Closed yuzeh closed 12 years ago
The function numeric.inv() allows you to invert an arbitrary dense matrix.
The functions numeric.cLU() and numeric.ccsLUP(), etc... are for more advanced usage with sparse matrices that are stored in suitable compressed matrix storage formats -- please see the documentation for details.
The conventional wisdom is that solving Ax = b is faster than doing inv(A) * b. I just timed my implementation of LU decomposition and I'll submit a pull request soon.
I noticed that numeric currently doesn't have a general linear solver (kind of like MATLAB's backslash operator). There is a
cLU
function, but it seems like that only takes matrices with up to 3 rows.Is there any plan on supporting this? If there is, I've implemented a non-optimized version of LU factorization with pivoting for square matrices, which is a step in that direction.