search

N-Wouda / Euro-NeurIPS-2022

OptiML's contribution to the EURO meets NeurIPS 2022 vehicle routing competition.
Other
16 stars 2 forks source link

Do not shuffle operators before running LS #107

Closed N-Wouda closed 2 years ago

N-Wouda commented 2 years ago

This appears to be worse than taking the order as given, and is - I think - the cause of the performance regression on the static problem.

N-Wouda commented 2 years ago

This was a fun experiment, but I think it's worse than what we had before (i.e., there is a dependency between the ops, although it is small). So we might just be better off accepting the ops as given, including the order they came in.

N-Wouda commented 2 years ago

@leonlan to the new submission results (particularly for the static case). I'm trying a few different things to figure out what caused the difference. This random shuffle stuff is pretty high on that list :-).