提出了一个修正的强次可行序列二次约束二次规划(SQCQP)算法.通过设计一个新的矩阵修正策略,算法在全局收敛性分析中不需要假设目标函数的(近似)Hesse阵正定或一致正定.在适当条件下,算法具备超线性收敛性.
This paper presents a modified strongly subfeasible sequential quadratically constrained quadratic programming (SQCQP) algorithm. By designing a new modification strategy of matrix, in the global convergence analysis the algorithm does not require that the (approximate) Hessian matrix of the objective function is positive denite or uniformly positive denite. Under suitable conditions, the algorithm possesses superlinear convergence.