Clingcon is an answer set solver for constraint logic programs, building upon the answer set solver clingo. It extends the high-level modeling language of ASP with constraint solving capacities. Constraints over finite domain integer variables can be used in logic programs. Clingcon adopts state-of-the-art techniques from the area of SMT, like conflict-driven learning and theory propagation. It uses lazy nogood and variable generation on the order encoding and features several preprocessing techniques.

Features

  • integer linear constraints
    • &sum{3*x; 4*y} >= z-7
  • global distinct constraint
    • &distinct{x;y;3*z}
  • can handle huge variables
    • no domains needed (-2^30 .. 2^30 as default domain)
    • &dom{1..10; 30..40} = x
  • multi-shot solving using python or lua
  • multi-objective optimization on constraint variables
    • &minimize{x;y;z}

Example

#include "csp.lp".
#const end=20.
#const stepsize=7.
step(0..end).

% initial state
&sum {at(0)} = 0.

% actions
{move(T)} :- step(T); T > 0.

% effects
&sum {at(T-1); stepsize} = at(T) :- move(T).

% frame axiom
&sum {at(T-1)} = at(T) :- not move(T); step(T); step(T-1).

% goals
:- &sum {at(end)} <= 100.

&show {at(X):step(X)}.
#show move/1.

More examples can be found in the example section.

Download

Building

See INSTALL for more details.

Benchmarks

See here for a comparison of clingcon 3.2.0 with other CP and CASP systems.

Citing

  • Description of clingcon version 2.x: pdf bibtex
  • Article for Constraint Answer Set Solving and clingcon 1.x: pdf bibtex

More information on the versions 1.x -2.x can be found on clingcon’s old home page.