Which algorithm does Gurobi use to solve the model.addGenConstrPow()?

Answered

Luis Moncayo

• June 20, 2023 01:45

I am solving an objective function with the term {\sqrt(x+t-y). I think I have done this, but I have to report a very general overview of the algorithm Gurobi uses to solve it.

Can somebody give me a hint?

Any help will be greatly appreciated.

Related to

• General Constraints

0

• 

  Maliheh Aramon

  Gurobi Staff

  • June 21, 2023 01:30

  Hi Luis, 

  Gurobi internally approximates the general function constraints including the power function as piecewise linear functions. In other words, Gurobi approximates these non-linear functions into linear MIP formulations and it then proceeds with its regular branch-and-cut approach to solve the approximated MIP formulation. 

  The piecewise linear approximation of general constraints involve a cost-vs-accuracy tradeoff. Using more points adds more pieces resulting in smaller errors. However, it increases the computational time. Gurobi includes parameters such as FuncPieces, FuncPieceLength, FuncPieceError to control the tradeoff. You can read more about them in the documentation of general constraints referred above.

  Best regards,

  Maliheh

  0
• 

  Luis Moncayo

  • June 21, 2023 16:55

  

  

  

  Dear Maliheh,

  Thank you very much; I appreciate your comment. That's exactly what I needed.

  Regards, :) 

  Luis 

  0

Please sign in to leave a comment.