В Вск, 21/01/2007 в 19:13 -0500, Kamaraju Kusumanchi пишет: > > 2. In the worst case, your algorithm scales as O(N^2). If I am not wrong, you > can do this in O(N log(N)) steps. If your N is large (say > 1000) this has a > huge benefit. The algorithm would be > a. use a quicksort technique to sort the input array There is qsort in stdlib.h (man 3 qsort) Complexity O(N * logN) But if you need to do it yourself -- "Donald Knuth. The Art of Computer Programming, Volume 3: Sorting and Searching" will help you > b. compare side by side elements in the sorted array to determine for > uniqueness. + O(N)
Rather fast, at least faster, then O(N^2)
