Research And Application Of Distributed Simulation System For Computational Sociology

The study of sociology is evolving in the direction of computational sociology.Researchers can more accurately evaluate the relevant studies in this field by introducing computational software to conduct simulation calculations.NetLogo is a commonly used programmable modeling software among scientific researchers.Sociologists often use its spatial function to explore group behavior studies.However,when the software performs large-scale parameter scans,it often encounters problems,such as insufficient computing power,limited number of tasks,limited number of simulation results,and data loss,affecting the validity and reliability of larger scale group behaviors.To address the above issues,this dissertation provides distributed computing architecture support based on the native NetLogo software by utilizing Python to realize the distributed simulation capability of NetLogo,and thereby support larger-scale model simulations.The main research work and contributions of this study include:(1)The feasibility of distributed simulation technology in computational sociology is analyzed and verified.The py NetLogo algorithm is optimized to reduce the waste of computing resources in simulation.The feasibility of Python calling the NetLogo API to operate the model and obtain model information was verified.The feasibility of NetLogo distributed execution was analyzed.The technical solution of Ray distributed framework was verified through the remote operation of local NetLogo software and the provision of remote task services.(2)Design and development of distributed simulation framework for computational sociology.Based on NetLogo software,a dynamic distributed calculation framework in the sociological simulation was designed to solve the problem of computing power scalability,support multi-level splitting,and distributed execution of simulation tasks.By redesigning the result storage format,The ability to dynamically add and restart tasks was implemented.Experimental results show that the distributed computing system has an excellent performance in large-scale model simulation,with a much lower calculation time than the original software simulation time.(3)Design and development of a visual research platform for computational sociology.A visualization research platform was designed and developed.The platform uses a dynamic distributed calculation framework of sociological simulation based on NetLogo software,designing and developing functional modules for optimal parameter calculation and robustness analysis.It also implements functions,such as verifying the completeness of NetLogo spatial data and data integration.