[R] singularity problem

[Rolf Turner]

On 10/14/13 23:31, Šárka Rusá wrote:
> Dear R,
>
> I have a problem concerning the inverse of a matrix. My matrix is quite
> large and I extracted just linearly independent columns from the matrix.
> However, the function solve() still cannot invert the matrix - I get an
> error that it is computationally singular. Is there any way how to extract
> some columns from the matrix so that I could get the inverse?
> The problém is I can extract a "nonsingular" matrix but there is no point
> in doing it since I don't get the inverse.

Your question makes no sense to me.  If you just select certain columns 
of your matrix
the resulting matrix will no longer be square and inverting it is 
meaningless.

If you have a (numerically) non-singular matrix, then it can, by 
definition be inverted.
(And solve() will invert it for you.)

What are you actually trying to accomplish?

     cheers,

     Rolf Turner