Gaussian quadrature for matrix valued functions on the real line (Q1917953): Difference between revisions

The author implements the Gaussian quadrature algorithm for matrix-valued functions on a multiprocessor SP1 based on RISC-processors. He gives a formula to determine the quadrature weights using the eigenvalues and components of the normalized eigenvectors of a symmetric block tridiagonal matrix. Because the elements of the main and first subdiagonal are the recursion coefficients of the corresponding orthonormal matrix polynomials, the block version of the modified Chebyshev algorithm to compute these recursion coefficients from the modified moments can be constructed.