We use an N-body simulation, constructed using GADGET-2, to investigate an accretion flow onto an astrophysical disk that is in the opposite sense to the disk's rotation. In order to separate dynamics intrinsic to the counter-rotating flow from the impact of the flow onto the disk, we consider an initial condition in which the counter-rotating flow is in an annular region immediately exterior the main portion of the astrophysical disk. Such counter-rotating flows are seen in systems such as NGC 4826 (known as the "Evil Eye Galaxy"). Interaction between the rotating and counter-rotating components is due to two-stream instability in the boundary region. A multi-armed spiral density wave is excited in the astrophysical disk and a density distribution with high azimuthal mode number is excited in the counter-rotating flow. Density fluctuations in the counter-rotating flow aggregate into larger clumps and some of the material in the counter-rotating flow is scattered to large radii. Accretion flow processes such as this are increasingly seen to be of importance in the evolution of multi-component galactic disks.

The Rossby wave instability (RWI) in non-self-gravitating discs can be triggered by a bump at a radius $r_0$ in the disc surface mass-density (which is proportional to the inverse potential vorticity). It gives rise to a growing non-axisymmetric perturbation [$\propto \exp(im\phi)$, $m=1,2..$] in the vicinity of $r_0$ consisting of anticyclonic vortices which may facilitate planetesimal growth in protoplanetary discs. Here, we analyze a continuum of thin disc models ranging from self-gravitating to non-selfgravitating. The key quantities determining the stability/instability are: (1) the parameters of the bump (or depression) in the disc surface density, (2) the Toomre $Q$ parameter of the disc (a non-self-gravitating disc has $Q\gg1$), and (3) the dimensionless azimuthal wavenumber of the perturbation $\bar{k}_\phi =mQh/r_0$, where $h$ is the half-thickness of the disc. For discs stable to axisymmetric perturbations ($Q>1$), the self-gravity has a significant role for $\bar{k}_\phi < \pi/2$ or $m<(\pi/2) (r_0/h)Q^{-1}$; instability may occur for a depression or groove in the surface density if $Q\lesssim 2$. For $\bar{k}_\phi > \pi/2$ the self-gravity is not important, and instability may occur at a bump in the surface density. Thus, for all mode numbers $m \ge 1$, the self-gravity is unimportant for $Q > (\pi/2)(r_0/h)$. We suggest that the self-gravity be included in simulations for cases where $Q< (r_0/h)$.