I'm working on a maximization problem. Here is some information about my problem.
From the information, I'm working on a QCP. In each optimization process, I will give a group of QC to obtain the optimal solution of decision variable Q. More precisely, the QC input for the next optimization is the optimal solution Q obtained from the previous optimization. Then, let me explain the problems I faced.
Preferred, I define QC as
qc = [[[-178, -178, 100], [-178, 178, 100], [178, -178, 100], [178, 178, 100]]]
When I set MIPFocus to 1, I can quickly get the first best solution Q. When I set MIPFocus to 3, I can't get the result for a long time. I set Method to 2, whether MIPFocus is 1 or 3, solver can get the result quickly.
- Why does barrier algorithm improve performance so much?
Although I can get the optimization of Q in the first optimization process, when Q is used as the input of the second optimization, gurobi can't appear the value of gap for a long time. Then I found a warning message.
- Is it max violation exceed tolerance problem that leads to the failure of the second optimization?
I check the problem by checking the slack of constraint.
Is there a problem with constraint C0 and how can I improve it?
When I take the first Q as an integer and manually take it as the input of the second optimization. Ps. For the first optimization, the optimal solution of Q is
qc = [[[-211, -211, 100], [-211, 211, 100], [211, -211, 100], [211, 211, 100]]]
The same result as the first one appears. When I set MIPFocus to 1, I can quickly get the first best solution Q. When I set MIPFocus to 3, I can't get the result for a long time. I set Method to 2, whether MIPFocus is 1 or 3, solver can get the result quickly. The third optimization can not be carried out smoothly.
Please sign in to leave a comment.