# Decomposition of a rectangular sparse matrix

All of the matrices in NLP are large sparse matrices (larger than $$1M$$ in $$850K$$). I want to decompose such matrices by the fastest method. Which one is the best method for decomposition of a sparse rectangular matrix: LU or Cholesky?

I mean: what's the best method in theoretical?

• Which software are you using? If it can handle sparse matrices it probably already has a decomposition function that is optimized for sparse matrices. – quarague Jul 12 at 12:02
• I use python for programming. But I need to know the theorical solution. – BarzanHayati Jul 12 at 12:14