1: \documentclass[12pt,preprint]{aastex}
2:
3: %Units of measurement (with spacing)
4:
5: \newcommand{\km}{\,{\textnormal{km}}}
6: \newcommand{\second}{\,{\textnormal{s}}}
7: \newcommand{\gram}{\,{\textnormal{g}}}
8: \newcommand{\cm}{\,{\textnormal{cm}}}
9: \newcommand{\kms}{{\km\second^{-1}}}
10: \newcommand{\solarmass}{{\,{\textnormal{M}}_\sun}}
11: \newcommand{\yr}{{\,\textnormal{yr}}}
12: \newcommand{\solarmassyear}{{\solarmass\yr^{-1}}}
13: \newcommand{\AU}{{\,\textnormal{AU}}}
14: \newcommand{\stellarmass}{{{\textnormal{M}}_\ast}}
15: \newcommand{\gauss}{\,{\textnormal{G}}}
16: \newcommand{\massden}{{\gram\cm^{-3}}}
17: \newcommand{\numden}{{\cm^{-3}}}
18: \def\lsim{\raise0.3ex\hbox{$<$}\kern-0.75em{\lower0.65ex\hbox{$\sim$}}}
19: \def\gsim{\raise0.3ex\hbox{$>$}\kern-0.75em{\lower0.65ex\hbox{$\sim$}}}
20:
21: %Units of measurement (without spacing)
22: \newcommand{\kmsns}{{{\textnormal{km}}\second^{-1}}}
23: \newcommand{\AUns}{{\textnormal{AU}}}
24: \newcommand{\secondns}{{\textnormal{s}}}
25:
26: \begin{document}
27:
28: \title{Magnetocentrifugal Winds in 3D: Nonaxisymmetric Steady State}
29: \author{Jeffrey M. Anderson\altaffilmark{1},
30: Zhi-Yun Li\altaffilmark{1}, Ruben Krasnopolsky\altaffilmark{2},
31: \& Roger D. Blandford\altaffilmark{3}}
32: \altaffiltext{1}{Astronomy Department, University of Virginia,
33: Charlottesville, VA 22904; jma2u, zl4h@virginia.edu}
34: \altaffiltext{2}{Department of Astronomy \& Astrophysics,
35: University of Toronto, Toronto, ON M5S 3H4, Canada}
36: \altaffiltext{3}{SLAC, M/S 75, 2575 Sandhill Rd, Menlo Park, CA 94025;
37: rdb3@stanford.edu}
38:
39: \begin{abstract}
40: Outflows can be loaded and accelerated to high speeds along rapidly
41: rotating, open magnetic field lines by centrifugal forces. Whether
42: such magnetocentrifugally driven winds are stable is a longstanding
43: theoretical problem. As a step towards addressing this problem,
44: we perform the first large-scale 3D MHD simulations that extend
45: to a distance $\sim 10^2$ times beyond the launching region,
46: starting from steady 2D (axisymmetric) solutions. In an attempt
47: to drive the wind unstable, we increase the mass loading on one
48: half of the launching surface by a factor of $\sqrt{10}$,
49: and reduce it by the same factor on the other half. The evolution
50: of the perturbed wind is followed numerically. We find no evidence
51: for any rapidly growing instability that could disrupt the wind
52: during the launching and initial phase of propagation, even when the
53: magnetic field of the magnetocentrifugal wind is toroidally dominated
54: all the way to the launching surface. The strongly perturbed wind
55: settles into a new steady state, with a highly asymmetric mass
56: distribution. The distribution of magnetic field strength is, in
57: contrast, much more symmetric. We discuss possible reasons for
58: the apparent stability, including stabilization by an axial
59: poloidal magnetic field, which is required to bend field lines
60: away from the vertical direction and produce a magnetocentrifugal
61: wind in the first place.
62:
63: \end{abstract}
64:
65: \keywords{ISM: jets and outflows --- MHD --- stars: formation}
66:
67: \section{Introduction}
68:
69: The jets and winds observed around young stellar objects (YSOs) are thought
70: to be driven magnetocentrifugally from disk surface (\citealt{bp82};
71: see \citealt{us85} for a related mechanism). The fluid
72: rotation winds the magnetic field up into a predominantly toroidal
73: configuration at large distances. The toroidal field is thought to be
74: able to collimate part of the wind into a jet through ``hoop'' stresses
75: \citep{shu95,hn89}.
76: It may, however, lead to instabilities that can
77: potentially disrupt the outflow \citep{e93,be98}.
78:
79: How narrow astrophysical jets maintain their stability over large distances
80: is a longstanding puzzle \citep{f98}.
81: Numerical simulations have demonstrated that hydrodynamical
82: jets are prone to disruption by Kelvin-Helmholtz (KH) instabilities (e.g.,
83: \citealt{bo98,h04}). Magnetic fields can add rigidity to a flow, and have a
84: stabilizing effect against KH instabilities. They may, however, introduce
85: current-driven (CD) instabilities (e.g., \citealt{ac92,lba00}).
86: \citet{bk02} studied
87: the interplay between KH and CD instabilities and concluded that large-scale
88: deformation of magnetic fields associated with the CD mode can effectively
89: saturate the KH surface vortices and thus aid in jet survival.
90: Whether magnetized
91: jets can indeed travel large distances without being disrupted remains
92: an area of active research (e.g., \citealt{nm04}).
93:
94: By comparison, the stability of magnetically driven jets and winds {\it during
95: launching and early propagation} is less explored. \citet{lb96} studied the
96: 3D stability of a jet accelerated and pinched by a purely
97: toroidal magnetic field. They found that the $m=1$ (kink) instability can
98: cause the tip of the jet to fold back upon itself.
99: The mode is stabilized by poloidal magnetic fields in the simulations
100: of \citet{lb97}, where the jet is squirted along the (initially
101: uniform) poloidal field lines by the high pressure created near the
102: central object through rapid equatorial infall. \citet{ocp03} examined
103: the 3D stability of a cold jet launched magnetically from a Keplerian
104: disk. They too adopted
105: an initially uniform magnetic field threading the disk vertically. The
106: differential rotation between the disk and a stationary (pressure-supported)
107: corona winds up the field lines, generating a larger toroidal field at
108: a smaller radius. The gradient
109: in the toroidal field can bend the initially vertical field lines away from
110: the disk axis by more than $30\degr$, enabling steady outflow through the
111: magnetocentrifugal mechanism. \citet{op99} showed that a relatively large
112: mass loading is required to generate a large enough toroidal field gradient
113: to open up the field lines for steady magnetocentrifugal wind driving in
114: 2D; lightly-loaded outflows remain unsteady,
115: generating knots episodically. The 3D jet of \citet{ocp03} has parameters
116: in this episodic regime. More recently, \citet{ks05} carried
117: out 3D simulations of disk-corona system threaded (again) by vertical
118: field lines. They found that a jet is produced by the Uchida-Shibata
119: mechanism, despite the non-axisymmetric perturbations imposed on the
120: disk rotation rate. In this letter, we are interested in the stability
121: of cold magnetocentrifugal winds accelerated steadily along field lines
122: inclined more than $30^\circ$ away from the axis, as in the original
123: picture of \citet{bp82}.
124:
125: \section{Simulation Setup and Numerical Results}
126: \label{desc}
127:
128: We simulate the disk-driven magnetocentrifugal wind using Cartesian
129: coordinate system, with the $z$-axis along the rotation axis, and
130: $x$- and $y$-axis in the disk plane. Our calculations are carried
131: out using an MPI-parallel version of the ZEUS-3D MHD code \citep{cnf94},
132: which we have previously used to simulate 2D axisymmetric winds
133: (\citealt{klb99,klb03a}; \citealt{a05}, Paper I
134: hereafter). The 2D simulations serve as the starting points for
135: our new, 3D calculations. They are specified by three functions on
136: the disk surface: the distributions of disk rotation speed $v_d(\varpi)$,
137: vertical field strength $B_z(\varpi)$, and rate of mass loading per
138: unit area $\rho(\varpi) v_z(\varpi)$, where $\varpi$ is the cylindrical
139: radius from the axis, and $\rho$ and $v_z$ are the density and vertical
140: component of the injection speed at the base of the wind. Inside a
141: sphere of radius $r=\varpi_g$, we
142: soften the gravitational field of the central point mass to avoid
143: singularity (eq.~[11] of Paper I). The softening yields an
144: equilibrium disk in sub-Keplerian rotation inside $\varpi_g$.
145: Outside $\varpi_g$, the disk rotation is
146: exactly Keplerian. The magnetocentrifugal wind is
147: launched from this portion of the disk, from $\varpi_g$ to
148: an outer radius $\varpi_0$. The material coming off of the
149: outer edge is assumed to slide outwards along the equatorial
150: plane to fill all available space.
151:
152: On the launching surface between $\varpi_g$ and $\varpi_0$, we impose
153: the Blandford-Payne distributions for density $\rho\propto
154: \varpi^{-3/2}$ and field strength
155: $B_z\propto \varpi^{-5/4}$; the latter is multiplied by a spline
156: function $S(\varpi)$
157: to bring it to zero at the outer edge of the launching region (see
158: eq.~[14] of Paper I), as demanded by the space-filling requirement.
159: Cold material is injected into the wind at a slow speed $v_z=0.1\
160: V_K (\varpi)\ S(\varpi)$, where $V_K$ is the Keplerian speed.
161: Inside the softening radius $\varpi_g$,
162: $B_z$ continues to increase slowly inwards, reaching a maximum value
163: at the origin. In
164: this sub-Keplerian region, the field lines are generally not
165: inclined by a large enough angle away from the axis
166: to drive a cold outflow centrifugally. Here, we inject a low-density
167: material along the field lines, with $v_z$ set to twice the local
168: escape velocity. This tenuous, fast-moving axial flow carries a
169: small fraction ($8.2\%$) of the total mass flux. It provides a
170: clean inner boundary to the magnetocentrifugally driven wind, the
171: focus of our study, and may represent the stellar wind inferred
172: in some young stellar objects from blue-shifted absorption lines
173: (e.g., \citealt{e06}).
174:
175: The 2D wind solutions are characterized by a dimensionless mass-loading
176: parameter
177: $
178: \mu_g= {4\pi \rho v_z V_K/B_z^2},
179: $
180: where the density, injection and rotation speeds, and field strength
181: are evaluated at the radius $\varpi_g$ on the disk. In ``light''
182: winds with $\mu_g \ll 1$, the magnetic field is initially dominated
183: by the poloidal component near the launching surface. It becomes
184: toroidally dominated only outside the Alfv\'en surface \citep{s96}.
185: As $\mu_g$ approaches unity, the winds become toroidally
186: dominated all the way to the launching surface. These are ``heavy''
187: winds. In Paper I, we have explored the structure of 2D winds with
188: $\mu_g$ varying from $6.25\times 10^{-4}$ to 6.25. Since our main
189: interest here is to determine whether a 2D wind
190: is stable in 3D to potential instabilities driven by toroidal field,
191: we decide to focus on the $\mu_g=0.625$ case, which is representative
192: of toroidally-dominated heavy winds. The heavy wind has the added
193: advantage of having a relatively low Alfv\'en speed, which allows for
194: a relatively large timestep, which in turn enables the simulation to
195: reach a later physical time than the lighter wind simulations that we
196: have also performed.
197:
198: Our simulations are carried out in dimensionless quantities. We set
199: the inner radius of the Keplerian disk to $\varpi_g=1$, and the outer
200: radius of the wind-launching region to $\varpi_0=5$. The simulation
201: box extends
202: far beyond the launching region, to $\pm 500$ in the $x$ and $y$
203: directions, and to $400$ in the $z$ direction. We adopt a
204: $240\times240\times132$ grid, with a uniform sub-grid of
205: $80\times80\times32$ covering the innermost $12\times12\times4$
206: region (which includes all of the launching surface), and the
207: remaining grid spaced logarithmically. As in the 2D
208: case, we impose conditions on the electromotive force at the
209: launching surface such that
210: the field lines are firmly anchored on the disk at their
211: footpoints while able to twist and bend freely in response to
212: the stresses in the
213: wind \citep{klb99,klb03a}. A technique based on Appendix A3
214: of \citet{ocp03} is used to ensure the anchoring of the field lines
215: on the disk to machine accuracy in a Cartesian coordinate system.
216: On the remaining boundaries, the standard outflow conditions in
217: ZEUS-3D are adopted.
218:
219: We perturb the initially steady axisymmetric wind through the boundary
220: conditions on the disk between $\varpi_g\le \varpi \le \varpi_0$. We
221: increase the
222: mass loading rate (through the density at the base of the wind) by
223: a factor of $\sqrt{10}$ on one half of the disk ($x \ge 0$) and
224: decrease it by the same factor on the other half ($x < 0$).
225: This perturbation is initiated at $t=0$, and is kept throughout
226: the simulation. The goal is to determine whether the large
227: non-axisymmetric perturbation imposed at the launching surface
228: can lead to disruption of the wind during its acceleration and
229: initial phase of propagation, particularly by the kink ($m=1$) mode.
230: The numerical results are shown in Figs.~\ref{fig:f1} and \ref{fig:f2}.
231:
232: Fig.~\ref{fig:f1}
233: shows the snapshots of column density in the $y$ direction at
234: four representative times, in units of the radius divided by the
235: Keplerian speed at $\varpi_g$. (In this unit, the rotation period
236: at the inner edge of the Keplerian disk is $2\pi$.) At time $t=0$,
237: the column density in the axisymmetric wind is well collimated at
238: large distances along the axis, as predicted by \citet{shu95}
239: from asymptotic analysis. As time progresses, an increasingly large
240: portion of the wind becomes distorted as the perturbation propagates
241: outwards from the launching surface. There is no evidence, however,
242: for any growth of instabilities that are commonly expected for such
243: a toroidally dominated wind. Indeed, the perturbed region appears to
244: settle quickly into a {\it nonaxisymmetric} steady state, as can be
245: seen by comparing the column density contours in the inner parts of
246: the last two frames.
247:
248: Fig.~\ref{fig:f2} displays selected properties of the apparent steady state.
249: In the upper panels, we show the distributions of volume density
250: in an $xy$-plane at height $z=30$ and in an $xz$-plane at $y=0$
251: for the time $t=165$, corresponding to $26.3$ times the rotation
252: period at the inner edge of the Keplerian disk. Clearly, the
253: density distribution is strongly asymmetric. In the $xy$-plane,
254: it is dominated by a trailing spiral arm outside the central
255: region -- the region occupied by the non-magnetocentrifugal
256: outflow injected near the axis (termed ``the axial column''
257: hereafter). The spiral is created by the smearing of the denser
258: material magnetocentrifugally launched from one half of
259: the disk surface by rotation. Wind rotation is evident from
260: the velocity vectors displayed, particularly in the central
261: region where the velocity field is dominated by rotation rather
262: than outflowing motion. Inside the axial column, there is some
263: hint of a 4-armed spiral structure in density distribution. We
264: believe the structure is numerical in origin, since the conditions
265: at the base of the axial column are kept axisymmetric. Most likely,
266: it is generated by the rectangular grid, which is not ideal for
267: following rotating motion near the axis. Nevertheless, the numerical
268: artifact appears localized inside the axial column. Between the
269: column and the
270: surrounding magnetocentrifugal wind lies a shell of high density.
271: Most likely, it is created by the squeezing of the strong toroidal
272: magnetic field in the magnetocentrifugal wind against the strong
273: poloidal field in the column. The compressed shell is
274: evident in the density distribution in the $xz$-plane.
275: The shell, which encases the axial column, is
276: tilted away from the $z$-axis. The dense ridge further to the
277: right of the axis (at an angle $\sim 45^\circ$) corresponds
278: to the large-scale spiral arm in the $xy$-plane, which has a
279: helical shape in 3D.
280:
281: The strong asymmetry in mass distribution all but disappears in
282: the distribution of magnetic field, as shown in the lower panels
283: of Fig.~\ref{fig:f2}. In the $xy$-plane, contours of constant {\it total}
284: field strength form nearly concentric rings. Close inspection
285: shows that the rings are shifted slightly in the positive $x$
286: direction. The white contour near the center marks the location
287: where $B_z=(B_x^2+B_y^2)^{1/2}$. Roughly speaking, it divides
288: the axial column of non-magnetocentrifugal outflow (inside the
289: contour) where the magnetic field is mainly poloidal from the
290: toroidally dominated magnetocentrifugal part of the wind; the
291: latter occupies most of the space. The bending of the axial
292: column can be seen more clearly in the $xz$-plot. Except for the
293: narrow region inside the white contour, the magnetic field
294: is toroidally dominated, including the launching surface. The
295: much more symmetric field distribution indicates that the mechanical
296: structure of the steady wind is controlled to a large extent
297: by magnetic stresses, rather than forces due to fluid motions.
298:
299: \section{Discussion: a Built-in Stabilizer for Magnetocentrifugal
300: Winds?}
301: \label{discussion}
302:
303: The magnetocentrifugal wind in our simulation appears stable in 3D.
304: There is no evidence for rapid growth of instabilities that would
305: lead to flow disruption, despite the fact that the magnetic field
306: in the wind is toroidally dominated all the way to the launching
307: surface. In particular, there is not hint of exponential growth of
308: the kink ($m=1$) mode, even though our perturbation at the base
309: of the wind is designed to maximize the $m=1$ component. The same
310: conclusion appears to hold for the more lightly loaded winds that
311: we have done as well (see also \citealt{klb03b}), although
312: in these cases we are unable to run the simulations for as long.
313:
314: The most likely reason for the stability is that, in our simulations,
315: the perturbed magnetocentrifugal part of the wind encloses a (light)
316: fast-moving outflow near the axis with a poloidally dominated magnetic
317: field. Mathematically, the fast axial flow is used to provide a
318: clean inner boundary to the outer part of wind driven through the
319: magnetocentrifugal
320: mechanism, which fails close to the axis because of unfavorable field
321: line inclinations. Physically, it may represent a flow driven
322: non-magnetocentrifugally along open field lines anchored on young
323: stars, perhaps by nonlinear Alfv\'en waves generated through magnetic
324: footpoint motions (e.g., \citealt{si05}). A magnetically
325: dominated stellar outflow is also seen in the simulations of
326: \citet{u06} for disk-magnetosphere interaction in the ``propeller''
327: regime. One may attempt to test the supposition by removing the
328: poloidally dominated fast outflow in the axial region.
329: However, if we were to do this, the field line originally at
330: the interface between the inner flow and outer wind would bend inwards,
331: pulling the field lines right outside of it to a more vertical
332: position. The magnetocentrifugal mechanism would shut off for
333: these field lines, leaving them loaded with little material. A
334: more tenuous outflow may still be possible along these unfavorably
335: inclined field lines, driven for example by Alfv\'en waves. We
336: would then be back to essentially the original
337: configuration: namely, a lighter and perhaps faster flow dominated
338: by poloidal magnetic field enclosed by a more heavily loaded wind
339: that becomes increasingly toroidally dominated at large distances.
340: Our simulations suggest that the same lightly loaded, nearly vertical
341: field lines that force
342: the open field lines further out on the disk to bend by more than
343: 30$^\circ$ away from the axis to make the magnetocentrifugal
344: wind-launching possible in the first place may stabilize the
345: launched wind at the same time. In other words, if a wind is
346: driven magnetocentrifugally, its stability may be guaranteed by the
347: built-in stabilizer. This two-component structure is intrinsic to
348: the X-wind theory, where the stabilizer is envisioned as the open
349: stellar field \citep{os95}. The same stabilizing mechanism
350: should work equally well, if not better, in the disk-wind picture
351: where, in addition to the stellar field, lightly loaded field
352: lines on the inner part of the disk can also contribute to wind
353: stabilization, especially if there is magnetic flux accumulation
354: due to disk accretion.
355:
356: Our simulations are limited to the region of acceleration and early
357: propagation of magnetocentrifugal winds, where they appear stable.
358: Whether such winds can stably propagate to much larger distances
359: remains to be determined. Kink instabilities are seen in some
360: simulations of MHD jet propagation (e.g., \citealt{nm04}).
361: The adopted jets are different from the magnetocentrifugally
362: driven jet in our picture, which is simply the densest part of
363: a space-filling wind that also includes an inner, axial region
364: dominated by a poloidal magnetic field and an outer, more tenuous
365: wide-angle component \citep{shu95}.
366: We have treated the disk as a fixed boundary. Allowing
367: the disk to evolve in response to angular momentum removal by
368: the wind may lead to instability in the coupled disk-wind
369: system (\citealt{l94}; see, however, \citealt{k04}). We plan
370: to address this important issue in the future through numerical
371: experiments.
372:
373: \acknowledgements
374: This work was supported in part by NASA grants NAG 5-7007, 5-9180,
375: 5-12102 and NNG05GJ49G, and NSF grants AST 00-93091 and 0307368.
376:
377: \begin{thebibliography}{}
378:
379: \bibitem[Anderson et al.(2005)]{a05}Anderson, J. M, Li, Z.-Y.,
380: Krasnopolsky, R. \& Blandford, R. D. 2005, \apj, 630, 945
381:
382: \bibitem[Appl \& Camenzind(1992)]{ac92} Appl, S., \& Camenzind, M.
383: 1992, \aap, 256, 354
384:
385: \bibitem[Baty \& Keppens(2002)]{bk02} Baty, H., \& Keppens, R. 2002,
386: \apj, 580, 800
387:
388: \bibitem[Begelman(1998)]{be98} Begelman, M. C. 1998, \apj, 493, 291
389:
390: \bibitem[Blandford \& Payne(1982)]{bp82} Blandford, R. D., \& Payne,
391: D. G. 1982, \mnras, 199, 883
392:
393: \bibitem[Bodo et al.(1998)]{bo98} Bodo, G., Rossi, P., Massaglia, S.,
394: Ferrari, A., Malagoli, A., Rosner, R. 1998, \aap, 133, 1117
395:
396: \bibitem[Clarke, Norman, \& Fiedler(1994)]{cnf94} Clarke, D. A.,
397: Norman, M. L., \& Fiedler, R. A. 1994, ZEUS-3D User Manual (Tech. Rep.
398: 015; Urbana-Champaign: National Center for Supercomputing Applications)
399:
400: \bibitem[Edwards et al.(2006)]{e06} Edwards, S., Fischer, W., Hillenbrand, L.
401: \& Kwan, J. 2006, \apj, 646, 319
402:
403: \bibitem[Eichler(1993)]{e93} Eichler, D. 1993, \apj, 419, 111
404:
405: \bibitem[Ferrari(1998)]{f98} Ferrari, A. 1998, \araa, 36, 539
406:
407: \bibitem[Hardee(2004)]{h04} Hardee, P. E. 2004, \apss, 293, 117
408:
409: \bibitem[Heyvaerts \& Norman(1989)]{hn89} Heyvaerts, J., \& Norman, C.
410: 1989, \apj, 347, 1055
411:
412: \bibitem[Kigure \& Shibata(2005)]{ks05}Kigure, K. \& Shibata, K. 2005, \apj,
413: 634, 879
414:
415: \bibitem[K\"onigl(2004)]{k04}K\"onigl, A. 2004, \apj, 617, 1267
416:
417: \bibitem[Krasnopolsky, Li, \& Blandford(1999)]{klb99} Krasnopolsky, R., Li,
418: Z.-Y., \& Blandford, R. 1999, \apj, 526, 631
419:
420: \bibitem[Krasnopolsky, Li, \& Blandford(2003a)]{klb03a} Krasnopolsky, R., Li,
421: Z.-Y., \& Blandford, R. 2003a, \apj, 595, 631
422:
423: \bibitem[Krasnopolsky, Li, \& Blandford(2003b)]{klb03b} Krasnopolsky, R., Li,
424: Z.-Y., \& Blandford, R. 2003b, \apss, 287, 75
425:
426: \bibitem[Lery, Baty, \& Appl(2000)]{lba00} Lery, T., Baty, H., \& Appl,
427: S. 2000, \aap, 355, 1201
428:
429: \bibitem[Lubow, Papaloizou, \& Pringle(1994)]{l94} Lubow, S.,
430: Papaloizou, J., \& Pringle, J. 1994, \mnras, 268, 1010
431:
432: \bibitem[Lucek \& Bell(1996)]{lb96} Lucek, S. G., \& Bell, A. R. 1996, \mnras,
433: 281, 245
434:
435: \bibitem[Lucek \& Bell(1997)]{lb97} Lucek, S. G., \& Bell, A. R. 1997, \mnras,
436: 290, 327
437:
438: \bibitem[Nakamura \& Meier(2004)]{nm04} Nakamura, M., \& Meier, D. L.
439: 2004, \apj, 617, 123
440:
441: \bibitem[Ostriker \& Shu(1995)]{os95} Ostriker, E. C. \& Shu, F. H. 1995,
442: ApJ, 447, 813
443:
444: \bibitem[Ouyed \& Pudritz(1999)]{op99} Ouyed, R., \& Pudritz, R. E.
445: 1999, \mnras, 309, 233
446:
447: \bibitem[Ouyed, Clarke, \& Pudritz(2003)]{ocp03} Ouyed, R., Clarke, D.
448: A., \& Pudritz, R. E. 2003, \apj, 582, 292
449:
450: \bibitem[Shu et al.(1995)]{shu95}Shu, F. H., Najita, J., Ostriker, E.
451: \& Shang, S. 1995, \apj, 455, L155
452:
453: \bibitem[Spruit(1996)]{s96} Spruit, H. C. 1996, Evolutionary Processes in
454: Binary Stars, NATO ASI Series C., 477, 249
455:
456: \bibitem[Suzuki \& Inutsuka(2005)]{si05} Suzuki, T. \& Inutsuka, S.
457: 2005, \jgr, 111, A06101, doi:10.1029/2005JA011502
458:
459: \bibitem[Uchida \& Shibata(1985)]{us85} Uchida, Y. \& Shibata, K.
460: 1985, \pasj, 36, 105
461:
462: \bibitem[Ustyugova et al.(2006)]{u06}
463: Ustyugova, G. V., Koldoba, A. V., Romanova, M. M. \& Lovelace, R. V. E.
464: 2006, \apj, 646, 304
465:
466: \end{thebibliography}
467:
468: \begin{figure}
469: \plotone{f1.eps}
470: \caption{Distributions of column density (log scale) in the $y$ direction
471: at four representative times, with 5 contours per decade. The inner
472: contours for the last two
473: times appear nearly identical, signaling the approach to a steady
474: state.
475: } \label{fig:f1}
476: \end{figure}
477: \begin{figure}
478: \plotone{f2.eps}
479: \caption{Properties of the non-axisymmetric steady state. Upper
480: panels show the density distribution (log scale) in an $xy$-plane
481: at height $z=30.14$ (left) and in the $xz$-plane at $y=0$ (right).
482: Superposed are vectors of velocity field in the plane, with the
483: length of arrow proportional to the magnitude of velocity.
484: Lower panels show the distribution of total field strength
485: (log scale) in the same $xy$- (left) and $xz$-plane (right), with
486: white contours marking the location where $B_z=(B_x^2+B_y^2)^{1/2}$.
487: Roughly speaking, the magnetic fields inside (outside) the
488: contours are poloidally (toroidally) dominated.
489: } \label{fig:f2}
490: \end{figure}
491:
492: \end{document}
493: