1: %\documentclass[12pt,preprint]{aastex}
2: \documentclass{emulateapj}
3: \usepackage{times}
4: \setlength{\parskip}{1ex plus 0.5ex minus 0.2ex}
5:
6: %\linespread{1}
7:
8: %\usepackage{amssymb}
9:
10: \newcommand{\D}[2]{\frac{\partial #2}{\partial #1}}
11: \newcommand\bb[1]{\mbox{\boldmath{$#1$}}}
12: \newcommand\del{\bb{\nabla}}
13: \newcommand\bcdot{\bb{\cdot}}
14: \newcommand\btimes{\bb{\times}}
15: \newcommand\real{{\rm Re}}
16:
17: \begin{document}
18: %\submitted{Submitted to the Astrophysical Journal, \today}
19: %\submitted{Draft version, \today}
20:
21: \shorttitle{\textsc{Viscous, Resistive Magnetorotational Modes}}
22: \shortauthors{\textsc{Pessah and Chan}}
23:
24: \title{\textsc{Viscous, Resistive Magnetorotational Modes}}
25:
26: \author{Martin E. Pessah}
27: \affil{School of Natural Sciences, Institute for Advanced Study, Princeton, NJ, 08540}
28: \and
29: \author{Chi-kwan Chan}
30: \affil{Institute for Theory and Computation,
31: Harvard-Smithsonian Center for Astrophysics,
32: 60 Garden Street, Cambridge, MA 02138}
33: \email{mpessah@ias.edu, ckchan@cfa.harvard.edu}
34:
35: \begin{abstract}
36: We carry out a comprehensive analysis of the behavior of the
37: magnetorotational instability (MRI) in viscous, resistive plasmas.
38: We find exact, non-linear solutions of the non-ideal
39: magnetohydrodynamic (MHD) equations describing the local dynamics of
40: an incompressible, differentially rotating background threaded by a
41: vertical magnetic field when disturbances with wavenumbers
42: perpendicular to the shear are considered. We provide a geometrical
43: description of these viscous, resistive MRI modes and show how their
44: physical structure is modified as a function of the Reynolds and
45: magnetic Reynolds numbers. We demonstrate that when finite
46: dissipative effects are considered, velocity and magnetic field
47: disturbances are no longer orthogonal (as it is the case in the
48: ideal MHD limit) unless the magnetic Prandtl number is unity. We
49: generalize previous results found in the ideal limit and show that a
50: series of key properties of the mean Reynolds and Maxwell stresses
51: also hold for the viscous, resistive MRI. In particular, we show
52: that the Reynolds stress is always positive and the Maxwell stress
53: is always negative. Therefore, even in the presence of viscosity
54: and resistivity, the total mean angular momentum transport is always
55: directed outwards. We also find that, for any combination of the
56: Reynolds and magnetic Reynolds numbers, magnetic disturbances
57: dominate both the energetics and the transport of angular momentum
58: and that the total mean energy density is an upper bound for the
59: total mean stress responsible for angular momentum transport. The
60: ratios between the Maxwell and Reynolds stresses and between
61: magnetic and kinetic energy densities increase with decreasing
62: Reynolds numbers for any magnetic Reynolds number; the lowest limit
63: of both ratios is reached in the ideal MHD regime.
64: \end{abstract}
65:
66: \keywords{black hole physics --- accretion, accretion disks --- MHD
67: --- instability --- turbulence}
68:
69:
70:
71: \section{Introduction}
72: \label{sec:intro}
73:
74: The magnetorotational instability \citep[MRI,][]{BH91, BH98} has been
75: widely studied in the inviscid and perfectly conducting,
76: magnetohydrodynamic (MHD) limit. The departures from this idealized
77: situation are usually parametrized according to the Reynolds ${\rm Re}
78: = vl/\nu$ and magnetic Reynolds ${\rm Rm} = vl/\eta$ numbers, where
79: $v$ and $l$ stand for the relevant characteristic velocity and
80: lengthscale and $\nu$ and $\eta$ stand for the kinematic viscosity and
81: resistivity. The ideal MHD regime is then formally identified with the
82: limit ${\rm Re}~,~{\rm Rm}~\rightarrow~\infty$. There are many
83: situations of interest in which the effects of dissipation need to be
84: considered.
85:
86: From the astrophysical point of view, accretion disks around young
87: stellar objects constitute one of the most compelling reasons for
88: investigating the MRI beyond the ideal limit. In particular, there is
89: great interest in understanding to what extent can MHD turbulence
90: driven by the MRI enable efficient angular momentum transport in cool,
91: poorly conducting, protoplanetary disks \citep[see, e.g.][]{BB94,
92: Jin96, Gammie96, SM99, SW05}. Most of the studies addressing the
93: effects of dissipation in non-ideal MRI have usually focused in
94: inviscid, resistive plasmas. However, accretion disks are
95: characterized by a wide range of magnetic Prandtl numbers, with ${\rm
96: Pm}=\nu/\eta$ varying by several orders of magnitude across the
97: entire disk \citep[see, e.g.,][]{BH08}. In order to understand the
98: behavior of the MRI under these conditions it is necessary to relax
99: the assumption of an inviscid plasma.
100:
101: A large fraction of the shearing box simulations addressing the
102: non-linear regime of the MRI have been carried out in the ideal MHD
103: limit, i.e., without including explicit dissipation in the codes
104: \citep[see, e.g.,][]{HGB95,Brandenburg95,Sanoetal04}. However, even
105: in the absence of explicit viscosity and resistivity, finite
106: difference discretization leads to numerical diffusion/dissipation.
107: Therefore, even in this type of simulations, it is necessary to
108: understand the impact of these numerical artifacts that lead to
109: departures from the ideal MHD regime and how similar they are when
110: compared with physical (resolved) dissipation.
111:
112: A handful of numerical studies with explicit resistivity but zero
113: physical viscosity have been carried out in order to understand the
114: effects of ohmic dissipation in the saturation of MRI-driven
115: turbulence \citep[see, e.g.,][]{SIM98, SI01, FSH00, Sanoetal04,
116: TSD07}. In particular, \citet{SS03} have shown that the saturation
117: level of the stresses increases with increasing magnetic Reynolds
118: number and seem to converge to an asymptotic value for magnetic
119: Reynolds numbers larger than unity.
120:
121: Recent work has pointed out problems with convergence in zero-net-flux
122: numerical simulations of ideal MHD driven by the MRI \citep{PCP07,
123: FPI07} implying the necessity of incorporating explicit dissipation
124: in the codes. Numerical studies with both resistivity and viscosity,
125: in the presence of a mean vertical magnetic field \citep{LL07} and in
126: the case of zero net flux \citep{FPII07}, have begun to uncover how
127: the characteristics of fully developed MRI-driven turbulence depends
128: on the Reynolds and magnetic Reynolds numbers. Even though the ranges
129: in Reynolds and magnetic Reynolds numbers that can be currently
130: addressed is still limited, the results obtained from the simulations
131: suggest that the magnetic and kinetic energies contained in turbulent
132: motions in the saturated regime depend on the values of the
133: microphysical viscosity and resistivity. In particular, the mean
134: angular momentum transport in the turbulent state increases with
135: increasing magnetic Prandtl number.
136:
137: From the experimental perspective, understanding the effects of
138: non-vanishing resistivity and viscosity in the behavior of the MRI
139: seems imperative, since the physical conditions achievable in the
140: laboratory depart significantly from the ideal MHD regime
141: \citep{JGK01, GJ02, Sisanetal04, LGJ06, SSW03}. Liquid metals (such
142: as sodium, gallium, and mercury) are often characterized by rather low
143: magnetic Prandtl numbers (${\rm Pm}\simeq 10^{-5}$--$10^{-7}$).
144: Although the regime of Reynolds numbers involved is still orders of
145: magnitude smaller than any astrophysical system with similar magnetic
146: Prandtl numbers, MRI experiments offer one of the few prospects of
147: studying anything close to MHD astrophysical processes in the
148: laboratory.
149:
150: A number of analyses addressing some aspects of the impact of
151: viscosity and resistivity on the MRI in various dissipative limits
152: appear scattered throughout the literature on theoretical, numerical,
153: and experimental MRI. More recently, \citet{LB07} have found
154: particular solutions of the viscous, resistive MHD equations
155: (including even a cooling term) in the shearing box approximation.
156: However, we are unaware of any comprehensive, systematic study
157: addressing how the MRI behaves in viscous, resistive, differentially
158: rotating magnetized plasmas for arbitrary combinations of the Reynolds
159: and magnetic Reynolds numbers. The aim of this work is to carry out
160: this analysis in detail.
161:
162: The rest of paper is organized as follows. In
163: \S~\ref{sec:assumptions}, we state our assumptions. In
164: \S~\ref{sec:solution}, we solve the eigenvalue problem defined by the
165: MRI for arbitrary Reynolds and magnetic Reynolds numbers. We provide
166: closed analytical expressions for the eigenfrequencies and the
167: associated eigenvectors. In \S~\ref{sec:mri_modes}, we address the
168: unexplored physical structure of MRI modes for finite Reynolds and
169: magnetic Reynolds numbers and derive simple analytical expressions
170: that describe these modes in various asymptotic regimes. In
171: \S~\ref{sec:stresses_energies}, we calculate the correlations between
172: magnetic and velocity MRI-driven perturbations that are related to
173: angular momentum transport and energy densities. We find that some key
174: results previously shown to hold in the ideal MHD limit \citep{PCP06}
175: are also valid in the non-ideal regime. In particular, we show that
176: even though the effectiveness with which the MRI disrupts the laminar
177: flow depends on the Reynolds and magnetic Reynolds numbers, the
178: instability always transports angular momentum outwards. We also find
179: that magnetic perturbations dominate both the energetics and the
180: transport of angular momentum for any combination of the Reynolds and
181: magnetic Reynolds numbers. In \S~\ref{sec:discussion} we summarize our
182: findings and discuss the implications of our study.
183:
184: \section{Assumptions}
185: \label{sec:assumptions}
186:
187: Let us consider a cylindrical, incompressible background characterized
188: by an angular velocity profile $\bb{\Omega}=\Omega(r)\check{\bb{z}}$,
189: threaded by a vertical magnetic field $\bb{\bar{B}} = \bar{B}_z
190: \check{\bb{z}}$. We work in the shearing box approximation, which
191: consist of a first order expansion in the variable $r-r_0$ of all the
192: quantities characterizing the flow at the fiducial radius $r_0$. The
193: goal of this expansion is to retain the most relevant terms governing
194: the dynamics of the MHD fluid in a locally-Cartesian coordinate system
195: co-orbiting and corrotating with the background flow with local
196: (Eulerian) velocity $\bb{v} = r_0\,\Omega_0 \check{\bb{\phi}}$. (For a
197: more detailed discussion on this expansion see \citealt{GX94} and
198: references therein.)
199:
200: The equations governing the dynamics of an incompressible MHD fluid
201: with constant kinematic viscosity $\nu$ and resistivity $\eta$ in the
202: shearing box limit are given by
203: \begin{eqnarray}
204: \label{eq:euler}
205: \D{t}{\bb{v}} + \left(\bb{v}\bcdot\del\right)\bb{v} & = &
206: - 2 \bb{\Omega}_0 \btimes \bb{v} \, +
207: \, q \Omega^2_0\del(r-r_0)^2 \nonumber \\
208: &-& \frac{1}{\rho}\del\left(P + \frac{\bb{B}^2}{8\pi}\right) +
209: \frac{(\bb{B}\bcdot\del)\bb{B}}{4\pi\rho} + \nu \del^2{\bb{v}} \,,
210: \nonumber \\ \\
211: %
212: \label{eq:induction}
213: \D{t}{\bb{B}} + \left( \bb{v} \bcdot \del \right)\bb{B}
214: & = & \left(\bb{B} \bcdot \del \right) \bb{v} + \eta \del^2{\bb{B}} \,,
215: \end{eqnarray}
216: where $P$ is the pressure, $\rho$ is the (constant) density, the
217: factor
218: \begin{eqnarray}
219: q\equiv-\left.\frac{d\ln\Omega}{d\ln r}\right|_{r_0} \,,
220: \end{eqnarray}
221: parametrizes the magnitude of the local shear, and we have defined the
222: (locally-Cartesian) differential operator
223: \begin{eqnarray}
224: \del & \equiv &
225: \check{\bb{r}} \, \frac{\partial}{\partial r} +
226: \frac{\check{\bb{\phi}}}{r_0}\,\frac{\partial}{\partial \phi} +
227: \check{\bb{z}} \, \frac{\partial}{\partial z} \,,
228: \end{eqnarray}
229: where $\check{\bb{r}}$, $\check{\bb{\phi}}$, and $\check{\bb{z}}$ are,
230: coordinate-independent, orthonormal vectors corrotating with the
231: background flow at $r_0$. The continuity equation reduces to
232: $\del \bcdot \bb{v}=0$ and there is no need for an equation of state
233: since the pressure can be determined from this condition.
234:
235: We focus our attention on the dynamics of perturbations that depend
236: only on the vertical coordinate. Under the current set of
237: assumptions, these types of perturbations are known to exhibit the
238: fastest growth rates in the ideal MHD case \citep{BH92, BH98, PP05}.
239: The equations governing the dynamics of these perturbations can be
240: obtained by noting that the velocity and magnetic fields given by
241: \begin{eqnarray}
242: \label{eq:mean_plus_perturbations_v}
243: \bb{v} &=&
244: \delta v_r(z) \check{\bb{r}} + [- q \Omega_0 (r-r_0)+ \delta v_\phi(z)]
245: \check{\bb{\phi}} + \delta v_z(z) \check{\bb{z}} \,, \\
246: \label{eq:mean_plus_perturbations_b}
247: \bb{B} &=& \delta B_r(z) \check{\bb{r}} +
248: \delta B_\phi(z) \check{\bb{\phi}} +
249: [\bar B_z + \delta B_z(z)] \check{\bb{z}} \,,
250: \end{eqnarray}
251: where the time dependence is implicit, constitute a family of exact,
252: non-linear, solutions to the viscous, resistive MHD equations
253: (\ref{eq:euler})-(\ref{eq:induction}). As noted in \citet{GX94}, even
254: in the dissipative case, the only non-linear terms, which are present
255: through the perturbed magnetic energy density, are irrelevant in the
256: case under consideration.
257:
258: We can further simplify
259: equations~(\ref{eq:euler})~and~(\ref{eq:induction}) by removing the
260: background shear flow\footnote{In the shearing box approximation, the
261: dependence of the background flow on the radial coordinate is
262: strictly linear and therefore viscous dissipation does not affect
263: its dynamics.} $\bb{v}_{\rm shear} = - q \Omega_0 (r-r_0)
264: \check{\bb{\phi}}$ and by realizing that we can take $\delta
265: v_z(z)=\delta B_z(z)=0$ without loss of generality. We then obtain
266: \begin{eqnarray}
267: \label{eq:vx}
268: \frac{\partial}{\partial t} \delta v_r
269: &=& 2 \Omega_0 \delta v_\phi + \frac{\bar B_z}{4\pi\rho} \,
270: \frac{\partial}{\partial z} \delta B_r + \nu \frac{\partial^2}{\partial z^2} \delta v_r \,, \\
271: \label{eq:vy}
272: \frac{\partial}{\partial t} \delta v_\phi
273: &=& - (2-q)\Omega_0 \delta v_r + \frac{\bar B_z}{4\pi\rho} \,
274: \frac{\partial}{\partial z} \delta B_\phi + \nu
275: \frac{\partial^2}{\partial z^2} \delta v_\phi \,, \\
276: \label{eq:bx}
277: \frac{\partial}{\partial t} \delta B_r &=& \bar B_z
278: \frac{\partial}{\partial z} \delta v_r + \eta \frac{\partial^2}{\partial z^2} \delta B_r\,, \\
279: \label{eq:by}
280: \frac{\partial}{\partial t} \delta B_\phi &=& - q \Omega_0 \delta B_r +
281: \bar B_z \frac{\partial}{\partial z} \delta v_\phi + \eta \frac{\partial^2}{\partial z^2} \delta B_{\phi}\,,
282: \end{eqnarray}
283: where the first term on the right hand side of equation (\ref{eq:vy})
284: is related to the epicyclic frequency
285: \begin{eqnarray}
286: \kappa\equiv\sqrt{2(2-q)}\, \Omega_0 \,,
287: \end{eqnarray}
288: at which the flow variables oscillate in a perturbed hydrodynamic
289: disk. For Keplerian rotation the parameter is $q=3/2$ and thus the
290: epicyclic frequency is $\kappa=\Omega_0$.
291:
292: It is convenient to define the new variables $\delta b_i \equiv \delta
293: B_i/\sqrt{4\pi\rho}$ for $i=r,\phi$, and introduce dimensionless
294: quantities by considering the characteristic time- and length-scales
295: set by $1/\Omega_0$ and $\bar{B}_z/(\sqrt{4\pi\rho}\,\Omega_0)$. The
296: equations satisfied by the dimensionless perturbations, $\delta
297: \tilde{v}_i$, $\delta \tilde{b}_i$, are then given by
298: \begin{eqnarray}
299: \label{eq:vx_nodim}
300: \partial_{\tilde t} \delta \tilde{v}_r &=& 2 \delta \tilde{v}_\phi +
301: \partial_{\tilde z} \delta \tilde{b}_r + \tilde{\nu} \partial_{\tilde z}^2 \delta \tilde{v}_r\,, \\
302: \label{eq:vy_nodim}
303: \partial_{\tilde t} \delta \tilde{v}_\phi &=& - (2-q) \delta \tilde{v}_r +
304: \partial_{\tilde z} \delta \tilde{b}_\phi + \tilde{\nu} \partial_{\tilde z}^2 \delta \tilde{v}_\phi \,, \\
305: \label{eq:bx_nodim}
306: \partial_{\tilde t} \delta \tilde{b}_r &=&
307: \partial_{\tilde z} \delta \tilde{v}_r + \tilde{\eta} \partial_{\tilde z}^2 \delta \tilde{b}_r\,, \\
308: \label{eq:by_nodim}
309: \partial_{\tilde t} \delta \tilde{b}_\phi &=& - q \delta \tilde{b}_r +
310: \partial_{\tilde z} \delta \tilde{v}_\phi + \tilde{\eta} \partial_{\tilde z}^2 \delta \tilde{b}_\phi\,,
311: \end{eqnarray}
312: where $\tilde t$ and $\tilde z$ denote the dimensionless time and
313: vertical coordinate, respectively.
314:
315: The dynamics of ideal MRI modes, with $\nu=\eta=0$, is completely
316: determined by the dimensionless shear $q$ \citep{PCP06}. The effects
317: of viscous and resistive dissipation introduce two new dimensionless
318: quantities that alter the characteristics and evolution of the MRI.
319: With our choice of characteristic scales, it is natural to define the
320: Reynolds and magnetic Reynolds numbers characterizing the MHD flow as
321: \footnote{In a compressible fluid, the sound speed, $c_{\rm s}$,
322: provides another natural characteristic speed to define the Reynolds
323: and magnetic Reynolds numbers. These definitions, e.g., ${\rm Re}'$
324: and ${\rm Rm}'$, are related to those provided in equations
325: (\ref{eq:re_number}) and (\ref{eq:rm_number}) via the plasma beta
326: parameter $\beta=(2/\Gamma) (c_{\rm s}/\bar{v}_{{\rm A}z})^2$, with
327: $\bar{v}_{{\rm A}z}= \bar{B}_z/\sqrt{4\pi\rho}$ the Alfv\'en speed
328: in the $z$ direction, simply by ${\rm Re}' = (\Gamma\beta/2) {\rm
329: Re}$ and ${\rm Rm}' = (\Gamma\beta/2) {\rm Rm}$ for a polytropic
330: equation of state $P=K \rho^\Gamma$, with $K$ and $\Gamma$
331: constants.}
332: \begin{eqnarray}
333: \label{eq:re_number}
334: {\rm Re} &\equiv& \frac{v_{{\rm A}z}^2}{\nu\Omega_0} = \frac{1}{\tilde\nu} \,, \\
335: \label{eq:rm_number}
336: {\rm Rm} &\equiv& \frac{v_{{\rm A}z}^2}{\eta\Omega_0} = \frac{1}{\tilde\eta} \,,
337: \end{eqnarray}
338: with associated magnetic Prandtl number
339: \begin{equation}
340: \label{eq:pm_number}
341: {\rm Pm} \equiv \frac{{\rm Rm}}{{\rm Re}}=\frac{\tilde\nu}{\tilde\eta} = \frac{\nu}{\eta} \,.
342: \end{equation}
343:
344: In order to simplify the notation, we drop hereafter the tilde
345: denoting the dimensionless quantities. In the rest of the paper, all
346: the variables are to be regarded as dimensionless, unless otherwise
347: specified.
348:
349:
350:
351: \section{The Eigenvalue Problem for the Non-ideal MRI\,:\\
352: A Formal Analytical Solution}
353: \label{sec:solution}
354:
355: In this section we provide a complete analytical solution to the set
356: of equations (\ref{eq:vx_nodim})--(\ref{eq:by_nodim}) as a function of
357: the shear parameter $q$, (or, equivalently, the epicyclic frequency,
358: $\kappa$) for any set of values $(\nu,\eta)$ defining the viscosity and
359: resistivity.
360:
361: It is convenient to work in Fourier space, as this provides the
362: advantage of obtaining explicitly the basis of modes that is needed to
363: construct the most general solution satisfying equations
364: (\ref{eq:vx_nodim})--(\ref{eq:by_nodim}). Taking the Fourier transform
365: of this set with respect to the $z$-coordinate, we obtain the matrix
366: equation
367: \begin{equation}
368: \partial_t \hat{\bb{\delta}}(k_n, t) = L \, \hat{\bb{\delta}}(k_n,t) \,,
369: \label{eq:matrix_form}
370: \end{equation}
371: where the vector $\hat{\bb{\delta}}(k_n, t)$ stands for
372: \begin{equation}
373: \hat{\bb{\delta}}(k_n, t) =
374: \left[\begin{array}{c}
375: \hat{\delta v_r}(k_n,t) \\ \hat{\delta v_\phi}(k_n,t) \\
376: \hat{\delta b_r}(k_n,t) \\ \hat{\delta b_\phi}(k_n,t)
377: \end{array}\right] \,,
378: \end{equation}
379: and $L$ represents the matrix
380: \begin{equation}
381: \label{eq:matrix_L}
382: L =
383: \left[\begin{array}{cccc}
384: -\nu k_n^2 & 2 & i k_n & 0 \\
385: -(2-q) & -\nu k_n^2 & 0 & i k_n \\
386: i k_n & 0 & -\eta k_n^2 & 0 \\
387: 0 & i k_n & -q & -\eta k_n^2
388: \end{array}\right] \,.
389: \end{equation}
390: The functions denoted by $\hat f(k_n, t)$ correspond to the Fourier
391: transform of the real functions, $f(z,t)$, and are defined via
392: \begin{equation}
393: \label{eq:ft_discrete}
394: \hat f(k_n,t) \equiv \frac{1}{2H}\int_{-H}^{H} f(z,t) \, e^{-ik_nz}
395: \,dz \,,
396: \end{equation}
397: where we have assumed periodic boundary conditions at $z=\pm H$, with
398: $H$ being the (dimensionless) scale-height and $k_n$ the wavenumber in
399: the $z$-coordinate,
400: \begin{equation}
401: \label{eq:invft_discrete}
402: k_n \equiv \frac{n\pi}{H} \,,
403: \end{equation}
404: where $n$ is an integer number. In order to simplify the notation,
405: hereafter we denote these wavenumbers simply by $k$.
406:
407:
408: \begin{figure*}[t]
409: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gamma_eta.eps}
410: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gamma_mu.eps}
411: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gamma_nu.eps}
412: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f1a.eps}
413: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f1b.eps}
414: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f1c.eps}
415: \caption{Growth rates $\gamma_+$,
416: eq.~(\ref{eq:eigenvalues_gamma_pm}), as a function of the vertical
417: wavenumber $k$ for different combinations of Reynolds and magnetic
418: Reynolds numbers for Keplerian rotation. In all three panels, the
419: thick solid line corresponds to the ideal MHD limit, i.e., ${\rm
420: Re}, {\rm Rm}\rightarrow\infty$. For any combination of the
421: Reynolds and magnetic Reynolds numbers, the growth rate has a well
422: defined, single maximum $\gamma_{\rm max}$ that corresponds to the
423: most unstable mode $k_{\rm max}$. The range of unstable modes,
424: $0<k<k_{\rm c}$, is always finite, the critical wavenumber $k_{\rm
425: c}$ satisfies eq.~(\ref{eq:dispersion_relation_nu_eta}) when
426: $\sigma\equiv 0$, see \S~\ref{sub:modes_marginal_fastest}.
427: \emph{Left}: Growth rate $\gamma_+$ for different values of the
428: magnetic Reynolds number in the inviscid limit, i.e., ${\rm
429: Re}\rightarrow \infty$. The thin solid lines, in decreasing
430: order, correspond to ${\rm Rm}=10,1,0.1$. \emph{Middle}: Growth
431: rate $\gamma_+$ for magnetic Prandtl number ${\rm Pm}={\rm
432: Rm}/{\rm Re}=1$. The thin solid lines, in decreasing order,
433: correspond to ${\rm Re}={\rm Rm}=10,1,0.1$. In all of the cases
434: shown in the left and middle panels, the critical wavenumber,
435: $k_{\rm c}$, below which unstable modes can exist decreases with
436: increasing resistivity, see \S~\ref{sub:modes_re_gg_rm} and
437: \S~\ref{sub:modes_re_eq_rm_ll_1} for analytic expressions of these
438: marginally stable modes. \emph{Right}: Growth rate $\gamma_+$ for
439: different values of the Reynolds number in the ideal conductor
440: limit, i.e., ${\rm Rm}\rightarrow \infty$. The various curves, in
441: decreasing order, correspond to ${\rm Re}=10,1,\ldots,10^{-3}$.
442: In this case, the range of unstable modes is insensitive to the
443: Reynolds number, all the modes with wavenumbers shorter than
444: $k_{\rm c}=\sqrt{2q}$ are unstable, see \S~\ref{sub:modes_ideal}.
445: It is evident that the growth rates and the characteristic scales,
446: both $k_{\rm max}$ and $k_{\rm c}$, are more sensitive to changes
447: in the resistivity than to changes in the viscosity. The
448: simultaneous analysis of all three panels leads to the conclusion
449: that viscous, resistive modes with magnetic Prandtl number equal
450: to unity resemble more closely inviscid, resistive modes rather
451: than viscous, conductive ones, see
452: \S~\ref{sub:modes_re_eq_rm_ll_1} for the explanation of this
453: behavior.}
454: \label{fig:growths}
455: \end{figure*}
456:
457: In order to solve the matrix equation (\ref{eq:matrix_form}), it is
458: convenient to find the eigenvector basis, $\{\mathbf{e}_j\}$ with
459: $j=1,2,3,4$, in which $L$ is diagonal. This basis exists for all
460: values of the wavenumber $k$ (i.e., the rank of the matrix $L$ is
461: equal to 4, the dimension of the complex space) with the possible
462: exception of a finite number of values of $k$. In this basis, the
463: action of $L$ over the set $\{\mathbf{e}_j\}$ is equivalent to a
464: scalar multiplication, i.e.,
465: \begin{equation}
466: L_{\rm diag} \, \mathbf{e}_{j} = \sigma_j \,\mathbf{e}_{j} \quad
467: \textrm{for} \quad j=1,2,3,4 \,,
468: \end{equation}
469: where $\{\sigma_j\}$ are complex scalars.
470:
471:
472: \subsection{Eigenvalues}
473:
474: In the eigenvector basis, the matrix $L$ has a diagonal
475: representation $L_{\rm diag}$ = diag$(\sigma_1, \sigma_2, \sigma_3,
476: \sigma_4)$. The eigenvalues $\{\sigma_j\}$, with $j=1,2,3,4$, are the
477: roots of the characteristic polynomial associated with $L$, i.e., the
478: dispersion relation associated with the non-ideal MRI, which can be
479: written in compact form as
480: \begin{equation}
481: \label{eq:dispersion_relation_nu_eta}
482: (k^2 + \sigma_\nu \sigma_\eta)^2 + \kappa^2 (k^2 + \sigma_\eta^2) -
483: 4 k^2 = 0 \,,
484: \end{equation}
485: where we have defined the quantities
486: \begin{eqnarray}
487: \label{eq:sigma_nu}
488: \sigma_\nu &\equiv& \sigma + \nu k^2 \,, \\
489: \label{eq:sigma_eta}
490: \sigma_\eta &\equiv& \sigma + \eta k^2 \,.
491: \end{eqnarray}
492: The dispersion relation (\ref{eq:dispersion_relation_nu_eta}) is a
493: fourth order polynomial with non-zero coefficients in $\sigma$ and
494: $\sigma^3$. In order to find its roots it is convenient to take this
495: polynomial to its depressed form. This can be achieved by defining the
496: new variables $\sigma_\mu$ and $\mu$ such that\footnote{A physical
497: interpretation of the variable $\mu$ is provided in
498: \S~\ref{sub:geometrical_representation}.}
499: \begin{eqnarray}
500: \label{eq:sigma_mu}
501: \sigma_\mu &\equiv& \frac{1}{2}(\sigma_\nu +\sigma_\eta) \,, \\
502: \label{eq:mu}
503: \mu &\equiv& \frac{1}{2}(\nu-\eta)k^2 \,.
504: \end{eqnarray}
505: The resulting polynomial can then be written as
506: \begin{eqnarray}
507: \label{eq:dispersion_relation_mu}
508: \sigma_\mu^4 + \alpha \sigma_\mu^2 +\beta \sigma_\mu + \lambda = 0 \,,
509: \end{eqnarray}
510: where the coefficients $\alpha$, $\beta$, and $\lambda$ are given by
511: \begin{eqnarray}
512: \alpha &\equiv& 2(k^2-\mu^2) + \kappa^2 \,, \\
513: \beta &\equiv& -2\mu\kappa^2 \,, \\
514: \lambda &\equiv& (k^2-\mu^2)^2 + \kappa^2(k^2 + \mu^2) - 4k^2 \,.
515: \end{eqnarray}
516:
517:
518: \begin{figure*}[t]
519: % \includegraphics[width=\columnwidth,trim=0 0 0 0]{geo_general.eps}
520: % \includegraphics[width=\columnwidth,trim=0 0 0 0]{geo_var.eps}
521: \includegraphics[width=\columnwidth,trim=0 0 0 0]{f2a.eps}
522: \includegraphics[width=\columnwidth,trim=0 0 0 0]{f2b.eps}
523: \caption{\emph{Left}: Geometrical representation of the velocity
524: field (black) and magnetic field (gray) perturbations for viscous,
525: resistive MRI modes. Note that this is a projection of a single
526: mode, which is inherently three-dimensional, onto the disk
527: mid-plane $(r,\phi,z=0)$. The velocity and magnetic field
528: components are always out of phase in the vertical direction $z$
529: by $\pi/4$, see eq.~(\ref{eq:delta_zt_unstable}). The angles
530: $\theta_{\rm v}$ and $\theta_{\rm b}$, defined in
531: eqs.~(\ref{eq:theta_v})~and~(\ref{eq:theta_b}), respectively,
532: correspond to the physical angles defining the planes
533: (perpendicular to the disk midplane) containing the MRI-driven
534: perturbations, see
535: eqs.~(\ref{eq:tan_theta_v})~and~(\ref{eq:tan_theta_b}). The
536: relative magnitude of velocity and magnetic field perturbations is
537: determined by eq.~(\ref{eq:b0_v0}). \emph{Right}: Evolution of
538: the geometrical representation of the fastest-growing, non-ideal
539: MRI mode, with associated wavenumber $k_{\rm max}$, with magnetic
540: Prandtl number equal to unity, as a function of the
541: Reynolds/magnetic Reynolds number. When the Reynolds/magnetic
542: Reynolds number varies according to ${\rm Re}={\rm
543: Rm}:\infty\rightarrow 0$, the angles evolve according to
544: $\theta_{\rm v}:\pi/4\rightarrow 0$ and $\theta_{\rm
545: b}:3\pi/4\rightarrow \pi/2$ and the relative amplitude of the
546: perturbations evolves according to $b_0/v_0:5/3\rightarrow
547: \infty$. Note that the velocity and magnetic field perturbations
548: are always orthogonal for ${\rm Pm}=1$, see
549: eq.~(\ref{eq:theta_bv}).}
550: \label{fig:geom_def}
551: \end{figure*}
552:
553:
554: The solutions to equation (\ref{eq:dispersion_relation_mu}) are
555: \begin{eqnarray}
556: \sigma_\mu = \pm_a (-\Lambda \mp_b \sqrt{\Delta})^{1/2} \pm_b \frac{\beta}{4\sqrt{\Delta}} \,,
557: \end{eqnarray}
558: where the subscripts $a$ and $b$ in the ``$+$'' and ``$-$'' signs
559: label the four possible combinations of signs and we have defined the
560: quantities\footnote{Defining the quantities $\Lambda$ and $\Delta$ in
561: this way allows us to show explicitly that in the limit $\nu, \eta
562: \rightarrow 0$ the solutions to equation
563: (\ref{eq:dispersion_relation_nu_eta}) converge smoothly to the
564: solutions found in the ideal MHD case (see Appendix
565: \ref{sec:appendix}).}
566: \begin{eqnarray}
567: \Lambda &=& \frac{3\alpha}{4} + \frac{y}{2} \,, \\
568: \Delta &=& (y + \alpha)^2 - \lambda \,,
569: \end{eqnarray}
570: and $y$ is any of the solutions to the cubic equation
571: \begin{equation}
572: y^3 + \frac{5\alpha}{2} y^2 + (2\alpha^2 - \lambda) y +
573: \left(\frac{\alpha^3}{2} - \frac{\alpha\lambda}{2} -
574: \frac{\beta^2}{8}\right) = 0 \,,
575: \end{equation}
576: which has closed analytic solutions
577: \begin{equation}
578: y = -\frac{5}{6} \alpha + \frac{1}{3}\frac{P}{U} -U \,,
579: \end{equation}
580: with
581: \begin{eqnarray}
582: P &=& -\frac{\alpha^2}{12}-\lambda \,, \\
583: Q &=& -\frac{\alpha^3}{108}+\frac{\alpha\lambda}{3}-\frac{\beta^2}{8} \,, \\
584: U &=& \left(\frac{Q}{2}\pm \sqrt{\frac{Q^2}{4}+\frac{P^3}{27}}\right)^{1/3} \,.
585: \end{eqnarray}
586: Note that the choice of either sign in $U$ is immaterial.
587:
588: It is now trivial to write the solutions of the dispersion relation
589: (\ref{eq:dispersion_relation_nu_eta}), $\sigma$, in terms of the
590: variable $\sigma_\mu$. Using equations
591: (\ref{eq:sigma_nu})--(\ref{eq:sigma_mu}) we obtain
592: \begin{eqnarray}
593: \sigma = \sigma_\mu - \frac{1}{2}(\nu+\eta)k^2 \,.
594: \end{eqnarray}
595: The eigenfrequencies of the viscous, resistive MRI modes
596: are then given by
597: \begin{eqnarray}
598: \label{eq:eigenvalues_sigma}
599: \sigma &=& \pm_a (-\Lambda \mp_b \sqrt{\Delta})^{1/2} \nonumber \\
600: &-&\frac{\nu}{2} k^2 \left(1 \pm_b \frac{\kappa^2}{2\sqrt{\Delta}} \right)
601: -\frac{\eta}{2} k^2 \left(1 \mp_b \frac{\kappa^2}{2\sqrt{\Delta}}
602: \right) \,.
603: \end{eqnarray}
604:
605: \subsection{Four classes of solutions}
606: \label{sub:classes_of_solutions}
607:
608: All of the quantities $\Lambda$, $\Delta$, and $y$, depend on the
609: viscosity $\nu$ and the resistivity $\eta$ only through $\mu^2 \propto
610: (\nu-\eta)^2$. This has a series of important implications, in
611: particular, there is always a range of wavenumbers for which the
612: discriminant in equation (\ref{eq:eigenvalues_sigma}) is positive,
613: i.e., $\sqrt{\Delta} - \Lambda > 0$. It can also be seen that the
614: last two terms between parentheses in equation
615: (\ref{eq:eigenvalues_sigma}) are always positive, i.e., $\sqrt{\Delta}
616: \ge \kappa^2/2$, and thus they always produce damping. Because of
617: this, we can classify the modes in four types: two (damped) growing
618: and decaying ``unstable'' modes with eigenvalues
619: \begin{eqnarray}
620: \label{eq:eigenvalues_gamma_pm}
621: \gamma_{\pm} &=& \pm (\sqrt{\Delta}-\Lambda)^{1/2} \nonumber \\
622: &-&\frac{\nu}{2} k^2 \left(1 - \frac{\kappa^2}{2\sqrt{\Delta}} \right)
623: -\frac{\eta}{2} k^2 \left(1 + \frac{\kappa^2}{2\sqrt{\Delta}}
624: \right) \,,
625: \end{eqnarray}
626: and two (damped) ``oscillatory'' modes with eigenvalues
627: \begin{eqnarray}
628: \label{eq:eigenvalues_omega_pm}
629: i \omega_{\pm} &=& \pm i (\sqrt{\Delta}+\Lambda)^{1/2} \nonumber \\
630: &-&\frac{\nu}{2} k^2 \left(1 + \frac{\kappa^2}{2\sqrt{\Delta}} \right)
631: -\frac{\eta}{2} k^2 \left(1 - \frac{\kappa^2}{2\sqrt{\Delta}}
632: \right) \,.
633: \end{eqnarray}
634: We arbitrarily label these eigenvalues as
635: \begin{eqnarray}
636: \label{eq:eigenvalues_all}
637: \sigma_{1} \equiv \gamma_{+} \,, \quad
638: %
639: \sigma_{2} \equiv \gamma_{-} \,, \quad
640: %
641: \sigma_{3} \equiv i \omega_{+} \,, \quad
642: %
643: \sigma_{4} \equiv i \omega_{-} \,.
644: \end{eqnarray}
645:
646: \begin{figure*}[t]
647: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kc_nu.eps}
648: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kc_mu.eps}
649: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kc_eta.eps}
650: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f3a.eps}
651: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f3b.eps}
652: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f3c.eps}
653: \caption{Critical wavenumber $k_{\rm c}$, see eq.~(\ref{eq:k_c}),
654: corresponding to the marginally stable MRI mode for Keplerian
655: rotation in different dissipative regimes. The horizontal lines
656: at $k_{\rm c}=\sqrt{3}$ represent the ideal MHD limit,
657: eq.~(\ref{eq:k_c_ideal}). \emph{Left}: Critical wavenumber
658: $k_{\rm c}$ as a function of the magnetic Reynolds number for
659: different values of the Reynolds number. The thick solid line
660: denotes the inviscid limit, i.e., ${\rm Re}\rightarrow \infty$.
661: Note that eq.~(\ref{eq:k_c_lim_reggrm}) describes this curve
662: exactly. The thin solid lines, in decreasing order, correspond to
663: ${\rm Re} = 1, 0.1, \ldots$. For small magnetic Reynolds number
664: $k_{\rm c} \propto {\rm Rm}$, see eq.~(\ref{eq:k_c_lim_reggrm}).
665: For finite Reynolds numbers, such that ${\rm Re\,Rm} \lesssim 1$
666: there is a transition between the regimes ${\rm Rm}\ll 1$ and
667: ${\rm Rm}\gg 1$ such that $k_{\rm c} \propto {\rm Rm}^{1/3}$, see
668: eq.~(\ref{eq:k_c_lim_rellrm}). \emph{Middle}: Critical wavenumber
669: $k_{\rm c}$ as a function of the Reynolds number for different
670: magnetic Prandtl numbers. ${\rm Pm}$ increases/decreases by an
671: order of magnitude for each curve to the left/right of the thick
672: solid line denoting the ${\rm Pm}=1$ case. The dashed lines
673: $k_{\rm c} \propto {\rm Re\,Pm} = {\rm Rm}$ are calculated
674: according to eq.~(\ref{eq:k_c_lim_reggrm}), which gives the
675: correct result even for ${\rm Pm} > 1$, provided that the Reynolds
676: number is sufficiently small. The dotted lines $k_{\rm c} \propto
677: ({\rm Re\,Rm})^{1/3}$ are calculated according to eq.~
678: (\ref{eq:k_c_lim_rellrm}). \emph{Right}: Critical wavenumber
679: $k_{\rm c}$ as a function of the Reynolds number for different
680: values of the magnetic Reynolds number. The thick solid line
681: corresponds to the ideal conductor limit, i.e., ${\rm
682: Rm}\rightarrow \infty$. The thin solid lines, in decreasing
683: order, correspond to ${\rm Re} =10^5, 10^4, \ldots, 10^{-3}$. For
684: small Reynolds numbers $k_{\rm c} \propto {\rm Re}^{1/3}$, see
685: eq.~(\ref{eq:k_c_lim_rellrm}), while for Reynolds numbers larger
686: than a few, the critical wavenumber is independent of ${\rm Re}$
687: regardless of the value of ${\rm Rm}$.}
688: \label{fig:k_c}
689: \end{figure*}
690:
691: Figure~\ref{fig:growths} shows the growth rate $\gamma_+$ as a
692: function of the vertical wavenumber $k$ for different combinations of
693: the Reynolds and magnetic Reynolds numbers for Keplerian rotation.
694: These growth rates are more sensitive to changes in the resistivity
695: than to changes in the viscosity. A qualitative understanding of this
696: behavior can be obtained by realizing that viscosity tends to quench
697: the instability, without altering the large scale magnetic field.
698: Thus, as long as the resistivity is negligible, the range of unstable
699: lenghtscales are the same in both ideal and viscous, perfectly
700: conducting fluids. On the other hand, resistivity tends to destroy
701: the magnetic field at small scales having a stronger impact on the
702: stability of the perturbations at these scales.
703:
704: Mathematically, the asymmetric response of the growth rate to changes
705: in the viscosity $\nu$ or the resistivity $\eta$ originates in the
706: different functional form of the terms that contribute to produce
707: damping, i.e., $(1-\kappa^2/2\sqrt{\Delta})$ and
708: $(1+\kappa^2/2\sqrt{\Delta})$, in the exponential growth characterized
709: by $\gamma_+$ in equation (\ref{eq:eigenvalues_gamma_pm}). If the
710: oscillatory modes, $\omega_{\pm}$ in equation
711: (\ref{eq:eigenvalues_omega_pm}), are considered instead, the roles of
712: the plus and minus signs in these terms are interchanged. From this
713: analysis we can infer that the ``oscillatory'' mode is affected
714: (damped) more strongly by viscosity than by resistivity.
715:
716: The simultaneous analysis of the various panels in
717: Figure~\ref{fig:growths} leads to the conclusion that viscous,
718: resistive unstable modes with magnetic Prandtl number equal to unity
719: resemble more closely inviscid, resistive modes rather than viscous,
720: conductive ones. In \S~\ref{sec:mri_modes} we provide analytical
721: expressions to support this conclusion.
722:
723: \subsection{Normalized Eigenvectors: Geometrical Representation}
724: \label{sub:geometrical_representation}
725:
726: The set of normalized eigenvectors, $\{\mathbf{e}_{\sigma_j}\}$,
727: associated with the eigenvalues (\ref{eq:eigenvalues_all}) are given by
728: \begin{eqnarray}
729: \label{eq:eigenvectors_initial}
730: \mathbf{e}_{\sigma_j} &\equiv& \frac{\mathbf{e}_{j}}{\|\mathbf{e}_{j}\|}
731: \quad \textrm{for} \quad j=1,2,3,4 \,,
732: \end{eqnarray}
733: where
734: \begin{equation}
735: \label{eq:e_sigma_j}
736: \mathbf{e}_{j}(k) =
737: \left[\begin{array}{c}
738: \sigma_{\eta j} \\
739: (k^2 + \sigma_{\nu j} \sigma_{\eta j})/2 \\
740: i k \\
741: - i k [2 \sigma_{\eta j} + q (\nu - \eta)k^2
742: ]/(k^2 + \sigma_{\nu j} \sigma_{\eta j})
743: \end{array}\right],
744: \end{equation}
745: $\sigma_{\nu j} = \sigma_j + \nu k^2$, $\sigma_{\eta j} = \sigma_j +
746: \eta k^2$, and the norms are given by
747: \begin{equation}
748: \|\mathbf{e}_{j}\|
749: \equiv \left[\sum_{l=1}^{4}
750: {\rm e}^{l}_{j}
751: {\rm e}^{l*}_{j}\right]^{1/2} \,.
752: \end{equation}
753: Here, ${\rm e}^{l}_{j}$ is the $l$-th component of the (unnormalized)
754: eigenvector associated with the eigenvalue $\sigma_j$.
755:
756: The set of four eigenvectors $\{\mathbf{e}_{\sigma_j}\}$, together
757: with the set of complex scalars $\{\sigma_j\}$ in equation
758: (\ref{eq:eigenvalues_all}), constitute the full solution to the
759: eigenvalue problem defined by the MRI for any combination of the
760: Reynolds and magnetic Reynolds numbers.
761:
762: A geometrical representation of the eigenvectors
763: (\ref{eq:eigenvectors_initial}) can be brought to light by defining
764: the angles $\theta_{{\rm v}j}$ and $\theta_{{\rm b}j}$ according to
765: \begin{eqnarray}
766: \label{eq:theta_v}
767: \tan \theta_{{\rm v}j} &\equiv& \frac{e^2_j}{e^1_j}
768: = \frac{k^2+ \sigma_{\nu j} \sigma_{\eta j}}{2\sigma_{\eta j}} \,, \\
769: \label{eq:theta_b}
770: \tan \theta_{{\rm b}j} &\equiv& \frac{e^4_j}{e^3_j}
771: = - \frac{2\sigma_{\eta j} + q (\nu-\eta)k^2}{k^2+ \sigma_{\nu j} \sigma_{\eta j}}\,.
772: \end{eqnarray}
773: It is important to remark that each of the four eigenvectors define,
774: in principle, four sets of angles $\{\theta_{{\rm v}j},\theta_{{\rm
775: b}j}\}$, for $j=1,2,3,4$. We label the angles associated with the
776: different types of modes discussed in
777: \S~\ref{sub:classes_of_solutions} according to
778: \begin{eqnarray}
779: \label{eq:theta_all}
780: \theta_{{\rm v}1} \equiv \theta_{\rm v}^{\gamma_+}, \quad
781: %
782: \theta_{{\rm v}2} \equiv \theta_{\rm v}^{\gamma_-}, \quad
783: %
784: \theta_{{\rm v}3} \equiv \theta_{\rm v}^{\omega_+}, \quad
785: %
786: \theta_{{\rm v}4} \equiv \theta_{\rm v}^{\omega_-}, \,\,\,\,\,
787: \end{eqnarray}
788: with similar definitions corresponding to $\theta_{{\rm b}j}$ for
789: $j=1,2,3,4$. Note that these angles are defined in \emph{spectral}
790: space and depend, in general, on the wavenumber $k$, the epicyclic
791: frequency, $\kappa$, the viscosity $\nu$, and the resistivity $\eta$.
792: The angles associated with the modes labeled by $\gamma_+$ and
793: $\gamma_-$ are always real while the ones associated with the modes
794: $\omega_+$ and $\omega_-$ are in general complex. For the sake of
795: brevity, in what follows we will refer to the set of angles describing
796: unstable MRI modes $\{\theta_{\rm v}^{\gamma_+}, \theta_{\rm
797: b}^{\gamma_+}\}$ simply as $\{\theta_{\rm v}, \theta_{\rm b}\}$.
798:
799:
800:
801: \begin{figure*}[t]
802: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kmax_nu.eps}
803: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kmax_mu.eps}
804: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{kmax_eta.eps}
805: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f4a.eps}
806: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f4b.eps}
807: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f4c.eps}
808: \caption{Wavenumber $k_{\rm max}$ corresponding to the fastest
809: growing non-ideal MRI modes for Keplerian rotation in different
810: dissipative regimes. The dot-dashed horizontal lines at $k_{\rm
811: max}=\sqrt{15/16}$ represent the ideal MHD limit,
812: eq.~(\ref{eq:k_max_ideal}). \emph{Left}: Fastest growing mode
813: $k_{\rm max}$ as a function of the magnetic Reynolds number for
814: different values of the Reynolds number. The thick solid line
815: denotes the inviscid limit, i.e., ${\rm Re}\rightarrow \infty$.
816: The thin solid lines, in decreasing order, correspond to ${\rm Re}
817: =1, 0.1, \ldots, 10^{-5}$. For magnetic Reynolds numbers larger
818: than unity, this wavenumber is independent of ${\rm Rm}$
819: regardless of the value of ${\rm Re}$. The dashed line,
820: calculated according to eq.~(\ref{eq:k_max_lim_reggrm}), provides
821: the correct asymptotic limit $k_{\rm max} \propto {\rm Rm}$ for
822: small magnetic Reynolds numbers. \emph{Middle}: Fastest growing
823: mode $k_{\rm max}$ as a function of the Reynolds number for
824: different values of the magnetic Prandtl number. From left to
825: right, the curves correspond to ${\rm Pm}=10^3, 10^2, \ldots, 1$
826: (thick solid line), $\ldots, 10^{-6}$. The dashed lines $k_{\rm
827: max} \propto {\rm Re\,Pm}={\rm Rm}$ are calculated according to
828: eq.~(\ref{eq:k_max_lim_reggrm}), which leads to the correct result
829: even for ${\rm Pm} \gtrsim 1$ provided that the Reynolds number is
830: sufficiently small. The dotted line $k_{\rm max} \propto {\rm
831: Re}^{1/2}$ results from eq.~(\ref{eq:k_max_lim_rellrm}).
832: \emph{Right}: Fastest growing mode $k_{\rm max}$ as a function of
833: the Reynolds number for different values of the magnetic Reynolds
834: number. The thick solid line corresponds to the ideal conductor
835: limit, i.e., ${\rm Rm}\rightarrow \infty$. The thin solid lines,
836: in decreasing order, correspond to ${\rm Rm} = 1, 0.1, 0.01$. For
837: Reynolds numbers larger than unity, the growth rate is independent
838: of ${\rm Re}$ regardless of the value of ${\rm Rm}$. The dotted
839: line, calculated according to eq.~(\ref{eq:k_max_lim_rellrm}),
840: provides the correct asymptotic limit $k_{\rm max} \propto {\rm
841: Re}^{1/2}$ for small Reynolds number.}
842: \label{fig:k_max}
843: \end{figure*}
844:
845:
846:
847: A normalized version of the MRI eigenvectors can be obtained by
848: multiplying the set of vectors in equation (\ref{eq:e_sigma_j}) by the
849: amplitudes
850: \begin{eqnarray}
851: A_j \equiv\sqrt{\frac{2}{q(k^2+\sigma_{\eta j}^2)}} \, \frac{v_0}{\sqrt{v_0^2+b_{0}^2}} \,.
852: \end{eqnarray}
853: where we have defined
854: \begin{eqnarray}
855: \label{eq:b0_v0}
856: b_{0}\equiv \frac{2k v_0}{k^2+\sigma_{\nu j} \sigma_{\eta j}} \, \left(1 +
857: \frac{(\nu-\eta)[4\sigma_{\eta j} +
858: q(\nu-\eta)k^2]k^2}{2(k^2+\sigma_{\eta j}^2)}\right)^{1/2} \,,
859: \nonumber \\
860: \end{eqnarray}
861: where, for the sake of simplicity, we have omitted the subscript $j$
862: on the left hand side. The expressions for the normalized eigenvectors
863: $\{\mathbf{e}_{\sigma_j}\}$, for $j=1,2,3,4$, are then given by
864: \begin{equation}
865: \label{eq:e_normalized}
866: \mathbf{e}_{\sigma_j}(k) =
867: \frac{1}{\sqrt{v_0^2+b_0^2}}
868: \left[\begin{array}{r}
869: v_0 \cos\theta_{{\rm v}j} \\
870: v_0 \sin\theta_{{\rm v}j} \\
871: i b_0 \cos\theta_{{\rm b}j} \\
872: i b_0 \sin\theta_{{\rm b}j}
873: \end{array}\right] \,.
874: \end{equation}
875:
876:
877: It is interesting to note that in this geometric representation the
878: dispersion relation (\ref{eq:dispersion_relation_nu_eta}) can be
879: obtained from the trigonometric identity
880: \begin{eqnarray}
881: \cos^2\theta_{{\rm v}j} + \sin^2\theta_{{\rm v}j}=1 \,,
882: \end{eqnarray}
883: where the expressions for
884: \begin{eqnarray}
885: \cos\theta_{{\rm v}j} &=& \sqrt{\frac{2 \sigma_{\eta j}^2}{q(k^2+\sigma_{\eta j}^2)}} \,, \\
886: \sin\theta_{{\rm v}j} &=& \frac{k^2+\sigma_{\nu j}\sigma_{\eta j}}{\sqrt{2q(k^2+\sigma_{\eta j}^2)}} \,,
887: \end{eqnarray}
888: can be obtained from the definition of the angle $\theta_{{\rm v}j}$ in
889: equation (\ref{eq:theta_v}).
890:
891: %\newpage
892: \subsection{Temporal Evolution}
893:
894: In physical space, the most general solution to the set of equations
895: (\ref{eq:vx_nodim})--(\ref{eq:by_nodim}), i.e.,
896: \begin{equation}
897: \bb{\delta}(z, t) =
898: \left[\begin{array}{c}
899: \delta v_r(z,t) \\ \delta v_\phi(z,t) \\
900: \delta b_r(z,t) \\ \delta b_\phi(z,t)
901: \end{array}\right] \,,
902: \end{equation}
903: evolves in time according to
904: \begin{equation}
905: \label{eq:solution_0}
906: {\bb \delta}(z,t) \equiv \sum_{k} \hat {\bb \delta}(k,t) \, e^{ikz} \,,
907: \end{equation}
908: where
909: \begin{equation}
910: \label{eq:solution_1}
911: \hat{\bb{\delta}}(k,t) = \sum_{j=1}^{4} a_j(k,0) \,
912: e^{\sigma_j t} \,\mathbf{e}_{\sigma_j} \,,
913: \end{equation}
914: with $\{\sigma_j\}$ and $\{\mathbf{e}_{\sigma_j}\}$, for $j=1,2,3,4$,
915: given by equations (\ref{eq:eigenvalues_all}) and
916: (\ref{eq:e_normalized}). The initial conditions $\bb{a}(k,0)$ are
917: related to the initial spectrum of perturbations,
918: $\hat{\bb{\delta}}(k,0)$, via $\bb{a}(k,0) = Q^{-1} \,
919: \hat{\bb{\delta}}(k,0)$. Here, $Q^{-1}$ is the matrix for the change
920: of coordinates from the standard basis to the normalized eigenvector
921: basis\footnote{The eigenvectors (\ref{eq:e_normalized}) are not in
922: general orthogonal, i.e., $\mathbf{e}_{\sigma_j} \bcdot \,
923: \mathbf{e}_{\sigma_{j'}} \ne 0$ for $j \ne j'$. If desired, an
924: orthogonal basis can be constructed using the Gram-Schmidt
925: orthogonalization procedure \cite[see, e.g.,][]{HK71}.} and can be
926: obtained by calculating the inverse of the matrix
927: \begin{equation}
928: Q = [
929: \mathbf{e}_{\sigma_1} \ \ \mathbf{e}_{\sigma_2} \ \
930: \mathbf{e}_{\sigma_3} \ \ \mathbf{e}_{\sigma_4}] \,.
931: \end{equation}
932:
933: The temporal evolution of a single MRI-unstable mode in physical space
934: can be obtained from a linear combination of
935: $\mathbf{e}_{\sigma_j}(k)$ and $\mathbf{e}_{\sigma_j}(-k)$ as defined
936: in equation (\ref{eq:e_normalized}). In particular, setting
937: $a_1(k,0)=a^*_1(-k,0)=-i/\sqrt{2}$ in equation~(\ref{eq:solution_0})
938: and substituting the result in equation~(\ref{eq:solution_1}) we
939: obtain
940: \begin{equation}
941: \label{eq:delta_zt_unstable}
942: \bb{\delta}(z, t) = \,
943: \frac{\sqrt{2} \, e^{\gamma_+ t}}{\sqrt{v_0^2+b_0^2}} \,
944: \left[\begin{array}{r}
945: v_0 \cos\theta_{{\rm v}}\, \sin(kz) \\
946: v_0 \sin\theta_{{\rm v}}\, \sin(kz) \\
947: b_0 \cos\theta_{{\rm b}}\, \cos(kz) \\
948: b_0 \sin\theta_{{\rm b}}\, \cos(kz)
949: \end{array}\right] \,.
950: \end{equation}
951:
952: \begin{figure*}[t]
953: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gmax_nu.eps}
954: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gmax_mu.eps}
955: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{gmax_eta.eps}
956: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f5a.eps}
957: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f5b.eps}
958: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f5c.eps}
959: \caption{Maximum growth rate $\gamma_{\rm max}$ for Keplerian
960: rotation in different dissipative regimes. The dot-dashed
961: horizontal lines at $\gamma_{\rm max}=3/4$ represent the ideal MHD
962: limit, eq.~(\ref{eq:gamma_max_ideal}). \emph{Left}: Maximum
963: growth rate as a function of the magnetic Reynolds number for
964: different values of the Reynolds number. The thick solid line
965: denotes the inviscid limit, i.e., ${\rm Re}\rightarrow \infty$.
966: The thin solid lines, in decreasing order, correspond to ${\rm Re}
967: =1, 0.1, \ldots, 10^{-6}$. For magnetic Reynolds numbers larger
968: than unity, the growth rate is independent of ${\rm Rm}$
969: regardless of the value of ${\rm Re}$. The dashed line,
970: calculated according to eq.~(\ref{eq:gamma_max_lim_reggrm}),
971: provides the correct asymptotic limit $\gamma_{\rm max} \propto
972: {\rm Rm}$ for small magnetic Reynolds numbers. \emph{Middle}:
973: Maximum growth rate $\gamma_{\rm max}$ as a function of the
974: Reynolds number for different values of the magnetic Prandtl
975: number. From left to right, the curves correspond to ${\rm
976: Pm}=10^3, 10^2, \ldots, 1$ (thick solid line), $\ldots,
977: 10^{-6}$. The dashed lines $\gamma_{\rm max} \propto {\rm
978: Re\,Pm}={\rm Rm}$ are calculated according to
979: eq.~(\ref{eq:gamma_max_lim_reggrm}), which leads to the correct result
980: even for ${\rm Pm} \gtrsim 1$ provided that the Reynolds number is
981: sufficiently small. The dotted line $\gamma_{\rm max} \propto
982: {\rm Re}^{1/2}$ results from eq.~(\ref{eq:gamma_max_lim_rellrm}).
983: \emph{Right}: Maximum growth rate $\gamma_{\rm max}$ as a function
984: of the Reynolds number for different values of the magnetic
985: Reynolds number. The thick solid line corresponds to the ideal
986: conductor limit, i.e., ${\rm Rm}\rightarrow \infty$. The thin
987: solid lines, in decreasing order, correspond to ${\rm Rm} = 1,
988: 0.1, 0.01$. For Reynolds numbers larger than unity, the growth
989: rate is independent of ${\rm Re}$ regardless of the value of ${\rm
990: Rm}$. The dotted line, calculated according to
991: eq.~(\ref{eq:gamma_max_lim_rellrm}), provides the correct
992: asymptotic limit $\gamma_{\rm max} \propto {\rm Re}^{1/2}$ for
993: small Reynolds number.}
994: \label{fig:gamma_max}
995: \end{figure*}
996:
997:
998:
999:
1000: These solutions are of particular importance for the linear late-time
1001: evolution of MRI modes. Note that any reasonable spectrum of initial
1002: perturbations of the type used in numerical simulations of shearing
1003: boxes will have a non-zero component along the unstable eigenvector
1004: $e_{\sigma_1}$. If the value of the magnetic field is such that the
1005: MRI can be excited for given values of the viscosity and resistivity
1006: then the exponentially growing perturbations in physical space will
1007: evolve towards a mode of the form (\ref{eq:delta_zt_unstable})
1008: dominated by the lengthscale $k=k_{\rm max}$ for which the growth rate
1009: reaches its maximum value $\gamma_{\rm max}$.
1010:
1011: Note that if a perturbation in physical space is composed by a single
1012: mode of the type described in \S~\ref{sub:classes_of_solutions}, no
1013: matter which class, then the angles defined in equations
1014: (\ref{eq:theta_v}) and (\ref{eq:theta_b}) are constant in time and are
1015: identical to the \emph{physical} angles between the planes containing
1016: magnetic and velocity perturbations in physical space, see
1017: Figure~\ref{fig:geom_def}, with
1018: \begin{eqnarray}
1019: \label{eq:tan_theta_v}
1020: \tan\theta_{{\rm v}j} &=& \frac{\delta v_\phi(z,t)}{\delta v_r(z,t)}={\rm const.} \,, \\
1021: \label{eq:tan_theta_b}
1022: \tan\theta_{{\rm b}j} &=& \frac{\delta b_\phi(z,t)}{\delta b_r(z,t)}={\rm const.}\,.
1023: \end{eqnarray}
1024:
1025: Finally, defining the angle $\theta_{{\rm bv}j}$ such that
1026: \begin{eqnarray}
1027: \theta_{{\rm bv}j} = \theta_{{\rm b}j} - \left(\theta_{{\rm v}j}+\frac{\pi}{2}\right) \,,
1028: \end{eqnarray}
1029: which implies that $\tan\theta_{{\rm b}j} \tan(\theta_{{\rm v}j}+\theta_{{\rm bv}j}) = -1$, and using the fact that
1030: \begin{eqnarray}
1031: \tan(\theta_1+\theta_2) = \frac{\tan\theta_1 + \tan\theta_2}{1-\tan\theta_1 \tan\theta_2} \,,
1032: \end{eqnarray}
1033: it is not difficult to show that
1034: \begin{eqnarray}
1035: \label{eq:theta_bv}
1036: \tan\theta_{{\rm bv}j} = -\mu = -\left(\frac{\nu-\eta}{2}\right)k^2 \,.
1037: \end{eqnarray}
1038: This means that $\mu\ne 0$ provides a measure of how non-orthogonal
1039: velocity and magnetic field perturbations are.
1040:
1041:
1042:
1043: It is evident that when the magnetic Prandtl number approaches unity
1044: viscous, resistive, MRI-driven magnetic and velocity perturbations tend
1045: to be orthogonal, i.e., $\tan \theta_{{\rm v}j}\tan \theta_{{\rm
1046: b}j}=-1$, and therefore
1047: \begin{eqnarray}
1048: \label{eq:theta_diff_Pmeq1}
1049: \theta_{\rm diff} \equiv \theta_{{\rm b}j} - \theta_{{\rm v}j} = \frac{\pi}{2} \quad
1050: \textrm{for} \quad {\rm Pm}=1 \,,
1051: \end{eqnarray}
1052: for every wavenumber $k$. This is illustrated in
1053: Figure~\ref{fig:geom_def} which shows the evolution of the angles
1054: $\theta_{{\rm b}}$ and $\theta_{{\rm v}}$ corresponding to the most
1055: unstable MRI mode as a function of the Reynolds/magnetic Reynolds
1056: number when the magnetic Prandtl number is equal to unity.
1057:
1058:
1059:
1060: \section{Physical Structure of MRI Modes}
1061: \label{sec:mri_modes}
1062:
1063: The evolution of the physical structure of a single growing MRI mode
1064: with wavenumber $0<k<k_{\rm c}$ is characterized by its growth rate
1065: $\gamma_+$, the relative magnitude between the amplitudes of magnetic
1066: and velocity field perturbations, $b_0/v_0$, and the two angles
1067: defining the planes containing them, $\theta_{\rm b}$ and $\theta_{\rm
1068: v}$. For any reasonable spectrum of initial perturbations the mode
1069: that exhibits the fastest exponential growth, $\gamma_{\rm max}$,
1070: which we refer to as $k_{\rm max}$, will dominate the dynamics of the
1071: late time evolution of the viscous, resistive MRI. It is therefore of
1072: particular interest to characterize the physical properties of this
1073: fastest growing mode in different dissipative regimes.
1074:
1075: \subsection{ Marginal and Fastest Growing MRI-modes}
1076: \label{sub:modes_marginal_fastest}
1077:
1078: Because the eigenvalue associated with the unstable growing mode,
1079: $\gamma_+$, is always real for any combination of the Reynolds and
1080: magnetic Reynolds numbers, it is possible to find the marginally
1081: stable mode $k_{\rm c}$ such that $\gamma_+(k_{\rm c}) \equiv 0$.
1082: Setting $\sigma=0$ in equation~(\ref{eq:dispersion_relation_nu_eta}),
1083: we obtain a polynomial in $k_{\rm c}$
1084: \begin{eqnarray}
1085: \label{eq:k_c}
1086: k_{\rm c}^2(1 + \nu \eta k_{\rm c}^2)^2 + \kappa^2(1 + \eta^2 k_{\rm c}^2) - 4 = 0 \,,
1087: \end{eqnarray}
1088: valid for any value of the viscosity and resistivity. Note that
1089: $k_{\rm c}$ sets the minimum domain height for numerical simulations
1090: of viscous, resistive MRI-driven turbulence. Figure~\ref{fig:k_c}
1091: shows the solutions of equation (\ref{eq:k_c}) in various dissipative
1092: regimes for Keplerian rotation. The analytic solutions of equation
1093: (\ref{eq:k_c}) are algebraically complicated but their asymptotic
1094: limits are rather simple. We find expressions for this critical
1095: wavenumber in several regimes of interest below.
1096:
1097: In the ideal MHD limit, it is straightforward to find simple analytical
1098: expressions for the most unstable wavenumber, $k_{\rm max}$, and its
1099: associated growth rate, $\gamma_{\rm max}$. However, the analytical
1100: expressions that we derived for the eigenfrequencies in the non-ideal
1101: case, equation (\ref{eq:eigenvalues_all}), are not amenable to the
1102: usual extremization procedure. More precisely, it is very challenging
1103: to find the values of $k_{\rm max}$ and $\gamma_{\rm max}$ that
1104: satisfy
1105: \begin{equation}
1106: \left.\frac{d\gamma_+}{dk}\right|_{k_{\rm max}}=0 \,.
1107: \end{equation}
1108: Figures~\ref{fig:k_max}~and~\ref{fig:gamma_max} show the solutions of
1109: this equation in various dissipative regimes.
1110:
1111: Another possible path to find the values of the wavenumber $k_{\rm
1112: max}$, and the associated growth rate, is to use the fact that
1113: $\gamma_{\rm max}$ satisfies simultaneously the dispersion relation
1114: (\ref{eq:dispersion_relation_nu_eta}) and its derivative to eliminate
1115: $k_{\rm max}$ between these two and obtain a polynomial in
1116: $\gamma_{\rm max}$. The largest of the roots of this polynomial is the
1117: desired maximum growth rate. It is possible to find $k_{\rm max}$
1118: following a similar methodology, but eliminating between the two
1119: polynomials $\gamma_{\rm max}$ instead. However, for arbitrary values
1120: of the viscosity and resistivity, both procedures lead to a seventh
1121: degree polynomial whose roots \emph{must} be found numerically,
1122: defeating altogether the attempt to find analytical expressions for
1123: $k_{\rm max}$ and $\gamma_{\rm max}$.
1124:
1125: Using as a guide the results shown in
1126: Figures~\ref{fig:k_max}~and~\ref{fig:gamma_max}, we follow an
1127: alternative procedure. The goal is to find simple analytical
1128: expressions to describe the asymptotic behavior of the most unstable
1129: mode, $k_{\rm max}$, and the maximum growth rate, $\gamma_{\rm max}$,
1130: in different dissipative regimes. It is evident from
1131: Figure~\ref{fig:growths} that $k_{\rm max}<1$ and $\gamma_{\rm max}<1$
1132: for all the non-ideal MRI modes. This information can be used to
1133: simplify the dispersion relation and its derivative so as to decrease
1134: their order without loosing vital information. This makes it possible
1135: to obtain manageable, but accurate, expressions for $k_{\rm max}$ and
1136: $\gamma_{\rm max}$ in different limiting regimes.
1137:
1138: Figure~\ref{fig:contours} shows contour plots for the critical
1139: wavenumber, $k_{\rm c}$, the most unstable wavenumber, $k_{\rm max}$,
1140: and the maximum growth rate, $\gamma_{\rm max}$, as a function of the
1141: Reynolds and magnetic Reynolds numbers for Keplerian rotation. In all
1142: three panels, lighter gray areas correspond to larger values of
1143: $k_{\rm c}$, $k_{\rm max}$, and $\gamma_{\rm max}$, respectively.
1144: Note that in all the cases, the functional form of the contours
1145: naturally divides the plane $({\rm Re}, {\rm Rm})$ in three
1146: distinctive regions that we denote according to ${\rm I}$ (ideal),
1147: ${\rm R}$ (resistive), and ${\rm V}$ (viscous). Note that when the
1148: most unstable wavenumber, $k_{\rm max}$, and the maximum growth rate,
1149: $\gamma_{\rm max}$, are considered, these regions can be associated
1150: with the regions where ${\rm Re, \,Rm} \gg 1$, ${\rm Re} \gg {\rm
1151: Rm}$, and ${\rm Re} \ll {\rm Rm}$, respectively. The overlap
1152: between these regions is not as clear when the critical wavenumber
1153: $k_{\rm c}$ is considered and some care is needed when deriving
1154: approximated expressions for it.
1155:
1156:
1157:
1158: \begin{figure*}[t]
1159: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{cont_kc.eps}
1160: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{cont_kmax.eps}
1161: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{cont_gmax.eps}
1162: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f6a.eps}
1163: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f6b.eps}
1164: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f6c.eps}
1165: \caption{Contour plots for the critical wavenumber, $k_{\rm c}$, the
1166: most unstable wavenumber, $k_{\rm max}$, and the maximum growth
1167: rate, $\gamma_{\rm max}$, for Keplerian rotation. In all three
1168: panels, lighter gray areas correspond to larger values of $k_{\rm
1169: c}$, $k_{\rm max}$, and $\gamma_{\rm max}$, respectively. The
1170: solid lines highlight the contours for $k_{\rm c} = 1, \ldots,
1171: 10^{-7}$ and $k_{\rm max} = 10^{-1}, \dots, 10^{-8}$. The labels I
1172: (ideal), R (resistive), and V (viscous), denote the three regions
1173: of the $({\rm Re}, {\rm Rm})$ plane where equations
1174: (\ref{eq:k_c_ideal}), (\ref{eq:k_max_ideal}), and
1175: (\ref{eq:gamma_max_ideal}); (\ref{eq:k_c_lim_reggrm}),
1176: (\ref{eq:k_max_lim_reggrm}) and (\ref{eq:gamma_max_lim_reggrm});
1177: and (\ref{eq:k_c_lim_rellrm}), (\ref{eq:k_max_lim_rellrm}), and
1178: (\ref{eq:gamma_max_lim_rellrm}) are valid, respectively. The
1179: dashed lines dividing the three regions are obtained by equating
1180: neighboring approximations for $k_{\rm c}$, $k_{\rm max}$, and
1181: $\gamma_{\rm max}$.}
1182: \label{fig:contours}
1183: \end{figure*}
1184:
1185:
1186: \subsection{Ideal MRI Modes}
1187: \label{sub:modes_ideal}
1188:
1189: Let us first demonstrate briefly how the formalism presented in
1190: \S~\ref{sec:solution} reduces to previously known results in the ideal
1191: MHD limit. In the absence of dissipation, the eigenvalues
1192: $\{\sigma_{0,j}\}$, with $j=1,2,3,4$, are the roots of the dispersion
1193: relation associated with the ideal MRI \citep{BH91,BH98},
1194: \begin{equation}
1195: \label{eq:dispersion_relation_ideal}
1196: (k^2+\sigma_{0,j}^2)^2 +
1197: \kappa^2( k^2 + \sigma_{0,j}^2) - 4 k^2 = 0 \,,
1198: \end{equation}
1199: and are given by \citep{PCP06}
1200: \begin{equation}
1201: \sigma_{0,j} = \pm \left(- \Lambda_0 \pm \sqrt{\Delta_0}\,\right)^{1/2} \,,
1202: \end{equation}
1203: where we have defined the quantities $\Lambda_0$ and $\Delta_0$ such that
1204: \begin{eqnarray}
1205: \label{eq:Lambda0}
1206: \Lambda_0 &\equiv& \frac{\kappa^2}{2} + k^2 \,, \\
1207: \label{eq:Delta0}
1208: \Delta_0 &\equiv& \frac{\kappa^4}{4} + 4 k^2 \,.
1209: \end{eqnarray}
1210:
1211: \begin{figure*}[t]
1212: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{diff_nu.eps}
1213: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{diff_mu.eps}
1214: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{diff_eta.eps}
1215: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f7a.eps}
1216: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f7b.eps}
1217: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f7c.eps}
1218: \caption{Opening angle, $\theta_{\rm diff}=\theta_{\rm
1219: b}-\theta_{\rm v}$, between the planes containing the fastest
1220: exponentially growing magnetic and velocity perturbations for
1221: Keplerian rotation in various dissipative regimes. In the ideal
1222: MHD limit the opening angle is $\theta_{\rm diff}=\pi/2$, see
1223: eqs.~(\ref{eq:theta_v_ideal})~and~(\ref{eq:theta_b_ideal}).
1224: \emph{Left}: opening angle $\theta_{\rm diff}$ as a function of
1225: the magnetic Reynolds number for different values of the Reynolds
1226: number. The thick solid line denotes the inviscid limit, i.e.,
1227: ${\rm Re}\rightarrow \infty$. The thin solid lines, in decreasing
1228: order according to $\theta_{\rm diff}$ for fixed ${\rm Rm}$,
1229: correspond to ${\rm Re} =10, 1, \ldots, 10^{-12}$. For magnetic
1230: Reynolds numbers much larger than unity, the opening angle is
1231: independent of ${\rm Rm}$ regardless of the value of ${\rm Re}$.
1232: For sufficiently small/large ${\rm Rm}$, $\theta_{\rm
1233: diff}\rightarrow \pi/2$ provided that ${\rm Re}\gg {\rm Rm}$.
1234: For sufficiently small/large ${\rm Rm}$, $\theta_{\rm
1235: diff}\rightarrow \pi/2-\arctan(\kappa/2)$ provided that ${\rm
1236: Rm}\ll {\rm Re}$. Note that this corresponds to $\theta_{\rm
1237: diff}=63\degr 26'$ for a Keplerian disk. The only conditions
1238: under which $\theta_{\rm diff}$ exceeds $\pi/2$ are such that
1239: ${\rm Re} \ge {\rm Rm} \simeq 1$. \emph{Middle}: opening angle
1240: $\theta_{\rm diff}$ as a function of the Reynolds number for
1241: different values of the magnetic Prandtl number. The thick solid
1242: line at $\theta_{\rm diff}=\pi/2$ corresponds to ${\rm Pm}=1$, see
1243: eq.~ (\ref{eq:theta_diff_Pmeq1}) and Fig.~\ref{fig:geom_def}. The
1244: thin solid lines with peaks at $\theta_{\rm diff}>\pi/2$
1245: correspond, from left to right, to ${\rm Pm}=10^{-1}, 10^{-2},
1246: \ldots$. The thin solid lines, with $\theta_{\rm diff}<\pi/2$, in
1247: decreasing order according to $\theta_{\rm diff}$ for fixed ${\rm
1248: Re}$, correspond to ${\rm Pm} =10, 10^2, \ldots$. Note that for
1249: small ${\rm Re}$, $\theta_{\rm diff}\rightarrow \pi/2$ for ${\rm
1250: Pm}\simeq 1$, while $\theta_{\rm diff}\rightarrow
1251: \pi/2-\arctan(\kappa/2)$ for ${\rm Pm}\gg 1$. \emph{Right}:
1252: opening angle $\theta_{\rm diff}$ as a function of the Reynolds
1253: number for different values of the magnetic Reynolds number. The
1254: thick solid line corresponds to the ideal conductor limit, i.e.,
1255: ${\rm Rm}\rightarrow \infty$. The thin solid lines, from right to
1256: left, correspond to ${\rm Rm} =10, \ldots, 10^{-3}$. For Reynolds
1257: numbers larger than unity, the opening angle is independent of
1258: ${\rm Re}$ regardless of the value of ${\rm Rm}$.}
1259: \label{fig:theta_diff}
1260: \end{figure*}
1261:
1262:
1263: The critical wavenumber for the onset of the ideal MRI is obtained by
1264: setting $\sigma_0=0$ in the dispersion relation
1265: (\ref{eq:dispersion_relation_ideal}), this leads to
1266: \begin{equation}
1267: \label{eq:k_c_ideal}
1268: k_{\rm c} = \sqrt{2q} = \sqrt{4-\kappa^2} \,.
1269: \end{equation}
1270: For all the modes with wavenumbers $k<k_{\rm c}$ the difference
1271: $\sqrt{\Delta_0}-\Lambda_0$ is positive and we can define the ``growth
1272: rate'' $\gamma_0$ and the ``oscillation frequency'' $\omega_0$ by
1273: \begin{eqnarray}
1274: \label{eq:gamma}
1275: \gamma_0 &\equiv& \left(\sqrt{\Delta_0} -\Lambda_0\right)^{1/2} \,, \\
1276: \label{eq:omega}
1277: \omega_0 &\equiv& \left(\sqrt{\Delta_0} + \Lambda_0\right)^{1/2} \,,
1278: \end{eqnarray}
1279: both of which are real and positive (for all positive values of the
1280: parameter $q$). This shows that two of the solutions of equation
1281: (\ref{eq:dispersion_relation_ideal}) are real and the other two are
1282: imaginary. We can thus write the four eigenvalues in compact notation
1283: as
1284: \begin{eqnarray}
1285: \label{eq:eigenvalues}
1286: \sigma_{0,1} = \gamma_0 \,, \quad
1287: %
1288: \sigma_{0,2} = -\gamma_0 \,,\quad
1289: %
1290: \sigma_{0,3} = i \omega_0 \,, \quad
1291: %
1292: \sigma_{0,4} = -i \omega_0 \,. \nonumber \\
1293: \end{eqnarray}
1294:
1295: In the ideal MHD limit, it is evident that the velocity and magnetic
1296: field perturbations are orthogonal for any mode, i.e., $\tan
1297: \theta_{{\rm v}} \tan \theta_{{\rm b}}=-1$, see
1298: equation~(\ref{eq:theta_bv}), and therefore
1299: \begin{eqnarray}
1300: \theta_{{\rm b}} = \theta_{{\rm v}} + \frac{\pi}{2} \,.
1301: \end{eqnarray}
1302: The temporal evolution of a single MRI-unstable mode in physical space
1303: reduces to
1304: \begin{equation}
1305: \bb{\delta}(z, t) =
1306: \frac{\sqrt{2} \, e^{\gamma_0 t}}{\sqrt{v_0^2+b_0^2}} \,
1307: \left[\begin{array}{r}
1308: v_0 \cos\theta_{{\rm v}} \sin(kz) \\
1309: v_0 \sin\theta_{{\rm v}} \sin(kz) \\
1310: b_0 \sin\theta_{{\rm v}} \cos(kz) \\
1311: - b_0 \cos\theta_{{\rm v}} \cos(kz)
1312: \end{array}\right] \,.
1313: \end{equation}
1314: These are essentially the (normalized) perturbations found in equation
1315: (4) in \citet{GX94} \footnote{Note that the angle $\gamma$ in
1316: \citet{GX94}, in our notation defined by $\tan \gamma = -\delta
1317: B_r/\delta B_\phi$, is such that $\gamma=0$ in the positive
1318: azimuthal axis and it takes increasingly positive values in the
1319: counter-clockwise direction.}.
1320:
1321: From the definition of the angle $\theta_{{\rm v}}$, see equation
1322: (\ref{eq:theta_v}), it can be seen that
1323:
1324: The maximum growth rate can be obtained by noting that $\gamma_{0} = q
1325: \sin\theta_{{\rm v}} \cos\theta_{{\rm v}}$ and therefore the maximum
1326: growth corresponds to
1327: \begin{equation}
1328: \label{eq:gamma_max_ideal}
1329: \gamma_{\rm max} = \frac{q}{2} = 1 - \frac{\kappa^2}{4}
1330: \,.
1331: \end{equation}
1332: It then follows that, in the absence of dissipation, the planes
1333: containing the exponentially growing velocity and magnetic field
1334: perturbations are characterized by the angles
1335: \begin{eqnarray}
1336: \label{eq:theta_v_ideal}
1337: \theta_{{\rm v}} &=& \frac{\pi}{4} \,, \\
1338: \label{eq:theta_b_ideal}
1339: \theta_{{\rm b}} &=& \frac{3\pi}{4} \,,
1340: \end{eqnarray}
1341: regardless of the value of the shearing parameter/epicyclic frequency.
1342:
1343:
1344: Finally, noting that the wavenumber for which the maximum growth rate
1345: is realized is
1346: \begin{equation}
1347: \label{eq:k_max_ideal}
1348: k_{\rm max} = \sqrt{1-\frac{\kappa^4}{16}} \,,
1349: \end{equation}
1350: and using equation~(\ref{eq:b0_v0}) for the ratio between the
1351: amplitudes of the magnetic and velocity fields we obtain
1352: \begin{equation}
1353: \label{eq:b0_v0_ideal}
1354: \frac{b_0}{v_0} =
1355: \sqrt{\frac{4+\kappa^2}{4-\kappa^2}}\,.
1356: \end{equation}
1357:
1358: In \S~\ref{sub:MRI_energetics} we derive equations for the MRI-driven
1359: Reynolds and Maxwell stresses, as well as the kinetic and magnetic
1360: energy densities associated with the perturbations. Equations
1361: (\ref{eq:mean_Rrphi}), (\ref{eq:mean_Mrphi}), (\ref{eq:mean_EK}), and
1362: (\ref{eq:mean_EM}), show why, in the ideal MHD limit, equations
1363: (\ref{eq:theta_v_ideal}), (\ref{eq:theta_b_ideal}), and
1364: (\ref{eq:b0_v0_ideal}) are the reason for which the ratio between the
1365: Maxwell to the Reynolds stresses is identical to the ratio between
1366: magnetic and kinetic energy densities for any shear parameter and
1367: equal to $5/3$ in the Keplerian case \citep{PCP06}.
1368:
1369:
1370: \begin{figure*}[t]
1371: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{mean_nu.eps}
1372: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{mean_mu.eps}
1373: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{mean_eta.eps}
1374: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f8a.eps}
1375: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f8b.eps}
1376: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f8c.eps}
1377: \caption{Mean angle, $\theta_{\rm mean}=(\theta_{\rm v}+\theta_{\rm
1378: b})/2$, defined by the fastest exponentially growing velocity
1379: and magnetic perturbations for Keplerian rotation in various
1380: dissipative regimes. Note that $\theta_{\rm mean}=\pi/2$ in the
1381: ideal MHD limit, see
1382: eqs.~(\ref{eq:theta_v_ideal})~and~(\ref{eq:theta_b_ideal}).
1383: \emph{Left}: mean angle $\theta_{\rm mean}$ as a function of the
1384: magnetic Reynolds number for different values of the Reynolds
1385: number. The thick solid line denotes the inviscid limit, i.e.,
1386: ${\rm Re}\rightarrow \infty$. There is a critical magnetic
1387: Reynolds number ${\rm Rm}_{\rm c} \lesssim 1$ that differentiates
1388: the asymptotic limits of $\theta_{\rm mean}$ for large and small
1389: Reynolds numbers. When the Reynolds number changes from ${\rm
1390: Re}\ll {\rm Rm}$ to ${\rm Re}\gg {\rm Rm}$, the mean angle
1391: evolves according to $\theta_{\rm
1392: mean}:[\pi/2+\arctan(\kappa/2)]/2 \rightarrow \pi/2$ for ${\rm
1393: Rm}>{\rm Rm}_{\rm c}$, while $\theta_{\rm
1394: mean}:[\pi/2+\arctan(\kappa/2)]/2 \rightarrow \pi/4$ for ${\rm
1395: Rm}<{\rm Rm}_{\rm c}$. Note that for Keplerian rotation
1396: $[\pi/2+\arctan(\kappa/2)]/2=58\degr 17'$. \emph{Middle}: opening
1397: angle $\theta_{\rm diff}$ as a function of the Reynolds number for
1398: different values of the magnetic Prandtl number. The thick solid
1399: line corresponds to ${\rm Pm}=1$. The thin solid lines to the
1400: right correspond to smaller values ${\rm Pm}=10^{-1}, 10^{-2},
1401: \ldots$. The thin solid lines to the left correspond to ${\rm Pm}
1402: =10, 10^2, \ldots$. Note that for small ${\rm Re}$, $\theta_{\rm
1403: mean}\rightarrow \pi/4$ for ${\rm Pm}\lesssim 1$, while
1404: $\theta_{\rm mean}\rightarrow [\pi/2+\arctan(\kappa/2)]/2$ for
1405: ${\rm Pm}\gg 1$. \emph{Right}: mean angle $\theta_{\rm mean}$ as
1406: a function of the magnetic Reynolds number for different values of
1407: the Reynolds number. The thick solid line corresponds to the
1408: ideal conductor limit, i.e., ${\rm Rm}\rightarrow \infty$. The
1409: thin solid lines, in decreasing order, correspond to ${\rm Re}
1410: =10, 1, \ldots$. For Reynolds numbers larger than unity, the mean
1411: angle is independent of ${\rm Re}$ regardless of the value of
1412: ${\rm Rm}$.}
1413: \label{fig:theta_mean}
1414: \end{figure*}
1415:
1416:
1417:
1418: \subsection{MRI Modes with ${\rm Re} \gg {\rm Rm}$}
1419: \label{sub:modes_re_gg_rm}
1420:
1421: Lets us consider the inviscid, poorly conducting limit described by
1422: $\nu=0$ and $\eta\gg 1$. In this case, the marginally stable mode
1423: satisfying equation (\ref{eq:k_c}) is given by
1424: \begin{eqnarray}
1425: \label{eq:k_c_lim_reggrm}
1426: k_{\rm c} = \sqrt{\frac{4-\kappa^2}{1+\eta^2\kappa^2}} \,.
1427: \end{eqnarray}
1428: The dependence of this critical wavenumber on the magnetic Reynolds
1429: number is shown on the left panel in Figure~\ref{fig:k_c}, which shows
1430: that for small magnetic Reynolds numbers $k_{\rm c} \propto Rm$.
1431:
1432: As discussed in \S~\ref{sub:modes_marginal_fastest}, finding an
1433: analytic expression for the maximum growth rate and wavenumber
1434: associated with it is not as straightforward. The left panel of
1435: Figure~\ref{fig:gamma_max} suggests that in the limit ${\rm
1436: Re}\rightarrow \infty$ and ${\rm Rm}\ll 1$, the maximum growth rate
1437: is linear in the magnetic Reynolds number, $\gamma_{\rm max} \propto
1438: {\rm Rm} \propto \eta^{-1}$. This information can be used to derive
1439: asymptotic expressions for the dispersion relation
1440: (\ref{eq:dispersion_relation_nu_eta}) and its derivative. The leading
1441: order contributions are given by
1442: \begin{eqnarray}
1443: \label{eq:disp_lim_reggrm}
1444: \kappa^2 \gamma_{\rm max}^2 + 2\kappa^2 \eta k_{\rm max}^2\gamma_{\rm max} +
1445: \kappa^2\eta^2k_{\rm max}^4 +(\kappa^2-4)k_{\rm max}^2= 0 \,,
1446: \nonumber \\
1447: \end{eqnarray}
1448: and
1449: \begin{eqnarray}
1450: \label{eq:disp_lim_reggrm_deriv}
1451: 2\kappa^2 \eta \gamma_{\rm max} +
1452: 2\kappa^2\eta^2k_{\rm max}^2 + \kappa^2-4= 0 \,,
1453: \end{eqnarray}
1454: respectively.
1455:
1456:
1457: Eliminating either $\gamma_{\rm max}$ or $k_{\rm max}$ between
1458: equations (\ref{eq:disp_lim_reggrm}) and
1459: (\ref{eq:disp_lim_reggrm_deriv}) we obtain
1460: \begin{eqnarray}
1461: \label{eq:k_max_lim_reggrm}
1462: k_{\rm max} = \frac{1}{\eta}\sqrt{\frac{4-\kappa^2}{4\kappa^2}} \,,
1463: \end{eqnarray}
1464: and
1465: \begin{eqnarray}
1466: \label{eq:gamma_max_lim_reggrm}
1467: \gamma_{\rm max} = \frac{1}{\eta}\frac{4-\kappa^2}{4\kappa^2} \,.
1468: \end{eqnarray}
1469: In this case, $\gamma_{\rm max}= \eta k_{\rm max}^2$ for any value of
1470: the epicyclic frequency $\kappa$.
1471:
1472: The dependence of both $k_{\rm max}$ and $\gamma_{\rm max}$ in this
1473: limiting case is shown with dashed lines in the left panels of
1474: Figures~\ref{fig:k_max} and \ref{fig:gamma_max}, respectively. The
1475: agreement between equations (\ref{eq:k_max_lim_reggrm}) and
1476: (\ref{eq:gamma_max_lim_reggrm}) and the solutions to the full
1477: dispersion relation (\ref{eq:dispersion_relation_nu_eta}) in the case
1478: $\nu=0$ and $\eta\gg1$ is excellent, only breaking down close to
1479: magnetic Reynolds numbers of order unity. Note that even though the
1480: equations (\ref{eq:k_max_lim_reggrm}) and
1481: (\ref{eq:gamma_max_lim_reggrm}) were derived under the assumption of
1482: an inviscid fluid, i.e., $\nu=0$, these expressions can describe the
1483: asymptotic behavior of both $k_{\rm max}$ and $\gamma_{\rm max}$ for
1484: finite Reynolds numbers provided that the conditions ${\rm Re} \gg
1485: {\rm Rm}$ and ${\rm Rm}\ll 1$ are satisfied.
1486:
1487: Substituting the asymptotic expressions for $k_{\rm max}$ and
1488: $\gamma_{\rm max}$ in equations (\ref{eq:k_max_lim_reggrm}) and
1489: (\ref{eq:gamma_max_lim_reggrm}) into equation (\ref{eq:b0_v0}) we obtain
1490: the ratio between the amplitudes of the magnetic and velocity field
1491: perturbations
1492: \begin{equation}
1493: \label{eq:b0_v0_reggrm}
1494: \frac{b_0}{v_0} =
1495: \frac{\eta\kappa^3}{\sqrt{4-\kappa^2}}\,.
1496: \end{equation}
1497: Therefore, inviscid, resistive MRI-unstable modes are dominated by
1498: magnetic field perturbations. Note that the ratio between amplitudes
1499: increases linearly with resistivity.
1500:
1501: The asymptotic behavior for the angles characterizing velocity and
1502: magnetic field perturbations, equations (\ref{eq:theta_v}) and
1503: (\ref{eq:theta_b}), are given by
1504: \begin{eqnarray}
1505: \tan \theta_{\rm v} = \frac{1}{2\kappa^2 \eta} \,,
1506: \end{eqnarray}
1507: and
1508: \begin{eqnarray}
1509: \tan \theta_{\rm b} = - \eta \kappa^2 \,.
1510: \end{eqnarray}
1511: In the limit ${\rm Re}\rightarrow\infty$ and ${\rm Rm}\rightarrow 0$, we obtain
1512: \begin{eqnarray}
1513: \lim_{\eta \rightarrow \infty} \theta_{\rm v} &=& 0 \,, \\
1514: \lim_{\eta \rightarrow \infty} \theta_{\rm b} &=& \frac{\pi}{2} \,.
1515: \end{eqnarray}
1516:
1517: We therefore conclude that in the regime of large Reynolds numbers and
1518: small magnetic Reynolds numbers, magnetic field perturbations are
1519: larger than velocity field perturbations, both fields tend to be
1520: orthogonal and aligned with the azimuthal and radial directions,
1521: respectively, see
1522: Figures~\ref{fig:theta_diff},~\ref{fig:theta_mean},~and~\ref{fig:geom_sketch}.
1523:
1524: \begin{figure*}[t]
1525: % \includegraphics[width=2\columnwidth,trim=0 0 0 0]{geo_sketch.eps}
1526: \includegraphics[width=2\columnwidth,trim=0 0 0 0]{f9.eps}
1527: \caption{Geometrical representation of viscous, resistive MRI modes
1528: for varying Reynolds and magnetic Reynolds numbers. The black/gray
1529: lines denote velocity/magnetic components of the most unstable
1530: mode.}
1531: \label{fig:geom_sketch}
1532: \end{figure*}
1533:
1534: %\newpage
1535:
1536:
1537: \subsection{MRI Modes with ${\rm Re}={\rm Rm}\ll 1$}
1538: \label{sub:modes_re_eq_rm_ll_1}
1539:
1540: When the magnetic Prandtl number is unity, and with $\eta=\nu\gg 1$,
1541: the marginally stable mode satisfying equation (\ref{eq:k_c}) is given
1542: by
1543: \begin{eqnarray}
1544: \label{eq:k_c_lim_reeqrm}
1545: k_{\rm c} = \frac{\sqrt{4-\kappa^2}}{\eta\kappa} \,.
1546: \end{eqnarray}
1547: The dependence of this critical wavenumber on the Reynolds number is
1548: shown in the middle panel in Figure~\ref{fig:k_c}. Incidentally,
1549: equation~(\ref{eq:k_c_lim_reeqrm}) corresponds to the asymptotic limit
1550: $\eta\gg 1$ of equation~(\ref{eq:k_c_lim_reggrm}).
1551:
1552: It is not hard to see that the leading order contributions to the
1553: dispersion relation and its derivative in the limit ${\rm Re}={\rm
1554: Rm}\ll 1$ are identical to the ones obtained in the case ${\rm
1555: Re}\rightarrow \infty$ and ${\rm Rm}\ll 1$. Therefore, all the
1556: expressions for $k_{\rm max}$, $\gamma_{\rm max}$, $\theta_{\rm v}$,
1557: and $\theta_{\rm b}$, derived in \S~\ref{sub:modes_re_gg_rm}, are also
1558: valid in this case. The dependence of both $k_{\rm max}$ and
1559: $\gamma_{\rm max}$ in this limiting case is shown with dashed lines in
1560: the middle panels of Figures~\ref{fig:k_max} and \ref{fig:gamma_max},
1561: respectively. The agreement between equations
1562: (\ref{eq:k_max_lim_rellrm}) and (\ref{eq:gamma_max_lim_rellrm}) and
1563: the solutions to the full dispersion relation
1564: (\ref{eq:dispersion_relation_nu_eta}) in the case $\nu=\eta$, i.e.,
1565: ${\rm Pm}=1$, is also excellent in this case.
1566:
1567: \subsection{MRI Modes with ${\rm Re} \ll {\rm Rm}$}
1568: \label{sub:modes_re_ll_rm}
1569:
1570: Lets us consider next the highly viscous, ideal conductor limit
1571: described by $\nu\gg1$ and $\eta=0$. In this case, the marginally
1572: stable mode satisfying equation (\ref{eq:k_c}) is given by
1573: \begin{eqnarray}
1574: \label{eq:k_c_lim_rellrm}
1575: k_{\rm c} = \left(\frac{\sqrt{4-\kappa^2}}{\nu\eta}\right)^{1/3} \,.
1576: \end{eqnarray}
1577:
1578: The right panel of Figure~\ref{fig:gamma_max} suggests that in the
1579: limit ${\rm Re}\ll 1$ and ${\rm Rm}\rightarrow \infty$, the dependence
1580: of the maximum growth on the Reynolds number is $\gamma_{\rm max}
1581: \propto {\rm Re}^{1/2} \propto \nu^{-1/2}$. This information can be
1582: used to derive asymptotic expressions for the dispersion relation
1583: (\ref{eq:dispersion_relation_nu_eta}) and its derivative. The leading
1584: order contributions are given by
1585: \begin{eqnarray}
1586: \label{eq:disp_lim_rellrm}
1587: (\kappa^2 +\nu^2k_{\rm max}^4)\gamma_{\rm max}^2 + (\kappa^2 -4)k_{\rm
1588: max}^2 =0 \,,
1589: \end{eqnarray}
1590: and
1591: \begin{eqnarray}
1592: \label{eq:disp_lim_rellrm_deriv}
1593: 2\nu^2k_{\rm max}^2 \gamma_{\rm max}^2 +\kappa^2-4= 0 \,,
1594: \end{eqnarray}
1595: respectively.
1596:
1597: Eliminating either $\gamma_{\rm max}$ or $k_{\rm max}$ between these
1598: equations we obtain
1599: \begin{eqnarray}
1600: \label{eq:k_max_lim_rellrm}
1601: k_{\rm max} = \sqrt{\frac{\kappa}{\nu}} \,,
1602: \end{eqnarray}
1603: and
1604: \begin{eqnarray}
1605: \label{eq:gamma_max_lim_rellrm}
1606: \gamma_{\rm max} = \sqrt{\frac{4-\kappa^2}{2\nu\kappa}} \,.
1607: \end{eqnarray}
1608: Eliminating the epicyclic frequency between equations
1609: (\ref{eq:gamma_max_lim_rellrm}) and (\ref{eq:k_max_lim_rellrm}) we
1610: obtain
1611: \begin{eqnarray}
1612: \gamma_{\rm max}^2 = \frac{4-\nu^2k_{\rm max}^4}{2\nu^2k_{\rm max}^2} \,.
1613: \end{eqnarray}
1614:
1615: The dependence of both $k_{\rm max}$ and $\gamma_{\rm max}$ in this
1616: limiting case is shown with dashed lines in the right panels of
1617: Figures~\ref{fig:k_max} and \ref{fig:gamma_max}, respectively. The
1618: agreement between equations (\ref{eq:k_max_lim_rellrm}) and
1619: (\ref{eq:gamma_max_lim_rellrm}) and the solutions to the full dispersion
1620: relation (\ref{eq:dispersion_relation_nu_eta}) in the case $\nu\gg 1$
1621: and $\eta = 0$ is excellent, only breaking down close to Reynolds
1622: numbers of order unity. Note that even though the equations
1623: (\ref{eq:k_max_lim_rellrm}) and (\ref{eq:gamma_max_lim_rellrm}) were
1624: derived under the assumption of a perfectly conducting fluid, i.e.,
1625: $\eta=0$, these expressions can describe the asymptotic behavior of
1626: both $k_{\rm max}$ and $\gamma_{\rm max}$ for finite Reynolds numbers
1627: provided that the conditions ${\rm Re} \ll {\rm Rm}$ and ${\rm Re}\ll
1628: 1$ are satisfied.
1629:
1630:
1631: Substituting the asymptotic expressions for $k_{\rm max}$ and
1632: $\gamma_{\rm max}$ in equations (\ref{eq:k_max_lim_rellrm}) and
1633: (\ref{eq:gamma_max_lim_rellrm}) into equation (\ref{eq:b0_v0}) we
1634: obtain the relative amplitude of the magnetic and velocity field
1635: perturbations
1636: \begin{equation}
1637: \label{eq:b0_v0_rellrm}
1638: \frac{b_0}{v_0} =
1639: 2\sqrt{\frac{\nu\kappa^{3}}{4+\kappa^2}}\,.
1640: \end{equation}
1641: Therefore, viscous, conducting MRI-unstable modes are also dominated
1642: by magnetic field perturbations. In this case, the ratio between
1643: amplitudes increases only with the square root of the viscosity.
1644:
1645: The asymptotic behavior for the angles characterizing velocity and
1646: magnetic field perturbations, equations (\ref{eq:theta_v}) and
1647: (\ref{eq:theta_b}), are given by
1648: \begin{eqnarray}
1649: \tan \theta_{\rm v} = \frac{\kappa}{2} +
1650: \frac{4+\kappa^2}{\sqrt{2\nu\kappa(4-\kappa^2)}} \,,
1651: \end{eqnarray}
1652: and
1653: \begin{eqnarray}
1654: \tan \theta_{\rm b} = \sqrt{\frac{\nu(4-\kappa^2)}{2\kappa}} \,.
1655: \end{eqnarray}
1656: In the limit ${\rm Re}\rightarrow0$ and ${\rm Rm}\rightarrow \infty$, we obtain
1657: \begin{eqnarray}
1658: \lim_{\nu \rightarrow \infty} \theta_{\rm v} &=& \arctan \left(\frac{\kappa}{2}\right) \,,\\
1659: \lim_{\nu \rightarrow \infty} \theta_{\rm b} &=& \frac{\pi}{2} \,.
1660: \end{eqnarray}
1661:
1662:
1663: For Keplerian rotation, the angle between the fastest growing velocity
1664: field perturbation and the radial direction is given by $\theta_{\rm
1665: v} = \arctan(1/2)=26\degr 34'$. Therefore, the opening angle between
1666: the planes containing velocity and magnetic field perturbations is
1667: $\theta_{\rm diff} = \theta_{\rm b} - \theta_{\rm v}= 63\degr 26'$ and
1668: their mean value is $\theta_{\rm mean} = (\theta_{\rm b} + \theta_{\rm
1669: v})/2 = 58\degr 17'$. The right panels of
1670: Figures~\ref{fig:theta_diff} and \ref{fig:theta_mean} show that both
1671: of these results are in perfect agreement with the asymptotic behavior
1672: of the full solutions derived directly from the original dispersion
1673: relation (\ref{eq:dispersion_relation_nu_eta}).
1674:
1675: We therefore conclude that in the regime of small Reynolds numbers and
1676: large magnetic Reynolds numbers, magnetic perturbations are larger
1677: than velocity perturbations. In this case, however, the perturbed
1678: magnetic and velocity fields are not orthogonal. The perturbed
1679: magnetic field tends to be aligned with the azimuthal direction but
1680: the velocity field perturbations do not tend to be aligned with the
1681: radial direction. The angle between both fields is determined entirely
1682: by the epicyclic frequency $\kappa$, see
1683: Figures~\ref{fig:theta_diff},~\ref{fig:theta_mean},~and~\ref{fig:geom_sketch}.
1684:
1685: \subsection{MRI Modes with ${\rm Re}={\rm Rm}\gg 1$}
1686: \label{sub:modes_re_eq_rm_gg_1}
1687:
1688: As we show in Appendix \ref{sec:appendix}, when the Reynolds and
1689: magnetic Reynolds numbers are large enough the solutions to the
1690: dispersion relation (\ref{eq:dispersion_relation_nu_eta}) tend
1691: smoothly to the solutions of the dispersion relation
1692: (\ref{eq:dispersion_relation_ideal}) for the idealized case.
1693: Therefore, in this limit we recover all the expressions derived in
1694: \S~\ref{sub:modes_ideal}.
1695:
1696: \section{Physics of Maximally Unstable MRI Modes}
1697: \label{sec:mri_physics}
1698:
1699: We have shown that the expressions derived for the most unstable
1700: wavenumber, $k_{\rm max}$, and its associated growth rate,
1701: $\gamma_{\rm max}$, obtained from the simplified equations
1702: (\ref{eq:disp_lim_reggrm}) and (\ref{eq:disp_lim_rellrm}) are good
1703: approximations to the solutions obtained directly from the dispersion
1704: relation~(\ref{eq:dispersion_relation_nu_eta}), in the limits ${\rm
1705: Re} \gg {\rm Rm}$ and ${\rm Re} \ll {\rm Rm}$, respectively. We can
1706: now identify the various terms in the original set of equations of
1707: motion~(\ref{eq:vx})--(\ref{eq:by}) that lead to
1708: equations~(\ref{eq:disp_lim_reggrm})~and~(\ref{eq:disp_lim_rellrm}).
1709: This allows us to better understand the forces that act to destabilize
1710: magnetized fluid elements.
1711:
1712: For the sake of clarity we write the equations in this section with
1713: physical dimensions. We represent the temporal derivatives with a dot
1714: and the derivatives with respect to the vertical coordinate $z$ with a
1715: prime.
1716:
1717:
1718:
1719: \subsection{MRI Modes with ${\rm Re} \gg {\rm Rm}$}
1720: \label{sub:mri_physics_re_gg_rm}
1721:
1722: The equations of motion that lead to the dispersion relation
1723: (\ref{eq:disp_lim_reggrm}) are given by
1724: \begin{eqnarray}
1725: \label{eq:vx_reggrm}
1726: \dot{\delta v_r} &=& 2 \Omega_0 \delta v_\phi + \frac{\bar
1727: B_z}{4\pi\rho} \,\delta B_r' \,, \\
1728: \label{eq:vy_reggrm}
1729: 0 &=& - (2-q)\Omega_0 \delta v_r + \frac{\bar B_z}{4\pi\rho} \,\delta B_\phi' \,, \\
1730: \label{eq:bx_reggrm}
1731: \dot{\delta B_r} &=& \bar B_z \delta v_r' + \eta \, \delta B_r'' \,, \\
1732: \label{eq:by_reggrm}
1733: \dot{\delta B_\phi} &=& - q \Omega_0 \delta B_r + \eta \, \delta B_\phi''\,.
1734: \end{eqnarray}
1735: Therefore, maximally unstable modes with ${\rm Re} \gg {\rm Rm}$ are
1736: characterized by motions with radial acceleration due to the Coriolis
1737: force acting on azimuthally displaced fluid elements and magnetic
1738: tension. Azimuthal force balance is attained via the joint action of
1739: the Coriolis force acting on radially displaced fluid elements, radial
1740: advection of background flow, and magnetic tension in the azimuthal
1741: direction. The rate of change of the radial magnetic field
1742: perturbations is due to the creation of radial field out of the
1743: vertical background frozen into the radial velocity field with a
1744: vertical gradient and field diffusion. The rate of change of the
1745: azimuthal magnetic field perturbations is due to the shearing of
1746: radial magnetic field perturbations and field diffusion.
1747:
1748: \subsection{MRI Modes with ${\rm Re} \ll {\rm Rm}$}
1749: \label{sub:mri_physics_re_ll_rm}
1750:
1751: The set of equations that lead to the dispersion relation
1752: (\ref{eq:disp_lim_rellrm}) are given by
1753: \begin{eqnarray}
1754: \label{eq:vx_rellrm}
1755: 0 &=& 2 \Omega_0 \delta v_\phi +\nu \delta v_r'' \,, \\
1756: \label{eq:vy_rellrm}
1757: 0 &=& - (2-q)\Omega_0 \delta v_r + \frac{\bar B_z}{4\pi\rho} \,\delta B_\phi' + \nu \delta v_\phi'' \,, \\
1758: \label{eq:bx_rellrm}
1759: \dot{\delta B_r} &=& \bar B_z \delta v_r' \,, \\
1760: \label{eq:by_rellrm}
1761: \dot{\delta B_\phi} &=& - q \Omega_0 \delta B_r \,.
1762: \end{eqnarray}
1763: In this case, maximally unstable modes with ${\rm Re} \ll {\rm Rm}$
1764: are characterized by fluid displacements that take place under force
1765: balance in both the radial and azimuthal directions. The Coriolis
1766: force acting on azimuthally displaced fluid elements is balanced by
1767: the viscous force in the radial direction. Azimuthal force balance is
1768: attained via the joint action of the Coriolis force acting on radially
1769: displaced fluid elements, radial advection of background flow,
1770: magnetic tension, and the viscous force in the azimuthal direction.
1771: The rate of change of the radial magnetic field perturbations is due
1772: to the creation of radial field out of the vertical background frozen
1773: into the radial velocity field with a vertical gradient. Finally, the
1774: rate of change of the azimuthal magnetic field perturbations is due to
1775: the shearing of radial magnetic field perturbations.
1776:
1777:
1778: \begin{figure*}[t]
1779: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{stress_nu.eps}
1780: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{stress_mu.eps}
1781: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{stress_eta.eps}
1782: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f10a.eps}
1783: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f10b.eps}
1784: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f10c.eps}
1785: \caption{Ratio between the Maxwell and the Reynolds stresses,
1786: $-\bar{M}_{r\phi}/\bar{R}_{r\phi}$, for Keplerian rotation in
1787: different dissipative regimes. \emph{Left}: Ratio between the
1788: Maxwell and the Reynolds stresses as a function of the magnetic
1789: Reynolds number for different values of the Reynolds number. The
1790: thick solid line denotes the inviscid limit, i.e., ${\rm
1791: Re}\rightarrow \infty$. The thin solid lines, in increasing
1792: order according to their asymptotic value at high magnetic
1793: Reynolds numbers correspond to ${\rm Re} =1, 0.1, \ldots$. For
1794: ${\rm Rm}>1$, the ratio is independent of ${\rm Rm}$ regardless of
1795: the value of ${\rm Re}$. \emph{Middle}: Ratio between the Maxwell
1796: and the Reynolds stresses as a function of the Reynolds number for
1797: different values of the magnetic Prandtl number. The various
1798: curves, from left to right, correspond to ${\rm Pm}=10^3, 10^2,
1799: \ldots$. The thick solid line corresponds to the case ${\rm
1800: Pm}=1$. \emph{Right}: Ratio between the Maxwell and the
1801: Reynolds stresses as a function of the Reynolds number for
1802: different values of the magnetic Reynolds number. The thick solid
1803: line corresponds to the ideal conductor limit, i.e., ${\rm
1804: Rm}\rightarrow \infty$. The thin solid lines, in increasing
1805: order according to their asymptotic value at high Reynolds numbers
1806: correspond to ${\rm Rm} = 1, 0.1, 0.01$. The Maxwell stress is
1807: larger than the Reynolds stress for any combination of the
1808: Reynolds and magnetic Reynolds numbers. The minimum value of this
1809: ratio is achieved in the ideal MHD regime, where
1810: $-\bar{M}_{r\phi}/\bar{R}_{r\phi} =5/3$, for Keplerian rotation.}
1811: \label{fig:stresses}
1812: \end{figure*}
1813:
1814:
1815:
1816: \section{Non-Ideal MRI-driven Stresses and Energy Densities}
1817: \label{sec:stresses_energies}
1818:
1819:
1820: The $i,j$ components of the mean Reynolds and Maxwell stresses
1821: associated with the velocity and magnetic field perturbations are given
1822: by
1823: \begin{eqnarray}
1824: \label{eq:mean_reynolds}
1825: \bar{R}_{ij}(t) & \equiv & \langle \delta v_i(z,t) \, \delta v_j(z,t) \rangle \, , \\
1826: \label{eq:mean_maxwell}
1827: \bar{M}_{ij}(t) & \equiv & \langle \delta b_i(z,t) \, \delta b_j(z,t) \rangle \, ,
1828: \end{eqnarray}
1829: where the brackets denote mean values obtained via integration over
1830: the disk scale-height, $2H$. These mean values can also be calculated
1831: directly from the perturbations in Fourier space according to
1832: \citep{PCP06}
1833: \begin{eqnarray}
1834: \label{eq:mean_reynolds_k}
1835: \bar{R}_{ij}(t) & \equiv & 2 \sum_{n=1}^{\infty} \;
1836: {\mathcal Re}[\,
1837: \hat{\delta v_i}(k_n,t) \, \hat{\delta v_j}\!\!^*\!(k_n,t)\,] \,, \\
1838: \label{eq:mean_maxwell_k}
1839: \bar{M}_{ij}(t) & \equiv & 2 \sum_{n=1}^{\infty} \;
1840: {\mathcal Re}[\,
1841: \hat{\delta b_i}(k_n,t) \, \hat{\delta b_j}\!\!^*\!(k_n,t)\,] \,.
1842: \end{eqnarray}
1843: where ${\mathcal Re}[\,]$ stands for the real part of the quantity between square
1844: brackets. The off-diagonal components of these stresses relate to
1845: angular momentum transport while their traces relate to the kinetic
1846: and magnetic energy in the perturbations.
1847:
1848: At late times, during the exponential growth of the instability, the
1849: branch of unstable modes will dominate the growth of the perturbations
1850: and the most important (secular) contribution to the mean stresses
1851: will be given by the most unstable mode. The leading order
1852: contribution to these stress components are thus obtained by
1853: considering the most unstable solutions to the set of equations
1854: (\ref{eq:vx})--(\ref{eq:by}), which are given by equation
1855: (\ref{eq:delta_zt_unstable}) when $k=k_{\rm max}$ with
1856: $\gamma_+=\gamma_{\rm max}$.
1857:
1858:
1859: \subsection{Non-ideal MRI Stresses}
1860: \label{sub:MRI_stresses}
1861:
1862: A measure of the angular momentum transport driven by the most
1863: unstable MRI modes and mediated by the correlated perturbations in the
1864: velocity and magnetic fields is obtained by setting
1865: $(i,\,j)=(r,\,\phi)$ in equations (\ref{eq:mean_reynolds}) and
1866: (\ref{eq:mean_maxwell}). To leading order in time we obtain
1867: \begin{eqnarray}
1868: \label{eq:mean_Rrphi}
1869: \bar{R}_{r\phi}(t)
1870: & = &
1871: \frac{1}{2} \,
1872: \frac{v_0^2 e^{ 2\gamma_{\rm max} t}}{v_0^2+b_0^2} \,
1873: \sin(2\theta_{{\rm v}}) \,,\\
1874: \label{eq:mean_Mrphi}
1875: \bar{M}_{r\phi}(t) & = &
1876: \frac{1}{2} \,
1877: \frac{b_0^2 e^{ 2\gamma_{\rm max} t}}{v_0^2+b_0^2} \,
1878: \sin(2\theta_{{\rm b}})\,.
1879: \end{eqnarray}
1880:
1881:
1882: The results derived in \S~\ref{sec:mri_modes}, together with
1883: Figures~\ref{fig:theta_diff},~\ref{fig:theta_mean},~and~\ref{fig:geom_sketch},
1884: show that the angles $\theta_{\rm v}$ and $\theta_{\rm b}$
1885: corresponding to the most unstable mode, $k_{\rm max}$, always satisfy
1886: \begin{eqnarray}
1887: 0 &\le& \theta_{\rm v} \le \arctan{\left(\frac{\kappa}{2}\right)} \le \frac{\pi}{4}\,, \\
1888: \frac{\pi}{2} &\le& \theta_{\rm b} \le \frac{3\pi}{4} \,.
1889: \end{eqnarray}
1890: Both of these inequalities show explicitly that the mean Reynolds and
1891: Maxwell stresses will be, respectively, positive and negative,
1892: \begin{equation}
1893: \bar{R}_{r\phi}(t) >0 \qquad \textrm{and} \qquad \bar{M}_{r\phi}(t) < 0 \,.
1894: \end{equation}
1895: This, in turn, implies that the mean total MRI-driven stress will be
1896: always positive, i.e.,
1897: \begin{equation}
1898: \bar{T}_{r\phi}(t) = \bar{R}_{r\phi}(t) - \bar{M}_{r\phi}(t) >0 \,,
1899: \end{equation}
1900: driving a net outward flux of angular momentum for any combination of
1901: Reynolds and magnetic Reynolds numbers.
1902:
1903: We conclude this section by calculating the ratio
1904: $-\bar{M}_{r\phi}(t)/\bar{R}_{r\phi}(t)$ at late times
1905: during the exponential growth of the instability.
1906: We obtain
1907: \begin{equation}
1908: \lim_{t\gg1}
1909: \label{eq:analytical_ratio}
1910: \frac{-\bar{M}_{r\phi}(t)}{\,\,\bar{R}_{r\phi}(t)} = -
1911: \frac{v_0^2}{b_0^2} \frac{\sin(2\theta_{\rm v})}{\sin(2\theta_{\rm
1912: b})} \,.
1913: \end{equation}
1914: Using the definitions for the ratio $v_0/b_0$ (eq.~[\ref{eq:b0_v0}]) and
1915: the angles $\theta_{\rm v}$ and $\theta_{\rm b}$
1916: (eqs.~[\ref{eq:theta_v}] and [\ref{eq:theta_b}]), it can be seen that
1917: the magnitude of the Maxwell stress, $-\bar{M}_{r\phi}(t)$, is always
1918: larger than the magnitude of the Reynolds stress,
1919: $\bar{R}_{r\phi}(t)$, provided that the flow is Rayleigh-stable, i.e.,
1920: \begin{equation}
1921: -\bar{M}_{r\phi}(t) > \bar{R}_{r\phi}(t) \quad \textrm{for} \quad 0<q<2 \,.
1922: \end{equation}
1923: Figure~\ref{fig:stresses} shows the ratio between the Maxwell and the
1924: Reynolds stresses in various dissipative regimes for Keplerian
1925: rotation. Note that when the Reynolds and magnetic Reynolds numbers
1926: are large enough we recover the result
1927: $-\bar{M}_{r\phi}/\bar{R}_{r\phi}=5/3$, which coincides with the value
1928: of this ratio in the ideal MHD case \citep{PCP06}.
1929:
1930:
1931:
1932: \subsection{Non-ideal MRI Energetics}
1933: \label{sub:MRI_energetics}
1934:
1935: The mean energy densities associated with the perturbations in the
1936: velocity and magnetic field are given by
1937: \begin{eqnarray}
1938: \label{eq:mean_EK}
1939: \bar{E}_K(t) &=&
1940: \frac{1}{2} \, (\bar{R}_{rr} + \bar{R}_{\phi\phi}) \,,\\
1941: \label{eq:mean_EM}
1942: \bar{E}_M(t) &=&
1943: \frac{1}{2} \, (\bar{M}_{rr} + \bar{M}_{\phi\phi}) \,.
1944: \end{eqnarray}
1945: Substituting the expressions for the most unstable MRI-driven
1946: perturbations from equation (\ref{eq:delta_zt_unstable}) into the
1947: definitions for the diagonal components of the Reynolds and Maxwell
1948: stresses, equations~(\ref{eq:mean_reynolds})--(\ref{eq:mean_maxwell}),
1949: respectively, we obtain
1950: \begin{eqnarray}
1951: \label{eq:mean_R}
1952: \bar{E}_K(t)
1953: & = & \frac{1}{2}\, \frac{v_0^2 e^{ 2\gamma_{\rm max} t}}{v_0^2+b_0^2} \,, \\
1954: \label{eq:mean_M}
1955: \bar{E}_M(t)
1956: & = & \frac{1}{2}\, \frac{b_0^2 e^{ 2\gamma_{\rm max} t}}{v_0^2+b_0^2} \,.
1957: \end{eqnarray}
1958:
1959: \begin{figure*}[t]
1960: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{energy_nu.eps}
1961: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{energy_mu.eps}
1962: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{energy_eta.eps}
1963: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f11a.eps}
1964: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f11b.eps}
1965: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f11c.eps}
1966: \caption{Ratio between the magnetic and the kinetic energy densities
1967: contained in MRI-driven perturbations, $\bar{E}_M/\bar{E}_K$, for
1968: Keplerian rotation in different dissipative regimes. \emph{Left}:
1969: Ratio between the magnetic and the kinetic energy densities as a
1970: function of the magnetic Reynolds number for different values of
1971: the Reynolds number. The thick solid line denotes the inviscid
1972: limit, i.e., ${\rm Re}\rightarrow \infty$. The thin solid lines,
1973: in increasing order according to their asymptotic value at high
1974: magnetic Reynolds numbers correspond to ${\rm Re} = 1, 0.1,
1975: \ldots$. For magnetic Reynolds numbers larger than unity, the
1976: ratio is independent of ${\rm Rm}$ regardless of the value of
1977: ${\rm Re}$. Moreover, the asymptotic value of this ratio for
1978: ${\rm Rm}\ll 1$ is independent of the Reynolds number.
1979: \emph{Middle}: Ratio between the magnetic and the kinetic energy
1980: densities as a function of the Reynolds number for different
1981: values of the magnetic Prandtl number. The various curves, from
1982: left to right, correspond to ${\rm Pm}=10^2, 10, \ldots$. The
1983: thick solid line corresponds to the case ${\rm Pm}=1$.
1984: \emph{Right}: Ratio between the Maxwell and the Reynolds stresses
1985: as a function of the Reynolds number for different values of the
1986: magnetic Reynolds number. The thick solid line corresponds to the
1987: ideal conductor limit, i.e., ${\rm Rm}\rightarrow \infty$. The
1988: thin solid lines, in increasing order according to their
1989: asymptotic value at high Reynolds numbers correspond to ${\rm Rm}
1990: = 1, 0.1, \ldots$. Unlike the ratio between stresses, the ratio
1991: between energy densities seems to decrease monotonically with
1992: ${\rm Re}$ for any value of the magnetic Reynolds number. The
1993: magnetic energy density is larger than the kinetic energy density
1994: for any combination of the Reynolds and magnetic Reynolds numbers.
1995: The minimum value of this ratio is achieved in the ideal MHD
1996: regime, where $\bar{E}_M/\bar{E}_K=5/3$, for Keplerian rotation.}
1997: \label{fig:energies}
1998: \end{figure*}
1999:
2000:
2001: Using the definitions for the ratio $v_0/b_0$,
2002: equation~(\ref{eq:b0_v0}), it can be seen that for non-ideal MRI modes
2003: the mean energy associated with magnetic perturbations is always
2004: larger than the mean energy corresponding to velocity perturbations as
2005: long as the flow is Rayleigh-stable. Figure~\ref{fig:energies} shows
2006: the ratio between the magnetic and the kinetic energy densities in
2007: various dissipative regimes for Keplerian rotation.
2008:
2009: In the limit of large Reynolds and magnetic Reynolds numbers, we
2010: recover the result $\bar{E}_M/\bar{E}_K=5/3$, which coincides with the
2011: value of this ratio in the ideal MHD case and also with the ratio
2012: between the magnitudes of the Maxwell and Reynolds stresses in the
2013: ideal case \citep{PCP06}. Note, however, that for arbitrary Reynolds
2014: and magnetic Reynolds numbers, it is no longer true that the ratio
2015: between mean magnetic and mean kinetic energies is equal to the ratio
2016: between the magnitude of the mean Maxwell and the mean Reynolds
2017: stresses. This can be seen by comparing
2018: Figures~\ref{fig:stresses}~and~\ref{fig:energies}.
2019:
2020: Finally, comparing equations (\ref{eq:mean_Rrphi}) and
2021: (\ref{eq:mean_Mrphi}) with (\ref{eq:mean_R}) and (\ref{eq:mean_M}), it
2022: immediately follows that
2023: \begin{eqnarray}
2024: \bar{R}_{r\phi}(t) &\le& \bar{E}_K(t) \,,\\
2025: -\bar{M}_{r\phi}(t) &\le& \bar{E}_M(t) \,.
2026: \end{eqnarray}
2027: This result, in turn, implies that the total mean energy associated
2028: with the perturbations, $\bar{E}(t) = \bar{E}_K(t) + \bar{E}_M(t)$,
2029: sets an upper bound on the total mean stress, i.e.,
2030: \begin{eqnarray}
2031: \bar{T}_{r\phi}(t) \le \bar{E}(t) \,,
2032: \end{eqnarray}
2033: for any Reynolds and magnetic Reynolds numbers.
2034:
2035:
2036:
2037: \section{Summary \& Discussion}
2038: \label{sec:discussion}
2039:
2040:
2041:
2042: We investigated the effects of viscosity and resistivity on the
2043: stability of differentially rotating plasmas threaded by a magnetic
2044: field perpendicular to the shear. We have shown that the most powerful
2045: incompressible MRI modes are exact solutions of the MHD equations for
2046: arbitrary combinations of the Reynolds and magnetic Reynolds numbers.
2047: We have derived analytical expressions for the eigenfrequencies as
2048: well as for the eigenmodes describing the MRI in viscous, resistive
2049: media and provided a detailed description of the physical properties
2050: of these modes in various dissipative regimes.
2051:
2052: We have shown that the scalings derived for the marginally stable
2053: mode, the most unstable wavenumber, and the maximum growth rate with
2054: magnetic Reynolds number, $k_{\rm c}, k_{\rm max}, \gamma_{\rm max}
2055: \propto {\rm Rm}$, valid for resistive, inviscid plasmas, see
2056: equations~(\ref{eq:k_c_lim_reggrm}),~(\ref{eq:k_max_lim_reggrm})~and~(\ref{eq:gamma_max_lim_reggrm}),
2057: as well as \citealt{SM99}, also hold when finite Reynolds numbers are
2058: involved. This is true as long as the magnetic Prandtl number is of
2059: order unity or smaller, as it is usually the case in many
2060: astrophysical systems (such as accretion disks around cataclysmic
2061: variables and young stellar objects, as well as the Sun) and also in
2062: MRI laboratory experiments. Furthermore, we have addressed in detail,
2063: for the first time to our knowledge, the physical properties of the
2064: MRI in highly viscous, slightly resistive media. These conditions are
2065: expected to be found in the hot, diffuse gas in galaxies and galaxy
2066: clusters. In this case, we found that the critical wavenumber for the
2067: onset of the MRI, the most unstable wavenumber, and the maximum growth
2068: rate scale with the Reynolds and magnetic Reynolds numbers according
2069: to $k_{\rm c}\propto ({\rm Re\,Rm})^{1/3}$ and $k_{\rm max},
2070: \gamma_{\rm max} \propto {\rm Re}^{1/2}$, see
2071: equations~(\ref{eq:k_c_lim_rellrm}),~(\ref{eq:k_max_lim_rellrm})~and~(\ref{eq:gamma_max_lim_rellrm}).
2072:
2073: We have provided a thorough geometrical description of the viscous,
2074: resistive MRI modes in terms of the angles that define the planes
2075: containing the velocity and magnetic field perturbations. In the
2076: ideal MHD limit, these planes are orthogonal, with the plane
2077: containing the velocity disturbances laying at $45\degr$ with respect
2078: to the radial direction. We have shown that velocity and magnetic
2079: field perturbations are still orthogonal if the magnetic Prandtl
2080: number is unity, but that the planes containing them tend to be
2081: aligned with the radial and azimuthal directions, respectively, when
2082: the Reynolds number increases. In the regime of large Reynolds numbers
2083: and small magnetic Reynolds numbers, magnetic and velocity field
2084: perturbations tend to be orthogonal and aligned with the azimuthal and
2085: radial directions, respectively. On the other hand, in the regime of
2086: small Reynolds numbers and large magnetic Reynolds numbers, the
2087: perturbed magnetic field tends to be aligned with the azimuthal
2088: direction but the velocity field perturbations do not tend to be
2089: aligned with the radial direction. The angle between both fields is
2090: determined entirely by the epicyclic frequency $\kappa$. It would be
2091: very interesting to understand to what extent this geometrical
2092: dependence of MRI modes on the Reynolds and magnetic Reynolds numbers
2093: influences the physical properties of kinetic and magnetic cells in
2094: fully developed viscous, resistive MHD turbulence.
2095:
2096: \begin{figure*}[t]
2097: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{Txy_nu.eps}
2098: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{Txy_mu.eps}
2099: % \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{Txy_eta.eps}
2100: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f12a.eps}
2101: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f12b.eps}
2102: \includegraphics[width=0.675\columnwidth,trim=0 0 0 0]{f12c.eps}
2103: \caption{Mean total stress responsible for angular momentum
2104: transport, $\bar{T}_{r\phi}=\bar{R}_{r\phi}-\bar{M}_{r\phi}$,
2105: calculated according to
2106: equations~(\ref{eq:mean_Rrphi})~and~(\ref{eq:mean_Mrphi}), in
2107: various dissipative regimes for Keplerian rotation. \emph{Left}:
2108: Mean total stress $\bar{T}_{r\phi}$ as a function of the magnetic
2109: Reynolds number for different values of the Reynolds number. The
2110: thick solid line denotes the inviscid limit, i.e., ${\rm
2111: Re}\rightarrow \infty$. The thin solid lines, in decreasing
2112: order, correspond to ${\rm Re} =10, 1, \ldots, 10^{-4}$. For
2113: magnetic Reynolds numbers larger than unity, this ratio is
2114: independent of ${\rm Rm}$ regardless of the value of ${\rm Re}$.
2115: Moreover, the asymptotic value of this ratio for ${\rm Rm}\ll 1$
2116: is independent of the Reynolds number. \emph{Middle}: Mean total
2117: stress $\bar{T}_{r\phi}$ as a function of the magnetic Prandtl
2118: number for different values of the Reynolds number. From left to
2119: right, the curves correspond to ${\rm Pm}=10^3, 10^2, \ldots, 1$
2120: (thick solid line), $\ldots, 10^{-6}$. \emph{Right}: Mean total
2121: stress $\bar{T}_{r\phi}$ as a function of the Reynolds number for
2122: different values of the magnetic Reynolds number. The thick solid
2123: line corresponds to the ideal conductor limit, i.e., ${\rm
2124: Rm}\rightarrow \infty$. The thin solid lines, in decreasing
2125: order, correspond to ${\rm Rm} =10, 1, 0.1$. Note that the stress
2126: scale in these plots is arbitrary.}
2127: \label{fig:Txy_RmPm}
2128: \end{figure*}
2129:
2130:
2131:
2132:
2133: In the ideal MHD limit, the exact (primary) MRI modes are known to be
2134: unstable to parasitic (secondary) instabilities \citep{GX94}. These
2135: parasitic modes have long been suspected to enable the mechanisms that
2136: disrupt the primary modes providing an avenue toward saturation. The
2137: solutions derived in this work describe the dynamics of primary MRI
2138: modes in viscous, resistive media enabling the study of parasitic
2139: instabilities for arbitrary combinations of Reynolds and magnetic
2140: Reynolds numbers. The modifications in the relative directions of the
2141: velocity and magnetic field perturbations characterizing the primary
2142: viscous, resistive MRI modes described above can have an important
2143: impact on the development and evolution of parasitic instabilities in
2144: the presence of dissipation.
2145:
2146:
2147: We have shown that, for any combination of the Reynolds and magnetic
2148: Reynolds numbers, the mean Reynolds stress, $\bar{R}_{r\phi}=\langle
2149: \delta v_r(z,t) \, \delta v_\phi(z,t) \rangle$, is always positive and
2150: the mean Maxwell stress, $\bar{M}_{r\phi}=\langle \delta b_r(z,t) \,
2151: \delta b_\phi(z,t) \rangle$, is always negative. This implies that the
2152: mean total stress, $\bar{T}_{r\phi}=\bar{R}_{r\phi}-\bar{M}_{r\phi}$
2153: is always positive, leading always to an outward transport of angular
2154: momentum. We have also demonstrated that both the ratio between
2155: magnetic and kinetic stresses, $-\bar{M}_{r\phi}/\bar{R}_{r\phi}$, and
2156: the ratio between magnetic and kinetic energy densities,
2157: $\bar{E}_M/\bar{E}_K$, are always dominated by the magnetic
2158: contribution. These last two statements, support a somewhat
2159: unexpected result since it is tempting to think that velocity
2160: perturbations would dominate both the transport of angular momentum
2161: and the energy density in highly resistive, inviscid plasmas. It
2162: would be very interesting to understand if and how the value of these
2163: ratios in the saturated turbulent state vary as a function of the
2164: Reynolds and magnetic Reynolds numbers.
2165:
2166: Sano and collaborators have studied the linear \citep{SM99} and
2167: non-linear \citep{SIM98, SI01, SS03, Sanoetal04} evolution of the MRI
2168: for inviscid, resistive MHD. The simulations in \citet{SS03} show
2169: that for small magnetic Reynolds numbers the stresses at saturation
2170: increase rapidly with increasing ${\rm Rm}$ and that there exists a
2171: critical magnetic Reynolds number, of order unity, beyond which
2172: turbulent stresses are rather insensitive to ${\rm Rm}$. The fact that
2173: this same behavior is indeed seen when the stresses are due to
2174: viscous, resistive MRI modes (see Figure~\ref{fig:Txy_RmPm}) rises the
2175: question of how strong is the influence of long-lived, channel-like
2176: modes on the fully developed turbulent state reached in shearing box
2177: simulations with net magnetic flux through the vertical boundaries.
2178:
2179: Systematic numerical studies of viscous, resistive MHD shearing flows
2180: have begun to uncover the dependencies of microphysical dissipation on
2181: the mean transport properties of MRI-driven turbulence. Numerical
2182: simulations with both zero \citep{FPII07} and non-zero net magnetic
2183: fluxes \citep{LL07} lead to the conclusion that angular momentum
2184: transport increases with increasing magnetic Prandtl number when the
2185: Reynolds number is held constant. This behavior can also be identified
2186: when examining the stresses due to viscous, resistive MRI modes,
2187: see~Figure~\ref{fig:Txy_RmPm}.
2188:
2189: The effects of varying the Reynolds number at fixed magnetic Prandtl
2190: number on the saturation of MRI-driven turbulence are currently rather
2191: uncertain \citep[see, in particular, the discussion in][]{LL07}. The
2192: global trends exhibited by the available simulations suggest that the
2193: stresses at saturation increase with increasing Reynolds number for
2194: magnetic Prandtl numbers smaller than unity while the opposite
2195: behavior is observed for magnetic Prandtl numbers larger than unity.
2196: If confirmed, these results suggest that the mechanisms leading to
2197: saturation might operate differently depending on whether the magnetic
2198: Prandtl number is larger or smaller than unity. In any case, having
2199: obtained a better understanding of the behavior of the most unstable
2200: MRI modes as a function the magnetic Prandtl number it would be very
2201: interesting to follow the evolution of the viscous, resistive MRI from
2202: the linear to the non-linear regime. By performing numerical
2203: simulations with the same Prandtl number (both larger and smaller than
2204: unity) and different Reynolds numbers we could see whether there is an
2205: inversion of the trends observed in the linear regime (i.e., higher
2206: stresses at higher Reynolds numbers for fixed magnetic Prandtl
2207: numbers) after the exact solutions break down. The comparison between
2208: the late time behavior of the viscous, resistive MRI modes and fully
2209: developed MHD turbulence with dissipation will shed light into the
2210: mechanisms that lead to the saturation of the MRI in different
2211: dissipative regimes.
2212:
2213: Finally, most current numerical algorithms employ finite difference
2214: methods (with constrained transport for the evolution of the magnetic
2215: field). The leading order errors in first-order upwind methods behave
2216: like diffusion, however, in second-order central difference methods,
2217: the leading order errors are dispersive (artificial viscosities are
2218: usually employed to damp unphysical oscillations near shocks). The
2219: comparison between numerical solutions from different algorithms can
2220: provide estimates of these errors. However, it is difficult to
2221: quantify these numerical artifacts based on analytical studies of
2222: ideal MHD. The analytical solutions derived in this paper, on the
2223: other hand, describe the effects of arbitrary combinations of
2224: viscosity and resistivity [see also \citet{LB07} who derived results
2225: to leading order in $(\eta-\nu)/k$]. It should now be possible to
2226: better measure the numerical viscosity and resistivity for a wide
2227: range of Reynolds and magnetic Reynolds numbers by comparing numerical
2228: solutions and analytical solutions of non-ideal MRI. The results
2229: presented in this paper provide ideal benchmarks to the study
2230: numerical artifacts generated by different algorithms in various
2231: dissipative regimes.
2232:
2233: \appendix
2234:
2235: \section{Analytical Solution to Quartic Equation and Ideal MHD Limit}
2236: \label{sec:appendix}
2237:
2238: The solutions to a depressed quartic equation of the form
2239: \begin{equation}
2240: \sigma^4 + \alpha \sigma^2 + \beta \sigma + \lambda = 0.
2241: \label{eq:quartic}
2242: \end{equation}
2243: are given by\footnote{Note that the two $\pm_b$'s have the same sign
2244: so there are only four solutions instead of eight.}
2245: \begin{equation}
2246: \sigma = \pm_a \sqrt{-\left(\frac{3\alpha}{4} + \frac{y}{2} \pm_b
2247: \frac{\beta/4}{\sqrt{\alpha/4 + y/2}}\right)} \pm_b \sqrt{\frac{\alpha}{4} + \frac{y}{2}} \,,
2248: \label{eq:solution_sigma}
2249: \end{equation}
2250: where $y$ is any of the solutions of the cubic equation
2251: \begin{equation}
2252: \label{eq:cubic}
2253: y^3 + \frac{5\alpha}{2} y^2 + (2\alpha^2 - \lambda) y +
2254: \left(\frac{\alpha^3}{2} - \frac{\alpha\lambda}{2} -
2255: \frac{\beta^2}{8}\right) = 0 \,.
2256: \end{equation}
2257:
2258:
2259: In the special case $\beta \rightarrow 0$, the solutions to equation
2260: (\ref{eq:cubic}) take simple forms. In order to see how the general
2261: solutions reduce to the simple cases, we write the cubic equation as
2262: \begin{equation}
2263: \left(y + \frac{\alpha}{2}\right)\left[(y + \alpha)^2 -
2264: \lambda\right] = \frac{\beta^2}{8} \,.
2265: \end{equation}
2266: It is easy to see that, if $y \ne -\alpha/2$,
2267: \begin{equation}
2268: \frac{\beta/4}{\sqrt{\alpha/4 + y/2}} = \sqrt{(y + \alpha)^2 - \lambda}.
2269: \end{equation}
2270: Using the above identity, the general
2271: solution~(\ref{eq:solution_sigma}) becomes
2272: \begin{equation}
2273: \sigma = \pm_a
2274: \sqrt{-\Lambda \pm_b \sqrt{\Delta}}
2275: \pm_b \, \frac{\beta/4}{\sqrt{\Delta}} \,,
2276: \end{equation}
2277: where we have defined
2278: \begin{equation}
2279: \Lambda = \frac{3\alpha}{4} + \frac{y}{2}
2280: \qquad \textrm{and} \qquad
2281: \Delta = \sqrt{(y + \alpha)^2 - \lambda} \,.
2282: \end{equation}
2283:
2284: If we choose the root so that
2285: \begin{equation}
2286: \lim_{\beta \rightarrow 0} y = -\frac{\alpha}{2},
2287: \end{equation}
2288: it is easier to take the limit
2289: \begin{equation}
2290: \lim_{\beta \rightarrow 0} \sigma =
2291: \lim_{\beta \rightarrow 0} \pm \sqrt{-\Lambda \pm \sqrt{\Delta}} =
2292: \pm \sqrt{- \frac{\alpha}{2} \mp \sqrt{\frac{\alpha^2}{4} - \lambda}}
2293: \equiv \pm \sqrt{-\Lambda_0\pm\sqrt{\Delta_0}} = \sigma_0
2294: \end{equation}
2295: where $\Lambda_0$ and $\Delta_0$ are defined in equations
2296: (\ref{eq:Lambda0}) and (\ref{eq:Delta0}).
2297:
2298: \acknowledgments{We thank Jeremy Goodman and Dimitrios Psaltis for
2299: valuable comments and discussions. We are grateful to Roman
2300: Shcherbakov for the initial discussions that lead to the idea of
2301: combining the dispersion relation and its derivative to obtain the
2302: asymptotic expressions derived in \S~\ref{sec:mri_modes}. MEP
2303: gratefully acknowledges support from the Institute for Advanced
2304: Study. CKC is supported through an ITC Fellowship at Harvard. MEP and
2305: CKC are grateful to the Harvard-Smithsonian Institute for Theory and
2306: Computation and the Institute for Advanced Study, respectively, for
2307: their hospitality during part of this work.}
2308:
2309:
2310: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2311:
2312: \begin{thebibliography}{}
2313:
2314: \bibitem[Balbus \& Hawley(1991)]{BH91}
2315: {Balbus, S.\ A. \& Hawley J.\ F. 1991, ApJ, 376, 214}
2316:
2317: \bibitem[Balbus \& Hawley(1992)]{BH92}
2318: {Balbus, S.\ A. \& Hawley J.\ F. 1992, ApJ, 392, 662}
2319:
2320: \bibitem[Balbus \& Hawley(1998)]{BH98}
2321: {---------. 1998, Rev. Mod. Phys., 70, 1}
2322:
2323: \bibitem[Balbus \& Henri(2007)]{BH08}
2324: {Balbus, S.\ A. \& Henri, P. 2008, ApJ, in press [arXiv0706.0828]}
2325:
2326: \bibitem[Blaes \& Balbus(1994)]{BB94}
2327: {Blaes, O.\ M., \& Balbus, S.\ A. 1994, ApJ, 421, 163}
2328:
2329: \bibitem[Brandenburg {et~al.}(1995) Brandenburg, Nordlund, Stein, \& Torkelsson]{Brandenburg95}
2330: {Brandenburg, A., Nordlund, A., Stein, R.\ F., \& Torkelsson, U. 1995, ApJ, 446, 741}
2331:
2332: \bibitem[Fleming, Stone, \& Hawley(2000)]{FSH00}
2333: {Fleming, T.\ P., Stone, J.\ M., Hawley 2000, ApJ, 530, 464}
2334:
2335: \bibitem[Fromang \& Papaloizou(2007)]{FPI07}
2336: {Fromang, S. \& Papaloizou, J. 2007, A\&A 476, 1113}
2337:
2338: \bibitem[Fromang {et~al.}(2007)Papaloizou, Lesur, \& Heinemann]{FPII07}
2339: {Fromang, S., Papaloizou, J., Lesur, G., \& Heinemann, T. 2007,
2340: A\&A, 476, 1123}
2341:
2342: \bibitem[Gammie(1996)]{Gammie96}
2343: {Gammie, C.\ F. 1996, ApJ, 462, 725}
2344:
2345: \bibitem[Goodman \& Xu(1994)]{GX94}
2346: {Goodman, J. \& Xu, G. 1994, ApJ, 432, 213}
2347:
2348: \bibitem[Goodman \& Ji(2002)]{GJ02}
2349: {Goodman, J. \& Ji, H. 2002, JFM, 462, 365}
2350:
2351: \bibitem[Hawley, Gammie, \& Balbus(1995)]{HGB95}
2352: {Hawley, J.\ F., Gammie, C.\ F., \& Balbus, S.\ A. 1995, ApJ, 440, 742}
2353:
2354: \bibitem[Hoffman \& Kunze(1971)]{HK71}
2355: {Hoffman K.M., Kunze R., 1971, Linear Algebra, 2nd ed. Prentice Hall, N.J.}
2356:
2357: \bibitem[Ji, Goodman \& Kageyama(2001)]{JGK01}
2358: {Ji, H., Goodman, J. \& Kageyama, A. 2001, MNRAS, 325, L1}
2359:
2360: \bibitem[Jin(1996)]{Jin96}
2361: {Jin, L. 1996, ApJ, 457, 798}
2362:
2363: \bibitem[Liu, Goodman, \& Ji(2006)]{LGJ06}
2364: {Liu, W., Goodman, J. \& Ji, H. 2006, ApJ, 643, 306}
2365:
2366: \bibitem[Lesur \& Longaretti(2007)]{LL07}
2367: {Lesur, G. \& Longaretti, P.\ Y. 2007, MNRAS, 378, 1471}
2368:
2369: \bibitem[Lesaffre \& Balbus(2007)]{LB07}
2370: {Lesaffre, P. \& Balbus, S.\ A. 2007, MNRAS, 381, 319}
2371:
2372: \bibitem[Pessah \& Psaltis(2005)]{PP05}
2373: {Pessah, M.\ E. \& Psaltis, D. 2005, ApJ, 628, 879}
2374:
2375: \bibitem[Pessah, Chan, \& Psaltis(2006)]{PCP06}
2376: {Pessah, M.\ E., Chan C.\ K., \& Psaltis, D. 2006, MNRAS, 372, 183}
2377:
2378: \bibitem[Pessah, Chan, \& Psaltis(2007)]{PCP07}
2379: {Pessah, M.\ E., Chan C.\ K., \& Psaltis, D. 2007, ApJ, 668, L51}
2380:
2381: \bibitem[R\"udiger, Schultz, Shalybkov(2003)]{SSW03}
2382: {R\"udiger, G., Schultz, M., \& Shalybkov, D. 2003, Phys. Rev. E, 67, 046312}
2383:
2384: \bibitem[Salmeron \& Wardle(2005)]{SW05}
2385: {Salmeron, R. \& Wardle, M. 2005, MNRAS, 361, 45}
2386:
2387: \bibitem[Sano \& Inutsuka(2001)]{SI01}
2388: {Sano, T. \& Inutsuka, S.\ I. 2001, ApJ, 561, L179}
2389:
2390: \bibitem[Sano, Inutsuka, \& Miyama(1998)]{SIM98}
2391: {Sano, T., Inutsuka, S.\ I., \& Miyama, S.\ M. 1998, ApJ, 506, L57}
2392:
2393: \bibitem[Sano {et~al.}(2004)Sano, Inutsuka, Turner, \& Stone]{Sanoetal04}
2394: {Sano, T., Inutsuka, S.\ I., Turner, N.\ J., \& Stone, J.\ M. 2004, ApJ, 605, 321}
2395:
2396: \bibitem[Sano \& Miyama(1999)]{SM99}
2397: {Sano, T. \& Miyama, S.\ M. 1999, ApJ, 515, 776}
2398:
2399: \bibitem[Sano \& Stone(2003)]{SS03} {Sano, T. \& Stone, J.\ M. 2003,
2400: in Scientific Frontiers in Research on Extrasolar Planets, ASPC,
2401: Vol 294, D., Deming \& S., Seager, eds. (San Francisco: ASP)}
2402:
2403: \bibitem[Sisan {et~al.}(2004)Sisan, Mujica, Tillotson, Huang, Dorland,
2404: Hassam, Antonsen, \& Lathrop]{Sisanetal04}
2405: {Sisan, D.\ R. Mujica, N., Tillotson, W.\ A., Huang, Y.\ M.,
2406: Dorland, W., Hassam, A.\ B., Antonsen, T.\ M., \& Lathrop, D.\ P.
2407: 2004, Phys. Rev. Lett., 93, 114502}
2408:
2409: \bibitem[Turner, Sano, \& Dziourkevitch(2007)]{TSD07}
2410: {Turner, N.\ J., Sano, T. \& Dziourkevitch, N. 2007, ApJ, 659, 729}
2411:
2412: \end{thebibliography}
2413:
2414: \end{document}
2415: