Re: lsqr: some guidance requested

[c.]

On 1 Dec 2016, at 19:55, Nicholas Jankowski <address@hidden> wrote:

> On Thu, Dec 1, 2016 at 9:59 AM, Marco Atzeri <address@hidden> wrote:
>> 
>> Left division already offer "a minimum norm solution"
>> 
>> https://www.gnu.org/software/octave/doc/interpreter/Simple-Examples.html#Matrix-Arithmetic
> 
> 
> in the interest of having a matlab compatible syntax and output is
> there a way a wrapper could be made to force \ to use a least squares
> solution algorithm? I assume it currently follows some decision tree.
> Also, more to Ruben's question, how optimized is \ for sparse A
> matrices?
> 

Here is the decision tree for mldivide on sparse matrices:
https://www.gnu.org/software/octave/doc/v4.0.0/Sparse-Linear-Algebra.html

Essentially all optimization is delegated to CHOLMOD or UMFPACK.

c.