|
|
Ublas : |
Subject: Re: [ublas] announcement: ublasJama (ublas port of Jama, including SVD and eigenvalue decomposition)
From: Frédéric Devernay (frederic.devernay_at_[hidden])
Date: 2010-05-08 14:09:17
Hi all,
Thank you for your interest in ublasJama.
Well, I was considering it to be public domain, since my work is incremental over public-domain software. Frankly, it just doesn't deserve a license of its own :-)
I guess Jama was public-domain for the same reason (there was a - good? - time, in the 70s and 80s, when a lot of FORTRAN code was just released as public domain, and products such as Matlab probably wouldn't exist without these).
I should probably make it more clear on the ublasJama page.
It should make no problem for a potential inclusion in boost, what do you think about this?
I did the same for cminpack http://devernay.free.fr/hacks/cminpack.html which is distributed with the original MINPACK copyright notice.
I love those old super-robust FORTRAN codes. I'm just hoping I didn't introduce a few bugs here and there ;-)
Fred
Le 1 mai 2010 à 10:00, David Bellot a écrit :
> In term of license, if we consider it's an original work (i.e. not just a single translation of jama in c++) then I would prefer to have the boost license on it so that not to bother with the complexity of licensing.
>
> If Frederic agree on it, I would like him to add this license to his work so that we can consider integration into ublas.
> I think we will have to better integrate things now, but it will be a pleasure to have such a collection of algorithms in ublas.
>
> David
>
> On Fri, Apr 30, 2010 at 17:19, Tom Vercauteren <tom.vercauteren_at_[hidden]> wrote:
> Hi Frédéric,
>
> This is indeed quite interesting. I was just wondering about the
> license of your code.
>
> As far as I understand it, the original JAMA is "public domain" hence
> even more permissaive than boost license:
> http://math.nist.gov/javanumerics/jama/#Authors
> http://cio.nist.gov/esd/emaildir/lists/jama/msg01295.html
> http://en.wikipedia.org/wiki/Public_domain
>
> Is it also the case for ublasJama? I was not able to see any license
> information on your webpage nor in the source code:
> http://devernay.free.fr/hacks/ublasJama-1.0.2.2.tar.gz
>
> It would be nice to add, for example, a LICENSE.txt file with the
> relevant information to your source package. The best for me would of
> course be to find an OSI approved license
> http://www.opensource.org/licenses/alphabetical
> such as the MIT, Boost or BSD one.
>
> Cheers,
> Tom
>
> On Fri, Apr 30, 2010 at 16:58, Nasos Iliopoulos <nasos_i_at_[hidden]> wrote:
> > Very nice!
> >
> > I find those algorithms could make it into uBlas like "extensions", yourself
> > and maintainers permitting of course. It could save a lot of people the time
> > to reinvent them or the need to bind to external libraries. What do all
> > think?
> >
> > I use my custom algorithms in some cases, and I was looking into
> > reimplementing JAMA, but never had the time. Nice work Fredric.
> >
> > Best
> > Nasos
> >
> >
> >> From: frederic.devernay_at_[hidden]
> >> Date: Fri, 30 Apr 2010 15:48:22 +0200
> >> To: ublas_at_[hidden]
> >> Subject: [ublas] announcement: ublasJama (ublas port of Jama, including
> >> SVD and eigenvalue decomposition)
> >>
> >> Hi,
> >>
> >> I translated Jama to ublas, mainly because I needed SVD. It's not optimal
> >> nor generic, but at least it's here. Note that the bug in the Jama SVD was
> >> fixed too.
> >>
> >> cat flames > /dev/null
> >>
> >> Fred
> >>
> >> Download it from http://devernay.free.fr/hacks/ublasJama.html
> >>
> >> * What is ublasJama?
> >>
> >> ublasJama is translation of the public domain Jama (Java Matrix Package)
> >> to C++ using boost.ublas for the linear algebra. Jama itself is a Java
> >> implementation of algorithms for dense matrices found in LINPACK and
> >> EISPACK.
> >>
> >> Author of the ublas translation: Frederic Devernay .
> >>
> >> Documentation: see the Jama documentation.
> >>
> >> Original Authors: see the Jama list of authors.
> >>
> >> No optimization of the original code was done, but the functionality is
> >> here:
> >>
> >> • Cholesky Decomposition of symmetric, positive definite matrices
> >> • LU Decomposition (Gaussian elimination) of rectangular matrices
> >> • QR Decomposition of rectangular matrices
> >> • Eigenvalue Decomposition of both symmetric and nonsymmetric square
> >> matrices
> >> • Singular Value Decomposition of rectangular matrices
> >> There are two test programs:
> >>
> >> • TestMatrix, in subdirectory test - the original Jama tests, where only
> >> the relevant tests were kept.
> >> • MagicSquareExample, in subdirectory examples
> >>
> >> _______________________________________________
> >> ublas mailing list
> >> ublas_at_[hidden]
> >> http://lists.boost.org/mailman/listinfo.cgi/ublas
> >> Sent to: nasos_i_at_[hidden]
> >
> > ________________________________
> > The New Busy think 9 to 5 is a cute idea. Combine multiple calendars with
> > Hotmail. Get busy.
> > _______________________________________________
> > ublas mailing list
> > ublas_at_[hidden]
> > http://lists.boost.org/mailman/listinfo.cgi/ublas
> > Sent to: tom.vercauteren_at_[hidden]
> >
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas
> Sent to: david.bellot_at_[hidden]
>
>
>
> --
> David Bellot, PhD
> david.bellot_at_[hidden]
> http://david.bellot.free.fr
>
> _______________________________________________
> ublas mailing list
> ublas_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/ublas
> Sent to: frederic.devernay_at_[hidden]