CMAEvolutionStrategyOptimizer.InitialStepSize Property

public double InitialStepSize { get; set; }

Public Property InitialStepSize As Double
	Get
	Set

public:
property double InitialStepSize {
	double get ();
	void set (double value);
}

member InitialStepSize : float with get, set

Property Value

The initial step size determines the scale of the initial search distribution and influences how far the candidate solutions can move in the search space during the initial iterations. A well-chosen initial step size can significantly impact the convergence speed and the ability of the algorithm to escape local optima. If the step size is too small, the algorithm may converge slowly or get stuck in local optima. Conversely, if the step size is too large, the algorithm may overshoot good solutions and fail to converge.

When choosing an optimal initial step size, consider the scale of the problem and the typical distance between good solutions. A common heuristic is to set the initial step size to a fraction of the typical range of the decision variables. For example, if the decision variables are expected to vary between 0 and 10, an initial step size of 0.5 to 1.0 might be appropriate. Additionally, it is often beneficial to experiment with different values and observe the algorithm's performance, adjusting the step size based on empirical results.

The default value is 0.5.

Reference