“…In the following experiment, we apply Algorithm 3.2 to the matrix pair (A, B) from (20). We emphasize that there exist algorithms for detecting a hyperbolic QEP that work directly with matrices M , D, and K, which reduces storage requirements and the cost per iteration, such as [3,52,54]. We recommend [52, preconditioned variant of Algorithm 2] for large quadratics 9 .…”