Continuous flows driving branching processes and their nonlinear evolution equations

Abstract We consider on M(ℝd) (the set of all finite measures on ℝd) the evolution equation associated with the nonlinear operator F↦ΔF′+∑k⩾1bkFk F \mapsto \Delta F' + \sum\nolimits_{k \geqslant 1} b_k F^k , where F′ is the variational derivative of F and we show that it has a solution represented by means of the distribution of the d-dimensional Brownian motion and the non-local branching process on the finite configurations of M(ℝd), induced by the sequence (bk)k⩾1 of positive numbers such that ∑k⩾1bk⩽1 \sum\nolimits_{k \geqslant 1} b_k \leqslant 1 . It turns out that the representation also holds with the same branching process for the solution to the equation obtained replacing the Laplace operator by the generator of a Markov process on ℝd instead of the d-dimensional Brownian motion; more general, we can take the generator of a right Markov process on a Lusin topological space. We first investigate continuous flows driving branching processes. We show that if the branching mechanism of a superprocess is independent of the spatial variable, then the superprocess is obtained by introducing the branching in the time evolution of the right continuous flow on measures, canonically induced by a right continuous flow as spatial motion. A corresponding result holds for non-local branching processes on the set of all finite configurations of the state space of the spatial motion.