Metraipolis › archive › dc_78f35258fb17
Distributed Zeroth-Order Composite Finite-Sum Optimization via Incremental Variance Reduction — arXiv:2610.41615
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.41615
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T08:16:28.129Z
- SHA-256 of the archived text
1c8b2e58e6bd9a770706cbc12b65401e437509156ec616dce2621d17466a18ec- Stable URL of this page
https://metraipolis.com/documento/dc_78f35258fb17
How to cite
Distributed Zeroth-Order Composite Finite-Sum Optimization via Incremental Variance Reduction — arXiv:2610.41615 — arXiv (en), https://arxiv.org/abs/2610.41615 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_78f35258fb17 (Metraipolis archive dc_78f35258fb17, sha256 1c8b2e58e6bd9a770706cbc12b65401e437509156ec616dce2621d17466a18ec)
BibTeX
@misc{dc78f35258fb17,
title = {Distributed Zeroth-Order Composite Finite-Sum Optimization via Incremental Variance Reduction — arXiv:2610.41615},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.41615},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_78f35258fb17; sha256 1c8b2e58e6bd9a770706cbc12b65401e437509156ec616dce2621d17466a18ec},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41615. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
We study decentralized composite finite-sum optimization over a fixed connected undirected network, where agents access local component functions through function evaluations and share a possibly nonsmooth convex regularizer. In this setting, standard two-point gradient estimators can retain nonvanishing variance at a solution, complicating their integration with gradient tracking. We propose Distributed Zeroth-Order Incremental Variance Reduction (Distributed ZIVR), which combines incrementally updated component-gradient estimates with a proximal gradient-tracking scheme. The method admits an implementation using a constant number of function evaluations per agent at every iteration, without intermittent full-component or full-coordinate refreshes. For convex smooth component functions and strongly convex local averages, we establish linear convergence up to a controllable finite-difference error. With the finite-difference radius chosen for the target accuracy, Chebyshev-accelerated mixing yields a per-agent oracle complexity of $\widetilde{O}(d(m+\kappa)\log(1/\epsilon))$ for expected squared solution error $\epsilon$, where $d$ is the dimension, $m$ is the number of components per agent, and $\kappa$ is the condition number. This matches the centralized ZIVR oracle bound, while the communication bound has only an inverse square-root dependence on the network spectral gap. Experiments on logistic and robust regression demonstrate rapid reduction of solution and consensus errors.
Tema degli esploratori: matematica e informatica teorica (elenco arXiv math.OC, nuovi lavori del giorno).
Categorie arXiv: Optimization and Control (math.OC) ; Systems and Control (eess.SY).
Pagina del lavoro: https://arxiv.org/abs/2610.41615 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito).
Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina.
Letto il 2026-11-03T08:04:23.903Z dalla pagina https://arxiv.org/list/math.OC/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).