Adaptive Decentralized Composite Optimization via Three-Operator Splitting
Xiaokai Chen, Ilya Kuruzov, Gesualdo Scutari · 2026
Read Adaptive Decentralized Composite Optimization via Three-Operator Splitting by Xiaokai Chen, Ilya Kuruzov, Gesualdo Scutari (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-17545.txt.
Description
Xiaokai Chen led a three-author team behind a 2026 arXiv paper on optimization spread across a network of agents. Each agent holds its own smooth convex loss, and all agents share one nonsmooth convex term. The paper proposes decentralized methods in which agents adjust their own step sizes by local backtracking, linked through light min-consensus protocols.
The design grows from a three-operator splitting factorization of a reformulated problem. A new BCV preconditioning metric, named for Bertsekas, O'Connor, and Vandenberghe, makes the decentralized form workable and lets step sizes adapt locally. The authors prove robust convergence guarantees. Under mere convexity, the methods converge sublinearly. With strong convexity of the sum and a partly smooth nonsmooth part, they prove linear convergence. Numerical tests back the theory and show the adaptive step size strategy working in practice.
Catalog record
| CREATOR | Xiaokai Chen, Ilya Kuruzov, Gesualdo Scutari |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 95 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-17545