Practical optimization and nonconvex problems

Mathematical Modelling: Methods, algorithms, technologies

Main aspects of optimization problem in real computing are described. Requirement to take into account such parameters of applied extremal problem as rigidity and nonconvexity of criterion functional is founded. It is established that classic Newtonian and quasi-Newtonian optimization procedures are strongly influenced by inaccuracies in this case and nonconvergent procedures may appear.