Originally posted by: JDalal

I am facing an interesting situation. When I declare some decision variable x>=0 as:

1) IloNumVar x(env,ILOFLOAT);

2) IloNumVar x(env);

there is a significant difference in result! Any suggestion or right declaration? Thanks in advance.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

> 1) IloNumVar x(env,ILOFLOAT);

This will result in the creation of a continuous variable with a lowerbound of 2 (x>=2).
This is because the ILOFLOAT is an enum which gets assigned a value of 2 since its Type=Float
The compiler will then try to match the closest constructor for IloNumVar and since the
second argument there is the lowerbound value, hence the result.

> 2) IloNumVar x(env);

This will result in the creation of a continuous variable with a lowerbound of 0 (x>=0).
When no bounds are explicitly mentioned, as in this case, CPLEX assigns a lowerbound of 0
and an upperbound of +IloInfinity. An equivalent call would be IloNumVar x(env,0,IloInfinity,ILOFLOAT);

I would suggest you to use ILOFLOAT only after explicitly specifying the LB and UB values to avoid mismatches.
Also, it might be a good idea to assign a name to the variable for tracking purposes, if you were to
export the model to a human readable LP/MPS format instead of the engine assigning names default names.
One such declaration could look like the following:

IloNumVar x(env,0,IloInfinity,ILOFLOAT,"x");

Hope this helps.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: JDalal

Thanks a lot. It answers my doubt. In fact for my specific problem instance, I was trying hard to find why x is taking 2 when it should be 0 !
#CPLEXOptimizers
#DecisionOptimization