We consider a contour integral-based eigensolver that finds eigenvalues in a given domain and the corresponding eigenvectors of the generalized eigenvalue problem. In the contour integral-based eigensolver, quadrature points are placed in the complex plane in order to approximate the contour integral. When eigenvalues exist near a quadrature point, the accuracy of other eigenvalues is deteriorated. We herein propose a method by which to recover the accuracy of the eigenpairs when eigenvalues exist near a quadrature point. A numerical experiment is conducted in order to verify that the proposed method is efficient.
The Cholesky QR algorithm is an ideal QR decomposition algorithm for high performance computing, but known to be unstable. We present error analysis of the Cholesky QR algorithm in an oblique inner product defined by a positive definite matrix, and show that by repeating the algorithm twice (called CholeskyQR2), its stability is greatly improved.