1: %\documentclass[12pt,preprint]{aastex}
2: %\newcommand{\eqref}[1]{(\ref{#1})}
3: \newif\ifjournal\journalfalse
4: %\journaltrue
5:
6: \documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
7:
8: \usepackage{graphicx}% Include figure files
9:
10: %\usepackage{color}
11: %\newcommand{\newrev}[1]{\textcolor{red}{#1}}
12: \newcommand{\newrev}[1]{{#1}}
13:
14: \newcommand{\grad}{\nabla}
15: \renewcommand{\div}{\nabla \cdot}
16: \newcommand{\rot}{\nabla \times}
17: \newcommand{\pp}[2]{\frac{\partial #1}{\partial #2}}
18: %
19: \newcommand{\Alfven}{Alfv\'{e}n }
20: \newcommand{\JS}{J\"{u}tter-Synge }
21: % Journals
22: \def\grl{{\itshape Geophys. Res. Lett.} }
23: \def\jgr{{\itshape J. Geophys. Res.} }
24: \def\apj{{\itshape Astrophys. J.} }
25: \def\prl{{\itshape Phys. Rev. Lett.} }
26: \def\pop{{\itshape Phys. Plasmas} }
27: \def\aap{{\itshape Astron. Astrophys.} }
28: \def\mnras{{\itshape Monthly Notices of the RAS} }
29: \def\nature{{\itshape Nature} }
30:
31: \newcommand{\myemail}{zenitani@lssp-mail.gsfc.nasa.gov}
32:
33: %\shorttitle{Reconnection and weibel instability}
34: %\shortauthors{Zenitani \& Hesse}
35:
36: \begin{document}
37:
38: \title{The role of the Weibel instability at the reconnection jet front \\
39: in relativistic pair plasma reconnection}
40:
41: \author{S. Zenitani}
42: \author{M. Hesse}
43: \affiliation{
44: %\affil{
45: NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA
46: }
47: \email{zenitani@lssp-mail.gsfc.nasa.gov}
48:
49: %\date{\today}
50: \ifjournal
51: \else
52: \date{submitted to \it{Physics of Plasmas}}
53: \fi
54:
55: \begin{abstract}
56: The role of the Weibel instability is investigated for the first time
57: in the context of the large-scale magnetic reconnection problem.
58: A late-time evolution of
59: magnetic reconnection in relativistic pair plasmas
60: is demonstrated by particle-in-cell (PIC) simulations.
61: In the outflow regions,
62: powerful reconnection jet piles up the magnetic fields
63: and then a tangential discontinuity appears there.
64: Further downstream, it is found that
65: the two-dimensional extension of
66: the relativistic Weibel instability
67: generates electro-magnetic fields,
68: which are comparable to the anti-parallel or piled-up fields.
69: In a microscopic viewpoint,
70: the instability allows plasma's
71: multiple interactions with the discontinuity.
72: In a macroscopic viewpoint,
73: the instability leads to rapid expansion of the current sheet
74: and then the reconnection jet front further propagates into the downstream.
75: Possible application to the three-dimensional case is briefly discussed.
76: \end{abstract}
77:
78: %\pacs{52.27.Ep, 52.27.Ny, 52.35.Vd, 95.30.Qd}
79: \keywords{magnetic fields --- plasmas --- instabilities --- relativity --- shock waves}
80: \maketitle
81:
82: \section{Introduction}
83:
84: Magnetic reconnection is widely recognized as
85: a fundamental physical mechanism in collisionless plasmas.
86: Consuming the magnetic field energy in the inflow region,
87: it releases the energy to the kinetic energy of plasma particles.
88: It is an effective engine for
89: magnetic dissipation, plasma heating, or particle acceleration.
90: Recently, the relativistic extension of magnetic reconnection
91: has received attention for its role
92: in various high-energy astrophysical places
93: --- active galactic nuclei \citep{lb97,dimatteo,birk01},
94: pulsars \citep{coro90,lyu01,kirk03}, gamma ray bursts \citep{dr02,drs02},
95: and magnetars \citep{thom95,lyut03a}.
96: %
97: The mechanism of relativistic reconnection still remains unclear
98: as well as the conventional non-relativistic reconnection,
99: but recent kinetic simulations start to reveal its features.
100: It is demonstrated that
101: relativistic pair plasma reconnection is a powerful particle accelerator
102: \citep{zeni01,zeni05b,zeni07a,zeni07b,claus04},
103: and that it keeps fast reconnection rate \citep{bessho07,hesse07}
104: despite the lack of the Hall physics \citep{birn01}.
105: In principle, magnetic reconnection is a relatively large scale process ---
106: the typical speed of reconnection jet is
107: the \Alfven velocity $V_A$ or the light speed $\sim c$,
108: and the typical time scale is several tens of characteristic time scale;
109: $\lambda/V_A$ or $\lambda/c$,
110: where $\lambda$ is the typical spatial scale of the field reversal.
111:
112: On the other hand,
113: in the context of gamma ray bursts \citep{piran,med99}
114: or extra galactic jets,
115: the Weibel-type two-stream instability
116: in relativistic plasmas has attracted recent attention, too.
117: The Weibel instability \citep{weibel} is an electromagnetic instability,
118: that arises from plasma anisotropy.
119: Since it quickly generates magnetic fields,
120: it is a likely origin of magnetic field
121: in the synchrotron source
122: near collisionless shocks or near relativistic jet fronts.
123: Series of PIC simulations
124: \citep{kazi98,silva03,nishikawa03,fre04,hed04,claus05,nishikawa06}
125: successfully demonstrate the magnetic generation
126: via jet penetration or plasma shell collision
127: in a weakly or nonmagnetized plasma.
128: The Weibel magnetic structure evolves into long-durated
129: filament-like magnetic structure,
130: whose energy is approximately 10\% of equi-partition energy.
131: In-situ particle acceleration is also reported,
132: but its detailed mechanism still remains unclear \citep{kazi98,hed04,nishikawa06}.
133: %These field potential scatterer for shock acceleration \citep{bo78,dr83}).
134: On the theoretical side, the conventional Weibel instability,
135: which propagates into the transverse direction from plasma anisotropy,
136: has been extended to relativistic temperatures \citep{yoon87,yang93,yoon07}.
137: Meanwhile, its two-dimensional extension,
138: the electromagnetic counter streaming instability, has been studied
139: in relativistic counter streaming conditions \citep{cal97,kazi98,saito}.
140: In general, the Weibel instabilities are microscale phenomena,
141: whose scales are characterized by the plasma frequency $\omega_p=[4\pi ne^2/m]^{1/2}$; $\omega_p^{-1}$ in time and by $(ck/\omega_p)$ in space.
142:
143: In the context of magnetic reconnection,
144: since magnetic reconnection expels powerful outflow jets
145: from the reconnecting region,
146: it is quite possible that the jets interact with pre-existing plasmas,
147: and then excite an anisotropy-driven instability.
148: In fact, \citet{dau07} reported generation of out-of-plane magnetic field
149: via firehose type instability in their non-relativistic pair plasma reconnection,
150: although its role in reconnection remains unclear.
151:
152: In the present paper, we study
153: the role of the Weibel instability in the reconnection context.
154: We carry out two-dimensional PIC simulations of
155: relativistic pair plasma reconnection, and
156: we find that the relativistic counter-streaming Weibel instability
157: generates out-of-plane magnetic fields
158: in the downstream region of reconnection outflow.
159: We discuss the properties of the instability,
160: and then we investigate how the Weibel instability affects
161: micro- and macro physics of magnetic reconnection.
162: The paper consists of the following sections.
163: In section II we describe our simulation setup.
164: In section III we present the two-dimensional simulation results,
165: and then in section IV we investigate the properties of the instability.
166: In section V we discuss how the Weibel instability effects plasma dynamics,
167: both in microscopic particle motion and macroscopic reconnection structure.
168: The last section contains discussion and the summary.
169: %In fact, we observe high-energy, non-thermal particles
170: %in extra galactic radio sources,
171: %supernova remnants, pulsar nebulae, \citep{bg84,koyama95,aha04},
172: %Many studies have been devoted to
173: %how charged particles are accelerated in collisionless environments.
174:
175: %One of the most important mechanisms is
176: %collisionless shock in plasmas.
177: %Fermi-type shock acceleration \citep{bo78}
178: %%
179: %In Fermi-type shock acceleration \citep{bo78,dr83},
180: %a particle scattering mechanism is
181: %important to enhance the acceleration efficiency.
182: %Parallel shock \citep{ch88}
183:
184: %The reconnection jet associated \citep{masuda94}.
185:
186:
187: \section{Simulation setup}
188:
189: We carry out two-dimensional particle-in-cell (PIC) simulations
190: in a current sheet configuration.
191: As an initial condition,
192: we employ a relativistic extension of the Harris model
193: in GSM-like geometry.
194: The magnetic field, plasma density and
195: plasma distribution functions are described by
196: $\vec{B} = B_0 \tanh(z/\lambda) \hat{\bm{x}}$,
197: $d(z) = d_0 \cosh^{-2} (z/\lambda) = (\gamma_{\beta} n_0) \cosh^{-2} (z/\lambda)$ and
198: $f_{s} \propto d(z) \exp[-\gamma_\beta\{\varepsilon - \beta_s u_y\} / T ] $.
199: In the above equations,
200: $B_0$ is the magnitude of antiparallel magnetic field,
201: $\lambda$ is the typical thickness of the current sheet,
202: \newrev{$d_0$ is the lab-frame number density in the current sheet,
203: $n_0$ is the proper number density,}
204: the subscript $s$ denotes the species (`p' for positrons, `e' for electrons),
205: $\beta_{p} = -\beta_{e} = \beta$ is the dimensionless drift velocity,
206: $\gamma_\beta$ is the Lorentz factor for $\beta$ ($\gamma_\beta = [1-\beta^2]^{-1/2}$),
207: $\varepsilon$ is the particle energy,
208: $\vec{u}$ is the relativistic four velocity of $\vec{u}= [1-(\vec{v}/c)^2]^{-1/2} \cdot \vec{v}$
209: and $T$ is the proper temperature including the Boltzmann constant.
210: We set $T = mc^2$ and $\beta = 0.3$, respectively.
211: In addition, a uniform background plasma is added to the system
212: \newrev{in order to supply plasmas in the reconnection inflow region.}
213: Its number density and temperature are
214: $d_{bg}/d_0 =5\%$ and $T_{bg}/mc^2=0.1$, respectively.
215: \newrev{
216: In general, the velocity of the reconnection outflow jet is known to be
217: approximately the \Alfven velocity in the inflow region.
218: In the present study, we choose a low-density (5\%) background population
219: to obtain fast outflow.
220: Notice that the Harris model with uniform background plasmas
221: exactly satisfies an equilibrium.}
222:
223: The system consists of $1568 (x) \times 768 (z)$ grids
224: and the typical scale of the current sheet $\lambda$ is set to 10 grids.
225: Since we consider periodic boundaries in the $x$ direction,
226: and since there are two current layers in the periodic $z$ direction,
227: the boundaries of the main simulation domain are located at
228: $x = \pm 76.8 \lambda$ and $z = \pm 19.2 \lambda$.
229: We use $7.5 \times 10^7$ super particles in this simulation.
230: One cell contains $6.6 \times 10^2$ particles
231: at the center of the current sheet.
232: \newrev{
233: During the very early stage
234: we impose a small artificial electric field $\tilde{E_y}$ around
235: $(x, z) = (0,\pm 3\lambda)$.
236: The resultant $\vec{E}\times\vec{B}$ flow
237: compresses the current sheet,
238: and then it triggers reconnection
239: around the center of the main simulation domain.
240: The typical spatial ranges of the trigger field $\tilde{E_y}$ are
241: set to $(\Delta x,\Delta z) \sim (\pm 2\lambda,\pm \lambda)$.}
242: The trigger field soon vanishes after $t/\tau_c=(10-15)$,
243: where $\tau_c=\lambda/c$ is the light transit time.
244: We discuss the physics of reconnection
245: in the late stage of $t/\tau_c=(60-120)$,
246: which is not influenced by this initial perturbation.
247: These conditions are similar to
248: the author's previous study \citep{zeni07a};
249: but we use a larger simulation box
250: to discuss late-time structure without boundary effects.
251: We call this reference run `run A'.
252: The total energy is conserved within an error of $0.1\%$
253: throughout the simulation run, after the initial trigger force vanishes.
254:
255:
256: \section{Results}
257:
258: \newrev{Due to the initial perturbation,
259: magnetic reconnection takes place around the center of
260: the main simulation domain.
261: Magnetic field lines start to reconnect at $t/\tau_c\sim50$,
262: and outflow jets into the $\pm x$ directions appear}.
263: Figure \ref{fig:snap}\textit{a} show a snapshot of
264: the right half of the main simulation domain at $t/\tau_c=80$.
265: \newrev{The left half ($-76.8\le x/\lambda \le 0$) is not presented.
266: The reconnection outflow is well developed at this stage, and its}
267: speed is up to $\sim 0.7c$.
268: Along the neutral line,
269: magnetic fields are piled-up
270: in front of the dense plasma region of the current sheet
271: around $x/\lambda \sim 18$.
272: Its peak amplitude is $B_z/B_0\sim 1.5$, and
273: there is a relatively sharp boundary between
274: the pileup magnetic field and
275: the pre-existing dense plasma in the downstream.
276: This boundary is a tangential discontinuity (hereafter TD in short),
277: and we discuss ``upstream'' and ``downstream'' based on the TD
278: throughout this paper.
279: The propagation speed of the TD ($V_{TD}\sim 0.65c$) is
280: slightly slower than the average velocity of local plasmas.
281: The typical plasma density in the simulation frame is
282: $d/d_0\sim 0.1-0.2$ in the upstream,
283: $d/d_0\sim 2.5$ at the downstream side of the TD
284: and then it decreases to $d/d_0 \sim 1$ in the further downstream region.
285: Figure \ref{fig:snap}\textit{b} shows
286: the out-of-plane magnetic field ($B_y$) structure and
287: the relevant current system
288: in the reconnection outflow region,
289: which is indicated by the rectangle in Figure \ref{fig:snap}\textit{a}.
290: The characteristic structure of $B_y$ is observed, and
291: its maximum amplitude is $B_y\sim 0.6 B_0$.
292: In Figure \ref{fig:snap}\textit{c}
293: we observe charge separation at the same place and
294: the vertical $E_z$ structure
295: \newrev{(Hereafter the term ``vertical'' means the $z$ direction)}.
296: The $E_z$ explains both
297: the motional field for $B_y$
298: and the electrostatic field by the charge separation.
299: %
300: The time development of the $B_y$ structure along the neutral plane
301: is presented in Figure \ref{fig:linear}.
302: These $B_y$ fields suddenly appear after $t/\tau_c\sim 64$ and then
303: they exponentially grow until they saturate after $t/\tau_c\sim 80$.
304: The instability looks like a convective mode, traveling into the $+x$-direction.
305: However, actually, it is nearly
306: non-convective purely-growing mode
307: in the frame of the plasma average flow.
308: The linear growth rate measured by $B_y$ growth is
309: $\tau_c\omega_i \sim 1.7$-$1.8 \times 10^{-1}$ or
310: $\omega_i/\Omega_p \sim 5.2$-$5.5 \times 10^{-2}$,
311: where $\Omega_p$ is the typical plasma frequency in the system.
312: The typical spatial scales are
313: $7\lambda$-$10\lambda$ ($x$) and $\sim 2\lambda$ ($z$).
314: Careful observation show that the instability has
315: a two-dimensional rectangular structure.
316: In Figure \ref{fig:snap}\textit{b},
317: we see the weak negative regions
318: on the upper side, on the lower side and on the right side
319: \newrev{[colored in light blue; $(x,z) \sim (24,\pm 1.5), (30,0)$ in unit of $\lambda$] }of
320: the characteristic positive region \newrev{[inner orange region; $\sim (24,0)$]}.
321: Similarly, weak positive regions are located in the vicinity of
322: the characteristic negative region \newrev{[inner blue region; $\sim (21,0)$]}.
323: %
324: We find that these structures are generated by
325: the two-dimensional Weibel-type instability.
326: In this case, plasmas are highly anisotropic along the $x$-direction,
327: mainly because the TD pushes away the pre-existing plasmas,
328: and because the reconnection outflow jet penetrates into this region.
329: Therefore, the situation is similar to
330: jet injection \citep{cal97,nishikawa03} or
331: relativistic counter-stream \citep{kazi98},
332: and magnetic generation near the shock \citep{med99} in pair plasmas.
333: The instability resides inside the current sheet,
334: where the plasma frequency is high.
335: In addition, the Weibel instability prefers an unmagnetized region,
336: and so an inner current sheet is an ideal place for the instability.
337: The current structure and the charge separation structure
338: indicates the nature of the Weibel-type activity.
339: As schematically explained in \citet{med99},
340: small $B_y$ fluctuation leads to
341: the $z$-displacement of $\pm x$-streaming plasmas,
342: and then the resultant \newrev{$x$-current structure $\pm \delta J_x$}
343: continues to enhance $\delta B_y$.
344: Thus, the Weibel instability generates magnetic field
345: which is perpendicular to the direction of the anisotropy,
346: and then it leads to the reduction of anisotropy.
347:
348: Figure \ref{fig:aniso} shows the plasma distribution function
349: of $1.4 \times 10^5$ particles in the Weibel active region
350: ($22 \le x/\lambda \le 26, -2 \le z/\lambda \le 2$) at $t/\tau_c=80$.
351: The left part of the distribution function is almost identical to
352: the initial distribution of pre-existing plasmas,
353: but the right part are highly elongated due to
354: both reflected plasmas and upstream-origin accelerated particles.
355: %The average plasma velocity is $[\Sigma{v_x}]/\Sigma 1 \sim 0.49c$,
356: \newrev{The average plasma velocity is $\langle v_x \rangle \sim 0.49c$,
357: %\newrev{The average plasma velocity is $\int f \vec{v} d\vec{u}^3/\int f d\vec{u}^3 \sim 0.49c$,
358: where $\langle ~ \rangle$ means the average value among the particles.}
359: On the contrary, the plasma fluid velocity
360: (a Lorentz transformation velocity to the rest frame,
361: where the plasma momentum flow is zero)
362: is $\sim 0.65c \sim V_{TD}$.
363: The two velocities differ due to highly asymmetric plasma distribution.
364: An integrated plasma temperature is as follows;
365: %$(T'_x,T'_y,T'_z)=([\Sigma{(mu'_x)v'_x}]/n'mc^2, [\Sigma{(mu'_y)v'_y}]/n'mc^2, [\Sigma{(mu'_z)v'_z}]/n'mc^2 ) = (2.3, 1.0, 1.5)$,
366: %where $T'$, $m\vec{u}'$, $\vec{v}'$ and $n'$ are
367: \newrev{$(T'_x,T'_y,T'_z)=(\langle{m u'_x v'_x}\rangle, \langle{m u'_y v'_y}\rangle, \langle{ mu'_z v'_z}\rangle ) = mc^2 (2.3, 1.0, 1.5)$,
368: where $T'$, $m\vec{u}'$ and $\vec{v}'$ are
369: the temperature, momentum and velocity in the rest frame of plasmas.}
370: After the reconnection jets start from the $X$-type region,
371: the plasma anisotropy in the downstream region grows in time,
372: until the Weibel instability appears.
373: The anisotropy stays at the same level after the instability occurs,
374: because $x$-momentum is continually supplied from the upstream side.
375:
376: Snapshots of the field properties along the neutral line;
377: the pileup field $B_z$ (\textit{bold line}),
378: the reconnection electric field $E_y$ (\textit{dashed line}),
379: and the Weibel magnetic field $B_y$ (\textit{thin line})
380: are presented in Figures \ref{fig:stack}\textit{a-d}.
381: Note that the Weibel fields are observed
382: in the local frame of plasma average flow.
383: They propagate to the $+x$ direction and
384: its speed is slightly slower than the speed of the TD ($V_{TD}$).
385: Therefore, sometimes the Weibel fields are caught up by the TD.
386: For example, the positive $B_y$ region
387: around $x/\lambda\sim 24$ at $t/\tau_c=80$
388: is nearly caught by the TD
389: around $x/\lambda\sim 34$ at $t/\tau_c=100$.
390: At the same time, new Weibel fields are continuously
391: generated in the further downstream region.
392: We discuss the late time development of the Weibel fields later.
393: The subpartition of Weibel field energy $(B_y^2+E_x^2+E_z^2)/8\pi$
394: to the local plasma kinetic energy saturates
395: around $8$-$12$\% in run A.
396:
397:
398: \section{Linear analysis}
399:
400: In order to study the properties of the instability,
401: we have solved the dispersion relation
402: by linearizing relativistic fluid equations.
403: \newrev{In the Weibel region,
404: plasmas mainly consist of three different components;
405: (i) pre-existing plasmas in the Harris current sheet,
406: (ii) current sheet plasmas, which are reflected by the TD,
407: and (iii) upstream-origin plasmas, which are originally from
408: the reconnection inflow region.
409: Since we set a low plasma density
410: in the reconnection inflow region (5\% of the Harris current sheet),
411: the third population is relatively smaller than the other two.
412: Therefore, we employ the counter-streaming model of (i) and (ii)
413: in order to evaluate the Weibel instability.}
414: \newrev{We extend \citet{kazi98}'s fluid theory
415: for counter-streaming four fluids
416: (streaming/counter-streaming positrons and electrons),
417: which was originally developed by \citet{cal97}.
418: Although \citet{kazi98} ignored the plasma pressure effect,
419: it is here included.}
420: We employ the following relativistic fluid equations;
421: \begin{equation}\label{eq:fluid}
422: \frac{\gamma_{sa}^2}{c^2} (p_{sa} + e_{sa})
423: ( \pp{}{t} + \vec{v}_{sa} \cdot \grad )
424: \vec{v_{sa}}
425: =
426: - \grad p_{sa} +
427: \gamma_{sa} q_{sa} n_{sa}
428: \Big( \vec{E} + \frac{\vec{v}_{sa}}{c} \times \vec{B} \Big)
429: - \frac{\vec{v}_{sa}}{c^2} ( \gamma_{sa} q_{sa} n_{sa} \vec{E} \cdot \vec{v}_{sa} + \pp{p_{sa}}{t} ),
430: \end{equation}
431: where $p$ is isotropic plasma pressure,
432: $e$ is the fluid internal energy,
433: the subscript $a$ denotes two kind of streams
434: (`$1$' for streaming fluids, and `$2$' for counter-streaming fluids),
435: and $\gamma_{sa} = [ 1-(\vec{v}_{sa}/c)^{2} ]^{-1/2}$ is
436: the relevant Lorentz factor.
437: We also use the continuity equation and Maxwell equations
438: \begin{equation}\label{eq:cont}
439: \pp{}{t}(\gamma_{sa} n_{sa}) + \div(\gamma_{sa} n_{sa} \vec{v}_{sa} ) = 0,
440: \end{equation}
441: \begin{equation}\label{eq:rotb}
442: \rot \vec{B} = \frac{ 4\pi }{c} \sum_{s=e,p} \sum_{a=1,2} \gamma_{sa} q_{sa} n_{sa} \vec{v}_{sa} + \frac{1}{c} \pp{ \vec{E}}{t}
443: \end{equation}
444: \begin{equation}\label{eq:rote}
445: \rot \vec{E} = - \frac{1}{c} \pp{ \vec{B}}{t}.
446: \end{equation}
447: We assume the adiabatic gas condition in order to close the equation
448: \begin{equation}\label{eq:p}
449: p_{sa} \propto n_{sa}^{\Gamma}
450: ,~
451: e_{sa} = n_{sa} m c^2 + \frac{1}{\Gamma -1} p_{sa}
452: \end{equation}
453: where $\Gamma = 5/3$-$4/3$ is the polytropic index of adiabatic gas,
454: We consider a two dimensional perturbation
455: $\delta f \propto \delta f \exp(ik_x x + ik_z z - i \omega t)$,
456: where $\vec{k}=(k_x,k_z)$ is the wavevector
457: and $\omega$ is the complex frequency,
458: and then we linearize all equations for four fluids.
459: Then, we numerically calculate the growth rate (Im $\omega$)
460: for arbitrary wavevector $\vec{k}=(k_x,k_z)$ by solving a matrix problem.
461: For simplicity, the following assumptions are used;
462: \begin{eqnarray}
463: v_{p1}=v_{e1}=V_{TD},~v_{p2}=v_{e2}=-V_{TD}\\
464: n_{p1}=n_{p2}=n_{e1}=n_{e2}=n_0\\
465: p_{p1}=p_{p2}=p_{e1}=p_{e2}=n_0 mc^2 .
466: \end{eqnarray}
467: We assume that plasma density is homogeneous,
468: two counter-streams are symmetric,
469: the frame is set to the co-moving frame of the TD,
470: considering that the TD completely reflects the momentum of
471: pre-existing plasmas. In the present case,
472: the simulation data shows
473: $\vec{k}=(k_x,k_z) \sim (\omega_p/c)(0.15,0.75)$
474: in the frame of interest.
475: %
476: Because of the complexity in the simulation system,
477: this analysis does not exactly describe the instability.
478: The density gradient of plasmas, the current sheet thickness,
479: the wavelength of the instability are all comparable,
480: the Weibel region moves to the $x$-direction
481: slightly slower than the TD ($V_{TD} \sim 0.65c$),
482: the local average velocity and the local fluid velocity differs,
483: and local plasma velocities depend on the distance from the TD.
484: However, the goal of our simple theory is to roughly understand the physics.
485:
486: Figure \ref{fig:weibel} shows
487: the dispersion relation of the two-dimensional mode for $k_z=5k_x$.
488: The linear analysis (\textit{bold line}) and the simulation data
489: are in good agreement.
490: Further investigation shows that
491: the maximum growth rate is
492: on an order of $0.01$-$0.02\omega_p$
493: with the relativistic temperature of $T=mc^2$,
494: and that the cut-off (decline of the growth rate) is
495: rather sensitive to the counter-streaming velocity.
496: %
497: The obtained mode is purely growing, and
498: it has an electromagnetic feature.
499: Because of the mathematical symmetry,
500: we obtain the other oblique modes for $(\pm k_x,\pm k_z)$
501: with the same growth rates.
502: Therefore, the two-dimensional rectangular structure is
503: obtained by superimposing these oblique modes.
504: The change separation structure (Figure \ref{fig:snap}\textit{c})
505: in the simulation frame can be explained by
506: the $z$-displacement by the instability.
507: It reflects both the density gradient inside the current sheet
508: and the Lorentz boost of the fast outflow streams.
509: The electrostatic component of the instability is relatively small.
510:
511: We can also obtain
512: \newrev{the growth rate of the instability}
513: in counter-streaming cold beams \newrev{\citep{kazi98}}
514: by dropping the plasma pressure effect.
515: (One can remove pressure-related terms from eq. \ref{eq:fluid}
516: and employ $e_{sa} = n_{sa} m c^2$ instead of eqs. \ref{eq:p}.)
517: For comparison, the growth rate of the cold-beam limit is also presented
518: in Figure \ref{fig:weibel} (\textit{dashed line}).
519: Obviously, the instability \newrev{grows substantially slower
520: than the cold-beam limit.}
521: One interpretation is that
522: imposing plasma pressure means the reduction of the anisotropy.
523: In a high temperature limit
524: where the four velocity of the counter-streams is relatively negligible,
525: the distribution becomes close to a single isotropic distribution.
526: Another interpretation is that
527: the relativistic pressure effect slows down \newrev{the growth rate},
528: as discussed in the relativistic studies
529: on the one-dimensional Weibel instability \citep{yoon87,yang93,yoon07}.
530: In relativistic temperature regime,
531: it is known that the Weibel instability is re-scaled by
532: Im $\omega \lesssim \omega_p/\hat{\gamma}^{1/2}$ in time
533: and $(\hat{\gamma}^{1/2} ck)/\omega_p$ in space,
534: where $\hat{\gamma}$ is the typical Lorentz factor of plasma maximum energy
535: \citep{yoon87}.
536: %
537: From the viewpoint of relativistic fluids,
538: the enthalpy term in equation \ref{eq:fluid} increases an effective inertia,
539: and then it slows down \newrev{the growth rate of the instability.}
540: In the present case, the term yields
541: $(\gamma_{sa}/c)^2(p_{sa} + e_{sa}) \sim (nm) \gamma_{sa}^2 ( 1 + [\Gamma/(\Gamma-1)] [T_{sa}/({mc^2})] )\sim 9n_{sa}m$.
