Originally posted by: user1234567

Hello, I'm a beginner in cplex.

I'm trying to solve Milp model of FJSP problem. I referred to sched_jobshop_flex example for the FJSP problem.  The code is not running after adding the two variable constraints X and Y:

forall (m in Mchs, j in Jobs,o in Ops, o1 in Ops, o2 in Ops, m1 in Modes, m2 in Modes: m1.mch==m && m2.mch==m){

(o2.pos==1+o1.pos) => 
(Y[o1,o2,m] == 1);
}

forall (j in Jobs, p in Modes, o in Ops, m in Mchs){
(p.pt != 0) =>
(X[m,j,o] == 1);
}

//Also it is not running after adding the other constraints of the Milp model (attached).

//This is the current code. 

using CP;

tuple paramsT{
    int nbJobs;
    int nbMchs;
};
paramsT Params = ...;    
int nbJobs = Params.nbJobs;
int nbMchs = Params.nbMchs;

range Jobs = 1..nbJobs;
range Mchs = 1..nbMchs; 

tuple Operation {
  int id;    // Operation id
  int jobId; // Job id
  int pos;   // Position in job
};

tuple Mode {
  int opId; // Operation id
  int mch;  // Machine
  int pt;   // Processing time
};

{Operation} Ops   = ...;
{Mode}      Modes = ...;

// Position of last operation of job j
int jlast[j in Jobs] = max(o in Ops: o.jobId==j) o.pos;

dvar interval ops  [Ops]; 
dvar interval modes[md in Modes] optional size md.pt;
dvar sequence mchs[m in Mchs] in all(md in Modes: md.mch == m) modes[md];

 

dvar int Y[o1 in Ops, o2 in Ops, m in Mchs];
dvar int X[m in Mchs, j in Jobs, o in Ops];

execute {
          cp.param.FailLimit = 10000;
}

minimize max(j in Jobs, o in Ops: o.pos==jlast[j]) endOf(ops[o]);
subject to {
    // Precedence constraints between consecutive operations o1,o2 of a job j
  forall (j in Jobs, o1 in Ops, o2 in Ops: o1.jobId==j && o2.jobId==j && o2.pos==1+o1.pos)
    endBeforeStart(ops[o1],ops[o2]);

forall (m in Mchs, j in Jobs,o in Ops, o1 in Ops, o2 in Ops, m1 in Modes, m2 in Modes: m1.mch==m && m2.mch==m){

(o2.pos==1+o1.pos) => 
(Y[o1,o2,m] == 1);
}

forall (j in Jobs, p in Modes, o in Ops, m in Mchs){
(p.pt != 0) =>
(X[m,j,o] == 1);
}

    // Alternative machines for a given operation o
  forall (o in Ops)
    alternative(ops[o], all(md in Modes: md.opId==o.id) modes[md]);
    
    // Operations on a given machine m cannot overlap
  forall (m in Mchs)
    noOverlap(mchs[m]);
}

execute {
  for (var m in Modes) {
    if (modes[m].present)
      writeln("Operation " + m.opId + " on machine " + m.mch + " starting at " + modes[m].start);   
  }
}

tuple solutionT{
    int operation;
    int machine;
    int start;
};
{solutionT} solution = {<m.opId, m.mch, startOf(modes[m])> | m in Modes : startOf(modes[m]) != 0};

//Attached you can find the variables used in this problem, and the constraints of the milp model.

Thank you.

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

can you attach your .dat file too ?

When you say it is not running do you mean you get errors or you do not get any solution after a while ?

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: user1234567

It was running before adding the two binary varibales X and Y. But after adding the binary variables, I do not get any solution.

// this is the .dat file

 Params = <3, 3>;

Ops = {
  <1,1,0>,
  <2,1,1>,
  <3,1,2>,
  <4,2,0>,
  <5,2,1>,
  <6,2,2>,
  <7,3,0>,
  <8,3,1>,
  <9,3,2>,
};

Modes = {
  <1,1,4>,
  <1,2,0>,
  <1,3,5>
  
  <2,2,0>,
  <2,3,0>,
  
  <3,1,6>,
  <3,2,5>,
  
  <4,1,0>,
  <4,3,4>,
  
  <5,1,0>,
  <5,2,5>,
  
  <6,2,4>,
  <6,3,7>,
  
  <7,1,5>,
  <7,2,3>,
  
  <8,2,4>,
  
  <9,1,4>,
  <9,2,5>,
  <9,3,3>,
};

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: user1234567

At the same time, I get this error (always) even when cplex runs and posts the solution : 

Exception IBM ILOG Concert : CP optimizer solves problems with a search space up to 2^1000.

#DecisionOptimization
#OPLusingCPLEXOptimizer

On my machine, your model works fine.

You get that error because you seem to use CPLEX Community Edition that is free and limited.

If you are a student you could move to CPLEX Academic Initiative that is free too.

Otherwise you could also try CPLEX free Cloud edition : https://twitter.com/AlexFleischer1/status/869882932734959616

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: user1234567

Thank you very much for your reply.

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: user1234567

Hello, I want to write these two decision variable expressions using the same code above on cplex, I got some errors when I used the above syntax.

Xjom={1 if machine m is selected for operation Ojo,

0 otherwise.

Yjoj'o'm={1 if operation Ojo precedes operation Oj'o' on machine m,

0 otherwise.

 

forall(m in Mchs, o in Ops, md in Modes)
if (md.pt != 0)

X [o][m] ==1;

//For example, when i use this expression, I get no solution!

forall(m in Mchs, o in Ops, md in Modes)
if (md.pt != 0)

X [o][m] ==1;
else 
X[o][m]==0;

//And when I delete the else statement, I got solution with an objective function equal to 3,and X takes always the value 1 :

X = [[1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]
             [1 1 1]];

Please help me to write the syntax correctly (the expression of the descision variables) in order to get a correct solution. Thank you in advance.

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

in your model you wrote

dvar int X[m in Mchs, j in Jobs, o in Ops];

but then today you wrote:

X [o][m] ==1;

So your X moved from a 3D array to a 2D array.

It will be difficult for other users to help you without the full picture.

Do not hesitate to attach .mod and .dat

Regards

#DecisionOptimization
#OPLusingCPLEXOptimizer

Originally posted by: user1234567

Excuse me, I forgot to attach the new files. Here is the new files .mod, and .dat (I just made some minor changes).

I'm not sure if I should use 2D or 3D array in order to get correct values. That's why I tried both of the two cases.

 

 

#DecisionOptimization
#OPLusingCPLEXOptimizer

Hi,

you re right , you made a mistake in

forall(m in Mchs, o in Ops, md in Modes)
if (md.pt != 0)

X [o][m] ==1;
else 
X[o][m]==0;

For a given tuple m,o, suppose you have md1 in Modes such as md1.pt!=0 and md2 in Modes such as md2.pt==0

Then you have both X[o][m] ==1 and X[o][m] ==0, which is not possible

regards

#DecisionOptimization
#OPLusingCPLEXOptimizer