Originally posted by: rkresnadi

Hi,
New to CP. I tried to enforce the precedence on my intervals and I got the following error: "Cannot use type string for <Ordid:string,TaskName:string,Worker:string,Duration:int>."

Here is my code:
{string} Ordid = ...; /* this is a query from a database from a table name called tblOrder*/

/*a tuple from a database from a table name: tbltask*/
tuple MyTask {
string Ordid;
string TaskName;
string Worker;
int Duration;
}

{MyTask} TaskData = ...;
int Duration d in TaskData = ...;

/*define precence*/
tuple Precedence {
string pre;
string post;
};

{Precedence} Precedences = ...;

/*define interval*/
dvar interval ordid o in Ordid in OrdTime[o]..(maxint) size 10;

dvar interval itvs o in Ordidt in TaskData size Duration[t];
/*CONSTRAINTS*/
subject to
{
forall(o in Ordid)
forall(p in Precedences)
endBeforeStart(itvs[o]http://p.pre, itvs[o]http://p.post);

This last line is where it gets the error from. Any idea? Thanks in advance.
#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

Hello,

I think that the issue is about the OPL syntax. Here is a trick that would solve it I guess:

1) Declare a key to the tuple My Task (note: that should be done only if the key uniquely defines a tuple)

tuple MyTask 
{ key string Ordid; string TaskName; string Worker; 

int Duration; 
}

Once this is done, it is possible to retreive an element of a set of MyTask, just by an order ID.
Because the second indexer of itvs is a set of MyTask, this feature can be used.

2) retrieve a element of the TaskData set for second indexer of itvs, using the

<key>

syntax

endBeforeStart(itvs[o][<p.pre>], itvs[o][<p.post>]);

Hope this helps.

Didier.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Didier,
Thank you for your help but the error is still there.

Do I need to modify the itvs declaration statement?
also, in my code I have int Duration d in TaskData = ...;. Do i still need this statement? If I don't declare it my itvs code couldn't find the "duration".

Thanks.
#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

Hi,

Do you mean that you still have the same error message ?

Cannot use type string 

for <Ordid:string,TaskName:string,Worker:string,Duration:int>

?

That would be strange because the type of the indexer is no longer a string in the example I gave you (it is a tuple of type MyTask).

I copied your code example in my environment and it worked well - I didn't have to change the definition of itvs.

Maybe there is an other issue I didn't see in your code - Can you copy your code statements between the code markups (see plain text help for all markups) - otherwise some characters of your code, such are brackets are transformed into something else (a link in this case) when your message is displayed.

Regards,

Didier.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Didier, I got that problem working now. I must have typed it incorrectly.

But I got another problem now. There are two key fields in that MyTask tupple: Ordid and TaskName and Cplex didn't like it because it has duplicate records under the Ordid field.

The data set looks like this:
Ordid TaskName
1 Task1
2 Task2
etc.

I may have to restructure the data. I just don't know how to do it yet.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Sorry, what I meant was:

The data set looks like this:
Ordid TaskName
1 Task1
1 Task2
etc.
#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

Then, if you need two fields to make a key and have only one in the precedence, it depends how you need to create the precedence constraints. For instance, if the preceddence constraints apply only for the tasks of a given order, it could look like (I assume that the string in a precedence is a taskname):

forall(o in ordid) forall (p in precedences) endBeforeStart(itvs[o][<o, p.pre>], itvs[o][<o, p.post>]);

But since taskData carries the orderId, it seems redundant to create the first indexer for the itvs intervals.

Didier.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Thanks Didier. It works fine without error, but as soon as I run the model I got another error: data element "Duration" not defined.

By the way, I created another tuple called OrderData. It is like this (since I thought this is the proper way):

/*From tblOrder*/
tuple MyOrder {
key string Ordid;
int OrdTime;
int EarlyArr2;
int LateArr2;
}
{MyOrder} OrderData = ...;

Now OPL does not like these syntax:
dvar interval ordid OrderData in (o in OrderData)http://<o.OrdTime>..(maxint) size 10;

dvar interval itvs TaskData size (t in TaskData)http://<t.Duration>;

by the way, how do you make the editor not to automatically create a URL?
#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

I cannot read your code - you didn't use the code markups and your text has been transformed.

I guess that the error is that you didn't declare the iteration variable (o or t) at the right place. It should be

dvar interval ordid [o in OrderData] in <o.OrdTime>..(maxint) size 10;

Didier,
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

it works, and now how to declare the itvs variable i tried to use similar method and didn
't work:   dvar interval itvs [t in TaskData] size [<t.Duration>];   now both interval ordid and itvs got errors: Formal parameter 
"t" not supported in 

this context, can not have iteration expression.

#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

I believe this is because the keyword size expects an integer (that will be evaluated for each interval). Not a tuple.

What about that syntax ?

dvar interval itvs [t in TaskData] size t.Duration;

Didier.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

OK, I changed those two inverval variables like you said:   dvar interval ordid [o in OrderData] in o.OrdTime..(maxint) size 10; dvar interval itvs [t in TaskData] size t.Duration;   Error is gone but I got another error (Error: not an array type) 

for my constraint:   forall(t in TaskData) forall(p in Precedences) endBeforeStart(itvs[t][<p.pre>], itvs[t][<p.post>]);

#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

This is probably because in your constraint you consider itvs as a 2 dimension array while you declared it as a one dimension array.

Didier.
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

yes but it did not work when i typed like 

this:   forall(t in TaskData) forall(p in Precedences) endBeforeStart(itvs[<p.pre>], itvs[<p.post>]);   the error says: 
"Cannot use type string as index in {MyTask}"

#CPOptimizer
#DecisionOptimization

Originally posted by: Didier Vidal

This is strange... The code below, that I took from your original example (I added dummy inline data) works for me with OPL 6.3

using CP;   
{string
} Ordid = 
{
"o1"
}; 
/* this is a query from a database from a table name called tblOrder*/   
/*a tuple from a database from a table name: tbltask*/ tuple MyTask 
{ key string Ordid; key string TaskName; string Worker; 

int Duration; 
}   
{MyTask
} TaskData = 
{ < 
/*comment*/ 
"o1",
"t1",
"w1",10>, < 
/*comment*/ 
"o1",
"t2",
"w1",20> 
}; 

int Duration [d in TaskData] = d.Duration;   
/*define precence*/ tuple Precedence 
{ string pre; string post; 
};   
{Precedence
} Precedences = 
{ < 
/*comment*/ 
"t1",
"t2">  
};   

int OrdTime[o in Ordid] = (o==
"o1") ? 3 : 1;   
/*define interval*/ dvar interval ordid [o in Ordid] in OrdTime[o]..(maxint) size 10;   dvar interval itvs [t in TaskData] size Duration[t];   
/*CONSTRAINTS*/ subject to 
{ forall(o in Ordid) forall(p in Precedences) endBeforeStart(itvs[<o,p.pre>], itvs[<o,p.post>]); 
}

#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Thank you Didier, I think the syntax is fine now. But now when I tried to solve I got "CP cannot extract expression" error. I think the error comes from the objective function. See all my updated code below:

using CP; 
/*From tblOrder*/ tuple MyOrder 
{ key string Ordid; 

int OrdTime; 

int EarlyArr2; 

int LateArr2; 
} 
{MyOrder
} OrderData = ...;   
/*From tblTask*/ tuple MyTask 
{ key string Ordid; key string TaskName; string Worker; 

int Duration; 
}   
{MyTask
} TaskData = ...;   tuple Precedence 
{ string pre; string post; 
};   
{Precedence
} Precedences = ...;   dvar interval ordid [o in OrderData] in o.OrdTime..(maxint) size 10; dvar interval itvs [t in TaskData] size t.Duration;     
/*OBJECTIVE*/ minimize sum(o in OrderData) (maxl(0, endOf(ordid[o]) - o.LateArr2)) + sum(o in OrderData) (maxl(0, o.EarlyArr2 - endOf(ordid[o])));   
/*CONSTRAINTS*/ subject to 
{ forall(o in OrderData) forall(p in Precedences) endBeforeStart(itvs[<o.Ordid,p.pre>], itvs[<o.Ordid,p.post>]); 
}

#CPOptimizer
#DecisionOptimization

Originally posted by: ChrisBr

Hello,

You should try to define your interval vars like this:

int maxinterval = (maxint div 2)-1; dvar interval ordid [o in OrderData] in o.OrdTime..(maxinterval) size 10;

following the OPL documentation:

dvar interval a [optional[(IsOptional)]] [in StartMin..EndMax] [size SZ | in SZMin .. SZMax] [intensity F]

Where:

int IsOptional, StartMin, EndMax, SZ, SZMin, SZMax; stepFunction F; -maxint/2 + 1 <= StartMin <= maxint/2 - 1 -maxint/2 + 1 <= EndMax <= maxint/2 - 1 0 <= SZ <= maxint/2 - 1 0 <= SZMin <= maxint/2 - 1 0 <= SZMax <= maxint/2 - 1

Hope this helps,

Chris
#CPOptimizer
#DecisionOptimization

Originally posted by: rkresnadi

Thank you. It works now.
#CPOptimizer
#DecisionOptimization