No label defined (Q4040698)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 193901
Language Label Description Also known as
default for all languages
No label defined
    English
    No label defined
    scientific article; zbMATH DE number 193901

      Statements

      0 references
      5 June 1993
      0 references
      singular value decomposition
      0 references
      Fundamentals of parallel computing
      0 references
      Numerical linear algebra
      0 references
      parallel computers
      0 references
      parallel programming
      0 references
      parallel sorting
      0 references
      symmetric eigenvalue problem
      0 references
      QR factorization
      0 references
      SVD
      0 references
      Jacobi methods
      0 references
      The book is divided into two distinct parts: Fundamentals of parallel computing (108 pages) and Numerical linear algebra (129 pages). Although it contains material from both numerical analysis and computer science, it is mainly a book about numerical methods. The first part of the book gives an introduction to some general principles of parallel computers, such as classification, graph theory and networks, embedding, and performance measurement, as well as a brief introduction to parallel programming techniques such as recursive doubling, cyclic odd-even reduction, slicing and crinkling. Included in the first part is also a chapter on parallel sorting algorithms that illustrate some of the above concepts. The second part of the book is devoted to topics in numerical linear algebra: solution of systems of linear equations, Jacobi methods for the symmetric eigenvalue problem, QR factorization, and SVD. Emphasis is primarily focused on the symmetric eigenvalue problem. This part of the book perfectly illustrates how important it is to `think in parallel'; i.e. to exploit all the parallelism in a particular algorithm and to find efficient arrangements of data in the memory. The most important contribution of the book is the discussion in chapter 5 of Jacobi methods for the symmetric eigenvalue problem. Here, both basic parallel techniques and many advanced techniques are presented, and the material covered is certainly up-to-date. It is obvious that some of the author's own interests are within this field. Unfortunately, the organization of the book could be a lot better. For example, the concept of `granularity', which ought to be defined in the beginning of Chapter 1, is defined in Chapter 2 in connection with the discussion of slicing and crinkling. It is disappointing that too much of the material - although fairly advanced - covers general aspects of the methods rather than particular `parallel' aspects (this is essentially true in Chapter 4, 6, and 7). It is also disappointing that most of the discussions of practical implementations are restricted to one particular computer, namely the Distributed Array Processor, and that comparisons of which algorithms are suited for particular architectures is completely missing. Thus, the book is not suited as a textbook in parallel computing, and neither is it of much use as a handbook in parallel methods for numerical linear algebra. As mentioned above, its main quality is as an introduction to concepts in parallel computing.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references