Currently it is not very optimal. See for example APEC 2010.
We should maybe select better suspicious cells.
When reaching a suspicious cell maybe we should first flood for target and then recalculate suspicious cell. Sometimes what was a good candidate is no longer.
In order to benchmark this we should generate statistics on all available classic mazes.
Currently it is not very optimal. See for example APEC 2010.
In order to benchmark this we should generate statistics on all available classic mazes.