Crout Versions of ILU for General Sparse Matrices

This paper presents an efficient implementation of the incomplete LU (ILU) factorization derived from the Crout version of Gaussian elimination. At step k of the elimination, the kth row of U and the kth column of L are computed using previously computed rows of U and columns of L. The data structure and implementation borrow from already known techniques used in developing both sparse direct solution codes and incomplete Cholesky factorizations. This version of ILU can be computed much faster than standard threshold-based ILU factorizations computed rowwise or columnwise. In addition, the data structure allows efficient implementations of more rigorous and effective dropping strategies.