1: % ****** Start of file apssamp.tex ******

2: %

3: %   This file is part of the APS files in the REVTeX 4 distribution.

4: %   Version 4.0 of REVTeX, August 2001

5: %

6: %   Copyright (c) 2001 The American Physical Society.

7: %

8: %   See the REVTeX 4 README file for restrictions and more information.

9: %

10: % TeX'ing this file requires that you have AMS-LaTeX 2.0 installed

11: % as well as the rest of the prerequisites for REVTeX 4.0

12: %

13: % See the REVTeX 4 README file

14: % It also requires running BibTeX. The commands are as follows:

15: %

16: %  1)  latex apssamp.tex

17: %  2)  bibtex apssamp

18: %  3)  latex apssamp.tex

19: %  4)  latex apssamp.tex

20: %

21: % \documentclass[twocolumn,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

22: \documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}

23: 

24: % Some other (several out of many) possibilities

25: %\documentclass[preprint,aps]{revtex4}

26: %\documentclass[preprint,aps,draft]{revtex4}

27: %\documentclass[prb]{revtex4}% Physical Review B

28: 

29: \usepackage{graphicx}% Include figure files

30: \usepackage{dcolumn}% Align table columns on decimal point

31: \usepackage{bm}% bold math

32: 

33: \def\VEC#1{\mbox{\boldmath $#1$}}

34: 

35: %\nofiles

36: 

37: \begin{document}

38: 

39: \preprint{APS/123-QED}

40: 

41: \title{Propagation of Electromagnetic Waves

42: in Resistive Pair Plasma and Causal Relativistic 

43: Magnetohydrodynamics}% Force line breaks with \\

44: 

45: \author{Shinji Koide}

46: %% \altaffiliation[Also at ]{Physics Department, XYZ University.}%Lines break automatically or can be forced with \\

47: %% \author{Second Author}%

48: %% \email{Second.Author@institution.edu}

49: \affiliation{%

50: Faculty of Science, Kumamoto University,

51: 2-39-1, Kurokami, Kumamoto 860-8555, Japan 

52: }%

53: 

54: %% \author{Charlie Author}

55: %% \homepage{http://www.Second.institution.edu/~Charlie.Author}

56: %%\affiliation{

57: %%Second institution and/or address\\

58: %%This line break forced% with \\

59: %%}%

60: 

61: \date{\today}% It is always \today, today,

62:              %  but any date may be explicitly specified

63: 

64: \begin{abstract}

65: We investigate the propagation of electromagnetic

66: waves in resistive e$^\pm$ pair plasmas using a one-fluid

67: theory derived from the relativistic two-fluid

68: equations. When the resistivity normalized by the electron/positron inertia

69: variable exceeds a critical value, the dispersion relation for

70: electromagnetic waves shows that the group velocity 

71: is larger than the light speed in vacuum.

72: However, in such a case, it also is found that the plasma parameter is less than unity: 

73: that is, the electron--positron pair medium no longer can be treated as plasma.

74: Thus the simple two-fluid approximation is invalid. This confirms that 

75: superluminal propagation of electromagnetic wave is forbidden

76: in a plasma ----- a conclusion

77: consistent with the relativistic principle of causality.

78: As an alternative, we propose a new set of equations

79: for ``causal relativistic magnetohydrodynamics", which both have non-zero

80: resistivity and yet are consistent with the causality principle.

81: 

82: %% with non-zero resistivity.

83: \end{abstract}

84: 

85: \pacs{52.27.Ny, 52.30.Cv, 52.30.Ex, 52.35.Hr}% PACS, the Physics and Astronomy

86:                              % Classification Scheme.

87: %\keywords{Suggested keywords}%Use showkeys class option if keyword

88:                               %display desired

89: \maketitle

90: 

91: \section{Introduction}

92: 

93: Relativistic magnetohydrodynamic (RMHD) numerical simulations

94: have been performed by a number of groups recently \cite{koide98,koide99,koide00,koide02,

95: koide03,koide04,koide06,mizuno04,komissarov04,komissarov07,gammie03,mckinney04,mckinney06}.

96: These numerical simulations revealed many important, interesting features

97: of relativistic plasmas, especially around rotating black holes in

98: active galactic nuclei (AGNs), microquasars, and gamma-ray bursts.

99: Regarding energy extraction from a rotating black hole,

100: the magnetohydrodynamic (MHD) Penrose process has been confirmed \cite{koide02,koide03}, 

101: and long-term simulations of relativistic jet formation

102: around a rotating black hole have been performed \cite{mckinney06}.

103: All of these RMHD simulations

104: were restricted by the ideal MHD condition, where

105: electric resistivity is zero. 

106: In spite of recent significant advancements in ideal RMHD simulations,

107: one with finite resistivity (resistive RMHD) 

108: have not been performed seriously except in a few cases 

109: (e.g. \cite{watanabe06,komissarov07b}).

110: This is reasonable because there has been concern that the inclusion of finite 

111: resistivity in the RMHD equations destroys their causality.

112: In fact, the group velocity of electromagnetic waves 

113: derived mathematically from the resistive RMHD equations 

114: is larger than the light speed in vacuum.

115: %% Here, the group velocity provides the velocity of communication

116: %% with electromagnetic waves and then this indicates 

117: This raises the possibility of superluminal communication, 

118: which is contradictory to the relativistic principle of causality.

119: The main purpose of this paper is to clarify and rectify this problem.

120: 

121: To fix this problem, we must reconsider the resistive RMHD

122: equations. Such a task was first performed

123: by \cite{ardavan76} using the Vlasov--Boltzmann equation

124: for a pulsar magnetosphere. It yielded a relativistic

125: version of the generalized Ohm's law and a new

126: condition for the validity of the MHD approximation

127: for a pulsar magnetosphere 

128: (where the Lorentz factor is much larger than unity).

129: A more generalized treatment, which included 

130: annihilation of electrons and positrons, radiation, Compton scattering,

131: and pair photoproduction was formulated by \cite{blackman93}

132: and \cite{gedalin96}.

133: Reconsideration of ideal MHD in a neutral cold plasma based on two-fluid

134: approximation was presented by \cite{melatos96}, who investigated the

135: conditions under which the MHD approximation

136: breaks down. For investigation of black hole magnetospheres,

137: \cite{khanna98} formulated the general relativistic version

138: of the two-fluid approximation in the Kerr metric.

139: An even more generalized version in a time-varying space-time 

140: was derived by \cite{meier04} from

141: the general relativistic Vlasov--Boltzmann equation.

142: 

143: In this present paper, we derive the one-fluid equations of an electron--positron

144: (pair) plasma based on the two-fluid equations with a new definition

145: of variable averaging for the two fluids (section \ref{sec2}).

146: %% We derive one-fluid equations

147: %% with two-fluid approximation to fix dispersion relation for 

148: %% electromagnetic waves in resistive plasmas in section \ref{sec2}. 

149: In section \ref{sec3}, we derive the dispersion relation for electromagnetic

150: waves in uniform, unmagnetized and magnetized pair plasmas. 

151: We then examine the situation where the group

152: velocity of electromagnetic waves in the resistive plasma is 

153: larger than the light speed in vacuum and  

154: show that this condition cannot be realized in a plasma

155: whose plasma parameter is larger than unity.

156: In section \ref{sec4}, we propose a simple set of resistive

157: RMHD equations, which are consistent with the principle of causality.

158: In section \ref{sec5}, we discuss phenomena with respect to

159: the superluminal propagation of wave packets ----- phenomena that cannot

160: be avoided when the RMHD equations are acausal.

161: Finally, our summary and discussion are presented in section \ref{sec6}.

162: %% Furthermore, we derive the

163: %% characteristic damping length of the electromagnetic waves and show

164: %% that the damping length is much shorter than the length of the wave packet,

165: %% which is the element of the communication with the electromagnetic waves.

166: %% In section 4, we discuss the possibility of the superluminal 

167: %% communication with the damping electromagnetic wave in the resistive

168: %% plasma. We also discuss the causality problems of the superluminal

169: %% communication, and provide (confirm) the validity of the resistive RMHD.

170: 

171: \section{Relativistic two--fluid model of pair plasma 

172: %% and resistive RMHD approximation

173: \label{sec2}}

174: 

175: To provide a solid base for resistive RMHD, we begin with

176: a relativistic two-fluid model of a pair plasma

177: in the Minkowski space-time $(x^0,x^1,x^2,x^3)=(t,x,y,z)$, where

178: the line element is given by

179: $ds^2=-(dx^0)^2+(dx^1)^2+(dx^2)^2+(dx^3)^2=\eta_{\mu\nu}dx^\mu dx^\nu$.

180: Throughout this paper (except for one paragraph in section \ref{sec3}), 

181: we use units in which the light speed, the dielectric constant,

182: and the magnetic permeability in vacuum all are unity: 

183: $c=1$, $\epsilon_0=1$, $\mu_0 = 1$.

184: The relativistic equations of the electron and positron

185: fluids are given as follows (e.g., \cite{misner70,weinberg72}):

186: %

187: \begin{eqnarray}

188: \partial_\mu (n_\pm U_\pm^\mu) &=& 0 , \label{4formnum} \\

189: \partial_\mu (h_\pm U_\pm^\mu U_\pm^\nu)  &=&

190: -\partial^\nu p_\pm \pm e n_\pm \eta^{\nu\sigma} U_\pm^\mu F_{\sigma\mu}

191: \pm R^\nu , \\

192: \partial_\mu \hspace{0.3em} ^*F^{\mu\nu} &=& 0 , \\

193: %% \partial_\mu F_{\nu\lambda} + \partial_\nu F_{\lambda\mu} 

194: %% + \partial_\lambda F_{\mu\nu} &=& 0 , \\

195: \partial_\mu F^{\mu\nu} & = & J^\nu , \label{4formmaxwel}

196: \end{eqnarray}

197: where a variable with subscript, plus (+) or minus (--),

198: is that of the positron and electron fluid, respectively,

199: $n_\pm$ is the proper particle number density, $U_\pm^\mu$ is the four-velocity,

200: $e$ is the electric charge

201: of positron, $p_\pm$ is the proper pressure, $h_\pm$ is the relativistic

202: enthalpy density, 

203: $F_{\mu\nu}$ is the electromagnetic field tensor, 

204: $\hspace{0.3em} ^*F^{\mu\nu}$ is the dual tensor density of $F_{\mu\nu}$,

205: $R^\mu$ is the frictional four-force density between the electron and

206: positron fluids, and $J^\mu$ is the four-current density. 

207: %% and $J^\mu = e(n_+ U_+^\mu - n_- U_-^\mu)$ is the four-current density. 

208: We often will write a set of the spacial components of the four-vector using a bold italic

209: font, e.g., $\VEC{U}_\pm = (U^1_\pm,U^2_\pm,U^3_\pm)$, $\VEC{J} = (J^1,J^2,J^3)$.

210: %% We also write $\gamma_\pm = U^0_\pm$ and $\rho_{\rm e} = J^0$.

211: Here we assume that the electron/positron fluids are heated only

212: by Ohmic heating and neglect pair creation and annihilation.

213: We also neglect radiation and quantum effects.

214: 

215: We further define the Lorentz factor $\gamma_\pm = U^0_\pm$, the three-velocity

216: $V^i_\pm=U^i_\pm/\gamma_\pm$, the electric field $E_i=F^{0i}$, the magnetic flux

217: density $B_i=\frac{1}{2} \sum_{jk} \epsilon_{ijk} F^{jk}$

218: ($\epsilon_{ijk}$ is the Levi--Civita tensor), and the electric charge density

219: $\rho_{\rm e} = J^0$. 

220: %% and current density $J_i = J^i$. 

221: Here, the alphabetic index ($i,j,k$) runs from 1 to 3.

222: Using the above relativistic equations (\ref{4formnum})--(\ref{4formmaxwel}), 

223: we obtain the vector form of the relativistic two-fluid equations,

224: \begin{eqnarray}

225: \frac{\partial}{\partial t} (\gamma_\pm n_\pm) +

226: \nabla \cdot (n_\pm \VEC{U}_\pm) &=& 0 , \label{twofluidnum} \\

227: %%\gamma_\pm h_\pm \left [ \frac{\partial}{\partial t} (\gamma_\pm \VEC{v}_\pm)

228: %%+(\VEC{v}_\pm \cdot \nabla ) (\gamma_\pm \VEC{v}_\pm) \right ] 

229: \frac{\partial}{\partial t} (h_\pm \VEC{U}_\pm)

230: + \nabla \cdot (h_\pm \VEC{U}_\pm \VEC{U}_\pm)

231: \hspace{-0.2cm} &=& \hspace{-0.2cm} 

232: -\nabla p_\pm \pm e \gamma_\pm n_\pm (\VEC{E} + \VEC{V}_\pm \times \VEC{B})

233: \pm \VEC{R}, \\

234: %%& & -\gamma m n_\pm \frac{d \hat{h}_\pm}{dt} \VEC{U}_\pm  , \\

235: \frac{\partial}{\partial t} (\gamma_\pm^2 h_\pm-p_\pm)

236: + \nabla \cdot (\gamma_\pm h_\pm \VEC{U}_\pm ) 

237: & = & \pm e n_\pm \VEC{U} \cdot \VEC{E} \pm R^0  , \\

238: %%% & = & \pm e n_\pm {\bf U} \cdot {\bf E} - k (\gamma_\pm - \gamma_\mp) , \\

239: %% \frac{\partial}{\partial t} \left ( \frac{p_\pm}{n_\pm^\Gamma} \right )

240: %% + (\VEC{v}_\pm \cdot \nabla) \left ( \frac{p_\pm}{n_\pm^\Gamma} \right ) &=& 0 , \\

241: \nabla \cdot \VEC{E} = \rho_{\rm e} &=& e(\gamma_+ n_+ - \gamma_- n_-)  , \\

242: \nabla \cdot \VEC{B} &=& 0 , \\

243: \frac{\partial \VEC{B}}{\partial t} &=& - \nabla \times \VEC{E} , \\

244: \frac{\partial \VEC{E}}{\partial t} + \VEC{J} &=& \nabla \times \VEC{B} .

245: \label{twofluidmaxwel}

246: \end{eqnarray}

247: %% where 

248: %% $d/dt$ is the convective derivative for 

249: %% each fluid.

250: The frictional four-force density between electrons and positrons is

251: (equation (\ref{frc4frc2}) in Appendix \ref{appena}),

252: \begin{equation}

253: R^\mu=-\frac{m\nu_{\rm ee}}{n}

254: (n_- \gamma_-^\prime n_+ U_+^\mu - n_+ \gamma_+^\prime n_- U_-^\mu)

255: + \frac{n_+ U_+^\mu + n_- U_-^\mu}{n_+ \gamma_+^\prime + n_- \gamma_-^\prime} 

256:  R^{0\prime} ,

257: \label{friction4frc}

258: %% R^\mu = \frac{m\nu_{\rm ee}}{ne^2} \gamma' \left [

259: %% J^\mu - \frac{U^\nu J_\nu}{-\gamma'^2} (1+\Theta) U^\mu \right ]

260: \end{equation} 

261: where $\nu_{\rm ee}$ is the electron--positron Coulomb collision frequency,

262: $m$ is the mass of an electron/positron particle,

263: the variables with primes are physical quantity observed in the

264: center-of-mass frame of the two fluids, and $R^{0\prime}$ is the energy gain rate of the positron

265: fluid due to the friction with electron fluid in the center-of-mass frame.

266: Note that the variables observed in the center-of-mass frame are 

267: proper variables (see Appendix \ref{appena}).

268: If we assume that relative velocity of the positron and electron fluids is

269: %%, $v_{\rm r}$, is

270: much smaller than thermal velocity of the fluids, the collision

271: frequency $\nu_{\rm ee}$ is proportional to the relative velocity

272: %% , $v_{\rm r}$, 

273: of the two fluids and its coefficient depends only on 

274: temperature and density of the two fluids.

275: On the other hand, if the relative velocity of the positron and electron fluids is 

276: relativistic, then the coefficient also depends on their relative velocity.

277: %% Here, $\gamma'$ is the Lorentz factor observed in the center--of--mass frame.

278: Note also that when the relative velocity of the electron and positron fluids

279: is nonrelativistic, $\gamma'_\pm \rightarrow 1$, while in the case of relativistic

280: relative velocity, $\gamma'_\pm > 1$.

281: Through this paper, we usually assume that the relative velocity of the two fluids

282: %% , $v_{\rm r}$, 

283: is smaller than the sound velocity of the plasma (which is  nonrelativistic).

284: %% except for the cases noted explicitly.

285: 

286: To derive the one-fluid equations of a pair plasma, we define average and difference

287: variables as follows:

288: \begin{eqnarray}

289: n &=& \frac{n_+ + n_-}{2}, \label{avenum} \\

290: \rho &=& 2mn, \label{avemas} \\

291: U^\mu &=& \frac{n_+ U_+^\mu + n_- U_-^\mu}{2n}  ,

292: \label{ave4vel} \\

293: J^\mu &=& e(n_+ U_+^\mu - n_- U_-^\mu) ,

294: \label{ave4cur}  \\

295: % h &=& \frac{n}{\gamma} 

296: % \left ( \frac{\gamma_+ h_+}{n_+} + \frac{\gamma_- h_-}{n_-} \right ), \\

297: \tilde{h}&=& n^2 \left ( \frac{h_+}{n_+^2} + \frac{h_-}{n_-^2} \right ), \\

298: % \Delta h&=&\frac{n}{\gamma} 

299: % \left ( \frac{\gamma_+ h_+}{n_+} - \frac{\gamma_- h_-}{n_-}\right ), \\

300: \Delta \tilde{h}&=&n^2 \left ( \frac{h_+}{n_+^2} - \frac{h_-}{n_-^2}\right ). 

301: \end{eqnarray}

302: From the relativistic two-fluid model of the 

303: pair plasma (\ref{twofluidnum})--(\ref{friction4frc}), 

304: we then can obtain the one-fluid equations of the pair plasma,

305: %% (see also Appendix \ref{appenb}),

306: \begin{eqnarray}

307: && \frac{\partial}{\partial t} (\gamma \rho) +

308: \nabla \cdot (\rho \VEC{U}) = 0  , \label{rmhdmass} \\

309: && \frac{\partial}{\partial t} \left [ \tilde{h} \left ( \gamma \VEC{U} + 

310: \frac{1}{(2ne)^2} \rho_{\rm e} \VEC{J} \right ) +

311: \frac{\Delta \tilde{h}}{2en} \left ( \gamma \VEC{J} + \rho_{\rm e} \VEC{U} \right ) 

312: \right ]

313: + \nabla \cdot \left [ \tilde{h} \left ( 

314: \VEC{UU} + \frac{1}{(2en)^2} \VEC{JJ} \right ) 

315: + \frac{\Delta \tilde{h}}{2en} (\VEC{UJ} + \VEC{JU}) \right ]  \nonumber \\

316: && = 

317: -\nabla p + \rho_{\rm e} \VEC{E} + \VEC{J} \times \VEC{B} , \label{rmhdmomentum} \\

318: && \VEC{E} + \VEC{V} \times \VEC{B} + \frac{1}{2en} \nabla (p_- - p_+)

319:  -  \eta \frac{\gamma'}{\gamma} \left [\VEC{J} - \frac{1 + \Theta}{\gamma^{\prime 2}} 

320: (\gamma \rho_{\rm e} - \VEC{J} \cdot \VEC{U} ) \VEC{U} \right ]  \nonumber \\

321: && = \frac{1}{4ne^2\gamma} \left [ \frac{\partial}{\partial t} \left ( 

322: \frac{\tilde{h}}{n} ( \gamma \VEC{J} + \rho_{\rm e} \VEC{U} ) 

323: + \Delta \tilde{h} \left \{ \gamma \VEC{U} 

324: + \frac{1}{(2ne)^2} \rho_{\rm e} \VEC{J} \right \} \right )

325: \right . \nonumber \\

326: && +  \left . 

327: \nabla \cdot \left \{ \frac{\tilde{h}}{n} (\VEC{UJ} + \VEC{JU}) 

328: + 2 n^2 e \Delta \tilde{h} \left ( \VEC{UU} + \frac{1}{(2en)^2} \VEC{JJ} \right )\right \}

329:  \right ] \label{rmhdohm}, \\

330: && \frac{\partial}{\partial t} \left [ \tilde{h} \left ( \gamma^2 + \frac{\rho_{\rm e}^2}{(2ne)^2} \right ) 

331: + \Delta \tilde{h} \frac{\gamma}{ne}  \rho_{\rm e} -p \right ] 

332: + \nabla \cdot \left [ \tilde{h} \left ( 

333: \gamma \VEC{U} + \frac{1}{(2ne)^2} \rho_{\rm e} \VEC{J} \right ) +

334: \frac{\Delta \tilde{h}}{2ne} \left ( \gamma \VEC{J} + \rho_{\rm e} \VEC{U} \right ) 

335: \right ]   \nonumber = \VEC{J} \cdot \VEC{E} , \nonumber \\

336: && 

337: \end{eqnarray}

338: \begin{eqnarray}

339: \nabla \cdot \VEC{E} &=& \rho_{\rm e} , \\

340: \nabla \cdot \VEC{B} &=& 0 , \\

341: \frac{\partial \VEC{B}}{\partial t} &=& - \nabla \times \VEC{E} , \\

342: \frac{\partial \VEC{E}}{\partial t} + \VEC{J} &=& \nabla \times \VEC{B}  ,

343: \label{rmhdampere}

344: \end{eqnarray}

345: where $\eta = m \nu_{\rm ee}/(n e^2)$ is the electric resistivity and

346: $\Theta$ is the equipartition factor of the thermal energy due to

347: the friction between the electron and positron fluids

348: given in Appendix \ref{appena}.

349: %%where $\tilde{m}= h/n$ and $\eta = m\nu_{\rm ee}/(ne^2)$.

350: %%Here we define one--fluid variables as $\rho=\rho_+ + \rho_-$,

351: %%$\VEC{U}=(\VEC{U}_+ + \VEC{U}_-)/2$, 

352: %%$\VEC{J}=e n_+ \VEC{U}_+ - e n_- \VEC{U}_-$, $h=h_+ + h_-$,

353: %%$\rho_{\rm e} = e\gamma_+ n_+ - e \gamma_- n_-$.

354: %%We assume the charge neutrality, $n = n_+ = n_-$.

355: 

356: Equation (\ref{rmhdohm}) corresponds to the generalized Ohm's law.

357: The classical formulation of the resistivity $\eta$ 

358: of a non-magnetized nonrelativistic plasma (due to 

359: Coulomb collision) is used, and the formulation for an electron--proton

360: plasma is confirmed as an appropriate expression for resistivity

361: of weakly magnetized nonrelativistic plasma in laboratory experiments

362: (e.g., in a ``Tokamak" thermonuclear fusion device \cite{bellan06}).

363: Note that the Hall effect disappears in a pair plasma.

364: 

365: 

366: \section{Dispersion relation for electromagnetic waves in resistive plasma

367: \label{sec3}}

368: 

369: In this section, we derive the dispersion relation for electromagnetic waves in a

370: pair plasmas using a linear analysis of equations

371: (\ref{rmhdmass})--(\ref{rmhdampere}). 

372: First, we assume that the background plasma is at rest, uniform, and non-magnetized:

373: $\rho=\rho_0$ ($n=n_0$), $\VEC{V}=\VEC{0}$, $p=p_0$ ($p_+=p_-=p_0/2$), 

374: $\tilde{h}=h_0$, $\VEC{B}=\VEC{E}=\VEC{0}$.

375: Perturbations due to the electromagnetic waves are so small that

376: the plasma motion is non-relativistic. 

377: Then we have the following linearized equations with respect to the perturbations,

378: $\rho_1=\rho - \rho_0$ ($n_1 = n - n_0$), $\VEC{V}_1=\VEC{V}$, $p_1=p-p_0$,

379: $\VEC{B}_1=\VEC{B}$, $\VEC{E}_1=\VEC{E}$, $\VEC{J}_1=\VEC{J}$,

380: $\rho_{\rm e 1} = \rho_{\rm e}$,

381: $h_1=\tilde{h}-h_0$,

382: \begin{eqnarray}

383: \frac{\partial}{\partial t} \rho_1 +

384: \nabla \cdot (\rho_0 \VEC{V}_1) &=& 0 ,\\

385: h_0 \frac{\partial \VEC{V}_1}{\partial t} &=& - \nabla p_1 ,\\

386: \VEC{E}_1 - \eta \VEC{J}_1 &=& \kappa \frac{\partial \VEC{J}_1}{\partial t} , 

387: \label{linearohm} \\

388: \frac{\partial}{\partial t} (h_1 - p_1) + \nabla \cdot (h \VEC{V}_1) &=& 0 , \\

389: \nabla \cdot \VEC{E}_1 &=& \rho_{\rm e 1} , \\

390: \nabla \cdot \VEC{B}_1 &=& 0 , \\

391: \frac{\partial \VEC{B}_1}{\partial t} &=& - \nabla \times \VEC{E_1} , \\

392: \frac{\partial \VEC{E}_1}{\partial t} + \VEC{J}_1 &=& \nabla \times \VEC{B}_1 ,

393: \end{eqnarray}

394: where $\kappa = h_0/(2n_0e)^2$, and we assume $p_+^1 \approx p_-^1$.

395: These linearized equations do not depend on the equipartition fraction 

396: of frictionally thermalized energy, $\theta$ ($0 \le \theta \le 1$).

397: When we consider transverse modes of the linearized equations, 

398: $\VEC{k} \cdot \VEC{V_1} = 0$, the dispersion relation for 

399: the electromagnetic waves can be written as

400: \begin{equation}

401: ( k^2 - \omega^2) \left ( 1 - \frac{i\omega}{2\nu_{\rm ee}'} \right )

402: = i\frac{\omega}{\eta}  ,

403: \label{thedisprel0}

404: \end{equation}

405: where $\VEC{k}$ is the wave number vector, $\omega$ the frequency of 

406: the electromagnetic wave, and

407: $\nu_{\rm ee}' = \eta/(2\kappa) = (mn_0/h_0) \nu_{\rm ee} $.

