The optimization by particulate swarms ( OEP or PSO in English) is a Métaheuristique of optimization, invented by Russel Ebenhart (engineer in electricity) and James Kennedy (socio-psychologist) in 1995.

This algorithm is inspired in the beginning by the world by the alive one. It is based in particular on a model developed by the biologist Craig Reynolds at the end of the years 1980, making it possible to simulate the displacement of a group of birds. Another source of inspiration, asserted by the authors, is the Socio-psychology.

This method of optimization is based on the collaboration of the individuals between them. Besides it has similarities with the algorithms of colonies of ants, which are also based them on the concept of Car-organization. This idea wants that a group of not very intelligent individuals can have a complex total organization.

Thus, thanks to very simple rules of displacement (in the space of the solutions), the particles can converge gradually towards a local minimum. This métaheuristique seems to however better function for spaces in continuous variables.

At the beginning of the algorithm each particle is thus positioned (by chance or not) in the Espace of research of the problem. Each iteration makes move the particles according to 3 components:

1. Its current speed,
2. Its best P_i solution,
3. the best solution obtained in its P_g vicinity.

That gives the following equation of motion:

V_k+1=a·V_k+b₁ (P_i-X_k) +b₂ (P_g-X_k)

X_k+1=X_k+V_k+1

With:

b₁ drawn by chance in : b₂ drawn by chance in ==Liens internes==

• Boids

External bonds

• Particle Swarm Central
• Particle Swarm Optimization