求二次规划问题全局解的新加速方法

A new accelerating method for solving quadratic programming

目的为求目标函数为一般二次函数的二次规划问题,提出一个新的加速算法。方法通过结合两个加速技巧,并将其置于分支定界算法框架下,给出一个新的全局优化算法。结果该方法可以有效地确定出不定二次规划问题的全局最优解。结论理论上证明了算法的收敛性,数值算例表明算法是有效可行的。

Aim For globally solving quadratic programming, an accelerating algorithm is presented. Methods Through combining two accelerating techniques, and putting it into the frame of branch and bound, a new global optimization algorithm is proposed. Results This method can be used to solve indefinite quadratic programming problem effectively. Conclusion Convergence of the algorithm is established and numerical results are given to show the feasibility and effectiveness.