1: %&latex
2: %&latex
3: \documentclass[namedreferences]{SolarPhysics}
4: \usepackage[optionalrh]{spr-sola-addons} % For Solar Physics
5: %\usepackage{epsfig} % For eps figures, old commands
6: \usepackage{graphicx} % For eps figures, newer & more powerfull
7: %\usepackage{courier} % Change the \texttt command to courier style
8: %\usepackage{natbib} % For citations: redefine \cite commands
9: %\usepackage{amssymb} % useful mathematical symbols
10: \usepackage{color} % For color text: \color command
11: \usepackage{url} % For breaking URLs easily trough lines
12: \def\UrlFont{\sf} % define the fonts for the URLs
13:
14:
15: % General definitions
16: % please place your own definitions here and don't use \def but
17: % \newcommand{}{} or
18: % \renewcommand{}{} if it is already defined in LaTeX
19:
20: \newcommand{\BibTeX}{\textsc{Bib}\TeX}
21: \newcommand{\etal}{{\it et al.}}
22:
23: % Definitions for equations
24: \newcommand{\deriv}[2]{\frac{{\mathrm d} #1}{{\mathrm d} #2}}
25: \newcommand{\rmd}{ {\ \mathrm d} }
26: \renewcommand{\vec}[1]{ {\mathbf #1} }
27: \newcommand{\uvec}[1]{ \hat{\mathbf #1} }
28: \newcommand{\pder}[2]{ \f{\partial #1}{\partial #2} }
29: \newcommand{\grad}{ {\bf \nabla } }
30: \newcommand{\curl}{ {\bf \nabla} \times}
31: \newcommand{\vol}{ {\mathcal V} }
32: \newcommand{\bndry}{ {\mathcal S} }
33: \newcommand{\dv}{~{\mathrm d}^3 x}
34: \newcommand{\da}{~{\mathrm d}^2 x}
35: \newcommand{\dl}{~{\mathrm d} l}
36: \newcommand{\dt}{~{\mathrm d}t}
37: \newcommand{\intv}{\int_{\vol}^{}}
38: \newcommand{\inta}{\int_{\bndry}^{}}
39: \newcommand{\avec}{ \vec A}
40: \newcommand{\ap}{ \vec A_p}
41: \newcommand{\bb}{ \vec B}
42: \newcommand{\jj}{ \vec j}
43: \newcommand{\rr}{ \vec r}
44: \newcommand{\xx}{ \vec x}
45:
46: % Definitions for the journal names
47: \newcommand{\adv}{ {\it Adv. Spa. Res.}}
48: \newcommand{\annG}{ {\it Annales Geophysicae}}
49: \newcommand{\aap}{ {\it Astron. Astrophys.}}
50: \newcommand{\aaps}{ {\it Astron. Astrophys. Suppl.}}
51: \newcommand{\aapr}{ {\it Astron. Astrophys. Rev.}}
52: \newcommand{\ag}{ {\it Ann. Geophys.}}
53: \newcommand{\aj}{ {\it Astronom. J.}}
54: \newcommand{\apj}{ {\it Astrophys. J.}}
55: \newcommand{\apjl}{ {\it Astrophys. J. Lett.}}
56: \newcommand{\apss}{ {\it Astrophys. Spa. Sci.}}
57: \newcommand{\cjaa}{ {\it Chinese J. Astron. Astrophys.}}
58: \newcommand{\gafd}{ {\it Geophys. Astrophys. Fluid Dyn.}}
59: \newcommand{\grl}{ {\it Geophys. Res. Lett.}}
60: \newcommand{\ijga}{ {\it Int. J. Geomag. Aeron.}}
61: \newcommand{\jastp}{ {\it J. Atmos. Sol. Terr. Phys.}}
62: \newcommand{\jgr}{ {\it J. Geophys. Res.}}
63: \newcommand{\mnras}{ {\it Mon. Not. Roy. Astron. Soc.}}
64: \newcommand{\nat}{ {\it Nature}}
65: \newcommand{\pasp}{ {\it Pub. Astron. Soc. Pac.}}
66: \newcommand{\pasj}{ {\it Pub. Astron. Soc. Japan}}
67: \newcommand{\pre}{ {\it Phys. Rev. E}}
68: \newcommand{\solphys}{{\it Solar Phys.}}
69: \newcommand{\sovast}{ {\it Sov. Astronom.}}
70: \newcommand{\ssr}{ {\it Space Sci. Rev.}}
71:
72:
73: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
74: \begin{document}
75:
76: \begin{article}
77:
78: \begin{opening}
79:
80: \title{Resonantly Damped Surface and Body MHD Waves in a Solar Coronal Slab with Oblique
81: Propagation}
82:
83: \author{I.~\surname{Arregui}$^{1}$\sep
84: J.~\surname{Terradas}$^{1,2}$\sep
85: R.~\surname{Oliver}$^{1}$\sep
86: J.~L.~\surname{Ballester}$^{1}$
87: }
88: \runningauthor{Arregui {\it et al.}}
89: \runningtitle{Resonantly Damped Surface and Body MHD Waves}
90:
91: \institute{$^{1}$ Departament de F\'{\i}sica, Universitat de les Illes Balears, E-07122
92: Palma de Mallorca, Spain
93: email: \url{inigo.arregui@uib.es}\\
94: $^{2}$ Centrum voor Plasma Astrofysica, KULeuven,
95: Celestijnenlaan 200B, 3001 Heverlee, Belgium
96: \\
97: }
98:
99: \begin{abstract}
100: The theory of magnetohydrodynamic (MHD) waves in solar coronal slabs in a zero-$\beta$
101: configuration and for parallel propagation of waves does not allow the existence of surface waves.
102: When oblique propagation of perturbations is considered both surface and body waves are
103: able to propagate. When the perpendicular wave number is larger than a certain value, the body kink
104: mode becomes a surface wave. In addition, a sausage surface mode is found below the internal cut-off
105: frequency. When non-uniformity in the equilibrium is included, surface and body modes are damped due
106: to resonant absorption. In this paper, first, a normal-mode analysis is performed and the period,
107: the damping rate, and the spatial structure of eigenfunctions are obtained. Then, the time-dependent
108: problem is solved, and the conditions under which one or the other type of mode is excited are
109: investigated.
110: \end{abstract}
111: \keywords{Magnetohydrodynamics; Waves, Propagation; Coronal Seismology}
112: \end{opening}
113: %-------------------------------------------------
114:
115: \section{Introduction}\label{introduction}
116:
117: Coronal seismology, such as first suggested by Uchida (1970) and Roberts, Edwin, and Benz (1984),
118: aims to determine unknown physical parameters in the solar corona by the combination of
119: observed and theoretical properties of waves and oscillations.
120: Flare-generated transverse coronal-loop oscillations have attracted particular interest,
121: since their first unambiguous detection with high resolution instruments on-board spacecraft,
122: such as SOHO and TRACE (Aschwanden {\it et al.} 1999; Nakariakov {\it et al.} 1999; Aschwanden {\it et al.} 2002;
123: Schrijver, Aschwanden, and Title, 2002). The observed motions have been interpreted in terms of fast
124: MHD kink modes of a cylindrical flux tube in their fundamental harmonic (Nakariakov {\it et al.} 1999),
125: opening the way of coronal seismology. Recent examples of the application of coronal seismology can be
126: found in Nakariakov and Ofman (2001); Goossens, Andries, and Aschwanden (2002); Aschwanden {et al.} (2003);
127: Andries, Arregui, and Goossens (2005); Verwichte, Foullon, and Nakariakov (2006c); Arregui {\it et al.} (2007b).
128:
129: The basic theoretical framework for MHD wave propagation in structured media was developed
130: well in advance of the observational evidence for oscillations. For example, Edwin and
131: Roberts (1982) studied wave propagation in coronal loops modelled as straight magnetic slabs.
132: Straight cylindrical flux tube models were used by Spruit (1981); Edwin and Roberts (1983);
133: Roberts (1983). More recent models have explored other effects, such as
134: the curvature of coronal loops (Van Doorsselaere {\it et al.} 2004; Brady and Arber, 2005;
135: D\'{\i}az, 2006; D\'{\i}az, Zaqarashvili, and Roberts, 2006; Terradas, Oliver, and Ballester, 2006;
136: Verwichte, Foullon, and Nakariakov, 2006a, b);
137: the non-circularity of their cross-sections (Ruderman, 2003) or the influence of a longitudinally-varying
138: density (Andries {\it et al.} 2005; Andries, Arregui, and Goossens 2005; Arregui {\it et al.} 2005;
139: McEwan {\it et al.} 2006; Dymova and Ruderman 2006). Recent comprehensive reviews on these and
140: other studies can be found in Nakariakov and Verwichte (2005) and Aschwanden (2006).
141:
142:
143: Among all these studies, surface waves have been the subject of considerable attention.
144: Jain and Roberts (1996) studied the properties of magneto-acoustic surface waves at a single
145: interface, for the case of non-parallel propagation, in the context of {\it f}-mode oscillations.
146: In a different context Zhelyazkov, Murawski, and Goossens (1996) considered the propagation of
147: magneto-acoustic surface waves with both non-parallel propagation and a sheared magnetic field.
148: Similar studies of surface waves can be found in Roberts (1981a) and Miles and Roberts (1989).
149: By surface wave we mean a wave with an evanescent tail that propagates on a sharp (discontinuous)
150: interface, although they can also propagate on a smoothly varying interface
151: (see Lee and Roberts, 1986; Hollweg, 1990a, b, 1991; Goossens, 1991, for example). Surface waves are
152: expected to occur in the solar atmosphere at places where physical parameters change abruptly.
153: Their propagation is directional and guided by the interface and their energy is confined to
154: within roughly a wavelength of the surface. Surface waves may also arise in more complex structures
155: than a single interface, such as slabs and flux tubes. However, the existence of surface waves
156: in a zero-$\beta$ equilibrium requires the presence of propagation in the direction perpendicular
157: to the magnetic field (Roberts, 1991).
158:
159: In this paper, damped coronal-loop oscillations (Aschwanden {\it et al.} 1999; Nakariakov {\it et al.} 1999)
160: are considered. We adopt a line-tied slab model for a coronal loop and study the MHD
161: wave properties, for oblique propagation of waves, by solving the normal-mode
162: problem as well as the time-dependent problem.
163: The non-uniformity of the equilibrium produces the resonant damping of these oscillations
164: (Goossens, Andries, and Arregui, 2006). Although the model adopted in this work is an oversimplification,
165: and is far from being an accurate representation of real coronal loops, the simplicity of the model
166: allows a detailed study of the problem and a comparison with the results in a cylindrical loop model.
167:
168:
169: The layout of the paper is as follows. In Section~\ref{eq}, the equilibrium configuration and the
170: basic MHD equations governing resonantly-coupled fast and Alfv\'en modes are presented. Next, in
171: Section~\ref{normal}, the normal mode properties are described. In Section~\ref{timesect}, we analyse
172: the temporal evolution of the perturbations after given initial disturbances. Finally, in
173: Section~\ref{conclusions}, our conclusions are drawn.
174:
175:
176: \section{Equilibrium Configuration, Linear MHD Waves, and Numerical Method}\label{eq}
177:
178: We model the equilibrium magnetic and plasma configuration of a solar coronal loop by means
179: of a one-dimensional, line-tied, over-dense slab in Cartesian geometry.
180: The magnetic field is straight and pointing in the $z$-direction, ${\bf B}=B\ \hat{\bf e}_z$.
181: For applications to the solar corona it is a good approximation to consider that
182: the magnetic pressure dominates over the gas
183: pressure. This classic zero plasma-$\beta$ limit
184: implies that the magnetic field is uniform and
185: that the density [$\rho(x)$] or Alfv\'en
186: speed [$v_{A}(x)$] profiles can be chosen arbitrarily.
187: The coronal slab is then modelled
188: using a varying equilibrium density
189: profile in the $x$-direction (see Figure~\ref{model}), by means of a density enhancement
190: of half-width $a$ centred about $x=0$. The
191: density inside the slab is constant [$\rho_i$] and
192: connected to the constant coronal environment, with density $\rho_e$, by transitional
193: non-uniform layers of thickness $l$. The explicit expression for the
194: equilibrium density considered is
195:
196: \begin{equation}
197: \rho(x)=\left\{\begin{array}{ll}
198: \rho_i \hspace{2.8cm} \textrm{$0\leq x \leq a-\frac{l}{2}$},\\\\
199: f(x) \hspace{1.55cm} \textrm{$a-\frac{l}{2}\leq x \leq a+\frac{l}{2}$},\\\\
200: \rho_e \hspace{3.5cm} \textrm{$x\geq a+\frac{l}{2}$},\\\\
201: \end{array}\right\}\label{profile2slabs}
202: \end{equation}
203:
204: \noindent
205: with $\rho(-x)=\rho(x)$.
206: The density profiles at the non-uniform transitional layers
207: have been chosen following Ruderman and Roberts (2002); Van Doorsselaere {\it et al.} (2004)
208: and are given by
209:
210: \begin{eqnarray}
211: f(x)&=&\frac{\rho_i}{2}\left[\left(1+\frac{\rho_e}{\rho_i}\right)-\left(1-\frac{\rho_e}{\rho_i}
212: \right)\sin{\frac{\pi\left(x-a\right)}{l}}\right].\\\nonumber
213: \end{eqnarray}
214:
215:
216: \noindent
217: We remark that the particular form of the density profile at the non-uniform layers has no
218: importance by itself for the results obtained.
219:
220:
221: In order to study small-amplitude oscillations of the previous equilibrium, the linear
222: resistive MHD equations with constant magnetic diffusivity [$\eta$] are considered.
223: The inclusion of dissipative terms in the MHD equations is needed for the computation of
224: resonantly-damped eigenmodes. Resistive dissipation is preferred over viscous dissipation
225: for the simplicity with which it enters into the equations. Both dissipation mechanisms
226: affect the thickness of the dissipation layers and the behaviour of the solutions in those layers,
227: but not the differences in the solutions across the layers, which in turn determine the damping rate of
228: the eigenmodes (Sakurai, Goossens, and Hollweg 1991).
229: As the equilibrium configuration only depends on the $x$-coordinate, a spatial dependence
230: of the form $\exp^{-i\left(k_y y+k_z z\right)}$ is assumed for all perturbed quantities,
231: with $k_y$ and $k_z$ being the perpendicular and parallel wave numbers. The photospheric line-tying effect
232: is then included by selecting the appropriate parallel wavenumber. This leads to the following set of
233: differential equations for the two components of the velocity perturbation [$v_x$ and $v_y$] and
234: the three components of the perturbed magnetic field [$b_{1x}$, $b_{1y}$, and $b_{1z}$]
235:
236: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
237:
238: \begin{figure}[!t]
239: \begin{center}
240: \includegraphics[width=7cm,angle=90]{fig1.ps}
241: \end{center}
242: \caption{Schematic representation of the density enhancement (shaded region) of half-width
243: $a$ located at $x=0$ used to model a coronal loop. The internal region with density $\rho_i$
244: is connected to the external medium, with density $\rho_e$, by transitional non-uniform
245: layers (light-shaded regions) of thickness $l$. Non-uniform layers of thickness $l/a=0.5$ have
246: been considered. }
247: \label{model}
248: \end{figure}
249:
250: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
251:
252:
253: \begin{eqnarray}
254: \frac{\partial v_x}{\partial t} &=& \frac{B}{\mu\rho}\left(-i k_z b_{1x}-
255: \frac{\partial b_{1z}}{\partial x}\right), \label{first}\\
256: \frac{\partial v_y}{\partial t} &=& \frac{B}{\mu\rho}\left(-i k_z b_{1y}+
257: i k_y b_{1z}\right),\\
258: \frac{\partial b_{1x}}{\partial t}&=&-i B k_z v_x +
259: \eta\left[\frac{\partial^2 b_{1x}}{\partial x^2}-\left(k^2_y+k^2_z\right) b_{1x}\right],\\
260: \frac{\partial b_{1y}}{\partial t}&=&-i B k_z v_y +\eta\left[\frac{\partial^2 b_{1y}}
261: {\partial x^2}-\left(k^2_y+k^2_z\right) b_{1y}\right],\\
262: \frac{\partial b_{1z}}{\partial t}&=&-B\left(\frac{\partial v_x}{\partial x}-i k_y v_y\right)+
263: \eta\left[\frac{\partial^2 b_{1z}}{\partial x^2}-\left(k^2_y+k^2_z\right) b_{1z}\right],\label{last}
264: \end{eqnarray}
265:
266: \noindent
267: where $\mu$ is the permeability of free space.
268: When $k_y\neq0$ and $l/a\neq 0$, Equations~(\ref{first})--(\ref{last}), describe the oscillatory properties of
269: coupled fast and Alfv\'en modes. As the plasma-$\beta$ = $0$, the slow mode is absent and there are
270: no motions parallel to the equilibrium magnetic field, $v_z=0$. In this paper, solutions to
271: these equations are obtained by performing two kinds of analysis. First the normal modes of
272: oscillation of the system are studied and then the time-dependent evolution of the slab after an
273: initial disturbance is investigated. When only oblique propagation is considered ($k_y\neq0$),
274: the normal-mode solutions can be obtained by solving an analytical dispersion relation. If, in addition,
275: transitional non-uniform layers are included ($l\neq0$), the solutions to these equations are,
276: in general, difficult to obtain. For this reason, numerical approximations are obtained using
277: PDE2D (Sewell, 2005), a general-purpose partial differential equation solver.
278: As for the boundary conditions, in both the normal mode analysis
279: as well as in time-dependent simulations, we impose the vanishing of the perturbed velocity
280: far away from the loop, hence ${\bf v}\rightarrow{ 0}$ as $x\rightarrow\pm\infty$.
281:
282:
283: \section{Normal Mode Analysis}\label{normal}
284:
285: We first consider the normal modes of oscillation of our equilibrium configuration. A
286: temporal dependence of the form $e^{i\omega t}$ is assumed for all perturbed quantities in
287: Equations~(\ref{first})--(\ref{last}), with $\omega=\omega_{R}+i \omega_{I}$ being the complex
288: frequency. These equations form an eigenvalue problem. The solutions for the simplest case,
289: with $k_y=0$ and $l=0$, are well-known and described by Edwin and Roberts (1982).
290: They can be classified, according to the parity of their eigenfunctions about $x=0$, as fast kink and
291: sausage modes. The corresponding dispersion relations are
292:
293: \begin{equation}\label{kink}
294: \tanh \kappa_i a=-\frac{\kappa_e}{\kappa_i},
295: \end{equation}
296:
297: \noindent
298: for kink modes ($v_x$ even about $x=0$) and
299:
300: \begin{equation}\label{sausage}
301: \coth \kappa_i a=-\frac{\kappa_e}{\kappa_i},
302: \end{equation}
303:
304: \noindent
305: for sausage modes ($v_x$ odd about $x=0$), where
306:
307: \begin{equation}
308: \kappa^2_e=k^2_z-\frac{\omega^2}{v^2_{Ae}} \mbox{\hspace{1cm}} \mbox{and}
309: \mbox{\hspace{1cm}} \kappa^2_i=k^2_z-\frac{\omega^2}{v^2_{Ai}}.
310: \end{equation}
311:
312: \noindent
313: Here $v_{Ai,e}=B\sqrt{1/\mu\rho_{i,e}}$ are the internal and external Alfv\'en speeds.
314: Trapped waves have real frequencies whereas modes with complex frequency represent leaky waves.
315: For the modes with the lowest number of internal nodes, the body sausage mode is leaky in
316: the long wavelength limit, $k_z a$$\ll$$1$, while the kink mode is trapped for all wavelengths.
317: All other branches, kink and sausage alike, are leaky in the long-wavelength limit.
318:
319: The propagation of fast magneto-acoustic waves in a plasma density inhomogeneity with smooth
320: transversal density profiles was considered by Edwin and Roberts (1988) and Nakariakov
321: and Roberts (1995). The latter found that the simple slab profile is a good general guide to the
322: behaviour in a smooth, sharply-structured, profile.
323: Here, a similar smooth transverse density profile is
324: considered and oblique propagation of perturbations included.
325: We will restrict our analysis to trapped modes, whereas for an analysis of the
326: leaky solutions with $k_y=0$ the reader is referred to Terradas, Oliver, and Ballester (2005).
327:
328:
329: \subsection{Undamped Surface and Body Waves for Oblique Propagation}\label{undamped}
330:
331: We first consider the changes of the oscillatory properties of fast modes due to the inclusion of oblique
332: propagation of perturbations.
333: %When oblique propagation of waves is considered the resulting solutions no longer represent
334: %pure fast magneto-acoustic waves, but are coupled to Alfv\'en waves.
335: %For the fundamental kink mode of oscillation in a cylinder this is always true since for this mode,
336: %the azimuthal wave number is $m=1$ and there is always an dependence of perturbations on this
337: %direction.
338: If the slab is connected to the external medium by discontinuities ($l=0$)
339: Equations~(\ref{first})--(\ref{last}) can be combined to give a single ordinary differential
340: equation for the transverse component of the perturbed velocity of the form
341:
342: \begin{equation}
343: \frac{{\mathrm d}\ }{{\mathrm d} x}\left[\frac{\rho v^2_{A}\left(k^2_z v^2_{A}-\omega^2\right)}
344: {(k^2_y+k^2_z) v^2_{A}-\omega^2} \frac{{\mathrm d}v_x }{{\mathrm d} x} \right]-\rho\left(k^2_z v^2_{A}-\omega^2\right)v_x=0.
345: \end{equation}
346:
347: \noindent
348: Solutions to this equation can readily be obtained, for $\rho$ and $v_{A}$ constant, by following
349: the usual procedure of matching different solutions in the internal and external regions
350: (Edwin and Roberts, 1982) and demanding the evanescence of perturbations far away
351: from the slab. This leads to the following dispersion relations
352:
353: \begin{equation}\label{kinkky}
354: \tanh m_i a=-\frac{\kappa^2_e}{\kappa^2_i}\frac{m_i}{m_e},
355: \end{equation}
356:
357: \noindent
358: for kink modes and
359:
360: \begin{equation}\label{sausageky}
361: \coth m_i a=-\frac{\kappa^2_e}{\kappa^2_i}\frac{m_i}{m_e},
362: \end{equation}
363:
364: \noindent
365: for sausage modes, where
366:
367: \begin{equation}\label{ms}
368: m^2_e=\left(k^2_y+k^2_z-\frac{\omega^2}{v^2_{Ae}}\right) \mbox{\hspace{1cm}}
369: \mbox{and} \mbox{\hspace{1cm}} m^2_i=\left(k^2_y+k^2_z-\frac{\omega^2}{v^2_{Ai}}\right).
370: \end{equation}
371:
372:
373:
374: \noindent
375: Equations~(\ref{kinkky}) and (\ref{sausageky}) reduce to Equations~(\ref{kink}) and
376: (\ref{sausage}) when $k_y$ = $0$, as expected. An interesting solution of the dispersion relations
377: can be obtained in the limit of quasi-perpendicular propagation, $k_y$$\gg$$k_z$, and additionally
378: assuming that $k^2_y$$\gg$$\left(k^2_z-\omega^2/v^2_{Ai}\right)$. Then, the left hand-side of the
379: dispersion relations given by Equations~(\ref{kinkky}) and (\ref{sausageky}) can be approximated to
380:
381: \begin{equation}
382: \lbrace^{\tanh}_{\coth}\rbrace m_i a\simeq 1.
383: \end{equation}
384:
385: \noindent
386: Let us define $\omega_{k}$ as the frequency in this limit. Then,
387: the following expression for the parallel phase speed is obtained
388:
389: \begin{equation}\label{kinkspeed}
390: \frac{\omega_k}{k_z}\simeq \sqrt{\frac{\rho_i v^2_{Ai}+\rho_e v^2_{Ae}}{\rho_i+\rho_e}}\equiv c_k,
391: \end{equation}
392:
393: \noindent
394: where $c_k$ is the kink speed. This quantity arises also in the description
395: of the kink mode of oscillation of a magnetic flux tube ({\it e.g.} Spruit, 1981;
396: Ryutova, 1990; Roberts, 1991) and in the propagation of surface waves in the incompressible
397: limit (Roberts, 1981b).
398:
399: Solutions to Equations~(\ref{kinkky}) and (\ref{sausageky}) can be found by means of a
400: simple numerical program. In our numerical calculations, we consider fixed values for the
401: density contrast, $\rho_i/\rho_e=10$, and for the parallel wavenumber, $k_za=\pi/50$.
402: For the observed transverse kink-mode oscillations, with a wavelength double the length of the loop,
403: this corresponds to a ratio of length to width of $L/2a=25$, which is a typical value for observed
404: oscillating coronal loops. Figure~\ref{disperplots} displays some solutions as a
405: function of the perpendicular wave number. The dispersion relations have two frequencies at which
406: the characteristics of wave
407: motions vary. Modes with a frequency above the external cut-off frequency
408: $\left[\omega_{ce}=v_{Ae}\sqrt{k^2_y+k^2_z}\right]$ are leaky. The internal cut-off frequency
409: $\left[\omega_{ci}=v_{Ai}\sqrt{k^2_y+k^2_z}\right]$ also plays an important role, since depending on whether
410: the frequency of the eigenmode is above or below this cut-off we have a body-like or a
411: surface-like mode.
412:
413: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
414: \begin{figure*}[!t]
415:
416: \includegraphics[width=10.0cm,angle=90]{fig2.ps}
417:
418: \caption{Dispersion diagram calculated from the numerical solution of the dispersion relations,
419: Equations~(\ref{kinkky}) and (\ref{sausageky}), for a slab model of coronal loop with
420: $k_z a=\pi/50$ and density contrast $\rho_i/\rho_e=10$. The frequency of the different solutions is
421: displayed as a function of the perpendicular wave number: kink mode (solid), sausage surface
422: mode (dotted) and sausage body (dash-dotted). The horizontal long-dashed line indicates the
423: frequency $\omega_k$, given by Equation~(\ref{kinkspeed}), to which both kink and sausage
424: surface waves approach for large $k_y$. The inset plot displays a broader view of the
425: solutions in order to show the behaviour
426: of the sausage body solution. The dashed lines represent the internal and external cut-off
427: frequencies, $w_{ci}=\sqrt{k^2_y+k^2_z}v_{Ai}$ and $\omega_{ce}=\sqrt{k^2_y+k^2_z}v_{Ae}$ respectively.
428: For large $k_y$, the sausage body mode approaches asymptotically $\omega_{ci}$ never crossing below
429: its value and the kink and sausage surface solutions go to the kink frequency given by
430: Equation~(\ref{kinkspeed}).}
431: \label{disperplots}
432: \end{figure*}
433: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
434:
435: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
436: \begin{figure*}[!t]
437: \hspace{-0.6cm}
438: \vbox{\hbox{
439: \includegraphics[width=5.cm,angle=90]{fig3a.ps}
440: \includegraphics[width=5.cm,angle=90]{fig3b.ps} }
441: \hbox{
442: \includegraphics[width=5.cm,angle=90]{fig3c.ps}
443: \includegraphics[width=5.cm,angle=90]{fig3d.ps}}
444: \hbox{
445: \includegraphics[width=5.cm,angle=90]{fig3e.ps}
446: \includegraphics[width=5.cm,angle=90]{fig3f.ps}}}
447: \caption{Real part of the transverse velocity component [$v_x$] and imaginary part of the perturbed
448: total pressure [$P_T$] for the fundamental kink mode ({\em top panels}), the sausage body mode
449: ({\em middle panels}) and the sausage surface mode ({\em bottom panels}) for an equilibrium
450: configuration with $k_z a=\pi/50$, $\rho_i/\rho_e=10$ and two values of the perpendicular wave number:
451: $k_y a=0.0$ (dotted lines) and $k_ya=0.5$ (solid lines) for the top and bottom panels and $k_y a=0.2$
452: (dotted lines) and $k_ya=0.5$ (solid lines) for the middle panel. Note that the character of the
453: eigenfunction for the kink mode changes from body-like to surface-like and that for $k_y a=0.5$ the
454: sausage body and surface mode solutions have an almost identical $v_x$. In all figures,
455: the shaded region represents
456: the density enhancement.}
457: \label{eigen}
458: \end{figure*}
459:
460: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
461: The kink-mode solution has a decreasing frequency with respect to
462: $k_y$. For small values of $k_y$ its frequency is above the internal cut-off frequency
463: ($m^2_i<0$, in Equation~(\ref{kinkky})) and the mode is a body wave.
464: Beyond a given value of $k_y$ ($k_ya\sim 0.15$ in Figure~\ref{disperplots}), however, the frequency crosses the
465: internal cut-off frequency ($m^2_i>0$) and the mode becomes a surface
466: solution. In addition, as soon as $k_y\neq0$, there is also a solution with frequency near
467: the internal cut-off frequency. This solution is always below the
468: internal cut-off frequency and, therefore, corresponds to a surface wave.
469: For this solution $v_x$ has odd parity with respect to $x=0$ and so is a sausage surface wave.
470: For increasing $k_y$ the sausage surface mode
471: approaches the kink mode solution and for large $k_y$ both
472: asymptotically tend to the frequency given by Equation~(\ref{kinkspeed}).
473: Therefore, for almost perpendicular propagation there are two surface modes corresponding to the
474: two parities allowed by the symmetry of the problem. Contrary to the case $k_y=0$, the mode with
475: lowest frequency is sausage and not kink. The inset plot in Figure~\ref{disperplots} displays a
476: broader view of the solutions. The sausage body solution is leaky for small values of $k_y$, but
477: it becomes non-leaky and its frequency is that of the external cut-off frequency near $k_ya=0.2$.
478: Then, for increasing $k_y$ its frequency increases. In contrast to the kink-mode solution the
479: frequency of the sausage body mode never crosses below the internal cut-off frequency, but approaches
480: asymptotically its value for increasing $k_y$. The reason for this behaviour is that there is
481: already a solution with odd parity with respect to $x=0$ below the internal cut-off frequency.
482:
483: We obtain further information on the properties of the normal modes from the spatial distribution
484: of the eigenfunctions. Figure~\ref{eigen} shows the perturbed transverse velocity [$v_x$] and the
485: perturbed magnetic pressure [$P_T$ obtained from $b_{1z}$], for two different
486: values of the perpendicular wave number and for the fundamental kink eigenmode, the sausage body mode
487: and the sausage surface mode. Significant changes of the properties of eigenfunctions are
488: clearly visible as the value of the perpendicular wave number is increased. For $k_y=0$, the kink
489: mode has its characteristic behaviour with a maximum of $v_x$ at the internal part of the slab, while
490: $v_x$ decreases exponentially for increasing distance from the loop. On the other hand, the total
491: pressure perturbation for this mode has the usual antisymmetric profile with a zero value at the
492: centre of the slab and maxima at its edges. When oblique propagation is included ($k_y\neq0$),
493: there is an increased confinement of the eigenfunctions, compared with non-oblique propagation.
494: This result, also found by D\'{\i}az, Oliver, and Ballester (2003) in the context of prominence
495: fibril oscillations, can be seen in terms of a sharper drop-off rate of the mode in the external
496: medium. The solutions for $k_ya\neq0$ over-plotted in Figure~\ref{eigen} correspond to a value for
497: which the kink solution is already below the internal cut-off frequency and show significant changes
498: with respect to the $k_y=0$ case. The kink mode becomes a surface-like mode and motions are more
499: confined to the edges of the slab. Further increase of $k_y$ produces a more marked confinement and a
500: lower amplitude inside the slab. The middle and bottom panels of Figure~\ref{eigen} show the
501: eigenfunctions for the body and surface sausage modes. The most important property is
502: the similarity in the transverse component of the velocity for both solutions when $k_y$ is
503: sufficiently large and also that the motions are very confined to the edges of the slab.
504: The magnetic-pressure perturbation distribution peaks at the centre of the slab for the body
505: solution, while it is maximum at the edges in the case of the surface solution.
506: Finally, oblique propagation of waves affects the magnitude of the total pressure
507: perturbation of the surface sausage mode, but not very much that of the body sausage mode
508: inside the density enhancement.
509:
510:
511: \subsection{Damping of Oscillations by Resonant Absorption}\label{damped}
512:
513: In this section the most general equilibrium configuration considered in this work is taken.
514: The uniform internal density of the slab is now connected to the external (coronal) medium
515: by means of non-uniform transitional layers. This produces damping by resonant coupling of fast
516: modes to Alfv\'en waves (Hollweg and Yang, 1988; Goossens, Andries, and Arregui, 2006).
517: In order to solve Equations~(\ref{first})--(\ref{last}), in this case we have to resort
518: to numerical techniques and the numerical code PDE2D is used.
519: The code uses finite elements and allows the use of non-uniformly distributed grids, which is needed
520: in order to better resolve the large gradients that arise in the vicinity of the resonant layers
521: and has been used successfully, in a similar problem, by Terradas, Oliver, and Ballester (2006).
522: For the normal-mode analysis, a small but finite value of resistivity has to be provided when
523: computing the damping of oscillations. Resistivity has to be small enough for the imaginary part
524: of the frequency to be independent of resistivity (Poedts and Kerner, 1991). This condition has
525: been checked to a high accuracy for the solutions presented in this paper and a value for the
526: magnetic Reynolds number, $R_{m}/v_{Ai}a$, between $10^6$ and $10^8$ has been sufficient.
527:
528:
529:
530: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
531: \begin{figure*}[!t]
532: \begin{center}
533: \includegraphics[width=10.0cm,angle=90]{fig4.ps}
534: \end{center}
535: \caption{Damping rate for the resonantly-damped fast kink surface mode (solid line), sausage
536: surface mode (dotted line) and sausage body mode (dash-dotted line) as a function of the
537: perpendicular wave number in a coronal slab model with $k_za=\pi/50$, $\rho_i/\rho_e=10$ and
538: non-uniform transitional layers with thickness $l/a=0.5$. A value for the magnetic Reynolds number
539: of $R_{m}/v_{Ai}a=10^{7}$ has been used in the computation of the solutions in a non-uniform grid
540: with $N_x=10\ 000$ points in the range
541: $-50$ $\le x/a \le$ $50$.}
542: \label{damping}
543: \end{figure*}
544: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
545: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
546: \begin{figure*}[!t]
547: \hspace{-0.6cm}
548: \vbox{\hbox{
549: \includegraphics[width=5.cm,angle=90]{fig5a.ps}
550: \includegraphics[width=5.cm,angle=90]{fig5b.ps} }
551: \hbox{
552: \includegraphics[width=5.cm,angle=90]{fig5c.ps}
553: \includegraphics[width=5.cm,angle=90]{fig5d.ps}}
554: \hbox{
555: \includegraphics[width=5.cm,angle=90]{fig5e.ps}
556: \includegraphics[width=5.cm,angle=90]{fig5f.ps}}}
557: \caption{Real part of the transverse velocity component, $v_x$ and imaginary part of the perturbed
558: total pressure for the fundamental kink mode ({\em top panels}), the sausage body mode
559: ({\em middle panels}) and the sausage surface mode ({\em bottom panels}) for an equilibrium
560: configuration with $k_z a=\pi/50$, $\rho_i/\rho_e=10$, $k_ya=0.5$ and non-uniform transitional
561: layers with thickness $l/a=0.5$. A value for the magnetic Reynolds number of
562: $R_{m}/v_{Ai}a=10^{7}$ has been used in the computation of the solutions in a non-uniform grid
563: with $N_x=10\ 000$ points in the range $-50$ $\le x/a \le$ $50$.
564: In all figures, the shaded region represents the density
565: enhancement and the light-shaded regions the non-uniform layers.}
566: \label{eigendamping}
567: \end{figure*}
568:
569: The same parameter values for the equilibrium configurations are considered. Non-uniform transitional
570: layers of thickness $l/a=0.5$ have been considered and we have computed the real and imaginary parts
571: of the frequency for the kink mode and the sausage surface and body modes, for varying perpendicular
572: wave number. The results are displayed in Figure~\ref{damping}. When we increase $k_y$,
573: the damping rates of the kink mode and the sausage surface mode increase. Both damping rates are
574: very similar, despite the different symmetry of the motions. The damping rate corresponding to
575: the sausage body mode is roughly four to five orders of magnitude smaller than the one corresponding to
576: the two surface modes and practically of the order of the used magnetic
577: diffusivity. The reason for the absence of resonant damping in the case of the sausage
578: body solution is that its frequency, for the case of $k_z a$ small, lies
579: outside the Alfven continua produced by the presence of the non-uniform
580: layers defined by the interval
581: $[V_{Ai}k_z,V_{Ae}k_z]$ (see
582: Figure~\ref{disperplots}).
583: In conclusion, surface-like kink and sausage oscillations are likely
584: to be damped out rapidly, by resonant conversion of energy, while
585: body-like sausage modes are unaffected by resonant absorption.
586:
587: Figure~\ref{eigendamping} displays example eigenfunctions obtained from the computed resonantly
588: damped eigensolutions. All three modes under consideration, the sausage surface and body modes and
589: the kink mode, are displayed. The spatial distribution of eigenfunctions is rather similar to the
590: ones shown in Figure~\ref{eigen}, but now peaks at the non-uniform transitional layers are
591: clearly visible at the perturbed transverse velocity of the kink mode and the sausage
592: surface mode, while they are absent in the case of the sausage body mode.
593: These peaks are an indication of resonant coupling to Alfv\'en waves at the
594: non-uniform layers. Aside from these peaks, the spatial distribution of the magnetic
595: pressure perturbation has extrema at the resonant layers in the case of the kink mode and
596: the sausage surface mode, while it is maximum at the centre of the slab in the case
597: of the sausage body mode. The total pressure perturbation has no extrema at the resonant layers in the case of the
598: sausage body solution. The behaviour of the eigenfunctions at the resonant layers and the fundamental conservation laws
599: that govern the resonant couplings were studied by Sakurai, Goossens,
600: and Hollweg (1991) in ideal MHD and by Goossens, Ruderman, and Hollweg (1995) in dissipative MHD.
601: These authors find that the derivative of the total pressure
602: perturbation is zero at the resonant positions. This behaviour is retrieved in the pressure perturbation
603: for our damped surface-like solutions, but not in the case of the
604: sausage body solutions, for which resonant couplings are absent.
605:
606: \section{Time-Dependent Analysis}\label{timesect}
607:
608: The normal-mode analysis has provided us with information on the spatial
609: scales of the eigenmodes of the system and their relation to the
610: frequency of the eigenmodes. The issue of the excitation of oscillations is now considered.
611: Observed transverse coronal-loop oscillations have been found to be mainly produced
612: by impulsive events, such as filament eruptions or flare-generated blast waves.
613: Previous studies have considered the temporal evolution of initial pulses in a magnetic slab.
614: For instance, Murawski and Roberts (1993a, b) studied the energy leakage associated with the
615: propagation of sausage and kink waves in smoothed slabs and the temporal evolution of
616: impulsively-generated linear waves, respectively. Murawski, Aschwanden, and Smith (1998) studied the
617: different propagation phases (periodic, quasi-periodic, and decay) of impulsively-generated
618: waves in solar coronal loops, with arbitrary plasma-$\beta$. More recently, Nakariakov, Pascoe,
619: and Arber (2005) have demonstrated the possibility of remote diagnostics of the steepness of
620: the transverse density profile and the density contrast in coronal loops, with impulsively-generated,
621: short-period, fast magneto-acoustic wave trains.
622: From a theoretical point of view, it is of interest to study the distribution of the energy
623: of an initial perturbation among the different normal modes of the system, as well as the study
624: of the conditions under which one or other type of mode of oscillation is excited in the system.
625: The amount of energy that is deposited in the trapped fast mode oscillation, for different
626: initial excitations, has been studied by Terradas, Andries, and Goossens (2007), for a
627: cylindrical magnetic tube. These authors find that the amount of energy deposited in the loop depends
628: on the shape and distance of the initial perturbation. In this paper, we are interested in the
629: conversion of energy from fast modes to Alfv\'en modes and the comparison of the oscillatory properties
630: of the excited excited disturbances, such as the period and the damping, with the results
631: of the normal mode analysis. For this reason, we show results of numerical simulations of the
632: linearised MHD equations performed with initial disturbances that deposit a considerable
633: amount of energy in the loop. As the normal modes studied in Section~\ref{normal}
634: are characterised by their symmetry around $x=0$, it is likely that an initial disturbance with a
635: given symmetry will excite one or the other types of mode. For this reason we have performed
636: numerical simulations with two different types of initial disturbances, namely $v_x$ symmetric and
637: antisymmetric with respect to $x=0$.
638:
639:
640:
641: To solve Equations~(\ref{first})--(\ref{last}) numerically we have used the time-dependent version
642: of the PDE2D code. Also, $\eta$ is now set to zero. The code uses a second-order
643: Crank-Nicholson method with adaptive time-step control. Since we are considering a finite domain,
644: reflections at the domain boundaries may affect the dynamics of the simulated loop. We have
645: solved this problem by locating the edges of the numerical domain far enough from the loop.
646: Given that the size of the domain is much larger than the loop thickness, we have used a non-uniform
647: grid with 10\ 000 grid points in the full domain, one fifth of them located inside the loop and
648: another fifth in each inhomogeneous layer.
649:
650: \begin{figure*}[!t]
651: \hspace{-0.7cm}
652: \vbox{\hbox{
653: \includegraphics[width=5.2cm,angle=90]{fig6a.ps}
654: \includegraphics[width=5.2cm,angle=90]{fig6b.ps} }
655: \hbox{
656: \includegraphics[width=5.2cm,angle=90]{fig6c.ps}
657: \includegraphics[width=5.2cm,angle=90]{fig6d.ps}}}
658: \caption{Transverse velocity component, $v_x$, at different times (in units of the
659: internal Alfv\'en transit time, $\tau_{Ai}=a/v_{Ai}$) for a symmetric disturbance given by
660: Equation~(\ref{kinkpert}), with $x_0=0$, $w=2a$ and $a=1$. The inset plot in the lower-sight frame
661: displays a detailed view of the transverse velocity component of the excited kink mode in order to
662: compare it with the eigenmode computation shown in Figure~\ref{eigendamping}. The grey-shaded regions
663: represent the loop.}
664: \label{timeevolkink}
665: \end{figure*}
666:
667: \begin{figure*}[!t]
668: \hspace{-0.6cm}
669: \includegraphics[width=5.4cm,angle=90]{fig7a.ps}
670: \hspace{-0.4cm}
671: \includegraphics[width=5.4cm,angle=90]{fig7b.ps}\\
672: \caption{{\em Left panel}: Transverse velocity [$v_x$] at the slab centre [$x=0$] as
673: a function of time for the simulation shown in Figure~\ref{timeevolkink}. After a very short
674: transient phase the loop reaches a stationary state, oscillating at the frequency predicted by the
675: normal mode analysis. The oscillation is damped due to resonant absorption. {\em Right panel}:
676: Periodogram of the signal, that peaks at $\omega a/v_{Ai}=0.1018$, in good agreement with
677: the normal mode result, $\omega_{R} a/v_{Ai}=0.1022$ (dashed-line).
678: An exponential fit of the signal for $v_x$ gives a damping time $\tau_{d}/\tau_{Ai}=225.5$,
679: that corresponds to an imaginary part of the frequency $\omega_{I} a/v_{Ai}=4.43$ $\times$ $10^{-3}$,
680: in good agreement with the normal mode result, $\omega_{I}a/v_{Ai}=4.38$ $\times$ $10^{-3}$.
681: Since both the peak power and the damping time agree with the normal-mode results, there
682: is a clear evidence that the fundamental fast kink body mode has been excited.}
683: \label{timekink}
684: \end{figure*}
685:
686:
687: \subsection{Symmetric Excitation}
688:
689: We first consider an initial disturbance on the transverse velocity component with a Gaussian of
690: the form
691:
692: \begin{equation}\label{kinkpert}
693: v_{x}(x,t=0)=v_{x0}\left\{\exp\left[-\left(\frac{x-x_0}{w}\right)^2\right]\right\},
694: \end{equation}
695:
696: \noindent
697: where $v_{x0}$ is the amplitude of the perturbation, $x_0$ the position of the Gaussian centre
698: and $w$ its width at half-height. This form of excitation aims to simulate the transverse
699: disturbance generated by a flare or filament eruption. Since the numerical computations apply to
700: the linear regime, the particular value of $v_{x0}$ is not relevant and has been set to $v_{Ai}$.
701:
702: In order to compare our results with the properties of the eigensolutions described in
703: Section~\ref{normal}, we have considered a slab model of half-width $a$, density contrast
704: $\rho_i/\rho_e=10$, and $k_za=\pi/50$. For the perpendicular wave number and the non-uniform
705: transitional layers thickness we take values of $k_ya=0.5$ and $l/a=0.5$. The system is then
706: disturbed with a perturbation given by Equation~(\ref{kinkpert}) with $x_0=0$ and $w=2a$.
707: The results of the simulation are shown in Figure~\ref{timeevolkink}, where the spatial distribution
708: of the transverse velocity component is plotted at different times. The initial perturbation
709: produces travelling disturbances to the left and right of its initial location and these
710: disturbances exhibit some dispersion as they propagate. After a short time it is seen that
711: the distribution of energy has a maximum at the centre of the system, with the amplitude
712: decreasing as we move away from the edges of the slab. This indicates that the fundamental
713: kink eigenmode may have been excited. The inset plot in the last shown frame supports this idea, since
714: the structure of $v_x$ is almost identical to that of the normal mode, shown in Figure~\ref{eigendamping}.
715: In Figure~\ref{timekink} (left)
716: the velocity at $x=0$ is plotted. The signal at this position clearly shows a damped oscillatory
717: behaviour. This damping is due to the resonant coupling and conversion of energy to Alfv\'enic
718: motions. We have performed a periodogram (Figure~\ref{timekink}, right) and found that the
719: dominant period agrees very well with the period of the fundamental kink eigenmode obtained
720: in Section~\ref{normal}. The extrema of the signal are then fitted to an exponential of the form $A\ e^{-\tau_d/ t}$.
721: The fitted function is also shown in Figure~\ref{timekink} (left).
722: From the periodogram and the exponential fit we find that the dominant period and damping time
723: agree very well with the oscillatory properties of the kink eigenfunction, which leads to the conclusion
724: that the resonantly damped fundamental kink mode has been excited.
725: Therefore, a symmetric excitation is likely to excite the fundamental kink mode of oscillation of
726: the system.
727:
728:
729:
730: \subsection{Antisymmetric Excitation}
731:
732: Next, we consider the excitation of the slab with an odd perturbation of the form
733:
734: \begin{equation}\label{sausagepert}
735: v_{x}(x,t=0)=v_{x0}\left\{\exp\left[-\left(\frac{x-x_0}{w}\right)^2\right]-
736: \exp\left[-\left(\frac{x+x_0}{w}\right)^2\right]\right\},
737: \end{equation}
738:
739: \noindent
740: where $v_{x0}$, $x_0$, and $w$ have the same meaning as before. The system is now disturbed with a
741: perturbation given by Equation~(\ref{sausagepert}) with $x_0= a$ and $w=2a$. As the surface sausage mode
742: is highly confined to the loop edges, short spatial scales are involved and these scales have to be
743: introduced either through the width of the disturbance or through the value of $k_y$.
744: The first option complicates the numerical computations, so a value $k_y a=1.4$ has now been taken.
745: The numerical solutions at different times (Figure~\ref{timeevolsausage}) show again that the
746: induced disturbances propagate away from the slab and contain different wavelengths due to the
747: dispersive nature of fast waves. After some time, the shape of the velocity around the slab
748: approaches the form of the sausage surface fast wave described in
749: Section~\ref{normal}. In contrast to the kink case, the velocity distribution in the slab is now more
750: complex, since more than one oscillatory mode
751: seems to be present in the signal (see, for example, the inset in the last shown frame).
752: The consequence of this can clearly be seen in Figure~\ref{timesausage} (left) where the
753: signal at $x/a=1$ is shown. The presence of two linearly superposed oscillations is apparent. A short
754: period oscillation is modulated in amplitude with a large period oscillation. The signal is again
755: damped in time. The computation of the dominant periods of the signal gives the result displayed in
756: Figure~\ref{timesausage} (right). There are two peaks whose frequencies are in perfect agreement with the
757: frequencies of the sausage surface and body modes, obtained from the normal mode analysis. The power
758: for the body sausage mode is larger, although this may change depending on the spatial scales involved.
759: The damping time of the signal, obtained with a similar exponential fit as in the previous case, does not correspond to
760: any of the two eigenmodes, since the two are present
761: at the same time, but is close to the damping of the sausage surface mode.
762: Therefore, we can conclude that an antisymmetric disturbance located at the edges of the slab
763: induces the excitation of both types of sausage modes. The reason is that the surface and body
764: sausage modes, although having very different frequencies, have very similar eigenfunctions,
765: when $k_y$ is sufficiently large.
766:
767:
768:
769: \begin{figure*}[!t]
770: \hspace{-0.7cm}
771: \vbox{\hbox{
772: \includegraphics[width=5.2cm,angle=90]{fig8a.ps}
773: \includegraphics[width=5.2cm,angle=90]{fig8b.ps} }
774: \hbox{
775: \includegraphics[width=5.2cm,angle=90]{fig8c.ps}
776: \includegraphics[width=5.2cm,angle=90]{fig8d.ps}}}
777: \caption{Transverse velocity component [$v_x$] at different times (in units of the internal
778: Alfv\'en transit time, $\tau_{Ai}=a/v_{Ai}$), for an antisymmetric disturbance given by
779: Equation~(\ref{sausagepert}), with $x_0=a$, $w=2a$, and $a=1$. The inset plot in the lower-sight frame
780: displays a detailed view of the transverse velocity component of the excited sausage modes in order to
781: compare it with the eigenmode computations shown in Figure~\ref{eigendamping}. The grey-shaded regions
782: represent the loop.}
783: \label{timeevolsausage}
784: \end{figure*}
785:
786:
787: \begin{figure*}[!t]
788: \hspace{-0.6cm}
789: \includegraphics[width=5.4cm,angle=90]{fig9a.ps}
790: \hspace{-0.4cm}
791: \includegraphics[width=5.4cm,angle=90]{fig9b.ps}\\
792:
793:
794: \caption{{\em Left panel}: Transverse velocity [$v_x$] at the slab edge [$x=a$] as a function
795: of time for the simulation shown in Figure~\ref{timeevolsausage}. After a very short transient phase
796: the loop oscillates with a linear superposition of two frequencies. The signal is damped due to resonant
797: absorption. A fit of the signal for $v_x$ gives a damping time $\tau_{d}/\tau_{Ai}=621.6$, that corresponds
798: to an imaginary part of the frequency $\omega_{I} a/v_{Ai}=1.6$$\times$$10^{-3}$. {\em Right panel}:
799: Periodogram of the signal that peaks at $\omega_1 a/v_{Ai}=0.0817$ and $\omega_2 a/v_{Ai}=1.9998$. The
800: dashed lines represent the frequencies obtained with the normal mode analysis; $\omega_{R} a/v_{Ai}=0.0778$ for
801: the sausage surface mode and $\omega_{R} a/v_{Ai}=2.0087$ for the sausage body mode. As the damping time of
802: the motion is a linear combination of the damping of the two modes the fitted damping is in between the
803: imaginary parts of the frequency for the sausage surface mode ($\omega_{I1} a/v_{Ai}=1.35$$\times$$10^{-2}$)
804: and the sausage body mode ($\omega_{I2} a/v_{Ai}=1.74$$\times$$10^{-6}$), for this case. Since the power
805: peaks agree with the normal mode results there is a clear evidence that both the surface and body sausage
806: modes have been excited.}
807: \label{timesausage}
808: \end{figure*}
809:
810:
811: \section{Summary and Conclusions}\label{conclusions}
812:
813: We have studied the oscillatory properties of surface and body MHD eigenmodes of a solar coronal loop
814: including oblique propagation of perturbations. For simplicity, a Cartesian slab model has been considered
815: and both the eigenvalue problem as well as the time-dependent problem have been solved.
816:
817: The inclusion of oblique propagation of perturbations produces some important effects on the properties of
818: eigenmodes. The resulting dispersion curves for non-zero perpendicular propagation show the existence of a
819: sausage surface mode in addition to the usual fast body kink and sausage solutions, even if a
820: zero-plasma-$\beta$ has been considered. This mode, not present if $k_y=0$, has the lowest frequency and
821: this frequency is always below the internal cut-off frequency. For small values of $k_y$, in the long-wave
822: limit, the kink eigenmode is a body wave with the usual distribution of the velocity perturbation with a
823: maximum at the centre and an evanescent behaviour outside the slab. The sausage surface mode has a velocity
824: perturbation which is more confined to the edges of the slab with a short penetration depth into the coronal
825: surroundings. When the value of the perpendicular wave number is increased, a marked decrease of the
826: oscillation frequency is produced in the case of the kink mode. The spatial structure of the eigenfunction
827: of this mode becomes much more confined to the slab. This leads to a change in the character of the
828: kink mode which changes from body-like to surface-like. For large values of $k_y$, both surface modes
829: (kink and sausage) approach the kink speed limit, that corresponds to surface waves in a magnetic interface
830: in the incompressible limit or to the kink eigenoscillations of a cylindrical flux tube in the long
831: wavelength limit. As for the sausage surface mode, there is no appreciable change in its eigenfunction
832: with $k_y$ and only a slight increase on the frequency is produced. The sausage body mode, which is leaky
833: in the long-wavelength limit, becomes trapped when, for a give value of $k_y$, its frequency equals
834: the external cut-off frequency, $\omega_{ci}$. This is due to the modification of the external
835: cut-off frequency produced by the inclusion of oblique propagation. The internal cut-off frequency is also modified by
836: oblique propagation, but the frequency of the sausage body mode is
837: always above this cut-off frequency, due to the existence of a surface mode with the same symmetry below that cut-off.
838: The eigenfunctions for both sausage solutions have a marked resemblance and this has important consequences on the
839: excitation of these modes. When the density is allowed to vary smoothly between the constant internal and the constant
840: external values, surface-like eigenmodes are damped by resonant conversion of energy. Both surface modes have a
841: similar damping rate, while the sausage body mode is unaffected by
842: resonant absorption.
843:
844:
845: In the second part of the paper we have studied the time-dependent behaviour of
846: our line-tied slab model for a coronal loop under different kinds of excitations.
847: The temporal evolution and damping by resonant conversion of wave energy of kink- and
848: sausage-type excitations has been studied for a typical coronal loop. The eigenmodes described in the
849: first part of the paper can be easily excited. A symmetric disturbance located at the centre of the slab
850: excites the fundamental symmetric kink mode. Depending on the value of $k_y$, the excited mode would have
851: body or surface wave properties. The period of the signal
852: and its damping time agree with the result obtained with the normal mode analysis. The velocity
853: distribution of the stationary phase coincides with the spatial distribution of the eigenfunction.
854: Interestingly, an antisymmetric initial disturbance with peaks at the edges of the slab excites both the
855: surface and the body sausage modes and a linear superposition of perturbations is obtained, with the sausage
856: body oscillation amplitude being modulated by the presence of the sausage surface mode. This can be explained
857: in terms of the involved eigenfunctions since oblique propagation confines the spatial distribution of the
858: body sausage mode and makes it very similar to that of the sausage surface mode.
859:
860: In this work, a magnetic slab has been used to model a coronal loop. Some of the results found in slab
861: geometry cannot be translated to cylindrical geometry. For instance, the sausage surface mode described in
862: this paper is not present in cylindrical geometry. Also, the sausage mode of oscillation of a cylinder
863: corresponds to an azimuthal wave number $m=0$, thus this mode cannot resonantly couple to Alfv\'en modes.
864: Slabs are known to be poor wave-guides, when compared to cylinders. However, our results indicate that the inclusion
865: of oblique propagation produces a sharper drop-off rate of the kink eigenfunction in the external medium and
866: a frequency that is a good approximation to the kink-mode frequency in a slender flux tube.
867:
868:
869:
870:
871: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
872: \begin{acks}
873: We thank the referee, Dr.\ Erwin Verwichte, for constructive and valuable comments that
874: have benefited this paper. The authors acknowledge the Spanish Ministerio de
875: Educaci\'on y Ciencia for the funding
876: provided under project AYA2006-07637 and the Conselleria d'Economia, Hisenda i
877: Innovaci\'o of the Government of the Balearic Islands for the funding provided under
878: grants PRIB-2004-10145 and PCTIB2005GC3-03. J. Terradas acknowledges the Spanish Ministerio de
879: Educaci\'on y Ciencia for the funding provided under a Juan de la Cierva fellowship.
880: \end{acks}
881:
882:
883:
884: \begin{thebibliography}{}
885: \bibitem[\protect\citeauthoryear{Andries et al.}{2005}]{AAG05}
886: Andries, J., Arregui, I., Goossens, M.: 2005, {\it Astrophys.
887: J.} {\bf 624},
888: L57.
889:
890: \bibitem[\protect\citeauthoryear{Andries et al.}{2005}]{Andries05}
891: Andries, J., Goossens, M., Hollweg, J.V., Arregui, I., Van
892: Doorsselaere, T.: 2005, {\it Astron. Astrophys.} {\bf 430}, 1109.
893:
894: \bibitem[\protect\citeauthoryear{Arregui et al.}{2005}]{Arregui05}
895: Arregui, I., Van Doorsselaere, T., Andries, J., Goossens, M.,
896: Kimpe, D.: 2005, {\it Astron. Astrophys.} {\bf 441}, 361.
897:
898:
899: \bibitem[\protect\citeauthoryear{Arregui et al.}{2007}]{Arregui07b}
900: Arregui, I., Andries, J., Van Doorsselaere, T., Goossens, M.,
901: Poedts, S.: 2007b, {\it Astron. Astrophys.} {\bf 463}, 333.
902:
903:
904: \bibitem[\protect\citeauthoryear{Aschwanden}{2006}]{Aschwanden06}
905: Aschwanden, M.J.: 2006, {\it Roy. Soc. London Phil. Trans.
906: Ser. A} {\bf 364}, 417.
907:
908: \bibitem[\protect\citeauthoryear{Aschwanden et al.}{2002}]{Aschwanden02}
909: Aschwanden, M.J., De Pontieu, B., Schrijver, C.J., Title, A.M.:
910: 2002, {\it Solar Phys.} {\bf 206}, 99.
911:
912: \bibitem[\protect\citeauthoryear{Aschwanden et al.}{1999}]{Aschwanden99}
913: Aschwanden, M.J., Fletcher, L., Schrijver, C.J., Alexander, D.:
914: 1999, {\it Astrophys. J.} {\bf 520}, 880.
915: \bibitem[\protect\citeauthoryear{Aschwanden et al.}{2003}]{Aschwanden03}
916: Aschwanden, M.J., Nightingale, R.W., Andries, J., Goossens, M., Van Doorsselaere, T.:
917: 2003, {\it Astrophys. J.} {\bf 598}, 1375.
918:
919: \bibitem[\protect\citeauthoryear{Brady and Arber}{2005}]{BA05}
920: Brady, C.S., Arber, T.D.: 2005, {\it Astron. Astrophys}. {\bf
921: 438}, 733.
922:
923:
924:
925: \bibitem[\protect\citeauthoryear{Diaz}{2006}]{toni06a}
926: D\'{\i}az, A.J.: 2006, {\it Astron. Astrophys.} {\bf 456}, 737.
927:
928:
929: \bibitem[\protect\citeauthoryear{Diaz et al.}{2003}]{toni03}
930: D\'{\i}az, A.J., Oliver, R., Ballester, J.L.: 2003, {\it Astron.
931: Astrophys.} {\bf 402}, 781.
932:
933: \bibitem[\protect\citeauthoryear{Diaz et al.}{2006}]{toni06b}
934: D\'{\i}az, A.J., Zaqarashvili, T.V., Roberts, B.: 2006, {\it Astron.
935: Astrophys.} {\bf 455}, 709.
936:
937:
938: \bibitem[\protect\citeauthoryear{Dymova and Ruderman}{2006}]{DR06}
939: Dymova, M.V., Ruderman, M.S.: 2006, {\it Astron. Astrophys.}
940: {\bf 457}, 1059.
941:
942: \bibitem[\protect\citeauthoryear{Edwin \& Roberts}{1982}]{ER82}
943: Edwin, P.M., Roberts, B.: 1982, {\it Solar Phys.} {\bf 79}, 239.
944:
945:
946: \bibitem[\protect\citeauthoryear{Edwin \& Roberts}{1983}]{ER83}
947: Edwin, P.M., Roberts, B.: 1983, {\it Solar Phys.} {\bf 88}, 179.
948:
949: \bibitem[\protect\citeauthoryear{Edwin \& Roberts}{1988}]{ER88}
950: Edwin, P.M., Roberts, B.: 1988, {\it Astron. Astrophys.} {\bf
951: 192}, 343.
952:
953:
954: \bibitem[\protect\citeauthoryear{Goossens}{1991}]{Goossens91}
955: Goossens, M.: 1991 In Ulmschneider, P., Priest E.R., Rosner, R.
956: (eds.), {\it Mechanisms of Chromospheric and Coronal Heating},
957: Springer-Verlag, Berlin, 480.
958:
959: \bibitem[\protect\citeauthoryear{Goossens et al.}{2006}]{GAA06}
960: Goossens, M., Andries, J., Arregui, I.: 2006, {\it Roy. Soc. London
961: Phil. Trans. Ser. A} {\bf 364}, 433.
962:
963: \bibitem[\protect\citeauthoryear{Goossens et al.}{2002}]{GAA02}
964: Goossens, M., Andries, J., Aschwanden, M.J..: 2002, {\it Astron. Astrophys.}
965: {\bf 394}, L39.
966:
967: \bibitem[\protect\citeauthoryear{Goossens et al.}{1995}]{GRH95}
968: Goossens, M., Ruderman, M.S., Hollweg, J.V.: 1995,
969: {\it Solar Phys.} {\bf 157}, 75.
970:
971:
972: %\bibitem[\protect\citeauthoryear{Goossens et al.}{2002}]{GAA02}
973: %Goossens, M., Andries, J., \& Aschwanden, M.~J.: 2002, {\it Astron. Astrophys.}, {\bf 394}, L39.
974:
975:
976: \bibitem[\protect\citeauthoryear{Hollweg \& Yang}{1988}]{HY88}
977: Hollweg, J.V., Yang, G.: 1988, {\it J. Geophys. Res.} {\bf 93}, 5423.
978:
979: \bibitem[\protect\citeauthoryear{Hollweg}{1990}]{Holl90a}
980: Hollweg, J.V.: 1990a, {\it Comp. Phys. Reports} {\bf 12}, 205.
981:
982: \bibitem[\protect\citeauthoryear{Hollweg}{1990}]{Holl90b}
983: Hollweg, J.V.: 1990b, In Russel, C.T., Priest, E.R., Lee, L.C.
984: (eds.), {\it Physics of Magnetic Flux Ropes}, (Washington, AGU), Geophys.
985: Mono. 58, 123.
986:
987: \bibitem[\protect\citeauthoryear{Hollweg}{1991}]{Holl91}
988: Hollweg, J.V.: 1991 In Ulmschneider, P., Priest E.R., Rosner, R.
989: (eds.), {\it Mechanisms of Chromospheric and Coronal Heating},
990: Springer-Verlag, Berlin, 423.
991:
992: \bibitem[\protect\citeauthoryear{Jain \& Roberts}{1994}]{JR94}
993: Jain, R., Roberts, B.: 1994, {\it Astron. Astrophys.} {\bf 286},
994: 243.
995:
996: \bibitem[\protect\citeauthoryear{Lee \& Roberts}{1986}]{LR86}
997: Lee, M.A., Roberts, B.: 1986, {\it Astrophys. J.} {\bf 301}, 430.
998:
999: \bibitem[\protect\citeauthoryear{McEwan et al.}{2006}]{McE06}
1000: McEwan, M.P., Donnelly, G.R., D\'{\i}az, A.J., Roberts, B.: 2006,
1001: {\it Astron. Astrophys.} {\bf 460}, 893.
1002:
1003: \bibitem[\protect\citeauthoryear{Miles \& Roberts}{1989}]{MR89}
1004: Miles, A.J., Roberts, B.: 1989, {\it Solar Phys.} {\bf 119}, 257.
1005:
1006:
1007: \bibitem[\protect\citeauthoryear{Murawski \& Roberts}{1993a}]{MR93a}
1008: Murawski, K., Roberts, B.: 1993a, {\it Solar Phys.} {\bf 143}, 89.
1009:
1010:
1011: \bibitem[\protect\citeauthoryear{Murawski \& Roberts}{1993b}]{MR93b}
1012: Murawski, K., Roberts, B.: 1993b, {\it Solar Phys.} {\bf 144}, 101.
1013:
1014:
1015: \bibitem[\protect\citeauthoryear{Murawski et al.}{1998}]{MR98}
1016: Murawski, K., Aschwanden, M.J., Smith, J.M.: 1998, {\it Solar
1017: Phys.} {\bf 179}, 313.
1018:
1019:
1020: \bibitem[\protect\citeauthoryear{Nakariakov \& Ofman}{2001}]{Nakariakov01}
1021: Nakariakov, V.M., Ofman, L.: 2001, {\it Astron. Astrophys.} {\bf
1022: 372}, L53.
1023:
1024: \bibitem[\protect\citeauthoryear{Nakariakov \& Roberts}{1995}]{Nakariakov95}
1025: Nakariakov, V.M., Roberts, B.: 1995, {\it Solar Phys.} {\bf
1026: 159}, 399.
1027:
1028: \bibitem[\protect\citeauthoryear{Nakariakov \& Verwichte}{2005}]{Nakariakov05}
1029: Nakariakov, V.M., Verwichte, E.: 2005, {\it Living Rev. Solar
1030: Phys.} {\bf 2}, 3. {\sf http://solarphysics.livingreviews.org}
1031:
1032: \bibitem[\protect\citeauthoryear{Nakariakov, Pascoe, and Arber}{2005}]{Nakariakov05b}
1033: Nakariakov, V.M., Pascoe, D.J., Arber, T.D..: 2005,
1034: {\it Space Sci Rev.} {\bf 121}, 115.
1035:
1036:
1037: \bibitem[\protect\citeauthoryear{Nakariakov et al.}{1999}]{Nakariakov99}
1038: Nakariakov, V.M., Ofman, L., DeLuca, E.E., Roberts, B., Davila,
1039: J.M.: 1999, {\it Science} {\bf 285}, 862.
1040:
1041:
1042: \bibitem[\protect\citeauthoryear{Poedts \& Kerner}{1991}]{POKE91}
1043: Poedts, S., Kerner, W.: 1991, {\it Phys. Rev. Lett.} {\bf 66}, 2871.
1044:
1045: \bibitem[\protect\citeauthoryear{Roberts}{1981}]{Roberts81a}
1046: Roberts, B.: 1981a, {\it Solar Phys.} {\bf 69}, 27.
1047:
1048: \bibitem[\protect\citeauthoryear{Roberts}{1981}]{Roberts81b}
1049: Roberts, B.: 1981b, {\it Solar Phys.} {\bf 69}, 39.
1050:
1051: \bibitem[\protect\citeauthoryear{Roberts}{1983}]{Roberts83}
1052: Roberts, B.: 1983, {\it Solar Phys.} {\bf 87}, 77.
1053:
1054:
1055: \bibitem[\protect\citeauthoryear{Roberts}{1991}]{Roberts91}
1056: Roberts, B.: 1991, In Ulmschneider, P., Priest E.R., Rosner, R.
1057: (eds.), {\it Mechanisms of Chromospheric and Coronal Heating},
1058: Springer-Verlag, Berlin, 494.
1059:
1060: \bibitem[\protect\citeauthoryear{Roberts et al.}{1984}]{REB84}
1061: Roberts, B., Edwin, P.M., Benz, A.O.: 1984, {\it Astrophys.
1062: J.} {\bf 279}, 857.
1063:
1064: \bibitem[\protect\citeauthoryear{Ruderman}{2003}]{Ruderman03}
1065: Ruderman, M.S.: 2003, {\it Astron. Astrophys.} {\bf 409}, 287.
1066:
1067:
1068: \bibitem[\protect\citeauthoryear{Ruderman \& Roberts}{2002}]{RR02}
1069: Ruderman, M.S. and Roberts, B.: 2002, {\it Astrophys. J.} {\bf
1070: 577}, 475.
1071:
1072: \bibitem[\protect\citeauthoryear{Ryutova}{1990}]{Ryutova90}
1073: Ryutova, M.~P.: 1990, In Stenflo, J.O. (ed.),
1074: {\it Solar Photosphere: Structure, Convection and Magnetic Fields},
1075: ( Reidel, Dordrecht), IAU Symp. 138, 229.
1076:
1077: \bibitem[\protect\citeauthoryear{Sakurai et al.}{1991}]{SGH91}
1078: Sakurai, T., Goossens, M., Hollweg, J.V.: 1991, {\it Solar
1079: Phys.} {\bf 133},
1080: 227.
1081:
1082:
1083: \bibitem[\protect\citeauthoryear{Sewell}{2005}]{Sewell05}
1084: Sewell, G.: 2005, {\it The Numerical Solution of Ordinary and Partial
1085: Differential Equations}, John Wiley \& Sons, Inc.,
1086: Hoboken, New Jersey.
1087:
1088:
1089: \bibitem[\protect\citeauthoryear{Schrijver et al.}{2002}]{SAT02}
1090: Schrijver, C.J., Aschwanden, M.J., Title, A.M.: 2002, {\it
1091: Solar Phys.} {\bf 206},
1092: 69.
1093: \bibitem[\protect\citeauthoryear{Spruit}{1981}]{Spruit81}
1094: Spruit, H.C.: 1981, In Jordan, S. (ed.), {\it The Sun as a Star},
1095: SP-450, NASA Washington, 385.
1096:
1097: \bibitem[\protect\citeauthoryear{Terradas, et al.}{2005}]{TOB05}
1098: Terradas, J., Oliver, R., Ballester, J.L.: 2005, {\it Astron.
1099: Astrophys.} {\bf 441}, 371.
1100:
1101:
1102:
1103: \bibitem[\protect\citeauthoryear{Terradas, et al.}{2006}]{TOB06}
1104: Terradas, J., Oliver, R., Ballester, J.L.: 2006, {\it
1105: Astrophys. J.} {\bf 650}, L91.
1106:
1107: \bibitem[\protect\citeauthoryear{Terradas, et al.}{2007}]{TAG07}
1108: Terradas, J., Andries, J., Goossens, M.: 2007, {\it Astron.
1109: Astrophys.}, {\bf 469}, 1135.
1110:
1111:
1112:
1113: \bibitem[\protect\citeauthoryear{}{}]{}
1114: Uchida, Y.: 1970, {\it Pub. Astron. Soc. Japan} {\bf 22}, 341.
1115:
1116: \bibitem[\protect\citeauthoryear{Van Doorsselaere et al.}{2004}]{tom04b}
1117: Van Doorsselaere, T., Andries, J., Poedts, S., Goossens, M.: 2004,
1118: {\it Astrophys. J.} {\bf 606}, 1223.
1119:
1120:
1121: \bibitem[\protect\citeauthoryear{Verwichte et al.}{2006a}]{verwichte06a}
1122: Verwichte, E., Foullon, C., Nakariakov, V.M.: 2006a, {\it
1123: Astron. Astrophys.} {\bf 446}, 1139.
1124:
1125: \bibitem[\protect\citeauthoryear{Verwichte et al.}{2006b}]{verwichte06b}
1126: Verwichte, E., Foullon, C., Nakariakov, V.M.: 2006b, {\it
1127: Astron. Astrophys.} {\bf 449}, 769.
1128:
1129:
1130: \bibitem[\protect\citeauthoryear{Verwichte et al.}{2006c}]{verwichte06c}
1131: Verwichte, E., Foullon, C., Nakariakov, V.M.: 2006c, {\it
1132: Astron. Astrophys.} {\bf 452}, 615.
1133:
1134:
1135: \bibitem[\protect\citeauthoryear{Zhelyazkov et al.}{1996}]{zhelyazkov96}
1136: Zhelyazkov, I., Murawski, K., Goossens, M.: 1996, {\it Solar
1137: Phys.} {\bf 165}, 99.
1138:
1139: \end{thebibliography}
1140:
1141:
1142: \end{article}
1143: \end{document}
1144: