An algorithm is developed for the generalized eigenvalue problem (A - λB)φ = O where A and B are real symmetric matrices. The matrices A and B are diagonalized simultaneously by a series of generalized Jacobi transformations and all eigenvalues and eigenvectors are obtained. A criterion expressed in terms of the transformation parameters is used to omit transformations leading to very small changes. The algorithm is described in pseudo code for lower triangular matrices A and B and implemented in the programming Language C.
- Jacobi Algorithm
- Eigenvalue Problems