Jacob Jin

To be more specific, I use the reset method in the callback function m_sub.reset(1)# m_sub.Params.Presolve = -1start = time.time()m_sub.optimize()end = time.time()print(m_sub.Runtime)print(m_sub.o...

Did you try explicitly setting the Presolve parameter inside your callback before solving the sub_dual model? You could also try calling the reset method before optimizing and setting the paramete...

Did you try changing objective coefficients via the variable Obj attribute instead of reconstructing the objective from scratch? I have tried before. The results are the same. You mentioned "the ...

Hi, Jaromil! I have tried your suggestion.  However, there are still some issues I want to talk about with you. 1. Threads. m_sub.Params.Threads = 4m_master.Params.Threads = 4 I have 8 threads in t...

Okay. Thank you for your quick response!!!! I am trying it(:  

Even though I reset the presolve parameter in the callback function(Method 2), it would take 10+ S to solve the LP problem (except the first time to solve the dual problem).   It is strange because...

def callBackFunction(model, where):    if where == GRB.Callback.MIPSOL: ### retrieve the current integer value     r_1 = np.array([[model.cbGetSolution(model.getVarByName('r_' + str(i) + ...

Hi, Dr Najman. I found that the sub_dual model didn't presolve when I only redefine its objective function in the callback function. That was why the sub_dual problem would take so much time. Outsi...

The result shows that for the solve time of a single sub dual model, it will takes more time if I use the second method.(only change the objective function in the callback function) For the total s...

I think you understood me in an opposite way.    In total, I have 2 ways to define the sub dual model need to be solved in the callback function. I can define the entire dual model in the callback ...