542: Since it replaces the mass term inside the plasma frequency,
543: the instability in a relativistic hot plasma
544: grows slower than
545: the instability in the cold beam limit by a factor of $\sqrt{9}\sim 3$.
546: By comparing the enthalpy term
547: in relativistically hot limit ($\sim 4p/c^2$)
548: and
549: in cold-beam limit ($\sim nm$),
550: we obtain the slow-down factor of ${p}^{1/2}$.
551: This is consistent with the scaling of the one-dimensional Weibel instability,
552: by a factor of $\hat{\gamma}^{1/2}$.
553: %In the case of ``water bag'' distribution,
554: %simple calculation yields $p \lesssim 3\hat{\gamma}nmc^2/8$ and then
555: %we obtain the factor is $(1.5 \hat{\gamma})^{1/2}$.
556: In summary, the counter-streaming Weibel-type instability slows down
557: by the inertia effect of relativistic pressure.
558: Roughly speaking, the instability
559: is similarly re-scaled by a factor of $\hat{\gamma}^{1/2}$,
560: as the one-dimensional Weibel instability.
561: %In fact, we can crudely estimate the growth rate
562: %by using one-dimensional Weibel instability, too \citep{yoon07}.
563: %Considering equal-mass contribution and local plasma density
564: %(typically $n/n_0 \sim 2$ in the Weibel-active region),
565: %we can replace the plasma frequency term
566: %$\omega_p^2 \rightarrow \sum_{e,p} (n/n_0) \Omega_{p}^2 \simeq (2\Omega_p)^2$.
567: %Then, assuming plasma temperature $T_{\perp}\sim 2, T_z\sim 1$
568: %with asymmetry index $\Delta \sim 1$,
569: %we assume Im $\omega/(2\Omega_p) \sim 0.01$-$0.02$ in Yoon 2007 (from Fig. 1; $\log \alpha_\parallel \sim 0$).
570: %Note that we use $T_{\perp} \sim 2$ here, but $T_y \sim 1$.
571:
572: The panels in Figure \ref{fig:weibel2} present
573: growth rates of the obtained unstable modes
574: as a function of $\vec{k}=(k_x,k_z)$.
575: Both relativistic pressure case (Fig. \ref{fig:weibel2}\textit{a}) and
576: the cold-beam limit (Fig. \ref{fig:weibel2}\textit{b}) are shown.
577: The one-dimensional mode along $k_x=0$ is
578: the conventional Weibel instability,
579: which has electromagnetic features.
580: On the other hand, the mode along $k_z=0$ is
581: the electrostatic counter-streaming instability.
582: The typical mode in our simulation is $(k_x,k_z)=(0.15,0.75)$
583: with some amount of ambiguity.
584: As seen in Figure \ref{fig:weibel2}\textit{a}),
585: the obtained mode is rather close to
586: the one-dimensional Weibel instability.
587: \newrev{It is important to note that the oblique mode grows
588: slightly faster than the one-dimensional Weibel instability,
589: and this is a signature of the counter-streaming Weibel-type instability.}
590: The central region and the right half
591: of Figure \ref{fig:weibel2}\textit{a} are
592: mainly occupied by the electrostatic-like mode.
593: Their growth rate is even faster, however,
594: since our theory depends on
595: the isotropic fluid pressure and the adiabatic condition (eq. \ref{eq:p}),
596: we think that our theory may be invalid,
597: especially in the short wavelength range of $(c/\omega_p) |\vec{k}| \gtrsim 1$.
598: In addition, in the high-pressure regime,
599: the two counter-streaming distributions overlap each other.
600: All these conditions are unfavorable to
601: describe the electrostatic modes in the parallel direction.
602: Meanwhile, the cold beam limit (Fig. \ref{fig:weibel2}\textit{b})
603: seems to be in good agreement with the \citet{saito}'s work,
604: based on \citet{kazi98}'s theory.
605: Their counter-streaming velocity $0.5c$ is comparable to ours of $0.65c$.
606:
607: %Let us compare these results to previous works.
608: %Compared with \citet{kazi98},
609: %we observe more elongated structure ($|k_z/k_x| \sim 5$)
610: %than theirs ($|k_x/k_y|\sim 2$).
611: %In other word, the unstable mode is rather transverse
612: %to the counter stream direction.
613: %It is known that
614: %transverse electromagnetic mode dominates
615: %under the faster counter-stream condition \citep{claus05},
616: %but the counter-streaming velocity is approximately same.
617: %We don't know whether this is due to
618: %the relativistic pressure effects or not.
619:
620: %so that generated Weibel magnetic fields scatter the plasma momentum
621: %into the perpendicular directions so that the anisotropy decreases.
622:
623:
624: \section{Effect of the Weibel instability}
625:
626: In this section, we investigate how the Weibel instability affects
627: the micro dynamics of plasmas motion and global dynamics of reconnection.
628: First we focus on the plasma motion near/in the downstream region,
629: because the Weibel instability occurs only in the downstream side of the TD.
630: Two characteristic regions will affect plasma motion ---
631: the TD and the Weibel region.
632: Before the Weibel instability appears,
633: particles in the downstream region are meandering in the current sheet.
634: Once they are hit or reflected by the TD,
635: they constantly travel into the $+x$-direction
636: because $x$-momentum is conserved
637: in the Harris current sheet configuration without $B_y$.
638:
639: So, what happens after the Weibel magnetic fields $B_y$ appears?
640: In order to study plasma motion around the two characteristic regions,
641: we select $10^5$ super particles
642: ($5 \times 10^4$ pairs)
643: that satisfies the following conditions;
644: they are
645: (i) found in the piled-up region
646: ($6\le x/\lambda \le 10$,$-2\le z/\lambda \le 2$)
647: at $t/\tau_c=60$
648: and (ii) found in the Weibel region
649: ($22\le x/\lambda \le 26 $,$-2 \le z/\lambda \le 2$)
650: at $t/\tau_c=80$.
651: The $x$-ranges of these regions are
652: indicated by arrows in Figures \ref{fig:stack}\textit{a,b}.
653: Then, the spatial distribution of the selected particles are investigated.
654: The panels in Figure \ref{fig:dist} show
655: the distribution of the selected positrons at $t/\tau_c=100$.
656: The right panel shows the distribution of fast positrons.
657: The relevant $x$-range is indicated by the dashed arrow
658: in Figure \ref{fig:stack}\textit{c}.
659: The left panel presents slow positrons,
660: whose $x$-velocity $v_x$ is slower than that of the TD; $v_{TD}=0.65 c$.
661: The panel contains $10^4$ slow positrons (20\% of positrons).
662: %Therefore, 20\% of positrons will be hit by the TD once again.
663: %
664:
665: Roughly speaking, these positrons can be classified
666: into the following three groups.
667: The first group is moving-away positrons,
668: who travels faster than the TD to the $+x$ direction.
669: The rectangle
670: (indicated by the white dashed line in Figure \ref{fig:dist}\textit{b})
671: is an approximate location of
672: the selected particles at $t/\tau_c=100$.
673: After they are hit by the TD, they continue to escape into the $+x$ region,
674: faster than $V_{TD}$.
675: %
676: The second group is found around $33<x/\lambda<40, z/\lambda \sim \pm 2$
677: in both two panels in Figure \ref{fig:dist}.
678: They are located along the magnetic field line,
679: which are connected to the TD.
680: They have relatively small population, and
681: they do not always escape into the $+x$ direction.
682: We discuss the field-line modulation and
683: the current sheet expansion
684: later in this section.
685: %
686: The last group is found along the neutral line ($z \sim 0$)
687: in Figure \ref{fig:dist}\textit{a}.
688: These particles are affected by the Weibel instability.
689: Their $z$-locations are positive around $40<x/\lambda<44$,
690: and negative around $36<x/\lambda<40$.
691: These $z$-displacements are due to the Weibel instability;
692: the effect of the out-of-plane field $B_y$
693: (Refer to Fig. \ref{fig:stack}\textit{c} for the polarity of $B_y$).
694: %
695: The high density region near the TD ($x/\lambda <36, z/\lambda \sim 2$ in Figure \ref{fig:dist}\textit{a}) contains
696: both the second class of positrons along the field lines and
697: the third class of Weibel-affected positrons.
698: They are soon reflected by the TD, and then
699: we see their reflection in the other high density region near the TD
700: ($x/\lambda <36, z/\lambda \sim -2$ in Figure \ref{fig:dist}\textit{b}).
701: We note that the magnetic field near the TD is not vertical,
702: but rather tilted into $+y$ direction,
703: because the TD hit the positive $B_y$ region at this time.
704: Since 20\% of selected positrons are in Figure \ref{fig:dist}\textit{b},
705: the third group has relatively large population.
706:
707: We pick up $1.7 \times 10^3$ positrons from them,
708: which are found in the vicinity of the TD
709: ($x/\lambda \le 33$, $-2 \le z/\lambda \le 2$) at $t/\tau_c=100$,
710: and then we examine their trajectories.
711: The selection mainly consists of
712: the third class of Weibel-modulated positrons,
713: because the second class has fewer population.
714: Properties of two typical trajectories are shown
715: in Figure \ref{fig:stack}\textit{e,f} as a function of $x$.
716: We call them positron $A$ (\textit{solid line}) and
717: positron $B$ (\textit{dotted line}).
718: Throughout the simulation period ($0 \le t/\tau_c\le 120$),
719: they stay in the narrow region of $-2<z/\lambda<2$.
720: In Figure \ref{fig:stack}\textit{e},
721: marks show the particle position at the selected time stages
722: for comparison with Figures \ref{fig:stack}\textit{a-d}.
