1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass[]{emulateapj}
3:
4:
5:
6: %%%%%%%%%%%%%%%%%%
7: % Macros
8: %%%%%%%%%%%%%%%%%%
9: \newcommand {\omegapez} {\omega_\mathrm{pe,0}}
10: %%%%%%%%%%%%%%%%%%
11:
12:
13:
14: \shorttitle{COLLISIONLESS SHOCKS IN UNMAGNETIZED PAIR PLASMAS}
15: \shortauthors{KATO}
16:
17: \begin{document}
18:
19:
20:
21:
22:
23: \title{Relativistic Collisionless Shocks in Unmagnetized Electron-Positron Plasmas}
24:
25: \author{Tsunehiko N. Kato\altaffilmark{1}}
26: \affil{National Astronomical Observatory of Japan,
27: 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan}
28: \altaffiltext{1}{
29: Present address:
30: Institute of Laser Engineering, Osaka University,
31: Yamada-oka 2-6, Suita, Osaka 565-0871, Japan
32: }
33: \email{kato-t@ile.osaka-u.ac.jp}
34:
35:
36:
37:
38:
39: \begin{abstract}
40: It is shown that collisionless shock waves can be driven
41: in unmagnetized electron-positron plasmas
42: by performing a two-dimensional particle-in-cell simulation.
43: At the shock transition region,
44: strong magnetic fields are generated by a Weibel-like instability.
45: The generated magnetic fields are strong enough to deflect
46: the incoming particles from upstream of the shock at a large angle
47: and provide an effective dissipation mechanism for the shock.
48: The structure of the collisionless shock propagates at an almost constant speed.
49: There is no linear wave corresponding to the shock wave and therefore
50: this can be regarded as a kind of ``instability-driven'' shock wave.
51: The generated magnetic fields rapidly decay in the downstream region.
52: It is also observed that
53: a fraction of the thermalized particles
54: in the downstream region
55: return upstream through the shock transition region.
56: These particles interact with the upstream incoming particles and
57: cause the generation of charge-separated current filaments
58: in the upstream of the shock as well as
59: the electrostatic beam instability.
60: As a result, electric and magnetic fields are generated
61: even upstream of the shock transition region.
62: No efficient acceleration processes of particles were observed in our simulation.
63: \end{abstract}
64:
65:
66:
67:
68: \keywords{shock waves --- plasmas --- instabilities --- magnetic fields --- acceleration of particles}
69:
70:
71:
72:
73: %%%%%%%%%%%%%%%%%%
74: \section{Introduction}
75: %%%%%%%%%%%%%%%%%%
76: Collisionless shocks are driven
77: in various situations in astrophysical plasmas.
78: They often accelerate particles and generate
79: nonthermal high-energy particles.
80: Such shocks in supernova remnants are known
81: as the accelerators of nonthermal electrons
82: \citep[e.g., SN1006;][]{Koyama95},
83: and are also believed to be the accelerators of cosmic rays
84: with energies up to the knee energy ($\sim 10^{15.5}$ eV).
85: Collisionless shocks also exist
86: in electron-positron plasmas,
87: for example, that in the Crab nebula,
88: and they are considered to generate nonthermal,
89: high-energy electrons and positrons as well.
90:
91:
92: Recently,
93: it has been suggested from observations that
94: the magnetic fields are amplified or generated
95: around collisionless shocks in several supernova
96: remnants \citep{Vink03, Bamba03, Voelk05}.
97: These magnetic fields may be generated by
98: the high-energy particles accelerated at the shocks \citep[e.g.,][]{Bell04}.
99: On the other hand,
100: such a generation mechanism
101: can be related to the microphysics of the collisionless shocks themselves.
102: For the generation of magnetic fields
103: in the relativistic shocks associated with
104: the afterglows of gamma-ray bursts,
105: \citet{Medvedev99} suggested that
106: the Weibel instability \citep{Weibel59, Fried59}
107: is driven at the shock and generates
108: strong magnetic fields.
109: This mechanism can work in shocks
110: in electron-positron plasmas as well.
111:
112:
113: For electron-positron plasmas,
114: the Weibel instability has been investigated in detail
115: by using numerical simulations.
116: Several particle-in-cell simulations showed that
117: in counterstreaming electron-positron plasmas,
118: the Weibel instability develops
119: and generates a strong, sub-equipartition magnetic field
120: \citep{Kazimura98, Silva03}.
121: Furthermore,
122: some authors found that
123: in their simulations with longer calculation times,
124: a shock-like structure that associates with a strong magnetic field
125: is formed
126: even in unmagnetized electron-positron plasmas
127: \citep{Gruzinov01, Haruki03}.
128: Thus,
129: it should be clarified whether
130: the shock-like structure can be regarded as a shock
131: from a macroscopic point of view,
132: for example,
133: whether the structure dissipates the upstream bulk kinetic energy
134: at the ``shock'' transition region
135: or propagates at a steady speed
136: into the upstream plasma, etc.,
137: as well as other detailed features.
