Packages for Ubuntu, Function library

[Alexander Hunziker]

I had requested Ubuntu packages from the GetDeb.net project, and
meanwhile they are available. Maybe you can put a link to your
Download page to make it easier for other Ubuntu users to find the
packages.

Otherwise, I'd like to say that I find FreeMat quite awesome. I threw
some old MATLAB scripts of mine at it, and at least part of them it
just executed without any complaints. Besides, the documentation is on
a very good level. I would reckon that one necessary improvement to
get more users would be a more complete and MATLAB compatible function
library. What are your plans there? Big parts could probably pretty
much be copy-pasted from Octave, seeing that it is GPL'ed and the
syntax being pretty much the same, no?

[Samit Basu]

Hi Alexander,

Thanks for the pointer (Demetrios, can you add the link?). As for the
Octave issue, it has been suggested before. There are some challenges
to a direct copy-and-paste, but we are working on expanding FreeMat's
function library. At the moment, our primary focus is on FreeMat 4,
which aims to increase the compatibility with MATLAB substantially,
but it is nearing completion, and once released, we will return to
expanding the function library.

Samit

[Soma S Dhavala]

Hey Guys:

I am new comer to this group and I have not used it either. What drew me to
FreeMat was: I was searching for an open source package that offers
functionality similar to MATLAB. The motivation for me is to develop
computing platform that is deployable on mobile devices such as a
smart-phones or PDAs.

Is it possible to re-code, re-implement FreeMat say targeting at Andriod or
Symbian etc.. Any ideas are welcome and if you share your knowledge in
developing FreeMat that would be highly appreciated.

regards,
-soma
PS: Samit, incidentally, I worked for General Electric JFWTC Bangalore and
as soon as I saw your name as the FreeMat developer/founder, I could
recognize you (though you don't know me).

[Eugene]

Alexander,

I've been looking at porting scripts from Octave for some time. The
problem is that most octave scripts use incompatible octave specific
syntax. I wrote a short python script that corrects most of the
obvious incompatibilities (endfor->end, endif->end, #->%, double
quotes->single quotes, etc..)

Unfortunately it still leaves a lot of manual work to port octave
scripts to standard syntax that is used by Freemat. Also, conversion
makes it hard to keep ported scripts up to date with the octave-forge
scripts.

I'm going to work more on that as release 4 gets under way. Also, hope
that user community can help us with this too.

Eugene

On Jul 30, 1:21 am, Alexander Hunziker <alex.hunzi...@gmail.com>
wrote:

[Eugene]

Soma,

From what I read Android supports only Java code and Freemat is
written in C++/Fortran. In addition JVM running on Android can't JIT
compile java code, so any computation will be painfully slow.

Symbian may be a better choice since you can run Qt under it (and
Freemat uses Qt a lot). However, you are still going to be limited by
the scant amount of memory that is available and fairly slow ARM cpu.

Finally, Freemat interface is really not designed for tiny screens.

Eugene

[Samit Basu]

Hi Soma,

I am pleased to meet you. Although I agree completely with Eugene, it
would still be very cool to get FreeMat running on e.g., an iPhone. =)

Samit

[bbl...@fh-lausitz.de]

Eugene schrieb:

Hi Soma,

There is a not so broadly known nice open source project jmathlib, a
Mathlab clone written in Java (http://www.jmathlib.de/) under the LGPL.

It is a german andress, but the content is mostly in english anyway. I
think with Java and swing you have a better chance to port to a small
device than with C/Fortran and Qt - if any. (But I have heard Nokia
bought Qt?)

Check yourself if that code might be ported to the KVM of Java ME. If,
say for instance a function plot of such a small low resolution device
could have meaning? Even MIDP 3.0 only offers 176x220 pixel with a 16
bit colour depth. Well, that is a minimum requirement, but does swing of
Java ME support higher resolutions? I don`t know.

Kind regards

BB

[Soma S Dhavala]

Thank you for the information and nice meeting you Samit. I really admire
you folks for developing a platform on your own, great job indeed.

I am very keen in developing computing platform for use on mobile OS or
those tiny devices. Do you guys suggest me learn Qt and "think" about
developing a miniature version of FreeMat.
Btw, I am doing PhD in Statisitcs at Texas A & M University, and offlate I
became very much interested in mobile/ubiquous computing and am an old user
of MATLAB :)

thank you,
-soma

[demetrios...@googlemail.com]

> I am very keen in developing computing platform for use on mobile OS or
> those tiny devices. Do you guys suggest me learn Qt and "think" about
> developing a miniature version of FreeMat.

I think not :).
The long term purpose(and requirement form many users) of FreeMat
would be to "abstract" and separate
the "engine" of FreeMat (i.e. that part should be than independent of
Qt), to be easily reusable
and integrated in other custom applications.

Please don't get me wrong, this doesn't mean ever to deliver FreeMat
without a GUI (as
it was interepreted in my previous posts last year :) ), since this is
the main agrument of FreeMat (that everything
is in one package that just works) but to have the logical separation,
so that when desired, to be very easily usable
with another GUI (mostly not as a concurrence to the actual QT GUI,
but as custom purpose GUI - e.g. a UI in Java for automotive, or for
some platform where QT does not even exists or is not good enough).

I think this approach would open the use of FreeMat engine in mobile
computing too, since the engine would
be practically the same, just the UI would be another one, adapted
more the lower resolutions (and other restrictions,
or .features like touchscreen that mobile devices have).

Demetrios.

[Alexander Hunziker]

I would be willing to invest some work into that, abilities permitting
(I know a little something about math and know the M syntax). Maybe
it's
best to wait until FreeMat 4 is out (what's the timeframe anyway)?
Also,
without a detailed plan it doesn't make sense to attack that. Ideally,
we'd have some user input about which function libraries are most
missed, and distribute little chunks of the task to people willing to
help? Also, if a test suite is available somewhere that checks if
functions perform as they should in all cases, the results should be
validated against that.

Cheers,
Alex