Asking for a poly-time algorithm for a problem sets a (very) low bar when asking for efficient algorithms. The question is: can we achieve even this f