Originally posted by: TRAN Van Ut

Hello, 
I have the MILP, file mod, file dat in file ZIP. I try to code by many cases but I only obtain 50347.  

I can not have the really optimal value 50346. Please show me what is the error in my code.

Thanks in advance for your help!CPLEX.zip|View Details

int nbJobs = ...;
range Process = 1..nbJobs;
int p     [Process] = ...;
int d     [Process] = ...;
int w     [Process] = ...;
dvar boolean x[Process][Process];
dvar int+ h;
minimize h;
subject to
{        
    forall(k in Process)
      sum(j in Process)x[j][k] == 1;                           
    forall(j in Process)
      sum(k in Process)x[j][k] == 1;                          
    forall (k in Process)
      sum(l in 1..k)sum(j in Process) (p[j]*x[j][l])- sum(j in Process)(d[j]*x[j][k])<= 0;
     sum(j in Process)(w[j]*(sum(k in Process)k*x[j][k]))== h;
}

Hi,

can you change your epgap ?

I would try

execute

{

cplex.epgap=1e-6;

}

regards

relative MIP gap tolerance

Sets a relative tolerance on the gap between the best integer objective and the objective of the best node remaining.

 

 

 

 

Originally posted by: TRAN Van Ut

Thank you very much,

With the time limited and the old model, all are very good.

However, with below model, I don't understand why the solving start at a very big value (For example, with n = 20, all the case start at Best Integer  36100; or with n = 100,  start at Best Integer  2.13266e+007 ). In fact the acceptable objective value can not be bigger than n(n-1)/2.(n is  "nbJobs" in the code).

But with n=60, it starts with Best Integer 1157

So, in 600s, most of results with n>60 have the unacceptable values (i.e they are bigger too much then n(n-1)/2)   

range Process = 1..nbJobs;

int p     [Process] = ...;
int d     [Process] = ...;
dvar int+ nn[Process][Process];
dvar boolean x[Process][Process];
dvar int+ h;
minimize h;
subject to
{
    forall(k in Process)
      sum(j in Process) x[j][k] == 1;
    forall(j in Process)
      sum(k in Process) x[j][k] == 1;
    forall(j in 1..(nbJobs-1),k in 1..(nbJobs-1))
        nn[j][k] >= sum(jp in (j + 1)..nbJobs,kp in (k + 1)..nbJobs)x[jp][kp] - nbJobs* (1 - x[j][k]);
    h == sum(j in Process, k in Process) nn[j][k];
    forall (k in Process)
        sum(j in Process,l in 1..k ) p[j]*x[j][l]- sum(j in Process)d[j]*x[j][k] <= 0;
}

Thanks in advance,

Best regards,