Solving a Semidefinite Programming using callback function and lazy constraints
I am writing a code that would be able to solve semidefinite programming such that after ignoring the semidefiniteness constraint, It will check all of the eigenvalues of my answer and if there is a negative eigenvalue it will add a lazy constraint in a callback function and repeat this procedure until it gets an answer which has just positive eigenvalues. I also set the lazy constraint and pre-crush parameter to 1, however, the code ignores the callback function, just solves the primary LP and terminate. I would appreciate if someone tells me what is missing here.
-
Hi,
Is this a continuous SDP or a Mixed-Integer SDP that you are looking at?
Please note, lazy constraints are used to cutoff infeasible integer solutions of a MIP. You cannot use them to do some outer approximation for a continuous cone. Because you would normally add them in a MIPSOL callback that is only called if a MIP solution is found.
Best regards,
Sonja
0
Please sign in to leave a comment.
Comments
1 comment