1: \documentclass[12pt,preprint]{aastex}
2: %\documentclass[manuscript]{aastex}
3: %% \documentclass[preprint2]{aastex}
4: %% \documentclass[preprint2,longabstract]{aastex}
5:
6: \newcommand{\bmax}{\beta_{\rho{\rm max}}}
7: \newcommand{\rmax}{\rho_{\rm max}}
8: \newcommand{\bl}[1]{\mbox{\boldmath$ #1 $}}
9:
10: \slugcomment{accepted by Astrophysical Journal Letters}
11:
12: \shorttitle{Magnetized Cloud Fragmentation}
13: \shortauthors{Kudoh \& Basu}
14:
15: \begin{document}
16:
17: \title{Three-dimensional Simulation of Magnetized Cloud Fragmentation
18: Induced by Nonlinear Flows and Ambipolar Diffusion}
19:
20: \author{Takahiro Kudoh\altaffilmark{1}
21: and Shantanu Basu\altaffilmark{2}}
22:
23: \altaffiltext{1}{Division of Theoretical Astronomy,
24: National Astronomical Observatory of Japan,
25: Mitaka, Tokyo 181-8588, Japan; kudoh@th.nao.ac.jp.}
26: \altaffiltext{2}{Department of Physics and Astronomy,
27: University of Western Ontario, London,
28: Ontario N6A 3K7, Canada; basu@astro.uwo.ca.}
29:
30: \begin{abstract}
31: We demonstrate that the formation of collapsing cores
32: in subcritical clouds is accelerated by nonlinear flows, by
33: performing three-dimensional non-ideal MHD simulations.
34: An initial random supersonic (and trans-Alfv\'enic) turbulent-like
35: flow is input into a self-gravitating
36: gas layer that is threaded by a uniform magnetic
37: field (perpendicular to the layer) such that the initial mass-to-flux
38: ratio is subcritical.
39: Magnetic ambipolar diffusion occurs very rapidly initially due to the
40: sharp gradients introduced by the turbulent flow. It subsequently occurs
41: more slowly in the traditional near-quasistatic manner, but in regions
42: of greater mean density than present in the initial state.
43: The overall timescale for runaway growth of the first core(s) is
44: several $\times\, 10^6$ yr, even though
45: previous studies have found a timescale of
46: several $\times\, 10^7$ yr when starting with linear perturbations and similar
47: physical parameters.
48: Large-scale supersonic flows exist in the cloud and provide an
49: observationally testable distinguishing characteristic from
50: core formation due to linear initial perturbations.
51: However, the nonlinear flows have decayed sufficiently that
52: the relative infall motions onto the first core are subsonic,
53: as in the case of starting from linear initial perturbations.
54: The ion infall motions are very similar to those of neutrals;
55: however, they lag the neutral infall
56: in directions perpendicular to the mean magnetic field direction and lead
57: the neutral infall in the direction parallel to the mean magnetic field.
58:
59: \end{abstract}
60:
61: \keywords{ISM: clouds -- ISM: magnetic fields -- MHD -- diffusion -- turbulence}
62:
63: \section{Introduction}
64:
65: In the standard model of low-mass star formation, molecular clouds have
66: an initially subcritical mass-to-flux ratio,
67: and spend a relatively long time ($\sim 10^7$ years) undergoing quasi-static
68: fragmentation by ambipolar diffusion until cores come to be supercritical.
69: After a core becomes supercritical, it collapses on a dynamical timescale
70: ($\sim 10^6$ years) to form either one or a small multiple system of
71: stars \citep[e.g.,][]{shu87,mos91}. Models of nonaxisymmetric
72: fragmentation by this mechanism have been presented recently in
73: several papers \citep[][hereafter KBOY]{ind00,bas04,cio06,kud07}.
74: This scenario is supported by the observed relatively low efficiency
75: of star formation in molecular clouds, i.e., only $1-5\%$ for
76: clouds as a whole but several times greater within cluster-forming cores
77: \citep[see][]{lad03}.
