Speed-up simulated annealing by parallel coordinates

High computational cost is an obstacle to the applications of simulated annealing, especially for moderate and high dimensional problems. Inspired by multidimensional visualization techniques, the notion of parallel coordinates system, or parallel coordinates in short, is introduced into the optimization research area to speed up the convergence rate of simulated annealing. Numerical studies demonstrate that the proposed method can converge to global solutions with reduced computational cost in terms of both the number of function evaluations and CPU time.