logo

FROG-EA


Description

FROG-EA is a framework enabling the distributed execution of initially serial implemented evolutionary algorithms in Globus 4.0 based computational grids with only little effort for the user. It supports the split up of the algorithm in subtasks, the synchronization, scheduling and submit of them to available compute resources of a grid. Likewise it is responsible for the dicovery and monitoring of available resources, transfer of in- and output files and fault handling. Two forms of parallelization are supported: the island model which splits up the population in several subpopulations that are processed in parallel and the distributed evaluation of the individuals. Thus, FROG-EA exploits the large computational power of grids for evolutionary algorithms and requires only minor knowlage of the user about the usage of such grids.

Slides

Contact

Address: Prof. Dr.-Ing. Dietmar Fey
Lehrstuhl für Informatik 3 (Rechnerarchitektur)
Martensstrasse 3
91058 Erlangen
Germany
Office: 07.156
Phone: +49 9131 85 27003
Fax: +49 9131 85 27912
Homepage: http://www3.informatik.uni-erlangen.de/Persons/fey/
E-Mail: dietmar.fey@informatik.uni-erlangen.de
Homepage: http://www3.informatik.uni-erlangen.de/Persons/limmer/
E-Mail: steffen.limmer@informatik.uni-erlangen.de

Publications

  • S. Limmer, D. Fey: Framework for Distributed Evolutionary Algorithms in Computational Grids. In: LNCS vol. 6382, Advances in Computation and Intelligence (5th International Symposium on Advances in Computation and Intelligence, Wuhan, China, October 2010). Berlin / Heidelberg: Springer Verlag, 2010, pp. 170-180.
  • Evolutionary Optimization of Layouts for High Density Free Space Optical Network Links. To be published in conference proceedings of GECCO 2011, Dublin, Ireland