Super Brownian motion with interactions. (Q2574599)

Let us denote by \(\mathcal M\) the space of all finite measures on \(\mathbb R^ {d}\) equipped with the topology of weak convergence and suppose that \(\theta : \mathcal M\to \mathbb R_ +\), \(b:\mathcal M \times \mathbb R^ {d}\to \mathbb R^ {d}\) and \(\sigma :\mathcal M\times \mathbb R^ {d}\to \mathbb R^ {d\times d}\) are bounded continuous functions, \(\theta \) and \(\sigma \) positive. Let \(A(\nu )\) denote the infinitesimal generator of the diffusion with drift \(b(\nu ,\cdot )\) and diffusion coefficient \(\sigma (\nu ,\cdot )\). Fix \(\mu \in \mathcal M\). It is proved that there exists an \(\mathcal M\)-valued process \(Y\) with continuous sample paths which solves the following martingale problem: \(Y_ 0 = \mu \), \((Y_ {t},\varphi ) = (Y_ 0,\varphi ) + \int ^ {t}_ 0 (Y_ {s},\theta (Y_ {s})A(Y_ {s})\varphi )\, ds + M(\varphi ) _ {t}\) for all \(\varphi \in C^ 2_ {b}(\mathbb R^ {d})\), \(M(\varphi )\) being a continuous martingale with quadratic variation \(\langle M (\varphi )\rangle _ {t} = 4\int ^ {t}_ 0 (Y_ {s},\theta (Y_ {s}) \varphi ^ 2)\, ds\). Uniqueness of solutions remains an open problem.