Geometrically convergent distributed optimization with uncoordinated step-sizes

Angelia Nedich, Alex Olshevsky, Wei Shi, Cesar A. Uribe

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Citations (Scopus)

Abstract

A recent algorithmic family for distributed optimization, DIGing's, have been shown to have geometric convergence over time-varying undirected/directed graphs [1]. Nevertheless, an identical step-size for all agents is needed. In this paper, we study the convergence rates of the Adapt-Then-Combine (ATC) variation of the DIGing algorithm under uncoordinated step-sizes. We show that the ATC variation of DIGing algorithm converges geometrically fast even if the step-sizes are different among the agents. In addition, our analysis implies that the ATC structure can accelerate convergence compared to the distributed gradient descent (DGD) structure which has been used in the original DIGing algorithm.

Original languageEnglish (US)
Title of host publication2017 American Control Conference, ACC 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages3950-3955
Number of pages6
ISBN (Electronic)9781509059928
DOIs
StatePublished - Jun 29 2017
Event2017 American Control Conference, ACC 2017 - Seattle, United States
Duration: May 24 2017May 26 2017

Other

Other2017 American Control Conference, ACC 2017
CountryUnited States
CitySeattle
Period5/24/175/26/17

Fingerprint

Directed graphs

ASJC Scopus subject areas

  • Electrical and Electronic Engineering

Cite this

Nedich, A., Olshevsky, A., Shi, W., & Uribe, C. A. (2017). Geometrically convergent distributed optimization with uncoordinated step-sizes. In 2017 American Control Conference, ACC 2017 (pp. 3950-3955). [7963560] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/ACC.2017.7963560

Geometrically convergent distributed optimization with uncoordinated step-sizes. / Nedich, Angelia; Olshevsky, Alex; Shi, Wei; Uribe, Cesar A.

2017 American Control Conference, ACC 2017. Institute of Electrical and Electronics Engineers Inc., 2017. p. 3950-3955 7963560.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Nedich, A, Olshevsky, A, Shi, W & Uribe, CA 2017, Geometrically convergent distributed optimization with uncoordinated step-sizes. in 2017 American Control Conference, ACC 2017., 7963560, Institute of Electrical and Electronics Engineers Inc., pp. 3950-3955, 2017 American Control Conference, ACC 2017, Seattle, United States, 5/24/17. https://doi.org/10.23919/ACC.2017.7963560
Nedich A, Olshevsky A, Shi W, Uribe CA. Geometrically convergent distributed optimization with uncoordinated step-sizes. In 2017 American Control Conference, ACC 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 3950-3955. 7963560 https://doi.org/10.23919/ACC.2017.7963560
Nedich, Angelia ; Olshevsky, Alex ; Shi, Wei ; Uribe, Cesar A. / Geometrically convergent distributed optimization with uncoordinated step-sizes. 2017 American Control Conference, ACC 2017. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 3950-3955
@inproceedings{ea8f7e4b0b42451c81378526e3d4a585,
title = "Geometrically convergent distributed optimization with uncoordinated step-sizes",
abstract = "A recent algorithmic family for distributed optimization, DIGing's, have been shown to have geometric convergence over time-varying undirected/directed graphs [1]. Nevertheless, an identical step-size for all agents is needed. In this paper, we study the convergence rates of the Adapt-Then-Combine (ATC) variation of the DIGing algorithm under uncoordinated step-sizes. We show that the ATC variation of DIGing algorithm converges geometrically fast even if the step-sizes are different among the agents. In addition, our analysis implies that the ATC structure can accelerate convergence compared to the distributed gradient descent (DGD) structure which has been used in the original DIGing algorithm.",
author = "Angelia Nedich and Alex Olshevsky and Wei Shi and Uribe, {Cesar A.}",
year = "2017",
month = "6",
day = "29",
doi = "10.23919/ACC.2017.7963560",
language = "English (US)",
pages = "3950--3955",
booktitle = "2017 American Control Conference, ACC 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
address = "United States",

}

TY - GEN

T1 - Geometrically convergent distributed optimization with uncoordinated step-sizes

AU - Nedich, Angelia

AU - Olshevsky, Alex

AU - Shi, Wei

AU - Uribe, Cesar A.

PY - 2017/6/29

Y1 - 2017/6/29

N2 - A recent algorithmic family for distributed optimization, DIGing's, have been shown to have geometric convergence over time-varying undirected/directed graphs [1]. Nevertheless, an identical step-size for all agents is needed. In this paper, we study the convergence rates of the Adapt-Then-Combine (ATC) variation of the DIGing algorithm under uncoordinated step-sizes. We show that the ATC variation of DIGing algorithm converges geometrically fast even if the step-sizes are different among the agents. In addition, our analysis implies that the ATC structure can accelerate convergence compared to the distributed gradient descent (DGD) structure which has been used in the original DIGing algorithm.

AB - A recent algorithmic family for distributed optimization, DIGing's, have been shown to have geometric convergence over time-varying undirected/directed graphs [1]. Nevertheless, an identical step-size for all agents is needed. In this paper, we study the convergence rates of the Adapt-Then-Combine (ATC) variation of the DIGing algorithm under uncoordinated step-sizes. We show that the ATC variation of DIGing algorithm converges geometrically fast even if the step-sizes are different among the agents. In addition, our analysis implies that the ATC structure can accelerate convergence compared to the distributed gradient descent (DGD) structure which has been used in the original DIGing algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85027031271&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85027031271&partnerID=8YFLogxK

U2 - 10.23919/ACC.2017.7963560

DO - 10.23919/ACC.2017.7963560

M3 - Conference contribution

AN - SCOPUS:85027031271

SP - 3950

EP - 3955

BT - 2017 American Control Conference, ACC 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -