On sufficient and necessary of existence for a class of singular optimal stochastic control (Q1433516): Difference between revisions

The paper deals with a class of discounted models of singular stocahstic optimal control. The state process \(X_t\), \(t\geq 0\) is the solution of a SDE with nonlinear drift \(\mu\) and diffusion \(\sigma^2\): \[ X_t= x+\int^t_0 \mu(X_s)\,ds+ \int^t_0 \sigma(X_s)\,dW_s+ \xi_t,\;t\geq 0;\;x=\text{const.} \] Here \(\xi= (\xi_t,\,t\geq 0)\in{\mathcal B}\), the set of left continuous and adapted processes of bounded variation. The structure of the cost function is quite general: \[ v(x,\xi)= {\mathbf E} \Biggl\{\int^\infty_0 e^{-\alpha t}[g(X_t)\,d\widehat\xi_t]+ h(X_t)]\,dt\Biggr\}. \] And the goal is to find the value function \(v(x)= \inf_\xi\,v(x,\xi)\), where \(\xi\in{\mathcal B}\). The authors have found conditions that are necessary and sufficient for the existence of the optimal control. The proofs are based on several results from stochastic analysis and variational inequalities. The authors intend to use the same ideas and techniques in their further work in the area of stochastic optimal control.