408: Here we also obtain $\rho_1 = 0$, $p_1 = 0$, $\rho_{\rm e 1} = 0$.

409: 

410: When we normalize the variables as $\hat{\omega} = \omega/ (2 \nu_{\rm ee}')$

411: and $\hat{k} = k/ (2 \nu_{\rm ee}')$, the dispersion relation (\ref{thedisprel0}) 

412: becomes

413: \begin{equation}

414: H(\hat{\omega}^2 - \hat{k}^2) (1 - i \hat{\omega}) + i \hat{\omega} = 0,

415: \label{dispcomlex}

416: \end{equation}

417: where $H = 2 \eta \nu_{\rm ee}' = \eta^2/\kappa$.

418: Note that the parameter $H$ is related to 

419: the Coulomb collision frequency $\nu_{\rm ee}=\eta n_0 e^2/m$ and

420: the electron plasma frequency $\omega_{\rm pe} = (n_0e^2/m)^{1/2}$ as

421: $H=2(mn_0/h_0) (\nu_{\rm ee}/\omega_{\rm pe})^2$.

422: %% where $H = (\nu_{\rm ee}'/\omega_{\rm pe})^2$.

423: Setting $\hat{\omega} = \Omega - i \gamma$ ($\Omega \ge 0, \gamma \in {\bf R}$),

424: we obtain the dispersion relation with respect to the real frequency $\Omega$

425: and the damping rate $\gamma$,

426: \begin{equation}

427: \gamma^3 - \gamma^2 + \frac{1}{4} \left ( 1 + \hat{k}^2 + \frac{1}{H} \right )\gamma

428: - \frac{1}{8H} = 0   ,

429: \label{disprealgamma}

430: \end{equation}

431: and 

432: \begin{equation}

433: \Omega^2 ( 4 \Omega^2 -C )^2 - \frac{1}{27} (C^3 + F^2) = 0   ,

434: \label{disprealomega}

435: \end{equation}

436: where $C = 3 (\hat{k}^2 + 1/H)-1$, $F = 9 [\hat{k}^2 - 1/(2H)]+1$.

437: In this section and Appendix \ref{appenb}, we use $\gamma$ to denote

438: the damping rate.

439: We also have the relation between $\Omega$ and $\gamma$, 

440: \begin{equation}

441: \Omega^2 = 3\gamma^2 - 2\gamma + \hat{k}^2 + \frac{1}{H} .

442: \label{omega2gamma0}

443: \end{equation}

444: The dispersion relations with various $H$ are shown in Fig. \ref{disprels8}.

445: Note that the determinant of the cubic equation (\ref{disprealgamma}) 

446: with respect to $\gamma$

447: is $D_\gamma = - (C^3 + F^2)/(9 \times 36^2)$. If $\gamma$ has 

448: three different real solutions, $D_\gamma > 0$, i.e.,

449: $\Omega^2 (4 \Omega^2 -C)^2 = (1/27) (C^3 + F^2) < 0$, then $\Omega$ has

450: no real solution. Therefore, we have to consider range of the single

451: $\gamma$ solution, $D_\gamma \le 0$.

452: This range is given by $\hat{k} > \hat{k}_{\rm crit}$, where

453: the critical wave number $\hat{k}_{\rm crit}$ is defined by $\Omega = 0$ 

454: if a solution of $\Omega=0$ exists and $\hat{k}_{\rm crit} = 0$

455: if there is no solution (see Fig. \ref{disprels8}).

456: %% When $H \gg 1$, $\hat{k}_{\rm crit}$ is given by $1/2H$ approximately.

457: These results clearly show that 

458: the group velocity $v_{\rm g} = \partial \Omega/\partial \hat{k}$ is larger 

459: than one when $H \agt 3$; that is, superluminal wave packet propagation is possible 

460: (see Appendix \ref{appenc}).

461: When $H \ge 3.5$, there are points of $\Omega = 0$ at 

462: $\hat{k} = \hat{k}_{\rm crit} >0$.

463: Figure \ref{calkcri} shows the value of $\hat{k}_{\rm crit}$ for each $H$ case.

464: The group velocity $\partial \Omega/\partial \hat{k}$ is infinity 

465: at $\hat{k} \rightarrow \hat{k}_{\rm crit} + 0$.

466: %% Then the superluminal propagation is realized most clearly when

467: %% $k > k_{\rm crit}$, $k \sim k_{\rm crit}$, $H > 3.5$.

468: On the other hand, in the cases of $H=1$ and $H=2$ 

469: (see Fig. \ref{disprels8}(a) and (b)), 

470: the derivative of $\Omega$ with respect to $\hat{k}$ increases monotonically, and 

471: $\partial \Omega/\partial k$ approaches unity when $\hat{k}$ becomes infinity 

472: ($\lim_{\hat{k} \longrightarrow \infty} \partial \Omega/\partial \hat{k} = 1$);

473: that is, the gradient remains less than unity as long as $\hat{k}$

474: remains finite. 

475: Furthermore, we prove that $\partial \Omega/\partial \hat{k} <1$ 

476: when $H<1.5$ (see Appendix \ref{appenb}). We find that there is no possibility of 

477: superluminal propagation of electromagnetic wave in the case of $H<2$,

478: while $\partial \Omega/\partial k$ is larger than unity in a certain range

479: of $\hat{k}$ when $H \ge 3$. (A detailed investigation produces a more

480: strict condition on superluminal propagation of $H \ge 2.3$.)

481: 

482: \begin{figure}

483: \includegraphics[width=0.85\textwidth]{fig1.eps}% Here is how to import EPS art

484: \caption{

485: Dispersion relation for electromagnetic waves in the

486: resistive pair plasma for various $H$.

487: The dotted line shows the dispersion relation for electromagnetic waves

488: in vacuum.

489: \label{disprels8}

490: }

491: \end{figure}

492: 

493: \begin{figure}

494: \includegraphics[width=0.5\textwidth]{fig2.eps}% Here is how to import EPS art

495: \caption{

496: Dependence of $\hat{k}_{\rm crit}$ on $H$. At $\hat{k} = \hat{k}_{\rm crit}+0$, 

497: $\partial \Omega/\partial \hat{k}$ becomes infinity for each $H$.

498: \label{calkcri}

499: }

500: \end{figure}

501: 

502: We now show that matter composed of electrons and positrons 

503: with $H \ge 1$ cannot be treated as a plasma.

504: This means that superluminal propagation of electromagnetic waves

505: is not permitted when the medium is a plasma; and, when it is not,

506: the medium must be treated in a different manner. 

507: Note that in this paragraph only, we shall use the SI unit system.

508: The plasma parameter is given by

509: \begin{equation}

510: N_{\rm p} = n \lambda_{\rm D}^3

511: = \sqrt{\frac{(\epsilon_0 T)^3}{n e^6}}   ,

512: \end{equation}

513: where $T$ is the temperature of the electron/positron fluids and

514: $\lambda_{\rm D}$ is the Debye length, 

515: $\lambda_{\rm D} = \epsilon_0 T/(n e^2)$ \cite{bellan06}.

516: For a plasma, $N_{\rm p}$ is (much) larger than unity

517: because charged particles are bound to

518: each other when $N_{\rm p} < 1$.

519: The frequency of electron--positron Coulomb collisions can be written

520: as,

521: \begin{equation}

522: \nu_{\rm ee} = \frac{n e^4 \ln \Lambda}{6 \sqrt{3} \pi

523: \epsilon_0^2 \sqrt{m} T^{3/2}}

524: =\frac{n e^4}{\epsilon_0^2 \sqrt{m} T^{3/2}} \ln \Lambda',

525: \end{equation}

526: where 

527: $\ln \Lambda$ is the Coulomb logarithm and $\ln \Lambda' = \ln \Lambda

528: /(6 \sqrt{3} \pi) \sim 1/3$ \cite{bellan06}. Here we used $\ln \Lambda \sim 10$. 

529: Then we find that 

530: $H= 2(mn/h)(\nu_{\rm ee}/\omega_{\rm pe})^2< 2 (\nu_{\rm ee}/\omega_{\rm pe})^2 

531: =2(\ln \Lambda)^2 n e^6/(\epsilon_0 T)^3$.

532: Finally, we get the relation between $H$ and $N_{\rm p}$,

533: \begin{equation}

534: H N_{\rm p}^2 < 2 (\ln \Lambda')^2 \sim \frac{2}{9}.

535: \end{equation}

536: Therefore, when we consider a plasma (i.e., $N_{\rm p} > 1$), 

537: we find that $H<(\ln \Lambda')^2/N_{\rm p}^2 \alt 2/9 < 1$.

538: This clearly shows superluminal propagation of electromagnetic 

539: wave is not permitted  in a true plasma (usually, $N_{\rm p} \gg 1$).

540: 

541: In the above discussion we used the rough approximation $\ln \Lambda \sim 10$.

542: If $\ln \Lambda$ were greater than 100 separately with other variables, 

543: $H$ would become larger than

544: 3 and then superluminal communication would become possible.

545: However, this situation can never be realized because there is strict relation

546: between $\Lambda$ and $N_{\rm p}$ as $\Lambda = 6 \pi N_{\rm p}$

547: (see \cite{bellan06}, page 24). So, if $\ln \Lambda$ becomes larger,

548: then $N_{\rm p}$ becomes much larger and $H$ decreases to a value much less

549: than unity.

550: 

551: Next, we discuss briefly the dispersion relation for electromagnetic waves 

552: in a uniformly magnetized pair plasma. We assume the background

553: plasma is the same as that of the previous unmagnetized case except

554: for a uniform magnetic field, $\VEC{B} = \VEC{B}_0 \ne \VEC{0}$.

555: Using the same procedure employed in the previous unmagnetized 

556: plasma case, we obtain the linearized equations,

557: \begin{eqnarray}

558: \VEC{E}_1 + \frac{i}{\omega h_0} (\VEC{J}_1 \times \VEC{B}_0)

559: \times \VEC{B}_0 &=& (\eta - i \kappa \omega) \VEC{J}_1, \label{magplalinele1} \\

560: \omega^2 \VEC{E}_1 + i \omega \VEC{J}_1 = k^2 \VEC{E}_1, \label{magplalincur1}

561: \end{eqnarray}

562: where we assume $\VEC{k} \cdot \VEC{V}_1 = 0$ 

563: and $\VEC{k} \cdot \VEC{E}_1 = 0$ to investigate transverse modes.

564: When we separate the perturbations of the electric field and current density into 

565: two components parallel and perpendicular to the background magnetic field

566: $\VEC{B}_0$,

567: \begin{eqnarray}

568: \VEC{E}_1 = \VEC{E}_\parallel + \VEC{E}_\perp, &

569: \VEC{E}_\parallel \parallel \VEC{B}_0, &

570: \VEC{E}_\perp \perp \VEC{B}_0, \\

571: \VEC{J}_1 = \VEC{J}_\parallel + \VEC{J}_\perp, &

572: \VEC{J}_\parallel \parallel \VEC{B}_0, & 

573: \VEC{J}_\perp \perp \VEC{B}_0, \\

574: \end{eqnarray}

575: equations (\ref{magplalinele1}) and (\ref{magplalincur1}) yield

576: \begin{eqnarray}

577: \VEC{E}_\perp - \frac{iB_0^2}{\omega h_0} \VEC{J}_\perp 

578: &=& (\eta - i \kappa \omega) \VEC{J}_\perp, \label{magplalinele2} \\

579: (k^2 - \omega^2) \VEC{E}_\perp &=& i\omega \VEC{J}_\perp, \\

580: \VEC{E}_\parallel &=& (\eta - i \kappa \omega) \VEC{J}_\parallel, \\

581: (k^2 - \omega^2) \VEC{E}_\parallel &=& i\omega \VEC{J}_\parallel. \\

582: \end{eqnarray}

583: Finally, we obtain the two dispersion relations,

584: \begin{eqnarray}

585: (k^2 - \omega^2) \left (\eta -i \kappa \omega + \frac{iB_0^2}{\omega h_0} 

586: \right ) &=& i\omega,

587: \label{dispmagpla1} \\

588: (k^2 - \omega^2) (\eta - i \kappa \omega) &=& i\omega .

589: \label{dispmagpla2}

590: \end{eqnarray}

591: Equations (\ref{dispmagpla1}) and (\ref{dispmagpla2}) are not

592: satisfied simultaneously when $B_0 \ne 0$. Equation (\ref{dispmagpla2})

593: is the same as that of the unmagnetized pair plasma case. Therefore,

594: we shall investigate the dispersion relation (\ref{dispmagpla1}).

595: 

596: When we set $\hat{\omega} = \kappa \omega/\eta$, $\hat{k}=\kappa k/\eta$,

597: we have,

598: \begin{equation}

599: H(\hat{\omega}^2 - \hat{k}^2)(i\hat{\omega} + \hat{\omega}^2 -\alpha) = \omega^2,

600: \end{equation}

601: where $H=\eta^2/\kappa$ and $\alpha = B_0^2 \kappa/(h_0 \eta^2) = u_{\rm A}^2/H$

602: ($u_{\rm A} \equiv B_0/\sqrt{h_0}$ is the Alfven four-velocity).

