Multivalued backward stochastic differential equations with local lipschitz drift (Q4347779)

The author considers the one-dimensional backward stochastic differential equation associated to the subdifferential \(\delta n\) of a lower semicontinuous function \(h\): \[ Y_t= \xi+\int^T_t f(s,Y_s,Z_s)ds- \int^T_t Z_sdW_s+ K_T- K_t.\tag{\(*\)} \] A triple \((Y,Z,K)\) of progressively measurable processes is solution of \((*)\) if it satisfies: (a) \(E\left(\sup_{0\leq t\leq T} |Y_t|^2+ \int^T_0|Z_s|^2ds\right)< \infty\), (b) \((Y_t, 0\leq t\leq T)\) is continuous and valued in \(\overline{\text{Dom}} h\), (c) \((K_t, 0\leq t\leq T)\) is continuous and has bounded variations with \(K_0= 0\) a.s., (d) for any optional process \(\{(\alpha_t,\beta_t), 0\leq t\leq T\}\) valued in \(\text{Gr}(\delta h)\), the measure \((Y_t- \alpha_t)(dK_t+ \beta_tdt)\) is a.s. negative. If \(\xi\) is bounded, and \(f\) locally Lipschitz in \((y,z)\), there exists a solution to \((*)\). The technique used is similar to \textit{S. Hamadene} [Ann. Inst. Henri Poincaré, Probab. Stat. 32, No. 5, 645-659 (1997)]. The uniqueness is obtained under the condition that \(f\) is locally Lipschitz in \(y\), and globally Lipschitz in \(z\).