Hi Sai,
The behavior of the Barrier algorithm on this instance is the following (thanks, EdKlotz!):
-
at iteration 18, the solution is not yet really feasible: primal infeasibility is still around 117
-
this infeasibility decreases during iterations 19-21 (divided by 10), at the expense of only a very slight increase of the primal objective value.
-
then primal objective is more or less stable, while primal infeasibility goes down to negligible
-
during all this process, dual objective value increases regularly, including during iterations 19-21.
The algorithm is making progress overall. Nothing is wrong, or even of any significance, here.
Best regards,
Xavier
#DecisionOptimization#OPLusingCPLEXOptimizer