Originally posted by: SystemAdmin

Hi everybody,

I am solving my LP by sifting like:
cplexSPD.setParam(IloCplex::RootAlg, CPX_ALG_SIFTING );

the results throw exception 3003, as comes in the following.
Sifting iteration 1 24720: Objective = 3172637.328263
Sifting iteration 2 24967: Objective = 2866431.199449
Sifting iteration 3 24823: Objective = 2860963.949237
Sifting iteration 4 24833: Objective = 2860729.852117
Sifting iteration 5 24834: Objective = 2860721.634061

SPD interior status: Infeasible
Warning: Output names have been modified to conform to LP format.
ERROR: CPLEX Error 3003: Not a mixed-integer problem.
While Cplex interactive optimizer reports optimal for the same model when extracted as lp file:

Iteration: 31012 Objective = 2865010.563650
Iteration: 31320 Objective = 2863247.359837
Iteration: 31632 Objective = 2861204.234382
Iteration: 31959 Objective = 2860740.969570
Iteration: 32054 Scaled infeas = 0.000003
Iteration: 32058 Objective = 2860721.334451
Removing shift (2799).
Iteration: 32061 Infeasibility = 0.000002
Iteration: 32070 Objective = 2865586.104266

Primal simplex - Optimal: Objective = 2.8607216341e+006
Solution time = 26.48 sec. Iterations = 32098 (20920)
Any comment is highly appreciated.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Sorry, my confusion:
The problem is not the error 3003 which is some thing else.
The question is why infeasibility against optimality!!?!
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: RWunderling

Shahin:

It seems that you are saving the problem in LP format in order to try again with the interactive optimizer. Could you please instead save it in SAV format?! Only then will you look at the exact same model. It could be that your problem is at the brink of infeasibility and a slight change, e.g. in the order of variables, changes the conclusion
CPLEX comes to for this model.

If that does not change things, could you please elaborate, on how you construct the model and what operations you
do before the solve in question.

Roland
#CPLEXOptimizers
#DecisionOptimization