Directional Search Algorithm and its Utilization of Simulated Annealing and Expand-Retract
Beom Jin An, Alexander S. Mentis · 2019
We present the Directional Search Algorithm (DSA), a novel metaheuristic optimization algorithm combining a highly-simplified version of Particle Swarm Optimization (PSO) with ideas adapted from simulated annealing, and a dynamic "Expand-Retract" method, a novel technique to balance exploration and exploitation. We compare the speed and accuracy of DSA to PSO in finding the global minimum of some common artificial landscapes for single-objective optimization and find that DSA outperforms PSO for both criteria on the selected functions.