scientific article; zbMATH DE number 3926637

From MaRDI portal
Publication:3701202





[For the entire collection see Zbl 0572.00022.] A method of feasible directions is presented for solving stochastic programming problems with a possibly non-smooth and nonconvex objective function. The algorithm uses stochastic subgradients which, by an auxiliary filter, are averaged and used in direction finding subproblems. The obtained feasible directions are then applied in the main (stochastic approximation type) iteration. The convergence of the procedure (with probability 1) is proved by means of a Lyapunov function technique.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3701202)