基于时域多分辨率的非球形气溶胶散射并行计算模型设计及验证

Design and Validation of Parallelized Computational Model for Nonspherical Aerosol Scattering Based on Multi-Resolution Time-Domain

为解决串行时域多分辨率(MRTD)散射模型运行时间长和内存消耗大的问题,基于消息传递接口(MPI)技术设计了一种非球形气溶胶散射并行计算模型。介绍了MRTD散射模型的基本框架和2种并行数据通信方案,并基于MPI重复非阻塞通信技术实现了MRTD散射模型的并行化设计;搭建了网络并行计算平台,实现了模型的并行化计算。将MRTD散射模型与Mie散射模型、T矩阵法进行了对比,验证了并行MRTD散射模型的计算准确性。结果表明,MRTD模型可较准确地模拟非球形粒子散射特性,并行计算技术可显著提高计算效率;电磁场分量同时交换的并行设计方案的计算效率略高于仅交换磁场分量的方案;通过增加中央处理器核数,程序的并行加速比随之增大,但单核运行效率却略有降低。随着粒子尺度参数的增大,单核计算效率随之增加,复折射率的改变并不会显著影响并行计算效率。

In order to decrease computational time and computer memory consumed by the serial multi-resolution time-domain（MRTD）scattering model,aparallel calculation model for nonspherical aerosol scattering is proposed based on message passing interface（MPI）technique.A basic frame of the MRTD scattering model and two parallelization data communication schemes are introduced,and the parallel design for MRTD scattering model is achieved by MPI repeated non-blocking communication technique.A network parallel computation platform is established for the parallel calculation.To validate the computational accuracy of the MRTD scattering model,the simulation results of parallelized MRTD scattering model are compared with that of Mie scattering model and T Matrix method.The results show that the MRTD model can calculate the scattering parameters of nonspherical particles accurately.Parallel computational technique can improve computation efficiency notably.The computationefficiency of the parallelization design scheme that exchanges electric field and magnetic field simultaneously is higher than that of the scheme that exchanges magnetic field simply.With increasing the number of central processing unit cores,the parallel acceleration ratio of program is incresing,while the computational efficiency of single processor is slightly decreasing.With the increasing of particle size parameter,the computational efficiency of single processor is increasing as well.It also can be found that the change of complex refractive index has no notable influence on parallel computational efficiency.