0710.1331/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: 
3: 
4: \shorttitle{Simulations on  a Galactic Gaseous Discs}
5: \shortauthors{Y\'a\~nez et al.}
6: 
7: 
8: \begin{document}
9: 
10: 
11: \title{Resonance Related Spiral Substructure in a Galactic Gaseous Disk}
12: 
13: 
14: \author{Miguel A. Y\'a\~nez,
15: Michael L. Norman}
16: \affil{Center for Astrophysics and Space Sciences, University of California at
17:     San Diego, La Jolla, CA 92093}
18: \email{myanez@ucsd.edu}
19: \email{mlnorman@ucsd.edu}
20: 
21: \author{Marco A. Martos}
22: \affil{Instituto de Astronom\'\i a, Universidad Nacional Aut\'onoma de 
23: M\'exico A. P.  70-264, M\'exico 04510, D. F.,  M\'exico}
24: \email{marco@astroscu.unam.mx}
25: 
26: \and
27: 
28: \author{John C. Hayes}
29: \affil{Lawrence Livermore National Laboratory, Livermore, CA 94550}
30: \email{jchayes@llnl.gov}
31: 
32: 
33: %
34: 
35: \begin{abstract}
36: 
37: We use high resolution (2048$^2$ zones) 2D hydrodynamic simulations to study
38:  the formation of spiral substructure in the gaseous disk  of a galaxy. 
39: The obtained gaseous response  is 
40: driven by a self-consistent non-axisymmetric potential  obtained from an 
41: imposed 
42: spiral mass distribution.
43: We
44: highlight the importance of ultraharmonic resonances in generating
45: these features. The temporal evolution of the system is followed with the 
46: parallel ZEUS-MP code, and we follow the steepening of perturbations 
47: induced by 
48: the spiral potential
49: until
50: large-scale shocks emerge. These shocks exhibit bifurcations that protrude 
51: from the 
52: gaseous arms and continue to steepen until  new shocks are 
53: formed.
54: When the contribution from the spiral potential relative 
55: to the axisymmetric background is increased from our default value, spurs 
56: protrude from the main arms after several revolutions of the gaseous disk.
57: Such spurs overlap
58:  on top of
59: the aforementioned  shocks. These results
60: support the hypothesis 
61: that a complicated gaseous response can coexist with an orderly spiral
62: potential term, in the sense that the underlying background potential can be 
63: smooth yet drive a gaseous response that is far more spatially complex.
64: \end{abstract}
65: 
66: 
67: 
68: \keywords{large scale galactic shocks: general --- ultraharmonic resonances,
69: interarm features, branches: individual(branches and spurs)}
70: 
71: 
72: \section{Introduction}
73: 
74: 
75: Young stars, H II regions and OB
76: associations delineate the structure that gives its name to spiral galaxies 
77: (Elmegreen 1981; Vall\'ee 2002, 2005). This implies a 
78: correlation between the 
79: spiral structure and
80: the process of star formation. The density wave theory attempts to explain the
81: large-scale structure of these galaxies in terms of a wave propagating in the 
82: disk of stars. Yet this stellar wave alone cannot directly explain the narrow 
83: spiral 
84: arms as delineated by the products of star formation.
85: 
86: Fujimoto (1968) first proposed that the dust lanes
87: observed on the concave side (inside corotation) of spiral arms might be the 
88: result of a Galactic 
89: shock. Large scale shocks propagating in the gaseous layer of the Galaxy could 
90: induce a sudden compression of the interstellar medium (ISM) and trigger the 
91: star formation process. 
92: Since the temporal scales involved in this process are large 
93: (of the order of the Galactic rotation period), an 
94: isothermal 
95: approximation of the ISM is a first step in modeling the gaseous response, 
96: thus rendering a highly 
97: compressible-supersonic medium in which a small amplitude disturbance, such as
98: that provided by an underlying spiral potential, may steepen into a shock 
99: wave. 
100: 
101: In a semi-analytical study of the gas flow in spiral density waves,
102: Roberts (1969) found nonlinear steady-state solutions containing shocks.
103: On the basis that these shocks were coincident with the imposed perturbing 
104: potential minima, he argued that the density jump could trigger star 
105: formation, and in
106: this way narrow bands of young stars could delineate the spiral arms.
107: 
108: Shu  et al. (1973) studied the gas flow in 
109: the context of the spiral density wave by adopting a two phase model for the 
110: ISM. In their study they carried out
111: a slightly nonlinear analysis of such flow and found that, at certain points 
112: in their 
113: solution, the amplitude  became infinite, and they called these 
114: positions ultraharmonic resonances. They argued that  additional 
115: prominent features could appear as a consequence of the ultraharmonic 
116: resonances, for instance: they found a secondary compression associated with
117: the first one.
118: 
119: However, these studies were carried out  assuming steady state flow
120: and tightly wound spirals. By removing the first constraint, Woodward (1975) 
121: showed how the convective steepening of the initial perturbation leads to 
122: large-scale shock formation (as a response to the driving potential) in a 
123: gaseous layer initially in 
124: circular orbit. 
125: He also found the effect of the first ultraharmonic resonance
126: as a secondary peak in the density field, but numerical viscosity in his 
127: calculations 
128: inhibited the formation of secondary shocks. 
129: 
130: In two papers, Contopoulos \&  
131: Grosb\o l (1986, 1988) removed the second constraint and
132: showed that nonlinear effects are not negligible
133:  in open spirals. Using numerical calculations, they 
134: demonstrated
135: that for open spirals the effect of the first 
136: ultraharmonic resonance (which they call the 4:1 resonance) is such that 
137: stellar orbits do not support the spiral perturbation beyond the position of 
138: this resonance 
139: and hence
140: the length  of the stellar spiral arms is limited by this position. In an 
141: analytical study of the effects of the ultraharmonic resonances,
142:  Artymowicz \& Lubow (1992) explained this result as a cumulative effect of 
143: higher-order resonances between  4:1 and the corotation radius. They further 
144: argued that in the vicinity of the former  the gas response is such that  it 
145: looks like a 4-arm structure with a smaller pitch angle  than that of 
146: the original 
147: two-arm pattern. 
148: 
149: It is clear that,  due to the nature of the problem, the highly nonlinear 
150: phenomena associated with the gaseous response to a spiral density wave are 
151: best studied via numerical simulations.
152: Results so obtained
153: add interarm features and spiral substructure on top of  large scale 
154: shocks. Patsis et al. (1997) found that  in open spirals  a 
155: bifurcation 
156: of the main spiral arms  takes place at the 4:1 resonance position. 
157: By analyzing numerical experiments that include self-gravity in the study 
158: of the 
159: effects of the ultraharmonic resonances on gaseous disks,
160: Chakrabarti et al. (2003, from here on CLS03) found secondary 
161: compressions, associated
162: with the first ultraharmonic resonance, in models
163: with a low Toomre parameter, Q, which were evolved only for a few revolutions 
164: of the disk. These compressions
165: eventually became branches.
166:  In high-Q models, evolved for several revolutions 
167: they found the appearance of leading structures which they identified with 
168: spurs. However, spurs and branches not related to resonances have also been 
169: obtained in numerical simulations (Dobbs \& Bonnell 2006; Wada \& Koda
170: 2004). By taking into 
171: account 
172: frozen magnetic fields and self-gravity in 
173: their local arm simulations, 
174: Kim \& Ostriker (2002, 2006) showed that gaseous spurs form as a consequence 
175: of 
176: gravitational 
177: instability inside the spiral arms, a 
178: result 
179: confirmed with global 2D simulations by Shetty \& Ostriker (2006). In those
180: simulations spurs jut out of the spiral arms at regular intervals.
181: 
182: In this paper we carry out two dimensional, global, high resolution 
183: hydrodynamic simulations 
184: to further study the formation of spiral substructure in galactic gaseous 
185: disks.
186: Our approach differs from previous work in that we employ a self-consistent
187: model for the spiral stellar potential (in the orbital sense). This potential 
188: accounts for its own self-gravity and thus is no longer a local arm 
189: approximation to the driving term, making it  more appropriate for 
190: simulations of 
191: open 
192: galaxies.  This paper is organized as follows: in \S 2 we describe the 
193: potential we employ, in \S 3 we present the obtained gaseous response and the 
194: type of 
195: substructure related to it, in \S 4 we discuss the results and compare with 
196: previous work, and in \S 5 we present our conclusions. 
197: 
198: 
199: \section{Modeling}
200: \subsection{The potential}
201: 
202: We employ a potential that consists of two parts. The first one is the 
203: axisymmetric background contribution, taken  from Allen 
204: \& Santill\'an (1991). This model assembles contributions from a bulge and
205: a flattened disk in the way proposed by Miyamoto and Nagai (1975), plus a 
206: massive spherical dark halo. This potential was chosen for its  
207: simplicity and easy mathematical manipulation. It renders a flat rotation
208: curve  of about 200 km s$^{-1}$ at a moderate radius and provides reasonable
209: values for Galactic parameters. 
210: 
211: The nonaxisymmetric contribution is more delicate.  
212: The usual approach
213: to model this term is via a logarithmic spiral potential. The rationale 
214: underlying this 
215: choice is two-fold: first, 
216: from the mathematical point of view it is tractable,
217: and second, no simpler form is known to represent this term. Yet, this choice 
218: has some limitations. It assumes  that only local effects of 
219: the potential can influence the gas dynamics; that is, it considers the 
220: potential minima to be in the geometrical center of the spiral arms. 
221: Such a potential is self-consistent only for small pitch angles 
222:  and does not consider its own 
223: self-gravity. 
224: To remove such
225: constraints and perform more adequate simulations in the presence of an open 
226: spiral, we choose a self-consistent potential 
227: (in the orbital sense) and with it performed our numerical experiments.
228: We employ 
229: a model derived from an imposed stellar spiral mass distribution
230:  given 
231: by Pichardo et al. (2003). 
232: In this paper self-consistency refers to the 
233: reinforcement of the spiral potential by the stellar orbits, i.e., the  orbits 
234: of the stars respond to the spiral potential in such a way as to reform the 
235: spiral locus of the initially imposed potential. 
236: 
237: 
238: In order to obtain the potential given by an imposed open spiral mass 
239: distribution, Pichardo et al. (2003) selected a possible spiral locus for the 
240: morphology of the Galaxy. Then they place a series of oblate spheroids along 
241: this locus to fit the mass distribution of our Galaxy. In this way the total 
242: potential at a given point is found by adding the contributions of every 
243: spheroid at that same point.
244: The characteristics of the spheroids are as follows: the minor
245: axis of the spheroids is perpendicular to the Galactic plane and extends up to 
246: 0.5 kpc. The major semiaxes  have a length of 1 kpc and lie in the Galactic 
247: plane. The locus on which 
248: they are placed fits the K band data of Drimmel (2000) with a pitch angle of 
249: 15.5$^{\circ}$. The central density of the spheroids follows an exponential law
250: with a scale length of 2.5 kpc, approximately that of the near-infrared 
251: Galactic disk (Freudenreich 1998). 
252: The self-consistency study of the potential was carried out following the same 
253: procedure as Contopoulos and Grosb\o l (1986). To obtain the density response 
254: to
255: the spiral perturbation, Pichardo et al.(2003) assume that stars with circular 
256: orbits and rotating in the same sense as the spiral pattern are trapped around 
257: the corresponding central periodic orbit in the presence of the spiral 
258: perturbing term. They compute a series of central periodic orbits and find the 
259: density response along their extension. They match the density response maxima 
260: with the center of the assumed spiral arms 
261: (the imposed spiral locus) and whenever they match, self-consistency can be 
262: claimed.
263: In this way they showed that self-consistency of their models was a 
264: function of two parameters: the angular 
265: pattern speed ($\Omega_p$) and the ratio of 
266: mass contained in the arms to that contained in the disk, M$_A$/M$_D$.
267: The
268: best-fit values of these parameters to achieve self-consistency were 
269: $\Omega_p=$20 km s$^{-1}$ kpc$^{-1}$ and  
270: M$_A$/M$_D$=0.0175, respectively.
271: 
272: 
273: \subsection{Model Parameters}
274: 
275: Our numerical calculations compute the 2D hydrodynamic gaseous response to an 
276: external, fixed, spiral mass distribution in a 3D  galactic disk 
277: model. We performed our simulations in a frame that corotates with
278: the imposed spiral pattern.
279: The gaseous layer was initialized
280: in centrifugal equilibrium with
281: the axisymmetric potential and then the spiral term was turned on. The 
282: nonaxisymmetric term was modeled as a rigidly rotating potential with an 
283: angular
284: speed  ${\bf \Omega_p}$. We investigated spiral substructure formation by 
285: solving the
286: hydrodynamical equations in polar coordinates in a frame rotating with this 
287: prescribed 
288: pattern  speed.
289:  In this frame the hydrodynamical equations are
290: 
291: \begin{equation}
292: \frac{\partial \rho}{\partial t} +  \nabla \cdot (\rho \mbox{\bf{\em v'}}) =0
293: \label{continuity}
294: \end{equation}
295: 
296: 
297: \begin{equation}
298: \rho \left[ \frac{\partial \mbox{{\bf{\em v'}}}}{\partial t} +(\mbox{\bf{\em v'}} \cdot \nabla)\mbox{\bf{\em v'}} \right] 
299: =-\nabla p -\rho(\nabla \Phi_{as}+\nabla \Phi_s +2{\bf\Omega_p}\times \mbox{\bf{\em v'}} 
300: +{\bf \Omega_p}\times({\bf \Omega_p} \times r))
301: \label{euler}
302: \end{equation} 
303: 
304: 
305: 
306: \begin{equation}
307: \rho\left[ \frac{\partial}{\partial t}\left( \frac{e}{\rho} \right) + {\mbox{\bf{{\em v'}}}}
308: \cdot \nabla \left( \frac{e}{\rho} \right) \right] + p\nabla \cdot {\mbox{\bf{\em v'}}}=0
309: \label{energy}
310: \end{equation}
311: 
312: In these equations $\rho$ is the gas density, {\bf {\em v'}} 
313: the velocity in the 
314: rotating frame, $p$ the 
315: pressure, $\Phi_{as}$ is the background axisymmetric potential, $\Phi_s$ the 
316: potential associated with the imposed spiral mass distribution, and $e$ the 
317: gaseous internal energy. 
318: 
319: An idealized ISM was assumed for these simulations in that we adopt an 
320: effectively isothermal
321: equation of state ($\gamma$ = 1.01) for the system. A constant sound speed, 
322: $c_s$, of 7 km s$^{-1}$ was 
323: used. We followed the evolution
324: of the gaseous response for several 
325: rotations; one rotation  is completed after $t_{orb}=2\pi /\Omega_0$ at a 
326: fiducial 
327: radius $R_0$. We present simulations that  evolve for 10 revolutions.
328: 
329: The effect of the first ultraharmonic resonance was studied by changing  
330: $\Omega_p$  so that the location of the resonance lay either inside or outside
331: the maximum radius of the spiral mass distribution (12 kpc). This was 
332: accomplished by selecting
333: $\Omega_p$ values of 20 and 10 km s$^{-1}$ kpc$^{-1}$, respectively. We also 
334:  explored the effects of a stronger spiral potential relative to the background
335:  axisymmetric contribution.
336:  Since the driving term
337: we use is obtained from   an imposed  mass distribution, we may control the 
338: relative force perturbation from the spiral potential by adjusting the mass 
339: contained in the spiral arms.
340: Using a constant pitch angle of 15.5$^{\circ}$ for the spiral locus, force 
341: perturbations of
342: 3\% and 10\% are obtained from mass ratios for the spiral arms to the disk
343:  (M$_A$/M$_D$) of 0.0175, 
344: 0.05,
345: respectively.
346: 
347: We follow the temporal evolution of the gaseous disk by solving the 
348: hydrodynamical equations~(\ref{continuity})-(\ref{energy}) in cylindrical 
349: polar 
350: coordinates (Z-R-$\phi$) with the ZEUS-MP
351: code (Hayes et al. 2006). ZEUS-MP differs from previous implementations of the 
352: ZEUS code in that it is designed for numerical calculations on massively 
353: parallel  computing platforms.
354: ZEUS-MP solves the equations of radiation magnetohydrodynamics in 1D, 2D or 3D
355: in cartesian, cylindrical or spherical  coordinates. The code
356: has been tested against an extensive bank of test problems 
357: documented in Hayes et al. (2006). For our simulations, we neglect  magnetic 
358: fields and radiation transport.
359: 
360: Our model uses
361: a computational mesh covering 2$\pi$ radians in azimuth and extending 
362: from 1 to 15 kpc in radius. Several boundary conditions were tried to ensure 
363: that wave reflections at the boundaries are not affecting our results. Inner 
364: inflow conditions prevent wave reflections from infalling gas and outer open 
365: conditions prevent gas from becoming stagnant. 
366: 2D models were computed in cylindrical polar coordinates by modifying the code
367: so that symmetry was imposed along the leading (Z) coordinate and all 
368: acceleration terms along the Z axis in the momentum equation were zeroed.
369: We present numerical experiments
370: in a computational mesh with 2048$^2$ zones in R and $\phi$; we are  thus 
371: able to resolve the 
372: formation of substructure on  scales $\leq$  10 pc in the radial direction. 
373: The minimum and maximum resolution achieved on the computational grid are
374:  21 pc$^2$ and 293 pc$^2$ respectively. 
