1: %\documentclass[12pt,preprint]{emulateapj}

2: \newif\ifjournal\journalfalse

3: %\journaltrue

4: \documentclass[12pt,preprint]{aastex}

5: 

6: \usepackage{graphicx}% Include figure files

7: 

8: \usepackage{bm}% bold math

9: \renewcommand{\vec}[1]{\boldsymbol{#1}}

10: 

11: \newcommand{\grad}{\nabla}

12: \renewcommand{\div}{\nabla \cdot}

13: \newcommand{\rot}{\nabla \times}

14: \newcommand{\pp}[2]{\frac{\partial #1}{\partial #2}}

15: %

16: \newcommand{\Alfven}{Alfv\'{e}n }

17: \newcommand{\JS}{J\"{u}tter-Synge }

18: 

19: % Journals

20: \def\grl{{\itshape Geophys. Res. Lett.} }

21: \def\jgr{{\itshape J. Geophys. Res.} }

22: \def\apj{{\itshape Astrophys. J.} }

23: \def\apjl{{\itshape Astrophys. J.} }

24: \def\prl{{\itshape Phys. Rev. Lett.} }

25: \def\pop{{\itshape Phys. Plasmas} }

26: \def\aap{{\itshape Astron. Astrophys.} }

27: \def\pra{{\itshape Phys. Rev. A} }

28: \def\pre{{\itshape Phys. Rev. E} }

29: \def\mnras{{\itshape MNRAS} }

30: 

31: \newcommand{\myemail}{zenitani@lssp-mail.gsfc.nasa.gov}

32: 

33: \shorttitle{Relativistic Guide Field Reconnection}

34: \shortauthors{Zenitani \& Hoshino}

35: 

36: \begin{document}

37: 

38: %\preprint{APS/123-QED}

39: 

40: \title{The Role of the Guide Field

41: in Relativistic Pair Plasma Reconnection}

42: 

43: \author{S. Zenitani}

44: %\affiliation{

45: \affil{

46: NASA Goddard Space Flight Center, Greenbelt, MD 20771;

47: zenitani@lssp-mail.gsfc.nasa.gov

48: }

49: \author{M. Hoshino}

50: %\affiliation{

51: \affil{

52: Department of Earth and Planetary Science, University of Tokyo,

53: 7-3-1, Hongo, Bunkyo, Tokyo, 113-0033 Japan

54: }

55: 

56: %\date{\today}

57: %\date{submitted to \it{Physics of Plasmas}}

58: 

59: \begin{abstract}

60: We study the role of the guide field

61: in relativistic magnetic reconnection

62: in a Harris current sheet of pair ($e^{\pm}$) plasmas,

63: using linear theories and particle-in-cell (PIC) simulations.

64: Two-dimensional PIC simulations exhibit

65: the guide field dependence to the linear instabilities;

66: the tearing or reconnection modes are relatively insensitive,

67: while the relativistic drift-kink instability (RDKI),

68: the fastest mode in a relativistic current sheet,

69: is stabilized by the guide field.

70: Particle acceleration in the nonlinear stage is also investigated.

71: A three-dimensional PIC simulation demonstrates

72: that the current sheet is unstable to the RDKI,

73: although a small reconnection occurs in the deformed current sheet.

74: Another three-dimensional PIC simulation with a guide field

75: demonstrates a completely different scenario.

76: Secondary magnetic reconnection is triggered by

77: nonlinear coupling of oblique instabilities,

78: which we call the relativistic drift-sausage tearing instability.

79: Therefore, particle acceleration by relativistic guide field reconnection

80: occurs in three-dimensional configuration.

81: Based on the plasma theories,

82: we discuss an important role of the guide field:

83: to enable non-thermal particle acceleration by magnetic reconnection.

84: \end{abstract}

85: 

86: \keywords{acceleration of particles --- magnetic fields

87: --- plasmas --- instabilities --- relativity}

88: \maketitle

89: 

90: 

91: \section{INTRODUCTION}

92: 

93: Magnetic reconnection in collisionless plasmas

94: is of strong interest in high-energy astrophysical places

95: such as active galactic nuclei (AGNs; \citet{dimatteo,birk01}),

96: extragalactic jets \citep{rom92,lb98,larra03},

97: pulsar winds \citep{michel82,michel94,coro90,lyu01,kirk03},

98: gamma-ray bursts \citep{dr02,drs02,uz06} and,

99: soft gamma repeaters \citep{thom95,thom01,lyut03a,lyut06}.

100: Since it rapidly releases

101: stored magnetic energy into plasma kinetic energy,

102: reconnection is considered as a possible underlying mechanism,

103: to explain particle acceleration or

104: bursty emission signatures in these sites,

105: regardless of plasma composition

106: (pair plasmas or ion-electron plasmas). 

107: 

108: For instance, in the case of the Crab pulsar,

109: magnetic reconnection is considered 

110: in the relativistic radial outflow of pair plasmas.

111: Since the central neutron star is a fast oblique rotator,

112: its strong magnetic fields ($\sim 10^{12}$ G) are highly striped

113: so that field lines are almost toroidal inside the flow.

114: Magnetic reconnection is expected to occur

115: in field reversal configuration of toroidal fields,

116: in order to dissipate the magnetic energy \citep{coro90}. 

117: % 

118: In the AGN context,

119: magnetic reconnection is expected

120: in association with magnetic loops

121: from a differentially rotating acceleration disk.

122: In this situation, the field configuration

123: involves magnetic fields perpendicular to the antiparallel component.

124: Indeed, reconnection in sheared configuration is

125: proposed as an acceleration site

126: of MeV/GeV electrons \citep{lb97,schopper,nodes03},

127: due to the field-aligned electric field $E_{\parallel}$. 

128: %

129: In the case of soft gamma repeaters,

130: a giant flare is expected

131: in ultrastrong magnetic environment ($\sim 10^{15}$ G)

132: around central neutron stars (magnetars).

133: Although there are various theoretical models,

134: it is expected that giant flares involve

135: magnetic reconnection in the relativistic pair plasma environment.

136: %Reconnection may accelerate particles or

137: %eject plasmas in ultrafast velocities.

138: 

139: However, the fundamental mechanism of

140: relativistic magnetic reconnection,

141: as well as the conventional non-relativistic counterpart,

142: is far from being understood. 

143: From the magnetohydrodynamics (MHD) viewpoint, steady state reconnection models

144: have been extended to the relativistic regime \citep{bf94b,lyut03b,lyu05},

145: although they do not fully agree. 

146: The relativistic resistive MHD simulation demonstrated

147: mildly-relativistic Petsheck reconnection \citep{naoyuki06}. 

148: %

149: From the kinetic viewpoint,

150: series of particle-in-cell (PIC) simulations demonstrated

151: that relativistic magnetic reconnection

152: is a powerful acceleration engine

153: \citep{zeni01,zeni05b,zeni07,claus04},

154: primarily due to the direct particle acceleration

155: by the reconnection electric field,

156: which is perpendicular to the reconnecting magnetic field lines. 

157: In fact, obtained plasma energy distribution is highly non-thermal,

158: approximated by the power-law distribution with an index of $-1 \sim -3$

159: \citep{rom92,zeni01,larra03,claus04}. 

160: Thus, magnetic reconnection seems to be

161: a favorable source of non-thermal particles 

162: at the sites of synchrotron radiation \citep{claus04apj}. 

163: However, it was reported that

164: the current sheet configuration is unstable to

165: the relativistic drift kink instability (hereafter the RDKI),

166: which occurs in the plane perpendicular to the reconnection plane \citep{zeni05a}.

167: The RDKI grows more rapidly than reconnection,

168: and it mainly contributes to plasma heating \citep{zeni07}. 

169: Thus, plasma heating by the RDKI overwrites the reconnection scenario;

170: nonthermal particle acceleration is unlikely to occur

171: in antiparallel field configuration

172: in relativistically hot pair plasmas.

173: 

174: Pair plasma reconnection has also attracted recent attention

175: as an equal-mass-limit example of ion-electron reconnection. 

176: For example, there is a long-standing problem regarding the reason

177: collisionless reconnection occurs ``faster'' than

178: predicted by the MHD theories.

179: It has been widely argued that

180: Hall physics plays an essential role in maintaining fast reconnection

181: (the GEM reconnection challenge; \citet{birn01,shay01}),

182: but fast reconnection in non-relativistic pair plasma

183: shows a counter-evidence

184: because it does not involve Hall effects \citep{bessho05}. 

185: Instead, although its physical interpretation needs refinement,

186: it seems that the off-diagonal part of the pressure tensor \citep{hesse99}

187: accounts for the electric field for fast reconnection \citep{bessho07,hesse07}.

188: Furthermore, it has recently been argued that

189: late-time dynamical structure may regulate fast reconnection \citep{dau07}.

190: 

191: Guide field configuration,

192: which contains a perpendicular magnetic field

193: to the antiparallel components,

194: is an important generalization for

195: studying the reconnection problem in a shear or twisted configuration.

196: Reconnection with the uniform guide field has long been studied

197: for better understanding of the reconnection structure

198: \citep{drake77,katanuma80,hoshino87,hori97,hesse99,hesse04}

199: and for applications to solar flares and

200: to the Earth's magnetopause \citep{shibata95,quest81,sonne81}.

201: Recent three-dimensional (3D) PIC simulations \citep{scholer03,drake03,ricci03,silin03,prit04}

202: discussed reduction of the cross-field activities \citep{scholer03,silin03},

203: detailed reconnection region structure \citep{prit04},

204: and enhanced electron acceleration \citep{drake03,ricci03}. 

205: We note that the acceleration mechanism differs

206: from the one in the antiparallel counterpart

207: because the reconnection electric field is

208: no longer perpendicular to magnetic lines;

209: guide field reconnection involves

210: a parallel electric field $E_{\parallel}$

211: to accelerate electrons \citep{ricci03}. 

212: %

213: In relativistic pair plasmas,

214: \citet{zeni05b} presented the idea that

215: the guide field stabilizes the RDKI

216: and that the new oblique instability

217: triggers secondary magnetic reconnection.  

218: Nonthermal particle acceleration by magnetic reconnection 

219: seems to be a likely process in a relativistic current sheet once again,

220: under the guide field condition.

221: %Thus, relativistic guide field reconnection is a possible acceleration engine

222: 

223: The purpose of this paper is to

224: comprehensively discuss the role of the guide field

225: in relativistic pair plasma reconnection.

226: Extending our recent papers

227: (\citet{zeni05b} and \citet{zeni07},

228: hereafter referred as ZH05b and ZH07, respectively),

229: we investigate the linear and non-linear development of

230: a relativistic current sheet in pair plasmas,

231: by using fully self-consistent kinetic PIC simulations. 

232: The conclusion of this paper is that

233: relativistic guide field reconnection is

234: a favorable process for nonthermal particle acceleration,

235: from the viewpoint of detailed plasma theory. 

236: %Secondary magnetic reconnection is triggered

237: %via nonlinear coupling of oblique instabilities,

238: %and then it particle acceleration occurs in three-dimensional configuration.

239: 

240: This paper consists of the following sections.

241: In \S 2 we describe the simulation setup. 

242: In \S 3 we study

243: the two-dimensional evolution of the current sheet with the guide field.

244: We investigate

245: how the guide field affects the linear instabilities in \S 3.1,

246: and then

247: we study

248: how particles are accelerated in nonlinear guide field reconnection

249: in \S 3.2.

250: In \S 4, we study the 3D evolution of

251: an anti-parallel current sheet. 

252: In \S 5, we study the 3D evolution of

253: a current sheet with a guide field.

254: We introduce new linear instabilities

255: in the oblique directions in \S 5.1,

256: and then we discuss the nonlinear development of

257: the current sheet in \S 5.2,

258: in relevance to the two-dimensional counterparts. 

259: Section 6 contains discussion and the summary.

260: 

261: 

262: \section{SIMULATION MODEL}

263: 

264: Throughout this study,

265: we employ a relativistic Harris model as an initial configuration.

