Lagrangian Dual, Subgradient Method

Answered

Farzane Ezzati

November 08, 2023 23:11

• 
• 
• 

Hello,

I want to solve a two stage stochastic programming model and I am using scenario decomposition and lagrangian relaxation of non-anticipativity constraints to solve the problem. I am mixing lagrangian dual with branch and bound. in the branch and bound tree, the problems I need to solve, should be solved by their lagrangian dual using subgradient method. My question is does GUROBI can be adjusted to solve a model by subgradient method? I want to avoid coding the whole iterations of subgradient method and want to focus only on the branch and bound tree. Thanks in advance for the help.

Bests, 

Farzane

Related to

• Branch-and-Cut
• Stochastic

0

• 

  Mario Ruthmair

  • Gurobi Staff

  November 09, 2023 15:49

  Hi Farzane,

  Gurobi does not include the subgradient method for solving node relaxations in the branch-and-bound tree. Node bounds are obtained by solving continuous relaxations by simplex or barrier only.

  So, if you want to use Lagrangian Relaxation in a branch-and-bound framework, Gurobi cannot help you here. You need to implement all ingredients by yourself or use other existing frameworks.

  Best regards,
  Mario

  0

Please sign in to leave a comment.