Distributed Scheduling with Decomposed Optimization Criterion: Genetic Programming Approach

Franciszek Seredyński; Jacek Koronacki; Cezary Z. Janikow

doi:10.1007/BFb0097900

Distributed Scheduling with Decomposed Optimization Criterion: Genetic Programming Approach

Franciszek Seredyński, Jacek Koronacki, Cezary Z. Janikow

University of Missouri-St. Louis

Research output: Contribution to conference › Presentation

Abstract

A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization criteria. For this purpose a program graph is interpreted as a multi-agent system. A game-theoretic model of interaction between agents is applied. Competetive coevolutionary genetic algorithm, termed loosely coupled genetic algorithm, is used to implement the multi-agent system. To make the algorithm trully distributed, decomposition of the global optimization criterion into local criteria is proposed. This decomposition is evolved with genetic programming. Results of succesive experimental study of the proposed algorithm are presented.

Original language	American English
DOIs	https://doi.org/10.1007/BFb0097900
State	Published - Apr 12 1999
Event	International Parallel Processing Symposium - Duration: Apr 12 1999 → …

Conference

Conference	International Parallel Processing Symposium
Period	4/12/99 → …

Keywords

Genetic Programming
Local Criterion
Schedule Problem
Task Graph
Total Execution Time

Disciplines

Digital Communications and Networking
Applied Mathematics
Computer Sciences

Access to Document

10.1007/BFb0097900License: Unspecified

Cite this

@conference{08582644e35048b888dca870829ee8ea,

title = "Distributed Scheduling with Decomposed Optimization Criterion: Genetic Programming Approach",

abstract = " A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization criteria. For this purpose a program graph is interpreted as a multi-agent system. A game-theoretic model of interaction between agents is applied. Competetive coevolutionary genetic algorithm, termed loosely coupled genetic algorithm, is used to implement the multi-agent system. To make the algorithm trully distributed, decomposition of the global optimization criterion into local criteria is proposed. This decomposition is evolved with genetic programming. Results of succesive experimental study of the proposed algorithm are presented. ",

keywords = "Genetic Programming, Local Criterion, Schedule Problem, Task Graph, Total Execution Time",

author = "Franciszek Seredy{\'n}ski and Jacek Koronacki and Janikow, {Cezary Z.}",

note = "A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization...; International Parallel Processing Symposium ; Conference date: 12-04-1999",

year = "1999",

month = apr,

day = "12",

doi = "10.1007/BFb0097900",

language = "American English",

}

TY - CONF

T1 - Distributed Scheduling with Decomposed Optimization Criterion: Genetic Programming Approach

AU - Seredyński, Franciszek

AU - Koronacki, Jacek

AU - Janikow, Cezary Z.

N1 - A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization...

PY - 1999/4/12

Y1 - 1999/4/12

N2 - A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization criteria. For this purpose a program graph is interpreted as a multi-agent system. A game-theoretic model of interaction between agents is applied. Competetive coevolutionary genetic algorithm, termed loosely coupled genetic algorithm, is used to implement the multi-agent system. To make the algorithm trully distributed, decomposition of the global optimization criterion into local criteria is proposed. This decomposition is evolved with genetic programming. Results of succesive experimental study of the proposed algorithm are presented.

AB - A new approach to develop parallel and distributed scheduling algorithms for multiprocessor systems is proposed. Its main innovation lies in evolving a decomposition of the global optimization criteria. For this purpose a program graph is interpreted as a multi-agent system. A game-theoretic model of interaction between agents is applied. Competetive coevolutionary genetic algorithm, termed loosely coupled genetic algorithm, is used to implement the multi-agent system. To make the algorithm trully distributed, decomposition of the global optimization criterion into local criteria is proposed. This decomposition is evolved with genetic programming. Results of succesive experimental study of the proposed algorithm are presented.

KW - Genetic Programming

KW - Local Criterion

KW - Schedule Problem

KW - Task Graph

KW - Total Execution Time

UR - https://rd.springer.com/content/pdf/10.1007%2FBFb0097900.pdf

U2 - 10.1007/BFb0097900

DO - 10.1007/BFb0097900

M3 - Presentation

T2 - International Parallel Processing Symposium

Y2 - 12 April 1999

ER -

Distributed Scheduling with Decomposed Optimization Criterion: Genetic Programming Approach

Abstract

Conference

Keywords

Disciplines

Access to Document

Other files and links

Cite this