面向两类典型分析仿真的多样本任务调度工具的设计

Design of a Multi-sample Scheduling Tool for Two Classes of Analytic Simulations

当在多台计算机上调度很多样本并行运行时,采用手工操作效率低且极易出错。设计一个自动化的任务分发与调度工具可以有效地提高仿真运行效率。蒙特卡罗仿真和粒子群优化算法仿真是两类典型的分析仿真,介绍了面向这两类仿真的多样本任务调度工具的实现原理和关键技术,这些方法具有很好的通用性和可扩展性,可以广泛应用于解决各类仿真任务的自动分发与调度问题。

When a large number of samples are being scheduled to run in dozens of computers,it is low in efficiency and easy to make mistakes for manual operation.Thus,an automatic tool for deploying and scheduling tasks can improve the analytic simulation＇s execution efficiency.The Monte Carlo simulation and the Particle Swarm Optimization algorithm simulation are two kinds of classical analytic simulations.This research introduces the rationales and key techniques for the multi-sample scheduling tool based on both simulations,which can be generally and widely used in resolving varieties of problems about the automatically deploying and scheduling of simulation tasks.