723: Particle $A$ starts from $x/\lambda \sim 33$ in the $-x$ direction.
724: Its energy is originally $\varepsilon \sim 3 mc^2$.
725: Around $t/\tau_c \sim 60$, it collides with the TD and then
726: it turns its way to the $+x$ direction.
727: Its energy increases to $\varepsilon \sim 5 mc^2$
728: via the interaction with the TD.
729: Figure \ref{fig:stack}\textit{g} show the field properties
730: $B_y$ (\textit{solid line}) and $E_y$ (\textit{dashed line})
731: at positron $A$'s position.
732: Near $x/\lambda \sim 22$,
733: it feels relatively strong $B_y$ in the Weibel-active region,
734: and then its $x$-momentum is transported to $z$ momentum
735: (also, to $y-$momentum through the meandering motion).
736: Consequently, its $x$-velocity starts to slow down,
737: as indicated by the arrow in Figure \ref{fig:stack}\textit{f}.
738: Since the TD travels relatively fast ($v_{TD} \sim 0.65c$),
739: the TD eventually catches up positron $A$ and hits it again.
740: This time, the positron $A$ gains more energy by the motional electric field $E_y$,
741: because more magnetic fields are piled up than during the first impact.
742: Now its energy goes up to $\varepsilon/mc^2 \sim 8$.
743: So, the Weibel field slow down the escaping particles, and then
744: it enables multiple interactions with the TD.
745: We even find three-time or four-time interactions with the TD
746: in selected positron trajectories.
747: For example, positron $B$ (\textit{dotted line}) are hit
748: by the TD twice ($x/\lambda \sim 31, 44$)
749: after the first interactions with the TD at $x/\lambda \sim 9$.
750: If we study further long-time evolution,
751: these positrons will be hit by the TD multiple times.
752: %but only a very small number of particles.
753:
754: Interestingly, it seems that
755: only low-energy particles are reflected by the Weibel region;
756: high-energy particles are insensitive to the Weibel fields and then
757: they easily escape to the $+x$ direction,
758: when their energy $\varepsilon/mc^2$ exceeds $7$-$8$.
759: The typical kinetic energy gain by the TD reflection
760: ranges up to $\varepsilon/mc^2 \sim 10$-$20$,
761: due to enhanced pileup electric field.
762: Therefore, among the reflected particles,
763: low-energy particles are trapped between the TD and the Weibel region,
764: and then they are heated by the multiple interactions with the TD.
765: They can not escape into the outflow region,
766: until they become energetic enough ---
767: their gyro radii exceeds the scale of the Weibel structure
768: ($\gamma c/\omega_c \gg \gamma^{1/2} c/\omega_p$).
769: So, this result indicates that
770: the Weibel instability enhances plasma heating in the downstream side of the TD,
771: while it is not likely to enhance high-energy particle acceleration.
772: The threshold energy will increase
773: as the system condition becomes more relativistic,
774: because the electron skin depth is relatively larger by default,
775: and because the typical scale of the Weibel instability
776: becomes even larger by a factor of $\hat{\gamma}^{1/2}$ or $p^{1/2}$.
777:
778: Next, in order to study the Weibel mode effect
779: to global reconnection structure,
780: we carried out another simulation.
781: The new run (run B) starts from
782: intermediate data of run A at $t/\tau_c=60$,
783: and then we artificially reduce $B_y$ to $0$ in run B.
784: The quadrupolar magnetic fields near the reconnecting region
785: \citep{hesse94,oe01,nagai01} do not appear
786: in pair plasma reconnection,
787: because Hall physics depends on different ion and electron masses.
788: In antiparallel configurations,
789: magnetic reconnection involves $B_x$ and $B_z$, and
790: only the Weibel instability or anisotropy-driven instabilities
791: generates $B_y$.
792: Because of the artificial reduction of $B_y$,
793: the system slightly loses energy.
794: \newrev{The total energy in run B is smaller than the total energy in run A
795: by $0.15\%$ at the end of the simulation ($t/\tau_c=120$).
796: This energy loss is much smaller than typical kinetic, magnetic,
797: and internal energies.}
798: %but an accumulated error is less than $0.15\%$ in the scope of this study.
799: The panels in Figure \ref{fig:t120} show
800: the late-time evolution of the two runs at $t/\tau_c=120$.
801: For comparison, the upper halves present the results of run A,
802: while the lower halves present those of run B.
803: \newrev{At this time, there should be no substantial boundary effects,
804: since the evolution time of the reconnection region is smaller than
805: a wave propagation time to the $z$-periodic boundaries.
806: In addition, the magnetic field lines at the $X$-type region
807: come from $z/\lambda\sim \pm 15$ around $x/\lambda = \pm 76.8$
808: as seen in Figure \ref{fig:t120}\textit{a}.
809: The reconnection still involves magnetic flux
810: inside the main simulation domain.}
811:
812: One can see the difference
813: in global structure in Figure \ref{fig:t120}\textit{a}.
814: The current sheet seems to be broadened in run A,
815: while it remains thin in run B.
816: The position of the TD front differs, too.
817: The TD is located at $x/\lambda \sim 46$ in run A.
818: On the contrary, the TD is located at $x/\lambda \sim 43$ in run B
819: --- the TD front can not penetrate into the $x$-direction as run A.
820: In Figures \ref{fig:stack}\textit{a-d},
821: the $B_z$ profile along the neutral line in run B
822: is also presented by a dotted line.
823: The peak plasma density at the downstream of the TD is
824: $d/d_0 \sim 2.4$ in run A,
825: while plasmas are much compressed near the TD; $d/d_0 \sim 5$ in run B.
826: These differences can be explained by
827: the current sheet expansion by the Weibel instability.
828: In run A, the Weibel instability transfers some of plasma $x$-momentum
829: into $z$-momentum (and also $y$-momentum via meandering motion).
830: Therefore, it reduces plasma $x$-pressure in the downstream region,
831: and the TD front can move further distance into $x$-direction.
832: The increased $z$-momentum leads to the current sheet expansion,
833: and then magnetic field lines become more round.
834: The current sheet continues to expand
835: as long as plasma $z$-momentum is continuously supplied
836: through the Weibel activity,
837: from reconnection outflow in the upstream region.
838: On the contrary, plasmas and anti-parallel field lines are
839: rather confined near the current sheet in run B.
840: The TD can not penetrate into the outflow direction as run A,
841: therefore,
842: the maximum amplitude of the pileup field is stronger
843: (See Figs. \ref{fig:stack}\textit{a-d} for the $B_z$ profiles in run B)
844: and
845: the inflow speed near the TD is slightly slower than in run A
846: (Fig. \ref{fig:t120}\textit{d}).
847: %The current sheet in the outflow side of the TD
848: %continues to expand for a while.
849: %The thickness is up to $8(\pm 4)$ at $t/\tau_c=120$.
850:
851: The total reconnected flux $\Sigma|B_z|$
852: along the neutral plane is the same in runs A and B,
853: because it is controlled by the physics of the upstream region;
854: magnetic reconnection near the $X$ points.
855: %
856: Around $(x,z)\sim(0,0)$, we observe
857: a small magnetic loop in the current sheet.
858: \newrev{This is not a projection of the $X$-point,
859: but a secondary magnetic island},
860: which appears after $t/\tau_c \sim 105$.
861: Its formation mechanism is unclear.
862: %Its typical length is much shorter than
863: %the wavelength of the conventional tearing mode.
864: %The fine structure of the island is
865: %slightly different in run A and run B.
866: %We also found growth of the density enhancement around $x/\lambda \sim 26$,
867: %which will be the initial stage of the secondary island structure.
868: %The small scale structure of the island slightly differs,
869: %$B_y$ or anisotropy-driven instabilities may play some role in it.
870: Since these secondary islands are also found in
871: non-relativistic reconnection in pair plasmas \citep{dau07},
872: the island formation will be common feature
873: in a low-density current sheet in large scale simulations.
874: Anyway, %except for the slight difference in the second islands,
875: we do not see noticeable difference ---
876: both run A and run B are almost same in the upstream region.
877: \newrev{At this stage the Weibel activity is not likely affecting
878: the central reconnection region,
879: because information can not have widely spread over
880: the central reconnection region.
881: For example, an information of large $B_y$-perturbation at $x/\lambda \sim 20$
882: at $t/\tau_c=80$ (Fig. \ref{fig:linear}) can not
883: arrive at the $X$-type region before $t/\tau_c = 100$.
884: On a longer time scale,
885: the Weibel instability may have an impact on the reconnection rate
886: since it changes the magnetic field line topology
887: by expanding the current sheet \cite{swisdak}.}
888:
889: In the case of run A,
890: Figure \ref{fig:t120}\textit{b} and
891: Figure \ref{fig:t120}\textit{c} show
892: the downstream field structure in more detail.
893: In Figure \ref{fig:t120}\textit{b},
894: the magnetic field lines near $x/\lambda \sim \pm 1$
895: at the right boundaries
896: are set to connect to $x/\lambda = \pm 1$ at periodic boundaries.
897: The field line shifts to $z/\lambda \sim 3$ at the thickest point
898: due to the current sheet expansion in run A,
899: while the field line stays around $z/\lambda \sim -1$ in run B.
900: In this stage, the Weibel instability is also active
901: outside the neutral line around $z/\lambda \sim 2$,
902: as well as along the neutral line.
903: The $xz$ current system is well developed around the $B_y$ regions.
904: Charge distribution (Fig. \ref{fig:t120}\textit{c}) is
905: correlated to the $x$-current system (Fig. \ref{fig:t120}\textit{b});
906: we see positron-rich $J_x>0$ region and electron-rich $J_x<0$ region.
