arXiv subjects
Publications and source records attributed to J. A. Sellwood.
The paper claimed that significant radial migration of stars in a stellar disk like that of the Milky Way could not occur. We now think that while the treatment of the effects of molecular clouds was correct, the paper seriously underestimated the ability of spiral arms to shift the radii of stars that corotate with them. Consequently, it is likely that significant radial migration_is_ possible.
The central densities of dark matter (DM) halos are much lower than predicted in cold DM models of structure formation. Confirmation that they have cores with a finite central density would allow us to rule out many popular types of collisionless particle as candidates for DM. Any model that leads to cusped halos (such as cold DM) is already facing serious difficulties on small scales and hot DM models have been excluded. Here I show that fermionic warm DM is inconsistent with the wide range of phase space densities in the DM halos of well-observed nearby galaxies.
We show that bars in galaxy models having halos of moderate density and a variety of velocity distributions all experience a strong drag from dynamical friction unless the halo has large angular momentum in the same sense as the disk. The frictional drag decreases the bar pattern speed, driving the corotation point out to distances well in excess of those estimated in barred galaxies. The halo angular momentum required to avoid strong braking is unrealistically large, even when rotation is confined to the inner halo only. We conclude, therefore, that bars are able to maintain their observed high pattern speeds only if the halo has a central density low enough for the disk to provide most of the central attraction in the inner galaxy. We present evidence that this conclusion holds for all bright galaxies.
Having once thought that we understood how some galaxies were barred and had difficulty accounting for the absence of bars in others, it now seems that we have the opposite problem. Most real galaxies have centres dense enough to inhibit bar formation, even if they have massive discs. This is also true for barred galaxies, which therefore could not have acquired their bars as a result of a self-excited, global instability. There is a hint from the morphology of galaxies in the Hubble Deep Fields that the growth of bars might be a slow or secular process. Here I discuss possible mechanisms that could form bars long after the disc is assembled.
A long-standing controversy in studies of spiral structure has concerned the lifetimes of individual spiral patterns. Much theoretical work has sought quasi-stationary spiral modes while N-body simulations have consistently displayed recurrent, short-lived patterns. The simulations manifest a recurrent cycle of true instabilities related to small-scale features in the angular momentum distribution of particles, with the decay of each instability seeding the growth of the next. Data from the recent Hipparcos mission seem to offer support for the recurrent transient picture.
Longitude-velocity (l-V) diagrams of H I and CO gas in the inner Milky Way have long been known to be inconsistent with circular motion in an axisymmetric potential. Several lines of evidence suggest that the Galaxy is barred, and gas flow in a barred potential could be consistent with the observed ``forbidden'' velocities and other features in the data. We compare the H I observations to l-V diagrams synthesized from 2-D fluid dynamical simulations of gas flows in a family of barred potentials. The gas flow pattern is very sensitive to the parameters of the assumed potential, which allows us to discriminate among models. We present a model that reproduces the outer contour of the H I l-V diagram reasonably well; this model has a strong bar with a semimajor axis of 3.6 kpc, an axis ratio of approximately 3:1, an inner Lindblad resonance (ILR), and a pattern speed of 42 km/s/kpc, and matches the data best when viewed from 34°to the bar major axis. The behavior of the models, combined with the constraint that the shocks in the Milky Way bar should resemble those in external barred galaxies, leads us to conclude that wide ranges of parameter space are incompatible with the observations. In particular we suggest that the bar must be fairly strong, must have an ILR, and cannot be too end-on, with the bar major axis at 35°+/- 5°to the line of sight. The H I data exhibit larger forbidden velocities over a wider longitude range than are seen in molecular gas; this important difference is the reason our favored model differs so significantly from other recently proposed models.
A model that accounts for the brief flaring of QSOs in the early stages of galaxy formation is proposed. I argue that a bar must develop early in the life of nearly every galaxy and that gas to create and fuel the QSO is driven into the center of the galaxy by the bar. The QSO lifetime, and the mass of its central engine, are also controlled by large-scale dynamics, since the fuel supply is shut off after a short period by the development of an inner Lindblad resonance. This resonance causes the gas inflow along the bar to stall at a distance of a few hundred parsecs from the center. The ILR develops as a result of previous inflow, making quasar activity self-limiting. The bars are weakened and can be destroyed by the central mass concentration formed in this way.
While the total interior mass of a galaxy is reasonably well determined by a good rotation curve, the relative contributions from disk, bulge and halo are only weakly constrained by one-dimensional data. Barred galaxies are intrinsically more complicated, but provide much tighter constraints on the disk masses and support the idea that most of the mass in the inner parts of bright galaxies is in their stars. There appears to be no systematic difference in dark matter content between barred and unbarred galaxies, consistent with the theoretical result that the global stability of galaxies with dense centers does not depend on their halo fraction. The rotation curve shapes of lower luminosity and low-surface-brightness galaxies, on the other hand, indicate significant mass in the DM halo even near their centers. We find that most DM halos appear to have large cores, inconsistent with the predictions from cosmological simulations. We also show that such large-core halos can result from compression by disk infall of physically reasonable initial halos. Maximum disks, while apparently required by the data, do seem to present some puzzles; most notably they re-open the old disk-halo ``conspiracy'' issue and incorrectly predict that surface brightness should be a second parameter in the Tully-Fisher relation.
In this review, I discuss just three aspects of the stability and evolution of galactic discs. (1) I first review our understanding of the bar instability and how it can be controlled. Disc galaxies in which the orbital speed does not decrease much towards the centre have no difficulty avoiding bars, even when dark matter makes an insignificant contribution to the inner part of the rotation curve. (2) I then briefly discuss interactions between disturbances in the discs of galaxies and the spherical components, which generally exert a damping effect through dynamical friction. The fact that bars in real galaxies appear to rotate quite rapidly, seems to require dark matter halos to have large, low-density cores. (3) In the remainder of the article, I consider the theory of spiral structure. The new development here is that the distribution function for stars in the Solar neighbourhood, as measured by HIPPARCOS, is far less smooth than most theoretical work had previously supposed. The strong variations in the values of the \DF over small ranges in angular momentum have the appearance of having been caused by scattering at Lindblad resonances with spiral patterns. This result, if confirmed when the radial velocity data become available, supports the picture of spiral patterns as dynamical instabilities driven by substructure in the \DF. The details of how decaying patterns might seed conditions for a new instability remain unclear, and deserve fresh attention.
A galaxy disk embedded in a rotating halo experiences a dynamical friction force which causes it to warp when the angular momentum axes of the disk and halo are misaligned. Our fully self-consistent simulations of this process induce long-lived warps in the disk which mimic Briggs's rules of warp behavior. They also demonstrate that random motion within the disk adds significantly to its stiffness. Moreover, warps generated in this way have no winding problem and are more pronounced in the extended \h1 disk. As emphasized by Binney and his co-workers, angular momentum misalignments, which are expected in hierarchical models of galaxy formation, can account for the high fraction of warped galaxies. Our simulations exemplify the role of misaligned spins in warp formation even when the halo density is not significantly flattened.
We propose that massive central objects form in the centers of the bars which must develop in young high-surface density galactic disks. Large-scale dynamics shuts off the growth of the central mass before it reaches $\sim 2%$ of the disk mass at the time, but this mass is sufficient to weaken the bar substantially. Subsequent evolution of the galaxy can either complete the destruction of the bar or cause it to recover, depending upon the angular momentum distribution of later infalling material. We produce massive, fully self-gravitating disks having roughly flat rotation curves which are quite stable. If at least part of the central masses we require constitute the engines of QSOs, then our picture naturally accounts for their redshift dependence, since the fuel supply is shut off by the development of an inner Lindblad resonance. A prediction is that massive objects should not be found in halo dominated galaxies, such as low-luminosity, or low-surface brightness galaxies.
When present, extended disks of neutral hydrogen around spiral galaxies show a remarkably uniform velocity dispersion of approx 6 km/s. Since stellar winds and supernovae are largely absent in such regions, neither the magnitude nor the constancy of this number can be accounted for in the classical picture in which interstellar turbulence is driven by stellar energy sources. Here we suggest that magnetic fields with strengths of a few microgauss in these extended disks allow energy to be extracted from galactic differential rotation through MHD driven turbulence. The magnitude and constancy of the observed velocity dispersion may be understood if its value is Alfvenic. Moreover, by providing a simple explanation for a lower bound to the gaseous velocity fluctuations, MHD processes may account for the sharp outer edge to star formation in galaxy disks.
We study the vertical heating and thickening of galaxy disks due to accretion of small satellites. Our simulations are restricted to axial symmetry, which largely eliminates numerical evolution of the target galaxy but requires the trajectory of the satellite to be along the symmetry axis of the target. We find that direct heating of disk stars by the satellite is not important because the satellite's gravitational perturbation has little power at frequencies resonant with the vertical stellar orbits. The satellite does little damage to the disk until its decaying orbit resonantly excites large-scale disk bending waves. Bending waves can damp through dynamical friction from the halo or internal wave-particle resonances; we find that wave-particle resonances dominate the damping. The principal vertical heating mechanism is therefore dissipation of bending waves at resonances with stellar orbits in the disk. Energy can thus be deposited some distance from the point of impact of the satellite. The net heating from a tightly bound satellite can be substantial, but satellites that are tidally disrupted before they are able to excite bending waves do not thicken the disk.
We use fully self-consistent N-body simulations of barred galaxies to show that dynamical friction from a dense dark matter halo dramatically slows the rotation rate of bars. Our result supports previous theoretical predictions for a bar rotating within a massive halo. On the other hand, low density halos, such as those required for maximum disks, allow the bar to continue to rotate at a high rate. There is somewhat meager observational evidence indicating that bars in real galaxies do rotate rapidly and we use our result to argue that dark matter halos must have a low central density in all high surface brightness disk galaxies, including the Milky Way. Bars in galaxies that have larger fractions of dark matter should rotate slowly, and we suggest that a promising place to look for such candidate objects is among galaxies of intermediate surface brightness.
We have examined the stability of a sequence of oblate elliptical galaxy models having the Stackel form suggested by Kuz'min & Kutuzov. We have employed the 2-integral DFs given by Dejonghe & de Zeeuw for which flattened non-rotating models are characterized by counter-streaming motion and are radially cool; we introduce net rotation in some models by changing the sign of the z-component of angular momentum for a fraction of the particles. We have found that all non-rotating and slowly rotating members of this sequence rounder than E7 are stable, and that even maximally rotating models rounder than E4 are stable. In the absence of strong rotation, the most disruptive instability, and the last to be stabilized by increasing thickness, is a lopsided (m=1) mode. This instability appears to be driven by counter-rotation in radially cool models. Its vigour is lessened as rotation is increased, but it remains strong even in models with net angular momentum 90% of that of a maximally rotating model before finally disappearing in maximally rotating models. Strongly rotating models are more unstable to bar-forming modes which afflict maximally rotating models with c/a >~ 0.5, but this mode is quickly stabilized by moderate fractions of counter-rotating particles. Bending instabilities appear not to be very important; they are detectable in the inner parts of the flatter models, but are less vigorous and more easily stabilized than the lop-sided or bar modes in every case. We briefly discuss the possible relevance of the lop-sided instability to the existence of many lop-sided disk galaxies.
We describe fully self-consistent N-body experiments of barred galaxies with massive halos. A rotating bar is braked through dynamical friction with the halo, which occurs on a short time scale when the central density of the halo is high. On the other hand, friction is weak in a model with a central halo density low enough that the disk dominates the rotation curve in the inner parts; this model supports a fast bar for a Hubble time. We conclude that real barred galaxies in which the bar rotates rapidly, which is believed to be the rule, must have close to maximum disks.
A strong bar rotating within a massive halo should lose angular momentum to the halo through dynamical friction, as predicted by Weinberg. We have conducted fully self-consistent, numerical simulations of barred galaxy models with a live halo population and find that bars are indeed braked very rapidly. Specifically, we find that the bar slows sufficiently within a few rotation periods that the distance from the centre to co-rotation is more than twice the semi-major axis of the bar. Observational evidence (meagre) for bar pattern speeds seems to suggest that this ratio typically lies between 1.2 to 1.5 in real galaxies. We consider, a number of possible explanations for this discrepancy between theoretical prediction and observation, and conclude that no conventional alternative seems able to account for it.
The stability of a galaxy model is most easily assessed through N-body simulation. Particle-mesh codes have been widely used for this purpose, since they enable the largest numbers of particles to be employed. We show that the functional expansion technique, originally proposed by Clutton-Brock for other simulation problems, is in fact superior for stability work. For simulations of linear evolution it is not much slower than grid methods using the same number of particles, and reproduces analytical results with much greater accuracy. This success rests on its ability to represent global modes with a modest number of basis functions; grid methods may be more effective for other applications, however. Our conclusions are based on implementations of functional expansion and grid algorithms for disk galaxies.