* GSL and ALGLIB
@ 2010-04-14 8:24 Sergey Bochkanov
2010-04-14 8:59 ` Brian Gough
0 siblings, 1 reply; 4+ messages in thread
From: Sergey Bochkanov @ 2010-04-14 8:24 UTC (permalink / raw)
To: gsl-discuss
Hello!
What do GSL developers think about using source code from ALGLIB in
GSL?
ALGLIB is an open source numerical analysis library distributed under
GPL 2+. It uses automatic translation from specially designed
pseudocode to provide same functionality in C++, C#, FreePascal and
other programming languages. Project aims to be highly portable,
compiler- and OS-independent. It is actively developed (new releases
each month).
ALGLIB contains many interesting algorithms which are missing in GSL.
For example, it includes:
* limited memory BFGS optimizer
* improved Levenberg-Marquardt optimizer
* bound constrained optimizer
* 2-dimensional interpolation (bilinear/bicubic splines)
* scattered N-dimensional interpolation/fitting with linearithmic
complexity.
It should be easy to write GSL-ish wrapper for ALGLIB. I can help if
someone wants to implement it. However it is unclear what is GSL devs
position on using external libraries. I've read past discussions on
FLAME and FFTW, but there was no decision.
Sergey Bochkanov.
--
With best regards,
Sergey mailto:sergey.bochkanov@alglib.net
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: GSL and ALGLIB
2010-04-14 8:24 GSL and ALGLIB Sergey Bochkanov
@ 2010-04-14 8:59 ` Brian Gough
2010-04-14 15:11 ` Sergey Bochkanov
0 siblings, 1 reply; 4+ messages in thread
From: Brian Gough @ 2010-04-14 8:59 UTC (permalink / raw)
To: Sergey Bochkanov; +Cc: gsl-discuss
At Tue, 13 Apr 2010 14:20:51 +0400,
Sergey Bochkanov wrote:
> What do GSL developers think about using source code from ALGLIB in
> GSL?
>
> ALGLIB is an open source numerical analysis library distributed under
> GPL 2+. It uses automatic translation from specially designed
> pseudocode to provide same functionality in C++, C#, FreePascal and
> other programming languages. Project aims to be highly portable,
> compiler- and OS-independent. It is actively developed (new releases
> each month).
>
> ALGLIB contains many interesting algorithms which are missing in GSL.
>
> For example, it includes:
> * limited memory BFGS optimizer
> * improved Levenberg-Marquardt optimizer
> * bound constrained optimizer
> * 2-dimensional interpolation (bilinear/bicubic splines)
> * scattered N-dimensional interpolation/fitting with linearithmic
> complexity.
>
> It should be easy to write GSL-ish wrapper for ALGLIB. I can help if
> someone wants to implement it. However it is unclear what is GSL devs
> position on using external libraries. I've read past discussions on
> FLAME and FFTW, but there was no decision.
Hello Sergey
Thanks for your email. There are two concerns here, licensing and
technical.
Regarding licensing, I think we corresponded about ALGLIB a couple of
years ago and I asked if the underlying converter was free software
(which it was not at that time). The GPL defines source code as "the
preferred form of the work for making modifications to it" -- which in
this case is the underlying algorithm definitions for the converter,
not the individual routines. We could only consider using other code
if all the associated software (and documentation) is free.
Whether it makes sense techically to use ALGLIB in GSL, it is
difficult for me to comment without the converter being free software
and available to study.
I do think the principle of converting to different languages from a
common source is a good one and would encourage you to consider
releasing everything under a free software license, whether or not we
might use parts of it.
--
best regards
Brian Gough
GNU Scientific Library -
http://www.gnu.org/software/gsl/
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: GSL and ALGLIB
2010-04-14 8:59 ` Brian Gough
@ 2010-04-14 15:11 ` Sergey Bochkanov
0 siblings, 0 replies; 4+ messages in thread
From: Sergey Bochkanov @ 2010-04-14 15:11 UTC (permalink / raw)
To: gsl-discuss
Hello, Brian.
You wrote 14 апреля 2010 г., 12:58:38:
> The GPL defines source code as "the
> preferred form of the work for making modifications to it" -- which in
> this case is the underlying algorithm definitions for the converter,
> not the individual routines. We could only consider using other code
> if all the associated software (and documentation) is free.
Yes, translator is still non-free. I think that some day I'll make it
free, but it isn't top priority for me. Just can't come to some
decision on this subject, so I am leaving this question open for a
while. And, actually, ALGLIB users rarely need something beyond the
version in the language they use.
> Whether it makes sense techically to use ALGLIB in GSL, it is
> difficult for me to comment without the converter being free software
> and available to study.
95% of C++ sources are as human-readable as original pseudocode.
Remaining 5% are places where C++ is used to emulate continuations -
feature lacking in many modern programming languages. However, it can
be made human-readable too (just one day or two to tweak translator)
if someone needs it. It wasn't deliberately obfuscated :)
Technically, ALGLIB can be linked to GSL and can be used under GPL
even without translator.
--
With best regards,
Sergey mailto:sergey.bochkanov@alglib.net
^ permalink raw reply [flat|nested] 4+ messages in thread
* GSL and ALGLIB
@ 2010-04-13 14:17 Sergey Bochkanov
0 siblings, 0 replies; 4+ messages in thread
From: Sergey Bochkanov @ 2010-04-13 14:17 UTC (permalink / raw)
To: gsl-discuss
Hello!
What do GSL developers think about using source code from ALGLIB in
GSL?
ALGLIB is an open source numerical analysis library distributed under
GPL 2+. It uses automatic translation from specially designed
pseudocode to provide same functionality in C++, C#, FreePascal and
other programming languages. Project aims to be highly portable,
compiler- and OS-independent. It is actively developed (new releases
each month).
ALGLIB contains many interesting algorithms which are missing in GSL.
For example, it includes:
* limited memory BFGS optimizer
* improved Levenberg-Marquardt optimizer
* bound constrained optimizer
* 2-dimensional interpolation (bilinear/bicubic splines)
* scattered N-dimensional interpolation/fitting with linearithmic
complexity.
It should be easy to write GSL-ish wrapper for ALGLIB. I can help if
someone wants to implement it. However it is unclear what is GSL devs
position on using external libraries. I've read past discussions on
FLAME and FFTW, but there was no decision.
--
With best regards,
Sergey mailto:sergey.bochkanov@alglib.net
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2010-04-14 15:11 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2010-04-14 8:24 GSL and ALGLIB Sergey Bochkanov
2010-04-14 8:59 ` Brian Gough
2010-04-14 15:11 ` Sergey Bochkanov
-- strict thread matches above, loose matches on Subject: below --
2010-04-13 14:17 Sergey Bochkanov
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).