magical
910da239c0
i'm not sure if this has been described in the literature exactly but it is similar to existing branch-and-bound techniques. it doesn't speed up the search directly (the number of visited nodes is the same) but it does cut the queue length down by an order of magnitude and shave a couple seconds off of day 16, presumably due to reduced memory pressure or something. |
||
|---|---|---|
| .. | ||
| bad.py | ||
| input | ||
| sample | ||
| sol.py | ||