1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3: %\documentclass[preprint2]{aastex}
4: %\documentclass[preprint2,longabstract]{aastex}
5:
6: \slugcomment{Submitted to The Astrophysical Journal}
7:
8: \shorttitle{Magnetically Arrested Disks}
9: \shortauthors{I. V. Igumenshchev}
10:
11: \begin{document}
12:
13: \title{MAGNETICALLY ARRESTED DISKS AND\\
14: ORIGIN OF POYNTING JETS: NUMERICAL STUDY}
15:
16: \author{Igor V. Igumenshchev}
17: \affil{Laboratory for Laser Energetics, University of Rochester\\
18: 250 East River Road, Rochester, NY 14623}
19: \email{iigu@lle.rochester.edu}
20:
21: \begin{abstract}
22:
23: The dynamics and structure of accretion disks, which accumulate the
24: vertical magnetic field in the centers, are investigated
25: using two- and three-dimensional MHD simulations.
26: The central field can be built up to the equipartition level and disrupts
27: a nearly axisymmetric outer accretion disk inside a magnetospheric radius,
28: forming a magnetically arrested disk (MAD).
29: In the MAD, the mass accretes in a form of irregular dense spiral streams
30: and the vertical field, split into separate bundles,
31: penetrates through the disk plane in low-density magnetic islands.
32: The accreting mass, when spiraling inward, drags the field and
33: twists it around the axis of rotation,
34: resulting in collimated Poynting jets in the polar directions.
35: These jets are powered by the accretion flow with the efficiency
36: up to $\sim 1.5\%$ (in units $\dot{M}_{in}c^2$).
37: The spiral flow pattern in the MAD
38: is dominated by modes with low azimuthal wavenumbers $m\sim 1-5$ and
39: can be a source of quasi-periodic oscillations in the outgoing radiation.
40: The formation of MAD and Poynting jets can naturally explain the
41: observed changes of spectral states in Galactic black hole binaries.
42: Our study is focused on black hole accretion flows;
43: however, the results
44: can also be applicable to accretion disks around nonrelativistic
45: objects, such as young stellar objects and stars in binary systems.
46:
47: \end{abstract}
48:
49: \keywords{accretion, accretion disks --- black hole physics --- galaxies:
50: jets --- gamma rays: bursts --- instabilities --- ISM:
51: jets and outflows --- magnetic fields --- MHD --- turbulence}
52:
53:
54: \section{Introduction}
55:
56: Accretion disks can carry small- and large-scale magnetic fields.
57: The small-scale field ($\ell\la R$, where $\ell$ is the field
58: scale length and $R$ measures the radial distance from the disk center)
59: can be locally generated by the MHD dynamo
60: (Brandenburg et al. 1995; Stone et al. 1996) supported by
61: the turbulence, which results from the
62: magneto-rotational instability (MRI, Balbus \& Hawley 1991).
63: This field can provide the outward transport of angular momentum
64: in the bulk of the disk with the help of local Maxwell stresses
65: (Shakura \& Sunyaev 1973; Hawley, Gammie, \& Balbus 1996).
66: The large-scale field ($\ell > R$) is unlikely produced in
67: accretion disks (however, see Tout \& Pringle 1996),
68: and can either be captured from the environment and
69: dragged inward by an accretion flow (Bisnovatyi-Kogan \& Ruzmaikin
70: 1974, 1976), or inherited from the past evolution (see \S 4).
71: The large-scale field can remove the angular momentum from accretion disks
72: by global Maxwell stresses through the magnetized disk corona (K\"onigl 1989).
73:
74: A large-scale bipolar field, unlike a small-scale field,
75: can not dissipate locally due to the magnetic diffusivity
76: and can not be absorbed by the central black hole.
77: In the case of inefficient outward diffusion of the bipolar field
78: through the disk
79: (see Narayan, Igumenshchev, \& Abramowicz 2003; Spruit \& Uzdensky 2005;
80: Bisnovatyi-Kogan \& Lovelace 2007; and
81: for other possibility, see van Ballegooijen 1989;
82: Lubow, Papaloizou, \& Pringle 1994;
83: Lovelace, Romanova, \& Newman 1994; Heyvaerts, Priest, \& Bardou 1996;
84: Agapitou \& Papaloizou 1996; Livio, Ogilvie, \& Pringle 1999),
85: this field is accumulated in the innermost region of an accretion disk
86: and forms a ``magnetically arrested disk,'' or MAD
87: (Narayan et al. 2003).
88: The MAD consists of two parts: the outer, almost axisymmetric,
89: Keplerian accretion disk and
90: the inner magnetically dominated region, in which the accumulated
91: vertical field disrupts the outer disk at the magnetospheric
92: radius $R_{\rm m}\sim 8\pi GM\rho/B^2$, where $M$ is the central mass,
93: $\rho$ is the accretion mass density, and $B$ is the magnetic induction.
94:
95: It is believed that the large-scale bipolar field in accretion disks
96: is responsible for the formation of jets
97: observed in a large variety of astrophysical objects
98: (e.g., Livio, Pringle, \& King 2003).
99: The magnetically driven jets can be of two types,
100: basically depending on a mass load by the disk matter
101: (e.g., Lovelace, Gandhi, \& Romanova 2005):
102: Poynting jets and hydromagnetic jets, which
103: have, respectively, small and large mass loads.
104: The hydromagnetic jets can be formed by two mechanisms:
105: the magneto-centrifugal mechanism
106: (Blandford \& Payne 1982; K\"onigl \& Pudritz 2000)
107: and the toroidal-field pressure generated by the disk rotation
108: (Lynden-Bell 2003; Kato, Mineshige, \& Shibata 2004).
109: The magneto-centrifugal mechanism
110: produces relatively wide outflows and, to be consistent with observations
111: of the collimated jets, requires an additional focusing mechanism.
112: Jets driven by the toroidal-field pressure can have
113: a high degree of collimation, but these jets are known to be kink
114: unstable (Eichler 1993; Appl 1996; Spruit, Foglizzo, \& Stehle 1997).
115: Poynting jets are naturally self-collimated and
116: expected to be marginally kink stable
117: (Li 2000; Tomimatsu, Matsuoka, \& Takahashi 2001).
118: These jets can originate in the innermost region of accretion disks
119: and powered either by the disks themselves (Lovelace, Wang, \& Sulkanen 1987;
120: Lovelace et al. 2002)
121: or by rotating black holes (Blandford \& Znajek 1977; Punsly 2001;
122: also see,
123: Takahashi et al. 1990; Komissarov 2005; Hawley \& Krolik 2006; McKinney 2006).
124:
125: Our study is based on two- and three-dimensional
126: (2D and 3D, respectively) MHD simulations
127: and has two main goals.
128: First, we investigate the dynamics and structure of MAD's.
129: We show that MAD's are formed in the accretion flows, which carry inward
130: large-scale poloidal magnetic fields.
131: Inside the magnetospheric radius $R_{\rm m}$,
132: the matter accretes as discrete streams and blobs, fighting its way
133: through the strong vertical magnetic field fragmented in separate bundles.
134: Because of rotation, the streams take spiral shapes.
135: Second, we demonstrate a link between the existence of MAD's and production
136: of powerful Poynting jets.
137: These jets should always be generated in MAD's
138: because of the interaction of the spiraling accretion flow
139: with the vertical magnetic bundles, which, as the result, are twisted around
140: the axis of rotation.
141:
142: This paper extends the work of Igumenshchev, Narayan, \& Abramowicz (2003)
143: by studying in more detail the radiatively inefficient accretion disks
144: with poloidal magnetic fields.
145: We employ a new version of our 3D MHD code, which can be utilized in
146: multi-processor simulations.
147: The paper is organized as follows:
148: We describe the solved equations, the numerical method used, and
149: initial and boundary conditions in \S 2.
150: We present our numerical results in \S 3, and
151: discuss and summarize them in \S 4.
152:
153:
154: \section{Numerical method}
155:
156: We simulate nonradiative accretion flows around
157: a Schwarzschild black hole of mass $M$ using the following equations
158: of ideal MHD:
159: \begin{equation}
160: {d\rho\over dt} + \rho{\bf\nabla\cdot v} = 0,
161: \end{equation}
162: \begin{equation}
163: \rho{d{\bf v}\over dt} = -{\bf\nabla} P - \rho{\bf\nabla}\Phi +
164: {1\over 4\pi}({\bf\nabla}\times{\bf B})\times {\bf B},
165: \end{equation}
166: \begin{equation}
167: {\partial\over\partial t}\left(\rho{v^2\over 2}+\rho\epsilon+
168: {B^2\over 8\pi}+\Phi\right)=-\nabla\cdot{\bf q},
169: \end{equation}
170: %\begin{equation}
171: %\rho{d\epsilon\over dt} = -(P_g+Q){\bf\nabla\cdot v} + {1\over
172: %4\pi}\eta{\bf J}^2,
173: %\end{equation}
174: \begin{equation}
175: {\partial{\bf B}\over \partial t} = {\bf\nabla}\times({\bf v}\times{\bf B}),
176: \end{equation}
177: where
178: %$\rho$ is the density,
179: ${\bf v}$ is the velocity, $P$ is the
180: gas pressure, $\Phi$ is the gravitational potential,
181: %${\bf B}$ is the magnetic induction,
182: $\epsilon$ is the specific internal energy of gas,
183: and ${\bf q}$ is the total energy flux per unit square
184: (see, e.g., Landau \& Lifshitz 1987).
185: We adopt the ideal gas equation of state,
186: \begin{equation}
187: P=(\gamma-1)\rho\epsilon,
188: \end{equation}
189: with an adiabatic index $\gamma=5/3$.
190: We neglect self-gravity of the gas
191: and employ a pseudo-Newtonian approximation (Paczy\'nski \&
192: Wiita 1980) for the black hole potential
193: \begin{equation}
194: \Phi=-{GM\over R-R_g},
195: \end{equation}
196: where $R_g=2GM/c^2$ is the gravitational radius of the black hole.
197: No explicit resistivity and viscosity are applied in equations (2)-(4).
198: However, because of the use of the total energy equation (3),
199: the energy released
200: due to the numerical resistivity and viscosity is consistently accounted
201: as heat in our simulations.
202:
203: The MHD equations (1)-(4) are solved employing the time-explicit
204: Eulerian finite-difference method, which is an extension
205: to MHD of the hydrodynamic piecewise-parabolic method by Colella \&
206: Woodward (1984).
207: We solve the induction equation (4) using the constrained transport
208: (Evans \& Hawley 1988; Gardiner \& Stone 2005), which preserves
209: the $\nabla\cdot{\bf B}=0$ condition.
210: In our method, we employ the approximate MHD Riemann solver by Li (2005).
211: Test simulations have shown that this solver is robust and provides
212: a good material interface tracking.
213:
214: We use the spherical coordinates $(R,\theta,\phi)$.
215: Our 3D numerical grid has $182\times 84\times 240$
216: zones in the radial, polar, and azimuthal directions, respectively.
217: The radial zones are spaced logarithmically from $R_{in}=2 R_{g}$
218: to $R_{out}=220 R_{g}$.
219: Both hemispheres are considered, in which polar cones with the opening
220: angle $\pi/8$ are excluded. Therefore, the polar domain extends
221: from $\theta=\pi/16$ to $15\pi/16$.
222: The grid resolution in the polar direction is gradually changed from
223: a fine resolution around the
224: equatorial plane to a coarse resolution near the poles
225: (with the maximum-to-minimum grid-size ratio $\approx 3$).
226: The azimuthal zones are uniform and cover the full $2\pi$ range
227: in $\phi$.
228: The absorption condition for the mass and
229: the condition of the zero-transverse magnetic field
230: are applied in the inner and outer radial boundaries,
231: providing that the mass and field can freely leave the computational
232: domain through these boundaries.
233: In the boundaries around the excluded polar regions,
234: we apply the reflection boundary conditions,
235: which means that no streamlines and magnetic lines can go through
236: these boundaries.
237:
238: At the beginning of our simulations,
239: the computational domain is filled with a very low-density,
240: nonmagnetized material.
241: The simulations are started in 2D, assuming the axial symmetry,
242: with an injection of mass in a slender torus, which is
243: located in the equatorial plane at $R_{inj}=210\, R_{g}$;
244: i.e., close enough to $R_{out}$.
245: This mass has the Keplerian angular momentum and specific internal energy
246: $\epsilon_{inj}=0.045\,GM/R_{inj}$.
247: After an initial period of simulation
248: without magnetic fields, the mass forms a steady
249: thick torus, which has the inner edge at $R\approx 150\,R_{g}$ and
250: which outer half is truncated at $R_{out}$.
251: The torus contains a constant amount of mass and is in a dynamic equilibrium:
252: all the injected mass flows outward through
253: $R_{\rm out}$ after a circulation inside the torus.
254: No accretion flow at this point is formed.
255: This hydrodynamic, steady, thick torus is used as an initial configuration
256: in our MHD simulations.
257:
258: The MHD simulations are started at $t=0$ from the steady, thick torus
259: by initiating the injection of a poloidal magnetic field
260: into the injection slender torus at $R_{inj}$.
261: The numerical procedure for the field injection is
262: similar to that described by Igumenshchev et al. (2003)
263: except for one modification:
264: now the strength of the injected field
265: can be limited by setting the minimum $\beta_{\rm inj}$,
266: which is the ratio of the gas pressure to
267: the magnetic pressure at $R_{\rm inj}$.
268: This modification allows us to better control the rate of field injection.
269: The entire volume of the thick torus is filled by the field during
270: about one orbital period, $t_{\rm orb}$, estimated at $R_{\rm inj}$.
271: Since this moment, $t\simeq t_{orb}$, the formation of accretion flow begins
272: as a result of redistribution of the angular momentum in the torus due to
273: Maxwell stresses.
274: In the following discussion, we will use the time normalized by
275: the time-scale $t_{orb}$, i.e. $t\rightarrow t/t_{\rm orb}$.
276:
277:
278: \section{Results}
279:
280: We present the results of combined axisymmetric 2D and
281: non-axisymmetric 3D simulations. The initial evolution
282: in our models has been simulated in 2D.
283: This allows us to consider longer evolution times
284: in comparison to those that can be obtained in 3D simulations,
285: because of the larger requirements for computational resources
286: in the latter case.
287: We initiate 3D simulations starting from developed
288: axisymmetric models. The results
289: have shown that non-axisymmetric motions are not very important
290: in the outer parts of the constructed accretion flows and, therefore,
291: the use of 2D simulations on the initial evolution stages is
292: the reasonable simplification.
293:
294: We consider three models, which differ by
295: the rates of field injection determined by
296: $\beta_{\rm inj}=10$, 100, and 1000, and we will refer to these models
297: as Model~A, B, and C, respectively. All other properties of the models,
298: including the injection radius $R_{inj}$, internal energy $\epsilon_{inj}$,
299: and numerical resolution, are the same (see \S 2).
300:
301:
302: \subsection{Accretion flows}
303:
304: The initial axisymmetric development of the models is
305: qualitatively similar: the inner edge of the thick torus
306: is extended toward the black hole, forming relatively
307: thin, almost Keplerian accretion disks.
308: The time of the disk development
309: is varied, depending on the strength of the injected field.
310: The accretion of the mass into the black hole begins at
311: $t\approx 0.7$ in Model~A,
312: $\approx 1.3$ in Model~B, and $\approx 4.2$ in Model~C
313: (time is given in units of the orbital period at $R_{inj}$, see \S 2).
314: At this stage, the evolution of the
315: disks is governed mainly by global Maxwell stresses
316: produced by the poloidal field component.
317: This component is advected inward with the accretion flow and,
318: because of the disks' Keplerian rotation,
319: generates relatively strong toroidal magnetic fields localized above and
320: below the mid-plane.
321: These toroidal fields form a highly magnetized disk corona with a typical
322: $\beta\sim 0.01$. Model~B and, especially Model~C, demonstrate
323: the development of 2D MRI.
324: This development is similar to that
325: observed by Stone \& Pringle (2001); in particular, in their ``Run C."
326: We have found the origin of the
327: channel solution (see Hawley \& Balbus 1992)
328: in the central regions of Models~B and C.
329: This solution consists of oppositely directed radial streams
330: and is the characteristic feature of the axisymmetric
331: non-linear MRI (Stone \& Norman 1994).
332: Analysis of the models shows that the channel solution is developed when the
333: wavelength $\lambda=2\pi V_A/\sqrt{3}\Omega$
334: of the fastest growing mode of the MRI is
335: well resolved on the numerical grid,
336: i.e. $\lambda\ga 5\Delta x$, where $\Delta x$
337: is the grid size, $V_A$ is the Alfv\'en velocity, and $\Omega$ is
338: the angular velocity.
339: Model~A shows no indication of the MRI, which can be attributed
340: to the strong magnetic fields, which suppress the instability.
341: In this model, the estimate of $\lambda$ typically exceeds
342: the disk thickness.
343: Model~A has some resemblance to nonturbulent
344: ``Run F'' of Stone \& Pringle (2001).
345: In spite of the mentioned similarities with the results of
346: Stone \& Pringle (2001),
347: our models show different behavior on the long evolution times.
348: Our simulation design with
349: the permanent injection of mass and magnetic field results in
350: accretion disks, which accumulate the poloidal field
351: in the center and form MAD's. The models of Stone \& Pringle (2001; also
352: %Hawley 2000; Hawley, Balbus, \& Stone 2001; Hawley \& Balbus 2002;
353: %Hawley, \& Krolik 2002;
354: De Villiers, Hawley, \& Krolik 2003; Hirose et al. 2004;
355: McKinney \& Gammie 2004; Hawley \& Krolik 2006)
356: did not form MAD's and did not show a long-time
357: accretion history, probably because of the initiation of simulations from
358: static magnetized tori, which contain
359: a limited amount of mass and magnetic flux of one sign.
360: We will concentrate on the results describing the formation, evolution,
361: and structure of MAD's in the following text.
362: Other aspects of our results will be reported elsewhere.