907: Compared with the early stage in Figure \ref{fig:snap},
908: these structures are rather elongated into $x$-direction.
909: This is consistent with many studies on Weibel instability;
910: elongated ``filament'' structure or current channels are
911: commonly observed in well-developed stage of the Weibel instability.
912: In run B, there is no current system in the $xz$ plane.
913: Regarding the outflow structure in Figure \ref{fig:t120}\textit{d},
914: plasma flow is rather bifurcated in run A,
915: due to the $z$-displacement of plasmas.
916: We can see a significant difference in the $J_y$ current structure
917: in Figure \ref{fig:t120}\textit{e}.
918: In run A, the current region is located in front
919: of the broadened plasma region.
920: The energy conversion $\vec{J}\cdot\vec{E}$ mainly takes place
921: in the vertical current front there.
922: In run B, the $y$-current structure is enhanced
923: around the small spot near the TD,
924: and then energy conversion takes place there.
925:
926: Regarding the composition of the energy in the system of interest,
927: two runs slightly differ in accordance with the field line topology;
928: the summary of the upstream reconnection field energy
929: $\Sigma{(B_x^2+B_z^2+E_y^2)/8\pi}$ is almost same in both two runs,
930: but run A has more field energy (120\%) in $\Sigma{(B_z^2+E_y^2)/8\pi}$
931: compared with run B.
932: %This is due to modulation of the magnetic fields around the TD.
933: The total plasma kinetic energy
934: $\Sigma{(\gamma-1)mc^2}$ is almost same, however,
935: in run A, a slight percentage of them ($0.5\%$) are converted again
936: into the Weibel-related field energy; $\Sigma{(B_y^2+E_x^2+E_z^2)/8\pi}$.
937: The total amount of Weibel-related field energy is equivalent to
938: $\sim 5.3 \lambda^2 (B_0^2/8\pi) $.
939: This is substantially smaller
940: ($10^{-1}$-$10^{-2}$) than that of the upstream-related field energies,
941: because the Weibel active region is relatively small.
942: %After $t/\tau_c=120$,
943: %the accumulated numerical error becomes comparable to
944: %the small , so it's not discuss it.
945: %We expect that the vertical current sheet structure
946: %enables efficient energy conversion from the magnetic fields
947: %to downstream plasmas energies.
948: %However, we do not observe significant difference of energy composition
949: %during the simulation time until $t/\tau_c=120$.
950: %Probably this stage is too early to observe such difference.
951:
952: Figure \ref{fig:espec} presents
953: energy spectra in the regions of interest.
954: The spectra of two runs look similar, too.
955: However, in order to distinguish the difference clearer,
956: these spectra are divided into two parts by the TD;
957: in the downstream region of the TD,
958: and in the upstream region of the TD.
959: Note that the TD is located in the further downstream in run A.
960: We observe a high-energy nonthermal tails in their spectra
961: in the upstream side.
962: This is due to $dc$ particle acceleration or piled-up acceleration
963: in the upstream side \citep{zeni01,zeni07a}.
964: In the mid-energy range ($20\lesssim \varepsilon/mc^2 \lesssim 60$),
965: run A has slightly more high-energy population.
966: We think this is due to the larger volume of the upstream region.
967: Since particles can stay longer
968: inside the larger upstream region,
969: or the main site of particle acceleration,
970: more particles are accelerated into high energy range in run A.
971: In the low-energy range around $\varepsilon/mc^2 \sim 10$,
972: We expected that plasma heating is enhanced through
973: multiple interaction by the TD in run A,
974: but the enhancement is too small (even in linear scaling).
975: One reason is that the Weibel region is too small.
976: Furthermore, the Weibel region not only reflects the escaping particles,
977: but also it hits the pre-existing particles.
978: So, as a result, the net effect will be small.
979: Meanwhile, in run B, plasma population is slightly enhanced
980: around $\varepsilon/mc^2 \sim 20$.
981: It is difficult to discuss this energy range,
982: because too many effects are relevant.
983:
984:
985: \section{Discussions and Summary}
986:
987: In ion-electron reconnection,
988: it is well known that quadropolar out-of-plane fields $B_y$
989: appears in the vicinity of the $X$-type region
990: \citep{hesse94,oe01,nagai01}.
991: However, quadropolar structure disappears
992: in pair plasmas with an equal temperature \citep{bessho07}.
993: In the present case, the out-of-plane fields are found
994: in the downstream of the reconnection outflow region.
995: \citet{dau07} reported similar structure
996: in non-relativistic pair plasma reconnection, and
997: they argued that it comes from some type of firehose instability.
998: Since both the firehose instability and the Weibel instability
999: belong to the anisotropy-driven instabilities,
1000: the generation of the out-of-plane fields in the downstream
1001: would be common feature in pair plasma reconnection.
1002: %
1003: We identified that
1004: the out-of-plane field $B_y$ is generated
1005: by the Weibel instability downstream of the TD.
1006: A comparison with another run (with $B_y$ artificially suppressed)
1007: demonstrates that
1008: the Weibel instability leads to a significant modulation
1009: of the downstream structure; further penetration of outflows,
1010: the current sheet expansion and the bifurcated downstream jets.
1011: We expect that
1012: the formation of ``T-shaped current sheet'' \citep{claus04}
1013: can be explained by the current sheet expansion
1014: by the Weibel instability.
1015:
1016: %Next, we discuss possible three dimensional extensions.
1017: In three dimensions,
1018: the Weibel instability also generates
1019: the vertical magnetic field $\pm \delta B_z$ and
1020: the out-of-plane electric field $\pm \delta E_y$.
1021: \newrev{The Weibel instability will lead to
1022: a filament-like development of the $x$-currents,
1023: involving small-scale reconnection of
1024: perturbed magnetic field lines \citep{cal01}.
1025: The plasma's $x$-momentum will be transferred
1026: to $y$-momentum as well as to $z$-momentum.
1027: We expect that
1028: the TD penetrates further downstream into the outflow region,
1029: because plasma $x$-pressure will be more efficiently scattered.
1030: Meanwhile, the current sheet expansion in the downstream region
1031: may be less apparent,
1032: because all of the $x$-momentum is not transformed into $z$-momentum.
1033: Regarding the particle acceleration,
1034: we will observe more high-energy particles,
1035: because the TD will further penetrate into downstream and then
1036: the upstream acceleration site expands.
1037: In addition, high-energy particles from the upstream region
1038: may also be affected by the Weibel fields.
1039: In two dimensional case, such high energy particles are
1040: rather insensitive to the Weibel region, and
1041: low-energy reflected particles bounce
1042: between the TD and the Weibel-active region.
1043: However, in three dimensions,
1044: the Weibel magnetic field can affect high energy particles,
1045: especially when it is antiparallel ($B_z<0$) to the pile-up field.
1046: Thus, some higher-energy particles may bounce
1047: between the TD and the Weibel-active region.
1048: Along with the expansion of the upstream acceleration site,
1049: particle acceleration is likely to be enhanced. }
1050: %In unmagnetized pair plasmas,
1051: %when the Weibel-generated magnetic fields are
1052: %perpendicular to the two-dimensional plane of interest,
1053: %PIC simulation addressed that
1054: %the shock-drift-like acceleration is slightly enhanced.
1055: % \citep{kato07}.
1056:
1057: Furthermore, we should consider
1058: all other instabilities in three dimensions.
1059: It is known that the relativistic drift kink instability (RDKI)
1060: quickly modulates the current sheet in a relativistic pair plasmas
1061: \citep{zeni05a,zeni07a}.
1062: Its typical growth rate in this configuration is
1063: $\tau_c\omega_i \sim 0.1$,
1064: while $\tau_c\omega_i \sim 0.03$ for the tearing instability.
1065: Although the RDKI grows slower than the Weibel instability,
1066: the RDKI is a macro instability,
1067: and it may inhibit the reconnection process
1068: by modulating the current sheet \citep{zeni05b},
1069: while the Weibel instability is
1070: the sub-product of the reconnection outflow.
1071: Since the RDKI slowly widens the current sheet,
1072: unmagnetized or weakly magnetized region becomes wider.
1073: Therefore, we expect that the Weibel instability is active
1074: in a wider region inside the modulated current sheet.
1075: \newrev{In addition, since the Weibel instability involves $y$-structure,
1076: repeated collision between the TD and the Weibel fields may lead to
1077: the instability of the TD front in the $xy$ plane
1078: %(e.g. the Richtmayer-Meshkov instability or
1079: (e.g. the interchange instability of the reconnection jet front \citep{naka02}).
1080: The Weibel instability in three dimensions
1081: will be an interesting problem to challenge. }
1082:
1083: The Weibel instability will also occur
1084: under non-relativistic temperature condition of $T \ll mc^2$.
1085: In this regime, usually the electron skin depth becomes
1086: smaller than the other scales like electron gyro radius, and so
1087: % the two-dimensional evolution of reconnection is
1088: % the fastest macro process, because the RDKI substantially slows down.
1089: the Weibel instability occurs
1090: in a shorter time/spatial scale in reconnection.
1091: As long as it occurs inside reconnection outflow structure,
1092: the physics will be the same.
1093:
1094: In ion-electron plasmas,
1095: the Weibel instability will work for electrons,
1096: and then it may contribute to quick electron heating.
1097: Although it is not clear whether
1098: the sharp TD is formed in the outflow region in ion-electron plasmas,
1099: the multiple interaction with the TD will also be possible.
1100: Similarly, enhanced heating may also occur
1101: near the fast shock or the other discontinuities.
