On the numerical solution of the problem of nonconvex optimization (Q2729361)

The article is devoted to the numerical study of the minimax problem, in particular, special attention is paid to the search of a global minimum for the quadratic functions with matrices of variable sign on a parallelepiped. The authors expose sufficient conditions for global optimality of the general nonconvex problem for d.~c. optimization. Using the conditions obtained, the authors present a global search strategy to the problem and, as a result, for the minimization problem of quadratic functions, they obtain certain realizations of the strategy. The test calculations illustrate the efficiency of the algorithm suggested.