Any smarter way to force a set of decision variables to be equal?
(If not feel free to use the solution)
Given the following set:
ID1 | ID2
------- | -------
A | AA
B | AA
C | BB
C | AA
C | CC
D | CC
e.g. initialized in OPL by
//Set ID
tuple ObjectID{
string ID1;
string ID2;
}
{ObjectID} ID = {
<"A", "AA">,
<"B", "AA">,
<"C", "BB">,
<"C", "AA">,
<"C", "CC">,
<"D", "CC">,
};
And a decision variable x[ID] to be declared in OPL as
dvar int+ x[ID]
The decision variable x[ID] should be equal if ID1 is equal for all ID2.
Example:
x[<"C", "BB">] == x[<"C", "AA">] == x[<"C", "CC">]
Pairwise comparison of all dvar's with identical ID1 and different ID2.
forall(
id_1 in ID, id_2 in ID:
id_1.ID1 == id_2.ID1 &&
id_1.ID2 != id_2.ID2
)
x[id_1] == x[id_2];
first improvement would be to use ordered to divide by 2 the number of equality constraints:
forall(ordered id_1,id_2 in ID :id_1.ID1 == id_2.ID1 )
x[id_1] == x[id_2];
and second improvement could be to move from (n-1)*n/2 constraints to (n-1) constraints
//Set ID
tuple ObjectID{
string ID1;
string ID2;
}
{ObjectID} ID = {
<"A", "AA">,
<"B", "AA">,
<"C", "BB">,
<"C", "AA">,
<"C", "CC">,
<"D", "CC">
};
dvar int+ x[ID];
{string} Id1s={i.ID1 | i in ID};
{string} Id2PerId1[id1 in Id1s]={i.ID2 | i in ID : i.ID1==id1};
subject to
{
forall(id1 in Id1s) forall(id2 in Id2PerId1[id1] diff {last(Id2PerId1[id1])})
x[<id1,id2>] == x[<id1,next(Id2PerId1[id1],id2)>];
}
`