The Application Research Of Parallel Adaptive Genetic Algorithm In The Big Data Processing Of Materials Science

Big data processing technology has been a hot topic in recent years,it is widely used in physics,astronomy,finance and other science research areas.In recent years,big data processing technology science has been also effectively applied in the field of materials.Since2011 Japan's Fukushima nuclear power plant explosion,many scientists believed that Si C material is a kind of more stable cladding material,The main work of this project is optimizing key functions of SiC material large data processing using improved genetic algorithm,to make the calculated data of SiC material more accurate.As the theoretical base of improved genetic algorithm,this paper introduces the basic principles and the main processes of genetic algorithm,commonly used coding methods,fitness function and structure of the way crossover and mutation methods.Given the adaptive genetic algorithm has many advantages,the paper chosen adaptive genetic algorithm as the initial algorithm to improve,and detailed analysis the advantages and disadvantages of the basic principles of a representative of the two adaptive genetic algorithm.In order to further curb the local optimization and premature convergence problem in adaptive genetic algorithm,the paper will apply a new crossover method into the traditional adaptive genetic algorithm,and improved the formula of adaptive genetic algorithm,an arithmetic based on adaptive genetic algorithm crossover,the optimization results of new algorithm has improved greatly than the conventional adaptive genetic algorithm.In Tersoff potential energy function,the parameters to be optimized as much as 30,in order to optimize to achieve the desired results,it is necessary to use a large-scale population,while increasing the number of iterations,which brings the algorithm execution time is too long.To solve this problem,we introduced the idea of parallel computing to the improved adaptive genetic algorithm,the proposed method based on arithmetic crossover parallel genetic algorithm,and use this algorithm to optimize Tersoff potential energy function on “Dawning-5000A”supercomputer,then implement the actual calculation,the calculated results are compared with the DFT calculations,the results demonstrate the effectiveness of the optimization algorithm.At the same time,the convergence of several genetic algorithms,running time and other data,is compared with the new genetic algorithm.Finally,we use the improved genetic algorithm developed SiC-He potential energy function and calculate the key characteristics ofthe material to prove the validity of the new algorithm.