symbolic equation solver

[nev]

Hi all!

I'm developing an educational math software, and I need a symbolic equation
solver. I have been searching for months, and found this code:
http://dream.inf.ed.ac.uk/software/press/press-src.tgz

It is Prolog Equation Solving System developed in 1985. I'm new in prolog,
and I failed to compile it. Please help me! (I'm using Win7)

I would like to advice on the primary issue: what is the proper way to
develop a symbolic equation solver in 2010?

Thank you!

Balazs Sebestyen
Budapest, Hungary

[Chip Eastham]

Hi, Balazs:

What kind of equations is the solver required to
handle? At some point any reasonably complex
software must stabilize around an equally clear
specification.

Solutions of polynomial equations in one variable
such as one might encounter in "gymnasium" (high
school) can be solved in deterministic fashion.
Perhaps this is what you have in mind. There are
more sophisticated problems tackled symbolically
by CAS packages like Maxima or Axiom:

[Maxima, A Computer Algebra System -- Sourceforge]
http://maxima.sourceforge.net/

[Axiom Computer Algebra System]
http://www.axiom-developer.org/

Software of this kind would seem to be a good match
for implementation in Prolog, but historically Lisp
has been the preferred language. "Classic" Axiom is
also Lisp dependent, but its implementation spawned
a higher level language spad as the means of defining
object types, e.g. multivariate polynomials with
coefficients over a custom defined ring. spad itself
gave rise to a new language Aldor, and there are a
couple of quite active "forks" of Axiom that emphasize
a change in implementation strategy:

[FriCAS - an advanced CAS (Sourceforge)]
http://fricas.sourceforge.net/

[OpenAxiom: The Open Scientific Computation Platform]
http://www.open-axiom.org/

While I'd be interested in helping out on a project
to do "symbolic equation" solving similar to this in
Prolog (perhaps in Logtalk to benefit from OO syntactic
sugar), a simpler tack might be to look into constraint
solvers built on Prolog, at least with regard to finite
domain problems:

[CLPFD Intro - SICStus Prolog]
http://www.sics.se/sicstus/docs/4.1.1/html/sicstus/CLPFD-Intro.html

regards, chip

[A.L.]

On Fri, 5 Feb 2010 20:53:23 -0800 (PST), Chip Eastham
<hard...@gmail.com> wrote:

>O

>
>[CLPFD Intro - SICStus Prolog]
>http://www.sics.se/sicstus/docs/4.1.1/html/sicstus/CLPFD-Intro.html
>

I would rather suggesr CLPQR

http://www.sics.se/sicstus/docs/4.1.1/html/sicstus/lib_002dclpqr.html#lib_002dclpqr

By the way, both CLP(FD) and COP(QR) are not symbolic in the same way
as, say, Mathematica.

Check here, these are tools that you could use in your project

http://issc.uj.ac.za/symbolic/symbolic.html

and here

http://www.isi.edu/isd/LOOM/Stella/

http://jscl-meditor.sourceforge.net/

A.L.

[nev]

hi Chip!

> What kind of equations is the solver required to
> handle?

It's not exactly specified. Polynomial, trigonometric, logarithmic,
exponential equations, equations containing roots, and their combinations...
The main problem is the equations usually are not in standard form, and
their type can not be determined. For example:

2^(x+3) - 4^x = 8

This is an exponential equation which is reducible to quadratic. Maple,
Derive, Mathematica and some other expensive (and not reusable) programme
can solve equations like above, but most of the CAS-s can not. According its
descripton Axiom works only with linear equations. I tried Maxima, and it
can't solve too.

PRESS (in my previous message) can solve it, but it is a 25 years old
programme... There is a short description how PRESS works (very
interesting):
http://books.google.hu/books?id=w-XjuvpOrjMC&lpg=PA439&ots=4WD0YIEZTt&pg=PA439#v=onepage&q=&f=false

> a simpler tack might be to look into constraint
> solvers built on Prolog, at least with regard to finite
> domain problems:

Thanks for the advice!

Best regards: Balazs

[jacinto...@gmail.com]

Hi Balazs,

Did you actually find anything definite solution after this?. Did you stay in Prolog's PRess?

Regards,

[cc.car...@gmail.com]

Hi Balazs

I think you could be interested to this SW (http://www.cs.cmu.edu/afs/cs/project/ai-repository/ai/lang/prolog/code/math/algebra/0.html). I'm sorry I haven't tried it.

HTH Carlo