Originally posted by: SystemAdmin

[polochon said:]

Hi,
I try to implemente the Golomb Rulers problem, I try it with a simple exemple (m=4) but I have an erroneous result (x=(0, 1, 2, 3))
to express  the inequality, we have the " !=" operator no?!
this is my code :

int m=...;

dvar int x[1..m] in 0..m*m;

minimize x[m];
subject to {
    forall (i in 1..m-1)
        x[i] < x&#91;i+1&#93;;<br />    forall (i in 1..m, j in 1..m,
           k in 1..m, l in 1..m: (i < j, k < l))<br />       x[j] - x[i] != x[l] - x[k];
}

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[alain.chabrier said:]

Hi,
missing parenthesis ?

  (x[j] - x[i]) != (x[l] - x[k]);

operators precedence are given somewhere in the documentation.

Alain
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

I try also with parenthesis but no way...
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

is there someone to execute this program and tell me the result?
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[shaw said:]

Hello,

It seems that you have an error in your model because
there are values of i,j,k,l for which j-i = l-k and so your
difference constraint is necessarily violated.

Perhaps you mean:

forall (i,j,k,l in 1..m : i < j && k < l && i < k && j < l)<br />  x[j] - x[i] != x[l] - x[k]


Paul

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

Hello,

thanks Paul, I think that I see the problem, I must have :

forall (i,j,k,l in 1..m : i < j && k < l && (i != k || j != l))<br />   x[j] - x[i] != x[l] - x[k]

we can also have for example : x[3]-x[5] != x[2]-x[5]

so, i != k  or j != l
[b]

I try it and it works fine :)
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[shaw said:]

Thanks for decoding my reply.  A typo seems to have corrupted
the last expression and its font.  I just wanted to follow up to say
that your filtering expression on the forall can be stronger than the
one you use.  For example, adding the difference constraint on

x[5] - x[2] != x[6] - x[2]

is pointless.  In addition, you don't break symmetries with your rule
and so you will add each difference constraint twice.  i.e. you would also add:

x[6] - x[2] != x[5] - x[2]


Paul
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

That's true, my goal is to shift from a model to another more performed, I reformulate the last constraint by:


forall (i in 1..m, j in 1..m: i < j)<br />d[i,j]=x[j]-x[i];

alldifferent(all(i in 1..m, j in 1..m: i < j) d&#91;i,j&#93;);<br />

thanks Paul,
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

Hello,

does the alldiff constraint (ex. alldifferent(all(i in 1..m, j in 1..m: i < j) d[i,j]) ) exist in CP Optimizer?
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[Didier Vidal said:]

does the alldiff constraint (ex. alldifferent(all(i in 1..m, j in 1..m: i < j) d&#91;i,j&#93;) ) exist in CP Optimizer?<br />
Yes it does. Search the keyword 'allDifferent' (with capitalized D) in the OPL documentation.

Didier.
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

Thanks Didier, it works fine with allDifferent.

I have another question :) , I'd like to define a matrix d[i,j] that will be initialised to x[j]-x[i] :


...
int d[1..m,1..m]=...;
...
subject to{
...
forall (i in 1..m, j in 1..m : i < j)<br /> d[i,j] == x[j]-x[i];
...
}


how to do that?
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[shaw said:]

Hello,

Perhaps you want this?

dexpr int d[i in 1..m][j in 1..m] = x[j]-x[i];



Paul
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

Thanks Paul,
I have now another problem :


dexpr int d[i in 1..m][j in 1..m] = x[j]-x[i];
.
.
.

allDifferent(all(i in 1..m, j in 1..m : i < j) d&#91;i,j&#93;);<br />

I have an error on the allDifferent, the message is :


Argument type mismatch calling function operatorALL (float):[range] with arguments range, dexpr int.

#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[Didier Vidal said:]

The allDifferent constraint works only on decision variables, not on decision expression. So You have to change a bit your model. See below an example that uses a tuple set to define only the variables you need.


using CP;

int m = 10;
dvar int x[1..m] in 0..1000;

tuple indexerTuple {
int i;
int j;
}
{indexerTuple} indexes = {<i, j> | i,j in 1..m : i < j};<br />// No longer define d as a dexpr
dvar int d[indexes];

constraints {
// Then force values of d by a constraint
forall(ind in indexes)
  d[ind] == x[ind.j]-x[ind.i];


// allDifferent statement applies on the array of decision variables
allDifferent(d);

}



Didier.
#DecisionOptimization
#OPLusingCPOptimizer

Originally posted by: SystemAdmin

[polochon said:]

Thanks Didier, it works fine :)

Nadjib.
#DecisionOptimization
#OPLusingCPOptimizer