Continuous Optimization – Algorithms, Applications, and Analysis
Org: Ahmet Alacaoglu, Michael Friedlander et Jiajin Li (University of British Columbia)
- AHMET ALACAOGLU, UBC
- HEINZ BAUSCHKE, UBC Okanagan
On the Bredies-Chenchene-Lorenz-Naldi algorithm [PDF]
-
Monotone inclusion problems are central in optimization and variational analysis, often solved using splitting methods featuring resolvents or proximal mappings. In 2022, Bredies, Chenchene, Lorenz, and Naldi introduced an elegant framework that unifies well-known algorithms, including Douglas-Rachford and Chambolle-Pock, with strong convergence results under certain conditions.
In this talk, I will report on joint work with Walaa Moursi, Shambhavi Singh, and Xianfu Wang. We extend the analysis of Bredies et al., providing new strong convergence results for linear relations. For the Chambolle-Pock algorithm, we prove convergence to the projection onto an intersection of linear subspaces. We also discuss algorithms by Ryu and by Malitsky and Tam.
- JAMES BURKE, University of Washington
- YING CUI, UC Berkeley
- JELENA DIAKONIKOLAS, University of Wisconsin
- DIMA DRUSVYATSKIY, University of Washington
- MICHAEL FRIEDLANDER, UBC
- NAOMI GRAHAM, UBC
- TIM HOHEISEL, McGill
- JIAJIN LI, UBC
- TIANYI LIN, Columbia
- ZHAOSONG LU, University of Minnesota
- YURA MALITSKY, University of Vienna
- DOMINIQUE ORBAN, Ecole Polytechnique
- COURTNEY PAQUETTE, McGill
- NICHOLAS RICHARDSON, UBC
- JOHANNES ROYSET, USC
- GESUALDO SCUTARI, Purdue
- HENRY WOLKOWICZ, Waterloo
- ALP YURTSEVER, Umea University
- ZIRUI ZHOU, Huawei Technologies