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 reconnection in a
14: stressed X-point collapse}
15: \author{D. Tsiklauri}
16: \author{T. Haruki}
17:
18: \affiliation{Joule Physics Laboratory,
19: Institute for Materials Research, University of Salford, Manchester, M5 4WT,
20: United Kingdom}
21: \date{\today}
22:
23: \begin{abstract}
24: Recently, magnetic reconnection during collisionless, stressed, X-point collapse was studied
25: using kinetic, 2.5D, fully electromagnetic, relativistic
26: Particle-in-Cell numerical code [D. Tsiklauri and T.
27: Haruki, Phys. Plasmas {\bf 14}, 112905 (2007)]. Here we finalise the investigation of this topic by addressing
28: key outstanding physical questions: (i) which term in the generalised Ohm's law is responsible
29: for the generation of the reconnection electric field?
30: (ii) how does the time evolution of the reconnected flux vary with the ion-electron mass ratio?
31: (iii) what is the exact energy budget of the reconnection process, i.e. in which proportion initial
32: (mostly magnetic) energy is converted into other forms of energy?
33: (iv) are there any anisotropies in the velocity distribution of the accelerated particles?
34: It has been established here that: (i) reconnection electric field is generated by
35: the electron pressure tensor off-diagonal terms, resembling to the case of tearing unstable
36: Harris current sheet studied by the GEM reconnection challenge;
37: (ii) For $m_i / m_e \gg 1$ the time evolution of the reconnected
38: flux is independent of ion-electron mass ratio;
39: also, in the case of $m_i / m_e = 1$ we show that reconnection proceeds slowly as the
40: Hall term is zero; when $m_i / m_e \gg 1$ (i.e. the Hall term is non-zero) reconnection is fast and
41: we conjecture that this is due to magnetic field being frozen into electron fluid, which
42: moves significantly faster than ion fluid;
43: (iii) within one Alfv\'en time, somewhat less than half ($\sim 40$\%) of the
44: initial total (roughly magnetic) energy is converted into the kinetic
45: energy of electrons, and somewhat more than half ($\sim 60$\%) into kinetic
46: energy of ions (similar to solar flare observations);
47: (iv) in the {\it strongly} stressed X-point case, in about one Alfv\'en time, a full
48: isotropy in all three spatial directions of the velocity distribution is seen for super-thermal
49: electrons (also commensurate to solar flare observations).
50: \end{abstract}
51:
52: \pacs{52.35.Vd; 96.60.Iv; 52.65.Rr; 45.50.Dd; 96.60.pf; 96.60.qe}
53:
54: \maketitle
55:
56: \section{Introduction}
57:
58: In many astrophysical or laboratory plasma situations (a) plasma beta is small, indicative of large
59: amounts of energy stored in a form of magnetic field and (b) there is a need to
60: explain or provide plasma heating, as well
61: as plasma particle acceleration. It is believed that in such situations magnetic reconnection, i.e. change
62: of connectivity of magnetic field lines that penetrate the
63: plasma, can serve as one of the important possible mechanisms.
64: There are different types of magnetic reconnection. One of the key descriptors is plasma collisionality,
65: i.e. if plasma is collisional then magnetic resistivity, $\eta$ (or more specifically $\eta {\vec j}$ term
66: in the generalised Ohm's law) is responsible for breaking the frozen-in condition
67: (enabling field line connectivity change).
68: However, if plasma is collisionless, then other terms in the generalised Ohm's law may be more important.
69: In this context, it is instructive to look at typical spatial
70: scales. Let us consider an example of solar coronal plasma.
71: Typical width of a Sweet-Parker current sheet is given by $\delta=S^{-1/2} L$ \cite{biskamp} (p. 54).
72: Fixing coronal temperature at 1 MK, Coulomb logarithm at 18.0948, the Lundquist number (using Spitzer
73: resistivity from \citet{NRLF}, p. 30) is $5.37933 \times 10^{12}$, which for $L=10$ Mm yields, $\delta=4.31157$ m.
74: Another way of looking at $\delta$ is associating it also with the
75: resistive length scale via Alfv\'en time scale $\tau_A = L / V_A$, where $V_A$ is the Alfv\'en speed
76: ($\approx 1$ Mm s$^{-1}$): i.e. $\delta=S^{-1/2} L=\eta \tau_A / \mu_0$.
77: Typical scale associated with the Hall term in the generalised Ohm's law at which deviation from
78: electron-ion coupled dynamics is observed is, $c/\omega_{pi} = 7.20064$ m (ion inertial length). Here
79: particle density of $n=10^{15}$ m$^{-3}$ is used.
80: The fact that $\delta / (c/\omega_{pi}) \leq 1 $ points to a necessity of
81: going beyond single fluid MHD approximation.
82: The Hall term itself cannot break the frozen-in condition, its inclusion into consideration
83: ensures that the magnetic field is frozen into electron fluid. Below, we shall use this argument to conjecture
84: why reconnection is fast when the Hall term is included.
85: Importance of different terms in the generalised Ohm's law is usually inferred
86: by comparing the spatial scales associated with them to the resistive length scale $\delta$.
87: E.g. one of the other noteworthy scales is
88: $c/\omega_{pe} = 0.16804$ m (electron inertial length) on which the electron inertia term operates.
89: When $\delta / (c/\omega_{pe}) < 1$ then electron inertia would dominate over resistive diffusion \cite{biskamp} (p. 200).
90: As can be seen from the above estimates $\delta / (c/\omega_{pe}) = 25.6576 \gg 1 $ in the solar
91: corona, thus electron inertia effects seem to be negligible.
92: However, electron inertia in contrary to the Hall term (as well as the electron pressure tensor)
93: can break the frozen-in condition and thus change the magnetic field connectivity.
94: Because, of the fact that with increase of $T$ (hot plasmas),
95: $\delta$ gets progressively smaller thus effects other than the resistivity should be included.
96: Indeed, collisionless (non-resistive) reconnection has recently attracted considerable
97: attention (see \citet{bp07} for a review).