1102: \newrev{
1103: In solar cases, it is reported that
1104: hard X-ray emission comes from the small spot
1105: near the loop top of magnetic field lines \citep{masuda94}.
1106: Shock-related electron heating may take place
1107: in the downstream of reconnection outflow, and
1108: the Weibel instability and the relevant bounce effect
1109: may contribute to the enhanced electron heating there.
1110: % Since the Weibel instability is active in the unmagnetized or
1111: % weakly magnetized region,
1112: % it is somewhat reasonable that
1113: % the enhanced heating occurs in the loop top,
1114: % not in the loop region where magnetic fields are strong.
1115: In addition, the Weibel instability may play a role in quick electron heating
1116: inside the nanoflare jets, in the context of coronal heating problem.
1117: %because the instability occurs in the small electron inertia scale.
1118: }
1119:
1120: In more generalized configuration of magnetic reconnection,
1121: for example, in magnetic reconnection with uniform guide field $B_y$,
1122: the situation will differ substantially.
1123: Since the guide field $B_y$ scatters $x$-momentum into $z$-momentum,
1124: the wavevector of the Weibel instability is likely to be in the $y$ direction.
1125: However, since the outflow is slower than the antiparallel case,
1126: and since the ambient magnetic field $B_y$ stabilizes the instability,
1127: the Weibel instability will be less active.
1128: The situation will be more complicated in relativistic pair plasmas,
1129: because charge neutrality often breaks down
1130: in the outflow region \citep{zeni07b}.
1131: Therefore, how plasma anisotropy disappears in the guide field case
1132: remains to be solved.
1133: %Although three dimensional evolution of the reconnection system
1134: %is not yet well understood \citep{claus04,zeni05b,zeni07b}.
1135:
1136: On the viewpoint of energetics,
1137: the ultimate energy source of the Weibel instability is
1138: the plasma bulk energy of the reconnection jet,
1139: which is expelled by the magnetic energy in the inflow region.
1140: Initial magnetic energy is converted to
1141: plasma energy of reconnection jet, and
1142: partially to magnetic energy of the Weibel region.
1143: Then, the Weibel activity modifies the downstream reconnection structure,
1144: which potentially changes the downstream energy conversion process once again!
1145:
1146:
1147: Finally, let us briefly summarize this paper.
1148: We investigated the role of the Weibel instability in the reconnection context.
1149: We demonstrated the following new results;
1150: (1) the Weibel instability occurs in the downstream of the reconnection outflow,
1151: (2) the counter-streaming Weibel instability is also affected by the relativistic pressure effect,
1152: and (3) the Weibel instability significantly modifies the downstream reconnection structure. Since the Weibel instability is a micro process, it may play a role in various macro instabilities, such as magnetic reconnection, the RDKI, the Kelvin-Helmholtz instability, as well as collisionless shocks.
1153:
1154: %study the properties relativistic
1155: %the energy of the instability is supplied by the plasma kinetic energy,
1156: %while the relevant processes (relativistic reconnection or
1157: %the relativistic drift kink instability) in the RCS
1158: %dissipates the magnetic energy into the plasma heat.
1159: %At present the RWI does not seem to affect the upstream physics,
1160: %but it slightly moderates the overall energy dissipation.
1161: %We found that the RWI works as a scattering process in the shock acc.
1162:
1163:
1164: \begin{acknowledgments}
1165: The authors enjoyed fruitful discussions with K. Schindler, M. Kuznetsova,
1166: S. Saito, M. Swisdak and Y. Liu.
1167: This research was supported by facilitates of JAXA
1168: and the NASA Center for Computational Sciences.
1169: \end{acknowledgments}
1170:
1171: \begin{thebibliography}{}
1172:
1173: \bibitem[Lesch \& Birk(1997)]{lb97} H. Lesch, G. T. Birk, \aap, 324, 461 (1997)
1174: \bibitem[di Matteo(1998)]{dimatteo} T.~{di Matteo}, \mnras, 299, L15 (1998)
1175: \bibitem[Birk et al.(2001)]{birk01} G. T. Birk, A. R. {Crusius-W{\"a}tzel}, H. {Lesch}, \apj, 559, 96 (2001)
1176: \bibitem[Coroniti(1990)]{coro90} F. V. Coroniti, \apj, 349, 538 (1990)
1177: \bibitem[Lyubarsky \& Kirk(2001)]{lyu01} Y. Lyubarsky, J. G. Kirk, \apj, 547, 437 (2001)
1178: \bibitem[Kirk \& Skj\ae raasen(2003)]{kirk03} J. G. Kirk, O. Skj\ae raasen, \apj, 591, 366 (2003)
1179: \bibitem[Drenkhahn(2002)]{dr02} G.~{Drenkhahn}, \aap, 387, 714 (2002)
1180: \bibitem[Drenkhahn \& Spruit(2002)]{drs02} G.~{Drenkhahn}, H.~C. {Spruit}, \aap, 391, 1141 (2002)
1181: \bibitem[Thompson \& Duncan(1995)]{thom95} C. Thompson, R. C. Duncan, \mnras, 275, 255 (1995)
1182: \bibitem[Lyutikov(2003)]{lyut03a} M. Lyutikov, \mnras, 346, 540 (2003)
1183: %\bibitem[Aharonian et al.(2004)]{aha04} F. Aharonian et al. 2004, \apj, 614, 897
1184: %\bibitem[Begelman et al.(1984)]{bg84} M. C. Begelman, R. D. Blandford and M. J. Rees 1984, Rev. Modern Phys., 56, 255
1185: \bibitem[Zenitani \& Hoshino(2001)]{zeni01} S. Zenitani, M. Hoshino, \apj, 562, L63 (2001)
1186: \bibitem[Jaroschek et al.(2004)]{claus04} C. H. Jaroschek, R. A. Treumann, H. Lesch, M. Scholer, \pop, 11, 1151 (2004)
1187: \bibitem[Zenitani \& Hoshino(2005b)]{zeni05b} S. Zenitani, M. Hoshino, \prl, 95, 095001 (2005)
1188: \bibitem[Zenitani \& Hoshino(2007)]{zeni07a} S. Zenitani, M. Hoshino, \apj, 670, 702 (2007)
1189: \bibitem[Zenitani \& Hoshino(2008)]{zeni07b} S. Zenitani, M. Hoshino, \apj, 677, 530 (2008)
1190: \bibitem[Bessho \& Bhattacharjee(2007)]{bessho07} N. Bessho, A. Bhattacharjee, \pop, 14, 056503 (2007)
1191: \bibitem[Hesse \& Zenitani(2007)]{hesse07} M. Hesse, S. Zenitani, \pop, 14, 112102 (2007)
1192: \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, P. L. Pritchett, \jgr, 106, 3715 (2001)
1193: %\bibitem[Blackman \& Field(1994)]{bf94b} E. G. Blackman and G. B. Field 1994, \prl, 72, 494
1194: \bibitem[Piran(1999)]{piran} T. Piran, Physics Reports, 314, 575 (1999)
1195: \bibitem[Medvedev \& Loeb(1999)]{med99} M. V. Medvedev, A. Loeb, \apj 526, 697 (1999)
1196: \bibitem[Weibel(1959)]{weibel} E. S. Weibel 1959, \prl 2, 83
1197: \bibitem[Kazimura et al.(1998)]{kazi98} Y. Kazimura, J. I. Sakai, T. Neubert, S. V. Bulanov, \apj, 498, L186 (1998)
1198: \bibitem[Silva et al.(2003)]{silva03} L. O. Silva, R. A. Fonseca, J. W. Tonge, J. M. Dawson, W. B. Mori, M. V. Medvedev, \apj, 596, L121 (2003)
1199: \bibitem[Nishikawa et al.(2003)]{nishikawa03} K. I. Nishikawa, P. Hardee, G. Richardson, R. Preece, H. Sol, G. J. Fishman, \apj, 595, 555 (2003)
1200: \bibitem[Frederiksen et al.(2004)]{fre04} J. T. Frederiksen, C. B. Hededal, T. Haugbolle, A. Nordlund, \apj, 608, L13 (2004)
1201: \bibitem[Hededal et al.(2004)]{hed04} C. B. Hededal, T. {Haugb{\o}lle}, J. T. {Frederiksen}, {\AA}. Nordlund, \apj, 617, L107 (2004)
1202: \bibitem[Jaroschek et al.(2005)]{claus05} C. H. Jaroschek, H. Lesch, R. A. Treumann, \apj, 618, 822 (2005)
1203: \bibitem[Nishikawa et al.(2006)]{nishikawa06} K. I. Nishikawa, P. E. Hardee, C. B. Hededal, G. J. Fishman, \apj, 642, 1267 (2006)
1204: \bibitem[Califano et al.(1997)]{cal97} F. Califano, F. Pegoraro, S. V. Bulanov, \pre, 56, 963 (1997)
1205: \bibitem[Saito \& Sakai(2004)]{saito} S. Saito, J. I. Sakai, \pop, 11, 859 (2004)
1206: \bibitem[Yoon \& Davidson (1987)]{yoon87} P. H. Yoon, R. C. Davidson, \pra, 35, 2718 (1987)
