It appears to me that what they've accomplished is to devise an algorithm that can, in polynomial time, find a solution to the ATSP with cost no worse than a co

It appears to me that what they've accomplished is to devise an algorithm that can, in polynomial time, find a solution to the ATSP with cost no worse than a constant multiple of the optimal solution. Can any domain experts confirm or deny my interpretation?