375: 
376: \section{RESULTS}
377: 
378: In Figure~\ref{om20} we show snapshots of the density distribution  
379: corresponding
380: to the case  $\Omega_p=20$ km s$^{-1}$ kpc$^{-1}$.  We  observe that at 
381:  $t/t_{orb} =0.1$ the gas responds to the external potential by tracing the 
382: potential minima, clustering around them. At this time  the gas is adjusting 
383: to 
384: the potential and the density perturbation is steadily steepening in response 
385: to 
386: the spiral arms. Slightly
387:  later, at $ t/t_{orb}=0.26$, the recently formed gaseous arms begin to 
388: bifurcate. This bifurcation occurs in the vicinity of the first ultraharmonic 
389: resonance at a radius of 7 kpc, which is marked with a dashed line. Gas 
390: continues to accumulate around 
391:  the potential minima and the newly formed bifurcations until two pairs of 
392: large-scale shock waves are formed. We show the final steady-state 
393: configuration in the third snapshot of Figure~\ref{om20}. Two shock waves  
394:  trace the underlying 
395: spiral potential while the other two shock waves are the result of the 
396: steepening of
397:  the bifurcations. Similar results were presented by Martos et al. (2004) at 
398: lower resolution and with a previous version of the code. 
399: In Figure~\ref{slice} we
400: have plotted a density cut at a radius of 6 kpc, in order to 
401: illustrate the arm to interarm contrast at  evolution times corresponding to 
402: snapshots (b) and (c) of Figure~\ref{om20}. The arm to interarm contrast is a 
403: function of radius and time, but generally speaking we can state that the 
404: shocks associated with the imposed spiral mass distribution have a higher 
405: contrast and are more extended in the radial direction than the shocks 
406: associated with the bifurcations. The compressions, $\rho/\rho_0$, induced by 
407: these shocks 
408: have a peak around 8, in the inner regions, and 3 on average. The new shocks 
409: arising from  bifurcations of the first shocks induce lower compressions with
410: average values around 2. These shocks also have a lower pitch angle and their 
411: radial 
412: extent is limited by the first ultraharmonic resonance position. They can be
413:  found in the range of 6-7.5 kpc.
414: 
415: 
416: \clearpage
417: \begin{figure}
418: \epsscale{.9}
419: \plotone{f1.eps}
420: \caption{Density snapshots for the case $\Omega_p$=20 km s$^{-1}$ kpc$^{-1}$. Surface density at (a) t/t$_{orb}$=0.1, (b) t/t$_{orb}$=0.26 the position of the first ultraharmonic resonance is marked by a dashed line, and (c) t/t$_{orb}$= 8.25. The density scale is shown in units of $\log  \Sigma/\Sigma_0$. The outer boundary for this simulation is located at 15 kpc.}
421: \label{om20}
422: \end{figure}
423: 
424: 
425: \begin{figure}
426: \epsscale{.9}
427: \plotone{f2.eps}
428: \caption{Density cut at 6 kpc. We have plotted the ratio of  
429: density to initial density, to show the arm to interarm contrast 
430:  attained in this simulation. The dotted line represents density
431: contrasts at t/t$_{orb}=0.26$ and the continuous line at 
432: t/t$_{orb}=8.5$.}
433: \label{slice}
434: \end{figure}
435: \clearpage
436: 
437: \subsection{Varying $\Omega_p$}
438: 
439: In order to emphasize the importance of the first ultraharmonic resonance on
440: the bifurcation of the spiral arms, we have performed simulations where the 
441: position of this resonance has been shifted beyond the termination of the 
442: imposed spiral mass distribution. If we maintain fixed all other parameters 
443: and vary $\Omega_p$ from 20 to 10 km s$^{-1}$ kpc$^{-1}$ then the new position
444: of the first ultraharmonic resonance is 14 kpc. In 
445: Figure~\ref{om10} 
446: we show the density distributions for this case. 
447: 
448: \clearpage
449: \begin{figure}
450: \epsscale{.9}
451: \plotone{f3.eps}
452: \caption{Density distribution snapshots for the case $\Omega_p$=10 km s$^{-1}$ kpc$^{-1}$. Inner boundary is at 1 kpc and outer radius at 22 kpc. The position of the first ultraharmonic resonance is 14 kpc. Surface density at (a) t/t$_{orb}$=0.05, (b) t/t$_{orb}$=1.03, and (c) t/t$_{orb}$= 10. The density scale is shown in units of $\log  \Sigma/\Sigma_0$.}
453: \label{om10}
454: \end{figure}
455: \clearpage
456: 
457: 
458: The evolution of the system is similar to before:
459: the induced perturbations in the gas steepen into large-scale shock waves and 
460: accumulate
461: around the potential minima. 
462: In snapshot (b) we observe transient substructure emerging as waves propagate 
463: radially. After a few revolutions of the potential, this substructure is 
464: smoothed. In snapshot (c) only a two-arm gaseous response remains and the 
465: density peaks have been shifted downstream relative to the previous case. We 
466: also observe waves propagating from the tips of the arms radially outwards. 
467: These waves have been reflected in the corotation region and amplified, yet 
468: they do not add to the large-scale shocks and remain in an intermediate 
469: region between the end of the imposed potential and the corotation radius.
470: 
471: 
472: \subsection{Higher Forcing}
473: 
474: A higher forcing case corresponds to either a more massive or a more open 
475: spiral 
476: driving term. Since the potential model employed is derived from an imposed
477: spiral mass distribution, we increased the mass  in the spiral distribution to 
478: achieve a higher forcing term. In the case described here we adopted a mass
479:  ratio of 
480: the arms to the disk of M$_A$/M$_D$=0.05. With this parameter 
481: we have an average relative force perturbation of $~$ 10$\%$. The gaseous 
482: response presented in Figure~\ref{highforcing} is such that, initially, 
483: density enhancements around the potential 
484: minima are obtained. At t/t$_{orb}=0.26$ (snapshot a) branches begin to emerge 
485: near the 
486: first
487:  ultraharmonic 
488: position (at 7 kpc) and eventually interact with the already formed gaseous 
489: arms. Waves 
490: reflect from the inner Lindblad resonance (at 3 kpc) and the corotation radius 
491: (located at 11 kpc) and interact
492: with the existing density enhancements as they steepen. In this way, we 
493: obtain a temporary ragged appearance, illustrated in snapshot (b) at 
494: t/t$_{orb}$=1. The system, however, evolves to a quasi-steady state, and
495: this rich substructure is eventually sheared away. In this simulation the 
496: shock 
497: structure does not remain fixed; the shocks oscillate and so does the spiral 
498: substructure. We followed this simulation  to $t/t_{orb}$=10, and 
499: both oscillations and overall shock structure were preserved. In  
500: this sense  we say that system achieved a quasi-steady state. 
501: 
502:  An  interesting outcome of this 
503: simulation is the appearance after a few revolutions of substructure in the 
504: internal regions of the computational mesh. Short streams extend out from the 
505: main gaseous arms; these streams are small compared with the bifurcations and 
506: have a greater pitch angle. Because these characteristics match those given by 
507: Elmegreen (1981) to classify spurs, so we will refer to  them as such.
508: In 
509: snapshot (c) we present the final configuration of the system which displays
510: the previously noted  4-arm shock structure plus spurs at an evolution time 
511: of t/t$_{orb}$=10. The spurs are marked with draw-in arrows. These spurs 
512: protrude from the main arms into the already steepened branches, and as can be 
513: seen in this snapshot, they wind in an opposite sense to the general gas 
514: rotation. The spurs are symmetric with respect to the main arms and originate
515:  close to inner Lindblad resonance.
516: 
517: \clearpage
518: \begin{figure}
519: \epsscale{.9}
520: \plotone{f4.eps}
521: \caption{Density contours snapshots for the case M$_A$/M$_D$=0.05. Surface density at (a) t/t$_{orb}$=0.26, (b) t/t$_{orb}$=1.55, and (c)
522: t/t$_{orb}$= 10. Spurs are delimited with solid lines, protruding from spiral arms. The density scale is shown in
523: units of $\log  \Sigma/\Sigma_0$.}
524: \label{highforcing}
525: \end{figure}
526: \clearpage
527: 
528: \section{Discussion}
529: 
530: Analytical work on the gaseous disk response to an external driving spiral 
531: potential
532: by Shu et al. (1973) and Artymowicz \& Lubow (1992) highlighted
533: the importance of the first ultraharmonic resonance as the main generator of
534: spiral substructure. Those  studies concluded that in a slightly nonlinear 
535: regime a second compression emerged in a gaseous disk that could account for
536: observed spiral substructure. But a more complete study is necessary to 
537: identify 
538:  the effect of resonances in a fully nonlinear regime. Due to the 
539: nature of the equations governing the evolution of such a disk, this task is 
540: best accomplished using numerical simulations.  
541: 
542: Numerical work on this subject by Patsis et al. (1997) and
543: CLS03
544: revealed that substructure arising in gaseous disks can be related to 
545: ultraharmonic 
546: resonances.
547: 
548: The gaseous response we have presented resembles that of previous numerical 
549: work: the initial response tends to associate with the underlying potential,
550: and once gaseous arms have formed, branches begin to protrude from them.
551:  But in our simulations these branches 
552: continue to steepen,  leading to the formation of new shock waves, with 
553: lower radial extension, pitch angle and Mach number, but large-scale shock 
554: waves 
555: nevertheless. The role of the first ultraharmonic resonance is such that,
556: if absent from the region of influence of the spiral term, no such 
557: additional shocks are formed.
558: 
559: We  also 
560: performed simulations with high forcing and  obtained additional 
561: substructure:  gaseous spurs. These spurs protrude from the arms after
562: several revolutions and are not a transient feature.
563:  We  wish to note  that these results were 
564: obtained without 
565: self-gravity in the gaseous disk and that spurs and bifurcations are obtained 
566: in the same long-timescale simulation.
567:  Such spurs  wind in the opposite sense  to the arms 
568: rotation  and are restricted to an area close to the inner Lindblad resonance.
569: CLS03 also obtained gaseous spurs employing a nonlinear perturbing potential 
570: with high forcing in a disk very stable against 
571: axisymmetric
572:  perturbations, but bifurcations did not appeared in those simulations.
573:  In our case we use a spiral term that ensures self-consistency and
574: employ the maximum value for forcing (5\%, a similar value as that used by 
575: CLS03 to obtain spurs) that still renders a self-consistent
576: driving term according to Pichardo et al. (2003). Our driving term and our 
577: algorithms to calculate the gaseous response have proven robust enough as to 
578: allow the development of branches and spurs in the same run. The high numerical
579: resolution and consequent low numerical dissipation of our simulation may also
580: have been important in obtaining this result.
581: CLS03 include self-gravity in 
582: the gaseous layer and the spurs they obtain are more pronounced. Although we 
583: demonstrated that these features can emerge in the 
584: absence of self-gravity in the gaseous layer, we are well aware that
585: self-gravity in the gas 
586:  will enhance the  features we have found giving rise to higher arm-interam
587: density contrasts.
588: 
589: 
590: It is worth remarking that the simulations presented here can be 
591: evolved for several revolutions of the driving term (more than t/t$_{orb}=10$)
592: and eventually achieve a steady state. Yet the isothermal condition we assumed
593: can only reproduce the formation of substructure, but not its subsequent 
594: fragmentation. To reproduce the latter
595:  we need to adopt a more realistic treatment of the
596: ISM, with special attention to its thermodynamic properties. Another line
597: of improvement is suggested by previous work  showing that the inclusion 
598: of frozen magnetic fields  leads to the formation of additional substructure. 
599: We
600: defer a discussion of the role of magnetic fields to a future paper.
601: 
602: 
603: 
604: \section{Conclusions}
605: 
606: A fully nonlinear treatment of the propagation of spiral density waves in a 
607: gaseous disk is tractable using  numerical simulations.  
608:  By considering an open spiral, in the absence of self-gravity in the gaseous 
609: layer, we are able to 
610: obtain rich substructure associated with an external spiral potential. Our work
611: differs from other published results in that we employ a self-consistent
612: driving term that considers its own gravity and removes the local arm 
613: approximation. We have
614: presented experiments showing the formation of a four-arm structure in 
615: response to a two-arm driving pattern. Initially the gas accumulates in the 
616: potential minima. After some time the gas responds to resonances and the main
617: arms bifurcate. These features have a considerable azimuthal extension and 
618: continue to 
619: steepen with  time. Eventually a four-arm shock structure  emerges, in the 
620: gaseous layer that 
621: coexist with 
622: the underlying two-arm potential, made up by the stars. This result appears to 
623:   agree with
624: observational data published by Drimmel (2000) where he concludes that, using
625: optical tracers, the spiral structure in our Galaxy is best fitted by a 
626: four-arm 
627:  structure, 
628: while in the infrared a two-arm structure dominates.
629: 
630: If we place the first ultraharmonic resonance outside the region of influence 
631: of the spiral (in our case by changing the value of $\Omega_p$) the main arms 
632: do not bifurcate and large-scale shocks are the only induced response in the 
633: gaseous layer. In this way we show that if the first ultraharmonic resonance is
634:  placed outside the region of influence of the perturbing term, no spiral 
635: substructure appears, thus emphasizing the connection between this resonance
636: and the bifurcations of the gaseous arms.
637: 
638: If we combine the effects of the resonances with a large 
639: forcing amplitude we obtain, on top of this four-arm structure, 
640: spurs protruding from the main arms in a region between the inner Lindblad 
641: resonance and the first ultraharmonic resonance. Overlapping of nonlinear 
642: effects take place in our long-timescale global simulations, showing that 
643: an orderly spiral density wave potential can produce a gaseous
644: response that is strongly disordered.
645: 
646: The inclusion of additional processes such as magnetic fields, self-gravity 
647: for 
648: the gaseous layer, and
649: thermal processes in the ISM may lead to the appearance of additional 
650: substructure and 
651: their 
652: consequent fragmentation into bound condensations. Numerical experiments 
653: addressing these topics are needed in order to improve our understanding of 
654: the 
655: global ISM in galaxies and its relation to large-scale processes.
656: 
657: 
658: 
659: \acknowledgments
660: M. Y\'a\~nez and M. Martos want to thank the UCMEXUS-CONACYT fellowship 
661: program for
662: economic support. M. Martos thanks the CASS-UCSD for their hospitality during 
663: a sabbatical visit which helped pursue this project.
664: The numerical simulations were carried out on the DataStar
665: system at the San Diego Supercomputer Center with LRAC allocation MCA98N020.
666: 
667: \begin{thebibliography}{}
668: 
669: 
670: \bibitem[]{AS} Allen, C. and Santill\'an, A. 1991, \rmxaa, 22, 255
671: 
672: \bibitem[]{AL} Artymowicz, P. and Lubow, S. H. 1992, \apj, 389, 129
673: 
674: \bibitem[]{CLS} Chakrabarti, S., Laughlin, G. and Shu, F. H. 2003, \apj, 596, 220
675: 
676: \bibitem[]{cont86} Contopoulos, G. and Grosb\o l, P. 1986, \aap, 155, 11
677: 
678: \bibitem[]{contopoulos} Contopoulos, G. and Grosb\o l, P. 1988, \aap, 197, 83
679: 
680: \bibitem[] {dobss} Dobbs, C. L. and Bonnell, I. A. 2006, \mnras, 367, 873
681: 
682: \bibitem[]{Drimmel} Drimmel, R. 2000, \aap, 117, 89
683: 
684: \bibitem[]{el81} Elmegreen, D. M. 1981, \apjs, 47, 229
685: 
686: \bibitem[]{Freu98} Freudenreich, H. T. 1998, \apj, 492, 495
687: 
688: \bibitem[Fujimoto(1968)]{fuj68} Fujimoto, M.  1968, in \emph{IAU Symposium} No.
689:  29, p. 453
690: 
691: \bibitem[]{zeusmp} Hayes, J. C., Norman, M. L., Fiedler, R. A., Bordner, J. O., Li, 
692: P. S., Clark, S. E., ud-Doula, A. and Mac Low, M. 2006, \apjs, 165, 188
693: 
694: %\bibitem[]{kalkerp} Kalberla, P. M. W. and Kerp, J. 1998, \aap, 339, 745
695: 
696: \bibitem[]{ko02} Kim, W. T. and Ostriker, E. 2002, \apj, 570, 132
697: 
698: \bibitem[]{ko06} Kim, W. T. and Ostriker, E. 2006, \apj, 646, 213
699: 
700: \bibitem[]{omp20} Martos, M., Hernandez, X., Y\'a\~nez, M., Moreno, E. and Pichardo, B. 2004, \mnras, 350, L47
701: 
702: \bibitem[]{mn75} Miyamoto, M. and Nagai, R. 1975, \pasj, 27, 533
703: 
704: \bibitem[]{pats97} Patsis, P. A., Grosb\o l, P. and Hiotelis, N. 1997, \aap, 323, 762
705: 
706: \bibitem[]{pichardoetal} Pichardo, B., Martos, M., Moreno, E. and Espresate, J. 2003, \apj,
707: 582, 230
708: 
709: \bibitem[]{roberts} Roberts, W. W. 1969, \apj, 158, 123
710: 
711: \bibitem[]{so06} Shetty, R. and Ostriker, E. 2006, \apj, 647, 997
712: 
713: \bibitem[SMR]{SMR} Shu, F. H., Milione, V. and Roberts, W. W. 1973, \apj, 183, 
714: 819
715: 
716: \bibitem[]{Val02} Vall\'ee, J. P. 2002, \apj, 566, 261
717: 
718: \bibitem[]{val05} Vall\'ee, J. P. 2005, \apj, 130, 569
719: 
720: \bibitem[]{wada} Wada, K. and Koda, J. 2004, \mnras, 349, 270
721: 
722: \bibitem[]{woodward} Woodward, P. 1975, \apj, 195, 61
723: 
724: 
725: \end{thebibliography}
726: 
727: \clearpage
728: 
729: 
730: 
731: \end{document}
732: 
733: %%
734: %% End of file `sample.tex'.
735: