Parallel Asynchronous Team Algorithms
Abstract Solution of today large complex problems may need the combination of several different methods, algorithms and techniques in a distributed computing system with heterogeneous processors, usually interconnected through a communication network. In this context, Team Algorithm is presented as a general technique used to combine different methods and algorithms in a distributed computing system composed of different workstations, personal computers and/or parallel computers. Moreover, experimental results have proved that in many real world problems, Team Algorithms can benefit from the use of asynchronous implementations, speeding up the whole process with an important synergy effect, in a new appealing technique known as Parallel Asynchronous Team Algorithms.
The main idea behind Parallel Asynchronous Team Algorithms is very simple: to partition a large complex problem in small sub-problems that can be solved in different processors of a distributed system with well known sequential methods, combining the partial results of each sub-problem in such a way that a good global solution is finally found. Because each processor works at its own speed and eventually, with its own algorithms, an asynchronous implementation eliminates idle synchronization times speeding up the whole process. Team Algorithms have been successfully applied in the resolution of many engineering problems in which a synergetic effect may take place between different processors.
KeywordsDistributed system asynchronous implementation parallelism network algorithm combination
Unable to display preview. Download preview PDF.
- A-Teams Project Home Page, Carnegie Mellon University. URL: http://www.cs.cmu.edu/afs/cs/project/edre-22/project/ateams/WWW/.
- B. Barán, A study ofparallel asynchronous team algorithms, PhD thesis, Federal Univ. of Rio de Janeiro, COPPE, CS Dept., Rio de Janeiro, Brasil, 1993 (in Portuguese). Google Scholar
- B. Barán and E. Chaparro, Algoritmos Asincronos combinados en un Ambiente Heterogéneo de Red, XXIII Conferencia Latinoamericana de Informática, Valparaiso, Chile, 1997 (in Spanish). Google Scholar
- B. Barán, E. Chaparro and N. Cáceres, A-Teams en la Optimización del Caudal Turbinado de una Represa Hidroeléctrica, Conferencia Iberoamericana de Inteligencia Artificial IBERAMIA-98, Lisbon, Portugal, 1998 (in Spanish). Google Scholar
- B. Barán, E. Chaparro and N. Cáceres, Hydroelectric optimization using A-Teams, International Conference in Electrical Power System Operation and Management (EPSOM-98), Zurich, Switzerland, 1998.Google Scholar
- B. Barán and F. Laufer, Topological Optimization of Reliable Networks using A-Teams, Symposium: Architecture, Tools and Algorithm for Networks, Parallel and Distributed Systems, Internacional Conference Systemics, Cybernetics and Informatics SCI’99, Orlando — Florida, United States of America, 1999.Google Scholar
- A. Bhaya, E. Kaszkurewicz, and F. C. Mota, Asynchronous block-iterative methods for almost-linear equations, Lin. Algebra Appl., 1991, pp. 487–508.Google Scholar
- C. J. Coulbourn, Reliability Issues in Telecommunication Network Planning, University of Vermont. URL: http://www.emba.uvm.edu/colbourn.
- C. J. Coulbourn, The Combinatorics of Network Reliability, Oxford Univ. Press, 1987.Google Scholar
- Y. P. Dusonchet, S. N. Talukdar, and H. E. Sinnot, Load flows using a combination of point Jacobi and Newton’s methods, IEEE Trans. Power Apparatus and Systems, PAS-90, 1971, pp. 941–949.Google Scholar
- D.M. Falcão, A. L. B. Bomfim, C. R. R. Dornellas, and G. N. Taranto, Genetics Algorithms in Power Systems Optimization, V SEPOE, Recife — Brazil, 1996.Google Scholar
- M.R. Garey and D. S. Johnson, Computers and Intractability: A Guide to the Theory of NP-Completeness, V SEPOE, Recife — Brazil, 1996.Google Scholar
- D.E. Goldberg, Genetic Algorithm in Search, Optimization and Machine Learning, San Francisco, Freeman, 1979.Google Scholar
- E. Kaszkurewicz, A. Bhaya, and D. D. Siljak, On the convergence of parallel asynchronous block-iterative computations, Lin. Algebra Appl., 1990, pp. 139–160.Google Scholar
- M. Mejia and E. Cantúi, DGENESIS: Software para la Ejecución de Algoritmos Genéticos Distribuidos, XX Conferencia Latinoamericana de Informática , Mexico City, Mexico, 1994 (in Spanish). Google Scholar
- N. H. Penny, Blackboard systems: the blackboard model ofproblem solving and the evolution ofblackboard architectures, The AI Magazine, 1986.Google Scholar
- P. S. Souza and S. N. Talukdar, Genetics Algorithms in Asynchronous Teams, ICGA — 91, San Diego — California, 1991, pp. 392–397.Google Scholar
- P. S. Souza and S. N. Talukdar, Genetic algorithms in asynchronous teams, in Proc. Fourth Intl. Conf. on Genetic Algorithms, California, 1991, pp. 392–397.Google Scholar
- H. S. Stone, High Performance ComputerArchitectures, Addison-Wesley, 1971.Google Scholar
- B. Stott, Review of load-flow calculation methods, Proc. IEEE, 1974, pp. 916–927.Google Scholar
- S. N. Talukdar, S. S. Pyo, and R. Mehrotra, Designing algorithms and assignments for distributed processing, Tech. Rep. EPRI EL-3317 RP17643, Electric Power Research Institute, 1983.Google Scholar
- S. N. Talukdar, V. C. Ramesh, and J. C. Nixon, A distributed system of control specialists for real-time operations, in Proc. Third Symp. Expert System Appl. to Power Systems, Japan, 1991.Google Scholar
- S. N. Talukdar, V. C. Ramesh, R. Quadrel, and R. Christie, Multiagent organizationsfor real-time operations, in Proc. of the IEEE, 1992, pp. 765–778.Google Scholar
- M. H. M. Vale, D. M. Falcão, and E. Kaszkurewicz, Electrical power network decomposition for parallel computations, in Proc. Int’l. Conf. Circuits and Systems (ISCAS), IEEE, San Diego, California, 1992.Google Scholar
- A. Vanelli, Solution techniques for 0–1 indefinite quadratic programming problems with applications to decomposition, PhD thesis, University of Waterloo, 1983.Google Scholar
- A. J. Wood and B. F. Wollenberg, Power Generation, Operation and Control, John and Sons, 1983.Google Scholar