Does Complex Metaheuristic Out-Perform Simple Hill-Climbing for Optimization Problems? A Simulation Evaluation

In this paper we compare the performance of metaheuristic methods, namely simulated annealing and Tabu Search, against simple hill climbing heuristic on a supply chain optimization problem. The benchmark problem we consider is the retailer replenishment optimization problem for a retailer selling multiple products. Computation and simulation results demonstrate that simulated annealing and Tabu search improve solution quality. However, the performance improvement is less in simulations with random noise. Lastly, simulated annealing appears to be more robust than Tabu search, and the results justify its extra implementation effort and computation time when compared against hill climbing.