I use CPLEX as an optimizer and JuMP apackage. I have a big IP that needs at least 2hours to generate a soultion.
I wonder is there a way to stop it early? when running a model that takes a long time to give you an optimal soultion:
1- is it possible to ste time limit?
2- is it possible to see soultions associated with upper and lower bound that CPLEX shows?
3- is it possiblr to see the cuts?
The lower bound is obtained by solving continuous relaxations (LPs), so there should be corresponding solutions (whether or not they’re available in a solver is another question).
The lower bound is obtained by solving continuous relaxations
A lower bound can obtained by solving continuous relaxations. But if CPLEX printed a particular bound, there might not be a continuous solution to go along with that. (It might have used some rounding heuristic from an LP solution.)
So perhaps the more nuanced answer to (2) is: if you use a callback then you can see the points in the continuous relaxation visited by CPLEX, as well as the current incumbent (upper bound) at each callback during the solve: