LU matrix factorization

`[L,U] = lu(A)`

`[L,U,P] = lu(A)`

`[L,U,P] = lu(A,outputForm)`

`[L,U,P,Q] = lu(S)`

`[L,U,P,Q,D] = lu(S)`

`[___] = lu(S,thresh)`

`[___] = lu(___,outputForm)`

`[___] = lu(`

specifies thresholds for the pivoting strategy employed by
`S`

,`thresh`

)`lu`

using any of the previous output argument
combinations. Depending on the number of output arguments specified, the default
value and requirements for the `thresh`

input are different.
See the `thresh`

argument description for details.

`[___] = lu(___,`

returns `outputForm`

)`P`

and `Q`

in the form specified by
`outputForm`

. Specify `outputForm`

as
`'vector'`

to return `P`

and
`Q`

as permutation vectors. You can use any of the input
argument combinations in previous syntaxes.

The LU factorization is computed using a variant of Gaussian elimination. Computing an
accurate solution is dependent upon the value of the condition number of the original
matrix `cond(A)`

. If the matrix has a large condition number (it is
nearly singular), then the computed factorization might not be accurate.

The LU factorization is a key step in obtaining the inverse with
`inv`

and the determinant with `det`

. It is also
the basis for the linear equation solution or matrix division obtained with the
operators `\`

and `/`

. This necessarily means that the
numerical limitations of `lu`

are also present in these dependent
functions.