Originally posted by: amedury

I am trying to code a program which involves repeatedly solving an optimization problem after modifying some pieces of the data. In particular, I have a multidimensional array (3) of the form:

int actns<i,j> in linksthttp://t in 1..40http://n in 1..100 = 0;

As per the help manual, I can modify arrays if they have been externally initialized. However, a 3+ dimensional array cannot be initialized externally.

I have tried to create a tuple to overcome this limitation in the following manner:
tuple action {
key link segment;
key int t;
key int n;
int act;
};

{action} actn = ...;

But when I try to modify the tuple in the scripting in main(), it gives an error that "the collection is immutable because it referenced by another element." I use the following script:

var actnsm = produce.actns; (actns is the array that initialized internally using multidimensional tuple)
var actnsi = produce.actn; (actn is the tuple of interest)

<omitting the rest of the script>

for(var x in actnsi){
x.act= actnsmhttp://x.segmenthttp://x.thttp://x.n;
data.actnshttp://x.segmenthttp://x.thttp://x.n=actnsmhttp://x.segmenthttp://x.thttp://x.n;
}
data.actn=actnsi;

I am also told that the syntax for tuples changes in the scripting process, so I would appreciate if you could help me with this issue. The big picture idea is that I would like to modify the multidimensional array/tuple and resolve the optimization.

Thanks!
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: amedury

Sorry about the random links (wasn't expecting them to come up). Hopefully this will work out:
for(var x in actnsi){
x.act= actnsm"";
data.actns"="actnsm"";
}
data.actn=actnsi;
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Hi,

Can you clarify what you mean by a 3D array cannot be initialized externally?
Here is an example of how a 3D array could be initialized in a .dat file:

actns = [ [[111,112],[121,122],[131,132],[141,142]], [[211,212],[221,222],[231,232],[241,242]], [[311,312],[321,322],[331,332],[341,342]] ];

with the array declared as

tuple Link 
{ 

int i; 

int j; 
} 
{Link
} links = 
{<1,1>, <1,2>, <1,3>
}; 

int actns[links][1..4][1..2]=...;

Then in your main script, you can update your array as follows:

main 
{ thisOplModel.generate(); var masterDef = thisOplModel.modelDefinition; var masterCplex = cplex; var dataElts = thisOplModel.dataElements; 

if (cplex.solve()) 
{ writeln(
"obj1 = ",cplex.getObjValue()); 
//update array actns 

for (var x in thisOplModel.actnsi) dataElts.actns[x.segment][x.t][x.n]=thisOplModel.actnsm[x.segment][x.t][x.n]+5; var opl2 = 

new IloOplModel(masterDef,masterCplex); opl2.addDataSource(dataElts); opl2.generate(); masterCplex.solve(); writeln(
"obj2 = ",masterCplex.getObjValue()); 
} 
}

If this doesn't answer your question, can you post a reduced sample that illustrates the issue you're facing?
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: ikucukkoc

Dear SystemAdmin

 

I am using CPLEX and I need to preprocess data before the optimisation.

In simplified version, I have a tuple "TaskTimes" as below and l need to modify the last column of every row by multiplying with a constant value, say 5.

 

Sample.mod file

{task} TaskTimes = ...;

 

Sample.dat file:

TaskTimes = {

 

Consequently; I should obtain following tuple:

 

TaskTimes = {

 

Your prompt help and support is highly acknowledged.

Many thanks.

Best Regards

Ibrahim

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Hi,
I am facing a similar issue.
I have

tuple TripleIndex{
int t;
int u;
int g;
}
{TripleIndex} Index = {<t,u,g> | t in T, u in U, g in 1..G[t][u]};
float T Index = ...;

I need to calculate some data before I process my MILP.
Here's what I want to do :

for(var i in thisOplModel.Index){
for (var a = i.g; a < Ghttp://i.thttp://i.u;a++) {
Turbinehttp://thisOplModel.TripleIndex(i.t,i.u,a);
}
}
}

I need to access the tuple indexed T.
So I have to build the coresponding tuple index.
How can I "create" a tuple in OPL script ?

Thanks !
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

for (var a = i.g ; a < G http:// i.t http:// i.u ; a++ ) {
Turbine http:// thisOplModel . TripleIndex ( i.t , i.u , a ) ;
}
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Hi,

In your code extract, "TripleIndex" is a tuple type definition, not a set so you can't use it to access elements of the sets Index.
You need to work with the set itself: Index.
So if "Turbine" is an array indexed by the set "Index", you would access the cell indexed by <i.t,i.u,a> with the following syntax:

Turbine[Index.find(i.t,i.u,a)] = /*the value you want to set*/;

The method find for tuple set is documented here: IDE and OPL > Optimization Programming Language (OPL) > IBM ILOG Script Reference Manual > OPL Classes > IloTupleSet

I am unclear what T represents in your code sample but the way you've declared it doesn't work as you are mixing float and tuple set.
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Hi, thanks a lot for your fast answer !
And yes you're right, it's actually a mistake...

It is float Turbine = ... ;
I just wanted to simplify the code, but I've done it too fast...

Thanks again
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: ikucukkoc

Dear SystemAdmin

I am using CPLEX and I need to preprocess data before the optimisation.

In simplified version, I have a tuple "TaskTimes" as below and l need to modify the last column of every row by multiplying with a constant value, say 5.

Sample.mod file

{task} TaskTimes = ...;

 

Sample.dat file:

TaskTimes = {

 

Consequently; I should obtain following tuple:

 

TaskTimes = {

 

Your prompt help and support is highly acknowledged.

Many thanks.

Best Regards

Ibrahim

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

 

could

{task} TaskTimes2 = {<e.h,e.j,e.i,e.x,5*e.pt> | e in TaskTimes};
execute
{
 
  TaskTimes2;
 
 
}

 

help ?

 

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer