Hi, I have a non convex MIQCP problem, where the non convexity is illustrated in bilinear constrains like :
x+yz=u where x,y,z,u are continuous variables, a general form of this topic is modeling or approaching the constraint : "xy=z" neither gurobi or cplex can handle it. I m wondering, should i use a non linear solver (like Baron)or is there a way to linearize the problem or to approach differently with gurobi or cplex ?
Thnaks for your answers.
Bilinear terms in equality constraints (non-convex MIQCP) xy=z?
This topic has been closed for comments
Already have an account? Login
Please use your business or academic e-mail address to register
Login to the community
No account yet? Create an account
Enter your username or e-mail address. We'll send you an e-mail with instructions to reset your password.
can you explain how it's done please?
As for Baron, do you know how efficient it is? At matter fact, i have a problem where the number of continuous variables can reach 3000, with eventually 1000 complementarity constrains (but i think i can take them off).
Great appreciation for your help,
The approximation has been explained in section 7.6 of the same document. x1x2 has been represented as y1^2 - y^2 and y1^2 is a separable function. See page number 82 for the lambda-formulation
The number of variables seems quite large for Baron. Can you share your project such that we can check whether there is a way to reformulate of solve it? If so, please add your project to a zip file. You can send it to our support by email if your project contains sensitive information.