Parallel Scientific Computation: A Structured Approach Using BSP

Parallel Scientific Computation provides a single unified approach to using parallel computers, from small desktops to massively parallel computers, and covers core problems in scientific computing and big data. It includes a theoretical section and practical section with parallel programs and numerical experiments, and is accompanied by a software package BSPedupack.

\n Format: Paperback / softback
\n Length: 416 pages
\n Publication date: 30 September 2020
\n Publisher: Oxford University Press
\n

Building upon the immense success of the first edition, Parallel Scientific Computation presents a comprehensive and unified approach to leveraging a diverse array of parallel computers, spanning from small desktop systems to massively parallel machines. The author elucidates the methodology for designing and implementing parallel algorithms in the domains of scientific computing and big data, employing the bulk synchronous parallel (BSP) model. This model serves as a framework for developing efficient solutions for complex problems.

The book encompasses a comprehensive treatment of core issues in these fields, commencing with a high-level problem description and progressing to sequential solution algorithms, parallel solution algorithms, and actual parallel programs written in the BSPlib library. Each chapter is divided into two sections: a theoretical section that delves into the underlying principles and concepts, and a practical section that showcases parallel programs and numerical experiments conducted on modern parallel computers. These experiments serve to validate the theoretical predictions and cost analysis presented in the text.

Furthermore, each chapter includes extensive bibliographical notes that provide additional discussions, pointers to relevant literature, and a multitude of exercises that serve as suitable projects for graduate students. The second edition of this book introduces new material that is particularly relevant to big-data science, including sorting and graph algorithms. Additionally, it adopts a BSP approach to address emerging hardware developments, such as hierarchical architectures with both shared and distributed memory. By employing a single, hybrid BSP system, the book enables efficient handling of both types of parallelism, eliminating the need for mastering two separate systems, as is common in alternative approaches.

Moreover, the second edition ensures that all algorithms are up-to-date, incorporating recent advancements in the field. It includes comprehensive coverage of high-performance linear system solving by LU decomposition and improved data partitioning for sparse matrix computations. To accompany the book, a software package named BSPedupack is freely available online. This package provides supplementary materials, including code examples, exercises, and additional resources, to enhance the learning experience for readers.

In summary, Parallel Scientific Computation serves as a valuable resource for researchers, scientists, and engineers seeking to harness the power of parallel computing for scientific and data-intensive applications. With its comprehensive coverage, practical insights, and up-to-date algorithms, the book empowers readers to develop efficient and scalable solutions to complex problems, enabling them to make significant contributions to their respective fields.

\n Weight: 660g\n
Dimension: 156 x 233 x 25 (mm)\n
ISBN-13: 9780198788355\n
Edition number: 2 Revised edition\n