Sig Numerical Optimization / Optimization methods comparison
From OpenFOAMWiki
Optimization Methods comparison
The following table aims to provide an overview of the main optimization methods included in Dakota, OpenFOAM and other optimization software.
| Capabilities | Description & usage | Methods | Tutorials & examples |
|---|---|---|---|
| Parameter study | Simplest method to explore parameter space. Useful for simple studies with defined, repetative structure. | vector / list / centered / multidimensional | ... |
| Design of Experiments (DOE) / Design and Analysis of Computer Experiments (DACE) | Sophisticated methods to explore parameter space. Recommended when global space-filling set of samples (multiple variables) is needed. DOE is largely used for physical experiments and tends to place samples to space's extremes, while DACE is more space-fulflling | Central Composite Design / Box-Behnken sampling / Monte Carlo (random) sampling / Latin hypercube sampling / Orthogonal array / Orthogonal array – Latin Hypercube Sampling / Grid Design / Psuade Moat | Blunt Body case |
| Optimization | |||
| Non-linear Least Squares | |||
| Surrogate-Based Minimization | |||
| Adjoint Shape Optimization |
