Counting lazy constraints

回答済み

Amirhossein Abbaszadeh

• 2024年03月29日 19:48

Hi all, 

I'm trying to apply Benders Decomposition to a problem using callback function in C++.

I understand that whenever Gurobi found a MIP solution (that is where == GRB_CB_MIPSOL), I must get the incumbent solution using getSolution, then solve the subproblem and check if a cut need to be added using addLazy. 

The condition that I check at each iteration is as follows:

if (varVal[varIndex->thetaOmegaIndex[1]] - subProblem.get(GRB_DoubleAttr_ObjVal) > 0)
{
                //So we need to add a cut!
                addLazy(theCutExpr, GRB_GREATER_EQUAL, 0.0);
                cutCounter++;
}

As you can see in the code, I've also defined the global variable cutCounter to count the cuts that are added to model. I know that the Gurobi gives me the numbers of lazy constraints that have been added to the model.  

The problem is that when I check the Gurobi report and compare it with the cutCounter, I get two different values! (Please see the last sentence of the following log file)

Cutting planes:
  Lazy constraints: 97

Explored 1569 nodes (188581 simplex iterations) in 30.06 seconds (38.02 work units)
Thread count was 1 (of 16 available processors)

Solution count 3: 1761 1508 1228

Time limit reached
Best objective 1.761000000000e+03, best bound 6.167127329585e+03, gap 250.2060%

User-callback calls 8903, time in user-callback 10.18 sec

---------------------------------------
cutCounter = 176

Why it's like that? Is it normal? or I'm missing something? 
I can provide more details if required.

Thanks

関連するエンティティ

• Bender's
• Lazy Constraints

0

• 

  Riley Clement

  Gurobi Staff

  • 2024年04月02日 10:27

  Hi Amirhossein,

  This is normal.  Sometimes a lazy constraint that has been added becomes redundant and is removed from the model, lowering the number of lazy constraints reported.

  - Riley

  0

サインインしてコメントを残してください。