78: An alternative model is that star formation begins in clouds with
79: supercritical mass-to-flux ratio that have additional support due to
80: turbulence \citep[e.g.,][]{nak98,har01}.
81: In this model, star formation occurs relatively rapidly,
82: as turbulence dissipates over a dynamical timescale,
83: and the large-scale magnetic field plays a minor role.
84: The rapid star formation model is supported by the observational
85: results that the age spreads of young stars in nearby
86: molecular clouds are often a few $\times \,10^6$ years \citep{elm00,har01,har03},
87: and the low fraction of clouds that are observed to be in a pre-star-formation
88: state.
89:
90: Recently, \citet{li04} and \citet{nak05} have modeled
91: mildly subcritical clouds and shown that the timescale of
92: cloud fragmentation is reduced by supersonic turbulence. They showed this by
93: performing two-dimensional simulations in the thin-disk approximation.
94: Such a model can explain both relatively rapid star formation
95: and the relatively low star formation efficiency in molecular clouds
96: that is not well explained if star formation starts from a supercritical cloud.
97: In this Letter, we study the three-dimensional extension of the
98: \citet{li04} model, by including a self-consistent calculation
99: of the vertical structure and dynamics of the cloud.
100: We confirm that
101: supersonic flows can significantly shorten the timescale of core formation in
102: three-dimensional subcritical clouds,
103: and clarify how the scaling of various physical quantities allows this
104: to occur.
105:
106: \section{Numerical Model}
107:
108: We solve the three-dimensional magnetohydrodynamic
109: (MHD) equations including self-gravity and ambipolar diffusion,
110: assuming that neutrals are much more numerous than ions.
111: Instead of solving a detailed energy equation, we assume isothermality
112: for each Lagrangian fluid particle \citep{kud03,kud06}.
113: For the neutral-ion collision time and associated quantities,
114: we follow \citet{bas94}. The basic equations
115: are summarized in KBOY.
116:
117: As an initial condition, we assume hydrostatic equilibrium of
118: a self-gravitating one-dimensional cloud along the $z$-direction in
119: a Cartesian coordinate system $(x,y,z)$.
120: Though nearly isothermal, a molecular cloud is usually surrounded
121: by warm material, such as neutral hydrogen gas.
122: Hence, we assume that the initial sound speed makes a transition
123: from a low value $c_{s0}$ to a high value $c_{sc}$ at
124: a distance $z=z_c$, with a transition length $z_d$ (see
125: eq. 16 of KBOY).
126: We take $c_{sc}^2=10\,c_{s0}^2$, $z_c=2H_0$, and $z_d=0.1H_0$,
127: where $c_{s0}$ is the initial sound speed at $z=0$,
128: $H_0=c_{s0}/\sqrt{2\pi G \rho_0}$, and $\rho_0$ is the initial density
129: at $z=0$. A numerical solution for the initial density distribution
130: shows that it is almost the same as the \citet{spi42}
131: solution for an equilibrium isothermal layer in the region $0 \leq z \leq z_c$.
132: We also assume that the initial magnetic field is uniform
133: along the $z$-direction.
134:
135: A set of fundamental units for this problem are
136: $c_{s0}$, $H_0$, and $\rho_0$. These yield a time
137: unit $t_0=H_0/c_{s0}$. The initial magnetic field
138: strength ($B_0$) introduces a dimensionless free parameter
139: \begin{equation}
140: \beta_0 \equiv \frac{8 \pi p_0}{B_0^2}
141: = \frac{8 \pi \rho_0 c_{s0}^2}{B_0^2}
142: = 2\, \frac{c_{s0}^2}{V_{A0}^2} ,
143: \end{equation}
144: the ratio of gas to magnetic pressure at $z=0$. In the above relation,
145: we have also used $V_{A0}\equiv B_0/\sqrt{4 \pi \rho_0}$,
146: the initial Alfv\'en speed at $z=0$.
