Expose FACTOR parameter instead of using magic number in scipy.optimize.nnls

[joshjohnston]

The Fortran code in scipy/optimize/nnls/nnls.f uses a parameter FACTOR, which is hardcoded to 0.01, to determine when to end the loop due to lack of linear independence. It would be much appreciated if this magic number was exposed as an option in both the Fortran and the scipy.optimize.nnls python wrapper.

[rgommers]

Hi @joshjohnston, do you have a code example perhaps where this would help?

[Kai-Striega]

@rgommers I'm not too familiar with the function - so I can't comment on whether it's a good idea or not - but the actual change should be pretty straight forward if you want someone to do it.

[osullivryan]

Hello @joshjohnston , I took a crack at it. Let me know how that looks!

[dschmitz89]

nnls was recently completely rewritten and the old Fortran source is gone (see #18570). The new code uses another algorithm that does not use this conversion criterion anymore, so closing this.