CPLEX OPL 在约束规划算法中使用决策变量作为浮点数

Question

I am trying to run the following code, but there is an error in dvar float+

using CP;
int rows=3;
int columns=3;
range para=1..rows*columns;
float model1[para]=...;
dvar float+ model0[para];
maximize 
sum(p in para) model0[p]*(lg(model0[p]/model1[p]));
  subject to{
    c1:
    sum(p in para)
    model0[p]==3.0;    
    }

Answer 1

Decision variables with CPOptimizer cannot be float.

As can be seen in the example floatexpr.mod you can use a float dexpr in order to use decimal values computed out of a integer decision variables.

In your case:

using CP;

execute
{
cp.param.timelimit=10;
}

int scale=1000;
int rows=3;
int columns=3;
range para=1..rows*columns;
float model1[p in para]=p/10;

dvar int+ scalemodel0[para];
dexpr  float model0[p in para]=scalemodel0[p]/scale;
maximize 
sum(p in para) model0[p]*(lg(model0[p]/model1[p]));
  subject to{
    c1:
    sum(p in para)
    model0[p]==3.0;    
    }