Randomized gossip algorithms for solving Laplacian systems

Anastasios Zouzias, Nikolaos Freris

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

Abstract

We consider the problem of solving a Laplacian system of equations Lx = b in a distributed fashion, where L is the Laplacian of the communication graph. Solving Laplacian systems arises in a number of applications including consensus, distributed control, clock synchronization, localization and calculating effective resistances, to name a few. We leverage our analysis on a randomized variant of Kaczmarz's algorithm to propose a distributed asynchronous gossip algorithm with expected exponential convergence. We quantify the convergence rate depending solely on properties of the network topology, and further propose an accelerated version that scales favorably for larger networks. Our approach naturally extends to least-squares estimation of general linear systems where each row/column is assigned to nodes of a given network. Last but not least, we show that average consensus is a special case in our framework.

Original languageEnglish (US)
Title of host publication2015 European Control Conference, ECC 2015
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages1920-1925
Number of pages6
ISBN (Electronic)9783952426937
DOIs
StatePublished - Nov 16 2015
EventEuropean Control Conference, ECC 2015 - Linz, Austria
Duration: Jul 15 2015Jul 17 2015

Other

OtherEuropean Control Conference, ECC 2015
CountryAustria
CityLinz
Period7/15/157/17/15

Fingerprint

Linear systems
Clocks
Synchronization
Topology
Communication

Keywords

  • Clock synchronization
  • Consensus
  • Cyberphysical systems
  • Distributed algorithms
  • Gossip algorithms
  • Laplacian systems
  • Randomized algorithms

ASJC Scopus subject areas

  • Control and Systems Engineering

Cite this

Zouzias, A., & Freris, N. (2015). Randomized gossip algorithms for solving Laplacian systems. In 2015 European Control Conference, ECC 2015 (pp. 1920-1925). [7330819] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ECC.2015.7330819

Randomized gossip algorithms for solving Laplacian systems. / Zouzias, Anastasios; Freris, Nikolaos.

2015 European Control Conference, ECC 2015. Institute of Electrical and Electronics Engineers Inc., 2015. p. 1920-1925 7330819.

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

Zouzias, A & Freris, N 2015, Randomized gossip algorithms for solving Laplacian systems. in 2015 European Control Conference, ECC 2015., 7330819, Institute of Electrical and Electronics Engineers Inc., pp. 1920-1925, European Control Conference, ECC 2015, Linz, Austria, 7/15/15. https://doi.org/10.1109/ECC.2015.7330819
Zouzias A, Freris N. Randomized gossip algorithms for solving Laplacian systems. In 2015 European Control Conference, ECC 2015. Institute of Electrical and Electronics Engineers Inc. 2015. p. 1920-1925. 7330819 https://doi.org/10.1109/ECC.2015.7330819
Zouzias, Anastasios ; Freris, Nikolaos. / Randomized gossip algorithms for solving Laplacian systems. 2015 European Control Conference, ECC 2015. Institute of Electrical and Electronics Engineers Inc., 2015. pp. 1920-1925
@inproceedings{f70be5752b2f490a9881f0862ee5e147,
title = "Randomized gossip algorithms for solving Laplacian systems",
abstract = "We consider the problem of solving a Laplacian system of equations Lx = b in a distributed fashion, where L is the Laplacian of the communication graph. Solving Laplacian systems arises in a number of applications including consensus, distributed control, clock synchronization, localization and calculating effective resistances, to name a few. We leverage our analysis on a randomized variant of Kaczmarz's algorithm to propose a distributed asynchronous gossip algorithm with expected exponential convergence. We quantify the convergence rate depending solely on properties of the network topology, and further propose an accelerated version that scales favorably for larger networks. Our approach naturally extends to least-squares estimation of general linear systems where each row/column is assigned to nodes of a given network. Last but not least, we show that average consensus is a special case in our framework.",
keywords = "Clock synchronization, Consensus, Cyberphysical systems, Distributed algorithms, Gossip algorithms, Laplacian systems, Randomized algorithms",
author = "Anastasios Zouzias and Nikolaos Freris",
year = "2015",
month = "11",
day = "16",
doi = "10.1109/ECC.2015.7330819",
language = "English (US)",
pages = "1920--1925",
booktitle = "2015 European Control Conference, ECC 2015",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

TY - GEN

T1 - Randomized gossip algorithms for solving Laplacian systems

AU - Zouzias, Anastasios

AU - Freris, Nikolaos

PY - 2015/11/16

Y1 - 2015/11/16

N2 - We consider the problem of solving a Laplacian system of equations Lx = b in a distributed fashion, where L is the Laplacian of the communication graph. Solving Laplacian systems arises in a number of applications including consensus, distributed control, clock synchronization, localization and calculating effective resistances, to name a few. We leverage our analysis on a randomized variant of Kaczmarz's algorithm to propose a distributed asynchronous gossip algorithm with expected exponential convergence. We quantify the convergence rate depending solely on properties of the network topology, and further propose an accelerated version that scales favorably for larger networks. Our approach naturally extends to least-squares estimation of general linear systems where each row/column is assigned to nodes of a given network. Last but not least, we show that average consensus is a special case in our framework.

AB - We consider the problem of solving a Laplacian system of equations Lx = b in a distributed fashion, where L is the Laplacian of the communication graph. Solving Laplacian systems arises in a number of applications including consensus, distributed control, clock synchronization, localization and calculating effective resistances, to name a few. We leverage our analysis on a randomized variant of Kaczmarz's algorithm to propose a distributed asynchronous gossip algorithm with expected exponential convergence. We quantify the convergence rate depending solely on properties of the network topology, and further propose an accelerated version that scales favorably for larger networks. Our approach naturally extends to least-squares estimation of general linear systems where each row/column is assigned to nodes of a given network. Last but not least, we show that average consensus is a special case in our framework.

KW - Clock synchronization

KW - Consensus

KW - Cyberphysical systems

KW - Distributed algorithms

KW - Gossip algorithms

KW - Laplacian systems

KW - Randomized algorithms

UR - http://www.scopus.com/inward/record.url?scp=84963799660&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84963799660&partnerID=8YFLogxK

U2 - 10.1109/ECC.2015.7330819

DO - 10.1109/ECC.2015.7330819

M3 - Conference contribution

AN - SCOPUS:84963799660

SP - 1920

EP - 1925

BT - 2015 European Control Conference, ECC 2015

PB - Institute of Electrical and Electronics Engineers Inc.

ER -