumber \\
668: && \left\{\phi_N(x_0, \cdots, x_{N+1}) \right. \nonumber \\
669: && - \left.
670: \phi_N(\cdots, x_{j-1}, y, x_j + x_{j+1} -y,\cdots)\right\} \ dy \nonumber \\
671: && + \left[\frac{\sigma}{4P^+ \sqrt{x_j x_{j+1}}}\right]
672: \phi_N(x_0, \cdots, x_{N+1})
673: \end{eqnarray}
674: (In the gauge theory there would also be particle number changing interactions
675: that are not present in the first quantized string theory).
676: The highly excited spectrum consists of wavefunctions $\phi_N$ that
677: oscillate rapidly and, following 't Hooft \cite{hoof2} and Kutasov
678: \cite{kutasov}, one may
679: simplify the integrand above in this limit. The integrals average to
680: zero except near the singularities of the integrand $x_j \approx y$,
681: in which case they may be effectively replaced by
682: \begin{eqnarray}
683: && \frac{\sigma}{\pi P^+}
684: \int_{-\infty}^{\infty} \frac{dz}{z^2} \ \left\{
685: \phi_N(x_0, \cdots, x_{N+1}) \right. \nonumber \\
686: && \left. - \phi_N(\cdots, x_{j-1}, x_j + z, x_{j+1}-z,\cdots)\right\} \ .
687: \end{eqnarray}
688: In the same regime, one can also assume that the mass terms are
689: negligible compared to the
690: excitation
691: energy. However, they do impose the boundary
692: condition that $\phi_N = 0$ whenever any $x_j = 0$.
693: Thus, in the $N=0$ sector the spectral
694: eigenvalue equation for highly
695: excited states becomes
696: \be
697: M^2 \phi_0 =
698: \frac{2\sigma}{\pi}
699: \int_{-\infty}^{\infty} dz \ \frac{\phi_0(x_0, x_{1}) -
700: \phi_N(x_0 + z, x_1-z)}{z^2} \ .
701: \eq
702: The solutions are of the form
703: \begin{eqnarray}
704: \phi_0(x_0,x_1) & = &
705: \sin s \pi x_0 \\
706: x_1 & = & 1-x_0 \\
707: M^2 & = & 2 \pi \sigma s
708: \end{eqnarray}
709: for large {\em positive} integers $s$. Comparing with eq(\ref{sc}), we
710: can identify $l_1 \equiv s$ in the $N=0$ sector. Note, however, that
711: the new analysis excludes the troublesome $l_1=0$ solution.
712:
713: Proceeding in the same way for $N=1$, the corresponding
714: eigenvalue equation for highly
715: excited states becomes
716: \begin{eqnarray}
717: M^2 \phi_1 & = &
718: \frac{2\sigma}{\pi}
719: \int_{-\infty}^{\infty} \frac{dz}{z^2} \
720: \left\{ \phi_1(x_0, x_{1}, x_2) \right. \nonumber \\
721: && - \phi_1(x_0 + z, x_1-z,x_2) \nonumber \\
722: & & +
723: \phi_1(x_0, x_{1}, x_2) \nonumber \\
724: && -
725: \left. \phi_1(x_0, x_1+z,x_2-z)\right\} \ .
726: \end{eqnarray}
727: The solutions in this case which respect the boundary conditions are
728: \begin{eqnarray}
729: \phi_1(x_0,x_1,x_2) & = &
730: \sin s_1 \pi x_0 \sin s_2 \pi x_2 \nonumber \\
731: && \pm \sin s_1 \pi x_2 \sin s_2 \pi x_0
732: \\
733: x_0 + x_1 + x_2 & = & 1 \\
734: M^2 & = & 2 \pi \sigma (s_1 + s_2)
735: \end{eqnarray}
736: for large positive integers $s_1$ and $s_2$ such that $s_2>s_1$.
737: The plus solution occurs when $s_2-s_1$ is odd,
738: the minus solution when it is even, corresponding to states of opposite
739: parity under orientation reversal of the open string. The identification
740: with eq(\ref{sc}) is made by $l_2 = s_1$, $l_1 + l_2 = s_2$. Again, the
741: solutions $l_1=0$ and $l_2=0$ are excluded, but this time one can use large
742: $s_1$ and $s_2$, where the analysis is valid, to exclude the former.
743:
744: \subsection{Asymptotic Density of States}
745:
746: The natural generalisation of these results is that in the $N$-sector
747: the excited spectrum is
748: \be
749: M^2 = 2 \pi \sigma (s_1 + s_2 + \cdots + s_{N+1})
750: \label{Nspec}
751: \eq
752: for large positive integers $s_{N+1} > s_{N} > \cdots > s_1$, with
753: \be
754: s_j = \sum_{n=N-j+2}^{N+1} l_n \ .
755: \eq
756: This matches (\ref{sc}) provided $l_n = 0$ is excluded.
757: Note that a similar asymptotic spectrum has been derived
758: for closed strings of bosonic and fermionic adjoint matter in
759: refs.\cite{kutasov,bvds}, although only
760: the solutions even under orientation reversal were found.
761: Numerical solutions for the low-lying spectrum
762: of two dimensional large-$N_c$ gauge theory coupled
763: to adjoint matter have also been obtained \cite{numerical}.
764: They provide additional support for the argument that each $N$-sector
765: should contribute
766: independently to the spectrum, even in the $\mu \to 0$ limit,
767: since no degeneracies across
768: different $N$-sectors are observed.
769:
770:
771: The asymptotic density of states corresponding to the spectrum (\ref{Nspec}),
772: including all $N$-sectors, can be obtained in a standard way from the
773: generating function
774: \be
775: G(w) \equiv \sum_{n=1}^{\infty} d_n w^n = \prod_{m=1}^{\infty} (1+w^m)
776: \eq
777: where $d_n$ is the number of states at level $M^2 = 2 \pi \sigma n$.
778: The large $n$ behaviour is obtained from the limit $w \to 1$:
779: \begin{eqnarray}
780: \ln G & = & - \sum_{m,q=1}^{\infty} \frac{(-w^m)^q}{q} \nonumber \\
781: & = & -\sum_{q=1}^{\infty} \frac{(-w)^q}{q(1-w^q)} \nonumber \\
782: & \to & \frac{1}{1-w} \sum_{q=1}^{\infty} \frac{(-w)^q}{q^2} \ \ ({\rm as} \
783: w \to 1) \nonumber \\
784: & = & \frac{\pi^2}{12(1-w)} \ .
785: \end{eqnarray}
786: Then $d_n$ can be obtained from the saddle point approximation to the
787: integral
788: \be
789: d_n = \frac{1}{2 \pi {\rm i}} \int_{\cal C} \frac{G(w)}{w^{n+1}}
790: \eq
791: where the contour ${\cal C}$ encircles the origin. The result
792: \be
793: d_n \sim {\rm exp}\left[ M \sqrt{\frac{\pi}{6 \sigma}} \right]
794: \eq
795: is characteristic of a $c=1/2$ conformal field theory (eq(\ref{exp})).
796:
797: In fact, one could have guessed this result without further calculation
798: by rewriting the spectrum (\ref{Nspec}), including all $N$-sectors, as
799: \be
800: M^2 = 2 \pi \sigma \sum_{n=1}^{\infty} n l
801: \eq
802: where $l \in \{0,1\}$. This is the spectrum of a tower of fermionic
803: harmonic oscillators! The non-linear massive bosonic longitudinal oscillations
804: of this string model contribute, in the asymptotic spectrum, just as
805: would a single free massless Majorana worldsheet fermion field.
806: This is the right
807: number of degrees of freedom to account for the lowering of the Hagedorn
808: temperature observed in large-$N_c$ lattice simulations.
809:
810: \section{Discussion}
811:
812: Lattice data that may be used
813: to constrain any purported string theory of QCD have been critically reviewed.
814: In general, one must extrapolate data to $N_c = \infty$ before comparing with
815: free-string formulas. From data on the Luscher coefficient
816: and Hagedorn temperature, in additional to the usual massless
817: degrees of freedom associated with oscillations in $D-2$ transverse
818: dimensions, it appears that
819: further massive modes contribute to the asymptotic spectrum.
820: Subject to the numerical accuracy of the data, they contribute
821: effectively like a $c \approx 1/2$ conformal worldsheet field.
822:
823: One obvious place to look for the extra modes is in the longitudinal
824: direction. The asymptotic spectrum of an `old'
825: massive relativistic string model in two dimensions was re-derived and
826: clarified.
827: It possesses only longitudinal oscillations. Despite being a
828: bosonic model, the counting of highly excited states matches
829: that of a $c=1/2$ worldsheet conformal field.
830: The observations made in this paper are also relevant for `new' approaches
831: to string theory. Longitudinal modes of oscillation will exist in the
832: bulk for holograhic models based on the ADS/CFT correspondence
833: \cite{brower}. The constraints
834: from lattice data may help to identify the correct background that
835: corresponds to confining gauge theory. In this and
836: other string models, there still
837: remains much to understand concerning the usual unitarity, space-time
838: symmetry, and groundstate stability expected of gauge theory.
839:
840: To test the hypothesis suggested in this paper, it would be useful to
841: have further lattice data at large-$N_c$ for
842: more detailed observables. For example, the thermodynamic pressure
843: may vary rapidly-enough close to the Hagedorn temperature for a quantitative
844: comparison to be made with the result based on the density of states
845: (\ref{exp}) with $c = (D-2) + 1/2$ and $D_\perp = 2$. In order to make
846: general statements, in this paper we have
847: essentially been considering data from the large-$l$ and minimum-$l$
848: behaviour of the groundstate functions $E_{c}(l)$, $E_o(l)$
849: in eqs.~(\ref{open})(\ref{closed}).
850: Of course,
851: it would be interesting to compare the
852: predictions of specific string models and
853: large-$N_c$ gauge theory at general $l$ and for low excited states of these
854: systems \cite{morning}.
855:
856:
857:
858: %\vspace{10mm}
859: %\noindent Acknowledgements:
860: \acknowledgments{
861: The work is supported by PPARC grant
862: PP/D507407/1. I thank B. Lucini for helpful discussions. }
863:
864:
865: \begin{thebibliography}{99}
866:
867:
868: \bibitem{nambu} Y. Nambu, Proc.\ Int.\ Conf.\ on Symm.\ and Quark
869: Model, Wayne State Univ.\ 1969, Gordon and Breach, London 1970;
870: Phys.\ Rev.\ D{\bf 10}, 4262 (1974);\\
871: L. Susskind, Phys.\ Rev.\ D{\bf 1}, 1182 (1970);\\
872: T. Goto, Prog.\ Theor.\ Phys.\ {\bf 46}, 1560 (1971).
873:
874: \bibitem{maldacena} J. Maldacena, Adv.\ Theor.\ Math.\ Phys.\ {\bf 2}, 231
875: (1998)
876:
877: \bibitem{polyakov} A. M. Polyakov, Int.\ J.\ Mod.\ Phys.\ A{\bf 14}, 645
878: (1999).
879:
880: \bibitem{chodos} A. Chodos and C. B. Thorn, Nucl.\ Phys.\
881: B{\bf 72}, 509 (1974).
882:
883: \bibitem{luscheretal}
884: M. Luscher, K. Symanzik and P. Weisz, Nucl.\ Phys.\ B{\bf 173}, 365 (1980).
885:
886: \bibitem{luscher} M. Luscher, Nucl.\ Phys.\ B{\bf 180}, 317 (1981).
887:
888: \bibitem{vafa} I. Kani and C. Vafa, Commun. Math. Phys. {\bf 130}, 529 (1990).
889:
890: \bibitem{hagedorn} R. Hagedorn, Nuovo Cimento Suppl.\ {\bf 3}, 147 (1965);\\
891: K. Huang and S. Weinberg, Phys.\ Rev.\ Lett.\ {\bf 25}, 895 (1970).\\
892: S. Frautchi, Phys.\ Rev.\ D{\bf 3}, 2821 (1971); \\
893: R. D. Carlitz, {\em ibid.}
894: {\bf 5}, 3232 (1972).
895:
896: \bibitem{parisi} N. Cabibbo and G. Parisi, Phys.\ Lett.\ B {\bf 59},
897: 67 (1975).
898:
899: \bibitem{olesen} P. Olesen, Nucl.\ Phys.\ B{\bf 267}, 539 (1986).
900:
901: \bibitem{broniowski} W. Broniowski, W. Florkowski, and L. Ya. Glozman,
902: Phys.\ Rev.\ D{\bf 70}, 117503 (2004).
903:
904: \bibitem{polyakov2} A. M. Polyakov, Phys.\ Lett.\ B{\bf 72}, 477 (1978).
905:
906: \bibitem{1storder}
907: B. Lucini, M. Teper and U. Wenger, Phys.\ Lett.\ B{\bf 545} 197 (2002);
908: JHEP {\bf 0401} 061 (2004).
909:
910: \bibitem{bern} K. Holland, M. Pepe, and U. J. Wiese, Nucl.\ Phys.\
911: B{\bf 694}, 35 (2004).
912:
913: \bibitem{teper05} B. Lucini, M. Teper, U. Wenger, JHEP {\bf 0502}, 033 (2005)
914:
915: \bibitem{bringoltz} B. Bringoltz and M. Teper, hep-lat/0508021
916:
917: \bibitem{deF} P. de Forcrand, G. Schierholz, H. Schneider, and M. Teper,
918: Phys.\ Lett.\ B {\bf 160}, 137 (1985).
919:
920: \bibitem{hoof} G. 't Hooft, Nucl.\ Phys.\ B{\bf 72}, 461 (1974).
921:
922: \bibitem{wilson} K. G. Wilson,, Phys.\. Rev.\ D {\bf 10}, 2445 (1974).
923:
924: \bibitem{thorn} P. Goddard, J. Goldstone, C. Rebbi, and C. Thorn, Nucl.\ Phys.\
925: B{\bf 56}, 109 (1973).