603: Setting $\hat{\omega} = \Omega -i \gamma$ ($\Omega \ge 0, \gamma \in {\bf R}$),

604: we obtain the dispersion relation for electromagnetic waves

605: in a magnetized pair plasma,

606: \begin{eqnarray}

607: H[(\Omega^2 -\gamma^2 -\hat{k}^2) (\Omega^2 - \gamma^2 - \alpha + \gamma)

608: + 2 \Omega^2 \gamma (1-2 \gamma)] &=& \Omega^2 -\gamma^2, \\

609: H[(\Omega^2 -\gamma^2 -\hat{k}^2) (1 - 2 \gamma) - 2 \gamma

610: (\Omega^2 - \gamma^2 -\alpha + \gamma) ] &=& -2 \gamma, 

611: \end{eqnarray}

612: where we assume $\Omega \ne 0$.

613: Figure \ref{dispmag} shows the dispersion relation for $\Omega$ 

614: in the case  $H=1$ and $\alpha = 0.1, 1$.

615: The figure clearly shows the group velocity of electromagnetic

616: waves in a magnetized pair plasma is less than the light speed in vacuum

617: when $H=1$. A detailed investigation shows that this is true when $H<1$ 

618: as in the unmagnetized pair plasma case.

619: However, when $H > 4$, the group velocity is larger than the speed of light

620: for some ranges of $\hat{k}$. 

621: 

622: \begin{figure}

623: \includegraphics[width=\textwidth]{fig3.eps}% Here is how to import EPS art

624: \caption{The dispersion relation of electromagnetic waves in uniform,

625: magnetized pair plasma (thick solid lines) and in unmagnetized plasma

626: for comparison ($\alpha = 0$; thin solid lines).

627: (a) Sub-relativistically strong magnetic field case, $H=1$, $\alpha = 0.1$.

628: (b) Relativistically strong magnetic field case, $H=1$, $\alpha = 1$.

629: \label{dispmag}

630: }

631: \end{figure}

632: 

633: \section{Causal Resistive RMHD Equations

634: \label{sec4}}

635: 

636: In the above discussion, we derived the one-fluid equations 

637: (\ref{rmhdmass})--(\ref{rmhdampere}) from the two-fluid ones.

638: The one-fluid equations confirmed that the superluminal

639: propagation is forbidden in a two-component medium that has a

640: plasma whose plasma

641: parameter greater than unity. That is, the one-fluid equations

642: of a pair plasma (\ref{rmhdmass})--(\ref{rmhdampere}) are causal.

643: When we neglect the first term of the right hand side in Ohm's law

644: equation (\ref{rmhdohm}), which comes from the inertial effect of the

645: positron and electron, the term, $-i\omega/(2\nu'_{\rm ee})$, on the

646: left hand side of the dispersion relation (\ref{thedisprel0}) drops out.

647: In this case, the group velocity becomes $v_{\rm g} = \partial \omega/

648: \partial k= 2k(4k^2-\eta^{-2})^{-1/2} > 1$, which means the group 

649: velocity is greater than the speed of 

650: light (superluminal). As shown in Appendix \ref{appenc}, when 

651: the group velocity is larger than the speed of light, superluminal 

652: communication would become possible, allowing us to develop

653: a device that could send information into the past. 

654: However, such a device would destroy

655: the causality of time--ordered events and, therefore, should not be possible. 

656: This means that, in order to preserve causality, we cannot neglect the inertial

657: term of the electron and positron in Ohm's law (\ref{rmhdohm}).

658: Recently, several groups performed simulations of resistive RMHD

659: including Ohm's law without the electron/positron inertia effect

660: \cite{watanabe06,komissarov07b}. As shown in the above results, unfortunately,

661: all of these calculations are acausal. Here, we propose a set of causal 

662: resistive RMHD equations in a simple form.

663: 

664: For simplicity, assuming that $|\Delta \tilde{h}| \ll \tilde{h}$

665: and $p_+ \approx p_-$, we obtain the following equations,

666: \begin{eqnarray}

667: \frac{\partial}{\partial t} (\gamma \rho) + \nabla \cdot (\rho \VEC{U}) &=& 0

668: \label{causalrmhdmass},\\

669: \frac{\partial}{\partial t} \left [ \tilde{h} \left ( 

670: \gamma  \VEC{U} + \frac{\rho_{\rm e}}{(2ne)^2} \VEC{J} \right ) \right ]

671: + \nabla \cdot \left [ \tilde{h} \left ( \VEC{UU} + \frac{1}{(2ne)^2} \VEC{JJ} \right ) 

672: \right ] &=& 

673: - \nabla p + \rho_{\rm e} \VEC{E} + \VEC{J} \times \VEC{B},

674: \label{causalrmhdmom}\\

675: \frac{\partial}{\partial t} \left [ \tilde{h} \left ( 

676: \gamma^2 + \frac{\rho_{\rm e}^2}{(2ne)^2} \right )  - p \right ] 

677: + \nabla \cdot \left [ \tilde{h} \left (

678: \gamma \VEC{U} + \frac{\rho_{\rm e}}{(2ne)^2} \VEC{J} \right ) \right ] 

679: &=& \VEC{J} \cdot \VEC{E},

680: \label{causalrmhdenr} \\

681: \VEC{E} + \VEC{V} \times \VEC{B} - \frac{\eta}{\gamma} [\VEC{J} - \gamma^2

682: (\rho_{\rm e} - \VEC{V} \cdot \VEC{J}) (1 + \Theta ) \VEC{V}] && \nonumber \\

683: = \frac{1}{4ne\gamma} \left [ 

684: \frac{\partial}{\partial t} \left ( 

685: \frac{\tilde{h}}{ne} (\gamma \VEC{J} + \rho_{\rm e} \VEC{U} ) \right ) 

686: + \nabla \cdot \left \{ 

687: \frac{\tilde{h}}{ne} (\VEC{UJ} + \VEC{JU})

688: \right \} \right ], && \label{causalrmhdohm} 

689: \end{eqnarray}

690: \begin{eqnarray}

691: \nabla \cdot \VEC{E} &=& \rho_{\rm e}, 

692: \label{causalrmhddive} \\

693: \nabla \cdot \VEC{B} &=& 0 ,\\

694: \frac{\partial \VEC{B}}{\partial t} &=& - \nabla \times \VEC{E},\\

695: \frac{\partial \VEC{E}}{\partial t} + \VEC{J} &=&  \nabla \times \VEC{B},

696: \label{causalrmhdampere}

697: \end{eqnarray}

698: where 

699: $\Theta = 2\theta (m/e)^2 (Q^2+W) \left /\left [\rho^2 - \left (mQ/e \right )^2 \right ] 

700: \right .$

701: (see Appendix \ref{appena}).

702: Here we have to assume $\gamma' \approx 1$ and $M=U_\nu U^\nu=-1$ 

703: in Appendix \ref{appena}, which means that the relative 

704: velocity of the electron fluid and positron fluid is nonrelativistic.

705: This condition also preserves $\gamma = 1/(1 - V^2)^{1/2}$.

706: In a pair plasma, we use $\theta=1$.

707: 

708: A covariant form for these one-component fluid equations

709: (\ref{causalrmhdmass})--(\ref{causalrmhdampere}) is as follows:

710: \begin{eqnarray}

711: \partial_\nu (\rho U^\nu) &=& 0, \\

712: \partial_\nu \left [ \tilde{h} \left ( U^\nu U^\mu + 

713: \frac{1}{(2ne)^2} J^\nu J^\mu \right ) \right ] &=&

714: -\partial^\mu p + J^\nu F_\nu^\mu, \\

715: U^\nu F_\nu^\mu - \eta [J^\mu + (U^\nu J_\nu) (1+\Theta) U^\mu]

716: &=& \frac{1}{4ne^2} \left [ \partial_\nu \left \{ \frac{\tilde{h}}{n}

717: \left ( U^\nu J^\mu + J^\nu U^\mu \right ) \right \} \right ], \\

718: \partial_\nu F^{\nu\mu} &=& J^\mu, \\

719: \partial_\nu \hspace{0.3em} ^* F^{\nu\mu} &=& 0 .

720: \end{eqnarray}

721: 

722: The difference between equations (\ref{causalrmhdmass})--(\ref{causalrmhdampere}) 

723: and the RMHD equations used by the 

724: previous acausal resistive RMHD simulations \cite{watanabe06,komissarov07b} is

725: mainly in Ohm's law, as expected and suggested by other articles 

726: \cite{ardavan76,blackman93,gedalin96,melatos96,khanna98,meier04}.

727: The linear analysis of the electromagnetic wave in a pair plasma

728: shows that the inertia effect of the electron and

729: positron is essential in preserving causality.

730: Here, the electron/positron inertia term of Ohm's law is the right

731: hand side of equation (\ref{causalrmhdohm}).

732: If we neglect the change of $\gamma h /n$,

733: Ohm's law simplifies to

734: \begin{equation}

735: \VEC{E} + \VEC{V} \times \VEC{B} - \frac{\eta}{\gamma} [\VEC{J} - \gamma^2

736: (\rho_{\rm e} - \VEC{V} \cdot \VEC{J} ) (1+\Theta) \VEC{V}]

737: = \kappa \left [ 

738: \frac{\partial}{\partial t} (\VEC{J} + \rho_{\rm e} \VEC{V})  + \nabla \cdot \left

739: (\VEC{VJ} + \VEC{JV}

740: \right ) \right ], \label{causalrmhdohmm}

741: \end{equation}

742: where $\kappa = \tilde{h}/(2en)^2$.

743: When $H=\eta^2/\kappa < 1$, that is $\eta < \sqrt{\kappa}$, equations

744: (\ref{causalrmhdmass})--(\ref{causalrmhdenr}), 

745: (\ref{causalrmhddive})--(\ref{causalrmhdampere}), (\ref{causalrmhdohmm}) 

746: are causal, and thus we 

747: call equations (\ref{causalrmhdmass})--(\ref{causalrmhdenr}),

748: (\ref{causalrmhddive})--(\ref{causalrmhdampere}),(\ref{causalrmhdohmm}) 

749: with $\eta < \sqrt{\kappa}$ the ``causal

750: resistive RMHD" equations. 

751: Among the causal resistive RMHD equations, equation (\ref{causalrmhdohmm})

752: is most important; we call it the ``causal Ohm's law".

753: When we set $\Theta =0$ , equation (\ref{causalrmhdohmm}) reduces to

754: the simpler Ohm's law,

755: \begin{equation}

756: \VEC{E} + \VEC{V} \times \VEC{B} - \frac{\eta}{\gamma} \left [\VEC{J} - \gamma^2

757: (\rho_{\rm e} - \VEC{V} \cdot \VEC{J} ) \VEC{V} \right ]

758: = \kappa \left [ 

759: \frac{\partial}{\partial t} (\VEC{J} + \rho_{\rm e} \VEC{V})  + \nabla \cdot \left

760: (\VEC{VJ} + \VEC{JV}

761: \right ) \right ], \label{causalrmhdohmmm}

762: \end{equation}

763: which is quite similar to the generalized Ohm's law derived by 

764: \cite{ardavan76} and \cite{meier04}, but not identical.

765: 

766: \section{Expected Phenomena related to superluminal wave packet

767: \label{sec5}}

768: 

769: In this section, we discuss phenomena related to superluminal propagation 

770: of electromagnetic wave packets that appeared in RMHD simulations 

771: that use an acausal Ohm's law with $H=\eta^2/\kappa \agt 3$.

772: First, we show that it is difficult to detect the superluminal propagation of a 

773: electromagnetic wave packet in an unmagnetized plasma at rest.

774: For simplicity, we use the acausal Ohm's law

775: with $\kappa = 0$ ($H \rightarrow \infty$). 

776: This is just the case of the previous studies with

777: resistive RMHD \cite{watanabe06,komissarov07b}. 

778: In this case, the dispersion relation becomes that of the telegraphic

779: equation,

780: \begin{equation}

781: \omega^2 + \frac{i}{\eta} \omega - k^2 =0 .

782: \end{equation}

783: The group velocity of the dispersion relation 

784: $v_{\rm g} = \partial \omega/\partial k = k/[k^2 -(2\eta)^{-2}]^{1/2} > 1$ 

785: is always greater than the light speed in vacuum. The damping time of the wave is

786: $\tau_{\rm damp} = 1/(-\Im(\omega)) = 2\eta$. The diffusion time of the wave 

787: packet is calculated by

788: \[

789: \tau_{\rm diff} = \frac{\sigma^2}{|D|} 

790: = \frac{\sigma^2}{|\partial^2 \omega/\partial k^2|}

791: =\sigma^2 (2 \eta)^2 [k^2 - (2\eta)^{-2}]^{3/2},

792: \]

793: where $\sigma$ is the width of the wave packet (see Appendix \ref{appenc}, 

794: equation (\ref{decaytime})).

795: The life time of the wave packet is estimated by

796: \begin{equation}

797: \tau = \left ( 

798: \frac{1}{\tau_{\rm damp}} + \frac{1}{\tau_{\rm diff}} \right )^{-1}

799: =\frac{\sigma^2 (2 \eta)^2 [k^2 - (2\eta)^{-2}]^{3/2}}

800: {1 + \sigma^2 (2\eta) [k^2 - (2\eta)^{-2}]^{3/2}}.

801: \end{equation}

802: The characteristic propagation length of the wave packet is

803: \[

804: l = v_{\rm g} \tau  

805: =\frac{k \sigma^2 (2 \eta)^2 [k^2 - (2\eta)^{-2}]}

806: {1 + \sigma^2 (2\eta) [k^2 - (2\eta)^{-2}]^{3/2}}.

807: \]

808: Using $N=k\sigma$ and $\chi = 2 \eta k$, we have

809: \begin{equation}

810: \frac{l}{\sigma} = \frac{N\chi^2 (\chi^2 -1 )}{\chi^2 + N^2 (\chi^2 -1)^{3/2}}.

811: \end{equation}

812: Note that the limitation $l/\sigma \rightarrow \infty$ 

813: ($\chi \rightarrow \infty$) means the wave packet propagates to a very long

814: distance compared to the scale of the wave packet itself in a highly

815: resistive plasma, where the situation is almost the same in a vacuum.

816: However, to detect the superluminal propagation of the wave packet, we have

817: to detect the difference between the propagation length of the wave packet 

818: and that of the light

819: in vacuum, $\Delta l = l - \tau$. The difference is estimated as

820: \begin{equation}

821: \frac{\Delta l}{\sigma} 

822: %% &=& \frac{(v_{\rm g} -1) \tau}{\sigma}

823: =\frac{l}{\sigma} \left ( 1 - \frac{1}{v_{\rm g}} \right )

824: = \frac{N(\chi^2 -1)}{[N^2 (\chi^2-1)^{3/2} + \chi^2] \left (1+\sqrt{1- \chi^{-2}}

825: \right )} \le \frac{2^{2/3}N}{3N^{4/3}+ 2^{2/3}}  .

826: %% < \frac{2^{2/3}}{3} \frac{1}{N^{1/3}} .

827: \label{impdetsupl}

828: \end{equation}

829: Because $N \gg 1$, equation (\ref{impdetsupl}) shows $\Delta l \ll \sigma$.

830: This means detection of the superluminal propagation of the wave

831: packet is difficult in the rest background plasma with a detector of 

832: ordinary sensitivity.

833: 

834: When we consider a moving plasma with relativistic speed, 

835: propagation of the superluminal wave packet changes drastically.

836: Here we consider the wave packet propagating along the $x$ direction

837: of a frame $(t,x)$ in a

838: uniform, unmagnetized plasma at rest (see Fig. \ref{funyphenom}(a)).

839: We assume that the wave packet propagates with the group velocity $v_{\rm g} >1$

840: and damps with the damping rate $\gamma_{\rm dmp}$.

841: Next, we consider a new frame $(t',x')$ moving with velocity $v_0 > 1/v_{\rm g}$

842: relative to the frame $(t,x)$,

843: where the $t'$-axis and $x'$-axis in the space-time $(t,x)$ 

844: are drawn as shown in Fig. \ref{funyphenom}(a). The world line of the wave

845: packet is located between the $x'$-axis and $x$-axis. When we ride on the 

846: new frame $(t',x')$, we see from time inversion arguments 

847: that the wave packet propagates from 

848: the right to the left as shown in Fig. \ref{funyphenom}(b).

849: Furthermore, the wave packet grows at the rate 

850: $\gamma'_{\rm grw} = \gamma_{\rm dmp} \sqrt{1-v_0^2}/(v_0 v_{\rm g}-1)$.

851: Here the points A, B, and C with respect to the wave packet are identified

852: with those at A', B', and C', respectively.

853: %

854: % To confirm this peculiar phenomenon, we calculated the growth rate

855: % of the electromagnetic wave in the uniform

856: % plasma moving relativistically. We found the positive growth

857: % rate of the wave, when $v_0$ is large enough. 

858: This suggests that we have to use the causal RMHD equations 

859: (\ref{causalrmhdmass})--(\ref{causalrmhdenr}),

860: (\ref{causalrmhddive})--(\ref{causalrmhdampere}), and

861: (\ref{causalrmhdohmm}) 

862: to avoid such a strange instability of the wave packet, at least 

863: in relativistic plasma flow, because wave

864: packets propagating in such a flow will grow explosively.

865: Relativistic flow exists around the black hole horizon in

866: the Kerr space-time, so 

867: {\it artificial radiation of electromagnetic wave packets 

868: from the horizon will occur in acausal RMHD calculations}.

869: On the other hand, the same acausal RMHD equations with $\kappa = 0$ 

870: cause no problem

871: for a non-relativistically moving plasma.

872: 

873: \begin{figure}

874: \includegraphics[width=\textwidth]{fig4.eps}

875: \caption{Propagation of electromagnetic wave packet in

876: the uniform, unmagnetized plasma.

877: (a) Case of rest plasma. (b) Case of relativistic flow of plasma.

878: \label{funyphenom}

879: }

880: \end{figure}

881: 

882: 

883: \section{Concluding Remarks and Discussion

884: \label{sec6}}

885: 

886: We have derived the dispersion relation for electromagnetic waves

887: in a resistive pair plasma based on the relativistic two-fluid model, 

888: and have shown that the group velocity

889: of electromagnetic waves in a resistive plasma

890: is smaller than the light speed within the plasma

891: condition ($N_{\rm p} > 1$). 

892: This shows that superluminal communication is impossible

893: in a resistive plasma,

894: thus confirming the causal nature of signals in plasmas.

895: %% The impossibility of the superluminal communication

896: %% with plasma confirms the causality.

897: Furthermore, the causality condition, $H = \eta^2/\kappa< 2(3N_{\rm p})^{-2}<1$,

898: provides an upper limit for electric resistivity in resistive RMHD: 

899: \begin{equation}

900: \eta < \sqrt{\kappa} = \sqrt{\frac{h_0}{2mc^2n_0}} \frac{1}{\epsilon_0\omega_{\rm pe}}

901: = 0.2 \left ( \frac{n_{\rm 0}}{10^{20} {\rm m}^{-3}} \right )^{-1/2} 

902: \left ( \frac{h_0}{2mc^2n_0} \right )^{1/2}

903: \, [\Omega \, {\rm m}].

904: \end{equation}

905: For simplicity, we assumed that the relative velocity of the positron

906: and electron fluids is much smaller than their internal thermal velocities.

907: This is consistent with the assumption of linear analysis.

908: In general, however, this assumption is not valid, especially

909: for relativistic plasma around black holes.

910: To deal with plasmas where the relative velocity of the electron 

911: and positron fluids is relativistic, we have to return to the relativistic 

912: Vlasov--Boltzmann

913: equation with collisional terms \cite{ardavan76,blackman93,gedalin96,meier04}

914: to obtain the resistive term in the causal Ohm's law.

915: In such a case, resistivity depends on current density.

916: 

917: We emphasize that the inertia effect is important for preserving 

918: causality, i.e., to forbid the superluminal propagation of 

919: electromagnetic waves in a resistive plasma. 

920: If we neglect the inertia term in the 

921: generalized Ohm's law (the first term of the right hand side of

922: equation (\ref{rmhdohm})), 

923: %% the right hand side of equation (\ref{linearohm})

924: %% becomes zero. Then the second term in the second parentheses

925: %% of the left hand side of equation (\ref{thedisprel0}),

926: %% $-i\omega/(2\nu_{\rm ee}^\prime)$, is dropped. 

927: then the resistive RMHD equations give a group velocity of electromagnetic

928: waves, $v_{\rm g} = \partial \omega/\partial k = 2k/(4k^2 - \eta^{-2})^{1/2}$,

929: that is greater than the light speed. This shows that the inertia 

930: effects of electrons and positrons should be considered to preserve causality.

931: We therefore proposed a set of causal

932: resistive RMHD equations in section \ref{sec4}.

933: Numerical techniques for simulating ``causal resistive

934: RMHD" flow should be developed quickly and be applied to 

935: astrophysical calculations ----- e.g., energy extraction from a rotating 

936: black hole by magnetic reconnection \cite{koide08}. 

937: To perform causal resistive RMHD simulations

938: of a black hole magnetosphere,

939: we are to use the general relativistic MHD equations along 

940: with the causal Ohm's law.

941: 

942: When we consider matter with a plasma parameter less

943: than one, we cannot use the simple two-fluid approximation,

944: because particles in the system are electrically bound to each other.

945: Metal, like iron, is an example for such matter.

946: To treat such a relativistic system, we ultimately must use 

947: relativistic quantum mechanics. However, we have no framework for that at present;

948: nevertheless it is an interesting and challenging field for future work.

949: In such an unknown framework, the group velocity of electromagnetic

950: waves in any medium should not be larger than the light speed 

951: (even if the wave damps quickly) to preserve causality. 

952: %% which provides restriction of the

953: %% forthcoming framework. 

954: %% With respect to the restriction,

955: %% Appendix \ref{appenc} shows that we can develop superluminal

956: %% communication tool using electromagnetic waves when

957: %% group velocity is greater than the light speed.

958: On the other hand, within the classical framework where we neglect 

959: quantum effects, we would show that the group velocity is always equal to or 

960: smaller than the speed of light when we treat the system properly. Here we cannot

961: use the method of smoothing the electromagnetic field, as is done in

962: traditional particle simulations, because the simple

963: smoothing destroys causality. Therefore, numerical calculations

964: may be more difficult than traditional plasma 

965: particle simulations.

966: %% In any case, in the medium with the plasma parameter less than one, 

967: It is interesting and important to investigate, 

968: in a medium with a small plasma parameter,

969: which effects (quantum or classical effects)

970: are more important for keeping the group velocity of electromagnetic wave

971: equal to or less than the speed of light. 

972: 

973: 

974: In this paper, we considered only a pair plasma; 

975: %%show impossibility of superluminal communication with the pair plasma.

976: we did not treat an electron--proton plasma. However, the similar conclusions

977: also should be drawn for the latter (at lease when the plasma is unmagnetized), 

978: because the linearized terms of resistive

979: RMHD in a pair plasma and in an electron--proton plasma

980: are expected to be similar.

981: %

982: It also is important to note the differences between

983: RMHD of a pair plasma and in an electron--proton plasma.

984: These come from the inequality between the mass ratios of 

985: the electron--positron and electron--proton.

986: With respect to equations (\ref{causalrmhdmass})--(\ref{causalrmhdenr}),

987: (\ref{causalrmhddive})--(\ref{causalrmhdampere}), (\ref{causalrmhdohmm}), 

988: it is expected that these equations are similar except for 

989: appearance of the second term in the brackets

990: on the left hand side of equation (\ref{causalrmhdmom}), $\VEC{JJ}/(2ne)^2$, and

991: the term 

992: in the brackets on the left hand side of equation (\ref{causalrmhdohmm}), 

993: $\gamma^2 (\rho_{\rm e} - \VEC{V} \cdot \VEC{J}) \Theta \VEC{V}$.

994: In the electron--proton plasma, the electron inertia term with $\VEC{JJ}$

995: is negligible compared to the proton inertia term with $\VEC{UU}$,

996: and $\Theta$ vanishes because of poor 

997: energy exchange between the electron and proton fluids. However, in

998: the pair plasma, $\Theta$ is not negligible.

999: %% appearance of the first

1000: %% term of the right hand side of equation (\ref{causalrmhdmom}),

1001: %% and the second term in the first large bracket of the right hand side

1002: %% of equation (\ref{causalrmhdohmm}). 

1003: Furthermore, the Hall effect disappears

1004: in Ohm's law (\ref{causalrmhdohmm}) in the pair plasma case.

1005: Note that all of the terms are nonlinear and that

1006: the coefficient $\kappa$ of the inertia term

1007: in the causal Ohm's law for the electron--proton plasma is

1008: much smaller than that of the pair plasma

1009: (by the ratio of the electron and proton masses).

1010: It is believed that an accretion disk in a black hole magnetosphere of an AGN

1011: will consist of an electron--proton plasma and a corona around 

1012: the disk and a relativistic

1013: jet from AGN consist of pair plasma \cite{wardle98}.

1014: % equivalence of the mass between the electron and positron, 

1015: % while the masses of the electron and proton have quite difference.

1016: Comparison between phenomena in relativistic pair plasmas

1017: and electron--proton plasmas is both interesting

1018: and necessary for understanding the physics of black hole

1019: magnetospheres where a relativistic jet may be produced. 

1020: %% from the accretion disk near the black hole.

1021: 

1022: %% To analyze the superluminal communication mathematically,

1023: %% the group velocity of the waves are important and useful

1024: %% parameter (tool) as shown in Appendix C.

1025: 

1026: 

1027: %% In spite of the careful treatment of the resistive RMHD in this paper,

1028: %% this conclusion should be criticized more strictly. For example,

1029: %% we neglect the discontinuity of the electron-positron plasma

1030: %% and its thermal noise. We may have to consider the disappearance

1031: %% of the electromagnetic signals into the thermal noise.

1032: %% The conclusion in this paper may be strongly contradictory

1033: %% to the standard common sense of the physics. We should fix this

1034: %% problem before we use the resistive RMHD.

1035: %% Eventually, we may have to perform the superluminal

1036: %% communication experimentally, with the electron-positron plasma

1037: %% (resistive plasma).

1038: 

1039: 

1040: \begin{acknowledgments}

1041: I thank Mika Koide, Takahiro Kudoh, 

1042: Dongsu Ryu, Masaaki Takahashi, and Satoshi Yajima for this study. 

1043: David L. Meier spent considerable effort checking my manuscript.

1044: I appreciate his important comments and suggestions.

1045: This work was supported in part by the Science Research Fund of the

1046: Japanese Ministry of Education, Culture, Sports, Science and Technology.

1047: \end{acknowledgments}

1048: 

1049: \appendix

1050: 

1051: \section{Derivation of frictional four-force density

1052: \label{appena}}

1053: 

1054: In this appendix, we derive the friction four-force density between electron

1055: and positron fluids, whose proper densities are $n_\pm$.

1056: We use $f_-^\mu$ and $f_+^\mu$ to denote the friction density

1057: of the electron and positron fluids, respectively.

1058: The principle of action--reaction is expressed as

1059: \begin{equation}

1060: f_+^i + f_-^i = 0 \verb!   ! (i=1,2,3) ,

1061: \end{equation}

1062: in any inertial frame $x^\mu$. When we consider any other inertial

1063: frame $X^\mu = A^\mu_\nu x^\nu$, the principle of action and reaction is

1064: \begin{equation}

1065: F_+^i + F_-^i = A^i_\nu (f_+^\nu + f_-^\nu) = A^i_0 (f_+^0 + f_-^0) = 0.

1066: \end{equation}

1067: Because $A^i_0 \neq 0$ in general, we have $f_+^0 + f_-^0 = 0$, or

1068: \begin{equation}

1069: f_+^\mu + f_-^\mu = 0  .

1070: \label{actcoa}

1071: \end{equation}

1072: Note that $f_+^0 + f_-^0 =0$ is the law of conservation of energy.

1073: We consider the center-of-mass frame of the two fluids 

1074: $x^{\mu '} = a^\mu_\nu x^\nu$

1075: where the four-velocity of the electron/positron fluids $U_\pm^{\mu '}$ satisfies

1076: \begin{eqnarray}

1077: n_+ U_+^{i '} + n_- U_-^{i '} = 0.

1078: \label{cenofmas}

1079: \end{eqnarray}

1080: With respect to the inverse transformation, $x^\mu = b^\mu_\nu x^{\nu \prime}$,

1081: we have

1082: \begin{equation}

1083: n_+ U_+^\mu + n_- U_-^\mu = b^\mu_\nu 

1084: (n_+ U_+^{\nu \prime} + n_- U_-^{\nu \prime})

1085: - b^\mu_0 (n_+ \gamma_+^\prime + n_- \gamma_-^\prime),

1086: \end{equation}

1087: where the prime denotes the variable observed in the center-of-mass frame.

1088: Using the definition $U^\mu = (n_+ U_+^\mu + n_- U_-^\mu)/(2n)$ and

1089: $\gamma = (n_+ \gamma_+ + n_- \gamma_-)/(2n)$, we have

1090: \begin{equation}

1091: b^\mu_0 = \frac{U^\mu}{\gamma'}.

1092: \label{bm0}

1093: \end{equation}

1094: %% These equations yield

1095: %% \begin{equation}

1096: %% a^i_\nu ( u_+^\nu + u_-^\nu ) = 0, \verb!   !

1097: %% a^0_\nu ( u_+^\nu - u_-^\nu ) = 0 .

1098: %% \end{equation}

1099: In the center-of-mass frame, the spacial components of the friction force density are

1100: \begin{equation}

1101: -f_-^{i '} = f_+^{i '} = 

1102: -m\sigma_{\rm ee} v_{\rm r} n_+ n_- \gamma_+^\prime \gamma_-^\prime 

1103: (v_+^{i \prime} -v_-^{i \prime}),

1104: %% -m \nu_{\rm ee} n \gamma'^2 (u_+^{i '} - u_-^{i '})

1105: \end{equation}

1106: where $\sigma_{\rm ee} $ is the electron/positron collisional cross section,

1107: which is a function of the thermal velocity. 

1108: The average relative velocity of the electrons and positrons, $v_{\rm r}$,

1109: is roughly given by the maximum of the thermal velocity and the relative

1110: velocity of the two fluids.

1111: We write the friction four-force density as

1112: \begin{eqnarray}

1113: -f_-^\mu = f_+^\mu = b^\mu_\nu f_+^{\nu \prime}

1114: &=& - m \sigma_{\rm ee}  v_{\rm r} n_+ n_- \gamma_+^\prime \gamma_-^\prime

1115: (v_+^{i \prime} - v_-^{i \prime}) + b^\mu_0 f_+^{0 \prime} \nonumber \\

1116: &=& - m \sigma_{\rm ee}  v_{\rm r} (n_-\gamma_-^\prime n_+ U_+^\mu 

1117: - n_+ \gamma_+^\prime n_- U_-^\mu) + b^\mu_0 f_+^{0 \prime} .

1118: \label{frc4frc1}

1119: \end{eqnarray}

1120: When we use the collision frequency of the electron and positron

1121: $\nu_{\rm ee} = \sigma_{\rm ee}  v_{\rm r} n$ and equation (\ref{bm0}), we have

1122: \begin{equation}

1123: f_+^\mu = - \frac{m\nu_{\rm ee}}{n} 

1124: (n_- \gamma_-^\prime n_+ U^\mu_+ - n_+ \gamma_+^\prime n_- U^\mu_-)

1125: + \frac{n_+ U^\mu_+ + n_- U^\mu_-}{n_+ \gamma_+^\prime + n_- \gamma_-^\prime} f_+^{0 \prime} .

1126: \label{frc4frc2}

1127: \end{equation}

1128: 

1129: Next we consider the energy gain rate of the positron fluid $f_+^{0\prime}$

1130: in the center-of-mass frame. The positron and electron fluids lose the 

1131: kinetic energy due to friction at the rate,

1132: \begin{eqnarray}

1133: -f_+^{i \prime} v_{+ i}' - f_-^{i \prime} v_{- i}' 

1134: &=& - f_+^{i \prime} v_{+ i}' - \frac{1}{n_- \gamma_-'} f_-^{i \prime} n_- U_{- i}' \nonumber \\

1135: &=& - f_+^{i \prime} v_{+ i}' - \frac{1}{n_- \gamma_-'} f_+^{i \prime} n_+ U_{+ i}'

1136: = - f_+^{i \prime} v_{+ i}' \left ( 1 + \frac{n_+ \gamma_+'}{n_- \gamma_-'} \right ).

1137: \end{eqnarray}

1138: In the above calculation, we employ the principle of action--reaction 

1139: (\ref{actcoa}), the condition of the center-of-mass frame (\ref{cenofmas}),

1140: and the assumption that

1141: the lost energy is thermalized. 

1142: %% and is distributed to the thermal energy of the positron and electron fluids. 

1143: A fraction $\theta$ of this thermalized energy 

1144: ($0 \le \theta \le 1$) is distributed to the positron and electron fluids,

1145: assuming equipartition, and other part is returned

1146: to the original fluid. Then the energy gain rate of the positron

1147: is calculated as

1148: \begin{eqnarray}

1149: f_+^{0 \prime} &=& \frac{- \theta f_+^{i \prime} v_{+ i}'

1150: - \theta f_-^{i \prime} v_{- i}'}{\gamma_+' n_+ + \gamma_-' n_-}

1151: \gamma_+' n_+ 

1152: +(1-\theta) (- f_+^{i \prime} v_{+ i}') - (- f_+^{i \prime} v_{+ i}) \\

1153: &=& - \theta  

1154: f_+^{i \prime} v_{+ i} \frac{\gamma_+' n_+ - \gamma_- n_-}{n_- \gamma_-'} .

1155: \label{enegainposi1}

1156: \end{eqnarray}

1157: Using the definition of the average four-velocity and four-current density

1158: (\ref{ave4vel}), (\ref{ave4cur}) and the friction force density expression 

1159: (\ref{frc4frc2}), 

1160: we have

1161: \begin{eqnarray}

1162: f_+^{i \prime} v_{+ i}' &=& - m \sigma_{\rm ee} v_{\rm r} n_+ n_- \gamma_+' \gamma_-'

1163: (v_+^{i \prime} - v_-^{i \prime}) v_{+ i}' \nonumber \\

1164: &=& - \frac{m \sigma_{\rm ee} v_{\rm r} n}{2 n_+ \gamma_+' \gamma' e^2} [Q^2 - WM] ,

1165: \label{fidpvdpi}

1166: \end{eqnarray}

1167: where $M=U_\nu U^\nu$, $Q=J_\nu U^\nu$, $W=J_\nu J^\nu$.

1168: Equations (\ref{enegainposi1}) and (\ref{fidpvdpi}) yield

1169: \begin{eqnarray}

1170: f_+^{0 \prime} &=& 

1171:  \frac{m \sigma_{\rm ee} v_{\rm r} n}{2 e^2 n_+ \gamma_+' \gamma' n_- \gamma_-'}

1172: (Q^2 - WM) (\gamma_+' n_+ - \gamma_-' n_-) \theta \nonumber \\

1173: &=& - \frac{m \sigma_{\rm ee} v_{\rm r} n^2/e^2}{(n^2M)^2- \left ( \frac{n}{2e}Q \right )^2}

1174: (Q^2 - WM) \frac{nQ}{2e} \theta.

1175: \label{eneexc}

1176: \end{eqnarray}

1177: Here we used

1178: \begin{eqnarray}

1179: \gamma_+' n_+ - \gamma_-n_- ' &=& \gamma' J^{0 \prime}

1180: = - J^{0 \prime} U_0' = - J^{\nu \prime} U_\nu' = - J^\nu U_\nu = -Q , \nonumber \\

1181: \gamma'^2 &=& - U^{0 \prime} U_0' = - U^{\nu \prime} U_\nu' 

1182: = - U^\nu U_\nu = -M , \nonumber \\

1183: \gamma'^2 n_+  \gamma_+' n_- \gamma_-' &=& 

1184: \gamma'^2 \left ( nU^{0 \prime} + \frac{J^{0 \prime}}{2e} \right )

1185: \left ( nU^{0 \prime} - \frac{J^{0 \prime}}{2e} \right ) \nonumber \\

1186: &=& \frac{1}{n^2} \left [

1187: (n^2 U_\nu U^\nu)^2 - \left ( \frac{n}{2e} J_\nu U^\nu \right )^2

1188:  \right ]  .

1189: \end{eqnarray}

1190: The equation of friction four-force density (\ref{frc4frc2}) reads 

1191: \begin{eqnarray}

1192: f_+^\mu &=& - m \sigma_{\rm ee} v_{\rm r} \left [

1193: n_- \gamma_-^\prime \left ( n U^\mu + \frac{1}{2e} J^\mu \right )

1194: - n_+ \gamma_+^\prime \left ( n U^\mu - \frac{1}{2e} J^\mu \right )

1195: \right ] + \frac{U^\mu}{\gamma'} f_+^{0 \prime} \\

1196: &=& - \frac{m\sigma_{\rm ee} v_{\rm r}n}{\gamma' e}

1197: \left [ -(U_\nu U^\nu) J^\mu + (U_\nu J^\nu) U^\mu \right ]

1198: + \frac{U^\mu}{\gamma'} f_+^{0 \prime} .

1199: \end{eqnarray}

1200: When we introduce the dimensionless factor with respect to the

1201: left hand side of equation (\ref{eneexc}),

1202: %% the variable with respect to the thermalization of 

1203: %% the kinetic energy to the thermal energy of positron and electron fluids,

1204: \begin{equation}

1205: \Theta = \frac{2m^2}{e^2} \frac{Q^2-MW}{(2mnM)^2-(mQ/e)^2} \theta  ,

1206: %% \Theta = 2 \frac{\left ( \frac{n}{2e} Q \right )^2 - \frac{1}{(2e)^2} n^2 WM }

1207: %% {(n^2M)^2 - \left ( \frac{n}{2e} Q \right )^2} \theta, \nonumber

1208: \end{equation}

1209: we finally obtain

1210: \begin{equation}

1211: f_+^\mu = - \frac{n \sigma_{\rm ee}  v_{\rm r} n}{e} \sqrt{-M}

1212: \left [ J^\mu - \frac{Q}{M} U^\mu (1+\Theta) \right ].

