Model infeasible but not sure why, formulation is sound

Answered

D

May 04, 2023 00:16

• 
• 
• 

Could someone help me pinpoint the issue of the model with this error result:

"

Gurobi Optimizer version 9.0.3 build v9.0.3rc0 (win64)
Optimize a model with 2922876 rows, 3868200 columns and 58611564 nonzeros
Model fingerprint: 0x2f72d501
Variable types: 600 continuous, 3867600 integer (3867600 binary)
Coefficient statistics:
  Matrix range     [9e-03, 6e+03]
  Objective range  [1e+00, 2e+04]
  Bounds range     [1e+00, 1e+00]
  RHS range        [1e+00, 6e+03]
Presolve removed 0 rows and 61500 columns
Presolve time: 7.48s

Explored 0 nodes (0 simplex iterations) in 8.03 seconds
Thread count was 1 (of 8 available processors)

Solution count 0

Model is infeasible
Best objective -, best bound -, gap -"

Related to

• Wrong Result

0

• 

  Simranjit Kaur

  • Gurobi Staff

  May 04, 2023 00:38 Edited

  You can run the Model.computeIIS() method on your model to examine why your model is infeasible. It will provide a minimal subset of constraints and variable bounds that, if isolated from the rest of the model, is still infeasible. You can obtain information about these constraints and variable bounds by writing an .ilp format file.

  model.compteIIS()
  model.write("model.ilp")

  Please refer to the article How do I determine why my model is infeasible for more details.

  0
• 

  D

  May 04, 2023 00:37

  • 
  • 
  • 

  This is the result I got: 

  which based on my understanding to IIS means none of the constraints is causing infeasibility

  IIS computed: 0 constraints, 2 bounds
  IIS runtime: 0.04 seconds

  0
• 

  Simranjit Kaur

  • Gurobi Staff

  May 04, 2023 00:42 Edited

  This means that there are two variables whose bounds are causing the infeasibility in your model. You can get information about these variables by writing the results in an .ilp format file.

  model.write("model.ilp")

  0
• 

  Simranjit Kaur

  • Gurobi Staff

  May 04, 2023 01:25

  You can open this file using any text editor and check the variables that have conflicting bounds.

   

  0
• 

  D

  May 04, 2023 01:42

  • 
  • 
  • 

  So, I got the following:

  Minimize
    0 CostPattern[T0] + 1.1283033680279922e+10 Constant
  Subject To
  Bounds
   CostPattern[T0] >= 1e+100
   Constant = 1
  End

   

  I think when I was defining the variable CostPattern, I added lower bound as lb=GRB.INFINITY,, which apparently corresponds to the positive infinity only. 

   

  How can I update this variable only without constructing the whole model again?

  I tried using model.update() 

  then adding the variable again and objective function and corresponding constraints, didn't work. 

  Does it overwrite the old variables or constraints or add them on top of each other?  because I did notice that my variables numbers increased. 

  0
• 

  Simranjit Kaur

  • Gurobi Staff

  May 04, 2023 04:30

  There is no need to add a new variable, you can simply update the lower bound of the existing variable. For instance,

  x.lb = 1

  will set the lower bound of variable x to 1. The next model.optimize() call will pick up the change.

   

  0

Please sign in to leave a comment.