The algorithm specifications - and potentially the rest of the library - use both forms for min and max. For example, \complexity At most \tcode{min(last1 - first1, last2 - first2)} applications... ([mismatch]) and \complexity ... \bigoh{\min(\tcode{last1 - first1}, \tcode{last2 - first2})} applications... ([alg.equal]). To me, it seems we should be using the "mathematical function" form for all cases that aren't actual code.