98:
99: One of the first studies of magnetic reconnection is stressed, X-point collapse \cite{d53} (also see Chap. 7.1 in
100: \citet{pf00}). The latter was using resistive MHD approach. We recently
101: revisited the problem in the regime of collisionless reconnection \cite{th07}.
102: In Ref. \cite{th07} we studied the magnetic reconnection during collisionless, stressed, X-point collapse
103: using kinetic, 2.5D, fully electromagnetic, relativistic Particle-in-Cell numerical code.
104: We considered two cases of weakly and strongly stressed X-point collapse.
105: Where descriptors weakly and strongly
106: refer to 20 \% and 124 \% unidirectional spatial compression of the X-point, respectively.
107: Amongst other interesting outcomes, we established that
108: within about one Alfv\'en time, 2\% and 20\% of the initial
109: magnetic energy can be converted into heat and accelerated particle energy in the
110: cases of weak and strong stress, respectively. However, open questions remained:
111: (i) which term in the generalised Ohm's law is responsible
112: for the generation of the reconnection electric field?
113: (ii) how does the time evolution of the reconnected flux vary with the ion-electron mass ratio?
114: (iii) what is the exact energy budget of the reconnection process, i.e. in which proportion initial
115: (mostly magnetic) energy is converted into other forms of energy?
116: (iv) are there any anisotropies in the velocity distribution of the accelerated particles?
117: Here we finalise the study of magnetic, collisionless reconnection of a stressed X-point by
118: providing answers to these questions.
119:
120: \section{The Model}
121:
122: The numerical code used here is 2.5D,
123: relativistic, fully electromagnetic PIC code, with the initial conditions
124: the same as in our previous work \cite{th07}. For completeness we re-iterate
125: key points:
126: Magnetic field configuration is an X-point without a guide-field
127: \begin{equation}
128: (B_x, B_y, B_z) = \frac{B_0}{L} (y, \alpha^2 x, 0),
129: \label{eq:init_b}
130: \end{equation}
131: where $B_0$ is magnetic field intensity at the distance $L$ from the X-point ($L$ is the global system scale).
132: $\alpha$ is the
133: {\it stress parameter}, which prescribes
134: the initial strength of magnetic pressure that collapses the system, due to lack of restoring force \cite{pf00}.
135: Using $\mu_0 \vec{j} = \nabla \times \vec{B}$,
136: a uniform current is imposed in the $z$ direction,
137: \begin{equation}
138: j_z = \frac{B_0}{\mu_0 L} (\alpha^2 - 1).
139: \label{eq:init_j}
140: \end{equation}
141: Electrons and ions have uniform spatial, and Maxwellian velocity distributions
142: throughout the system.
143: For $\alpha = 1$, magnetic field
144: geometry is completely symmetric, $j_z$ current zero
145: (see Eqs.~(\ref{eq:init_b})-(\ref{eq:init_j})), and thus, such
146: magnetic configuration is stable.
147: For $\alpha > 1$, stressed X-point starts collapse in the $x$ direction
148: because of the absence of a restoring force, causing time-transient
149: magnetic reconnection.
150: The main parameters of the standard simulation
151: model are as follows.
152: The length of the system in
153: two dimensions is $L_x = L_y = 400 \Delta$ (this is excluding so-called ghost cells),
154: where $\Delta = 1$ is the simulation grid size corresponding
155: to electron Debye length, $\lambda_D = v_{te} / \omega_{pe} = 1 \Delta$ ($v_{te}$ is
156: electron thermal velocity and $\omega_{pe}$ is electron plasma frequency).
157: The global reconnection scale is set $L = 200 \Delta$.
158: The number density is fixed at $n_0 = 100$ electron-ion pairs per cell.
159: Hence the total number is $1.6 \times 10^7$ pairs.
160: The simulation time step is $\omega_{pe} \Delta t = 0.05$.
161: Ion-to-electron mass ratio is fixed at
162: $m_i / m_e = 100$ (which is varied in Fig. 2 (subsection III.B) only).
163: The electron thermal velocity to speed of light ratio is $v_{te} / c = 0.1$.
164: The electron and ion skin depths are $c / \omega_{pe} = 10 \Delta$ and $c / \omega_{pi} = 100 \Delta$, respectively.
165: The electron cyclotron frequency to plasma
166: frequency ratio is $\omega_{ce} / \omega_{pe} = 1.0$ for magnetic field intensity, $B = B_0$.
167: This ratio is
168: close to unity in the solar
169: corona, while it is much
170: bigger than unity in the Earth magnetosphere.
171: The electron and ion Larmor
172: radii are $v_{te} / \omega_{ce} = 1 \Delta$ and $v_{ti} / \omega_{ci} = 10 \Delta$, where $v_{ti}$
173: is the ion thermal velocity.
174: Initial temperatures of electrons and ions are initially
175: set the same, $T_e = T_i$.
176: At the boundary ($B = B_0$ at the distance $L$ from the X-point), the plasma $\beta = 0.02$ and
177: Alfv\'en velocity, $V_{A0} / c = 0.1$.
178: Naturally these vary across the
179: simulation box as the background magnetic field is a function of $x$ and $y$.
180:
181: The boundary conditions on EM-fields are zero-gradient and
182: also, tangential component of electric field
183: was forced to zero, while normal component of magnetic field was kept
184: constant, both at the boundary.
185: This ensures that there is no change in magnetic flux through the simulation box,
186: i.e. the system is isolated.
187: When colliding with boundaries particles are reflected. Thus our
188: boundary conditions ensure there is no magnetic influx or
189: mass transport across the boundaries. It has been also confirmed that
190: the total energy in the system is conserved during the simulations to a good accuracy.
191:
192: \section{Results}
193:
194: Before we address the outstanding questions, we refer reader to
195: \citet{th07} for a more detailed description of dynamics of EM-fields, currents, and particles.
196: It is our intention to focus on the outstanding questions here.
197: In brief, the previous results can be summarised as follows:
198: when $\alpha > 1$, the stressed X-point collapses in the $x$ direction
199: due to the absence of a restoring force, and hence time-transient
200: magnetic reconnection occurs. The fast reconnection regime is achieved.
201: Initially uniform out-of-plane current becomes localised,
202: peaking at many times its initial value in a time transient manner.
203: Also, out-of-plane quadruple
204: magnetic field is generated.
205: High energy part of the electron distribution
206: function exhibits a power-law behaviour. Sizable fraction of
207: initial magnetic energy is converted into other forms of energy.
208:
209: \subsection{Source of the reconnection electric field}
210:
211: \begin{figure}
212: \includegraphics[scale = 0.5]{fig1.eps}% Here is how to import EPS art
213: \caption {\label{fig:ohms_law} Line plots of different terms in the
214: generalised Ohm's law along
215: $y$ direction, in $x = 0$, at $\omega_{pe} t = 170$ (time-transient reconnection peak)
216: for $\alpha = 1.20$.
217: Solid lines in panels (a)-(d) show the different terms as follows:
218: (a) $v_{ey} B_x$,
219: (b) $- (d_i / n) \partial P_{exz} / \partial x$,
220: (c) $- (d_i / n) \partial P_{eyz} / \partial y$ and (d) $- d_i (m_e / m_i) v_{ey} \partial v_{ez} / \partial y$.
221: The reconnection electric field $E_z(0,y)$ (normalised to $V_{A0} B_0$) is
222: shown with the dashed line in all four panels. Here $y$ is normalised to $c/\omega_{pi}$, and thus varies between
223: $-2<y<2$.}
224: \end{figure}
225:
226: In order to understand details of the collisionless reconnection process, we now
227: focus on the question: which term in the generalized Ohm's law is responsible
228: for the generation of the reconnection electric field? We adopt an appoach used by
229: \citet{pritchett01}. The generalized Ohm's law
230: can be written as (e.g. \cite{bp07} p. 108)
231: \begin{equation}
232: \vec{E} = - \vec{v}_e \times \vec{B}
233: - \frac{\nabla \cdot \vec{P}_e}{n_e e}
234: - \frac{m_e}{e} (\frac{\partial \vec{v}_e}{\partial t}
235: + \left( \vec{v}_e \cdot \nabla) \vec{v}_e \right),
236: \label{eq:ohms_law}
237: \end{equation}
238: where $\vec{E}$ and $\vec{B}$ are electric and magnetic fields, $\vec{v}$ is plasma
239: velocity, $\vec{P}$ is pressure tensor ($3 \times 3$ matrix), $n$ is
240: plasma number density, $m$ is mass and $e$ is electric charge.
241: The subscript $e$ refers to an electron.
242: Normalising space coordinates by global reconnection scale
243: $L$, fluid velocity by Alfv\'en speed $V_A$, time by
244: Alfv\'en transit time $\tau_A (= L / V_A)$,
245: magnetic field by $B_0$, number density by $n_0$ and
246: pressure tensor by $B_0^2 / \mu_0$, a dimensionless
247: version of Eq.(\ref{eq:ohms_law}) can be obtained
248: \begin{eqnarray}
249: \vec{E} = - \vec{v}_e \times \vec{B}
250: - d_i \frac{\nabla \cdot \vec{P}_e}{n_e}
251: - d_i \frac{m_e}{m_i} (\frac{\partial \vec{v}_e}{\partial t}
252: + \left( \vec{v}_e \cdot \nabla) \vec{v}_e \right),
253: \label{eq:nd_ohms_law}
254: \end{eqnarray}
255: where $d_i$ is the normalised ion skin depth ($d_i=c/\omega_{pi}L$).
256: Note that strictly speaking we should have used tildes in Eq.(\ref{eq:nd_ohms_law})
257: to denote dimensionless quantities, but we omit them for brevity.
258:
259: Let us focus on the out-of-plane component of the electric field $E_z$ at the magnetic null,
260: which is a measure of the reconnection rate. It is given by,
261: \begin{eqnarray}
262: E_z &=& - (v_{ex} B_y - v_{ey} B_x)
263: - d_i \frac{1}{n} \left( \frac{\partial P_{exz}}{\partial x}
264: + \frac{\partial P_{eyz}}{\partial y} \right) \nonumber \\
265: & & - d_i \frac{m_e}{m_i} \left( \frac{\partial v_{ez}}{\partial t}
266: + v_{ex} \frac{\partial v_{ez}}{\partial x}
267: + v_{ey} \frac{\partial v_{ez}}{\partial y} \right),
268: \label{eq:nd_ez}
269: \end{eqnarray}
270: where $\partial / \partial z = 0$ is assumed because of 2D reconnection model.
271:
272: The pressure tensor is defined as $P_{ij} = m \int v^{'}_i v^{'}_j f (\vec{r}, \vec{v}, t) d\vec{v}$,
273: where $m$ is mass, $v^{'}$ is random velocity,
274: the subscript $i$ and $j$ denote the components $x$, $y$ or $z$,
275: $f$ is the particle velocity distribution function,
276: $\vec{r}$ is position, and $\vec{v}$ is velocity.
277: In order to get the pressure tensor, number density is calculated first,
278: from $n(\vec{r}, t) = \int f (\vec{r}, \vec{v}, t) d\vec{v}$.
279: Mean velocity is also obtained
280: via $\vec{V} (\vec{r}, t) = (1 / n) \int \vec{v} f (\vec{r}, \vec{v}, t) d\vec{v}$.
281: For pressure tensor calculation, the number density, $n$ and
282: the mean velocity, $\vec{V} (\vec{r}, t)$, is calculated by
283: counting number of individual particles
284: per cell
285: and by computing the average
286: velocity in each cell, respectively.
287: We then estimate the random velocity, $\vec{v^{'}} = \vec{v} - \vec{V}$, which is
288: used in the above definition of the
289: pressure tensor $P_{ij}$. In PIC simulations,
290: in practise, the summation of $m v^{'}_i v^{'}_j$ over all individual particles is used.
291:
292: Figure \ref{fig:ohms_law} shows $y$-profiles of different terms in the
293: generalised Ohm's law at $x = 0$, for $\omega_{pe} t = 170$ (time-transient reconnection peak).
