python?

[argriffing]

This might be a dumb question, but is there a Python interface? This possibility was mentioned in https://media.readthedocs.org/pdf/matrix-repository/latest/matrix-repository.pdf which seems to have been merged into MatrixDepot.jl.

[weijianzhg]

Hi @argriffing, thanks for the question. But unfortunately, I haven't found time to implement it. I think I will work on it this summer.

[argriffing]

Even if this is not implemented, would you know the best way to get access to matrix sets 1--4 mentioned in section 6 of http://core.ac.uk/download/pdf/17664.pdf in a way that could be included in MIT or BSD licensed code (not GPL)? I've noticed a typo in my implementation of that expm code which wasn't caught by less comprehensive testing.

[weijianzhg]

Hi,

You may want to try rogues (https://pypi.python.org/pypi/rogues/0.2.0), which is MIT licensed.

Best wishes,

Weijian

[argriffing]

Thanks for the link! It does say that it's MIT licensed, but its sources appear to be incompatibly licensed (TOMS-licensed and GPL-licensed). @macd?

[macd]

Well, that is a tricky question. I did these awhile ago when I was first learning numpy / scipy. As it says in the readme, I mostly used version 3.0, which was downloaded from Prof. Higham website. I didn't see how that version was licensed. So that is unclear to me. I didn't use any of the 1989 TOMS code, but I did look at it.

So, the situation is murky, at best. I know some folks have a hard time with the older TOMS license (the newer version seems OK?). I don't believe there is anything here that is GPL'd however. @weijianzhang how is the Julia version licensed exactly? You must have used the old MATLAB implementations from Prof. Higham? You are in Prof. HIgham's research group? I see you have an MIT "expat" license. Anything special done for that licensing?

[argriffing]

As it says in the readme, I mostly used version 3.0, which was downloaded from Prof. Higham website. [...] I don't believe there is anything here that is GPL'd however.

The website says that both the Matrix Computation Toolbox and the Test Matrix Toolbox are distributed under the terms of the GNU General Public License (version 3 of the License, or any later version) as published by the Free Software Foundation. But this doesn't necessarily mean that they are not also distributed under other licenses. Separately, I'm not sure about possible TOMS-related issues.

The Matrix Computation Toolbox is also hosted on Matlab Central where No BSD License is explicitly stated.

[macd]

Ah, well, I haven't looked at that website in awhile. I certainly don't remember that to be the case in 2008-2009 time frame and none of the MATLAB code includes a license. But that is not to say that it wasn't GPL'd then, only I don't remember.

On Tue, Jun 9, 2015 at 8:54 AM, argriffing notifications@github.com wrote:

As it says in the readme, I mostly used version 3.0, which was downloaded from Prof. Higham website. [...] I don't believe there is anything here that is GPL'd however.

The website http://www.maths.manchester.ac.uk/%7Ehigham/mctoolbox/ says that both the Matrix Computation Toolbox and the Test Matrix Toolbox are distributed under the terms of the GNU General Public License http://www.gnu.org/copyleft/gpl.html (version 3 of the License, or any later version) as published by the Free Software Foundation. But this doesn't necessarily mean that they are not also distributed under other licenses. Separately, I'm not sure about possible TOMS-related issues.

— Reply to this email directly or view it on GitHub https://github.com/weijianzhang/MatrixDepot.jl/issues/6#issuecomment-110412832 .

[weijianzhg]

Hi @macd,

how is the Julia version licensed exactly? You must have used the old MATLAB implementations from Prof. Higham? You are in Prof. HIgham's research group? I see you have an MIT "expat" license. Anything special done for that licensing?

MatrixDepot is distributed under MIT "expat" license. I did look at Higham's MATLAB implementations of many matrix generators. But the Julia implementations are quite different and I coded the test matrices based on the original papers. Also in Julia, loops are fast and my implementations use loops instead of the vectorized expressions of MATLAB.

Yes, I am in Prof. Nick Higham's research group.

Best wishes,

Weijian

[argriffing]

Guess I'd better start learning Julia. It's from MIT and their media relations team says that everybody's doing it! It has multiple dispatch! Its loops are fast and it solves the two-language problem!

[weijianzhg]

@argriffing Yeah :)