Presolve Yields sub-Optimal Result (Bug tied to GenConstraint?)

回答済み

Alexander Paskov

2022年12月31日 06:54 編集済み

• 
• 
• 

I'm solving the problem

 

I've gone through the forums and implemented it in Python as such

 

However, as shown in the following log, the pre-solve yields a sub-Optimal result. Is this possibly a bug related to the presolve? For reference, the optimal solution yields objective 0.363804 by setting x[1,0]=x[1,1]=1.

Any help would be greatly appreciated.

関連

• General Constraints
• Python

0

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2026年11月13日 01:50

  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.
• 

  Marika Karbstein

  • Gurobi Staff

  2023年01月02日 14:43

  Did you check the model you created? You could write the LP file with model.write("model.lp"). What are the values for p and v/w? Can you post the lp file here? 

  0
• 

  Alexander Paskov

  2023年01月02日 17:12 編集済み

  • 
  • 
  • 

  Thank you for the response! I can't seem to upload the LP file here (only jpeg, gif, or png are allowed), so here is a screenshot of the model.lp file itself:

  The values are as follows:

  P = [0.15539925, 0.06792077

         0.39277593, 0.26142169]

  V = [0.16961872, 0.08829835]

  This is a different instance than the one posted above, but the result is still the same — the x variables are all set to 0 and Gurobi claims the "optimal objective" is 0.2579. However, doing it by hand, setting x[0,0]=x[1,1]=1 is clearly feasible and yields a better objective of 0.2084.

   

  EDIT-APPEND: For the gen constraint, I added "FuncPieces=-2 FuncPieceError=0.0001" requirement to ensure it's not an approximation error of the log, but the results didn't change.

   

  0
• 

  Marika Karbstein

  • Gurobi Staff

  2023年01月03日 06:55

  For the variables 

  a = model.addVars(m, vtype=gp.GRB.CONTINUOUS, name='a')

  no lower bound is defined. The default lower bound is 0. Hence, the solution you mention is not feasible.

  a = model.addVars(m, vtype=gp.GRB.CONTINUOUS, lb=-1, name='a')

  will give you x[0,0]=x[1,1]=1 as optimal solution.

  0

投稿コメントは受け付けていません。