Pony package for Particle Swarm Optimization

An implementation of the Particle Swarm Optimization algorithm[1] in Pony, with support for dissipative variations[2] and inertia weight strategies[3]. PSO is a population based global stochastic optimization technique inspired by social behavior of bird flocking or fish schooling.

Building

This software depends on the development version of ponyc, the Pony language compiler. You should build and install the compiler from ponyc's source repository.

Execute ./build.sh -e to compile all the examples. You will find the executables in the bin/ folder.

Usage

use "pso"

actor Main
  new create(env: Env) =>

    let params = recover val
        let p = SwarmParams(2)
        p.max = [5000, 5000]
        p.min = [-5000, -5000]
        consume p
    end

    let sw = Swarm(params,
        SwarmLog(env),
        object is FitnessFunc
            fun apply(x: Array[F64]): F64 ? =>
                (x(0) - 200).abs() + (x(1) - 200).abs()
        end)

    sw.solve()

Supported Parameters

c1 : Cognitive factor. Usually c1 equals to c2 and ranges from [0, 4].

c2 : Social factor.

cv : Chaos velocity factor, in the range [0, 1].

cl : Chaos location factor, in the range [0, 1].

max : Maximum values of the search space.

min : Minimum values of the search space.

vmax : Maximum velocity.

particles : Number of particles. Typical range is [20, 40]. Actually for most of the problems 10 particles is large enough to get good results. For some difficult or special problems, one can try 100 or 200 particles as well.

precision : Number of decimal figures per dimension.

inertia : Inertia weight function.

stagnation : Maximum iterations without a global fit. Termination condition.

target : Target cost value for the optimization problem. Termination condition.

iterations : Maximum number of iterations. Termination condition.

Examples

Take a look to the source code in the examples/ folder; v.g., running bin/sphere solves the sphere problem for 60 dimensions and prints an output close to this:

PSO Sphere function
f(x) = sum(x[]^2)

Execution Results
----------------
Best            0
X1:             0
X2:             0
X3:             0
X4:             0
...
X60:            0
Epoch:          328
Reason:         Target

References

[1] J. Kennedy and R. C. Eberhart. “Particle swarm optimization,” Proc. IEEE Int. Conf. on Neural Networks, pp. 1942-1948, 1995.

[2] Xiao-Feng Xie, Wen-Jun Zhang and Zhi-Lian Yang. “Dissipative particle swarm optimization,” in Evolutionary Computation, 2002. CEC '02. Proceedings of the 2002 Congress on , vol.2, no., pp.1456-1461, 2002

[3] Bansal, J.C., Singh, P.K., Saraswat, M., Verma, A., Jadon, S.S. and Abraham, A. “Inertia Weight strategies in Particle Swarm Optimization,” in Nature and Biologically Inspired Computing (NaBIC), 2011 Third World Congress on , vol., no., pp.633-640, 19-21 Oct. 2011