The authors of this paper chose this specific optimization technique. Why do you think they did that, and what are the alternatives?
