Asking the community for advice on how the best spread a number of activities over an 8 days cycle:
7: just 1 day min apart
6:
Method 1:
-- Take the last 4 activities as a group, add a 2 days min day constraint to the group
-- All 6 has a 1 day min days constraint
-- All 6 has a 6 days max days constraint (to avoid the first and last activities being back-to-back in the cycle
Method 2:
-- Add two dummy activities the make it a group of 8
-- Add 1 day min days for the group of 8
-- add 2 days min days for the dummy, to avoid the dummy (free day) being back to back
5:
Method 1: same as 6
Method 2: same as 6, just three dummies added
4: 2 days min day for the group, 6 days max day
3:
-- a1 and a3 min 5 days apart, max 6 days apart
-- a1,a2,a3 min 2 days apart, max 3 days apart (I think this will ensure that there will never be 3 consecutive free days)
My main question is, is method 1 or method 2 for the >5 cases more efficient for FET?
Or any other ideas?
Thanks.
I think your idea 2nd method with 6 activities + 2 dummy activities is "wrong", because in your variant it might happen that the 2 dummies are at the first and the last day. So in fact also "consecutive". So you might also add a max 7 days constraint for that 2 dummy activities. I guess this "fixed" method 2 is faster than method 1.
If you ask me: There are so many other things constraints that are more important, you should care about your idea only if you have got an simple data set and all other constraints are successfully added. Don't forget that there are so many things that can affect your "good spreading". Like an ill teacher, an ill student, a public holiday, a project day, ... . In my opinion you will spent too much time into spreading it "too well". Just use a simple min n day constraint and it is fine.
Would it be max 6 days, or 7 days as you said, for the dummies to avoid them being on first and last day
Wouldn't max 7 days allow this: x 2 3 4 5 6 7 x
I tried Method 2, the adding dummies approach, and find the all the dummy activities are placed at the end of the initial order. Wouldn't it make it very difficult to place those dummy to honor the constraints? If this dummy approach is to work, they would need to be considered as a group with the rest of the activities in that group, am I right?
Yes, I think Volker means max 6, not 7, for dummies.
I always prefer not to add dummy activities, they involve additional computation and are not elegant. But if you use them, yes, I think you should group them in the initial order with the real ones.
I would prefer for number 6 the Method 1.
I don't have other ideas.