1207: \bibitem[Yang et al.(1993)]{yang93} T. Y. B. Yang, Y. Gallant, J. Arons, A. B. Langdon, \textit{Physics of Fluids B}, 5, 3369 (1993)
1208: %\bibitem[Yoon (1989)]{yoon89} P. H. Yoon 1989, Physics of Fluids B, 1, 1336
1209: \bibitem[Yoon (2007)]{yoon07} P. H. Yoon, \pop, 14, 024504 (2007)
1210: \bibitem[Daughton \& Karimabadi(2007)]{dau07} W. Daughton, H. Karimabadi, \pop, 14, 072303 (2007)
1211: %\bibitem[Sakai \& Kawata(1980)]{sakai} J. Sakai, T. Kawata, Physical Society of Japan Journal, 49, 747 (1980)
1212: %\bibitem[Blandford \& Ostriker(1978)]{bo78} R. D. Blandford and J. P. Ostriker 1978, \apj, 179, 573
1213: %\bibitem[Drury(1983)]{dr83} L. O. Drury 1983, Reports of Progress in Physics, 46, 973
1214: %\bibitem[Chiueh(1988)]{ch88} T. Chiueh 1988, \apj, 333, 366%
1215: %\bibitem[Mandt et al.(1994)]{mandt94} M. E. Mandt, R. E. Denton, J. F. Drak, \grl, 21, 73 (1994)
1216: \bibitem[Hesse \& Winske(1994)]{hesse94} M. Hesse, D. Winske, \jgr, 99, 11177 (1994)
1217: \bibitem[{\O}ieroset et al.(2001)]{oe01} M. {\O}ieroset, T. D. Phan, M. Fujimoto, R. P. Lin, R. P. Lepping, \nature, 412, 414 (2001)
1218: \bibitem[Nagai et al.(2001)]{nagai01} T. Nagai, I. Shinohara, M. Fujimoto, M. Hoshino, Y. Saito, S. Machida, T. Mukai, \jgr, 106, 25929 (2001)
1219: %\bibitem[Koyama et al.(1995)]{koyama95} K. Koyama, R. Petre, E. V. Gotthelf, U. Hwang, M. Matsuura, M. Ozaki and S. S. Holt 1995, Nature, 378, 255
1220: %\bibitem[Lyubarsky \& Eichler(2006)]{lyu06} Y. Lyubarsky and D. Eichler 2006, \apj, 647, 1250
1221: %\bibitem[Loureiro et al.(2007)]{lou07} N.~F. Loureiro, A. A. {Schekochihin} and S. C. {Cowley} 2007, \pop, 14, 703
1222: \bibitem[Califano et al.(2001)]{cal01} F. Califano, N. Attico, F. Pegoraro, G. Bertin, V. Bulanov, \prl, 86, 5293 (2001)
1223: \bibitem[Zenitani \& Hoshino(2005a)]{zeni05a} S. Zenitani, M. Hoshino, \apj, 618, L111 (2005)
1224: \bibitem[Swisdak(submitted)]{swisdak} M. Swisdak, Y. H. Liu, J. F. Drake, ``Development of a turbulent outflow during electron-positron magnetic reconnection'', \apj (submitted)
1225: \bibitem[Nakamura et al.(2002)]{naka02} M. S. Nakamura, H. Matsumoto, M. Fujimoto, \grl, 29, 88 (2002)
1226: \bibitem[Masuda et al.(1994)]{masuda94} S. Masuda, T. Kosugi, S. Tsuneta, Y. Ogawa, \nature, 371, 495 (1994)
1227: %\bibitem[Parker(1988)]{parker88} E. N. Parker, \apj, 330, 474 (1988)
1228: %\bibitem[TanDokoro \& Fujimoto(2005)]{tan05} R. TanDokoro and M. Fujimoto 2005, \grl, 32, 23
1229: %\bibitem[Schaefer-Rolffs et al.(2006)]{sc06} U. Schaefer-Rolffs, I. Lerche, and R. Schlickeiser 2006, 13, 2107
1230:
1231: \end{thebibliography}
1232:
1233:
1234: \clearpage
1235:
1236: \begin{figure}[thp]
1237: \begin{center}
1238: \ifjournal
1239: \includegraphics[width={\columnwidth},clip]{f1.eps}
1240: \else
1241: \includegraphics[width={\columnwidth},clip]{f1.png}
1242: \fi
1243: \caption{(Color online)
1244: \label{fig:snap}
1245: (a) Snapshot of the right half of the main simulation domain. Magnetic field lines (\textit{contour}), plasma density (\textit{color contour})
1246: and plasma flow (\textit{arrows}).
1247: (b) Out-of-plane field structure ($B_y$) in the selected region and the electric current system (\textit{arrows}) in the $xz$ plane.
1248: (c) Charge distribution $\rho = [d_p-d_e]/d_0$ in color.
1249: The dotted line shows $E_z=0$ and the solid lines are contour of electric field $E_z$ with $\Delta E_z=0.1 B_0$.
1250: }
1251: \end{center}
1252: \end{figure}
1253:
1254: \clearpage
1255:
1256: \begin{figure}[thp]
1257: \begin{center}
1258: \ifjournal
1259: \includegraphics[width={\columnwidth},clip]{f2.eps}
1260: \else
1261: \includegraphics[width={\columnwidth},clip]{f2.pdf}
1262: \fi
1263: \caption{(Color online)
1264: \label{fig:linear}
1265: Time development of out-of-plane magnetic field $B_y$
1266: along the neutral plane ($z=0$).
1267: Profiles at three stages ($t/\tau_c=68,74,80$) are indicated
1268: by \newrev{solid} color lines.
1269: }
1270: \ifjournal
1271: \includegraphics[width={0.8\columnwidth},clip]{f3.eps}
1272: \else
1273: \includegraphics[width={0.8\columnwidth},clip]{f3.png}
1274: \fi
1275: \caption{(Color online)
1276: \label{fig:aniso}
1277: Plasma distribution function in the Weibel active region
1278: ($22 \le x/\lambda \le 26, -2 \le z/\lambda \le 2$) at $t/\tau_c=80$
1279: in the $x$-$z$ four-velocity space, normalized by $c$.
1280: }
1281: \end{center}
1282: \end{figure}
1283:
1284: \clearpage
1285:
1286: \begin{figure}[thp]
1287: \begin{center}
1288: \ifjournal
1289: \includegraphics[width={0.75\columnwidth},clip]{f4.eps}
1290: \else
1291: \includegraphics[width={0.75\columnwidth},clip]{f4.pdf}
1292: \fi
1293: \caption{
1294: \label{fig:weibel}
1295: Dispersion relation of
1296: the two-dimensional purely-growing mode for $k_z=5k_x$.
1297: The growth rate for three counter-streaming velocities ($v=0.6$,$0.65$,$0.7c$),
1298: the cold-beam limit counterpart for $v=0.65c$ (\textit{dashed line}),
1299: and observed rate (\textit{white square}) are presented.
1300: }
1301: \ifjournal
1302: \includegraphics[width={\columnwidth},clip]{f5.eps}
1303: \else
1304: \includegraphics[width={\columnwidth},clip]{f5.png}
1305: \fi
1306: \caption{
1307: \label{fig:weibel2}
1308: \textit{(a)}
1309: Growth rate of the two-dimensional purely-growing mode
1310: as a function of $\vec{k}=(k_x,k_z)$.
1311: \textit{(b)} The same, but for the cold-beam limit case of $p=0$.
1312: }
1313: \end{center}
1314: \end{figure}
1315:
1316: \clearpage
1317:
1318: \begin{figure}[thp]
1319: \begin{center}
1320: \ifjournal
1321: \includegraphics[width={\columnwidth},clip]{f6.eps}
1322: \else
1323: \includegraphics[width={\columnwidth},clip]{f6.pdf}
1324: \fi
1325: \caption{
1326: \label{fig:stack}
1327: \textit{(a-d)} Field properties along the neutral plane ($z=0$);
1328: $B_z/B_0$ (\textit{thick line}), $B_y/B_0$ (\textit{thin line}),
1329: and $E_y/B_0$ (\textit{dashed line}) are presented.
1330: The dotted line shows $B'_z/B_0$, obtained from a simulation without $B_y$.
1331: \textit{(e-g)} Properties of selected particles;
1332: \textit{(e)} energy,
1333: \textit{(f)} velocity, and
1334: \textit{(g)} fields at its position are shown.
1335: }
1336: \end{center}
1337: \end{figure}
1338:
1339: \clearpage
1340:
1341: \begin{figure}[thp]
1342: \begin{center}
1343: \ifjournal
1344: \includegraphics[width={\columnwidth},clip]{f7.eps}
1345: \else
1346: \includegraphics[width={\columnwidth},clip]{f7.png}
1347: \fi
1348: \caption{(Color online)
1349: \label{fig:dist}
1350: \textit{(a)} Spatial distribution of selected positrons, whose $x$-velocity is slower than TD.
1351: \textit{(b)} Spatial distribution of positrons, faster than TD.
1352: }
1353: \ifjournal
1354: \includegraphics[width={\columnwidth},clip]{f8.eps}
1355: \else
1356: \includegraphics[width={\columnwidth},clip]{f8.png}
1357: \fi
1358: \caption{(Color online)
1359: \label{fig:t120}
1360: \textit{(a-c)} Same as Fig. 1, but for $t/\tau_c=120$ in run A (\textit{upper half}) and in run B (\textit{buttom half}).
1361: \textit{(d)} $x$-velocity and \textit{(e)} $y$-current at $t/\tau_c=120$ in run A (\textit{upper half}) and in run B (\textit{buttom half}).
1362: }
1363: \end{center}
1364: \end{figure}
1365:
1366: \begin{figure}
1367: \begin{center}
1368: \ifjournal
1369: \includegraphics[width={0.9\columnwidth},clip]{f9.eps}
1370: \else
1371: \includegraphics[width={0.9\columnwidth},clip]{f9.pdf}
1372: \fi
1373: \caption{
1374: \label{fig:espec}
1375: Energy spectra in the right half main simulation domain.
1376: }
1377: \end{center}
1378: \end{figure}
1379:
1380: \end{document}
1381: