I wrote some script to remove students with duplicate activities set from the Students list of each activties. However, I didn't remove the students from the Students list. So, I have a bunch (173/540) of students with no activities.
When generating the timetable, FET still tells me I have 173 duplicate sets, but the are really all empty sets. Does it still affect the speed of generation? Or impact just a little? Should I remove the students all together. I kind of want to keep the student names in the group even they don't have activities.
I think it does not affect the speed, or maybe 0.01%.
Maybe you could check and let us know as well. Test starting with the same random seed, with and without students, and check the seeds at the end to be the same.
Surprise result! The one with most (only 4 students with different deactivated activities were not removed) students removed from Students_List took 3:29:51. The one with 163 "duplicates empty" (all these students have no activities, but remained in Students_List) took a wooping 5:30:28!! There are 534 students in total.
I have done a diff between the files and confirm that the only differences are the extra "inactive" students in Students_List. The results generated are also identitical (I just manually checked a few students). Both generations uses all 1's in the custom seeding dialog.
I would attached the files for you to check... but don't see an attachment option here.
Thank you!
You can attach (archived if big), by pressing "Click or drag files here to attach them." right below.
I am interested if the random seeds after are identical.
Sorry, false alarm. I forgot that I had a charge limiter app running on my MacBook to keep my battery healthy. It basically takes the power off my computer constantly. So, when I walked away (which I did more on the 2nd run), it puts to computer into sleep mode or power saving mode... resulting in much slower generating time.
I ran them again, with full power. I ran them side-by-side, simultaneously. The one with 4 dups finishes in 2:33:04, the one with 16 dups finished just 50 sec behind.
I have kept an eye on the progress, both generation have thresholds at the exact same number of activities, with the one with 163 dups gradually falling behind as it progress.
Thank you for letting us know! But it is very easy to check that they have the same random seed after: there are two files, or you can go to view the random seed in FET GUI immediately after the generation.
To attach a file here, you need to press "Reply", otherwise the quick reply dialog seems not to have the option "Click or drag files here to attach them."
Hello
Quote from: yush on February 10, 2025, 04:52:40 AMSurprise result! The one with most (only 4 students with different deactivated activities were not removed) students removed from Students_List took 3:29:51. The one with 163 "duplicates empty" (all these students have no activities, but remained in Students_List) took a wooping 5:30:28!! There are 534 students in total.
Did you wait 3 hours and 5 hours, respectively, for successful generation?
I ask this because my average wait time rounds 40~50 minutes, and I think it is too much :o
My files has time constraints that isn't very optimized. In particular, I needed every students' schedule to end the day with a lesson, not a free period. On top of that, it is a very early version with zero optimization in terms of removing unneccesary constraints. But I remember the days when I was using i5 MacBook and the files takes over 20 hours to get a solutions... I have some pretty complicated timetable.