Sparse Matrices and Vectors in C-XSC
Michael Zimmer, Walter Krämer, Werner Hofschuster · 2010
C-XSC is a C++ library for reliable scientific computing, which provides data types for dense vectors and matrices with real, complex, real interval and complex interval entries. These data types are easy to use and provide many helpful functionalities such as the ability to work with submatrices and subvectors. However, when dealing with sparse vectors, and especially with sparse matrices, these data types are inefficient. C-XSC version 2.4.0 added special types for sparse vectors and matrices that take advantage of the sparsity, both for performance and for memory consumption. This paper explains the data structures for and the implementation of these new types. Many examples and some performance tests with sparse matrices from real world applications are included.