Abstract:
A bilinear programming problem with uncoupled variables is considered. First, a special technique for generating test bilinear problems is considered. Approximate algorithms for local and global search are proposed. Asymptotic convergence of these algorithms is analyzed, and stopping rules are proposed. In conclusion, numerical results for randomly generated bilinear problems are presented and analyzed.
Key words:nonconvex optimization, bilinear programming, test problem generation, local search, critical point, stopping rule, global search, numerical experiment.