Issue No.01 - January/February (2000 vol.2)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/5992.814658
A matrix decomposition is a factorization of a matrix into the product of simpler matrices. The introduction of matrix decomposition into numerical linear algebra in the years from 1945 to 1965 revolutionized matrix computations. This article outlines the decompositional approach, comments on its history, and surveys the six most widely used decompositions: the Cholesky decomposition, the pivoted LU decomposition, the QR decomposition, the spectral decomposition, the Schur decomposition, and the singular value decomposition.
G.w. Stewart, "The Decompositional Approach to Matrix Computation", Computing in Science & Engineering, vol.2, no. 1, pp. 50-59, January/February 2000, doi:10.1109/5992.814658