CMAEvolutionStrategyOptimizer.BoundaryHandling Property

public BoxConstraintHandlingMethod BoundaryHandling { get; set; }

Public Property BoundaryHandling As BoxConstraintHandlingMethod
	Get
	Set

public:
property BoxConstraintHandlingMethod BoundaryHandling {
	BoxConstraintHandlingMethod get ();
	void set (BoxConstraintHandlingMethod value);
}

member BoundaryHandling : BoxConstraintHandlingMethod with get, set

Property Value

The BoundaryHandling property determines how the optimizer handles candidate solutions that fall outside the defined boundaries of the search space. Different boundary handling strategies can significantly impact the performance and behavior of the optimization process.

When set to None, the optimizer does not enforce any boundary constraints, allowing candidate solutions to explore the entire search space without restrictions. This can be useful for unconstrained optimization problems but may lead to invalid solutions in constrained scenarios.

The Reflect strategy reflects candidate solutions that exceed the boundaries back into the feasible region. This approach helps maintain diversity in the population and prevents solutions from getting stuck at the boundaries. It is particularly useful for problems with well-defined boundaries.

The Resample strategy resamples candidate solutions that fall outside the boundaries until they lie within the feasible region. This method ensures that all solutions remain valid but may require additional computational effort due to repeated sampling. It is suitable for problems where boundary constraints are strict and must be enforced at all times.

Choosing the optimal boundary handling strategy depends on the nature of the optimization problem and the importance of boundary constraints. For problems with strict constraints, the Resample strategy is recommended. For problems where maintaining diversity is crucial, the Reflect strategy may be more effective. Experimenting with different strategies and observing their impact on the optimization process can help determine the best approach for a given problem.

Reference