Both observationally and theoretically the bulk of the radio emission from Sgr A^* appears to come from close to the event horizon. At small radii the assumption of spherical symmetry is probably invalid. Therefore, we would like to find self-consistent solutions to the equations of motion which include differentially rotating gas flows. We discuss two such models here: the Keplerian flow dynamo model and the sub-Eddington two-temperature accretion model. Other commonly used models, such as the well-known thin-disk model (Shakura and Sunyaev 1973) probably do not apply to Sgr A^* since they predict substantial infrared emission which is not seen in the Galactic Center. In fact, combined with the presence of the stellar winds, it seems unlikely that any true large-scale disk exists around Sgr A^* (Coker et al 1999). However, the sub-Eddington two-temperature accretion model discussed results in a disklike accretion flow; it has been argued that this flow incorporates the stellar winds without either emitting significant infrared or being destroyed (Narayan et al 1998).

This apparent absence of a disk and any associated jet sets Sgr A^* apart from many other black hole systems such as luminous AGNs and XRBs. The stars that feed Sgr A^* are fairly uniformly distributed around the black hole so that it is probably accreting relatively little angular momentum. The direction of the accreting angular momentum vector is likely to be time variable as well. Thus, the complete picture probably requires a combination of large-scale spherical accretion with a small-scale (and as yet unobserved) disk and/or jet.

A rough example of what the large-scale flow might look like near Sgr A^*. The 10 wind sources in this hydrodynamical model (see Coker and Melia 1997 for details) produce large-scale shock fronts and cavities with time-dependent characteristics. In reality there are at least two dozen (Genzel et al 1996) stellar sources and some of them may be <1 R_A from Sgr A^*. Also, the stars move relative to one another while in this simulation they are stationary. Thus, the flow is likely to be even more highly non-spherical than is shown in the figure. In fact, if the sources are rotating as a cluster (Genzel et al 2000), the flow may have sufficient angular momentum to circularize at a radius as large as 10⁴r_s.