1:
2: \documentclass[12pt,preprint]{aastex}
3: %\documentclass[manuscript]{aastex}
4: %\documentclass[preprint2]{aastex}
5: %\usepackage{graphicx,amsmath,amssymb}
6: %\slugcomment{submitted to the Astrophysical Journal}
7:
8: \shorttitle{Accretinng Gas Sphere}
9: \shortauthors{HANAWA AND SOEDA}
10:
11: \begin{document}
12:
13: \title{Critical Accretion Rate for Triggered Star Formation}
14:
15: \author{Tomoyuki Hanawa\altaffilmark{1}
16: and Akihito Soeda\altaffilmark{1}}
17: \altaffiltext{1}{Center for Frontier Science,
18: Chiba University, Inage-ku, Chiba, 263-8522, Japan}
19: \email{hanawa@cfs.chiba-u.ac.jp}
20:
21: \begin{abstract}
22: We have reexamined the similarity solution for a self-gravitating
23: isothermal gas sphere and examined implication to star formation
24: in a turbulent cloud. When parameters are adequately chosen,
25: the similarity solution expresses an accreting isothermal gas sphere
26: bounded by a spherical shock wave. The mass and radius of the sphere
27: increases in proportion to the time, while the central density
28: decreases in proportion to the inverse square of time.
29: The similarity solution is specified by the accretion rate and
30: the infall velocity. The accretion rate has an upper limit
31: for a given infall velocity. When the accretion rate is
32: below the upper limit, there exist a pair of similarity
33: solutions for a given set of the accretion rate and infall
34: velocity. One of them is confirmed to be unstable against
35: a spherical perturbation. This means that the gas sphere
36: collapses to initiate star formation only when the accretion
37: rate is larger than the upper limit. We have also examined
38: stability of the similarity solution against non-spherical
39: perturbation. Non-spherical perturbations are found to be
40: damped.
41: \end{abstract}
42:
43: \keywords{accretion --- hydrodynamics ---
44: shock waves --- stars: formation}
45:
46: \section{INTRODUCTION}
47:
48: Similarity solutions have contributed very much to our
49: understanding of star formation process. The classical
50: similarity solution by \citet{larson69} and \citet{penston69}
51: elucidated the runaway nature of gravitational collapse.
52: The density increases in proportion to the inverse square
53: of the time during the runaway collapse phase. We learned
54: from the similarity solutions of \citet{shu77} and \citet{hunter77}
55: that the accretion rate of a protostar is of the order of
56: $ c _s {}^3 / G $ where $ G $ and $ c _s $ denote the
57: gravitational constant and the isothermal sound speed of gas.
58:
59: The similarity solution is also used to evaluate the effects
60: of rotation and magnetic field. The similarity solutions of
61: \citet{narita84} and \citet{saigo98} indicated that the runaway
62: collapse cannot be prevented by rotation if once initiated.
63: Collapse of a rotating magnetized gas cloud is described
64: by the similarity solution of \citet{krasnopolsky02}.
65: Ambipolar diffusion is taken into account in the the
66: similarity solution of \citet{adams07}.
67: \citet{tsai95} extended the similarity solution to include
68: shock wave. \citet{shu02} extended the similarity solution
69: involving a shock wave for application to champagne phase
70: of an \ion{H}{2} region.
71:
72: \citet{tsai95} found two classes of similarity solution;
73: the first class describes accretion onto protostar while the
74: second one does failure of star formation. The central density
75: decreases in proportion to the inverse square of the
76: time, $ \rho _c \, \propto \, t ^{-2} $, in the second class
77: solution. Although this solution has not gained much
78: attention thus far, it provides an insight on dynamical
79: compression of a molecular cloud core.
80: If a dense clump of gas is compressed by an external
81: force, the temporal increase in the density may trigger
82: gravitational collapse and star formation. One can surmise
83: existence of threshold of gravitational collapse. If the
84: dynamical compression is either weak or short, the clump
85: will bounce back to expansion. A shock wave will be formed
86: when accreting gas is stoped by the expansion \citep[see, e.g.]
87: {adams07}.
88: The similarity solution of
89: \cite{tsai95} demonstrated that a spherical cloud can expand
90: even when it is steadily compressed by a shock wave.
91: On the other hand, the shock compressed gas sphere will
92: collapse owing to its self gravity if the shock is strong
93: and lasts for a long enough period.
94:
95: In this paper we reexamine the similarity solution of
96: \cite{tsai95} while keeping its negative implication
97: in mind. We find the condition for existence of similarity
98: solution describing expansion of a gas sphere.
99: Conversely it will tell us condition for a shock compressed
100: gas sphere to collapse by its self gravity.
101: We also study stability of the similarity solution.
102: The similarity solution denies collapse due to the self
103: gravity only when it is stable.
104:
105: We review the similarity solution in \S 2.1 and show
106: the method of linear stability analysis in \S 2.2.
107: Technical details on the stability analysis are given
108: in Appendix. Properties of similarity solutions, such as
109: accretion rate and infall velocity, are shown in \S 3.1.
110: Stability of the similarity solution is given in \S 3.2
111: and \S 3.3. We discuss implications of our analysis in \S 4.
