0808.1208/pm.tex
1: \documentclass[twocolumn,showpacs,amsmath,amssymb]{revtex4}
2: %\documentclass[preprint,showpacs,preprintnumbers,amsmath,amssymb]{revtex4}
3: % Some other (several out of many) possibilities
4: %\documentclass[preprint,aps]{revtex4}
5: %\documentclass[preprint,aps,draft]{revtex4}
6: %\documentclass[prb]{revtex4}% Physical Review B
7: 
8: \usepackage{graphicx}% Include figure files
9: \usepackage{dcolumn}% Align table columns on decimal point
10: \usepackage{bm}% bold math
11: 
12: \begin{document}
13: \title{Physics of collisionless phase mixing}
14: \author{D. Tsiklauri}
15: \author{T. Haruki}
16: 
17: \affiliation{Joule Physics Laboratory,
18: Institute for Materials Research, University of Salford, Manchester, M5 4WT,
19: United Kingdom}
20: \date{\today}
21: 
22: \begin{abstract}
23: Previous studies of phase mixing of ion cyclotron (IC), Alfv\'enic, waves in the collisionless
24: regime have established the generation of parallel electric field and hence
25: acceleration of electrons in the regions of transverse density inhomogeneity.
26: However, outstanding issues were left open. 
27: Here we use 2.5D, relativistic, fully electromagnetic PIC (Particle-In-Cell) 
28: code and an analytic MHD (Magnetohydrodynamic) formulation, to establish the following points:
29: (i) Using the generalised Ohm's law we find that the parallel electric field is supported mostly by
30: the electron pressure tensor, with a smaller contribution from the electron inertia term.
31: (ii) The generated parallel electric field and the fraction of accelerated electrons are 
32: independent of the IC wave
33:  frequency remaining at a level of six orders of magnitude larger 
34:  than the Dreicer value and approximately
35: 20 \% respectively. The generated parallel electric field and the fraction of accelerated electrons 
36: increase with the increase of IC wave  amplitude. The generated parallel electric field seems to be
37: independent of  plasma beta, while the fraction of accelerated electrons strongly increases 
38: with the decrease of plasma beta (for plasma beta of 0.0001 the fraction of accelerated electrons
39: can be as large as 47 \%).
40: (iii) In the collisionless regime 
41: IC wave dissipation length (that is defined as the distance over which the wave damps) 
42: variation with the driving frequency shows a deviation from the
43: analytical MHD result, which we attribute to 
44: a possible frequency  dependence  of the effective resistivity.
45: (iv) Effective anomalous resistivity, inferred from our numerical simulations, is at least four orders of
46: magnitude larger than the classical Spitzer value.
47: \end{abstract}	
48: 
49: \pacs{52.35.Hr; 96.50.Tf; 04.30.Nk; 96.50.Ci; 96.60.P-;52.65.Rr; 96.60.pf; 96.60.qe}
50: 
51: \maketitle
52: 
53: %===============================================================================
54: % Introduction
55: %===============================================================================
56: \section{Introduction}
57: 
58: Phase mixing is a mechanism of enhanced dissipation of 
59: Alfv\'en waves due to inhomogeneity of Alfv\'en speed in a
60: direction transverse to a local magnetic field.
61: This mechanism originally was studied in the
62: fusion and laboratory plasma context by a number of 
63: authors \cite{1972PhFl...15.1673U,1973ZPhy..261..203T,1973ZPhy..261..217G,1974PhRvL..32..454H,
64: 1974PhFl...17.1399C,1975JPlPh..13...87T} and subsequently applied to
65: the solar corona \cite{1983A&A...117..220H}. Most of the 
66: large amount of work done in the field of phase mixing 
67: was in the resistive MHD (Magnetohydrodynamic) regime. Recently, a few works looked at the
68: same mechanism in the collisionless regime in the context of Earth 
69: magnetosphere \cite{1999JGR...10422649G, 2004AnGeo..22.2081G} and solar corona
70: \cite{2005A&A...435.1105T,2005NJPh....7...79T}. The main findings of these
71: works include the generation of electric field that is parallel to the ambient 
72: magnetic field in the regions of transverse density inhomogeneity,
73: as well as associated electron acceleration. It should be mentioned that
74: these studies considered circularly polarised ion cyclotron (IC) waves
75: which in the low frequency regime become Alfv\'en waves. 
76: We use terms Alfv\'en or IC interchangeably, but reader should bear in mind
77: we always refer to {\it waves with frequencies $< \omega_{ci}$} (with $\omega_{ci}$ being ion cyclotron frequency).
78: The exact mechanism of generation of the parallel electric field has stimulated
79: a debate \cite{2006A&A...449..449M,2007NJPh....9..262T}, and even MHD regime option was explored
80: \cite{2006A&A...455.1073T}.
81: Continuing this investigation, here we apply technique used in the
82: collisionless reconnection \cite{pritchett01,th08}.
83: Namely, in section 3.1 we use generalised Ohm's law to find out
84: which term generates the parallel electric field.
85: 
86: Solar flare observations \cite{2005SSRv..121..141F} trigger one's 
87: interest in how effectively plasma particles are accelerated.
88: Hence, in section 3.2 we look into how 
89: the generated parallel electric field and the fraction of accelerated electrons
90: depend on model parameters such as IC wave frequency, amplitude, and plasma beta.
91: 
92: Ref. \cite{1983A&A...117..220H} provides a simple analytical expression
93: how Alfv\'en wave amplitude should decay in space due to the phase mixing.
94: Despite the fact that their formula is derived in the resistive MHD
95: regime, we still apply it to our collisionless, kinetic simulation
96: and see what does the comparison yield (Section 3.3). 
97: This is done in the light of previous results of \citet{2005A&A...435.1105T}
98: who established that in the collisionless, kinetic regime
99: Alfv\'en wave amplitude in the density gradient regions decays with
100: distance (from where it is driven) according to collisional MHD formula of 
101: \citet{1983A&A...117..220H}. Here we stretch the 
102: MHD-kinetic analogy further to test $\omega_d^2$ dependence under the exponent.
103: 
104: In Sect 3.4 we estimate the effective "resistivity" (again the spirit of MHD-kinetic analogy).
105: The quotation marks are needed to signify that PIC (Particle-In-Cell)
106: simulation code is collisionless and hence no resistive effects
107: exist as such. However, scattering of particles by magnetic
108: fields plays effective role of collisions.
109: 
110: 
111: 
112: %===============================================================================
113: % Simulation Model
114: %===============================================================================
115: \section{Simulation Model}
116: 
117: In our numerical simulations we use 2.5D, relativistic, fully electromagnetic PIC code. 
118: The initial conditions, basic parameters and boundary conditions are exactly the same as 
119: in the previous work by \citet{2005A&A...435.1105T}.
120: In particular, 
121: the uniform magnetic field is in $x$-direction, the transverse density inhomogeneity is across $y$-direction.
122: $z$ is the spatially ignorable coordinate, i.e. $\partial / \partial z = 0$. However, we retain
123: all three components of velocity, $v_x$, $v_y$ and $v_z$. 
124: The system size without ghost cells 
125: in two dimensions is $L_x = 5000 \Delta$ and $L_y = 200 \Delta$ where $\Delta (= 1)$ is 
126: the simulation grid size, corresponding 
127: to the electron Debye length, $\lambda_D = v_{te} / \omega_{pe} = 1 \Delta$ ($v_{te}$ is 
128: electron thermal velocity and $\omega_{pe}$ is electron plasma frequency).
129: The total particle number is $2.39 \times 10^8$ electron-ion pairs.
