1: \documentclass[aps,showpacs,msmath,amssymb,prl,twocolumn]{revtex4}
2: %\usepackage{pslatex}
3: \usepackage{epsfig}
4:
5:
6: %%% NEW COMMANDS %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
7: \newcommand\bb[1] {\mbox{\boldmath{$#1$}}}
8: \newcommand\del{\bb{\nabla}}
9: \newcommand\bcdot{\bb{\cdot}}
10: \newcommand\btimes{\bb{\times}}
11: \newcommand\real{{\rm Re}}
12: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
13:
14: \begin{document}
15:
16: \title{\textsc{A Local Model for Angular Momentum Transport in
17: Accretion Disks \\ Driven by the Magnetorotational Instability}}
18: \author{Martin E. Pessah$^{1,2}$\email{mpessah@as.arizona.edu},
19: Chi-kwan Chan$^2$, and Dimitrios Psaltis$^{2,1}$} \affiliation{
20: $^{1}$Department of Astronomy, University of Arizona, Tucson, AZ 85721
21: \\ $^{2}$Department of Physics, University of Arizona, Tucson, AZ 85721}
22:
23: \email{mpessah@as.arizona.edu}
24:
25: \begin{abstract}
26: We develop a local model for the exponential growth and saturation
27: of the Reynolds and Maxwell stresses in turbulent flows driven by
28: the magnetorotational instability. We first derive equations that
29: describe the effects of the instability on the growth and pumping of
30: the stresses. We highlight the relevance of a new type of
31: correlations that couples the dynamical evolution of the Reynolds
32: and Maxwell stresses and plays a key role in developing and
33: sustaining the magnetorotational turbulence. We then supplement
34: these equations with a phenomenological description of the triple
35: correlations that lead to a saturated turbulent state. We show that
36: the steady-state limit of the model describes successfully the
37: correlations among stresses found in numerical simulations of
38: shearing boxes.
39: \end{abstract}
40:
41:
42:
43: \pacs{97.10.Gz, 95.30.Qd, 52.35.Ra, 52.30.Cv}
44: %97.10.Gz Accretion and accretion disks
45: %95.30.Qd Magnetohydrodynamics and plasmas (see also 52.30.Cv
46: % and 52.72.+v in physics of plasmas)
47: %52.35.Ra Plasma turbulence
48: %52.30.Cv Magnetohydrodynamics (including electron
49: % magnetohydrodynamics) (see also 47.65.-d Magnetohydrodynamics
50: % and electrohydrodynamics in fluid dynamics; for MHD generators,
51: % see 52.75.Fk; see also 95.30.Qd Magnetohydrodynamics and plasmas
52: % in astrophysics)
53: \maketitle
54:
55: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
56:
57: Since the early days of accretion disk theory, it has been recognized
58: that molecular viscosity cannot account for a number of observational
59: properties of accreting objects. Shakura and Sunyaev \cite{SS73}
60: introduced a parametrization of the shear stress that has been widely
61: used since. Much of the success of their model lies on that many disk
62: observables are determined mostly by energy balance and depend weakly
63: on the adopted prescription \cite{BP99}. However, this
64: parametrization leaves unanswered fundamental questions on the origin
65: of the anomalous transport and its detailed characteristics.
66:
67: Strong support for the relevance of magnetic fields in accretion disks
68: arose with the realization that differentially rotating flows with
69: radially decreasing angular velocities are unstable when threaded by
70: weak magnetic fields \cite{BH91-98}. Since the discovery of this
71: magnetorotational instability (MRI), a variety of local
72: \cite{HGB95-96,Sanoetal04} and global \cite{H00-01,SP01,HK01}
73: numerical simulations have confirmed that its long-term evolution
74: gives rise to a sustained turbulent state and outward angular momentum
75: transport. However, global simulations also demonstrate that angular
76: momentum transfer in turbulent accretion disks cannot be
77: adequately described by the Shakura \& Sunyaev prescription. In
78: particular, there is evidence that the turbulent stresses are not
79: proportional to the local shear \cite{ABL96} and are not even
80: determined locally \cite{Armitage98}.
81:
82: Some attempts to eliminate these shortcomings have been made within
83: the formalism of mean-field magnetohydrodynamics \cite{KR80}.
84: This has been a fruitful approach for modeling the growth of mean
85: magnetic fields in differentially rotating media \cite{BS05}. However,
86: it has proven difficult to use a dynamo model to describe the
87: transport of angular momentum in these systems \cite{Blackman01,
88: Brandenburg05}. This is especially true in turbulent flows driven by
89: the MRI, where the fluctuations in the magnetic energy are larger than
90: the mean magnetic energy, and the turbulent velocity and magnetic
91: fields evolve simultaneously \cite{BH91-98}.
92:
93: In order to overcome this difficulty various approaches have been
94: taken in different physical setups. Blackman and Field~\cite{BF02}
95: derived a dynamical model for the nonlinear saturation of helical
96: turbulence based on a damping closure for the electromotive force in
97: the absence of shear. On the other hand, Kato and Yoshizawa
98: \cite{KY93-95,Kato98} and Ogilvie~\cite{Ogilvie03} derived a set of
99: closed dynamical equations that describe the growth and saturation of
100: the Reynolds and Maxwell stresses in shearing flows in the absence of
101: mean magnetic fields.
102:
103: In this {\em Letter}, we relax some of the key assumptions made in
104: previous works and develop the first local model for the dynamical
105: evolution of the Reynolds and Maxwell tensors in turbulent magnetized
106: accretion disks that incorporates explicitly the MRI as their source.
107:
108: The equation describing the dynamical evolution of the mean angular
109: momentum density, $\bar{l}$, of a fluid element in an accretion disk
110: with tangled magnetic fields is
111: \begin{equation}
112: \label{eq:angular_momentum_mean_weak}
113: \partial_t \bar l + \del \bcdot (\bar l \bar{\bb{v}}) = - \del \bcdot
114: (r \bar{\bb{\mathcal F}}) \,.
115: \end{equation}
116: Here, the over-bars denote properly averaged values, $\bar{\bb{v}}$ is
117: the mean flow velocity, and the vector $\bar{\bb{\mathcal F}}$
118: characterizes the flux of angular momentum. Its radial component,
119: $\bar{\mathcal F}_r \equiv \bar{R}_{r\phi} - \bar{M}_{r\phi}$, is
120: equal to the total stress acting on a fluid element. The stresses due
121: to correlations in the velocity field, $\bar{R}_{r\phi} \equiv \langle
122: \rho \delta\!v_r \delta\!v_\phi\rangle$, and magnetic field
123: fluctuations, $\bar{M}_{r\phi} \equiv \langle\delta\!B_r
124: \delta\!B_\phi\rangle/4\pi$, are the Reynolds and Maxwell stresses,
125: respectively. A self-consistent accretion disk model based on the
126: solution of the equations for the mean velocities requires a closed
127: system of equations for the temporal evolution of these mean stresses.
128:
129: In a recent paper \cite{PCP06}, we identified the signature of the
130: magnetorotational instability in the mean Maxwell and Reynolds
131: stresses due to correlated fluctuations of the form $\delta
132: \bb{v}=[\delta v_r(t,z), \delta v_\phi(t,z), 0]$ and $\delta
133: \bb{B}=[\delta B_r(t,z), \delta B_\phi(t,z), 0]$ in an incompressible,
134: cylindrical, differentially rotating flow, threaded by a mean vertical
135: magnetic field, $\bar{B}_z$. We demonstrated that a number of
136: properties of the mean stresses during the initial phase of
137: exponential growth of the MRI are approximately preserved in the
138: saturated state reached in local three-dimensional numerical
139: simulations. As a first step in our calculation, we aim to derive
140: dynamical equations for the mean stresses that describe this result.