112:
113: \section{Model and Methods of Computation}
114:
115: \subsection{Similarity Solution}
116:
117: We consider an isothermal gas of which distribution is spherically
118: symmetric. Then the hydrodynamical equations are expressed as
119: \begin{equation}
120: \frac{\partial \rho}{\partial t} \; + \;
121: \frac{1}{r^2} \, \frac{\partial}{\partial r} \, (r^2 \rho v)
122: \; = \; 0 \, , \label{hydro1}
123: \end{equation}
124: \begin{equation}
125: \frac{\partial v}{\partial t} \; + \;
126: v \, \frac{\partial v}{\partial r} \; + \;
127: \frac{1}{\rho} \, \frac{\partial P}{\partial r} \; + \;
128: \frac{G M_r}{r ^2} \; = \; 0 \, , \label{hydro2}
129: \end{equation}
130: and
131: \begin{equation}
132: \frac{\partial M _r}{\partial r} \; = \; 4 \pi r ^2 \, \rho
133: \label{hydro3}
134: \end{equation}
135: where
136: \begin{equation}
137: P \; = \; c _{\rm s} ^2 \rho \, .
138: \end{equation}
139: Here, the symbols, $ \rho $, $ v$, $ P $, $ M _r $, $ G $, and
140: $ c _{\rm s} $ denote the density, velocity, pressure,
141: mass inside the radius $r $, gravitational constant,
142: and the isothermal sound speed, respectively.
143: As originally shown by Larson (1969)
144: and Penston (1969), the hydrodynamical equations have
145: similarity solutions,
146: \begin{eqnarray}
147: \rho (r, \, t) & = & \frac{\varrho (\xi)}{4 \pi G t ^2}
148: \, , \label{sim1}\\
149: v \, (r, \, t) & = & \frac{u (\xi)}{c _s} \, , \label{sim2}\\
150: M _r (r, \, t) & = & \frac{c _{\rm s} ^3 \, t}{G} \, \mu (\xi)
151: \label{sim3}
152: \end{eqnarray}
153: where
154: \begin{equation}
155: \xi \; = \; \frac{r}{c _{\rm s} \, t} \; .
156: \label{sim4}
157: \end{equation}
158:
159: We restrict ourselves in the case of $ t \, > \, 0 $ in the
160: following. Substituting Equations (\ref{sim1}) through (\ref{sim4}) into
161: Equations (\ref{hydro1}) through (\ref{hydro3}) we obtain
162: \begin{eqnarray}
163: \frac{\partial \varrho}{\partial \xi} & = &
164: - \, \frac{\left[ \mu \, - \, 2 \, (\xi \, - \, u) ^2 \right] \, \varrho}
165: {\left[(u \, - \, \xi) ^2 \, - \, 1 \right] \, \xi} \label{sim5}\\
166: \frac{\partial u}{\partial \xi} & = &
167: \frac{(\mu \, - \, 2) \, (\xi \, - \, u)}
168: {\left[(u \, - \, \xi) ^2 \, - \, 1 \right] \, \xi} \label{sim6}
169: \end{eqnarray}
170: and
171: \begin{equation}
172: \mu \; = \; \xi \, (\xi \, - \, u) \, \varrho \, .
173: \end{equation}
174:
175: We assume that the density is finite at the center since
176: we are interested in application to a starless core, i.e.,
177: case of no star formation.
178: Then the density and velocity should be expressed as
179: \begin{eqnarray}
180: \varrho & = & \varrho _0 \; - \; \frac{\rho _0}{6}
181: \left( \varrho _0 \, - \, \frac{2}{3} \right) \, \xi ^2
182: \, + \, \frac{\rho _0}{45} \, \left( \varrho _0 \, - \,
183: \frac{2}{3} \, \right) \, \left( \varrho _0 \, - \,
184: \frac{1}{2} \right) \, + \, {\cal O} \, (\xi ^6) \\
185: u & = & \frac{2 \xi}{3} \, - \, \frac{\varrho _0}{45} \,
186: \left( \varrho _0 \, - \, \frac{2}{3} \right) \, \xi ^3 \, + \,
187: {\cal O} \, (\xi ^5)
188: \end{eqnarray}
189: near the center. Following \citet{tsai95} we assume that
190: the flow has a shock wave at $ \xi \, = \, \xi _{\rm sh} $.
191: Then the Rankine-Hugonio relation gives us the condition,
192: \begin{equation}
193: \frac{\varrho _+}{\varrho _-} \; = \;
194: \frac{u _+ \, - \, \xi _{\rm sh}}{u _- \, - \, \xi _{\rm sh}}
195: \, ,
196: \end{equation}
197: and
198: \begin{equation}
199: (u _+ \, - \, \xi _{\rm sh}) \, (u _+ \, - \, \xi _{\rm sh})
200: \; = \; 1 \, .
201: \end{equation}
202: where $ \varrho _+ $ and $ \varrho _- $ denote the
203: the densities of the pre- and post-shocked gases,
204: respectively, and$ u _+ $ and $ u _- $ do the velocities of them,
205: respectively.
206:
207: In the region far from the origin, a solution of Equations
208: (\ref{sim5}) and (\ref{sim6}) approaches the asymptotic solution,
209: \begin{eqnarray}
210: \varrho & = & \frac{\dot{M}}{v _{\rm inf} \xi ^2} \; + \;
211: {\cal O} \, (\xi ^{-3}) \, , \\
212: u & = & - \, v _{\rm inf} \; + \; {\cal O} \, (\xi ^{-1}) \, .
213: \end{eqnarray}
214: The symbol, $ v _{\rm inf}, $ denotes the infall velocity
215: at the infinity while $ \dot{M} $ denotes the accretion
216: rate.
217: The similarity solution can be specified either by
218: ($ \varrho _c $,~$ \xi _{\rm sh} $) or by
219: ($ v _{\rm inf}$,~$\dot{M}$).
220:
221: Note that Larson-Penston solution
222: has the same asymptotic form. Only when
223: $ v _{\rm inf} $ and $ \dot{M}$ vanish,
224: the solution Equations (\ref{sim5})
225: and (\ref{sim6}) have a different asymptotic form,
226: \lq \lq plus solutions\rq \rq of \citet{shu77}.
227:
228: We integrate equations (\ref{sim5}) and (\ref{sim6}) by
229: the 4th order Runge-Kutta method from $ \xi \, = \, 0 $
230: for a given set of $ \varrho _c $ and $ \xi _{\rm sh}$
231: to obtain a similarity solution. The infall velocity and
232: accretion rate are obtained numerically as a function of
233: $ \varrho _c $ and $ \xi _{\rm sh}$. We also obtain the
234: mass enclosed in the shock front,
235: \begin{equation}
236: M _c \; = \; \int _0 ^{\xi _{\rm sh}} \varrho \,
237: \xi ^2 \, d \xi \, .
238: \end{equation}
239:
240: \subsection{Spherical and Non-spherical Perturbations}
241:
242: We have performed a normal mode analysis to examine the stability
243: of the similarity solution. In the analysis,
244: the density is assumed to be expressed as
245: \begin{equation}
246: \rho \, (r,~t) \; = \;
247: \frac{\varrho (\xi) \, + \, t ^\sigma \,
248: \delta \varrho (\xi) \, Y _{\ell} ^m \, (\theta,~\varphi)}
249: {4 \pi G t ^2} \, , \label{density-p}
250: \end{equation}
251: where $ Y _{\ell} ^m \, (\theta,~\varphi)$ denotes the
252: spherical harmonic function. This particular form is
253: chosen because the similarity solution has no specific
254: timescale. An eigenmode has a growth timescale and
255: an unstable perturbation grows exponentially,
256: when the unperturbed state is stationary and has a
257: specific timescale. When the physical quantities
258: vary according to a power law in time in an unperturbed state,
259: the eigenmode grows (or decay) in proportion to a power of
260: time, $ | t | ^\sigma$. See \citet{hanawa99} for the
261: justification of Equation (\ref{density-p}). They analyzed
262: linear stability of the Larson-Penston solution against
263: a non-spherical perturbation, using the coordinates,
264: $ (\xi,~\theta,~\varphi,~\ln~t)$ instead of
265: the ordinary spherical coordinates, $(r,~\theta,~\varphi,~t)$.
266: It is shown that the similarity solution can be expressed
267: as a steady state and an unstable mode grows in propotion
268: to $ \exp~(\sigma~\ln~t) \, = \, t ^\sigma $ in the coodinates.
269:
270: The power index, $ \sigma $,
271: is obtained as an eigenvalue of the perturbation equations
272: as shown in Appendix. When the real part of
273: $ \sigma $ is positive, the mode grows in time and the
274: similarity solution is unstable. We call the real part
275: of $\sigma$, the growth index in the following according
276: to the suggestion by F.~H. Shu, the referee of this paper.
277: When $ \sigma $ is complex, the mode grows (or decays) while
278: oscillating. The imaginary part of $ \sigma $ is related with
279: the frequency of oscillation, although the physical
280: oscillation period increases with time.
281:
282: \section{Results}
283:
284: \subsection{Similarity Solutions}
285:
286: First we obtained a series of similarity solutions for a given
287: $ \varrho _c $ by increasing $ \xi _{\rm sh} $.
288: The infall velocity, $ v _{\rm inf} $, decreases
289: monotonically with increase in $ \xi _{\rm sh}$.
290: It reaches $ v _{\rm inf} \, = \, 0 $ at some $ \xi _{\rm sh} $
291: and the similarity solution terminates. By compiling
292: the similarity solutions, we obtained $ \dot{M} $ as a function
293: of $ v _{\rm inf} $ and $ \xi _{\rm sh} $.
294: The curves denote $ \dot{M} $ as a function of
295: $ \xi _{\rm sh} $ for given $ v _{\rm inf} $ in Figure \ref{dMdt}.
296:
297: The accretion rate, $ \dot{M} $, is maximum at a certain
298: $ \xi _{\rm sh}$ for a given $ v _{\rm inf}$. It increases
299: with increase in $ \xi _{\rm sh}$ at a small $ \xi _{\rm sh} $
300: while it decreases with increase in $ \xi _{\rm sh} $ at a
301: large $ \xi _{\rm sh}$. The former is denoted by
302: thin curves while the latter is by thick ones. This means
303: that there exists two similarity solutions for a given set
304: of $ v _{\rm inf}$ and $ \dot{M}$.
305:
306: Figure \ref{Solution} shows two similarity solutions having
307: $ v _{\rm inf} \, = \, 1.4 $ and $ \dot{M} \, = \, 1.204 $.
308: The solid curves denote the solution of
309: $ \xi _{\rm sh} \, = \, 0.986 $ while the dashed curves do
310: that of $ \xi _{\rm sh} \, = \, 0.574 $. Both the solutions have
311: the same density and velocity distributions in the region
312: of $ \xi \, \ge \, 0.986 $. The main difference is the
313: location of the shock front. The shock compressed gas
314: sphere is denser and expands more slowly in the solution
315: denoted by the dashed curve. Since the expansion is slow,
316: the solution is similar to the \citet{bonnor56}-\citet{ebert55}
317: solution for a self-gravitating isothermal gas sphere.
318: As well as the Bonnor-Ebert sphere, the latter solution
319: is shown to be unstable (\S 3).
320:
321: Figure \ref{dMdt-max} indicates that the similarity solution has
322: an upper limit on $ \dot{M} $ for a given $ v _{\rm inf} $.
323: The upper limit is highest $ \dot{M} _{\rm max} \, = \, 1.312 $
324: at $ v _{\rm inf} \, = \, 1.64 $ .
325: Similarity solutions do not exist for $ \dot{M} \, > \, 1.312$.
326: The implication of non-existence is discussed in \S 4.
327:
328: \subsection{Spherical Perturbations}
329:
330: Figure \ref{l=0} shows the growth index of the spherical
331: perturbation, $ \sigma _r $, as a function of $ \xi _{\rm sh} $ for a
332: series of similarity solutions having a given $ v _{\rm inf}$.
333: The solid curves denote the growth index of modes having
334: real index. The dashed lines denote the real part of
335: complex indecies. One of the index is positive
336: ($ \sigma _r \, > \, 0$) and the similarity solution is unstable
337: only when the shock radius ($ \xi _{\rm sh} $) is smaller than a
338: critical value. The condition of neutral stability coincides with
339: that of maximum accretion rate for a given $ v _{\rm inf}$ as
340: expected. When $ \xi _{\rm sh} $ is a little larger than
341: the critical, the solution is stable and the most slowly damping
342: mode has a real index. When $ \xi _{\rm sh} $ is large enough,
343: the solution is stable and the most slowly damping mode has
344: a complex index.
345:
346: Our survey is limited to modes having low indecies, i.e.,
347: in the range $ |\sigma _i | \, \le \, 1.0 $. It is however
348: unlikely that we have missed unstable spherical perturbations.
349: A spherical perturbation induces only sound waves and they are
350: confined within the gas sphere since it is bounded by the
351: shock wave. A high frequency sound wave has a shorter wavelength
352: and is unlikely to be Jeans unstable.
353:
354: \subsection{$ \ell $~=~2 Mode}
355:
356: We have studied $ \ell $~=~2 mode as a typical non-spherical
357: perturbation. This is in part because the dipole ($\ell \, = \, 1 $)
358: mode is unlikely to be excited. If the dipole mode grows, the
359: inner and outer parts of the gas sphere should move in the
360: opposite direction each other to keep the center of gravity.
361:
362: Figure~\ref{l=2v=1} denotes the eigenfrequencies of the
363: $ \ell \, = \, 2 $ modes for
364: similarity solutions of $ v _{\rm inf} $~=~1.0.
365: The abscissa denotes $ \xi _{\rm sh } $ while the ordinate
366: denotes the index, $ \sigma $. All the modes are
367: damping ($ \sigma _i \, < \, -1 $). The solid lines denote
368: the mode having the smallest damping index. The mode has a
369: small imaginary part ($\sigma _i \, \simeq \, \pm 0.1 $).
370: The mode having the second smallest damping index is denoted by
371: the dashed line in Figure~\ref{l=2v=1} and has a real
372: eigenfrequency (pure damping). The mode having the third
373: smallest damping index has an imaginary part in the
374: eigenfrequency. The imaginary part is similar to that
375: of the mode having the smallest damping index. These
376: three modes have a similar damping index of
377: $ -1.1 \, \le \, \sigma _r \, \le \, -1.0 $. The other
378: mode (dash-dotted line in Fig.~\ref{l=2v=1}) has a much
379: larger damping index.
380:
381: Figure~\ref{l=2v=4} is the same as Figure~\ref{l=2v=1} but
382: for $ v _{\rm inf} $~=~4.0. Again, all the modes are damping.
383: The damping index is larger than unity ($ \sigma _r \, < \, -1 $).
384: The oscillation frequency of the smallest damping mode is
385: larger than those of $ v _{\rm inf} $~=~1.0.
386:
387: One might ask the reason why the similarity solution is
388: stable against a bar ($\ell~=~2$) mode. We think that
389: the bar mode is damped by expansion of the gas sphere.
390: Since the radius of the shock front increases with the time,
391: the asphericity of the shock compressed gas sphere
392: decreases unless the displacement grows faster than
393: the radius. When a gas sphere is collapse, the bar
394: mode can be excited as shown by \citet{lin65} for the
395: pressure less gas and \citet{hanawa99} for an isothermal gas.
396:
397: We have not yet studied the modes of $ \ell \, \ge \, 3 $.
398: However they are also unlikely to be unstable, since
399: the self-gravity does not excite a short wave perturbation.
400:
401: \section{Discussion}
402:
403: We obtained the critical accretion rate above which there
404: exits no similarity solution. The critical rate can be
405: interpreted as the minimum accretion rate for a high
406: density clump to initiate self-gravitational collapse.
407: The critical accretion rate can be rewritten as
408: \begin{equation}
409: \left. \frac{dM}{dt} \right| _{\rm cr} \; = \;
410: \frac{3.6 \, c _s ^4}{G v} \, , \label{critical-phys}
411: \end{equation}
412: for $ v \, \ga \, 3 c _s$ in the dimensional form.
413:
414: Equation (\ref{critical-phys}) gives us an estimate
415: for a converging flow to initiate gravitational
416: collapse. We shall consider a spherical region of
417: which surface is surrounded by a converging flow.
418: The radius, inflow velocity, and density are assumed
419: to be $ r$, $ v $ and $ \rho $, respectively. Then
420: the gravitational collapse will be initiated when
421: the mass accretion rate exceeds the critical,
422: \begin{equation}
423: 4 \pi r ^2 \, \rho \, v \; > \;
424: \frac{3.6 \, c _s ^4}{G \, v} \, . \label{Jeans1}
425: \end{equation}
426: Equation (\ref{Jeans1}) can be rewritten as
427: \begin{equation}
428: \frac{2r}{\lambda _{\rm J}} \; > \;
429: \frac{0.3 \, c _{\rm s}}{v} \, , \label{Jeans2}
430: \end{equation}
431: where
432: \begin{equation}
433: \lambda _{\rm J} \; = \; \frac{2 \pi \, c _s}
434: {\displaystyle \sqrt{4 \pi G \rho}} \; .
435: \end{equation}
436: Since $ \lambda _{\rm J} $ denotes the Jeans length,
437: Equation (\ref{Jeans2}) means that the effective Jeans
438: length reduces in proportion to the inverse of the
439: Mach number.
440:
441: The Jeans mass is proportional to the cube of the Jeans length
442: for a given density. Thus the effective Jeans mass
443: should reduce to
444: \begin{eqnarray}
445: M _{\rm J,~eff} & = & M _{\rm J} \, \left(
446: \frac{\lambda _{\rm J,~eff}}{\lambda _{\rm J}}\right) ^3 \\
447: & = & M _{\rm J} \, \left( \frac{v}{0.3~c_s}\right) ^{-3} \, .
448: \end{eqnarray}
449: This implies that compression of sub Jeans mass clump may
450: result in gravitational collapse in the region of flow
451: convergence. Note that the effective Jeans mass is
452: several order of magnitude smaller than the classical
453: one when $ v \, \ga \, 3 \, c _s $.
454:
455: The compression should continue for a certain timescale for
456: a dynamically compressed clump to collapse by its self gravity.
457: If we evaluate the minimum timescale to be the effective Jeans length
458: divided by the flow velocity, it is shorter than the free-fall
459: timescale by a factor of the Mach number squared,
460: \begin{equation}
461: \tau _{\rm comp} \; \simeq \; \frac{\lambda _{\rm J,~eff}}{v} \;
462: \simeq \; \tau _{\rm ff} \, \left( \frac{v}{c _s}\right) ^{-2} \, .
463: \end{equation}
464:
465: The timescale can be translated into the wavelength of perturbation.
466: A compressed clump can collapse by the self gravity if the wavelength
467: of velocity perturbation is longer than the effective Jeans length.
468: If turbulence contains velocity perturbations of long wavelengths,
469: gravitational collapse due to dynamical compression will take place
470: somewhere in the cloud. In such case we can expect a number of
471: clumps of which masses are much smaller than the classical Jeans mass.
472:
473: \acknowledgments
474:
475: We thank T. Matsumoto and F. Nakamura for discussion and
476: valuable comments on the original manuscript.
477: We also thank F.~H. Shu for his valuable comments as
478: the referee.
479: We have added comparison of
480: \citet{tsai95} solution with Bonnor-Ebert
481: solution in the revised manuscript according to the comments
482: given to the original manuscript.
483: This study is financially supported in part by the
484: Grant-in-Aid for Scientific Research on Priority Area (19015003) of
485: The Ministry of Education, Culture, Sports, Science, and
486: Technology (MEXT).
487:
488:
489: \appendix
490:
491: \section{Linear Stability Analysis}
492:
493: Since the density is given by Equation (\ref{density-p})
494: in our linear stability analysis, the gravitational potential
495: should be expressed as
496: \begin{equation}
497: \Phi \,(r,~t) \; = \; c _s ^2 \, \left[
498: \psi \, (\xi) \; + \; t ^\sigma \, \delta \psi (\xi) \,
499: Y _{\ell} ^m (\theta,~\varphi) \right] \, .
500: \end{equation}
501: The velocity, $ \mbox{\boldmath$v$} \, = \,
502: (v _r,~v_\theta,~v_\varphi) $,
503: is assumed to be expressed as
504: \begin{eqnarray}
505: v _r (r,~t) & = & c _s \, \left[ u (\xi) \, + \, t ^\sigma \,
506: \delta u _r \, Y _{\ell} ^m (\theta,~\varphi) \right] \, ,\\
507: v _\theta (r,~t) & = & - \, c _s \, t ^\sigma \,
508: \frac{\delta u _\theta (\xi)}{\ell \, + \, 1} \,
509: \frac{\partial}{\partial \theta}
510: Y _\ell ^m (\theta,~\varphi) \, , \\
511: v _\varphi (r,~t) & = & - \, c _s \, t ^\sigma \,
512: \frac{\delta u _\theta (\xi)}{(\ell \, + \, 1) \, \sin \theta} \,
513: \frac{\partial}{\partial \varphi}
514: Y _\ell ^m (\theta,~\varphi) \, ,
515: \end{eqnarray}
516: in the spherical coordiantes.
517: We assume that the perturbation is vanishingly small in the
518: region very far from the center. In other words, we restrict
519: ourselves to search for an instability due to the
520: internal structure of the shock compressed gas sphere.
521: Thus the flow is assumed to have no
522: vorticity,
523: \begin{equation}
524: \left( \ell \, + \, 1 \right) \, \delta u _r \; + \;
525: \frac{d}{d\xi} \, \left(\xi \, \delta u _\theta \right)
526: \; = \; 0 \, .
527: \end{equation}
528:
529: The assumption of no vorticity is based on the Kelvin's
530: circulation theorem.
531: It says the circulation,
532: \begin{equation}
533: \Gamma \; \equiv \; \oint _{C} \mbox{\boldmath$u$}\cdot
534: d\mbox{\boldmath$u$}
535: \end{equation}
536: is conserved for an isothermal (barotropic) gas
537: \citep[see, e.g., Chapter 6 of][]{shu92}.
538: Since the Lagrangian loop {\it C} expands, the
539: circulation velocity should decrease with the time
540: to conserve $ \Gamma $, although it can grow
541: with the time if measured at a given $ \xi $.
542: See \citet{hanawa00} for more details on the mathematical
543: treatment of perturbations having vorticity.
544:
545: After some manipulation, the perturbation equations are
546: written as
547: \begin{equation}
548: \left(\sigma \, + \, 1 \right) \, \delta \varrho \; + \;
549: \frac{1}{\xi ^2} \, \frac{~d}{d\xi} \,
550: \left( \xi ^2 \, \delta f \right) \; + \;
551: \frac{\ell \, \varrho _0 \, \delta u _\theta}{\xi} \; = \; 0 \, ,
552: \label{delta-rho}
553: \end{equation}
554: \begin{eqnarray}
555: \left(\sigma \, + \, 2 \right) \, \delta f & + &
556: \frac{1}{\xi ^2} \, \left\{ \xi ^2 \,
557: \left[ 2 w \, \delta f \, + \, (1 \, - \, w ^2) \right]
558: \right\} \nonumber \\
559: & \; & - \, \left(\varrho _0 w \, + \, \frac{2}{\xi}\right)
560: \, \delta \varrho \; + \; \varrho _0 \delta \gamma
561: \; + \; \frac{\ell \varrho _0 w \delta u _\theta}{\xi} \;
562: = \; 0 \, ,
563: \end{eqnarray}
564: \begin{equation}
565: \frac{~d}{d\xi} \, \delta \psi \; = \;
566: \delta \gamma \, ,
567: \end{equation}
568: \begin{equation}
569: \frac{~d}{d\xi} \, \delta \gamma \; = \;
570: \delta \varrho \; - \; \frac{2 \, \delta \gamma}{\xi}
571: \; - \; \frac{\ell \, (\ell \, + \, 1)}{\xi ^2} \,
572: \delta \psi \label{delta-gamma}
573: \end{equation}
574: where
575: \begin{eqnarray}
576: w & = & u \, - \, \xi \, , \\
577: \delta u _\theta & = &
578: \frac{\ell \, + \, 1}{(\sigma \, + \, 1) \, \xi} \,
579: \left( w \, \delta u _r \; + \; \frac{\delta \varrho}{\varrho _0}
580: \; + \; \delta \psi \right) \, , \\
581: \delta f & = & \varrho _0 \, \delta u _r \; + \; w \,
582: \delta \varrho \, ,\\
583: \delta \gamma & = & \frac{\partial}{\partial \xi} \,
584: \delta \psi \, .
585: \end{eqnarray}
586: The perturbation equations do not contain the azimuthal
587: wavenumber, $ m $, and accordingly thus the growth rate
588: does not depend on $ m $. This is because the unperturbed
589: state (similarity solution) is spherically symmetric.
590:
591: We must consider the jump condition at the shock front
592: as well as the boundary conditions to solve the perturbation
593: equation. The density perturbation is discontinuous
594: between the pre- and post-shocked flows and contains
595: change due to shift of the shock front. The latter
596: introduces the Dirac's delta function proportional
597: to the shift and the density difference between the
598: pre- and post flows. After some manipulation we obtain
599: the jump condition,
600: \begin{eqnarray}
601: \left(w _+ \, - \, w _-\right) \, \left[ \delta f \right]
602: \; - \; \left(\varrho _+ \, - \, \varrho _- \right) \,
603: \left[w \delta u _r \, + \, \frac{\delta \varrho}{\varrho _0} \right]
604: & = & 0 \, , \\
605: \left(\sigma \, + \, 1 \right) \, \left[\delta \gamma \right] \;
606: + \; \left[ \delta f \right] & = & 0 \,
607: \end{eqnarray}
608: where the bracket denotes the difference between
609: $ \xi \, = \, \xi _{\rm sh} \, \pm \, \varepsilon $.
610: The symbols with the subscripts, $ + $ and $ - $, denote the
611: values at $ \xi \, = \, \xi _{\rm sh} \, \pm \, \varepsilon $,
612: respectively. The perturbation in the potential, $ \delta \psi $,
613: is continuous even at $ \xi \, = \, \xi _{\rm sh}$.
614: These jump conditions enable us to connect the perturbation
615: in the post-shocked flow ($ \xi \, < \, \xi _{\rm sh}$) and
616: that in the pre-shocked flow ($\xi \, > \, \xi _{\rm sh}$).
617:
618: When the perturbation is spherically symmetric ($ \ell \, = \, 0 $),
619: Equations (\ref{delta-rho}) and (\ref{delta-gamma}) are linearly
620: dependent and we obtain
621: \begin{equation}
622: \delta \gamma \; = \; - \, \frac{\delta f}{\sigma \, + \, 1}
623: \end{equation}
624: for $ \ell \, = \, 0$. Thus we need two independent boundary
625: conditions for $ \ell \, = \, 0 $ and three for $ \ell \, \ne \, 0 $.
626:
627: The spherical perturbation should vanish in the pre-shocked
628: since the unperturbed flow is supersonic. Otherwise the
629: perturbation diverges when the mode is unstable.
630: Thus the effective outer boundary is set at $ \xi \, = \,
631: \xi _{\rm sh} $ and the jump condition is applied there.
632: The other boundary condition is set so that the velocity
633: perturbation vanishes at the origin, $ \xi \, = \, 0 $
634: and is proportional to $ \xi $ near the origin.
635: We have integrated the perturbation equation numerically
636: with the Runge-Kutta method from the
637: origin and searched the eigenvalue, $ \sigma $, by
638: try and error.
639:
640: Non-spherical perturbations should also be regular at the origin
641: and vanishingly small at infinity. We calculated asymptotic solutions
642: around the origin and those around the infinity to examine the
643: condition for perturbations to be regular according to
644: \citet{hanawa99}.
645: Some asymptotic solutions are divergent since the perturbation
646: equations are singular at the origin and infinity.
647: When $ \ell \, \ne \, 0 $, the perturbation should be
648: expressed by the asymptotic solution,
649: \begin{eqnarray}
650: \delta \varrho & = & B \varrho _0 \, \xi ^\ell \, ,
651: \label{delta-rho-0}\\
652: \delta u _r & = & - \, A \ell \xi ^{\ell \, - \, 1}
653: \; - \; C \, (\ell \, + \, 2) \, \xi ^{\ell \, + \, 1} \, ,
654: \label{delta-u-0}\\
655: \delta \psi & =& \left[A \, \left(\sigma \, + \, 1 \, - \,
656: \frac{\ell}{3} \right) \, - \, B\right] \, \xi ^\ell \, ,
657: \label{delta-psi-0}
658: \end{eqnarray}
659: where
660: \begin{equation}
661: C \; = \; \frac{1}{4 \ell \, + \, 6} \,
662: \left[ \frac{\ell}{3} \, \left(\varrho _0 \, - \, \frac{2}{3}
663: \right)\, A \; + \; \left(\sigma \, - \, \frac{\ell \,- \, 1}{3}
664: \right)\, B \right] \, .
665: \end{equation}
666: Here the symbols, $ A $ and $ B $, denote arbitrary constants.
667: Equations (\ref{delta-rho-0}) and (\ref{delta-psi-0}) mean that
668: the density and potential perturbations can be expanded by
669: a series of polynomials starting from the $\ell$-th power
670: in the Cartesian coordinates. Otherwise the perturbation
671: diverges at the origin. The velocity perturbation is also
672: expanded by another series of polynomials starting from
673: the $ (\ell - 1) $-th order since it should be balanced
674: with gradient of the potential perturbation.
675:
676: The other boundary condition is given by the asymptotic
677: solution,
678: \begin{eqnarray}
679: \delta \varrho & = & \frac{2 D \, (\ell \, + \, 1) \, \varrho _0}
680: {(\sigma \, + \, \ell \, + \, 2)(\sigma \, + \, \ell \, + \, 3) \,
681: \xi ^{\ell \, + \, 3}} \, ,
682: \label{delta-rho-inf}
683: \\
684: \delta u _r & = & \frac{D \, (\ell \, + \, 1)}
685: {(\sigma \, + \, \ell \, + \, 2) \, \xi ^{\ell \, + \,2}} \, ,
686: \label{delta-u-inf}
687: \\
688: \delta \psi & = & \frac{D}{\xi ^{\ell\, + \,1}} \, ,
689: \label{delta-psi-inf}
690: \end{eqnarray}
691: for $ \xi \, \gg \, 1$. Here, the symbol $D $ denotes an
692: arbitrary constant. Equation (\ref{delta-psi-inf}) means
693: that the potential perturbation in the region
694: $ \xi \, \gg \, 1 $ is dominated by the
695: aspherical density distribution near the center.
696: In other words the density perturbation is too small to
697: affect the gravitational potential. Equation (\ref{delta-u-inf})
698: means that the velocity perturbation is induced by the gravitational
699: perturbation. Equation (\ref{delta-rho-inf}) means that
700: the density perturbation is induced by the velocity perturbation.
701: Thus our boundary conditions gurantee that the perturbation
702: is induced by internal change in the flow.
703:
704: We integrated the perturbation equations
705: both from the origin ($ \xi \, = \, 10 ^{-2} $)
706: and a very large $ \xi \,(\simeq \, 100)$
707: with the Runge-Kutta method.
708: We surveyed the eigenvalue, $ \sigma $, by examining whether
709: the numerically solutions from both ends satisfy the jump
710: condition at the shock front.
711:
712:
713: \begin{thebibliography}{}
714: \bibitem[Adams \& Shu(2007)]{adams07} Adams, F.~C., \& Shu, F.~H.
715: 2007, \apj, 671, 497
716: \bibitem[Bonnor(1956)]{bonnor56} Bonnor, W.~B. 1956, \mnras, 116, 351
717: \bibitem[Ebert(1955)]{ebert55} Ebert. R. 1955, \zap, 37, 217
718: \bibitem[Hanawa \& Matsumoto(1999)]{hanawa99}
719: Hanawa, T., \& Matsumoto, T. 1999, \apj, 521, 703
720: \bibitem[Hanawa \& Matsumoto(2000)]{hanawa00}
721: Hanawa, T., \& Matsumoto, T. 2000, \apj, 540, 962
722: \bibitem[Hunter(1977)]{hunter77} Hunter, C. 1977, \apj, 218, 834
723: \bibitem[Krasnopolsky \& K\"onigl(2002)]{krasnopolsky02}
724: Krasnopolsky, R., \& K\"onigl, A. 2002, \apj, 580, 987
725: \bibitem[Larson(1969)]{larson69} Larson, R.~B. 1969,
726: \mnras, 145, 271
727: \bibitem[Larson(2003)]{larson03}
728: Larson, R. B. 2003, Reports on Progress in Physics, 66, 1651
729: \bibitem[Lin, Mestel, \& Shu(1965)]{lin65} Lin, C.~C.,
730: Mestel, L., \& Shu, F.~H. 1965, \apj, 143, 1431
731: \bibitem[Narita, Hayashi, \& Miyama(1984)]{narita84}
732: Narita, S., Hayashi, C., Miyama, S.~M. 1984, Prog. Theor. Phys.,
733: 72, 1118
734: \bibitem[Penston(1969)]{penston69} Penston, M.~V. 1969,
735: \mnras, 144, 425
736: \bibitem[Saigo \& Hanawa(1998)]{saigo98}
737: Saigo, K., \& Hanawa, T. 1998, \apj, 493, 342
738: \bibitem[Shu(1977)]{shu77} Shu, F.~H. 1977, \apj, 214, 488
739: \bibitem[Shu(1992)]{shu92} Shu, F.~H. 1992, The Physics
740: of Astrophysics Volume II Gas Dynamics, (Mill Valley, University
741: Science Books)
742: \bibitem[Shu et al.(2002)]{shu02} Shu, F.~H., Lizano, S.,
743: Galli, D., Cant\'{o}, J. \& Laughlin, G. 2002, \apj, 580, 969
744: \bibitem[Tsai \& Hsu(1995)]{tsai95}
745: Tsai, J.~C., Hsu, J.~J.~L. \apj, 448, 774
746: \end{thebibliography}
747: %-------------------------
748:
749: \clearpage
750:
751: \begin{figure}[!hp]
752: \centering
753: \plotone{f1.eps}
754: \caption{The accretion rate, $ \dot{M}$, is shown as a function
755: of $ xi _{\rm sh}$ for a given $ v _{\rm inf}$.
756: \label{dMdt}}
757: \end{figure}
758:
759: \begin{figure}[!hp]
760: \centering
761: \plotone{f2.eps}
762: \caption{The density and velocity distributions are
763: shown for two similarity solutions having the same $ v _{\rm inf} $
764: and $ dM/dt $.\label{Solution}}
765: \end{figure}
766:
767:
768: \begin{figure}[!hp]
769: \centering
770: \plotone{f3.eps}
771: \caption{The critical accretion rate is shown as a function of
772: $ v _{\rm inf} $.\label{dMdt-max}}
773: \end{figure}
774:
775: \begin{figure}[!hp]
776: \centering
777: \plotone{f4.eps}
778: \caption{Each curve denotes the real part of eigenfrequency,
779: $ \sigma _r $, of a spherical perturbation as a function of the
780: shock radius, $ \xi _{\rm sh} $ for a given $ v _{\rm inf} $.
781: \label{l=0}}
782: \end{figure}
783:
784: \begin{figure}[!hp]
785: \centering
786: \plotone{f5.eps}
787: \caption{The same as Fig.~\ref{l=0} but for the imaginary
788: part, $ \sigma _i $.\label{l=0imag}}
789: \end{figure}
790:
791:
792: \begin{figure}[!hp]
793: \centering
794: \plotone{f6.eps}
795: \caption{The eigenfrequency, $ \sigma $, is shown as a
796: function of $ \xi _{\rm sh} $ for similarity solutions having
797: $ v _{\rm inf} \, = \, 1.0 $. The solid curves denote the
798: real part of $\sigma $ for non-spherical perturbations
799: having $ \ell \, = \, 2 $, while dashed curves do that of
800: imaginary part. Mode a has complex eigenfrequencies while
801: modes b and c have real eigenfrequencies.\label{l=2v=1}}
802: \end{figure}
803:
804: \begin{figure}[!hp]
805: \centering
806: \plotone{f7.eps}
807: \caption{The same as Fig,~\ref{l=2v=1} but for
808: $ v _{\rm inf} $~=~4.\label{l=2v=4}}
809: \end{figure}
810:
811: \end{document}
812: %%
813: %% End of file
814: