No label defined (Q3972976)

[For the entire collection see Zbl 0719.00020.] The article deals with constructing quadratic dual bounds for some nonconvex problems. In the introduction the process of getting dual bounds with the use of a Lagrange function is described. It is shown that finding the dual bound can be considered as a coordinating problem in decomposition scheme with respect to constraints. Finding optimum in a coordinating problem is usually a non-smooth problem and subgradient type methods with space dilatation proved to solve it well. The procedure of finding a dual bounds, for a quadratic non-convex optimization problem is studied and an algorithm for its evaluation is described. Several examples demonstrate how to use this technique for finding a global minimum of a polynomial function and to solve a maximum stable set problem. The properties of quadratic dual bounds for the maximum stable set problem and the problem of finding a global minimum of a polynomial function are given alongside with results of numerical tests.