294: Here $\alpha = 1.20$. Solid lines in panels (a)-(d) indicate
295: different terms as follows: (a) $v_{ey} B_x$,
296: (b) $- (d_i / n) \partial P_{exz} / \partial x$,
297: (c) $- (d_i / n) \partial P_{eyz} / \partial y$ and (d) $- d_i (m_e / m_i) v_{ey} \partial v_{ez} / \partial y$.
298: The reconnection electric field $E_z(0,y)$ is shown with the dashed line in all four panels.
299: A boxcar average scheme with a width of 7 mesh points is applied for smoothing data.
300: The other terms in Eq.~(\ref{eq:nd_ez}) are negligibly small.
301: Fig.~\ref{fig:ohms_law} is analogous to figure 5 from \citet{pritchett01}.
302: Fig.~\ref{fig:ohms_law}(a) shows that in all regions {\it except} the magnetic
303: null, (0,0), contribution to $E_z(0,y)$ from the $\vec v_e \times \vec B$ term
304: is significant. However, $v_{ey} B_x$ is zero at the X-point (the magnetic null).
305: As seen in Fig.~\ref{fig:ohms_law}(b-c), the off-diagonal components of the
306: electron pressure tensor are major contributors to $E_z(0,0)$.
307: The electron inertia term also generates the electric field {\it away} from the X-point
308: (see Fig.~\ref{fig:ohms_law} (d)). Thus, we conclude that the reconnection electric field is generated by
309: the electron pressure tensor off-diagonal terms; and hence the latter are responsible for breaking
310: the frozen-in condition. A similar conclusion was reached by \citet{pritchett01}. This
311: coincidence seems somewhat unexpected, because X-point collapse considered here and
312: onset of tearing instability considered by \citet{pritchett01} are physically different.
313: Similarity of the cause of breaking of the frozen-in condition in both cases can only point to
314: a universal nature of this mechanism.
315:
316: \subsection{Effect of variation of the ion-electron mass ratio
317: and conjecture of fast reconnection}
318:
319: \begin{figure}
320: \includegraphics[scale = 0.5]{fig2.eps}% Here is how to import EPS art
321: \caption{\label{fig:mass_depend}
322: Time evolution of the magnetic flux difference between the O and X points (i.e. amount of reconnected flux).
323: The solid lines with progressively increasing thickness
324: show cases of $m_i / m_e = 9, 25, 64$ and $100$, respectively.
325: The dotted line shows electron-positron plasma case ($m_i / m_e = 1$).
326: The magnetic flux difference $\Delta \psi$ is normalised by $(B_0 c / \omega_{pi})$ and
327: then a unity is subtracted to start from zero.
328: Time is normalised by the ion cyclotron frequency $\omega_{ci} = e B / m_i$.
329: Here $\alpha=1.2$.}
330: \end{figure}
331:
332: The next question we consider is: how does the time
333: evolution of the reconnected flux vary with the ion-electron mass ratio?
334: Such question historically was relevant because of the inability of
335: performing realistic ion-electron mass ratio (1836) numerical simulation, due to lack
336: of computational resources. Although within our reach computationally, we do not
337: show here results $m_i / m_e > 100$ because the total energy conservation
338: error (which is defined as $(E(\omega_{pe} t = 250) - E(\omega_{pe} t = 0))/E(\omega_{pe} t = 0)$
339: and is indicative
340: of the code accuracy) starts to deteriorate to values of circa 10\% for $m_i / m_e = 400$, while
341: for $m_i / m_e = 100$ it is 0.04\% (both for $\alpha=1.2$).
342: In order to be able to compare our results with the previous work \cite{hesse99},
343: when varying $m_i / m_e $, we accordingly adjust number of spatial grid points
344: and total time step. Such adjustments insure that spatial scale of the simulation box, is
345: $L_x = L_y = 4 c / \omega_{pi}$, and
346: the time scale, $\omega_{ci} t = 25$.
347: Thus when setting $m_i / m_e =1, 9, 25, 64$ and $100$,
348: Accordingly, the system size is adjusted to $40, 120, 200, 320$ and $400 \Delta$.
349: The global reconnection size is fixed at $L = 200 \Delta$.
350: Ion cyclotron frequency for each case is
351: defined using by the magnetic intensity at the boundary.
352:
353: In 2D the magnetic flux function can be defined as
354: $\psi = - \int B_x dy = \int B_y dx$.
355: In our simulation, X-point is located at the centre of system $(x, y) = (0, 0)$,
356: while O-points are at $(x, y) = (0, -2)$ and $(0, 2)$.
357: Note that spatial coordinates here are normalised by ion skin depth, and $L_x = L_y = 4 c / \omega_{pi}$.
358: Therefore we can use the same definition of the
359: reconnected flux as in the case tearing mode-unstable Harris current sheet \cite{hesse99}.
360: Figure \ref{fig:mass_depend} shows time evolution of the magnetic flux difference
361: between O and X points (reconnected flux)
362: for different ion-to-electron mass ratios, $m_i / m_e = 1, 9, 25, 64$ and $100$.
363: We gather from this graph that time dynamics of the reconnected flux
364: does not depend on $m_i / m_e$ when $m_i / m_e \gg 1$ and that reconnection
365: is fast. In fact, the time derivative
366: of the reconnected flux is the reconnection rate. Thus, the conclusion is that
367: the reconnection rate is independent of the mass ratio (when $m_i / m_e \gg 1$). As with above
368: conclusion (in previous subsection) that the reconnection electric field is generated by
369: the electron pressure tensor off-diagonal terms; again similarity with the tearing-unstable
370: Harris current sheet holds, i.e. \citet{hesse99} came to the same conclusion in their case.
371:
372: As a further test, we performed a numerical run with $m_i / m_e = 1$ (case of
373: electron-positron plasma). One of the main conclusions of \citet{birn01} was that
374: as long as Hall term is included, the reconnection is fast. i.e.
375: when electron and ion dynamics can be distinguished. They showed that slow
376: reconnection occurs only in the case of single fluid resistive
377: MHD (in which there is no
378: distinction in the electron-ion dynamics). However, in two-fluid MHD or
379: PIC simulation it is possible to switch off the Hall term by setting $m_i / m_e = 1$
380: as this will make electron-ion dynamics indistinguishable. The result is given by
381: the dotted line in Fig. (\ref{fig:mass_depend}). It can be clearly seen that
382: the amount of reconnected flux grows very slowly with time, indicating that
383: the reconnection is slow, as expected.
384:
385: We propose the following conjecture to explain why the reconnection is fast when
386: the Hall term is included. Inclusion of the latter means that
387: in the reconnection inflow magnetic field is frozen into {\it electron} fluid.
388: As it was previously shown in \citet{th07} (see their Figs.(7) and (11))
389: speed of electrons, during the reconnection peak time, is
390: at least 4-5 times greater than that of ions. This means that electrons can
391: bring in / take out the magnetic field attached to them into / away from
392: the diffusion region
393: much faster than in the case of single fluid MHD which
394: does not distinguish between
395: electron-ion dynamics. In fact, in Fig.(2) the amount of
396: reconnected flux attained by $ \omega_{ci} t=25$
397: in the cases of $m_i / m_e \gg 1$ and $m_i / m_e = 1$
398: has the same ratio ($0.11/0.03 \approx 4$) as is
399: the ratio of electron and ion speeds ($\approx 4-5$).
400:
401: It should be mentioned that although the importance of the
402: Hall term for providing the fast reconnection has been firmly established,
403: some results indicate \cite{bb07,dk07,hz07} that the fast reconnection
404: without it is still possible. This indicates that the issue of
405: which physical factor(s) uniquely guarantee the fast rate is still open.
406:
407:
408: \subsection{Energy budget of the reconnection process}
409:
410: The next question we address is: what is the exact energy
411: budget of the reconnection process, i.e. in which proportion initial magnetic
412: energy is converted into other forms of energy?
413:
414: \begin{figure}
415: \includegraphics[scale = 0.5]{fig3.eps}
416: \caption{\label{fig:conv_a120}
417: Time evolution of (a) magnetic field energy,
418: (b) electric field energy, relativistic (c) electron and
419: (d) ion kinetic energies of the whole system for $\alpha = 1.20$.
420: These energies are normalised by the initial total energy.
421: Time is normalised by the electron plasma frequency $\omega_{pe}$.}
422: \end{figure}
423:
424: \begin{figure}
425: \includegraphics[scale = 0.5]{fig4.eps}
426: \caption{\label{fig:conv_a224}
427: The same as in Fig.~\ref{fig:conv_a120}, but for $\alpha = 2.24$.}
428: \end{figure}
429:
430: Figure~\ref{fig:conv_a120} shows time evolution
431: of (a) magnetic field energy, (b) electric field energy,
432: relativistic (c) electron and (d) ion kinetic energies of
433: the whole system for $\alpha = 1.20$.
434: According to the previous results \cite{th07}, in this case
435: the normalised reconnection rate peaks at $E_z = 0.11$ at time $ \omega_{pe} t = 170$.
436: Initially magnetic field energy is dominant, which constitutes $96$\%
437: of the total energy of system.
438: The rest $4$\% goes to
439: the initial electron and ion kinetic energies
440: because we impose a non-zero current $j_z$ at $t=0$
441: according to Eq.(\ref{eq:init_j}) (as $\alpha > 1$).
442: We gather from Fig.~\ref{fig:conv_a120}(a) that as the reconnection proceeds
443: magnetic field energy is converted into other forms of energy.
444: As it is also stated in \citet{th07} at $\omega_{pe} t=250$, which corresponds to
445: about 1.25 Alfv\'en times, $(0.96-0.94)/0.96 = 2$ \% of the initial magnetic energy is released.
446: Here we explore partition into which other forms this energy goes into.
447: Panels (b)-(d) in Fig.~\ref{fig:conv_a120} show that all other forms of energy increase
448: as time progresses. In particular, electric field energy that starts from
449: zero, attains value of 0.0024, i.e. $(0-0.0024)/(0.96-0.94)=12$\% of the consumed magnetic energy.
450: One can conjecture that ultimately this energy will go into particle kinetic energy
451: (as particles would be easily accelerated by electric fields).
452: Relativistic kinetic energy of electrons attains $(0.0205-0.0145)/(0.96-0.94)=30$\%,
453: while the same for ions $(0.037-0.0255)/(0.96-0.94)=58$\%.
454: Given that electrons have small inertia and thus
455: are more influenced by the electric field, we conjecture that within few Alfv\'en times
456: electron-ion kinetic energy partition (as the percentage of consumed magnetic energy)
457: will be roughly 40\% - 60\%.
458: It should be mentioned that, in general, it is not easy for an electric
459: field to accelerate particles, unless it is
460: parallel to the magnetic field or the magnetic field is small, such that the
461: particles are non-adiabatic (loosely tied to magnetic field lines). The
462: latter condition is more stringent for electrons.
463: Movies of electric field in our numerical simulation
464: show complicated, and yet coherent, oscillatory patterns.
465: We have not performed detailed analysis of identification of nature of these
466: waves, but based on previous experience, the case without the
467: guide field considered here,
468: would support excitation of whistler waves as the X-point collapses.
469: In turn, we conjecture that the whistler waves are ultimately responsible for
470: the particle acceleration. More detailed analysis of this is needed in the future.
471: Returning to the issue of the established 40\% - 60\% energy partition, \citet{emslie04} showed that the energy
472: of accelerated electrons is comparable to that of accelerated ions. However,
473: they admit to large uncertainties in the ion energy spectrum. Despite of this,
474: our simulation results broadly agree with the solar flare observations \cite{emslie04}.
475:
476: Previously we also considered strongly stressed X-point ($\alpha = 2.24$) \cite{th07}. In this case
477: in 1.25 Alfv\'en times, $(0.9-0.72)/0.9 = 20$\% of the initial magnetic energy is converted
478: into other forms of energy (this is equivalent of
479: $(0.9-0.72) = 18$\% of the initial {\it total} energy; and as we saw in the
480: weakly stressed case, the difference between the two is negligible. It is only with
481: the increase of $\alpha$ the difference between initial {\it magnetic} energy and
482: initial {\it total} energy becomes noticeable, because stronger initial currents
483: (i.e. initial kinetic energy of particles) need to be imposed according to Eq.(\ref{eq:init_j})).
484: This 18\% decrease in the magnetic energy is also corroborated in panel (a) in Fig.~\ref{fig:conv_a224}.
485: Exact break down (partition) of the latter is as follows (based on panels (b)-(d)):
486: electric field energy that starts from
487: zero, peaks and then settles at 0.006, i.e. $(0-0.006)/(0.9-0.72)=3$\% of the consumed magnetic energy.
488: Relativistic kinetic energy of electrons attains $(0.075-0.002)/(0.9-0.72)=41$\%,
489: while the same for ions $(0.2-0.1)/(0.9-0.72)=56$\%.
490: As in the weakly stressed case, within 1.25 Alfv\'en times,
491: electron-ion kinetic energy partition (as the percentage of total energy,
492: which for a solar flare would be the total energy released by flare)
493: is roughly 40\% - 60\%. This again is in accord to solar flare observations \citet{emslie04}.
494:
495: \subsection{Properties of velocity distribution of the accelerated particles}
496:
497:
498: The final question we address is: are there any anisotropies in the velocity distribution of the
499: accelerated particles? This question naturally comes to one's mind due to a recent study of
500: \citet{kontar06}, who surprisingly found
501: near-isotropic electron distributions in solar flares, which contrast
502: strongly with the expectations from the standard model
503: that invokes strong downward beaming, including the collisional thick-target model.
504:
505: \begin{figure}
506: \includegraphics[scale = 0.5]{fig5.eps}
507: \caption{\label{fig:vd_a120}
508: (a-c) Electron and (d-f) ion velocity distribution
509: functions in $x$, $y$ and $z$ directions near the
510: current sheet at the initial stage $t = 0$ (dashed line),
511: the peak reconnection stage $t = 170$ (dotted line) and
512: the final simulation time $t = 250$ (solid line) for $\alpha = 1.20$.
513: As in Ref.\cite{th07}, here data are
514: produced using the region of the current sheet
515: ($-2 (c/ \omega_{pe}) \leq x \leq 2(c/ \omega_{pe}), -8(c/ \omega_{pe}) \leq y \leq 8(c/ \omega_{pe})$).
516: $f_e$ and $f_i$ are the number of electrons and ions, respectively.
517: Velocity and time are normalised by light speed $c$
518: and $\omega_{pe}$, respectively.}
519: \end{figure}
520:
521: \begin{figure}
522: \includegraphics[scale = 0.5]{fig6.eps}
523: \caption{\label{fig:vd_a224}
524: (a-c) Electron and (d-f) ion velocity distribution
525: functions in $x$, $y$ and $z$ directions near the
526: current sheet at the initial stage $t = 0$ (dashed line),
527: the peak reconnection time $t = 45$ (dotted line) and
528: the final simulation time $t = 250$ (solid line) for $\alpha = 2.24$.
529: Data are produced using the region of the current sheet
530: ($-1 (c/ \omega_{pe}) \leq x \leq 1 (c/ \omega_{pe}), -16 (c/ \omega_{pe})\leq y \leq 16 (c/ \omega_{pe})$).
531: The normalisation is same as in Fig~\ref{fig:vd_a120}.
532: }
533: \end{figure}
534:
535: As in \citet{th07}, here we consider two cases of the weakly and strongly stressed X-point.
536: The results are shown in Figures \ref{fig:vd_a120} and \ref{fig:vd_a224}. The following observations can be
537: made:
538:
539: In the weakly stressed case, for electrons we see appearance of
540: super-thermal electrons towards the end of simulation time (shortly
541: after the peak of time-transient reconnection) mostly in $y$ and $z$ velocity distribution function
542: components.
543: Dynamics of the flows and currents is presented in detail in Ref.\cite{th07}.
544: Here we only mention that the reconnection inflow is in $x$ direction, while the
545: outflow is in $y$ direction. Thus based on panels (a)-(c) in Fig.(\ref{fig:vd_a120}) we gather
546: that accelerated electrons (focus on solid and dotted curves) are due to reconnection outflow (in reconnection plane)
547: as well as out-of-plane flow (which is triggered by the out-of-plane electric
548: field generated at the magnetic null). For ions, at later
549: stages of the reconnection, in panels (d)-(f) in
550: Fig.(\ref{fig:vd_a120}) we see (focus on solid and dotted curves) a
551: superposition of two Maxwellian distributions in both reconnection
552: inflow (along $x$) and outflow (along $y$). These seem to be created
553: by reconnection flow dynamics. In $z$-direction we see a shifted
554: (also somewhat broadened by the heating) Maxwellian, which is due
555: to out-of-plane ion beam (localized current).
556:
557: We gather from panels (a)-(b) in Fig.(\ref{fig:vd_a224}) (focus now only on solid curves) that in the
558: strongly stressed X-point case, in about one Alfv\'en time, super-thermal
559: electrons show a full isotropy in all three spatial directions of
560: the velocity distribution. In solar flare observations \citet{kontar06} report that
561: electron distributions are also nearly isotropic, which seems to contradict to
562: what is expected from the standard model flare models that
563: invoke strong downward beaming of electrons. In this respect,
564: the match of our simulation results with the flare observations
565: seem encouraging, in that stressed X-point collapse seems to be a
566: viable mechanism acting during solar flares. For ions (panels (d)-(f)
567: in Fig.(\ref{fig:vd_a224})) behaviour is not so much different from
568: the weakly stressed case (panels (d)-(f) in Fig.(\ref{fig:vd_a120})),
569: except for much higher velocities attained and distribution functions
570: modified by kinetic, wave-particle interaction instabilities.
571: The latter can be judged by sign changes in the slope of the distribution
572: function, which can only occur when waves and particles exchange
573: energy and momentum.
574:
575:
576: \section{Conclusions}
577:
578: By and large, the present work closes our initial study of
579: stressed X-point collapse in the collisionless regime started in
580: Ref.\cite{th07}, by bridging gaps in the understanding of key physical
581: aspects. The main findings can be listed as following:
582:
583: (i) despite significant differences of the initial setup between
584: tearing unstable Harris current sheet \cite{pritchett01} and stressed
585: X-point considered here, in both cases source of the reconnection
586: out-of-plane electric field at the magnetic null is provided by
587: off-diagonal terms of the electron pressure tensor.
