An Integrated Stochastic Method for Global Optimization of Continuous Functions

Stochastic methods have attracted growing interest in the recent past as they require less computational effort to provide the global optima. Some of the well known methods are Genetic Algorithm (GA), Differential Evolution (DE) and Tabu search (TS). Each of these methods has a unique feature of escaping from the local minima and/or improved computational efficiency. Though each of these methods has its own advantage(s), they may be trapped in the local minima at times because of the highly non-linear nature of the objective function. In this work, an integrated stochastic method (ISM) is proposed by identifying and then integrating the strong features of DE and TS. A local optimization technique is used at the end to improve the accuracy of the final solution and computational efficiency of the algorithm. The performance of ISM is tested on many benchmark problems and challenging phase equilibrium calculations. The former contain a few to hundreds of local minima whereas the latter has comparable minima. The results show that the performance of ISM is better compared to DE and TS.