Approximate Bregman proximal gradient algorithm with variable metric Armijo--Wolfe line search
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915876076257280 |
|---|---|
| author | Fujiki, Kiwamu Takahashi, Shota Takeda, Akiko |
| author_facet | Fujiki, Kiwamu Takahashi, Shota Takeda, Akiko |
| contents | We propose a variant of the approximate Bregman proximal gradient (ABPG) algorithm for minimizing the sum of a smooth nonconvex function and a nonsmooth convex function. ABPG is known to converge globally to a stationary point even when the smooth part of the objective function does not have a globally Lipschitz continuous gradient, and its iterates can often be expressed in closed form. However, ABPG relies on an Armijo line search to guarantee global convergence, which can slow down its practical performance. To address this issue, we propose a variant of ABPG with a variable metric Armijo--Wolfe line search. Under the variable metric Armijo--Wolfe condition, we establish global subsequential convergence of the algorithm. Moreover, assuming the Kurdyka--Łojasiewicz property, we also prove that the algorithm globally converges to a stationary point. Numerical experiments on $\ell_p$-regularized least squares problems and nonnegative linear inverse problems demonstrate that the proposed algorithm outperforms existing algorithms. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_06615 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Approximate Bregman proximal gradient algorithm with variable metric Armijo--Wolfe line search Fujiki, Kiwamu Takahashi, Shota Takeda, Akiko Optimization and Control We propose a variant of the approximate Bregman proximal gradient (ABPG) algorithm for minimizing the sum of a smooth nonconvex function and a nonsmooth convex function. ABPG is known to converge globally to a stationary point even when the smooth part of the objective function does not have a globally Lipschitz continuous gradient, and its iterates can often be expressed in closed form. However, ABPG relies on an Armijo line search to guarantee global convergence, which can slow down its practical performance. To address this issue, we propose a variant of ABPG with a variable metric Armijo--Wolfe line search. Under the variable metric Armijo--Wolfe condition, we establish global subsequential convergence of the algorithm. Moreover, assuming the Kurdyka--Łojasiewicz property, we also prove that the algorithm globally converges to a stationary point. Numerical experiments on $\ell_p$-regularized least squares problems and nonnegative linear inverse problems demonstrate that the proposed algorithm outperforms existing algorithms. |
| title | Approximate Bregman proximal gradient algorithm with variable metric Armijo--Wolfe line search |
| topic | Optimization and Control |
| url | https://arxiv.org/abs/2510.06615 |