3 ms·
As others have said, Gaussian elimination, or if the matrix is not invertible (square and singular or not square) then the Moore-Penrose pseudoinverse can be us
by SandB0x 17y ago
As others have said, Gaussian elimination, or if the matrix is not invertible (square and singular or not square) then the Moore-Penrose pseudoinverse can be used to find the least squares solution.
- jedbrown 17y agoExcept that the normal equations are rarely a good way to compute the action of the pseudoinverse (though explicitly computing it would be even worse). QR is the workhorse for under- and over-determined dense systems, with SVD as a fallback for especially nasty ones.