Iterative local search (ILS)
Iterative local search escapes local optima with perturbation and acceptance criteria — visualized alongside tabu search on continuous landscapes.
Iterative local search (ILS) improves a solution by alternating intense local search with controlled perturbation. After a local optimum is found, the incumbent is shaken just enough to leave its basin, then local search runs again. An acceptance rule decides whether the new local optimum replaces the incumbent — a simple structure that competes well on many combinatorial problems.
Related ideas in the demo
The AlgoScape optimization view also includes tabu search, which forbids recently visited moves so the search does not cycle. Watching ILS and tabu side by side makes the tradeoff concrete: how hard you perturb, how long you forbid moves, and when you accept a worse candidate.
Try it
Explore the optimization algorithms demo (ILS and tabu search on a searchable landscape).