Originally posted by: SystemAdmin

Hi, I´m working with a very difficult formulation for a scheduling problem and I need to insert a initial solution for improve my results, I already read and implement the example of the OPL with de vectors in the main script and It was great , but what I actually discover and where resides my question, is that when a decision variable is set as float or float+, the OPL doesn´t take the initial solution and starts whit no initial solution, my question is: The problem is only of OPL or is a problem of CPLEX, and there is a way to resolve these problem?
Regards
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

In the warm star example, why if I make these modifications to the decisions variables the program don´t take the initial solution:

range r = 1..10;
dvar float+ x[r];
dvar float+ y[r];
// The following array of values (defined as data) will be used as
// a starting solution to warm-start the CPLEX search.
float valuesi in r = (i==5)? 10.5 : 0;
minimize
sum( i in r ) x[i] + sum( j in r ) y[j];
subject to{
ctSum:
sum( i in r ) x[i] >= 10.5;
forall( j in r )
ctEqual:
y[j] == j;
}

main{
thisOplModel.generate();
var def = thisOplModel.modelDefinition;
// Default behaviour
writeln("Default Behaviour");
var opl1 = new IloOplModel(def, cplex);
opl1.generate();
cplex.solve();
writeln(opl1.printSolution());
// Setting initial solution
writeln("Setting initial solution");
var opl2 = new IloOplModel(def, cplex);
opl2.generate();
var vectors = new IloOplCplexVectors();
// We attach the values (defined as data) as starting solution
// for the variables x.
vectors.attach(opl2.x,opl2.values);
vectors.setVectors(cplex);
cplex.solve();
writeln(opl2.printSolution());

opl1.end();
opl2.end();
0;
}

You can see in the results that the program don´t take the initial solution:

Default Behaviour
// solution (optimal) with objective 65.5
// Quality There are no bound infeasibilities.
// There are no reduced-cost infeasibilities.
// Maximum Ax-b residual = 0
// Maximum c-B'pi residual = 0
// Maximum |x| = 10.5
// Maximum |slack| = 0
// Maximum |pi| = 1
// Maximum |red-cost| = 1
// Condition number of unscaled basis = 1.0e+000
//

x = [0
0 0 0 0 0 0 0 0 10.5];
y = 1 2 3 4 5 6 7 8 9 10;

Setting initial solution
// solution (optimal) with objective 65.5
// Quality There are no bound infeasibilities.
// There are no reduced-cost infeasibilities.
// Maximum Ax-b residual = 0
// Maximum c-B'pi residual = 0
// Maximum |x| = 10.5
// Maximum |slack| = 0
// Maximum |pi| = 1
// Maximum |red-cost| = 1
// Condition number of unscaled basis = 1.0e+000
//

x = [0
0 0 0 0 0 0 0 0 10.5];
y = 1 2 3 4 5 6 7 8 9 10;
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Changing the dvars from int+ (non-negative integers) to float+ (non-negative real numbers) changes the problem. You can see this in the results:

x[10] = 10.5

which is not an integer.

Setting the initial solution is different from fixing the decision variables. In the second run, "values" is still used as the initial guess. The optimization algorithm then determines that x(10)=10.5 and all others =0 has a higher objective function.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Note that the integer+ problem gives an objective function of 65, and your float+ problem gives an objective function of 65.5.
#DecisionOptimization
#OPLusingCPLEXOptimizer