Abstract: With the development of linear programming,the decomposition technology has been playing a significant role in the implementation of the algorithm. In the traditional simplex method,product form of inverse(PFI) and LU decomposition keep the sparsity of the A matrix and decrease the round-off error. Decomposition of dynamic matrix becomes a new research topic as soon as the LP dynamic factorization and kernel matrix came out.This paper discusses the decomposition and storage of kernel matrix by applying the LU decomposition method,which makes a good result on matrix sparsity and round-off error.Addidonally,the permutation elimination method proposed in this paper makes the LU update process more efficient.

Key words: linear programming, kernel matrix, dynamic decomposing, LU deco mposition