I have a MILP model with continuous variables ($y_{ij}$) and binaries ($x_{ij}$). I am using Java and cplex 12.8 for implementation.
Is there a way by which - using the Java-cplex API - I would be able to write only one model that can be solved first with the binary $x_{ij}$'s and then for the continuous variables ($0 \leq x_{ij} \leq 1$)? I want to see the gap between LP-R and MIP on the fly - and don't want to build two different LP and MILP models.