This document studies time-varying partially connected topologies for particle swarm optimization. It compares different population structures, including local best (lbest), global best (gbest), and a dynamic Von Neumann neighborhood structure. The results show that the dynamic partially connected structure offers the best balance, finding solutions faster and with fewer iterations than the static lbest and gbest models. The optimal grid size is around twice the number of particles. Overall, the dynamic topology performed best in successfully finding solutions.