363:
364: Figure~1 shows example snapshots of the axisymmetric density distribution in
365: Model~B from 2D simulations at two successive moments,
366: %$t=5.452$ and $5.487$
367: $t=5.1153$ and $5.1458$.
368: %(given in units of the orbital period at $R_{inj}$, see \S 2).
369: The accretion flow is nonuniform because of the development of turbulence.
370: The turbulence results from the combined effect
371: of the MRI and current sheet instability. The latter instability locally
372: releases heat due to reconnections of the oppositely directed toroidal
373: magnetic fields.
374: The reconnection heat makes a significant contribution to the local energy
375: balance in the central regions of the flow,
376: because of the relatively high energy density of the field, which
377: is comparable to the gravitational energy density of the accretion mass.
378: The thick disk structure observed in Fig.~1 is explained by
379: convection motions supported by the reconnection heat.
380: Note that the case, in which the turbulence is
381: supported by only convection motions
382: from the reconnections,
383: without the effects of rotation and MRI,
384: had been demonstrated in simulations of spherical magnetized
385: accretion flows (Igumenshchev 2006).
386: In the case of disk accretion,
387: %Note that similar,
388: almost axisymmetric convection motions, similar to those found here,
389: had been observed in 3D models
390: % of accretion disks
391: with toroidal magnetic fields (Igumenshchev et al. 2003).
392: The convection motions in Models~B and C make
393: these models relevant to
394: convection-dominated accretion flows (Narayan, Igumenshchev, \&
395: Abramowicz 2000; Quataert \& Gruzinov 2000).
396: Our simulations show that the poloidal
397: field is transported inward in axisymmetric turbulent flows
398: and accumulated in the vicinity of the black holes. When the central poloidal
399: field reaches some certain strength (about equipartition with
400: the gravitational energy of the accreting mass), the accretion flow
401: becomes unstable (Narayan et al. 2003).
402: In axisymmetric simulations, the instability takes the form of cycle
403: accretion, in which the more-extended periods of halted accretion
404: (see Fig.~1a) are followed by
405: the relatively short periods of accretion (see Fig.~1b).
406: In the case of the halted accretion period, the inner accretion disk is
407: truncated at the magnetospheric radius $R_{m}$, which is
408: $\approx 15\,R_g$ in Fig.~1a.
409: The pressure of the strong central vertical field (see Fig.~2a) prevents
410: the mass accumulated at $R_{m}$ from falling into the black hole.
411: The accretion begins as soon as the gravity of
412: the accumulated mass overcomes the magnetic pressure.
413: During the accretion period, the whole magnetic flux,
414: which is localized inside $R_{\rm m}$ in the
415: period of halted accretion, is moved on the black hole horizon
416: (see Fig.~2b). Note that similar structural features of the inner MHD
417: flows in accretion disks related to the model of gamma-ray bursts
418: were discussed by Proga \& Zhang (2006).
419:
420: Figure~3 illustrates the time dependence of the accretion flow in
421: Model~B, showing the evolution of the mass accretion rate $\dot{M}_{in}$
422: and magnetic fluxes
423: in the midplane inside the five specific radii: $210\,R_g$
424: ($=R_{inj}$), $100\, R_g$, $50\, R_g$, $25\, R_g$,
425: and $2\, R_g$ ($=R_{inj}$).
426: This figure shows the evolution, which has been simulated
427: in 2D from $t=0$ to $2.14$
428: and in 3D after $t=2.14$.
429: The vertical dashed line in Fig.~3 indicates the moment of initiation
430: of the 3D simulations.
431: The cycle accretion in the 2D simulations begins at $t\approx 1.4$
432: and is clearly seen as a sequence of spikes
433: in the time-dependence of $\dot{M}_{in}$ (see Fig.~3a).
434: Spikes, which are related to the same cycle
435: accretion, are also observed in the variation of magnetic flux inside
436: $R=2\, R_g$ (see Fig.~3b).
437: The magnetic fluxes inside the other selected radii are gradually
438: increased with time because of the inward advection of the vertical field.
439: The time dependence of these fluxes is not significantly influenced
440: by the cycle accretion.
441:
442: The structure and dynamics of the inner region in Model~B are drastically
443: changed in the 3D simulations.
444: %in which non-axisymmetric perturbations are allowed.
445: Shortly after the initiation of the 3D simulations at $t=2.14$,
446: the axisymmetric distribution of mass near $R_{\rm m}$
447: undergoes the Rayleigh-Taylor and, possibly,
448: Kelvin-Helmholtz instability (see Kaisig, Tajima, \& Lovelace 1992;
449: Spruit, Stehle, \& Papaloizou 1995;
450: Chandran 2001; Li \& Narayan 2004)
451: with the fastest growing azimuthal mode number $m\simeq 50$
452: (the latter is probably determined by our grid resolution).
453: As a result,
454: the empty region inside $R_{\rm m}$ is quickly filled,
455: on the free-fall time scale estimated at $R_{\rm m}$, with the large
456: number of density spikes moved almost radially toward the center.
457: These spikes quickly disappear in the black hole and, at a later time,
458: the inner disk structure is modified toward establishing a
459: dynamic quasi-steady state. This state is characterized by
460: a low $m$-mode ($m\simeq 1$-5) spiral-flow structure,
461: which results from the interaction of the accreting mass with
462: the strong vertical magnetic field.
463: Note that the similar low $m$-mode flow structure
464: was found in the simulations
465: of accretion flows onto a magnetic dipole (Romanova \& Lovelace 2006).
466:
467: The non-axisymmetric inner flow is highly time-variable and experiences a
468: quasi-periodic behavior.
469: Figure~4 shows an example of the flow structure inside 50 $R_g$
470: in Model~B, at two successive moments: $t=2.2767$ and 2.2867.
471: The flow is essentially 3D inside the magnetically
472: dominated region limited by the radius $\simeq 35\,R_g$
473: and remains almost axisymmetric on the outside of this radius.
474: In the magnetically dominated region,
475: the flow forms moderately tightened spirals of dense
476: matter, which are clearly seen in Figs~4a and 4b.
477: This matter is quickly
478: accreted into the black hole with the radial velocity, which is
479: $\sim 0.5$ a fraction of the free-fall velocity.
480: Such a relatively fast infall is explained by the efficient loss
481: of the angular momentum by the mass during its interaction with
482: the vertical field.
483: The field is distributed nonuniformly in the disk plane,
484: concentrating in bundles that penetrate through
485: the plane in very low density,
486: magnetically dominated (with $\beta\sim 0.01$) regions,
487: or magnetic ``islands.''
488: The rotating mass interacts with magnetic bundles
489: and forces them to twist around the disk's rotational axis.
490: In the simulations, this twist is observed as the rotation of
491: magnetic islands around the center in the disk plane.
492: The rotational velocity of the islands typically has the
493: reduced rotational velocity by the factor of $\sim0.5-1$
494: in comparison with the velocity of the surrounding accretion matter.
495: This can be explained by the resistance
496: of the large-scale vertical field to such a twist.
497: The faster rotation of accretion matter and
498: slower rotation of magnetic islands produces a shear flow.
499: The shear flow plays two roles in our simulations.
500: First, it provides the exchange of momentum and energy between
501: the accreting mass and vertical field.
502: Second, the shear flow
503: results in an ablation of the islands caused by magnetic
504: diffusivity, making each island a temporal structure.
505: An example evolution of magnetic islands can be seen in Fig.~4:
506: the magnetic islands observed as low-density spiral arms
507: above the center in Fig.~4a are observed below the center in Fig.~4b,
508: after about half a revolution in the clockwise direction.
509: In the latter figure,
510: the islands are apparently reduced in size due to the ablation.
511:
512: The vertical field ablated from magnetic islands
513: is carried inward by the accretion flow and accumulates on the
514: black hole horizon. This accumulation results in
515: quasi-periodic eruptions of the field outward from the horizon
516: as soon as the field pressure overcomes
517: the dynamic pressure of the accreting mass.
518: The eruptions typically take the form of high-velocity narrow streams
519: (in the equatorial cross-section) of a low-density,
520: magnetically dominated medium fountained outward from the black hole.
521: In Fig.~4b, four magnetic islands observed as low-density regions
522: inside $R\approx 15\, R_g$
523: result from such eruptions and the eruption of one of these islands
524: (to the right from the center; see also the steam that produced it)
525: still continues at the moment shown.
526: In the consequent evolution, these islands are pushed outward
527: and stretched in the azimuthal direction by the accretion flow, and
528: take the spiral shape similar to that shown in Fig.~4a.
529:
530: Model~A evolves faster and accumulates a larger magnetic flux at the
531: center in comparison with Model~B.
532: Figure~5 shows the evolution of the accretion rate $\dot{M}_{\rm in}$
533: and magnetic fluxes in Model~A.
534: Qualitatively, the evolution of these quantities is similar to
535: the evolution of those in Model~B (see Fig.~3).
536: Quantitatively, however, Model~A demonstrates
537: significantly larger
538: time-averaged accretion rates (by about two orders of magnitude) and
539: longer quasi-periods of the cycle accretion
540: (represented by the intervals between spikes
541: in the time-dependence of $\dot{M}_{\rm in}$ in Fig.~5a)
542: in the 2D simulations.
543: By the end of the 2D simulations at $t\approx 1.7$, this model has
544: the maximum $R_m\simeq 30-40\,R_g$.
545:
546: In the 3D simulations,
547: Model~A experiences the initial transient period,
548: similar to the period of the development of
549: the Rayleigh-Taylor instability in Model~B (see above), in which
550: the non-axisymmetric, small-scale structures quickly appear
551: and disappear.
552: Figure~6 shows an example of the developed low $m$-mode spiral structure
553: in Model~A obtained after the transient period.
554: This structure is clearly dominated by the $m=1$ mode.
555: The magnetically dominated region is extended up to
556: $R\simeq 70\, R_g$.
557: Note that the spiral-density arms seen in Fig.~6
558: are more open than the arms in Model~B
559: (see Fig.~4a). This could be due to the stronger central field
560: in Model~A.
561:
562: Model~C is our slowest evolving model and,
563: accordingly, shows the slowest rate of accumulation
564: of the central vertical field.
565: This model has been calculated only in 2D and
566: demonstrated the qualitative similarity to the axisymmetric
567: evolution of Models~A and B.
568: The cycle accretion, which is caused by the accumulated field,
569: begins at $t\approx 4.8$ in Model~C.
570: The model demonstrates more-efficient turbulent motions in the
571: accretion flow.
572: This can be attributed to weaker magnetic fields, which
573: suppress less the MRI
574: % (see Stone \& Pringle 2001)
575: and convection motions. At the end of simulation
576: at $t\approx 6$, the model has the maximum $R_m\simeq 6\, R_g$.
577:
578:
579: \subsection{Poynting Jets}
580:
581: The 3D simulations of Models~A and B
582: show that the vertical field penetrated the central
583: magnetically dominated regions in MAD's is twisted around
584: the axis of rotation by the rotating accretion flows.
585: The field twist generates electromagnetic perturbations, which
586: propagate outward and transport the released energy
587: in the form of a Poynting flux (e.g., Landau \& Lifshitz 1987)
588: \begin{equation}
589: {\bf S}={1\over 8\pi}(({\bf v}\times{\bf B})\times{\bf B}).
590: \end{equation}
591: The Poynting flux is distributed nonuniformly in the polar angles,
592: basically showing two components:
593: a jet-like concentration of the flux near the poles and
594: a wide-spread distribution of the flux in the equatorial and
595: mid-polar-angle directions (from $\theta\sim\pi/4$ to $\sim 3\pi/4$).
596:
597: Figures~7 and 8 show example $\theta$-distributions of the
598: radial Poynting flux (solid lines),
599: \begin{equation}
600: S_R=v_R{B^2\over 4\pi}-{B_R\over 4\pi}({\bf v}\cdot {\bf B}),
601: \end{equation}
602: %normalized to the accretion power $\dot{M}_{in}c^2$
603: at six different radii, 5 $R_g$, 10 $R_g$, 25 $R_g$, 50 $R_g$,
604: 100 $R_g$, and 220 $R_g$,
605: which cover most of the radial domain
606: in Models~B and A, respectively. The shown distributions are averaged in
607: the azimuthal direction and in time over the interval $\Delta t\simeq 0.05$,
608: using a set of data files stored during the simulations.
609: Outside of the
610: inner magnetically dominated region (at $R\ga 35\,R_g$ in Model~B and
611: $R\ga 70\,R_g$ in Model~A),
612: the outward (positive) Poynting flux in the equatorial and
613: mid-polar-angle directions
614: is supported mostly by the rotation of the outer,
615: almost axisymmetric disks, in which the poloidal field component is frozen in.
616: Such a flux is present in both the axisymmetric 2D and 3D simulations.
617: The 3D simulations introduce new important features in
618: the Poynting flux distribution:
619: an increase of the flux at the equatorial and mid-polar-angle
620: directions inside the magnetically dominated region, and
621: at the polar directions outside this region (see Figs~7 and 8).
622: The equatorial flux inside the magnetically dominated region
623: is generated due to the twist of the vertical field by the spiraling
624: non-axisymmetric accretion flows. In Model~B, this flux,
625: represented by bumps in the $\theta$-distributions, gradually
626: deviates from the equatorial plane toward the poles as the radial
627: distance increases (see Figs~7a-7d). At $R\ga 100\,R_g$, the flux
628: is collimated into bi-polar Poynting jets
629: (see Figs~7e and 7f).
630: In the case of Model~A, the process of jet collimation
631: is less evident and somewhat different; but still, one can observe the
632: formation of narrow bi-polar Poynting jets starting from $R\simeq 10\,R_g$
633: and further development of these jets at large radii (see Figs~8b-8f).
634:
635: Figures~7 and 8 show $\theta$-distributions of the kinetic
636: flux (dashed lines),
637: \begin{equation}
638: F_R=v_R\rho{v^2\over 2},
639: \end{equation}
640: for comparison with the Poynting flux.
641: Typically, the kinetic flux is comparable,
642: but does not exceed the Poynting flux in the polar jets
643: (except in the outermost region in Model~A, see Fig.8f).
644: Accordingly, the jet velocity is mostly sub-Alfv\'enic.
645: However, the value of the kinetic flux is relatively large and
646: this is in some contradiction
647: with our expectations that MAD's can develop Poynting flux dominated jets.
648: The problem of the excessive kinetic flux in our simulations
649: can probably be explained
650: by the action of the numerical magnetic diffusivity (see \S 4),
651: which results in an unphysically large mass load of the Poynting jets
652: and the consequent excessive kinetic flux in them.
653:
654: The Poynting jets are powered by
655: the released binding energy of the accretion mass.
656: To estimate quantitatively the amount of energy going into the jets,
657: we calculate the Poynting jet ``luminosity" $\dot{E}_{jet}$
658: as a function of the radius R,
659: \begin{equation}
660: \dot{E}_{jet}(R)=\int R^2 S_R\,d\Omega,
661: \end{equation}
662: where
663: the integration is taken over the solid angles $\Omega$
664: occupying the polar regions with $\theta < \pi/4$
665: and $\theta > 3\pi/4$ (excluding the boundary polar cones, see \S 2).
666: For comparison, we also calculate the Poynting total luminosity
667: $\dot{E}_{tot}$, which is defined analogously to $\dot{E}_{jet}$,
668: but with the integration in eq.~(9) taken over the whole sphere.
669: Figures~9 and 10 show the radial profiles of the normalized
670: $\dot{E}_{jet}$ (solid lines) and $\dot{E}_{tot}$ (dashed lines)
671: in Models~B and A, respectively.
672: The jet luminosity $\dot{E}_{jet}$ weakly depends on the radius
673: at $R\ga 50\,R_g$ in both models and equals to
674: $\approx 1.5\%$ in Model~B and $\approx 0.5\%$ in Model~A.
675: Here, we quantify the luminosity in the units of
676: accretion power $\dot{M}_{in}c^2$.
677: The total luminosity $\dot{E}_{tot}$ includes the flux from
678: the bi-polar Poynting jets and wide equatorial Poynting outflow.
679: % from the outer axisymmetric disk.
680: The latter component of $\dot{E}_{tot}$ exceeds $\dot{E}_{jet}$
681: by the factor of $\sim 3$ at large radii (see Figs~9 and 10).
682: This can be the consequence of the employed simulation design
683: (see \S 2), in which
684: the disk accretion at outer radii
685: is mostly provided by the global Maxwell stresses.
686:
687: The smaller value of the final (at large radii)
688: relative $\dot{E}_{jet}$ in Model~A,
689: in comparison with that in Model~B,
690: can be explained by the different structure of the
691: inner magnetically dominated region in these models.
692: Model~A has the less tighten spiral density arms (see Fig.~6),
693: in which the mass accretes with larger radial velocity and, therefore,
694: delivers less energy to the field.
695: %Other difference in these models, which can be also attributed
696: %to the discussed difference in the innermost structure.
697: Note, also, that the value of $\dot{E}_{jet}$
698: %at the inner boundary $R_{in}$
699: in Model~A takes the relatively large
700: finite value right at the inner boundary $R_{in}$
701: (see Fig.~10), whereas $\dot{E}_{jet}$
702: in Model~B begins from a small value at $R_{in}$
703: and gradually increases outward (see Fig.~9).
704: This difference in the behavior of $\dot{E}_{jet}$
705: can be attributed to the discussed difference of the innermost structure
706: in the considered models.
707:
708:
709:
710: \section{Discussion and Conclusions}
711:
712: We have performed a numerical study of the formation and
713: evolution of quasi-stationary MAD, which is
714: characterized by a strong vertical magnetic field
715: accumulated at the disk center.
716: We employ the simulation design, in which the
717: poloidal magnetic field of one sign is permanently injected into the
718: computational domain at the outer numerical boundary and
719: the unipolar vertical field is transported inward by the accretion flow.
720: The accumulated field has a significant impact on the inner flow structure and
721: dynamics in both 2D and 3D simulations.
722: In the axisymmetric 2D simulations,
723: the field pressure can temporarily halt
724: the mass falling into the black hole, resulting in
725: the cycle accretion, in which the longer periods of accumulation
726: of the mass at the magnetospheric radius $R_m$
727: are followed by the short periods of accretion.
728: The 3D simulations have shown, however, that the axisymmetric cycle accretion
729: is not realized.
730: Instead, the accumulated field
731: causes the mass to accrete quasi-regularly in the form of non-axisymmetric
732: spiral streams and blobs.
733: We have demonstrated that 3D MAD's can be efficient sources of
734: collimated, bipolar Poynting jets,
735: which originate in the vicinity of the central black hole.
736: These jets develop due to and are powered by
737: the interaction of the spiral mass inflows
738: with the central field split into separate magnetic bundles.
739: The efficiency of conversion of the accretion energy $\dot{M}_{in} c^2$
740: into the Poynting jet energy is up to 1.5\% in our simulations.
741: This estimate may not be accurate
742: (we believe, underestimated) because of the use
743: of the pseudo-Newtonian approximation (see \S 2).
744: The better estimate of the efficiency can be obtained using
745: general relativistic MHD simulations.
746:
747: We have presented the simulation results from
748: three models of radiatively inefficient accretion disks,
749: which differ by the strength of
750: the injected field. In accordance with the previous studies
751: (e.g., Stone \& Pringle 2001), the structure of the outer disks
752: in these models is determined by the field strength.
753: In Model~A, which has the largest injected field,
754: the MRI is suppressed and the accretion flow is driven by
755: global Maxwell stresses, which transport
756: the excessive angular momentum outward
757: from the disk through the disk corona.
758: In Models~B and C, which have the smaller injected fields,
759: the MRI and turbulent motions are developed.
760: The turbulence in these models is axisymmetric and
761: partially supported by the efficient convection motions resulting from
762: dissipation of toroidal and small-scale, poloidal magnetic fields.
763: These motions cause the increase of the disk thicknesses
764: in comparison with non-turbulent Model~A.
765: The 3D simulations of Models~A and B have demonstrated that
766: non-axisymmetric motions are not important in the outer parts of the disks,
767: outside the inner magnetically dominated region (the disks remain
768: almost axisymmetric),
769: but very important inside this region,
770: resulting in the development of the spiral accretion flows
771: and bi-polar Poynting jets.
772: %Bi-polar Poynting jets, which are developed in Models~A and B, are
773: %originated due to and powered by the non-axisymmetric accretion flows
774: %in the inner regions.
775:
776: Numerical magnetic dissipations and reconnections
777: result in a magnetic diffusivity, which
778: influences the structure and dynamics of our models.
779: The spatial scale, on which the diffusivity
780: occurs in our simulations,
781: is determined by the gridsize, which greatly exceeds
782: the scales of various resistive mechanisms (e.g., Coulomb collisions,
783: dissipative plasma instabilities)
784: in the relevant astrophysical conditions.
785: Therefore, the magnetic diffusivity is significantly overestimated
786: in our models and results in an excessive
787: slippage of an accretion flow through magnetic field.
788: This slippage reduces the ability of the flow to drug
789: inward the vertical field, but, however, the numerical diffusivity
790: is not efficient enough
791: to totally prevent the field accumulation at the disk center.
792: The numerical diffusivity suppresses the MRI on the scales
793: of the gridsize, and, therefore, prevents
794: the development of turbulence in the outer regions of our models,
795: where the gridsize is increased.
796: Other effect of the numerical magnetic diffusivity is
797: the enhancement of the ablation of
798: magnetic islands, which are found in the 3D simulations (see \S 3.1).
799: To test the sensitivity of our models to
800: %effect of numerical resolution on
801: magnetic dissipations,
802: % and overall development of the models,
803: we have performed a 2D simulation of the model, which is
804: similar to Model~A, but has the double number of grid points in
805: the $R$- and $\theta$-directions.
806: The simulation has demonstrated the qualitative similarity of
807: axisymmetric evolution
808: of the high resolution model and Model~A:
809: both models show the formation of accretion disks,
810: accumulation of the vertical field in the disk centers, and
811: development of the cycle accretion.
812: The high resolution model forms a thiner laminar accretion disk.
813: Unfortunately, a more detailed quantitative
814: comparison of these models meets some difficulties because of
815: the different properties of the mass
816: and field injection region (see \S 2),
817: which are changed with the change of the resolution.
818:
819: The main results of our study,
820: the formation of MAD's and Poynting jets,
821: have been obtained under the assumption of radiatively inefficient flows,
822: but, we believe that these results
823: can also be applied to the radiatively efficient, dense
824: accretion disks (e.g., Kato, Fukue, \& Mineshige 1998).
825: %which we are not able to currently model due to the absence of
826: %radiation losses in our numerical consideration.
827: The formation of MAD's should not be affected by the
828: radiative cooling as soon as the central field satisfies
829: the equipartition condition,
830: \begin{equation}
831: {B^2\over 8\pi} \sim {GM\rho\over R_g},
832: \end{equation}
833: where $\rho$ is the mass density in the innermost region.
834: The radiative losses results in the higher $\rho$ and, therefore,
835: the larger B is necessary to obtain radiative MAD's.
836: We expect that the qualitatively similar spiral
837: structure of the inner magnetically dominated region,
838: to that found in our simulations, can be developed
839: in the case of the radiative disks.
840: Poynting jets should be a necessary attribute of the radiative MAD's as well.
841:
842: The formation of MAD's in the radiative disks can be used to explain
843: the observations of the low/hard state in black hole binaries
844: (for a review, see Remillard \& McClintock 2006).
845: Here, we briefly discuss basic moments of this application of MAD's and leave
846: more quantitative considerations for future works.
847: We assume, for example, the development of the MAD in the
848: radiation pressure dominated accretion disk
849: at the subcritical regime (see Shakura \& Sunyaev 1973).
850: In such a disk, the radiation diffusion time scale
851: $t_{rad}\simeq H^2\sigma_T\rho/c$
852: can significantly exceed the Keplerian time
853: $t_{K}= 2\pi R^{3/2}/\sqrt{GM}$ at small values of the $\alpha$-parameter, $\alpha \la 0.1$,
854: in virtu of the relation
855: \begin{equation}
856: {t_{K}\over t_{rad}} \simeq 3.4\alpha,
857: \end{equation}
858: which follows from the Shakura-Sunyaev solution.
859: Here, we denote $H$ to be the disk half-thickness and $\sigma_T=0.4$ cm$^2$/g to be
860: the Thomson scattering cross-section.
861: As soon as $t_{rad}\gg t_{K}$ in the outer Shakura-Sunyaev disk,
862: $t_{rad}$ can also significantly exceed the accretion velocity $t_{accr}\sim t_{K}$
863: in the inner spiral flow in the MAD, in virtu of the relation
864: \begin{equation}
865: {t_{accr}\over t_{rad}}\propto R,
866: \end{equation}
867: which is satisfied in accretion flows with the scaling law of the accretion velocity
868: $v_{accr}\propto R^{-1/2}$ and $H\propto R$.
869: %obtained assuming the almost free-fall accretion and
870: %the half-thickness of the flow $H\propto R$.
871: Having $t_{rad}\gg t_{accr}$, one concludes that the radiation is traped
872: inside the spiral flow on the accretion time scale and, therefore,
873: this flow is radiatively inefficient.
874: From the point of view of an observer, which detects
875: the softer part of the spectrum of outgoing radiation (below $\sim$ eVs),
876: the MAD will look like a Shakura-Sunyaev disk truncated at the inner radius
877: $R_{tr}$, which coinsides with the transition radius between the inner
878: magnetically dominated region and outer axisymmetric accretion disk.
879: Typically, in observations, $R_{tr}$ is in the interval from a few tens
880: to hundreds of $R_g$ (e.g., in Cyg X-1, see Done \& Zycki 1999), which
881: is consistent with that obtained in our Models~A and B.
882: The observed specta of black hole binaries in the low/hard state
883: are dominated by the hard x-ray component
884: (e.g., Done, Gierli\'nski, \& Kubota 2007) and this
885: can be explained by the radiation from
886: the hot, optically thin magnetized medium, which surrounds the accreting spiral flows
887: and in which the binding energy of these flows is released.
888: %the release of the binding energy of the
889: %accreting spiral flows in the optically thin, magnetized medium surrounded these flows.
890: The synchrotron radiation from the magnetized medium and Poynting jets in the MAD
891: (see Goldston, Quataert, \& Igumenshchev 2005)
892: can be used to explain the observed radio luminosity
893: in the low/hard state; this luminosity is believed to be due to steady jets
894: (e.g., Corbel et al. 2003; Gallo, Fender, \& Pooley 2003).
895:
896: Our simulations assume accretion disks
897: around black holes and can be relevant to objects with
898: relativistic jets containing accreting
899: stellar-mass black holes (e.g., in micro-quasars; black holes resulted from
900: type Ib/c supernova explosions and mergers of two compact objects) and
901: supermassive black holes (in galactic centers).
902: However, we believe that our main results
903: %the formation of MAD's and powerful Poynting jets,
904: can also be relevant to nonrelativistic
905: astrophysical objects, in which accretion disks and jets are observed.
906: These objects include, for example, young stellar objects
907: and accreting stars (e.g., white dwarfs) in binary systems.
908: Qualitatively, we expect that the structure and dynamics of MAD's and
909: Poynting jets are similar in the both relativistic and nonrelativistic cases.
910: We expect, however, large quantitative differences
911: in these two cases
912: because of the different energy-density scales involved
913: in the regions of jet formation.
914: Black holes, which are capable of launching jets almost from
915: the event horizon, can produce ultra relativistic jets
916: (e.g., McKinney 2006).
917: Jets from nonrelativistic objects,
918: which have the surface radius $R_*\gg R_g$,
919: are limited by the velocities $v\sim\sqrt{GM/R_*}\ll c$.
920: For example, the latter formula gives the upper estimate of the jet velocity
921: $\sim 400\,km/s$
922: from the solar-type stars.
923:
924: The problem of inward transport and amplification of the vertical field
925: in turbulent accretion disks was intensively discussed
926: in past and recent years (see, e.g., Spruit \& Uzdensky 2005).
927: The solution of this problem can help to discriminate models of
928: accretion disks, which are consistent with observations
929: (e.g., Meier \& Nakamura 2006; Schild, Leiter, \& Robertson 2006).
930: Our simulation results show that the vertical field is transported inward
931: and amplified
932: independent of the disk structure, either laminar or turbulent.
933: It is worth noting, however,
934: that magnetic fields in our models are imposed and
935: relatively large. The assumed strength of these fields
936: exceeds the possible strength of the self-sustained magnetic fields that
937: could be developed due to the MRI (Sano et al. 2004).
938: Therefore, these results should be considered with some caution, because
939: they do not represent the case
940: of weak vertical magnetic fields.
941:
942: The strong vertical field in the center of accretion disks
943: can be, in principal,
944: a relic field that is inherited from the previous evolution.
945: This field can appear, for example, in the merger scenario
946: (merger of two magnetized neutron stars or
947: a black hole with a magnetized neutron star, e.g., Berger et al. 2005) or
948: in the course of
949: the gravitational collapse of an extended ``proto" object
950: (e.g., proto-stellar cloud, supernova progenitor),
951: which produces a significantly more compact object
952: (protostar, black hole).
953: In the latter case,
954: the proto object may contain some amount of the poloidal field,
955: which will be amplified and accumulated at the center during the collapse.
956: After the formation of the compact object, the remained
957: noncollapsed mass can still
958: move inward, forming an accretion disk and confining
959: the field in the vicinity of the object.
960: Depending on the relative strength of this relic field
961: and the mass accretion rate, MAD's and Poynting jets can be developed.
962: The considered scenario can be applied
963: to young stellar objects (T-Tauri stars, e.g., Donati et al. 2007)
964: and the hyper-accretion model for gamma-ray bursts
965: (Woosley 1993; Paczy\'nski 1998).
966:
967: The spiral-flow pattern in MAD's rotates with about
968: the same angular velocity at all radial distances; i.e.,
969: it rotates almost as a rigid body.
970: %This means that MAD's can be sources of
971: Such a rotation can result in
972: quasi-periodic oscillations (QPO's)
973: in the emitted radiation, if the disk's axis is inclined
974: to the line of view of an observer
975: (e.g., Alpar \& Shaham 1985; Lamb et al. 1985; Strohmayer et al. 1996;
976: Lamb \& Miller 2001; Titarchuk 2003).
977: The frequency of these QPO's should be related to the
978: rotation of the spiral pattern,
979: %angular velocity of the spiral pattern,
980: which angular velocity is defined by the radius
981: of the magnetically dominated region and can be a
982: fraction ($\sim 0.5-1$) of the orbital velocity at this radius.
983: More investigations are required to make quantitative
984: predictions about QPO's from MAD's.
985:
986:
987: \acknowledgments
988:
989: %The author gratefully thanks ... for discussions.
990: This work was supported by
991: the U.S. Department of Energy (DOE) Office of Inertial Confinement
992: Fusion under Cooperative Agreement No. DE-FC52-92SF19460, the
993: University of Rochester, the New York State Energy Research and
994: Development Authority.
995:
996:
997:
998: \clearpage
999:
1000: \begin{thebibliography}{999}
1001:
1002: \bibitem[]{} Agapitou, V., \& Papaloizou, J.C.B. 1996,
1003: Astrophys. Lett. Commun., 34, 363
1004: \bibitem[]{} Alpar, M. A., \& Shaham, J. 1985, Nature, 316, 239
1005: \bibitem[]{} Appl, S. 1996, A\&A, 314, 995
1006: \bibitem[]{} Balbus, S. A., \& Hawley, J. F. 1991, \apj, 376, 214
1007: \bibitem[]{} Berger, E. et al. 2005, Nature, 438, 988
1008: \bibitem[]{} Bisnovatyi-Kogan, G. S., \& Ruzmaikin, A. A. 1974, Ap\&SS, 28, 45
1009: \bibitem[]{} Bisnovatyi-Kogan, G. S., \& Ruzmaikin, A. A. 1976, Ap\&SS, 42, 401
1010: %\bibitem[]{} Bisnovatyi-Kogan, G. S., \& Lovelace, R. V. E. 1997, \apj, 486, L43
1011: %\bibitem[]{} Bisnovatyi-Kogan, G. S., \& Lovelace, R. V. E. 2000, \apj, 529, 978
1012: \bibitem[]{} Bisnovatyi-Kogan, G. S., \& Lovelace, R. V. E. 2007, \apj, 667, L167
1013: \bibitem[]{} Blandford, R.D., \& Znajek, R.L. 1977, \mnras, 179, 433
1014: \bibitem[]{} Blandford, R.D., \& Payne, D.G. 1982, \mnras, 199,883
1015: \bibitem[]{} Brandenburg, A., Nordlund, A., Stein, R.F., \& Torkelsson, U.
1016: 1995, \apj, 446, 741
1017: \bibitem[]{} Cabral, B., \& Leedom, L. 1993, Computer Graphics: Proceedings:
1018: Annual Conference Series 1993: SIGGRAPH 93 (New York: Association for
1019: Computing Machinery), 263
1020: \bibitem[]{} Chandran, B. D. G. 2001, \apj, 562, 737
1021: \bibitem[]{} Colella, P., \& Woodward, P.R. 1984, J. Comp. Phys., 54, 174
1022: \bibitem[]{} Corbel, S., Nowak, M. A., Fender, R. P., Tzioumis, A. K., \& Markoff, S.
1023: 2003, Astron. Astrophys., 400, 1007
1024: \bibitem[]{} De Villiers J.-P., Hawley, J. F., \& Krolik, J. H.
1025: 2003, \apj, 599, 1238
1026: \bibitem[]{} Donati, J.-F., Jardine, M. M., Gregory S. G., Petit, P.,
1027: Bouvier, J., Dougados, C., M\'enard, F., Cameron, A. C., Harries, T. J.,
1028: Jeffers, S. V., \& Paletou, F. 2007, \mnras, 380, 1297
1029: \bibitem[]{} Done, C., \& Zycki, P. T. 1999, \mnras, 305, 457
1030: \bibitem[]{} Done, C., Gierli\'nski, M., \& Kubota, A. 2007, Astron. Astrophys. Rev., in press
1031: \bibitem[]{} Eichler, D. 1993, \apj, 419, 111
1032: \bibitem[]{} Evans, C.R., \& Hawley, J.F. 1988, \apj, 332, 659
1033: %\bibitem[]{} Frank, J., King, A., \& Raine, D. 1992, Accretion Power in
1034: % Astrophysics (Cambridge Univ. Press, Cambridge)
1035: \bibitem[]{} Gallo, E., Fender, R. P., \& Pooley, G. G. 2003, \mnras, 344, 60
1036: \bibitem[]{} Gardiner, T.A., \& Stone, J.M. 2005, J. Comp. Phys., 205, 509
1037: \bibitem[]{} Goldston, J. E., Quataert, E., \& Igumenshchev, I. V. 2005, \apj, 621, 785
1038: %\bibitem[]{} Hawley, J. F. 2000, \apj, 528, 462
1039: \bibitem[]{} Hawley, J. F., \& Balbus, S. A. 1992, \apj, 400, 595
1040: %\bibitem[]{} Hawley, J. F., Balbus, S. A., \& Stone, J. M. 2001, \apj, 554, L49
1041: %\bibitem[]{} Hawley, J. F., \& Balbus, S. A. 2002, \apj, 573, 738
1042: \bibitem[]{} Hawley, J.F., Gammie, C.F., \& Balbus, S.A. 1996, \apj, 464, 690
1043: %\bibitem[]{} Hawley, J. F., \& Krolik, J. H. 2002, \apj, 566, 164
1044: \bibitem[]{} Hawley, J. F., \& Krolik, J. H. 2006, \apj, 641, 103
1045: \bibitem[]{} Heyvaerts, J., Priest, E.R., \& Bardou, A. 1996, \apj, 473, 403
1046: \bibitem[]{} Hirose, S., Krolik, J.H., De Villiers, J.-P., \& Hawley, J.F.
1047: 2004, \apj, 606, 1083
1048: %\bibitem[]{} Igumenshchev, I. V., \& Abramowicz, M. A. 1999, \mnras, 303, 309
1049: %\bibitem[]{} Igumenshchev, I. V., \& Narayan, R. 2002, \apj, 566, 137 (IN)
1050: \bibitem[]{} Igumenshchev, I. V. 2006, \apj, 649, 361
1051: \bibitem[]{} Igumenshchev, I. V., Narayan, R., \& Abramowicz, M. A. 2003,
1052: \apj, 592, 1042
1053: \bibitem[]{} Kaisig, M., Tajima, T., \& Lovelace, R. V. E. 1992, \apj, 386, 83
1054: \bibitem[]{} Kato, S., Fukue, J., \& Mineshige, S. 1998, Black-Hole
1055: Accretion Disks (Kyoto: Kyoto Univ. Press)
1056: \bibitem[]{} Kato, Y., Mineshige, S., \& Shibata, K. 2004, \apj, 605, 307
1057: \bibitem[]{} Komissarov, S.S. 2005, \mnras, 359, 801
1058: \bibitem[]{} K\"onigl, A. 1989, \apj, 342, 208
1059: \bibitem[]{} K\"onigl, A., \& Pudritz, R. E. 2000,
1060: in Protostars and Planets IV, ed. V. Mannings, A. Boss, \& S. Russell
1061: (Arizona: Univ. Arizona Press), p.759
1062: (astro-ph/9903168)
1063: %\bibitem[]{} Krasnopolsky, R., Li, Z.-Y., \& Blandford, R. 1999, \apj, 526, 631
1064: \bibitem[]{} Lamb, F. K., \& Miller, M. C. 2001, \apj, 554, 1210
1065: \bibitem[]{} Lamb, F. K., Shibazaki, N., Alpar, M. A., \& Shaham, J.
1066: 1985, Nature, 317, 681
1067: \bibitem[]{} Landau, L. D., \& Lifshitz, E. M. 1987, Electrodynamics of
1068: Continuous Media. Pergamon Press, Oxford
1069: \bibitem[]{} Li, L.-X. 2000, \apj, 531, L111
1070: \bibitem[]{} Li, L.-X., \& Narayan, R. 2004, \apj, 601, 414
1071: \bibitem[]{} Li, S. 2005, J. Comp. Phys., 203, 344
1072: \bibitem[]{} Livio, M., Ogilvie, G.I., \& Pringle, J.E. 1999, \apj, 512, 100
1073: \bibitem[]{} Livio, M., Pringle, J.E., \& King, A.R. 2003, \apj, 593, 184
1074: \bibitem[]{} Lovelace, R. V. E., Wang, J. C. L., \& Sulkanen, M. E. 1987,
1075: \apj, 315, 504
1076: \bibitem[]{} Lovelace, R. V. E., Li, H., Koldoba, A. V., Ustyugova, G. V., \&
1077: Romanova, M. M. 2002, \apj, 572, 445
1078: \bibitem[]{} Lovelace, R. V. E., Gandhi, P. R., \& Romanova, M. M. 2005,
1079: Ap\&SS, 298, 115
1080: \bibitem[]{} Lovelace, R. V. E., Romanova, M. M., \& Newman, W. I. 1994,
1081: \apj, 437, 136
1082: \bibitem[]{} Lubow, S.H., Papaloizou, J.C.B., \& Pringle, J.E. 1994,
1083: \mnras, 267, 235
1084: \bibitem[]{} Lynden-Bell, D. 2003, \mnras, 341, 1360
1085: \bibitem[]{} Meier, D.L., Nakamura, M. 2006, ASPC, 350, 195
1086: \bibitem[]{} McKinney, J. C., \& Gammie, C. F. 2004, \apj, 611, 977
1087: \bibitem[]{} McKinney, J. C. 2006, \mnras, 368, 1561
1088: %\bibitem[]{} Miller, K. A. \& Stone, J. M. 2000, \apj, 534, 398
1089: \bibitem[]{} Narayan, R., Igumenshchev, I. V., \& Abramowicz, M. A.
1090: 2000, \apj, 539, 798
1091: \bibitem[]{} Narayan, R., Igumenshchev, I. V., \& Abramowicz, M. A.
1092: 2003, PASJ, 55, L69
1093: \bibitem[]{} Paczy\'nski, B., \& Wiita, P.J. 1980, A\&A, 88, 23
1094: \bibitem[]{} Paczy\'nski, B. 1998, \apj, 494, L45
1095: \bibitem[]{} Proga, D., \& Zhang, B. 2006, \mnras, 370, L61
1096: \bibitem[]{} Punsly, B. 2001, Black Hole Gravitohydromagnetics
1097: (New York: Springer)
1098: %\bibitem[]{} Quataert, E. 2002, \apj, 575, 855
1099: %\bibitem[]{} Quataert, E., \& Gruzinov, A. 1999, \apj, 520, 248
1100: \bibitem[]{} Quataert, E., \& Gruzinov, A. 2000, \apj, 539, 809
1101: \bibitem[]{} Remillard, R. A., \& McClintock, J. E. 2006, Annu. Rev. Astron Astrophys., 44, 49
1102: \bibitem[]{} Romanova, M. M., \& Lovelace, R. V. E. 2006, \apj, 645, L73
1103: \bibitem[]{} Sano, T., Inutsuka, S., Turner, N.J., Stone, J.M. 2004,
1104: \apj, 605, 321
1105: \bibitem[]{} Schild, R.E., Leiter, D.J., \& Robertson, S.L. 2006, \aj, 132, 420
1106: \bibitem[]{} Shakura, N. I., \& Sunyaev, R.A. 1973, A\&A, 24, 337
1107: \bibitem[]{} Spruit, H.C., Foglizzo, T., \& Stehle, R. 1997, \mnras, 288, 333
1108: \bibitem[]{} Spruit, H. C., Stehle, R., \& Papaloizou, J. C. B. 1995,
1109: \mnras, 275, 1223
1110: \bibitem[]{} Spruit, H.C., \& Uzdensky, D.A. 2005, \apj, 629, 960
1111: \bibitem[]{} Stone, J.M., Hawley, J.F., Gammie, C.F., \& Balbus, S.A. 1996,
1112: \apj, 463, 656
1113: \bibitem[]{} Stone, J. M., \& Pringle, J. E. 2001, \mnras, 322, 461
1114: \bibitem[]{} Stone, J. M., \& Norman, M. L. 1994, \apj, 433, 746
1115: \bibitem[]{} Strohmayer, T. E., Zhang, W., Swank, J. H., Smale, A.,
1116: Titarchuk, L., Day, C., \& Lee, U. 1996, \apj, 469, L9
1117: \bibitem[]{} Takahashi, M., Nitta, S., Tatematsu, Y., \&
1118: Tomimatsu, A. 1990, \apj, 363, 206
1119: %\bibitem[]{} Thorne, K. S., Price, R. H., \& MacDonald, D. H. 1986,
1120: % Black Holes: The Membrane Paradigm
1121: \bibitem[]{} Titarchuk, L. 2003, \apj, 591, 354
1122: \bibitem[]{} Tomimatsu, A., Matsuoka, T., \& Takahashi, M. 2001,
1123: Phys. Rev. D, 64, 123003
1124: %\bibitem[]{} Toropin, Yu. M., Toropina, O. D., Savelyev, V. V.,
1125: % Romanova, M. M., Chechetkin, V. M., \& Lovelace, R. V. E. 1999,
1126: % \apj, 517, 906
1127: %\bibitem[]{} Toropina, O. D., Romanova, M. M., Toropin, Yu. M., \&
1128: % Lovelace, R. V. E. 2003, \apj, 953, 472
1129: \bibitem[]{} Tout, C. A., \& Pringle, J. E. 1996, \mnras, 281, 219
1130: \bibitem[]{} van Ballegooijen, A.A. 1989, in Proc. European Phys. Soc.
1131: Study Conf., Accretion Disks and Magnetic Fields in Astrophysics,
1132: ed. G. Belvedere (Dordrecht: Kluwer), 99
1133: \bibitem[]{} Woosley, S.E. 1993, \apj, 405, 273
1134:
1135: \end{thebibliography}
1136:
1137:
1138: \clearpage
1139:
1140: \begin{figure}
1141: %\epsscale{1.0}
1142: \epsscale{.60}
1143: %%\plottwo{dens_aab383.eps}{dens_aab386.eps}
1144: %\plottwo{dens_aac005.eps}{dens_aac008.eps}
1145: %\plottwo{fig1a.eps}{fig1b.eps}
1146: \plotone{see_jpg.eps}
1147: \caption{Distributions of density in the meridional plane in Model~B
1148: at two successive moments
1149: %(a) $t=5.452$ and (b) $t=5.487$ from axisymmetric 2D simulations.
1150: (a) $t=5.1153$ and (b) $t=5.1458$ from axisymmetric 2D simulations.
1151: The black hole is located on the left and
1152: the small empty circle there corresponds to the inner
1153: boundary around the black hole at $R_{in}=2\,R_g$.
1154: The axis of rotation is in the vertical direction.
1155: The domain shown has the radial extent $100\,R_g$ along the equatorial plane
1156: and represents a fraction of the full computational domain
1157: with $R_{out}=220\, R_g$.
1158: The time is in units of the orbital period at $R_{inj}=210\,R_g$.
1159: The color bars on the bottom indicate the scales for $\log\rho$
1160: (in arbitrary units).
1161: The model develops the thick turbulent accretion disk, which is seen
1162: as the nonuniform mass concentration near the equator.
1163: The low-density polar regions are filled with the strong vertical
1164: magnetic field. This field causes the cycle accretion into the black hole.
1165: (a) The halted accretion period
1166: when the mass is accumulated at the magnetospheric radius
1167: $R_m\approx 15\,R_g$ and (b) the accretion period.
1168: \label{fig1}}
1169: \end{figure}
1170:
1171: %\clearpage
1172:
1173: \begin{figure}
1174: %\epsscale{1.0}
1175: \epsscale{.60}
1176: %%\plottwo{licm_aab383.eps}{licm_aab386.eps}
1177: %\plottwo{licm_aac005.eps}{licm_aac008.eps}
1178: %\plottwo{fig2a.eps}{fig2b.eps}
1179: %\plotone{see_jpg.eps}
1180: \caption{Snapshot of magnetic lines in Model~B
1181: at the same moments and on the same spatial scales as in Fig.~1.
1182: The component of the field lines parallel to the meridional plane
1183: is shown. The accretion disk transports the vertical magnetic
1184: flux inward, which accumulates in the vicinity of the black hole.
1185: Small-scale magnetic loops are the result of turbulent motions
1186: in the disk and disk corona.
1187: (a) The halted accretion period (see Fig.~1a), in which
1188: most of the accumulated magnetic flux
1189: is outside the black hole horizon.
1190: (b) The accretion period (see Fig.~1b), in which
1191: all the accumulated flux goes through the horizon.
1192: The lines have been plotted using the method of Cabral \& Leedom (1993).
1193: \label{fig2}}
1194: \end{figure}
1195:
1196: %\clearpage
1197:
1198: \begin{figure}
1199: \epsscale{.60}
1200: %\plotone{time4paper_B.eps}
1201: %\plotone{fig3.eps}
1202: %\plotone{fig3_lr.eps}
1203: \caption{Evolution of mass accretion rate and magnetic fluxes
1204: in Model~B.
1205: The time is in units of the orbital period at $R_{inj}$.
1206: The simulations are performed in 2D, assuming axisymmetry,
1207: from $t=0$ to 2.14 (the latter moment indicated by the vertical dashed line)
1208: and in 3D after that.
1209: (a) The accretion into the black hole begins at $t\approx 1.3$.
1210: Starting from $t\approx 1.4$, the cycle accretion is developed
1211: (seen as a sequence of spikes).
1212: In 3D, the cycle accretion disappears.
1213: (b) Magnetic fluxes (in arbitrary units) through the equatorial plane
1214: inside five fixed radii: $2\,R_g$ ($=R_{in}$), $25\,R_g$, $50\,R_g$,
1215: $100\,R_g$, and $210\,R_g$ ($=R_{inj}$). The variability of the flux
1216: at $R_{in}$ is related to the variability of the accretion rate in (a).
1217: \label{fig3}}
1218: \end{figure}
1219:
1220: %\clearpage
1221:
1222: \begin{figure}
1223: %\epsscale{1.0}
1224: \epsscale{.60}
1225: %\plottwo{dens_aac31.eps}{dens_aac34.eps}
1226: %\plottwo{fig4a.eps}{fig4b.eps}
1227: %\plotone{see_jpg.eps}
1228: \caption{Distributions of density in the equatorial plane
1229: in Model~B at two successive moments
1230: (a) $t=2.2767$ and (b) $t=2.2867$ from 3D simulations.
1231: The black hole is located in the center and the central
1232: black circle corresponds to the inner boundary at $R_{in}=2\,R_g$.
1233: The shown domain has the extension 100 by $100\,R_g$.
1234: The time is in units of the orbital period at $R_{inj}=210\,R_g$.
1235: The color bars on the bottom indicate the scales for $\log\rho$
1236: (in arbitrary units).
1237: The accretion flow rotates in the clockwise direction and
1238: forms the spiral structure inside $R\approx 35\,R_g$.
1239: The low-density regions correspond to the magnetically dominated islands,
1240: through which the vertical field penetrates the disk.
1241: (a) The developed structure of spiral density inflows and
1242: magnetic islands stretched in the azimuthal direction.
1243: (b) The moment after about half a revolution of the spiral
1244: magnetic pattern seen in (a); new magnetic islands inside $R\approx 15\,R_g$
1245: are the result of an
1246: eruption of the vertical magnetic flux from the inner boundary.
1247: \label{fig4}}
1248: \end{figure}
1249:
1250: %\clearpage
1251:
1252: \begin{figure}
1253: \epsscale{.60}
1254: %\plotone{time4paper_A.eps}
1255: %\plotone{fig5.eps}
1256: %\plotone{fig5_lr.eps}
1257: \caption{Evolution of mass accretion rate and magnetic fluxes
1258: in Model~A. The notations used are the same as in Fig.~3.
1259: Note the earlier beginning of accretion,
1260: significantly larger accretion rate and magnetic fluxes,
1261: and longer quasi-periods between the accretion events
1262: in the 2D simulations
1263: in comparison with those in Model~B (see Fig.~3).
1264: These are the result of a stronger field injection in Model~A.
1265: \label{fig5}}
1266: \end{figure}
1267:
1268: %\clearpage
1269:
1270: \begin{figure}
1271: \epsscale{.60}
1272: %\plotone{dens_aac27_A.eps}
1273: %\plotone{fig6.eps}
1274: %\plotone{see_jpg.eps}
1275: \caption{Distribution of density in the equatorial plane
1276: in Model~A at $t=1.82$ from 3D simulations.
1277: The notations used are the same as in Fig.~4.
1278: The shown domain has the extension 300 by $300\,R_g$.
1279: The model is characterized by the relatively large magnetic flux accumulated
1280: at the center. The magnetically dominated region has the outer
1281: radius $R\simeq 70\,R_g$.
1282: The accretion flow demonstrates the one-arm
1283: ($m=1-$ mode) spiral structure in the innermost region.
1284: This structure rotates in the clock-wise direction and is split onto the three
1285: arms at larger radial distances.
1286: %The other two large density spirals (seen in the lower part
1287: %on the left) are developed.
1288: There are other two large density streams in the process of development
1289: (below the center and on the left from the center).
1290: \label{fig6}}
1291: \end{figure}
1292:
1293: %\clearpage
1294:
1295: \begin{figure}
1296: \epsscale{.60}
1297: %\plotone{poynting_theta_B.eps}
1298: %\plotone{fig7.eps}
1299: \caption{Distribution of Poynting $S_R$ (solid lines) and kinetic
1300: $F_R$ (dashed lines) radial fluxes in the $\theta$-direction in Model~B.
1301: The fluxes are averaged in the azimuthal direction and in time
1302: over the interval $\Delta t=0.05$, beginning from $t=2.27$, and
1303: normalized to the accretion power $\dot{M}_{in}c^2$.
1304: The distributions are shown at the radial distances
1305: 5 $R_g$, 10 $R_g$, 25 $R_g$, 50 $R_g$, 100 $R_g$, and 220 $R_g$
1306: ($=R_{out}$) in the panels from (a) to (f), respectively, and
1307: illustrate the collimation of Poynting flux into bi-polar jets.
1308: %See test for discussions.
1309: \label{fig7}}
1310: \end{figure}
1311:
1312: %\clearpage
1313:
1314: \begin{figure}
1315: \epsscale{.60}
1316: %\plotone{poynting_theta_A.eps}
1317: %\plotone{fig8.eps}
1318: \caption{Same as in Fig.~7, but for Model~A.
1319: The fluxes are averaged in time over the interval $\Delta t=0.04$,
1320: beginning from $t=1.78$.
1321: \label{fig8}}
1322: \end{figure}
1323:
1324: %\clearpage
1325:
1326: \begin{figure}
1327: \epsscale{.60}
1328: %\plotone{poynting_rad_B.eps}
1329: %\plotone{fig9.eps}
1330: \caption{Radial distribution of Poynting jet
1331: $\dot{E}_{jet}$ (solid line) and total $\dot{E}_{tot}$
1332: (dashed line) luminosities in Model~B. The luminosities are normalized
1333: to the accretion power $\dot{M}_{in}c^2$ and averaged in time
1334: as explained in the capture to Fig.7.
1335: \label{fig9}}
1336: \end{figure}
1337:
1338: %\clearpage
1339:
1340: \begin{figure}
1341: \epsscale{.60}
1342: %\plotone{poynting_rad_A.eps}
1343: %\plotone{fig10.eps}
1344: \caption{Same as in Fig.~9, but for Model~A.
1345: The luminosities are averaged in time
1346: as explained in the capture to Fig.8.
1347: \label{fig10}}
1348: \end{figure}
1349:
1350:
1351: \end{document}
1352:
1353: