Terms more even for a teacher

[mikkojoo]

I'll check your file now, thanks

[mikkojoo]

So, I,m having just the same constraint as you. I only put days == 25, so it covers all days in our 5x5 days week.

It it solves, everything is ok with this. And if not, I will try something lower. Ok?

[Liviu Lalescu]

Yes, I think like you; it might be more efficient if you make min days = 25, rather than 0, as me, if all days are supposed to be non-KOKO-empty.

[Liviu Lalescu]

Yes, I think so.

[mikkojoo]

Thanks again!

I'm having 1242 out of 1558 in 2min 25s. It seems to solve, maybe in two hours I'd say.

[Liviu Lalescu]

I am generating on 8 threads, max is 1232/1558 after 30 minutes. So, it might solve. I am generating on the 0 min occupied days, because I want to check the behavior of FET.

[Liviu Lalescu]

I left it overnight. It generated, I attach the archive.

[mikkojoo]

Thanks, same here.
We got quite good timetable. Anyway, we decuded to have

max activities with tag KOKO = 2, days =25
and
min hours daily for all students =3

So, we can have R-activity in beginning and/or end AND min 2 KOKO activities in the middle. So all students will always have min 2 KOKO activities and possibly R-activity in both ends

[mikkojoo]

Good morning!
It seems to me that some constraints get broken, which is quite obvious with this file, but it does not give eny information about that in result statistics. Just wondering this.

[Liviu Lalescu]

Good morning!

Please share the locked file (file_data_and_timetable.fet), and tell me where do you think it is a bug (the broken constraints)! I'll check.

Do you mean constraints with weight less than 100% are broken and not reported in the soft conflicts, or constraints with weight 100% which are broken? What type of constraints?

[Liviu Lalescu]

Hello again!

I think I know your problem, because I checked your file and saw it. It is a feature, not a bug: you have 50% subject preferred room, generate first time on the unlocked time, and you have one such broken constraint (for activity A, which has no room). Then you generate again on the locked file - this is where the feature appears: FET will allocate this activity A to a good room, respecting the constraint of preferred room, because it finds that this room is free for A at A's time.

The reason is that during the first computation FET will swap activities and not try again to place A in a good room. The locking does not add any preferred room for A, but generating on a locked file will allow A to be placed in a suitable room, if possible.

[mikkojoo]

We have quite good timetable, some final editing still with constraints (mainly about teachers not available times).
It seems that we always get one 8th grade FY or KE not according to constraints:
max 2 terms (ok)
max 3 hours (ok)
max days between 9 (not) - to keep in two terms, no empty terms between
In this one FY 8D has an empty term between two terms. It is something with the constraints or/and complexity of the file, because the problem stays but group changes in every generation.

This is no big problem, just wondering why this happens.

[Liviu Lalescu]

I will check now.

[Liviu Lalescu]

No, Mikko, FET is right. You added max 10 days between those activities, and consider that it can be in day 1 and day 11; also, these days may be not first of each term.

I think you need to use max 1 terms between activities.

[mikkojoo]

Ok, thanks.
So, one between does not mean 'one empty between'?