Re: C library for numerical analysis and math
See my comments below.
On Fri, 2004-02-06 at 19:04, Halim Boukaram wrote:
> I don't know enough to really know, but it sure sounds
> impressive.
>
> Presumably you are aware of the various GNU projects for these
> sorts of
> libraries?
>
> If not, go hunting from www.gnu.org -> software or something
> like that.
>
> If so, what does your library bring to the table that the
> existing
> libraries don't have?
>
> What license are you planning to use?
>
> If you are planning to use GPL or LGPL or MIT/X style license,
> is there
> room for cooperation/ integration with existing (namely GNU)
> projects
> that are similar?
>
> And a big welcome aboard to Debian, The Universal Operating
> System (TM).
>
> cheers
> zen
>
> --
> * Debian Enterprise: http://debian-enterprise.org/
> * Homepage: http://soulsound.net/
> * PGP Key: http://soulsound.net/zen.asc
> * Please respect the confidentiality of this email as sensibly
> warranted.
>
>
> --
> To UNSUBSCRIBE, email to
> debian-mentors-request@lists.debian.org
> with a subject of "unsubscribe". Trouble? Contact
> listmaster@lists.debian.org
>
>
> ////////////////////////////////////////////////////////////////////////////////////////////////////////
>
> //////////////////////////////////////////////////////////////////////////////////////////////////////
>
> Hi
>
> The closest thing to my own library is Octave.
>
> the main differences are:
>
> pluses for mine:
>
> 1) Octave only allows Ordinary differential equations of a
> very simple form. My library can take any ode of any form
>
> 2) there is no support for general least-squares methods in
> Octave mine done.
>
> 3) octave can only solve equations, differientail equations,
> integrals, by one method. mine can solve them using many
> different numerical methods.
>
> 4) octave han only handle continous integrals and fourier
> transforms mine can also handle discrete.
>
> 5) octave seems to have many dependancies not included in
> default libraries and programs mine is programmed from
> scratch.
>
> negatives for mine:
>
> 1) no financial, image or audio stuff.
>
> 2) no statististical stuff(yet)
>
> ///////////////////////////////////////////////////////////////////////////////////////////////////////
>
> I'm fairly sure that my library is more powerful for
> mathematical stuff and pure numerical solutions.
>
> as mentioned before my library has more than one way of
> integrating or differentiating ex i have 6 seperate functions
> for integrating which are computer simulations of known rules
> such as trapizoidal, boole, simpson, simpson3/8,
> trapizoidal-4,trapizoidal-8.
>
> /////////////////////////////////////////////////////////////////////////////////////////////////////
If you want decent comments and feedback, you have to reply-all to the
list, so that those who actually know about maths libraries can comment.
I can't do this because I don't know about them. If you do reply-all, it
is generally better to remove the address of the person you are replying
to (unless they ask for that) so that they don't get two copies of your
message (which is annoying).
> how do i get the library evaluated?
Sell it, or give demo copies away, or make it Free Software and set up a
web page so that other people in the Free Software community will be
interested in helping out/ supporting - but that will only happen if you
use a Free Software license.
> its all done with help file and working examples.
>
> If possible id like to keep it closed source. i dont want any
> one ripping me off. Thats happened to me alot and is really
> frustrating having someone else claim the glory for your work.
>
> is there such a thing as closed source under linux? if not
> then i'll take the closest licence.
You might try packaging for debian "non-free" archive, but there are
plans to remove that from Debian.
> PS: I could make head or tail of the GPL, LGPL MIT/X licences
> that you pointed out and thats the third time that i've read
> them in detail.
That's probably because you don't have clear goals - what do you want to
achieve - eg. fame and glory, money, resume kudos, etc ??? All of these
can be achieved with Free Software, but you might have to be a little
bit inventive and do a bit of research.
> power to the penguin
Juding by your conflict of licensing desires, you are perhaps "power to
popularity". It's all well and good, but the point of Free Software, and
GNU/Linux operating systems such as Debian, is to create software that
has certain Freedoms available with it. You have read the GPL so you
should at least start to understand this.
Once you really Get It (TM), you would be much more likely to put your
effort to work with existing projects, rather than try to compete - no
matter what you think, your library will have bugs, and it takes many
people and many many hours for bugs to be found and fixes to be made.
Either you sell your library as Proprietary software and pay for the
work that way, or you make it Free Software, and get the community
involved. If you sit on the fence in the middle, you will get the worst
of both worlds - I recommend the community approach for a library such
as yours, since there are already plenty of commercial alternatives you
will probably find it hard to sell...
As to licenses, the most open Free Software license is X/MIT license -
it is closest to public domain.
The most closed Free Software license is the GNU General Public License
(GPL). It gives you most protection, assuming that matches your goals.
> thanks for your help.
I really recommend that you read a bunch of the philosophy pages on the
http://www.gnu.org/ web site (there is a link to "GNU Philosophy").
There are a lot of papers, and they will give you a good background
understanding that you don't get in the GPL. Then you will (hopefully)
know much more clearly what you want to achieve.
Finally, to help you decide which license to use, one paper The Zen
Meister recommends you to read is this:
http://www.gnu.org/licenses/why-not-lgpl.html
Good luck,
Zen
Reply to: