Originally posted by: char507

Hello,

I am slowly building a complex data structure with tuples containing other tuples. For example, say I have defined the following tuples and sets of tuples:

{code}
tuple Asset{
key string id;
float cost;
}

tuple Customer{
key string id;
float budget;
}

tuple CustomerUsesAsset{
Customer c;
Asset a;
}

{Asset} listOfAssets = …;
{Customer} listOfCustomers = …;
{CustomerUsesAsset} CustomerAssetLinks = …;

{/code}

In the long run, I will be initializing the data set through external code but for now i'm trying to use a .dat file to test out the model.

My problem is that I do not know how to initialize the "CustomerUsesAsset" tuple. These objects already exist in their respective sets but I don't know how to link them to the third tuple set. Is there a way to do this in the .dat file? In script maybe?

The purpose of building these new data types is to cut down on memory usage so obviously I do not want to instantiate each tuple object more than once (plus they need to be unique). I just can't figure out how to make those links.

Any help would be greatly appreciated! Complicated tasks such as this do not seem to be present in the documentation...
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: SystemAdmin

Not sure I fully understand your question.

Here is how to fill a tupleSet with subTuple via comprehension, via direct data, via scripting.

//X and X1 are just to make the sample work.
{string} X = {"f0", "f1", "f2", "f3", "f4", "f5", "f6", "f7", "f8", "f9", "f10"}
;
{string} X1 = {"f0", "f2", "f4", "f6", "f8", "f10"};
tuple Asset{
key string id;
int cost;
}
tuple Customer{
key string id;
int budget;
}
tuple CustomerUsesAsset{
key Customer c;
key Asset a;
}

{Asset} listOfAssets = {<item(X,i),i+5> | i in 1..10};
{Customer} listOfCustomers = {<item(X, i),i+10> | i in 1..10};;

// fill with via iterators with some filtering.
{CustomerUsesAsset} C2 = {<<id1, cost>, <id2, budget>> | <id1, cost> in listOfAs
sets, <id2, budget> in listOfAssets : id1 not in X1 && id2 not in X1};

//fill it with direct data (from .dat file, the separator must be removed in some places)
{CustomerUsesAsset} C3 = {<<"f1", 6>, <", 12>>, <<"f2",7>, <",13>>};
// another equivalent formula:
{CustomerUsesAsset} C4 = {<t1, t2> | t1 in listOfAssets, t2 in listOfAssets : t1.id not in X1 && t2.id not in X1};

//fill with via scripting: give all the components.
execute{
C2.add("f3",8,"f4",14);
}
execute{
writeln(C2);
writeln("\n");
writeln(C3);
writeln("\n");
writeln(C4);
writeln("\n");
}

// to find tuples in the tupleSet, you can use the keys:
// in general, you can use the keys to find and iterate on tupleSets.
execute{
writeln(C2.find("f3", "f4"));
}
#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: char507

Okay, I have a few updates to try and help everyone better understand the problem:

The objective here is to initialize a large set of data in this form with minimal memory consumption. I don't think I will be able to use generic initialization such as
"{Asset} listOfAssets = {<item(X,i),i+5> | i in 1..10};"
or
"{CustomerUsesAsset} C4 = {<t1, t2> | t1 in listOfAssets, t2 in listOfAssets : t1.id not in X1 && t2.id not in X}"
because these relationships are specifically defined by business logic. For example, each asset has a very specific cost associated with it that is pulled from another data source. In terms of the {CustomerUsesAsset} relationships, I would like to only define the relationships that exist in the real world in order to cut down on memory consumption.

In regards to populating these relationships from a .dat file in the manner described:
{CustomerUsesAsset} C3 = {<<"f1", 6>, <"f2",7>, <"f1" and some asset #, does this create another customer object in memory or does it refer to the "f1" customer object that already exists in the {Customer} Customers list? I.E. when I create relationships between customer "f1" and some number of assets, I only want to store Customer "f1" data one time.
#DecisionOptimization
#OPLusingCPLEXOptimizer