Can't get the result of optimization problem in an acceptable time(more than 10 hours).

進行中

songtao wang

2023年02月14日 15:00

• 
• 
• 

The above message is provided by matlab2022a with gurobi10.0. Now I have some questions. Firstly, I think that I don't set any quadratic constraint but the log shows that there is a quadratic constraint. I want to konw why dose this happen. Secondly, My problem is big in the number of constaints, but at the last time, It takes me 10 hours and still doesn't get the result. The nodes that are not searched are more and more, and I can't get the Gap. I sincerely hope that you can help me and tell me the reason. If you need more details about the program or the code, I will sent it as soon as possible.

0

• 正式なコメント

  

  Simranjit Kaur

  • Gurobi Staff

  2026年11月13日 01:49

  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.
• 

  Riley Clement

  • Gurobi Staff

  2023年02月15日 00:35

  Hi Songtao,

  Thanks for the log output, but it will be hard to help without seeing the code, or math formulation, you used to produce your model.

  If you send it through, then someone may be able to provide you advice.

  - Riley

  0
• 

  songtao wang

  2023年02月15日 06:57

  • 
  • 
  • 

  My code is consist of three fuctions, and most of them have more than 100 rows. So I want to send the constraints and objective firstly. Hoping to get your instruction.

  u=sdpvar(T*dim_input,1,'full');
  
  
  
  constr=[Pu_bar*u<=pu_bar];
  
  
  
  for q=1:num_model
  
  pi{q}=sdpvar(size(K(:,:,q),1),size(M(:,:,q),1),'full');
  
  delta{q}=sdpvar(1,1,'full');
  
  z{q}=sdpvar(2*dim_state+2*T*dim_mu+2*(T+1)*dim_eta,1,'full');
  
  omega1{q}=sdpvar(2*(T+1)*dim_output,1,'full');
  
  omega2{q}=sdpvar(size(M(:,:,q),1),1,'full');
  
  g1{q}=sdpvar(2*(T+1)*dim_output,1,'full');
  
  g2{q}=sdpvar(size(M(:,:,q),1),1,'full');
  
  
  
  
  
  
  constr=[constr;pi{q}*m(:,:,q)<=k(:,:,q)-Px_title(:,:,q)*Gammaxu_title(:,:,q)*u];
  
  constr=[constr;pi{q}*M(:,:,q)==K(:,:,q)];
  
  constr=[constr;pi{q}>=0];
  
  
  
  constr=[constr;delta{q}>=xi];
  
  
  
  %KKT condition
  
  %primal
  
  constr=[constr;H(:,:,q)*z{q}+C_title(:,:,q)*Gammaxu_title(:,:,q)*u+C_title(:,:,q)*Gammax_title(:,:,q)-delta{q}*ones(size(H(:,:,q),1),1)<=0];
  
  constr=[constr;M(:,:,q)*z{q}-m(:,:,q)<=0];
  
  
  
  %dual
  
  constr=[constr;omega1{q}>=0];
  
  constr=[constr;omega2{q}>=0];
  
  
  
  %slackness
  
  for l=1:2*(T+1)*dim_output
  
  constr=[constr;g1{q}(l)==H(l,:,q)*z{q}+C_title(l,:,q)*Gammaxu_title(:,:,q)*u+C_title(l,:,q)*Gammax_title(:,:,q)-delta{q};sos1([g1{q}(l),omega1{q}(l)],[1,10])];
  
  
  
  
  end
  
  for l=1:size(M(:,:,q),1)
  
  
  
  
  constr=[constr;g2{q}(l)==M(l,:,q)*z{q}-m(l,:,q);sos1([g2{q}(l),omega2{q}(l)],[1,10])];
  
  end
  
  
  
  %stationary
  
  constr=[constr;sum(omega1{q})==1];
  
  
  
  for l=1:2*dim_state+2*T*dim_mu+2*(T+1)*dim_eta
  
  constr=[constr;omega1{q}'*H(:,l,q)+omega2{q}'*M(:,l,q)==0];
  
  
  
  end
  
  end
  
  
  
  obj=norm(u,2);
  
  
  
  options=sdpsettings('verbose',1,'debug',1,'solver','gurobi');
  
  sol=optimize(constr,obj,options);

  0

投稿コメントは受け付けていません。