Université des Sciences et Techniques de Lille
E-mail : El-Ghazali.Talbi@lifl.fr
(PARAllel and DIStributed Evolving Objects). ParadisEO is a free C++ white-box
object-oriented framework dedicated to the reusable design of parallel
metaheuristics for (multi-objective) optimization. It is basically an extension
of the EO evolutionary computation framework (http://eodev.sourceforge.net).
It provides a broad range of new features including local searches (Hill
Climbing, Simulated Annealing
and Tabu Search), the most common parallel models (based on the walk, the
solution and the objective function) and some hybridization mechanisms.
ParadisEO is based on a clear conceptual separation of the
solution methods from the problems they are intended to solve. This separation
confers to the user a maximum code and design reuse. A first implementation
relies on a multi-programmation layer (Posix threads) and some communication
libraries (LAM-MPI or PVM) for execution on dedicated parallel and/or
distributed computational resources. Another implementation relies on Athapascan
and Inuktitut for the dynamic scheduling on a dedicated grid environment. A last
be available soon. It is based on Condor and the Master/Worker API for High
Throughput Computing and Grid Computing on volatile non dedicated resources.
GUIMOO (a Graphical User Interface for Multi Objective Optimization).
Guimoo is a free software dedicated to the analysis of results in
Its main features enable - The on-line visualization of approximative Pareto
frontiers. Such information could be used by the expert to build more efficient
metaheuristics. A Pareto frontier may be characterized by its (dis)continuity, (dis)convexity,
modality, - some metrics for quantitative and qualitative performance evaluation
(contribution, entropy, generational distance, spacing, size of the dominated
space, coverage of two sets and coverage difference) GUIMOO aims to be generic.
Yet, its architecture permits to easily customize it in order to provide the
user more functionalities, as a
specific problem is tackled (it was done for radio network
optimization). Some 'problems' with related files are supplied for
demonstration. They deal with the 'Vehicle Routing Problem' and the 'Flow Shop
Scheduling'. The latest release of Guimoo is available either as an executable
file on Windows or a tarball of sources on Linux.
[Back to EWG-MCDA Software page]