Defining the planning horizon in the RCPSP

Problems with modeling
Post Reply
abb.omidi
User
User
Posts: 39
Joined: 6 years ago

Defining the planning horizon in the RCPSP

Post by abb.omidi »

Dear support team,

I am trying to write an optimization model related to the resource-constrained project schedule based on Pritsker et al. formulation. The model is as follows:
test.jpg
When I have defined the objective function as shown in the model, the final solution is incorrect but, when I write it in the following form, it seems to work fine.
The first attempt:

Code: Select all

z =e= sum(t,(ord(t))*x("end",t));
The second attempt: (works fine)

Code: Select all

z =e= sum(t,(ord(t)-1)*x("end",t));
Also, it is about the second constraint on loop for index (t2):
The first attempt:

Code: Select all

sum((i,tt)$(ord(tt)> (max(0,(ord(t))-p(i)+1)) and ord(tt)< ord(t)+1), d(i,r)*x(i,tt)) =l= cap(r);
The second attempt: (works fine)

Code: Select all

sum((i,tt)$(ord(tt)> (max(0,(ord(t)-1)-p(i)+1)) and ord(tt)< ord(t)+1), d(i,r)*x(i,tt)) =l= cap(r);
What I am interested to know is:
  • Why we will have to define the set index (t) from (t-1)?
  • In the second constraint to loop over (t2), why we will need to use (</>) in contrast to (<=/>=)? (In the second form, the solution is incorrect.).
It should be noted that in the python-based equivalent formulation, (with another solver and the same notation as the above original model), it works well without any changing in the (t) ordered.


Thanks in advance
Abbas
Post Reply