Martingale solutions and Markov selections for stochastic partial differential equations (Q1016638)

This paper treats the stochastic porous medium equations of the type \[ \begin{cases} d x_t = ( \partial_{ij}^2 a^{ij}(u, x_t) + \partial_i b^i(u, x_t) + c(u, x_t) ) dt + \sigma_i(u, x_t) d W_t^i \\ x_t(u) = 0, \quad (t,u) \in {\mathbb R}_+ \times \partial {\mathcal O} \\ x(0) = x_0 \in L^2( {\mathcal O}) \end{cases} \tag{1} \] with Dirichlet boundary condition, and the stochastic Navier-Stokes equations of the type \[ d {\mathbf u}_t = ( \varDelta {\mathbf u}_t - ( {\mathbf u}_t \cdot \nabla ) {\mathbf u}_t + \nabla p(t) + {\mathbf f}(x, {\mathbf u}_t) ) dt + ( \nabla \tilde{p}_i(t) + {\mathbf h}_i(x, {\mathbf u}_t ) ) d W_t^i \tag{2} \] with the incompressibility condition \(\text{div} {\mathbf u}_t=0\), Dirichlet boundary condition \({\mathbf u}_t (x)=0\) for \((t,x) \in{\mathbb R}_+ \times \partial {\mathcal O}\) and with the initial condition \({\mathbf u}_0\) for a bounded domain \({\mathcal O}\subset{\mathbb R}^d\). The purpose of this paper is twofold. The first aim is to present a general existence result of solutions for a large class of stochastic partial differential equations (SDDEs) of evolutionary type in the sense of martingale problem. The second aim is to prove the existence of almost sure Markov selections for abstract stochastic partial differential equation of evolutionary type. The motivation of the first aim consists in the strong desire to treat the stochastic porous medium equations without monotonicity conditions. To this end, the authors propose a quite general existence theorem on solutions for SPDE which also includes the stochastic Navier-Stokes equations over a bounded domain in all dimension with multiplicative noise. On the other hand, without any local weak monotonicity conditions on the coefficients, one cannot expect to prove the uniqueness of martingale solutions. However, one can expect to prove the existence of Markov selections, cf. \textit{F. Flandoli} and \textit{N. Romito} [Probab.\ Theory Relat.\ Fields 140, No. 3--4, 407--458 (2008; Zbl 1133.76016)]. That's why, as the second aim, the authors show the existence of almost sure Markov selections in their framework. In fact, their construction of almost sure Markov selections differs from that in Flandoli-Romito's work, because the abstract Markov selection theorem in this paper is stated in a Polish space. Therefore, this enables them to deal with more general stochastic equations.