Talk:Gaussian elimination: Difference between revisions

:Basically, Gaussian elimination is the same as LU decomposition followed by backsubstitution. To solve the system AX=B with LU decomposition, you multiply on the left by P: PAX=PB, hence LUX=PB. Then you solve first the lower triangular system LZ=PB, which gives a vector Z (or matrix if B is a matrix, which happens if you want to solve several systems at the same time with the same matrix A). Then you solve the upper triangular system UX=Z, and you are done.

:The difference lies in the order the operations are done: in Gaussian elimination, the lower triangular matrix L is not stored, since the computation of Z (that is, the solution of the lower triangular system LZ=PB) is done once and for all in the first step. Likewise, the permutation matrix P is not needed because the rows of B are swapped at the same time the rows of A are swapped. The backsubstitution step is identical in Gaussian elimination and LU decomposition.