On the superlinear convergence of the variable metric proximal point algorithm using Broyden and BFGS matrix secant updating

Citation
Jv. Burke et Mj. Qian, On the superlinear convergence of the variable metric proximal point algorithm using Broyden and BFGS matrix secant updating, MATH PROGR, 88(1), 2000, pp. 157-181
Citations number
20
Categorie Soggetti
Mathematics
Journal title
MATHEMATICAL PROGRAMMING
ISSN journal
00255610 → ACNP
Volume
88
Issue
1
Year of publication
2000
Pages
157 - 181
Database
ISI
SICI code
0025-5610(200006)88:1<157:OTSCOT>2.0.ZU;2-W
Abstract
In previous work, the authors provided a foundation for the theory of varia ble metric proximal point algorithms in Hilbert space. In that work conditi ons are developed for global, linear, and super-linear convergence. This pa per focuses attention on two matrix secant updating strategies for the fini te dimensional case. These are the Broyden and BFGS updates. The BFGS updat e is considered for application in the symmetric case, e.g., convex program ming applications, while the Broyden update can be applied to general monot one operators. Subject to the linear convergence of the iterates and a quad ratic growth condition on the inverse of the operator at the solution, supe r-linear convergence of the iterates is established for both updates. These results are applied to show that the Chen-Fukushima variable metric proxim al point algorithm is super-linearly convergent when implemented with the B FGS update.