OpLyX - The Optimization Platform
OpLyX is a generally applicable optimization platform for parameter optimizations. OpLyX incorporates a powerful and versatile optimization algorithm which has proven its effectiveness for many engineering applications. This built-in optimization engine is based on an Evolutionary Algorithm which can handle arbitrary solution processes and a mix of discrete and continuous optimization variables. The Evolutionary Algorithm is equipped with a unique constraining method that allows to set-up the optimization problem efficiently.
OpLyX works with arbitrary simulation tools such as CATIA, ABAQUS, ANSYS, NASTRAN, MATLAB, etc. Furthermore, several simulation tools can be combined to a sequence to solve highly sophisticated engineering problems.
Optimization runs with OpLyX can either be set up on a single computer or in a multi processor environment. Such parallel computations use standard queuing systems or in an office environment OpLyX can set up a workstation cluster. OpLyX can be expanded with various optimization strategies such design of experiment, gradient based strategies, etc.
OpLyX stores all ever created individuals in a database together with computed results. As default backend OpLyX uses a MySQL server to store the individual solutions to an optimization task. Algorithm Daemons interact with the database, create new solutions to the optimization task and write them to the database. Evaluator deamons search the database for solutions which have not yet been evaluated and distribute the evaluation task to idle computers.
All interactions with the database and the daemons are performed via a user friendly Python interface. Online monitoring of individual evaluation processes and of the entire optimization run is available.
Key features of OpLyX
Storage of all solutions which have ever been evaluated
Parallel evaluation on workstation clusters or with a queuing system
Online monitoring of all running jobs
Online monitoring of the optimization process
Works with arbitrary simulation tools
Handles entire sequences of simulation tools for complex evaluations
Key features of the Evolutionary Algorithm
Robust search algorithm even in noisy design spaces
Mix of discrete and continuous optimization variables allows to set-up the optimization close to its real counterpart
Smart constraining method handles multiple constraints simultaneously
32 and 64 bit Windows and Linux
Queuing systems LSF, PBS, SGE
OpLyX Optimized Structures