home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!gatech!concert!duke!ola
- From: ola@duke.cs.duke.edu (Owen L. Astrachan)
- Newsgroups: comp.theory
- Subject: big-O, big-Theta
- Keywords: asymptotic analysis
- Message-ID: <716574072@mantis.cs.duke.edu>
- Date: 15 Sep 92 16:21:14 GMT
- Organization: Duke University Computer Science Dept.; Durham, N.C.
- Lines: 49
-
-
- Texts aimed at students in introductory courses most often use big-O
- notation exclusively, no mention is made of Omega or Theta. Often an
- informal definition of big-O is given, one that emphasizes "throwing out
- constants and ignoring all but the leading term in a polynomial
- expression".
-
- For example, in Hopcroft and Ullman's _Foundations of Computer Science_
- there is a substantial section on big-O including using big-O to give
- "tight big-O bounds". In Cormen/Leiserson/Rivest 's _Algorithms_ we see
-
- " Many people continue to use the O-notation where the $\Theta$-notation
- is more technically precise."
-
- I'm interested in whether there is a consensus as to the correct
- approach. The Advanced Placement Computer Science exam is revising it's
- curricular guidelines and using something perceived as anachronistic is
- not our most fervid desire. We run into trouble when we have multiple
- choice questions since we must do things like:
-
-
- < code or algorithm description >
-
- which of the following is the best description of the runtime of the ___
- above?
-
- (A) O(log n)
- (B) O(n)
- (C) O(n log n)
- (D) O(n^2)
- (E) O(n^2 log n)
-
- Note that we can't say "which of the following is correct", because
- picking the largest expression always works. Using Theta could
- alleviate some of these problems, but it would go against the practice
- employed in most intro books aimed at programming courses but which
- sometimes include informal asymptotic analysis.
-
- ----
-
- Thanks for any help, opinions, advice. I'd prefer email responses, I'll
- summarize to this group if there's interest.
-
- Owen
-
- Owen L. Astrachan internet: ola@cs.duke.edu
- Dept.of Computer Science Phone (919) 660-6522
- Duke University fax (919) 660-6519
- Durham NC 27706
-