Generalized Eigenvalue - Support for positive semi-definite but singular matrices

[brtang63]

Hi @yixuan,

For the generalized eigenvalue problem $Ax = \lambda Bx$, the current implementation of Generalized Eigen Solvers seems to require either A or B to be positive definite. Is it possible to support the case when both A and B are symmetric positive semi-definite (but singular) matrices? I've encountered such a situation in a statistical scenario, where both matrices are covariance matrices.