Originally posted by: banun

Hi,

in my "model.lp" file I get this weird line :

id244: >= 0

(like something is missing from the left of ">=", it seems)

What is that ?
Thanks
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: John Cui

Yes， you are right, the coefficients of this constraint are zero.
0x1 + 0x2 + 0x3 + ... + 0xn >= 0
So you need to check the coefficients of this constraint in your program.

John Cui
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

You have created a constraint with an empty left-hand side. Like John said, this is the same as constraint with all-zero coefficients:

0x + 0y + 0z >= 0

If the right-hand side is 0 then the constraint will always be satisfied and CPLEX will eliminate this constraint during presolve.
However, since the empty left-hand side seems to be unexpected you should check your program and figure out why you created a constraint with empty left-hand side.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: banun

Thanks for you answers,

what about this :

i240372: id434555#259226 = 1 <-> z(82,(516,522,0)p7)#12167
          - z(84,(50,565,0)p5)#12725  = 0

?

It looks strange too..
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: banun

or is the second line a continuation of the first ??
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

This is an indicator constraint. The '<->' represents "if and only if". And yes, the second line is a continuation of the first one.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: banun

another question :

I have :

IloIntVar va = getVar(v);
        IloIntVar var1 = getVar(first_var);
        model.add(va != var1);

but in my model.lp file I don't see " != " constraint at all.
Is it possible this constraint is converted to something like that :

id1581088: id1581094#170096  = 0
       i151242: id1581094#170096 = 1 <-> z(64,(877,663,0)p13)#8100 - z(37,(839,667,0)p13)#4002  = 0
        0 <= id1581094#170096 <= 1

(id1581094#170096 is under "Binaries")

?

thanks
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Yes, that would make perfect sense. CPLEX (and integer linear programming) does not directly support "not equal" constraints. Hence such constraints must be reformulated as something that CPLEX supports. And what you have given is a reformulation of "not equal" using indicator constraints.
#CPLEXOptimizers
#DecisionOptimization