A New DIRECT-Type Algorithm Based on Bisection of Rectangles and Diagonal Sampling with the Pareto Approach

A novel DIRECT-type algorithm is proposed to tackle box-constrained optimization problems. The algorithm incorporates bisection partitioning and diagonal sampling procedures, utilizing the Pareto approach to identify potential hyperrectangles. Furthermore, each hyperrectangle’s size is determined based on the length of its longest side, in accordance with the infinity norm. This combination enhances convergence speed, facilitating the generation of a global optimal solution. The proposed algorithm’s effectiveness is evaluated through numerical experiments, with detailed results demonstrating its efficacy