Decision Optimization

Originally posted by: DiegoEm

Hi, I have a C# code solving a problem with int variables. I tried to change the variables type to float and the code stopped working. this is my code:

CP cp = new CP();

INumVar[] assignment = null;
INumExpr profit = null;

assignment = new INumVarhttp://parameters.NumberOfItems;
INumExpr total = cp.NumExpr();
for (int i = 0; i < parameters.NumberOfItems; i++)
{
assignment[i] = cp.NumVar(0, parameters.UnitsAvaialble[i], NumVarType.Float);
total = cp.Sum(total, assignment[i]);
}

cp.Add(cp.Le(total, 15));

profit = cp.NumExpr();

//Adding revenue for each delivered item
for (int i = 0; i < parameters.NumberOfItems; i++)
profit = cp.Sum(profit, cp.Prod((parameters.RevenueForItem[i]), assignment[i]));

cp.Add(cp.Maximize(profit));

cp.SetParameter(CP.IntParam.LogVerbosity, CP.ParameterValues.Verbose);
cp.Solve();
Console.WriteLine("profit: " + cp.GetValue(profit));

for (int i = 0; i < parameters.NumberOfItems; i++)
{
Console.Write(" Item " + (int)(i + 1) + ": " + cp.GetValue(assignment[i]));
}

Console.ReadLine();

This code fails throwing a "variable is not fixed" exception while trying to get the profit value. In the console I noticed the number of variables is 0.

Changing NumVarType.Float to NumVarType.Int fixes it. But I'd like my variables to be float.

What am I missing here???

Thanks,
Diego.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

It looks like you are using CP optimizer and not the CPLEX solver. Is this correct? If so then you might find better help in the CP optimizer forums.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: GuangFeng

CP Optimizer does not support float decision variables, only float expressions. So the symptom looks to be expected, although I am not sure whether the error message is meaningful to you. For further inquires, please post a question by following the URL in Daniel's post.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Diego,

CP Optimizer can solve problems on integer decision variables only. Floatting points variables are admitted but they must behave as an expression. That is they must be fixed to a value once all the integer variables are fixed. They cannot be decision variables.

Apparently, you want to solve optimization problems with floatting-point decision variables. In this case, the way to go is to use CPLEX.

Regards,

Philippe
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: DiegoEm

Thanks you for your reply. Actually I'm trying to solve an integert problem but was trying to run the relaxation to see how long does it take.

I moved to use Cplex and it solves both problems. It solves the integers problem much faster the CP. Can you please add some details on CP vs Cplex?

Thanks again,
Diego.
#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

Hi Diego,

If your model has linear constraints, then using CPLEX is the way to go because CP solvers cannot do interesting inferences on these models with little structural information. If you solve, say a scheduling problem, and its structure fits the the CP modelling capabilities (with resources constraints, makespan, non-overlap constraint,...), then CP solvers are able to exploit this information to solve the problem more efficiently.

You can have a look at the CP optimizer dcumentation to see the the examples provided and the special modelling objects that can be exploited.

Philippe
#CPLEXOptimizers
#DecisionOptimization