Krylov type subspace methods for matrix polynomials (Q2368740)

The authors develop and analyze Arnoldi and Lanczos type methods for model reduction problems associated with quadratic matrix polynomials \( \lambda^2 I - \lambda A - B\), where \(A\) and \(B\) are sparse large matrices. The methods are Krylov type projection methods that generate quadratic matrix polynomials with coefficient matrices of much smaller size. Applications include second order linear input-output systems.