一类修正BFGS算法的局部超线性收敛性
Local Q-superlinear Convergence of a Modified BFGS Algorithm
-
摘要: 基于Hiroshi Yabe提出的新拟牛顿方程,给出了一类更广泛的修正BFGS算法(MBFGS算法),并在一定的假设条件下,结合Wolfe搜索准则,证明了MBFGS算法具有局部超线性收敛性.数值试验结果表明,对于一般无约束优化, MBFGS算法是正确和有效的.Abstract: In this paper, we present a more general modified BFGS(MBFGS)method based on the new qua-si-Newton equation Hiroshi Yabe proposed. Under some suitable conditions, we prove local q-superlinear convergence of our method with Wolfe linear search. Meanwhile, numerical experiment results also show that compared with the generally used method, our modified BFGS presented is correct and efficient for unconstrained optimization.
-
-
计量
- 文章访问数: 476
- HTML全文浏览数: 227
- PDF下载数: 0
- 施引文献: 0