130: The ion to electron mass ratio is $m_i / m_e = 16$ due to the limitation of 
131: computer resources and speed.
132: \citet{2005A&A...435.1105T} use the fixed driving wave 
133: amplitude  $\delta B / B_0 = 0.05$, and plasma $\beta = 0.02$.
134:  The dimensionless 
135: ion and electron density inhomogeneity is described by
136: $$
137:  {n_i(y)}=
138: {n_e(y)}=1+3 \exp\left[-\left(\frac{y-100\Delta}{50 \Delta}\right)^6\right]
139: \equiv F(y).
140: $$
141: These are normalised to some background constant value ($n_0$).
142: Here all plasma parameters are quoted at the boundary, 
143: away from the density inhomogeneity region.
144: In the central region (across $y$-coordinate), the density is
145: smoothly enhanced by a factor of 4, and there are the 
146: strongest density gradients having 
147: a width of about ${51 \Delta}$ around the 
148: locations $y=51.5 \Delta$ and $y=148.5 \Delta$.
149: Below, in Eqs.(4) and (5) we shall be using $y=51.5-51/2=26\Delta$
150: and $y=51.5+51/2=77 \Delta$ as the boundaries of one of the
151: transverse density gradients.
152: The background temperature of ions and electrons, 
153: and their thermal velocities
154: are varied accordingly
155: $$
156: {T_i(y)}/{T_0}=
157: {T_e(y)}/{T_0}=F(y)^{-1},
158: $$
159: $$
160:  {v_{th,i}}/{v_{i0}}=
161: {v_{th,e}}/{v_{e0}}=F(y)^{-1/2},
162: $$
163: such that the thermal pressure remains constant. Since the background magnetic field
164: along the $x$-coordinate  is also constant, the total pressure remains constant too.
165: Then we impose a current of the following form
166: $$
167: {\partial_t E_y}=-J_0\sin(\omega_d t)\left(1-\exp\left[-(t/t_0)^2\right]\right),
168: $$
169: $$
170: {\partial_t E_z}=-J_0\cos(\omega_d t)\left(1-\exp\left[-(t/t_0)^2\right]\right).
171: $$
172: In Ref. \cite{2005A&A...435.1105T} the driving frequency was fixed at $\omega_d=0.3\omega_{ci}$.
173: Here,  we also use driving frequencies that satisfy 
174: $\omega_d < \omega_{ci}$ so that no significant ion-cyclotron resonant 
175: damping takes place. $\partial_t$ denotes the time derivative.
176: $t_0$ is the onset time of the driver, which was fixed at $50 /\omega_{pe}$
177: i.e.  $3.125 / \omega_{ci}$. This means that the driver onset time is about 3 ion-cyclotron
178: periods. Imposing such a current on the system results in the generation of
179: left circularly polarised IC (Alfv\'enic) wave, which is driven at the left 
180: boundary of simulation box and has spatial driver width of $1 \Delta$.
181: The wave propagates along $x$-coordinate and generates the parallel
182: electric field in the density gradient regions (for more details see \citet{2005A&A...435.1105T}).
183: The parameters used are commensurate to what is seen in solar
184: corona by e.g. Doppler broadening of emission lines.
185: The observed values of the Alfv\'en waves at heights of 
186: $R = 1.04 R_{\rm sun}= 28$ Mm are about 
187: 50 km s$^{-1}$ (see e.g. \cite{moran01}), 
188: which for a typical Alfv\'en speed of 
189: 1000 km s$^{-1}$ makes $\delta B / B_0$ equal to 0.05.
190: 
191: As one of our goals is to investigate the parameter space of the
192: problem, the following range of physical parameters was used:
193:  we vary the frequency, amplitude of IC wave or the plasma beta.
194: The driving wave frequency $\omega_d / \omega_{ci}$ ranged from $0.1$ to $0.5$ with a step of $0.1$.
195: The wave amplitude $\delta B / B_0$ was also set at $0.01, 0.05, 0.10, 0.15, 0.20$ and $0.25$.
196: The plasma beta varies
197: from $\beta = 10^{-4}$ to $10^{-2}$ by controlling the electron thermal velocity.
198: Specifically, the plasma beta was set 
199: at $0.0001, 0.0003, 0.0010, 0.0030, 0.0100, 0.0200$ and $0.0300$.
200: Each beta value corresponds 
201: to $v_{te} / c = 0.007, 0.012, 0.022, 0.039, 0.071, 0.100$ and $0.122$, respectively, 
202: where $v_{te}$ is taken from the low density region and $c$ is speed of light.
203: Each numerical run (each data point
204: in subsequent Figs. (2)-(5) typically takes about 7-10
205: days on 64 parallel processors.
206: %===============================================================================
207: % Simulation Results
208: %===============================================================================
209: \section{Simulation Results}
210: 
211: \subsection{Source of the parallel electric field}
212: 
213: % spatial scale
214: % Lx = 5000 (= L)
215: % Ly = 200
216: % maximum density gradient region Lg = 51.5
217: %
218: % Lx /di = 5000 / 40 = 125
219: % Lx / L = 1.0
220: % Lg  / di = 51.5 / 40 = 1.2875
221: % Lg / L = 51.5 / 5000 = 0.00103
222: 
223: % time scale
224: % ts = 17500
225: % tA = L / va = 5000 / (0.25 c) = 40000
226: %
227: % wci ts = 0.003125 * 17500 = 54.6875
228: % ts / tA = 17500 / 40000 = 0.4375
229: 
230: \begin{figure}
231: \includegraphics[scale = 0.5]{fig1.eps}
232:   \caption
233:   {
234: Different term profiles along $x$-coordinate from the generalised Ohm's law:
235:     (a) $E_x$ (normalised to $v_{A0} B_0$),
236:     (b) $- (d_i / n) \partial P_{exx} / \partial x$, 
237:     (c) $- (d_i / n) \partial P_{eyx} / \partial y$, and 
238:     (d) $- d_i (m_e / m_i) \partial v_{ex} / \partial t$ 
239:     in the maximum density gradient region $y = 0.0103$ at $t = 0.4375$
240:     for the case, $\omega_d / \omega_{ci} = 0.3$, $\delta B / B_0 = 0.25$ and $\beta = 0.02$.
241:     The time derivative term in (d) is calculated from $t = 0.4370$ to $0.4375$
242:     (time interval corresponds to the inverse of electron plasma frequency $\omega^{-1}_{pe}$).
243:      Refer to text for the normalisation used.
244:   }
245:   \label{fig1}
246: \end{figure}
247: 
248: In order to understand details of the parallel electric field generation process, 
249: we now focus on the question: which term in the generalised Ohm's law is responsible for 
250: the generation of the parallel electric field?
251: The generalised Ohm's law can be written as
252: \begin{equation}
253:   \vec{E} = - \vec{v}_e \times \vec{B} 
254:             - \frac{\nabla \cdot \vec{P}_e}{n_e e}
255:             - \frac{m_e}{e} (\frac{\partial \vec{v}_e}{\partial t} 
256:             + \left( \vec{v}_e \cdot \nabla) \vec{v}_e \right), 
257: \label{g_ohms}
258: \end{equation}
259: where $\vec{E}$ and $\vec{B}$ are electric and magnetic fields, 
260: $\vec{v}$ is plasma velocity, $\vec{P}$ is pressure tensor ($3 \times 3$ matrix), 
261: $n$ is plasma number density, $m$ is  mass and $e$ is electric charge.
262: The subscript $e$ refers to an electron.
263: Normalising space coordinate by $L_x = 5000 \Delta$, 
264: fluid velocity by Alfv\'en speed (at the lowest density region)
265: $v_{A0}$, time by Alfv\'en transit time $\tau_A (= L / v_{A0})$, 
266: magnetic field by $B_0$, electric field by $v_{A0} B_0$,
267: number density by $n_0$ and pressure tensor by $B_0^2 / \mu_0$, 
268: a dimensionless version of Eq.~(\ref{g_ohms}) can be obtained
269: \begin{eqnarray}
270:   \vec{E} = - \vec{v}_e \times \vec{B} 
271:             - d_i \frac{\nabla \cdot \vec{P}_e}{n_e}
272:             - d_i \frac{m_e}{m_i} (\frac{\partial \vec{v}_e}{\partial t} 
273:             + \left( \vec{v}_e \cdot \nabla) \vec{v}_e \right), 
274: \label{nd_g_ohms}
275: \end{eqnarray}
276: where $d_i$ is the normalised ion skin depth ($d_i = c / \omega_{pi} L$).
277: Note that strictly speaking we should have used tildes in Eq.~(\ref{nd_g_ohms}) to denote 
278: dimensionless quantities, but we omit them for brevity.
279: 
280: Let us focus on the parallel electric field, $E_x$, 
281: which is generated  in the density gradient regions 
282: by phase mixing. It is given by, 
283: \begin{eqnarray}
284: \nonumber
285:   E_x = - (v_{ey} B_z - v_{ez} B_y) 
286:           - d_i \frac{1}{n} \left(
287:             \frac{\partial P_{exx}}{\partial x}
288:          + \frac{\partial P_{eyx}}{\partial y} \right) \\
289:           - d_i \frac{m_e}{m_i} \left(
290:               \frac{\partial v_{ex}}{\partial t}
291:             + v_{ex} \frac{\partial v_{ex}}{\partial x}
292:             + v_{ey} \frac{\partial v_{ex}}{\partial y} \right), 
293: \label{g_ohms_ex}
294: \end{eqnarray}
295: where $\partial / \partial z = 0$ is assumed because of spatially 2 dimensional model.
296: 
297: Fig.~\ref{fig1} shows the different term profiles (along the uniform
298: magnetic field in $x$-direction) in the generalised Ohm's law (Eq.~(\ref{g_ohms_ex})): 
299: (a) $E_x$,
300: (b) $- (d_i / n) \partial P_{exx} / \partial x$, 
301: (c) $- (d_i / n) \partial P_{eyx} / \partial y$, and 
302: (d) $- d_i (m_e / m_i) \partial v_{ex} / \partial t$ 
303: in the maximum density gradient region $y = 0.0103$ at $t = 0.4375$
304: for the case, $\omega_d / \omega_{ci} = 0.3$, $\delta B / B_0 = 0.25$ and $\beta = 0.02$.
305: The time derivative term in Fig.~\ref{fig1}(d) is calculated from $t = 0.4370$ to $0.4375$ 
306: (time interval corresponds to the inverse of electron plasma frequency $\omega^{-1}_{pe}$).
307: The other terms in right-hand side of Eq.~(\ref{g_ohms_ex}) are negligible.
308: In Fig.~\ref{fig1}(a), the parallel electric field is clearly observed 
309: in the density gradient regions where phase mixing can occur.
310: It should be noted that no parallel electric field is seen
311: away from the density gradient regions.
312: According to the generalised Ohm's law,
313: there has to be a term on the 
314: right-hand side of Eq.~(\ref{g_ohms_ex}) that supports this electric field.
315: 
316: By comparing Fig.~\ref{fig1}(a) to  Fig.~\ref{fig1}(b-d) it is clear that 
317: the parallel electric field (Fig.~\ref{fig1}(a)) is supported mostly by
318: the electron pressure tensor (Fig.~\ref{fig1}(b)), with a smaller contribution 
319: from the electron inertia term (Fig.~\ref{fig1}(d)). It is interesting to note that
320: previous results on collisionless reconnection both in tearing unstable 
321: Harris current sheet \cite{hesse99,birn01,pritchett01} and stressed X-point collapse 
322: \cite{th07,th08} have shown that the term in the generalised Ohm's law 
323: that is responsible for breaking the frozen-in condition, i.e. that
324: supports out-of-plane electric field is electron pressure tensor.
325: Here the electron pressure tensor supports (generates) the parallel
326: electric field. Thus, one should note a universal importance of
327: the electron pressure tensor in relation to supporting 
328: the electric fields in collisionless plasmas.
329: 
330: \subsection{Parametric study of the generated parallel electric 
331: field and the fraction of accelerated electrons}
332: 
333: To estimate how efficiently 
334: the parallel electric field is generated by phase mixing, 
335: as a function of model parameters, we introduce 
336: the average of the absolute value of the parallel electric field in the density gradient region:
337: \begin{equation}
338:   \frac{E^*}{E_0} = \frac{1}{L_x L_y} \int_{x = 0}^{L_x} 
339:   \int_{y = 26\Delta}^{77 \Delta} \frac{|E_x(x, y)|}{E_0} dx dy,
340: \label{e_star}
341: \end{equation}
342: where $E_0 = m_e c \omega_{pe}/ e$.
343: Note that in what follows the normalisation of the electric field is 
344: different from Section 3.1 where, usual for the
345: generalised Ohm's law, "fluid" normalisation ($v_{A0} B_0$) is used.
346: Normalisation $E_0 = m_e c \omega_{pe}/ e$ is usually referred to as "electrostatic".
347: By using the definition given by Eq.(\ref{e_star}) 
348: we can evaluate quantitatively the electric field generated 
349: in the density gradient region.
350: Although there are two density regions in our simulation box because of 
351: the periodic boundary condition, we focus on the lower one (the physics of the upper and 
352: the lower regions is essentially the same).
353: The range from $y = 26 \Delta$ to $77 \Delta$ indicates the 
354: density gradient. See \citet{2005A&A...435.1105T} for details.
355: 
356: Also, in order to investigate the fraction of accelerated electrons by the generated 
357: parallel electric field, we use particle data in the density gradient at lower side 
358: ($26 \Delta \le y \le 77 \Delta$). We count 
359: the number of electrons with velocities greater than 
360: the thermal velocity ($v_{te}<v_x < c $) in the electron velocity 
361: distribution function, in the $x$-direction, 
362: at the final time snapshot $\omega_{ci} t = 54.69$,
363: and divide this by the total number of particles (with $0<v_x<c$)
364: in the same domain:
365: \begin{equation}
366:   \frac{N}{N_0} = \frac
367:     {
368:      \int_{v_x = v_{te}}^{c} \int_{x = 0}^{L_x} \int_{y = 26\Delta}^{77 \Delta} 
369:      f(v_x) dv_x dx dy
370:     }
371:     {
372:      \int_{v_x = 0}     ^{c} \int_{x = 0}^{L_x} \int_{y = 26\Delta}^{77 \Delta} 
373:      f(v_x) dv_x dx dy
374:     }.
375: \label{n_acc}
376: \end{equation}
377: Here it was to enough to integrate only positive region in this distribution
378: because electron acceleration was symmetrical in the $x$-direction.
379: Note that the initial velocity distribution function is nearly Maxwellian.
380: To maintain the balance of the total kinetic pressure throughout the system, 
381: the particle thermal velocity in the dense plasma region ($y = 100 \Delta$) is 
382: lower than the outside region ($y = 0 \Delta$ or $200 \Delta$) (see for details
383: Fig.(4) from \citet{2005A&A...435.1105T}).