138: If they are considered as shocks,
139: it is also important to elucidate
140: the role of the strong magnetic fields
141: generated by the Weibel-like instability
142: in the dissipation of the shock.
143:
144:
145: In this study,
146: we further investigate the dynamics of collisionless shocks
147: in unmagnetized electron-positron plasmas in detail
148: by performing a high-resolution two-dimensional particle-in-cell simulation
149: and confirm that a kind of collisionless shocks indeed forms
150: mainly due to the magnetic fields generated by the Weibel-like instability.
151: However, we also find
152: that no efficient particle acceleration occurs in the shock
153: and the generated magnetic fields rapidly decay within the shock transition region,
154: while in the real world, particles are accelerated at the shocks and
155: large-scale, sub-equipartition magnetic fields are generated
156: around the shocks.
157: The reasons why these processes are not observed in our simulation may be
158: related to the background magnetic field, the spatial and temporal scales
159: of the simulation, or the composition of the plasma.
160:
161:
162:
163:
164: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
165: \section{Simulation}
166: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
167: In order to investigate the collisionless shocks
168: in electron-positron plasmas without background magnetic fields,
169: we performed numerical simulations.
170: The simulation code is a relativistic, electromagnetic, particle-in-cell code
171: with two spatial and three velocity dimensions (2D3V)
172: developed based on a standard method described by \cite{Birdsall}.
173: Thus,
174: the basic equations of the simulation are
175: the Maxwell's equations (in Gaussian units):
176: %
177: \begin{equation}
178: \frac{1}{c} \frac{\partial \mathbf{E}}{\partial t} = \nabla \times \mathbf{B} - \frac{4\pi}{c} \mathbf{J},
179: \qquad
180: \frac{1}{c} \frac{\partial \mathbf{B}}{\partial t} = -\nabla \times \mathbf{E},
181: \end{equation}
182: with
183: \begin{equation}
184: \nabla \cdot \mathbf{E} = 4\pi \rho,
185: \qquad
186: \nabla \cdot \mathbf{B} = 0,
187: \end{equation}
188: %
189: as constraints,
190: where $c$ is the speed of light;
191: $\mathbf{E}$ and $\mathbf{B}$, the electric and magnetic fields, respectively;
192: $\mathbf{J}$, the current density;
193: and $\rho$, the charge density,
194: together with the equation of motion of particles:
195: %
196: \begin{equation}
197: \frac{d \mathbf{u}}{dt}
198: =
199: \frac{q}{mc} \left( \mathbf{E} + \frac{\mathbf{u} \times \mathbf{B}}{\gamma} \right),
200: \end{equation}
201: %
202: where $q$ and $m$ are the charge and mass of the particle, respectively;
203: $\gamma$, the Lorentz factor of the particle;
204: and $\mathbf{u} \equiv \gamma \mathbf{v}/c$, the 4-velocity of the particle
205: (where $\mathbf{v}$ is the ordinary 3-velocity).
206: Particle velocities are often denoted in terms of the 4-velocity
207: throughout this paper.
208:
209:
210: In the following,
211: we consider $\tau = \omegapez^{-1}$ as the unit of time
212: and the electron skin depth $l_0 = c \omegapez^{-1}$ as the unit of length,
213: where $\omegapez \equiv (4\pi n_{e0} e^2 / m_e)^{1/2}$
214: is the electron plasma frequency defined for the mean electron
215: number density $n_{e0}$;
216: $m_e$ is the electron mass,
217: and $-e$ is the electron charge.
218: The units of electric and magnetic fields are both taken as
219: $E_* = B_* = c (4\pi n_{e0} m_e)^{1/2}$;
220: they are defined so that
221: their corresponding energy densities
222: are both equivalent to the half of
223: the mean electron rest mass energy density.
224:
225:
226: The simulations were performed
227: on a grid of $N_x \times N_y = 4096 \times 512$
228: with $3.2 \times 10^8$ particles for each species.
229: The physical size of the simulation box is $L_x \times L_y = 480 \times 60$
230: in the unit of $l_0$.
231: Thus, the size of a cell is $(\Delta x, \Delta y) = (0.12, 0.12)$ in the same unit.
232: The time step is taken as $\Delta t = 0.025$.
233:
234:
235: Since we consider collisionless shocks in
236: unmagnetized plasmas,
237: the electromagnetic fields were initially set to zero
238: over the entire simulation box.
239: The boundary condition of the electromagnetic field
240: is periodic in each direction.
241:
242:
243: In this simulation,
244: a collisionless shock wave is driven
245: according to the well-known ``injection method.''
246: There are two walls at $x=30$ and $450$.
247: They reflect particles specularly,
248: but let the electromagnetic waves pass through freely;
249: the electromagnetic waves propagating
250: outside the walls ($0 < x < 30$ and $450 < x < 480$)
251: are dissipated by means of Ohmic dissipation
252: so that they cannot reach the wall at the other side.
