4.6 Article

A trust region algorithm with a worst-case iteration complexity of for nonconvex optimization

期刊

MATHEMATICAL PROGRAMMING
卷 162, 期 1-2, 页码 1-32

出版社

SPRINGER HEIDELBERG
DOI: 10.1007/s10107-016-1026-2

关键词

Unconstrained optimization; Nonlinear optimization; Nonconvex optimization; Trust region methods; Global convergence; Local convergence; Worst-case iteration complexity; Worst-case evaluation complexity

资金

  1. U.S. Department of Energy, Office of Science, Office of Advanced Scientific Computing Research, Applied Mathematics, Early Career Research Program [DE-SC0010615]
  2. U.S. National Science Foundation [DMS-1217153, DMS-1319356]

向作者/读者索取更多资源

We propose a trust region algorithm for solving nonconvex smooth optimization problems. For any , the algorithm requires at most iterations, function evaluations, and derivative evaluations to drive the norm of the gradient of the objective function below any . This improves upon the bound known to hold for some other trust region algorithms and matches the bound for the recently proposed Adaptive Regularisation framework using Cubics, also known as the arc algorithm. Our algorithm, entitled trace, follows a trust region framework, but employs modified step acceptance criteria and a novel trust region update mechanism that allow the algorithm to achieve such a worst-case global complexity bound. Importantly, we prove that our algorithm also attains global and fast local convergence guarantees under similar assumptions as for other trust region algorithms. We also prove a worst-case upper bound on the number of iterations, function evaluations, and derivative evaluations that the algorithm requires to obtain an approximate second-order stationary point.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.6
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据