PSO Neighborhood Topology: Global-Best vs Ring
Watch a real Particle Swarm Optimization run search a Rastrigin fitness landscape, and switch between global-best (star) and local-best (ring) neighborhood topologies to see the classic exploration-vs-premature-convergence trade-off.
Particle Swarm Optimization borrows its update rule from flocking behavior: every candidate solution ("particle") nudges its velocity toward the best position it has personally found and toward the best position known within its neighborhood, then moves. This simulator runs a real PSO search over the Rastrigin benchmark function, rendered as a 3D terrain riddled with local minima around one true global optimum, and lets you switch the neighborhood structure that decides which particles influence each other — a fully-connected global-best swarm versus a sparse local-best ring — to see firsthand how topology trades convergence speed against resistance to getting stuck in a local optimum.
Watch a real Particle Swarm Optimization run search a 3D Rastrigin fitness landscape, and switch between global-best (star) and local-best (ring) neighborhood topologies to see the classic exploration-vs-premature-convergence trade-off.
3D · Three.js / WebGL renderer · 60 FPS target · runs fully client-side, no install