253: Initially, both the electrons and positrons are loaded uniformly
254: in the region between the two walls
255: with a bulk velocity of $u_x = 2.0$
256: (thus, the corresponding bulk Lorentz factor is $\Gamma = 2.24$).
257: Their thermal velocities were given so that
258: the distribution of each component of the 4-velocity $u_i$ ($i = x, y, z$)
259: obeys the Gaussian distribution
260: with a standard deviation of $\sigma_\mathrm{th} = 0.1$ in the plasma
261: rest frame, or the ``fluid'' frame.
262: (In effect, this generates the Maxwell distribution.)
263: At the initial stage of the simulation,
264: particles that were located near the right wall (at $x = 450$)
265: were reflected by the wall
266: and then interacted with the incoming particles,
267: i.e., the upstream plasma.
268: This interaction causes some instabilities
269: and then a collisionless shock is formed.
270: It should be noted that the frame of the simulation
271: corresponds to the downstream rest frame.
272: Thus,
273: we observed the propagation of the shock
274: from right to the left in the downstream rest frame.
275:
276:
277:
278:
279: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
280: \section{Results}
281: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
282: The following figures show the quantities at the end of the simulation ($t = 233$)
283: otherwise stated.
284:
285:
286: %%%%%
287: \subsection{Basic Features of Collisionless Shock}
288: %%%%%
289: Figure~\ref{fig:ne} shows the electron number density
290: normalized to the initial number density,
291: or the far upstream number density,
292: measured in the downstream rest frame, $n_{e0}$.
293: In this figure,
294: the left-hand side and the right-hand side
295: correspond to upstream and downstream
296: of the shock, respectively.
297: We observe that the number density rapidly increases
298: while crossing the transition region, which extends roughly from $x=350$ to $400$,
299: and then it becomes homogeneous by moving further downstream.
300: The transition region does not have
301: a one-dimensional structure
302: but a two-dimensional filamentary structure with
303: density fluctuations along the $y$ direction.
304: The typical size of the filaments
305: is in the order of several electron skin depths.
306: %%%%%
307: \begin{figure}
308: \plotone{f1.eps}
309: \caption{
310: \label{fig:ne}
311: Electron number density at $t=233$.
312: The value is normalized to the far upstream number density
313: measured in the downstream rest frame (i.e., the simulation frame).
314: The left-hand side and the right-hand side of this figure correspond to
315: upstream and downstream of the shock, respectively.
316: The transition region extends from $x=350$ to $400$.
317: }
318: \end{figure}
319: %%%%%
320:
321:
322:
323: Figure~\ref{fig:ne_history} shows the time development of the electron number density.
324: The horizontal and vertical axes represent the $x$ coordinate and time, respectively.
325: The plotted number density is averaged over the $y$ direction.
326: We observe that
327: the transition region, which is visible as the jump in the number density,
328: propagates upstream, i.e., to the left, with time at an almost constant speed.
329: From this figure,
330: the propagation speed measured in the downstream rest frame
331: is estimated to be $V_\mathrm{sh,d} \sim -0.39c$.
332: %%%%%
333: \begin{figure}
334: \plotone{f2.eps}
335: \caption{
336: \label{fig:ne_history}
337: Time development of the electron number density.
338: The horizontal axis is the $x$ coordinate
339: and the vertical axis is the time.
340: The plotted electron number density is
341: averaged over the $y$ direction.
342: [\textit{See the electronic edition of the Journal for a
343: color version of this figure}.]
344: }
345: \end{figure}
346: %%%%%
347:
348:
349:
350:
351: The thin structure propagating upstream at almost the speed of light,
352: which is visible in the lower-left portion of this figure,
353: is due to the particles reflected at the right wall
354: at a very early stage of the simulation.
355: This is of course the consequence of the initial condition,
356: but such a situation may be realized in real cases.
357: Anyway,
358: this structure fades away with time,
359: and it would not affect the structure around the transition region later.
360:
361:
362:
363: Figure~\ref{fig:profile} shows the $y$-averaged profiles of three quantities
364: around the shock transition region:
365: (a) the electron number density, $n_e / n_{e0}$;
366: (b) the mean velocity of electrons in the $x$-direction, $V_x / c$;
367: and (c) the magnitude of electric and magnetic fields, $|E| / E_*$
368: and $|B| / B_*$.
369: In Fig.~\ref{fig:profile}a,
370: the solid curve and the dashed curve represent the number densities
371: at $t = 233$ and $t=200$, respectively.
372: Comparing the two curves,
373: we see that the transition region propagates upstream keeping its averaged structure almost unchanged.
374: The compression ratio measured in the downstream rest frame is approximately $3.3$.
375: The mean velocity of electrons in the $x$ direction, $V_x$,
376: rapidly approaches zero within the transition region (Fig.~\ref{fig:profile}b).
377: In Fig.~\ref{fig:profile}c,
378: it is notable that a strong magnetic field exists
379: within the transition region ($350 < x < 400$).
380: We also see that
381: behind the shock transition region,
382: the strength of the magnetic field decreases
383: and strong electric and magnetic fields exist
384: even in the upstream region.
385: These points will be mentioned later.
386: %%%%%
387: \begin{figure}
388: \plotone{f3.eps}
389: \caption{
390: \label{fig:profile}
391: The $y$-averaged profiles of quantities at $t=233$:
392: (a) the electron number density $n_e/n_{e0}$
393: (solid curve for $t=233$ and dashed curve for $t=200$),
394: (b) the mean velocity in the $x$-direction $V_x/c$,
395: and (c) the magnetic field $|B|/B_*$ (solid curve)
396: and electric field $|E|/E_*$ (dashed curve).
397: }
398: \end{figure}
399: %%%%%
400:
401:
402:
403: Figure~\ref{fig:Bz_Jx}a shows
404: the $z$ component of the magnetic field, $B_z$.
405: It is clear that a strong magnetic field exists
406: within the transition region ($350 < x <400$),
407: as indicated in Fig.~\ref{fig:profile}c.
408: Figure~\ref{fig:Bz_Jx}b shows
409: the current density in the $x$ direction, $J_x$.
410: We observed that
411: there are considerable number of current filaments that carry currents in the $x$ direction
412: within the transition region.
413: The filamentary structure observed in the number density (Fig.~\ref{fig:ne})
414: is in fact due to the existence of these current filaments.
415: They generate a magnetic field fluctuating in the $y$ direction,
416: as shown in Fig.~\ref{fig:Bz_Jx}a.
417: This situation
418: in which strong magnetic fields fluctuating perpendicular to the direction
419: of streaming are generated is similar to those in the simulations
420: of the Weibel instability in counterstreaming plasmas
421: \citep{Kazimura98, Silva03}.
422: Therefore,
423: it is reasonable to consider that
424: in the transition region
425: in which the upstream plasma is mixed up with the downstream plasma,
426: a large velocity dispersion in the $x$ direction and hence
427: a large anisotropy in the particle velocity distribution are induced;
428: further, the anisotropy drives
429: the Weibel-type instability, which generates the magnetic field.
430: The typical magnitude of the magnetic field in this region is $|B| \sim 0.7 B_*$
431: and the corresponding Larmor radius for the incoming upstream particles with $u=2.0$
432: is $r_g \sim 2.9$,
433: which is comparable to the size of the magnetic field structure,
434: i.e., the current filaments, in the transition region.
435: Therefore,
436: the incoming upstream particles should be isotropised in this region
437: because they are deflected at a large angle by the strong magnetic field at this location.
438: This isotropisation provides an effective dissipation mechanism for the upstream bulk kinetic energy.
439: %%%%%
440: \begin{figure}
441: \plotone{f4.eps}
442: \caption{
443: \label{fig:Bz_Jx}
444: (a) The $z$ component of the magnetic fields $B_z$ and
445: (b) the $x$ component of the current density $J_x$ at $t=233$.
446: }
447: \end{figure}
448: %%%%%
449:
450:
451:
452:
453:
454: Figure~\ref{fig:UB} shows the $y$-averaged profile
455: of the magnetic energy density normalized by the upstream
456: bulk kinetic energy density. We see that
457: the magnetic energy density
458: reaches about 10\% of the upstream bulk kinetic energy density
459: at the peak and decays downstream of the shock.
460: The saturation and decay mechanism of the magnetic field
461: would be similar to those of the Weibel instability \citep{Kato05}.
462: The current and the magnetic field generated by the instability in each filament
463: saturates when the magnetic field becomes strong enough to deflect
464: the current-carrying particles in the filament.
465: This predicts the sub-equipartition magnetic field
466: at saturation and it is consistent with the simulation result.
467: After the saturation,
468: the current filaments will coalesce with each other to
469: form larger filaments with the decreasing magnetic field strength.
470: We observe the evolution of the structure to a larger scale
471: together with the decay of the magnetic field strength
472: downstream of the shock transition region
473: in Figs.~\ref{fig:Bz_Jx}a, \ref{fig:profile}c, and \ref{fig:UB}.
474: %%%%%
475: \begin{figure}
476: \plotone{f5.eps}
477: \caption{
478: \label{fig:UB}
479: The $y$-averaged profile of the magnetic energy density
480: normalized by the upstream bulk kinetic energy density at $t = 233$,
481: $\epsilon_B / \Gamma n_{e0} m_e c^2$.
482: The magnetic field strength reaches a sub-equipartition level
483: at the peak and rapidly decays in the downstream region.