266: Magnetic field and plasma distribution functions are set

267: in the following way:

268: \begin{equation}

269: \vec{B} = B_0 \tanh(z/\lambda) \vec{\hat{x}} + \alpha B_0 \vec{\hat{y}},

270: \end{equation}

271: \begin{equation}

272: f_{s} = \frac{n_0\cosh^{-2}(z/\lambda)}{4\pi m^2cT K_2(mc^2/T)}

273: \exp\big[

274: \frac{ -\gamma_s (\varepsilon - \beta_s mc u_y) }{ T }

275: \big]

276: + \frac{n_{bg}}{4\pi m^2cT_{bg} K_2(mc^2/T_{bg})} \exp\big[-\frac{\varepsilon}{T_{bg}} \big],

277: \end{equation}

278: where $B_0$ is the antiparallel magnetic fields,

279: $\alpha$ is the relative amplitude of the guide field,

280: $n_0$ is the plasma number density of the current sheet in the proper frame,

281: $T$ is the plasma temperature including the Boltzmann constant in the proper frame,

282: $m$ is the positron/electron mass,

283: $c$ is the light speed,

284: $K_2(x)$ is the modified Bessel function of the second kind,

285: the subscript $s$ denotes the species

286: ($p$ for positrons and $e$ for electrons),

287: $c\beta_s$ is the drift speed of the species, and $u$ is the four velocity.

288: The parameters $T$ and $n_0$ are defined in the proper frame.

289: $T_{bg}$ and $n_{bg}$ are the temperature and

290: the number density  of background plasmas.

291: In this paper, we choose $\beta_p=0.3$, $\beta_e=-0.3$

292: (relevant Lorentz factor is $\gamma_{\beta}=1.048$),

293: $T/mc^2=1$, $T_{bg}/T=0.1$ and $n_{bg}/(\gamma_{\beta}n_0) = 0.05$.

294: Notice that the Harris model with the uniform guide field

295: exactly satisfies an equilibrium.

296: We assume that $B_y$ is negative,

297: but we will observe the same (mirror) results in the positive cases,

298: because of the mathematical symmetry in pair plasmas. 

299: %

300: We use 3D particle-in-cell (PIC) code

301: with periodic boundaries in all three directions.

302: Because of the field reversal,

303: we set double current sheets in the $z$ direction.

304: The typical scale of the current sheet $\lambda$ is set to 10 grids.

305: The size of the simulation box depends on the problem.

306: Their sizes are presented in Table \ref{table}.

307: These conditions are same as our recent studies;

308: runs R3 and D3 are identical to ones in ZH07 and

309: runs 3D-A and 3D-B are identical to run A and run B in ZH05b.

310: 

311: In many cases, we impose small artificial electric fields around

312: $(x, y, z) = (0,0,\pm 3\lambda)$ in the very early stage of simulation runs,

313: in order to ``force'' reconnection around the center of the main simulation domain.

314: Typical spatial ranges for the artificial fields are

315: set to $(\Delta x,\Delta z) \sim (\pm 2\lambda,\pm \lambda)$ in two-dimensional cases

316: and $(\Delta x,\Delta y,\Delta z) \sim (\pm 2\lambda,\pm 2\lambda,\pm \lambda)$

317: in 3D cases.

318: This force is often strong enough and then

319: reconnection process immediately breaks up soon after short linear growth stage.

320: In Table \ref{table}, the letter F means that

321: we observe such sudden breakup of forced reconnection.

322: S* means that we observe quasi-spontaneous evolution,

323: because the trigger force is not strong enough.

324: In these cases, it takes some time

325: until nonlinear reconnection breaks up, and

326: we observe the linear growth of the instabilities for a while.

327: In S cases, we set no trigger force.

328: In two-dimensional runs for the RDKI,

329: we do not need the trigger force

330: because the instability quickly grows from thermal noise.

331: 

332: 

333: \section{GUIDE FIELD EFFECT TO THE TWO-DIMENSIONAL PROCESSES}

334: \label{sec:2Dguide}

335: 

336: \subsection{Linear Growth Rates}

337: 

338: First, we investigate the guide field effect on

339: the two-dimensional reconnection instabilities:

340: relativistic reconnection in the $x$-$z$ plane and

341: the RDKI in the $y$-$z$ plane.

342: We carry out a series of two-dimensional simulations:

343: five cases of reconnection

344: ($\alpha = B_y/B_0 = 0,-0.5,-1,-1.5,-5$)

345: and four cases of the RDKI

346: ($B_y/B_0 = 0,-0.25,-0.5,-1$).

347: Then, we calculate the linear growth rates of

348: the fastest growing modes

349: from the perturbed magnetic fields $\delta \vec{B}$ in the neutral plane. 

350: %calculated from the $\delta B_x$ perturbation in the neutral plane.

351: In reconnection cases,

352: it is sometimes difficult to pick up the ``linear'' mode,

353: because the linear stage is rather short.

354: It seems that the linear stage takes longer time,

355: when the initial artificial impact is zero or weak. 

356: Therefore, for the most difficult case of $B_y/B_0 = -1.5$ (run R3-C),

357: we also carry out the relevant spontaneous run (run R3-D)

358: to confirm the linear growth rate. 

359: We notice that the linear growth rate is

360: not identical to the energy release rates in the nonlinear stage.

361: Once the nonlinear reconnection breaks up,

362: the linear perturbations quickly cascade into

363: the longer-wavelength perturbations, which grows more explosively,

364: $2$-$3$ times faster than the linear or quasi-linear modes.

365: 

366: Along with the PIC simulations,

367: we evaluate the linear growth rates of

368: the two-dimensional instabilities by using eigenvalue analysis,

369: in the same way as our previous work (Appendix B in ZH07). 

370: In this analysis, 

371: we use the relativistic fluid equations

372: \begin{equation}\label{eq:fluid}

373: \frac{\gamma^2_s}{c^2}

374: (p_s + e_s) 

375: ( \pp{}{t} + \vec{v}_s \cdot \grad )

376: \vec{v_s}

377: =

378: - \grad p_s +

379: \gamma_s q_s n_s 

380: \Big( \vec{E} +  \frac{\vec{v}_s}{c} \times \vec{B}  \Big)

381: - \frac{\vec{v}_s}{c^2} ( \gamma_s q_s n_s \vec{E} \cdot \vec{v}_s + \pp{p_s}{t} ),

382: \end{equation}

383: where $p$ is the isotropic plasma pressure 

384: and $e$ is the internal energy that contains the rest mass energy.

385: In addition, we use the continuity, Maxwell equations, and adiabatic gas condition

386: to close equations.

387: We assume the perturbation for

388: arbitrary wavevector $\vec{k}=(k_x,k_y)$;

389: $\delta f \propto \delta f(z) \exp(ik_x x + ik_y y + \omega_i t)$,

390: where $\omega_i$ is the linear growth rate

391: (the imaginary part of the wave frequency),

392: and then

393: we solve the linearized equations as an eigenvalue problem. 

394: 

395: Let us discuss two-fluid theory in the $x$-$z$ plane

396: ($k_y=0$ or $\partial/\partial y=0)$.

397: By ignoring the relativistic drift speed effect

398: (third term of the right-hand side of eq. [\ref{eq:fluid}]),

399: simple assumptions ($q_p=-q_e=q$,

400: $\gamma_p=\gamma_e=\gamma$, $p_p=p_e=p$, $e_p=e_e=e$, and $n_p=n_e=n$)

401: yield the $y$ component of the generalized Ohm's law

402: \begin{equation}\label{eq:ohm}

403: \Big( \vec{E} + \frac{\vec{V}}{c} \times \vec{B} \Big)_y

404: =

405: \frac{\gamma (p + e) }{2 c^2 q n }

406: \Big(\pp{}{t}v_{py} - \pp{}{t}v_{ey} \Big),

407: %+ \Big( \frac{v_{py}^2+v_{ey}^2}{2c^2} E_y

408: %+ \frac{v_{py}-v_{ey}}{2c^2\gamma q n}\pp{p}{t} \Big),

409: \end{equation}

410: where $\vec{V}=(1/2)(\vec{v_p}+\vec{v_e})$ is

411: the bulk velocity of pair-plasma fluids.

412: This indicates that

413: the fluid inertia of electrons and positrons

414: account for the effective resistivity

415: to break down the frozen-in condition and to drive the tearing instability.

416: The Hall terms are not included,

417: because they vanish in electron-positron plasmas. 

418: As discussed in the previous works

419: on the relativistic tearing instability \citep{zelenyi79}

420: and the conventional tearing instability,

421: we obtain purely-growing modes

422: in the wavevector range of $0 < k_x\lambda < 1$.

423: We employ the maximum growth rates in the above parameter range

424: as theoretical eigen growth rates of the reconnection mode. 

425: % 

426: In the $y$-$z$ plane

427: ($k_x=0$ or $\partial/\partial x=0)$,

428: \citet{dau99} demonstrated that

429: two-fluid theory and Vlasov theory are in good agreement

430: to explain the drift kink instability in ion-electron plasmas,

431: in long-wavelength range of $k_y\lambda \lesssim 0.7$.

432: Similarly, we know that 

433: our two-fluid theory is in agreement with PIC simulations

434: in long-wavelength rage of the RDKI; $k_y\lambda \lesssim 0.7$

435: \citep{zeni05a,zeni07}.

436: Therefore, we employ $k_y\lambda=0.7$ to

437: estimate the typical eigen growth rates of the RDKI.

438: We keep in mind that our theory assumes isotropic fluids

439: and that it does not contain any kinetic effects nor wave-particle interactions.

440: These approximations will break down in a thin current sheet,

441: in which gyro radii is much larger,

442: and so

443: kinetic-based theories are more favorable for describing instabilities

444: (e.g. Vlasov analysis by \citet{bri95}).

445: However, in the scope of the present paper,

446: two-fluid theory gives us a plausible estimate of the growth rates.

447: 

448: The obtained growth rates of the reconnection mode

449: are presented in black lines

450: in Figure \ref{fig:rec_vs_dki_by}

451: as a function of $|B_y/B_0|$.

452: In PIC simulations, the fastest reconnection mode

453: (Fig. \ref{fig:rec_vs_dki_by}, $\textit{black line}$)

454: has wavenumbers of $k_x\lambda \sim 0.4-0.7$,

455: which correspond to those

456: of the relativistic tearing instability.

457: The maximum eigen growth rates of

458: the relativistic tearing instability

459: are presented with the dashed line in Figure \ref{fig:rec_vs_dki_by}.

460: Both growth rates are in agreement. 

461: It seems that the growth rates are

462: relatively insensitive to the guide field.

463: In the extreme guide field case ($|B_y/B_0|=5.0$),

464: the growth rate is still the half of that of the antiparallel case.

465: This tendency is consistent with

466: reconnection studies in ion-electron plasmas;

467: an analysis \citep{bri95} and PIC simulations \citep{hori97}

468: reported that the linear growth rate decreases by a factor of 2 or 3

469: under strong guide field conditions (up to $|B_y/B_0| = 4$).

470: It was also reported that

471: the guide field reduces the quasi-steady reconnection rate

472: by a factor of $\sim 2$

473: in PIC simulations and Hall MHD simulations \citep{prit04,huba05}.

474: 

475: The gray lines in Figure \ref{fig:rec_vs_dki_by} show

476: the growth rates of the RDKI

477: (gray line by PIC simulation; gray dashed line by the theory). 

478: The RDKI is rather sensitive to the guide field.

479: In PIC simulations,

480: the RDKI spontaneously arises from the thermal noise

481: before $t/\tau_c=100$

482: under no/weak guide field conditions

483: (Runs D3 and D3-A; $|B_y/B_0|=0,0.25$).

484: However,

485: we could not observe any sign of the RDKI before $t/\tau_c=200$,

486: when the guide field is strong enough, $|B_y/B_0| \ge 0.5$.

487: We also failed to obtain the unstable eigen solutions

488: when guide field amplitude exceeds a threshold;

489: $|B_y/B_0| \gtrsim (0.45-0.5)$.

490: In the physical sense, 

491: since the wavevector is parallel to the guide field lines,

492: the magnetic tension of the guide field prevents the RDKI from growing.

493: Previous ion-electron work \citep{prit04} reported

494: that the guide field $B_{y}/B_0=0.5$ is

495: sufficient to stabilize the kink modes.

496: 

497: In summary, in two-dimensional linear regime,

498: the RDKI grows faster than

499: the reconnection or the relativistic tearing instability

500: under no/weak guide field conditions.

501: However, the situation changes

502: when we introduce a finite amount of the guide field.

503: The reconnection is rather insensitive to the guide field,

504: but the RDKI is suddenly stabilized by the guide field.

505: Therefore, reconnection will dominate

506: in the guide field conditions of $|B_{y}/B_0| \gtrsim 0.5$.

507: 

508: \subsection{2D Guide Field Reconnection}

509: \label{sec:2DPIC}

510: 

511: Next, we study the nonlinear development of

512: the relativistic guide field reconnection.

513: Among several simulation runs,

514: we present the results of run R3-C ($|B_y/B_0|=1.5$),

515: in which the guide field features are well developed.

516: 

517: We force magnetic reconnection around the center

518: and then reconnection quickly breaks up.

519: Figure \ref{fig:t100} shows snapshots at $t/\tau_c=100$,

520: where $\tau_c=\lambda/c$ is the light transit time.

521: Figure \ref{fig:t100}\textit{a} shows plasma density,

522: average flow and magnetic field lines.

523: The maximum outflow speed is $\sim 0.37c$ along the $x$-axis and

524: it is substantially slower than the antiparallel case of $\sim (0.7$-$0.8)c$.

525: The composition of plasma flow depends on the quadrants in the reconnection region;

526: electron flow into the $(+x,-y)$ direction in the $(+x,+z)$ quadrants,

527: positron flow into the $(-x,+y)$ direction in the $(-x,+z)$ quadrants, 

528: electron flow into the $(-x,-y)$ direction in the $(+x,-z)$ quadrants

529: and

530: positron flow into the $(+x,+y)$ direction in the $(+x,-z)$ quadrants

531: are dominant. 

532: The $X$-type layer along the separatrix region is also characteristic.

533: Figure \ref{fig:t100}\textit{b} shows the $y$-component of the positron current.

534: The current is strong around the $O$-type region,

535: while it is weak or reversed in the very center of the $O$-type region,

536: where the plasma density is at a maximum.

537: Around the reconnecting region,

538: there is a broadened structure of weak $y$-current.

539: The most characteristic thing is

540: the inclined current layer

541: along the lower side of the separatrix,

542: which is indicated by the dashed lines.

543: This inclined layer corresponds to one of the $X$-type density regions

544: in Figure \ref{fig:t100}\textit{a}.

545: The electron current structure is upside down;

546: it is slightly inclined in the counter-clockwise direction and

547: it is on the upper side of the separatrix.

548: In ion-electron plasmas,

549: several simulations of guide field reconnection reported

550: an inclined electron current layer \citep{hori97,drake03,hesse04,prit04}

551: near the $X$-type region.

552: However, the ion current layer is hardly recognized

553: due to the ion's large spatial scale.

554: In our simulation, we observe

555: both current layers for positrons and electrons.

556: Positrons are driven by the reconnection electric field $E_y$

557: around the $X$-type region,

558: and then

559: they tend to escape in the lower right (or upper left) direction

560: along 3D reconnected field lines.

561: Similarly, electrons tend to escape from the $X$-type region;

562: they tend to escape in the upper right (or lower left) direction.

563: Therefore, there is charge non-neutrality

564: around the outflow region (Fig. \ref{fig:t100}\textit{c}).

565: Positrons are localized in the lower right side of the reconnection region.

566: %

567: Figure \ref{fig:t100}\textit{d} shows the reconnection electric field $E_y$.

568: Its typical amplitude is $E_y/B_0 \sim (0.05-0.1)$

569: around the $X$-type region and

570: it is substantially weaker than antiparallel case; $E_y/B_0 \sim (0.2-0.3)$.

571: The $E_x$ component is comparable or slightly larger than $E_y$,

572: but consistent with plasma inflows.

573: Its typical value is $E_x/B_0 \sim 0.1$ in the upper inflow region

574: and $E_x/B_0 \sim -0.1$ in the lower inflow region.

575: The vertical electric field is

576: substantially large: $E_z/B_0 \sim 0.7$ in Figure \ref{fig:t100}\textit{e}.

577: It is consistent with the outflow jet and guide magnetic field $B_y$,

578: which is compressed around the $O$-type region.

579: The non-neutral charge distribution is consistent

580: with the $E_z$ structure, too.

581: 

582: Figure \ref{fig:espec} shows

583: the energy spectra over the main simulation domain.

584: As reconnection evolves, particle acceleration takes place

585: and then the high-energy tail continues to grow.

586: The late-time spectrum at $t/\tau_c=200$ is quasi-stable.

587: Although it is difficult to discuss the spectral index

588: in such a temporally and spatially limited system,

589: the power-law index is $\sim -2.9$ in the last stage.

590: 

591: In order to study particle acceleration,

592: we track the trajectories of the highest energy positrons,

593: whose energy exceeds $40 mc^2$ at $t/\tau_c=100$

594: (indicated by gray shading in Fig. \ref{fig:espec}).

595: Over the entire simulation domain

596: they are only found along the inclined current layer,

597: both in the upper left side and

598: in the lower right side in the reconnection region.

599: The high-energy electrons are found in the other current layer. 

600: Gray points in Figure \ref{fig:HE}\textit{a} show

601: spatial distribution of selected positrons

602: around the lower right current layer.

603: We pick up two typical trajectories

604: from $t/\tau_c=0$ to $t/\tau_c=200$,

605: which are indicated by solid lines in Figure \ref{fig:HE}.

606: We call them positron A and positron B, respectively.

607: The diamond or circle signs show their positions at $t/\tau_c=100$. 

608: %

609: Positron A belongs to the majority.

610: Starting from the left outside the presented region,

611: the positron visits the $X$-type region

612: and then escapes toward the lower right,

613: gyrating around the reconnected field lines.

614: The path is along the inclined positron current layer.

615: While it travels a long distance in the $y$-direction,

616: the particle mainly gains its energy

617: from the reconnection electric field $E_y$ near the $X$-type region.

618: Notice that the scaling of the $y$-direction is

619: extremely larger than the $x$- and $z$-directions in Figure \ref{fig:HE}\textit{b}. 

620: The energy gained from $E_x$ and $E_z$ is substantially low. 

621: %Since particles are threaded by the guide field,

622: %most of particle energy is stored in $y$-momentum;

623: %typical $y$-momentum $\langle p_y \rangle/ \langle mc^2 \rangle$ exceeds 20

624: %in the inclined current layer.

625: %This is substantially larger than that of the current sheet in the antiparallel case. 

626: % 

627: Positron B is a rare sample.

628: It starts near the center

629: and then it moves into the $X$-type region.

630: As reconnection goes on,

631: it travels in the $y$-direction

632: along the $X$ line in the $x$-$z$ plane

633: because it is trapped by the guide field $B_y$ around the $X$ line.

634: At $t/\tau_c=200$, it still rotates around the $X$ line

635: and its energy increases from $40.8 mc^2 (t/\tau_c=100)$

636: to $75.5 mc^2 (t/\tau_c=200)$.

637: The acceleration continues

638: until it finds a way to escape toward

639: the upper left or the lower right.

640: 

641: In the antiparallel reconnection,

642: the reconnection electric field $E_y$ is a main player

643: of particle acceleration, too.

644: Note that the reconnection electric field is

645: perpendicular to the magnetic field lines

646: in the antiparallel case, $\vec{E} \cdot \vec{B} \sim 0$.

647: Therefore, particle acceleration occurs

648: in the acceleration region

649: where the frozen-in condition is never satisfied

650: ($|\vec{E}| > |\vec{B}|$; \citet{zeni01}), 

651: or in the piled-up regions where

652: the electro-magnetic fields have strong peaks (ZH07). 

653: In the present case of guide field reconnection,

654: particle acceleration occurs

655: because $\vec{E} \cdot \vec{B} \ne 0$ around the reconnection region.

656: The guide field $B_y$ traps the particles,

657: preventing them from escaping in the $x$- or $z$-directions,

658: so that they are accelerated by $E_y$.

659: The particle acceleration occurs

660: even when they gyrate around the magnetic field lines,

661: because it relies on the parallel electric field $E_{\parallel}$.

662: 

663: We have also compared the nonlinear evolutions of

664: two-dimensional reconnection runs with various guide field amplitudes. 

665: Generally speaking, as the guide field increases,

666: signatures of guide field reconnection become more apparent;

667: electron/positron flows become separated, and then

668: the inclined current layers and charge separation structure appear.

669: The reconnection electric field also becomes weaker

670: because outflow jets becomes slower.

671: In addition, in spontaneous or nearly spontaneous cases,

672: it takes a longer time until the nonlinear reconnection process breaks up,

673: as the stronger guide field is imposed. 

674: 

675: Figure \ref{fig:3d_nonth} compares

676: the acceleration efficiency in various reconnection runs.

677: The ratio of the nonthermal kinetic energy to

678: the total kinetic energy are presented as a function of time.

679: The ratio is calculated from the energy spectra

680: in the same way as ZH07 (Appendix A in ZH07);

681: we assume

682: an equivalent energy spectra of the relativistic Maxwellian distribution,

683: which carries the same amount of kinetic energy

684: as the energy spectra in simulations,

685: and then we obtain the nonthermal kinetic energy

686: by subtracting the equivalent thermal spectra

687: from the simulation spectra in their high-energy tail.

688: This method may underestimate the amount of the nonthermal energy,

689: but we can quantitatively compare the acceleration efficiency in a simple way.

690: %Notice that $\varepsilon_2$ is the

691: In Figure \ref{fig:3d_nonth},

692: the simulation time for run R3-A ($|B_y/B_0|=0.5$),

693: in which the trigger force is slightly weak,

694: is re-arranged by $\Delta t=35\tau_c$

695: so that we can directly compare it to the other cases. 

696: Here we focus on the two-dimensional (2D) reconnection runs;

697: we discuss the 3D results later again in \S 5.2. 

698: %

699: Roughly speaking,

700: the acceleration efficiency seems to

701: decrease as the guide field increases.

702: The antiparallel run (run R3) is

703: the most efficient accelerator among the three runs.

704: The maximum accelerated energy is

705: also a function of the guide field

706: during the similar timescale

707: (See Table \ref{table}).

708: This is due to its strong reconnection electric field

709: $E_y \sim 0.3B_0$,

710: and the reconnection electric field $E_y$ becomes weaker

711: in the guide field cases.

712: In run R3-C,

713: the ``acceleration region'' nearly smears out

714: due to the weak reconnection field $E_y \sim 0.1B_0$.

715: On the other hand,

716: the trapping effect of the guide field

717: contributes to the acceleration efficiency.

718: Even though particle acceleration by $E_{\perp}$ becomes less active,

719: parallel acceleration by $E_{\parallel}$ works instead.

720: Thus, guide field reconnection (run R3-C) still produces

721: a substantial amount of nonthermal energy $\sim 15$\%.

722: Considering the limitation of our simple method,

723: more kinetic energy will be carried by nonthermal particles. 

724: In general, guide field reconnection seems to be

725: an efficient particle accelerator,

726: which releases more than 15\%-30\% of the plasma energy

727: into the nonthermal kinetic energy.

728: 

729: %***

730: %In this stage, multiple reconnecting region appears

731: %due to the spontaneous growth of the tearing mode,

732: %and then the accelerated particles

733: %start to enter the neighboring reconnection regions.

734: %Probably the presence of multiple reconnection regions

735: %improves the acceleration efficiency in the moderate guide field case,

736: %but it is out of the scope of the present paper.

737: In run R3-A, we notice that

738: the secondary particle acceleration takes place

739: in the very late stage (at $t/\tau_c \gtrsim 250$),

740: and then particles are accelerated up to $\gamma\sim 200$.

741: The spectral index evolves to a harder value of $\sim -2.2$

742: in the very late stage (at $t/\tau_c=400$).

743: Since periodic boundary effects arise in that late stage,

744: particle acceleration may be enhanced

745: due to interactions with the multiple reconnection structures.

746: The secondary acceleration is only found

747: in the weak guide field case (run R3-A),

748: and so its long-term evolution needs further investigation. 

749: 

750: When we compare runs R3-D and R3-C ($|B_y/B_0|=1.5$),

751: the system evolution in run R3-D

752: delays by approximately $145\tau_c$-$150\tau_c$

753: in terms of the global energy distribution.

754: In run R3-D, we observe six tearing islands around $t/\tau_c=200$,

755: and then they start to collide with each other until

756: one reconnection region dominates around $t/\tau_c=300$.

757: The $O$-type islands often have onion-ring density structure,

758: and the $X$-type current layers connect to the outermost rings.

759: However, the global energy evolution and

760: the acceleration efficiency are almost the same.

761: They seem to be rather insensitive to the initial evolution

762: whether the system evolves from forced single reconnection

763: or from multiple tearing islands,

764: in the moderate guide field cases of $|B_y/B_0|\sim 1.5$.

765: 

766: 

767: \section{3D EVOLUTION WITH NO GUIDE FIELD}

768: \label{sec:3DA}

769: 

770: In this section, we study

771: the 3D evolution of the current sheet

772: in antiparallel configuration. 

773: We look at the simulation results of run 3D-A.

774: As stated in \S 2,

775: we imposed small external electric fields around the center,

776: but the trigger fields are not strong enough to

777: force reconnection.

778: 

779: We briefly review the linear evolution,

780: which was discussed in our previous paper (ZH05b).

781: Figure \ref{fig:3D-A}\textit{a} is a snapshot at $t/\tau_c=80$.

782: The current sheet is between the two gray surfaces and

783: the plasma density at the neutral plane ($z=0$) is

784: projected into the bottom wall,

785: with color shading from black (empty) to red (dense; $n=1.2 n_0$). 

786: Figure \ref{fig:3DEy} presents

787: the $E_y$ structure at $t/\tau_c=80$.

788: The red regions have the positive polarity of $E_y > 0$,

789: while the blue regions are negative: $E_y < 0$.

790: Apparently these profiles exhibit

791: a quasi 2D evolution of the RDKI.

792: One can see the polarity change along the $x$-axis

793: in the $x$-$z$ plane (the right surfaces in Fig. \ref{fig:3DEy}),

794: but this is due to small $y$-displacement of the structure.

795: The wave-number of the RDKI is $k_y\lambda \sim 0.74$ (mode 3),

796: while mode 4 is observed in the relevant 2D run D3.

797: However, both mode 3 and mode 4 are reasonable with the linear theories.

798: The observed growth rate $\tau_c\omega_i=0.06$ is

799: slightly slower than the expected rate $\tau_c\omega_i\sim0.1$,

800: but it is still faster than that of the relativistic tearing instability.

801: 

802: Figures \ref{fig:3D-A}\textit{b} and \ref{fig:3D-A}\textit{c}

803: show the nonlinear development of the current sheets

804: at $t/\tau_c=110$ and $140$. 

805: The current sheet is strongly folded at $t/\tau_c=110$,

806: and then its wave fronts start to collide each other.

807: At $t/\tau_c=110$,

808: the gray density surfaces are reset to $n=1/3n_0$,

809: so that we can see the low-density hole around the center. 

810: %

811: We discover that magnetic reconnection takes place

812: in this central hole.

813: Figure \ref{fig:3D-A}\textit{d} is a zoomed-in view

814: around the neutral plane at $t/\tau_c=110$. 

815: The gray lines are magnetic field lines,

816: traced from the following six start points:

817: $(x,y,z) = (-5,0,-0.5),(-5,2,-0.5) \cdots (-5,10,-0.5)$ in units of $\lambda$.

818: Magnetic field lines are reconnected around the central low-density region.

819: Contrary to the 2D reconnection in the $x$-$z$ plane,

820: reconnection structure involves the $x$-$y$ plane,

821: because the current sheet is highly folded by the RDKI.

822: Plasma inflows mainly come from the $\pm y$-directions,

823: and then

824: outflows flow into the $\pm x$-direction.

825: Inflows from the $\pm z$-direction are difficult to identify.

826: Interestingly, the central reconnection region contains

827: multiple field reversals between the folded current sheets.

828: Since plasmas are expelled away from the reconnection region,

829: there is a low-density hole around the central region.

830: The reason reconnection takes place in the center

831: may be the initial trigger impact.

832: Around the central region,

833: the RDKI is invoked earlier than other locations, 

834: %probably due to the triggering perturbation,

835: and then the folded structure first appears there.

836: Reconnection occurs in such a well-developed folded region. 

837: In this simulation, the central reconnection region is

838: finally overwhelmed by the current sheet corruption,

839: and then the system evolves into the turbulent state

840: at $t/\tau_c=140$.

841: This is consistent with the 2D RDKI picture.

842: The $x$-$y$ cross sections of the plasma density structure in the upper wall

843: indicate the plasma mixing across the double periodic boundaries.

844: Eventually, most energy in the system is

845: converted into plasma heat,

846: 2 times hotter than the original state.

847: Particle acceleration by the RDKI or reconnection is negligible.

848: 

849: 

850: \section{3D EVOLUTION WITH GUIDE FIELD}

851: 

852: Next, we study a 3D evolution of

853: the current sheet with a guide field in run 3D-B.

854: The guide field amplitude is $|B_y/B_0|=0.5$.

855: All other conditions are the same as

856: those of run 3D-A in \S \ref{sec:3DA}. 

857: Throughout the simulation run ($0 \le t/\tau_c \le 220$),

858: the total energy is conserved within an error of 0.6\%. 

859: %Snapshots at four different stages

860: %are presented in the panels in Figure \ref{fig:3D-B}:

861: %(\textit{a}) $t/\tau_c=120$, (\textit{b}) $170$,

862: %(\textit{c}) $200$ and (\textit{d}) $220$]

863: %Gray surfaces show the plasma density.

864: 

865: \subsection{Linear Evolution: Relativistic Drift Sausage/Kink Tearing Instability}

866: 

867: First, we study the linear structure of the mode in detail.

868: Compared with run 3D-A,

869: the current sheet seems to be more stable in the early stage.

870: It takes $t/\tau_c \sim 100$

871: until we observe visible changes in the current sheet.

872: At $t/\tau_c=120$, we observe a purely growing flute like mode

873: in the oblique direction on the upper side of the current sheet

874: in Figure \ref{fig:3D-B}\textit{a}.

875: We think

876: this is a generalized mode between

877: the relativistic tearing instability and

878: the RDKI-type instabilities (ZH05b). 

879: The mode is $(1,1)$ or its wavevector is $\vec{k}_1 \lambda = (0.25,0.25)$. 

880: The two panels in Figure \ref{fig:cut} show

881: the plasma density slices at $z=\pm\lambda$.

882: We see the structure with $\vec{k}_1$

883: on the upper side of the current sheet at $z=\lambda$,

884: while we observe another oblique mode

885: $(1,-1): \vec{k}_2 \lambda= (0.25,-0.25)$

886: on the lower side of the current sheet at $z=-\lambda$.

887: On both sides,

888: it seems that the oblique lines are slightly disconnected

889: around $(x,y)\sim(\pm 12.8\lambda, 0)$.

890: Probably the fastest growing modes do not match the system size,

891: and then they adjust into the nearest periodic mode $(1,\pm 1)$.

892: Therefore, it takes a longer time before we observe the visible change.

893: The most powerful Fourier modes are $(1,\pm1)$ in the simulation data,

894: and Figure \ref{fig:sim} shows $z$-profiles of

895: the density perturbation of the two modes. 

896: The amplitude of the two modes is nearly same,

897: but their peaks are in the other side of the current sheet.

898: We confirmed that these oblique perturbations are

899: purely growing unstable modes.

900: Carrying out a supplemental 2D PIC simulation

901: under the same conditions in that particular angle, 

902: % (Run 2D-C in Tab. \ref{table}),

903: we observe similar a similar asymmetric profile, mode $(2,\pm2)$,

904: while we obtain mode $(1,\pm 1)$ in run 3D-B.

905: 

906: In order to study the nature of the instability,

907: we have also calculated eigen profiles of

908: the oblique instability in the current sheet

909: by using the relativistic two-fluid theory.

910: We assume the perturbation of

911: $\delta f \propto f(z) \exp(ik_x x + ik_y y + \omega_i t)$

912: as introduced in \S 3.1, and then

913: we consider the oblique case of $k_x,k_y \ne 0$.

914: % 

915: In Figure \ref{fig:rdkti_eigen},

916: the eigen functions for $k_x \lambda = 0.25$,

917: $k_y \lambda = 0.25$ and $B_y/B_0 = -0.5$

918: are presented as functions of $z$.

919: Perturbed magnetic field profiles ($\delta B_{x}, \delta B_{y}, i\delta B_{z}$)

920: and electric field profiles ($\delta E_{x}, \delta E_{y}, i\delta E_{z}$) employ

921: the same normalization.

922: The other three panels display

923: the current profiles ($\delta J_{x}, \delta J_{y}, i\delta J_{z}$),

924: the bulk velocity perturbations ($i\delta V_{+x}, i\delta V_{+y}, \delta V_{+z}$,

925: where $\delta \vec{V}_{+} = \delta \vec{v}_{p} + \delta \vec{v}_{e}$),

926: and density profiles

927: ($\delta D_{+} = \delta d_{p} + \delta d_{e}$ and $\delta D_{-} = \delta d_{p} - \delta d_{e}$)

928: are normalized by their maximum values.

929: The current profiles ($\delta J_{x}, \delta J_{y}, i\delta J_{z}$)

930: are calculated from other perturbed values: $\delta D_{+}, \delta\vec{V}_{-}$,

931: where $\delta \vec{V}_{-} = \delta \vec{v}_{p} - \delta \vec{v}_{e}$.

932: These profiles are consistent with the simulation data.

933: For example, the density perturbation $\delta D_{+}$

934: in Figure \ref{fig:rdkti_eigen}

935: is in excellent agreement with the relevant perturbation

936: in simulation data in Figure \ref{fig:sim}.

937: In general, the perturbation profiles are not

938: symmetric nor anti-symmetric with $z$.

939: As observed in $\delta B_x$ or $\delta D_{+}$,

940: the mode seems to be substantially localized in the upper half of $z > 0$.

941: The other mode ($\vec{k}_2$; $k_x \lambda = 0.25$, $k_y \lambda = -0.25$)

942: has an opposite structure;

943: it is localized in the lower half of $z < 0$.

944: 

945: Figure \ref{fig:illust} schematically illustrates

946: the structure of the oblique mode ($\vec{k}_1$).

947: Although we called this mode

948: relativistic drift-kink tearing instability (RDKTI) in ZH05b, 

949: we find that

950: the structure is rather similar to

951: that of the relativistic drift sausage instability (RDSI),

952: a cousin mode of the RDKI.

953: Therefore, we shall rename it

954: relativistic drift-sausage tearing instability (RDSTI).

955: For simplicity, we assume that the perturbed structure

956: is localized in the upper half of the current sheet in the top panel

957: in Figure \ref{fig:illust}. 

958: In addition,

959: the two conventional instabilities in antiparallel configuration;

960: the RDSI/RDKI and the relativistic tearing instability

961: are illustrated in the bottom panels

962: of Figure \ref{fig:illust}. 

963: The small arrows represent the perturbed vectors.

964: The phase of the linear perturbation is set to zero at the origin.

965: Therefore, the dashed line in the neutral sheet is phase $\pi$ and

966: the imaginary perturbations (phases ${\pi}/{2}$ and ${3\pi}/{2}$)

967: are found between the origin and the dashed line.

968: The dashed line (phase $\pi$) in the neutral plane

969: is equivalent to the $X$-line of the tearing mode.

970: If we look at the eigen profiles

971: in the upper vicinity of the neutral plane ($z \gtrsim 0$),

972: plasma bulk inflow into the $X$ line ($-\delta v_z < 0$),

973: the diverging outflows in $i\delta v_{x,y}$,

974: the perpendicular magnetic field $i\delta B_z$,

975: and the current enhancement $J_y$ are

976: all consistent with the those of the tearing mode.

977: In addition, roughly speaking,

978: the density $\delta D_+$ is positive

979: in the $O$-like region and

980: negative in the $X$-like region.

981: The reconnection electric field $E_y$ is positive around the $X$ line,

982: although it not strong enough to penetrate into the lower half domain.

983: There is also $E_z$ structure in $i\delta E_z$,

984: which is between the $X$- and $O$-type regions.

985: The $E_z$ structure and the relevant charge separation in $i\delta D_-$

986: are signatures of the guide field tearing mode or guide field reconnection,

987: as presented in Figures \ref{fig:t100}\textit{c} and \ref{fig:t100}\textit{e}.

988: 

989: Next, we compare the perturbed structure with

990: that of the RDSI/RDKI in the $y$-$z$ plane.

991: We compare the eigen profiles with

992: those of the RDKI and the RDSI,

993: which are presented in Figures 14 and 21 in ZH07. 

994: In both cases of the RDSTI and the RDSI,

995: the two-peak structure in density profile $\delta D_{+}$

996: stands for sausage-type modulation,

997: while reverse peaks stand for kink-type modulation.

998: The charge separation and $E_z$ structure of the RDSTI

999: is well connected to the those of the RDSI. 

1000: %

1001: One specific feature of the RDSTI is

1002: the perturbed plasma flow structure.

1003: The flow direction changes as a function of $z$,

1004: and then the flow is parallel to

1005: the background magnetic field in the topmost layer.

1006: Overall, the structure of the RDSTI is complicated,

1007: but it is well consistent with the relevant 2D instabilities.

