Home

Wiki

How use?

Consider a simple example. We have a target function - Rastrigin function:

We want to find minimum of function with constraints:

We know that function has global minimum equal 0 in point (0,0). We trying to find approximate solution.

At first, we define target function and constraints.

Func<double[], double> func = (x) => 20 + (x[0] * x[0] - 10 * Math.Cos(2 * Math.PI * x[0])) +
               (x[1] * x[1] - 10 * Math.Cos(2 * Math.PI * x[1]));

double[] constr1 = { -5.12, -5.12 };
double[] constr2 = { 5.12, 5.12 };

Each pair (constr1[i], constr2[i]) determine bounds for coordinate.

At second, we define instances of methods.

IOptimizer[] opts = {
          new BBBCOptimizer(),
          new FireworksOptimizer(),
          new GEMOptimizer()
};

Remarks. Methods use random number generator and if you have own implementation, then can use overloaded constructors. Main requirement is implement corresponding interface.

Now, we define parameters for methods.

// Distance between points need for Fireworks method.
// It is squared Euclidean distance.
Func<PointND, PointND, double> distance = (a, b) => (a[0] - b[0]) * (a[0] - b[0]) + 
      (a[1] - b[1]) * (a[1] - b[1]);

object[] parameters =  {
      new BBBCParams(20, 100, 0.4, 0.5),
      new FireWorksParams(20, 50, distance, 20),
      new GEMParams(1, 20, 50, 2 * Math.Sqrt(2), 100)
};

Remarks. Recommendation by choosing parameters is no. You can test methods with different combinations parameters and choose optimum.

At finally, we search solution optimization problem.

string[] names =
 {
    "BBBC",
    "Fireworks",
    "GEM"
};

// The definition for general parameters: target function, constraints.
GeneralParams param = new GeneralParams(func, constr1, constr2);

for (int i = 0; i < opts.Length; i++)
{
   opts[i].InitializeParameters(parameters[i]);
   opts[i].Optimize(param);

   Console.WriteLine($"Method: {names[i]}.");
   Console.WriteLine(opts[i].Solution);
   Console.WriteLine();
}

Remarks. The convergence is not proven. Consequently, you can get more best or worst solution. For more confidence you need run several times method and choose best solution.

Full code

Here full code of example.