基于Spark的并行遗传算法在温稠密物质状态方程构建中的应用

Application of Spark-based parallel genetic algorithm to the construction of state equation of warm dense matter

针对温稠密物质状态方程参数的确定问题,提出一种基于Spark的并行遗传算法,把参数问题转化为函数最优化问题。使用Scala语言编写算法程序,从温稠密物质的数据库中选取部分温度和压强数据,再用遗传算法产生新的参数值,采用最小二乘法确定适应度函数,进而确定温稠密物质状态方程的参数。与矩阵法相比,并行遗传算法运算结果的精度和稳定性更优。实验结果表明,基于Spark的并行遗传算法可以用来解决温稠密物质状态方程参数确定的问题。

In this paper, to determine the parameters of the state equation of warm dense matter, a parallel genetic algorithm based on Spark is proposed. The parameter problem is transformed into the function optimization problem. The algorithm program is written in Scala language, and the physical experiment is selected from the database of warm and dense matter. Part of the temperature and pressure data is obtained and genetic algorithm is used to generate new parameter values. The least squares method is used to define the fitness function to determine the parameters of the temperature equation for the dense matter. Compared with the matrix method, the precision and stability of the parallel genetic algorithm results are better. The experimental results show that the parallel genetic algorithm based on Spark can be used to determine the parameters of the state equation of warm and dense matter.