multi-objective logging
AnsweredHi,
I'm trying to solve a hierarchical multi-objective ILP with gurobi 10.0.
When the model is solved, I retrieve the objective values with:
nSolutions = model.SolCount
nObjectives = model.NumObj
for s in range(nSolutions):
model.params.SolutionNumber = s
print('Solution', s, ':', end='\n')
for o in range(nObjectives):
model.params.ObjNumber = o
print(' ',model.ObjNVal, end='\n')
This gives me the following solution:
Solution 0 :
1074757567.0
44910059.0
Solution 1 :
1122697741.0
46681612.0
Solution 2 :
1126893891.0
47306774.0
As I also solved the single-objective problem before, I know that the returned objective value of objective 1 of the first solution is right.
When I look at the gurobi log, however, the incumbent seems to be even better and none of the solutions in "solution count" is returned by model.ObjNVal.
Optimize a model with 8236 rows, 17755 columns and 68217 nonzeros
Variable types: 0 continuous, 17755 integer (17749 binary)
Coefficient statistics:
Matrix range [8e-03, 3e+04]
Objective range [1e+03, 7e+07]
Bounds range [1e+00, 1e+00]
RHS range [1e+00, 2e+02]
---------------------------------------------------------------------------
Multi-objectives: starting optimization with 2 objectives ...
---------------------------------------------------------------------------
Multi-objectives: applying initial presolve ...
---------------------------------------------------------------------------
Presolve removed 8229 rows and 12014 columns
Presolve time: 0.02s
Presolved: 7 rows and 5741 columns
---------------------------------------------------------------------------
Multi-objectives: optimize objective 1 (LCC) ...
---------------------------------------------------------------------------
Found heuristic solution: objective 1.083384e+09
Presolve removed 0 rows and 3772 columns
Presolve time: 0.00s
Presolved: 7 rows, 1969 columns, 4073 nonzeros
Found heuristic solution: objective 1.079187e+09
Variable types: 0 continuous, 1969 integer (1897 binary)
Root relaxation: objective 1.024916e+09, 5 iterations, 0.00 seconds (0.00 work units)
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 1.0249e+09 0 2 1.0792e+09 1.0249e+09 5.03% - 0s
0 0 1.0309e+09 0 1 1.0792e+09 1.0309e+09 4.48% - 0s
H 0 0 1.031247e+09 1.0309e+09 0.03% - 0s
0 0 1.0309e+09 0 1 1.0312e+09 1.0309e+09 0.03% - 0s
Explored 1 nodes (9 simplex iterations) in 0.08 seconds (0.05 work units)
Thread count was 8 (of 8 available processors)
Solution count 3: 1.03125e+09 1.07919e+09 1.08338e+09
Optimal solution found (tolerance 0.00e+00)
Best objective 1.031247217000e+09, best bound 1.031247217000e+09, gap 0.0000%
---------------------------------------------------------------------------
Multi-objectives: optimize objective 2 (CO2-eq) ...
---------------------------------------------------------------------------
Loaded user MIP start with objective 1.52736e+07
Presolve removed 0 rows and 3781 columns
Presolve time: 0.03s
Presolved: 8 rows, 1960 columns, 6005 nonzeros
Variable types: 0 continuous, 1960 integer (1890 binary)
Root relaxation: objective 1.527358e+07, 28 iterations, 0.00 seconds (0.00 work units)
Nodes | Current Node | Objective Bounds | Work
Expl Unexpl | Obj Depth IntInf | Incumbent BestBd Gap | It/Node Time
0 0 1.5274e+07 0 2 1.5274e+07 1.5274e+07 0.00% - 0s
Explored 1 nodes (28 simplex iterations) in 0.11 seconds (0.07 work units)
Thread count was 8 (of 8 available processors)
Solution count 1: 1.52736e+07
Optimal solution found (tolerance 1.00e-04)
Best objective 1.527357700000e+07, best bound 1.527357611111e+07, gap 0.0000%
---------------------------------------------------------------------------
Multi-objectives: solved in 0.13 seconds (0.07 work units), solution count 3
Do you have any idea/hint, why model.ObjNVal and logging values differ?
Thanks for your help!
-
Official comment
This post is more than two years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum, or try Gurobot, our chatbot interface offering instant, expert-level support. -
Hi Nathalie,
This is a bug in 10.0.0 an will be fixed with 10.0.1. Please refer to Why does Gurobi 10.0.0 report an incorrect objective value for my multi-objective problem? for more details.
Best regards,
Jaromił0 -
Thanks for the quick response!
0
Post is closed for comments.
Comments
3 comments