147: In the sheet-like equilibrium cloud with a vertical
148: magnetic field, $\beta_0$ is related to the mass-to-flux ratio
149: for Spitzer's self-gravitating layer. The mass-to-flux ratio
150: normalized to the critical value is
151: $\mu_S \equiv 2\pi G^{1/2} \Sigma_S / B_0$,
152: where
153: $\Sigma_S= 2\rho_0 H_0$
154: is the column density of the Spitzer cloud.
155: Therefore,
156: $\beta_0=\mu_S^2$.
157: Although the initial cloud we used is not exactly
158: the same as the Spitzer cloud, $\beta_0$ is a good
159: indicator of whether or not the magnetic field
160: can prevent gravitational instability.
161: For the model presented in this Letter, we choose $\beta_0=0.25$ so that
162: the cloud is slightly subcritical. For this choice,
163: $V_{A0} = 2.8\,c_{s0}$.
164: Dimensional values of all quantities can be found
165: through a choice of $\rho_0$ and $c_{s0}$.
166: For example, for $c_{s0}=0.2$ km s$^{-1}$ and $n_0=\rho_0/m_n=10^4$ cm$^{-3}$,
167: we get $H_0=0.05$ pc, $t_0=2.5 \times 10^5$ yr, and
168: $B_0=40\,\mu$G if $\beta_0=0.25$.
169:
170: The level of magnetic coupling in the partially ionized gas is
171: characterized by numerical values of the ion number density
172: $n_{\rm i}$ and neutral-ion collision timescale $\tau_{\rm ni}$.
173: From eqs. (8) and (9) of KBOY, and using standard values
174: of parameters in that paper as well as the values of units used
175: above, we find an initial midplane ionization fraction
176: $x_{\rm i,0} = n_{\rm i,0}/n_0 = 9.5 \times 10^{-8}$ and
177: a corresponding neutral-ion collision time $\tau_{\rm ni,0} = 0.11 t_0$.
178: The ionization fraction $x_{\rm i}$ and timescale $\tau_{\rm ni}$
179: at other densities can be found from the initial midplane values given
180: that they both scale $\propto \rho^{-1/2}$ \citep{elm79}.
181:
182: In this equilibrium sheet-like gas layer, we input a nonlinear perturbation
183: to $v_x$ and $v_y$ at each grid point. Independent realizations of
184: $v_x$ and $v_y$ are generated
185: in Fourier space with amplitudes drawn from a Gaussian
186: distribution and consistent with power spectrum $v_k^2 \propto k^{-4}$.
187: Here, $k = (k_x^2+k_y^2)^{1/2}$ is the total wavenumber.
188: Our adopted power spectrum means that the large scale has more energy
189: than the small scale.
190: This perturbation is consistent with that implemented in \citet{li04}.
191: The rms value of the initial velocity perturbation in physical space, $v_a$,
192: is about $3c_{so}$, so that $v_a \simeq V_{A0}$ as well.
193:
194: The method of solution and boundary conditions are described
195: by KBOY \citep[see also][]{kud99,oga04}.
196: The computational region is $|x|,|y| \leq 8\pi H_0$ and
197: $0 \leq z \leq 4H_0$, with a number of grid points for
198: each direction $(N_x,N_y,N_z)=(64,64,40)$.
199:
200:
201:
202: \section{Results}
203:
204: The top panel of Figure 1 shows the time evolution of the maximum density
205: $\rmax$ at $z=0$.
206: The simulation is stopped when $\rmax = 30\rho_0$.
207: The solid line shows the result when we input an initially nonlinear
208: supersonic perturbation to the $\beta_0=0.25$ model.
209: The dashed line ($\beta_0=0.25$) and the dash-dotted line ($\beta_0=4$)
210: show the result in the case of the small (linear) initial perturbation
211: that we have applied in KBOY.
212: This figure shows that the timescale
213: of collapsing core formation for the nonlinear perturbation case is much shorter
214: than that for the linear perturbation case, when $\beta_0$ is the same.
215: Even when the initial cloud is subcritical ($\beta_0=0.25$), the core
216: formation occurs on almost the same timescale as that of
217: the supercritical ($\beta_0=4$) linear perturbation case
218: \footnote{When the initial cloud is supercritical and the perturbation is supersonic,
219: the collapsing core formation happens quickly, at $t \simeq t_0$, from the initial flow.
220: This may be {\it too} rapid to agree with observations.}.
221: The bottom panel of Figure 1 shows the time evolution of the maximum value of
222: density at $z=0$ ($\rmax$) and $\beta$ at the location of maximum density ($\bmax$).
223: At first, $\bmax$ increases rapidly
224: up to $\sim 0.9$ due to rapid ambipolar diffusion in the highly compressed
225: regions caused by the initial supersonic perturbation.
226: However, there is enough stored magnetic energy in the compressed region
227: that it rebounds and starts oscillations, with $\bmax$ around $0.7$
228: and increasing gradually. Eventually, $\bmax$ becomes
229: $> 1$ and the dense region collapses to form a core.
230: This figure implies that $\beta$ is a good indicator
231: to see whether a subregion of the cloud is supercritical or not.
232: The evolution of $\rmax$
233: confirms that there is an initial compression followed
234: by a rebound to a lower density (still greater than the initial background
235: value) and subsequent oscillations until a runaway collapse starts when
236: continuing ambipolar diffusion has created a region with $\beta > 1$.
237:
238: Figure 2 shows an image of the logarithmic density at the last snapshot
239: ($t=20.5t_0$, when $\rmax=30\rho_0$) for the nonlinear perturbation case of
240: the subcritical cloud ($\beta_0=0.25$).
241: The top panel shows the cross section at $z=0$, and the bottom panel
242: shows the cross section at $y=-5.9H_0$. The value of $y$ for the bottom
243: panel is chosen so that the vertical cut passes through the maximum
244: density point. A collapsing core is located in the
245: vicinity of $x=-20H_0,y=-5H_0$.
246: The size of the core is similar to that created by linear initial
247: perturbations (see Fig. 2 in KBOY), although the shape is
248: notably less circular.
249:
250: The top panel of Figure 3 shows the density and $x$-velocity
251: of neutrals and ions
252: along an $x$-axis cut at $y=-5.9H_0, z=0$, taken from the snapshot
253: illustrated in Figure 2.
254: The $x$-velocities show infall motion toward the center of the core,
255: although the core itself is moving with nonzero negative $x$-velocity.
256: The relative infall speed to the core is subsonic and about $0.35c_{s0}$.
257: It is comparable to the case of
258: initial linear perturbation (see Fig. 9 in KBOY).
259: However, there are systematic motions throughout the simulation region
260: that are still supersonic at this time, even though the initial turbulent
261: energy has decayed somewhat. This is qualitatively distinct
262: from the corresponding case with linear perturbations,
263: and should be observationally testable.
264: For the single core that is formed in this simulation, the
265: systematic core motion is about $0.5c_{s0}$ in the $x$-direction.
266: However, it can be even larger in other realizations.
267: The ion velocity $\bl{v}_i$ can be related to the neutral velocity
268: $\bl{v}$ and the magnetic acceleration $\bl{a}_M$ in the assumed limit
269: of low ion inertia by the relation
270: \begin{equation}
271: \bl{v}_i = \bl{v} + \bl{a}_M\, \tau_{\rm ni}.
272: \end{equation}
273: Figure 3 reveals that the ion infall motions are smaller in magnitude than the
274: neutral motions due to the retarding magnetic acceleration; however
275: the relative drift in the $x$-direction between ions and neutrals
276: is typically within $0.05c_{s0}$.
277: In order to see the force balance in the core, we plot
278: the $x$-accelerations in the bottom panel of Figure 3
279: along the same $x$-axis as the top panel.
280: These values are normalized by the computational units.
281: The inward gravitational acceleration ($a_G$) dominates around the core.
282: The acceleration from the magnetic force ($a_M$) and
283: thermal pressure force ($a_T$) resist the contraction.
284: The net acceleration ($a_{\rm net}$) is working toward
285: the center of the core.
286:
287: The top panel of Figure 4 shows the density and $z$-velocities
288: along a $z$-axis cut at $x=-20H_0,y=-5.9H_0$ from the snapshot illustrated
289: in Figure 2. The $z$-velocities also show infall motion toward
290: the center of the core. The relative neutral infall speed to the core
291: is also subsonic and about $-0.35c_{s0}$. The ion velocity in the vertical
292: direction is almost identical to the neutral velocity but actually
293: very slightly {\it greater} in magnitude at the peak.
294: The bottom panel of Figure 4 shows the $z$-accelerations along
295: the same $z$-axis as in the top panel. In the $z$-direction,
296: $a_G$ is nearly balanced by $a_T$, and $a_M$ is almost negligible,
297: {\it but working in the same direction as gravity}.
298: The net acceleration ($a_{\rm net}$) is also working toward
299: the center of the core.
300: The inward pointing vertical magnetic acceleration associated
301: with an hourglass-type
302: magnetic field morphology explains why the ion velocity in the
303: $z$-direction is greater in magnitude than the neutral velocity.
304:
305: \section{Discussion and Conclusions}
306:
307: Our three-dimensional MHD simulations have shown that the supersonic nonlinear
308: flows significantly reduce the timescale of collapsing core formation
309: in subcritical clouds. It is of order several $\times \,10^6$ years
310: for typical parameters, or $\sim 10$ times less than
311: found in the linear initial
312: perturbation studies of \cite{bas04}, \cite{cio06}, and KBOY.
313: Our supersonic perturbations are also trans-Alfv\'enic, in agreement with
314: analysis of observed magnetic field strengths \citep{bas00}.
315: It is also in line with
316: the theoretical result of \citet{kud03,kud06} that turbulent
317: clouds will settle into
318: a trans-Alfv\'enic state when global motion/expansion is allowed,
319: even if driven with turbulence that is initially super-Alfv\'enic.
320:
321: To see how accelerated ambipolar diffusion can occur, we note
322: that the magnetic induction equation (see, e.g., eq. 3 of KBOY)
323: and our assumptions of ionization balance
324: can be used to estimate the diffusion time $\tau_d \propto
325: \rho^{3/2}L^2/B^2$, where $L$ is the gradient length scale introduced
326: by the initial turbulent compression and $B$ is the magnetic field
327: strength.
328: Because the compression by the nonlinear flow is nearly one-dimensional,
329: the magnetic field scales roughly as $B \propto L^{-1}$ within the
330: flux freezing approximation.
331: If the compression is rapid enough that vertical hydrostatic
332: equilibrium cannot be established (unlike in previous calculations
333: using the thin-disk approximation), then $\rho \propto L^{-1}$ as well
334: (i.e., one-dimensional contraction without vertical settling),
335: and $\tau_d \propto L^{5/2}$. This means that diffusion can occur
336: quickly (and lead to a rapidly rising value of $\beta$)
337: if the turbulent compression creates small values of $L$.
338: If diffusion is so effective during the first turbulent compression
339: that a dense region becomes magnetically supercritical, then it
340: will evolve directly into collapse. Alternately, the stored magnetic
341: energy of the compressed (and still subcritical) region may lead to a
342: reexpansion of the
343: dense region. The timescale for this, in the flux-freezing limit,
344: is the Alfv\'en time $\tau_A \propto L\rho^{1/2}/B$, which scales
345: $\propto L^{3/2}$ for the above conditions. Thus, $\tau_d$ decreases
346: more rapidly than $\tau_A$, and sufficiently small turbulent-generated
347: values of $L$ may lead to enough magnetic diffusion that collapse occurs
348: before any reexpansion can occur.
349: See \citet{elm07} for some similar
350: discussion along these lines.
351: Ultimately, whether or not reexpansion
352: of the first compression can occur depends on the strength of the
353: turbulent compression, mass-to-flux ratio of the initial cloud,
354: and neutral-ion collision time.
355: In a preliminary study of such effects, we found that keeping
356: $v_a$ and $\beta_0$ fixed but allowing significantly poorer
357: or stronger neutral-ion coupling yielded
358: differing results. For $\tau_{\rm ni,0} = 0.3 t_0$ (poorer coupling),
359: a collapsing
360: core formed due to the first compression, at $t\simeq 1.4t_0$,
361: while for $\tau_{\rm ni,0} = 0.05 t_0$ it happened after more
362: oscillations than in our standard model, at $t\simeq 85t_0$.
363: A full parameter study will be required to elucidate.
364:
365: If reexpansion of the initial compression does occur, as in the standard model
366: presented in this paper, then there is enough time for the vertical
367: structure to settle back to near-hydrostatic equilibrium, in which
368: case $B \propto \rho^{1/2}$. Since the compressed and reexpanded region
369: executes oscillations about a new mean density, it is convenient to
370: analyze the scalings in terms of the density $\rho$. The diffusion time
371: now scales as $\tau_d \propto \rho^{-1/2}$.
372: This yields a scaling of $\tau_d$
373: that is the traditionally used one (and is satisfied by
374: design in the thin-disk approximation). However, the diffusion
375: occurs more rapidly than it would in the initial state due to the elevated
376: value of $\rho$ in the compressed but oscillating region (see
377: Figure 1).
378:
379: \acknowledgments
380:
381: SB was supported by a grant from NSERC.
382: Numerical computations were done mainly on the VPP5000
383: at the National Astronomical Observatory.
384:
385:
386: \begin{thebibliography}{}
387:
388: \bibitem[Basu(2000)]{bas00}
389: Basu, S. 2000, \apjl, 540, L103
390:
391: \bibitem[Basu \& Mouschovias(1994)]{bas94}
392: Basu, S., \& Mouschovias, T. Ch. 1994, \apj, 432, 720
393:
394: \bibitem[Basu \& Ciolek(2004)]{bas04}
395: Basu, S., \& Ciolek, G. E. 2004, \apjl, 607, L39
396:
397: \bibitem[Ciolek \& Basu(2006)]{cio06}
398: Ciolek, G. E., \& Basu, S. 2006, \apj, 652, 442
399:
400: \bibitem[Elmegreen(1979)]{elm79}
401: Elmegreen, B. G. 1979, \apj, 232, 729
402:
403: \bibitem[Elmegreen(2000)]{elm00}
404: Elmegreen, B. G. 2000, \apj, 530, 277
405:
406: \bibitem[Elmegreen(2007)]{elm07}
407: Elmegreen, B. G. 2007, \apj, 668, 1064
408:
409: \bibitem[Hartmann(2001)]{har01}
410: Hartmann, L. 2001, \aj, 121, 1030
411:
412: \bibitem[Hartmann(2003)]{har03}
413: Hartmann, L. 2003, \apj, 585, 398
414:
415: \bibitem[Indebetouw \& Zweibel(2000)]{ind00}
416: Indebetouw, R., \& Zweibel, E. G. 2000, \apj, 532, 361
417:
418: \bibitem[Kudoh et al.(1999)]{kud99}
419: Kudoh, T., Matsumoto, R., \& Shibata, K. 1999,
420: Computational Fluid Dynamics Journal, 8, 56
421:
422: \bibitem[Kudoh \& Basu(2003)]{kud03}
423: Kudoh, T., \& Basu. S. 2003, \apj, 595, 842
424:
425: \bibitem[Kudoh \& Basu(2006)]{kud06}
426: Kudoh, T., \& Basu. S. 2006, \apj, 642, 270
427:
428: \bibitem[Kudoh et al.(2007)]{kud07}
429: Kudoh, T., Basu. S., Ogata, Y., \& Yabe, T. 2007, \mnras, 380, 499 (KBOY)
430:
431: \bibitem[Lada \& Lada(2003)]{lad03}
432: Lada, C. J., \& Lada, E. A. 2003, \araa, 41, 57
433:
434: \bibitem[Li \& Nakamura(2004)]{li04}
435: Li, Z.-Y., \& Nakamura, F. 2004, \apjl, 609, L83
436:
437: \bibitem[Mouschovias(1991)]{mos91}
438: Mouschovias, T. Ch. 1991, in The Physics of Star Formation and Early Stellar Evolution,
439: ed. C. J. Lada \& N. D. Kylafis (Dordrecht: Kluwer), 449
440:
441: \bibitem[Nakamura \& Li(2005)]{nak05}
442: Nakamura, F., \& Li, Z.-Y. 2005, \apj, 631, 411
443:
444: \bibitem[Nakano(1998)]{nak98}
445: Nakano, T. 1998, \apj, 173, 87
446:
447: \bibitem[Ogata et al.(2004)]{oga04}
448: Ogata, Y., Yabe, T., Shibata, K., \& Kudoh, T. 2004,
449: International Journal of Computational Methods, 1, 201
450:
451: \bibitem[Shu et al.(1987)]{shu87}
452: Shu, F. H., Adams, F. C., \& Lizano, S. 1987, \araa, 25, 23
453:
454: \bibitem[Spitzer(1942)]{spi42}
455: Spitzer, L., Jr. 1942, \apj, 95, 329
456:
457: \end{thebibliography}
458:
459: \clearpage
460:
461: \begin{figure}
462: \epsscale{.80}
463: \plotone{f1.eps}
464: \caption{
465: {\it Top}: the time evolution of maximum densities at $z=0$.
466: The solid line shows the evolution for an initially nonlinear
467: supersonic perturbation and $\beta_0=0.25$. The dashed line ($\beta_0=0.25$) and
468: the dash-dotted line ($\beta_0=4$) show the evolution for models with
469: a linear initial perturbation as calculated by KBOY.
470: The dotted line shows the evolution
471: for an initially nonlinear supersonic perturbation and $\beta_0=0.25$,
472: but without ambipolar diffusion.
473: {\it Bottom}: evolution of the maximum density (solid line) at $z=0$ of
474: the simulation box for the model with $\beta_0=0.25$ and
475: nonlinear initial perturbation, and evolution of $\beta$ at
476: the location of maximum density (dashed line).
477: }
478: \end{figure}
479:
480: \clearpage
481:
482: \begin{figure}
483: \plotone{f2.eps}
484: \caption{
485: Logarithmic density image at $t=20.5t_0$ for the nonlinear
486: perturbation case of the subcritical cloud ($\beta_0=0.25$).
487: The top panel shows the cross section at $z=0$, and the bottom panel
488: shows the $x-z$ cross section at $y=-5.9H_0$.
489: An mpeg animation of the evolution of the density image from $t=0$ to $t=20.5t_0$ is
490: available online.
491: }
492: \end{figure}
493:
494: \begin{figure}
495: \plotone{f3.eps}
496: \caption{
497: {\it Top}: the density (solid line), $x$-velocity of neutrals
498: (dashed line), and $x$-velocity of ions (dotted line) along an $x$-axis cut
499: at $y=-5.9H_0,z=0$, in the snapshot shown in Fig. 2.
500: The $x$-positions are measured by offset from
501: $x_c=-20H_0$, which is the maximum density point for the core.
502: {\it Bottom}: the $x$-accelerations along
503: the same $x$-axis as the top panel.
504: The solid line is the gravitational acceleration ($a_G$),
505: the dotted line is the magnetic acceleration ($a_M$),
506: the dashed line is acceleration due to thermal pressure ($a_T$),
507: and the dash-dotted line is the net acceleration ($a_{\rm net}$).
508: All quantities are normalized by the computational units.
509: }
510: \end{figure}
511:
512: \begin{figure}
513: \plotone{f4.eps}
514: \caption{
515: {\it Top}: the density and $z$-velocity of neutrals and ions
516: along a $z$-axis cut at $x=-20H_0,y=-5.9H_0$,
517: in the snapshot shown in Fig. 2.
518: {\it Bottom}: the $z$-accelerations along
519: the same $z$-axis as the top panel.
520: The line styles are the same as those in Fig. 3.
521: }
522: \end{figure}
523:
524:
525: \end{document}
526: