spent about 40 minutes trying to optimize my spatial partitioning code again this morning. still a lot of mystery garbage collection caused by converting floats to ints even though the procedure that does so has been memoized. also hash table access may also be a problem? I can't really tell. maybe alists would perform better.