A QR-type reduction for computing the SVD of a general matrix product/quotient (Q1402183)

The authors study the computation of the singular value decomposition (SVD) of a general matrix product/quotient sequence. First, they reduce the sequence by QR-factorizations. Then using the Matlab command \(gsvd\) they produce the SVD. An advantage of the proposed QR-factorization is its flexibility of adding one more matrix from left or right of the original matrix product/quotient. This feature is very useful for the applications like the estimation of Lyapunov exponents of dynamic systems. Some numerical examples illustrating the proposed method are given.