384: However, in order to use Eq.~(\ref{n_acc}),
385: we had to estimate an appropriate velocity corresponding to
386: the thermal velocity in the Maxwellian.
387: Fortunately, initial electron velocity distribution function did not deviate much from 
388: the exact Maxwellian.
389: Therefore, we adopted the standard thermal velocity which was set to 
390: $36.8$\% of $f(v=0)$.
391: Recall that $f(v=v_{te}) = n_0 \exp(-v^2/v_{te}^2) = n_0 \exp(-1.0) = 0.368 n_0$, 
392: where $n_0$ is the peak number at $v = 0$ in Maxwellian.
393: 
394: \begin{figure}
395: \includegraphics[scale = 0.5]{fig2.eps}
396:   \caption{
397:            (a) $E^* / E_0$ vs. $\omega_d / \omega_{ci}$   and
398:            (b) $N / N_0$ vs. $\omega_d / \omega_{ci}$ 
399:            for $\delta B / B_0 = 0.05$ and $\beta = 0.02$.
400:           }
401:   \label{fig2}
402: \end{figure}
403: 
404: The choice of diagnostic for characterizing the degree of particle 
405: acceleration needs a further clarification.
406: Eq.(5) provides number of electrons with speed $v_x$ 
407: exceeding the thermal speed $v_{te}$ as a fraction of total distribution value.
408: However, in a Maxwellian plasma 16\% of the electrons have $v > v_{te}$ (for 
409: $v_{te} = \sqrt{2 k_B T_e/m_e}$). Thus, it may be tempting 
410: to either: (i) subtract this 16\% from our diagnostic in Eq.(5) or 
411: (ii) try to fit a Maxwellian to the electron distribution at the
412: final simulation time step (after all the acceleration has taken place)
413: and then count the number of electrons that have speeds $v_x > v_{te}$.
414: Our motivation to keep the definition given by Eq.(5) is two-fold:
415: (i) In context of the solar flare observations in X-rays, one always infers 
416: the integrated spectrum
417: averaged over some volume $V$, i.e. $<f(v)nV>$ (where $n$ is electron number density)  
418: and it is impossible to subtract the above mentioned 16\% without introducing 
419: additional uncertainties due to the unknown density and 
420: poorly known emitting volume, due to a line of sight effect (Dr. E. Kontar of 
421: University of Glasgow, private communication), see also Refs. \cite{bek03,bk05}.
422: Hence the definition given by Eq.(5) is more appropriate for comparison of
423: theory with the observations. Also, this is particularly timely
424: because the acceleration of electrons by Alfv\'enic waves in flares have been
425: recently studied \cite{2008ApJ...675.1645F}.
426: (ii) Although the electron distribution function at the initial time step is
427: nearly Maxwellian (despite density and temperature transverse inhomogeneities),
428: at the final stages of the simulation  the deviations from the 
429: Maxwellian form are large and hence a fit to a Maxwellian, in order to calculate
430: $v_{te}$, and in turn to count the super-thermal particles above that value ($v_x > v_{te}$),
431: is impractical.
432: 
433: Fig.~(\ref{fig2}) shows the generated parallel electric 
434: field, $E^*/E_0$, and the fraction of accelerated electrons,
435: $N/N_0$ as function of driving frequency of the IC wave (normalised to $\omega_{ci}$).
436: The wave frequency $\omega_d / \omega_{ci}$ was set at $0.1, 0.2, 0.3, 0.4$ 
437: and $0.5$. 
438: The wave amplitude and plasma beta were fixed at $\delta B / B_0 = 0.05$ and 
439: $\beta = 0.02$.
440: We gather from Fig.~(\ref{fig2}) that the generated parallel 
441: electric field and the fraction of accelerated electrons are 
442: independent of the IC wave
443: frequency remaining at a level of six orders of magnitude larger 
444: than the Dreicer value and approximately
445: 20 \% respectively. This conclusion is based on the
446: following estimate  $E^*/E_0 \approx 0.0018$, i.e. $E^*=5473.4337$ V m$^{-1}$
447: (note that here solar coronal number density of 
448:  $n=10^{15}$ m$^{-3}$ was used). The Dreicer electric field (which
449:  is associated with the particle acceleration run-away regime \cite{dreicer}), 
450: $ E_d=(n e^3 \ln \Lambda)/(4 \pi \epsilon_0^2 k_B T)$, for $T=1$ MK (and hence 
451:  $\ln \Lambda=18.095$) is  $E_d=0.00547$ V m$^{-1}$. Thus, $E^*/E_d=10^6$.
452:  This result should be taken with caution because it is obtained
453:  with the ion-electron mass ratio of 16. As shown by \citet{2007NJPh....9..262T} (see their
454:  Figure 7) that attained amplitude of the generated parallel electric
455:  field scales strongly as  $\propto 1/(m_i/m_e)$.
456: 
457: 
458: \begin{figure}
459: \includegraphics[scale = 0.5]{fig3.eps}
460:   \caption{
461:            (a) $E^* / E_0$  vs. $\delta B / B_0$  and
462:            (b) $N / N_0$ vs.    $\delta B / B_0$  
463:            for $\omega_d / \omega_{ci} = 0.3$ and $\beta = 0.02$.
464:           }
465:   \label{fig3}
466: \end{figure}
467: 
468: Next, we explore how the generated parallel electric 
469: field and the fraction of accelerated electrons
470: depend on driving IC wave amplitude.
471: The latter, $\delta B / B_0$, was set at $0.01, 0.05, 0.10, 0.15, 0.20$ 
472: and $0.25$.
473: The wave frequency and plasma beta were fixed at $\omega_d / \omega_{ci} = 0.3$ 
474: and $\beta = 0.02$.
475: We gather from Fig.~\ref{fig3} that the generated parallel electric 
476: field and the fraction of accelerated electrons
477:  increase with the increase of IC wave  amplitude.
478:  This seems as a reasonable result because larger amplitude waves
479:  have more energy to give to electrons. Also, 
480:  non-linear effects would be progressively important.
481: 
482: 
483: \begin{figure}
484: \includegraphics[scale = 0.5]{fig4.eps}
485:   \caption{
486:            (a) $E^* / E_0$ vs. $\beta$   and
487:            (b) $N / N_0$ vs.  $\beta$  
488:                for $\omega_d / \omega_{ci} = 0.3$ and $\delta B / B_0 = 0.05$.
489:           }
490:   \label{fig4}
491: \end{figure}
492: 
493: We also explored the plasma beta dependence.
494: The plasma beta was set at $0.0001, 0.0003, 0.0010, 0.0030, 0.0100, 0.0200$ 
495: and $0.0300$.
496: The wave frequency and amplitude were fixed at $\omega_d / \omega_{ci} = 0.3$ 
497: and $\delta B / B_0 = 0.05$.
498: The plasma beta is defined as $\beta = 2 \mu_0 p / B^2 
499: = 2 (v_{te} / c)^2 / (\omega_{ce} / \omega_{pe})^2$.
500: We altered the plasma beta by changing the electron thermal velocity, 
501: affecting the plasma kinetic pressure.
502: Therefore, in this simulation, magnetic pressure is kept constant 
503: while plasma kinetic pressure varies.
504: Fig.~\ref{fig4}(a) does not show any correlation between plasma beta and 
505: the parallel electric field generation.
506: Incidentally, \citet{2002A&A...395..285T} investigated plasma beta dependence 
507: of the fast magnetosonic wave amplitude, which is
508: generated in a transversely inhomogeneous medium  when
509: an Alfv\'enic pulse is launched (using MHD numerical simulation).
