comp.lang.prolog Google Group

Re: Is Prolog the best language to solve this kind of problem?

janbu...@fastmail.fm (Jan Burse) — Sun, 21 Mrz. 2010 00:19:34 UT

A.L. schrieb:
Well the following should be Folk knowledge:
1) It is pretty clear that the problem cannot be solved declaratively
with normal Prolog, since the declarative semantic corresponds
to Horn Clauses with a restricted form of search. And Horn Clause
don't know about algebraic formulas.
2) Further it is pretty clear that any algorithmic problem can be

Re: Quintus --> Sicstus

ulr...@mips.complang.tuwien.ac.at (Ulrich Neumerkel) — Sat, 20 Mrz. 2010 21:22:44 UT

ISO 13211-1 demands in 5.1 e such a strict mode that rejects *all*
implementation specific extensions. However, the precise way how this
strict mode is provided is not specified. That is, a conforming
implementation has to document how to enable strict mode in accordance
with 5.4.
The absence of one particular flag for strict mode is for very good

Re: Is Prolog the best language to solve this kind of problem?

milan.babus...@gmail.com (Milan Babuskov) — Sat, 20 Mrz. 2010 21:29:06 UT

Reading that book I have learned how to solve this (now I'm off to
Amazon to buy a copy).
I wish to thank everyone who replied. In managed to get correct
solutions for my small test problem. Soon, I will try it out on few
thousand variables to see how it scales. If MIP shows to be slow I
might look into other solutions you guys have suggested (BTW, I did

Re: Is Prolog the best language to solve this kind of problem?

kishs...@yahoo.com (Kish Shen) — Sat, 20 Mrz. 2010 20:19:26 UT

A.L. wrote:
I am not sure I understand what you mean by 'defining problem on the
matrix level'. Can you expand on this? As a normal user of eplex, you
certainly don't need to normalise your constraints or know how the
constraints are mapped to the matrix that a MP solver use for solving
the problem.

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Sat, 20 Mrz. 2010 18:51:35 UT

On Sat, 20 Mar 2010 18:10:41 +0100, Jan Burse <janbu...@fastmail.fm>
wrote:
Do you know what you are taking about?... Wnd what the original
question was?...
A.L.

Re: Is Prolog the best language to solve this kind of problem?

janbu...@fastmail.fm (Jan Burse) — Sat, 20 Mrz. 2010 17:10:41 UT

A.L. schrieb:
Yes, full formulation would be:
P(x1,y1,z) & ~exists x2,y2 (P(x2,y2,z) & y2<y1))
where z are the true parameters. z and x1, and thus
x2 can of course be vectors of variables. y1, respectively
y2 is the objective variable.
I guess the above formulation is valid. Leave it to the
system to optimize. Can be extended to multi objective

Re: Quintus --> Sicstus

j...@hppc323.few.vu.nl (Jan Wielemaker) — Sat, 20 Mrz. 2010 16:43:04 UT

Many implementors want to be cover the ISO-Standard and meaningful
extensions and/or extensions found in other implementations. That
makes sense. It makes porting programs to an implementation easier.
As long as the implementors watch each other (and I have the
impression that is increasingly the case), it also establishes new

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Sat, 20 Mrz. 2010 15:03:52 UT

On Sat, 20 Mar 2010 15:17:05 +0100, Jan Burse <janbu...@fastmail.fm>
wrote:
Quote: "Please note that the above is only an example. In real
program, I would use up to 10000 variables .."
A.L.

Re: Is Prolog the best language to solve this kind of problem?

janbu...@fastmail.fm (Jan Burse) — Sat, 20 Mrz. 2010 14:17:05 UT

Milan Babuskov schrieb:
The if-then-else should not so much be a problem, since:
f = 0 v g = 0
Is the same as:
f * g = 0
Use a Prolog that can handle Quantifier Elimination.
Minimizing a predicate can be reformulated with quantifiers:
find the x that minimizes the y of P(x,y)
Is the same:

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Sat, 20 Mrz. 2010 13:29:08 UT

By the way, look for topics with title like "how to linearize this..."
A.L.

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Sat, 20 Mrz. 2010 13:05:20 UT

On Sat, 20 Mar 2010 00:37:58 +0000, Kish Shen <kishs...@yahoo.com>
wrote:
With all respect, but problem interface in ECLiPSE is rather
primitive. It requires defining problem on the matrix level, what is
good for small problems and problems without internal structure. Try
to program one of more complex examples from AMPL book or Williams

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Sat, 20 Mrz. 2010 12:47:05 UT

On Sat, 20 Mar 2010 02:05:50 -0700 (PDT), Milan Babuskov
Right group to ask quetions like yours is sci.op-research (Operations
Research group). Questions about linearizing nonlinear/logical
expressions for LP pop up at least once a month. You can find a lot of
information just browsing old posts. You can also ask specific

Re: Is Prolog the best language to solve this kind of problem?

milan.babus...@gmail.com (Milan Babuskov) — Sat, 20 Mrz. 2010 09:05:50 UT

Thanks A.L.
I'm just reading the book and it really explains some problems I
encountered so far. I still haven't found the solution to my problem,
but hopefully I will by the time I read it completely.
Regards,
M.

Re: Is Prolog the best language to solve this kind of problem?

kishs...@yahoo.com (Kish Shen) — Sat, 20 Mrz. 2010 00:37:58 UT

A.L. wrote:
ECLiPSe has the eplex library, which interfaces to MP solvers (CPLEX,
Xpress-MP, and the open-source COIN-OR's CLP/CBC solvers). In
particular, the solving of a MP problem can be triggered in a
data-driven manner (i.e. behaving in a similar way to constraints in
Constraint Programming), all using ECLiPSe/Prolog syntax, and other

Re: Is Prolog the best language to solve this kind of problem?

alewa...@aol.com (A. L.) — Fri, 19 Mrz. 2010 23:50:55 UT

On Fri, 19 Mar 2010 14:49:20 -0700 (PDT), Chip Eastham
Linear programming solvers usually come with their own languages that
are integrated with the solver and have been designed specifically for
LP (MIP) problems. These are OPL for CPLEX, MOSEL for Xpress-MP and
many others. There are also "generic" LP languages that work with