1213: \end{equation}

1214: We also calculate the resistive term in Ohm's law,

1215: \begin{equation}

1216: \frac{f_+^\mu}{en\gamma} = - \eta \frac{\sqrt{-M}}{\gamma}

1217: \left [ J^\mu - \frac{Q}{M} U^\mu (1+\Theta) \right ],

1218: \end{equation}

1219: where $\eta \equiv m \sigma_{\rm ee}  v_{\rm r}/e^2$ is resistivity.

1220: When we use the collision frequency $\nu_{\rm ee} = \sigma_{\rm ee}  v_{\rm r} n$,

1221: we can write $\eta = m \nu_{\rm ee}/(ne^2)$.

1222: 

1223: %% Time-component of the four-force is

1224: %% \begin{equation}

1225: %% f_+^{0 '} = f_-^{0 '} = 0  ,

1226: %% \end{equation}

1227: %% because of symmetry $p_+^{0 '} = p_-^{0 '}$, and $f_+^{0 '} = d p_+^{0 '}/d\tau$,

1228: %% $f_-^{0 '} = d p_-^{0 '}/d\tau$, where $p_\pm^\mu$ is four momentum density

1229: %% of the electron/positron fluids.

1230: %% Then, when we write the inverse transformation 

1231: %% $x^\mu = b^\mu_\nu x^{\nu '}$, we have

1232: %% \begin{eqnarray}

1233: %% f_+^\mu & = & - f_-^\mu = b^\mu_\nu f_+^{\nu '} = b^\mu_0 f_+^{0 '} + b^\mu_i f_+^{i '} \nonumber \\

1234: %% & = &  - m \nu_{\rm ee} n b^\mu_i (u_+^{i '} - u_-^{i '}) \\

1235: %% & = &  - m \nu_{\rm ee} n (u_+^\mu - u_-^\mu). \nonumber

1236: %% \end{eqnarray}

1237: 

1238: \section{Forbidden range of superluminal communication

1239: \label{appenb}}

1240: 

1241: We prove that $\partial \Omega/\partial k < 1$ in the dispersion 

1242: relation (\ref{disprealomega}) when $H < 1.5$.

1243: In this appendix, we omit the hat over $\hat{k}$.

1244: The determinant of the cubic equation (\ref{disprealgamma}) with

1245: respect to $\gamma$ is $D_\gamma = -(C^3+F^2)/(9 \times 36^2)$.

1246: If $\gamma$ had three different real solutions, it would yield $D_\gamma > 0$, i.e.,

1247: $\Omega^2 (4\Omega^2 -C)^2 = (C^3+F^2)/27<0$ 

1248: ($\Omega$ would be pure imaginary).

1249: %% This means that $\Omega$ has no real solution.

1250: Then $\gamma$ has only one real solution, so that $\Omega$ has a real solution.

1251: When we consider a function of the left hand side of equation (\ref{disprealgamma}),

1252: \[

1253: f(\gamma) = \gamma^3 - \gamma^2 - \frac{1}{4} \left (

1254: 1 + k^2 + \frac{1}{H} \right )\gamma - \frac{1}{8H},

1255: \]

1256: we have $f(0) = -1/(8H)<0$ and $f(1/2)=k^2/8>0$, and then

1257: the single solution of $\gamma$ should be $0< \gamma < 1/2$.

1258: Then we have

1259: \begin{equation}

1260: k^2 + \frac{1}{H} - \frac{1}{3} < \Omega^2 < k^2 + \frac{1}{H}  ,

1261: \label{rangeomega2}

1262: \end{equation}

1263: because

1264: \begin{equation}

1265: \Omega^2 = 3\gamma ( \gamma - \frac{2}{3}) + k^2 + \frac{1}{H}

1266: =3 \left ( \gamma - \frac{1}{3} \right )^2 - \frac{1}{3} + k^2 + \frac{1}{H}.

1267: \label{omega2gamma}

1268: \end{equation}

1269: From equation (\ref{omega2gamma0}), we have

1270: \begin{equation}

1271: \Omega \frac{d \Omega}{dk} = (3\gamma -1) \frac{d \gamma}{dk} + k.

1272: \label{domegadk}

1273: \end{equation}

1274: Using equation (\ref{disprealgamma}), we obtain

1275: \begin{equation}

1276: \frac{d \gamma}{dk} = k \left [ 1 - \frac{1}{2} 

1277: \frac{3\gamma^2 -\gamma}{3 \gamma^2 - 2\gamma + \frac{1}{4} (1 + k^2 + 1/H)}

1278: \right ].

1279: \label{dgammadk}

1280: \end{equation}

1281: When $H < 3$, the denominator in equation (\ref{dgammadk}) is positive because of

1282: the right side equation of equation (\ref{omega2gamma}).

1283: From equations (\ref{domegadk}) and (\ref{dgammadk}), we have

1284: \begin{equation}

1285: \frac{d\Omega}{dk} = \frac{k}{2\Omega}

1286: \frac{\Omega^2 - \frac{k^2}{2} - \frac{1}{2H} + \frac{1}{2} -\gamma }

1287: {\Omega^2 - \frac{3k^2}{4} - \frac{3}{4H} + \frac{1}{4} }   .

1288: \label{domegadk2}

1289: \end{equation}

1290: % The denominator of equation (\ref{domegadk2}) is positive when $H<3$.

1291: We consider the difference between the numerator and positive denominator

1292: of equation (\ref{domegadk2}),

1293: \begin{equation}

1294: \Delta = k \left (\Omega^2 - \frac{k^2}{2} - \frac{1}{2H} + \frac{1}{2} -\gamma \right )

1295: -2 \Omega \left  (\Omega^2 - \frac{3k^2}{4}) -\frac{3}{4H} + \frac{1}{4} \right )  .

1296: \end{equation}

1297: After some algebraic calculations, we have

1298: \begin{equation}

1299: \Delta = (k-\Omega) \left ( \Omega^2 -k^2 - \frac{1}{2H} \right )

1300: -\frac{\Omega}{4} \left ( 4 \Omega^2 - 4k^2 - \frac{4}{H} +2 \right )

1301: + \left ( \frac{1}{2} - \gamma \right ) k

1302: - \frac{3}{4} \Omega k^2 + (k-\Omega) \frac{k^2}{2}.

1303: \end{equation}

1304: From the left inequality in equation (\ref{rangeomega2}), we find

1305: $\Omega > k$ when $H<3$. And then we have

1306: \begin{equation}

1307: \Delta \le (k-\Omega) \left ( \frac{1}{2H} - \frac{1}{3} \right )

1308: + (k-\Omega) \frac{k^2}{2} - \frac{\Omega}{6} 

1309: + \left ( \frac{1}{2} - \gamma \right ) k - \frac{3}{4} k^2.

1310: \end{equation}

1311: Using equation (\ref{omega2gamma0}), we obtain

1312: \begin{equation}

1313: \Delta \le (k-\Omega) \left ( \frac{1}{2H} - \frac{1}{3} \right )

1314: +\frac{k-\Omega}{6} + \frac{k}{2} \left [

1315: \frac{2}{3} - \frac{1}{H} + \Omega \left ( \Omega - \frac{5k}{2} \right)

1316: -3 \gamma^2 \right ].

1317: \end{equation}

1318: When $k \ge 2/\sqrt{21H}$, we get $\Omega^2 < k^2 + 1/H \le 25k^2/4$

1319: using equation (\ref{rangeomega2}). Then we have $\Omega \ge 5k/2$ and

1320: $\Delta < 0$ when $H<3/2$.

1321: 

1322: On the other hand, when $k < 2/\sqrt{21H}$, we obtain

1323:  $\Omega^2 < k^2 + 1/H < 25/(21H)$ and then $\Omega < 5/\sqrt{21H}$.

1324: After some calculations, we have

1325: \begin{eqnarray}

1326: \Delta & \le &  (\Omega -k) \left ( 

1327: \frac{1}{6} -\frac{1}{2H} + \frac{k\Omega}{2} -\frac{3k^2}{4} \right )

1328: + \frac{k}{2} \left ( \frac{2}{3} - \frac{1}{H} \right )

1329: -\frac{3k^2}{4} - \frac{3\gamma^2 k}{2} \nonumber \\

1330: & < & (\Omega -k) \left ( 

1331: \frac{1}{6} -\frac{11}{42H} -\frac{3k^2}{4} \right )

1332: + \frac{k}{2} \left ( \frac{2}{3} - \frac{1}{H} \right ) 

1333: -\frac{3k^2}{4} - \frac{3\gamma^2 k}{2} < 0   ,

1334: \end{eqnarray}

1335: when $H \le 3/2$. Summarizing above calculations, we conclude that

1336: $\Delta < 0$ when $H \le 3/2$. This shows that 

1337: $v_{\rm g} = \partial \Omega/\partial k < 1$ when $H<3/2$.

1338: 

1339: \section{Propagation and damping of electromagnetic wave packets

1340: \label{appenc}}

1341: 

1342: Here we consider the propagation of a packet of electromagnetic

1343: waves in a resistive plasma.

1344: The wave packet is regarded as an element for communication in the medium.

1345: First, we use the analytic approximation of a wave packet with a large width. 

1346: 

1347: 

1348: \subsection{An analytic approximation solution}

1349: 

1350: Any variable perturbation of the electromagnetic wave packet 

1351: in resistive pair plasma, $f_1$, is given by,

1352: \begin{equation}

1353: f_1 = \int_{-\infty} ^\infty F(k) e^{ikx - i\omega(k) t} dk  ,

1354: \label{analyticsolution0}

1355: \end{equation}

1356: where $F(k)$ is the Fourier transformation of the variable $f_1$. 

1357: We take the Gaussian distribution of the wave packet to be

1358: \begin{equation}

1359: F(k) \propto \frac{\sigma}{\sqrt{2\pi}} e^{-\frac{\sigma^2}{2} (k-k_0)^2}  ,

1360: \label{gaussdistr}

1361: \end{equation}

1362: where $\sigma$ is the width of the wave packet and $k_0$ is 

1363: the characteristic wave number.

1364: The initial profile of the variable $f_1$ is proportional to

1365: $\exp[-x^2/(2\sigma^2)] \exp(ik_0x)$.

1366: 

1367: When $1/\sigma$ is much smaller than the characteristic scale $\Delta k$

1368: of the dispersion relation with respect to $\omega(k)$, we use an approximation

1369: \begin{equation}

1370: f_1 = \int_{k_0-}^{k_0+} F(k) e^{ikx - i\omega(k)t} dk

1371: \approx \int_{k_0-}^{k_0+} F(k) e^{ikx - i\left [ 

1372: \omega(k_0) + \frac{\partial \omega}{\partial k} (k_0) (k - k_0) 

1373: + \frac{1}{2} \frac{\partial^2 \omega}{\partial k^2} (k_0) (k - k_0)^2

1374: \right ] t} dk ,

1375: \end{equation}

1376: where we have expanded to 2nd order in $k - k_0$.

1377: When we write $u=\partial k / \partial \omega (k_0)$,

1378: $D=\partial^2 \omega / \partial k^2 (k_0)$,

1379: \begin{eqnarray}

1380: f_1 \propto \frac{\sigma}{\sqrt{2\pi}} e^{ik_0x - i \omega(k_0) t}

1381: \int_{0-}^{0+} e^{-\frac{1}{2} (\sigma^2 + iDt) k'^2 

1382: + i(x-ut)k' } dk' \\

1383:  \approx \frac{\sigma}{\sqrt{2\pi}} e^{ik_0x - i \omega(k_0) t}

1384: \int_{-\infty}^{+\infty} e^{-\frac{1}{2} (\sigma^2 + iDt) k'^2 

1385: + i(x-ut)k' } dk' \\

1386: = \frac{\sigma}{\sqrt{2\pi}} e^{ik_0x - i \omega(k_0) t}

1387: \sqrt{\frac{\pi}{|\sigma^2+iDt|^2+\Re({\sigma^2+iDt})}}

1388: \left ( 1 + \frac{\sigma^2-iD^*t}{|\sigma^2+iDt|^2} \right ) \nonumber \\

1389: \times \exp \left [-\frac{1}{2} (\sigma^2 + iDt) (x-ut)^2 \right ]  .

1390: \end{eqnarray}

1391: Here we used a formula,

1392: \begin{equation}

1393: \int_{-\infty}^\infty e^{-\frac{c}{2}x^2} dx

1394: = \sqrt{\frac{2\pi}{c}}

1395: = \sqrt{\frac{\pi}{|c| + \Re{c}}} \left ( 1+ \frac{c^*}{|c|} \right )

1396: \verb!   ! (c \in {\bf R})  ,

1397: \end{equation}

1398: where $c$ is an arbitrary complex constant.

1399: When we write $u=u_{\rm r} + i u_{\rm i}$ and $D=D_{\rm r} + i D_{\rm i}$

1400: ($u_r, u_i, D_r, D_i \in \VEC{R}$),

1401: after some algebraic calculations, we have 

1402: \begin{eqnarray}

1403: f_1 &= \frac{1}{\sqrt{2}} 

1404: \left [ 1 + \frac{\sigma^2-iD_{\rm i}t -iD_{\rm r}t}

1405: {\sqrt{(\sigma^2 - D_{\rm i} t)^2+(D_{\rm r}t)^2}} \right ] 

1406: \frac{1}{\sqrt{\sqrt{(1-D_{\rm i}t/\sigma^2)^2+(D_{\rm r}t/\sigma^2)^2}

1407: +1-D_{\rm i}t/\sigma^2}}  \\

1408: & \times \exp \left [ ik_0 x - i\omega(k_0) t 

1409: -i \frac{2(x-u_{\rm r}t)u_{\rm i}(\sigma^2-D_{\rm i}t )t + 

1410: D_{\rm r} \{ (x-u_{\rm r}t)^2-(u_{\rm i}t)^2 \} t}

1411: {2\{ (\sigma^2 -D_{\rm i} t)^2 + (D_{\rm r} t)^2 \}} \right ] \\

1412: & \times \exp \left [ - \frac

1413: {(\sigma^2 -D_{\rm i}t) \left ( x - u_{\rm r} t 

1414: - \frac{D_{\rm r} u_{\rm i}}{\sigma^2 - D_{\rm i} t} t^2 \right )^2

1415: -\frac{(D_{\rm r} u_{\rm i})^2}{\sigma^2 - D_{\rm i} t} t^4

1416: -u_{\rm i}^2 (\sigma^2 - D_{\rm i} t)t^2}

1417: {2\{ (\sigma^2 - D_{\rm i}t)^2 + (D_{\rm r}t)^2) \}}

1418: \right ]  .

1419: \end{eqnarray}

1420: The width of the wave packet is approximately given by 

1421: \begin{equation}

1422: \Delta x \sim \sqrt{\frac{(\sigma^2 - D_{\rm i}t)^2+(D_{\rm r}t)^2}

1423: {\sigma^2-D_{\rm i}t}}.

1424: \end{equation}

1425: When $t \ll \frac{u_{\rm r} \sigma^2}{u_{\rm r} D_{\rm i} + D_{\rm r} u_{\rm i}}$,

1426: the propagation velocity of the wave packet is

1427: \begin{equation}

1428: v_{\rm packet} \sim u_{\rm r}.

1429: \end{equation}

1430: The diffusion time scale of the wave packet, $T_{\rm D}$, 

1431: is found from the condition

1432: $(\Delta x)^2 = 2 \sigma^2$, which yields

1433: \begin{equation}

1434: T_{\rm D} = \frac{\sigma^2}{\sqrt{D_{\rm r}^2 + D_{\rm i}^2}}

1435: =\frac{\sigma^2}{|D|}  .

1436: \label{decaytime}

1437: \end{equation}

1438: The propagation velocity, $v_{\rm packet}$, has meaning

1439: only when $t \ll T_{\rm D}$ or $\sigma \gg \sqrt{t|D|}$.

1440: 

1441: \subsection{A method of numerical integration and application for imaginary

1442: superluminal propagation of wave packet}

1443: 

1444: We show a numerical solution for an electromagnetic wave packet

1445: propagation in resistive pair plasma, using Simpson's formula.

1446: The variable perturbation of the electromagnetic wave packet $f_1$ given by 

1447: equations (\ref{analyticsolution0}) and (\ref{gaussdistr}) is calculated as

1448: \begin{equation}

1449: \Re(f_1) \propto e^{-\gamma_0 t} P(x,t)  ,

1450: \end{equation}

1451: % where

1452: \begin{equation}

1453: P(x,t) = \frac{\sigma}{\sqrt{2\pi}}

1454: \int_{k_0-M/\sigma}^{k_0+M/\sigma}

1455: e^{-\frac{\sigma^2}{2} (k-k_0)^2 - (\gamma(k) - \gamma_0) t}

1456: \cos(kx-\Omega(k)t) dk  ,

1457: \end{equation}

1458: where $\gamma_0$ is the characteristic damping rate of the wave packet.

1459: When the number $M$ is large enough, the integration becomes the exact value.

1460: Usually we set $M=$4--10, which gives precise enough evaluation.

1461: To calculate the profile of the wave packet, we evaluate the profile function

1462: $P(x,t)$.

1463: 

1464: Figure \ref{timevol} shows mathematically the imaginary time evolution 

1465: of a superluminal electromagnetic wave packet with

1466: $\sigma=50$, $k_0=0.25$ in the pair plasma when $H=4$ in equation (\ref{dispcomlex}).

1467: The magnification rate of the variable $f_1$ is indicated by the factor

1468: beside the ordinate.

1469: The propagation velocity of the wave packet is found to be $v_{\rm packet} \sim 1.4$,

1470: i.e., superluminal. The group velocity $u_{\rm r}=1.47$ gives a good approximation

1471: to the propagation velocity of the wave packet.

1472: When $t=4,000 \alt T_{\rm D} = 5,000$, the wave packet

1473: begins diffuse. The value of $T_{\rm D}$ gives a good estimate of 

1474: wave packet break-down.

1475: This numerical calculation clearly shows that superluminal 

1476: propagation of a wave packet

1477: is possible if the group velocity of electromagnetic wave 

1478: exceeds the speed of light.

1479: 

1480: %% \subsection{Image of superluminal communication}

1481: %%

1482: %% Here we discuss an imaginary method of the superluminal communication with

1483: %% the electromagnetic wave in the resistive plasma if the group velocity

1484: %% exceeds the light speed.

1485: %% To send the information, we use the protocol of ``Morse signals"

1486: %% with the superluminal wave packet discussed in the above subsection.

1487: %% That is, we use the long and short wave packet to send the signal

1488: %% ``1" and ``0", respectively (Fig. \ref{commun}).

1489: %% The amplitude of the wave packet damps very rapidly as shown in the 

1490: %% previous section,

1491: %% however, the amplitude is always finite and we can detect the length

1492: %% of the wave packet principally before it is destroyed at 

1493: %% $x \alt v_{\rm g} T_{\rm D} = (\partial \Omega/\partial k)

1494: %% (\sigma^2/[\partial^2 \Omega/\partial k^2])$

1495: %% when we can avoid the hidden

1496: %% of the signal behind the thermal noise of the resistive plasma.

1497: %% Then we can realize the superluminal communication by the 

1498: %% electromagnetic wave packet in the resistive plasma using

1499: %% ``Morse code" signals, if the group velocity of the waves exceeds

1500: %% the light speed and we can detect the electromagnetic small signals.

1501: 

1502: \begin{figure}

1503: \includegraphics[width=0.9\textwidth]{fig5.eps}% Here is how to import EPS art

1504: \caption{

1505: \label{timevol}

1506: Imaginary time evolution of wave packet superluminal propagation 

1507: in pair plasma, with $\sigma = 50$, $k_0=0.25$, $H=4$.

1508: Here $f_1$ is normalized by the maximum initial value.

1509: }

1510: \end{figure}

1511: 

1512: %%%% \begin{figure}

1513: %%%% \includegraphics[width=\textwidth]{plasmaparam.eps}% Here is how to import EPS art

1514: %%%% \caption{

1515: %%%% The state of the pair plasma in the various electric density $n_{\rm e}$

1516: %%%% and the temperature $T_{\rm e}$.

1517: %%%% \label{plasmaparam}

1518: %%%% }

1519: %%%% \end{figure}

1520: 

1521: %% \begin{figure}

1522: %% \includegraphics[width=0.7 \textwidth]{commun.eps}% Here is how to import EPS art

1523: %% \caption{

1524: %% Image of ``Morse signals".

1525: %% \label{commun}

1526: %% }

1527: %% \end{figure}

1528: 

1529: 

1530: %% \newpage %Just because of unusual number of tables stacked at end

1531: \bibliography{apssamp}% Produces the bibliography via BibTeX.

1532: 

1533: \begin{thebibliography}{}

1534: %% \bibitem[Koide et al. 1998]{koide98}

1535: \bibitem{koide98}

1536: S. Koide, K. Shibata, and T. Kudoh, \apj Lett., 495, L63 (1998).

1537: %% \bibitem[Koide et al. 1999]{koide99}

1538: \bibitem{koide99}

1539: S. Koide, K. Shibata, and T. Kudoh, \apj, 522, 727 (1999).

1540: %% \bibitem[Koide et al. 2000]{koide00}

1541: \bibitem{koide00}

1542: S. Koide, D. L. Meier, K. Shibata, and T. Kudoh, \apj, 536, 668 (2000).

1543: %% \bibitem[Koide et al. 2002]{koide02}

1544: \bibitem{koide02}

1545: S. Koide, K. Shibata, T. Kudoh, and D. L. Meier, Science, 295, 1688 (2002).

1546: %% \bibitem[Koide 2003]{koide03}

1547: \bibitem{koide03}

1548: S. Koide, Phys. Rev. D 67, 104010 (2003).

1549: %% \bibitem[Koide 2004]{koide04}

1550: \bibitem{koide04}

1551: S. Koide, Astrophys. J. Lett, 606, L45 (2004).

1552: %% \bibitem[Koide et al. 2006]{koide06}

1553: \bibitem{koide06}

1554: S. Koide, K. Shibata, and T. Kudoh, Phys. Rev. D, 74, 044005 (2006).

1555: %% \bibitem[Mizuno et al. 2004]{mizuno04a}

1556: \bibitem{mizuno04}

1557: Y. Mizuno, S. Yamada, S. Koide, and K. Shibata,

1558: \apj, 606, 395 (2004).

1559: %% \bibitem[Komissarov 2004]{komissarov04}

1560: \bibitem{komissarov04}

1561: S. S. Komissarov, Mon. Not. R. Astron. Soc., 350, 1431 (2004).

1562: %% \bibitem[Komissarov et al. 2007]{komissarov07}

1563: \bibitem{komissarov07}

1564: S. S. Komissarov, and J. C. McKinney, Mon. Not. R. Astron. Soc., 377, L49 (2007).

1565: %% \bibitem[Gammie et al. 2003]{gammie03}

1566: \bibitem{gammie03}

1567: C. F. Gammie, J. C. McKinney, and G. Toth, \apj, 589, 444 (2003).

1568: %% \bibitem[McKinney et al. 2004]{mckinney04}

1569: \bibitem{mckinney04}

1570: J. C. McKinney and C. F. Gammie, \apj, 611, 977 (2004).

1571: %% \bibitem[McKinney 2006]{mckinney06}

1572: \bibitem{mckinney06}

1573: J. C. McKinney, Mon. Not. R. Astron. Soc., 368, 1561 (2006).

1574: %% \bibitem[Watanabe et al. 2006]{watanabe06}

1575: \bibitem{watanabe06}

1576: N. Watanabe and T. Yokoyama, Astrophys. J. Lett, 647, L123 (2006).

1577: \bibitem{komissarov07b}

1578: S. S. Komissarov, Mon. Not. R. Astron. Soc. 382, 995 (2007).

1579: \bibitem{ardavan76}

1580: H. Ardavan, \apj, 203, 226 (1976).

1581: \bibitem{blackman93}

1582: E. G. Blackman, and G. B. Field, Phys. Rev. Lett. 71, 3481 (1993).

1583: \bibitem{gedalin96}

1584: M. Gedalin, Phys. Rev. Lett. 76, 3340 (1996).

1585: \bibitem{melatos96}

1586: A. Melatos and D. B. Melrose, Mon. Not. R. Astron. Soc.

1587: 279, 1168 (1996).

1588: \bibitem{khanna98}

1589: R. Khanna, Mon. Not. R. Astron. Soc. 294, 673 (1998).

1590: \bibitem{meier04}

1591: D. L. Meier, \apj, 605, 340 (2004).

1592: \bibitem{weinberg72}

1593: S. Weinberg, {\it Gravitation and Cosmology} 

1594: (John Wiley \& Sons, New York, 1972).

1595: \bibitem{misner70}

1596: C. W. Misner, K. S. Thorne, and J. A. Wheeler,

1597: {\it Gravitation}, (W. H. Freeman and Company, New York, 1970).

1598: \bibitem{bellan06}

1599: P. M. Bellan, {\it Fundamentals of Plasma Physics},

1600: (Cambridge University Press, Cambridge, 2006).

1601: \bibitem{koide08}

1602: S. Koide and K. Arai, \apj, 682, 1124 (2008).

1603: \bibitem{wardle98}

1604: J. F. Wardle, D. C. Homan, R. Ojha, and D. H. Roberts,

1605: Nature 395, 457 (1998).

1606: %% \bibitem[Meier et al. 2001]{meier01}

1607: %%\bibitem{meier01}

1608: %%Meier, D. L., Koide, S., Uchida, Y. 2001, Science, 291, 84

1609: %%\bibitem{misner70}

1610: %%Misner, C. W., Thorne, K. S., \& Wheeler, J. A. 1970,

1611: %%Gravitation. (San Francisco: Freeman)

1612: %%%

1613: %%% \bibitem[McKinney et al. 2004]{mckinney04}

1614: %%% McKinney, J. C. \& Gammie, C. F. 2004, \apj, 611, 977

1615: %%% \bibitem[Komissarov et al. 2007]{komissarov07}

1616: %%% Komissarov, S. S., Barkov, M., \& Lyutikov, M. 2007,

1617: %%% MNRAS, 374, 415

1618: \end{thebibliography}

1619: 

1620: \end{document}

1621: