When one wants to solve a convex Quadratic Problem in Cplex with a given time limit, how is it possible to get lower and upper bounds on the actual optimal value ?
The primal and dual objective value should provide this information (and they should be equal if the convex problem is solved), but I can't figure a way to access them.
When using Cplex's CLI, I see an "Objective" value that is returned after optimization ; I know it is not a primal nor dual bound as it can be above or below the actual optimum of the problem depending on the time limit provided to Cplex. I can't find any definition for this value when optimality isn't reached. Similarly I can't see any guaranties on what is returned by the cplex.getObjValue() method in C++.
I see a lot of information on how to get bounds for an Mixed Integer Problem (for instance getMIPRelativeGap() ), but nothing on a simple fractional problem.
Is it possible to get upper/lower bounds on the optimum of a problem after "partially solving" a QP (and stopping before convergence because of a time limit) ?
Thanks you for your help!
