> > MIP optimzers are not fast solvers for constraint programming > problems. > > That's a shame. > What would you suggest for solving a 10x10x10 binary array subject to > ~200 restrictions? > It's not suitable for DLX. > > Also, I suspect that any insights I have into my problems will not > improve on the T-D heuristic. >
If your instance is pure 0-1 and all constraint coefficients are integer, you may try to solve it with the minisat solver supported by glpsol. You need to specify the '--minisat' option, in which case glpsol transforms your instance to a satisfiability problem and solves the latter with minisat. For more details please see the glpk reference manual. See also an example model (the "paint-by-numbers" puzzle) in the subdirectory glpk/examples/pbn. _______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
