[1-20] Survey of CLP with Non-Linear Constraints

by Olga Caprotti <Olga.Caprotti@risc.uni-linz.ac.at>

Systems (discussed individually in Part 2 of this FAQ):

	CAL
	CLP(F)
	GDCC
	ILOG Solver
	Newton
	QUAD-CLP(R)
	RISC-CLP(Real)

Architectures (discussed here):

	Cooperative Constraint Solvers	
	Symbolic Representation Scheme



----------------------------------------------------------------
Cooperative Constraint Solvers
------------------------------
Michel Rueher <rueher@essi.fr>

For solving constraints over the reals, we have defined a cooperating
architecture based on an asynchronous communication between
heterogeneous solvers. It enables to put together both symbolic and
numerical solvers for tackling systems of constraints that none of them
could solve alone. Solutions provided by such a cooperating system are
always at least as accurate than the one which could individually be
computed by the different solvers. A prototype for a cooperative system
including a solver of linear equations and inequalities, a solver of
polynomial equalities, and a solver based on interval propagation is
under development.

References:

[1] Michel Rueher. An Architecture for Cooperating Constraint Solvers
on Reals. In "Constraint Programming: Basics and Trends". LNCS 910,
231-250, March 1995.
[2] P. Marti and M. Rueher. A Distributed Cooperating Constraints
Solving System. Special issue of IJAIT (International Journal on
Artificial Intelligence Tools), 4(1-2):93--113, June 1995.
(ps available from: http://wwwi3s.unice.fr/~rueher/IJAIT.ps)

----------------------------------------------------------------
Symbolic Representation Scheme
------------------------------
Leon Sterling <leon@ces.cwru.edu>

This paper describes experience in using constraint logic programming
to reason about mechanical parts. A prototype program was written in
the language CLP(R) which verified whether a part met specific
tolerances in its dimensions. The program is interesting in that the
same code can be used with any mixture of symbolic and numeric values.
A symbolic representation scheme, underlying the program, which allows
both symbolic and numeric values, is described. Examples are given of
defining generic parts and toleranced parts, and checking whether a
part meets its tolerances. Finally, a design rule checker is sketched
to show how the logical representation of logic representation
languages facilitates higher level reasoning.

Reference:

[1] L.S. Sterling. Of Using Constraint Logic Programming for Design of
Mechanical Parts. Intelligent Systems - Concepts and Applications, (ed.
L. Sterling), 107-116, Plenum Press, 1993