QR decomposition in mpmath?

[Ken]

Does mpmath have any built-in routines to calculate QR matrix
decompositions? All I'm able to find is qr_solve but that doesn't
actually calculate Q or R which I need for constrained least squares.
Any suggestions?

Thanks, Ken

[Fredrik Johansson]

Hi Ken,

Unfortunately, there's no such function provided currently. It would
be a welcome contribution.

Fredrik

[Ken]

I'm working on a QR function that should be ready soon. How would I
go about submitting it for consideration?

Ken

[Fredrik Johansson]

On Wed, Mar 28, 2012 at 19:01, Ken <ken....@sbcglobal.net> wrote:
> I'm working on a QR function that should be ready soon.  How would I
> go about submitting it for consideration?

Great! You can attach a patch to the issue tracker (or send it to this
list), or preferably send a pull request on github
(https://github.com/fredrik-johansson/mpmath).

Fredrik

[Ken]

OK, I attached a file to the issue tracker (sorry, github looked way
too complicated). The algorithm factors A -> Q, R or A -> L, Q
depending on the dimensions of A. The only restriction is m > 1 and n
> 1. The matrix A can have any content (even all zeros). I've tested
it on over 200k random sized matrices (m > n, m = n, m < n) with
random integers and floats, sprinkled with rows and columns of zeros,
duplicate rows and columns, and all zeros. So far no errors. I hope
the algorithm is a useful contribution.
Ken