Meta-heuristic methods are global optimization algorithms which are widely used in the engineering issues, nowadays.
The main problem with the classical optimization algorithms is their slow rate of convergence to time-consuming
mathematical calculations. In this paper, a new stochastic search for optimization is presented using variable variance
Guassian distribution sampling. The main idea of searching in this algorithm is to regenerate new samples around each
solution with a Guassian distribution. The proposed algorithm is applied to four popular test functions for
optimizations (Griewank, Booth, Rosenbrock, Rastrigin). Numerical simulations have revealed that the new presented
algorithm outperformed simulated annealing and genetic algorithms.
کلید واژگان :
Optimization, Gaussian distribution, covariance matrix, stochastic search, variance reduction, Probability Density Function (PDF, hereafter)
