SwarmOps Pattern Search (PS)
SwarmOps algorithms are being removed and may not be compatible with future releases of Analyzer. Users should be cautious when using them in models
Note: SwarmOps algorithms are being removed and may not be compatible with future releases of Analyzer. Users should be cautious when using them in models
Version
SwarmOps 3.0
Description
SwarmOps Pattern Search is a constrained optimization that is a variation of Hooke-Jeeves Pattern Search. Without using gradient information, the algorithm may perform better for noisy objective functions. The algorithm is expected to be less efficient for smooth object functions compared to other gradient based methods.
More Information
It starts with a randomly selected design. For a randomly selected dimension, it perturbs the current design. If the perturbed design improves the current design, it becomes the new current design. Otherwise, the step size is reduced and a new design is tried again. These steps are repeated until any of the termination criteria is met.
References
SwarmOps Manual (pp. 15)
Control Parameters
| Name | Default Value | Description |
|---|---|---|
| Optimization Parameters | ||
| Seed | Random number generator seed value (optional). Specifying the same seed value between two different optimization runs would help generate identical results provided all other parameters stay the same. This is a good way to analyze the effect of different parameters on the optimization. | |
| Stopping Criteria | ||
| AbsoluteConvergenceTolerance | 0.0001 | Maximum absolute change in fitness value between successive evaluations to indicate convergence. The value specified must be greater than 0. |
| ConsecutiveFunctionEvaluations | 200 | The number of consecutive iterations for which the convergence criteria must be met to indicate convergence. Thus option must have a positive integer value. |
| MaxFunctionEvaluations | 3000 | Maximum number of iterations. Thus option must have a positive integer value. |