484: }
485: \end{figure}
486: %%%%%
487:
488:
489:
490:
491:
492: Figure~\ref{fig:Vx_KE} shows (a) the fluid velocity of electrons
493: in the $x$-direction, $V_x/c$,
494: and (b) the mean kinetic energy of the electrons measured
495: in the \textit{local} fluid rest frame of the electrons
496: expressed in terms of electron rest mass, $\langle\gamma - 1\rangle_\mathrm{fluid}$.
497: It is evident that
498: the incoming plasma from upstream
499: is rapidly decelerated within the transition region,
500: where the strong magnetic field exists
501: (see also Fig.~\ref{fig:Bz_Jx}a),
502: and its bulk kinetic energy is converted into
503: random or ``thermal'' kinetic energy in the downstream rest frame.
504: %%%%%%
505: \begin{figure}
506: \plotone{f6.eps}
507: \caption{
508: \label{fig:Vx_KE}
509: (a) The $x$ component of the electron mean velocity $V_x/c$ and
510: (b) the mean kinetic energy of the electrons measured
511: in the local plasma rest frame of the electrons,
512: $\langle \gamma - 1 \rangle_\mathrm{fluid}$,
513: at $t=233$.
514: It is evident that the bulk kinetic energy of the upstream plasma
515: is converted into the internal kinetic energy
516: within the transition region.
517: [\textit{See the electronic edition of the Journal for a
518: color version of this figure}.]
519: }
520: \end{figure}
521: %%%%%%
522:
523:
524:
525:
526: Figure~\ref{fig:sgmx_sgmy} shows
527: the standard deviations of (a) $u_x$ of electrons, $\sigma_x$,
528: and (b) $u_y$ of electrons, $\sigma_y$,
529: both measured in the local plasma rest frame
530: of the electrons.
531: We observe that the particles are isotropised
532: in the transition region due to
533: the strong magnetic field existing there.
534: It should be noted that,
535: in this two-dimensional simulation,
536: particles are isotropised only on the $x-y$ plane
537: and not in the $z$ direction
538: \citep[c.f.][]{Haruki03}
539: because only the Weibel modes with wave vectors
540: on the $x-y$ plane can develop
541: and hence only $B_z$ is generated.
542: In a three-dimensional simulation,
543: the Weibel modes with wave vectors
544: in the $z$ direction should develop equally
545: and particles are isotropised in all directions.
546: %%%%%
547: \begin{figure}
548: \plotone{f7.eps}
549: \caption{
550: \label{fig:sgmx_sgmy}
551: Standard deviation of (a) $u_x$ and (b) $u_y$
552: of electrons both measured in the local plasma rest frame
553: of the electrons.
554: The electrons are isotropised when they pass through the transition region.
555: [\textit{See the electronic edition of the Journal for a
556: color version of this figure}.]
557: }
558: \end{figure}
559: %%%%%
560:
561:
562:
563:
564: In a macroscopic view,
565: the features shown in the above figures
566: would be observed as those of shock waves;
567: the shock waves dissipate the upstream bulk kinetic energy
568: in the shock transition region
569: and propagate at an almost constant speed
570: into the upstream plasma.
571: In this sense, we can regard this as a collisionless shock.
572: In this shock,
573: the magnetic field in the transition region
574: generated by the Weibel-like instability
575: plays an essential role in the dissipation process
576: of the collisionless shock in unmagnetized plasmas.
577: It should be noted that
578: the shock has no corresponding fundamental linear wave,
579: which is required to define the Mach number,
580: in contrast to, for example, the perpendicular shocks in magnetized plasmas,
581: in which the magnetosonic wave is the fundamental
582: linear wave of the shock.
583: This point is clear from the fact that
584: the wave vector of the magnetic field in the transition region
585: is in the $y$ direction and not in the $x$ direction
586: or the shock normal in the present case
587: and hence it has an essentially a two-dimensional structure.
588: Thus,
589: it can be said that this is a class of ``instability-driven''
590: shock waves.
591: For such shocks,
592: the concept of the Mach number
593: would not be applicable, and
594: the propagation speed of the shock
595: and the width of the transition region
596: would be essentially determined by the nature
597: of the instability that provides the dissipation mechanism
598: at the shock transition region.
599:
600:
601:
602:
603:
604:
605:
606: %%%%%
607: \subsection{Charge-separated Current Filaments and Backward Flowing Particles}
608: %%%%%
609:
610: From Fig.~\ref{fig:Bz_Jx}b,
611: we observe that there are current filaments even in the upstream region ($x < 350$).
612: Figure~\ref{fig:rho_Ex_Ey}a shows the charge density $\rho$.
613: Comparing these two figures,
614: it is evident that these filaments have a net space charge.
615: Since in each filament, $J_x$ and $\rho$ have the same sign,
616: these filaments are mainly composed of
617: the incoming upstream particles with $u_x > 0$.
618: Figure~\ref{fig:rho_Ex_Ey}b shows
619: the $y$ component of the electric field, $E_y$.
620: We observe that $E_y$ is generated between these filaments
621: due to their space charge.
622: As shown in Fig.~\ref{fig:Bz_Jx}a,
623: these filaments also generate magnetic field $B_z$ around them.
624: Thus,
625: rather strong electric and magnetic fields exist
626: in the upstream region, as shown in Fig.~\ref{fig:profile}c.
627: Figure~\ref{fig:rho_Ex_Ey}c shows
628: the $x$ component of the electric field.
629: We observed that strong electric fields exist
630: at the upstream edge of the transition region
631: and that there is a coherent structure in $E_x$
632: in the upstream of the transition region ($x < 350$).
633: %%%%%
634: \begin{figure}
635: \plotone{f8.eps}
636: \caption{
637: \label{fig:rho_Ex_Ey}
638: (a) Charge density $\rho$,
639: (b) the $y$ component of the electric field $E_y$ and
640: (c) the $x$ component of the electric field $E_x$ at $t=233$.
641: }
642: \end{figure}
643: %%%%%
644:
645:
646: The region where the charge-separated current filaments exist
647: would correspond to ``the charge separation layer''
648: suggested by \citet{Milosavljevic06}.
649: As they pointed out,
650: these filaments need a small fraction
651: of the particles flowing backward to stabilize themselves.
652: Figure~\ref{fig:electron_x_ux} shows
653: the $x-u_x$ phase-space plot of electrons at $t=233$.
654: We observe that
655: the particles flowing backward ($u_x < 0$) are supplied
656: around the shock transition region ($x \sim 350$).
657: These particles would stabilize
658: the charge-separated current filaments in the charge separation layer.
659: They can also drive the electrostatic beam instability.
660: The coherent electrostatic field $E_x$
661: upstream of the transition region
662: ($x < 350$; see Fig.~\ref{fig:rho_Ex_Ey}c)
663: would be induced by this instability.
664: These particles should also contribute to drive the Weibel-type instability
665: around the transition region.
666: %%%%%
667: \begin{figure}
668: \plotone{f9.eps}
669: \caption{
670: \label{fig:electron_x_ux}
671: The $x-u_x$ phase-space plot of electrons at $t=233$.
672: Upstream of the shock transition region ($x < 350$),
673: a fraction of particles flow backward, which causes
674: two-stream type instabilities.
675: In the downstream region ($x > 370$),
676: electrons are thermalized (along the $x$ direction).
677: }
678: \end{figure}
679: %%%%%
680:
681:
682:
683:
684: Figure~\ref{fig:backward_electron_x_t} shows
685: the time development of
686: the number density of electrons again
687: as in Fig.~\ref{fig:ne_history}, but with contours.
688: We observe that there is a ``precursor'' region in which the number density is
689: within a range of $1.1 < n_e/n_{e0} < 1.9$
690: in front of the transition region.
691: In this region,
692: the increase in the number density from that in the upstream plasma $n_{e0}$
693: would be mainly due to the electrons flowing backward
694: (see also Fig.~\ref{fig:electron_x_ux}).
695: The width of the precursor region
696: is almost constant ($\Delta x \sim 30$) after $t \sim 100$.
697: %%%%%
698: \begin{figure}
699: \plotone{f10.eps}
700: \caption{
701: \label{fig:backward_electron_x_t}
702: Time development of the electron number density
703: as in Fig.~\ref{fig:ne_history}, but with contours.
704: The width of the precursor region ($1.1 < n_e/n_{e0} < 1.9$),
705: where the density of the electrons flowing backward is high,
706: is almost constant ($\Delta x \sim 30$) after $t \sim 100$.
707: [\textit{See the electronic edition of the Journal for a
708: color version of this figure}.]
709: }
710: \end{figure}
711: %%%%%
712:
713:
714:
715: As already shown in Fig.~\ref{fig:Bz_Jx}b,
716: there are current filaments
717: in the transition region.
718: However,
719: they are charge-separated only at the filament edges
720: except those at the downstream side
721: that are continuously connected to the thermalized downstream region
722: (see Fig.~\ref{fig:rho_Ex_Ey}a).
723: The strength of the magnetic and electric fields,
724: current density, and charge density
725: all attain their maximum values in this region,
726: and the upstream kinetic energy
727: is dissipated mainly by the magnetic field
728: and partly by the electric field.
729:
730:
731: A fraction of the downstream particles flowing backward with $u_x < u_\mathrm{sh,d} \sim -0.42$,
732: which is the 4-velocity corresponding to $V_\mathrm{sh,d}$,
733: can return upstream through the filaments
734: that carry the currents in the same direction as the particles.
735: On the other hand,
736: if the particles return through the filament carrying opposite currents,
737: the magnetic fields in the filaments
738: repel the particles from the filaments.
739: Figure~\ref{fig:electron_m2_x_y} shows the distribution of electrons
740: that are flowing upstream with $u_x < -2$.
741: By comparing with $J_x$ in Fig.~\ref{fig:Bz_Jx}b,
742: we observed that these electrons indeed flow
743: in the current filaments with $J_x > 0$.