510: According to Fig.~9(b) in  \citet{2002A&A...395..285T}, 
511: the maximum fast magnetosonic wave amplitude also does not depend on plasma beta.
512: \citet{2006A&A...455.1073T} alluded to the relation between
513: the non-linear fast magnetosonic wave and parallel electric field generation.
514: Hence it is not surprising that  in our Fig.~\ref{fig4}(a) we do not
515: see plasma beta dependence. What is surprising, at the first glance,
516: in Fig.~\ref{fig4}(b), is that the fraction of accelerated particles strongly
517: depends on beta. In particular, a decrease in beta (for $\beta = 0.0001$) yields as
518: large percentage as  $N / N_0 = 0.472 \sim 47$\%!
519: One can conjecture that this is due to the fact the in the
520: case of small plasma beta, magnetic effects dominate
521: over thermal ones, and because IC wave is essentially a magnetic-type
522: perturbation, electrons respond better to the wave influence and
523: accelerate more efficiently.
524: 
525: \subsection{Amplitude decay law in the kinetic regime}
526: 
527: \begin{figure}
528: \includegraphics[scale = 0.5]{fig5.eps}
529:   \caption{
530:            A log-log plot of the
531: 	   dissipation length $L_d / (c/\omega_{pi})$
532: 	    vs. driving IC wave frequency $\omega_d / \omega_{ci}$ (open symbols). 
533:            The solid line is the least squares fit with a slope of $-1.10$.
534:                      }
535:   \label{fig5}
536: \end{figure}
537: 
538: 
539: \citet{2005A&A...435.1105T} established that in the collisionless, kinetic regime
540: Alfv\'en (IC) wave amplitude in density gradient region decays with
541: distance (from where it is driven) according to collisional MHD formula of 
542: \citet{1983A&A...117..220H}
543: \begin{equation}
544:   B_z \sim \exp \left[- 
545:   \frac{\eta \omega_d^2 (\partial_y v_A)^2} {6 v_A^5} x^3 \right],
546:   \label{mhd_decay_law}
547: \end{equation}
548: where $\eta$ is resistivity (divided by $\mu_0$, i.e. by $\eta$
549: we mean $1/(\sigma \mu_0)$, $\omega_d$ is (driving) wave frequency, $v_A$ is 
550: Alfv\'en velocity and $x$ is the axis which AW propagates.
551: Recall that plasma density inhomogeneity is a function of $y$ and 
552: wave propagates in the $x$ direction.
553: In particular, (see for details left panel of Fig.6 in \citet{2005A&A...435.1105T})
554: it was shown that, at a fixed time instance corresponding to well-developed phase mixing,
555: $B_z(x) \propto \exp\left(-(x/L_d)^3\right)$. In other words, one can define
556: an empirical dissipation length, $L_d$, according to collisional MHD formula 
557: (Eq.(\ref{mhd_decay_law})) in the
558: collisionless, kinetic regime. Applicability of the MHD formula
559: in the kinetic regime was rather surprising.
560: We now turn back to the previous results in order to see how far this
561: MHD-kinetic analogy can be stretched. In particular,
562: one can e.g. check the $\omega_d^2$ scaling in the Eq.(\ref{mhd_decay_law}).
563: For this purpose, after simple algebra applied to $B_z \propto \exp[-(x/L_d)^3]$, 
564: for fixed $\eta, \partial_y v_A$ and $v_A$,
565: one can obtain the following scaling $\log_{10} L_d = -2/3 \log_{10} \omega_d + C$,
566: where $C$ is some constant.
567: Hence the slope of a log-log graph of
568: dissipation length $L_d$ versus $\omega_d$ is expected to be $-2/3 = -0.67$.
569: Fig.~\ref{fig5} presents the dissipation length (distance over which the wave damps)
570: dependence on the IC wave frequency.
571: The values in Fig.~\ref{fig5} were obtained from line data of magnetic field 
572: $B_z$ in the maximum plasma gradient $Y / (c / \omega_{pe}) = 14.8$ at 
573: $\omega_{ci} t = 82.0$ for $\omega_d / \omega_{ci} = 0.2$, 
574: $\omega_{ci} t = 54.7$ for $\omega_d / \omega_{ci} = 0.3$, 
575: $\omega_{ci} t = 41.0$ for $\omega_d / \omega_{ci} = 0.4$ and 
576: $\omega_{ci} t = 32.8$ for $\omega_d / \omega_{ci} = 0.5$, respectively.
577: The reason for the different snapshot times
578: is that we only consider well-developed phase mixing, i.e. when IC wave
579: is fully damped. As in left panel of Fig.6 in \citet{2005A&A...435.1105T})
580: we fit $B_z(x)$ to $\propto \exp\left(-(x/L_d)^3\right)$ and obtain empirical dissipation length $L_d$.
581:  we gather from Fig.~\ref{fig5} that the slope is $-1.10$ contrary to the above
582:  prediction of $-2/3 = -0.67$.
583:  
584:  To address the inconsistency we conjecture that the resistivity might be variable.
585: One can estimate the resistivity for each case of driving wave frequency considered,
586: by calculating $\eta = 6 v_A^5 / (\omega_d^2 (\partial_y v_A)^2 L_d^3)$.
587: We reiterate that strictly speaking PIC
588: simulation code is collisionless and hence no resistive effects
589: exist. However, scattering of particles by magnetic
590: fields plays effective role of collisions. When resistivity is mentioned
591: we refer to "effective" resistivity.
592: Normalising the frequency by IC frequency $\omega_{ci}$,
593: length by ion skin depth $c / \omega_{pi}$ and 
594: velocity by speed of light $c$,
595: the dimensionless resistivity is given by, 
596: \begin{eqnarray}
597:   \frac{\eta}{c^2 \omega_{pi}/ \omega_{ci}^2} &=&
598:   \frac{2}{27} 
599:   \left( \frac{\omega_d}{\omega_{ci}} \right)^{-2}
600:   \left( \frac{L}{c / \omega_{pi}}    \right)^2
601:   \left( \frac{L_d}{c / \omega_{pi}}  \right)^{-3}
602:   \left( \frac{v_A}{c}                \right)^3 \nonumber \\
603:   &&
604:   \sqrt{1 + 3 \exp \left[- \left(
605:   \frac{y - y_c}{L}\right)^6 \right]} \nonumber \\
606:   &&
607:   \exp \left[2 \left( \frac{y - y_c}{L} \right)^6 \right]
608:   \left( \frac{y - y_c}{L} \right)^{-10}.
609:   \label{eta}
610: \end{eqnarray}
611: We can now put in the known parameters $L / (c / \omega_{pi}) = 1.25$, 
612: $v_A / c = 0.25$ and $(y - y_c) / L = (148.5 - 100) / 50 = 0.97$ 
613: into Eq.~(\ref{eta}).
614: Here $L$ and $y_c$ are the width and the centre of plasma density gradient, 
615: respectively.
616: Therefore the dimensionless resistivity can be estimated using
617: \begin{equation}
618:   \frac{\eta}{c^2 \omega_{pi}/ \omega_{ci}^2} = 1.97 \times 10^{-2}
619:   \left( \frac{\omega_d}{\omega_{ci}} \right)^{-2}
620:   \left( \frac{L_d}{c / \omega_{pi}}  \right)^{-3}.
621:   \label{eta_pic}
622: \end{equation}
623: % Wd/Wci  Ld/(c/wpi)  eta/(c^2wpi/wci^2)  eta
624: % 0.2     43.75       5.88E-6             2.76E4
625: % 0.3     28.75       9.21E-6             4.32E4
626: % 0.4     21.25       1.28E-5             6.02E4
627: % 0.5     15.75       2.02E-5             9.46E4
628: As the driving wave frequency and the dissipation length for each case are given by choice and empirically,
629: respectively, one can obtain the resistivity by substituting above values into 
630: Eq.~(\ref{eta_pic}).
631: The result is $\eta / (c^2 \omega_{pi}/ \omega_{ci}^2) = 5.88 \times 10^{-6}$ 
632: (for $\omega_d / \omega_{ci} = 0.2$), 
633: $9.21 \times 10^{-6}$ (for $\omega_d / \omega_{ci}=0.3$), 
634: $1.28 \times 10^{-5}$ (for $\omega_d / \omega_{ci}=0.4$) and 
635: $2.02 \times 10^{-5}$ (for $\omega_d / \omega_{ci}=0.5$).
636: Thus our initial conjecture that the effective resistivity depends on the
637: driving IC wave frequency turns out to be correct.
638: But the main conclusion of this analysis is that 
639: despite collisional MHD scaling Eq.(\ref{mhd_decay_law}) being applicable
640: to the collisionless, kinetic regime of phase mixing, i.e. 
641: $B_z(x) \propto \exp\left(-(x/L_d)^3\right)$ scaling holds,
642: stretching the MHD-kinetic analogy further to $\omega_d^2$ dependence
643: under the exponent
644: is not valid (due to the effective resistivity being a function of $\omega_d$).
645: 
646: 
647: \subsection{Effective  anomalous resistivity}
648: 
649: Issue of anomalous resistivity is central for many space and laboratory 
650: plasma applications. It can facilitate fast magnetic reconnection via Petschek type
651: mechanism (if $\eta$ is not spatially uniform), or have significant implications for wave heating models of
652: solar corona where normal Spitzer resistivity is too small to produce any sizable effect.
653: Ref.\cite{anres} presented plasma resistivity measurements in the reconnection 
654: current sheet of Magnetic Reconnection 
655: Experiment (MRX) (see ref.\cite{yamada97} for details of the
656: experimental setup). They established that in the collisionless regime
657: measured resistivity values can be more than an order of magnitude larger than the
658: Spitzer value \cite{anres}.
659: 
660: Let us apply our PIC simulation results to see
661: if there is any evidence for the anomalous effective
662: resistivity. We fix physical parameters corresponding to 
663: solar coronal plasmas: $B = 0.01 T$ and 
664: plasma number density $n_0 = 2 \times 10^{15} m^{-3}$, i.e.
665: $\omega_{ci} = eB / (16 m_e) = 1.10 \times 10^8 rad/s$ and 
666: $\omega_{pi} = \sqrt{n_0 e^2 / (16 m_e \epsilon_0)} = 6.30 \times 10^8 rad/s$.
667: Eq.~(\ref{eta_pic}) can be rewritten as:
668: \begin{equation}
669:   \eta = 9.24 \times 10^{7}
670:   \left( \frac{\omega_d}{\omega_{ci}} \right)^{-2}
671:   \left( \frac{L_d}{c / \omega_{pi}}  \right)^{-3}.
672:   \label{eta_real}
673: \end{equation}
674: Similarly to the previous calculation in Eq.~(\ref{eta_pic}), 
675: now using Eq.~(\ref{eta_real}) we  obtain
676: $\eta = 2.76 \times 10^4$ (for $\omega_d/ \omega_{ci} = 0.2$),
677: $4.32 \times 10^4$ (for $\omega_d/ \omega_{ci} =0.3)$,
678: $6.02 \times 10^4$ (for $\omega_d/ \omega_{ci} =0.4)$ and
679: $9.46 \times 10^4$ (for $\omega_d/ \omega_{ci} =0.5)$.
680: Here units of the resistivity are  m$^2$ sec$^{-1}$.
681: We gather that all values are in the range of $\approx 10^4-10^5$ m$^2$ sec$^{-1}$.
682: Spitzer resistivity (normalised to 
683: $\mu_0$) for the above parameters and $T=1$ MK is 1.83 m$^2$ sec$^{-1}$.
684: Thus, we conclude that our numerical simulations provide
685: effective resistivity values of $\approx 10^4$ times
686: larger than Spitzer value, which is indicative of
687: the anomalous resistivity.
688: It should be mentioned that these results were obtained for
689: the ion-to-electron mass ratio of 16. Clearly one
690: would expect some dependence of the obtained effective resistivity 
691: on the mass ratio. Thus, the obtained results should be taken with caution.
692: %===============================================================================
693: % Conclusions
694: %===============================================================================
695: \section{Conclusions}
696: 
697: Let us summarise the above findings:
698: 
699: We used the generalised Ohm's law and found that the parallel electric field, 
700: which is generated by propagation of IC (Alfv\'enic) wave in a transversely inhomogeneous plasma,
701: is supported mostly by
702: the electron pressure tensor, with a smaller contribution from the electron inertia term.
703: Surprisingly, this result resembles closely to the 
704: previous results on collisionless reconnection both in tearing unstable 
705: Harris current sheet \cite{hesse99,birn01,pritchett01} and stressed X-point collapse 
706: \cite{th07,th08}. However, in the latter two cases, the generated electric field is in
707: the plane perpendicular to the magnetic field.
708: Thus, a universal importance of
709: the  electron pressure tensor in relation to supporting 
710: the electric fields in collisionless plasmas should be noted.
711: 
712:  
713: We explored physical parameter space of the problem with regards
714: to the efficiency of generation of parallel electric field
715: and acceleration of electrons. We found that
716: the generated parallel electric field and the fraction of accelerated electrons are 
717: independent of the IC wave
718:  frequency staying at a level that is $10^6$ times larger 
719:  than the Dreicer value and approximately
720: 20 \% respectively. The generated parallel electric field and the fraction of accelerated electrons 
721: increase with the increase of IC wave  amplitude. The generated parallel electric field seems to be
722: independent of  plasma-$\beta$. However, the fraction of accelerated electrons strongly increases 
723: with the decrease of plasma-$\beta$, e.g.  for plasma $\beta=0.0001$ the fraction of accelerated electrons
724: can be as large as 47 \%.
725: 
726: Previously it was established  that in the collisionless, kinetic regime
727: phase-mixed Alfv\'en (IC) wave amplitude damps with distance of propagation
728: according to $\propto \exp[-(x/L_d)^3]$ \cite{2005A&A...435.1105T}, which
729: resembles closely to collisional MHD result of \citet{1983A&A...117..220H}.
730: We tried to stretch this analogy further by investigating 
731: how the dissipation length $L_d$ scales with the IC driving frequency.
732: We found that the scaling is different from the MHD result.
733: We have shown that this discrepancy can be attributed to 
734: the frequency dependence of the effective resistivity.
735: 
736: We also found that  the effective resistivity, albeit for unrealistic mass ratio,
737: still is as large as $10^4$ times the classical Spitzer value.
738: 
739: \begin{acknowledgements}
740: Numerical calculations of this work were performed using the MHD Cluster 
741: at University of St Andrews. 
742: Author acknowledges useful discussion of solar flare 
743: observational aspects with Dr. E. Kontar.
744: This research was supported by the United Kingdom's 
745: Science and Technology Facilities Council (STFC).
746: \end{acknowledgements}
747: 
748: %%%%%%%%%%%%%%\bibliography{pm} 
749: 
750: \begin{thebibliography}{28}
751: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
752: \expandafter\ifx\csname bibnamefont\endcsname\relax
753:   \def\bibnamefont#1{#1}\fi
754: \expandafter\ifx\csname bibfnamefont\endcsname\relax
755:   \def\bibfnamefont#1{#1}\fi
756: \expandafter\ifx\csname citenamefont\endcsname\relax
757:   \def\citenamefont#1{#1}\fi
758: \expandafter\ifx\csname url\endcsname\relax
759:   \def\url#1{\texttt{#1}}\fi
760: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
761: \providecommand{\bibinfo}[2]{#2}
762: \providecommand{\eprint}[2][]{\url{#2}}
763: 
764: \bibitem[{\citenamefont{{Uberoi}}(1972)}]{1972PhFl...15.1673U}
765: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{{Uberoi}}},
766:   \bibinfo{journal}{Phys. Fluids} \textbf{\bibinfo{volume}{15}},
767:   \bibinfo{pages}{1673} (\bibinfo{year}{1972}).
768: 
769: \bibitem[{\citenamefont{{Tataronis} and {Grossmann}}(1973)}]{1973ZPhy..261..203T}
770: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Tataronis}}} \bibnamefont{and}
771:   \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{{Grossmann}}},
772:   \bibinfo{journal}{Zeitschrift fur Physik} \textbf{\bibinfo{volume}{261}},
773:   \bibinfo{pages}{203} (\bibinfo{year}{1973}).
774: 
775: \bibitem[{\citenamefont{{Grossmann} and {Tataronis}}(1973)}]{1973ZPhy..261..217G}
776: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{{Grossmann}}} \bibnamefont{and}
777:   \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Tataronis}}},
778:   \bibinfo{journal}{Zeitschrift fur Physik} \textbf{\bibinfo{volume}{261}},
779:   \bibinfo{pages}{217} (\bibinfo{year}{1973}).
780: 
781: \bibitem[{\citenamefont{{Hasegawa} and {Chen}}(1974)}]{1974PhRvL..32..454H}
782: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Hasegawa}}} \bibnamefont{and}
783:   \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{{Chen}}},
784:   \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{32}},
785:   \bibinfo{pages}{454} (\bibinfo{year}{1974}).
786: 
787: \bibitem[{\citenamefont{{Chen} and {Hasegawa}}(1974)}]{1974PhFl...17.1399C}
788: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{{Chen}}} \bibnamefont{and}
789:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Hasegawa}}},
790:   \bibinfo{journal}{Phys. Fluids} \textbf{\bibinfo{volume}{17}},
791:   \bibinfo{pages}{1399} (\bibinfo{year}{1974}).
792: 
793: \bibitem[{\citenamefont{{Tataronis}}(1975)}]{1975JPlPh..13...87T}
794: \bibinfo{author}{\bibfnamefont{J.~A.} \bibnamefont{{Tataronis}}},
795:   \bibinfo{journal}{J. Plasma Phys.} \textbf{\bibinfo{volume}{13}},
796:   \bibinfo{pages}{87} (\bibinfo{year}{1975}).
797: 
798: \bibitem[{\citenamefont{{Heyvaerts} and {Priest}}(1983)}]{1983A&A...117..220H}
799: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Heyvaerts}}} \bibnamefont{and}
800:   \bibinfo{author}{\bibfnamefont{E.~R.} \bibnamefont{{Priest}}},
801:   \bibinfo{journal}{Astron. Astrophys.} \textbf{\bibinfo{volume}{117}},
802:   \bibinfo{pages}{220} (\bibinfo{year}{1983}).
803: 
804: \bibitem[{\citenamefont{{G\'enot} et~al.}(1999)\citenamefont{{G\'enot},  {Louarn}, and {Le Qu\'eau}}}]{1999JGR...10422649G}
805: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{{G\'enot}}},
806:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{{Louarn}}}, \bibnamefont{and}
807:   \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Le Qu\'eau}}},
808:   \bibinfo{journal}{J. \ Geophys. \ Res.} \textbf{\bibinfo{volume}{104}},
809:   \bibinfo{pages}{22649} (\bibinfo{year}{1999}).
810: 
811: \bibitem[{\citenamefont{{G\'enot} et~al.}(2004)\citenamefont{{G\'enot},   {Louarn}, and {Mottez}}}]{2004AnGeo..22.2081G}
812: \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{{G\'enot}}},
813:   \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{{Louarn}}}, \bibnamefont{and}
814:   \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{{Mottez}}},
815:   \bibinfo{journal}{Ann. Geophysicae} \textbf{\bibinfo{volume}{22}},
816:   \bibinfo{pages}{2081} (\bibinfo{year}{2004}).
817: 
818: \bibitem[{\citenamefont{{Tsiklauri}   et~al.}(2005{\natexlab{a}})\citenamefont{{Tsiklauri}, {Sakai}, and {Saito}}}]{2005A&A...435.1105T}
819: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}},
820:   \bibinfo{author}{\bibfnamefont{J.-I.} \bibnamefont{{Sakai}}},
821:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{{Saito}}},
822:   \bibinfo{journal}{Astron. Astrophys.} \textbf{\bibinfo{volume}{435}},
823:   \bibinfo{pages}{1105} (\bibinfo{year}{2005}{\natexlab{a}}).
824: 
825: \bibitem[{\citenamefont{{Tsiklauri} et~al.}(2005{\natexlab{b}})\citenamefont{{Tsiklauri}, {Sakai}, and {Saito}}}]{2005NJPh....7...79T}
826: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}},
827:   \bibinfo{author}{\bibfnamefont{J.-I.} \bibnamefont{{Sakai}}},
828:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{{Saito}}},
829:   \bibinfo{journal}{New J. Physics} \textbf{\bibinfo{volume}{7}},
830:   \bibinfo{pages}{79} (\bibinfo{year}{2005}{\natexlab{b}}).
831: 
832: \bibitem[{\citenamefont{{Mottez} et~al.}(2006)\citenamefont{{Mottez}, {G{\'e}not}, and {Louarn}}}]{2006A&A...449..449M}
833: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{{Mottez}}},
834:   \bibinfo{author}{\bibfnamefont{V.}~\bibnamefont{{G{\'e}not}}},
835:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{P.}~\bibnamefont{{Louarn}}},
836:   \bibinfo{journal}{Astron. Astrophys.} \textbf{\bibinfo{volume}{449}},
837:   \bibinfo{pages}{449} (\bibinfo{year}{2006}).
838: 
839: \bibitem[{\citenamefont{{Tsiklauri}}(2007)}]{2007NJPh....9..262T}
840: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}},
841:   \bibinfo{journal}{New J. Physics} \textbf{\bibinfo{volume}{9}},
842:   \bibinfo{pages}{262} (\bibinfo{year}{2007}).
843: 
844: \bibitem[{\citenamefont{{Tsiklauri}}(2006)}]{2006A&A...455.1073T}
845: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}},
846:   \bibinfo{journal}{Astron. Astrophys.} \textbf{\bibinfo{volume}{455}},
847:   \bibinfo{pages}{1073} (\bibinfo{year}{2006}).
848: 
849: \bibitem[{\citenamefont{{Pritchett}}(2001)}]{pritchett01}
850: \bibinfo{author}{\bibfnamefont{P.~L.} \bibnamefont{{Pritchett}}},
851:   \bibinfo{journal}{J. \ Geophys. \ Res.} \textbf{\bibinfo{volume}{106}},
852:   \bibinfo{pages}{3783} (\bibinfo{year}{2001}).