1008: The RDKTI has almost same structure as the RDSTI,

1009: except that perturbation is kink-like

1010: in the lower side of the current sheet in the $y$-$z$ plane.

1011: 

1012: %In the lower half of the current sheet,

1013: %two 2D structures no longer connect each other,

1014: %primary because $0$th order magnetic field contains

1015: %both reverse ($B_x$) and constant ($B_y$) components.

1016: %This also represents the magnetic tension effects.

1017: 

1018: We investigate the eigen modes over various parameters:

1019: the guide field amplitude $B_y/B_0=0, -0.5, -1, -1.5$ and

1020: the wavevectors of $(0 \le k_x\lambda \le 1, 0 \le k_y\lambda \le 1)$.

1021: Their fastest growth rates ($\tau_c\omega_i$) are 

1022: presented in contour maps in top panels in Figure \ref{fig:map}.

1023: We take $\Delta k_{x,y}\lambda = 0.05$ so that

1024: $21^{2}$ parameters per map are presented. 

1025: All of the modes are purely growing.

1026: Due to the mathematical symmetry,

1027: we can obtain eigen modes and growth rates for $(k_x, k_y)$

1028: from their counterparts in $(|k_x|, |k_y|)$. 

1029: We repeat the validity of our linear theory again.

1030: As discussed in \S 3.1,

1031: our two-fluid approximation will be valid

1032: only in the long-wavelength range

1033: (e.g. $|k_y\lambda| \lesssim 0.7$ for the RDKI/RDSI).

1034: Therefore, the results in Figure \ref{fig:map}

1035: will be reliable

1036: only when $|k\lambda| \lesssim 0.7$.

1037: The bottom panels in Figure \ref{fig:map}

1038: show the type of the fastest eigen mode.

1039: By comparing the peak-structure in their density profiles,

1040: we classify the obtained fastest eigen modes into

1041: the following three types:

1042: kink type (A in Fig. \ref{fig:map}),

1043: sausage type (B), and neither of them (C).

1044: The tearing mode is classified in sausage-type modes,

1045: because its density perturbation is

1046: symmetric with the neutral plane ($z=0$).

1047: Notice that both

1048: kink-type mode and sausage-type mode coexist

1049: in the same point,

1050: and that we discuss the type of the fastest mode

1051: in the specific parameter range.

1052: It seems that the RDKI/RDKTI is faster

1053: only in the shorter wavelength region along the $k_y$-axis.

1054: In the other region, the RDSI/RDSTI replaces the RDKI/RDKTI.

1055: 

1056: In the guide field case of $B_y/B_0=-0.5$,

1057: the RDKI/RDSI along the $k_y$-axis is stabilized by

1058: the magnetic tension of the guide field

1059: as discussed in \S 3.

1060: The oblique RDSTI modes,

1061: which are along the background magnetic field lines,

1062: survive instead. 

1063: Some oblique modes grow even faster in the guide field case

1064: because they are weakly stabilized

1065: by the magnetic tension in antiparallel case.

1066: As the guide field becomes stronger, 

1067: the most dominant modes change direction,

1068: in accordance with the background magnetic field lines.

1069: Since the RDKI/RDSI component of the instability

1070: is driven by the $\vec{k}$-aligned component of the current,

1071: the growth rates of the oblique modes decrease.

1072: On the other hand, growth rates of the tearing mode component

1073: seems to be rather insensitive to $|B_y/B_0|$. 

1074: % 

1075: In the strong guide field case of $|B_y/B_0|\gtrsim 1$,

1076: we can no longer classify some oblique modes

1077: to kink-type or sausage-type modes (C in Fig. \ref{fig:map}).

1078: These modes are highly localized on one side of the current sheet,

1079: and their perturbation is very small on the other side of the current sheet.

1080: 

1081: 

1082: \subsection{Nonlinear Evolution}

1083: 

1084: After the linear stage,

1085: the current sheet becomes very thin at several points,

1086: where two RDSTI/RDKTI modes compress

1087: the current sheet from the upper and lower sides.

1088: The thinning point evolves into

1089: a big plasma hole at the center of the simulation box

1090: at $t/\tau_c=170$ in Figure \ref{fig:3D-B}\textit{b}.

1091: The secondary magnetic reconnection takes place there. 

1092: Once reconnection breaks up, it continues to grow.

1093: Figure \ref{fig:3Dcut} shows a 2D slice at $y=0$ at $t/\tau_c=170$.

1094: The upper panel shows the typical reconnection structure.

1095: The outflow velocity is up to $0.6c$,

1096: which is between the antiparallel case ($0.8c$) and

1097: the strong guide field case ($0.4c$) of $B_y/B_0=-1.5$.

1098: The typical inflow velocity is $\sim 0.1c$. 

1099: The current sheet still looks similar to

1100: a Sweet-Parker current sheet, but

1101: careful observation shows

1102: several signatures of the guide field reconnection.

1103: The bottom panel shows the positron current structure in color and the $E_z$ structure in contour, respectively.

1104: The positron current layer is slightly inclined in a clockwise direction.

1105: However, the electron current layer is

1106: inclined in a counter clockwise direction.

1107: So, the two current layers coexist in

1108: a Sweet-Parker-like current sheet.

1109: In addition, the vertical electric field $E_z$ is not negligible.

1110: Its amplitude is $\sim 0.4B_0$ in the right side

1111: and $\sim -0.4B_0$ in the left side,

1112: while the typical reconnection electric field is $E_y\sim 0.17B_0$.

1113: The $x$ component of the inflow electric field is $E_x \sim \pm0.05B_0$. 

1114: These signatures are almost same in the relevant 2D run: run 3D-A.

1115: The width (in $y$) of the reconnection region seems to be limited

1116: by the scale of the system or the scale length of the oblique modes. 

1117: %

1118: Figure \ref{fig:3D-B}\textit{c} shows a snapshot at $t/\tau_c=200$.

1119: The oblique bridges are blown away from the neutral sheet

1120: due to the intense plasma pressure,

1121: and then the dense points around $(x, y) \sim (0, \pm 12.8 \lambda )$

1122: are no longer observed.

1123: In addition, plasmas are drawn into the central reconnecting point,

1124: and there is a plasma hole along the $X$ line: $x=0,z=0$.

1125: The typical speed of reconnection jets is up to $\sim 0.71c$.

1126: Figure \ref{fig:3D-B}\textit{d} is

1127: the last snapshot of our simulation at $t/\tau_c=220$.

1128: The system structure is highly turbulent,

1129: but we observe the filament-like structures in the $y$-directions.

1130: Considering that reconnection dissipates

1131: the field energy of antiparallel magnetic fields,

1132: it is reasonable that we observe structures

1133: which are threaded by the guide field $B_y$.

1134: 

1135: After magnetic reconnection occurs,

1136: the magnetic energy tends to be converted

1137: to the nonthermal part of the plasma kinetic energy,

1138: due to the particle acceleration around the reconnection region.

1139: Figure \ref{fig:espec3} compares

1140: the energy spectra in the system

1141: for two 3D runs and the relevant 2D runs of guide field reconnection.

1142: The initial state of two 3D runs is almost similar to

1143: the spectrum of run 3D-B at $t/\tau_c=140$

1144: (Fig. \ref{fig:espec3}, \textit{dotted line}).

1145: In the case of run 3D-A (\textit{bold line}),

1146: plasma energy is converted into plasma heat,

1147: due to magnetic dissipation by the RDKI.

1148: The nonthermal tail of the spectrum ($\varepsilon \gtrsim 30mc^2$)

1149: is slightly enhanced due to the particle acceleration.

1150: The footpoint of the nonthermal tail is approximately

1151: same as that of the 2D RDKI case,

1152: but the tail itself is not as apparent as in the 2D case. 

1153: Due to the irregularity along the $x$-direction or other 3D effects,

1154: particle acceleration by the RDKI

1155: works less effectively than the ideal 2D case. 

1156: In the case of run 3D-B,

1157: the energy spectrum is almost unchanged

1158: until relativistic reconnection breaks up.

1159: At $t/\tau_c = 140$

1160: (Fig. \ref{fig:espec3}, \textit{dotted line}),

1161: plasmas are slightly heated by 3\%

1162: from the initial state.

1163: In the late stage of $t/\tau_c=220$

1164: (Fig. \ref{fig:espec3}, \textit{dash-dotted line}),

1165: the nonthermal tail is enhanced

1166: due to the particle acceleration by reconnection.

1167: The nonthermal slope is well-described

1168: by the power law with the index of $-2.8$

1169: in a range of $8<\varepsilon/mc^2<20$.

1170: This is nearly same as the spectral index in the 2D runs.

1171: The power-law spectral index is $\sim -2.7$ at $t/\tau_c=200$ in run R3-A,

1172: and $\sim -2.9$ at $t/\tau_c=200$ in run R3-C. 

1173: 

1174: The time history of the nonthermal ratio parameter,

1175: the ratio of plasma nonthermal energy to plasma kinetic energy,

1176: is presented in Figure \ref{fig:3d_nonth}.

1177: In the case of run 3D-A (Fig. \ref{fig:3d_nonth}, \textit{thick line}),

1178: there is a small peak after $t/\tau_c=80$ 

1179: due to the $dc$ acceleration by the RDKI,

1180: but eventually the nonthermal ratio is less than 2\%.

1181: This is consistent with

1182: 2D simulations on the RDKI,

1183: which reports the nonthermal ratio of far less than $5\%$ (ZH07).

1184: However, in the case of run 3D-B

1185: (Fig. \ref{fig:3d_nonth}, \textit{thick dashed line}),

1186: shortly after the central reconnection region appears at $t/\tau_c=170$,

1187: more than $14\%$ of the kinetic energy consists of the nonthermal energy. 

1188: This ratio is approximately half of the value

1189: in the relevant 2D run

1190: (run R3-A; Fig. \ref{fig:3d_nonth}, \textit{thin gray line}).

1191: We think this is due to the limited volume of

1192: reconnection region in 3D configuration.

1193: For example, along the neutral line ($X=0$)

1194: at $t/\tau_c=200$ (Fig. \ref{fig:3D-B}c),

1195: reconnection and the relevant particle acceleration is active

1196: around the center $y \sim 0$,

1197: while we do not observe reconnection flow structure

1198: around $y/\lambda \sim \pm 12.8$.

1199: Since reconnection can utilize half of the system volume

1200: along the $y$ direction, the nonthermal ratio is small in run 3D-B,

1201: compared with the 2D counterpart (run R3-A).

1202: If we employ larger simulation box

1203: to reduce the periodic limitation in the $y$-direction,

1204: we may observe wider reconnection region and then

1205: the nonthermal ratio may increase.

1206: Anyway, run 3D-B more efficiently generates

1207: the nonthermal energy than the antiparallel case of run 3D-A,

1208: because underlying physical mechanism is completely different.

1209: Most of magnetic energy is dissipated into

1210: the thermal energy by the RDKI in the anti-parallel case (run 3D-A).

1211: However, in the guide field case (run 3D-B)

1212: a substantial amount of magnetic energy

1213: is dissipated into the nonthermal component of plasma kinetic energy,

1214: associated with magnetic reconnection.

1215: 

1216: 

1217: \section{DISCUSSION}

1218: 

1219: First, we discuss the system evolution in an antiparallel configuration.

1220: We are interested in which process dominates, the reconnection or the RDKI,

1221: because it greatly changes the energy distribution in the system;

1222: reconnection involves nonthermal particle acceleration,

1223: but the RDKI leads to plasma heating.

1224: Comparison of linear growth rates (ZH07) suggested

1225: that the RDKI dominates in the relativistic regime of $T/mc^2 \gtrsim 1$.

1226: Our results in \S 4 basically support this argument.

1227: Although we imposed the external trigger force,

1228: it was not strong enough to evoke reconnection

1229: before the RDKI modulates the current sheet.

1230: The $dc$ acceleration by relativistic reconnection is not likely to evolve;

1231: then magnetic dissipation and plasma heating by the RDKI

1232: would be the main signature of a relativistic current sheet. 

1233: However, we discovered small reconnection regions

1234: inside the folded current sheet structure.

1235: The reconnection generates a density hole around the center,

1236: and then the density hall may evolve into

1237: a global reconnection structure in the larger system.

1238: It is true that

1239: the RDKI grows faster than the reconnection

1240: in our relativistic regime of $T/mc^2 \sim 1$,

1241: but once reconnection is initiated,

1242: we do not know whether it overwhelms the outside RDKI structures.

1243: 

1244: Next, we discuss the guide field effect on the system evolution.

1245: We showed that the RDKI is completely stabilized by

1246: the finite amount of the guide field in \S 3.

1247: We do not yet understand

1248: how much guide field is necessary to stabilize the RDKI,

1249: but we expect that reconnection mode dominates again

1250: under the guide field conditions.

1251: In \S 5, we showed that the evolution is more complicated than expected.

1252: We considered the oblique instabilities (RDSTI/RDKTI)

1253: to understand the linear evolution of the current sheet.

1254: They grow in two oblique directions (e.g. $\vec{k}_1$ and $\vec{k}_2$),

1255: which can be interpreted as twin extensions of the conventional RDSI/RDKI.

1256: In the antiparallel case

1257: the kink-type mode dominates, and

1258: there were no signatures of sausage-type modes.

1259: In the guide field case,

1260: we found that the sausage-type branch (RDSTI) dominates in run 3D-B.

1261: In some sense this is quite reasonable,

1262: because both the RDSI and the tearing instability modulate the current sheets

1263: in a symmetric way with the neutral plane,

1264: while the RDKI modulates the current sheets in an asymmetrical way.

1265: On the other hand,

1266: since the RDSTI/RDKTI have asymmetric structure

1267: and since they always appear as twins in the guide field cases,

1268: it does not matter whether the oblique modes are sausage-like or kink-like.

1269: The important point is that

1270: the twin oblique modes,

1271: whose wave fronts are parallel to the twisted background magnetic fields,

1272: initiates magnetic reconnection.

1273: Since relativistic guide field reconnection

1274: involves particle acceleration,

1275: the guide field turns on the nonthermal particle acceleration

1276: in the 3D system. 

1277: When we impose a stronger guide field,

1278: one can see that an angle between two RDKTI/RDSTI branches

1279: ($\vec{k}_1$ and $\vec{k}_2$)

1280: becomes wider in accordance with the lobe magnetic field lines.

1281: The RDSTI/RDKTI slow down,

1282: but their growth rates are still faster than the relativistic tearing instability

1283: (Fig. \ref{fig:map}).

1284: Under the extreme guide field condition,

1285: since the dominant RDSTI/RDKTI modes are

1286: inclined to the $\pm x$-directions,

1287: coupling between two oblique waves may directly lead to

1288: quasi-2D growth of the tearing instability.

1289: 

1290: On the system size limitation,

1291: the late-time evolution of our 3D runs may be somewhat artificial,

1292: due to the periodic boundary effects.

1293: In run 3D-A, magnetic diffusion is enhanced by the plasma transport

1294: across the double periodic boundaries in $z$.

1295: In run 3D-B, the oblique mode (RDSTI) seems to be bounded by the system length.

1296: Since we see the remnant of the other modes (Fig. \ref{fig:cut}),

1297: we probably failed to observe the most unstable modes.

1298: However, as long as similar oblique modes dominate

1299: they will trigger the secondary magnetic reconnection

1300: in the similar way.

1301: If we set larger simulation box,

1302: do we observe multiple reconnection points

1303: in accordance with the spatial structure of the RDSTI/RDKTI?

1304: If so, how are particles accelerated in multiple reconnection regions?

1305: Or does the $X$ line extend and then

1306: does the system evolve into the 2D reconnection?

1307: Furthermore, recent and ongoing works on

1308: large-scale evolution of magnetic reconnection

1309: exhibit more dynamical behaviors than expected

1310: (e.g. plasmoid formation and collisions; \citet{dau07}).

1311: Large-scale, 3D evolution of

1312: relativistic pair plasma reconnection

1313: still remains an open issue,

1314: regardless of the presence of the guide field.

1315: Future simulations may reveal various dynamical behaviours

1316: beyond our linear and early nonlinear results.

1317: 

1318: Regarding the particle acceleration in the guide field reconnection,

1319: we studied particle acceleration

1320: by the parallel electric field $E_{\parallel}$ in the 2D case,

1321: and we observed similar acceleration signatures in the 3D case, too. 

1322: In general, as long as we investigated

1323: we observed nonthermal energy spectra with indexes of $-2 \sim -3$

1324: by magnetic reconnection, regardless of the amplitude of guide field,

1325: in both 2D and 3D cases.

1326: The upper limit of the accelerated energy remains to be solved,

1327: or it seems unlimited

1328: as long as reconnection continues to consume magnetic energy.

1329: We confirmed that maximum lepton energy easily exceeds

1330: the Lorentz factor of $80$-$200$ ($40$-$100$MeV; see Table \ref{table}).

1331: This will also be influenced by large-scale evolution of the system.

1332: 

1333: Let us briefly discuss potential astrophysical applications,

1334: based on physical insights from our results.

1335: In pulsar winds with relativistically hot pair plasmas

1336: \citep{coro90,lyu01,kirk03},

1337: it is unlikely that current sheets contain substantial amount of guide fields

1338: because magnetic field lines are highly striped or toroidal.

1339: Therefore, plasma heating by the RDKI is the most likely process,

1340: unless plasma temperature drops down to nonrelativistic one.

1341: On the contrary, in the AGN context,

1342: magnetic reconnection is quite likely

1343: to involve the guide field component.

1344: Several authors demonstrated

1345: electron acceleration due to the field-aligned electric field $E_{\parallel}$

1346: by means of test particle simulations \citep{schopper,nodes03}.

1347: We demonstrated that

1348: relativistic particle acceleration by $E_{\parallel}$

1349: in a fully self-consistent way,

1350: including feedbacks from accelerated electrons and other kinetic effects,

1351: and so our results provide a theoretical proof of

1352: the ultra relativistic particle acceleration in the MeV/GeV range,

1353: at least in electron-positron pair plasma reconnection. 

1354: %

1355: In the case of soft gamma repeaters,

1356: it is unclear how giant flares occur at magnetars.

1357: The present models discuss

1358: the crustal ``quakes'' \citep{thom95,thom01}

1359: or the flux tube corruption

1360: in analogy with solar flares and coronal mass ejections

1361: to trigger giant flares.

1362: In addition, the tearing instability

1363: in the force-free magnetar corona is proposed to explain

1364: the subsequent bursting activities \citep{lyut03a}.

1365: In these models, magnetic energy is stored in

1366: a magnetic spiral in the star core or the flux lopes,

1367: or coronal magnetic shear. 

1368: Therefore, once a flare occurs,

1369: it is likely that magnetic reconnection involves

1370: a magnetic shear or out-of-plane magnetic field.

1371: The guide field reconnection will occur, and 

1372: reconnection will be a yet another favorable source

1373: of nonthermal particles

1374: as well as ultra-relativistic shock fronts.

1375: 

1376: Recently, relativistic MHD models have been developed

1377: to investigate astrophysical plasma problems.

1378: Our results warn that an MHD approximation is no longer valid

1379: in the case of relativistic guide field reconnection. 

1380: As plasma outflow becomes an order of $c$,

1381: charge separation structure becomes apparent in the outflow region,

1382: and then it breaks down the charge neutral assumption of one-fluid theory.

1383: Instead, multi-fluid simulations, which deal with

1384: positively charged fluids and negatively charged fluids independently,

1385: are favorable to study with guide field reconnection problems.

1386: 

1387: Let us summarize this paper.

1388: First, we investigated how the guide field affects 2D instabilities;

1389: the RDKI is stabilized by a finite amount of the guide field,

1390: while the reconnection/tearing mode is rather insensitive to the guide field.

1391: Then, we studied the nonlinear evolution of

1392: relativistic guide field reconnection.

1393: Characteristic field structure and particle acceleration process

1394: were investigated.

1395: Next, we studied 3D evolution of the current sheet.

1396: As predicted by 2D studies,

1397: the RDKI dominates and dissipates the magnetic energy,

1398: but we also discovered that

1399: reconnection occurs inside the folded current sheet.

1400: Finally, we studied 3D evolution

1401: with a guide field condition.

1402: The properties of the oblique RDSTI/RDKTI mode and

1403: the nonlinear evolution of secondary reconnection is discussed,

1404: in association with the 2D counterparts.

1405: Due to the guide field, nonthermal particle acceleration,

1406: which generates power-law energy spectra with an index of $-2 \sim -3$,

1407: occurs in the relativistic pair plasmas.

1408: Our results show that the guide field reconnection is

1409: a favorable acceleration engine in high-energy astrophysical plasmas.

1410: 

1411: \begin{acknowledgments}

1412: The authors express their gratitude

1413: to T. Yokoyama, M. Hesse, I. Shinohara, P. L. Pritchett and Y. E. Nakagawa

1414: for helpful comments.

1415: The author is also grateful to the anonymous referee

1416: for helping to improve this manuscript.

1417: This work was supported by the facilitates of JAXA

1418: and the Solar-Terrestrial Environment Laboratory, Nagoya University.

1419: \end{acknowledgments}

1420: 

1421: \begin{thebibliography}{}

1422: 

1423: \bibitem[Bessho \& Bhattacharjee(2005)]{bessho05} N. Bessho and A. Bhattacharjee 2005, \prl, 95, 245001

1424: \bibitem[Bessho \& Bhattacharjee(2007)]{bessho07} N. Bessho and A. Bhattacharjee 2007, \pop, 14, 056503

1425: \bibitem[Birk et al.(2001)]{birk01} G. T. Birk, A. R. {Crusius-W{\"a}tzel} and H. {Lesch} 2001, \apj, 2001, 559, 96

1426: \bibitem[Birn et al.(2001)]{birn01} J. Birn, J. F. Drake, M. A. Shay, B. N. Rogers, R. E. Denton, M. Hesse, M. Kuznetsova, Z. W. Ma, A. Bhattacharjee, A. Otto and P. L. Pritchett 2001, \jgr, 106, 3715

1427: \bibitem[Blackman \& Field(1994)]{bf94b} E. G. Blackman and G. B. Field 1994, \prl, 72, 494

1428: \bibitem[Brittnacher et al.(1995)]{bri95} M. Brittnacher, K. B. Quest and H. Karimabadi 1995, \jgr, 100(A3), 3551

1429: %\bibitem[Bucciantini et al.(2006)]{buc06} N. Bucciantini, T. A. {Thompson}, J. {Arons}, E. {Quataert} and L. {Del Zanna} 2006, \mnras, 368, 1717

1430: \bibitem[Coroniti(1990)]{coro90} F. V. Coroniti 1990, \apj, 349, 538

1431: \bibitem[Daughton (1999)]{dau99} W. Daughton 1999, \jgr, 104, 28701

1432: %\bibitem[Daughton \& Karimabadi(2005)]{dau05} W. Daughton and H. Karimabadi 2005, \jgr, 110, A03217

1433: \bibitem[Daughton \& Karimabadi(2007)]{dau07} W. Daughton and H. Karimabadi 2007, \pop, 14, 072303

1434: \bibitem[di Matteo(1998)]{dimatteo} T.~{di Matteo} 1998, \mnras, 299, L15

1435: \bibitem[Drake \& Lee(1977)]{drake77} J. F. Drake and Y. C. Lee 1977, Physics of Fluids, 20, 1341

1436: \bibitem[Drake et al.(2003)]{drake03} J. F. Drake, M. Swisdak, C. Cattell, M. A. Shay, B. N. Rogers and A. Zeiler 2003, Science, 299, 873

1437: %\bibitem[Drake et al.(2005)]{drake05} J.~F. Drake, M. A. Shay, W. Thongthai and M. Swisdak 2005, \prl, 94, 095001

1438: %\bibitem[Drake et al.(2006)]{drake06} J.~F. Drake, M. Swisdak, H. Che and M. A. Shay 2006, Nature, 443, 553

1439: \bibitem[Drenkhahn(2002)]{dr02} G.~{Drenkhahn} 2002, \aap, 387, 714

1440: \bibitem[Drenkhahn \& Spruit(2002)]{drs02} G.~{Drenkhahn} and H.~C. {Spruit} 2002, \aap, 391, 1141

1441: %\bibitem[Heyvaerts et al.(1988)]{hey88}

1442: %J.~{Heyvaerts}, C.~{Norman} and R.~E. {Pudritz} 1988,

1443: %\apj, 330, 718

1444: \bibitem[Hesse et al.(1999)]{hesse99} M. Hesse, K. Schindler, J. Birn and M. Kuznetsova 1999, \pop, 6, 1781

1445: \bibitem[Hesse et al.(2004)]{hesse04} M. Hesse, M. Kuznetsova and  J. Birn 2004, \pop, 11, 5387

1446: \bibitem[Hesse \& Zenitani(2007)]{hesse07} M. Hesse and S. Zenitani 2007, \pop, 14, 112102

1447: \bibitem[Hoshino(1987)]{hoshino87} M. Hoshino 1987, \jgr, 92, 7368

1448: \bibitem[Horiuchi \& Sato(1997)]{hori97} R. Horiuchi and T. Sato 1997, \pop, 4, 277

1449: \bibitem[Huba(2005)]{huba05} J.~D. {Huba} 2005, \pop, 12, 012322

1450: \bibitem[Jaroschek et al.(2004a)]{claus04apj} C. H. Jaroschek, H. {Lesch} and R. A. Treumann 2004a, \apj, 605, L9

1451: \bibitem[Jaroschek et al.(2004b)]{claus04} C. H. Jaroschek, R. A. Treumann, H. Lesch and M. Scholer 2004b, \pop, 11, 1151

1452: \bibitem[Katanuma \& Kamimura(1980)]{katanuma80} I. Katanuma and T. Kamimura 1980, Physics of Fluids, 23, 2500

1453: \bibitem[Kirk \& Skj\ae raasen(2003)]{kirk03} J. G. Kirk and O. Skj\ae raasen 2003, \apj, 591, 366

1454: \bibitem[Larrabee et al.(2003)]{larra03} D.~A. {Larrabee}, R.~V.~E. {Lovelace} and M.~M. {Romanova} 2003, \apj, 586, 72

1455: \bibitem[Lesch \& Birk(1997)]{lb97} H. Lesch and G. T. Birk 1997, \aap, 324, 461

1456: \bibitem[Lesch \& Birk(1998)]{lb98} H. Lesch and G. T. Birk 1998, ApJ, 499, 167

1457: \bibitem[Lyubarsky \& Kirk(2001)]{lyu01} Y. Lyubarsky and J. G. Kirk 2001, ApJ, 547, 437

1458: \bibitem[Lyubarsky(2005)]{lyu05} Y. Lyubarsky 2005, \mnras, 358, 113

1459: \bibitem[Lyutikov(2003)]{lyut03a} M. Lyutikov 2003, \mnras, 346, 540

1460: \bibitem[Lyutikov \& Uzdensky(2003)]{lyut03b} M. Lyutikov and D. {Uzdensky} 2003, \apj, 589, 893

1461: \bibitem[Lyutikov(2006)]{lyut06} M. Lyutikov 2006, \mnras, 367, 1594

1462: \bibitem[Michel(1982)]{michel82} F. C. {Michel} 1982, Reviews of Modern Physics, 54, 1

1463: \bibitem[Michel(1994)]{michel94} F. C. {Michel} 1994, \apj, 431, 397

1464: %\bibitem[Ono et al.(1997)]{ono97} Y. Ono, M. Inomoto, T. Okazaki and Y. Ueda 1997, \pop, 4, 1953

1465: \bibitem[Nodes et al.(2003)]{nodes03} C. Nodes, G. T. {Birk}, H. {Lesch} and R. {Schopper} 2003, \pop, 10, 835

1466: \bibitem[Pritchett \& Coroniti(2004)]{prit04} P. L. Pritchett and F. V. Coroniti 2004, J.\ Geophys.\ Res., 109, 1220

1467: \bibitem[Quest \& Coroniti(1981)]{quest81} K. B. Quest and F. V. Coroniti 1981, \jgr, 86, 3289

1468: \bibitem[Ricci et al.(2003)]{ricci03} P. Ricci, G. Lapenta and J. U. Brackbill 2003, \pop, 10, 3554

1469: \bibitem[Romanova \& Lovelace(1992)]{rom92} M. M. Romanova and R. V. E. Lovelace 1992, \aap, 262, 26

1470: \bibitem[Scholer et al.(2003)]{scholer03} M. Scholer, I. {Sidorenko}, C.~H. Jaroschek, R.~A. {Treumann} and A. {Zeiler} 2003, \pop, 10, 3521

1471: \bibitem[Shay et al.(2001)]{shay01} M. A. Shay, J. F. Drake, B. N. Rogers and R. E. Denton 2001, \jgr, 106, 3759

1472: \bibitem[Shibata et al.(1995)]{shibata95} K. Shibata, S. Masuda, M. Shimojo, H. Hara, T. Yokoyama, S. Tsuneta, T. Kosugi and Y. Ogawara 1995, \apj, 451, L83

1473: \bibitem[Schopper et al.(1998)]{schopper} R. Schopper, H. Lesch and G. T. Birk 1998, \aap, 335, 26

1474: \bibitem[Silin \& {B{\"u}chner}(2003)]{silin03} I. Silin and J. {B{\"u}chner} 2003, \pop, 10, 3561

1475: \bibitem[Sonnerup et al.(1981)]{sonne81} B. U. O. Sonnerup, G. Paschmann, I. Papamastorakis,N. Sckopke, G. Haerendel, S. J. Bame, J. R. Asbridge, J. T. Gosling and C. T. Russell 1981, \jgr, 86, 10049

1476: \bibitem[Thompson \& Duncan(1995)]{thom95} C. Thompson and R. C. Duncan 1995, \mnras, 275, 255

1477: \bibitem[Thompson \& Duncan(2001)]{thom01} C. Thompson and R. C. Duncan 2001, \apj, 561, 980

1478: \bibitem[Uzdensky \& MacFadyen(2006)]{uz06} D. A. Uzdensky and A. I. MacFadyen 2006, \apj, 647, 1192

1479: \bibitem[Watanabe \& Yokoyama(2006)]{naoyuki06} N. {Watanabe} and T. {Yokoyama} 2006, \apj, 647, L123

1480: \bibitem[Zelenyi \& Krasnoselskikh(1979)]{zelenyi79} L. M. {Zelenyi} and V. V. {Krasnoselskikh} 1979, Astronomicheskii Zhurnal, 56, 819

1481: \bibitem[Zenitani \& Hoshino(2001)]{zeni01} S. Zenitani and M. Hoshino 2001, \apj, 562, L63

1482: \bibitem[Zenitani \& Hoshino(2005a)]{zeni05a} S. Zenitani and M. Hoshino 2005a, \apj, 618, L111

1483: \bibitem[Zenitani \& Hoshino(2005b)]{zeni05b} S. Zenitani and M. Hoshino 2005b, \prl, 95, 095001

1484: \bibitem[Zenitani \& Hoshino(2007)]{zeni07} S. Zenitani and M. Hoshino 2007, \apj, 670, 702

1485: 

1486: \end{thebibliography}

1487:  

1488: %\begin{thebibliography}{}

1489: %\end{thebibliography}

1490: 

1491: % ****** End of file apssamp.tex ******

1492: 

1493: \clearpage

1494: 

1495: \begin{figure}[htbp]

1496: \begin{center}

1497: \ifjournal

1498: \includegraphics[width={0.7\columnwidth},clip]{f1.eps}

1499: \else

1500: \includegraphics[width={0.7\columnwidth},clip]{f1.pdf}

1501: \fi

1502: \caption{

1503: Guide field dependence ($|B_y/B_0|$) of

1504: the growth rate $\omega_i$,

1505: normalized in the light transit time $\tau_c$.

1506: The fastest eigen growth rate of the relativistic tearing instability,

1507: linear growth rate of the fastest growing modes

1508: in relativistic magnetic reconnection,

1509: eigen growth rate of typical RDKI,

1510: and linear growth rate of the RDKI are shown.

1511: \label{fig:rec_vs_dki_by}}

1512: %\includegraphics[width={0.5\columnwidth},clip]{dki_vs_by.eps}

1513: %\caption{\label{fig:dki_vs_by}

1514: %Guide field dependence ($|B_y/B_0|$) of

1515: %the eigen growth rate ($\tau_c\omega_i$) of

1516: %the relativistic drift kink mode.

1517: %}

1518: \end{center}

1519: \end{figure}

1520: 

1521: \clearpage

1522: 

1523: \begin{figure}

1524: \begin{center}

1525: \ifjournal

1526: \includegraphics[width={\columnwidth},clip]{f2.eps}

1527: \else

1528: \includegraphics[width={\columnwidth},clip]{f2.jpg}

1529: \fi

1530: \caption{

1531: Snapshots of run R3-C ($|B_y/B_0|=1.5$) at $t/\tau_c=100.0$.

1532: (\textit{a}) Plasma density and flows,

1533: (\textit{b}) positron current density $J_{yp}$,

1534: (\textit{c}) charge non-neutrality [$\sigma=[n_p - n_e]/[n_p + n_e]$],

1535: (\textit{d}) reconnection electric field $E_y$, and

1536: (\textit{e}) vertical electric field $E_z$ are presented, respectively.

1537: The background counter lines show the magnetic field lines in the $x$-$z$ plane.

1538: \label{fig:t100}}

1539: \end{center}

1540: \end{figure}

1541: 

1542: \clearpage

1543: 

1544: \begin{figure}

1545: \begin{center}

1546: \ifjournal

1547: \includegraphics[width={0.6\columnwidth},clip]{f3.eps}

1548: \else

1549: \includegraphics[width={0.6\columnwidth},clip]{f3.pdf}

1550: \fi

1551: \caption{

1552: \label{fig:espec}

1553: Energy spectra of run R3-C at characteristic stages:

1554: $t/\tau_c=50,70,100$, and $200$.}

1555: \end{center}

1556: \begin{center}

1557: \ifjournal

1558: \includegraphics[width={0.95\columnwidth},clip]{f4.eps}

1559: \else

1560: \includegraphics[width={0.95\columnwidth},clip]{f4.pdf}

1561: \fi

1562: \caption{

1563: (\textit{a}) Spatial distribution of high-energy particles

1564: (\textit{gray points}; $\varepsilon \ge 40 mc^2$) at $t/\tau_c=100$.

1565: Two typical trajectories are projected in the same plane.

1566: The signs show the relevant positions at $t/\tau_c=100$.

1567: (\textit{b}) Trajectories in the $x$-$y$ plane.

1568: \label{fig:HE}}

1569: \end{center}

1570: \end{figure}

1571: 

1572: \clearpage

1573: 

1574: \begin{figure}

1575: \begin{center}

1576: \ifjournal

1577: \includegraphics[width={\columnwidth},clip]{f5.eps}

1578: \else

1579: \includegraphics[width={\columnwidth},clip]{f5.jpg}

1580: \fi

1581: \caption{\label{fig:3D-A}

1582: Snapshots of the current sheet in run 3D-A

1583: at (\textit{a}) $t/\tau_c=80$, (\textit{b}) $110$ and (\textit{c}) $140$.

1584: The gray surfaces show the density surface of

1585: (\textit{a}) $n=2/3n_0$, (\textit{b}) $n=1/3n_0$, and (\textit{c}) $n=1/5n_0$, respectively.

1586: The plasma density at the neutral plane ($z=0$) is projected into the bottom wall,

1587: with color from black (empty) through blue (sparse)  to red (dense; $n\sim1.2 n_0$).

1588: Panel (\textit{d}) is a zoomed-in view

1589: around the neutral plane ($-3.2<z/\lambda<3.2$)

1590: at $t/\tau_c=110$.  The gray lines trace the magnetic field lines.

1591: The plasma flow in the 2D planes ($x=0$,$y=0$, and $z=0$)

1592: are presented as white arrows in the three walls.

1593: The light speed ($v=c$) is projected to the arrow length $4$.

1594: \label{fig:3Drec}}

1595: \end{center}

1596: \end{figure}

1597: 

1598: \clearpage

1599: 

1600: \begin{figure}

1601: \begin{center}

1602: \ifjournal

1603: \includegraphics[width={0.8\columnwidth},clip]{f6.eps}

1604: \else

1605: \includegraphics[width={0.8\columnwidth},clip]{f6.png}

1606: \fi

1607: \caption{\label{fig:3DEy}

1608: The $E_y$ structure in run 3D-A at $t/\tau_c=80$.

1609: }

1610: \end{center}

1611: \end{figure}

1612: 

1613: \clearpage

1614: 

1615: \begin{figure}

1616: \begin{center}

1617: \ifjournal

1618: \includegraphics[width={\columnwidth},clip]{f7.eps}

1619: \else

1620: \includegraphics[width={\columnwidth},clip]{f7.jpg}

1621: \fi

1622: \caption{\label{fig:3D-B}

1623: Snapshots of the current sheet in run 3D-B

1624: with a guide field configuration ($B_y/B_0 = - 0.5$) at

1625: (\textit{a}) $t/\tau_c=120$, (\textit{b}) $170$,

1626: (\textit{c}) $200$ and (\textit{d}) $220$.

1627: Gray surfaces show the plasma density of

1628: (\textit{a}) $n=2/3 n_0$, (\textit{b}) $2/3 n_0$, (\textit{c}) $1/2 n_0$ and (\textit{d}) $1/3 n_0$,

1629: respectively.

1630: The bottom walls show the plasma density structure

1631: in/under the neutral plane at (\textit{a-b}) $z=-\lambda$ or

1632: at (\textit{c-d}) $z=0$.

1633: }

1634: \end{center}

1635: \end{figure}

1636: 

1637: \clearpage

1638: 

1639: \begin{figure}[htbp]

1640: \begin{center}

1641: \ifjournal

1642: \includegraphics[width={\columnwidth},clip]{f8.eps}

1643: \else

1644: \includegraphics[width={\columnwidth},clip]{f8.jpg}

1645: \fi

1646: \caption{\label{fig:cut}

1647: Slices of the simulation domain in run 3D-B at $t/\tau_c=120$.

1648: The plasma density is represented by gray shading.

1649: }

1650: \ifjournal

1651: \includegraphics[width={0.6\columnwidth},clip]{f9.eps}

1652: \else

1653: \includegraphics[width={0.6\columnwidth},clip]{f9.pdf}

1654: \fi

1655: \caption{

1656: The $z$-profiles of the density perturbation

1657: in run 3D-B at $t/\tau_c=120$.

1658: Two dominant modes (1,1) and (1,-1) are presented.

1659: \label{fig:sim}}

1660: \end{center}

1661: \end{figure}

1662: 

1663: \clearpage

1664: 

1665: \begin{figure}[htbp]

1666: \begin{center}

1667: \ifjournal

1668: \includegraphics[width={\columnwidth},clip]{f10.eps}

1669: \else

1670: \includegraphics[width={\columnwidth},clip]{f10.pdf}

1671: \fi

1672: \caption{

1673: Typical eigen functions of

1674: the relativistic drift-sausage tearing instability

1675: as a function of $z$,

1676: for the normalized wavenumbers

1677: $k_x \lambda = 0.25$ and $k_y \lambda = 0.25$.

1678: Perturbed magnetic fields: $\delta B_{x}, \delta B_{y}, i\delta B_{z}$;

1679: electric fields: $\delta E_{x}, \delta E_{y}, i\delta E_{z}$;

1680: electric currents: $\delta J_{x}, \delta J_{y}, i\delta J_{z}$;

1681: bulk velocities: $i\delta V_{+x}, i\delta V_{+y}, \delta V_{+z}$;

1682: and

1683: density fluctuations: $\delta D_{\pm} = \delta d_{p} \pm \delta d_{e}$

1684: are presented, respectively.

1685: \label{fig:rdkti_eigen}}

1686: \end{center}

1687: \end{figure}

1688: 

1689: \clearpage

1690: 

1691: \begin{figure}[htbp]

1692: \begin{center}

1693: \ifjournal

1694: \includegraphics[width={\columnwidth},clip]{f11.eps}

1695: \else

1696: \includegraphics[width={\columnwidth},clip]{f11.pdf}

1697: \fi

1698: \caption{

1699: Schematic illustration of the 3D relativistic drift-sausage tearing instability.

1700: Bottom panels show the relevance to

1701: the conventional 2D instabilities.

1702: \label{fig:illust}}

1703: \end{center}

1704: \end{figure}

1705: 

1706: \clearpage

1707: 

1708: \clearpage

1709: 

1710: \begin{figure}

1711: \begin{center}

1712: \ifjournal

1713: \includegraphics[width={\columnwidth},clip]{f12.eps}

1714: \else

1715: \includegraphics[width={\columnwidth},clip]{f12.pdf}

1716: \fi

1717: \caption{\label{fig:map}

1718: \textit{Top}: Growth rate ($\tau_c\omega_i$) of the instabilities

1719: in wavevector spaces of

1720: $(0\le k_x\lambda \le 1, 0\le k_y\lambda \le 1)$,

1721: as a function of the guide field amplitude $B_y/B_0=\alpha=0,-0.5,-1,-1.5$.

1722: \textit{Bottom}: Three classes of the unstable modes are mapped;

1723: A: kink-type modes; B: sausage-type modes;

1724: and C: others (neither or intermediate).

1725: }

1726: \end{center}

1727: \end{figure}

1728: 

1729: 

1730: \clearpage

1731: 

1732: \begin{figure}

1733: \begin{center}

1734: \ifjournal

1735: \includegraphics[width={\columnwidth},clip]{f13.eps}

1736: \else

1737: \includegraphics[width={\columnwidth},clip]{f13.jpg}

1738: \fi

1739: \caption{\label{fig:3Dcut}

1740: Snapshots of 2D cut plane at $y=0$ in run 3D-B at $t/\tau_c=170$.

1741: (\textit{a}) Plasma density, flows, 2D magnetic field lines and

1742: (\textit{b}) positron current $J_{py}$ (colored shading),

1743: vertical electric field $E_z$ (contours) are presented.

1744: The contour step $\Delta E_z$ is set to $0.1 B_0$.

1745: }

1746: \end{center}

1747: \end{figure}

1748: 

1749: \clearpage

1750: 

1751: \begin{figure}

1752: \begin{center}

1753: \ifjournal

1754: \includegraphics[width={0.65\columnwidth},clip]{f14.eps}

1755: \else

1756: \includegraphics[width={0.65\columnwidth},clip]{f14.pdf}

1757: \fi

1758: \caption{

1759: \label{fig:espec3}

1760: Energy spectra of 3D/2D runs.

1761: Plasma count number for 3D runs are re-arranged for comparison with 2D runs.}

1762: \end{center}

1763: \end{figure}

1764: 

1765: \begin{figure}[htbp]

1766: \begin{center}

1767: \ifjournal

1768: \includegraphics[width={\columnwidth},clip]{f15.eps}

1769: \else

1770: \includegraphics[width={\columnwidth},clip]{f15.pdf}

1771: \fi

1772: \caption{

1773: Time histories of nonthermal ratio parameters for reconnection runs;

1774: runs R3, R3-A, R3-C, 3D-A and 3D-B.

1775: \label{fig:3d_nonth}}

1776: \end{center}

1777: \end{figure}

1778: 

1779: \clearpage

1780: 

1781: \begin{deluxetable}{l|ccccccccccccc}

1782: \tabletypesize{\scriptsize}

1783: \rotate

1784: \tablecaption{\label{table} List of Simulation Runs}

1785: \tablewidth{0pt}

1786: \tablehead{

1787: \colhead{Run} &

1788: \colhead{$L_x$} &

1789: \colhead{$L_y$} &

1790: \colhead{$L_z$} &

1791: \colhead{$B_y/B_0$} &

1792: \colhead{${n_{bg}}/{(\gamma_{\beta}n_{0})}$} &

1793: \colhead{$n_{tot}$} &

1794: \colhead{${K_{nth}}/{K}$} &

1795: \colhead{$\tau_c \omega_i$} &

1796: \colhead{$\gamma_{max}$} &

1797: \colhead{Type}

1798: }

1799: \startdata

1800: R3 &

1801: 102.4 &

1802: -- &

1803: 51.2 &

1804: 0 &

1805: 5\% &

1806: $1.7 \times 10^{7}$ &

1807: $>3.7 \times 10^{-1} $ &

1808: $3.0 \times 10^{-2}$ &

1809: $> 158 $ &

1810: F &

1811: \\

1812: R3-A &

1813: 102.4 &

1814: -- &

1815: 51.2 &

1816: -0.5 &

1817: 5\% &

1818: $1.7 \times 10^{7}$ &

1819: $>4.0 \times 10^{-1} $ &

1820: $3.1 \times 10^{-2}$ &

1821: $> 122$

1822: ($\sim 210$) &

1823: S* &

1824: \\

1825: R3-B &

1826: 102.4 &

1827: -- &

1828: 51.2 &

1829: -1.0 &

1830: 5\% &

1831: $1.7 \times 10^{7}$ &

1832: $> 1.7 \times 10^{-1}$ &

1833: $2.8 \times 10^{-2}$ &

1834: $> 80 $ &

1835: S* &

1836: \\

1837: R3-C &

1838: 102.4 &

1839: -- &

1840: 51.2 &

1841: -1.5 &

1842: 5\% &

1843: $1.7 \times 10^{7}$ &

1844: $1.7 \times 10^{-1}$ &

1845: $2.3 \times 10^{-2}$ &

1846: $> 86 $ &

1847: F &

1848: \\

1849: R3-D &

1850: 102.4 &

1851: -- &

1852: 51.2 &

1853: -1.5 &

1854: 5\% &

1855: $1.7 \times 10^{7}$ &

1856: $>1.8 \times 10^{-1}$ &

1857: $2.3 \times 10^{-2}$ &

1858: $> 79 $ &

1859: S &

1860: \\

1861: R3-E &

1862: 102.4 &

1863: -- &

1864: 51.2 &

1865: -5.0 &

1866: 5\% &

1867: $1.7 \times 10^{7}$ &

1868: - &

1869: $1.5 \times 10^{-2}$ &

1870: $> 42 $ &

1871: S* &

1872: \\

1873: D3 &

1874: -- &

1875: 25.6&

1876: 51.2 &

1877: 0 &

1878: 0 &

1879: $4.2 \times 10^{6} $ &

1880: $7.2 \times 10^{-2}$ &

1881: $1.1 \times 10^{-1}$ &

1882: $48 $ &

1883: S &

1884: \\

1885: D3-A &

1886: -- &

1887: 25.6&

1888: 51.2&

1889: -0.25 &

1890: 0 &

1891: $4.2 \times 10^{6} $ &

1892: - &

1893: $1.0 \times 10^{-1}$ &

1894: $38 $ &

1895: S &

1896: \\

1897: D3-B &

1898: -- &

1899: 25.6&

1900: 51.2&

1901: -0.5 &

1902: 0 &

1903: $4.2 \times 10^{6} $ &

1904: - &

1905: N/A &

1906: $22 $ &

1907: S &

1908: \\

1909: D3-C &

1910: -- &

1911: 25.6&

1912: 51.2&

1913: -1.0 &

1914: 0 &

1915: $4.2 \times 10^{6} $ &

1916: - &

1917: N/A &

1918: $22 $ &

1919: S &

1920: \\

1921: 3D-A &

1922: 25.6 &

1923: 25.6 &

1924: 25.6 &

1925: 0 &

1926: 5\% &

1927: $5.4 \times 10^{8} $ &

1928: $1.8 \times 10^{-2}$ &

1929: $6 \times 10^{-2}$ &

1930: $ 52 $ &

1931: S* &

1932: \\

1933: 3D-B &

1934: 25.6 &

1935: 25.6 &

1936: 25.6 &

1937: -0.5 &

1938: 5\% &

1939: $5.4 \times 10^{8} $ &

1940: $> 1.4 \times 10^{-1}$ &

1941: - &

1942: $> 45 $ &

1943: S* &

1944: %\\

1945: %2D-C &

1946: %36.2 &

1947: %-- &

1948: %25.6 &

1949: %-0.5 &

1950: %5\% &

1951: %$6.9 \times 10^{6} $ &

1952: %- &

1953: %- &

1954: %S &

1955: \enddata

1956: \tablecomments{

1957: Initial parameters and obtained physical values of simulation runs.

1958: R3, D3, and 3D runs are

1959: 2D simulations for magnetic reconnection,

1960: 2D simulations for the RDKI,

1961: and 3D simulations, respectively. 

1962: %Run 2D-C is a supplemental two dimensional run for 3D-B,

1963: The system width $L_x$ is assumed to be

1964: in the oblique direction ($x$-$y$).

1965: The system size

1966: ($L_x,  L_y, L_z$ in units of $\lambda$),

1967: the guide field amplitude ($B_y/B_0$),

1968: the ratio of the background plasma density ($n_{bg}$)

1969: to the plasma density in the current sheet ($\gamma_{\beta}n_{0}$),

1970: the total number of the particles ($n_{tot}$),

1971: the nonthermal ratio ($K_{nth} / K$)

1972: the linear growth rate ($\tau_c \omega_i$)

1973: and

1974: the Lorentz factor of the highest energy particle ($\gamma_{max}$)

1975: are presented.

1976: }

1977: \end{deluxetable}

1978: 

1979: \end{document}

1980: %

1981: % ****** End of file apssamp.tex ******