Issues with outlev and logfile in Gurobi version 10.0.1
AnsweredHi,
I tried to solve a MISOCP problem formulated in AMPL with the recent version of gurobi (10.0.1), but when I set "outlev=1" the optimization progress does not appear. Then, I added the setting "logfile=test.log" after which the progress appears but indicating that the user provided a MIP start (without producing incumbent). Obviously, I didn't provide such a MIP start. Additionally, when I run my code in another machine with Gurobi 9.5.2 everything happens without problems. The output is shown as follows:
Presolve eliminates 181785 constraints and 177380 variables.
Adjusted problem:
466266 variables:
1392 binary variables
1122 integer variables
385200 nonlinear variables
78552 linear variables
397410 constraints; 2195271 nonzeros
122400 nonlinear constraints
275010 linear constraints
147252 equality constraints
250158 inequality constraints
1 linear objective; 1398 nonzeros.
Gurobi 10.0.1: Set parameter OutputFlag to value 1
Set parameter InfUnbdInfo to value 1
Gurobi Optimizer version 10.0.1 build v10.0.1rc0 (win64)
CPU model: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz, instruction set [SSE2|AVX|AVX2]
Thread count: 6 physical cores, 12 logical processors, using up to 12 threads
Optimize a model with 275010 rows, 466266 columns and 1718631 nonzeros
Model fingerprint: 0x08b3cb2a
Model has 122400 quadratic constraints
Variable types: 463752 continuous, 2514 integer (0 binary)
Coefficient statistics:
Matrix range [6e-12, 1e+02]
QMatrix range [1e+02, 1e+02]
Objective range [1e-02, 1e+03]
Bounds range [9e-01, 2e+01]
RHS range [5e-05, 4e+01]
User MIP start did not produce a new incumbent solution
User MIP start violates constraint R148565 by 0.000308847
Presolve removed 72492 rows and 85083 columns
Presolve time: 8.86s
Presolved: 202518 rows, 381183 columns, 1388091 nonzeros
Presolved model has 115920 quadratic constraint(s)
Variable types: 379142 continuous, 2041 integer (919 binary)
Root relaxation presolve removed 27 rows and 27 columns
Root relaxation presolve time: 10.11s
Root relaxation presolved: 898011 rows, 612996 columns, 2547237 nonzeros
Root relaxation presolved model has 115920 second-order cone constraints
Root barrier log...
Ordering time: 9.44s
Ordering time: 18.68s
Barrier statistics:
Dense cols : 277
AA' NZ : 2.200e+07
Factor NZ : 1.678e+08 (roughly 2.0 GB of memory)
Factor Ops : 2.562e+11 (roughly 2 seconds per iteration)
Threads : 6
Objective Residual
Iter Primal Dual Primal Dual Compl Time
0 1.01704956e+06 -1.46452764e+07 9.38e+01 7.08e+00 2.77e+02 35s
1 6.91205163e+05 -3.73003410e+07 6.18e+01 1.05e+00 9.65e+01 37s
2 4.58684428e+05 -1.96503192e+07 4.12e+01 3.75e-01 4.22e+01 39s
3 3.38516691e+05 -7.25132048e+06 2.80e+01 5.45e-02 1.61e+01 42s
4 1.62817759e+05 -3.27783284e+06 1.22e+01 1.65e-02 6.41e+00 44s
5 1.14700377e+05 -1.01083971e+06 8.18e+00 2.04e-03 2.81e+00 47s
6 5.54621818e+04 -6.22177983e+05 2.88e+00 7.60e-04 1.12e+00 50s
7 3.13507558e+04 -2.23986268e+05 8.46e-01 9.22e-05 3.01e-01 54s
8 2.63551292e+04 -7.41350904e+04 4.73e-01 1.73e-05 1.08e-01 57s
9 2.00460945e+04 -1.18389587e+04 6.79e-02 3.87e-06 2.56e-02 59s
10 1.85098977e+04 7.05833655e+03 2.17e-02 1.32e-06 8.75e-03 62s
11 1.80574440e+04 1.19394967e+04 1.51e-02 6.62e-07 4.64e-03 64s
12 1.77402398e+04 1.40187435e+04 1.06e-02 3.73e-07 2.81e-03 66s
13 1.73897671e+04 1.51017388e+04 6.23e-03 2.27e-07 1.72e-03 68s
14 1.71786187e+04 1.59116576e+04 3.87e-03 1.12e-07 9.46e-04 71s
15 1.70206885e+04 1.62710498e+04 2.30e-03 6.28e-08 5.58e-04 73s
16 1.68573933e+04 1.65759965e+04 6.84e-04 2.14e-08 2.09e-04 76s
17 1.68221853e+04 1.66415645e+04 3.82e-04 1.39e-08 1.34e-04 79s
18 1.67886707e+04 1.67121473e+04 1.93e-04 4.16e-09 5.67e-05 82s
19 1.67718160e+04 1.67326223e+04 9.45e-05 1.99e-09 2.91e-05 84s
20 1.67660261e+04 1.67402977e+04 6.63e-05 1.17e-09 1.91e-05 87s
21 1.67605205e+04 1.67440412e+04 3.96e-05 7.77e-10 1.22e-05 89s
22 1.67566832e+04 1.67474846e+04 1.97e-05 4.33e-10 6.82e-06 92s
23 1.67550325e+04 1.67512550e+04 1.14e-05 1.05e-10 2.80e-06 94s
24 1.67538860e+04 1.67520778e+04 5.45e-06 4.33e-11 1.34e-06 97s
25 1.67533350e+04 1.67524425e+04 2.56e-06 3.29e-11 6.60e-07 100s
26 1.67531281e+04 1.67526256e+04 1.65e-06 6.80e-11 3.72e-07 102s
27 1.67530409e+04 1.67526828e+04 1.24e-06 1.17e-10 2.65e-07 104s
28 1.67529752e+04 1.67527039e+04 9.24e-07 1.60e-10 2.01e-07 107s
29 1.67529349e+04 1.67527220e+04 7.26e-07 2.02e-10 1.57e-07 109s
30 1.67529006e+04 1.67527483e+04 5.47e-07 2.96e-10 1.13e-07 112s
31 1.67528683e+04 1.67527744e+04 3.63e-07 4.89e-10 6.94e-08 115s
32 1.67528327e+04 1.67527939e+04 1.35e-07 7.57e-10 2.86e-08 118s
33 1.67528195e+04 1.67528037e+04 6.06e-08 2.04e-09 1.17e-08 121s
Barrier solved model in 33 iterations and 121.35 seconds (109.67 work units)
Optimal objective 1.67528195e+04
Root relaxation: objective 1.675282e+04, 0 iterations, 112.06 seconds (103.90 work units)
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 16752.8195 0 1967 - 16752.8195 - - 142s
Another try with MIP start
0 0 16752.8195 0 1919 - 16752.8195 - - 43230s
Explored 1 nodes (0 simplex iterations) in 43230.45 seconds (17273.32 work units)
Thread count was 12 (of 12 available processors)
Solution count 0
Time limit reached
Thanks
-
The problem with outlev=1 not showing any log output was caused by a glitch in the build of Gurobi 10.0.1 for AMPL. It is fixed in the most recent build (dated 20230330).
The problem with a MIP start being tried, when no starting values were provided by the user, appears to be a bug in the new AMPL-solver interface starting with Gurobi 10.0. The AMPL interface to Gurobi 9.5.2 used our older interface, which did not have this problem. For the case where you are not specifying any MIP start, a workaround is to add mipstart=0 to your gurobi_options string.
0
Please sign in to leave a comment.
Comments
1 comment