Mathics

[Ralf Hemmecke]

Not that I ever used Mathics, but I find the way they present themselves
impressive.

https://mathics.org/

Obviously, the poster of this message
https://groups.google.com/g/sage-devel/c/XJUnM4HEUG4/m/LXUU2iLdAAAJ
found another project to invest his time and efforts.

Ralf

[Waldek Hebisch]

I wonder what impressed you? I recently looked at available
parsers for Mathematica syntax. In the process also took
a quick look at Matics. I was unimpressed.

--
Waldek Hebisch

[Ralf Hemmecke]

>> https://mathics.org/

> I wonder what impressed you? I recently looked at available parsers
> for Mathematica syntax. In the process also took a quick look at
> Matics. I was unimpressed.

I am not saying that Mathics is better than FriCAS. Probably not.

However, the website looks more professional than that of FriCAS.
(I bet that more people use other browsers than Lynx.)

Do we have descriptions of all the Graphics demos in
https://mathics.org/docs/mathics-latest.pdf ? I even wonder whether
FriCAS can do them at all.

Number of active developers.

As far as I can feel, FriCAS is well known for its integration
capabilities and thus used as a subprogram in FriCAS, but is someone
using it as a general purpose computer algebra system? Do you see the
latter as a goal for FriCAS?

Ralf

[Waldek Hebisch]

On Fri, Sep 10, 2021 at 04:20:04PM +0200, Ralf Hemmecke wrote:
> >> https://mathics.org/
>
> > I wonder what impressed you? I recently looked at available parsers
> > for Mathematica syntax. In the process also took a quick look at
> > Matics. I was unimpressed.
>
> I am not saying that Mathics is better than FriCAS. Probably not.
>
> However, the website looks more professional than that of FriCAS.
> (I bet that more people use other browsers than Lynx.)
>
> Do we have descriptions of all the Graphics demos in
> https://mathics.org/docs/mathics-latest.pdf ? I even wonder whether
> FriCAS can do them at all.

ATM we do not have bar charts and pie charts, user would have to
synthetise them from primitives. Few others are superposition
of two operations.

> Number of active developers.
>
> As far as I can feel, FriCAS is well known for its integration
> capabilities and thus used as a subprogram in FriCAS, but is someone
> using it as a general purpose computer algebra system? Do you see the
> latter as a goal for FriCAS?

Well, I do not only develop FriCAS, I also use it. I find it better
than open source alternatives. Most what I do could be done using
other systems. But I think not all. For example, I fequently
use equation solving. I Maxima list from time to time come
questions "those eqations take too mauch time or run out of
memeory, what can be done to solve them faster (or at all)".
I have tried few such things and they were no problem for
FriCAS. Most recent example was:

eq1 := N = a1 + b1 * x + c1 * y + d1 * x * y + e1 * x^2 + f1 * y^2
eq2 := P = a2 + b2 * x + c2 * y + d2 * x * y + e2 * x^2 + f2 * y^2
solve([eq1,eq2],[x,y])

Maxima folks were unable to compute anwer and "explained" that
this system is too complicated and one should not expect answer
from computer. In FriCAS on my machine it takes 9.5 seconds
(answer takes several screens). In the past I did some
experiments with Sympy and largish linear systems with symbolic
coefficients which worked fine in FriCAS were too hard for Sympy.
Sage has some very good specialized solvers, but AFAIK once things
get general enough it sends problems to Maxima or Sympy.

Of course, there things missing from FriCAS, but IMO FriCAS
is pretty strong as "general purpose CAS".

--
Waldek Hebisch

[Ralf Hemmecke]

> Maxima folks were unable to compute anwer and "explained" that
> this system is too complicated and one should not expect answer
> from computer. In FriCAS on my machine it takes 9.5 seconds
> (answer takes several screens). In the past I did some
> experiments with Sympy and largish linear systems with symbolic
> coefficients which worked fine in FriCAS were too hard for Sympy.
> Sage has some very good specialized solvers, but AFAIK once things
> get general enough it sends problems to Maxima or Sympy.

I would be very much in favour of collecting such "benchmark" examples
where FriCAS performs better and other systems and make that public so
that we can have a "selling argument".

Would you be able to show-case a number of your comparison experiments.

> Of course, there things missing from FriCAS, but IMO FriCAS
> is pretty strong as "general purpose CAS".

Yes, yes, once you know FriCAS, it is easy to see that. But other people
are not easily convinced. It's like convincing someone to change from
emacs to vi or the other way round. And we know that learning FriCAS is
not that easy so we should put more emphasis on showing a lot of
convincing arguments.

Ralf

[Waldek Hebisch]

On Fri, Sep 10, 2021 at 10:34:02PM +0200, Ralf Hemmecke wrote:
> > Maxima folks were unable to compute anwer and "explained" that
> > this system is too complicated and one should not expect answer
> > from computer. In FriCAS on my machine it takes 9.5 seconds
> > (answer takes several screens). In the past I did some
> > experiments with Sympy and largish linear systems with symbolic
> > coefficients which worked fine in FriCAS were too hard for Sympy.
> > Sage has some very good specialized solvers, but AFAIK once things
> > get general enough it sends problems to Maxima or Sympy.
>
> I would be very much in favour of collecting such "benchmark" examples
> where FriCAS performs better and other systems and make that public so
> that we can have a "selling argument".
>
> Would you be able to show-case a number of your comparison experiments.

I have put one or two cases on FriCAS Wiki. However, getting each
to presentable form requires effort and IMO it is better to spent
effort on other things.

> > Of course, there things missing from FriCAS, but IMO FriCAS
> > is pretty strong as "general purpose CAS".
>
> Yes, yes, once you know FriCAS, it is easy to see that. But other people
> are not easily convinced. It's like convincing someone to change from
> emacs to vi or the other way round. And we know that learning FriCAS is
> not that easy so we should put more emphasis on showing a lot of
> convincing arguments.

I think that we need more examples and "recipes": seeing how to
do something that user want to do is IMO more convincing than
other arguments.

--
Waldek Hebisch

[Alasdair McAndrew]

As far as I can tell, Mathics uses Mathematica syntax with an engine that
runs Python SymPy. It's nice looking, but I don't think (comparing Sympy
and FriCAS) that FriCAS has anything to worry about. However, a nice
interface goes a long way to winning people, and the best system in the
world will always remain a hobby of a very few unless it lifts its external
game.

--
0432 854 858
https://numbersandshapes.net