Support theorem for stochastic variational inequalities (Q616307)

Assume that \(\varphi\) is a proper lower semicontinuous convex function on \(\mathbb R^m\) (satisfying a few other technical assumptions), \(\partial\varphi\) is its subdifferential, \(w\) is a \(d\)-dimensional Wiener process on some stochastic basis and \(b\) and \(\sigma\) are \(C^3_b\)-smooth functions from \(\mathbb R^m\) to \(\mathbb R^m\) and to \((m\times d)\)-matrices, respectively. A pair \((X,K)\) of continuous adapted processes is said to be a solution of a stochastic variational inequality \[ dX(t)\in b(X(t))\,dt+\sigma(X(t))\circ dw(t)-\partial\varphi(X(t))\,dt,\qquad X(0)=x\in\overline{D(\partial\varphi)} \] if \(X(t)\in\overline{D(\partial\varphi)}\) a.s., paths of \(K\) have bounded variation a.s., \(dK(t)\in\partial\varphi(X(t))\,dt\) and \[ dX(t)=b(X(t))\,dt+\sigma(X(t))\circ dw(t)-dK(t),\quad X(0)=x,\quad K(0)=0. \] It is known that, under the above assumptions, there exists a unique solution \((X,K)\) and the authors describe the support of its law \(P_x\) on \(C(\mathbb R_+;\mathbb R^{2m})\) in the following way. Denote by \(\xi_{h,x}\) the unique solution of \[ d\xi(t)\in b(\xi(t))\,dt+\sigma(\xi(t))\dot h(t)\,dt-\partial\varphi(\xi(t))\,dt,\qquad \xi(0)=x\in\overline{D(\partial\varphi)} \] and define \[ \eta_{h,x}(t)=-\xi_{h,x}(t)+\int_0^t[b(\xi_{h,x}(s))+\sigma(\xi_{h,x}(s))\dot h(s)]\,ds. \] Then \[ \text{supp}(P_x)=\overline{\{(\xi_{h,x},\eta_{h,x}):h\text{ smooth}\}}=\overline{\{(\xi_{h,x},\eta_{h,x}):h\text{ piecewise smooth}\}}. \]