[Cosmo-torun] copying algorithms: GNU Scientific Library vs Numerical Recipes

Śro, 17 Lis 2004, 14:08:56 CET

On Mon, 15 Nov 2004, Michal Frackowiak wrote:

> Oh, I have just read this email. This is somehow terrifying.
> So the answer should be "don't care" I suppose. In this case the authors
> of GSL are the one to blame and not the user of GSL.
>
> I agree the copyright of algorithms' code is tricky. I wonder if anyone
> understands it....

i guess we have to agree to disagree on this. To me, all the stuff i've
previously cited says "algorithms are not copyrighted, only the way they
are expressed (especially arbitrary and stylistic choices) is copyrighted".

So to me the *principle* is clear, it's just the application in practice
which is difficult.

So the question is: what part is algorithm, what part is expression,
especially arbitrary and stylistic choices?

> I think GSL violates the NR license. But it is their problem.

OK, so you are consistent. :)

Maybe you should write to Richard Stallman ;)

> eeee..... perhaps we should just use it? ;-)

Well, we probably have consensus on that. :)

Anyway, any concrete comments, bug reports (or revised versions or forked
distributions ;) of
http://cosmo.torun.pl/GPLdownload/dodec/isolat-0.1.11.tar.gz
would be welcome.

As long as there's only a small number of developers, we probably
don't need a cvs system.

pozdr
boud

> Boud Roukema wrote:
> > Witam,
> >   What can be legally (and morally) copied from a program
> > under a restrictive copyright is "the algorithms, the ideas" but not
> > the particular way of expressing them.
> >
> > So the practical question is: where is the borderline?
> >
> > In the GSL (GNU Scientific Library), http://www.gnu.org/software/gsl/,
> > GNU provides, e.g. in Version 1.1.1, the version below of NR's
> > ran0 routine.
> >
> > i'm sure most of us have access to a legal copy of the NR routine,
> > or else a little googling shows e.g.
> >
> > http://cc.oulu.fi/~tf/tiedostot/pub/nrf/ran0.f
> >
> > Well, to me these look extremely similar to one another. Even the
> > constants:
> >
> > static const long int m = 2147483647, a = 16807, q = 127773, r = 2836;
> > static const unsigned long int mask = 123459876;
> >
> > are exactly identical to those in the NR routines, they even have the
> > same names, except for the "i".  Is this really a different "expression"?
> >
> >
> > In any case, IMHO noone can deny that the authors of the GSL routines
> > (there is also ran1, ran2, ran3) have very likely had access to the NR
> > source code.
> >
> > Sure, they translated it to C and added some stuff, but it's the
> > same algorithm, it's essentially the same sequence of steps.
> >
> >
> > Anyway, GSL has lots of really cool stuff. :)   Let's defend scientific
> > software freedom!
> >
> >
> > pozdr
> > boud
> >
> >
> > ----------------------------------------------------------------------
> >
> > /* rng/ran0.c
> >  *
> >  * Copyright (C) 1996, 1997, 1998, 1999, 2000 James Theiler, Brian Gough
> >  *
> >  * This program is free software; you can redistribute it and/or modify
> >  * it under the terms of the GNU General Public License as published by
> >  * the Free Software Foundation; either version 2 of the License, or (at
> >  * your option) any later version.
> >  *
> >  * This program is distributed in the hope that it will be useful, but
> >  * WITHOUT ANY WARRANTY; without even the implied warranty of
> >  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> >  * General Public License for more details.
> >  *
> >  * You should have received a copy of the GNU General Public License
> >  * along with this program; if not, write to the Free Software
> >  * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
> >  */
> >
> > #include <config.h>
> > #include <stdlib.h>
> > #include <gsl/gsl_errno.h>
> > #include <gsl/gsl_rng.h>
> >
> > /* This is an implementation of the algorithm used in Numerical
> >    Recipe's ran0 generator. It is the same as MINSTD with an XOR mask
> >    of 123459876 on the seed.
> >
> >    The period of this generator is 2^31.
> >
> >    Note, if you choose a seed of 123459876 it would give a degenerate
> >    series 0,0,0,0, ...  I've made that into an error. */
> >
> > static inline unsigned long int ran0_get (void *vstate);
> > static double ran0_get_double (void *vstate);
> > static void ran0_set (void *state, unsigned long int s);
> >
> > static const long int m = 2147483647, a = 16807, q = 127773, r = 2836;
> > static const unsigned long int mask = 123459876;
> >
> > typedef struct
> >   {
> >     unsigned long int x;
> >   }
> > ran0_state_t;
> >
> > static inline unsigned long int
> > ran0_get (void *vstate)
> > {
> >   ran0_state_t *state = (ran0_state_t *) vstate;
> >
> >   const unsigned long int x = state->x;
> >
> >   const long int h = x / q;
> >   const long int t = a * (x - h * q) - h * r;
> >
> >   if (t < 0)
> >     {
> >       state->x = t + m;
> >     }
> >   else
> >     {
> >       state->x = t;
> >     }
> >
> >   return state->x;
> > }
> >
> > static double
> > ran0_get_double (void *vstate)
> > {
> >   return ran0_get (vstate) / 2147483647.0 ;
> > }
> >
> > static void
> > ran0_set (void *vstate, unsigned long int s)
> > {
> >   ran0_state_t *state = (ran0_state_t *) vstate;
> >
> >   if (s == mask)
> >     {
> >       GSL_ERROR_VOID ("ran0 should not use seed == mask",
> > 				GSL_EINVAL);
> >     }
> >
> >   state->x = s ^ mask;
> >
> >   return;
> > }
> >
> > static const gsl_rng_type ran0_type =
> > {"ran0",			/* name */
> >  2147483646,			/* RAND_MAX */
> >  1,				/* RAND_MIN */
> >  sizeof (ran0_state_t),
> >  &ran0_set,
> >  &ran0_get,
> >  &ran0_get_double};
> >
> > const gsl_rng_type *gsl_rng_ran0 = &ran0_type;
> >
> >
> > ----------------------------------------------------------------------
> >
> > _______________________________________________
> > Cosmo-torun mailing list
> > Cosmo-torun at cosmo.torun.pl
> > http://cosmo.torun.pl/mailman/listinfo/cosmo-torun
>
> --
> --------------------------------------------------------
> Michal Frackowiak
> mail: michalf at ncac.torun.pl
> www: http://www.ncac.torun.pl/~michalf
> jabber im: michal_frackowiak at jabber.pl
> phone: +48 (56) 6219319 int. 22, fax: +48 (56) 6219381
> --------------------------------------------------------
> Nicolaus Copernicus Astronomical Center
> Department of Astrophysics in Torun (CAMK Torun)
> ul. Rabianska 8, 87-100 Torun, Poland
> http://www.ncac.torun.pl, http://www.camk.edu.pl
> --------------------------------------------------------
>