I have a linear MIP problem for which Gurobi finds the solution in 10 iterations.
To actually prove that the solution is optimal, it takes much more time.
The log is below.
Is there a way to tell Gurobi to stop?
I tried the automated tuning tool.
It tells me to set Heuristics=0.
If I follow this advice, the total running time to find a solution decreases.
But this total time is much more than the time of the 10 iterations with heuristics on.
I'm new to MIP, so , from the log, I don't really know, which parameter will be a good stopping criterion (GAP, BestBound, ...) .
Optimize a model with 434 rows, 380 columns and 1332 nonzeros
Found heuristic solution: objective -0.667665
Presolve removed 74 rows and 72 columns
Presolve time: 0.00s
Presolved: 360 rows, 308 columns, 1428 nonzeros
Variable types: 188 continuous, 120 integer (120 binary)
Root relaxation: objective 1.454681e+00, 383 iterations, 0.00 seconds
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 1.45468 0 80 -0.66764 1.45468 318% - 0s
H 0 0 -0.2958055 1.45468 592% - 0s
0 0 1.33723 0 87 -0.29581 1.33723 552% - 0s
H 0 0 -0.0360081 1.33723 3814% - 0s
0 0 1.32350 0 88 -0.03601 1.32350 3776% - 0s
0 0 1.31284 0 62 -0.03601 1.31284 3746% - 0s
0 5 1.31284 0 62 -0.03601 1.31284 3746% - 0s
H 407 237 -0.0223424 1.12204 5122% 13.3 0s
H 606 348 -0.0139589 1.09397 7937% 12.8 0s
H 1209 691 0.0000905 1.00647 - 12.2 0s
H 1543 852 0.0000935 1.00647 - 15.4 1s
12464 8280 0.31259 37 45 0.00009 0.83003 - 16.1 5s
32517 21750 cutoff 44 0.00009 0.75633 - 15.8 10s
41026 27530 0.15234 45 67 0.00009 0.40720 - 15.7 15s
67008 28123 0.00079 87 9 0.00009 0.00252 2599% 12.1 20s
123660 32561 0.00088 82 13 0.00009 0.00197 2008% 8.0 25s
183205 53085 0.00111 80 14 0.00009 0.00175 1766% 6.5 30s
242669 70749 0.00115 82 13 0.00009 0.00160 1611% 5.6 35s
300464 86096 0.00016 83 14 0.00009 0.00150 1499% 5.2 40s
360002 99530 0.00116 77 12 0.00009 0.00141 1407% 4.8 45s
419747 111348 0.00092 82 11 0.00009 0.00134 1330% 4.5 50s
479404 121404 0.00094 78 18 0.00009 0.00128 1265% 4.4 55s
538670 130127 0.00061 86 9 0.00009 0.00122 1206% 4.2 60s
599541 137721 0.00071 87 10 0.00009 0.00117 1152% 4.1 65s
659419 143977 0.00049 81 13 0.00009 0.00113 1104% 4.0 70s
719366 148872 0.00090 82 7 0.00009 0.00108 1058% 3.9 75s
778800 152645 cutoff 81 0.00009 0.00104 1015% 3.8 80s
838419 155900 0.00064 82 12 0.00009 0.00101 975% 3.7 85s
898257 157892 0.00038 82 11 0.00009 0.00097 937% 3.7 90s
959133 158950 0.00064 82 9 0.00009 0.00093 898% 3.6 95s
1019118 158672 cutoff 86 0.00009 0.00090 863% 3.6 100s
1077389 157263 0.00034 79 16 0.00009 0.00087 828% 3.5 105s
1136559 154819 0.00015 83 6 0.00009 0.00084 795% 3.5 110s
1197408 151286 0.00033 79 11 0.00009 0.00080 760% 3.5 115s
1256981 146998 0.00058 85 11 0.00009 0.00077 726% 3.4 120s
1315053 141986 0.00015 87 9 0.00009 0.00074 693% 3.4 125s
1369901 136123 cutoff 84 0.00009 0.00071 662% 3.4 130s
1423732 129573 0.00042 84 11 0.00009 0.00068 631% 3.3 135s
1483143 120871 0.00036 86 11 0.00009 0.00065 593% 3.3 140s
1541197 111293 0.00020 84 11 0.00009 0.00061 553% 3.3 145s
1598804 100832 0.00030 81 15 0.00009 0.00057 511% 3.3 150s
1655909 89315 0.00039 84 11 0.00009 0.00053 466% 3.2 155s
1704245 77614 0.00018 82 15 0.00009 0.00049 420% 3.2 160s
1750024 63910 0.00014 83 12 0.00009 0.00044 367% 3.2 165s
1795438 46988 cutoff 78 0.00009 0.00037 299% 3.2 170s
1847433 21718 0.00012 82 10 0.00009 0.00026 178% 3.2 175s
Cutting planes:
Gomory: 54
MIR: 14
Flow cover: 28
Explored 1875647 nodes (5924527 simplex iterations) in 178.11 seconds
Thread count was 4 (of 4 available processors)
Optimal solution found (tolerance 1.00e-04)
Best objective 9.353429694370e-05, best bound 9.353429694481e-05, gap 0.0%
If you want to prove that the solution is optimal, do not terminate early. The grbtune parameter tuning tool is very useful, but it looks like you may be able to do better by focusing on moving the bound. For suggestions on how to move the bound, see the MIP section of the Parameter Tuning Guidelines. For example, I would try increasing the Cuts parameter, setting MIPFocus to 2 or 3, and/or setting Presolve to 2.
Disclaimer: I'm in charge of Gurobi technical support.