744: In a similar manner,
745: the positrons flow backward in the current filaments,
746: but with $J_x < 0$ (not shown here).
747: However,
748: only a small fraction of the backward flowing
749: particles can move to the upstream region ($x < 350$)
750: because large potential barriers exist,
751: caused by the charge density
752: at the upstream edge of the filaments
753: (see also Figs.~\ref{fig:rho_Ex_Ey}a and \ref{fig:rho_Ex_Ey}c).
754: The particles that could pass through to the upstream region
755: would contribute to stabilize the charge-separated current filaments
756: in the charge separation layer, as mentioned earlier.
757: %%%%%
758: \begin{figure}
759: \plotone{f11.eps}
760: \caption{
761: \label{fig:electron_m2_x_y}
762: Distribution of backward flowing electrons with $u_x < -2$.
763: They move upstream through the current filaments with $J_x > 0$.
764: This is evident for $x > 350$.
765: }
766: \end{figure}
767: %%%%%
768:
769:
770:
771:
772: %%%%%
773: \subsection{Particle Energy Distribution and Acceleration}
774: %%%%%
775:
776: Figure~\ref{fig:hist_KE_downstream}
777: shows the energy histogram of the electrons
778: accumulated over the area between $x=430$ and $x=442$
779: in the downstream region (the solid histogram).
780: The dashed curve and the dot-dashed curve
781: represent the Maxwell distributions
782: in three and two dimensions, which are defined by
783: %
784: \begin{equation}
785: N_{3D}(\gamma) = n_d \frac{\theta^2 \gamma (\gamma^2-1)^{1/2}}{2 K_1(\theta)
786: + \theta K_0(\theta)} \exp(-\gamma \theta),
787: \end{equation}
788: %
789: and
790: %
791: \begin{equation}
792: N_{2D}(\gamma) = n_d \frac{\theta^2 \gamma}{\theta + 1}
793: \exp\left[ -(\gamma-1) \theta\right],
794: \end{equation}
795: %
796: respectively.
797: Here, $n_d$ is the number density in the downstream region,
798: $K_i$ is the modified Bessel function of the second kind \citep{Abramowitz},
799: and $\theta$ is determined
800: so that $\int_1^\infty \gamma N(\gamma) d\gamma = \langle \gamma \rangle \sim \sqrt{5}$,
801: which is expected if the upstream bulk kinetic energy is completely converted
802: into the internal kinetic energy in the downstream rest frame.
803: From Fig.~\ref{fig:profile}a, we consider $n_d \sim 3.3\ n_{e0}$.
804: It is observed that the electron distribution
805: fits neither of the Maxwell distributions.
806: %%%%%
807: \begin{figure}
808: \plotone{f12.eps}
809: \caption{
810: \label{fig:hist_KE_downstream}
811: Histogram of the kinetic energy of the downstream electrons
812: within $430 < x < 442$ at $t=233$ (solid line).
813: The horizontal axis is the kinetic energy normalized by
814: the electron rest mass energy, $\gamma - 1$.
815: The dashed curve and the dot-dashed curve
816: represent the Maxwell distribution
817: for three and two dimensions, respectively.
818: }
819: \end{figure}
820: %%%%%
821:
822:
823:
824: Figure~\ref{fig:hist_KE_all} shows
825: the energy histogram of all electrons
826: in the simulation box by a solid histogram.
827: For reference,
828: we plotted the three-dimensional Maxwell distribution
829: by the dotted curve.
830: Their normalizations are arbitrary,
831: and there is no significant particle acceleration.
832: %%%%%
833: \begin{figure}
834: \plotone{f13.eps}
835: \caption{\label{fig:hist_KE_all}
836: Histogram of the kinetic energy of all the electrons
837: in the simulation box (solid line).
838: The three-dimensional Maxwell distribution
839: is also plotted as a reference (dashed curve).
840: There is no significant particle acceleration.
841: }
842: \end{figure}
843: %%%%%
844:
845:
846:
847:
848:
849: %%%%%
850: \section{Conclusions}
851: %%%%%
852: The simulation results clearly show
853: that collisionless shocks can be driven
854: in unmagnetized electron-positron plasmas.
855: The structure of the collisionless shock
856: propagates at an almost constant speed.
857: The dissipation of the upstream bulk kinetic energy
858: is mainly due to the deflection of particles
859: by the strong magnetic field
860: generated around the shock transition region
861: by the Weibel-type instability.
862: It is remarkable that
863: this shock has no corresponding linear wave and therefore
864: it can be regarded as a kind of ``instability-driven'' shock wave.
865: We also observed the electric and magnetic field generation
866: even in the upstream of the shock transition region
867: due to the charge-separated current filaments
868: together with the electrostatic beam instability.
869: We found that a fraction of the downstream thermalized particles
870: return upstream through the shock transition region.
871: These particles would interact with the upstream incoming plasma
872: and cause
873: the generation of the charge-separated current filaments
874: as well as
875: the electrostatic beam instability
876: upstream of the transition region.
877: We did not observe any efficient particle acceleration processes in our simulation,
878: whereas we observed relatively rapid decay of the generated magnetic field
879: within the shock transition region.
880: On the other hand, in the real world, there are many evidences to suggest that
881: particles are accelerated at shocks, and there are several indications that large-scale,
882: sub-equipartition magnetic fields may be generated around these shocks.
883: The reasons why these processes were not observed in our simulation may be that
884: (1) the background magnetic field is essential for them,
885: (2) the spatial and temporal scales of the simulation are too small to deal with them,
886: or
887: (3) these processes are inefficient in the electron-positron shocks.
888:
889:
890:
891:
892:
893:
894: \acknowledgments
895: I would like to thank F. Takahara, M. Hattori, Y. Fujita, and S. Nagataki
896: for their helpful discussions.
897: This research was supported in part by
898: the Japan Science and Technology Agency.
899: Numerical computations were carried out on VPP5000 at the Astronomical Data Analysis Center, ADAC,
900: of the National Astronomical Observatory of Japan.
901:
902:
903:
904:
905:
906: %%%%%%%%%%%%
907: % References
908: %%%%%%%%%%%%
909: \begin{thebibliography}{}
910: %
911: \bibitem[Abramowitz \& Stegun (1965)]{Abramowitz}
912: Abramowitz,~M., \& Stegun,~I.A. 1965,
913: \textit{Handbook of Mathematical Functions}
914: (New York: Dover).
915: %
916: \bibitem[Bamba et al.~(2003)]{Bamba03}
917: Bamba,~A., Yamazaki,~R., Ueno,~M., \& Koyama,~K. 2003,
918: \apj, 589, 827
919: %
920: \bibitem[Bell (2004)]{Bell04}
921: Bell,~A.~R. 2004,
922: \mnras, 353, 550
923: %
924: \bibitem[Birdsall \& Langdon (1991)]{Birdsall}
925: Birdsall,~C.~K., \& Langdon,~A.~B. 1991,
926: Plasma Physics via Computer Simulation (IOP Publishing: Bristol).
927: %
928: \bibitem[Fried (1959)]{Fried59}
929: Fried,~B.~D. 1959,
930: Phys. Fluids, 2, 337
931: %
932: \bibitem[Gruzinov (2001)]{Gruzinov01}
933: Gruzinov,~A. 2001, preprint (astro-ph/0111321)
934: %
935: \bibitem[Haruki \& Sakai (2003)]{Haruki03}
936: Haruki,~T., \& Sakai,~J. 2003,
937: Phys. Plasmas, 10, 392
938: %
939: \bibitem[Kato (2005)]{Kato05}
940: Kato,~T.~N. 2005,
941: Phys. Plasmas, 12, 080705
942: %
943: \bibitem[Kazimura et al.~(1998)]{Kazimura98}
944: Kazimura,~Y., Sakai,~J.~I., Neubert,~T., \& Bulanov,~S.~V. 1998,
945: \apjl, 498, L183
946: %
947: \bibitem[Koyama et al.~(1995)]{Koyama95}
948: Koyama,~K., Petre,~R., Gotthelf,~E.~V.,
949: Hwang,~U., Matsuura,~M., Ozaki,~M., \& Holt,~S.~S. 1995,
950: \nat, 378, 255
951: %
952: \bibitem[Medvedev \& Loeb (1999)]{Medvedev99}
953: Medvedev,~M.~V., \& Loeb,~A. 1999,
954: \apj, 526, 697
955: %
956: \bibitem[Milosavljevi\'{c} et al.~(2006)]{Milosavljevic06}
957: Milosavljevi\'{c},~M., Nakar.~E., \& Spitkovsky.,~A. 2006,
958: \apj, 637, 765
959: %
960: \bibitem[Silva et al.~(2003)]{Silva03}
961: Silva,~L.~O., Fonseca,~R.~A., Tonge,~J.~W., Dawson,~J.~M.,
962: Mori,~W.~B., \& Medvedev,~M.~V. 2003
963: \apjl, 596, L121
964: %
965: \bibitem[Vink \& Laming~(2003)]{Vink03}
966: Vink,~J., \& Laming,~J.~M. 2003,
967: \apj, 584, 758
968: %
969: \bibitem[V\"{o}lk et al.~(2005)]{Voelk05}
970: V\"{o}lk,~H.~J., Berezhko,~E.~G., \& Ksenofontov,~L.~T. 2005,
971: \aap, 433, 229
972: %
973: \bibitem[Weibel (1959)]{Weibel59}
974: Weibel,~E.~S. 1959,
975: \prl, 2, 83
976: %
977: \end{thebibliography}
978:
979:
980:
981:
982:
983:
984: \end{document}
985:
986: