| Dynamics on complex networks is about the study on the dynamics of networked systems wherein the graphs are used to describe the relationships among numerous agents.The study on dynamics on complex networks involves lots of different disciplines such as artificial intelligence,mathematics,computer science,and information theory,and has both major theoretical and practical significance.Agent-based modeling and simulation are one of the most important means to study the dynamics on complex networks nowadays.By computer modeling and simulation tools,the dynamic behaviors on complex networks can be studied intuitively.However,the common modeling and simulation platforms are not friendly for users,have a high learning cost and low computing efficiency,and thus cannot meet the requirements of scholars and students in the field of complex systems as well as of large-scale modeling and simulation of dynamics on complex networks.In addition,due to the complexity of the connections between individuals on complex networks,some dynamics cannot be quantitatively characterized and analyzed by theoretical methods.Therefore,it is necessary to design and develop a virtual platform for modeling and simulations of dynamics on complex networks with advantages,such as more friendly user interface and high computing efficiency.To accomplish this,we analyze the advantages and disadvantages of the agent-based modeling and simulation platforms available nowadays,and then consider the realistic requirements of users,and finally design and develop a platform based on the framework of Py Qt.Specifically,by using the simulation techniques for dynamics on complex networks,we obtain the following three main results:First,the platform is developed by layered architecture,which composed of three major layers:presentation layer,business logic layer and data access layer.Each layer involves a relatively independent sub-problem.The collaboration between different layers not only simplifies the development process,but also enhances the scalability as well as maintainability of the platform.And we design several fundmental functional modules including module for the visualization of dynamics on complex networks,module for the display of statistical variables and module for tool expansion.Furthermore,in order to provide users with comfortable user-experience,we also enable the software to be cross-platform,multi-threaded and parallel computing.By module diagram,we introduce and design each functional module in detail,which construct a solid foundadtion for the realization of the platform.Secondly,the layout of the functional modules on the main interface is expounded,which reduces the user's learning cost and is easy to operate,and gives the experimental results of each functional module.Herein,the realization functions of each new module are as follows: The platform is applicable in the mainstream operating systems,including Windows,mac OS,Linux and other system environments.The multi-threaded operation provides the function of synchronously building and modifying projects.In the framework of CUDA,GPU-based parallel computing is used to optimize the algorithms for the dynamics on the regular networks,which largely accelerates the efficiency of computation as well as optimizes user experiences.Finally,the Jmeter test tool is used to test the function and performance of the platform.The test results show that the functions designed by this platform can run smoothly,and can achieve the design goals of the platform.In this platform,the experimental efficiency has been significantly improved and the workload of users has been reduced,which thus meet the actual needs of daily experiments. |
PDF Full Text Request