1: %\documentclass[draft,12pt]{JHEP3}
2: \documentclass[12pt]{JHEP3}
3:
4: %% \usepackage{makeidx}
5: %% \usepackage{acronym}
6: %% \usepackage{epsfig}
7: %% \usepackage{subfigure}
8:
9: \usepackage{amsmath}
10: \usepackage{amssymb}
11: \usepackage{cite}
12: \usepackage{array}
13: %\usepackage[notref,notcite]{showkeys} %% just for draft versions
14:
15: \usepackage[german,american]{babel}
16: \usepackage[latin1]{inputenc}
17:
18: \usepackage{fancyhdr}
19: % \pagestyle{fancy}
20:
21: %% Settings for "plain" style
22: \fancypagestyle{plain}{%
23: \fancyhf{} % clear all header and footer fields
24: \cfoot{\thepage} % except the center
25: \renewcommand{\headrulewidth}{0pt}
26: \renewcommand{\footrulewidth}{0pt}}
27:
28: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
29: %% %%
30: %% LOADING THE MACROS %%
31: %% %%
32: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
33:
34: \input{macros.tex}
35:
36: %% Defining the Dirac and hermitian conjugates of the spinors
37:
38: \newcommand{\chib}{\overline{\chi}}
39: \newcommand{\chid}{\chi^\dagger}
40:
41: %% left-right-derivatives with lower indices
42:
43: \newcommand{\plr}[1]{\overleftrightarrow{\partial_{#1}}}
44:
45: %% Absätze nicht einrücken
46:
47: \setlength{\parindent}{0mm}
48:
49:
50: %% Author, Abstract, Title, Affiliation
51:
52: \author{René Meyer\\ Institute for Theoretical Physics, University of Leipzig\\ Augustusplatz 10-11, D-04109 Leipzig, Germany\\ E-mail: \email{rene.meyer@itp.uni-leipzig.de}}
53:
54: \abstract{Generalized Dilaton Theories in two dimensions coupled to
55: Dirac fermions are subjected to constraint analysis. Three first
56: class secondary constraints are found, corresponding to one local
57: Lorentz symmetry and two diffeomorphisms. Moreover, the system also
58: yields second class constraints from the fermions. The algebra of
59: first class constraints is calculated in some detail, and is found
60: to be related to the classical Virasoro algebra.}
61:
62: \title{Constraints in two-dimensional Dilaton Gravity with Fermions}
63:
64:
65:
66: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
67: %% %%
68: %% DOCUMENT STARTS HERE %%
69: %% %%
70: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
71:
72:
73: \begin{document}
74:
75: \section{Introduction}\label{sec:intro}
76:
77: Generalized Dilaton Theories (GDTs) in two dimensions arise from
78: several fields of physics (for an exhaustive review, see
79: \cite{Grumiller:2002nm}). Prominent examples would be spherical
80: reduced Einstein-Hilbert gravity from $D>2$ dimensions, the ``string
81: inspired'' dilaton model (CGHS model, \cite{Callan:1992rs}) or the
82: models put forward by Jackiw \& Teitelboim
83: \cite{Jackiw:1984,Teitelboim:1984} and Katanaev \& Volovich
84: \cite{Katanaev:1986wk} in the 1980s. All these models can be subsumed
85: into one action \cite{Russo:1992yg,Odintsov:1991qu,Banks:1991mk}
86: %
87: \beq\label{dilatonaction}
88: S^{(GDT)} = \int d²x \sqrt{-g} \left[\frac{R}{2}X - \frac{U(X)}{2}(\nabla X)²+V(X) \right]
89: \eeq
90: %
91: with $X$ being the dilaton field, $R$ the Ricci scalar and $U(X)$ and
92: $V(X)$ arbitrary potentials specifying the model.
93:
94: It turns then out that there exists a classically equivalent
95: formulation of \eqref{dilatonaction} in terms of Cartan variables,
96: namely the Vielbein $e^a=e^a_\mu dx^\mu$, the spin connection
97: $\omega^{ab}=\omega_{\mu}^{ab}dx^\mu=:\omega\epsilon^{ab}$; the dilaton $X$ and
98: additional auxiliary fields $X^\pm$. This First Order Gravity (FOG)
99: action reads \cite{Schaller:1994es}
100: %
101: \beq\label{FOGaction}
102: S_{FOG} = \int\limits_{{\cal M}_2} \left[ X^a(De)_a+Xd\omega+\epsilon{\cal V}(X^+X^-;X)\right]
103: \eeq
104: %
105: Here we allow also for non-vanishing torsion $T^a = (De)^a$ terms, and
106: the proof of classical equivalence (for potentials ${\cal V}(X^+X^-;X)
107: = U(X) X^a X_a + V(X)$), which can be found in ch. 2 of
108: \cite{Grumiller:2002nm}, amounts to using the equations of motion for
109: the $X^a$ to eliminate the torsion dependent part of the spin
110: connection.
111:
112: The FOG action is our starting point for the analysis presented. We
113: will explain the coupling to Dirac fermions in sec.
114: \ref{sec:fermions}. In Sec. \ref{sec:analysis} we will analyse the
115: constraints of the theory and obtain the constraint algebra. Sec.
116: \ref{sec:witt} relates this algebra with the well known classical
117: Virasoro algebra (or Witt algebra) in two dimensions, and Sec. 5
118: contains some discussion of the main results. The conventions used in
119: this article can be found in Appendix \ref{app:A}. Appendices
120: \ref{app:B} and \ref{app:C} contain equations necessary for proving
121: the main results of this paper.
122:
123: It should be noted that a similar analysis has been carried out for
124: scalar fields coupled to FOG in \cite{Grumiller:2001ea} and for
125: massless, not self interacting and minimally coupled fermions
126: \cite{wal01}. We extend this analysis to the general case of massive,
127: self-interacting Dirac fermions with non-minimal coupling. (What is
128: meant by (non-)minimal coupling will be explained in the next
129: section.) One of the first works considering GDTs (with $U(X)=0$) coupled to fermions was \cite{Cavaglia:1998yp}.
130: Even before, specific models were used as toy
131: models for Black Hole evaporation \cite{Nojiri:1992st,Ori:2001xc} and
132: more recently in a paper by Thorlacius et. al. \cite{Frolov:2005ps}
133: For another remark on this, see also Sec. \ref{sec:conclusions}.
134:
135:
136:
137: \section{Fermions}\label{sec:fermions}
138:
139:
140:
141:
142: We add to \eqref{FOGaction} an action for the fermion fields
143: consisting of the well known kinetic term and a general self
144: interaction\footnote{Note that $\ast e^a \neq e^a$.}
145: %
146: \beqa\label{wholeaction}
147: S & = & S_{FOG} + S_{kin} + S_{SI}\\\nonumber
148: S_{kin} & = & - \frac{i}{2} \int\limits_{{\cal M}_2} f(X)\; (*e^a) \wedge (\chib \ga_a \overleftrightarrow{\mathrm{d}}\chi)\\\label{fermionkinetic}
149: & = & \frac{i}{2}\int d²x (e) f(X) {e_a}^\mu (\chib \ga^a \plr{\mu} \chi)\\
150: \nonumber
151: S_{SI} & = & - \int\limits_{{\cal M}_2} \epsilon h(X) g(\chib\chi)\\\label{fermionSI}
152: & = & \int d^2x (e) h(X) g(\chib\chi)
153: \eeqa
154:
155: Here all possible boundary terms coming from the fermions have been
156: omitted. The functions $f(X)$ and $h(X)$ introduce a coupling to the
157: dilaton field. If both are constant, i.e. $f \propto h =
158: \mathrm{const.}$, we call the fermions minimally coupled, and
159: non minimally coupled otherwise. Because spinors are anti commuting
160: Grassmann fields, a Taylor expansion of $g$ yields at at most a
161: quartic term, $g(\chib\chi) = g_0 + m\chib\chi + \la (\chib\chi)²$;
162: and the constant term can always be absorbed into $V(X)$.
163:
164: Note that in two dimensions the kinetic term is independent of the
165: spin connection: In arbitrary dimension, the action for the kinetic
166: term would be\footnote{h.c. means hermitian conjugate}
167: \cite{Birrell:1989}
168: \beqs
169: \frac{i}{2} \int d^n x \det(e^a_\mu) \left[ f(X) e_a^\mu (\chib \ga^a (\partial_\mu + {\om_\mu}^{bc}\Si_{bc})\chi) + \mathrm{h.c.}\right]\eeqs
170: For $n=2$ however, there is only one independent generator of
171: Lorentz transformations $\Si_{01} = \frac{1}{4}[\ga_0,\ga_1]=-\frac{\ga_*}{2}$,
172: and with $\{\ga_a,\ga_*\}=0$ the terms in \eqref{fermionSI} containing the spin connection
173: vanish
174: %
175: \beqas
176: & & -\frac{i}{4}f(X) (\ast e^a) \wedge \om \chid(\ga^0\ga_a\ga_* - \ga_*\ga_a^\dagger \ga^0)\chi\\
177: & = & -\frac{i}{4}f(X) (\ast e^a) \wedge \om \chid\underbrace{(\ga^0\ga_a - \ga_a^\dagger \ga^0)}\limits_{= 0} \ga_*\chi = 0
178: \eeqas
179:
180:
181:
182: \section{Hamiltonian Analysis}\label{sec:analysis}
183:
184: For the sake of better memorability, we henceforth denote the
185: canonical coordinates and momenta by
186: %
187: \beqas\nonumber
188: \ol{q}^i & = & (\om_0,e_0^-,e_0^+)\\
189: q^i & = & (\om_1,e_1^-,e_1^+),\hspace{9.5mm} i=1,2,3 \\
190: p_i & = & (X,X^+,X^-)\\\label{Qj}
191: Q^\al & = & (\chi_0,\chi_1,\chid_0,\chid_1),\hspace{3mm} \al=0,1,2,3
192: \eeqas
193: %
194: The canonical structure on the phase space is given by Poisson
195: brackets
196: %
197: \beqa\label{poisson}
198: \spoiss{q^i(x)}{p_j(y)} & = & \de^i_j \de(x-y) \\ \nonumber
199: \spoiss{Q^\al(x)}{P_\be(y)} & = & - \de^\al_\be \de(x-y)
200: \eeqa
201: where the $P_\be$ are canonical momenta for the spinors, and not
202: present explicitly in \eqref{fermionkinetic} and \eqref{fermionSI}.
203:
204: \subsection{Primary and Secondary Constraints}
205:
206: In our system there occur both primary first and second class
207: constraints. A look at the Lagrangian written in components
208: %
209: \beqas
210: {\cal L}_{FOG} & = & \tilde{\epsilon}^{\mu \nu}(X^+(\partial_\mu - \omega_\mu)e^-_\nu + X^-(\partial_\mu + \omega_\mu)e^+_\nu \\
211: & & + X \partial_\mu \omega_\nu) - (e){\cal V}(X^+X^-;X) \\
212: {\cal L}_{kin} & = & \frac{i}{\sqrt{2}} f(X)\left[ -e_0^+(\chid_0\plr{1}\chi_0)
213: +e_0^-(\chid_1\plr{1}\chi_1)\right.\\
214: & & \hspace{19mm}\left. +e_1^+(\chid_0\plr{0}\chi_0)
215: -e_1^-(\chid_1\plr{0}\chi_1)
216: \right]\\
217: {\cal L}_{SI} & = & (e_0^- e_1^+ - e_0^+e_1^-) h(X) g(\chid_1 \chi_0 + \chid_0 \chi_1)
218: \eeqas
219: %
220: shows that there do not occur any time (i.e. $x^0$) derivatives of the
221: $\ol{q}^i$ and thus the $\ol{p}_i$ are constrained to zero, $\ol{p}_i
222: \approx 0$, where $\approx$ means weakly equal to zero.
223:
224: Because the kinetic term for the fermions is of first order in the
225: derivatives, the fermion momenta $P_\al := \pld{\cal L}{\dot{Q}^\al}$ give
226: rise to primary constraints
227: %
228: \beqa\label{Phi0}
229: \Phi_0 & = & P_0 + \frac{i}{\sqrt{2}}f(p_1) q^3 Q^2 \approx 0\\\label{Phi1}
230: \Phi_1 & = & P_1 - \frac{i}{\sqrt{2}}f(p_1) q^2 Q^3 \approx 0\\\label{Phi2}
231: \Phi_2 & = & P_2 + \frac{i}{\sqrt{2}}f(p_1) q^3 Q^0 \approx 0\\\label{Phi3}
232: \Phi_3 & = & P_3 - \frac{i}{\sqrt{2}}f(p_1) q^2 Q^1 \approx 0
233: \eeqa
234: %
235: They have a non vanishing Poisson bracket\footnote{$\spoiss{.}{.}$
236: denotes the graded Poisson bracket. For the definition c.f. App. A}
237: with each other,
238: %
239: \beqa\label{Calphabeta}\nonumber
240: C_{\al\be}(x,y) & := & \spoiss{\Phi_\al(x)}{\Phi_\be(y)} \\
241: & = & i\sqrt{2}f(X)
242: \left(
243: \begin{matrix}
244: 0 & 0 & -e_1^+ & 0 \\
245: 0 & 0 & 0 & e_1^- \\
246: -e_1^+ & 0 & 0 & 0 \\
247: 0 & e_1^- & 0 & 0
248: \end{matrix}
249: \right)\de(x-y)
250: \eeqa
251: %
252: and thus are \textbf{of second class}, according to Dirac's original
253: classification of constraints \cite{Dirac:1996}. The $\Phi_\al$,
254: however, are independent of the $\ol{q}^i$ and thus commute with the
255: $\ol{p}_i$.
256:
257: Having computed all the momenta, we obtain the \textbf{Hamiltonian
258: density}
259: %
260: \beqa\nonumber
261: \cal H & = & \dot{Q}^\al P_\al + p_i \dot{q}^i - \cal L \\\label{wholehamiltonian}
262: & =: & {\cal H}_{FOG} + {\cal H}_{kin} + {\cal H}_{SI} \\\nonumber
263: {\cal H}_{FOG} & = & X^+ (\partial_1 - \om_1)e_0^- + X^-(\partial_1 + \om_1)e_0^+ + X\partial_1 \om_0 + (e){\cal V}\\\nonumber
264: & & +(X^+e_1^- - X^-e_1^+)\om_0\\\nonumber
265: {\cal H}_{kin} & = & \frac{i}{\sqrt{2}}f(X)\left[e_0^+(\chid_0 \lrpd{1} \chi_0) - e_0^-(\chid_1 \lrpd{1} \chi_1) \right]\\\nonumber
266: {\cal H}_{SI} & = & -(e)h(X)g(\chib\chi)
267: \eeqa
268:
269: To deal with the second class constraints, we pass to the \textbf{Dirac bracket}
270: \cite{Dirac:1996,Henneaux:1992}
271: %
272: \beq\label{diracbracket}
273: \dirac{f(x)}{g(y)} := \spoiss{f}{g} - \int dzdw \; \spoiss{f(x)}{\Phi_\al(z)} C^{\al\be}(z,w) \spoiss{\Phi_\be(w)}{g(y)}
274: \eeq
275: %
276: with $C^{\al\be}(x,y)$ being the inverse of the matrix-valued
277: distribution
278: %
279: \footnote{The inverse $C^{\al\be}(x,y)$ of a matrix valued distribution $C_{\al\be}(x,y)$ is defined such that
280: $\int dy \left(\int dx \varphi(x) C_{\al\ga}(x,y)\right)\left(\int dz \psi(z) C^{\ga\be}(y,z)\right) =
281: \int dx \varphi(x) \int dz \psi(z) \delta_\al^\be \delta(x-z)$ for all test functions $\varphi,\psi$.
282: When $C_{\al\be}(x,y) = C_{\al\be}(x)\delta(x-y)$, then the inverse is $C^{\al\be}(x,y)=C^{\al\be}(x)\delta(x-y)$
283: with $C_{\al\ga}(x)C^{\ga\be}(x)=\delta_\al^\be \; \forall x$}
284: %
285: \eqref{Calphabeta}.
