Annals of Mathematical Sciences and Applications

In this work, we present a new structure-preserving method to compute the structured Schur form of a dense complex $\mathsf{T}$ Hamiltonian matrix $\mathscr{H}$ of moderate size. Origination of the complex $\mathsf{T}$ Hamiltonian eigenvalue problem outside the control theory is briefly discussed. Specifically, our method consists of three main stages. At the first stage, we compute eigenvalues of $\mathscr{H}$ using the $\mathsf{T}$ symplectic URV-decomposition of complex $\mathscr{H}$ followed up with the complex periodic QR algorithm to thoroughly respect the $(\lambda,-\lambda)$ pairing of eigenvalues. At the second stage, we construct the $\mathsf{T}$ isotropic invariance subspace of $\mathscr{H}$ from suitable linear combination of columns of $U$ and $V$ matrices from the first stage. At the third stage, we find a $\mathsf{T}$ symplectic-orthogonal basis of this invariance subspace, which immediately provides the structured Schur form of $\mathscr{H}$. Several numerical results are presented to demonstrate the effectiveness and accuracy of our method.