The superlinear convergence of the primal variables under milder conditions is proved.
讨论了利用NCP函数将KKT条件转化为与之等价的一个半光滑等式,并针对求解这个半光滑KKT等式的混合拟牛顿算法,在比较弱的条件下,证明了算法所计算的序列中原问题变量的超线性收敛性。
However,in the proof of the superlinear convergence,their method still completely relies on this condition.
本文改进了这一结果,在对原假设进行分析的基础上,给出了比严格互补性假设更弱的条件,证明在这一新假设下仍然可以得到超线性收敛性。
By choosing the proper parameters,we can prove that the new method has global and superlinear convergence properties under suitable conditions.
通过选择适当的参数,证明了改进的BFGS方法对于一类更广的搜索准则保持局部收敛性,在Wolfe搜索准则下方法还具有超线性收敛性。
At each iteration, only one system of linear equations needs to be solved, and its global linear convergence and local quadratic convergence are proved.
对P0矩阵线性互补问题提出了一个基于Chen Harker Kanzow Smale光滑函数的非内点连续算法,该算法在每次迭代时只需求解一个线性等式组,并证明了算法的全局线性收敛性和局部二次收敛性。
Finally the Q-linear convergence of the dual algorithm, which isbased on the potential function, is proved.
基于Carroll(1961)建立的罚函数,本文给出了不等式约束优化问题的一个势函数,并且讨论了该函数的性质,最后证明了在此基础上建立的对偶算法具有Q-线性收敛性。
It was proved that,when the objective function was uniformly convex,this algorithm possessed superlinear convergence.
证明该算法在目标函数为一致凸时具有局部超线性收敛性。