Optimizer: Difference between revisions
From SUMOwiki
Jump to navigationJump to search
(New page: == Optimizer == TODO: stub === DirectOptimizer === === MatlabOptimizer ===) |
mNo edit summary |
||
| Line 1: | Line 1: | ||
== Optimizer == | == Optimizer == | ||
The SUMO toolbox contains a whole hierarchy of optimizers that can be used by several | |||
components (for instance the InfillSamplingCriterion). Currently the following optimization methods are supported: | |||
=== DirectOptimizer === | === DirectOptimizer === | ||
DIviding RECTangles is a global optimization algorithm based on the divide & conqueror philosophy. It is quite efficient at finding accurate, global optimums of computationel cheap functions. It is one of the standard choices to optimize the [[SampleSelector#InfillSamplingCriterion|expected improvement function]]. | |||
=== MatlabOptimizer === | === MatlabOptimizer === | ||
This optimizer is a wrapper around Matlab's own optimization methods: fminunc and fmincon. | |||
Revision as of 16:17, 19 February 2008
Optimizer
The SUMO toolbox contains a whole hierarchy of optimizers that can be used by several components (for instance the InfillSamplingCriterion). Currently the following optimization methods are supported:
DirectOptimizer
DIviding RECTangles is a global optimization algorithm based on the divide & conqueror philosophy. It is quite efficient at finding accurate, global optimums of computationel cheap functions. It is one of the standard choices to optimize the expected improvement function.
MatlabOptimizer
This optimizer is a wrapper around Matlab's own optimization methods: fminunc and fmincon.
