Can't get Pi (shadow prices) of a continuos model.

Answered

Ilija Dukovski

March 17, 2022 15:39

• 
• 
• 

Hi,

I have a continuous model (the code is in Java). After optimization, I can get the X attributes, but when I try Pi, I get this error message:

Error code: 10005. Unable to retrieve attribute 'Pi'

The lines of Java code are:

double[] xvals= model.get(GRB.DoubleAttr.Pi, model.getConstrs());

or

GRBConstr[] constrains=model.getConstrs();

double sp = constrains[1].get(GRB.DoubleAttr.Pi); 

Again, this is a continuous model. This is how I add the variables:

for(int i=0;i<l.length;i++){
            objective[i]=0;
            types[i]=GRB.CONTINUOUS;
        }

rxnFluxes = model.addVars(l, u, objective, types, null);
model.update();

 

Thanks for your help,

Ilija

0

• 

  Gwyneth Butera

  • Gurobi Staff

  March 17, 2022 16:02

  • Pi values are only available for convex, continuous problems.  
  • If you are solving a QCP, there are a couple of extra steps to get your duals.  To get the duals of linear constraints, set the QCPDual parameter is set to 1.  To get the duals for quadratic constraints, you will want to use QCPi attribute rather than the Pi attribute.

  If this doesn't help, can you tell us a bit more about your model and how you are trying to access Pi values?

  Gwyneth

  0
• 

  Ilija Dukovski

  March 17, 2022 17:30

  • 
  • 
  • 

  Thanks for your help. That's probably the problem. Most likely my solution is not unique, and I don't get Pis in that case. I will try secondary optimization to make the problem convex.

  Thanks,

   

  0

Please sign in to leave a comment.