One of the primary complaints toward particle swarm optimization (PSO) is the occurrence of premature convergence because the diversity of the particles rapidly comedown. In order to improve diversity of the particles, a dynamical particle swarm optimization (DPSO) is proposed for global optimization, which adds a memory mechanism conceptually derived from the principle free entropy minimization. DPSO is that all particles in a swarm are running and searching with their swarm evolving driven by a new selecting mechanism. This mechanism simulates the principle of molecular dynamics to keep the diversity of the particles and obtain a good balance between aggressive exploration and detailed search. In order to verify the effectiveness of the propose scheme, DPSO is applied to solving some typical global optimization problems. The experimental results show the DPSO is feasible and reliable.