Originally posted by: gabbersepp

Hello,

i am relatively new to ILOG/CPLEX/OPL and i'm running into a weird issue.

Given following model:

int H = 10000000;
range NRange = 1..2;
range MRange = 1..1;
int t[NRange][MRange] = [
        [10],
                [11]
];
int IM[NRange][NRange] = [
                [0, 1],
                [1, 0]
];
                                                                                                                        
dvar int Y[NRange][NRange][MRange] in 0..1;
dvar int X[NRange][MRange]; 

minimize (
        0
);
  
subject to {
        forall(i in NRange)
          forall(h in NRange)
            forall(k in MRange)
              (X[h][k] - X[i][k])*IM[i][h] >= (t[i][k] - (H + t[i][k])*(1 - Y[i][h][k]))*IM[i][h];
              
        forall(i in NRange)
          forall(h in NRange)
            forall(k in MRange)
              akl: (X[i][k] - X[h][k])*IM[i][h]  >= (t[h][k] - (H + t[h][k])*Y[i][h][k])*IM[i][h];
}

i get different results, with respect to H.

If I select H = 10000000, the decision variable Y contains only '0'.

If i take a further look at the result, the IDE tells me, that "0 >= 11" (see attachment).

If i change H to 1000, everything seems OK.

My question:

1) Why behaves the solver different, depending on H?

2) Why want the solver tell me that 0 >= 11?

Thanks for your help.

best regards,

Josef

Hi

which version do you use ?

With your model if I add

execute
{
writeln("Y=",Y);

}

I get

Y= [[[0]
             [1]]
         [[0]
             [0]]]
 

regards

Originally posted by: gabbersepp

Using the executionblock i get:

Y= [[[0]
             [0]]
         [[0]
             [0]]]

My version:

IBM ILOG CPLEX Optimization Studio

Version: 12.5
Build-ID: 20121025-0121

My OS is a 32Bit Windows 7 .

Originally posted by: gabbersepp

For those who are interested:

I have tested the modell on another System:

Version: 12.6.1, 64Bit

Windows 7 64Bit

With this Konfiguration everything is fine.