853: 
854: \bibitem[{\citenamefont{{Tsiklauri} and {Haruki}}(2008)}]{th08}
855: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}} \bibnamefont{and}
856:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{{Haruki}}},
857:   \bibinfo{journal}{{\it "Physics of collisionless reconnection in a stressed
858:   X-point collapse"}, Physics of Plasmas (in press)}  (\bibinfo{year}{2008}).
859: 
860: \bibitem[{\citenamefont{{Fletcher}}(2005)}]{2005SSRv..121..141F}
861: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{{Fletcher}}},
862:   \bibinfo{journal}{Space Science Reviews} \textbf{\bibinfo{volume}{121}},
863:   \bibinfo{pages}{141} (\bibinfo{year}{2005}).
864: 
865: \bibitem[{\citenamefont{Moran}(2001)}]{moran01}
866: \bibinfo{author}{\bibfnamefont{T.~G.} \bibnamefont{Moran}},
867:   \bibinfo{journal}{Astron. \ Astrophys.} \textbf{\bibinfo{volume}{374}},
868:   \bibinfo{pages}{L9} (\bibinfo{year}{2001}).
869: 
870: \bibitem[{\citenamefont{{Hesse} et~al.}(1999)\citenamefont{{Hesse}, {Schindler}, {Birn}, and {Kuznetsova}}}]{hesse99}
871: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Hesse}}},
872:   \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{{Schindler}}},
873:   \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Birn}}}, \bibnamefont{and}
874:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Kuznetsova}}},
875:   \bibinfo{journal}{Phys. Plasmas} \textbf{\bibinfo{volume}{6}},
876:   \bibinfo{pages}{1781} (\bibinfo{year}{1999}).
877: 
878: \bibitem[{\citenamefont{{Birn} et~al.}(2001)\citenamefont{{Birn}, {Drake}, {Shay}, {Rogers}, {Denton}, {Hesse}, {Kuznetsova}, {Ma},{Bhattacharjee}, {Otto} }}]{birn01}
879: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Birn}}},
880:   \bibinfo{author}{\bibfnamefont{J.~F.} \bibnamefont{{Drake}}},
881:   \bibinfo{author}{\bibfnamefont{M.~A.} \bibnamefont{{Shay}}},
882:   \bibinfo{author}{\bibfnamefont{B.~N.} \bibnamefont{{Rogers}}},
883:   \bibinfo{author}{\bibfnamefont{R.~E.} \bibnamefont{{Denton}}},
884:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Hesse}}},
885:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Kuznetsova}}},
886:   \bibinfo{author}{\bibfnamefont{Z.~W.} \bibnamefont{{Ma}}},
887:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Bhattacharjee}}},
888:   \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Otto}}}, 
889:   \bibinfo{journal}{J. \ Geophys. \ Res.}
890:   \textbf{\bibinfo{volume}{106}}, \bibinfo{pages}{3715} (\bibinfo{year}{2001}).
891: 
892: \bibitem[{\citenamefont{{Tsiklauri} and {Haruki}}(2007)}]{th07}
893: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}} \bibnamefont{and}
894:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{{Haruki}}},
895:   \bibinfo{journal}{Phys. Plasmas} \textbf{\bibinfo{volume}{14}},
896:   \bibinfo{pages}{112905} (\bibinfo{year}{2007}).
897: 
898: \bibitem[{\citenamefont{{Brown} et~al.}(2003)\citenamefont{{Brown}, {Emslie},  and {Kontar}}}]{bek03}
899: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{{Brown}}},
900:   \bibinfo{author}{\bibfnamefont{A.~G.} \bibnamefont{{Emslie}}},
901:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{E.~P.}
902:   \bibnamefont{{Kontar}}}, \bibinfo{journal}{Astrophys. J.}
903:   \textbf{\bibinfo{volume}{595}}, \bibinfo{pages}{L115} (\bibinfo{year}{2003}).
904: 
905: \bibitem[{\citenamefont{{Brown} and {Kontar}}(2005)}]{bk05}
906: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{{Brown}}} \bibnamefont{and}
907:   \bibinfo{author}{\bibfnamefont{E.~P.} \bibnamefont{{Kontar}}},
908:   \bibinfo{journal}{Adv. Space Res.} \textbf{\bibinfo{volume}{35}},
909:   \bibinfo{pages}{1675} (\bibinfo{year}{2005}).
910: 
911: \bibitem[{\citenamefont{{Fletcher} and {Hudson}}(2008)}]{2008ApJ...675.1645F}
912: \bibinfo{author}{\bibfnamefont{L.}~\bibnamefont{{Fletcher}}} \bibnamefont{and}
913:   \bibinfo{author}{\bibfnamefont{H.~S.} \bibnamefont{{Hudson}}},
914:   \bibinfo{journal}{Astrophys. J.} \textbf{\bibinfo{volume}{675}},
915:   \bibinfo{pages}{1645} (\bibinfo{year}{2008}).
916: 
917: \bibitem[{\citenamefont{Dreicer}(1959)}]{dreicer}
918: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{Dreicer}},
919:   \bibinfo{journal}{Phys. Rev.} \textbf{\bibinfo{volume}{115}},
920:   \bibinfo{pages}{238} (\bibinfo{year}{1959}).
921: 
922: \bibitem[{\citenamefont{{Tsiklauri} et~al.}(2002)\citenamefont{{Tsiklauri},   {Nakariakov}, and {Arber}}}]{2002A&A...395..285T}
923: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}},
924:   \bibinfo{author}{\bibfnamefont{V.~M.} \bibnamefont{{Nakariakov}}},
925:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{T.~D.}
926:   \bibnamefont{{Arber}}}, \bibinfo{journal}{Astron. Astrophys.}
927:   \textbf{\bibinfo{volume}{395}}, \bibinfo{pages}{285} (\bibinfo{year}{2002}).
928: 
929: \bibitem[{\citenamefont{{Trintchouk} et~al.}(2003)\citenamefont{{Trintchouk}, {Yamada}, {Ji}, {Kulsrud}, and {Carter}}}]{anres}
930: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{{Trintchouk}}},
931:   \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Yamada}}},
932:   \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{{Ji}}},
933:   \bibinfo{author}{\bibfnamefont{R.~M.} \bibnamefont{{Kulsrud}}},
934:   \bibnamefont{and} \bibinfo{author}{\bibfnamefont{T.~A.}
935:   \bibnamefont{{Carter}}}, \bibinfo{journal}{Phys. Plasmas}
936:   \textbf{\bibinfo{volume}{10}}, \bibinfo{pages}{319} (\bibinfo{year}{2003}).
937: 
938: \bibitem[{\citenamefont{{Yamada} et~al.}(1997)\citenamefont{{Yamada}, {Ji}, {Hsu}, {Carter}, {Kulsrud}, {Bretz}, {Jobes}, {Ono}, and {Perkins}}}]{yamada97} 
939: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Yamada}}},
940:   \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{{Ji}}},
941:   \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{{Hsu}}},
942:   \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{{Carter}}},
943:   \bibinfo{author}{\bibfnamefont{R.}~\bibnamefont{{Kulsrud}}},
944:   \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{{Bretz}}},
945:   \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{{Jobes}}},
946:   \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{{Ono}}}, \bibnamefont{and}
947:   \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{{Perkins}}},
948:   \bibinfo{journal}{Phys. Plasmas} \textbf{\bibinfo{volume}{4}},
949:   \bibinfo{pages}{1936} (\bibinfo{year}{1997}).
950: 
951: \end{thebibliography}
952: 
953: \end{document}
954: