Originally posted by: SystemAdmin

[fk185 said:]

Hello,
I'm new to cplex, I hope my question isn't stupid.
I'm trying to solve a simple bounded least squares problem: min|Cx-d| where x is variable and C and d are Vectors. lb <= x <= ub.<br />As cplex doesn't support least squares (please correct me if i'm wrong), i try to write the problem as min((sum_over_all_i(a_i*x - b_i) )^2)
(no square root).
I used QCPex1.java as a starting point. i don't use any constraints, only upper and lower bound.
I deleted all constraints stuff and edited populateByRow like this:

private IloNumVar[] populateByRow(IloMPModeler model) throws IloException
{
double[] lb = new double[ C_Vector.size() ];
double[] ub = new double[ C_Vector.size() ];
for(int i = 0; i != lb.length; i++)
{
lb[i] = lowerBound;
ub[i] = upperBound;
}

IloNumVar[] x  = model.numVarArray(C_Vector.size(), lb, ub);

// sum_i = sum(C_i*x - d_i) )

// sum_0 = ((C_i*x) - d_i)^2
IloNumExpr sum_i = model.square(model.diff(model.prod( C_Vector.get(0), x[0]), d_Vector[0]));
// add x_i to others
for(int i = 1; i != d_Vector.length; i++ )
{
// sum in step i
IloNumExpr sumInStep = model.square(model.diff(model.prod( C_Matrix.get(i, x[0]), d_Vector[i]));

// add sumInStep to other sums
sum_i = model.addEq(sum_i, sumInStep);
}

model.minimize(sum_i);
return x;
}


When i call cplex.solve() i get an exception i don't understand: CPLEX Error 6002: Sense must be 'L' or 'G' .
Can anybody tell me what's wrong?

Thank you,
Florian

#CPLEXOptimizers
#DecisionOptimization

Originally posted by: SystemAdmin

[fk185 said:]

I think the mistake was

// add sumInStep to other sums
sum_i = model.addEq(sum_i, sumInStep);

it should be

// add sumInStep to other sums
sum_i = model.sum(sum_i, sumInStep);



#CPLEXOptimizers
#DecisionOptimization