The dynamical evolution of dense stellar systems is simulated using a two-dimensional Fokker-Planck method, with the goal of providing a model for the formation of supermassive stars which could serve as seed objects for the supermassive black holes of quasars. This work follows and expands on earlier 1-D studies of spherical clusters of main-sequence stars. The 2-D approach allows for the study of rotating systems, as would be expected due to cosmological tidal torquing; other physical effects included are collisional mergers of stars and a bulk stellar bar perturbation in the gravitational potential. The 3 Myr main-sequence lifetime for large stars provides an upper limit on simulation times. Two general classes of initial systems are studied: Plummer spheres, which represent stellar clusters, and \gamma=0 spheres, which model galactic spheroids. At the initial densities of the modeled systems, mass segregation and runaway stellar collisions alone are insufficient to induce core collapse within the lifetime limit if no bar perturbation is included. However, core collapse is not a requirement for the formation of a massive object: the choice of stellar initial mass function is found to play a crucial role. When using an IMF similar to that observed for dense stellar clusters the simulations show that the stellar system forms massive (250M_\odot) objects by collisional mergers; in almost all such cases the presence of a stellar bar allows for sufficient additional outward transport of angular momentum that a core-collapse state is reached with corresponding further increase in the rate of formation of massive objects. In contrast, simulations using an IMF similar to that observed for field stars in general (which is weighted more towards lower masses) produce no massive objects, and reach core collapse only for initial models which represent the highest-density galactic spheriods.

We have developed a two-dimensional orbit averaged Fokker-Planck model of stellar clusters which expands on spherically symmetric one-dimensional models to include rotation and ellipticity. Physical effects such as collisions, finite stellar lifetimes and bar formation (i.e., a non-axisymmetric component of the potential) can also be included. The first use of the model is to study the evolution of dense clusters (rho(r=0)=~10^7 M_sun/pc^3) that may be expected to have existed at the centres of newly-forming galaxies, with the goal of verifying that angular momentum can be removed from the core of the cluster quickly enough so that rotation no longer prevents the formation of a massive (~10^2 M_sun) object. This could act as the seed black hole for the formation of an AGN.