The generalized eigenvalue problem and acoustic surface wave computations (Q580900)

Eigenvalue problems linear in the vector but nonlinear in the eigenvalue are considered. It is talked about algorithms that find zeroes of the smallest diagonal element of the triangular part of a QR or LU decomposition of the matrix, and said that this is better than finding zeroes of the determinant. Numerical experiments on a problem from acoustics leading to a second degree matrix polynomial are reported.