Hi,

I'm working on a MINLP problem using BONMIN but there is a question.

The output shows:

the process, however, doesn't going further for one or two hours. I shortened the time in model and it goes very smoothly.

I'm wondering what's the reason and what should I do.Should I keep running it or change another solver or change my model.

Thank you

Shouyi Hu

## Stuck in the final part of MINLP using BONMIN

### Re: Stuck in the final part of MINLP using BONMIN

Bonmin solves many subproblems. The "EXIT: Optimal Solution Found" is from a subproblem (solver IPOPT) and does not mean that your original problem has been solved. The default algorithm in Bonmin is branch-and-bound. Even if you hold an integer solution it might still take a long time to prove optimality of this solution. You should carefully inspect the "Solution summary" when your Bonmin run terminates and see what it says.

-Michael

-Michael

### Re: Stuck in the final part of MINLP using BONMIN

Dear Micheal,

Thank you for your reply.

There is a quadratic function in my model and I changed it into a linear one but the model is still a minlp. After the change it turns out to be ok.

I'm wondering why the square ones make the problem so hard. If it is not the problem, I should check my model again(つД`)

-Shouyi Hu

Thank you for your reply.

There is a quadratic function in my model and I changed it into a linear one but the model is still a minlp. After the change it turns out to be ok.

I'm wondering why the square ones make the problem so hard. If it is not the problem, I should check my model again(つД`)

-Shouyi Hu

### Re: Stuck in the final part of MINLP using BONMIN

Perhaps it makes your discrete variables to take non-integral values in the relaxed optimal solution (something you don't have with the linear variant) and the B&B has to work much harder. Without the model it's just guessing.

-Michael

-Michael