A Jacobi-type systolic algorithm for Riccati and Lyapunov equations (Q1344537)

The paper presents an algorithm aimed at numerically solving algebraic Riccati and Lyapunov equations which is based on a modification of the LU factorization approach proposed by \textit{J.-P. Charlier} and \textit{P. Van Dooren} [Math. Control Signals Syst. 2, No. 2, 109-136 (1989; Zbl 0675.65032)] for the recursive matrix sign algorithm. The modification makes use of Jacobi transformations for the inversions in the recursions. The operational count shows that the proposed modification requires on a systolic array \({8 \over 3}\) the computation time required by the LU factorization approach, but involving eight times less processors.