Date of Graduation
Statler College of Engineering and Mineral Resources
Lane Department of Computer Science and Electrical Engineering
Matthew C. Valenti
Vinod K. Kulathumani
Yenumula V. Reddy
Monte Carlo simulation is an effective way to analyze models of sophisticated problems, but often suffers from high computational complexity. Distributed computing is an effective technology that can be used for compute-intensive applications, such as Monte Carlo simulation. The goal of this thesis is to combine the concepts of Monte Carlo simulation and distributed computing in an effort to develop an efficient system capable of rapidly executing computationally-demanding simulations.;When distributed computing is used to support the simulations of multiple users, a scheduling algorithm is required to allocate resources among the users' jobs. In this thesis, a scheduling algorithm is developed that is suitable for Monte Carlo simulation and utilizes the available distributed-computing resources. The unified framework for scheduling is capable of accommodating classic scheduling algorithms such as equal job share, first-in first-out (FIFO), and proportional fair scheduling. The behavior of the scheduler can be controlled by just three parameters. By choosing appropriate parameter values, individual users and their jobs can be assigned different priorities. By introducing an appropriate analytical model, the role of these parameters on system behavior is thoroughly investigated. Using insights obtained by studying the analytical model, a complete distributed Monte Carlo system is designed and presented as a case study.
Bommagani, Aruna Sri, "Distributed Monte Carlo Simulation" (2011). Graduate Theses, Dissertations, and Problem Reports. 675.