Sam Coveney
Sam Coveney
It's been over 6 months since I opened the original PR. In that time, I have rebased many times, and identified some outright bugs in other PRs (which broke this...
Thanks @skoudoro - like last time, let me know if there are any things in other PR I can help with
> Hello! I am in the midst of reviewing the code and it looks overall quite good. Could you please say a bit more about how RESTORE was reimplemented here?...
Thanks for checking. Many things, I will clean up. Some things though, I think at this point just need to not be fiddled with, such as aesthetic concern over types,...
Last comment: I have a lot of other work to get on with, I will get to this when I can. When are you planning the release?
On the matter of the mixed types for the weights argument, I still disagree. The provided weights need to be either: - an array (use these weights) - True -...
> I think the answer to my latter question is Collier, Q., Veraart, J., Jeurissen, B., den Dekker, A., Sijbers, J., 2015. Iterative reweighted linear least squares for accurate, fast,...
I think I've addressed everything, except for any pep8 issues I've accidentally made in the process of fixing. I will check tomorrow. One other thing: I need to finish off...
Some of these tests are failing here, because they are complaining about the test `test_nnls_jacobian_func` specifically that `nlls.err_func` should return a scaler variable. For DiPy, due to `scip.opt.leastsq`, we do...
> On a bit of further thought, I don't know if this PR is a good idea, even if I can make it work. I think that we should, as...