Mixed real/complex factorization (power systems)

A mixed real/complex sparse matrix factorization and solution scheme applied to a large matrix problem is described. Large system eigenanalysis and frequency domain methods will directly benefit from the proposed scheme, which can reduce both memory and CPU time requirements when compared to conventional complex-only solutions. The application in hand is the small signal electromechanical stability analysis of large power systems. The savings obtained are significant considering the CPU-intensive nature of these matrix problems.<
>