9216308 Alvarado This research concerns topics in sparse matrices. The first is the use of iterative methods based on approximate partitioned inverses (the explicit inverses of the L and U factors of A) used as preconditioners for iterative solvers. Another research topic is sparse matrix primitization: it is possible to enlarge the size of a sparse matrix and improve its condition number as well as reduce computation and enhance parallelism. A third topic is on- the-fly compensation. This uses one of several variants of the Matrix Inversion Lemma. The fourth research topic is a further look into ordering algorithms. Of particular interest is the investigation of methods suitable for augmented eigenvalue formulations and for orthogonal factorization. ***