588:
589: (ii) we find that when $m_i / m_e \gg 1$ reconnection rate is independent of the
590: ion-electron mass ratio and it is fast, which is also witnessed by \citet{hesse99}.
591: However, when electron-ion mass ratio is unity, i.e. the Hall term
592: is switched off, we show that reconnection rate is indeed slow.
593: This broadly agrees with the results of \citet{birn01} (However see also Refs.\cite{bb07,dk07,hz07}
594: for alternative view).
595: When the Hall physics is included, we also conjecture that the reconnection is fast
596: because the magntic field (being frozen into electron fluid, which
597: moves significantly faster than ion fluid, as shown in Ref.\cite{th07}) is
598: transported in and out of the
599: diffusion region much faster than in the case of single fluid resistive MHD.
600: We show that the amount of
601: reconnected flux attained by $ \omega_{ci} t=25$
602: in the cases of $m_i / m_e \gg 1$ and $m_i / m_e = 1$
603: has the same ratio ($\approx 4$) as is
604: the ratio of electron and ion speeds ($\approx 4-5$).
605:
606: (iii) we find that within one Alfv\'en time, roughly $\sim 40$\% of
607: the
608: initial total energy (which is mostly stored in the magnetic field)
609: is converted into the kinetic energy electrons, and somewhat
610: more than half ($\sim 60$\%) into kinetic energy of ions.
611: In solar flare observations a similar behaviour is seen \cite{emslie04}.
612:
613: (iv) When X-point is stressed {\it strongly}, in about one Alfv\'en time, a full
614: isotropy in all three spatial directions of the velocity distribution is
615: seen for super-thermal electrons. Again similar behaviour is reported
616: in solar flare observations \cite{kontar06}.
617:
618: Resuming aforesaid, it seems that collisionless, stressed X-point
619: collapse is a viable mechanism for solar flares. Also, its behaviour
620: is remarkably similar to tearing unstable Harris current sheet which
621: is thought to be more relevant for the Earth geomagnetic tail and
622: generally to magnetospheric applications.
623:
624: We close this paper with some words of caution:
625:
626: (i) realising that
627: size of the numerically simulated area $40 \times 40$
628: electron skin depths ($c / \omega_{pe}=0.16804$ m in solar corona), i.e.
629: $6.7\times 6.7$ m is a tiny proportion of the actual solar flare site
630: which can be tens of Mm. Thus, {\it a priori} it is
631: not clear whether this mechanism is suitable
632: to accelerate enough electrons in a much larger volume.
633:
634: (ii) whether e.g. the obtained electron-ion kinetic energy partition
635: (as the percentage of total released energy)
636: being roughly 40\% - 60\%, would hold for a different mass ratio?
637: We remind the reader that except in subsection III.B
638: the ion-electron mass ratio was kept
639: constant at $m_i/m_e=100$. We can only conjecture that
640: this 40\% - 60\% partition
641: holds for a different mass ratio,
642: because we also found that the amount of reconnected flux versus time
643: graph (Fig.~2) showed no mass ratio dependence.
644:
645: \begin{acknowledgments}
646: This research was supported by the United Kingdom's Science and
647: Technology Facilities Council (STFC).
648: \end{acknowledgments}
649:
650:
651:
652: %%%%%%%%%%%%%%%\bibliography{th}
653:
654: \begin{thebibliography}{14}
655: \expandafter\ifx\csname natexlab\endcsname\relax\def\natexlab#1{#1}\fi
656: \expandafter\ifx\csname bibnamefont\endcsname\relax
657: \def\bibnamefont#1{#1}\fi
658: \expandafter\ifx\csname bibfnamefont\endcsname\relax
659: \def\bibfnamefont#1{#1}\fi
660: \expandafter\ifx\csname citenamefont\endcsname\relax
661: \def\citenamefont#1{#1}\fi
662: \expandafter\ifx\csname url\endcsname\relax
663: \def\url#1{\texttt{#1}}\fi
664: \expandafter\ifx\csname urlprefix\endcsname\relax\def\urlprefix{URL }\fi
665: \providecommand{\bibinfo}[2]{#2}
666: \providecommand{\eprint}[2][]{\url{#2}}
667:
668: \bibitem[{\citenamefont{{Biskamp}}(2005)}]{biskamp}
669: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Biskamp}}},
670: \emph{\bibinfo{title}{{Magnetic reconnection in Plasmas}}}
671: (\bibinfo{publisher}{Cambridge University Press}, \bibinfo{year}{2005}).
672:
673: \bibitem[{\citenamefont{{Huba}}(2004)}]{NRLF}
674: \bibinfo{author}{\bibfnamefont{J.~D.} \bibnamefont{{Huba}}},
675: \emph{\bibinfo{title}{{NRL plasma formulary}}} (\bibinfo{publisher}{Naval
676: Research Laboratory, Washington DC 20375}, \bibinfo{year}{2004}).
677:
678: \bibitem[{\citenamefont{{Birn} and {Priest}}(2007)}]{bp07}
679: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Birn}}} \bibnamefont{and}
680: \bibinfo{author}{\bibfnamefont{E.~R.} \bibnamefont{{Priest}}},
681: \emph{\bibinfo{title}{{Reconnection of magnetic fields: magnetohydrodynamics
682: and collisionless theory and observations}}} (\bibinfo{publisher}{Cambridge
683: University Press}, \bibinfo{year}{2007}).
684:
685: \bibitem[{\citenamefont{{Dungey}}(1953)}]{d53}
686: \bibinfo{author}{\bibfnamefont{J.~W.} \bibnamefont{{Dungey}}},
687: \bibinfo{journal}{Phil. \ Mag.} \textbf{\bibinfo{volume}{44}},
688: \bibinfo{pages}{725} (\bibinfo{year}{1953}).
689:
690: \bibitem[{\citenamefont{{Priest} and {Forbes}}(2000)}]{pf00}
691: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{{Priest}}} \bibnamefont{and}
692: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{{Forbes}}},
693: \emph{\bibinfo{title}{{Magnetic reconnection: MHD theory and applications}}}
694: (\bibinfo{publisher}{Cambridge University Press}, \bibinfo{year}{2000}).
695:
696: \bibitem[{\citenamefont{{Tsiklauri} and {Haruki}}(2007)}]{th07}
697: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{{Tsiklauri}}} \bibnamefont{and}
698: \bibinfo{author}{\bibfnamefont{T.}~\bibnamefont{{Haruki}}},
699: \bibinfo{journal}{Physics of Plasmas} \textbf{\bibinfo{volume}{14}},
700: \bibinfo{pages}{112905} (\bibinfo{year}{2007}).
701:
702: \bibitem[{\citenamefont{{Pritchett}}(2001)}]{pritchett01}
703: \bibinfo{author}{\bibfnamefont{P.~L.} \bibnamefont{{Pritchett}}},
704: \bibinfo{journal}{J. \ Geophys. \ Res.} \textbf{\bibinfo{volume}{106}},
705: \bibinfo{pages}{3783} (\bibinfo{year}{2001}).
706:
707: \bibitem[{\citenamefont{{Hesse} et~al.}(1999)\citenamefont{{Hesse},
708: {Schindler}, {Birn}, and {Kuznetsova}}}]{hesse99}
709: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Hesse}}},
710: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{{Schindler}}},
711: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Birn}}}, \bibnamefont{and}
712: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Kuznetsova}}},
713: \bibinfo{journal}{Physics of Plasmas} \textbf{\bibinfo{volume}{6}},
714: \bibinfo{pages}{1781} (\bibinfo{year}{1999}).
715:
716: \bibitem[{\citenamefont{{Birn} et~al.}(2001)\citenamefont{{Birn}, {Drake},
717: {Shay}, {Rogers}, {Denton}, {Hesse}, {Kuznetsova}, {Ma}, {Bhattacharjee},
718: {Otto} }}]{birn01}
719: \bibinfo{author}{\bibfnamefont{J.}~\bibnamefont{{Birn}}},
720: \bibinfo{author}{\bibfnamefont{J.~F.} \bibnamefont{{Drake}}},
721: \bibinfo{author}{\bibfnamefont{M.~A.} \bibnamefont{{Shay}}},
722: \bibinfo{author}{\bibfnamefont{B.~N.} \bibnamefont{{Rogers}}},
723: \bibinfo{author}{\bibfnamefont{R.~E.} \bibnamefont{{Denton}}},
724: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Hesse}}},
725: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Kuznetsova}}},
726: \bibinfo{author}{\bibfnamefont{Z.~W.} \bibnamefont{{Ma}}},
727: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Bhattacharjee}}},
728: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Otto}}},
729: \bibinfo{journal}{J. \ Geophys. \ Res.}
730: \textbf{\bibinfo{volume}{106}}, \bibinfo{pages}{3715} (\bibinfo{year}{2001}).
731:
732: \bibitem[{\citenamefont{{Bessho} and {Bhattacharjee}}(2007)}]{bb07}
733: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{{Bessho}}} \bibnamefont{and}
734: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Bhattacharjee}}},
735: \bibinfo{journal}{Physics of Plasmas} \textbf{\bibinfo{volume}{14}},
736: \bibinfo{pages}{056503} (\bibinfo{year}{2007}).
737:
738: \bibitem[{\citenamefont{{Daughton} and {Karimabadi}}(2007)}]{dk07}
739: \bibinfo{author}{\bibfnamefont{W.}~\bibnamefont{{Daughton}}} \bibnamefont{and}
740: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{{Karimabadi}}},
741: \bibinfo{journal}{Physics of Plasmas} \textbf{\bibinfo{volume}{14}},
742: \bibinfo{pages}{072303} (\bibinfo{year}{2007}).
743:
744: \bibitem[{\citenamefont{{Hesse} and {Zenitani}}(2007)}]{hz07}
745: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{{Hesse}}} \bibnamefont{and}
746: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{{Zenitani}}},
747: \bibinfo{journal}{Physics of Plasmas} \textbf{\bibinfo{volume}{14}},
748: \bibinfo{pages}{112102} (\bibinfo{year}{2007}).
749:
750: \bibitem[{\citenamefont{{Emslie} et~al.}(2004)\citenamefont{{Emslie},
751: {Kucharek}, {Dennis}, {Gopalswamy}, {Holman}, {Share}, {Vourlidas}, {Forbes},
752: {Gallagher}, {Mason} }}]{emslie04}
753: \bibinfo{author}{\bibfnamefont{A.~G.} \bibnamefont{{Emslie}}},
754: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{{Kucharek}}},
755: \bibinfo{author}{\bibfnamefont{B.~R.} \bibnamefont{{Dennis}}},
756: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{{Gopalswamy}}},
757: \bibinfo{author}{\bibfnamefont{G.~D.} \bibnamefont{{Holman}}},
758: \bibinfo{author}{\bibfnamefont{G.~H.} \bibnamefont{{Share}}},
759: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{{Vourlidas}}},
760: \bibinfo{author}{\bibfnamefont{T.~G.} \bibnamefont{{Forbes}}},
761: \bibinfo{author}{\bibfnamefont{P.~T.} \bibnamefont{{Gallagher}}},
762: \bibinfo{author}{\bibfnamefont{G.~M.} \bibnamefont{{Mason}}},
763: \bibinfo{journal}{Journal of Geophysical Research
764: (Space Physics)} \textbf{\bibinfo{volume}{109}}, \bibinfo{pages}{10104}
765: (\bibinfo{year}{2004}).
766:
767: \bibitem[{\citenamefont{{Kontar} and {Brown}}(2006)}]{kontar06}
768: \bibinfo{author}{\bibfnamefont{E.~P.} \bibnamefont{{Kontar}}} \bibnamefont{and}
769: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{{Brown}}},
770: \bibinfo{journal}{Astrophys. \ J. \ Lett.} \textbf{\bibinfo{volume}{653}},
771: \bibinfo{pages}{L149} (\bibinfo{year}{2006}).
772:
773: \end{thebibliography}
774:
775:
776: \end{document}
777:
778: