hep-th0009204/pol4
1: \documentstyle[12pt,titlepage,epsf]{article}  % Specifies the document style.
2: \setlength{\textwidth}{17.0cm}  % Specifies the size of the printed part of
3: \setlength{\textheight}{23.5cm} % page and correct margins for the paper
4: \setlength{\topmargin}{-2.5cm}      % sets the appropriate left and top margins.
5: \setlength{\oddsidemargin}{0in}
6: \setlength{\topsep}{5ex}      % sets amount of separation between text & formula
7: \setlength{\textfloatsep}{0.5in}  % sets amount of space between fig and text.
8: \setlength{\intextsep}{0.5in} % also sets amount of space between fig and text.
9: \setcounter{secnumdepth}{4}     % Numbers upto paragraphs
10: \setcounter{tocdepth}{4}        % Lists upto paragraph in the contents
11: \setlength{\parindent}{0.5in}
12: \setlength{\unitlength}{1cm}
13: \newcommand{\nin}{\noindent}
14: \newcommand{\be}{\begin{equation}}
15: \newcommand{\ee}{\end{equation}}
16: \newcommand{\bea}{\begin{eqnarray}}
17: \newcommand{\eea}{\end{eqnarray}}
18: \newcommand{\br}{\hskip .25cm/\hskip -.25cm}
19: \newcommand{\hf}{\frac{1}{2}}
20: \newcommand{\nonu}{\nonumber\\}
21: \newcommand{\la}{\hskip .25cm^\leftarrow\hskip -.25cm}
22: \newcommand{\dg}{^\dagger}
23: \newcommand{\ol}{\overline}
24: 
25: \begin{document}
26: 
27: \hfill   NTUA-97/00
28: 
29: \begin{center}
30: 
31: {\Large Vacuum polarization in thermal QED\\
32: with an external magnetic field}
33: 
34: \vspace{1cm}
35: 
36: J.Alexandre\footnote{jalex@central.ntua.gr}
37: 
38: Department of Physics, National Technical University of
39: Athens,
40: 
41: Zografou Campus, 157 80 Athens, Greece
42: 
43: \vspace{15mm}
44: 
45: Abstract
46: \end{center}
47: 
48: The one-loop vacuum polarization tensor is computed in QED with an external, 
49: constant, homogeneous magnetic field at finite temperature. The Schwinger
50: proper-time formalism is used and the computations
51: are done in Euclidian space. The well-known results are
52: recovered when the temperature and/or the magnetic field are switched off
53: and the effect of the magnetic field on the Debye screening is discussed. 
54: 
55: 
56: \section*{Introduction}
57: 
58: 
59: The question of dynamical chiral symetry breaking 
60: in thermal $QED$ with an external
61: magnetic field (magnetic catalysis) has been studied 
62: in the context of the electroweak 
63: transition \cite{electroweak} and also, with $QED_3$ (in 2+1 dimensions),
64: in the framework of effective 
65: descriptions of planar superconductors \cite{superconductors},\cite{afk2}.
66: Recent studies of the magnetic catalysis at zero temperature \cite{afk1}
67: showed that it is essential to take into account the momentum
68: dependence of the fermion self-energy since the dynamical mass
69: given by the constant self-energy
70: approximation proved to be too small, 
71: by several orders of magnitude 
72: in the case of $QED$. These studies have been made with the analysis of
73: the gap equation provided by the Schwinger-Dyson equation, where 
74: the photon propagator was truncated at the one-loop level. 
75: The polarization tensor in the presence of an external magnetic field
76: was used in its lowest Landau level approximation,
77: as was done in \cite{miransky}.
78: The study of the magnetic catalysis at finite temperature
79: taking into account the momentum dependence of the 
80: fermion self-energy has been done in $QED_3$ \cite{afk2} but not
81: in $QED$ for which only the constant self-energy approximation
82: has been done \cite{gusynin}, \cite{lee}. 
83: With a study including the momentum dependence,       
84: we can still expect the critical temperature for 
85: the magnetic catalysis to be of the order of
86: the dynamical mass found at zero temperature \cite{gusynin}, 
87: but where the latter is given by a momentum-dependent
88: analysis as was made in \cite{afk1}. 
89: As a first step in this direction, we compute here
90: the one-loop polarization tensor in finite temperature
91: $QED$ in the presence of a external, constant, homogeneous magnetic field.
92: 
93: The computation will be done in Euclidian space, 
94: using the proper-time formalism 
95: introduced by Schwinger \cite{schwinger} which 
96: takes into account the complete
97: interaction between the fermion and the external, classical  field.
98: The same computation has been done at zero temperature in 
99: the paper \cite{tsai}
100: which will be often cited in the present article and the 
101: generalization to any external constant field is done in \cite{schubert},
102: using the 'string-inspired' technique.
103: We note that the derivation of the Heisenberg-Euler lagrangian
104: has been done at finite temperature with the same formalism
105: \cite{dittrich}, as well as the generalization to any 
106: external constant electromagnetic field \cite{gies}.
107: 
108: Section 1 will introduce the notations and recall the 
109: characteristics of fermions in an external magnetic field.
110: Section 2 will be devoted to the computation of the 44-component
111: of the polarization tensor: this presentation is chosen
112: for the sake of clarity since the external environmement
113: strongly breaks the symmetry between the Lorentz indices such
114: that the computation is not straightforward. The 
115: technical details of the method will be explained and we will recover the
116: well-known results in the limit where the temperature and/or the
117: magnetic field go to zero.
118: The other components will be computed in section 3 where the
119: transversality of the polarization tensor will be checked.
120: The section 4 will give the strong field approximation of the 
121: 44-component of the polarization tensor, consitent with the lowest 
122: Landau level approximation. Finally, the conlusion will show the Debye screening 
123: obtained through these computations.
124: 
125: 
126: 
127: \section{Fermions in a constant magnetic field}
128: 
129: 
130: To fix our notations we shortly review here the characteristics of
131: fermions in a external, constant, homogeneous magnetic field 
132: at zero temperature. 
133: 
134: The model we are going to consider is described by the Lagrangian density:
135: 
136: \be
137: {\cal L}=-\frac{1}{4}F_{\mu\nu}F^{\mu\nu}+i\ol\Psi D_\mu\gamma^\mu\Psi 
138: -m\ol\Psi\Psi,
139: \ee
140: 
141: \nin where $D_\mu = \partial_\mu+i e A_\mu + i e A^{ext}_\mu$, 
142: $A_\mu$ is the abelian quantum gauge field, $F_{\mu \nu}$ its
143: corresponding field strength, and $A^{ext}_\mu$ describes 
144: the external magnetic field. 
145: We recall the usual definition $e^2\equiv 4\pi\alpha$.
146: 
147: We will choose the symmetric gauge for the external field ($\vec B$ is 
148: in the direction 3)
149: 
150: \be\label{symgauge}
151: A^{ext}_0(x)=0,~~A^{ext}_1(x)=-\frac{B}{2}x_2,
152: ~~A^{ext}_2(x)=+\frac{B}{2}x_1,~~A^{ext}_3(x)=0
153: \ee
154: 
155: \nin for which we know from the work by Schwinger 
156: \cite{schwinger} that the fermion propagator is given by:
157: 
158: \be
159: S(x,y)=e^{iex^\mu A^{ext}_\mu(y)}\tilde S(x-y),
160: \label{sphase}
161: \ee
162: 
163: \nin where the translational invariant propagator $\tilde S$ 
164: has the following Fourier transform in the proper-time formalism:
165: 
166: \bea\label{schwingerrep}
167: \tilde S(p)&=&\int_0^\infty ds e^{is\left(p_0^2-p_3^2-p_\bot^2
168: \frac{\tan(|eB|s)}{|eB|s}-m^2\right)}\nonu
169: &\times&
170: \left[(p^0\gamma^0-p^3\gamma^3+m)(1+\gamma^1\gamma^2\tan(|eB|s))
171: -p^\bot\gamma^\bot
172: (1+\tan^2(|eB|s))\right]
173: \eea
174: 
175: \nin where $p^\bot=(p^1,p^2)$ is the transverse momentum and the 
176: same notation holds for the gamma matrices. 
177: 
178: Let us now turn to the finite temperature case.
179: We will note the fermionic Matsubara modes $\hat\omega_l=(2l+1)\pi T$ 
180: and the bosonic ones $\omega_n=2n\pi T$. The translational invariant part
181: of the bare fermion propagator reads in Euclidian space 
182: ($p_0\to i\hat\omega_l$) and with 
183: the substitution $s\to -is$:
184: 
185: \bea\label{freeTprop}
186: \tilde S_l(\vec p)&=&
187: -i\int_0^\infty ds e^{-s\left(\hat\omega_l^2+p_3^2+p_\bot^2
188: \frac{\tanh (|eB|s)}{|eB|s}+m^2\right)}\nonu
189: &\times&
190: \left[(-\hat\omega_l\gamma^4-p^3\gamma^3+m)(1-i\gamma^1\gamma^2\tanh(|eB|s))
191: -p^\bot\gamma^\bot(1-\tanh^2(|eB|s))\right]
192: \eea
193: 
194: \nin where the Euclidian gamma matrices satisfy the 
195: anticommutation relation
196: $\{\gamma^\mu,\gamma^\nu\}=-2\delta^{\mu\nu}$, with $\mu,\nu=1,2,3,4$
197: and $\vec p=(p_\bot,p_3)$.
198: 
199: Finally, the one-loop polarization tensor is 
200: 
201: \be\label{poltensor}
202: \Pi^{\mu\nu}_n(\vec k)=
203: -4\pi\alpha T\int\frac{d^3\vec p}{(2\pi)^3}
204: \sum_{l=-\infty}^\infty\mbox{tr}\left\{\gamma^\mu \tilde S_l(\vec p)
205: \gamma^\nu\tilde S_{l-n}(\vec p-\vec k)\right\}
206: +Q^{\mu\nu}(k)
207: \ee
208: 
209: \nin where $Q^{\mu\nu}$, usually called the 'contact term', 
210: cancels the ultraviolet divergences and
211: therefore does not depend on the temperature or on the
212: magnetic field since these give finite effects. 
213: The addition of this contact term is equivalent to the 
214: addition of the counterterm $(1-Z_3)F^{\mu\nu}F_{\mu\nu}/4$
215: in the original Lagrangian and
216: the usual ultraviolet divergences appear in the 
217: proper-time formalism as singularities in $s=0$, as will be seen 
218: in the next section. With this proper-time method, 
219: a cut-off $0<\varepsilon<s$ provides
220: a gauge invariant regularization which will be used in what 
221: follows. The limit $\varepsilon\to 0$ will be taken after
222: computing the contact term $Q^{\mu\nu}$.\\
223: We note that the $A^{ext}_\mu$-dependent phase of the fermion propagator
224: does not contribute to the polarization tensor since in coordinate space
225: 
226: \be
227: \exp\left\{ie\left(x^\mu A^{ext}_\mu(y)+
228: y^\mu A^{ext}_\mu(x)\right)\right\}=1
229: \ee
230: 
231: \nin with the specific choice of gauge (\ref{symgauge}).
232: If we choose another potential $A^{ext}_\mu$, 
233: it is shown in \cite{gies} that the
234: change of gauge is equivalent to the introduction of a 
235: chemical potential.
236: 
237: 
238: 
239: \section{44-component}
240: 
241: 
242: 
243: 
244: 
245: With the expression (\ref{freeTprop}) of the fermion propagator, we obtain
246: for the 44-component of the polarization tensor
247: after the integration over $\vec p$
248: 
249: \bea
250: &&\Pi^{44}_n(\vec k)=
251: \frac{-2\alpha T}{\sqrt\pi}|eB|\int
252: \frac{dsd\sigma}{\sqrt{s+\sigma}(\tanh(|eB|s)+\tanh(|eB|\sigma))}\\
253: &&\times\sum_{l=-\infty}^\infty
254: e^{-\frac{k_\bot^2}{|eB|}
255: \frac{\tanh(|eB|s)\tanh(|eB|\sigma)}{\tanh(|eB|s)+\tanh(|eB|\sigma)}
256: -[(s+\sigma)(\hat\omega_l^2+m^2)
257: +s\omega_n(\omega_n-2\hat\omega_l)+\frac{s\sigma}{s+\sigma}k_3^2]}\nonu
258: &&\times
259: \Bigg[k_\bot^2\frac{\tanh(|eB|s)\tanh(|eB|\sigma)}
260: {(\tanh(|eB|s)+\tanh(|eB|\sigma))^2}(1-\tanh(|eB|s))
261: (1-\tanh(|eB|\sigma))\nonu
262: &&~~~~~~~~-|eB|\frac{(1-\tanh(|eB|s))(1-\tanh(|eB|\sigma))}
263: {\tanh(|eB|s)+\tanh(|eB|\sigma)}\nonu
264: &&+\left(\hat\omega_l(\hat\omega_l-\omega_n)-m^2+
265: \frac{s\sigma}{(s+\sigma)^2}k_3^2-
266: \frac{1}{2(s+\sigma)}\right)
267: \left(1+\tanh(|eB|s)\tanh(|eB|\sigma)\right)\Bigg]+Q^{44}(k)\nonumber
268: \eea
269: 
270: \nin In finite temperature computations,
271: one usually first does the summation over Matsubara modes and then
272: the integration over momenta. In this formalism, what is important 
273: as we will see below is to do the summation over Matsubara modes
274: before the integration over the proper-time parameters, when
275: the cut-off is removed (i.e. $\varepsilon\to 0$).
276: As in \cite{tsai}, we make the change of variable  $s=u(1-v)/2$ and
277: $\sigma=u(1+v)/2$ to obtain
278: 
279: \bea\label{pol1}
280: &&\Pi^{44}_n(\vec k)=
281: \frac{-\alpha T}{\sqrt\pi}|eB|
282: \int_\varepsilon^\infty du\sqrt{u}\int_{-1}^1 dv\sum_{l=-\infty}^\infty
283: e^{-\frac{k_\bot^2}{|eB|}\frac{\cosh\overline u-\cosh\overline uv}
284: {2\sinh\overline u}-
285: u[\hat\omega_l^2+m^2
286: +(1-v)\omega_n(\omega_n/2-\hat\omega_l)+\frac{1-v^2}{4}k_3^2]}\nonu
287: &\times&\left[\left(\hat\omega_l(\hat\omega_l-\omega_n)+
288: \frac{1-v^2}{4}k_3^2-\frac{1}{2u}
289: -m^2\right)\coth\overline u
290: -\frac{|eB|}{\sinh^2\overline u}+
291: k_\bot^2\frac{\cosh\overline u-\cosh\overline uv}{2\sinh^3\overline u}
292: \right]+Q^{44}(k)\nonumber
293: \eea
294: 
295: \nin where $\overline u=|eB|u$. We make the integration by parts over $u$
296: (we note $\phi(u)$ the exponent)
297: 
298: \be
299: |eB|\int_\varepsilon^\infty du e^{-\phi(u)}
300: \frac{\sqrt u}{\sinh^2\overline u}\longrightarrow
301: \int_\varepsilon^\infty du
302: e^{-\phi(u)}\sqrt u\coth\overline u
303: \left(\frac{1}{2u}-\frac{d\phi(u)}{du}\right)
304: \ee
305: 
306: \nin where we disregard the surface term \cite{tsai}. 
307: We then obtain the final expression
308: 
309: \bea\label{pol44}
310: &&\Pi^{44}_n(\vec k)=\frac{-\alpha T}{\sqrt\pi}|eB|
311: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv\sum_{l=-\infty}^\infty
312: e^{-\frac{k_\bot^2}{|eB|}\frac{\cosh\overline u-\cosh\overline uv}
313: {2\sinh\overline u}
314: -u[m^2+W_l^2+\frac{1-v^2}{4}(\omega_n^2+k_3^2)]}\\
315: &&\times
316: \left[\frac{k_\bot^2}{2}\frac{\cosh\overline uv-
317: v\coth\overline u\sinh\overline uv}{\sinh\overline u}
318: -\coth\overline u\left(\frac{1}{u}-2W_l^2+v\omega_nW_l-
319: \frac{1-v^2}{2}k_3^2\right)\right]+Q^{44}(k)\nonumber
320: \eea
321: 
322: \nin where $W_l=\hat\omega_l-\frac{(1-v)}{2}\omega_n$.
323: We note for the purpose of consistency
324: that the integrand in (\ref{pol44}) is an even function 
325: of the parameter $v$ since $W_l(-v)=-W_{n-l-1}(v)$ and therefore
326: $\sum_le^{-uW_l^2}$ is even in $v$, which ensures the symmetry
327: between the proper-times $s$ and $\sigma$. Thus it is important to 
328: perform the summation over Matsubara modes before doing the 
329: integrations over the proper-time parameters. Another reason to 
330: do the summation over Matsubara modes first
331: is to avoid artificial divergences in the temperature-dependent
332: part of the polarization tensor (which should be finite), as will
333: be seen at the end of this section.
334: 
335: Let us now determine the contact term $Q^{44}(k)$. Since it does not 
336: depend on the temperature or the magnetic field, it will be
337: determined after taking the limit $T\to 0$ and 
338: $|eB|\to 0$ of (\ref{pol44}).
339: If we set $T=0$ in (\ref{pol44}), 
340: we recover the zero-temperature results given 
341: in \cite{tsai} since the substitutions
342: $W_l\to p_4$ and $T\sum_l\to (2\pi)^{-1}\int dp_4$ lead to 
343: 
344: \bea
345: \lim_{T\to 0}&&T\sum_{l=-\infty}^\infty e^{-uW_l^2}=
346: \frac{1}{2\sqrt{\pi u}}\nonu
347: \lim_{T\to 0}&&T\sum_{l=-\infty}^\infty \left(\frac{1}{u}-2W_l^2\right)
348: e^{-uW_l^2}=0\nonu
349: \lim_{T\to 0}&&T\sum_{l=-\infty}^\infty v\omega_nW_l e^{-uW_l^2}=0
350: \eea
351: 
352: \nin and therefore ($\omega_n\to k_4$)
353: 
354: \bea\label{piindep}
355: &&\lim_{T\to 0}\Pi_n^{44}(\vec k)=
356: \frac{-\alpha|eB|}{4\pi}
357: \int_\varepsilon^\infty du\int_{-1}^1 dv
358: e^{-\frac{k_\bot^2}{|eB|}\frac{\cosh\overline u-\cosh\overline uv}
359: {2\sinh\overline u}
360: -u[m^2+\frac{1-v^2}{4}(k_4^2+k_3^2)]}\nonu
361: &&\times \left[k_\bot^2
362: \frac{\cosh\overline uv-v\coth\overline u\sinh\overline uv}
363: {\sinh\overline u}+k_3^2(1-v^2)\coth\overline u\right]
364: +Q^{44}(k)
365: \eea
366: 
367: \nin However it is important to keep a non zero fermion mass
368: if we wish to recover this infrared limit:
369: we will see in the conclusion that if we set $m=0$,
370: $\Pi^{44}_0(0)$ reaches
371: a non-zero value in the limit $T\to 0$. 
372: Thus we
373: can commute the limit $T\to 0$ and the integration over the
374: proper-time $u$ to find a consistent zero temperature result
375: only if $m\ne 0$, at least as long as $|eB|>0$.
376: This condition is consistent since the magnetic catalysis
377: generates a dynamical mass when the temperature is lower than
378: the critical one, for any value of
379: the gauge coupling, which has been proven at least for
380: strong magnetic fields \cite{miransky}. Therefore if the fermion
381: should be massless, the use of the dynamically generated
382: mass $m_{dyn}$ instead of $m=0$ in 
383: (\ref{schwingerrep}), corresponding to a resummation of 
384: graphs for the fermion propagator, would lead us back to the massive case.
385: Thus we will consider $m\ne 0$ in what follows.
386: 
387: If we now wish to take the limit of zero magnetic field
388: of (\ref{piindep}), we take $\overline u\to 0$ and obtain
389: 
390: \be
391: \lim_{T\to 0,|eB|\to 0}\Pi_n^{44}(\vec k)=
392: \frac{-\alpha}{4\pi}
393: \int_\varepsilon^\infty \frac{du}{u}\int_{-1}^1 dv      
394: e^{-u[m^2+\frac{1-v^2}{4}k^2]} 
395: (1-v^2)\vec k^2+Q^{44}(k)
396: \ee
397: 
398: \nin where $k^2=k_4^2+\vec k^2$.
399: Then if we take the contact term
400: 
401: \be
402: Q^{44}(k)=\frac{\alpha}{4\pi}
403: \int_\varepsilon^\infty \frac{du}{u}\int_{-1}^1 dv
404: e^{-um^2}(1-v^2)\vec k^2
405: \ee
406: 
407: \nin we obtain finally when $\varepsilon\to 0$
408: 
409: \bea
410: \lim_{T\to 0,|eB|\to 0}\Pi_n^{44}(\vec k)&=&
411: \frac{-\alpha}{4\pi}
412: \int_0^\infty \frac{du}{u}\int_{-1}^1 dv(1-v^2)
413: \left(e^{-u[m^2+\frac{1-v^2}{4}k^2]}
414: -e^{-um^2}\right)\vec k^2\nonu
415: &=&\frac{\alpha}{4\pi}
416: \int_{-1}^1 dv(1-v^2)
417: \ln\left(1+\frac{1-v^2}{4m^2}k^2\right)
418: \vec k^2
419: \eea
420: 
421: \nin which is a result obtained by standard methods
422: \cite{zuber} with the Feynman parameter $z=(1+v)/2$.
423: 
424: To finish the comparison with results already established,
425: let us take the zero magnetic field 
426: limit of (\ref{pol44}). 
427: The limit $\overline u\to 0$ 
428: leads to
429: 
430: \bea\label{limebzero}
431: \lim_{|eB|\to 0}\Pi^{44}_n(\vec k)&=&\frac{-\alpha T}{\sqrt\pi}
432: \int_\varepsilon^\infty \frac{du}{\sqrt u}\int_{-1}^1 dv
433: \sum_{l=-\infty}^\infty
434: e^{-u[m^2+W_l^2+\frac{1-v^2}{4}(\omega_n^2+\vec k^2)]}\nonu
435: &\times&
436: \left[\frac{\vec k^2}{2}(1-v^2)
437: -\frac{1}{u}+2W_l^2-v\omega_nW_l\right]+Q^{44}(k)
438: \eea
439: 
440: \nin For $|eB|=0$, we can take a massless fermion ($m=0$) 
441: since there is no magnetic catalysis and
442: the Debye screening is then given by
443: 
444: \be\label{thermmB0}
445: M_{|eB|=0,m=0}^2(T)=-\lim_{\vec k^2\to 0}\Pi^{44}_0(\vec k)|_{|eB|=0}
446: =c~\alpha T^2
447: \ee
448: 
449: \nin with ($\varepsilon\to 0$)
450: 
451: \be\label{constant}
452: c=2\sqrt\pi\int_0^\infty \frac{du}{\sqrt u}
453: \sum_{l=-\infty}^\infty e^{-u(2l+1)^2}\left[2(2l+1)^2-\frac{1}{u}\right]
454: \ee 
455: 
456: \nin To compute $c$, we use the Poisson resummation 
457: \cite{dittrich}:
458: 
459: \be\label{poiss}
460: \sum_{l=-\infty}^\infty e^{-a(l-z)^2}=
461: \left(\frac{\pi}{a}\right)^{1/2}
462: \sum_{l=-\infty}^\infty e^{-\frac{\pi^2l^2}{a}-2i\pi zl}
463: \ee
464: 
465: \nin which shows that it is essential to perform the summation over
466: Matsubara modes before doing the integration over
467: the proper-time $u$ to avoid the
468: singularity $\int duu^{-3/2}$ in (\ref{constant}) since
469: we obtain 
470: 
471: \be
472: \sum_{l=-\infty}^\infty e^{-u(2l+1)^2}
473: \left[2(2l+1)^2-\frac{1}{u}\right]
474: =\frac{\pi^{5/2}}{2 u^{5/2}} 
475: \sum_{l\ge 1} (-1)^{l+1}l^2 e^{-\frac{\pi^2l^2}{4u}}
476: \ee
477: 
478: \nin such that
479: 
480: \be
481: c=\pi^3\int_0^\infty\frac{du}{u^3}\sum_{l\ge 1}(-1)^{l+1}
482: l^2e^{-\frac{l^2\pi^2}{4u}}
483: =\frac{16}{\pi}\int_0^\infty dx x e^{-x}
484: \sum_{l\ge 1}\frac{(-1)^{l+1}}{l^2}
485: =\frac{4\pi}{3}
486: \ee
487: 
488: \nin which gives the well known
489: result for the one-loop Debye screening with massless
490: fermions \cite{lebellac}
491: for which higher order corrections can be found in \cite{blaizot}.
492: We note that we can commute the integration over the proper-time
493: and the summation over the Matsubara modes after doing the Poisson 
494: resummation.
495: 
496: Using again the Poisson resumation (\ref{poiss}),
497: we can give another form of $\Pi_n^{44}(\vec k)$ which 
498: splits the temperature independent part from the temperature dependent one.
499: A straightforward computation leads to
500: 
501: \be\label{pitot}
502: \Pi_n^{44}(\vec k)=\Pi_n^0(\vec k)+\Pi_n^T(\vec k)
503: \ee
504: 
505: \nin where $\Pi_n^0(\vec k)$ is the zero temperature part (\ref{piindep})
506: (with $k_4\to\omega_n$) 
507: and $\Pi^T_n(\vec k)$ the temperature dependent part
508: 
509: \bea\label{pidep}
510: &&\Pi^T_n(\vec k)=\frac{-\alpha}{2\pi}|eB|
511: \int_0^\infty du\int_{-1}^1 dv
512: e^{-\frac{k_\bot^2}{|eB|}\frac{\cosh\overline u-\cosh\overline uv}
513: {2\sinh\overline u}
514: -u[m^2+\frac{1-v^2}{4}(\omega_n^2+k_3^2)]}\\
515: &&\times\sum_{l\ge 1}(-1)^le^{-\frac{l^2}{4uT^2}}
516: \left[\left(k_\bot^2\frac{\cosh\overline uv-
517: v\coth\overline u\sinh\overline uv}{\sinh\overline u}+
518: k_3^2(1-v^2)\coth\overline u\right)\cos\pi nl(1-v)\right.\nonu
519: &&\left.~~~~~~~~~~~~~~~~~~~~~~~~
520: -\frac{\coth\overline u}{u}\left(\frac{l^2}{uT^2}\cos\pi nl(1-v)
521: -2\pi vnl\sin\pi nl(1-v)\right)\right]\nonumber
522: \eea
523: 
524: \nin where we took $\varepsilon\to 0$ since the temperature dependent
525: part is finite. We see again that after this Poisson resumation every term 
526: of the Masubara series gives a finite integration over the 
527: proper-time $u$.
528: 
529: 
530: 
531: 
532: \section{Other components and transversality}
533: 
534: 
535: 
536: 
537: We now compute the other components in a similar way and 
538: will give only the important steps.
539: We first give the diagonal components of the polarization tensor
540: which all need integrations by parts to lead to the good limit 
541: when $T\to 0$. We set
542: 
543: \be
544: \phi_l(u,v)=\frac{k_\bot^2}{2|eB|}\frac{\cosh\overline u-
545: \cosh\overline uv}{\sinh\overline u}
546: +u\left[m^2+W_l^2+\frac{1-v^2}{4}(\omega_n^2+k_3^2)\right]
547: \ee
548: 
549: Let us start with $\Pi^{33}_n(\vec k)$:
550: the same steps as the ones used for the computation of
551: $\Pi^{44}_n(\vec k)$ and the same 
552: integration by parts lead to
553: 
554: \bea\label{pol33}
555: &&\Pi^{33}_n(\vec k)=\frac{-\alpha T}{\sqrt\pi}|eB|
556: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
557: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}\nonu
558: &&\times\left[v\omega_nW_l\coth\overline u
559: +\frac{k_\bot^2}{2}\frac{\cosh\overline uv-
560: v\coth\overline u\sinh\overline uv}{\sinh\overline u}
561: +\omega_n^2\frac{1-v^2}{2}\coth\overline u\right]+Q^{33}(k)
562: \eea
563: 
564: The computation of $\Pi_n^{ii}(\vec k)$, $i=1,2$ (without summation 
565: over $i$) is slightly different. 
566: After the integration over the loop momentum $\vec p$,
567: the change of variable $s=u(1-v)/2$
568: and $\sigma=u(1+v)/2$ leads to
569: 
570: \bea
571: \Pi^{ii}_n(\vec k)&=&\frac{-\alpha T}{\sqrt\pi}|eB|
572: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
573: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}
574: \left[\frac{\cosh\overline uv}{\sinh\overline u}
575: \left(v\omega_nW_l-\frac{1}{2u}
576: -W_l^2-m^2\right)\right.\nonu
577: &-&
578: \left.\frac{2k_i^2-k_\bot^2}{2}\frac{\cosh\overline u-
579: \cosh\overline uv}{\sinh^3\overline u}
580: +(\omega_n^2+k_3^2)\frac{1-v^2}{4}\frac{\cosh\overline uv}
581: {\sinh\overline u}\right]
582: +Q^{ii}(k)
583: \eea
584: 
585: \nin Then we make the integration by parts over $u$
586: 
587: \bea
588: &&\int_\varepsilon^\infty du
589: e^{-\phi_l(u,v)}m^2\sqrt u\frac{\cosh\overline uv}{\sinh\overline u}
590: \longrightarrow\\
591: &&\int_\varepsilon^\infty du
592: e^{-\phi_l(u,v)}\sqrt u\frac{\cosh\overline uv}{\sinh\overline u}
593: \left[\frac{1}{2u}+|eB|(v\tanh\overline uv-\coth\overline u)
594: -\frac{d}{du}\left(\phi_l(u,v)-
595: um^2\right)\right]\nonumber
596: \eea
597: 
598: \nin followed by the intergation by parts over $v$
599: 
600: \bea
601: &&\int_{-1}^1dve^{-\phi_l(u,v)}\frac{\cosh\overline uv}{\sinh\overline u}
602: \left[\frac{1}{u}
603: +|eB|\left(v\tanh\overline uv-
604: \coth\overline u\right)\right]
605: \longrightarrow\nonu
606: &&\int_{-1}^1dve^{-\phi_l(u,v)}\frac{v\cosh\overline uv
607: -\coth\overline u \sinh\overline uv}{\sinh\overline u}
608: \left[\omega_nW_l-\frac{k_\bot^2}{2}
609: \frac{\sinh\overline uv}{\sinh\overline u}-
610: \frac{v}{2}(\omega_n^2+k_3^2)\right]
611: \eea
612: 
613: \nin where we again disregarded the surface terms. We finally obtain 
614: 
615: \bea\label{polii}
616: &&\Pi^{ii}_n(\vec k)=\frac{-\alpha T}{\sqrt\pi}|eB|
617: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
618: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}
619: \left[\coth\overline u\frac{\sinh\overline uv}{\sinh\overline u}
620: \omega_nW_l\right.\nonu 
621: &+&
622: \left.(k_\bot^2-k_i^2)\frac{\cosh\overline u-\cosh\overline uv}
623: {\sinh^3\overline u}
624: +\frac{\omega_n^2+k_3^2}{2}\frac{\cosh\overline uv-
625: v\coth\overline u\sinh\overline uv}{\sinh\overline u}\right]
626: +Q^{ii}(k)
627: \eea
628: 
629: Now let us go to the off-diagonal components of the polarization tensor.
630: What differs from the diagonal components is that we do not make any 
631: integration by parts and we obtain directly the final results with the 
632: expected limit when $T\to 0$:
633: 
634: \bea\label{polother}
635: \Pi^{34}_n(\vec k)&=&\frac{\alpha T}{\sqrt\pi}|eB|
636: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
637: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}
638: k_3\left[vW_l+\frac{1-v^2}{2}\omega_n\right]\coth\overline u
639: +Q^{34}(k)\nonu
640: \Pi^{12}_n(\vec k)&=&\frac{\alpha T}{\sqrt\pi}|eB|
641: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
642: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}
643: k_1k_2\frac{\cosh\overline u-\cosh\overline uv}{\sinh^3\overline u}
644: +Q^{12}(k)\nonu
645: \Pi^{i4}_n(\vec k)&=&\frac{\alpha T}{\sqrt\pi}|eB|
646: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
647: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}\\
648: &&~~~~~~~~~~~~~~~~\times
649: k_i\left[W_l\coth\overline u\frac{\sinh\overline uv}
650: {\sinh\overline u}
651: +\frac{\omega_n}{2}\frac{\cosh\overline uv-
652: v\coth\overline u\sinh\overline uv}{\sinh\overline u}\right]
653: +Q^{i4}(k)\nonu
654: \Pi^{i3}_n(\vec k)&=&\frac{\alpha T}{\sqrt\pi}|eB|
655: \int_\varepsilon^\infty du\sqrt u\int_{-1}^1 dv
656: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}
657: \frac{k_ik_3}{2}\frac{\cosh\overline uv-
658: v\coth\overline u\sinh\overline uv}{\sinh\overline u}
659: +Q^{i3}(k)\nonumber
660: \eea
661: 
662: \nin where $i=1,2$. It is easy to check that all the components of 
663: the polarization tensor
664: give the results found in \cite{tsai} when $T\to 0$. The contact terms 
665: are determined in the same way as $Q^{44}(k)$
666: and can be summarised as
667: 
668: \bea\label{ct}
669: Q^{\mu\nu}(k)&=&\frac{\alpha}{4\pi}
670: \int_\varepsilon^\infty \frac{du}{u}\int_{-1}^1 dv
671: e^{-um^2}(1-v^2)
672: \left(\delta^{\mu\nu}k^2-k^\mu k^\nu\right)\nonu
673: &=&\frac{\alpha}{3\pi}
674: \int_\varepsilon^\infty \frac{du}{u}
675: e^{-um^2}
676: \left(\delta^{\mu\nu}k^2-k^\mu k^\nu\right)
677: \eea
678: 
679: \nin as in \cite{tsai}.
680: 
681: It is important now to check the transversality of the polarization tensor,
682: which is not obvious since $\Pi^{\mu\nu}$ contains terms which are not
683: explicitely proportional to any external momentum component.
684: The contact term (\ref{ct}) is obviously transverse and
685: with the expressions (\ref{pol44}), (\ref{pol33}), (\ref{polii}) and
686: (\ref{polother}), we obtain 
687: 
688: \bea\label{transverse}
689: k_\nu\Pi_n^{\nu\mu}(\vec k)&=&
690: \omega_n\Pi^{4\mu}_n(\vec k)+k_3\Pi^{3\mu}_n(\vec k)+
691: k_i\Pi^{i\mu}_n(\vec k)\nonu
692: &=&\delta^{4\mu}\frac{\alpha T}{\sqrt\pi}|eB|
693: \int_\varepsilon^\infty du\sqrt u\coth\overline u\int_{-1}^1 dv
694: \sum_{l=-\infty}^\infty e^{-\phi_l(u,v)}\nonu
695: &&~~~~~~~~~~~~\times\left[\frac{\omega_n}{u}
696: +W_l\left(v(\omega_n^2+k_3^2)+
697: \frac{\sinh\overline uv}{\sinh\overline u}k_\bot^2-
698: 2\omega_nW_l\right)\right]\nonu
699: &=&2\delta^{4\mu}\frac{\alpha T}{\sqrt\pi}|eB|
700: \int_\varepsilon^\infty \frac{du}{\sqrt u}\coth\overline u\int_{-1}^1 dv
701: \frac{d}{dv}\left(\sum_{l=-\infty}^\infty W_le^{-\phi_l(u,v)}\right)\nonu
702: &=&\mbox{surface term}
703: \eea
704: 
705: \nin so that the polarization tensor is transverse, since the above sum
706: is zero up to surface terms which are normally omitted in this formalism.
707: 
708: 
709: 
710: \section{Strong field approximation}
711: 
712: 
713: 
714: We give here the strong field approximation of the 44-component of
715: the polarization tensor that can be used in a strong field study
716: of the magnetic catalysis. 
717: 
718: The strong field asymptotic form of the fermion propagator (\ref{freeTprop}) can be 
719: found by taking the limit $|eB|\to\infty$ in the integrand, neglecting 
720: the shrinking region of integration where the product $s|eB|$ goes to 
721: zero \cite{calucci}. We obtain then
722: 
723: \bea\label{propLLL}
724: \tilde S_l(\vec p)&\simeq& -i\int_0^\infty ds e^{-(\hat\omega_l^2+p_3^2+
725: \frac{p_\bot^2}{|eB|s}+m^2)}
726: \left(-\hat\omega_l\gamma^4-p^3\gamma^3+m\right)\left(1-i\gamma^1\gamma^2\right)\nonu
727: &=&ie^{-\frac{p_\bot^2}{|eB|}}\frac{\hat\omega_l\gamma^4+p^3\gamma^3-m}
728: {\hat\omega_l^2+p_3^2+m^2}\left(1-i\gamma^1\gamma^2\right)
729: \eea
730: 
731: \nin which is the well-known lowest Landau level approximation for the fermion 
732: propagator \cite{miransky} that we can obtain by truncating the expansion of
733: the propagator over the Landau levels to the dominant term \cite{chodos}. 
734: We will take the same limit $|eB|\to\infty$ in the expressions (\ref{piindep}) and 
735: (\ref{pidep}) to find the asymptotic form of $\Pi^{44}$. We note that the polarization
736: tensor does not contain divergences in the limit $|eB|\to\infty$, the magnetic field 
737: playing the role of a cut-off for the momenta. This is due to the exponential decrease
738: of the transverse degrees of freedom as can be seen in (\ref{propLLL}). Therefore we will not
739: consider any contact term here.
740: 
741: The temperature-independent part (\ref{piindep}) reads in the strong field limit 
742: ($\overline u\to \infty$) 
743: 
744: \bea
745: \Pi_n^0(\vec k)&\simeq& -\frac{\alpha |eB|}{4\pi}\int_0^\infty du\int_{-1}^1 dv
746: e^{-\frac{k_\bot^2}{2|eB|}-u[m^2+\frac{1-v^2}{4}(\omega_n^2+k_3^2)]}\nonu
747: &&\times\left\{k_\bot^2\left[(1-v)e^{-\overline u(1-v)}+(1+v)e^{-\overline u(1+v)}\right]
748: +k_3^2(1-v^2)\right\}\nonu
749: &=&-\frac{\alpha |eB|}{4\pi}e^{-\frac{k_\bot^2}{2|eB|}}\int_{-1}^1 dv\left[
750: \frac{(1-v^2)k_3^2}{m^2+\frac{1-v^2}{4}(\omega_n^2+k_3^2)}+
751: {\cal O}\left(\frac{k_\bot^2}{|eB|}\right)\right]
752: \eea
753: 
754: \nin The integration over $v$ of the dominant term leads then 
755: to the following expression that was already 
756: derived in \cite{calucci} where the authors started the computation 
757: with the propagator (\ref{propLLL}):
758: 
759: \be
760: \Pi_n^0(\vec k)\simeq-\frac{2\alpha}{\pi}|eB|\frac{k_3^2}{k_\|^2}
761: e^{-\frac{k_\bot^2}{2|eB|}}
762: \left[1-\frac{2m^2}{\sqrt{k_\|^2(4m^2+k_\|^2)}}
763: \ln\left(\frac{\sqrt{4m^2+k_\|^2}+\sqrt{k_\|^2}}
764: {\sqrt{4m^2+k_\|^2}-\sqrt{k_\|^2}}\right)\right]
765: \ee
766: 
767: \nin where $k_\|^2=\omega_n^2+k_3^2$.
768: 
769: For the temperature-dependent part (\ref{pidep}), the limit $\overline u\to \infty$ 
770: followed by the change of variable $u\to u/|eB|$ leads to the dominant term
771: 
772: \be
773: \Pi_n^T(\vec k)\simeq-\frac{\alpha}{2\pi}\frac{|eB|^2}{T^2}e^{-\frac{k_\bot^2}{2|eB|}}
774: \int_0^\infty \frac{du}{u^2}\int_{-1}^1 dv 
775: e^{-u\frac{\mu^2(v)}{4|eB|}}
776: \sum_{l\ge 1}(-1)^{l+1}l^2 e^{-\frac{l^2|eB|}{4uT^2}}
777: \cos\pi nl(1-v)
778: \ee
779: 
780: \nin where $\mu^2(v)=4m^2+(1-v^2)k_\|^2$. We recognize here the Bessel function $K_1$ since
781: 
782: \be
783: \int_0^\infty \frac{du}{u^2} e^{-au-\frac{b}{u}}=
784: \sqrt\frac{a}{b}\int_0^\infty due^{-(u+\frac{1}{u})\sqrt{ab}}
785: =2\sqrt\frac{a}{b}K_1(2\sqrt{ab})
786: \ee
787: 
788: \nin where we defined
789: 
790: \be
791: a=\frac{\mu^2(v)}{4|eB|}~~~~\mbox{and}~~~~b=\frac{l^2|eB|}{4T^2}
792: \ee
793: 
794: \nin The temperature-dependent part can finaly be written
795: 
796: \be
797: \Pi_n^T(\vec k)\simeq-\frac{2\alpha}{\pi}|eB|e^{-\frac{k_\bot^2}{2|eB|}}
798: \sum_{l\ge 1}(-1)^{l+1}\int_{-1}^1 dv
799: \frac{l\mu(v)}{2T}K_1\left(\frac{l\mu(v)}{2T}\right)\cos\pi nl(1-v)
800: \ee
801: 
802: \nin We will see in the conclusion that the previous sum and integrals can be 
803: evaluated for the computation of the Debye screening in the regime where 
804: $m<<T<<\sqrt{|eB|}$.
805: 
806: 
807: 
808: 
809: \section*{Conclusion: Debye screening in a magnetic field}
810: 
811: 
812: 
813: 
814: To conclude, we look in more details at the Debye screening obtained
815: in this computation. From equation (\ref{pidep}), 
816: we find for the Debye mass
817: 
818: \be\label{debyem}
819: M_{|eB|}^2(T)=-\lim_{\vec k^2\to 0}\Pi^{T}_0(\vec k)
820: =\frac{\alpha|eB|}{\pi T^2}\int_0^\infty \frac{du}{u^2}\coth\overline u
821: e^{-um^2}\sum_{l\ge 1}(-1)^{l+1}l^2e^{-\frac{l^2}{4uT^2}}
822: \ee
823: 
824: \nin The zero-magnetic field limit is
825: 
826: \be
827: M_{|eB|=0}^2(T)=
828: \frac{\alpha}{\pi T^2}\int_0^\infty \frac{du}{u^3}
829: e^{-um^2}\sum_{l\ge 1}(-1)^{l+1}l^2e^{-\frac{l^2}{4uT^2}}
830: \ee
831: 
832: \nin For given values of $|eB|$ and $m$, 
833: in figure \ref{thma} we compare
834: the ratios $M_{|eB|}^2/|eB|$ and $M_{|eB|=0}^2/|eB|$ as functions
835: of $T/\sqrt{|eB|}$, such that all the dimensionful quantities
836: are rescaled in units of the magnetic field ($[eB]=2$).
837: For high temperatures the curves converge towards the  
838: result (\ref{thermmB0}) (rescaled by $|eB|$)
839: since $m<<T$ and $\sqrt{|eB|}<<T$,
840: but for strong magnetic field
841: $T<<\sqrt{|eB|}$, a strong Debye sceening is generated compared to the
842: one without external field. As long as the temperature remains 
843: greater than the fermion mass, the Debye screening follows a plateau
844: when the temperature decreases.
845: We note that if we had $m=0$, the limit of the Debye screening
846: when $T\to 0$ would be (after the change of variable $u\to u/T^2$ in
847: (\ref{debyem}))
848: 
849: \bea\label{plateau}
850: \lim_{T\to 0}M_{|eB|,m=0}^2(T)&=&
851: \frac{\alpha|eB|}{\pi}\int_0^\infty \frac{du}{u^2}
852: \sum_{l\ge 1}(-1)^{l+1}l^2e^{-\frac{l^2}{4u}}\nonu
853: &=&\frac{4\alpha}{\pi}|eB|\int_0^\infty dx
854: \sum_{l\ge 1}(-1)^{l+1}le^{-xl}\nonu
855: &=&\frac{4\alpha}{\pi}|eB|\lim_{x\to 0}
856: \sum_{l\ge 1}(-1)^{l+1}e^{-xl}\nonu 
857: &=&\frac{4\alpha}{\pi}|eB|\lim_{x\to 0} 
858: \left(1-\frac{1}{1+e^{-x}}\right)\nonu
859: &=&\frac{2\alpha}{\pi}|eB|
860: \eea
861: 
862: \nin But when $T<m$ the fermion mass forces the screening
863: to vanish with the temperature, so that the value (\ref{plateau})
864: of the plateau is valid only if $m<<T<<\sqrt{|eB|}$. 
865: 
866: We note that a more unexpected behaviour has been observed in $QED_3$
867: at finite temperature 
868: in an external magnetic field \cite{afk2}: $M^2_{|eB|}$
869: first increases when the temperature decreases 
870: (in the region $T<<\sqrt{|eB|}$),
871: reaches a maximum when $T\simeq m$ and then decreases to 0 when $T\to 0$.
872: 
873: These behaviours of the Debye screening are a consequence of the dimensional 
874: reduction of the fermion dynamics in a strong magnetic field, as can be seen
875: with the propagator (\ref{propLLL}).
876: 
877: \begin{figure}
878: \epsfxsize=10cm
879: \epsfysize=8cm
880: \centerline{\epsfbox{thma.fig}}
881: \caption{$M_{|eB|}^2/|eB|$ and $M_{|eB|=0}^2/|eB|$ versus 
882: $T/\sqrt{|eB|}$ for $\alpha=.001$ and $m/\sqrt{|eB|}=.1$}
883: \label{thma}
884: \end{figure}
885: 
886: To conclude, we note again the consistency between the necessity to 
887: have a massive fermion to obtain the good zero temperature limits
888: (as long as $|eB|>0$) and the occurence of 
889: the magnetic catalysis which generates dynamically this mass.
890: 
891: 
892: 
893: 
894: 
895: \section*{Acknowledgements}
896: 
897: This work has been done within the
898: TMR project 'Finite temperature phase transitions in particle Physics',
899: EU contract number: FMRX-CT97-0122, and I would like to thank 
900: K.Farakos and G.Koutsoumbas for guiding me through the subject of
901: magnetic catalysis. Finally, I would like to thank G.Tiktopoulos 
902: for helpful advice.
903: 
904: 
905: \begin{thebibliography}{99}
906: 
907: \bibitem{electroweak}
908: J.Ambjorn, P.Olesen, hep-ph/9304220, in the
909: {\it Proceedings of the 4th Hellenic School on Elementary Particle Physics},
910: Corfu, September 1992;
911: M.Giovannini, M.E.Shaposhnikov, Phys.Rev.{\bf D57} (1998),
912: 2186; K.Kajantie, M.Laine, J.Peisa, K.Rummukainen, M.E.Shaposhnikov,
913: Nuc.Phys. {\bf B544} (1999), 357; V.Skalozub, M.Bordag, hep-ph/9807510,
914: hep-ph/9904333.
915: 
916: \bibitem{superconductors} 
917: K.Farakos, N.E.Mavromatos, Int.Journ.Mod.Phys.{\bf B12}
918: (1998), 809.
919: 
920: \bibitem{afk2} J.Alexandre, K.Farakos, G.Koutsoumbas, hep-th/0010211,
921: to appear in Phys.Rev.D.
922: 
923: \bibitem{afk1} J.Alexandre, K.Farakos, G.Koutsoumbas, 
924: Phys.Rev.{\bf D62} (2000), 105017.
925: 
926: \bibitem{miransky} V.P.Gusynin, V.A.Miransky, I.A.Shovkovy, Nucl.Phys.
927: {\bf B462} (1996), 249; V.P.Gusynin, V.A.Miransky, I.A.Shovkovy,
928:  Nucl.Phys.{\bf B563} (1999), 361. 
929: 
930: \bibitem{gusynin} V.P.Gusynin, I.A.Shovkovy, Phys.Rev.{\bf D56}
931: (1997), 5251.
932: 
933: \bibitem{lee} D.-S.Lee, C.N.Leung, Y.J.Ng, Phys.Rev.{\bf D55} 
934: (1997), 6504.
935: 
936: \bibitem{schwinger} J.Schwinger, Phys.Rev.{\bf 82} (1951), 664.
937: 
938: \bibitem{tsai} W.Tsai, Phys.Rev.{\bf D10} (1974), 2699.
939: 
940: \bibitem{schubert} C.Schubert, Nucl.Phys. {\bf B585} (2000), 407
941: and 429.
942: 
943: \bibitem{dittrich} W.Dittrich, Phys.Rev.{\bf D10} (1979), 2385.
944: 
945: \bibitem{gies} H.Gies, Phys.Rev.{\bf D60} (1999), 105002.
946: 
947: \bibitem{zuber} C.Itzyckson, J.-B.Zuber, {\it Quantum Field Theory},
948: Mc Graw-Hill (1980), section 7.1.1.
949: 
950: \bibitem{lebellac} M.Le Bellac, {\it Thermal Field Theory}, 
951: Cambridge (1996), section 6.3.1.
952: 
953: \bibitem{blaizot} J.-P.Blaizot, E.Iancu, R.R.Parwani, Phys.Rev. {\bf D52}
954: (1995), 2543.
955: 
956: \bibitem{calucci} G.Calucci, R.Ragazzon, J.Phys. {\bf A27} (1994), 2161.
957: 
958: \bibitem{chodos} A.Chodos, K.Everding, D.Owen, Phys.Rev.{\bf D42} (1990), 2881.
959: 
960: \end{thebibliography}
961: 
962: 
963: 
964: \end{document}
965: 
966: 
967: 
968: 
969: 
970: 
971: \end{document}
972: 
973: 
974: