Do you have a paper on approximating NP complete problems and the bounds on the error in the solution?