Unified Compression Algorithm for Distributed Nonconvex Optimization: Generalized to 1-Bit, Saturation, and Bounded Noise

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Haonan, Liwang, Minghui, Hong, Yiguang, Johansson, Karl H., Yi, Xinlei
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866914466925379584
author Wang, Haonan
Liwang, Minghui
Hong, Yiguang
Johansson, Karl H.
Yi, Xinlei
author_facet Wang, Haonan
Liwang, Minghui
Hong, Yiguang
Johansson, Karl H.
Yi, Xinlei
contents In this paper, we propose a unified compression algorithm for distributed nonconvex opitmization with both the locally- and globally-bounded communication compressors, including 1-bit compressors, saturating quantizers, and the globally-bounded compressors with both relative and absolute compression errors, as well as additional arbitrary bounded noise. We provide a rigorous convergence analysis in nonconvex settings and establish linear convergence under the Polyak-Lojasiewicz (P-L) condition. Notably, we establish an $\mathcal{O}(1/\sqrt{T})$ convergence rate for the locally-bounded class in the distributed nonconvex setting, matching that achieved by the centralized algorithms with 1-bit compressors, where $T$ denotes the total number of iterations. Moreover, one initial uncompressed communication round further yields an order-wise improvement to $\mathcal{O}(1/T^{2/3})$. For the P-L setting and the globally-bounded class, we recover state-of-the-art convergence rates.
format Preprint
id arxiv_https___arxiv_org_abs_2604_10615
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Unified Compression Algorithm for Distributed Nonconvex Optimization: Generalized to 1-Bit, Saturation, and Bounded Noise
Wang, Haonan
Liwang, Minghui
Hong, Yiguang
Johansson, Karl H.
Yi, Xinlei
Optimization and Control
In this paper, we propose a unified compression algorithm for distributed nonconvex opitmization with both the locally- and globally-bounded communication compressors, including 1-bit compressors, saturating quantizers, and the globally-bounded compressors with both relative and absolute compression errors, as well as additional arbitrary bounded noise. We provide a rigorous convergence analysis in nonconvex settings and establish linear convergence under the Polyak-Lojasiewicz (P-L) condition. Notably, we establish an $\mathcal{O}(1/\sqrt{T})$ convergence rate for the locally-bounded class in the distributed nonconvex setting, matching that achieved by the centralized algorithms with 1-bit compressors, where $T$ denotes the total number of iterations. Moreover, one initial uncompressed communication round further yields an order-wise improvement to $\mathcal{O}(1/T^{2/3})$. For the P-L setting and the globally-bounded class, we recover state-of-the-art convergence rates.
title Unified Compression Algorithm for Distributed Nonconvex Optimization: Generalized to 1-Bit, Saturation, and Bounded Noise
topic Optimization and Control
url https://arxiv.org/abs/2604.10615