Particle Swarm Global Optimization Search Algorithm

Abstract

Particle Swarm Optimization (PSO) was originally intended for simulating social behaviour, as a stylized representation of the movement of organisms in a bird flock or fish school (en.wikipedia.org/wiki/Particle_swarm_optimization and all references therein). “The key idea is to have a swarm of interacting particles, each representing a candidate solution to a given optimization problem. Thus, particles are embedded in the search space and explore the solution space by flying around. Moreover, the particles are also attracted to high fit regions located by other particles” (Luo et al., J Am Chem Soc 133(40):16285–16290, 2011). Particle has position, velocity and acceleration, but we may look at it as without volume and mass; the PSO derivative-free global optimization algorithm can be used in molecular crystal structure prediction and materials’ construction. One successful promise of PSO in crystal structure determination/identification or in designing the multi-functional materials is the package CALYPSO (Guillaume et al., Nat Phys 7(3):211–214, 2011; Li et al., J Phys Chem C 114(49):21745–21749, 2010; Liu et al., Phys Lett A 375(3):771–774, 2011; Luo et al., J Am Chem Soc 133(40):16285–16290, 2011; Lv et al., Phys Rev Lett 106:15503–15506, 2011; Wang et al., Phys Rev B 82(9):094116, 2010; Wang et al., J Chem Phys 137(22):224108, 2012; Zhao et al., ACS Nano 5(9):7226–7234, 2011; Zhu et al., Phys Rev Lett 106(14):145501–145504, 2011), which requires only chemical compositions and their given external conditions (e.g., pressure and temperature).