Abstract: "Mesh adaptation is a powerful tool for efficient unstructured-grid computations but causes load imbalance on multiprocessor systems. To address this problem, we have developed PLUM, an automatic portable framework for performing adaptive large-scale numerical computations in a message-passing environment. This paper presents several experimental results that verify the effectiveness of PLUM on sequences of dynamically adapted unstructured grids. We examine portability by comparing results between the distributed-memory system of the IBM SP2, and the Scalable Shared-memory MultiProcessing (S2MP) architecture of the SGI/Cray Origin2000. Additionally, we evaluate the performance of five state-of-the-art partitioning algorithms that can be used within PLUM. Results indicate that for certain classes of unsteady adaption, globally repartitioning the computational mesh produces higher quality results than diffusive repartitioning schemes. We also demonstrate that a coarse starting mesh produces high quality load balancing, at a fraction of the cost required for a fine initial mesh. Finally, we show that the data redistribution overhead can be significantly reduced by applying our heuristic processor reassignment algorithm to the default partition-to-processor mapping given by partitioners."

This IMA Volume in Mathematics and its Applications GRID GENERATION AND ADAPTIVE ALGORITHMS is based on the proceedings of a workshop with the same title. The work shop was an integral part of the 1996-97 IMA program on "MATHEMAT ICS IN HIGH-PERFORMANCE COMPUTING. " I would like to thank Marshall Bern (Xerox, Palo Alto Research Cen ter), Joseph E. Flaherty (Department of Computer Science, Rensselaer Polytechnic Institute), and Mitchell Luskin (School of Mathematics, Uni versity of Minnesota), for their excellent work as organizers of the meeting and for editing the proceedings. I also take this opportunity to thank the National Science Founda tion (NSF), Department of Energy (DOE), and the Army Research Office (ARO), whose financial support made the workshop possible. Willard Miller, Jr. , Professor and Director v PREFACE Scientific and engineering computation has become so complex that traditional numerical computation on uniform meshes is generally not pos sible or too expensive. Mesh generation must reflect both the domain geometry and the expected solution characteristics. Meshes should, fur thermore, be related to the solution through computable estimates of dis cretization errors. This, suggests an automatic and adaptive process where an initial mesh is enriched with the goal of computing a solution with prescribed accuracy specifications in an optimal manner. While automatic mesh generation procedures and adaptive strategies are becoming available, major computational challenges remain. Three-dimensional mesh genera tion is still far from automatic.

A unique investigation of the state of the art in design, architectures, and implementations of advanced computational infrastructures and the applications they support Emerging large-scale adaptive scientific and engineering applications are requiring an increasing amount of computing and storage resources to provide new insights into complex systems. Due to their runtime adaptivity, these applications exhibit complicated behaviors that are highly dynamic, heterogeneous, and unpredictable—and therefore require full-fledged computational infrastructure support for problem solving, runtime management, and dynamic partitioning/balancing. This book presents a comprehensive study of the design, architecture, and implementation of advanced computational infrastructures as well as the adaptive applications developed and deployed using these infrastructures from different perspectives, including system architects, software engineers, computational scientists, and application scientists. Providing insights into recent research efforts and projects, the authors include descriptions and experiences pertaining to the realistic modeling of adaptive applications on parallel and distributed systems. The first part of the book focuses on high-performance adaptive scientific applications and includes chapters that describe high-impact, real-world application scenarios in order to motivate the need for advanced computational engines as well as to outline their requirements. The second part identifies popular and widely used adaptive computational infrastructures. The third part focuses on the more specific partitioning and runtime management schemes underlying these computational toolkits. Presents representative problem-solving environments and infrastructures, runtime management strategies, partitioning and decomposition methods, and adaptive and dynamic applications Provides a unique collection of selected solutions and infrastructures that have significant impact with sufficient introductory materials Includes descriptions and experiences pertaining to the realistic modeling of adaptive applications on parallel and distributed systems The cross-disciplinary approach of this reference delivers a comprehensive discussion of the requirements, design challenges, underlying design philosophies, architectures, and implementation/deployment details of advanced computational infrastructures. It makes it a valuable resource for advanced courses in computational science and software/systems engineering for senior undergraduate and graduate students, as well as for computational and computer scientists, software developers, and other industry professionals.

Contains 17 papers written by an international group of academic and industrial specialists in computer science. Some of the topics addressed include the design and implementation of video servers in video-on-demand systems; a framework for the development of globally convergent adaptive learning rate algorithms; a vector-based approach to analysis of file space properties; load balancing for unstructured mesh applications; musical composition based on genetic algorithms and fuzzy transformations of traditional Greek music patterns; and frequency-adaptive join for shared nothing machines. Most papers consist of an abstract, key words, an introduction, discussion, conclusions, suggestions for future research, and references. Several contributions are printed in a rather dark, compacted font that is difficult to read. c. Book News Inc.

Proceedings -- Parallel Computing.

With considerations such as complex-dimensional geometries and nonlinearity, the computational solution of partial differential systems has become so involved that it is important to automate decisions that have been normally left to the individual. This book covers such decisions: 1) mesh generation with links to the software generating the domain geometry, 2) solution accuracy and reliability with mesh selection linked to solution generation. This book is suited for mathematicians, computer scientists and engineers and is intended to encourage interdisciplinary interaction between the diverse groups.

Abstract: "The computational requirements for an adaptive solution of unsteady problems change as the simulation progresses. This causes workload imbalance among processors on a parallel machine which, in turn, requires significant data movement at runtime. We present a new dynamic load-balancing framework, called JOVE, that balances the workload across all processors with a global view. Whenever the computational mesh is adapted, JOVE is activated to eliminate the load imbalance. JOVE has been implemented on an IBM SP2 distributed-memory machine in MPI for portability. Experimental results for two model meshes demonstrate that mesh adaption with load balancing gives more than a sixfold improvement over one without load balancing. We also show that JOVE gives a 24-fold speedup on 64 processors compared to sequential execution."