286:
287: Demanding that the primary first class constraints should not change
288: during time evolution, i.e.\footnote{Henceforth, a prime in a Dirac or Poisson bracket means
289: evaluation of the function at a point $x'$.}
290: %
291: $G_i := \dot{\ol{p}}_i = \dirac{\ol{p}_i}{{\cal H}'} = \spoiss{\ol{p}_i}{{\cal
292: H}'} \approx 0$, leads us to \textbf{secondary constraints}
293: %
294: \beqa \label{G1}
295: G_1 & = & G_1^g \\\label{G2}
296: G_2 & = & G_2^g + \frac{i}{\sqrt{2}}f(X)(\chid_1 \lrpd{1} \chi_1) + e_1^+ h(X) g(\chib\chi) \\\label{G3}
297: G_3 & = & G_3^g - \frac{i}{\sqrt{2}}f(X)(\chid_0 \lrpd{1} \chi_0) - e_1^- h(X) g(\chib\chi)
298: \eeqa
299: with
300: \beqas
301: G_1^g & = & \partial_1 X + X^- e_1^+ - X^+ e_1^- \\
302: G_2^g & = & \partial_1 X^+ + \om_1 X^+ - e_1^+ \cal V \\
303: G_3^g & = & \partial_1 X^- - \om_1 X^- + e_1^- \cal V
304: \eeqas
305: %
306: The Hamiltonian density now turns out to be constrained to zero, as
307: expected for a \textbf{generally covariant system}\footnote{Up to a
308: boundary term $\int\limits_{\partial \cal M} p_i \ol{q}^i$ coming
309: from ${\cal L}_{FOG}$.} \cite{Henneaux:1992}.
310: %
311: \beq\label{constrainedhamiltonian}
312: {\cal H} = -\ol{q}^i G_i
313: \eeq
314: %
315: This already lets us to expect that the $G_i$ are related to the
316: generators of the three gauge symmetries in the system, namely local
317: Lorentz symmetry and the two diffeomorphisms.
318:
319: The secondary constraints commute with the $\ol{p}_i$ because both the
320: $G_i$ and $\Phi_\al$ are independent of the $\ol{q}^i$. They also
321: trivially commute with the primary second class constraints,
322: $\dirac{\Phi_\al}{G_j'} = 0$, because of the definition of the Dirac
323: bracket. For the same reason the $\Phi_\al$ do not give rise to new
324: secondary constraints.
325:
326:
327: \subsection{Algebra of the secondary constraints}
328:
329: \textbf{Dirac conjectured} \cite{Dirac:1996} that every first class
330: constraint generates a gauge symmetry. The proof of this conjecture is
331: possible in a very general setting \cite{Gitman:1990}, but some
332: additional assumptions (see paragraph 3.3.2 of \cite{Henneaux:1992})
333: to rule out ``pathological'' examples make it easier. These
334: assumptions are fulfilled in our case, because 1. every constraint
335: belongs to a well defined generation; 2. the Dirac bracket ensures
336: that the primary second class constraints do not generate new ones
337: and, as will be seen below, the secondary constraints are first class
338: and there are no ternary constraints; and 3. every primary first class
339: constraint $\ol{p}_i = 0$ generates one $G_i$.
340:
341: To show that the system doesn't admit any ternary constraints, it is
342: sufficient to show that the algebra of secondary constraints closes,
343: i.e. $\dirac{G_i}{G_j '} = \left.C_{ij}\right.^k (x)G_k\;\de(x-x')$, and thus
344: the secondary constraints are preserved under the time evolution
345: generated through the Dirac bracket,
346: %
347: \beqs
348: \dot{G}_i = \dirac{G_i}{\left.{\cal H}\right.' } = - \left.\ol{q}^j\right. ' \dirac{G_i}{G_j'} \approx 0
349: \eeqs
350: %
351:
352: To calculate all the Dirac brackets, one first needs the Poisson
353: brackets $\spoiss{\Phi_\al}{G_j}$. They are rather lengthy and thus
354: listed in Appendix \ref{app:B}. The resulting \textbf{algebra} then
355: reads
356: %
357: \beqa\label{GiGi}
358: \dirac{G_i}{G_i '} & = & 0 \hspace{5mm} i = 1,2,3 \\\label{G1G2}
359: \dirac{G_1}{G_2 '} & = & - G_2 \,\de \\\label{G1G3}
360: \dirac{G_1}{G_3 '} & = & G_3 \,\de \\\label{G2G3}
361: \dirac{G_2}{G_3 '} & = & \left[ - \sum\limits_{i=1}^3 \td{{\cal V}}{p_i} G_i + \left(g h' - \frac{h}{f} f'g'\cdot (\chib\chi)\right)G_1 \right]\de
362: \eeqa
363: %
364:
365: We'd like to comment on some technical points. Only obtaining
366: \eqref{G2G3} needs some care, the others brackets are rather
367: straightforward, using the Poisson structure of our phase space
368: \eqref{poisson}. However, one should keep in mind that the $Q^\al$ are
369: anti commuting. The tricky part of \eqref{G2G3} is actually not the
370: $C^{\al\be}$-term in the Dirac bracket, but the bracket
371: $\spoiss{G_2[\varphi]}{G_3[\psi]}$ and therein the integrations by
372: part during calculation, which have to be performed using smeared
373: constraints, i.e.
