Parameters tuning
Answered
Hi,
I have a large integer model which has near 13000 binary variables and 100 integer variables(after presolve).
I have test different parameters' value. But didn't get any feasible solution. could anybody help to tune parameters to get a feasible solution? here is the log:
Optimize a model with 202879 rows, 30005 columns and 870071 nonzeros
Variable types: 0 continuous, 30005 integer (29898 binary)
Coefficient statistics:
Matrix range [1e+00, 9e+06]
Objective range [1e+00, 4e+01]
Bounds range [1e+00, 1e+00]
RHS range [1e+00, 3e+06]
Presolve removed 45042 rows and 17000 columns
Presolve time: 4.52s
Presolved: 157837 rows, 13005 columns, 745539 nonzeros
Variable types: 0 continuous, 13005 integer (12898 binary)
Presolve removed 43 rows and 70 columns
Presolved: 12962 rows, 170772 columns, 755729 nonzeros
Presolve removed 12962 rows and 170772 columns
Root simplex log...
Iteration Objective Primal Inf. Dual Inf. Time
0 4.0980400e+03 0.000000e+00 2.415470e+05 11s
28457 4.1097061e+03 0.000000e+00 1.574963e+06 15s
51855 4.1155907e+03 0.000000e+00 1.828101e+06 20s
73614 4.1195006e+03 0.000000e+00 1.408128e+07 25s
91648 4.1228688e+03 0.000000e+00 4.653613e+06 30s
108546 4.1274614e+03 0.000000e+00 1.789445e+08 35s
123740 4.1302029e+03 0.000000e+00 1.297128e+06 40s
138184 4.1122067e+03 8.553554e+00 0.000000e+00 45s
143621 4.1116232e+03 4.444469e+00 0.000000e+00 50s
145832 4.1116232e+03 1.043372e+00 0.000000e+00 55s
146305 4.1116233e+03 0.000000e+00 0.000000e+00 57s
146305 4.1116233e+03 0.000000e+00 0.000000e+00 57s
146305 4.1116233e+03 0.000000e+00 0.000000e+00 57s
Root relaxation: objective 4.111623e+03, 146305 iterations, 46.45 seconds
Total elapsed time = 62.10s
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 4111.62333 0 997 - 4111.62333 - - 64s
----------------------------------------
and gurobi keeps stopping here until my halt.
Best,
Amir,
-
I forgot to mention that there is no time in which Memory usage exceeds 30%
0 -
I let the machine keep solving and the logs are here:
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time0 0 4111.62333 0 997 - 4111.62333 - - 65s
0 0 4118.37586 0 2128 - 4118.37586 - - 395s
0 0 4118.73753 0 1789 - 4118.73753 - - 401s
0 0 4118.73753 0 1793 - 4118.73753 - - 402s
0 0 4121.50600 0 2313 - 4121.50600 - - 625s
0 0 4121.70667 0 2020 - 4121.70667 - - 628s
0 0 4121.70667 0 1978 - 4121.70667 - - 629s
0 0 4122.04000 0 1964 - 4122.04000 - - 642s
0 0 4122.04000 0 2138 - 4122.04000 - - 644s
0 0 4122.37333 0 1721 - 4122.37333 - - 669s
0 0 4122.37333 0 1653 - 4122.37333 - - 671s
0 0 4122.37333 0 1682 - 4122.37333 - - 677s
0 0 4122.37333 0 1662 - 4122.37333 - - 679s
0 0 4122.41500 0 1521 - 4122.41500 - - 701s
0 0 4122.41500 0 1623 - 4122.41500 - - 704s
0 0 4122.45667 0 1640 - 4122.45667 - - 710s
0 0 4122.45667 0 1617 - 4122.45667 - - 711s
0 0 4122.45667 0 1723 - 4122.45667 - - 717s
0 0 4122.45667 0 1628 - 4122.45667 - - 719s
0 0 4122.45667 0 1565 - 4122.45667 - - 755s
0 0 4122.45667 0 1613 - 4122.45667 - - 757s
0 0 4122.45667 0 1582 - 4122.45667 - - 779s
0 0 4122.45667 0 1484 - 4122.45667 - - 802s
0 2 4122.45667 0 1484 - 4122.45667 - - 860s
1 4 4122.45667 1 1482 - 4122.45667 - 2946 870s
3 8 4122.45667 2 1490 - 4122.45667 - 2742 883s
7 16 4122.45667 3 1489 - 4122.45667 - 3017 900s
15 32 4122.45667 4 1483 - 4122.45667 - 2674 935s
31 52 4122.45667 5 1582 - 4122.45667 - 2731 994s
------------------------------
---------- and gurobi doesn't converge.
0 -
after 2813 second:
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time0 0 4111.62333 0 997 - 4111.62333 - - 65s
0 0 4118.37586 0 2128 - 4118.37586 - - 395s
0 0 4118.73753 0 1789 - 4118.73753 - - 401s
0 0 4118.73753 0 1793 - 4118.73753 - - 402s
0 0 4121.50600 0 2313 - 4121.50600 - - 625s
0 0 4121.70667 0 2020 - 4121.70667 - - 628s
0 0 4121.70667 0 1978 - 4121.70667 - - 629s
0 0 4122.04000 0 1964 - 4122.04000 - - 642s
0 0 4122.04000 0 2138 - 4122.04000 - - 644s
0 0 4122.37333 0 1721 - 4122.37333 - - 669s
0 0 4122.37333 0 1653 - 4122.37333 - - 671s
0 0 4122.37333 0 1682 - 4122.37333 - - 677s
0 0 4122.37333 0 1662 - 4122.37333 - - 679s
0 0 4122.41500 0 1521 - 4122.41500 - - 701s
0 0 4122.41500 0 1623 - 4122.41500 - - 704s
0 0 4122.45667 0 1640 - 4122.45667 - - 710s
0 0 4122.45667 0 1617 - 4122.45667 - - 711s
0 0 4122.45667 0 1723 - 4122.45667 - - 717s
0 0 4122.45667 0 1628 - 4122.45667 - - 719s
0 0 4122.45667 0 1565 - 4122.45667 - - 755s
0 0 4122.45667 0 1613 - 4122.45667 - - 757s
0 0 4122.45667 0 1582 - 4122.45667 - - 779s
0 0 4122.45667 0 1484 - 4122.45667 - - 802s
0 2 4122.45667 0 1484 - 4122.45667 - - 860s
1 4 4122.45667 1 1482 - 4122.45667 - 2946 870s
3 8 4122.45667 2 1490 - 4122.45667 - 2742 883s
7 16 4122.45667 3 1489 - 4122.45667 - 3017 900s
15 32 4122.45667 4 1483 - 4122.45667 - 2674 935s
31 52 4122.45667 5 1582 - 4122.45667 - 2731 994s
------------------------------
------ but no convergence!
0 -
It could be anything. Try reading the suggestions in the MIP models section of the parameter guidelines.
0
Please sign in to leave a comment.
Comments
4 comments