Optimization Methods
Numerical methods available for an Optimization approach.
| Method | Input Variable Model Restrictions | Input Variable Constraint Restrictions | Distribution Role | # of Objectives | Exploration Type | Accuracy | Efficiency | Comments |
|---|---|---|---|---|---|---|---|---|
| Adaptive Response Surface Method (ARSM) | None | None | Deterministic | Single | Local | ✩ | ✩✩✩ | Default method for single objective problems. |
| ARSM-Based Sequential Optimization and Reliability Assessment (SORA_ARSM) | Continuous only | Input variable constraints are not allowed | Probabilistic | Single | Local | ✩ | ✩✩✩ | More efficient than Sequential Optimization and Reliability Assessment, but not as
accurate. It is not recommended to use ARSM-Based Sequential Optimization and Reliability Assessment with a Fit. |
| Genetic Algorithm (GA) | None | None | Deterministic | Single | Global | ✩✩ | ✩ | Significantly expensive. Use Genetic Algorithm if the simulation is affordable or if you have a good Fit. |
| Global Response Search Method (GRSM) | None | None | Deterministic | Single or Multiple | Global | ✩✩✩ | ✩✩ | Default method for multi objective problems. Preferred method when the number of design variables is large. Optimizing can start with just a few number of points independent of the number of design variables. |
| Method of Feasible Directions (MFD) | Continuous only | Input variable constraints are not allowed | Deterministic | Single | Local | ✩✩ | ✩✩ | May work more efficiently for problems with a large number of constraints. |
| Multi - Objective Genetic Algorithm (MOGA) | None | None | Deterministic | Multiple | Global | ✩✩ | ✩ | Significantly more expensive. Use Multi - Objective Genetic Algorithm if the simulation is affordable or if you have a good Fit. |
| Sequential Optimization and Reliability Assessment (SORA) | Continuous only | Input variable constraints are not allowed | Probabilistic | Single | Local | ✩✩✩ | ✩ | Use if the simulation is affordable or if you have a good Fit. |
| Sequential Quadratic Programming (SQP) | Continuous only | Input variable constraints are not allowed | Deterministic | Single | Local | ✩✩✩ | ✩✩ | Use if the simulation is affordable or if you have a good Fit. |
| System Reliability Optimization (SRO) | None | None | Probabilistic | Single | Global | ✩✩✩ | ✩✩ | Default method for probabilistic problems. In robust optimization, it provides the trade-off between the nominal value and variance of the objective. |
| Xopt (User-Defined Optimization Engine) | None | None | Deterministic | Single |