Originally posted by: expir55

Originally posted by: GGR

Hi

The day off is part of the calendar constraints on your resource. You have several facilities in CP Optimizer for such a constraint please take a look to the page concepts Intensity and size and forbidden values on the CP Optimizer reference manual:

http://www-01.ibm.com/support/knowledgecenter/SSSA5P_12.6.2/ilog.odms.ide.help/refcppopl/html/interval_variables.html?cp=SSSA5P_12.6.2%2F1-4-1-1-37&lang=en 

and

http://www-01.ibm.com/support/knowledgecenter/SSSA5P_12.6.2/ilog.odms.ide.help/OPL_Studio/opllang_quickref/topics/tlr_oplsch_forbidExtent.html?lang=en

In Your case you must use an forbidExtent(x, F) constraint whose function F  is !=0 on the time segments at which the worker j is on dayoff and x one of the interval of agent i.

Hope that helps

PS By the way I am surprise your intervals are not optional?

Originally posted by: expir55

Hi Rogerie,

First of all, thank you so much for your answer.

I wrote the wrong decision variable (My decision variable contain "optional".)

"dvar assign interval [j in agents, ie in days, m in shifts] optional in the start [m] .. (start [m] + duration [m]) size duration [m]"

I have looked at the page and the concepts Intensity and size and forbidden values on the CP Optimizer reference manual.

But i could not apply these functions to my model.

I am writing as a constraint for dayoff (Because, I consider dayoffs of each agent (greater than equal to 1 day) for each 7 day weekly planning period. ):

"forall (j in agents) sum (i in days, m in shifts) presenceOf (assign [j][i][m])<=(Nbday-1)"

But there are conflicts in solution.

Must I write the decision variables as follows?:

dvar interval assign[j in agents, i in days, m in shifts] optional in start[m]..(start[m]+duration[m]) size duration[m] intensity F[j][i][m];

and stepfunction?:

stepFunction F[j in agents, i in days, m in shifts] = stepwise { 0->Nbday; 100};

where/how should I use forbidextent function?

I would be glad, if you may help me to solve.

Your sincerely,

Originally posted by: expir55

Additional information:

In my case's model decides on which agent is dayoff in which day (it is unknown that which agent is dayoff in which day).

It means my model have any data as follows:

"{Vacation} Vacations[employees] = ...;"

How can i use this function (forbidExtent(x, F))?

How can i write stepFunction?

I would be glad, if you may help me to solve.

Your sincerely,

Originally posted by: GGR

Hi

If I understand you, you problem is about resource timetabling. The schedule of the task being here to assure the feasibility of the timetabling. The right model depends upon how the task modeling.

Note I suppose when you say day-off, the worker does not actually work for all the day. In other word you.

I have realized that your alternative for task is indexed by the day. Does it means your time is discretized with a one day time unit for the task also. This would mean that the problem is rather a packing one and you should use a Mixed Integer Model.

If the time for operation cannot not be  discretized that means other constraints holds on or between tasks that cannot be told within a daily time unit. In practice we need the kind of constraints and objective on task to design the right model. For example it is not sure the alternative of day is required.

Hope that helps

Originally posted by: expir55

Hi,

Thank you for your interest.

Model gave a solution when i added  followings code to model:

"stepFunction Fassign[j in agents,i in days,m in shifts] = stepwise {0->start[m];100->(start[m]+duration[m]);0};"

"dvar interval assign[j in agents, i in days, m in shifts] optional in start[m]..(start[m]+duration[m]) size duration[m] intensity Fassign[j][i][m];"

Constraint:

forall(j in agents,i in days, m in shifts) forbidExtent(assign [j][i][m], Fassign[j][i][m]);

Thank you again.

Your sincerely,