We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
A modified Broyden family algorithm with global convergence under a weak Wolfe-Powell line search for unconstrained nonconvex problems.
- Authors
Yuan, Gonglin; Wang, Zhan; Li, Pengyuan
- Abstract
The Quasi-Newton method is one of the most effective methods using the first derivative for solving all unconstrained optimization problems. The Broyden family method plays an important role among the quasi-Newton algorithms. However, the study of the convergence of the classical Broyden family method is still not enough. While in the special case, BFGS method, there have been abundant achievements. Yuan et al. (Appl Math Model. 47:811–825, (2017)) presented a modified weak Wolfe-Powell line search and obtained the convergence of BFGS method for general functions under this line search. Motivated by their works, a new modified weak Wolfe-Powell line search technique is proposed for unconstrained problems. We assume that the objective function is nonconvex and the global convergence of the restricted Broyden family method is established. Preliminary numerical results including the classical optimization problems and the Muskingum model show that the presented algorithm is promising.
- Subjects
ALGORITHMS; QUASI-Newton methods; FAMILIES
- Publication
Calcolo, 2020, Vol 57, Issue 4, pN.PAG
- ISSN
0008-0624
- Publication type
Article
- DOI
10.1007/s10092-020-00383-5