Optimal stopping and free boundary characterizations for some Brownian control problems (Q2378635): Difference between revisions

A singular stochastic control problem with state constraints in \(\mathbb R_{+}^2\) is studied for the state process \( X(t)=x+B(t)+Y(t)\) where \(x \in \mathbb R_{+}^2, t \geq 0 \), \(B\) is a two dimensional Brownian motion with a nondegenerate covariance matrix and the control \(Y\) is a nondecreasing, right continuous with left limits adapted process. The value function \(V(x)\) is the infimum of \(J(x,Y)= E \int_0^{\infty}e^{- \gamma t} l(X(t)) \,dt\) over all admissible controls, where \( l(z_1,z_2)\) equals \(\langle(\alpha_1,\alpha_2), (z_1,z_2)\rangle\) (respectively \(\langle(\beta_1, \beta_2),(z_1,z_2)\rangle\)) if \(z_2 \geq c z_1\) (respectively \(z_2 < c z_1\)), \((\alpha_1, \alpha_2) \in \mathbb R^2\), \((\beta_1, \beta_2) \in \mathbb R^2\) and \(c>0\). The \(C^1 \)-property of \( V(x)\) is proved in Section 3. In Section 4, the authors introduce an optimal stopping problem whose the value function equals a directional derivative of \(V(x)\). The associated free boundary is introduced in Section 5 to show that an optimally controlled process with \(\alpha_1 < 0\), \(\alpha_2 > 0\), \(\beta_1 > 0\), \(\beta_2 \geq 0\) is a Brownian motion in a no-action region with reflection at the free boundary.