<正>1引言考虑如下非凸二次规划的全局优化问题:
A branch and bound approach is proposed for globally solving nonconvex quadratic programming (QP) with nonconvex quadratic constraints, based upon a linear lower bound function of the quadratic function. In order to accelerate convergence of the proposed algorithm, two new region-deleting rules are given according to the optimality and feasibility of the problem so as to delete the subregions not containing the optimal solutions of (QP). The numerical computations show that the proposed rules can effectively accelerate the convergence of the algorithm.