A stochastic equation based on a Poisson system for a class of measure- valued diffusion processes (Q1174386)

Two classes of atomic measure-valued processes are constructed. The first is a measure-valued branching process with a spatial motion given by a bounded operator (which results in an atomic measure-valued process) and a nonlinear population dependent immigration term. It is the inclusion of this nonlinear term which is the important contribution of this paper. To handle this the author develops a construction of measure-valued branching diffusions with immigration using a Poisson point process of excursions of continuous branching diffusion. In order to incorporate the nonlinear term he introduces a stochastic equation based on this point process. Existence under fairly general conditions and uniqueness under a Lipschitz-type condition are established. In addition there is an extension to a spatially homogeneous setting involving a countable family of independent Poisson point processes of the above type. A second interesting class of measure-valued processes is obtained from by a transformation of the measure-valued branching with state dependent immigration using a time change and normalization. This produces a class of probability measure-valued processes of the Fleming-Viot type.