Fast Solvers for Mesh-based Computations

The author presents an alternative method of constructing multi-frontal direct solver algorithms for mesh-based computations based on the additional available knowledge about the structure of the computational mesh. The book follows the structures of the matrices, from tri-diagonal matrices resulting from one-dimensional mesh-based methods, through multi-diagonal or block-diagonal matrices, and ending with general sparse matrices. Each chapter addresses how to design and implement a parallel sparse direct solver specific to a particular structure of the matrix, and derives the complete Java or Fortran code of the parallel sparse direct solver or presents modifications of the source code from previous chapters. The book includes exemplary element frontal matrices for two and three-dimensional mesh-based computations and a chapter on numerical examples. Readers should understand basic algebra transformation and the basics of Java and Fortran programming languages. Annotation c2016 Ringgold, Inc., Portland, OR (protoview.com)

Maciej Paszynski, PhD, Department of Computer Science, Electronics and Telecommunications, AGH University of Science and Technology, Krakow, Poland