Originally posted by: savant_ning

Hi  I have this problem, I don't know what's the warning meaning, the warning see attachment

thanks

Do you mean

Warning -- Non-integral bounds for integer variables rounded.

? Isn't that self-explanaining? It says that you have integer variables that have non-integer bounds. Since integer variables can only take integer values, their bounds should also be integers and CPLEX has rounded the bounds appropriately. Having non-integer bounds on integer variables is likely to be a modeling or data error, hence CPLEX emits a warning.

Originally posted by: savant_ning

but my variables is all defined type :"B" or "C"

so  this error means my variables defined by type "B"  data error or relative constraints have error?

'B' is a special kind of integer variable (one that can take only values 0 and 1).

Originally posted by: savant_ning

yeah, "B" is a binary variables, the lb = 0.0; ub = 1.0.   my variables is correct,

and below the warning, it give me ROW "cxxx" infeasible, all entries at implied bounds.

I use problem.conflict.get()  and problem.conflict.all_constraints()

it print :   no conflict  available

and I check my constraints,  I think its no problem,  and the software has run some time,  about 20.82sec.

and then it give me no solution exists.....

How do I solve this problem?

best regards

You have to call Cplex.conflict.refine() before you can query any conflict. By default CPLEX does not compute a conflict because that may be time consuming.

Please take a look at reference documentation of the refine function here, in particular at the code example given there.

You may also call Cplex.write('model.sav.gz') and then post the model.sav.gz here. Then other people can take a look as well.

Originally posted by: savant_ning

okay, the model.sav.gz  in the first attachment, I don't know how to open it, I use SPSS and R to try open it, but it doesn't work

and the second is the conflict print, is it means my variables value is outside range of my [lb,ub]?

Only CPLEX can read SAV files.

The error message looks odd, assuming you called Cplex.conflict.refine(). Could it be that you called Cplex.conflict.refine_MIP_start() instead?

In any case, the conflict in your model is this (you did not assign names to variables/constraints, so the SAV file only has default names):

Subject To
 c518:  - x2 + x16 <= -700
 c5033: - 5182 x1 - x2 + x7201 >= 0
Bounds
 0 <= x1 <= 1
      x2 Free
      x16 >= 0
 0 <= x7201 <= 1
Binaries
 x1  x7201
Generals
 x2  x16

The first constraints requires x2>=700 but with the bounds on the variables there is no way to satisfy the second constraint in case x2>=700 (the second constraint requires x2>=-1).

Originally posted by: savant_ning

yup, I called cplex.conflict.refine_MIP_start(), and I call cplex.conflict.refine() again, it give me this output. see attachment

and I checked my code, the variables initial bounds are no problem...

maybe the bounds in my constraints have problem?

I'm checking my constraints now

very thanks

best regards

Didn't you follow the code example in the reference documentation of the refine() function like I suggested? Your output looks good but it seems you are still trying to access a solution. This does not exist since your model is infeasible! The code example also shows how to display the constraints that participate in the conflict.

I would not know whether your problem is with variable bounds or with the constraints since I don't know what you are trying to model. I suggest to assign names to all variables and constraints and then check whether the offending bounds and constraints look as expected.

Originally posted by: savant_ning

okay，very thank u.

I have get optimal solution~

My data have conflict with my constraint, I regenerate mt data, and the model has solution.

best regrads