374: %
375: \beqs
376: G_i[\varphi] = \int dx \; \varphi(x) G_i(x)
377: \eeqs
378: %
379: The bracket itself reads with \eqref{G2}, \eqref{G3}
380: %
381: \beqa\nonumber
382: \spoiss{G_2[\varphi]}{G_3[\psi]} & = & \iint dx dz \varphi(x) \psi(z) \left( \spoiss{G_2^{old}(x)}{G_3^{old}(z)}
383: + \spoiss{q^3(x)h(x)g(x)}{G_3^{old}(z)}\right.\\\label{sG2G3}
384: & & \hspace{37mm} \left.- \spoiss{G_2^{old}(x)}{q²(z)h(z)g(z)}\right)
385: \eeqa
386: %
387: Here we denote with $G^{old}$ the constraints with $h=0$,
388: %
389: \beqas
390: G_1^{old} & = & G_1^g = G_1 \\
391: G_2^{old} & = & G_2^g + \frac{i}{\sqrt{2}}f(X)(\chid_1 \lrpd{1} \chi_1)\\
392: G_3^{old} & = & G_3^g - \frac{i}{\sqrt{2}}f(X)(\chid_0 \lrpd{1} \chi_0)
393: \eeqas
394: %
395: and
396: \beqs
397: \spoiss{G_2^{old}(x)}{G_3^{old}(z)} = - \sum\limits_{i=1}^3 \td{{\cal V}}{p_i} G_i^{g}
398: \eeqs
399: %
400: The tricky parts are the second and third bracket in \eqref{sG2G3}
401: \footnote{The points in space where the functions are evaluated are denoted by subscript here, e.g. $h_x(p_1) := h(p_1(x))$}:
402: %
403: \beqas
404: \spoiss{(q³h(p_1)g(\chib\chi))[\varphi]}{G_3^{old}[\psi]}
405: %& = & \iint dx dz \varphi_x \psi_z \spoiss{q^3_x h_x(p_1)g_x(\chib\chi)}{G_{3,z}^{old}} \\
406: & = & \iint dx dz \varphi_x \psi_z g_x(\chib\chi) \poiss{q^3_xh_x(p_1)}{G_{3,z}^g} \\
407: & = & \iint dx dz \varphi_x \psi_z g_x(\chib\chi) \left[ (\partial_z \de(x-z))h_x(p_1) \right.\\
408: & & \hspace{15mm}\left.- (q^1h-q^3p_3 h' -q^2p_2 U h)_x \;\de(x-z) \right]\\
409: \spoiss{G_2^{old}[\varphi]}{(q²h(p_1)g(\chib\chi))[\psi]}
410: % & = & \iint dx dz \varphi_x \psi_z \spoiss{G_{2,x}^{old}}{q^2_z h_z (p_1) g_z(\chib\chi)} \\
411: & = & \iint dx dz \varphi_x \psi_z g_z(\chib\chi) \poiss{G_{2,x}^g}{q^2_z h_z(p_1)} \\
412: & = & \iint dx dz \varphi_x \psi_z g_z(\chib\chi) \left[ (-\partial_x \de(x-z))h_z(p_1) \right.\\
413: & & \hspace{15mm}\left.- (q^1h-q^2p_2 h' -q^3p_3 U h)_x\;\de(x-z) \right]
414: \eeqas
415: %
416: \beqas
417: & \Rightarrow & \hspace{12mm} \spoiss{(q³h(p_1)g(\chib\chi))[\varphi]}{G_3^{old}[\psi]} -
418: \spoiss{G_2^{old}[\varphi]}{(q²h(p_1)g(\chib\chi))[\psi]} \\
419: & & \hspace{3mm} = \hspace{5mm} \iint dx dz \varphi_x \psi_z [(\underbrace{(\partial_z \de(x-z))}\limits_{= -\partial_x \de(x-z)} g_x(\chib\chi) h_x(p_1) + (\partial_x \de(x-z)) g_z(\chib\chi) h_z(p_1)) \\
420: & & \hspace{35mm} \left. -g (q^2p_2-q³p_3)(h'(p_1)-U(p_1) h(p_1)) \de(x-z)\right]\\
421: & & \stackrel{int.p.p.}{=} \iint dx dz \varphi_x \psi_z \de(x-z)[ \partial_x(hg) - g(q²p_2-q³p_3)(h'-Uh) ]
422: \eeqas
423: Thus we get for the graded Poisson bracket of $G_2$ and $G_3$
424: \beqas\nonumber
425: \spoiss{G_2}{G_3 '} & = & \left[ -\td{{\cal V}}{p_i} G_i^g + \frac{i}{\sqrt{2}} f' [p_3(Q³\lrpd{x}Q^1)-p_2(Q²\lrpd{x}Q^0)]\right. \\
426: & & \hspace{5mm}\left. + \partial_x(hg) - g(q²p_2-q³p_3)(h'-Uh)\right]\delta
427: \eeqas
428:
429: The $C^{\al\be}$-terms of the Dirac bracket are (with $\partial_xg = g'\partial_x(\chib\chi)$, $\partial_x f = f' \partial_x p_1$ and $p_2q²-p_3q³-\partial_x p_1 = -G_1$)
430: \beqa\nonumber
431: -\frac{i}{\sqrt{2}}[f'+Uf][p_3(Q³\lrpd{x}Q^1)-p_2(Q²\lrpd{x}Q^0)] - \frac{h}{f}f'g'\cdot(\chib\chi) G_1 - h(\partial_x g)
432: \eeqa
433:
434: With these results, we obtain (omitting the $\de(x-x')$)
435: %
436: \beqa\nonumber
437: \dirac{G_2}{G_3 '} & = & -\td{\cal V}{p_1}G_1
438: -\td{\cal V}{p_2}\left(G_2^g + \frac{i}{\sqrt{2}} f (Q³\lrpd{x}Q^1) + q³hg\right)\\\nonumber
439: & & -\td{\cal V}{p_3}\left(G_3^g - \frac{i}{\sqrt{2}} f (Q³\lrpd{x}Q^1) - q²hg\right)\\\nonumber
440: & & +\underbrace{\partial_x(hg)-h(\partial_x g) -gh'(\partial_x p_1)}\limits_{=0}
441: + g h' G_1\\\nonumber
442: & & - \frac{h}{f}f' g'\cdot (\chib\chi)G_1\\\nonumber
443: & = & - \td{\cal V}{p_i}G_i + (gh' - \frac{h}{f}f' g'\cdot (\chib\chi))G_1
444: \eeqa
445:
446:
447:
448: \section{Relation to the Conformal algebra}\label{sec:witt}
449:
450:
451: As first noted in \cite{Katanaev:1994qf}, certain linear
452: combinations of the $G_i$ fulfil the Witt algebra. In that work FOG
453: coupled to scalar matter was considered. However the same result holds
454: in our case. New generators $G = G_1$, $H_{0/1} = q^1 G_1 \mp q^2 G_2
455: + q^3 G_3$ fulfil an algebra (with $\de' = \pd{\de(x-x')}{x'}$)
456:
457: \beq
458: \begin{array}{ll}
459: \dirac{G}{G'} = 0 & \qquad \dirac{H_i}{H_i'} = (H_1 + H_1')\de' \\
460: \dirac{G}{H_i} = -G \de' & \qquad \dirac{H_0}{H_1'} = (H_0 + H_0')\de'
461: \end{array}
462: \eeq
463:
464: Some Dirac brackets needed for calculating this algebra are listed in
465: App. \ref{app:C}. Fourier transforming the light cone combinations
466: $H^\pm = H_0\pm H_1$ according to $H^+(x) = \int \frac{dk}{2\pi} L_k
467: e^{ikx}$; $H^-(x) = \int \frac{dk}{2\pi} \overline{L}_k e^{ikx}$ shows
468: that the $L_k$ (and equally $\ol{L}_k$) fulfil the classical Virasoro
469: algebra
470:
471: \beq\label{witt}
472: \dirac{L_k}{L_m} = i(k-m) L_{k+m}
473: \eeq
474:
475: \section{Discussion \& Outlook}\label{sec:conclusions}
476:
477: From \eqref{GiGi} - \eqref{G2G3} its clear that the algebra of
478: secondary constraints closes with delta functions. This implies the
479: absence of ternary constraints. The $G_i$ on-shell generate three
480: gauge symmetries, namely one local Lorentz symmetry ($G_1$) and
481: two diffeomorphisms ($G_2$, $G_3$), which can be seen from
482: \beqa
483: \dirac{G_1}{X^\pm} & = & \mp X^\pm \de \\
484: \dirac{G_{2/3}}{X} & = & \pm X^\pm \de
485: \eeqa
486: by comparing with the transformation property of $X^\pm$ under Lorentz
487: transformations and the Lorentz scalar $X$ under diffeomorphisms.
488:
489: The nontrivial part of the algebra of first class constraints still
490: closes like in the case of a compact Lie groups,
491: $[\de_A,\de_B]={f^C}_{AB}(x)\de_C$, but rather with structure
492: functions than with constants. This is especially important when
493: considering BRST symmetry, because the homologic perturbation series
494: for the BRST charge then terminates at Yang-Mills level
495: \cite{Grumiller:2006}.
496:
497: The second term in \eqref{G2G3} deserves some remarks: First, it
498: vanishes for minimal coupling, i.e. for $h=f=const.$ Second, if $h
499: \propto f$, it becomes proportional to $f'(g-g'\cdot\chib\chi)G_1\de =
500: - f'\la(\chib\chi)²G_1\de$. Thus a mass term $m\chib\chi$ doesn't
501: change the constraint algebra at all. Third, it doesn't contain
502: derivatives of $\chi$. This is different from the case of scalar
503: matter (see eq. E.31 in \cite{Grumiller:2001ea}), where the additional
504: contribution to $\poiss{G_2}{G_3}$ is proportional to
505: $\frac{f'}{f}{\cal L}_{scalar}$.
506:
507: Boundary contributions both to the dilaton and the fermionic action
508: have been omitted this work. Dilaton theories with boundaries (but
509: without matter) have been considered recently in detail
510: \cite{Bergamin:2005pg}, with the result that a consistent variational
511: principle can be defined. We don't expect problems from the interplay
512: of fermion boundary terms and gravitational ones. Nevertheless this
513: point still has to be worked out.
514:
515: As noted in Sec. \ref{sec:intro}, another motivation for this work
516: stems from the recent paper by Frolov, Kristjánsson and Thorlacius
517: \cite{Frolov:2005ps}, who considered a two-dimensional Schwinger model
518: on a curved background manifold to investigate the effect of
519: pair-production on the global structure of black hole space times. To
520: this end they used the quantum equivalence of the Schwinger model in
521: 1+1 dimensions and the Sine-Gordon model found by Coleman, Jackiw \&
522: Susskind \cite{Coleman:1975pw} to do calculations on the Sine-Gordon
523: side. It is an interesting question whether Bosonisation still shows
524: up in a quantum theory with dynamical gravitational background. For
525: matter less generalised dilaton theories \eqref{dilatonaction} and for
526: ones coupled to scalar fields an exact path integral quantisation of
527: the geometric sector is known \cite{Kummer:1996hy} and gives rise to a
528: nonlocal vertices effective theory which turn out to be interesting
529: intermediary states like Virtual Black Holes \cite{Grumiller:2000ah}.
530: A similar analysis for our case is in preparation
531: \cite{Grumiller:2006} and will shed some light on the question posted
532: above.
533:
534:
535: \acknowledgments
536:
537: This text is a contribution to the proceedings of the International
538: V.A. Fock School for Advances in Physics (IFSAP-2005) held in St.
539: Petersburg in November this year. The author would like to thank the
540: organisers of the school and especially Prof. Victor Novozhilov for
541: his sacrificial work and for the opportunity of giving a talk on the
542: results presented here. Special thanks deserves Dmitri Vassilevich for
543: recommending me for a UNESCO short term fellowship.
544:
545: The results presented here are part of my diploma thesis, and I am
546: very grateful to my advisors, Daniel Grumiller and Prof. Gerd
547: Rudolph, for their support of my studies, and in particular to Daniel
548: \& Dmitri for many discussions on physics and two-dimensional gravity.
549:
550: Last but not least I want to thank Prof. Wolfgang Kummer for the
551: invitation to talk in Vienna in October and Luzi Bergamin for
552: financial support.
553:
554: \begin{appendix}
555:
556: \section{Conventions}\label{app:A}
557:
558: For the Levi-Civit{\'a} symbols both in tangent space $\epsilon^{ab}$and
559: on the manifold $\tilde{\epsilon}^{\mu\nu}$, we fix $\epsilon^{01} :=
560: +1$. This is necessary to retain $\epsilon^{\mu\nu}=e_a^\mu e_b^\nu
561: \epsilon^{ab}$, with $\epsilon^{\mu\nu}$ now being the Levi-Civit{\'a}
562: tensor. In the tangent space we use light cone coordinates $X^\pm =
563: \frac{1}{\sqrt{2}}(X^0 \pm X^1)$, and thus $\eps^{ab}_{LC} =
564: -\eps^{ab}$. For the square root of the determinant of the metric we
565: sometimes write $(e):=e_0^- e_1^+ - e_0^+e_1^- = -\det({e_\mu}^a) =
566: \sqrt{-\det(g_{\mu\nu})}$, whereas the volume 2-form is $\epsilon =
567: -(e)d^2x = \det({e_\mu}^a)dx^0 \wedge dx^1$. The Hodge star is defined
568: as in \cite{Grumiller:2002nm}. The two-sided derivative is
569: $a\overleftrightarrow{\mathrm{d}}b := a (\mathrm{d} b) -
570: (\mathrm{d}a)b$
571:
572: Our \textbf{Dirac matrices} are
573: %
574: \beqs
575: \begin{array}{ll}
576: \ga^0 = \ \;\left(
577: \begin{array}{rr}
578: 0 & 1 \\
579: 1 & 0
580: \end{array}\right)
581: & \ \ \
582: \ga^1 = \left(
583: \begin{array}{rr}
584: 0 & 1 \\
585: -1 & 0
586: \end{array}\right) \\
587: \ga^+ = \left(
588: \begin{array}{rr}
589: 0 & \sqrt{2} \\
590: 0 & 0
591: \end{array}\right)
592: & \ \ \
593: \ga^- = \left(
594: \begin{array}{rr}
595: 0 & 0 \\
596: \sqrt{2} & 0
597: \end{array}\right)
598: \end{array}
599: \eeqs
600: The analogue of $\ga_5$ is defined as $\ga_* := \ga^0\ga^1 =
601: \frac{1}{2}[\ga^0,\ga^1] $.
602: %
603: Because our fermion fields are Grassmann variables, throughout this
604: article we use the \textbf{graded Poisson bracket} defined as
605: \cite{Henneaux:1992} ($\partial^L$ is the usual left derivative)
606: %
607: \beqas
608: \spoiss{F}{G} & = & \int dz \left[ \pd{F}{q^i(z)} \pd{G}{p_i(z)} - \pd{F}{p_i(z)} \pd{G}{q^i(z)} \right] \\
609: & & + (-)^{\eps(F)} \left[ \pld{F}{Q^\al(z)} \pld{G}{P_\al(z)} - \pld{F}{P_\al(z)} \pld{G}{Q^\al(z)} \right]
610: \eeqas
611: %
612: with $(q,p)$ and $(Q,P)$ being a set of bosonic ($\eps(q)=\eps(p)=0$)
613: and fermionic ($\eps(Q)=\eps(P) =1$), and $\eps(F)$ the Grassmann
614: parity of $F$. Its main
615: properties used here are
616: %
617: \beqas
618: \spoiss{F}{G} & = & (-)^{\eps(F)\eps(G) + 1} \spoiss{G}{F} \\
619: \spoiss{F}{G_1 G_2} & = & \spoiss{F}{G_1} G_2 + (-)^{\eps(F)\eps(G_1)} G_1 \spoiss{F}{G_2}
620: \eeqas
621: %
622: All these properties carry over to the corresponding Dirac bracket
623: defined by \eqref{diracbracket}.
624:
625: \section{Brackets of the secondary with the second class constraints}\label{app:B}
626:
627: To calculate the Dirac brackets, we need all the graded Poisson
628: brackets of the $G_i$ with the $\Phi_\al$. They are easily obtained by
629: using the algebraic properties of the graded Poisson bracket (see App.
630: \ref{app:A}).
631: %
632: \beqas
633: \spoiss{G_1}{\Phi_0'} & = & -\frac{i}{\sqrt{2}} f e_1^+ \chid_0 \;\de(x-x')\\
634: \spoiss{G_1}{\Phi_2'} & = & -\frac{i}{\sqrt{2}} f e_1^+ \chi_0 \;\de(x-x')\\
635: \spoiss{G_1}{\Phi_1'} & = & -\frac{i}{\sqrt{2}} f e_1^- \chid_1 \;\de(x-x')\\
636: \spoiss{G_1}{\Phi_3'} & = & -\frac{i}{\sqrt{2}} f e_1^- \chi_1 \;\de(x-x')\\
637: \spoiss{G_2}{\Phi_0'} & = & \frac{i}{\sqrt{2}} \left[ f'+Uf \right] X^+ e_1^+ \chid_0\;\de(x-x')\\
638: & & -\; e_1^+ h g' \chid_1 \;\de(x-x')\\
639: \spoiss{G_2}{\Phi_2'} & = & \frac{i}{\sqrt{2}} \left[ f'+Uf \right] X^+ e_1^+ \chi_0\;\de(x-x')\\
640: & & +\; e_1^+ h g' \chi_1 \;\de(x-x')
641: \eeqas
642:
643: However one must be careful with integrating by parts the derivatives
644: of the delta distributions. This is most easily done by smearing the
645: fields with appropriate test functions
646: %
647: \footnote{After integrating by parts one obtains
648: \beqs
649: \int dx \varphi(x)\left[ f(x) - f(y) \right]\partial_x\de(x-y) = - \int dx \varphi(x) (\partial_x f(x)) \delta(x-y)
650: \eeqs},
651: %
652: \beqas
653: \spoiss{G_2}{\Phi_1'} & = & \frac{i}{\sqrt{2}} \left[ \chid_1(\om_1 f - X^+ e_1^- f' - X^- e_1^+ U f)
654: + 2 (\partial_x \chid_1)f + (\partial_x f)\chid_1 \right] \de(x-x')\\
655: & & -\; e_1^+ h g' \chid_0 \;\de(x-x')\\
656: \spoiss{G_2}{\Phi_3'} & = & \frac{i}{\sqrt{2}} \left[ \chi_1(\om_1 f - X^+ e_1^- f' - X^- e_1^+ U f)
657: + 2 (\partial_x \chi_1)f + (\partial_x f)\chi_1 \right] \de(x-x')\\
658: & & +\; e_1^+ h g' \chi_0 \;\de(x-x')\\
659: \spoiss{G_3}{\Phi_0'} & = & \frac{i}{\sqrt{2}} \left[ \chid_0(\om_1 f - X^- e_1^+ f' - X^+ e_1^- U f)
660: - 2 (\partial_x \chid_0)f - (\partial_x f)\chid_0 \right] \de(x-x')\\
661: & & +\; e_1^- h g' \chid_1 \;\de(x-x')\\\nonumber
662: \spoiss{G_3}{\Phi_2'} & = & \frac{i}{\sqrt{2}} \left[ \chi_0(\om_1 f - X^- e_1^+ f' - X^+ e_1^- U f)
663: - 2 (\partial_x \chi_0)f - (\partial_x f)\chi_0 \right] \de(x-x')\\
664: & & -\; e_1^- h g' \chi_1 \;\de(x-x')\\\nonumber
665: \spoiss{G_3}{\Phi_1'} & = & \frac{i}{\sqrt{2}} \left[ f'+Uf \right] X^- e_1^- \chid_1\;\de(x-x')\\
666: & & +\; e_1^- h g' \chid_0 \;\de(x-x')\\\nonumber
667: \spoiss{G_3}{\Phi_3'} & = & \frac{i}{\sqrt{2}} \left[ f'+Uf \right] X^- e_1^- \chi_1\;\de(x-x')\\
668: & & -\; e_1^- h g' \chi_0 \;\de(x-x')
669: \eeqas
670:
671: \section{Dirac Brackets needed for eq. \eqref{witt}}\label{app:C}
672:
673: \beqas
674: \dirac{G_1}{q^1} & = & -\partial_1 \de \\
675: \dirac{G_1}{q^2} & = & q^2 \de \\
676: \dirac{G_1}{q^3} & = & -q^3 \de \\
677: \dirac{G_2}{q^1} & = & q^3\left[ \pd{\cal V}{p_1} - \left(h' g - \frac{f'}{f}h g' \cdot (\chib\chi)\right) \right] \de\\
678: \dirac{G_2}{q^2} & = & - \left[ \partial_1 + q^1 - q^3\pd{\cal V}{p_2}\right]\de \\
679: \dirac{G_2}{q^3} & = & q^3 \pd{\cal V}{p_3} \de
680: \eeqas
681: \beqas
682: \dirac{G_3}{q^1} & = & -q^2\left[ \pd{\cal V}{p_1} - \left(h' g - \frac{f'}{f}h g' \cdot (\chib\chi)\right) \right] \de \\
683: \dirac{G_3}{q^2} & = & -q^2 \pd{\cal V}{p_2} \de \\
684: \dirac{G_3}{q^3} & = & - \left[ \partial_1 - q^1 + q^2\pd{\cal V}{p_3}\right]\de
685: \eeqas
686: \beqas
687: \dirac{q^iG_i}{q^iG_i} & = & -(\partial_1\de) q^i G_i \ \ \mathrm{(no\ summation)} \\
688: \dirac{q^1G_1}{q^2G_2} & = & -q^2q^3\left[ \pd{\cal V}{p_1} - \left(h' g - \frac{f'}{f}h g' \cdot (\chib\chi)\right) \right] G_1 \de
689: = - \dirac{q^1G_1}{q^3G_3} = \dirac{q^2G_2}{q^3G_3}
690: \eeqas
691:
692: \end{appendix}
693:
694: \input{constraintanalysis.bbl.fix}
695: %\bibliographystyle{JHEP}
696: %\bibliography{constraintanalysis}
697:
698:
699: \end{document}
700: