> Does anybody have tried to solve a Sudoko instance with a BigM method
> (to say "all of the digits of a line must be different").
Using integer variables (like 0 <= x[i,j] <= 9) to model sudoku
solutions together with "all different" constraints leads to a poor
formulation which is hard to solve with b&b. Much more efficient
formulation can be obtained by using binary variables x[i,j,k] that
model a mapping from the grid to {0,...,9}. See an example model
included in the glpk distribution (examples/sudoku.mod).
> There is solution without this method.
> How can someone know where and when GLPK is "able" or not (actually
> the Simplex method used) to solve such problems. Any help on this
> would be appreciable.
>
_______________________________________________
Help-glpk mailing list
[email protected]
https://lists.gnu.org/mailman/listinfo/help-glpk
