Different solutions returned by gurobi602 and gurobi604!

89 views
Skip to first unread message

H Pouya

unread,
Oct 8, 2015, 11:26:53 AM10/8/15
to Gurobi Optimization
Hi everyone;

I am running the exact same projects (copied and pasted from one computer to the other one) written in C++ on two computers one with gurobi602 and the other one with gurobi604 installed on. The results are different. Any help about the origin of this issue?

Please let me know if you need more info. Sorry, I cannot share the code for 2 reasons:
1- It is big.
2- The confidentiality problems.


Any help is appreciated in advance.

Hamed

Greg Glockner

unread,
Oct 8, 2015, 11:30:05 AM10/8/15
to gur...@googlegroups.com
Please post the last ~20 lines of the logs from the two tests.

H Pouya

unread,
Oct 8, 2015, 11:57:12 AM10/8/15
to Gurobi Optimization
Sorry if I am not that smart. I didn't get you. What do you mean by "last ~20 lines of the logs"?

Greg Glockner

unread,
Oct 8, 2015, 1:28:17 PM10/8/15
to gur...@googlegroups.com
> Sorry if I am not that smart. I didn't get you. What do you mean by "last ~20 lines of the logs"?

Can you run both versions of the solver, and take the log file (gurobi.log) and post the last few lines of the logs (approximately 20) to this discussion thread?

H Pouya

unread,
Oct 8, 2015, 1:30:54 PM10/8/15
to Gurobi Optimization
Oh...yeah, sure... I will do that. I will send them ASAP.

Thank you very much.
Message has been deleted

H Pouya

unread,
Oct 23, 2015, 9:56:11 AM10/23/15
to Gurobi Optimization
Here are the last line for the 2 log files got form 2 PCs with different Gurobi versions:

1-

Gurobi 6.0.5 (win64) logging started 10/22/15 22:49:34

Optimize a model with 698 rows, 3577 columns and 51107 nonzeros
Coefficient statistics:
  Matrix range    [3e-01, 1e+00]
  Objective range [1e+03, 2e+03]
  Bounds range    [1e-01, 4e+00]
  RHS range       [1e+00, 1e+00]

Concurrent LP optimizer: primal simplex, dual simplex, and barrier
Showing barrier log only...

Presolve removed 174 rows and 885 columns
Presolve time: 0.09s
Presolved: 524 rows, 2692 columns, 49928 nonzeros

Ordering time: 0.00s

Barrier statistics:
 AA' NZ     : 3.146e+04
 Factor NZ  : 5.084e+04 (roughly 2 MBytes of memory)
 Factor Ops : 6.776e+06 (less than 1 second per iteration)
 Threads    : 14

                  Objective                Residual
Iter       Primal          Dual         Primal    Dual     Compl     Time
   0   4.38449634e+06 -3.29777743e+06  6.01e+02 2.27e-13  3.87e+03     0s
   1   2.05411980e+06 -1.50396561e+06  2.78e+02 2.39e-12  1.51e+03     0s
   2   8.09554904e+04 -3.36190451e+05  6.16e+00 1.43e-09  8.98e+01     1s
   3   2.50223843e+04 -6.73002760e+04  2.30e-01 2.74e-10  1.70e+01     1s
   4   2.31263289e+04 -4.27383689e+03  1.29e-01 6.02e-11  4.93e+00     1s
   5   1.86479982e+04  2.95039624e+03  3.68e-02 3.79e-11  2.83e+00     1s
   6   1.72636678e+04  9.60463076e+03  1.88e-02 1.78e-11  1.38e+00     1s
   7   1.60923896e+04  1.22745381e+04  5.66e-03 9.27e-12  6.90e-01     1s

Barrier performed 7 iterations in 1.16 seconds
Barrier solve interrupted - model solved by another algorithm


Solved with dual simplex
Solved in 2525 iterations and 1.17 seconds
Optimal objective  1.525169509e+04

Gurobi 6.0.5 (win64) logging started 10/22/15 22:49:36

Optimize a model with 698 rows, 3577 columns and 51107 nonzeros
Coefficient statistics:
  Matrix range    [3e-01, 1e+00]
  Objective range [1e+03, 2e+03]
  Bounds range    [1e-01, 4e+00]
  RHS range       [1e+00, 1e+00]
Presolve removed 698 rows and 3577 columns
Presolve time: 0.98s

Explored 0 nodes (0 simplex iterations) in 1.06 seconds
Thread count was 1 (of 16 available processors)

Optimal solution found (tolerance 1.00e-04)
Best objective 8.483193743438e+04, best bound 8.483193743438e+04, gap 0.0%


2-
Gurobi 6.0.4 (win64) logging started 10/22/15 21:22:59

Optimize a model with 1204 rows, 1138 columns and 7232 nonzeros
Coefficient statistics:
  Matrix range    [5e-01, 1e+03]
  Objective range [5e-13, 7e+02]
  Bounds range    [1e+00, 1e+00]
  RHS range       [1e+00, 5e+02]
Found heuristic solution: objective 592.733
Presolve removed 967 rows and 841 columns
Presolve time: 0.12s
Presolved: 237 rows, 297 columns, 1257 nonzeros
Variable types: 0 continuous, 297 integer (297 binary)

Root relaxation: objective -2.485705e+03, 210 iterations, 0.00 seconds

    Nodes    |    Current Node    |     Objective Bounds      |     Work
 Expl Unexpl |  Obj  Depth IntInf | Incumbent    BestBd   Gap | It/Node Time

     0     0 -2485.7050    0   58  592.73276 -2485.7050   519%     -    0s
H    0     0                     412.2904956 -2485.7050   703%     -    0s
H    0     0                      90.4289952 -2485.7050  2849%     -    0s
     0     0 -904.71190    0   39   90.42900 -904.71190  1100%     -    0s
H    0     0                      84.4121449 -904.71190  1172%     -    0s
     0     0 -337.17382    0   31   84.41214 -337.17382   499%     -    0s
H    0     0                       0.0000000 -337.17382      -     -    0s
     0     0 -103.78120    0   20    0.00000 -103.78120      -     -    0s
     0     0  -78.71920    0   35    0.00000  -78.71920      -     -    0s
     0     0  -78.71920    0   26    0.00000  -78.71920      -     -    0s
     0     0  -78.71920    0   47    0.00000  -78.71920      -     -    0s
     0     0  -22.82118    0   10    0.00000  -22.82118      -     -    0s
     0     1  -19.21246    0   10    0.00000  -19.21246      -     -    0s

Explored 3 nodes (591 simplex iterations) in 0.41 seconds
Thread count was 1 (of 1 available processors)

Optimal solution found (tolerance 1.00e-04)
Best objective 1.136868377216e-13, best bound -1.734948221390e+01, gap -

Gurobi 6.0.4 (win64) logging started 10/22/15 21:23:00

Optimize a model with 698 rows, 3531 columns and 49281 nonzeros
Coefficient statistics:
  Matrix range    [3e-01, 1e+00]
  Objective range [1e+03, 2e+03]
  Bounds range    [1e-01, 4e+00]
  RHS range       [1e+00, 1e+00]
Presolve removed 130 rows and 846 columns
Presolve time: 0.02s
Presolved: 568 rows, 2685 columns, 48141 nonzeros

Iteration    Objective       Primal Inf.    Dual Inf.      Time
       0    0.0000000e+00   1.530000e+02   0.000000e+00      0s
    2592    1.5238784e+04   0.000000e+00   0.000000e+00      0s

Solved in 2592 iterations and 0.50 seconds
Optimal objective  1.523878447e+04

Gurobi 6.0.4 (win64) logging started 10/22/15 21:23:00

Optimize a model with 698 rows, 3531 columns and 49281 nonzeros
Coefficient statistics:
  Matrix range    [3e-01, 1e+00]
  Objective range [1e+03, 2e+03]
  Bounds range    [1e-01, 4e+00]
  RHS range       [1e+00, 1e+00]
Presolve removed 698 rows and 3531 columns
Presolve time: 1.03s

Explored 0 nodes (0 simplex iterations) in 1.08 seconds
Thread count was 1 (of 1 available processors)

Optimal solution found (tolerance 1.00e-04)
Best objective 7.654257229907e+04, best bound 7.654257229907e+04, gap 0.0%




On Thursday, October 8, 2015 at 1:28:17 PM UTC-4, Greg Glockner wrote:

Greg Glockner

unread,
Oct 23, 2015, 10:09:27 AM10/23/15
to gur...@googlegroups.com
These aren't the same models:

> Gurobi 6.0.5 (win64) logging started 10/22/15 22:49:34
>
> Optimize a model with 698 rows, 3577 columns and 51107 nonzeros
> Coefficient statistics:
> Matrix range [3e-01, 1e+00]
> Objective range [1e+03, 2e+03]
> Bounds range [1e-01, 4e+00]
> RHS range [1e+00, 1e+00]

This has more columns and nonzeros than the following model:

H Pouya

unread,
Oct 23, 2015, 10:36:44 AM10/23/15
to Gurobi Optimization
I just copied and pasted from one PC to the other one. How is it possible?!
Reply all
Reply to author
Forward
0 new messages