141:
142: In order to work with dimensionless variables we consider the
143: characteristic time- and length-scales set by $1/\Omega_0$ and
144: $\bar{v}_{{\rm A} z}/\Omega_0$. Here, $\Omega_0$ and $\bar{v}_{{\rm A}
145: z}=\bar{B}_z/\sqrt{4\pi\rho_0}$ denote the local values of the angular
146: frequency and the (vertical) Alfv\'en speed in the disk with local
147: density $\rho_0$ at the fiducial radius $r_0$. The equations
148: governing the local dynamics of the (dimensionless) MRI-driven
149: fluctuations in Fourier space, are then given by \cite{PCP06}
150: \begin{eqnarray}
151: \label{eq:ft_vx_nodim}
152: \partial_t \hat{\delta v_r} &=& 2 \hat{\delta v_\phi} +
153: ik_n \hat{\delta b_r} \,, \\
154: \label{eq:ft_vy_nodim}
155: \partial_t \hat{\delta v_\phi} &=& (q-2) \hat{\delta v_r} +
156: ik_n \hat{\delta b_\phi} \,, \\
157: \label{eq:ft_bx_nodim}
158: \partial_t \hat{\delta b_r} &=& ik_n \hat{\delta v_r} \,, \\
159: \label{eq:ft_by_nodim}
160: \partial_t \hat{\delta b_\phi} &=& - q \hat{\delta b_r} +
161: i k_n \hat{\delta v_\phi} \,,
162: \end{eqnarray}
163: where $\hat{\delta v_i}$ and $\hat{\delta b_i}$ stand for the Fourier
164: transform of the dimensionless physical fluctuations $\delta v_i
165: /\bar{v}_{{\rm A} z}$ and $\delta B_i/\bar{B}_z$. The wavenumber,
166: $k_n$, denotes the mode with $n$ nodes in the vertical direction and
167: the parameter $q\equiv-d\ln\Omega/d\ln r|_{r_0}$ is a measure of the
168: local shear.
169:
170: Using the fact that the modes with vertical wavevectors dominate the
171: fast growth driven by the MRI, we obtain a set of equations to
172: describe the initial exponential growth of the Reynolds and Maxwell
173: stresses. We start from the equations for the fluctuations
174: (\ref{eq:ft_vx_nodim})--(\ref{eq:ft_by_nodim}) and use the fact that
175: the mean value of the product of two functions $f$ and $g$, with zero
176: means, is given by \cite{PCP06}
177: \begin{eqnarray}
178: \label{eq:mean_ft}
179: \langle f g\rangle (t) \equiv 2 \sum_{n=1}^{\infty}
180: \; {\rm Re}[\,\hat{f}(k_n,t)\, \hat{g}^*(k_n,t)\,] \,.
181: \end{eqnarray}
182: By combining different moments of equations
183: (\ref{eq:ft_vx_nodim})--(\ref{eq:ft_by_nodim}) we obtain the
184: dimensionless set
185: \begin{eqnarray}
186: \label{eq:mean_Rrr}
187: \partial_t \bar{R}_{rr} &=& 4 \bar{R}_{r\phi}+2 \bar{W}_{r\phi}
188: \,, \\
189: %
190: \label{eq:mean_Rrphi}
191: \partial_t \bar{R}_{r\phi} &=&
192: (q-2) \bar{R}_{rr} + 2 \bar{R}_{\phi\phi} - \bar{W}_{rr} +
193: \bar{W}_{\phi\phi} \,, \\
194: %
195: \label{eq:mean_Rphiphi}
196: \partial_t \bar{R}_{\phi\phi} &=& 2(q-2) \bar{R}_{r\phi}
197: - 2 \bar{W}_{\phi r} \,, \\
198: %
199: %
200: \label{eq:mean_Mrr}
201: \partial_t \bar{M}_{rr} &=& -2 \bar{W}_{r\phi}
202: \,, \\
203: %
204: \label{eq:mean_Mrphi}
205: \partial_t \bar{M}_{r\phi} &=& -q \bar{M}_{rr}
206: + \bar{W}_{rr} - \bar{W}_{\phi\phi}
207: \,, \\
208: %
209: \label{eq:mean_Mphiphi}
210: \partial_t \bar{M}_{\phi\phi} &=& -2q \bar{M}_{r\phi}
211: +2 \bar{W}_{\phi r}
212: \,,
213: \end{eqnarray}
214: where we have defined the tensor $\bar{W}_{ij}$ with components
215: \begin{eqnarray}
216: \bar{W}_{rr} &\equiv& \langle\delta v_r \delta j_r\rangle =
217: \,\,\,\,\, \langle\delta b_\phi \delta \omega_\phi\rangle\,,
218: \label{eq:W_rr}\\
219: %
220: \bar{W}_{r\phi} &\equiv& \langle \delta v_r \delta j_\phi \rangle =
221: -\langle\delta b_r \delta \omega_\phi\rangle\,, \\
222: %
223: \bar{W}_{\phi r} &\equiv& \langle \delta v_\phi \delta j_r \rangle =
224: -\langle\delta b_\phi \delta \omega_r\rangle\,, \\
225: %
226: \bar{W}_{\phi \phi} &\equiv& \langle \delta v_\phi \delta j_{\phi}\rangle =
227: \,\,\,\,\, \langle\delta b_r \delta \omega_r\rangle\,.
228: \label{eq:W_pp}
229: \end{eqnarray}
230: Here, $\delta j_i$ and $\delta \omega_i$, for $i=r,\phi$, stand for
231: the components of the induced current $\delta \bb{j} = \del \btimes
232: \delta \bb{b}$ and vorticity $\delta \bb{\omega} = \del \btimes \delta
233: \bb{v}$ fluctuations. Note that the components of the tensor
234: $\bar{W}_{ij}$ are defined in terms of the correlations between the
235: velocity and current fields. However, for the case under
236: consideration, these are identical to the corresponding correlations
237: between the magnetic field and vorticity fluctuations
238: (eqs.~[\ref{eq:W_rr}]-[\ref{eq:W_pp}]).
239:
240: Equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi}) show that the
241: MRI-driven growth of the Reynolds and Maxwell tensors can be described
242: formally {\it only} via the correlations $\bar{W}_{ij}$ that connect
243: the equations for their temporal evolution. Note that, in contrast to
244: the correlations $\langle \delta v_i \delta \omega_k\rangle$ and
245: $\langle \delta b_i \delta j_k \rangle$, that appear naturally in
246: helical dynamo modeling and transform as tensor densities (i.e., as
247: the product of a vector and an axial vector), the correlations
248: $\langle \delta v_i \delta j_k\rangle$ and $\langle \delta b_i \delta
249: \omega_k \rangle$ transform as tensors. Moreover, the tensor
250: $\bar{W}_{ij}$ cannot be recast in terms of the cross-helicity tensor
251: $\bar{H}_{ij}\equiv\langle\delta v_i \delta b_j\rangle$ because, for
252: the unstable MRI modes, the ratio $\bar{H}_{ij}/\bar{W}_{ij}$
253: approaches zero at late times. Neither can $\bar{W}_{ij}$ be
254: expressed in terms of the turbulent electromotive force $\langle\delta
255: \bb{v} \btimes \delta \bb{b}\rangle$, as the latter vanishes under our
256: set of assumptions, implying that no mean magnetic field is generated.
257:
258: In order to describe the MRI-driven exponential growth of the stresses
259: in equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi}) we need to
260: write an additional set of dynamical equations for the evolution of
261: the tensor $\bar{W}_{ij}$. Using appropriate combinations of different
262: moments of equations (\ref{eq:ft_vx_nodim})--(\ref{eq:ft_by_nodim}) we
263: obtain
264: \begin{eqnarray}
265: \label{eq:mean_Wrr}
266: \partial_t \bar{W}_{rr} \!\!&=&\!\! q \bar{W}_{r \phi} + 2 \bar{W}_{\phi r} +
267: (\bar{k}^R_{r\phi})^2 \bar{R}_{r\phi} - (\bar{k}^M_{r\phi})^2 \bar{M}_{r\phi}\,, \\
268: %
269: \label{eq:mean_Wrphi}
270: \partial_t \bar{W}_{r\phi} \!\!&=&\!\! 2 \bar{W}_{\phi \phi} -
271: (\bar{k}^R_{rr})^2 \bar{R}_{rr} + (\bar{k}^M_{rr})^2 \bar{M}_{rr}\,, \\
272: %
273: \label{eq:mean_Wphir}
274: \partial_t \bar{W}_{\phi r} \!\!&=&\!\! (q-2) \bar{W}_{rr} + q \bar{W}_{\phi\phi} +
275: (\bar{k}^R_{\phi\phi})^2 \bar{R}_{\phi\phi} - (\bar{k}^M_{\phi\phi})^2 \bar{M}_{\phi\phi}\,,\ \ \ \ \ \ \\
276: %
277: \label{eq:mean_Wphiphi}
278: \partial_t \bar{W}_{\phi\phi} \!\!&=&\!\! (q-2) \bar{W}_{r\phi} -
279: (\bar{k}^R_{r\phi})^2 \bar{R}_{r\phi} + (\bar{k}^M_{r\phi})^2 \bar{M}_{r\phi}\,,
280: \end{eqnarray}
281: where we have defined the set of mean wavenumbers
282: \begin{eqnarray}
283: (\bar{k}^R_{ij})^2 \equiv
284: \frac{\displaystyle \sum_{n=1}^{\infty} k_n^2 \real[\hat{\delta v_i}
285: \hat{\delta v_j}\!\!^*]}
286: {\displaystyle \sum_{n=1}^{\infty} \real[\hat{\delta v_i}
287: \hat{\delta v_j}\!\!^*]} \,,\,
288: (\bar{k}^M_{ij})^2 \equiv
289: \frac{\displaystyle \sum_{n=1}^{\infty} k_n^2 \real[\hat{\delta b_i} \hat{\delta b_j}\!\!^*]}
290: {\displaystyle \sum_{n=1}^{\infty} \real[\hat{\delta b_i}
291: \hat{\delta b_j}\!\!^*]} \,. \ \ \ \ \nonumber
292: \label{eq:k_ave}
293: \end{eqnarray}
294:
295: The system of equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi})
296: and (\ref{eq:mean_Wrr})--(\ref{eq:mean_Wphiphi}) describes the
297: temporal evolution of the stresses during the exponential growth of
298: the MRI in a way that is formally correct, with no approximations.
299: Motivated by the similarity between the ratios of the stresses during
300: the exponential growth of the MRI and during the saturated turbulent
301: state~\cite{PCP06}, we propose to use the right-hand sides of
302: equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi}) and
303: (\ref{eq:mean_Wrr})--(\ref{eq:mean_Wphiphi}) as a local model for the
304: source of turbulence in MRI-driven magnetohydrodynamic flows. Of
305: course, in the turbulent regime, the various average wavenumbers
306: $\bar{k}^{R,M}_{ij}$ will depend on the spectrum of velocity and
307: magnetic field fluctuations. As the lowest order model for these
308: wavenumbers we choose
309: \begin{eqnarray}
310: \label{eq:kbar}
311: (\bar{k}^{R,M}_{ij})^2 = \zeta^2 k^2_{\rm max}
312: = \zeta^2 \left(q - \frac{q^2}{4}\right) \quad \textrm{for} \quad i,j =
313: r,\phi \,, \ \ \ \
314: \end{eqnarray}
315: where $\zeta$ is a parameter of order unity and $k_{\rm max}$
316: corresponds to the wavenumber at which the growth rate of the {\it
317: fluctuations} reaches its maximum value, $\gamma_{\rm max} \equiv
318: q/2$.
319:
320: \begin{figure}[t]
321: \epsfig{file=f1.eps, width=0.825\columnwidth,trim=0 10 0 10}
322: \caption{\footnotesize Correlations between the Maxwell and Reynolds
323: stresses and mean magnetic energy density at saturation in
324: MRI-driven turbulent shearing boxes \cite{HGB95-96}. The lines
325: show the result obtained with our model in the asymptotic limit
326: for $\zeta=0.3$.}
327: \label{fig:stresses_vs_emagnetic}
328: \end{figure}
329:
330:
331:
332: By construction, the set of
333: equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi}) and
334: (\ref{eq:mean_Wrr})--(\ref{eq:mean_Wphiphi}) leads to the expected
335: exponential growth of the mean stresses driven by modes with
336: wavevectors perpendicular to the disk midplane. These modes, however,
337: are known to be subject to parasitic instabilities, which transfer
338: energy to modes in the perpendicular ($k_r,k_\phi$) directions
339: \cite{GX94}. The initial fast growth experienced by the stresses will
340: eventually be slowed down by the combined effects of non-linear
341: couplings between modes and of dissipation at the smallest scales of
342: interest.
343:
344: The terms accounting for these interactions would appear in the
345: equations for the stresses as triple correlations between components
346: of the velocity and magnetic fields, i.e., they would be of the form
347: $\langle\delta v_i \delta v_j \delta b_k\rangle$ and $\langle\delta
348: v_i \delta b_j \delta b_k\rangle$. These types of non-linear terms
349: have also been considered by Ogilvie \cite{Ogilvie03}, who proposed
350: scalings of the form
351: \begin{equation}
352: \langle\delta v_i \delta v_j \delta b_k\rangle
353: \sim \langle\delta v_i \delta v_j\rangle \, \langle\delta b_k \delta
354: b_k\rangle^{1/2} \sim \bar{M}_{kk}^{1/2}\bar{R}_{ij} \,.
355: \end{equation}
356:
357: In the absence of a detailed model for these correlations and
358: motivated again by the similarity of the stress properties during the
359: exponential growth of the MRI and the saturated turbulent
360: state~\cite{PCP06}, we introduce a phenomenological description of the
361: non-linear effects on the evolution of the various stresses. In
362: particular, denoting by $\bar{X}_{ij}$ the $ij$-component of any one
363: of the three tensors, i.e., $\bar{R}_{ij}$, $\bar{M}_{ij}$, or
364: $\bar{W}_{ij}$, we add to the equation for the temporal evolution of
365: that component the sink term
366: \begin{equation}
367: \left.\frac{\partial\bar{X}_{ij}}{\partial t}\right\vert_{\rm sink} \equiv
368: -\sqrt{\frac{\bar{M}}{\bar{M}_0}} \bar{X}_{ij} \,.
369: \label{eq:sink}
370: \end{equation}
371: Here, $\bar{M}/2 = (\bar{M}_{rr} + \bar{M}_{\phi\phi})/2$ is the mean
372: magnetic energy density in the fluctuations and $\bar{M}_0$ is a
373: parameter.
374:
375:
376: Adding the sink terms to the system of
377: equations~(\ref{eq:mean_Rrr})--(\ref{eq:mean_Mphiphi}) and
378: (\ref{eq:mean_Wrr})--(\ref{eq:mean_Wphiphi}) leads to a saturation of
379: the stresses after a few characteristic timescales, {\em preserving\/}
380: the ratio of the various stresses to the value determined by the
381: exponential growth due to the MRI and characterized by the parameter
382: $\zeta$. These non-linear terms dictate only the saturated level of
383: the mean magnetic energy density according to $\lim_{t\rightarrow
384: \infty}{\bar{M}}= \Gamma^2 \, \bar{M_0}$, where $\Gamma$ is the growth
385: rate for the {\it stresses}, which, in the case of a Keplerian disk,
386: with $q=3/2$, is given by $\Gamma^2=2\left[\sqrt{1 + 15\zeta^2} -
387: \left(1+15\zeta^2/8\right) \right]$.
388:
389:
390: \begin{figure}[t]
391: \epsfig{file=f2.eps, width=0.825\columnwidth,trim=0 10 0 10}
392: \caption{\footnotesize The mean magnetic energy density in terms of a
393: saturation predictor found at late times in numerical simulations of
394: MRI-driven turbulence \cite{HGB95-96}. The line shows the result
395: obtained with our model in the asymptotic limit for
396: $\zeta=0.3$~and~$\xi=11.3$.}
397: \label{fig:saturator_predictor}
398: \end{figure}
399:
400:
401: We infer the dependence of the energy density scale $\bar{M}_0/2$ on
402: the four characteristic scales in the problem $\Omega_0$, $H$ (the
403: vertical length of the box), $\rho_0$, and $\bar{v}_{{\rm A} z}$ using
404: dimensional analysis. We obtain $\bar{M}_0/2 \propto \rho_0 H^\delta
405: \Omega_0^{\delta} \bar{v}_{{\rm A}z}^{2-\delta}$ which leads, with the
406: natural choice $\delta=1$, to $\bar{M}_0/2 \equiv \xi \rho_0 H\Omega_0
407: \bar{v}_{{\rm A} z}$, where we show the dimensional quantities
408: explicitly and introduce the parameter $\xi$.
409:
410: Our expression for $\bar{M}_0$ describes the same scaling between the
411: magnetic energy density during saturation and the various parameters
412: characterizing the disk found in a series of shearing box simulations
413: threaded by a finite vertical magnetic field and with a Keplerian
414: shearing profile \cite{HGB95-96}. By performing a numerical study of
415: the late-time solutions of the proposed model, we found a unique set
416: of values ($\zeta, \xi$) such that its asymptotic limit describes the
417: correlations found in these numerical simulations.
418:
419: Figure~\ref{fig:stresses_vs_emagnetic} shows the correlations between
420: the $r\phi$-com\-po\-nents of the Maxwell and Reynolds stresses and
421: the mean magnetic energy density found during the saturated state in
422: numerical simulations \cite{HGB95-96}. In our model, this ratio
423: depends only on the parameter $\zeta$ and the shear $q$ (held fixed at
424: $q=3/2$ in the simulations). It is evident that, in the numerical
425: simulations, the ratios of the stresses to the magnetic energy density
426: are also practically independent of any of the initial parameters in
427: the problem that determine the magnetic energy density during
428: saturation (i.e., the $x$-axis in the plot). This is indeed why we
429: required for our model of saturation to preserve the stress ratios
430: that are determined during the exponential phase of the MRI. Assigning
431: the same fractional uncertainty to all the numerical values for the
432: stresses, our model describes both correlations simultaneously for
433: $\zeta=0.3$.
434:
435: Figure~\ref{fig:saturator_predictor} shows the mean magnetic energy
436: density in terms of a saturation predictor found in numerical
437: simulations \cite{HGB95-96}. For $\zeta=0.3$ and assigning the same
438: fractional uncertainty to all the numerical values for the stresses,
439: we obtain the best fit for $\xi=11.3$. These values for the parameters
440: complete the description of our model.
441:
442: In summary, in this {\it Letter} we developed a local model for the
443: evolution and saturation of the Reynolds and Maxwell stresses in
444: MRI-driven turbulent flows. The model is formally complete when
445: describing the initial exponential growth and pumping of the
446: MRI-driven stresses and, thus, satisfies, by construction, all the
447: mathematical requirements described by Ogilvie~\cite{Ogilvie03}.
448: Although it is based on the absolute minimum physics (shear, uniform
449: $\bar{B}_z$, $2D$-fluctuations) for the MRI to be at work, the model
450: is able, in its asymptotic limit, to recover successfully the
451: correlations found in three-dimensional local numerical simulations
452: \cite{HGB95-96}.
453:
454: Finally, the local model described in this {\em Letter} contains an
455: unexpected feature. The mean magnetic field in the vertical direction
456: couples the Reynolds and Maxwell stresses via the correlations between
457: the fluctuations in the velocity field and the fluctuating currents
458: generated by the perturbations in the magnetic field. These second
459: order correlations, that we denoted by the tensor $\bar{W}_{ij}$, play
460: a crucial role in driving the exponential growth of the mean stresses
461: and energy densities observed in numerical simulations. To our
462: knowledge this is the first time that their relevance has been pointed
463: out in either the context of dynamo theory or MRI-driven turbulence.
464:
465: \bigskip
466:
467: We thank Eric Blackman, Gordon Ogilvie, Jim Stone, and an anonymous
468: referee for useful discussions. MEP acknowledges the hospitality of
469: the Institute for Advanced Study during part of this work. MEP is
470: supported through a Jamieson Fellowship at the Astronomy Department at
471: the UA. This work was partially supported by NASA grant NAG-513374.
472:
473: %------------------------------------------------------%
474:
475: \bibliographystyle{apsrev}
476: \begin{thebibliography}{99}
477:
478: \bibitem{SS73}
479: N. I. Shakura and R. A. Sunyaev, {\em Astron. Astrophys.}, {\bf 24}, 337 (1973)
480:
481: \bibitem{BP99}
482: S. A. Balbus and J. C. B. Papaloizou, {\em Astrophys. J.}, {\bf 521}, 650 (1999)
483:
484: \bibitem{BH91-98}
485: S. A. Balbus and J. F. Hawley, {\em Astrophys. J.}, {\bf 376}, 214
486: (1991); {\em Rev. Mod. Phys.}, {\bf 70}, 1 (1998)
487:
488: \bibitem{HGB95-96}
489: J. F. Hawley, C. F. Gammie, and S. A. Balbus, {\em Astrophys. J.},
490: {\bf 440}, 742 (1995); {\em Astrophys. J.}, {\bf 464}, 690 (1996)
491:
492: \bibitem{Sanoetal04}
493: T. Sano, S. Inutsuka, N. J. Turner, and J. M. Stone, {\em Astrophys. J.}, {\bf 605}, 321 (2004)
494:
495: \bibitem{H00-01}
496: J. F. Hawley, {\em Astrophys. J.}, {\bf 528}, 462 (2000); {\em Astrophys. J.}, {\bf 554}, 534 (2001)
497:
498: \bibitem{SP01}
499: J. M. Stone, and J. E. Pringle, {\em Mon. Not. R. Astron. Soc.}, {\bf 322}, 461 (2001)
500:
501: \bibitem{HK01}
502: J. F. Hawley and J. H. Krolik, {\em Astrophys. J.}, {\bf 548}, 348 (2001)
503:
504: \bibitem{ABL96}
505: M. Abramowicz, A. Brandenburg, and J. P. Lasota, {\em Mon. Not. R. Astron. Soc.}, {\bf 281}, L21 (1996)
506:
507: \bibitem{Armitage98}
508: P. J. Armitage, {\em Astrophys. J., Lett}, {\bf 189}, 501 (1998)
509:
510: \bibitem{KR80}
511: F. Krause and K. H. R\"adler, {\em Mean-Field Magnetohydrodynamics and Dynamo Theory}, Pergamon (1980)
512:
513: \bibitem{BS05}
514: A. Brandenburg and K. Subramanian, {\em Phys. Rep.}, {\bf 417}, 1 (2005)
515:
516: \bibitem {Blackman01}
517: E. G. Blackman, {\em Mon. Not. R. Astron. Soc.}, {\bf 323}, 497 (2001)
518:
519: \bibitem{Brandenburg05}
520: A. Brandenburg, {\em Astron. Nachr.}, {\bf 326}, 787 (2005)
521:
522: \bibitem {BF02}
523: E. G. Blackman and G. B. Field, {\em Phys. Rev. Lett.}, {\bf 89}, 265007 (2002)
524:
525: \bibitem{KY93-95}
526: S. Kato and A. Yoshizawa, {\em Publ. Astron. Soc. Jap.}, {\bf 45}, 103
527: (1993); {\em Publ. Astron. Soc. Jap.}, {\bf 47}, 629 (1995)
528:
529: \bibitem{Kato98}
530: S. Kato, J. Fukue, and S. Mineshige, {\em Black Hole Accretion Disks},
531: Kioto Univ. Press. (1998)
532:
533: \bibitem{Ogilvie03}
534: G. I. Ogilvie, {\em Mon. Not. R. Astron. Soc.}, {\bf 340}, 969 (2003)
535:
536: \bibitem{GX94}
537: J. Goodman and G. Xu, {\em Astrophys. J.}, {\bf 432}, 213 (1994)
538:
539: \bibitem{PCP06}
540: M. E. Pessah, C.K. Chan, and D. Psaltis, {\em Mon. Not. R. Astron. Soc.}, {\bf 372}, 183 (2006)
541:
542: \end{thebibliography}
543:
544: \end{document}
545:
546: