Hi, everybody:
Still testing and trying to get as much as possible from FET. I have another question. Maybe someone nicely can help me save some time. Fingers crossed.
This is the issue. We have a room, let's say A4, which we want to be the prefered room for spanish language from all groups and years. The problem is the overall number of hours is greater than the 30 hours available weekly in this room.
My appropach. I added a constraint for the subject in each year to assign A4 as prefered room with 99% weight. Then for the same each subject, I added a 100% set of prefered rooms for those subjects. My thought was, ok, try with A4 and if not possible, use one room in the prefered set. The problem is it seems to really make the process very slow, specially when adding other constraints or trying to make some harder to get. I tried to lower the weight to 90% and it really seems to improve the time of generation, and get almost same results as with 99%, but still slow when adding more constraints.
So, I wonder, maybe is there a way that I still didn't think of doing the same thing, and get similar results without using the prefered room with a weight under 100?
Thanks in advance to everyone interested in this.
Hmm, I cannot think of another alternative. Try to lower to 75%.
Since i have got same problem, i just do it this way;
add 2 rooms with 100% room constraint.
as soon as i found a solution (since my timetable is very very difficult it take hours to get a solution; since in my opinion lower teacher gaps is more important then this minor room "problem"); i do this variant:
unlock all rooms and activate a second constraint now: this time with 99.999% preferred single room for that subject.
so now i get a solution in just a few seconds.
with this "trick" you will have in worst case only have of the activities in the preffered room.
you can avoid this by placing other (pseudo) activities into the second room if you search the first solution (so not in the second small optimation - in that case it is too late for this trick); so the room is 100% filled. In that case you will get even a perfect solution (but i don't do it that ways, since my dataset is too difficult).