Originally posted by: SystemAdmin

The attached is a small LP problem I solved with CPLEX12.4. There are two bounded variables, x4 and x5, whose sum is the objective. The problem solves with either the primal or the dual methods, and CPLEX returns an "optimal" solution as well as an "optimal" value in both cases. Strangely, the solution obtained by using the dual method violates the upper bound of x5, while the primal method returns an incorrect "optimal" value. Although the LP problem is practically infeasible, why CPLEX does not report the infeasibility.
Thank you for your help.

Amar
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: T_O

In the interactive optimizer of CPLEX 12.5, primopt, tranopt and baropt report infeasibility.

In the API, you can sometimes obtain solutions from infeasible bases (I guess if a factorization exists). There it does not matter whether the basis is primal/dual feasible. You should check the status before obtaining the solution.

Best regards,
Thomas
#CPLEXOptimizers
#DecisionOptimization