SwarmOps Differential Evolution with Dithered Parameters (DESuite)

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 DESuite is a constrained optimizer that handles continuous design variables. This is a variation of SwarmOps DE that introduces a random weight when combining current designs to generate a next candidate design. The idea is to make the optimizer less susceptible to the choice of the weighting factor.

More Information
This is a variation of SwarmOps DE that uses a random differential weight F, generated from a random distribution of:

The idea of this modification was to make the algorithm less susceptible to a particular choice of F. There is an option to control how often to compute new F. By default it is computed for each generation. Other options include per agent (design) or per dimension. Another option is to control the crossover scheme. By default, the best design is used as the base design. If the random option is selected, a randomly selected design is used for the base design.

References

Tuning & Simplifying Heuristical Optimization, Pedersen, E. H. P., PhD thesis, University of Southampton, 2010, pp. 68-69

Control Parameters

Name	Default Value	Description
Optimization Parameters
CR	0.9	Probability of crossover being performed. Thus the option must have a value between 0 and 1.
Crossover	Best	Crossover scheme. Best: Use the best design for base Random: Use a randomly selected design for base
Dithering	Generation	How often the differential weight is computed Generation: per each generation Agent: per each agent Dimension: per each dimension
FMid	0.75	Average value of differential weight. Option must have a value greater than 0.
FRange	0.25	Range of differential weight. This value must be greater than 0.
NP	14	Number of agents or population size. Option must have a positive integer value.
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	1E-05	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 absolute or relative convergence criteria must be met to indicate convergence. Thus option must have a positive integer value.
MaxFunctionEvaluations	2000	The maximum number of iterations. Thus option must have a positive integer value.