Zhang Kai. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2005. / Includes bibliographical references (leaves 99-102). / Abstracts in English and Chinese. / Chapter 1 --- Introduction --- p.6 / Chapter 1.1 --- The Generalized PML Theory --- p.6 / Chapter 1.1.1 --- Background --- p.6 / Chapter 1.1.2 --- Derivation --- p.8 / Chapter 1.1.3 --- Reflection Properties --- p.11 / Chapter 1.2 --- Unified Formulation --- p.12 / Chapter 1.2.1 --- "Face-, Edge- and Corner-PMLs" --- p.12 / Chapter 1.2.2 --- Unified PML Equations in 3D --- p.15 / Chapter 1.2.3 --- Unified PML Equations in 2D --- p.16 / Chapter 1.2.4 --- Examples of PML Formulations --- p.16 / Chapter 1.3 --- Inhomogeneous Initial Conditions --- p.23 / Chapter 2 --- Numerical Analysis of PMLs --- p.25 / Chapter 2.1 --- Continuous PMLs --- p.26 / Chapter 2.1.1 --- PMLs for Wave Equations --- p.27 / Chapter 2.1.2 --- Finite PMLs for Wave Equations --- p.31 / Chapter 2.1.3 --- Berenger's PMLs for Maxwell Equations --- p.33 / Chapter 2.1.4 --- Finite Berenger's PMLs for Maxwell Equations --- p.35 / Chapter 2.1.5 --- PMLs for Acoustic Equations --- p.38 / Chapter 2.1.6 --- Berenger's PMLs for Acoustic Equations --- p.39 / Chapter 2.1.7 --- PMLs for 1-D Hyperbolic Systems --- p.42 / Chapter 2.2 --- Discrete PMLs --- p.44 / Chapter 2.2.1 --- Discrete PMLs for Wave Equations --- p.44 / Chapter 2.2.2 --- Finite Discrete PMLs for Wave Equations --- p.51 / Chapter 2.2.3 --- Discrete Berenger's PMLs for Wave Equations --- p.53 / Chapter 2.2.4 --- Finite Discrete Berenger's PMLs for Wave Equations --- p.56 / Chapter 2.2.5 --- Discrete PMLs for 1-D Hyperbolic Systems --- p.58 / Chapter 2.3 --- Modified Yee schemes for PMLs --- p.59 / Chapter 2.3.1 --- Stability of the Yee Scheme for Wave Equation --- p.61 / Chapter 2.3.2 --- Decay of the Yee Scheme Solution to the Berenger's PMLs --- p.62 / Chapter 2.3.3 --- Stability and Convergence of the Yee Scheme for the Berenger's PMLs --- p.67 / Chapter 2.3.4 --- Decay of the Yee Scheme Solution to the Hagstrom's PMLs --- p.70 / Chapter 2.3.5 --- Stability and Convergence of the Yee Scheme for the Hagstrom's PMLs --- p.75 / Chapter 2.4 --- Modified Lax-Wendroff Scheme for PMLs --- p.80 / Chapter 2.4.1 --- Exponential Decays in Parabolic Equations --- p.80 / Chapter 2.4.2 --- Exponential Decays in Hyperbolic Equations --- p.82 / Chapter 2.4.3 --- Exponential Decays of Modified Lax-Wendroff Solutions --- p.86 / Chapter 3 --- Numerical Simulation --- p.93 / Bibliography --- p.99

Optimal partitioning of three-dimensional (3-D) mesh applications necessitates dynamically determining and optimizing for the most time-inhibiting factors, such as load imbalance and communication volume. One challenge is to create an analytical model where the programmer can focus on optimizing load imbalance or communication volume to reduce execution time. Another challenge is the best individual performance of a specific mesh refinement demands precise study and the selection of the suitable computation strategy. Very-large-scale finite element method (FEM) applications require sophisticated capabilities for using the underlying parallel computer's resources in the most efficient way. Thus, classifying these requirements in a manner that conforms to the programmer is crucial. / This thesis contributes a simulation-based approach for the algorithm analysis and design of parallel, 3-D FEM mesh refinement that utilizes Petri Nets (PN) as the modeling and simulation tool. PN models are implemented based on detailed software prototypes and system architectures, which imitate the behaviour of the parallel meshing process. Subsequently, estimates for performance measures are derived from discrete event simulations. New communication strategies are contributed in the thesis for parallel mesh refinement that pipeline the computation and communication time by means of the workload prediction approach and task breaking point approach. To examine the performance of these new designs, PN models are created for modeling and simulating each of them and their efficiencies are justified by the simulation results. Also based on the PN modeling approach, the performance of a Random Polling Dynamic Load Balancing protocol has been examined. Finally, the PN models are validated by a MPI benchmarking program running on the real multiprocessor system. The advantages of new pipelined communication designs as well as the benefits of PN approach for evaluating and developing high performance parallel mesh refinement algorithms are demonstrated.

No description available.

No description available.