Particle Swarm Optimization (PSO) in C#


Particle Swarm Optimization (PSO) is a computational method for optimization of a problem by simulating a set of moving particles that move around a search-space. Now let’s say that you want to find the (global) minimum of the square function: f(x) = x[0]^2 + x[1]^2.

 var solverConfig = PSOSolverConfig.CreateDefault(
            numberParticles: 500,
            maxEpochs: 1000,
            lowerBound: new double[] {-20, -20},
            upperBound: new double[] {10, 10},
            isStoppingCriteriaEnabled: false); // If not enabled, the optimization, will run over the total number of epochs (maxEpochs).
            
  Func<double[], double> func = x => (x[0] * x[0]) + (x[1] * x[1]);
  var solver = new ParticleSwarmMinimization(func, solverConfig);

  solver.OnAfterEpoch += (s, d) =>
        {
            Console.WriteLine($"Iteration: {d.Iteration}\n" +
                              $"Best fitness: {d.BestFitness:##.000} \n" +
                              $"Best global position: " +
                              $"x1: {d.BestPosition[0]:##.000}" +
                              $"x2: {d.BestPosition[1]:##.000}");
        };

   var solution = solver.Solve();
   Assert.AreEqual(0, solution.BestPosition[0], 1E-6);
   Assert.AreEqual(0, solution.BestPosition[1], 1E-6);

The source code can be found on https://github.com/isachpaz/OptimizationSharp

Leave a Reply

Your email address will not be published.