Posted by robusk on 8/1/2023 2:15:00 PM (view original):
As a general comment, I am testing out a new script leveraging some new ortools / a refactored objective function to see if I can improve upon our division distribution model and reduce occurrences of opponents sharing the same division twice. I will do some experiments and compare the outputs with the existing model and if it looks better we can replace it with the new output.
Something is off with how I have the constraints modeled... even though I warm start it with the existing solution, every time I run it, it always gets to point in its evaluation before it can't identify any new solutions even though its current one is not viable. Not sure why it gets hung up in the same spot every time.