> The additional constraint was set to > minimize the number of fake boxes Shame, then I made my life harder than it needed be. If the stack height is simply restricted to be at most 12 boxes, then I can remove the stack height constraints from my program and add a symmetry breaking constraint that orders the box class indexes in the bottom row and get the following results: Prob Size Sol Time 0: 10 4 0.0s 1: 10 3 0.0s 2: 10 3 0.0s 3: 10 4 0.0s 4: 12 4 0.1s 5: 12 4 0.1s 6: 14 3 0.2s 7: 12 3 0.1s 8: 17 5 0.3s 9: 16 6 0.2s 10: 16 5 0.2s 11: 15 4 0.2s 12: 11 4 0.0s 13: 19 4 0.4s 14: 18 4 0.3s 15: 21 5 0.6s 16: 17 6 0.3s 17: 20 5 0.5s 18: 12 4 0.1s 19: 10 3 0.0s 20: 23 5 0.9s 21: 26 5 1.4s 22: 15 5 0.2s 23: 13 4 0.1s 24: 30 6 2.2s 25: 32 6 2.8s 26: 16 4 0.3s 27: 14 5 0.1s 28: 30 7 2.0s 29: 35 6 4.3s 30: 27 6 1.4s 31: 29 8 1.7s 32: 26 6 1.1s 33: 31 6 2.6s 34: 39 8 5.4s 35: 30 6 2.3s 36: 19 6 0.4s 37: 25 5 1.1s 38: 11 5 0.0s 39: 41 8 6.7s I will later test other variable selection strategies to see if some instances can be improved. Cheers, ThorstenReceived on Mon Mar 22 2010 - 08:33:03 CET
This archive was generated by hypermail 2.3.0 : Wed Sep 25 2024 - 15:13:20 CEST