# top right submatrix of a matrix in reduced row echelon form

In linear algebra, I remember that there was something special about the submatrix in the top right of an rref'd matrix.

1 0 0 |  0 1
0 1 0 | -1 0
0 0 1 |  1 0
-------------
0 0 0    0 0


You would ammend an identity matrix (in this case $2\times2$) above the submatrix and get two vectors in 5 dimensions. I think they say something about the kernel or so, but I do not really remember.

What do those two vectors tell me about the original matrix?

-
@Michael: It seems you accidentally entered "spelling" as a tag instead of an edit summary? – joriki Dec 14 '11 at 17:40

Perhaps you mean this: If you negate that submatrix and add the identity matrix below it, you get a basis for the kernel. In your example, you can choose $x_4$ and $x_5$ arbitrarily, and then a general element of the kernel is
$$x_4\pmatrix{0\\1\\-1\\1\\0}+x_5\pmatrix{-1\\0\\0\\0\\1}\;.$$