Added the coordinate descent method from scikit-learn, supporting sparsity

[vilim]

This is a more-or-less direct translation of the Python Scikit-learn NMF method, which hopefully respects Julian idioms.

I am not sure how I should reference the original implementation, so any comments on attribution, licences and copyright are welcome.

[kmsquire]

Here's an example of some publicly available code that was translated from Matlab and placed in Images.jl.

Scikit-learn code is available under a BSD-style license. This package is MIT-licensed. Generally, these licenses are compatible with each other, so as long as the core maintainers of this package are open to it, a typical solution would be to note an exception in LICENSE.md, which states that the package is available under the MIT license (already there), except as noted at the top of individual files, and then to include the copyright and BSD license text at the top of coorddesc.jl.

As long as you include the copyright and license somewhere, any variation on above should be fine.

[ararslan]

except as noted at the top of individual files

Note that BSD style licenses are not copyleft, so we don't need to have an exception in our license for code derived from BSD-licensed code; BSD style licenses permit licensing derived works under any license. We can simply acknowledge and link to the license for the code from which this was derived directly in the file.

[vilim]

@ararslan I have already updated the file referencing the original implementation and the license, is there anything else to do before it can be merged? The same Travis test error is also present on the master branch.

[ararslan]

I have already updated the file referencing the original implementation and the license

I saw, that looks good. My comment about the license was in response to Kevin's.

[vilim]

@ararslan Thanks for the review! I have implemented all the changes.

[ararslan]

Looks like the Project.toml changes weren't implemented.

[vilim]

@ararslan sorry, I switched to VS Code recently and am still getting used to file status indicators; it wasn't committed. It is fixed now.

[vilim]

@ararslan is there anything else I need to do before this can be merged?

[VPetukhov]

Hi @ararslan ! Can I ask what's blocking from merging this PR? Looks like the requested changes were resolved. And I can't reproduce the test failure: now it works absolutely fine. Here is travis the successful report from my fork (without any changes): https://travis-ci.com/VPetukhov/NMF.jl/builds/120496869 . So can you please maybe trigger rebuild and check if it still fails?

[VPetukhov]

Thank you!