926:
927: \bibitem{polchinski} J. Polchinski and A. Strominger, Phys.\ Rev.\ Lett.\
928: {\bf 67}, 1681 (1991).
929:
930: \bibitem{luscher2} M. Luscher and P. Weisz, JHEP {\bf 0207}, 049 (2002).
931:
932: \bibitem{sommer} S. Necco and R. Sommer, Nucl.\ Phys.\ B{\bf 622}, 328 (2002).
933:
934: \bibitem{caselle} M. Caselle, M. Hasenbusch and M. Panero, JHEP {\bf 0301}, 057 (2003); JHEP {\bf 0503}, 026 (2005).
935:
936: \bibitem{wako} K. J. Juge, J. Kuti, and C. Morningstar,
937: Proceedings of Wako 2003, {\em Color confinement and hadrons in quantum
938: chromodynamics} 233 [hep-lat/0401032].
939:
940: \bibitem{lucini} B. Lucini and M. Teper, Phys.\ Rev.\
941: D{\bf 64} 105019 (2001);\\
942: M. Caselle, M. Pepe and A. Rago, JHEP {\bf 0410}, 005 (2004).
943:
944: \bibitem{meyer} H. Meyer and M. Teper, JHEP {\bf 0412}, 031 (2004).
945:
946: \bibitem{lattice}
947: J. Engels, F. Karsch, E. Laermann, C. Legeland, M. Lutgemeier,
948: B. Petersson and T.
949: Scheideler, Nucl.\ Phys.\ Proc.\ Suppl.\ {\bf 53}, 420 (1997).
950:
951: \bibitem{teper93} M. Teper, Phys.\ Lett.\ B{\bf 313} 417 (1993) and
952: unpublished.
953:
954: \bibitem{prl} S. Dalley and B. van de Sande, Phys.\ Rev.\ Lett.\
955: {\bf 95}, 162001 (2005).
956:
957: \bibitem{dienes} K. R. Dienes and J-R. Cudell,
958: Phys.\ Rev.\ Lett.\ {\bf 72}, 187 (1994).
959:
960: \bibitem{morning} K. J. Juge, J. Kuti and C. Morningstar, Phys.\ Rev.\ Lett.\
961: {\bf 90}, 161601 (2003).
962:
963: \bibitem{liddle} J. Liddle and M. Teper,
964: 23rd International Symposium on Lattice Field Field: Lattice 2005,
965: Trinity College, Dublin, Ireland, 25-30 Jul 2005.
966: Published in PoS LAT2005:188,2005 [hep-lat/0509082].
967:
968: \bibitem{holland} K. Holland, hep-lat/0509041.
969:
970: \bibitem{teper98} M. Teper, Phys.\ Rev.\ D{\bf 59}, 014512 (1999).
971:
972: \bibitem{meyer2} H. Meyer and M. Teper, Phys.\ Lett.\ B{\bf 605}, 344 (2005).
973:
974: \bibitem{goddard} P. Goddard and C. Thorn, Phys.\ Lett.\ B{\bf 40}, 235
975: (1972);\\
976: R. C. Brower, Phys.\ Rev.\ D{\bf 6}, 1655 (1972).
977:
978: \bibitem{polyakov3} A. M. Polyakov, Phys.\ Lett.\ B{\bf 103}, 207 (1981).
979:
980: \bibitem{nielsen} H. B. Nielsen and P. Olesen, Nucl.\ Phys.\ B{\bf 61}, 45
981: (1973).
982:
983: \bibitem{gross} D. Gross and P. Mende, Phys.\ Lett.\ B{\bf 197}, 129 (1987).
984:
985: \bibitem{bardeen} W. A. Bardeen, I. Bars, A. J. Hanson, and R. D. Peccei,
986: Phys.\ Rev.\ D{\bf 13}, 2364 (1976).
987:
988: \bibitem{numerical} S. Dalley and I. Klebanov, Phys.\ Rev.\ D{\bf 47}, 2517
989: (1993);\\
990: K. Demeterfi, I Klebanov, and G. Bhanot, Nucl.\ Phys.\ B{\bf 418}, 15 (1994).
991:
992: \bibitem{hoof2} G. 't Hooft, Nucl.\ Phys.\ B{\bf 75}, 461 (1974).
993:
994: \bibitem{kutasov} D. Kutasov, Nucl.\ Phys.\ B{\bf 414}, 33 (1994).
995:
996: \bibitem{bvds} S. Dalley and B. van de Sande, Phys.\ Rev.\ D{\bf 56},
997: 7917 (1997).
998:
999:
1000: \bibitem{brower} R. C. Brower, C-I. Tan, and E. Thompson,
1001: Int.\ J.\ Mod.\ Phys.\ A{\bf 20}, 4508 (2005).
1002:
1003:
1004: \end{thebibliography}
1005: \end{document}
1006: